Re-assessing the Present Perfect - WordPress.com · Collins, Johan Elsness, Andreas Eriksson,...

Valentin Werner, Elena Seoane, Cristina Suárez-Gómez (Eds.)Re-assessing the Present Perfect

Topics in English Linguistics

EditorsElizabeth Closs TraugottBernd Kortmann

Volume 91

Re-assessing the Present Perfect

Edited byValentin WernerElena SeoaneCristina Suárez-Gómez

ISBN 978-3-11-044311-0e-ISBN (PDF) 978-3-11-044353-0e-ISBN (EPUB) 978-3-11-043532-0ISSN 1434-3452

Library of Congress Cataloging-in-Publication DataA CIP catalog record for this book has been applied for at the Library of Congress.

Bibliographic information published by the Deutsche NationalbibliothekThe Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie;detailed bibliographic data are available on the Internet at http://dnb.dnb.de.

© 2016 Walter de Gruyter GmbH, Berlin/BostonCover image: Brian Stablyk/Photographer’s Choice RF/Getty ImagesTypesetting: PTP-Berlin, Protago-TEX-Production GmbH, BerlinPrinting and binding: CPI books GmbH, Leck♾ Printed on acid-free paperPrinted in Germany

www.degruyter.com

Table of contentsAcknowledgements | vii

List of abbreviations | ix

Valentin Werner, Elena Seoane, and Cristina Suárez-Gómez0 Introduction: The present perfect – a re-assessment | 1

Part I: Diachronic and synchronic perspectives on the perfect in native varieties of English

Berit Johannsen 1 From possessive-resultative to perfect? Re-assessing the meaning of

[hæbb- + past participle] constructions in Old English prose | 23

Jill Bowie and Sean Wallis2 The to-infinitival perfect: A study of decline | 43

Markku Filppula3 Expression of the perfect in two contact varieties of English | 95

Sophie Richard and Celeste Rodríguez Louro4 Narrative-embedded variation and change: The sociolinguistics of the

Australian English narrative present perfect | 119

Part II: Perfects across varieties of English

Bertus van Rooy5 Present perfect and past tense in Black South African English | 149

Julia Davydova 6 The present perfect in New Englishes: Common patterns in situations of

language contact | 169

vi Table of contents

Elena Seoane7 The perfect space in creole-related varieties of English: The case of

Jamaican English | 195

Robert Fuchs 8 The frequency of the present perfect in varieties of English around the

world | 223

Valentin Werner9 Rise of the undead? be-perfects in World Englishes | 259

Part III: Building bridges

Robert Fuchs, Sandra Götz, and Valentin Werner10 The present perfect in learner Englishes: A corpus-based case study on L1

German intermediate and advanced speech and writing | 297

Björn Rothstein11 Afterthought: Some brief remarks on autonomous and speaker-centered

linguistic approaches to the present perfect | 339

Subject index | 351

AcknowledgementsThe present volume can very much be seen as a collective effort and would not have become a reality without the help of a great many people that were involved at various stages.

First of all, we would like to say thank you to the organizers of the ICAME 35 conference, titled “Corpus Linguistics, Context and Culture”, which was hosted at the University of Nottingham in April 2014. We were more than happy that Michaela Mahlberg and her team (Gavin Brookes, Kathy Conklin, Rachele de Felice, Dave Evans, Kat Gupta, Kevin Harvey, Tony Fisher, Lorenzo Mastropierro, Rebecca Peck, Ana Pellicer-Sánchez, Rein Ove Sikveland, and Viola Wiegand) offered us a venue and the opportunity to run a thematic pre-conference work-shop on corpus-based analyses of the present perfect and related constructions.

It is obvious that a workshop could not go ahead without the contributors. We thank all of them for creating a productive and friendly atmosphere, and in particular Stephanie Hackert, John M. Kirk, Gloria Otchere and Marije Van Hattum, who unfortunately could not contribute to this book directly, but whose presentations and comments helped to shape the ideas presented in the individ-ual chapters of this volume. All other authors are to be thanked for their efforts, which have gone into developing and revising their workshop contributions.

We also received invaluable support and advice from a number of scholars who volunteered as external and internal referees. For their constructive comments and recommendations we would like to thank Carolin Biewer, Laurel Brinton, Peter Collins, Johan Elsness, Andreas Eriksson, Stephanie Hackert, Raymond Hickey, Marianne Hundt, Jim Miller, Joseph Roy, Edgar W. Schneider, Julia Schlüter, Norbert Schlüter, Nicolas Smith, Marije Van Hattum, Gerard Van Herk, Jim Walker, Xinyue Yao, and Yuri Yerastov. It is not overstated to claim that the expertise of this for-midable group could easily fill a second volume on the present perfect. In addi-tion to the burden of their own contributions, Jill Bowie, Julia Davydova, Robert Fuchs, Sophie Richard and Celeste Rodríguez Louro acted as referees for other chapters.

For help and advice during the publication process, we would like to extend our gratitude to the De Gruyter Mouton team, in particular to Julie Miess and Birgit Sievert, who dealt extremely quickly and efficiently with our requests and ensured that everything ran smoothly. The series editors, Elizabeth Closs Trau-gott and Bernd Kortmann, have supported the project from its fledgling stage. The latter, in charge of the present volume, has provided numerous diligent com-ments on both structure and content, which have greatly improved it. We are proud that the book became part of their Topics in English Linguistics series.

viii Acknowledgements

Finally, for generous financial support, we would like to express our grat-itude to the Spanish Ministry of Economy and Competitiveness (grants FFI FFI2014-53930-P and FFI2014-51873-REDT), the Regional Government of Galicia (grant GPC2014/060), and the University of Bamberg (FNK travel grant).

List of abbreviationsAmE American EnglishAusE Australian EnglishBanE Bangladesh EnglishBrE British EnglishBSAfE Black South African EnglishCanE Canadian EnglishEAfE East African EnglishGhaE Ghanaian EnglishHebE Hebridean English HKE Hong Kong EnglishIndE Indian EnglishIrE Irish EnglishJamC Jamaican CreoleJamE Jamaican EnglishKenE Kenyan EnglishMalE Malaysian EnglishNigE Nigerian EnglishNZE New Zealand EnglishPakE Pakistan English PhiE Philippine EnglishSAfE South African EnglishScE Scottish EnglishSinE Singapore EnglishSLE Sri Lanka EnglishStE Standard EnglishTanE Tanzanian English

Valentin Werner, Elena Seoane, and Cristina Suárez-Gómez0 Introduction: The present perfect –

a re-assessmentIt may be argued that the present perfect (henceforth PP) has never ceased to be a highly topical issue in English linguistics, the main general rationale being the central position of the notion of temporality as an anthropological fundamen-tal (Comrie 1985: 7). As a consequence, formal means for establishing time ref-erence in natural languages commonly form an essential part of grammars, with reference to the past representing an important subfield. Similar to a number of related languages, chiefly morphosyntactic structures are used in English to encode reference to the past, the PP being no exception.

However, this alone does not explain the eminence of the PP in the linguis-tic literature on English. Its prominence in research on grammar is likely also due to its “Januslike nature” (Michaelis 2002: 10) in multiple respects. Therefore, debates on its theoretical status as a grammatical category and on its potential semantic interpretations have a longstanding tradition (see, e.g., Labov 1978: 13; Veloudis 2003: 385; Klein 2009: 54).

As to the former, the main area of disagreement is whether the PP from a theoretical point of view should be seen as a tense or an aspect, and a multi-tude of arguments have been presented in favor of one or the other position (see Werner 2014: 50–56 for an overview). Most notably, this contrast is represented in differing positions taken in influential reference grammars of English, with The Cambridge Grammar of the English Language (Huddleston and Pullum 2002: 116, 142) advocating a “perfect as tense” analysis¹ and the Comprehensive Grammar of the English Language (Quirk et al. 1985: 189) as well as the Longman Grammar of Spoken and Written English (Biber et al. 1999: 460) advocating a “perfect as aspect” view. Others consider both categorizations descriptively inadequate and therefore have proposed alternative labels such as “phase” (Joos 1964), “status” (Bauer 1970), “perfect qua perfect” (Vermant 1983; Kortmann 1995) or “ori-entation” (Kortmann 1991). While we cannot discuss the implications of these approaches in detail, suffice it to say that the majority of these labels are set on a par with tense and aspect, so that an additional grammatical category is estab-lished (Werner 2014: 57). A radically different approach is presented in Hübler (1998). Hübler proposes that the PP, rather than assigning it to a grammatical

1 A subordinate point of disagreement here is whether the PP should be viewed as a present (e.g. Jespersen 1931; Reichenbach 1947) or past tense (e.g. Huddleston and Pullum 2002) or whether it merits a separate tense label altogether (e.g. Declerck 2006).

2 Valentin Werner, Elena Seoane, and Cristina Suárez-Gómez

category, should be seen as a pragmatic device expressing “adoptive attachment” (Hübler 1998: 127) of the speaker to a situation, a function that is especially salient in older stages of English, but thought to be relevant in present-day forms as well.

As noted above, a second contested ground is the issue of how many seman-tic interpretations or “readings” can be assigned to the PP. We can broadly dis-tinguish between monosemous and polysemous accounts (with up to seven different shades of meaning being claimed; see Werner 2014: 72 and additional discussion in Yao 2014: 29–37), while recent assessments of the applicability of these predominantly introspective approaches have been critical (Werner 2014: 106; Yao 2014: 54).

A less long-standing, yet equally salient theoretical issue is the “present perfect puzzle”, first raised in Klein (1992). The “puzzling” thing about the English PP from a typological view is the constraints that operate on its combinability with a number of definite temporal adverbials (e.g. last week/month/year, three hours ago, yesterday, etc.),² given that a similar situation is not observable for a range of European languages (e.g. French or German) where perfects have extended their functional scope at the expense of traditional narrative tenses. This incompat-ibility has also been the focus of later studies such as Boogart (1999), Kiparsky (2002), Portner (2003), and Rothstein (2008), and has been explained in terms of a “current relevance” or “extended-now” property specific to the PP.

Against the backdrop of this long-standing tradition of theoretical linguis-tic investigations, it is conspicuous that the PP (and related constructions, see below) have again received increased scholarly attention in recent years, both in terms of individual papers and book-length studies. Corpus linguists have been particularly active, using empirical data both to revisit the issues presented in the foregoing and for opening the doors to new areas of enquiry.

The starting point for these re-assessments is the wider availability of corpora and the technical opportunities to access and analyze these electronic data from the 1990s onwards. An early proponent (see also Meyer 1992, 1995) of this first wave of re-assessments is Elsness (1997), who studies the alternation between PP and simple past (SP) usage by examining material from Brown-family corpora, from the Survey of English Usage, and a specifically compiled historical corpus. He also includes experimental elicitation tests to complement the corpus perspec-tive, and identifies frequency differences between American (AmE) and British English (BrE) – overall the more PP-friendly variety – and between different text

2 This typological oddity has resulted in a number of publications contrasting perfects in En-glish with comparable forms in other languages (see, e.g., Engel 1998 for French; Boogart 1999 for Dutch; Miller 2004a for Russian; Molsing 2006 for Portuguese; Rothstein 2008 for Swedish and German).

Introduction: The present perfect – a re-assessment 3

types. In addition, he diagnoses an overall decreasing frequency of the PP from a diachronic perspective.

In contrast, Wynne (2000), in another corpus-based analysis of American and British material, argues that PP frequencies are not decreasing when truly variable contexts are investigated (that is, instances where the PP and the SP are in free variation without any shift in meaning; see Van Herk 2010: 50–51 for discussion). Relating to the debate on the grammatical status of the PP as either tense or aspect, Wynne develops a multi-layered alternative, including a pragmatic dimension. He also considers the presence of temporal specification through adverbs and finds that patterns in AmE and BrE are similar.

Likewise, Schlüter (2002) compares AmE and BrE, for instance with the help of Brown-family corpus data and material from the London-Lund Corpus and the British National Corpus, and develops a corpus-based alternative to introspective semantic models for the PP, where he identifies reference to an indefinite point in time as the main function of the PP. He further finds differences in mode in terms of higher PP rates in speech (see also Miller 2000) and claims that register effects on PP rates are stronger than effects of regional variation.

A more recent influential study in the tradition of contrasting the two major native varieties of English is Hundt and Smith (2009). It considers short-term change in the development of the PP by way of investigating PP and SP frequen-cies mainly in 1960s and 1990s data from the Brown-family corpora (with spoken material from other sources), and shows that variation between the PP and SP can be described as “stable layering” (Hundt and Smith 2009: 58), as PPs overall have only marginally decreased and PP/SP ratios have remained constant over the period studied. However, the study concurs with previous ones in empha-sizing the influence of regional and register effects (see also Biber and Conrad 2009). In addition, it finds that variation between the PP and the SP is pervasive in contexts temporally specified through indefinite adverbials, and that viola-tions of the non-combinability constraint with definite temporal adverbials (as discussed in Klein 1992) may occur, particularly in spoken data.

In sum, studies contrasting PP usage in AmE and BrE evidently have to be credited for informing studies of other varieties (see below). They have also con-tinued to spark scholarly interest, as new corpus material (such as the Corpus of Historical American English [COHA] and the Corpus of Contemporary Ameri-can English [COCA]) has allowed the development of additional diachronic and synchronic perspectives (see, e.g., Elsness 2009a, 2014; Bowie, Wallis, and Aarts 2013), and as increased corpus sizes have facilitated the description of quantita-tively more marginal phenomena such as infinitival (Bowie and Aarts 2012) and non-present perfects (Yao and Collins 2013a).


Quite naturally, however, a second wave of re-assessments has emerged. This is a result of the broadening of the research focus of English linguistics beyond AmE and BrE, notably involving the analysis of a variety of additional corpus materials – most prominent among them the components of the International Corpus of English (ICE). Studies either extend the view to single postcolonial varieties of English or provide contrastive views of multiple varieties of different types, occasionally including learner Englishes.³

On a related note, it is worth mentioning that these studies often employ a broader conceptualization of “perfect” to be better able to account for the enve-lope of variation. This means that researchers do not only take a form-to-function approach and restrict their analyses to the canonical variant have + past parti-ciple (Biber et al. 1999: 467) and its alternation with the SP, as was often done in earlier studies. In contrast, they use function-to-form approaches and consider the full range of formal variants used in the “perfect space”, that is, in contexts where a different (non-standard) surface realization is used to convey a meaning typically associated with the PP in standard varieties.⁴

To exemplify this briefly, it is helpful to first have a look at a set of single-va-riety studies with a function-to-form approach. Most notably, the perfect space of Irish English (IrE) has extensively been discussed (see, e.g., Siemund 2004; McCafferty 2005; Kirk and Kallen 2007; O’Keeffe and Amador Moreno 2009; Pietsch 2009; Van Hattum 2012), and six different formal variants have been identified for this variety, potentially emerging through Irish substrate influence. It has been suggested that these variants cover different semantic-pragmatic niches, so that theoretical semantic models may have to be revised and adapted in the face of such empirical data.

Further single-variety studies have been conducted both for other native and for indigenized postcolonial varieties. As to the former group, besides IrE, Cana-dian English (CanE) has received scholarly attention. For instance, Roy (2014) studies adverbial specification of the PP in CanE. To this end, Roy analyzes mate-rial from the Quebec English Corpus and further considers the issue of grammati-calization of the PP, thus offering a perspective of its diachronic semantic exten-sion. Furthermore, Yerastov (2015) analyzes the transitive be-perfect in CanE from a construction grammar point of view and contrasts it with related constructions.

3 Note that the PP usually also receives considerable attention in general grammatical descrip-tions of individual postcolonial varieties, such as Hundt (1998) on New Zealand English, Filppu-la (1999) on Irish English, or Bautista (2008) on Philippine English.4 This does not imply, however, that studies on the perfect space in AmE and BrE do not exist at all. See, e.g., Kjellmer (2003) for a study of forms with auxiliary ellipsis based on the Cobuild Corpus.


The alternation between the PP and the SP in Australian (AusE) and New Zealand English (NZE) have also been investigated, for instance synchronically in Elsness (2009b), both synchronically and diachronically in Yao (2014, 2015), and in a series of studies (e.g. Engel and Ritz 2000; Ritz and Engel 2008; Ritz 2010; Rodríguez Louro and Ritz 2014) devoted to narrative uses of the PP in domains such as police reports, radio interviews and personal narratives. Again, it is evident that data for this variety yield a different picture than what earlier accounts of the incompatibility of the PP with definite temporal adverbials have suggested.

Single-variety studies for indigenized (L2) Englishes are rarer, but do exist. Take, for instance, Winford (1993) on the PP in Trinidadian English, Davydova (2013) on the perfect space in Singapore English (SinE) and its relation to IrE (see above) as an input variety, Elsness (2016) on past-referring verbs in South African English, and Werner and Fuchs (2016) on the PP/SP alternation in Nigerian English.

Scholars recently have been quite prolific as regards studies in which mul-tiple varieties are compared and contrasted. Some of these studies (Seoane and Suárez-Gómez 2013; Suárez-Gómez and Seoane 2013) focus on specific regions, such as South East Asia (Hong Kong [HKE], Indian [IndE], SinE, and Philippine English [PhilE]), and try to determine whether substrate influence on the perfect space can be traced or whether other factors such as learner mechanisms or diffu-sion also exert some influence. A group of others (Hundt and Biewer 2007; Biewer 2008) investigates the PP/SP alternation as a case study to test whether geograph-ically close native varieties (AusE and NZE) emerge as new “epicenters” for South Pacific varieties (SinE, PhilE, Fiji, Samoa, and Cook Islands English). In addition, these studies also find variation between the PP and the SP in contexts specified by indefinite temporal adverbials and identify occurrences where the temporal adverbial constraint is violated.

Yet another group considers sets of varieties that differ in their sociolinguis-tic status and are regionally more diversified. Van Rooy (2009), in an analysis of BrE, HKE and East African English (EAfE), explicitly re-addresses the issues of the grammatical status as well as of semantic categorizations of the PP, and develops an alternative corpus-based semantic model. Van Rooy further suggests that central meanings of the PP are more tense-like, while extended meanings are more aspect-like. Therefore, he tags the arguments surrounding the grammatical status and the semantics of the PP as being part of a “fake debate” (Van Rooy 2009: 329), and proposes that we can arrive at a better understanding of English grammar in view of findings from indigenized varieties.

Davydova (2011) broadly conceives of “non-native Englishes” and investigates the perfect space in both indigenized Englishes (EAfE, IndE, SinE) and learner Englishes, with BrE used as reference variety. This study views the canonical PP


as a complex construction only mastered to a certain extent in non-native variet-ies. However, there is a trade-off in terms of a complex perfect space in non-native varieties with an extended number of surface variants, and a variety of factors determining the choice between these variants is tested. Davydova (2011) further advocates a unified analysis of indigenized and learner Englishes, as there seems to be a continuum along variety types rather than absolute distinctions, and crit-icizes earlier semantic and pragmatic accounts of the PP as already mentioned.

Large-scale cross-variety comparisons have been presented in Yao and Collins (2012) and in a series of studies by Werner (2013, 2014, 2016, forthcoming). The former investigate the alternation between the PP and the SP in ten native and indigenized varieties of English and try to determine the degree of PP-friendliness of these varieties. AmE and BrE emerge as the least and most PP-friendly varieties respectively, so that lower PP-friendliness values, particularly in temporally spec-ified contexts, may be interpreted in terms of Americanization (Yao and Collins 2012: 399). In a related study (Yao and Collins 2013b), they moreover indicate the fuzziness of established semantic categories and describe narrative uses of the PP, thus providing additional evidence for leveling between the PP and the SP in favor of the PP.

Variation in the amounts of PP and SP rates in adverbially specified contexts in twelve varieties is the focus of Werner (2013), who further identifies a relax-ation of the definite adverbial constraint in natural language, particularly in the speech of indigenized varieties of English (see also Rastall 1999). Werner (2014) considers internal variability of the data in terms of effects of mode of discourse and register on PP rates. In this study, it is found that written text types overall are much more variable, also as regards distributions of contextual factors (such as Aktionsart of the main verbs, sentence type, etc.) and that variety types only exert a limited influence. In addition, alternative surface realizations in the perfect space as a kind of spoken and informal “register versal[s]” (Werner 2014: 377) are investigated.

PP research we have not mentioned up to this point are diachronic studies as well as analyses of the PP in minor non-standard varieties and learner Englishes. While there is some disagreement on the exact time of emergence of the PP as a periphrastic form (Rissanen 1999: 215; Werner 2014: 20), issues pertaining to the diachronic development and grammaticalization of the PP have been debated much less often compared to the abundance of publications on synchronic issues. Thus, observers appear to agree that PP usage patterns had regularized at least from the end of the Early Modern Period onwards (Werner 2014: 21).⁵ The alter-

5 Note that earlier stages of the English language had a two-tense system merely distinguishing past vs. non-past (preterite vs. non-preterite).


nation between the PP and the SP in historical stages and the related grammat-icalization and frequency increase of the PP (e.g. Tagliamonte 2000) as well as the recession of variants such as the be-perfect (McFadden and Alexiadou 2010; Anderwald 2014; McCafferty 2014) have received some attention, however.

Both synchronic and diachronic studies of traditional regional dialects and informal speech have illustrated the alternation between surface variants in the perfect space (see, e.g., Miller 2000, 2004b; Van Herk 2008), and the influence of creole structures has also been discussed (see, e.g., Winford 1993; Tagliamonte 2000; Poplack and Tagliamonte 2001).

Another particular type of non-standard usage in an informal domain, some-what parallel to the narrative use of the PP in AusE, is the “footballer’s perfect” in BrE, that is, PP use in post-match interviews where the SP would typically be expected (Walker 2008, 2011). On the basis of his findings, Walker (2011) posits the narrative use as an additional semantic reading of the PP, and speculates on the reversal of the functional split between PP and SP.

Learner Englishes constitute a final area where recent contributions have given rise to re-assessments of earlier views.⁶ The PP is commonly considered as a “learner-hard” feature, since the complex morphosyntactic surface form has to be learned in addition to adequate semantic and pragmatic usage contexts (Slobin 1994; Bardovi-Harlig 1997, 2000; Liszka 2003). It seems intuitively plau-sible that the first language of the learners constitutes an important variable in the learning process. However, recent corpus work, building on resources such as the International Corpus of Learner English (ICLE; see, e.g., Eriksson 2008; Rogatcheva 2014) and other data (e.g. Davydova 2011) has shown that the effects of the native language of the learners may be contrary to expectations, as native languages typologically and structurally close to English do not necessarily lead to faster and more accurate target-language production of the PP.

All in all, what this extensive (yet, in all probability, still not exhaustive) liter-ature review of newer works shows is that, in general, both the area of the PP and the perfect space constitute lively fields of linguistic research from a multitude of theoretical angles and, more specifically, that variation in this realm of English grammar is far more pervasive than assumed in earlier studies.

Corpus-based and other work on the PP has led to a number of re-assess-ments and refinements of earlier theory (for instance, as regards grammatical and semantic categorizations of the PP, or the association of particular temporal adverbs with either the PP or the SP), and has established new areas of debate: should the PP be seen as a receding form or rather as one increasing its functional

6 A related topic not treated here is the (late) first-language acquisition of the PP (see, e.g., the classic studies Nussbaum and Naremore 1975 or Fletcher 1981).


scope, as the definite temporal adverbial constraint seems to be weakening in some varieties? How pervasive are alternative surface realizations in the perfect space? Which role do register effects and lexical effects play? We further submit that going beyond AmE and BrE as well as exploiting the affordances of struc-tured corpus data has enabled researchers to take a broader, descriptively more adequate perspective on and, at the same time, to present a more fine-grained picture of English grammar.

The contributions in this volume can be seen in this spirit and represent a third wave of re-assessments for one principal reason. It is a commonplace that new data permit new linguistic insights, but this has never been truer than at present. The (English) linguistics community at large, including students of the PP and the perfect space, now is in the fortunate situation to be able to rely on a number of resources that allow access to naturalistic data of a size and scope not known before. To mention just a few examples, Mark Davies’s COCA, COHA (chapters by Bowie and Wallis; Johannsen), and the Corpus of Global Web-based English (chapters by Fuchs; Johannsen; Werner) exploit the opportunities offered by optical character recognition and extensive digitization of archives (e.g. from newspapers or literature) as well as the widespread availability of language data from the Internet. Not only are these resources large in size, which facilitates the analysis of more marginal phenomena (chapters by Bowie and Wallis; Werner; see also Hundt 2016). Arguably, they are also much more accessible and user-friendly than traditional offline corpus tools, as open access is provided through online interfaces, and they widen the perspective to electronic registers that have not been included in other investigations.

At the same time, it goes without saying that there are other invaluable resources for probing the area of grammatical variation. The list comprises smaller corpora, both diachronic (chapters by Johannsen; Van Rooy) and synchronic, such as the ever-growing ICE-family (chapters by Bowie and Wallis; Davydova; Seoane; Van Rooy; Werner), which still can be seen as highly instrumental for the study of individual native and indigenized varieties as well as for cross-variety comparisons. A family of corpus components on learner Englishes that is similarly vibrant and constantly expanding is represented by ICLE (chapter by Fuchs, Götz, and Werner). Specialized and self-compiled corpora serve an additional import-ant role as they allow researchers to access patterns that would otherwise remain hidden in particular (sub-)varieties (chapters by Filppula; Richard and Rodríguez Louro; Van Rooy) and registers, such as casual conversation (chapters by Richard and Rodríguez Louro; Van Rooy). On a related note, it is clear that relying on part-of-speech tagged corpora has become a standard procedure in data preparation to increase the sophistication of corpus queries, leading to greater accuracy in identifying relevant occurrences and consequently more reliable results.


Special mention needs to be made of another open-access resource, the elec-tronic World Atlas of Varieties of English (eWAVE; Kortmann and Lunkenheimer 2013). Its main purpose is to allow the assessment of morphosyntactic variation from a typological perspective, both in terms of attestation rates of grammatical features across different varieties of English as well as in terms of pervasiveness, that is, ratings on the obligatory or non-obligatory nature of a particular feature. The perfect space figures prominently in the tense and aspect section (see also Lunkenheimer 2012) of eWAVE (features 97–102, 104–109). This enables scholars to compare their empirical findings to the ratings provided by the database. A number of contributions in the present volume capitalize on this opportunity (chapters by Davydova; Filppula; Fuchs; Fuchs, Götz, and Werner; Richard and Rodriguez Louro; Seoane; Van Rooy; Werner), thus allowing the results to be better contextualized.

With the help of these resources, and by way of using both well-established (e.g. testing for statistical significance and effect size, variable rule analyses) and innovative statistical methods (e.g. logistical regression modelling with random factors) and related visualizations (such as dot plots and graphs including infor-mation on confidence intervals), this volume can be viewed as representative of the state-of-the-art in research on variation in the perfect space from different methodological and theoretical angles (see the afterthought by Rothstein) in a multitude of varieties.

The first part of the book contains diachronic and synchronic views on the perfect in native varieties of English.

The opening chapter is Berit Johannsen’s “From possessive-resultative to perfect? Re-assessing the meaning of [hæbb- + past participle] constructions in Old English prose”, which aims at challenging the traditional account of the development of the English PP as a change from a possessive-resultative con-struction into a temporal-aspectual one with perfect-anterior meaning. For this purpose, the author adopts a semasiological perspective to re-assess the meaning of the [hæbb- + past participle] structure in the large York-Toronto-Helsinki Parsed Corpus of Old English Prose, which contains nearly 1.5 million words. Her hypoth-esis is that a semantic shift from resultative to perfect-anterior meaning can be observed in Old English data but only for some of the five semantically distinct [hæbb- + past participle] structures she finds. At the heart of this contribution, we find the methodological claim that we need to adopt a systematic, corpus-based approach to the study of early stages of English rather than base our discussions and conclusions on singular, isolated sentences. She illustrates this claim with a “thought experiment” in which the reader is faced with real Present-Day English examples illustrating how ambiguous have + past participle structures can be,


and, by extension, how imprecise it can be to draw general conclusions from indi-vidual examples.

Also from a diachronic perspective, Jill Bowie and Sean Wallis explore a fairly neglected perfect construction, the to-infinitival perfect (e.g. She is said to have seen it). In “The to-infinitival perfect: A study of decline”, they focus on its dis-tribution in written American English over the last 200 years, and find that its frequency against the baseline of PP and SP verbs decreases dramatically. Using the parsed ICE-GB to automatically obtain the most common contexts where the construction appears, Bowie and Wallis establish that to-infinitival perfects feature mainly in catenative constructions, and examine these in COHA. In this 400 million-word corpus, 30 catenative verbs account for 95 % of the examples, and these verbs are classified into four broad semantic groups. The authors show that whereas the “seeming” verbs (seem, appear), by far the most common of all, and the “prospective” verbs (mean, wish) decline according to a logistic curve, the other categories do not follow suit and exhibit varying patterns of decline. They also explore possible patterns of replacement for the to-infinitival perfect as possible determinants for its decline, and find that only some semantic groups and verbs have clear alternants, such as the broader “prospective group”, which tends to use a non-perfect infinitival complement now (I expected to have been vs. I expected to be). In an overall assessment of the results, the authors conclude that they illustrate how “grammatical change tends to be lexically constrained and gradually diffusional in nature”, thus in keeping with the postulates of con-struction grammar.

Turning to synchronic studies, Markku Filppula’s “Expression of the perfect in two contact varieties of English” explores and contrasts the perfect space in two high-contact varieties, the traditional vernacular dialects of Ireland (IrE) and western Scotland (Hebridean English, HebE), as represented in recordings made with speakers of these varieties. Both are shown to exhibit a complex mixture of have-perfect and non-standard means (SP, present – simple and continuous – and the after-perfect), non-standard means being restricted to express indefi-nite-anterior, extended-now, and hot news perfect meanings respectively. Filp-pula examines his results against the backdrop of previous theoretical accounts of language contact. Most notably, his results run counter to Trudgill’s (2011) claim that high contact should lead to simplification, since Filppula’s examples show explicit processes of complexification at work. However, his findings do corrobo-rate the generalization proposed by Davydova et al. (2011) that the have-perfect is closely associated with resultative and extended-now contexts, whereas in expe-riential and recent past contexts this association is less robust. Overall, what his study most strongly brings to the fore is that it is continuing influence from the substrate Celtic languages (Irish and Scottish Gaelic) that shapes the grammati-


cal systems of IrE and HebE, an influence which overrides other factors, such as geographical proximity.

Another synchronic study closes the first part of the book. Sophie Richard and Celeste Rodríguez Louro’s chapter “Narrative-embedded variation and change: The sociolinguistics of the Australian English narrative present perfect” carries out an exhaustive quantitative sociolinguistic analysis of the narrative PP in a corpus of 220 performed Labovian narratives produced by 57 AusE speakers. The authors demonstrate that a variety of linguistic factors constrain the use of the narrative PP and offer a detailed comparison of their findings with earlier research. The most important of these are (i) priming, (ii) occurrence with third person subjects, (iii) pragmatic context, since the narrative PP tends to appear in the middle of the complicating action due to its foregrounding function, and (iv) lexical effects. As for the latter, there is a pronounced tendency for the narrative PP to occur after quotative go (but not after other quotative verbs) and with pred-icates expressing atelic activities. The authors take a step further to examine and underscore the role that social factors play in the use of the narrative PP in AusE. This construction features prominently in the speech of non-professional males (who are also associated with the use of quotative go) and older speakers, which is interpreted as a sign that the narrative PP is not a change in progress, but a socially (and grammatically) constrained phenomenon.

In the second part of this volume, the perfect space is considered in indige-nized varieties of English and from a cross-variety perspective.

Exploring the “Present perfect and past tense in Black South African English”, Bertus van Rooy considers register differences in the expansion of the SP, con-trasting the language of newspapers (from an ICE-equivalent source) as a heavily edited text type with unmonitored conversation (from the Xhosa-English Corpus) and second-language student writing (from the Tswana Learner Corpus). At the same time, he provides a diachronic perspective on newspaper language from the end of the nineteenth to the twenty-first century. Surprisingly, his initial hypoth-esis that a normative constraint operating in edited newspaper texts should favor higher PP rates is not borne out by the data. In fact, PP rates are decreasing in newspaper writing over time, and are comparatively high in the other two regis-ters in the synchronic view. Van Rooy thus concludes that news writing may be leading change, and that a psycholinguistic process of “explicitation”, that is, increased reliance on the analytic PP form due to easier processing, may operate in conversation and student writing. Van Rooy’s results for Black South African English further support findings of studies showing (i) leveling between the PP and the SP (in both directions) and (ii) a correlation between reduced PP-friend-liness values and progress along the phases of Schneider’s (2003, 2007) Dynamic Model for other varieties.


In the chapter “The present perfect in New Englishes: Common patterns in situations of language contact”, Julia Davydova deals with the perfect space in three varieties, namely Indian, Singapore and East-African English. She uses spoken data from ICE, and compares them with British English, on the basis of data from the London-Lund Corpus. Despite the acknowledged differences in the surface forms used in perfect contexts in these varieties, repeatedly justified as contact-induced innovations, in this paper she concentrates on the common pat-terns displayed by the analyzed varieties. She accounts for them in terms of diffu-sion, motivated by the shared core they have as a consequence of the expansion of the British Empire. In her opinion, this common core is further reinforced by the classroom context in which English is learned, which arguably inhibits lan-guage variation. Following a function-based approach, she shows that the two variants which compete in the perfect space across the board are the have-perfect and the SP, and she concludes that this has indeed to be linked to the classroom environment in which this linguistic feature is acquired.

Elena Seoane’s contribution “The perfect space in creole-related varieties of English: The case of Jamaican English” extends the scope further, and discusses the perfect space in Jamaican English (JamE) on the basis of the relevant ICE data. Building on earlier work, she starts from the observation that distributions between surface variants are conspicuous in this variety, as the SP, as well as be-perfects, bare participles (such as she gone) and morphologically unmarked forms (such as she go) feature prominently. As JamE is closely related to Jamaican Creole, this leads her to the investigation of whether these patterns are contact-in-duced. To approach this issue, she looks at the distribution of variant realiza-tions of a set of ten high-frequency verbs, and exploits the ICE layout to contrast writing and speech as well as individual registers. Seoane indeed traces effects of Jamaican Creole influence. However, considering individual variant realiza-tions, she argues that additional factors such as general cognitive constraints in language contact situations (mainly, simplification strategies) as well as the his-torical input variety also play a role. In addition, she finds considerable lexical effects, as the distributions between the variant realizations diverge markedly for individual lexical items, and describes differences between modes and registers. In contrast, her results suggest that the semantic interpretation only marginally influences the choice between the PP and alternative surface forms.

In “The frequency of the present perfect in varieties of English around the world”, Robert Fuchs investigates the periphrastic have-perfect in twenty national varieties of English, using data from the Corpus of Global Web-based English (GloWbE). By implementing linear regression models he discovers the independent factors which determine the differences in the frequency of use of the have-perfect in these varieties. Unlike previous research, which gives impor-


tance to factors such as degree of formality, evolutionary development following Schneider’s Dynamic Approach, and status in terms of Kachru’s Circle Model, in his research geographic proximity counts as the strongest factor. For instance, the PP is used with similar frequencies in the varieties spoken in the macro-region of South and South East Asia, in comparison with other varieties such as JamE or Kenyan English. Although they deserve further research, he finds other factors relevant to explain differences between varieties, such as the local identities por-trayed by the different countries. Another factor is language contact, determined both by a heterogeneous superstrate and the different substrates, as already pointed out in previous research. The author proposes that his results should be taken into account in English language teaching, and that the norms adopted in the classroom should be adapted to the learners’ communicative needs.

Valentin Werner, in “Rise of the undead? be-perfects in World Englishes”, presents an investigation of a low-frequency phenomenon commonly considered merely from a diachronic perspective. He contrasts findings of the be-perfect being a recessive structure with opposing claims of its persistence and productiv-ity in present-day English. He bases his observations on a quantitative and qual-itative cross-variety analysis of data from ICE and GloWbE, and finds that both transitive and intransitive uses of the be-perfect persist, albeit to different extents in different varieties. Variety type apparently is an important factor here, as the be-perfect is particularly vital in L2 varieties. In addition, his results suggest that certain (e.g. lexical) restrictions claimed to be in operation for this perfect variant in fact do not apply. In the discussion of the findings, Werner argues that the be-perfect should better be analyzed as a revitalized structure that unites both conservative and innovative properties, and that its spread in World Englishes is more pervasive than suggested in earlier research. In terms of methodology, he emphasizes the significant contribution that the analysis of web-based data as a locus of both change and persistence can make in the study of variation, and thus encourages the future study of this type of discourse.

In the third part, we introduce additional perspectives on the perfect, and build bridges to its study in applied linguistics and from the viewpoint of non-us-age based theories.

Starting with applied linguistics, in “The present perfect in learner English: A corpus-based case study on L1 German intermediate and advanced speech and writing”, Robert Fuchs, Sandra Götz, and Valentin Werner demonstrate the use of a corpus approach to inform the instruction of English as a foreign language. They consider the PP as a learner-hard feature, and present a quasi-longitudi-nal case study of the alternation between the PP and the SP in German-speak-ing learners of English. They use a variety of spoken and written corpus data to contrast learner English from both intermediate and advanced learners with


usage patterns by American and British speakers. From a quantitative perspec-tive, their data yield that the PP is more frequent in writing and can be viewed as a late-emerging and, by implication, challenging structure for learners. Their results also suggest that only the most advanced learners reach target-level dis-tributions between the PP and the SP, with some indication that an early start in English language teaching may lead to native-like usage frequencies. Other factors, such as a stay abroad, only seem to play a subordinate role, however. The authors further propose that a combination of factors, such as L1 influence, the structural complexity as well as the low input frequency of the PP, contribute to the observed underuse of the PP, and provide concrete suggestions concerning how to address this issue in the field of teaching English as a foreign language.

Björn Rothstein’s contribution, “Afterthought: Some brief remarks on auton-omous and speaker-centered linguistic approaches to the present perfect”, closes the volume with a discussion of the two main approaches to the PP, namely the “autonomous” (or “speaker-free”) linguistic approach and the “speaker-cen-tered” approach. In general terms, the former is a time-relational approach which distinguishes a speech time, an event time and a reference time, and the different times are organized according to the different meanings of the perfect (e.g. “extended-now”, “result”, or “current relevance”). This model is seman-tically-based and is essentially built on structural regularities of the language, without making reference to its speakers. By contrast, the speaker-centered approach, dominant in sociolinguistic analyses, relies heavily on the speakers and concentrates on the contexts in which the PP occurs. The author revises a comprehensive list of references dealing with the PP in English and shows how most of them can be ascribed to one of these two general approaches, with scarce attempts to combine both. Rothstein’s paper ends with a plea for linguists to inte-grate both sociolinguistic and autonomous grammatical factors in order to get a more comprehensive architecture of the language, following recent research by Adli, García García, and Kaufmann (2015).

With the contributions in this volume, we hope to have sharpened the view on the English perfect. As we are aware that new corpus (and other) material will be made available, and that different perspectives will be taken in the future, we also hope that the contributions in the present book provide a point of departure for these prospective re-assessments.


ReferencesAdli, Aria, Marco García García & Göz Kaufmann (eds.). 2015. Variation in language: System-

and usage-based approaches. Berlin: Mouton de Gruyter.Anderwald, Lieselotte. 2014. The decline of the be-perfect, linguistic relativity, and grammar

writing in the nineteenth century. In Marianne Hundt (ed.), Late Modern English syntax, 13–37. Cambridge: Cambridge University Press.

Bardovi-Harlig, Kathleen. 1997. Another piece of the puzzle: The emergence of the present perfect. Language Learning 47. 375–422.

Bardovi-Harlig, Kathleen. 2000. Tense and aspect in second language acquisition: Form, meaning, and use. Oxford: Blackwell.

Bauer, Gero. 1970. The English ‘perfect’ reconsidered. Journal of Linguistics 6. 189–198.Bautista, Maria L. S. 2008. Investigating the grammatical features of Philippine English. In

Maria L. S. Bautista & Kingsley Bolton (eds.), Philippine English: Linguistic and literary perspectives, 201–218. Hong Kong: Hong Kong University Press.

Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad & Edward Finegan. 1999. Longman grammar of spoken and written English. Harlow: Longman.

Biber, Douglas & Susan Conrad. 2009. Register, genre, and style. Cambridge: Cambridge University Press.

Biewer, Carolin. 2008. South Pacific Englishes: Unity and diversity in the usage of the present perfect. In Terttu Nevalainen, Irma Taavitsainen, Päivi Pahta & Minna Korhonen (eds.), The dynamics of linguistic variation, 203–219. Amsterdam: Benjamins.

Boogart, Ronny. 1999. Aspect and temporal ordering: A contrastive analysis of Dutch and English. Amsterdam: Vrije Universiteit dissertation.

Bowie, Jill & Bas Aarts. 2012. Change in the English infinitival perfect construction. In Terttu Nevalainen & Elizabeth Closs Traugott (eds.), The Oxford handbook of the history of English, 200–210. Oxford: Oxford University Press.

Bowie, Jill, Sean Wallis & Bas Aarts. 2013. The perfect in spoken British English. In Bas Aarts, Joan Close, Geoffrey Leech & Sean Wallis (eds.), The verb phrase in English: Investigating recent language change with corpora, 318–350. Cambridge: Cambridge University Press.

Comrie, Bernard. 1985. Tense. Cambridge: Cambridge University Press.Davydova, Julia. 2011. The present perfect in non-native Englishes: A corpus-based study of

variation. Berlin: Mouton de Gruyter.Davydova, Julia. 2013. Detecting historical continuity in a linguistically diverse urban area:

The present perfect in modern Singapore English. In Joana Duarte & Ingrid Gogolin (eds.), Linguistic superdiversity in urban areas: Research approaches, 193–225. Amsterdam: Benjamins.

Davydova, Julia, Michaela Hilbert, Lukas Pietsch & Peter Siemund. 2011. Comparing varieties of English: Problems and perspectives. In Peter Siemund (ed.), Linguistic universals and language variation, 291–323. Berlin: Mouton de Gruyter.

Declerck, Renaat. 2006. The grammar of the English tense system: A comprehensive analysis. Berlin: Mouton de Gruyter.

Elsness, Johan. 1997. The perfect and the preterite in contemporary and earlier English. Berlin: Mouton de Gruyter.


Elsness, Johan. 2009a. The perfect and the preterite in Australian and New Zealand English. In Pam Peters, Peter Collins & Adam Smith (eds.), Comparative studies in Australian and New Zealand English: Grammar and beyond, 89–114. Amsterdam: Benjamins.

Elsness, Johan. 2009b. The present perfect and the preterite. In Günter Rohdenburg & Julia Schlüter (eds.), One language, two grammars? Differences between British and American English, 228–245. Cambridge: Cambridge University Press.

Elsness, Johan. 2014. The present perfect and the preterite in Late Modern and contemporary English: A longitudinal look. In Kristin Davidse, Caroline Gentens, Lobke Ghesquìere & Lieven Vandelanotte (eds.), Corpus interrogation and grammatical patterns, 81–104. Amsterdam: Benjamins.

Elsness, Johan. 2016. English in South Africa: The case of past-referring verb forms. In María José López-Couso, Belén Méndez-Naya, Paloma Núñez-Pertejo & Ignacio M. Palacios-Martínez (eds.), Corpus linguistics on the move: Exploring and understanding English through corpora, 181–203. Leiden: Brill.

Engel, Dulcie M. 1998. A perfect piece? The present perfect and passé composé in journalistic texts. Belgian Journal of Linguistics 12. 129–147.

Engel, Dulcie M. & Marie-Eve A. Ritz. 2000. The use of the present perfect in Australian English. Australian Journal of Linguistics 20(2). 119–140.

Eriksson, Andreas. 2008. Tense and aspect in advanced Swedish learners’ written English. Göteborg: University of Göteborg.

Filppula, Markku. 1999. The grammar of Irish English: Language in Hibernian style. London: Routledge.

Fletcher, Paul. 1981. Description and explanation in the acquisition of verb-forms. Journal of Child Language 8(1). 93–108.

Hübler, Axel. 1998. The expressivity of grammar: Grammatical devices expressing emotion across time. Berlin: Mouton de Gruyter.

Huddleston, Rodney & Geoffrey K. Pullum. 2002. The Cambridge grammar of the English language. Cambridge: Cambridge University Press.

Hundt, Marianne. 1998. New Zealand English grammar, fact or fiction? Amsterdam: Benjamins.Hundt, Marianne. 2016. Error, feature, (incipient) change – or something else altogether?

In Elena Seoane & Cristina Suárez-Gómez (eds.), World Englishes: New theoretical and methodological considerations, 37–60. Amsterdam: Benjamins.

Hundt, Marianne & Carolin Biewer. 2007. The dynamics of inner and outer circle varieties in the South Pacific and Asia. In Marianne Hundt, Nadja Nesselhauf & Carolin Biewer (eds.), Corpus linguistics and the web, 249–269. Amsterdam: Rodopi.

Hundt, Marianne & Nicolas Smith. 2009. The present perfect in British and American English: Has there been a change, recently? ICAME Journal 33. 45–63.

Jespersen, Otto. 1931. A modern English grammar on historical principles. Heidelberg: Winter.Joos, Martin. 1964. The English verb. Madison: University of Wisconsin Press.Kirk, John M. & Jeffrey L. Kallen. 2007. Assessing celticity in a corpus of standard Irish English.

In Hildegard L. C. Tristram (ed.), The Celtic languages in contact: Papers from the workshop within the framework of the XIII International Congress of Celtic Studies, 270–298. Potsdam: Universitäts-Verlag Potsdam.

Kiparsky, Paul. 2002. Event structure and the perfect. In David I. Beaver, Luis D. Casillas Martinez, Brady Z. Clark & Stefan Kaufmann (eds.), The construction of meaning, 113–136. Stanford: CSLI.


Kjellmer, Göran. 2003. On nonoccurring perfective have in Modern English. Studia Neophil-ologica 75(1). 11–20.

Klein, Wolfgang. 1992. The present perfect puzzle. Language 68(3). 525–552.Klein, Wolfgang. 2009. How time is encoded. In Wolfgang Klein & Ping Li (eds.), The expression

of time, 39–81. Berlin: Mouton de Gruyter.Kortmann, Bernd. 1991. The triad ‘tense-aspect-Aktionsart’. Belgian Journal of Linguistics 6.

9–30.Kortmann, Bernd. 1995. Compositionality and the perfect. In Wolfgang Riehle & Hugo Keiper

(eds.), Anglistentag 1994 Graz: Proceedings, 183–199. Tübingen: Niemeyer.Kortmann, Bernd & Kerstin Lunkenheimer (eds.). 2013. The electronic world atlas of varieties of

English. Leipzig: Max Planck Institute for Evolutionary Anthropology. http://ewave-atlas.org (accessed 06 December 2015).

Labov, William. 1978. The study of nonstandard English. Urbana: National Council of Teachers of English.

Liszka, Sarah A. 2003. The development of tense and the present perfect in second language English. Colchester: University of Essex dissertation.

Lunkenheimer, Kerstin. 2012. Tense and aspect. In Raymond Hickey (ed.), Areal features of the Anglophone world, 329–353. Berlin: Mouton de Gruyter.

McCafferty, Kevin. 2005. Future, perfect – and past? Changing uses of be after V-ing in Irish English. In Gulbrand Alhaug, Endre Mørck & Aud-Kristi Pedersen (eds.), Mot rikare mål å trå: Festskrift til Trove Bull, 241–251. Oslo: Novus.

McCafferty, Kevin. 2014. ‘[W]ell are you not got over thinking about going to Ireland yet’: The be-perfect in eighteenth- and nineteenth-century Irish English. In Marianne Hundt (ed.), Late Modern English syntax, 333–351. Cambridge: Cambridge University Press.

McFadden, Thomas & Artemis Alexiadou. 2010. Perfects, resultatives, and auxiliaries in earlier English. Linguistic Enquiry 41(3). 389–425.

Meyer, Matthias. 1992. Das englische Perfekt [The English perfect]. Tübingen: Niemeyer.Meyer, Matthias. 1995. Past tense and present perfect in the Brown and the LOB corpora. In

Wolfgang Riehle & Hugo Keiper (eds.), Anglistentag 1994 Graz: Proceedings, 201–228. Tübingen: Niemeyer.

Michaelis, Laura. 2002. Aspectual grammar and past time reference. London: Routledge.Miller, Jim. 2000. The perfect in spoken and written English. Transactions of the Philological

Society 98(2). 323–352.Miller, Jim. 2004a. Perfect and resultative constructions in spoken and non-standard English.

In Olga Fischer, Muriel Norde & Harry Perridon (eds.), Up and down the cline: The nature of grammaticalization, 229–246. Amsterdam: Benjamins.

Miller, Jim. 2004b. Problems for typology: Perfects and resultatives in spoken and non-standard English and Russian. In Bernd Kortmann (ed.), Dialectology meets typology: Dialect grammar from a cross-linguistic perspective, 305–334. Berlin: Mouton de Gruyter.

Molsing, Karina V. 2006. The present perfect at the semantics/pragmatics interface: American English and Brazilian Portuguese. In Christian Ebert & Cornelia Endriss (eds.), Proceedings of the Sinn und Bedeutung 10: Annual meeting of the Gesellschaft für Semantik, 239–250. Berlin: ZAS.

Nussbaum, N. Jo & Rita C. Naremore. 1975. On the acquisition of present perfect “have” in normal children. Language and Speech 18( 3). 219–226.

O’Keeffe, Anne & Carolina P. Amador Moreno. 2009. The pragmatics of the be + after + V-ing construction of Irish English. Intercultural Pragmatics 6(4). 517–534.


Pietsch, Lukas. 2009. Hiberno-English medial-object perfects reconsidered: A case of contact-induced grammaticalisation. Studies in Language 33(3). 528–568.

Poplack, Shana & Sali Tagliamonte. 2001. African American English in the diaspora. Malden: Blackwell.

Portner, Paul. 2003. The (temporal) semantics and (modal) pragmatics of the perfect. Linguistics and Philosophy 26. 459–510.

Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A comprehensive grammar of the English language. Harlow: Longman.

Rastall, Paul. 1999. Observations on the present perfect in English. World Englishes 18(1). 79–83.Reichenbach, Hans. 1947. Elements of symbolic logic. London: Macmillan.Rissanen, Matti. 1999. Syntax. In Roger Lass (ed.), The Cambridge history of the English

language, Vol. III, 1476–1776, 187–331. Cambridge: Cambridge University Press.Ritz, Marie-Eve A. 2010. The perfect crime? Illicit uses of the present perfect in Australian police

media releases. Journal of Pragmatics 42. 3400–3417.Ritz, Marie-Eve A. & Dulcie M. Engel. 2008. ‘Vivid narrative use’ and the meaning of the present

perfect in spoken Australian English. Linguistics 46(1). 131–160.Rodríguez Louro, Celeste & Marie-Eve A. Ritz. 2014. Stories down under: Tense variation at the

heart of Australian English narratives. Australian Journal of Linguistics 34(4). 549–565.Rogatcheva, Svetlomira. 2014. Aspect in learner writing: A corpus-based comparison of

advanced Bulgarian and German learners’ written English. Giessen: University of Giessen dissertation.

Rothstein, Björn. 2008. The perfect time span: On the present perfect in German, Swedish and English. Amsterdam: Benjamins.

Roy, Joseph. 2014. The perfect approach to adverbs: Applying variation theory to competing models. Ottawa: University of Ottawa dissertation.

Schlüter, Norbert. 2002. Present perfect: Eine korpuslinguistische Analyse des englischen Perfekts mit Vermittlungsvorschlägen für den Fremdsprachenunterricht [Present perfect: A corpus-linguistic analysis of the English perfect with suggestions for foreign language teaching]. Tübingen: Narr.

Schneider, Edgar W. 2003. The dynamics of New Englishes: From identity construction to dialect birth. Language 79(2). 233–281

Schneider, Edgar W. 2007. Postcolonial English: Varieties around the world. Cambridge: Cambridge University Press.

Seoane, Elena & Cristina Suárez-Gómez. 2013. The expression of the perfect in South-East Asian Englishes. English World-Wide 34(1). 1–25.

Siemund, Peter. 2004. Substrate, superstrate and universals: Perfect constructions in Irish English. In Bernd Kortmann (ed.), Dialectology meets typology: Dialect grammar from a cross-linguistic perspective, 401–434. Berlin: Mouton de Gruyter.

Slobin, Dan I. 1994. Talking perfectly: Discourse origins of the present perfect. In William Pagliuca (ed.), Perspectives on grammaticalization, 119–133. Amsterdam: Benjamins.

Suárez-Gómez, Cristina & Elena Seoane. 2013. They have published a new cultural policy that just come out: Competing forms in spoken and written New Englishes. In Gisle Andersen & Kristin Bech (eds.), English corpus linguistics: Variation in time, space and genre, 163–182. Amsterdam: Rodopi.

Tagliamonte, Sali. 2000. The grammaticalization of the present perfect in English: Tracks of change and continuity in a linguistic enclave. In Olga Fischer, Anette Rosenbach & Dieter


Stein (eds.), Pathways of change: Grammaticalization in English, 329–354. Amsterdam: Benjamins.

Trudgill, Peter. 2011. Sociolinguistic typology. Oxford: Oxford University Press.Van Hattum, Marije. 2012. “A cannot get a loan now for more than six years now”: The

relationship between modal verbs and past time reference in Irish English. In Bettina Migge & Máire Ní Chiosáin (eds.), New perspectives on Irish English, 131–152. Amsterdam: Benjamins.

Van Herk, Gerard. 2008. Letter perfect: The present perfect in early African American correspondence. English World-Wide 29(1). 45–69.

Van Herk, Gerard, 2010. Aspect and the English present perfect: What can be coded? In James A. Walker (ed.), Aspect in grammatical variation, 49–64. Amsterdam: Benjamins.

Van Rooy, Bertus. 2009. The shared core of the perfect across Englishes: A corpus-based analysis. In Thomas Hoffmann & Lucia Siebers (eds.), World Englishes – Problems, properties and prospects, 309–330. Amsterdam: Benjamins.

Veloudis, Ioannis. 2003. Possession and conversation: The case of the category perfect. In Artemis Alexiadou, Monika Rathert & Arnim von Stechow (eds.), Perfect explorations, 381–399. Berlin: Mouton de Gruyter.

Vermant, Stefan. 1983. The English present perfect: A dynamic-synchronic approach. Wilrijk: Universiteit Antwerpen.

Walker, Jim. 2008. The footballer’s perfect – are footballers leading the way? In Eva Lavric, Gerhard Pisek, Andrew Skinner & Wolfgang Stadtler (eds.), The linguistics of football, 295–303. Tübingen: Narr.

Walker, Jim. 2011. The emergence of the narrative present perfect in British English: Reality or illusion? Groninger Arbeiten zur germanistischen Linguistik 53(2). 71–87.

Werner, Valentin. 2013. Temporal adverbials and the present perfect/past tense alternation. English World-Wide 34(2). 202–240.

Werner, Valentin. 2014. The present perfect in World Englishes: Charting unity and diversity. Bamberg: University of Bamberg Press.

Werner, Valentin. 2016. Overlap and divergence – aspects of the present perfect in World Englishes. In Elena Seoane & Cristina Suárez-Gómez (eds.), World Englishes: New theoretical and methodological considerations, 113–142. Amsterdam: Benjamins.

Werner, Valentin. Forthcoming. The present perfect as a core feature of World Englishes. In Anna Mauranen, Juhani Klemola, Svetlana Vetchinnikova & Markku Filppula (eds.), Changing English: Global and local perspectives. Berlin: Mouton de Gruyter.

Werner, Valentin & Robert Fuchs. 2016. The present perfect in Nigerian English. English Language and Linguistics.

Winford, Donald. 1993. Variability in the use of perfect have in Trinidadian English: A problem of categorical and semantic mismatch. Language Variation and Change 5. 141–187.

Wynne, Terence. 2000. The present perfect: A corpus-based investigation. Stirling: University of Stirling dissertation.

Yao, Xinyue. 2014. The present perfect in English: Meaning, interpretation and use. Sydney: University of New South Wales dissertation.

Yao, Xinyue. 2015. The present perfect and the preterite in Australian English: A diachronic perspective. In Peter Collins (ed.), Grammatical change in English world-wide, 247–268. Amsterdam: Benjamins.

Yao, Xinyue & Peter Collins. 2012. The present perfect in World Englishes. World Englishes 31(3). 386–403.


Yao, Xinyue & Peter Collins. 2013a. Recent change in non-present perfect constructions in British and American English. Corpora 8(1). 115–135.

Yao, Xinyue & Peter Collins. 2013b. Functional variation in the English present perfect: A cross-varietal study. Language and Computers 77(1). 91–111.

Yerastov, Yuri. 2015. A construction grammar analysis of the transitive be perfect in present-day Canadian English. English Language and Linguistics 19(1). 157–178.

Part I: Diachronic and synchronic perspectives on the perfect in native varieties of English

Berit Johannsen 1 From possessive-resultative to perfect?

Re-assessing the meaning of [hæbb- + past participle] constructions in Old English prose

Abstract: The development of the have-perfect is often given as a prime example of a grammaticalization path. The generally accepted account of the development of the English [have + past participle] construction is that it changed from a pos-sessive-resultative construction into a temporal-aspectual construction with per-fect-anterior meaning at some time in the Old English period. This study seeks to test the hypothesis that a semantic shift from resultative to perfect-anterior meaning can be observed in early English data. It is based on corpus data from the York-Toronto-Helsinki Parsed Corpus of Old English Prose. All instances of [hæbb- + past participle] are categorized according to their meaning, and impli-cations for possible source structures of the have-perfect are discussed. Finally, a look on Present-Day English helps to sound a note of caution on drawing conclu-sions from singular examples.

1 IntroductionThe structure [habban + past participle] and the have-perfect (henceforth HP)¹ in Old English (henceforth OE) have been debated for a long time (Harrison 1887; Caro 1896; Hoffmann 1934; Kuryłowicz 1965; Traugott 1972; Visser 1973; Mitchell 1985; Brinton 1988; Carey 1994; Wischer 2004, 2008; Kilpiö 2007). Many of these treatments have focused on the emergence of the structure as a temporal-aspec-tual marker, claiming that it developed from a possessive-resultative construc-tion. Although several studies have already dated this development to preliter-ary times (Brinton 1988: 99–102; Wischer 2004; de Acosta 2013: 33–34), some of the literature on the perfect at least gives the impression that this change can be observed in OE data (Bybee, Perkins, and Pagliuca 1994: 68; Heine and Kuteva 2006: 156; Yao 2014: 309).

1 The OE be-perfect will not be treated here (see Werner, this volume).

Berit Johannsen, Kiel University

24 Berit Johannsen

Taking a form-based approach, this study will reassess the meaning(s) of the [hæbb- + past participle]² structure in OE and discuss the implications for the origin of the HP. In a first step, I will discuss the possible source construction(s) and clarify the notions of possessive and resultative meaning, as well as perfect meaning. In a second step, I will analyze the meaning of all occurrences of [hæbb- + past participle] in the York-Toronto-Helsinki Parsed Corpus of Old English Prose (YCOE). Choosing a corpus-based method comes, first of all, out of necessity: Studies on OE are necessarily corpus-based in the widest sense. As introspection and surveys or experiments with native speakers of OE are impossible, all our hypotheses on the nature of OE rely on written sources. Most of the discussion on the OE HP so far has been based on singular, isolated sentences, or as Mitch-ell (1985: 298) puts it: “we are reduced to arguing about individual examples”. I claim, however, that discussing individual examples only marginally helps to clarify the status of the [hæbb- + past participle] structure in OE and that we need to adopt a more systematic corpus-based approach, which takes frequencies into account.

Corpus-based studies that have systematically analyzed all occurrences of the HP or perfect-like structures in a collection of texts include Harrison (1887; six texts), Caro (1896; 16 texts), and Wischer (2004, 2008; OE part of the Helsinki Corpus). They have, however, focused on different aspects of the HP or have used a different semantic categorization. Harrison describes the inflection of the participle and the types of objects occurring with [have + past participle]. Caro extensively studies the use of the preterite with perfect meaning, the inflection of the participle and word order. Wischer (2004) takes into account different types of objects, word order and inflection, and then addresses different functions of the have-periphrasis. In her (2008) article, Wischer analyzes the different func-tions of the structure in more detail. The semantic categorization into “anterior”, “resultative”, “past” and “perfective” is not very thoroughly explained and the categories are applied to structures with present and past have at the same time, which may have biased the results. It is therefore reasonable to conduct another study, which focuses on [hæbb- + participle], uses a clear semantic categorization and is based on a larger corpus. With 1,450,376 words, the YCOE is more than three times as large as the OE part of the Helsinki Corpus, which has 413,250 words (Kytö 1996), and can therefore help to draw a broader picture of the [hæbb- + past participle] structure in OE than previous studies have done.

2 Only structures with present forms of habban ‘have’ are analyzed. These are represented by hæbb-.

From possessive-resultative to perfect? 25

2 Meanings of [have + past participle]The potential source construction for the HP has been described in different ways. According to Kuryłowicz (1965: 58), it consists of the full verb have meaning ‘to own, possess’, a direct object and a past participle. The source construction “stresses a state resulting from a previous action (as expressed by the past participle)”, whereas the perfect stresses “the previous action inherent in the result” (Kuryło-wicz 1965: 58). Traugott (1972: 93–94) claims that the source is a “possessive con-struction” where the participle is an adjectival form of the verb, as in example (1).

(1) ða þas boc hæfde geleornodethen these books had learned‘then [he] had these books in-a-state-of-learnedness’³(BoProem: 1.6)⁴

Facilitated by the fact that the adjectival ending on participles was not always overtly marked, habban “was reinterpreted as a marker of perfectiveness and the adjectival nature of the participle was lost” (Traugott 1972: 94). Visser (1973: 2189) claims that have originally was “a notional verb denoting possession” and the past participle “a complement or attribute to the object [which] had a good deal of adjective force”. He paraphrases the original construction as “I have my work done = I possess or have my work in a done or finished condition”, that is, a state as a result. From this, the meaning of ‘antecedent action’ was inferred. Wischer (2004: 244) explains that a reanalysis from haveLexV + [NP + PartMod]Acc to haveLexV + NP + PartLexV must have taken place at some time.

As the notions possessive and resultative are not used consistently in these accounts, and as it is not always clear to what kind of constructions they refer, I use a categorization based on de Acosta (2013: 36–39). In his discussion of the potential source structure of the perfect, he distinguishes four different meanings that the pattern [have + NP.acc + past participle]⁵ can have apart from the perfect and gives examples of Present-Day English (PDE), reproduced in (2).⁶

3 Following Traugott’s (1972: 94) free translation.4 Sources of OE data throughout the chapter are indicated by Dictionary of Old English short titles and line numbers (diPaolo Healey et al. 2000).5 De Acosta has “<have + noun.acc + perfect participle>”, which is misleading because the structure includes a noun phrase (which may or may not be a single noun). De Acosta’s examples (shown as (2)) all have a noun phrase consisting of determiner and noun.6 As one of the reviewers notes, this bears the risk of being biased by a present-day perspective and we cannot be sure that there are no additional meanings for this construction in OE. This has to be kept in mind, but in the YCOE data at least I did not find any additional meanings.

26 Berit Johannsen

(2) a. Mary has a shirt made in China. (Adnominal)b. Mary has her opponent cornered. (Attained State)c. Mary had a rock thrown at her. (Affectee)d. Mary had the papers graded by an assistant. (Causative)

In my classification of the YCOE data I adopt these four types of usage for struc-tures with an accusative noun phrase. In order to shed light on the use of [hæbb- + past participle] structures in general, I also analyze those without an accusative noun phrase, that is, with other types of complements or without any comple-ments. Later, I will distinguish between structures with and without an accusa-tive noun phrase.

According to de Acosta, the adnominal use is “a collocation of lexical have and a noun phrase containing a participle functioning as an attributive adjective” (2013: 36). The use of the term attributive here is slightly misleading. From de Acosta’s account it is not clear whether he would subsume cases such as Mary has a loaded gun under adnominal uses, since the only example he gives is (2a). Both Quirk et al. (1985: 416–421) and Huddleston and Pullum (2002: 528–529) reserve the term attributive adjective for adjectives (or adjective phrases) premodifying the head of the noun phrase and distinguish them from postpositive adjectives, which follow the head of the noun phrase they modify. The participle made in (2a) would thus be classified as a postpositive adjective. In the following, I subsume participles functioning either as attributive or as postpositive adjectives under the adnominal category.⁷ Consequently, (3a–c) all exemplify the adnominal use of the [have + NP.acc + past participle] structure.

(3) a. I still have a letter written in pencil from his camp at Laisamis on his way to Nairobi in 1914. (BNC: H0A 715)

b. The US, of course, has a written constitution (BNC: K5C 2135)

c. unless the design you are knitting from has a specially written Form program (BNC: CA2 944)

In the attained state use, the syntactic subject of have is the agent (as in 4a) or cause (as in 4b) of the action implied by the participle. The situation described is the resultant state of a previous action, which may persist for some duration,

7 It is reasonable to take over this distinction for OE, where word order of nouns and their mod-ifiers is very similar to present-day usage (Quirk and Wrenn 1958: 87–89).


“usually under the control of the subject” (de Acosta 2013: 37). De Acosta stresses that have here does not signify ‘possess’ or ‘hold’.

(4) a. Once you have all your lists of goals written somewhere at the front of your diary, you are ready to start. (BNC: AD0 1879)

b. This news had them puzzled. (BNC: CD8 96)

In the affectee use, the subject of have never performs the role of agent, cause, patient, or theme of the action implied by the participle. It can have one of a number of “secondary” thematic roles: “beneficiary, sufferer, recipient, experi-encer, source, possessor, person for whom the statement holds true” (de Acosta 2013: 38). The agent can be expressed in a by-phrase. Additionally, de Acosta claims that the subject must be an animate sentient being (2013: 39). Examples like (5a) and (5b), however, show that the subject in the affectee use can also be inanimate.

(5) a. Or if I am called to a meeting that does not have a purpose written on the agenda (BNC: K6V 667)

b. Standard disc drives can not duplicate the ‘I can not read that’ instruction because the unreadable sectors have data written on them at twice the normal speed. (BNC: B7G 1129)

De Acosta excludes the causative type from his discussion since he considers it, in accordance with Visser (1973: 2269), a sixteenth- and seventeenth-century development. I do not discard the causative a priori, as Kilpiö (2013) has shown that causative habban has been used in OE already. In the causative use, the subject of have is an indirect causer, while the agent of the action implied by the participle can again be expressed in a by-phrase. In example (6a), the agent is expressed, but it is not expressed in (6b).

(6) a. To this end we had a feasibility study written by Ken [gap:name] who is Devon’s Community Recycling Network Coordinator. (BNC: HYJ 156)

28 Berit Johannsen

b. The job of a producer is to originate ideas, gather the material, work out the costs, keeping within a given budget, have the scripts written, rehearse the principals and finally preside over the dress rehearsal and the live produc-tion on the air. (BNC: B11 739)

Further differences between these structures, which are not mentioned by de Acosta, can be found. The adnominal is the only one in which the participle is a modifier of the accusative noun. This can be tested by turning the participle (or participial phrase) into a relative clause that modifies the noun. Only the transformations of the adnominals in (7a) have the same meaning or at least a meaning similar to the original, while (7b–d) change the meaning of the original or do not even turn out to be a very meaningful sentence. It is therefore only in the adnominal that have means ‘possess anything denoted by NP’, even if it is possession in its widest sense.⁸

(7) a. Mary has a shirt, which is made in China; or Mary has a gun, which is loaded.

b. ?Mary has her opponent, which is cornered.c. ? Mary had a rock, which was thrown at her.d. ?Mary had the papers, which were graded by an assistant.

The attained state use is best analyzed as a predicative construction, where the participle is not a modifier but an object-oriented predicative complement with the accusative noun phrase as object and predicand. We can further distin-guish resultative predicative complements, which typically occur with verbs that describe a change of state and which “denote the state of the predicand argument at the end of the process”, from depictive predicative complements, which give “a property of the predicand argument at the time of the situation under consid-eration, without any such factor of change” (Huddleston and Pullum 2002: 251). As Huddleston and Pullum show, have occurs with both types of complements: “We had half the children sick (depictive) vs. We had the meal ready in half an hour (resultative)” (2002: 266). The examples of attained state uses given so far can all be analyzed as containing resultative predicative complements.

8 According to Langacker (1999: 183), the senses of the possessive element have can range from “immediate physical control; to ownership, implying the possibility of physical access whenever desired; to more abstract kinds of ownership and access; to situations where the subject interacts with the object without in any way controlling it; to instances where the subject’s role is essen-tially limited to its reference point function”.


The causative and affectee uses again have a different internal structure according to Huddleston and Pullum (2002: 1236). They analyze these uses as cat-enative constructions consisting of the catenative verb have and a past-particip-ial non-finite clause, which is the catenative complement. The syntactic object of the matrix clause (e.g. a rock or the papers) is a raised object, that is, it is seman-tically an object of the catenative clause, not of the matrix clause.

In sum, what seems to be one and the same structure on the surface turns out to be four rather different structures on closer inspection. To which of these struc-tures do the initially described source constructions correspond now? Wischer seems to refer to the adnominal use, as she states that the participle is originally modifying the noun phrase. Both Kuryłowicz and Visser stress that in the original construction have means ‘to own, possess’, which indicates that they also refer to the adnominal. On the other hand, they both underline the ‘state resulting from a previous action’-meaning of the overall construction, which is found in the attained state use. None of them seems to refer to the affectee or causative use.

Finally, the fifth possible use of the [hæbb- + past participle] structure in OE is the perfect. Its meaning includes all four uses described by Comrie (1976: 56–60) for the English present perfect subsumed under the general meaning “continuing present relevance of a past situation”:

– perfect of result: Bill has gone to America (‘he is still there’) – experiential perfect: Bill has been to America (‘at some time until now’) – perfect of persistent situation: We’ve lived here for ten years (‘until now’) – perfect of recent past: I have recently learned that the match is to be postponed

Using CorpusSearch 2 (Randall 2010), I searched for all instances of present habban and a past participle in the same clause. Additionally, I searched for present habban co-occurring with a participial phrase or a small clause includ-ing a past participle. In order to find participles that are tagged as adjectives, I searched for all instances of present habban co-occurring with an accusative noun phrase consisting of at least a noun and an adjective or an adjective phrase, and manually selected all cases in which the adjective is a participle. All the extracted [hæbb- + past participle] sequences were categorized according to these five different meanings: adnominal, attained state, affectee, causative, perfect. If a structure was ambiguous between two categories, it was marked as such.

30 Berit Johannsen

3 ResultsTable 1: Frequency of uses of [hæbb- + past participle] in the YCOE.

N %

Adnominal 11 1.42Attained State 2 0.26Affectee 2 0.26Perfect 717 92.51Attained State/Affectee 1 0.13Attained State/Causative 1 0.13Attained State/Perfect 41 5.29

Totals 775 100.00

As Table 1 shows, all the uses introduced above occur in the OE data. The vast majority of uses of the [hæbb- + past participle] pattern have an unambiguous perfect meaning. Examples are given in (8a–e). In (8a), the temporal adverb ær (‘before’) indicates that the focus is on an anterior action, whereas in (8b) the meaning ‘anterior action’ is obvious without a temporal adverb. While (8a–b) include an accusative noun phrase and thus potentially allow all the other uses, examples (8c–e) include structures that do not allow the adnominal, attained state, affectee or causative reading. None of these have an accusative noun phrase but a complement clause (8c) or no complement at all (8d–e). These cases always have an unambiguous perfect reading.

(8) a. Gif he ær hæfþ attor gedruncenif he before has poison drunk‘If he has drunk poison before’(Lch II [3]: 43.1.3)

b. Gif he hwæt yfla gedon hæbbeif he something evil done have.sbjv‘If he has done something evil’(LawAf 1: 37.2)

c. Ðær ic hæbbe getæht hwelc hierde bion sceal.There I have taught who pastor be must‘There I have taught of which kind a pastor must be.’(CP: 65.467.19)


d. Ic hæbbe nu gewifod.I have now married‘I have married now.’(ÆCHom II, 26: 215.71)

e. þis folc hæfð gesyngodthis people has sinned‘This people has sinned’(ÆHom 21: 47)

The results above may be biased by the fact that occurrences without an accusa-tive noun phrase do not allow any of the non-perfect readings. Table 2 therefore shows the frequencies of the different uses for [hæbb- + NP.acc + past participle] only. This leaves us with the same overall picture. As should be expected, the number of clear perfects decreases. There were, however, five cases with genitive objects that could also be interpreted as either attained state or perfect, e.g. Nu we ealles heron habbað gefangen (ByrM 1 [Baker-Lapidge]: 1.2.251), which leads to a reduced number of cases that are ambiguous between an attained state and a perfect reading. The other absolute numbers remain stable.

Table 2: Frequency of uses of [hæbb- + NP.acc + past participle] in the YCOE.

N %

Adnominal 11 1.83Attained State 2 0.33Affectee 2 0.33Perfect 548 91.18Attained State/Affectee 1 0.17Attained State/Causative 1 0.17Attained State/Perfect 36 5.99

Totals 601 100.00

In the YCOE data, there are eleven adnominal uses of the [hæbb- + past participle structure], eight of which are attributive, such as (9b), and three of which are postpositive, such as (9a). As these are the only use with habban as a lexical verb meaning ‘possess’ in the widest sense, we can discard the assumption that [hæbb- + past participle] was only or mainly used in “possessive contexts” (Heine and Kuteva 2006: 156) in OE. The fallacy of conceiving of the core or orig-inal meaning of have as ‘possess’ or ‘hold’ has been hinted at by Brinton (1988: 100–101) and de Acosta (2013: 53–55). They also underline the polysemy of have, which must have existed in OE already.

32 Berit Johannsen

(9) a. Se læce […] hæfð on his agnum nebbe opene wunde unlacnodethe doctor has on his own face open wound uncured‘The doctor […] has on his own face an open uncured wound.’(CP: 9.61.2)

b. ðu hafast besmitene handayou have contaminated hands‘You have contaminated hands.’(Mart 5 [Kotzor]: Ja 24, A.5)

There are only two cases that I interpret as clear attained state uses. In (10a), the subject he, the king, is the agent of the action ‘to man his land (well)’ described by the participle, and the sentence describes the result of this action. In (10b), the unexpressed subject, the addressee of this handbook for Anglo-Saxon healers, is the agent of the action ‘to hang this same herb on his neck’, and again the situa-tion described by the whole sentence is the result of this action.

(10) a. þæt bið þonne Cyninges andweorc & his tol mid to ricsianne,that is the king’s material and his tool with to reignþæt he hæbbe his lond fullmonnadthat he have.sbjv his land well-manned‘That is the king’s material and his tool to reign with, that he may have his land well-manned.’(Bo: 17.40.15)

b. & hæbbe ðas sylfan wyrte on his swyran ahangene⁹and have.sbjv this same herb on his neck hung‘and (you) may have this same herb hung on his neck.’(Lch I [Herb]: 61.1)

There are two affectee uses, one of which is shown in (11a). In this passage from a law book, the subject (‘anyone committing the aforementioned crime’) is the sufferer of the action described by the participle, that is, it performs one of the secondary thematic roles. Furthermore, there is one affectee use which is ambigu-ous with an attained state reading. The agent of ‘to blind your hearts’ in (11b) can either be ge (‘you’) or somebody/something else not mentioned. In the first case it would be an attained state reading, in the second case it would be the affectee

9 Correlation of usage type with inflection on the participle and word order has not been taken into account in this study. Mitchell (1985: 291–292), Brinton (1988: 101–102), and Wischer (2004: 244–246) have shown that these are no reliable indicators of perfect or non-perfect usage.


reading.¹⁰ The adverb gyt (‘still’) seems to support the attained state interpreta-tion, but it is also compatible with the affectee reading. Similarly, the agent of ‘to write the names of those giving over their altar’ in (11c) can either be the subject ða mæssepreostas (‘the priests’) or someone else, meaning that they let someone else write the names. In the first case, it would be an attained state reading; in the second case, the priests would be the causer in a causative reading.

(11) a. næbbe his agne forfongennot.have.sbjv his own forfeited‘he may not have his property forfeited’(LawAf 1: 2.1)

b. Gyt ge habbað eowre heortan geblende?still you.pl have your.pl hearts blinded‘Do you still have your hearts blinded?’(Mk [WSCp]: 8.17)

c. Ða mæssepreostas underfon þæs folces teoðuncga &the priests receive.sbjv the.gen people.gen tithe andhabbon ealra þæra syllendra naman gewriten ofer heorahave.sbjv all.gen those.gen giving.gen names written over theirweofod.altar‘The priests may receive the people’s tithe and may have the names of all those giving written over their altar.’(ChrodR 1: 73.1)

Cases where [hæbb- + past participle] is ambiguous between an attained state and a perfect reading are more numerous. They account for 5.29 % of all the [hæbb- + past participle] data and for 5.99 % of the data with an accusative noun phrase. In (12), two examples are given. In both cases, the subject is the agent of the action implied by the participle. They could both, however, either focus on the state that is the result of the action or on the previous action that has current relevance. This shows how intimately related these two uses, the attained state and the perfect use, are with regard to their semantics. In many cases, it depends on how the reader interprets the structure. Even the linguistic context does not help. Brinton (1988: 102) has already pointed out that “problems of interpreting

10 De Acosta (2013: 46) excludes the affectee reading because in his view the context makes the subject ge the only possible agent. I claim that the context in this bible passage (Mark 8) does not necessarily lead to this interpretation.

34 Berit Johannsen

the perfect constructions of OE as either stative or actional result from the lack of contextual clues”.

(12) a. syllað me ealle þa hæftnedlingas, þe ge nu gefangen habbaðgive me all the captives that you.pl now caught have‘Give me all the captives that you have now caught.’(GDPref and 3 [C]: 37.255.17)

b. & þone mæstan dæl his hæfð sæ oferseten.and the biggest part it.gen has sea taken.over‘and the sea has covered the biggest part of it [the earth]’(Bo: 18.41.28)

Although the interpretation of the individual ambiguous examples may be debat-able, the overall picture shows us that unambiguous perfect uses were in the great majority. As a matter of fact, every text in the YCOE that has a [hæbb- + past participle] structure has at least one perfect use. There is no text that uses the structure with one or several of the other meanings only (see the Appendix). Vari-ables such as time period or dialect are not included in this study, but from the distribution in the YCOE texts we can hypothesize that speakers of OE at differ-ent times¹¹ at different places did not use the [hæbb- + past participle] structure exclusively with one of the non-perfect meanings.

The above results do not confirm a semantic shift from one of the [hæbb- + past participle] uses to the perfect construction in the documented OE period. They neither disprove the hypothesis that the HP evolved out of a [hæbb- + past participle] construction with non-perfect meaning. If one of the above-described [hæbb- + past participle] uses is the source construction of the perfect, the most likely candidate is the attained state use, given the considerable number of ambiguous cases. This is also proposed by de Acosta (2013: 46).

A note of caution has to be added, though. Just because both of these struc-tures exist in OE and are ambiguous in many cases, it does not mean that one evolved necessarily out of the other. They may also have been formed simulta-neously or have been calqued from other languages. Latin has been discussed as a possible source of the periphrastic perfect in several European languages, and I refer the reader to other studies for further details on the discussion (Traugott 1972: 92; Heine and Kuteva 2006: 152–157; Drinka 2003, 2011, 2013).

11 The earliest text in the YCOE is from the period 800–899, the last is from the period 1150–1199.


4 A thought experimentMoreover, it is dangerous to simply date putative developments to preliterary times. These accounts of the origin of the HP carry first of all “the burden of proof” (de Acosta 2013: 53). A little thought experiment is supposed to show why these theories are problematic, especially if they are based only on individual examples.

Let us assume that we are neither native nor foreign language speakers of English but have some passive knowledge of the language. Due to some unknown reason, we have no evidence of English before 1800. In Late Modern English (LModE) and PDE sources we find, among others, the examples in (13).

(13) a. You can put red lamps round it at night, and have the place manned all the time. (BNC: H0D 1289)

b. if ever I escape from this labyrinth, I mean to have it fairly written down(COHA: HowTryLover, 1817)

c. He has works written on the different building materials, and architects must be well acquainted with the nature of timber and stone, if they desire their works to be durable. (COHA: PsychologyOrAView, 1941)

d. Chelsea have already captured Marko Marin and Belgian boy-wonder Eden Hazard for next season(GloWbE: mirror.co.uk)

e. Radio, linked up with the movies and what is left of vaudeville, has now captured almost the entire concert management business(COHA: Atlantic, 1932)

f. IBM Corp has now wrapped up the deal under which it hands over MAPICS to Marcam Corp (BNC: CND 246)

g. She has fastened the more firmly around him, that chain, which the love of riches, or a thirst for fame, had already drawn till it corroded his immortal part (COHA: YoungMaiden, 1840)

h. When we sleep, one of us has it chained to our wrist. (BNC: HH5 1480)

i. and that they have their ensigns and flags hung out at the door of their meetings (BNC: HXC 1342)

36 Berit Johannsen

As well as examples like those in (14).

(14) a. Recently I have experienced serious and potentially fatal fevers. (BNC: A00 198)

b. He has shown the power of his arm (BNC: HU9 415)

c. Conservative policy on the inner cities has diverged in some ways from the earlier policies of the 1970s (BNC: AS6 1010)

d. Several of them have said that they now understand how children feel when adults do not want, or choose not to, believe them. (BNC: ARW 1226)

On the basis of examples (14a–d) we can claim that there is a structure [have + past participle] with perfect meaning, that is these sentences denote previous actions with current relevance. Turning to examples (13a–b), we conclude that there is a similar structure where the meaning describes the resultant state of a previous action. On top of that, we are faced with examples (13c–g), which can be interpreted both ways: they either focus on the resultant state or on the previous action. Finally, in (13h–i) we find examples that are ambiguous between a resul-tant state reading and a causative reading. Since there is a close semantic relation between the resultant state and the previous action reading, we conclude that the more grammatical perfect structure must have evolved out of the resultant state structure. We date this development to some point in a period of around 200 years before our documentation sets in.

This kind of reasoning is analogous to the kind of reasoning often used in studies on OE, partly because there is no other possibility. Based on individual examples drawn from a small, corrupt corpus of texts, they make claims about the origin of certain structures. As documentation does not predate the seventh century, with the exception of runic inscriptions, the proposed development is then dated to the undocumented OE period, where it cannot be proved. Of course, we know that the perfect has been used long before 1800. That is why, on the basis of this admittedly rudimentary analogy, I claim that it should be considered a likely scenario that the perfect had already been used in the undocumented OE period.


5 ConclusionThe re-assessment of the meanings of [hæbb- + past participle] structures in OE has yielded additional insights on the present perfect in OE. The analysis of all occurrences of [hæbb- + past participle] in the YCOE has shown that perfect uses by far outnumber other uses of the [hæbb- + past participle] structure in the OE period covered by the YCOE. In adopting a clear categorization of the different meanings that the [hæbb- + past participle] structure can have, it was shown that an adnominal structure with hæbb- as a lexical verb meaning ‘posses, hold’ is an unlikely candidate for the predecessor of the HP. The considerable number of cases that are ambiguous between an attained state and a perfect reading lends support to the hypothesis that the HP evolved from a [have + past participle] structure with attained state meaning. However, a thought experiment based on examples drawn from LModE and PDE has shown that this claim stands on very unstable ground and can hardly be proved.

AppendixDifferent uses of [hæbb- + past participle] in YCOE texts

Adno-minal

Affectee Attained State

Attained State/Affectee

Attained State/Causative

Attained State/Perfect

Perfect Totals

coaelhom 1 – – – – – 21 22coaelive – – – – – 1 26 27coalex – – – – – – 3 3coapollo – – – – – 1 10 11cobede – – – – – – 4 4cobenrul – – – – – 1 5 6coblick – – – – – 1 21 22coboeth 1 – 1 – – 10 95 107cobyrhtf – – – – – 1 13 14cocanedgD – – – – – – 2 2cocanedgX – – – – – – 1 1cocathom1 – – – – – 2 38 40cocathom2 1 – – – – – 21 22cochdrul – – – – 1 – 18 19cochristoph – – – – – – 2 2cochronD – – – – – – 2 2cochronE – – – – – – 3 3cochronE – – – – – – 4 4cocura 2 – – – – 11 66 79

38 Berit Johannsen

Different uses of [hæbb- + past participle] in YCOE texts

Adno-minal





Perfect Totals

cocuraC – – – – – – 8 8codicts – – – – – – 2 2codocu1 – – – – – – 1 1codocu2 – – – – – – 4 4codocu3 – – – – – – 14 14coeluc1 – – – – – – 6 6coeuphr – – – – – – 7 7coeust – – – – – – 1 1cogenesiC – – – – – 1 10 11cogregdC – – – – – 3 19 22cogregdH – – – – – – 6 6coherbar 4 – 1 – – – 7 12coinspolD – – – – – – 2 2coinspolX – – – – – – 3 3cojames – – – – – – 3 3colaece – – – – – – 4 4colaw1cn – – – – – – 2 2colaw2cn – – – – – – 6 6colaw5atr – – – – – – 4 4colaw6atr – – – – – – 2 2colawaf – 2 – – – – 1 3colawine – – – – – 1 5 6coleofri – – – – – – 1 1colsigewB – – – – – – 1 1colsigewZ – – – – – – 7 7colwsigeXa 1 – – – – – 2 3colwstan1 – – – – – – 1 1colwstan2 – – – – – – 2 2comargaC – – – – – – 10 10comargaT – – – – – – 1 1comart3 1 – – – – – 1 2comarvel – – – – – – 1 1comary – – – – – – 1 1conicodA – – – – – 1 18 19conicodC – – – – – – 11 11conicodD – – – – – – 4 4conicodE – – – – – – 3 3coorosiu – – – – – – 13 13cootest – – – – – 1 32 33coprefcura – – – – – – 1 1copreflives – – – – – – 1 1coprefsolilo – – – – – – 2 2


Different uses of [hæbb- + past participle] in YCOE texts

Adno-minal





Perfect Totals

coquadru – – – – – – 1 1corood – – – – – – 14 14cosevensl – – – – – 1 10 11cosolilo – – – – – – 38 38cotempo – – – – – – 2 2coverhom – – – – – – 33 33covinsal – – – – – – 5 5cowsgosp – – – 1 – – 5 6cowulf – – – – – 5 24 29

Totals 11 2 2 1 1 41 717 775

AcknowledgementsI would like to thank two anonymous reviewers and the participants of the work-shop “Perfect and perfectivity re-assessed through corpus studies” at ICAME 35 in Nottingham, where an early version of this paper was presented, for their com-ments and insightful discussion. I am also grateful to Ferdinand von Mengden for comments on an earlier draft of this paper.

SourcesBNC = The British National Corpus. Version 3 (BNC XML Edition). 2007. Distributed by Oxford

University Computing Services on behalf of the BNC Consortium. http://www.natcorp.ox.ac.uk/ (accessed 25 February 2015).

COHA = Davies, Mark. 2010. The Corpus of Historical American English: 400 million words, 1810–2009. http://corpus.byu.edu/coha (accessed 25 February 2015).

GloWbE = Davies, Mark. 2013. Corpus of Global Web-Based English: 1.9 billion words from speakers in 20 countries. http://corpus2.byu.edu/glowbe (accessed 25 February 2015).

YCOE = Taylor, Ann, Anthony Warner, Susan Pintzuk & Frank Beths (eds.). 2003. The York-To-ronto-Helsinki Parsed Corpus of Old English Prose. First edition. University of York: Department of Linguistics. http://ota.ahds.ac.uk./desc/2462 (accessed 7 January 2014).

40 Berit Johannsen

ReferencesAcosta, Diego de. 2013. The Old English have-perfect and its congeners. Journal of English

Linguistics 41(1). 33–64. doi: 10.1177/0075424212441706.Brinton, Laurel. 1988. The development of English aspectual systems. Cambridge: Cambridge

University Press.Bybee, Joan, Revere Perkins & William Pagliuca. 1994. The evolution of grammar: Tense, aspect,

and modality in the languages of the world. Chicago: University of Chicago Press.Carey, Kathleen. 1994. The grammaticalization of the perfect in Old English: An account based

on pragmatics and metaphor. In William Pagliuca (ed.), Perspectives on grammatical-ization, 103–117. Amsterdam: Benjamins.

Caro, George. 1896. Zur Lehre vom altenglischen Perfectum [On the teaching of the Old English perfect]. Anglia 18. 386–449.

Comrie, Bernard. 1976. Aspect: An introduction to the study of verbal aspect and related problems. Cambridge: Cambridge University Press.

diPaolo Healey, Antonette, Joan Holland, Ian McDougall & Peter Mielke. 2000. The dictionary of Old English corpus in electronic form. Toronto: DOE Project 2000.

Drinka, Bridget. 2003. The formation of periphrastic perfects and passives in Europe: An areal approach. In Barry J. Blake & Kate Burridge (eds.), Historical Linguistics 2001: Selected papers from the 15th International Conference on Historical Linguistics, Melbourne, 13–17 August 2001, 105–128. Amsterdam: Benjamins.

Drinka, Bridget. 2011. The sacral stamp of Greek: Periphrastic constructions in New Testament translations of Latin, Gothic, and Old Church Slavonic. Oslo Studies in Language 3(3). 41–73.

Drinka, Bridget. 2013. Sources of auxiliation in the perfects of Europe. Studies in Language 37(3). 599–644. doi: 10.1075/sl.37.3.06dri.

Harrison, James A. 1887. The Anglo-Saxon perfect participle with “habban”. Modern Language Notes 2(6). 134–135.

Heine, Bernd & Tania Kuteva. 2006. The changing languages of Europe. Oxford: Oxford University Press.

Hoffmann, Gerhard. 1934. Die Entwicklung des umschriebenen Perfektums im Altenglischen und Frühmittelenglischen [The development of the circumscribing perfect in Old and Early Middle English]. Ohlau: University of Breslau dissertation.

Huddleston, Rodney D. & Geoffrey K. Pullum. 2002. The Cambridge grammar of the English language. Cambridge: Cambridge University Press.

Kilpiö, Matti. 2007. Auxiliation in progress: Diachronic grammaticalisation changes in Old and Early Middle English have perfects. In Matti Rissanen, Marianna Hintikka, Leena Kahlas-Tarkka & Rod McConchie (eds.), Change in meaning and the meaning of change: Studies in semantics and grammar from Old to Present-Day English, 323–343. Helsinki: Société Néophilologique.

Kilpiö, Matti. 2013. Causative habban in Old English: Tracing the development of a budding construction. In Gabriele Diewald, Leena Kahlas-Tarkka & Ilse Wischer (eds.), Comparative studies in early Germanic languages: With a focus on verbal categories, 101–126. Amsterdam: Benjamins.

Kuryłowicz, Jerzy. 1965. The evolution of grammatical categories. Diogenes 13(51). 55–71.


Kytö, Merja. 1996. Manual to the diachronic part of the Helsinki Corpus of English Texts: Coding conventions and lists of source texts. University of Helsinki: Department of English. http://clu.uni.no/icame/manuals/HC/INDEX/HTM (accessed 25 February 2015).

Langacker, Ronald W. 1999. Grammar and conceptualization. Berlin: Mouton de Gruyter.Mitchell, Bruce. 1985. Old English syntax: Concord, the parts of speech, and the sentence,

Vol. 1. Oxford: Clarendon Press.Quirk, Randolph, Sydney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A comprehensive

grammar of the English language. Harlow: Longman.Quirk, Randolph & Charles Leslie Wrenn. 1958. An Old English grammar. London: Methuen.Randall, Beth. 2010. CorpusSearch 2. http://corpussearch.sourceforge.net (accessed 7 January

2014).Traugott, Elizabeth Closs. 1972. A history of English syntax: A transformational approach to the

history of English sentence structure. New York: Holt, Rinehart and Winston.Visser, Fredericus Th. 1973. An historical syntax of the English language: Syntactical units with

two and more verbs, Vol. 3.2. Leiden: Brill.Wischer, Ilse. 2004. The “have”-perfect in Old English. In Christian Kay, Simon Horobin &

Jeremy Smith (eds.), New perspectives on English historical linguistics: Syntax and morphology, Vol. 1, 243–255. Amsterdam: Benjamins.

Wischer, Ilse. 2008. Grammaticalization of periphrastic constructions. In Elisabeth Verhoeven, Stavros Skopeteas, Yong-Min Shin, Yoko Nishina & Johannes Helmbrecht (eds.), Studies on grammaticalization, 241–250. Berlin: Mouton de Gruyter.

Yao, Xinyue. 2014. Developments in the use of the English present perfect: 1750–present. Journal of English Linguistics 42(4). 307–329. doi: 10.1177/0075424214549560.

Jill Bowie and Sean Wallis2 The to-infinitival perfect: A study of

declineAbstract: The English to-infinitival perfect (as in She claims to have seen him) has not received the same attention as the present perfect. In this paper we examine its changing use in written American English over the last 200 years, using data from the Corpus of Historical American English (COHA). This reveals a reduction of about 80 % over this period, against a baseline of other past-referring forms (present perfect and past tense verbs). Secondly, we examine contexts of decline, focusing on the most frequent verb collocates of the to-infinitival perfect in COHA (such as claim in the example above) on the premise that these collocates iden-tify the semantic contexts in which the to-infinitival perfect may be used. Collo-cates are divided into subgroups based on semantic and grammatical criteria, including possible alternation patterns to the to-infinitival perfect. This proce-dure exposes a rich variation in the behaviour of both subgroups and individual verbs.

1 IntroductionMany studies in English linguistics have focused on the present perfect, whereas other types of perfect have received less attention. In this study we look at the to-infinitival perfect, as in A witness is said to have seen the whole event, con-sidering the changing usage of this structure in written American English (AmE) over the last 200 years.

We first need to set this structure in context as one subtype of the English perfect. In all perfect subtypes, the perfect auxiliary have is followed by a verb in the past participle form. The auxiliary occurs in the present tense in the present perfect (as in She has spoken to them), but it can also occur in the past tense, giving the past perfect (She had spoken to them). Additionally, it can occur in non-finite forms, within gerund-participial clauses (Having spoken to them, she went home), bare-infinitival clauses (She must have spoken to them), or to-infini-tival clauses (She seems to have spoken to them).

All subtypes of the English perfect typically function to express anteriority (i.e. “pastness” relative to a reference point). This supports a treatment of the

Jill Bowie and Sean Wallis, University College London

44 Jill Bowie and Sean Wallis

perfect as an analytically expressed tense system (e.g. Huddleston and Pullum 2002). However, some grammarians analyse the perfect in terms of aspect (e.g. Quirk et al. 1985), and the correct analysis has been much debated, largely because of the added semantic complexity associated with the present perfect (see, e.g., Comrie 1976; Elsness 1997; Werner 2014; Werner, Seoane, and Suárez-Gómez, this volume). The present perfect generally presents a situation as occurring within (or even continuing through) a time span beginning in the past and leading up to the present, with an added dimension of “current relevance” (i.e. a focus on the present repercussions of the situation). This specialised use contrasts with that of the inflectionally marked past tense (used to represent a time as wholly in the past rather than connected to the present; see the contributions in this volume by Fuchs; Fuchs, Götz, and Werner; Seoane; Van Rooy).

This contrast, however, is neutralised in the other types of perfect, which can correspond to either a present perfect or a simple past (Huddleston and Pullum 2002: 146). For instance, compare (1) and (2).

(1) The government claims to have achieved great things since the election.

(2) The neighbour claims to have heard three gunshots shortly after midnight.

The content of the claim in (1) would be rendered as “We have achieved great things since the election” (with a present perfect), while that of (2) would be “I heard three gunshots shortly after midnight” (with a simple past).

A number of corpus-based studies have looked at variation and change in the present perfect (including, among others, Elsness 1997, 2014; Hundt and Smith 2009; Yao and Collins 2012; Bowie, Wallis, and Aarts 2013a). Most of the evidence indicates that its frequency has declined since 1800 in AmE, and some of the evi-dence also suggests a decline in British English (BrE) – though this is less clear – while for the present day there is general agreement that the structure is more frequent in BrE than in AmE.

There has been more limited corpus investigation of other (non-present) types of perfect, which are of much lower frequency than the present perfect. The results show some parallels with the findings for the present perfect, in terms of American versus British contrasts and of trends of decline over time. Elsness (1997) reports written data showing a higher frequency of the past and infinitival perfect forms in BrE than in AmE of the present day, and a decline in the fre-

The to-infinitival perfect: A study of decline 45

quency of these forms in AmE since 1800.¹ Yao and Collins (2013) report written data showing a higher frequency of all non-present perfect types in BrE than in AmE of both the 1960s and 1990s, and a decline of the past perfect in both vari-eties from the 1960s to the 1990s. In our own previous research (Bowie and Aarts 2012; Bowie, Wallis, and Aarts 2013a), we investigated change over recent time (1960s to 1990s) in all forms of the perfect in spoken BrE in the Diachronic Corpus of Present-Day Spoken English (DCPSE), finding a decline over this period in the past and infinitival perfects. Thus, the research to date on non-present perfect structures shows that they are worthy of further investigation in terms of varia-tion and change.

Our focus in the present study is the to-infinitival perfect. For this form, our previous research on spoken BrE from the 1960s to 1990s found a steep decline (52 % ± 26 %)² in frequency relative to that of to-infinitives in general (Bowie and Aarts 2012). This decline presents a puzzle. It is not in general possible to simply omit the perfect without changing the meaning: there is a clear temporal contrast between He is believed to be in Paris and He is believed to have been in Paris, since in the latter the perfect expresses anteriority to the present time referred to by is. If people are using fewer to-infinitival perfects, how might they express the meanings conveyed by this form?

As the to-infinitival perfect is a relatively infrequent form (with only 127 occurrences in DCPSE, a corpus of c. 885,000 words), we need a larger corpus in order to investigate it in more detail. It is also of interest to look at longer-term trends in the use of this form. For the present study we use the 400-million-word Corpus of Historical American English (COHA; Davies 2012), which covers the period from 1810 to 2009 and includes text from the following genres: fiction, newspaper, magazine, and a fourth category, “other non-fiction”. Our research questions are as follows:

– Do we see a decreasing inclination to use the to-infinitival perfect in written AmE over the last 200 years?

– What are the most frequent structural and/or collocational contexts for this construction? Are there alternants for these which express a similar meaning? If so, are these alternants increasing in frequency at the expense of the con-struction we are investigating?

1 For the infinitival perfect forms, we performed our own calculations based on combinations of the figures provided by Elsness for several different constructions involving either the bare infinitival perfect or the to-infinitival perfect (1997: 104, 267–268).2 Errors are estimated using the Wilson score interval at a 0.05 error rate, i.e. predicting that it is 95 % probable that the true “population” value is within this range (Wallis 2013).


Our study is in three stages. First, we investigate frequency variation of the to-in-finitival perfect in COHA over time. We discuss different ways of measuring rates of change, going beyond the commonly used baseline of words (typically, per-million-word adjusted frequency measures) to more linguistically relevant base-lines which allow us to test for the robustness of the patterns we find. In previous studies using DCPSE and its software, ICECUP (Nelson, Wallis, and Aarts 2002), we were able to exploit the parse annotation to construct precise queries for the studied perfect form and a baseline category (such as tensed past-marked VPs; see Bowie, Wallis, and Aarts 2013a: 328). Although COHA is not a parsed corpus, the part-of-speech tagging and flexible search options offered by the web inter-face do permit a careful researcher to devise search query strings which approxi-mate to constructional queries.

Second, we wish to identify the limited set of contexts within which the to-in-finitival perfect occurs, since it is in these contexts that replacement with alter-nate forms is possible. To help us achieve this, we draw on data from a smaller, parsed corpus which allows us to carry out precisely specified searches for gram-matical structures. The corpus used is ICE-GB, the British component of the Inter-national Corpus of English (Nelson, Wallis, and Aarts 2002).

Finally, we return to COHA to examine these contexts, and where feasible, enumerate plausible grammatical and lexical alternation patterns. For instance, we find that in COHA forms of the verb seem are the most frequent left collocates of the to-infinitival perfect. Enumerating the limited number of verbs found in this position, we examine potential alternation patterns such as (3) versus (4).

(3) He seems to have known about it.

(4) It seems that he knew about it.

In (3) we find a to-infinitival perfect, while in (4) seems takes as its complement a finite clause instead of an infinitival clause (and has impersonal it as subject). An obvious lexical alternant to be considered here is appear, which is semantically close to seem and grammatically able to replace it in structures like (3) and (4). In fact there is a wide range of alternative expressions that speakers might use to express reasonably similar shades of meaning. These may differ more radically from (3) in terms of grammatical structure and/or lexical content, so that there is no clear-cut limit: I suspect he knew about it; The evidence suggests he knew about it; He apparently knew about it; and so on. It is not feasible in this study to investigate the full range of possible alternative expressions for every context in which a to-infinitival perfect occurs, and we therefore focus on a selection of closer structural and lexical alternants.


The plan for the paper is as follows. In Section 2 we introduce COHA and report on frequency variation of the to-infinitival perfect in this corpus over time. Section 3 describes ICE-GB and presents a frequency profile for the various gram-matical contexts in which the to-infinitival perfect occurs in this corpus. We then go on to examine the most frequent contexts in COHA in Section 4, focusing on verbal left collocates, and considering some possible alternation patterns. Our conclusion is given in Section 5.

2 The decline of the to-infinitival perfect in COHAThis section briefly introduces COHA (2.1) before reporting on frequency variation of the to-infinitival perfect in this corpus over time (2.2).

2.1 COHA

The Corpus of Historical American English (COHA) is a part-of-speech tagged megacorpus of written English. It contains over 400 million words of written AmE sampled for every year from 1810 to 2009, with approximately 2 million words per year. It was released in late 2010 and is freely available over a web interface (http://corpus.byu.edu/coha).³ Davies (2012) provides an overview of the corpus.

COHA comprises four main text genres. Figure 1 depicts the proportion of words collected for each of these genres over time (see Appendix A for the raw word frequencies).⁴ It can be seen that the sampling is not fully consistent over time. Fiction comprises approximately half the material over the decades, and magazines contribute about a quarter from the 1820s onwards. Newspapers and non-fiction provide the remaining quarter, but their proportions are less evenly distributed over the earlier decades. Newspapers are included only from the 1860s onwards, and their share increases in later decades. The total quantity of material sampled per decade is also somewhat variable, with less than 20 million words for some of the early decades.

When we performed our analyses, we found that the data for the 1810s, much smaller in size in any case (at 1.18 million words), obtained observations that were

3 More recently, in March 2015, the corpus has also become available to purchase for download. For the research reported here, we used the old web interface, currently at http://corpus.byu.edu/coha/old.4 The data are based on raw word frequencies under “Composition of the corpus”.


inconsistent with the remainder of the data – a fact that was true across many dif-ferent analyses. The simplest explanation for this can be seen in the graph: the sampling for this decade is quite distinctive and this fact may skew the results. For this reason we decided to omit the first decade from our analyses.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

1810

1820

1830

1840

1850

1860

1870

1880

1890

1900

1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

NF BOOKSNEWSPAPERSMAGAZINESFICTION

Figure 1: Distribution of text samples over the decades in COHA.

At the time of writing (2015), the web interface provides four different search and display options. To study frequency variation, we used the “Chart” option. This outputs a frequency breakdown by decade, providing both raw and per-mil-lion-word figures. From this display, the researcher can select any decade to see examples of the search item within their sentence context. COHA is annotated for parts-of-speech (POS), using the C7 tagset of the CLAWS tagger.⁵ Search options include specifying words, lemmas and POS tags, for all of which a simple wild-card system can be employed. As well as the usual lexical wildcard, applying a wildcard to a POS tag supports simple grammatical abstraction, such as extract-ing all verbs irrespective of any other coded features.

5 For the list of tags, see http://ucrel.lancs.ac.uk/claws7tags.html.


2.2 Frequency of the to-infinitival perfect over time

To search for the to-infinitival perfect we used the search string “to have [v?n*]”, i.e. the sequence (to + have + past participle verb). The third element in the string matches all items tagged as past participle verb forms. Note that this will not identify every example of the to-infinitival perfect: any string including one or more intervening words will not be found, so to have started will be included, whereas to have already started or to already have started will not. This represents an underestimate of approximately 2–3 % of genuine examples.⁶

Figure 2 plots the results of this search (n = 66,219) as a simple proportion per word (the web interface per-million-word graph produces a similar picture, without error bars). This graph indicates an almost linear decline in the overall use of the form (an “exposure rate”) over the period. Note that the expected dis-tribution of a simple decline would be an “S” curve (a “logistic curve”). A linear decline in probability is steeper than this, and implies that the baseline is less than optimal.

This graph does not simply describe this sample: we can establish that the observed trend represents a genuine decline in AmE. Each point includes an “I”-shaped error bar, which identifies the region where the actual proportion in the population of comparable American written English is most likely to be found. These error bars are computed using the Wilson score interval method (Wallis 2013) with a 95 % range. This means we are 95 % certain that the true pop-ulation value is within this range.⁷

To “read” the intervals in these graphs, note that where intervals do not overlap, the difference between observations is guaranteed to be statistically sig-nificant, i.e. a resample from the population will find a difference 19 times out

6 When sequences with an intervening word in either position are examined, it can be seen that the great majority are either not genuine examples of the to-infinitival perfect (e.g. to have it known), or involve instances with an additional past participle (perfect + passive, e.g. to have been loved) which are also included in the search string without intervening words. If only sequences with intervening words tagged as adverbs are considered, then most examples are genuine. Such sequences would represent 2 % of a corrected total that included them. Longer intervening sequences (such as multiple adverbs or preposition phrases like in fact, at that stage) also occur, but string searches show them to be much less frequent. We can compare results from ICE-GB (see Section 3), a parsed corpus, where it is possible to identify all examples of the required structure regardless of intervening words: in that corpus, only 1 % of instances of the to-infinitival perfect (2 out of 192) involve intervening words (a single adverb in each instance).7 An anonymous reviewer asks whether this takes account of the missing 2–3 % of cases. The answer is no: it cannot do so, as it is a statistical measure based on the data we actually used (the cases without intervening words).


of 20. Slightly overlapping intervals may represent a significant difference, but a separate statistical test (the Newcombe-Wilson test) is required to determine this. Any pair of intervals where one point falls in the other’s interval range should be considered effectively constant (i.e. the difference is non-significant). Thus in Figure 2, we cannot claim that the rate falls between 1840 and 1850, or that it rises again in the 1860s. For more information see Wallis (2013).

0.0000

0.0001

0.0002

0.0003

1820

1830

1840

1850

1860

1870

1880

1890

1900

1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

p(to-infinitive perfect | word)

Figure 2: The to-infinitival perfect measured per word in COHA, by decade, with 95 % Wilson score intervals for each observation representing the likely range in the population.

Although per-million-word measures are commonly used in corpus linguistic studies, it is usually preferable to use other baselines. When any studied form is varying, the opportunity to use that form will almost certainly also vary.⁸ To obtain reliable results we must first factor out variation due to opportunity in

8 For example, Bowie, Wallis, and Aarts (2013b) show that the frequency of tensed verb phrases per word in the DCPSE corpus varies over genre and time. Suppose we were to study a subtype of tensed verb phrase which always had a constant 20 % share of cases. The correct conclusion would be that the proportion was stable over time and genre. But were we to use a per-million-word baseline to study this subtype, we would instead see the original undulating and varied pattern!


order to focus on the choices that speakers and writers make. Conflating these two types of variation, in the opportunity to choose and in the choices made, has a further undesirable consequence: it makes the reproduction of results between corpora more difficult than otherwise.

In our case we are interested in the grammatical choices made by writers, rather than their overall authorial choices, or, potentially, the effect of external constraints. We wish to investigate whether US writers have decreasingly chosen to use the to-infinitival perfect when the opportunity arises. We therefore address our hypothesis by examining contexts where the option to use this form arose in the first place, i.e. either (a) by setting observed frequencies against those of a set of alternants, or (b) by comparing them against the observed frequency distribu-tion of a meaningful superordinate category. In this section, we employ method (b), as we are concerned with the overall data and have not yet considered the range of contexts within which the structure occurs.

We first consider a baseline of to-infinitives in general, since this represents the structural context within which the form in question can occur (the to-infini-tival perfect is, logically, a subset of to-infinitives). This eliminates variation that may be due to changes in the use of to-infinitival complementation. The non-fi-nite complementation system in English is still undergoing change. Mair (2006: 215) notes that “there is every indication that usage of infinitives, participles and gerunds has continued changing with unabated vigour […] in the past two centu-ries”. Studies have looked, for example, at the long-term tendency for gerund-par-ticipial clauses to replace infinitival clauses (e.g. Fanego 2007; Vosberg 2009) and at synchronic variation between bare infinitival and to-infinitival complements (Callies 2013). A baseline of to-infinitives allows us to discount the possibility that the apparent decline in the to-infinitive perfect is simply a corollary of variation in the use of to-infinitival complementation in general.

To test this hypothesis we employed the search string “to [v?i*]” (to + infini-tive verbs) to retrieve baseline frequencies, and then plotted the proportion p for all perfect cases, which we might write as p(to-infinitive perfect | to-infinitive). This obtains an almost identical result to that obtained with a per-million-word baseline. The observed decline is not due to an overall decline in to-infinitive fre-quency. On the other hand, the resulting plot is still apparently linear.

However, as discussed earlier, there is generally a temporal contrast between the to-infinitival perfect and a corresponding non-perfect infinitival structure, since the perfect expresses anteriority. A more reliable baseline for comparison purposes might be all past tense verbs (cf. Bowie, Wallis, and Aarts 2013a), which we can obtain using the simple search string “[v?d*]”. To this baseline we add the smaller set of present perfect constructions, because (as noted in Section 1 above) the meaning conveyed by the to-infinitival perfect can sometimes be rendered by


a finite construction containing a present perfect rather than a morphologically marked past tense.⁹ It can also be rendered in some instances by a past perfect, but those are already retrieved by our past tense search.

In this approach, we are considering the to-infinitival perfect as one of a group of verbal forms or constructions that can express anteriority. This is import-ant because texts vary considerably in their temporal orientation (their concern with the past, present, or future). An infinitival perfect is likely to be used more frequently in texts with greater focus on past time, reflected in a greater use of past tense verbs. We therefore wish to test whether the apparent decline reflects a genuine change in preferences for the form, or reflects variation in the temporal orientation of texts sampled (which may not be controlled for). Our baseline rep-resents a meaningful superordinate category for this purpose.¹⁰

The results of this comparison against past tense + present perfect are pre-sented in Figure 3. This finds a smoother, more consistent decline against this baseline that approximates to a logistic regression line (popularly known as an “S-curve”, in this case with a close fit, r2 = 0.9603).¹¹ Measured in this way the fall is huge: a decline of around 80 % from 1820 to 2000.

More generally, although the confidence intervals are small, joining the dots still obtains a “bumpy” line. The peak in 1880 is likely to be genuine (it appears in graphs with different baselines), but it is distinct from the overall logistic trend. This suggests that, for the to-infinitival perfect at least, the opportunity to use the form varies according to other uncontrolled factors, not just temporal orientation,

9 For the present perfect, we combined two searches using POS tags to allow for either a third-person singular or general present tense form of have followed by a past participle: the strings “[vhz] [v?n*]” and “[vh0] [v?n*]”.10 An alternative perspective, which has a similar outcome, focuses on a more specific potential alternation. This is applicable to examples that occur as complements to catenative verbs, which comprise a large majority of the total (see sections 3 and 4). We could consider the probability that a writer decided to insert a catenative verb and adjust the original tensed verb to become the complement of the catenative, i.e., to transform “subject + V-ed” → “subject + catenative verb + perfect infinitive have + V-en” (e.g. he said → he seems to have said). Similar transformations would also apply to “subject + present/past perfect”. In this kind of transformation, the meaning changes as a result of the insertion, but the decision to insert a catenative is not heavily con-strained (except, naturally, as the sum of all meanings potentially expressible by catenative + perfect). The decision to use a particular catenative verb is one that will likely be sensitive to context, implying the need for additional analyses with different alternation patterns for each verb. The focus here is the overall trend, however, implying that all we need assume is that the total set of contexts that would permit insertion is stable over time.11 For more information on the method see http://corplingstats.wordpress.com/2015/04/24/ logistic-regression.


which might vary from text to text within the sample. It is necessary to dig deeper by examining the structural and semantic contexts in which the perfect is used.

We also compared the frequency of the to-infinitival perfect against the base-line of past tense verb plus present perfect constructions for each of the four genres in COHA. We found that all four genres show significant declines over the period, and we employed logistic regression to identify each underlying trend for the purposes of comparison.

The structure is consistently the least frequent in fiction, falling from around 0.4 % of cases in 1820 to a quarter of that rate by 2000, and declines less steeply in that genre than magazines and newspapers, which fall to as much as a seventh of their starting point. Although newspaper data is absent in the early decades, from 1860 onwards the remaining pattern is remarkably consistent with that for mag-azines. The category of “non-fiction books” sees the most gentle decline, from around 1 % to 0.4 % over the same period, becoming the genre containing the highest proportion of to-infinitival perfect cases from 1950 on.

0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

1820

1830

1840

1850

1860

1870

1880

1890

1900

1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

p(to-infinitive perfect | present perfect + past)

Figure 3: Relative frequency of the to-infinitival perfect against a baseline consisting of present perfect constructions plus past tense verbs, over the decades in COHA.


3 Identifying contexts of occurrenceA further investigation of the decline in the to-infinitival perfect requires us to consider the contexts in which it occurs. To-infinitival clauses occur in a wide range of grammatical structures, as described in English reference grammars. These are structures within which the perfect form in question can potentially occur.¹² Huddleston and Pullum (2005: 212) provide a convenient list with exam-ples, several of which are cited below, with the clauses in bold.

(5) To turn back now would be a mistake.

(6) Her parents intend to buy her a car.

(7) I go to the gym to keep fit.

(8) We found a big box to keep the CDs in.

(9) He was anxious to make a good impression.

In (5) we have a subject; more common is an extraposed subject (It would be a mistake to turn back now). The other functions illustrated are complement of a verb (6), clausal adjunct (7), modifier within a noun phrase (8), and complement of an adjective (9). Other functions listed by Huddleston and Pullum include extraposed object, complement of preposition, and complement of noun.

As can be seen from the above examples, it would be extremely hard to iden-tify all such contexts in an unparsed corpus like COHA. Instead, we propose to determine in which contexts a perfect is most frequently found. To obtain a rea-sonable estimate of the frequency distribution for the perfect in COHA we use the parsed corpus ICE-GB, assuming that the differences between AmE and BrE are likely to be minor in this respect. Employing this relatively small parsed corpus allows us to do trial runs where we can obtain a complete set of results by auto-matic search before attempting to perform analogous searches on the unparsed COHA.

12 This does not mean that they can occur in all instances of such structures. In particular, Hud-dleston and Pullum (2002: 161) note that “Most but not all catenative verbs allow their non-fi-nite complement to be headed by the perfect auxiliary have.” For instance, they contrast He seemed/*began to have offended them.


3.1 ICE-GB

The British Component of the International Corpus of English (ICE-GB; Nelson, Wallis, and Aarts 2002) is a 1-million-word parsed corpus of BrE, synchronically sampled in the early 1990s and comprising both spoken and written material (60 % and 40 % respectively). It was grammatically analysed, using a phrase structure grammar based on Quirk et al. (1985), by a small team of linguist annotators at the UCL Survey of English Usage, and was first published in 1998. The parsing of all 84,000 “sentences” (i.e. punctuated sentences in writing and putative “sentences” in the speech data) involved a combination of automatic methods, specialist computer software, and patient manual correction, and took around 15 person years. Such an effort would clearly be impossible in the case of the 400-times larger COHA corpus.

For our present purposes, the main advantage of this corpus is not simply that the corpus has been parsed systematically according to a grammar scheme, but that human annotators have been able to address structures unanticipated by a formal grammar or parsing algorithm. They have also been able to disambig-uate alternative possible analyses using pragmatic and semantic world knowl-edge. Human errors will undoubtedly exist, but are non-deterministic, unlike the errors of an automatic parser. From a statistical point of view, human annotation substantially improves overall precision by covering all cases, while errors do not introduce a systematic bias to results. Wallis (2014) discusses the classes of evidence that it is possible to derive from a parsed corpus: here we are interested in frequency evidence.

ICE-GB and its sister corpus, DCPSE, are supplied with the ICECUP research tool, which uses a powerful grammatical query system termed Fuzzy Tree Frag-ments (FTFs). These are visual grammatical query structures that the program matches against parsed trees and sentences in the corpus to retrieve results. The software visually exposes both the query and matched sentences, so that it is very easy to see how any given query matches, or fails to match, a particular tree in the corpus.

As a result, it is simple to experiment with numerous alternative queries and learn the grammar scheme as one works with the corpus. It is even pos-sible to identify some portion of a parse tree and abstract an FTF from it. This “cyclical exploration” method allows novice and expert users to gain a high level of research competence and also permits grammarians committed to another scheme (e.g. that of Huddleston and Pullum 2002) to translate their query into the ICE (i.e. the Quirk et al. 1985) grammar framework to obtain data. Even when one framework does not have a one-to-one relationship with another, multiple queries can be performed. In summary, using ICE-GB as a test micro-corpus has


two big advantages: it is fully parsed, and it is possible to refine queries in the light of the results they obtain.

Figure 4: The ICECUP interface with an example FTF (left) representing a to-infinitival perfect, matching the ICE-GB tree (right). Trees are drawn from left to right for reasons of space.

3.2 Frequency profile of contexts of occurrence in ICE-GB

We used two FTF searches to identify the relevant structures in ICE-GB: – Particle to followed by a VP containing an auxiliary with the features “perf,

infin”, with the particle and the VP being children of a clausal node (see Figure 4, left).

– A VP containing (i) an auxiliary with the feature “semi” and containing the lexical form to, followed by (ii) an auxiliary with the features “perf, infin”, with the VP as child of a clausal node. The FTF for this is depicted in Figure 5.

The second search finds examples where the perfect follows a verb treated as a “semi-auxiliary”, e.g. seem, appear, ought, as in (10).

(10) Ah we seem to have lost James there but never mind. (ICE-GB S2A-008)¹³

In the ICE grammatical analysis, seem is treated as a semi-auxiliary within a verbal group headed by the main verb lost. The entire verbal group seem to have

13 Examples from ICE-GB are cited with their identifying text codes. An initial “S” or “W” indi-cates spoken or written material, respectively. Note that, although COHA contains written data only, this includes a considerable amount of dialogue in fiction and plays.


lost follows Quirk et al. by being labelled as “VP”, but excluding elements fol-lowing the main verb such as objects and clausal adjuncts. The first search finds examples within a range of different structures, for example following a verb other than a semi-auxiliary (as in (11)) or following an adjective (as in (12)).

Figure 5: FTF for semi-modal to-infinitive structures.

(11) A scientist named Bridgeman claimed to have shown that there was a devia-tion from Ohm’s Law at very high current densities […] (ICE-GB S2A-041)

(12) I am sorry to have hounded you over the past few days […] (ICE-GB W1B-020)

In the present research we follow an analysis where auxiliary verbs (such as the modals and progressive be) and “semi-auxiliaries” (like seem) are treated, together with verbs like claim in (11), as catenative verbs, i.e. verbs taking a non-finite complement (see, e.g., Huddleston and Pullum 2002: 1209–1220 for a discussion of alternative analyses and arguments in favour of the catenative analysis). Therefore, for our purposes we combine the results of the two searches above. This obtains nearly 200 cases, which we examine and classify according to the function of the to-infinitival perfect clause. The combined results are pre-sented in Table 1.¹⁴ The label “postmodifier” is used, as in the corpus analysis, to include both complements and modifiers within the phrase. The distinction is sometimes difficult to apply and is not necessary for present purposes.

The overwhelming majority of examples (84 %) function as complement of a verb. Another 11 % function as postmodifier within an adjective phrase or noun phrase, while the remaining 6 % fill several different clausal functions. This sug-gests that a collocational analysis of verbs occurring with the to-infinitival perfect

14 Four examples were excluded from the final counts as their status was dubious.


will be worthwhile: something that can be readily carried out in an unparsed corpus like COHA.

Table 1: Functions of to-infinitival perfect clauses in ICE-GB.

Function N %

complement of verb 161 84postmodifier in AJP 14 7postmodifier in NP 7 4extraposed subject 6 3adjunct 3 2subject complement 1 1

Total 192 100

Catenative verbs taking to-infinitival complements can occur in different struc-tural patterns. Further examination of the complement-of-verb examples from ICE-GB provided additional relevant information about the frequencies of differ-ent structural types.

We follow the analysis of Huddleston and Pullum (2002: 1177–1178), which divides catenative structures into “simple” and “complex” according to whether an intervening noun phrase occurs between the catenative verb and the to-infini-tive in the active construction.¹⁵ This noun phrase can often become the subject of a passive in the matrix clause, and such structures are also included as “complex” catenatives although there is no intervening noun phrase in the actual surface structure. This gives us the three structural possibilities illustrated in (13)–(15).

(13) The Iraqis claim to have shot down fourteen aircraft. (ICE-GB S2B-008) (active, simple catenative)

(14) Does he consider sampling to have been an important innovation? (ICE-GB S2B-023) (active, complex catenative)

(15) Candidates are usually expected to have reached a standard well above the pass mark in their qualifying examinations. (ICE-GB W2D-007) (passive, complex catenative)

15 This is an area of the grammar where there are competing analyses of the grammatical status of the intervening noun phrases with different verbs. Discussion of these issues is beyond the scope of this paper. The division between “simple” and “complex” is, however, valuable for our present purposes regardless of the grammatical analysis chosen.


Example (13) is active and there is no intervening noun phrase, so this is a simple catenative structure. Example (14) is active and has an intervening noun phrase (sampling), so this is a complex catenative structure. Finally, (15) is passive and can be compared with an active counterpart with an intervening noun phrase (e.g. We usually expect candidates to have reached…), so is classified as a complex catenative structure. The frequencies of these structures are presented in Table 2.

Table 2: Catenative structures containing to-infinitival perfect clauses in ICE-GB.

Structure N %

active, simple catenative 126 78passive, complex catenative 29 18active, complex catenative 6 4

Total 161 100

The majority of examples (78 %) involve simple catenative structures with the matrix clause in the active voice. Another 18 % involve complex catenative struc-tures with the matrix clause in the passive voice. This means that 96 % have no intervening noun phrase in the surface string. This is useful information for plan-ning data retrieval in an unparsed corpus like COHA, where it is difficult to allow for intervening noun phrases of widely varying lengths without also including irrelevant examples. For instance, consider the intervening noun phrase (indi-cated in square brackets) in the following example:

(16) He estimated [the slope at the time in nineteen eighty-four] to’ve been some-thing like one in four. (ICE-GB S2A-067)

In our parsed corpus, searches can allow for any intervening noun phrase, even one longer than the eight-word string above, but this is not possible in an unparsed corpus.

Finally, we also identified the verbs which occurred in the 161 examples of verbs taking to-infinitival complements with the perfect. There were 29 differ-ent verb lexemes, but the two most frequent accounted for over half of the data: seem (34 %) and appear (22 %). The next most frequent verbs, ought and claim, accounted for only 4 % each. This suggests that searches for the most frequent verb collocates of our target structure in COHA will enable us to retrieve large proportions of the relevant structures.


4 Collocations and semantic categories in COHAOur ICE-GB results lead us to expect that the great majority of to-infinitival clauses containing a perfect auxiliary in COHA will occur as the complement of a verb, i.e. in catenative constructions (the proportion in ICE-GB being over 80 %). We further expect that only a small minority of these catenative constructions will include an intervening noun phrase between the catenative verb and its infini-tival complement (the proportion in ICE-GB being only 4 %).

We cannot specify the proportions in COHA as precisely as for ICE-GB, but we can give reasonable estimates by modifying our original search string for the to-infinitival perfect. If we add an immediately preceding verb to the string (using the tag “[v*]” to find any verb), the results comprise 67 % of the total found by the original string. That search excludes not only those catenative constructions with intervening noun phrases, but also those with other intervening material between the catenative verb and to (e.g. seems to me to have, claimed not to have). We estimate that genuine examples of these kinds comprise no more than another 10 % of the total found by the original string with no preceding verb.¹⁶ Thus the overall proportion of to-infinitival perfects occurring in catenative con-structions (around 77 %) seems to be rather lower than in ICE-GB, but of a similar order of magnitude.¹⁷

As we saw with ICE-GB, we have good reason to believe that the lexical seman-tics of the preceding catenative verb paradigmatically frames possible alternation patterns. Our next step therefore involved identifying these collocates.

We retrieved the left-hand verb collocates of the to-infinitival perfect string using the corpus interface “List” function – search string: “to have [v?n*]”; col-locates: “[v*]” (i.e. any verb), left: 2, right: 0; group by lemmas; increase hits fre-quency above the 100 default until a full list is displayed (see Appendix B for a screenshot).¹⁸ This obtained a list of 402 verb lemmas (or lexemes) in order of declining frequency, with a total number of 44,321 tokens. This, as one would expect, is almost identical to the number of tokens retrieved by the string search

16 This estimate is based on an examination of data retrieved by search strings that allow for intervening material of various kinds in this position, with account taken of irrelevant examples found by such searches. Note that these comments on proportions apply to COHA taken as a whole. Strictly, we cannot assume that the frequencies of the patterns remain completely stable over this 190-year period. For instance, while the overall frequency in COHA of our to-infinitival string immediately preceded by a verb is 67 %, the proportions per decade range from 59 % to 71 %, with a tendency to gradually increase over time.17 Recall that ICE-GB represents 1990s BrE, with 60 % being spoken data.18 That is, it was necessary to experiment with increasing the default hit frequency of 100 until no further examples were found by further increasing the limit (in this case, to 500, say).


“[v*] to have [v?n*]” mentioned above; the data again exclude those examples with intervening material between the verb and to. Examples include both active, simple catenative structures and passive, complex catenative structures.

The results of our search present a lexical profile of the to-infinitival perfect in the catenative construction. Table 3 lists the top 30 lemmas in rank order with individual frequencies, cumulative frequencies, and cumulative percentages of the overall data. These 30 items account for 95 % of retrieved tokens – the “top 95 percentile” of the data. We reasoned that although the remaining “long tail” of 5 % of tokens may contribute to the overall pattern of change, precisely because

Table 3: Catenative verb immediate collocates of the to-infinitival perfect in COHA: top 95 percentile.

Rank Lemma Category Frequency Cumulative %

1 SEEM seeming 17,174 17,174 39 2 APPEAR seeming 5,060 22,234 50 3 SAY cognition and saying 4,964 27,198 61 4 OUGHT modality 3,998 31,196 70 5 BE modality 1,791 32,987 74 6 REPORT cognition and saying 1,271 34,258 77 7 CLAIM cognition and saying 970 35,228 79 8 BELIEVE cognition and saying 957 36,185 82 9 KNOW cognition and saying 953 37,138 8410 LIKE – 594 37,732 8511 THINK cognition and saying 429 38,161 8612 UNDERSTAND cognition and saying 420 38,581 8713 ALLEGE cognition and saying 399 38,980 8814 REMEMBER – 385 39,365 8915 FIND cognition and saying 334 39,699 9016 HAVE modality 264 39,963 9017 PROVE – 256 40,219 9118 HAPPEN – 244 40,463 9119 PROFESS cognition and saying 165 40,628 9220 PRETEND – 156 40,784 9221 SHOW cognition and saying 150 40,934 9222 CONSIDER cognition and saying 140 41,074 9323 REPUTE cognition and saying 121 41,195 9324 MEAN prospective meaning 120 41,315 9325 WISH prospective meaning 119 41,434 9326 ESTIMATE cognition and saying 115 41,549 9427 PRESUME cognition and saying 112 41,661 9428 EXPECT prospective meaning 106 41,767 9429 HOPE prospective meaning 102 41,869 9430 SEE cognition and saying 98 41,967 95


they are low in frequency, they are unable to explain the scale of the change seen in Figure 3. If they declined to zero occurrences, they could only explain at most a 5 % decline.

We assigned most of the top 30 collocates to groups based on their meaning and grammatical properties, as summarised below. Five verbs were not assigned to any grouping, either because of grammatical or semantic differences, and/or because they showed two distinctly different uses in the data and so did not fit any one group. The rationale behind the groupings will be explained below. The distribution of the lemmas is plotted in Figure 6, and the distribution of the groups in Figure 7.

By far the most frequent collocate is seem, accounting for 39 % of the data. The next most frequent is appear, with 11 %. Thus, these two semantically similar items account for half the data.¹⁹ Similar to appear in frequency is say (11 %), followed by ought (9 %). After that there is another steep drop to be (4 %) and thereafter an exponential decline. This distribution, visualised in Figure 6, can be thought of as an example of Zipf’s law applied to a single grammatical “slot”.

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

SEE

M A

PPEA

R S

AY O

UGHT

REP

ORT

CLA

IM B

ELIE

VE K

NOW

LIK

E T

HINK

UND

ERST

AND

ALL

EGE

REM

EMBE

R FI

ND H

AVE

PRO

VE H

APPE

N P

ROFE

SS P

RETE

ND S

HOW

CON

SIDE

R R

EPUT

E M

EAN

WIS

H E

STIM

ATE

PRE

SUM

E E

XPEC

T H

OPE

SEE

F (lemma)

BE

Figure 6: Frequency distribution of the top 95 percentile of catenative verb lemmas (as in seem to have, etc.) in COHA (all data).

It is notable that many of our top 30 verbs are broadly concerned with the areas of evidentiality and modality: i.e. with indicating the source or reliability of the evidence for a proposition, or expressing the speaker’s commitment to the truth

19 These two items were also the most frequent in ICE-GB, where the share for seem was 34 %, while that for appear was 22 %, twice as high as in COHA.


of a proposition or attitude to the desirability of a situation being actualised. They are concerned with what is believed, known, said or reported to be the case, what appears to be the case or is shown to be the case, with what should be done, and so on. A related semantic-grammatical property shared by a striking number of the verbs is that of occurring with a raised subject (either in the active, or in the passive where it is a “derived” raised subject).

A raised subject differs from an ordinary subject in that it does not have a semantic role in relation to the matrix verb, only having one in relation to the verb in the subordinate clause, and this semantic distinction is reflected in various aspects of grammatical behaviour (see, e.g., Huddleston and Pullum 2002: 1194–1198, 1200–1203).

For example, the “seeming” group verbs take a raised subject in the active, simple catenative structure (described in Section 3.2). In He appears to know them, it is not “he” who appears, but rather “he to know them” that appears to be the case (compare It appears that he knows them, with it as “dummy” subject). It is as if he has been “raised” into subject position out of the lower clause where he semantically belongs. In contrast, a verb like offer takes an ordinary subject. In He offered to help us, the subject he has a semantic role in relation to offered as well as to help.

The verbs in the “cognition and saying” group take a raised object, which differs from an ordinary object in an exactly parallel way. Thus, in She believed him to be dead, it is not “him” that she believes, but rather “him to be dead” (compare She believed that he was dead). It is as if him has been “raised” to object position out of the lower clause where him semantically belongs. By contrast, in She persuaded him to leave, him has a semantic role in relation to persuaded as well as to leave. In the passive counterpart of our raised-object example, He was believed to be dead, we can consider he to be a “derived” raised subject. Such passives are frequent in our data.

The raised subject property seems to be semantically connected to the “modalising” or “evidentialising” contribution that many of the collocate verbs make in relation to the situation described in the infinitival complement clause. We can see this kind of contribution by comparing a simple construction like He was dead with He seemed to be dead or He was believed to be dead. The catenative verbs relate semantically to the entire situation of the complement clause, and do not assign any further semantic roles to explicitly expressed participants; there is an implicit “believer” role in the example with believed, but its importance is downplayed by its being left implicit. By contrast, if we passivise our ordinary- object example to get He was persuaded to leave, we see that the verb persuaded does assign a semantic role to the subject he (the same role as when it is the object in the active counterpart).


In categorising and describing our verbs, we have taken account of properties used or noted in the classification of catenative verbs provided in Huddleston and Pullum (2002: 1225–1245), including:

– occurrence in the simple and complex catenative constructions (see Section 3.2 above): simple only, both simple and complex, or complex only,²⁰

– raised versus ordinary subject or object, – possibilities for passivisation, – alternation with structures involving finite subordinate clauses.

Huddleston and Pullum also cover the formal type of non-finite complement taken (to-infinitival, bare infinitival, gerund-participial, past-participial, or more than one of these). This dimension will be largely ignored here, as our focus is on to-infinitivals (although the possibility of alternation with a gerund-participial clause is relevant for some verbs).

A further relevant property concerns the temporal interpretation that catena-tive verbs allow for their complement in the absence of the perfect, as discussed elsewhere in Huddleston and Pullum (2002: 160–162). For instance, they contrast the interpretations of I remembered/enjoyed/intended going to Sydney at Christ-mas. The time of going to Sydney at Christmas is anterior to that of remembering, simultaneous to that of enjoying, and posterior to that of intending (where the going may not have been realised). With some verbs, the choice of temporal inter-pretation depends on the aspectuality of the complement clause. For instance, compare She may be in her office (imperfective: simultaneous) and She may resign (perfective: posterior). The temporal interpretation of the non-perfect is relevant to the effect of applying the perfect.

It should be noted that some verbs have two or more different uses or mean-ings which are assigned to different categories in Huddleston and Pullum’s classi-fication. This is the case for some of our collocate verbs, as we note below. In some instances, such divided usage means we have not assigned a verb to any category; in others, a verb has been assigned to a category based on the use involved in the great majority of examples.

We used the above criteria, along with other semantic properties, to formu-late principal semantic-grammatical classes (here termed “verb groups”) for each verb lemma, taking the most frequent verbs first. Although we are not principally

20 For our purposes, we are considering occurrence in the simple and/or complex catenative constructions with a to-infinitival complement only. Huddleston and Pullum’s classification considers other formal types of non-finite complement also (e.g. gerund-participial). A verb can occur in both simple and complex constructions, but with different formal types of complement (e.g. gerund-participial and to-infinitival).


interested in the possibility of alternation between lemmas (e.g. choosing seem over appear and vice versa), by grouping categories in this way we attempt to isolate contexts of variation between the to-infinitival perfect and other struc-tures, thereby allowing us to pose the question: where are the perfects going?

These categories, like the lemmas, are similarly exponentially distributed. Half the data is explained by the first, “seeming” group, and most of the remain-der falls under the next three groups. Figure 7 summarises the top of this distri-bution.

0 5,000 10,000 15,000 20,000

seeming

cognition and saying

modality

prospective meaning

F (group)

Figure 7: Frequency distribution of the top four verb semantic groups, COHA (95 percentile, all data).

The proportion of cases in each of these semantic groups is not constant over time, however, as Figure 8 reveals. There may be a number of potential reasons why the “seeming” category, while consistently dominant, falls to a minimum in the 1930s before rising again, whereas the three “modality” verbs peak in the 1870s. In each group there are two types of potential explanation: that the contextual opportunity for writers to use this type of expression varies, or that the decision to express “seeming”, “modality” etc. with the perfect is varying. This change in share is substantial. Both the “modality” and the “cognition and saying” groups vary by as much as 100 % (10–20 % and 20–40 % respectively).

However, we also know that this variation within the share of perfects is found within a declining form (see Figure 3). Figure 9 allows us to visualise the decline of each group relative to the decline of the perfect to-infinitive, expressed here by the 95 percentile “Total” line. From this perspective, we can see different verb groups decline over different periods and at different rates: first “seeming” (1820–1930: ~60 % fall), then the “modality” group (1880–1950: 75 %), and finally “cognition and saying” (1970–2000: 60 %, alongside a further 40 % decline in “seeming”). The probabilities of each group (or individual verb) sum to the total.

In the following subsections we look at each of these groupings in turn, and then consider the remaining five uncategorised verbs in the 95 percentile.


0%

10%

20%

30%

40%

50%

60%18

20

1830

1840

1850

1860

1870

1880

1890

1900

1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

“seeming”

“cognition and saying”

“modality”

p(verb group | to-infinitive perfect)

Figure 8: The changing meaning of the to-infinitive perfect: variation in the semantic distribu-tion of the catenative verb group over time (as proportion of the top 95 percentile).

0.000

0.001

0.002

0.003

0.004

“seeming”

“cognition and saying”

“modality”“prospective meaning”

Total

1820

1830

1840

1850

1860

1870

1880

1890

1900

1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

p(verb group | present perfect + past)

Figure 9: Perfect meaning variation (95 percentile total, catenative verb groups and remaining lemmas) plotted against the present perfect + past baseline.


4.1 The “seeming” group

Seem and appear are both semantically and grammatically similar. Here are a couple of examples from the corpus, with the catenative verbs in bold:

(17) His old age seems to have been as quiet and happy as his youth was restless and miserable.(1821:MAG NorthAmRev)

(18) By then, the addiction appeared to have ended Howe’s ability to earn major league money again […](1991:MAG SportsIll)

Both verbs occur in the simple catenative construction only and take a raised subject. In the absence of a perfect infinitive to express anteriority, the time of the complement situation is interpreted as being simultaneous with that of the matrix situation (e.g. He seems/appears to like her). Both verbs can also take a finite subordinate clause in the impersonal construction.²¹

Compare the above examples with the following:

(17’) […] it seems that his old age was as quiet and happy as his youth was restless and miserable.

(18’) […] it appeared that the addiction had ended Howe’s ability to earn major league money again.

Happen is the only other verb in the top 30 list which shares all the above proper-ties, but as it is different in meaning and far less frequent than seem and appear we have not included it in the same group.

We will first plot the proportion of both verbs out of the present perfect + past baseline over the time period. Figure 10 shows that both falls are reasonably con-sistent with a logistic decline (r2 = 0.8952 and 0.7966 respectively).

21 This construction is sometimes treated as an instance of extraposition, but differs in that the subordinate clause cannot fill the actual subject position: *That Dunlap was governed chiefly by caprice seems. See Huddleston and Pullum (2002: 960–962).


0.0000

0.0005

0.0010

0.0015

0.002018

20

1830

1840

1850

1860

1870

1880

1890

1900

1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

p(to-infinitive perfect | present perfect + past) “seeming”

SEEM

APPEAR

Total

Figure 10: “Seeming” perfect to-infinitive over time, as a proportion of the present perfect + past baseline, COHA.

0.0000

0.0005

0.0010

0.0015

0.0020

1820

1830

1840

1850

1860

1870

1880

1890

1900 1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

p(to-infinitive perfect | present perfect + past) “seeming”

past

present

Total (present+past)

Figure 11: “Seeming” verbs, present tense and past tense, plotted against the present perfect + past baseline, COHA.


Although these appear to be approximately parallel, it turns out that the ratio between seem and appear in the to-infinitive perfect is not constant over time. Seem increases its share of the “seeming” verbs from 62 % to 85 % between 1820 and 1910, before stabilising and gradually declining from 1950 onwards.

Present tense forms of seem comprise 64 % of the data, against 28 % for the past tense and just 8 % for other, non-finite forms. So another way of subdividing the data would be to examine present and past tense forms.

Figure 11 compares the to-infinitival perfect decline following present and past tense forms with the total of the two. This shows a steeper decline with present tense contexts than past tense (over the period 1820–2000: around 80 % for present, 60 % for past). As a result, the proportion of past forms climbs from around 25 % of all “seeming” perfects in 1820 to about 35 % by 2000.²²

How might we study the alternation of individual verbs between perfect and non-perfect patterns? The non-perfect is not in general able to replace the perfect in the infinitival complements of the seeming verbs without a change of meaning (involving loss of anteriority). For instance, His old age seems to be quiet and happy would imply he is still alive and enjoying his old age, unlike (17) above (with seems to have been). Nor, to take an example with a non-stative lower verb, could we replace The train seems to have stopped with The train seems to stop.

We therefore need to explore other possible alternates. One possibility is that finite complement clauses, as in (17’) and (18’) above, are being used instead of infinitival complement clauses. It is not possible in an unparsed corpus to retrieve full and precise data for finite complement clauses containing a past tense verb or present perfect construction, but we were able to use multiple search strings to obtain an indicative sample of these data.²³

For both seem and appear considered separately, we evaluated the proba-bility of selecting the to-infinitival perfect rather than the finite construction, i.e. the probability of selecting examples like (17) rather than (17’) and (18) rather than (18’). We found that seem and appear behave slightly differently: the prob-ability of choosing the to-infinitival perfect with seem declines from 85–90 % to around 75 % of cases, with the main fall from 1950 to 2000; with appear, on the other hand, the probability is relatively stable. The picture is clearly not one of simple replacement: this alternation cannot explain the decline in the to-infini-tival perfect with the “seeming” verbs.

22 All three lines fit logistic curves with approximately 90 % accuracy, but this does not mean that past and present have a constant ratio.23 Our searches allowed for examples with various subordinating conjunctions (that, like, as if/though) or none, and were restricted to those with pronoun subjects, which gave more reliable results.


If they are not subject to simple replacement by finite complements, what is happening to these to-infinitive perfect “seeming” cases? Authors may be choos-ing to express this type of meaning less frequently or they may be finding other ways to express similar meanings. Possible alternative forms include adverbs such as seemingly and apparently (cf. Apparently, the addiction had ended Howe’s ability…), or parenthetical finite clauses with seem (as in He had forgotten, it seemed), to mention just a few. As acknowledged earlier, there is no clear limit to the set of possibilities, and investigation of them all is beyond the scope of the current paper.

4.2 The “cognition and saying” group

We categorised 16 verbs as belonging to what we termed a “cognition and saying” group, borrowing a term used informally by Huddleston and Pullum (2002: 1234). This group accounts for the second largest share of the data after the two “seeming” verbs. Most of these verbs occur in the complex catenative construc-tion only, take a raised object, and tend to occur frequently in the passive. In fact, the passive is obligatory with some of these verbs, including say, the high-est-ranking of this group in our top 30 list. Here are a few examples from our data:

(19) She was said to have become insane on receiving the news of her son’s ship-wreck […](1867:FIC AmbroseFecitThe)

(20) He is reported to have enlisted at Marseilles […](1844:FIC EdmondDantès)

(21) More than 2 million North Koreans are believed to have died of famine from 1994 to 1997.(2003:NEWS USAToday)

(22) If a man is known to have killed another, the killer likewise is put to death.(1941:MAG NatGeog)

Recall that our COHA search string could only find immediate verb collocates of the to-infinitival perfect, which does not allow for an intervening noun phrase as typically found in an active complex catenative construction. As a result, passive constructions dominate the data for this group of verbs to an even greater extent than otherwise, so that constructions with “derived” raised subjects are domi-


nant. Like the “seeming” verbs, these verbs can also occur in constructions with a finite subordinate clause complement and dummy it as subject (extraposition constructions in this case). Compare (19) and (20) above with (19’) and (20’).

(19’) It was said that she {became/had become} insane on hearing the news of her son’s shipwreck.

(20’) It is reported that he (has) enlisted at Marseilles.

The “cognition and saying” verbs are also similar to the “seeming” verbs in that the time of the catenative complement clause is, in the absence of the perfect, interpreted as simultaneous with that of the matrix clause (e.g. She was said to be living in Paris; He is known to be in the building).

We do find some active, complex constructions in our data, in instances where a potentially intervening noun phrase occurs in an “extracted” position, due for instance to relativisation (23) or to preposing to pre-subject position (24).²⁴

(23) The man whom you believe to have stolen a million is my own brother. (1869:FIC TriflesChristmas)

(24) This person he knew to have been an active confederate […] (1835:FIC HorseShoeRobinson)

While most of the verbs in this group occur with a to-infinitival complement in the complex catenative construction only, there are two which can occur in either simple or complex constructions: claim and profess. The overwhelming major-ity of our data for these verbs involves the simple construction, as in (25) and (26); example (27) shows one of a small number of passives found for claim (none were found in the data for profess).

(25) He claims to have been at work in the Park stables at the time of the assault.(1885:NEWS NYT-Reg)

(26) […] Dickens professed to have been greatly inspired and influenced by Field-ing and Smollett […](1925:MAG Harpers)

24 The use of terms like “extracted” and “preposing” is not intended to imply a literal derivation in the speaker’s mind which starts from an underlying structure with the noun phrase in ques-tion in the intervening position.


(27) Take one of the principal points of argument, the evil which is claimed to have resulted from the attempt to study so many of the practical sciences.(1851:MAG NewEngYaleRev)

The simple construction with these verbs involves an ordinary rather than a raised subject, so they differ from the other verbs in this group where most of the examples involve passive constructions with (derived) raised subjects. However, they bear semantic similarities to other members of the group, and share with them the ability to occur with a finite subordinate clause, and the temporal inter-pretation (in the absence of the perfect) of simultaneity of the infinitival subor-dinate clause and the matrix clause. We have therefore included them in this group.

Note that the verb see is included in this group because, in the kind of usage in our data, it often has the sense of mental inference rather than sensory per-ception, and in this usage has similar grammatical properties to the rest of the group, as pointed out by Huddleston and Pullum (2002: 1236–1237). This usage is illustrated in (28).

(28) Projected into the past, the Earth’s rotation is seen to have been a very little faster than its present rate.(1951:MAG Harpers)

A number of other verbs in the group seem to go beyond the central semantic area of “cognition and saying” to refer to processes of examining or presenting evidence for something: find, show, and estimate. This is a semantically linked area, and they share the other properties of the group. Examples are given in (29) to (31).

(29) Roger was asked to leave college because he was found to have been black-mailing the wife of one of the business men in town. (1932:FIC MardiGrasMurders)

(30) Many were themselves accused of witchcraft; and noble ladies were shown to have dabbled in mystic arts […] (1869:NF MemoirsExtraordinary)

(31) The value of the ore shipped from Creede in February is estimated to have been $850,000. (1893:NEWS NYT-Ed)


The first task in examining variation in this group is to plot it over time against the general present perfect + past baseline, as in Figure 12. Against this baseline, the total for the group falls to about 30 % of its 1920 value (it declines by 70 %) from 1920 to 2000.²⁵

Over the same period, the predominant form, say, which initially accounts for half the forms in the group, declines to a quarter-share by 2000. The result is an increase in diversity of other forms, with first report rising (1910–1940: +150 %) before falling, and then claim rising (1950–1980: +75 %), to the point that claim and say are not significantly different in frequency by 2000.²⁶

0.0000

0.0001

0.0002

0.0003

0.0004

0.0005

0.0006

0.0007

0.0008Total

SAY

REPORT CLAIMBELIEVEKNOW

1820

1830

1840

1850

1860

1870

1880

1890

1900

1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

p(to-infinitive perfect | present perfect + past) “cognitionand saying”

Figure 12: “Cognition and saying” perfect to-infinitive over time, against the general present perfect + past baseline, in COHA.

25 Before 1920, the total line fluctuates, possibly due to text-sampling variation, but is most meaningfully interpreted as being reasonably steady. The same cannot be said for the sharp de-cline in the twentieth century.26 Note that with multiple competing forms, we should not be surprised that these do not all follow a simple logistic “S-curve” rise or fall. See http://corplingstats.wordpress.com/2015/05/27/logistic-multinomial.


Given that we have a complex pattern of change here, it may be worth considering whether in fact verbs are more or less likely to be used in a perfect construction than in a general to-infinitive. That is, suppose we compare say in the perfect to-infinitive (e.g. said to have been) versus the same verb with a non-perfect to-in-finitive (e.g. said to be). Are some verbs more likely to be used in the first con-struction than other verbs?

If we examine the distribution of lemmas synchronically (i.e. for all periods of COHA combined), we obtain the distribution in Figure 13. For the purposes of comparison, we have maintained the same ordering as in Figure 6, i.e. say has more to-infinitival perfect cases than the others, etc. We can see that about 50 % of to-infinitival cases of allege and report are found in perfect constructions, which we might think of as a simple measure of their “attraction” to the perfect.

If we reorder this synchronic sequence according to attraction probability, and then take the difference in rank between the two sequences, we find that allege, repute and estimate increase in rank most substantially, and think and find fall. These verbs are relatively infrequent and so do not affect the overall pattern of diachronic change plotted in Figure 12. Nonetheless, they indicate that there is a lot of variation within this one group, and that some verbs may be co-oc-curring with the perfect as an “idiom chunk”.

0.0

0.1

0.2

0.3

0.4

0.5

SAY

REPO

RT

CLAI

M

BELI

EVE

KNOW

THIN

K

UNDE

RSTA

ND

ALLE

GE

FIND

PROF

ESS

SHOW

CONS

IDER

REPU

TE

ESTI

MAT

E

PRES

UME

SEE

p(to-infinitive perfect | to-infinitive) “cognition and saying”

Figure 13: Plotting “attraction” to the perfect. Synchronic distribution of the probability of each “cognition and saying” verb being used in a perfect to-infinitival construction, given it is used in a to-infinitival construction. Comparison of column heights against the 95 % Wilson score intervals indicates that this variation is wide and substantial.

As say is the dominant form and falls so sharply, we tested this verb for possible alternation of the to-infinitival perfect with a finite complement clause contain-


ing a past tense or present perfect (cf. (19’) above). An indicative sample of such clauses was obtained by using several search strings. The results of the compari-son do not show a consistent downward trend. Therefore, the alternation cannot explain the decline observed for say + to-infinitival perfect. Rather, this decline seems to be the result of the be said to-infinitival perfect halving its share of all to-infinitival perfect forms over the period 1940–2000.

4.3 The “modality” group

The “modality” group accounts for the third largest portion of data. It includes the verbs ought, be, and have, which rank at 4, 5, and 16 on the list respectively (see (32) to (34)).

(32) “[…] You ought to have let him do as he wished with his own wigwam. You were unjust”. (1857:FIC RolloInWoods)

(33) He was to have come to me for a music lesson this afternoon, but as usual he sent an excuse. (from dialogue in a play; 1879:FIC LastWord)

(34) To give examples I should have to have taken notes […] (1896:FIC Embarrassments)

These verbs, like the “seeming” verbs, occur only in the simple catenative con-struction, and have raised subjects. However, they differ from that group, and from the “cognition and saying” group, in that they do not take finite clausal complements. They also differ from those groups in terms of possibilities for tem-poral interpretation of the infinitival complement in the absence of the perfect. There are complex patterns involved, but it is noteworthy that the interpretation often involves posteriority rather than simultaneity (e.g. We ought/are/have to visit her).

Each of the three “modality” verbs has changed in frequency over the two centuries captured by COHA. Figure 14 depicts the proportion of the past + present perfect baseline for these three verbs. Was/were is plotted here, as past tense forms account for the great majority of the data for be; see further below.

The variation in ought (the predominant form) explains the verb group pattern until the 1950s, with the other two forms being nearly constant. In the second half of the twentieth century, have increases from nearly zero occur-


rences to converge with the falling ought. Meanwhile was/were declines by 70 % from 1920 to 1990.

As these three verbs are rather disparate in meaning, we consider each in turn.

0.0000

0.0001

0.0002

0.0003

0.0004

0.0005

0.0006

ought

was/were

HAVE

Totalmodals

1820

1830

1840

1850

1860

1870

1880

1890

1900

1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

p(to-infinitive perfect | present perfect + past) “modality”

Figure 14: The rise and fall of the ought to-infinitive perfect, and other “modality” verbs in COHA.

4.3.1 Ought

Ought is considered to be primarily deontic in meaning, though it can also have epistemic readings (e.g. Collins 2009: 53–55); many of our examples, like (32) above, are clearly deontic in meaning, but there are some which are not, as in (35).

(35) “[…] We really ought to have smashed to pieces. It’s strange that the ship with-stood it”. (1989:FIC Eden)


This conveys that one would have reasonably expected the ship and its passen-gers to smash to pieces, given the conditions: an epistemic, or perhaps dynamic (circumstantial), reading. Ought can convey an implication of non-actualisation, as seen in our examples with the perfect: you in (32) did not let him do as he wished; we in (35) did not smash to pieces.

An obvious semantic competitor for ought is should, which, as a modal verb, takes a bare infinitival complement. These items are generally viewed as being semantically very similar, and often interchangeable (e.g. Collins 2009: 52–53). We examine this alternation in Figure 15.

0.00

0.05

0.10

0.15

0.20

(to-) infinitive perfect

all infinitive

1820

1830

1840

1850

1860

1870

1880

1890

1900 1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

p(ought to | {ought to, should})

Figure 15: The rise and fall of ought to against its modal alternate should in COHA.

The dotted line represents the alternation within the perfect (ought to-infinitive perfect vs. should bare infinitive perfect). The dashed line represents the corre-sponding infinitive alternation. The baseline in each case is the set of both alter-nates, ought to and should.

The perfect line correlates with the ought line in Figure 14, suggesting that the ought to-infinitive is indeed seeing a pattern of replacement. In this pattern, ought to increases relative to should over the period 1830–1880 before declining in the


twentieth century. This peak presages a parallel pattern of modal replacement in the infinitive form overall, which outnumbers the perfect by around 10 : 1, shown as a darker dashed line in Figure 15.

4.3.2 Be

Be²⁷ can be used to express deontic necessity (e.g. You are to come home at once), but the examples typical of our data, like (33) above, express meanings involving schedules or plans – meanings that are perhaps “more temporal than modal”, in the words of Collins (2009: 85). As Collins notes, the perfect can be used to imply non-actualisation: this certainly characterises many of our examples, including (33) above, (36) and (37).

(36) “I see not here your son, young Henry Winthrop,” said the Lady Arabella. “I thought he was to have been of our company”. (1846:FIC MyrtisWithOther)

(37) […] she had been left by a man the very day she was to have been married. (1909:FIC ThreeLives)

Often, the non-actualisation is obvious from context. Alternation with a non-per-fect seems possible with examples like these, at least in present-day English, e.g. I thought he was to be of our company; […] the very day she was to be married.

To test this possible area of alternation, we examined the proportion over time of to-infinitival clauses that contained a perfect when preceded by was/were (since be occurs in the past tense in the great majority of our examples with the to-infinitival perfect: about 1,600 out of 1,791 examples). This does not show a clear diachronic trend, so does not support the hypothesis that, over time, speak-ers have increasingly chosen to use a non-perfect in such examples. Be followed by the to-infinitive shows a steady decline per million words in COHA during the twentieth century, which amounts to a decline in a potential context of occur-rence for the to-infinitival perfect.

27 The use of be where it is followed by a to-infinitival complement, being idiomatic, is often referred to as “be to”. However, as we are treating all the verbs under consideration here as cat-enatives taking to-infinitival complements, for consistency we simply refer to be. A similar point applies to the idiomatic use of have (“have to”) discussed in Section 4.3.3.


4.3.3 Have

Have can also be used to express deontic necessity, but this kind of meaning is likely to be infrequent in connection with the anteriority expressed by the perfect – though not impossible, as shown by examples like (38), which describes a requirement relating to anterior time.

(38) To qualify, firms had to have been independent and privately held through 1997. (1998:NEWS CSMonitor)

Many of the examples in our data are somewhat hard to categorise in terms of the type of necessity expressed. Example (39) below is perhaps best categorised as epistemic: it is from a fictional passage which appears to be taking “his” view-point, so we could take it as reporting a judgement on his part. Example (40) could be said to involve dynamic necessity, a situational requirement.

(39) She had to have heard him, but she didn’t move. (1996:FIC SouthernRev)

(40) Besides, in order to be blackmailed one has to have done something disgrace-ful, and I know Harold didn’t. (1941:FIC MurderMakesUsGay)

We examined the proportion of to-infinitival clauses that contained a perfect when preceded by have, and found that it increases significantly from the 1950s to the 2000s. Overall, have followed by a to-infinitival clause increases per million words over the COHA period, but the increase tails off in the last 50 years. The increase in the proportion with a perfect may be due to a rise in epistemic and dynamic uses at the expense of deontic uses; further examination of the data would be needed to test this.

4.4 The “prospective meaning” group

The final group to be considered is the “prospective meaning” group, which accounts for a far smaller share of the data than the other groups. This group comprises the four verbs mean (in the sense “intend”), wish, expect, and hope, as exemplified in (41) to (44).


(41) We meant to have sown a little barley to-day, but the ground is too dry […] (1848:FIC Stranger)

(42) I wished to have purchased a mule, but, though I offered thirty pounds for a sorry one, I could not obtain her […] (1843:NF BibleInSpainJourneys)

(43) “I expected to have been here in time, but these trains are never to be depended on.” (1871:FIC LucyRaymond)

(44) “[…] I hoped to have hidden from you what I have suffered. But it is too late now.” (1851:MAG Harpers)

With these verbs, the time of the infinitival clause situation is, in the absence of perfect marking, posterior to (later than) that of the matrix clause (e.g. I mean to tell him the truth; He hopes to get a pay rise). In the examples, the temporal relationship is also one of posteriority, and the perfect in the subordinate clause expresses counterfactual meaning (e.g. We did not sow barley; I was not able to purchase a mule). Such examples are frequent in the earlier data.

In each of the above, we would be more likely to find a non-perfect in pres-ent-day English: We meant to sow a little barley today; I expected to be here in time, and so on. The decline in counterfactual usages of the infinitival perfect may have been influenced by prescriptivism: such usages were condemned in the eighteenth and nineteenth centuries on the basis that the Latin perfect infinitive expressed only anteriority (Molencki 2003).²⁸ However, we do find some recent examples of the perfect infinitive with “prospective” verbs in our data, such as (45).

(45) Of those having children in the preceding five years, 35 percent said they had not wished to have gotten pregnant when they did […](1994:MAG Americas)

These examples involve the catenative verb occurring in a past tense form, or in a past participle form as complement of perfect have (example (45)). They rep-

28 Molencki (2003) notes that counterfactuality was in fact the earliest meaning of the English perfect infinitive: it was only after a few hundred years, around 1380, that it began to be used for non-counterfactual anteriority.


resent the simple catenative structure, with ordinary subject. The four verbs can also occur in the complex catenative structure (with the intervening noun phrase introduced by for in some instances, e.g. I was hoping for him to change his mind), but this is uncommon in our data except in the case of expect.

Although examples with present tense catenatives are possible, the data for present tense verbs in this group turn out, unusually, to include a considerable number of irrelevant examples, often involving the kind of structure seen in (46).

(46) “Well, then, ask me the question that you wish to have answered.” (1876:FIC Ishmael InDepths)

Here the string to have answered is not a genuine instance of the perfect – an unrelativised equivalent would be you wish to have the question answered.

The overall pattern of change against the “present perfect + past” baseline is a consistent fall, matching a logistic decline of more than 90 % over the period (r2 = 0.8650). This pattern can be glimpsed at the bottom of Figure 9. This fall mirrors the global decline of the perfect of some 80 %.

When considering alternation patterns we decided to focus on past-marked prospective verbs only (i.e. those in past tense form, or in past participle form fol-lowing perfect have).²⁹ This enabled us to exclude the “noisy” data found for the present tense forms, and to focus on examples like (41)–(45) above, since we have good reason to suspect that the use of the perfect has declined in such contexts over time (given the contrasts between earlier and present-day English noted above). We therefore looked at the proportion of to-infinitivals which contained a perfect in the contexts following a past-marked catenative verb. The results, shown in Figure 16, show a decline in this proportion, supporting the hypothesis that speakers’ inclination to use a perfect in these contexts has declined during this period. Given this choice, the decline is even sharper: comparing points on the logistic curve, by 2000 the proportion of perfect forms is only 3 % of the same figure in 1820, or a decline of 97 %.

The fall matches a logistic regression line closely (r2 = 0.9117), indicating that these verbs are being retained but writers are increasingly choosing to use the simple infinitive in preference to the perfect.

29 We used the search string “-[be] wished|hoped|expected|meant” for the context, which in-cludes past tense and past participle forms (identical for these verbs), but excludes a preceding passive be by using the “not” symbol.


0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08“ prospective

meaning”18

20

1830

1840

1850

1860

1870

1880

1890

1900

1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

p(to-infinitive perfect | {perfect, non-perfect})

past

Figure 16: Decline in past-marked “prospective meaning” verbs, evaluated as an alternation between the to-infinitival perfect and to-infinitive, COHA.

4.5 The verb like

The verb like ranks tenth in our top 30 list. Two distinct uses are found in our data. The first use, illustrated in (47), is not found in present-day standard English.

(47) But my father came in so close upon me, he liked to have cut his foot with the pieces. (1833:FIC PoemsProseWritings)

This example appears to mean “he nearly cut his foot, he seemed very likely to cut his foot”, with an implication of non-actualisation (he did not cut his foot), and a raised subject. The OED lists similar examples under like, v.2, 2b: “To look like or be near to doing (something) or to being treated (in a specified manner). Now vulgar and dial. (U.S.)”. Variant forms are found in the data in reference to past time, in addition to liked: like, had like, had liked, as in (48) to (50).


(48) […] he like to have paid dear for his conduct […] (1827:FIC TennesseanANovel)

(49) These exertions had like to have been fatally injurious to Constantia. (1827:FIC Novels)

(50) “I’ve something to tell you, which I had liked to have forgotten!” (1856:FIC DredATaleGreat)

Of 50 instances of had like(d) in the corpus,³⁰ none occur after 1908, but we do find some later instances of like(d), e.g. I like to have died when she asked me that (2000:FI SouthernRev). If we consider all variants together, only about a fifth of the examples occur in the twentieth century. This use of like is clearly a context of potential occurrence for the to-infinitival perfect which has declined over the COHA time period.

The second use of like in the data typically involves a preceding would or should. This use is normal in present-day standard English, but is subject to variation in the occurrence of the perfect at both matrix and subordinate level. Compare (51) to (53).

(51) He would like to have been a KGB agent […] (1987:NEWS Chicago)

(52) “[…] I would have liked to have seen you one last time,” I said. (2007:FIC BloomsdayDead)

(53) “Now, honey,” he began uneasily, “do you think Steve would have liked to see you acting this way?” (1970:FIC PickUpSticks)

The last example is not from our data, but illustrates the possibility of using a perfect in the higher clause and not the lower. The “pastness” expressed by the perfect applies semantically to the higher clause, as reflected in the structure used in (53). However, the perfect is sometimes placed in the lower clause instead (as in (51)) or in both clauses (as in (52), where we might perhaps think of the lower perfect as expressing counterfactuality). Given this variation, it is worth

30 Some of these instances have like tagged as an adjective, so are not included in our collocates data. This applies to (49), where like is indeed arguably an adjective rather than a verb. The OED mentions the use of had like to under like, adj. 9b.


testing whether usage has changed over time to favour a non-perfect in the lower clause. The results, shown in Figure 17, confirm that this is the case, although there is still variation.³¹

The confidence intervals are wide due to the small amount of data in the overall set, particularly in the early decades. We can see that the non-perfect becomes the dominant form around 1850, and its proportion continues to climb until around 1910. The peak in 1950 may be a text-sampling “blip”.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

would|should like to + perfect

1820

1830

1840

1850

1860

1870

1880

1890

1900

1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

p(form | {like perfect, have liked perfect, non-perfect})

would|should have liked to + non-perfect

have liked to + perfect

Figure 17: Variation in alternation of the perfect with would/should + like, in COHA, by decade.

31 We combined the results of two search strings to obtain figures for the variant with a non-per-fect in the lower clause: “would|should have liked to [v?i*] -[v?n*]”, which excludes an infinitive followed by a past participle, and “would|should have liked to be [v?n*]”, to include any passives excluded by the first search.


4.6 The verb remember

The verb remember ranks 14th in our top 30 list. Although it has a cognitive meaning, we treat it separately from the “cognition and saying” group because of its different behaviour.

There are a few examples in our data of remember in the complex catenative construction, such as the passive in (54), which has a raised subject.

(54) Elizabeth Scott Hardin was remembered to have hidden in a cave with her children […] (2003:NF WhereIWasFrom)

However, the vast majority of examples in our data belong to the simple catena-tive construction, with an ordinary subject, as in (55) and (56).

(55) I remember to have felt some one pull me by the hair, before I was utterly senseless. (1835:FIC InfidelFallMexico)

(56) He remembered to have seen a photograph of this celebrity […] (1911:FIC Tante)

These last two examples differ from present-day English usage, where such “ret-rospective” meanings are expressed with a gerund-participial complement, often but not always without the perfect, e.g. He remembered {seeing/having seen} a photograph of this celebrity. The to-infinitival complement in the simple construc-tion is typically used in a different way in present-day English, as in I remem-bered to lock the door (meaning that I remembered that I should lock the door and did so) – a “prospective” meaning which does not lend itself to an expression of anteriority with the perfect. This contrast between the “retrospective” use of the gerund-participial complement and the “prospective” use of the infinitival com-plement with remember is well known by grammarians of present-day English (see, e.g., Quirk et al. 1985: 1193).

The rate of remember followed by a to-infinitival perfect shows a clear decline in COHA (whether per-million-words or against our verb baseline), falling to almost zero by the 1940s. The most recent examples of the kind shown in (55) and (56) above occur in the 1930s (four examples) and 1940s (one example) – apart from an example from a biography published in 1997 which involves quoted speech attributed to a man born in 1782.


Plotting the to-infinitival perfect against the gerund-participial³² in the context of a preceding lemma remember, we can show a temporal pattern of replacement (see Figure 18). By the 2000s, we find 707 instances with the ger-und-participial, as against only one instance with the to-infinitival perfect, which in fact is one of the very few examples of the complex catenative construction in our data (and thus not a genuine alternant in any case). The decline shows a close match to a logistic curve, mirroring in microcosm the overall decline of the perfect (cf. Figure 3).

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

1820

1830

1840

1850

1860

1870

1880

1890

1900

1910

1920

1930

1940

1950

1960

1970

1980

1990

2000

remember

p(perfect | {perfect, gerund participial})

Figure 18: The decline of the perfect to-infinitive after remember, alternation against the gerund-participial.

Other scholars have reported on the changing complementation patterns of remember. For example, Mair (2006) presents data from the OED quotation base for 1700 to 2000, having found available corpora unsuitable for the purpose (for example, ARCHER contained too few instances of the relevant constructions). On

32 The gerund-participial was searched for using “[v?g]”, which includes both perfect and non-perfect instances.


the one hand, Mair traces the origin of the gerund-participial complement back to the late eighteenth century, citing two quotations from the 1770s. On the other, he notes two authentic examples of the “retrospective” use of the to-infinitival perfect occurring as recently as the 1960s and 1970s. It is hard to resist quoting one of these: It is the first play in which I remember to have encountered an actor with a jockstrap which squeaks when pushed (Daily Telegraph, 7 March 1969).

4.7 The remaining verbs

The remaining ungrouped verbs are prove, happen, and pretend, ranking at 17, 18, and 20 respectively in our top 30 list. Prove is ungrouped because it occurs with two different senses in our data, illustrated in (57) and (58).

(57) The Moscow “failure” might yet prove to have been a successful first step. (1977:MAG Time)

(58) The defendant was proved to have struck a horse twenty-five times upon the head with a heavy club […] (1873:NEWS NYT-Ed)

In (57), where prove has the sense of “turn out”, we have a simple catenative structure with a raised subject, as found with the “seeming” verbs. In (58), we have a passive complex catenative structure with a raised subject, as found with the cognition and saying group. The simple structure seems to be dominant in our data, with a substantial minority of the complex.

We have already mentioned that happen shares many properties with the “seeming” verbs, but as it is different in meaning and far less frequent, we have not included it in the same group. (59) and (60) are a couple of examples.

(59) “Meta happens to have fallen in love with a young gentleman named Edmund Innis […]” (1872:FIC DoctorVandykeA)

(60) He happened to have brought along a poem of his on the white wines of Sparta. (1970:FIC WreckageAgathon)

Finally, pretend occurs in our data mainly in the simple catenative construction, with an ordinary subject, as in (61) and (62).


(61) They pretend to have invented the printing press […] (1841:NF PsychologyOrAView)

(62) I hit the floor, pretending to have fainted dead away. (2003:FIC Fudge-a-mania)

These examples seem to represent two somewhat different senses, the first being close to “claim” in meaning. The two senses are not always clearly distinct, however; a clear “claim” sense may belong to earlier usage. The “claim” sense also allows a complex construction, as in (63), one of very few passives found in the data, all from the nineteenth century.

(63) No aid and no countenance can the labors of the apostles be pretended to have received from any secular power. (1842:NF AnInquiryIntoFoundation)

Despite the small figures and large error bars for these individual verbs, we can identify a small decline over the entire period (1820–2000) against the past tense + present perfect baseline for the to-infinitival perfect for both prove and pretend. However, there is insufficient data to identify a similar trend for happen.

5 ConclusionsOur study has focused on the to-infinitival perfect, a subtype of the perfect that has received far less attention than the present perfect. We looked at changing use of this structure in written AmE over the period 1820–2009. Against a baseline of present perfect and past tense verbs, we found a huge overall decline of around 80 % over this time period. We chose this baseline because perfect to-infinitive forms correspond with past-referring contexts, whether expressed by a present perfect or by a past tense verb. Other potential generic baselines, such as words or to-infinitives in general, risk introducing both additional noise and skewing by other factors, such as variation in the temporal orientation of texts sampled.

In order to investigate the decline in more detail, we needed to examine the contexts in which the to-infinitival perfect occurs. We used ICE-GB, with its parse analysis and powerful search tools, to obtain a reliable snapshot of comparable data (albeit 1990s BrE, 60 % spoken) and thereby to ascertain the different kinds of structures we could expect to find in COHA, a much larger corpus with more limited annotation and tools. We found a large majority of examples occurring


as complements of catenative verbs. Our diversion into ICE-GB further allowed us to anticipate the distribution of simple and complex catenative constructions, including passive complex ones. It enabled us to design POS-tag query strings for COHA to obtain results, which, while not guaranteed to cover all cases, were reasonably reliable.

Use of COHA, with its 400 million words covering two centuries of written AmE, then provided ample data for a detailed diachronic study of the catena-tive verb collocates of the to-infinitival perfect. Given the exponential frequency distribution of these collocates, we opted to focus on the most numerous verbs, because changes affecting the “long tail” of 5 % of tokens could not numerically explain the scale of the overall decline in the perfect.

We found that many of the “top 30” collocates express meanings in the broad areas of evidentiality and modality, and so could be thought of as making “evi-dentialising” and “modalising” contributions to the situation described in the infinitival complement clause. Within this broad area, however, we could iden-tify some distinct subsets. We used criteria drawn from Huddleston and Pullum (2002) to group verb lemmas, both semantically and grammatically, taking into account their possible alternation patterns (i.e. the options available to an author as alternatives to using the perfect to-infinitive).

By grouping verbs in this way and studying their change over time, we found a range of different patterns. Some cases – including the “seeming” group, the “prospective” group, and the verb remember – decline according to a logistic curve: the simplest, most consistent form of decline. They therefore conform to the overall pattern of decline, although they make differing quantitative contri-butions to this trend.

Since the “seeming” verbs are most frequent (covering 50 % of the data), their apparently regular decline might lead one to conclude that all verbs would behave like this. This is not the case. Examining the “cognition and saying” group reveals a number of different changes taking place at the same time. First know declines, at the beginning of the period examined. say, the predominant form, holds more-or-less steady until around 1920, and then starts to fall quite sub-stantially. Report and claim expand in use in the late twentieth century, such that by the end of the period the overall decline is combined with a growth in the diversity of forms. Closer study, including genre comparison, would be needed to investigate whether these patterns reflect stylistic changes, or extra-linguistic, social changes of some kind. This group also revealed a striking degree of vari-ation in their “strength of attraction” to the perfect, considered synchronically: another aspect worth further investigation.

Similarly, the “modality” group also covers three verbs which behave very differently. Most obviously, ought rises before falling – a pattern attributed to its


replacement by the core modal should. By contrast, was/were and have are rela-tively constant in their proportion (of present perfect and past verbs) for around a century, until the 1920s, when was/were declines and have increases.

The groups and verbs also show marked contrasts in terms of the alterna-tion processes in which they participate over time. For the “seeming” group, and say (the predominant “cognition and saying” verb), we were unable to identify any clear pattern of replacement of the to-infinitival perfect by a particular form: the most obvious potential alternant, the finite complement clause, could not account for the declines. By contrast, for other groups and verbs, clear patterns of replacement could be identified. We saw replacement by non-perfect to-infin-itival complements for the “prospective” group (earlier I expected to have been here in time vs. present-day I expected to be here in time), and for like preceded by would/should (have), where ongoing variation remains despite the increase in proportions of the non-perfect complement with a superordinate perfect (as in I would have liked to go). A different pattern again was seen with remember, with to-infinitival complements clearly being replaced over time by gerund-participial complements (earlier I remember to have seen him there vs. present-day I remem-ber seeing him there).

Given the wide range of different patterns we have found, what can we con-clude about the decline in this form, and about the way language changes in general? It may appear that the observed decline is simply the “accidental” result of various unconnected changes in the language which combine to create this effect.

However, the fact that the overall trend is progressively downwards suggests that there is indeed a decreasing general inclination to use the to-infinitival perfect, a trend which manifests itself in various ways across different construc-tions and collocational contexts. Some contexts may buck the trend because of other influences, whether linguistic or extra-linguistic.

This view is reinforced by the observation that the same form declines signifi-cantly in COHA in other grammatical contexts besides those we focused on in this study. For instance, further searches show that it also declines following adjective collocates, and in complex active catenative constructions with an intervening pronoun (excluded by our main data). Furthermore, there is a steep decline over the COHA period in another category of non-finite perfect, the gerund-participial perfect (as in Having finished her work, she left early). These observations about declines in related structures suggest there is some more general tendency at work.

Our findings of lexical variation within a more general overall trend fit quite well with recent strands of work on language change within a construc-tional approach (e.g. Hilpert 2013), which stress that grammatical change tends


to be lexically constrained and gradually diffusional in nature rather than rapid and systemic. This is certainly a fertile area for further research. Tracing such complex, lexically conditioned change processes has become much more feasible with the availability of megacorpora spanning long time periods such as COHA.

AcknowledgementsWe are grateful to two anonymous reviewers for their helpful, detailed comments. We would also like to thank Seth Mehl and Martin Hilpert for their comments on drafts of the paper, and Mark Davies for his extensive answers to queries about COHA.

ReferencesBowie, Jill & Bas Aarts. 2012. Change in the English infinitival perfect construction. In Terttu

Nevalainen & Elizabeth Closs Traugott (eds.), The Oxford handbook of the history of English, 200–210. Oxford: Oxford University Press.

Bowie, Jill, Sean Wallis & Bas Aarts. 2013a. The perfect in spoken British English. In Bas Aarts, Joanne Close, Geoffrey Leech & Sean Wallis (eds.), The verb phrase in English: Investigating recent language change with corpora, 318–352. Cambridge: Cambridge University Press.

Bowie, Jill, Sean Wallis & Bas Aarts. 2013b. Contemporary change in modal usage in spoken British English: Mapping the impact of “genre”. In Juana I. Marín Arrese, Marta Carretero, Jorge Arús Hita & Johan van der Auwera (eds.), English modality: Core, periphery and evidentiality, 57–94. Berlin: Mouton de Gruyter.

Callies, Marcus. 2013. Bare infinitival complements in Present-Day English. In Bas Aarts, Joanne Close, Geoffrey Leech & Sean Wallis (eds.), The verb phrase in English: Investigating recent language change with corpora, 239–255. Cambridge: Cambridge University Press.

Collins, Peter. 2009. Modals and quasi-modals in English. Amsterdam: Rodopi.Comrie, Bernard. 1976. Aspect: An introduction to the study of verbal aspect and related

problems. Cambridge: Cambridge University Press.Davies, Mark. 2012. Expanding horizons in historical linguistics with the 400-million word

Corpus of Historical American English. Corpora 7(2). 121–157.Elsness, Johan. 1997. The perfect and the preterite in contemporary and earlier English. Berlin:

Mouton de Gruyter.Elsness, Johan. 2014. The present perfect and the preterite in Late Modern and contemporary

English: A longitudinal look. In Kristin Davidse, Caroline Gentens, Lobke Ghesquière & Lieven Vandelanotte (eds.), Corpus interrogation and grammatical patterns, 81–103. Amsterdam: Benjamins.

Fanego, Teresa. 2007. Drift and the development of sentential complements in British and American English from 1700 to the present day. In Javier Pérez-Guerra, Dolores González-


Álvarez, Jorge L. Bueno-Alonso & Esperanza Rama-Martínez (eds.), ‘Of varying language and opposing creed’: New insights into Late Modern English, 161–235. Bern: Peter Lang.

Hilpert, Martin. 2013. Constructional change in English: Developments in allomorphy, word formation, and syntax. Cambridge: Cambridge University Press.


Huddleston, Rodney & Geoffrey K. Pullum. 2005. A student’s introduction to English grammar. Cambridge: Cambridge University Press.

Hundt, Marianne & Nicholas Smith. 2009. The present perfect in British and American English: Has there been any change, recently? ICAME Journal 33. 45–63.

Mair, Christian. 2006. Nonfinite complement clauses in the nineteenth century: The case of remember. In Merja Kytö, Mats Rydén & Erik Smitterberg (eds.), Nineteenth-century English: Stability and change, 215–228. Cambridge: Cambridge University Press.

Molencki, Rafał. 2003. Proscriptive prescriptivists: On the loss of the ‘pleonastic’ perfect infinitive in counterfactual constructions in Late Modern English. In Marina Dossena & Charles Jones (eds.), Insights into Late Modern English, 175–196. Bern: Peter Lang.

Nelson, Gerald, Sean Wallis & Bas Aarts. 2002. Exploring natural language: Working with the British component of the International Corpus of English. Amsterdam: Benjamins.


Vosberg, Uwe. 2009. Non-finite complements. In Günter Rohdenburg & Julia Schlüter (eds.), One language, two grammars? Differences between British and American English, 212–227. Cambridge: Cambridge University Press.

Wallis, Sean. 2013. Binomial confidence intervals and contingency tests: Mathematical fundamentals and the evaluation of alternative methods. Journal of Quantitative Linguistics 20(3). 178–208.

Wallis, Sean. 2014. What might a corpus of parsed spoken data tell us about language? In Ludmila Veselovská & Markéta Janebová (eds.), Complex visibles out there. Proceedings of the Olomouc Linguistics Colloquium 2014: Language use and linguistic structure, 641–662. Olomouc: Palacký University.


Yao, Xinyue & Peter Collins. 2012. The present perfect in world Englishes. World Englishes 31(3). 386–403.

Yao, Xinyue & Peter Collins. 2013. Recent change in non-present perfect constructions in British and American English. Corpora 8(1). 115–135.


Appendix ATable of COHA sampling distribution (number of words), drawn from http://corpus.byu.edu/coha/old under “Composition of the corpus”. See also Figure 1. This distribution affects our results in the following two main ways: The lower total frequencies for the earlier decades tend to obtain wider confidence inter-vals (lower certainty). The low proportion of magazines in the first decade (com-pounding the absence of newspaper data) causes us to discount this data point.

Decade Fiction Magazines Newspapers NF books Total

1810 641,164 88,316 0 451,542 1,181,0221820 3,751,204 1,714,789 0 1,461,012 6,927,0051830 7,590,350 3,145,575 0 3,038,062 13,773,9871840 8,850,886 3,554,534 0 3,641,434 16,046,8541850 9,094,346 4,220,558 0 3,178,922 16,493,8261860 9,450,562 4,437,941 262,198 2,974,401 17,125,1021870 10,291,968 4,452,192 1,030,560 2,835,440 18,610,1601880 11,215,065 4,481,568 1,355,456 3,820,766 20,872,8551890 11,212,219 4,679,486 1,383,948 3,907,730 21,183,3831900 12,029,439 5,062,650 1,433,576 4,015,567 22,541,2321910 11,935,701 5,694,710 1,489,942 3,534,899 22,655,2521920 12,539,681 5,841,678 3,552,699 3,698,353 25,632,4111930 11,876,996 5,910,095 3,545,527 3,080,629 24,413,2471940 11,946,743 5,644,216 3,497,509 3,056,010 24,144,4781950 11,986,437 5,796,823 3,522,545 3,092,375 24,398,1801960 11,578,880 5,803,276 3,404,244 3,141,582 23,927,9821970 11,626,911 5,755,537 3,383,924 3,002,933 23,769,3051980 12,152,603 5,804,320 4,113,254 3,108,775 25,178,9521990 13,272,162 7,440,305 4,060,570 3,104,303 27,877,3402000 14,590,078 7,678,830 4,088,704 3,121,839 29,479,451

Total 207,633,395 97,207,399 40,124,656 61,266,574 406,232,024


Appendix BCOHA search for left-hand verb collocates of the to-infinitival perfect (screenshot fragment).

Markku Filppula3 Expression of the perfect in two contact

varieties of EnglishAbstract: This chapter discusses the means of expression of the present perfect in Irish English (IrE) and Hebridean English (HebE). Both of these can be consid-ered “high-contact varieties”, having emerged as a result of language shift from the indigenous Celtic languages to English. Because of the close affinity between the substrate languages, IrE and HebE can be expected to display similarities in their grammars, including the ways in which perfects are expressed. The data for this study are drawn from recordings of IrE and HebE dialect speakers in the late 1970s and early 1980s. The results show that these varieties share several features in their ways of using perfect variants. One such turns out to be unique amongst varieties of English and therefore poses a challenge to some of the mainstream developments in this domain of grammar. The close Celtic parallels suggest them-selves as a principal source for the systems of perfects in these two contact vari-eties.

1 IntroductionPerfects and perfectivity are two notions that feature centrally in studies of contact between two or more languages and/or dialects of a language. This should come as no surprise as both are an essential part of the tense and aspect systems of any language and belong to those areas of grammar that are known to be particu-larly susceptible to contact effects (see, e.g., chapters by Davydova and Van Rooy, this volume). Creoles – whether based on English or some other languages – provide ample evidence of such effects (see, e.g., Holm 1988: 148–168; Winford 2003: 324–326; Matras 2009: 282; Seoane, this volume). However, similar effects are also a regular feature of non-creole varieties such as different regional variet-ies that have developed in conditions of contact between, say, English and some other language(s). The two contact varieties at issue in this chapter, viz. Irish English (IrE) or Hiberno-English, as it is also called (especially in some earlier works on this variety), and Hebridean English (HebE), a Scottish variety spoken in the Western Isles of Scotland, are examples of this latter type. What makes these dialects of English a particularly interesting pair for linguistic comparisons

Markku Filppula, University of Eastern Finland

96 Markku Filppula

is the close linguistic affinity between the two Celtic substrate languages, Irish in Ireland (sometimes termed Irish Gaelic, especially in non-Irish contexts and in contrast to Scottish Gaelic) and Scottish Gaelic in the western Scottish context.

The aim of this chapter is to compare the means of expression of the present perfect in the Irish and western Scottish dialects of English in the light of evidence drawn from recordings made with speakers of traditional vernacular dialects in Ireland and in the Hebrides (see Section 4 on the database below for details). I have already given a brief description of some of the grammatical similarities and differences between these varieties in Filppula (1997), in which the features covered included tense and aspect, focus constructions, subordinating uses of the conjunction and, “absolute” uses of reflexive pronouns, and some aspects of prepositional usage.

In the present study, the HebE database is significantly larger and all the transcripts are now in digital form (and not just hand-written notes picked out from typed transcripts as in the earlier study). Also, the aims of the study are different from what they were in Filppula (1997). Here, the focus is on perfects and perfectivity only. Furthermore, the evidence obtained from the HebE and IrE datasets will be put in a global perspective, relying on the recent surveys of vari-eties of English worldwide. It will also be assessed against some of the general trends of development in this domain of grammar observed in recent linguistic literature. Among these, particularly relevant in this context are some of the gen-eralisations put forward by Davydova et al. (2011) and Trudgill (2011). The former identify the following three factors as the most important ones affecting the use of the have-perfect (HP) as opposed to other means of expression:

– Nature and degree of contact: in language contact situations, the nature and degree of contact with the target language are the major determinants of the types of expression used for the perfect.

– Stage of development: percentage of the standard HP gradually diminishes from acrolect through upper-mesolect and mesolect to basilect.

– Semantic context: the standard HP is closely associated with resultative and extended-now contexts, whereas in experiential and recent past contexts this association is less robust (Davydova et al. 2011: 302).

Trudgill (2011), in turn, distinguishes between two types of contact and their typical linguistic outcomes: “low contact”, which typically leads to complexifi-cation of grammar, and “high contact”, which has simplification of grammar as its usual outcome. The duration of the contact, age of the learner and the method of transmission act as additional variables determining the linguistic outcome of the contact. Simplification occurs most in short-term contacts involving non-na-tive language learning by adults in an untutored context; the most extreme

Expression of the perfect in two contact varieties of English 97

example according to Trudgill is pidginisation, which involves recourse to “easier for adults to learn”, mainly analytical, structures (Trudgill 2011: 33–35). By con-trast, childhood bilingualism – especially that preceding the “critical threshold” of language acquisition – and long-term co-territorial contacts generally lead to more complexity in the grammar and phonology of a language or variety. As an example of this kind of situation Trudgill mentions linguistic areas such as the Balkan Sprachbund and the languages of Amazonia, which have come to share a large number of morphosyntactic features (Trudgill 2011: 33–34, 40–43). Trudgill’s (2011) theory about two types of contact, additional variables, and their typical linguistic outcomes can thus be summarised in a simplified form as follows:

– Low contact: leads to complexification – High contact: leads to simplification – Duration of contact, age of learner and method of transmission as additional

variables: simplification occurs most in short-term untutored non-native adult learning whereas early childhood bilingualism leads to more complex-ity (Trudgill 2011: 33–43).

In the following, I will first present a brief description of the main differences and similarities between the language contact settings in Ireland and Scotland, especially the Western Isles of Scotland. (Note that I will be using “Celtic” in the following as a convenient cover term for both Irish and Scottish Gaelic.) This will be followed by an equally brief introduction to the linguistic characteristics of the Irish and the Hebridean English varieties, after which I will explain the more precise aims of my study and the details of my databases.

2 The language contact settings in Ireland and the Hebrides compared

The introduction of the English language into Ireland took place in two major waves. The first started around the mid-twelfth century along with invasions by Anglo-Normans, the then rulers of England. At first, the English language faced competition not only from Irish, the indigenous language of Ireland, but from Latin and French. The latter was the language of the Anglo-Norman nobility, and together with Latin it was long used as the language of administration and education in Ireland as well as in England. However, it was Irish which emerged victorious from this first round of battle with English and the other languages: the sheer number of the Irish-speaking population in mediaeval Ireland grad-ually pushed the English language into a decline in Ireland, and eventually the

98 Markku Filppula

English-speakers were almost entirely assimilated into the Irish language and culture. By 1600 English survived only in some of the major towns like Dublin and in few scattered rural areas in the east and south-east of Ireland (see, e.g., Bliss 1977: 8, 1979: 12–18). In some of the more recent scholarship, the lack of con-tinuity between the language of the Old English of Ireland and that of the “New English”, brought into Ireland by the new settlements from the latter half of the sixteenth century onwards (Bliss 1979: 17–18), has been called into question. For example, Kallen (1994: 155–156) concludes on the basis of some contemporary reports that mediaeval English survived to a much greater extent especially in the so-called English Pale in the eastern coastal regions around Dublin than Bliss (1979) and some other scholars had assumed.

Despite the (near-)demise of mediaeval English in Ireland, a new phase in the encounter between Irish and English began with the late sixteenth-century plantations of Ulster and parts of Munster. They were followed by the large-scale plantations under Cromwell in the mid-seventeenth century. Yet, Irish was able to sustain the pressure from English right up to the end of the eighteenth century. It was the early part of the nineteenth century which then saw the tipping of the scales in favour of English. The process of language shift, once it got under way, proceeded at a pace hardly paralleled in linguistic history, especially when the size of the shifting population is concerned, and by the middle of the century English had made deep inroads into the Irish-speaking communities throughout the country, excepting the coastal areas in the west of Ireland and some rather iso-lated pockets inland. A number of factors promoted the rise of English and sped up the decline of the Irish language. They included the setting up of a system of “National Schools” in 1831 with English as the medium of instruction, the choice of English as the main vehicle of the Catholic Emancipation movement, followed by the Great Famine of the 1840s and the subsequent emigration of about one million Irish people, many of them Irish-speaking. The result could be described as a “mass flight” from Irish, which led to a drastic drop in the numbers of espe-cially monoglot speakers of Irish (for more detailed discussion, see, e.g., Filppula 1999: 6–11 and the references there).

In Scotland, too, the contact history goes back to the Middle Ages, but the Scottish situation is characterised by a much sharper geographical division than is the case in Ireland. Already by the end of the mediaeval period English had spread in a wave-like manner from the south-east, gradually pushing the indig-enous Gaelic language north-westwards. Eventually, it led to the formation of a linguistic (and social) boundary called the “Highland Line”, which cut across Scotland from around present-day Glasgow in the south-west to an area east of Inverness in the north-east (Withers 1979: 51).


In the modern period, the advance of English in the western parts of Scot-land has been steady but relatively slow up until quite recently, despite measures aimed at rooting out the use of Gaelic. Among these, particularly important were the Statutes of Iona, signed in 1609, which were explicitly directed against the use of Gaelic as a medium of education and promoting English as the language of literacy in all parts of Scotland. Withers (1979: 45) writes that it was educational policies like these which sent Gaelic into not only a spatial but also social decline, with English replacing it first as the language of the upper levels of society. From there it spread gradually into other contexts, involving strata of society such as trade and later even religious services. The present-day situation is such that Gaelic is now holding out in the north-west and in the islands off the north-west coast of Scotland but its use among the younger generations, in particular, is diminishing all the time. Relevant statistics and a detailed discussion of the pros-pects of survival of Scottish Gaelic are given in MacKinnon (1993), for instance. Areas in mainland Scotland have long been almost completely English-speaking.

The dialects of English spoken in Scotland retain some archaic features from the old Anglian dialect of English, which had developed into a distinct variety usually referred to by the term “Scots” towards the end of the Middle Ages. For a time, Scots occupied the position of a standard literary language in Scotland, but from the sixteenth century onwards it began to absorb more and more influences from southern English and was eventually replaced by what is now called “Scot-tish English” (ScE) with its many subvarieties (for further discussion, see McClure 1994; Macafee and Ó Baoill 1997).

Yet another type of English in the complex linguistic set-up of Scotland is the variety which has evolved in the Gaelic-speaking areas of the north-west and the Western Isles: the terms used for this variety, or varieties rather, are “High-land English” and “Island English”. The most Gaelic-influenced subvariety of the latter is the one spoken in the Hebrides, known as “Hebridean English”. Sabban (1984: 6) defines this as “the variety of English spoken by bilingual inhabitants of the Islands”. It must be understood as referring to the post-war and earlier period when a large part of the population in the Hebrides was still brought up in Gaelic while English was first learnt in school (cf. Sabban 1984: 5). It is these Gaelic-influenced Englishes which are of the greatest interest to us here. It is true that Scots and ScE, too, contain some traces of influence from Gaelic, but the majority of scholarly opinion considers the Celtic input to them rather minimal as compared with the Highland and Island Englishes, in which the presence of the Gaelic substratum, that is, features originating from Gaelic, is much more notice-able (see, e.g., Sabban 1982, 1984; Shuken 1984; Macafee and Ó Baoill 1997).

To summarise, the following are the main differences between the Irish and Hebridean contact situations:

100 Markku Filppula

– A longer history of contact between English and Celtic in Ireland, reaching back into the twelfth century; in the Western Isles of Scotland, English did not spread on a larger scale until the nineteenth century.

– Different pace and timing of a large-scale language shift: a relatively rapid but large-scale shift in early nineteenth-century Ireland, leading to an almost complete shift by the end of the nineteenth century; a slightly later and more gradual spread and adoption of English as the main medium of communica-tion in everyday life and education in the Western Isles.

– Different roles of schooling and of the method of transmission of English: in Ireland, mainly naturalistic, especially in the days of the most intense lan-guage shift in the first half of the nineteenth century, with only few opportu-nities for formal education for the majority of the population; in the Western Isles, a major role for formal instruction through the schooling system.

The main similarities between these settings, in turn, consist of the following: – Active measures by the English-speaking rulers to impose the English lan-

guage on the Celtic-speaking population, including all kinds of oppressive legislative means and policies.

– Historically close and linguistically very similar substrate languages, which enables an assessment of the role of contact influence vs. other factors on perfects in these varieties.

Taken together, the differences and similarities between these two contact set-tings make them particularly worthwhile and almost unique objects of compar-ison.

3 Linguistic characteristics of the Irish and Hebridean varieties of English

There is a constantly growing body of literature on the Irish dialects of English, their background, and characteristic features (see, e.g., Harris 1993; Filppula 1999; Hickey 2007; Corrigan 2010; Kallen 2013). Among the present-day Englishes, IrE is generally considered to be one of the L1 or Inner Circle Englishes, and by and large, its morphology and syntax follow the patterns found in the other Englishes of the British Isles. This holds especially for educated IrE but the picture is differ-ent and much more diversified with respect to non-standard regional and social varieties. They can be better described as contact varieties, as they still exhibit many grammatical and lexical features which distinguish IrE from most other


British Isles varieties. Four major factors have affected the development of both southern and northern dialects of IrE: 1. Conservatism: IrE retains some features of earlier “mainstream” English that

are now mostly archaic or defunct in British English (BrE).2. Dialect contact with other varieties of English spoken especially in the British

Isles: of particular importance here is the diffusion of influences from the Scottish varieties of English to northern IrE.

3. Contact influences from Irish: for centuries these have exercised a consider-able amount of substratal influence upon IrE, especially in phonology and syntax.

4. Universal features associated with second-language acquisition: these are the expected outcome in the kind of intense language shift conditions which existed in Ireland especially from the early nineteenth century onwards.

The combined effect of these makes IrE an interesting mixture of linguistic features derived from one or the other of the mentioned sources. The areas of grammar most affected by substratal influences include article usage, use of reflexive pronouns, focus constructions such as clefting and fronting, preposi-tional usage, as well as tense and aspect systems, including perfects.

As compared with IrE, HebE has been studied far less, but on the basis of the few existing descriptions (see, especially, Sabban 1982, 1984; Shuken 1984; Clement 1997; Odlin 1997a; Filppula 1997, 1999), it is strikingly similar to Irish English in its grammar, though not at all so much in lexis or phonology. This in itself is not surprising in view of the structural similarity between the two Celtic substrate languages and the presence of Scottish Gaelic in the Western Isles up until very recently (and, of course, continuing in some localities and social con-texts even now). Thus, one finds an almost identical set of syntactic features in HebE as in IrE. The system of perfects is but one example of these: in both varieties we can witness a complex mixture of the “standard” HP and other, non-standard, means that are used for almost all meanings: resultative, experi-ential, recent past, and also extended-now. The wider use of the progressive in stative and habitual contexts is another shared feature; so is the prominent use of clefting as a means of thematic highlighting, as well as some uses of the definite article (for further details, see the references above).

102 Markku Filppula

4 Aims and databasesAs stated above, the aim of this chapter is to compare the means of expression of the present perfect in the Irish and Hebridean varieties of English. The distinc-tive nature of the IrE system(s) of perfects has been thoroughly described in the literature (e.g. in Kallen 1989; Harris 1993; Filppula 1999; Ronan 2005; Hickey 2007). For HebE, a similarly detailed description has been given in Sabban (1982). Shuken (1984: 155), relying on Sabban’s account, only notes “the relatively infre-quent use of perfective verb forms” in sentences like I’m a widower now for six years (‘I have been…’).

My focus here will be on three features which are shared by IrE and HebE and which I will call the “indefinite anterior perfect”, the “extended-now perfect”, and the “after-perfect” in accordance with the terminology adopted in Filppula (1999: 90). These three were selected as objects of study as all have close parallels in the tense and aspect systems of the corresponding Celtic substrate languages, Irish and Scottish Gaelic. It is worth noting here that on the basis of the recently published electronic World Atlas of Varieties of English (eWAVE; Kortmann and Lunkenheimer 2013), the after-perfect is found to be unique for IrE and some other varieties with heavy Irish input through Irish immigration, such as New-foundland English. By contrast, the indefinite anterior perfect and the extend-ed-now perfect are much more widespread (though pervasive in only few variet-ies in the case of the latter according to eWAVE).

The database for Irish English consists of transcribed recordings made by myself and a number of other fieldworkers in the late 1970s and the early 1980s. The method was informal interviews with traditional vernacular speakers from four different regions in the south of Ireland: Counties Clare, Kerry, and Wicklow, and the city of Dublin. All informants had English as their first language, but those from the (south-)west of Ireland, in particular, had some knowledge of Irish, too. The size of this corpus is c. 158,000 words (for further details, see Filp-pula 1999: 37–42).

For HebE I have used transcribed dialect recordings made in Tiree in the Inner Hebrides in the late 1960s and early 1970s. The fieldworker was Eric Cregeen, a full-time collector working for the School of Scottish Studies at the University of Edinburgh. The size of the corpus (in its present composition), which I will refer to as the Cregeen corpus, is c. 50,000 words (excluding the interviewer’s questions and other contributions).¹ This material is complemented by a similar corpus

1 This was based on an estimate arrived at by calculating the mean number of words uttered by the interviewer in randomly selected five pages of the transcripts from each interviewee. The same method was applied to the transcripts of the tapes from Annette Sabban.


collected by the German scholar Annette Sabban in North and South Uist in the Outer Hebrides and Skye in the late 1970s. The size of this corpus is c. 80,000 words, and taken together the two corpora thus consist of some 130,000 words.

5 Perfects and perfectivity in Irish English and Hebridean English

As noted above, both varieties make use of standard and non-standard features in their systems of perfects and perfectivity. Seen from a wider global perspective, the non-standard means of expression found in them can be divided into two types: on the one hand, there is plenty of evidence for features that are increas-ingly common in varieties of English throughout the world; on the other, they exhibit features that can be considered unique or distinctive to them. I will start off with the former type, a particularly good example of which is the levelling of the present perfect/past tense distinction and frequent use of the past tense to express perfectivity in the indefinite anterior or “experiential” sense.

5.1 The indefinite anterior perfect

In Harris’s (1984: 308) description of the indefinite anterior perfect (henceforth abbreviated as IAP), it is defined as a construction referring to events or states of affairs taking place “at (an) unspecified point(s) in a period leading up to the present”. In the linguistic literature, several different terms have been used for this type of time reference, for example: “experiential perfect” (Comrie 1976: 58–59), “indefinite past” (Leech 1971: 32–33), “existential perfect” (McCawley 1971: 104; all quoted here from Brinton 1988: 11). In Standard English (StE), the usual means of expression is the HP; Brinton’s paradigmatic examples include sentences like I have been abroad several times and I have read that novel (Brinton 1988: 10).

eWAVE treats the IAP under the heading of “levelling of the difference between present perfect and simple past: simple past for StE present perfect” (feature 99). The example sentence is Were you ever in London? This type is attested in 59 % of the varieties, and its rate of pervasiveness in these amounts to 61 %, thus making it cross-dialectally a common feature. The eWAVE survey of the varieties where it is found reveals that in non-standard IrE this feature is

104 Markku Filppula

“pervasive or obligatory”.² Somewhat surprisingly, the rating for ScE is the com-plete opposite: “attested absence of feature”. For dialects in the north of England the corresponding rating is slightly less absolute: “feature exists, but is extremely rare”. On the basis of the eWAVE ratings, the closest “relatives” of IrE with respect to this feature are found in Colloquial American English and – interestingly – in Newfoundland English, in both of which the levelling between present perfect and simple past is “pervasive or obligatory”.

In both my IrE and HebE data, the IAP can be said to be a pervasive feature. It very often occurs with a time adverbial, such as never, ever, always, or since. (1) and (2) are examples from my IrE corpus.

(1) Sure that would be a natural […] if = if you could employ every man the = and woman = that could work. That never happened in this world yet. (MF IrE corpus, Kerry: M.C.)

(2) […] I always had me, me health, I was always lucky that way. (MF IrE corpus, Dublin: J.O’B.)

Similar examples are plentiful in HebE and are illustrated here by (3) and (4).

(3) They were keeping the 7th January, they were keeping the 7th January for Christmas Day, and keeping the other day for the New Year’s Day. But I did not see it in any history or any writing at all, any book at all which was written, whether it was the 25th January or the 7th January.(Cregeen corpus, SA 1970/104/B: H.K.)

(4) I didn’t have an operation at all but they did something inside. And I was so well since.(Sabban corpus, No. 16, Portree, Skye: I.C.)

The IAP can be considered very pervasive in both varieties and is clearly preferred to the standard HP in the kinds of context illustrated in the examples above. In a global perspective, the same trend is particularly common in American English (AmE), which is generally considered to lead the development towards the use of the past tense instead of the HP. Indeed, the increasing use of the past tense in this function in other varieties, including BrE itself, is often put down to AmE

2 It has to be noted here that the eWAVE rating for this and the other IrE syntactic features was done by myself. To avoid being too subjective, I relied on corpus-based evidence and existing studies of IrE usage as much as possible.


influence (see, e.g., Vanneck 1958; Visser 1963–1973: 754; Hundt and Smith 2009), but that can hardly explain the pervasiveness of this feature in such an isolated variety as HebE, which unsurprisingly does not exhibit AmE influences in other areas of its grammar, phonology or lexis (cf. Sabban 1982: 106–107).

Historically, IrE has had much more contact with AmE, not least because of the extensive and large-scale waves of emigration from Ireland to the US, especially from the mid-nineteenth century onwards. However, even in the Irish context the use of the past tense with indefinite-anterior time reference goes back to the nine-teenth century and even earlier, when AmE could hardly have exercised the same kind of influence from that direction as it is doing in the present day. Examples (5) and (6) are taken from letters written in the 1850s by a mother and her daughter to a son who had emigrated to America (for details of the Oldham Letters and other letters from nineteenth-century Ireland, America, and Australia, see Filp-pula 1999: 42–47).

(5) […] her son is gone to Cork but I did not hear from her since […] (The Oldham Letters, No. 12, N.O., Rossmore, Co. Cork, 1855)

(6) My dear Son I am unwell in my health this length of time if I do be one day up I do be two days lying down and I never wrote that to any of ye in any of your letters. (The Oldham Letters, No. 19, N. & B.O., Rossmore, Co. Cork, 1861)

The use of the IAP in nineteenth-century Irish letters is systematic and wide-spread enough to make it clear that the IrE usage derives from something other than influence from AmE. Retention of the older (mainly Early Modern) English usage where the past tense could be used for indefinite anterior time reference is a possible source, but its likelihood in the Irish setting is not particularly strong. This is because of the dating of the formative period of present-day IrE dialects, which is now commonly accepted to have taken place relatively late, viz. in the nineteenth century (see, e.g., Garvin 1977: 100; Filppula 1999: 31–32; Hickey 2007: 278–279). A more obvious candidate therefore is substratal influence from Irish, which has no formal parallel to the English HP; instead, it uses the preterite (past tense) form of verbs to refer to experiences, events or states in indefinite past time. Ó Sé (1992: 55–56) provides a comprehensive description of the Irish per-fects. He illustrates the Irish usage with examples such as (7) and (8).

(7) Níor léigh mé an leabhar sin riamh.‘I [have] never read that book’.

106 Markku Filppula

(8) Chuala mé an t-amhrán sin cúpla uair.‘I [have] heard that song a couple of times’.

If the Irish system of perfects is indeed the principal source of the IrE usage, the contact influence has most probably been of a reinforcing or preservative nature rather than direct, considering that the Early Modern English use of the past tense must have lingered on to some degree in nineteenth-century “mainstream” English, too (as it is in some varieties and contexts even today). Grammarians of English like Visser, for example, have been well aware of this usage and link it particularly with AmE and sentences containing the time adverbs always, never, and ever (Visser 1963–1973: 754).

In order to assess the possible extent of superstratal influences on IrE, in Filppula (1999) I examined the use of the IAP in a corpus of traditional English dialects, based on data from the Survey of English Dialects tape recordings. Briefly, the result was that the IAP, though used to some extent by traditional dialect speakers in England, was virtually restricted to contexts that involved the adverbs always, never and ever. In other contexts, the HP was clearly the preferred choice. In IrE, about one third of the tokens had no overt adverbs, and the overall rates of use of the IAP as opposed to the HP were around 90 % in all of the variet-ies examined (Filppula 1999: 96–97). The IrE predilection for the IAP is also con-firmed by Edwards and Weltens (1985: 112), whose survey of the dialects spoken in the British Isles revealed that this type of perfect was a feature of only Irish English and the dialect of West Wirral in the northwest Midlands. These findings suggest that any influence that may have come from superstratal sources on IrE cannot have been so strong as to provide the principal model for the Irish learners of English.

A further factor speaking for the same conclusion is the method of transmis-sion of English amongst the Irish people, which was mainly naturalistic, with most people, especially in the early part of the nineteenth century, picking up their English from their countrymen and -women rather than from school (see Bliss 1977: 16–17). This goes hand in hand with the slow spread of literacy in nineteenth-century Ireland. Thus, on the basis of the 1851 census figures, Odlin (1997b: 3–6) has calculated that the proportion of the illiterate bilingual popu-lation was very high in most western counties of Ireland up until the 1850s and 1860s. For example, in County Mayo the numbers of illiterate bilinguals ranged between some 55 to 73 % of the estimated bilingual population (Odlin 1997b: 5).

Much of what has been said about the background of the use of the IAP in IrE also applies to HebE, mutatis mutandis. Of the earlier writers on this variety, Sabban considers both possible influences from the earlier English usages and the possible role of the Scottish Gaelic substrate, which behaves similarly to Irish


in this respect. She concludes in cautious terms that the latter has in all likeli-hood favoured the use of the IAP in HebE but that the earlier English parallels may also have contributed to it (Sabban 1982: 111–112).

To sum up so far, the IAP is a good example of a feature which, though wide-spread in varieties of English all over the world, may nevertheless have different backgrounds depending on the specifics of each language or dialect contact situ-ation. While superstratal influences cannot be ruled out in the Irish or Hebridean settings, these contact varieties most probably derive their usage of the IAP from their substratal parallels.

5.2 The extended-now perfect

Here, too, I will start off with a definition of the extended-now perfect (hence-forth ENP). Briefly, it is used to refer to an event, a state of affairs or an activity that has begun in the past but persists up to the present moment, or rather, the moment of utterance. Like the IAP, this type of perfect is accompanied by a time adverbial, which is compulsory in this case, however. Also, it expresses duration rather than frequency or a time point, as was the case with the IAP. A notable dif-ference between these two is that, instead of the past tense form as in the case of the IAP, the ENP has the verb in the present tense but with a clear extended-now time reference conveyed by the time adverbial. Brinton’s (1988: 11) description of the terminology used in the literature includes terms such as “inclusive past-and-present” (Jespersen 1924: 271–272), “continuative perfect” (Bauer 1970: 189), “universal perfect” (McCawley 1971: 104), and “perfect of persistent situation” (Comrie 1976: 60). Brinton’s examples for this type are: We have known him since he was a child and He has sung in the choir for years (Brinton 1988: 10). In the liter-ature on IrE, the most common terms are “extended present perfect” (e.g. Kallen 1989) and “extended-now perfect” (e.g. Harris 1984).

The description used by eWAVE for the ENP is “simple present for contin-uative or experiential perfect” (feature 101). The rates of attestation and perva-siveness are 49 and 61 %, respectively, thus making this type of perfect cross-di-alectally somewhat less common than the IAP (attested in 59 % of varieties). The typical examples cited in eWAVE are: I’m in here about four months; I know her since she was a child; I’m here for twenty years.

Just as the IAP, this type of perfect is rated to be “pervasive or obligatory” in IrE, and indeed, there is evidence to show that it is not only a feature of non-standard speech, but is commonly used by educated speakers of IrE and some-times finds its way into written texts as well (see, e.g., Filppula 2008: 331). Again, ScE presents a completely different picture: the ENP is not attested there at all.

108 Markku Filppula

This time, the northern English dialects are no different from ScE, and the same appears to hold for Welsh English. In fact, it seems that Manx English, the English dialects in the Southwest of England and Channel Islands English are the only other varieties in the British Isles in which the ENP is found to have some level of usage. In Manx English it is rated in eWAVE as “neither pervasive nor extremely rare”, whereas in the Southwest and in the Channel Islands it is classified as being “extremely rare”. On the other side of the Atlantic, it is “neither pervasive nor extremely rare” in Newfoundland English, while in Colloquial AmE it is deemed to be existent but extremely rare.

The IrE uses of the ENP are exemplified by (9) and (10).

(9) And = they’re fighting out ten years in the North for an all-Ireland republic.(MF IrE corpus, Kerry: M.McG.)

(10) Hugh Curtin is buried with years an’ years but his grandchildren are there now.(MF IrE corpus, Clare: C.O’B.)

Despite the absence of the ENP from ScE dialects in general, as the results of the eWAVE survey show, exactly similar examples to the IrE construction can be found in the HebE material. There they are frequent enough to confirm that the ENP is an established construction in HebE. Sabban (1982) devotes a whole section to a detailed description of this type type of perfect, its various forms of realisation, as well as its backgound (Sabban 1982: 56–112). The examples in (11) and (12) can serve as illustrations of the ENP in HebE.

(11) My father is seventy-five years dead. Well, he was sixty nine when he died.(Cregeen corpus, SA 1969/165/A: D.S.)

(12) [Interviewer:] And did they ever come over again?Oh yes, this lady she’s working in a bank […]. She be [?] working in a bank before my son saw her, and they’re fourteen or fifteen years married now.(Sabban corpus, No 64: F.C.)

Next, I will turn to the question of the background of these uses in the two vari-eties at issue. As we have seen, the geographical distribution of the ENP is such that it does not point to roots in the other varieties spoken in the British Isles. Yet, a parallel construction is found in earlier English and even further back in Conti-nental Germanic languages, many of which preserve it even today. In the history of English, the establishment of the present-day functional distinction between


the use of the present tense vs. the present perfect construction with have was a gradual process starting in the Middle English period and continuing through the Early Modern period. According to Visser (1963–1973: 737), this process came to an end in the nineteenth century when the periphrastic construction finally prevailed over the use of the present tense except in some dialects. It is interest-ing to note that Visser (1963–1973: 737) mentions here the Irish English dialects as an example of modern dialects that still employ the present tense in a perfective function.

This leaves open the possibility that the IrE and HebE uses of the ENP are retentions from the earlier system in which the present tense could be used in a perfect sense. While this is something that has to be considered in the case of IrE, it is not so likely in the case of HebE, which is of later vintage than IrE and has hardly been modelled on Early Modern English parallels. It is doubtful even in the case of IrE. In Filppula (1999), data from conservative English regional dialects were used as one of the points of comparison, along with earlier English paral-lels. The results showed the ENP to be extremely rare in these varieties (Filppula 1999: 125). This, coupled with the observed geographical distribution in the other British Isles varieties and beyond, suggests that the primary source of the ENP has to be sought in the Celtic substrate languages. Yet another factor supporting this line of argumentation is the co-occurrence of the IrE ENP with a non-stan-dard adverbial phrase clearly modelled on the corresponding Irish usage. Thus, several instances were found both in the present-day and the nineteenth-century data where the adverbial element was governed by the preposition with, used in a temporal sense indicating duration. Consider the example in (10) above and those from Kerry speech in (13) and from a nineteenth-century emigrant’s letter in (14).

(13) He was the chief of police, he, oh, he’s dead with long […] (MF IrE corpus, Kerry: D.B.)

(14) My brother Patt is out the Bush with the last 14 Mounths.(The Burke Letters, No. 3, 1884; quoted from Fitzpatrick 1994: 156)

An obvious explanation for the peculiar meaning of with in these examples is the dual character of the Irish preposition le, combining both the instrumental ‘with’ and the temporal ‘for (some length of time)’ meanings.³ It is thus plausible to assume that the combination of a present verb and the time adverbial governed

3 Ó Sé (1992: 55) illustrates the temporal meaning of le with the following example: Táim anseo le bliain ‘I have been here for a year’ (lit. ‘I-am here with [a] year’).

110 Markku Filppula

by the preposition were transferred into the speech of Irish learners of English more as a whole construction rather than as individual lexical-syntactic items. As regards HebE, no instances with the preposition with occur in my data; neither does Sabban offer any comment on this construction.

In the next section I will discuss a third type of perfect, which differs from the previous ones in that its distribution across varieties of English is extremely limited and on the basis of eWAVE is virtually unique for IrE.

5.3 The after-perfect

The after-perfect (henceforth AP) is a well-established term used to refer to a sit-uation – an event or an activity – that has obtained or happened in the (more or less) recent past but the effects of which persist in some way or other up to the moment of utterance (cf. Filppula 1999: 99; Ronan 2005: 253, 257). Because of the recentness of the temporal point of reference, it has also been described as “hot news” perfect (Harris 1984: 313). Other studies, terms and descriptions of its range of uses can be found, for example, in Kallen (1989: 10–12, 2013: 95–100), McCafferty (2006), and Hickey (2007: 197–208).

The term “after-perfect” is also the one used in eWAVE for this type (feature 98). The example given is She’s after selling the boat ‘She has just sold the boat’. The rarity of this construction is revealed by its rate of attestation, which is only 5 %, but this is counterbalanced by its high degree of pervasiveness, 72 %. It is clearly above the ratings for the other non-standard perfect realisations, which underlines its special and particularly entrenched nature in those few varieties that have it in their grammatical repertoire. The list of these varieties is extremely short, consisting of only IrE and Newfoundland English, in both of which it is classified as being “pervasive or obligatory”. It is also mentioned as a “neither pervasive nor extremely rare” feature in the entries for Sri Lankan English, but nowhere else apart from these three. Thus it is absent from the closest neighbour-ing varieties of IrE, such as ScE (in general), northern English dialects, and Welsh English.

By contrast, there is no question about the existence of the AP in IrE or HebE, where it also occurs despite the eWAVE rating for ScE. The examples in (15) to (18) illustrate the uses of this peculiar construction in these two varieties.

(15) An’ there was a house you’re after passin’, there was fifteen, sixteen children in = in the house. (MF IrE corpus, Clare: J.N.)


(16) You’re after ruining me, you’re after ruining me.(MF IrE corpus, Dublin: M.L.)

(17) Now when I went out last Monday – aye last Monday – there was quite a few crabs after going into them [i.e. creels used for catching lobsters].(Sabban corpus, No 44: L.C., Skye)

(18) [H.K.:] I think it will be a good morning.[Interviewer:] Will it?[H.K.:] Aye, the mist is after clearing.(Cregeen corpus, SA 1970/110/B: H.K.)

In IrE, the first examples of APs occur in some seventeenth-century fictional Irish texts. In these, however, the construction appears in different forms and functions, and partly for this reason they have often been considered to repre-sent “stage-Irish” and thus to be somewhat doubtful as far as their authenticity as depictions of the IrE usage in that period is concerned. Bliss (1979: 301–303) provides a detailed discussion of some of the earliest examples of what he terms the “after writing construction” and notes the difference between them and the present-day IrE AP. Yet, he concludes that since at least some of the writers of these early texts were English, and since in other respects the after writing con-structions “reflect Irish-English usage with such accuracy” (Bliss 1979: 300), they should be considered trustworthy. In later research, several studies have been conducted on the AP and its development through the past few centuries (see, e.g., Ronan 2005; McCafferty 2006; Hickey 2007: 197–208). It seems that, instead of the Early Modern period, present-day scholarship looks to the nineteenth-cen-tury large-scale language shift in Ireland as the most probable period when the AP – just as several other Irishisms – became established in their present-day forms and functions in IrE (see also Montgomery 1995: 35–36 for a similar con-clusion).

There is no shortage of examples from written records from that period. Example (19), from the early nineteenth century fictional text by John Banim, should make it clear that already at that point of time the construction could convey the same meaning of recentness of an event, state or activity as it has in present-day non-standard speech, in which it has become some kind of a shibbo-leth of Irishness.

(19) “[…] I’m sure of that in the heart within; for you’re after breakin’ Peery’s heart, Peggy Nowlan, an’ […]”(Banim 1826/1992: 86)

112 Markku Filppula

The AP construction in both IrE and HebE can be considered a particularly clear example of substrate influence and calques on the Irish and Scottish Gaelic con-structions. No similar constructions with the same kind of functions exist in other modern varieties of English nor in earlier English, whereas close parallels can be found in both Irish and Scottish Gaelic. In Irish, the corresponding structure consists of the “substantive” (as opposed to “copular”) verb tá ‘be’, followed by the subject, the preposition tréis ‘after’ (derived from tar éis) and the so-called verbal noun. Greene (1979) provides the example in (20) as an illustration of this construction, which he terms “P I”. It is noteworthy that in the same connection he comments on the IrE AP, which he considers to be a calque on the Irish P I because of its “surface agreement” and because of the fact that no other variety of English can be put forward as an alternative source (Greene 1979: 122).

(20) Tá sé tréis imeacht.(lit. ‘He’s after going.’)‘He has just gone.’

Scottish Gaelic has an exactly similar construction, which here serves to confirm the substratal origin of the AP in the two contact varieties at issue. This becomes clear from Sabban’s discussion of the HebE AP, where she notes the parallelism between Scottish Gaelic and HebE in this respect and, importantly, also points out the lack of any similar constructions in languages outside the Celtic family (Sabban 1982: 161). Her conclusion is that the AP in HebE “ist eindeutig eine Übernahme aus dem G[älischen]” (‘is without doubt a transfer from Gaelic’ – my translation) (Sabban 1982: 161).

To conclude this part, the AP is of wider linguistic interest as a construction, because it represents a contact-induced change that adds to the complexity of a grammar rather than simplifies it, which would have been the expected outcome in such language contact situations as we have witnessed in Ireland and the Heb-rides. As will be remembered, Trudgill’s (2011) thesis on the typical outcomes of language contacts predicts simplification of grammar rather than complexifica-tion in circumstances that are characterised by untutored non-native adult learn-ing. As was described above, this was the setting in the crucial formative periods of both IrE and HebE in nineteenth-century rural Ireland and in the north-west-ern parts of Scotland. It has to be noted, though, that the language shift in these two areas was not exactly “short-term” although it was very intense in the first decades of the nineteenth century in Ireland, but considerably slower and later in the Hebridean context. Also, it did not affect the adult population alone: in the years of the most rapid shift to English in nineteenth-century Ireland, bilin-gualism must have been widespread among children and the youth, who were


literally pushed towards learning English by their parents and by various oppres-sive measures adopted by the English administration in Ireland and introduced into schools. All this, in turn, would have worked towards complexification of the grammar rather than simplifying it, if we are to follow Trudgill’s prediction.

This may at least partly explain why both IrE and HebE have developed a par-ticularly complex system of perfects in their vernacular forms. The uniqueness of the AP cross-dialectally and cross-linguistically is but one sign of this complexity. Added to the other special features of perfects in IrE and HebE, these systems mark a clear departure from that of standard English or any other variety of English, for that matter, with the exception of Newfoundland English. This variety, however, inherits its system of perfects from the English of the large numbers of Irish immi-grants to that part of the world and their impact on the kind of English spoken in Newfoundland. Thus, Clarke (1997: 217) writes that the AP “has been probably the most successful of the IrE features brought to the area [of Newfoundland as a whole]”. She goes on to note that its use is not confined to those communities whose members are predominantly of Irish descent but has nowadays spread to almost all communities on the island (Clarke 1997: 216).

6 Discussion and conclusionsIn this chapter I have discussed the use of three types of perfects in two varieties of English, one well-researched and the other much less, but both of which rep-resent high-contact varieties insofar as their sociohistorical genesis is concerned. In both varieties one can witness a complex mixture of the “standard” HP and other, non-standard, means used for the meanings of what I have here called indefinite-anterior, extended-now, and hot news (i.e. recentness of an event, state or activity).

The examined features of the IrE and HebE systems of perfects provide good examples of complexification rather than simplification in varieties that have evolved in conditions of intense language contacts (see also Werner, this volume). Seen from this perspective, both varieties appear to have worked against the generalisation proposed by Trudgill (2011) that high contact should lead to simplification in the type of circumstances that prevailed during the emergence of these two high-contact varieties. On the other hand, Trudgill’s generalisation is aimed at describing non-native adult learning and, as noted above, the process of language shift must also have involved widespread bilingualism amongst chil-dren. This, in turn, would have allowed for more complex structures such as the after-perfect in particular to become a part of the evolving new system of perfects.

114 Markku Filppula

Apart from the issue of complexity vs. simplification, the vernacular Irish and Scottish varieties and their systems of perfects examined here confirm the gener-alisation expressed by Davydova et al. (2011), who argue that in language contact situations the nature and degree of contact with the target language are the major determinants of the types of expression used for the perfect. According to them, the percentage of the standard HP gradually diminishes from acrolect through upper-mesolect and mesolect to basilect. In the case at hand, no “crosslectal” comparative data was available for IrE or HebE, and thus no systematic compari-sons between acrolectal, mesolectal, and basilectal varieties could be made. Yet, for IrE we know from previous studies that the proportions of use of the standard HP as against those of the non-standard perfects vary from one region to another and from one type of perfect to another (for details, see Filppula 1999: 90–129).

Furthermore, even a quick search through the Irish component of the Interna-tional Corpus of English reveals that in educated IrE usage the HP predominates in all the contexts examined here. This is confirmed by a detailed study of the occur-rence of non-standard perfects in this corpus by Kirk and Kallen (2007: 274–283). For Hebridean English, no such comparison in terms of regional or register varia-tion can be made, but the eWAVE ratings for the non-standard types of perfects in ScE in general make it clear that the HP is virtually the only choice on the highest echelons of the lectal scale.

My results also provide support for the third generalisation proposed by Davydova et al. (2011: 302), that is, that the HP is closely associated with resulta-tive and extended-now contexts, whereas in experiential and recent past contexts this association is less robust. I have treated the last-mentioned context under the heading of indefinite-anterior perfect and found the use of the past tense form of verbs in this function to be a very common, if not the prevailing, means of expres-sion in both IrE and HebE speech. The resultative context is outside the scope of this study (for IrE, see Filppula 1999: 107–116) but the extended-now context has, as we have seen, revealed a clear tendency in both IrE and HebE to use the pres-ent-tense verb + time adverbial combination in this function. In this respect, the two varieties studied here distinguish themselves from many other varieties. The obvious reason for this – as for the other types of perfects I have discussed – is the existence of strikingly close parallels in the Celtic substrate languages that have “come in handy” for learners of English in both the Irish and Hebridean language contact settings.

Further support for the role of the substrate languages in both contact set-tings comes from some studies on perfects in the standard varieties of English in different parts of the world. Thus, Fuchs (in this volume) compares the use of the present HP between 20 varieties, including BrE and IrE. He concludes that, although geographical proximity and “areoversals” generally constitute a


primary factor accounting for similarities between varieties in their use of present perfects, they do not explain the relationship between BrE and IrE, which accord-ing to him are noticeably different with respect to their levels of use of present perfects. Additional factors must be considered, as he notes, and one such is national identity. I would like to add to this the role of the national language(s): whether dominant or receding or even lost under the pressure of another lan-guage, they may continue to play a role as a substrate language. In the case at hand and in the light of the evidence from IrE and HebE, it is clear that geograph-ical proximity has been overridden by continuing influence from the substrate Celtic languages and their tense and aspect systems.

AcknowledgementsThis research and the writing of this article were supported by the Academy of Finland funding for the research consortium entitled “Changing English: users and learners worldwide” (grant no. 269385), which I gratefully acknowledge. I would like to thank Cathlin Macaulay and Caroline Milligan of the School of Scottish Studies, University of Edinburgh, for making it possible for me to gain access and use the Hebridean English audio recordings and transcripts. I am also grateful to two anonymous reviewers whose comments and suggested corrections have been very helpful. Of course, any possible errors or other shortcomings will remain my own responsibility.

References:Banim, John. 1992 [1826]. The Nowlans. Belfast: Appletree Press.Bauer, Gero. 1970. The English ‘perfect’ reconsidered. Journal of Linguistics 6. 189–198.Bliss, Alan J. 1977. The emergence of modern English dialects in Ireland. In Diarmaid Ó Muirithe

(ed.), The English language in Ireland, 7–19. Dublin: The Mercier Press.Bliss, Alan J. 1979. Spoken English in Ireland, 1600–1740. Dublin: The Dolmen Press.Brinton, Laurel J. 1988. The development of English aspectual systems. Cambridge: Cambridge

University Press. Clarke, Sandra. 1997. The role of Irish English in the formation of new World Englishes: The

case from Newfoundland. In Jeffrey L. Kallen (ed.), Focus on Ireland, 207–225. Amsterdam: Benjamins.

Clement, David. 1997. Highland English. In Hildegard L. C. Tristram (ed.), The Celtic Englishes: Proceedings of the Potsdam Colloquium on Celtic Englishes, 28–30 September 1995, 301–307. Heidelberg: Winter.

Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press.

116 Markku Filppula

Corrigan, Karen P. 2010. Irish English. Vol. 1: Northern Ireland. Edinburgh: Edinburgh University Press.

Davydova, Julia, Michaela Hilbert, Lukas Pietsch & Peter Siemund. 2011. Comparing varieties of English: Problems and perspectives. In Peter Siemund (ed.), Linguistic universals and language variation, 291–323. Berlin: Mouton de Gruyter.

Edwards, Viv & Bert Weltens. 1985. Research on non-standard dialects of British English: Progress and prospects. In Wolfgang Viereck (ed.), Focus on: England and Wales, 97–139. Amsterdam: Benjamins.

Filppula, Markku. 1997. Cross-dialectal parallels and language contacts: Evidence from Celtic Englishes. In Raymond Hickey & Stanisław Puppel (eds.), Language history and linguistic modelling: A Festschrift for Jacek Fisiak on his 60th birthday, 943–957. Berlin: Mouton de Gruyter.

Filppula, Markku. 1999. The grammar of Irish English: Language in Hibernian style. London: Routledge.

Filppula, Markku. 2008. Irish English: Morphology and syntax. In Bernd Kortmann & Clive Upton (eds.), Varieties of English. Vol. 1: The British Isles, 328–359. Berlin: Mouton de Gruyter.

Fitzpatrick, David. 1994. Oceans of consolation: Personal accounts of migration to Australia. Cork: Cork University Press.

Garvin, John. 1977. The Anglo-Irish idiom in the works of major Irish writers. In Diarmaid Ó Muirithe (ed.), The English language in Ireland, 100–114. Dublin: The Mercier Press.

Greene, David. 1979. Perfects and perfectives in Modern Irish. Ériu 30. 122–141.Harris, John. 1984. Syntactic variation and dialect divergence. Journal of Linguistics 20(2).

303–327.Harris, John. 1993. The grammar of Irish English. In James Milroy & Lesley Milroy (eds.), Real

English: The grammar of English dialects in the British Isles, 139–186. London: Longman.Hickey, Raymond. 2007. Irish English. Cambridge: Cambridge University Press.Holm, John. 1988. Pidgins and Creoles. Cambridge: Cambridge University Press.Hundt, Marianne & Nicholas Smith. 2009. The present perfect in British and American English:

Has there been any change, recently? ICAME Journal 33. 45–63. Jespersen, Otto. 1924. The philosophy of grammar. London: George Allen.Jespersen, Otto. 1931. Modern English grammar on historical principles, Part IV. Heidelberg:

Winter.Kallen, Jeffrey L. 1989. Tense and aspect categories in Irish English. English World-Wide 10(1).

1–39.Kallen, Jeffrey L. 1994. English in Ireland. In Robert Burchfield (ed.), The Cambridge history of

the English language. Vol V: English in Britain and overseas: Origins and development, 148–196. Cambridge: Cambridge University Press.

Kallen, Jeffrey L. 2013. Irish English. Vol. 2: The Republic of Ireland. Berlin: Mouton de Gruyter.Kirk, John M. & Jeffrey L. Kallen. 2007. Assessing Celticity in a corpus of Irish Standard English.

In Hildegard L. C. Tristram (ed.), The Celtic languages in contact: Papers from the workshop within the framework of the XIII International Congress of Celtic Studies, Bonn, 26–27 July 2007, 270–298. Potsdam: Universitätsverlag Potsdam.

Kortmann, Bernd & Kerstin Lunkenheimer (eds.). 2013. The electronic world atlas of varieties of English. Leipzig: Max Planck Institute for Evolutionary Anthropology. http://ewave-atlas.org (accessed 02 Jan 2016).

Leech, Geoffrey N. 1971. Meaning and the English verb. London: Longman.


Macafee, Caroline & Colm Ó Baoill. 1997. Why Scots is not a Celtic English. In Hildegard L. C. Tristram (ed.), The Celtic Englishes: Proceedings of the Potsdam Colloquium on Celtic Englishes, 28–30 September 1995, 245–286. Heidelberg: Winter.

MacKinnon, Kenneth. 1993. Scottish Gaelic today: Social history and contemporary status. In Martin Ball (ed.), The Celtic languages, 491–535. London: Routledge.

Matras, Yaron. 2009. Language contact. Cambridge: Cambridge University Press.McCafferty, Kevin. 2006. Be after V-ing on the past grammaticalisation path: How far is it after

coming? In Hildegard L. C. Tristram (ed.), The Celtic Englishes IV: The interface between English and the Celtic languages, 130–151. Potsdam: Universitätsverlag Potsdam.

McCawley, James D. 1971. Tense and time reference in English. In Charles J. Fillmore & D. Terrence Langendoen (eds.), Studies in linguistic semantics, 96–113. New York: Holt, Rinehart and Winston.

McClure, J. Derrick. 1994. English in Scotland. In Robert Burchfield (ed.), The Cambridge history of the English language. Vol V: English in Britain and overseas: Origins and development, 23–93. Cambridge: Cambridge University Press.

Montgomery, Michael. 1995. The linguistic value of Ulster emigrant letters. Ulster Folklife 41. 26–41.

Odlin, Terence. 1997a. Bilingualism and substrate influence: A look at clefts and reflexives. In Jeffrey L. Kallen (ed.), Focus on Ireland. 35–50. Amsterdam: Benjamins.

Odlin, Terence. 1997b. Hiberno-English: Pidgin, creole, or neither? Dublin: Trinity College Dublin.

Ó Sé, Diarmuid. 1992. The perfect in Modern Irish. Ériu 43. 39–67. Ronan, Patricia. 2005. The after-perfect in Irish English. In Markku Filppula, Juhani Klemola,

Marjatta Palander & Esa Penttilä (eds.), Dialects across borders, 253–270. Amsterdam: Benjamins.

Sabban, Annette. 1982. Gälisch-Englischer Sprachkontakt [Gaelic-English language contact]. Heidelberg: Julius Groos.

Sabban, Annette. 1984. Investigations into the syntax of Hebridean English. Scottish Language 3. 5–32.

Shuken, Cynthia R. 1984. Highland and Island English. In Peter Trudgill (ed.), Language in the British Isles, 152–166. Cambridge: Cambridge University Press.

Trudgill, Peter. 2011. Sociolinguistic typology. Oxford: Oxford University Press.Vanneck, Gerard. 1958. The colloquial preterite in modern American English. Word 14. 237–242.Visser, Frederik Th. 1963–1973. An historical syntax of the English language, 4 vols. Leiden:

Brill.Winford, Donald. 2003. An introduction to contact linguistics. Oxford: Wiley-Blackwell.Withers, Charles W. J. 1979. The language geography of Scottish Gaelic. Scottish Literary

Journal 9. 40–53.

Sophie Richard and Celeste Rodríguez Louro4 Narrative-embedded variation and

change: The sociolinguistics of the Australian English narrative present perfect

Abstract: In this paper we examine sociolinguistic constraints on the narra-tive present perfect (PP) in Australian English (e.g. He couldn’t talk. That’s how freaked out he was. And I’ve just yelled like, “Everyone, get out of the water! Massive shark!”). The dataset consists of 220 performed Labovian narratives produced by 57 Australian English speakers. Multivariate analysis of 2,954 narrative clauses shows the narrative PP is primed by the tense used in the previous complicating action clause, and most frequently headed by quotative go. It is favoured with the third person and in the middle of the narrative complication. Our findings also show that non-linguistic factors play a significant role in narrative PP usage. The narrative PP does significant albeit specific work within a sub-section of the Aus-tralian English narrative genre and – crucially – it is indexical of older, non-pro-fessional, male speech.

1 The (Australian) English present perfect: What we know, and what we yet need to learn

Cross-linguistically, perfects follow a developmental path from resultative to anterior and anterior to perfective (Bybee, Perkins, and Pagliuca 1994: 68, 81), with languages such as French featuring a perfect construction (the passé composé) to refer to finished past situations (as in J’ai reçu une carte postale de mes parents hier ‘I received a postcard from my parents yesterday’).¹ Conversely, the English present perfect (PP) is predominantly an anterior form, used to estab-lish a link between a past situation and the present moment and crucially not

1 Cross-linguistic tense-aspect categories appear in small caps (e.g. present perfect) while lan-guage-specific tenses are spelled with lowercase (e.g. passé composé, narrative present perfect).

Sophie Richard and Celeste Rodríguez Louro, University of Western Australia

120 Sophie Richard and Celeste Rodríguez Louro

encoding perfectivity (Comrie 1976: 52).² A corollary of this is the purported inabil-ity of the PP to move the narrative sequence forward or foreground key events in the story (Dahl 1985: 139).³

This long-held assumption for the English PP has been challenged in research by Engel and Ritz (2000), Ritz and Engel (2008), and Ritz (2007, 2010) who – drawing on publicly available data such as radio chat show programs and police reports – document narrative uses of the Australian English (AusE) PP where the PP is used innovatively to sequence events in narrative. In example (1), from Ritz and Engel (2008: 153–154), a radio caller who is a professional shark feeder shares a work-related story with the listeners.

(1) And umm, I’ve ducked underand I’ve looked back and, and she’s gone past and I’ve gone, “Okay, that, that was all good”. Another one’s come down, I’ve thrown this fish out, and he’s started snapping on it, and I’m like, “Ohh, Thank God for that”. And then I’ve looked at, at the tunnel, at the kids, and all the little eyes are just like Christmas,and the, the tour guide in the tunnel’s just like lost itshe’s just throwing her hands in the air.(Ritz and Engel 2008: 153–154)

Labov and Waletzky (1967: 32–40) divide oral narratives of personal experience into five sections: orientation (scene-setting), complication (core sequence of events unfolding), evaluation (justifying the point of the narrative: how and why it is remarkable), resolution (what finally happened), and coda (the moral of the story, returning the perspective to the present). The majority of the complicat-ing action clauses in the narrative presented in (1) feature the PP and – contrary to canonical English usage – are able to advance the story line. Because of this,

2 For a treatment of the evolution of the have + past participle construction from resultative to anterior in Old English written texts see Johannsen, this volume. 3 However, aspect is sometimes deployed to foreground main events in narrative, see Couper-Kuhlen (1995) and Mesthrie (2013) on the use of the present progressive in, respectively, Ameri-can English narratives and in South African Indian English.

Narrative-embedded variation and change 121

Engel and Ritz (2000: 119) attribute a foregrounding function to the narrative PP and argue that the AusE PP “is undergoing an extension of its meaning”.⁴

According to the electronic World Atlas of Varieties of English (eWAVE; Kort-mann and Lunkenheimer 2013), the “levelling of the difference between present perfect and simple past: present perfect for StE [Standard English] simple past” (feature 100) is neither pervasive nor extremely rare in AusE, suggesting that use of the PP in lieu of the simple past (SP) is recorded in the variety but may not be encroaching on SP territory. Elsness (2009: 99) notes that the ratio of the PP to the SP is high in AusE in comparison to American English and British English, and twice as high in the spoken corpora. Elsness (2009: 101–102) finds that most instances of the present perfect with definite past temporal adverbials (ago, yes-terday, last night/week/month/year/X-day) stem from the Australian Corpus of English (AusCorp), the Australian section of the International Corpus of English and the Australian Radio Talkback Corpus rather than the British English, Amer-ican English or New Zealand English corpora used in his study. Yao’s (2015) dia-chronic study of the present perfect and preterite in AusE (based on the Corpus of Oz Early English and AusCorp) reveals that the ratio between the two forms has remained stable over time. Yao (2015: 263) contends that “the stability of the PP vs. SP ratio in the AusE data is suggestive not so much of a functional extension of the PP to past tense contexts as of a retention of patterns in an earlier stage of the English language”. However, Yao’s (2015) analysis is crucially not based on narrative. In fact, narrative sections stemming from dialogues were excluded as they “were found to exert a near-categorical effect in favour of the SP, with complicating action clauses massively disfavouring the PP” (Yao 2015: 253–254). Yao’s (2015) findings are likely to be due to the nature of the corpora under study: fictional works, rather than conversational narratives.

Ritz (2007) examines the discourse features and pragmatic factors at the heart of the semantic extension of the PP from an anterior to a narrative tense. Drawing on a rhetorical analysis of chat show program narratives on several radio stations in Australia featuring narrative PP uses, Ritz concludes that the narrative PP expresses temporal progression, “signalling a retrospective look at a situation […] and providing a post-time in which other events can be located” (Ritz 2007: 139). These claims have also been put to the test by Ritz (2010) on a corpus of West Australian and Queensland police reports, and Ritz notes that overall usage trends in the police data are parallel to those of the radio chat show

4 We follow Walker (2011) in using the term “narrative present perfect” to refer to PP tokens which encode foregrounded story events and advance temporal progression. These have also been described in the literature as “vivid narrative uses” (Ritz and Engel 2008) and “historical present perfect” (Rodríguez Louro and Ritz 2014).


program corpora, although the police report data show a wider range of definite past temporal adverbials modifying the PP (as illustrated in A male person aged between 25 to 30 years has entered the bank at about 12:45 pm on Friday 29th April 2005, from Ritz 2010: 3406) and no instances of the historical present.⁵ Moreover, for the police corpus, Ritz (2010: 3400) suggests that the PP gives rise to mirative effects, with the PP used to introduce new or unexpected information.

The rich suite of semantic and discourse-pragmatic changes documented by Ritz and colleagues begs the question of (i) how this variation in AusE storytell-ing plays out when the narrative PP is analysed based on a collection of squarely defined narratives and in the context of its variants, including the SP and the historical present and (ii) which social constraints (specifically, occupation, age and sex) operate on its usage. To fully grasp the magnitude of this linguis-tic change, Ritz and colleagues call for “a detailed quantitative analysis” (Engel and Ritz 2000: 137) and “a comprehensive sociolinguistic study of PP usage in Australian English” (Ritz 2010: 3416). In this chapter we engage expressly with these calls and offer a quantitative sociolinguistic analysis of the narrative PP in an attempt to establish its function, structure and social meaning. We also focus on the extent to which corpus type and methods used in data analysis impact on findings for tense variation and the narrative PP.

The chapter is organised as follows. In Section 2 we revisit the latest research on narrative and tense switching. Section 3 reviews studies on the narrative PP across Englishes. The unique narrative corpus used in this research and the various methodological decisions are presented in Section 4. Section 5 includes a detailed analysis of the findings. In Section 6 we contextualise these findings, highlighting our research contribution and pinpointing areas that still require further study.

2 Narratives and tense switchingThe narrative PP is intimately linked to storytelling as a discourse type. Labov (1972a: 359–360) defines a narrative as “one method of recapitulating past expe-rience by matching a verbal sequence of clauses to the sequence of events which actually occurred”. Hopper (1979: 213) views the organisation of narrative dis-course as a universal, proposing that “in any extended text a distinction is made

5 For further insights into the use of temporal adverbials with the present perfect across English varieties see Hundt and Smith (2009), Yao and Collins (2012), Seoane and Suárez-Gómez (2013), Werner (2013), and Yao (2015).


between the language of the actual story line and the language of supportive mate-rial which does not itself narrate the main events”. Hopper refers to the former as foreground and the latter as background. Similarly, Comrie (1985: 28) explains that “[a] narrative is by definition an account of a sequence of chronologically ordered events (real or imaginary), and for a narrative to be well formed it must be possible to work out the chronological order of events from the structure of the narrative with minimal difficulty”. From a functional perspective, then, a narra-tive has a referential function. It relays past experiences but the basic narrative units recapitulate the experience in the same order as the original events. This is illustrated in (2) where dropping the clutch precedes skidding it.

(2) a. I was pulling out of a street,b. dropped the clutch,c. skidded it,d. maybe overcorrected a little bit,e. then undercorrected.f. And then you know, just tapped the kerb. (Male, 18, trade assistant)⁶

In other words, a narrative contains narrative clauses – “the smallest unit of lin-guistic expression which defines the function of narrative” (Labov and Waletzky 1967: 13) – that in turn relate past events in chronological order. The narrative clauses form the complicating action, the skeletal structure of a narrative. A number of stories about past events do not qualify as Labovian narratives. For instance, the bolded sentence in (3) exemplifies how past events can be presented using syntactic embedding.

(3) But she landed face down in water um and luckily came to quick enough to- sorta get out of the water, and not drown herself um but the first thing that she – ’cause it was just a puddle – the first thing that she called up when the teachers were like, “Maxine, are you ok?” was “My jumper’s wet”.(Female, 18, student)

6 Unless otherwise indicated, all examples stem from our corpus of West Australian English and speaker’s sex, age and occupation/profession are indicated parenthetically (e.g. Male, 27, me-chanic). To protect speaker anonymity, names in the examples are pseudonyms. Further details about the corpus used in the study are offered in Section 4.1.


This syntactic complexity is arguably not expected in narratives. Rather, a narra-tive retelling would have presented events as shown in (4).

(4) a. She called upb. And the teachers were like, “Maxine, are you ok?”c. And she said, “My jumper’s wet”.

Labov (1972a: 377) points to the fundamental simplicity of narrative syntax, which is related to the straightforward temporal sequence of the narrative. Since it follows the chronological order of events, there are no flashbacks and the audience discovers events when the narrator provides evidence for their existence (Labov 2004: 36). Because of this, the past perfect does not feature in narrative clauses as this tense indicates that the narrator operates a flash-back of events and that s/he is reporting his/her experience non-narratively, as shown in (5).

(5) And luckily there were some police officers there. They intervened very very quickly. I think Michelle had run outside, [had] grabbed the police officers, and [had] run back in.(Male, 28, health administration manager)

In a canonical narrative, events would have been reported chronologically with the police officers’ intervention introduced after Michelle’s actions, as in (6).

(6) a. And luckily there were some police officers there. b. Michelle ran outside, c. grabbed the police officers, d. and ran back in.e. And they intervened very very quickly.

Crucially, narratives are also evaluative. Rather than a mere reporting of events, a narrative is about a “most reportable event” (Labov 1997: 405), which the nar-rator evaluates as being (at least slightly) unusual or exceptional. The notion of “reportability” or “tellability” (Norrick 2005: 323) is relative to the social situa-tion, including the relationship between the narrator and his/her audience, and the setting of the interaction. Some events, given their highly climactic nature (e.g. danger of death stories) are reportable in almost every situation (but see, for example, Trudgill 1974: 52 and Milroy 1980: 24 for unsuccessful responses to the danger of death question in Norwich and Belfast). Mundane events may be reportable in some settings such as a family dinner (cf. Blum Kulka 1993) but fewer


events would be deemed reportable to a committee of Congress (Labov 2004: 42). Telling a narrative is thus a social act or “social transaction” (Smith 1981: 228), and, to be acceptable, the story must be reportable relative to the social context in which it is told (Labov 2003: 66).

A specific discourse subtype of essence to the narrative PP has been dis-tinguished in the literature and identified as “performed” (Wolfson 1978: 216), “reportive” (Maranhão 1984: 235) or “vivid” (Ritz and Engel 2008: 132). Wolfson (1978: 216) defines performed stories based on six performance features: (i) direct speech; (ii) asides; (iii) repetition; (iv) expressive sounds; (v) sound effects; (vi) motions and gestures. The more features of performance used, the livelier the narrative. A performed story should not be a mere oral report of past events but an enactment of the said events. As Maranhão (1984: 253) notes, the best narra-tors are those who – endowed with “histrionic gifts” – are able to adopt the tone, body positions, facial expressions and typical gestures of their characters. The theatrical aspect of stories is hence largely conferred by those features of perfor-mance. It complements the chronological rendering of events, which imbues the narrative with historical flavour.

A hallmark of orally performed narratives is tense switching (Fleischman 1990: 8). Wolfson’s (1978, 1982) research, which deals with contemporary American English narratives, shows that tense switching occurs only in those narrations in which the speaker breaks through into performance (see Hymes 1975: 13). Additionally, Fleischman (1990: 3) proposes that the use of tenses in narrative is “anomalous with respect to a language’s normal use of tenses”. She investigates what appears to be non-prototypical tense usage – notably the use of the historical present – in the narration of past situations. The func-tion of tense switching in narrative is foregrounding: there is a mise en relief of unknown and unexpected (hence more salient) events; there is also mise en relief of evaluated events (Fleischman 1990: 184). Tense switching works as an internal evaluation device (Schiffrin 1981; Silva-Corvalán 1983). The narrator chooses to simply report events, leaving interpretation and judgment to the lis-teners so that events speak for themselves (Fleischman 1990: 149). Tense alter-nation also marks off different events within the story, giving it structure, as shown in (7) where variation between the SP and the historical present is used to single out key narrative events (verb phrases heading complicating action clauses are bolded for clarity).


(7) The funniest story I had <LAUGHTER> was I was in my office at the mine site. And this occupational health and safety officer came in with a bag. He said to me, “I’ve got a snake in the bag um I found it”. And I’m like, “Oh yeah. Sure”. I thought it would be like a lolly snake, you know, like a- a candy snake. I thought he was joking, because nobody would be that stupid to bring a snake into the office. Actually catch it from outside and bring it in. He said, “Oh I saw the snake and I thought it was dangerous so I thought I’d bring it to you and you could tell me what to do with it”. I’m like, “Ok. Yeah, sure. You’re joking”. And I was about to put my hand in it. And I opened it up and I saw there was a king brown snake in there. (Male, 35, mining consultant)

The SP occupies a central role in the narrative genre (Maranhão 1984: 249; Fleis-chman 1990: 24). It situates events squarely in past time and presents them aspec-tually as unanalysable wholes. However, as suggested earlier, other tenses also feature in narration. Wolfson (1978: 217–220) observes that the alternation between the historical present and the SP is a feature of performance (cf. example (7) above). The switch marks important events in the story, highlighting some events at the expense of others (Wolfson 1978: 222).

3 The narrative present perfect across English varieties

As stated in Section 1, because the present perfect signals a relation between two points in time but is not “about” the earlier situation, it is traditionally consid-ered unfit for narrative progression (Comrie 1976: 52). In this regard, Fleischman (1990: 30) argues that the perfect “establishes a connection to the speaker’s now that cannot be neutralized. Yet narratives are not about the speaker’s present, they are about the past”. However, the non-canonical PP uses documented for AusE narratives by Ritz and colleagues and presented in Section 1 have also been reported for British English and New Zealand English.

Levey (2006) considers variation between the SP, the historical present and the narrative PP in the complicating action portion of 56 stories told by seven to eleven-year-old working-class preadolescents from Redbridge, Southeast of England. Drawing on his own original corpus of oral narratives of personal expe-rience, Levey (2006: 140) notes an overall usage frequency of 9 % (50/571) for the narrative PP. Levey (2006: 148) concludes that the narrative PP is a pragmatic phe-nomenon rather than a grammaticalised substitute for the SP. Non-standard uses


of the PP in British English have also been noted in research by Walker (2008a, 2008b, 2011) who analyses the language of footballers and narratives. The PP attested in these corpora, he argues, represents a new “type” of perfect: the “narrative perfect”, defined as “the use of the present perfect form of the verb to recount past events, without supporting past time adverbials, in contexts where most accounts of standard English would predict a simple past form” (Walker 2011: 74). Walker (2011) also proposes a pragmatically entrenched mirative value for this form motivated by “some kind of personal or emotional involvement in the narration of the events being described” (Walker 2011: 77).

Non-standard PP usage has also been documented for New Zealand English. Cox (2005) analyses data from a New Zealand police reality television show (Police Ten 7) and argues that, among other functions, this form encodes definite past time and is able to reset narrative reference time (Cox 2005: 111).⁷ A clause headed by a verb in the narrative PP is able to introduce a new reference time in narrative discourse, so that the situation depicted is interpreted as being subse-quent to, not simultaneous with, the event of the previous clause (cf. Partee 1984: 261; Couper-Kuhlen 1987: 19; Kamp and Reyle 1993: 521–528). In other words, sequences of clauses in the narrative PP are able to encode temporal progression.

The dearth of sociolinguistic studies on tense variation in AusE has recently been addressed by Rodríguez Louro and Ritz (2014). Drawing on a corpus of 100 naturalistic narratives of personal experience arising in conversation with 38 AusE native speakers raised in metropolitan Perth and aged 12–62 years old, they note an almost negligible use of the narrative PP (1 % [9/654]) amongst the older speaker cohort (aged 54–62). Rodríguez Louro and Ritz (2014) claim that the increase of quotative be like has brought about radical changes to how nar-ration is constructed by younger speakers (aged 12–28); specifically in terms of (i) how the historical present is deployed in narrative and (ii) the pragmatically- motivated uses of be like and other quotatives (e.g. say) to depict first-person as opposed to third-person stance. What remains to be established, however, is how tense variation plays out in a large corpus of strictly defined Labovian narra-tives and the extent to which linguistic and social variables are implicated in the semantic extension of the PP.

7 Metalinguistic awareness studies are tangential to our treatment, but the reader is referred to Bauer (1989) and Ellis (2012) for attitudinal research on the narrative PP in, respectively, New Zealand and Australian English.


4 Data and method

4.1 The corpus

The original talk-in-interaction data used in this research stem from a large corpus of 220 strictly defined Labovian narratives produced by fifty-seven 13- to 69-year-old West Australian males (N = 29) and females (N = 28) representing pro-fessional (N = 18 males; N = 18 females) and non-professional (N = 11 males; N = 10 females) backgrounds. The data consist of naturalistic, casual interaction with native speakers of AusE and were collected in metropolitan Perth between 2011 and 2015.

The data stem from Rodríguez Louro’s (under construction) UWA Corpus of English in Australia, a collection of casual conversations and sociolinguistic inter-views obtained by University of Western Australia (UWA) students and research-ers between 2011 and 2015 (see Rodríguez Louro 2013: 56–57 for further details) and Richard’s (under construction) UWA Narrative Corpus, which currently con-sists of 132 narratives produced by 44 native speakers of AusE, collected in met-ropolitan Perth between 2013 and 2015. Richard’s Narrative Corpus is built on the assumption that the narrative PP occurs when stories are performed. It also rests on the observation that Ritz and Engel (2008) find such innovative uses of the PP in radio narratives which are delivered with almost no interruption and mono-logically rather than conversationally (Romano, Porto, and Molina 2013: 73–74). Narratives are elicited in storytelling sessions between the researcher and partici-pants. The latter are provided with a list of topics/questions (e.g. Did you ever get caught sneaking out? Have you ever witnessed a terrible accident or tragic event?) and explicitly invited to share a few personal stories and anecdotes of their choice. The prompts used are based on (i) Labov’s (1984: 34–36) conversational modules within the sociolinguistic interview, and (ii) a selection of questions successfully used on interactive radio programs (e.g. Craziest person to knock on your door; What went wrong when you travelled alone? What happened at the end of your big night out?) – both proven means to obtain vivid narratives of personal experience. As we note in Section 5, the use of the zero quotative (see example (9)) and tense variation as depicted in Figure 1 attest to the performed nature of the narratives in the corpus.


4.2 Tense variation and accountability

Tense variation is a key feature of performed narratives of personal experience (Wolfson 1978: 222; Schiffrin 1981: 47) and narratives provide an ideal constrained environment for systematic and controlled variation analysis (Schiffrin 1981: 45). However, existing research remains mostly mum about both the degree to which the data considered are narrative in nature, and the linguistic features that define complicating action clauses and how these can be clearly operationalized for transparency and replicability.

Our narrative corpus is unique in its inclusion of narratives which closely fit in with the Labovian framework (e.g. Labov 1972a) structurally and also display performativity. Only narratives which display tense variation in the complication (i.e. build up to the climax of the story) were included in the analysis. Although Labov’s (1972a) model accounts for what are termed “minimal narratives” (Labov 1972a: 360–361) – sequences of two temporally ordered clauses with a temporal juncture between clauses (e.g. The boy punched me and I punched back) – these were deliberately not included in our corpus as they do not feature a well-devel-oped complicating action sequence and therefore provide limited room for tense variation.

A minimum of two and up to six performed Labovian narratives were selected per speaker. All narrative clauses in the complication were then manu-ally extracted from the corpus. To ensure consistency in the extraction of tokens, an inter-rater reliability analysis was performed between the researchers and an independent rater for 20 % of randomly selected narratives. Using the Kappa sta-tistic (Cohen 1960), the inter-rater agreement was established at 0.88. A sample narrative appears in (8), where the extracted tense variants are bolded for illus-tration purposes.

(8) a. So she’s come in, b. and she’s gone, “Oh I need to fill out this form. Can you like get me a pen

and like sort me out a spot on the table?”c. I was like, “Yeah yeah, no problem”.d. So I moved all this stuff out of the way,e. and I’ve set this chair up,f. and it’s a nice, clean table,g. but she goes and sits on the side that was all dirty.h. And I’m like, “Ah, whatever”. (Male, 33, glazier)


In addition to the narrative tenses documented in the literature – including the SP, the historical present and the narrative PP – we also extracted complicating action clauses whose verbal predicate was elided, as in (9). These contexts are entirely represented by the zero quotative (Mathis and Yule 1994) and, although this categorical distribution disqualifies it from multivariate analysis (see Section 5), we extracted complicating action clauses headed by the zero quotative in the first instance in line with Labov’s principle of accountability (Labov 1972b: 72).

(9) And um I got to security and the laptop had to come out of the bag. And it was- it was just rushed and nuts. And the security guard, ZERO “You- you’re gonna have to take that out”. (Male, 22, student support officer)

Once extracted, all tokens were coded for grammatical and social factors. The lin-guistic constraints were established following Levey (2006) and Rodríguez Louro and Ritz (2014). These include:(i) Position of narrative clause in complicating action sequence: The

location of particular tense forms within the sequence of complicating action clauses, i.e. whether the tense form occurs clause-initially, medially or finally.

(ii) Tense in previous complicating action clause: The tense of the verb in the preceding clause, given the clustering/priming effect or persistence noted by scholars such as Tamminga and Ecay (2013) to refer to speakers’ prefer-ence for constructions recently utilised in the discourse.

(iii) Grammatical person: The grammatical person of the subject associated with each verb form as Elsness (1997: 342) reports that the (standard) PP tends to collocate with first person subjects while the SP tends to appear with the third person.

(iv) Quotative or non-quotative verb: Whether the predicate is headed by a quotative verb – quotative go favours the narrative PP (Engel and Ritz 2000: 136) and quotative be like is strongly associated with the historical present in AusE (Winter 2002; Rodríguez Louro 2013) – and, if so, the kind of quotative verb involved (e.g. say, think, go, be like).

The social variables under consideration include age – operationalized as decade of birth (1940s to 2000s) for clarity, sex (female, male), and socio-economic status (professional, non-professional). Following Docherty, Hay, and Walker (2006: 378), speakers are classified into the broad categories professional and non-professional according to both educational and occupational criteria. We also follow Ash (2003) in considering profession/occupation as a proxy for social status: “if social class is determined by a combination of features, the single


indicator that accounts for by far the greatest portion of the variance is occu-pation” (Ash 2003: 419) in Western cultures. Examples of professional occupa-tions include teacher, engineer and IT expert while non-professional occupations encompass careers such as electrician, cook or waiter. Because of their educa-tional achievements and career prospects, university students were assigned to the professional cohort (as they are soon-to-be professionals).

Excluded from the analysis were false starts, shown in (10) and clauses headed by epistemic/evidential phrases such as I think, illustrated in (11). Epis-temic/evidential phrases encode the speaker’s subjective viewpoint on a sit-uation (Rodríguez Louro and Harris 2013; Rodríguez Louro 2015) and, as such, express a stance rather than advance the narrative forward.

(10) And then I just said to her- I was like, “Look, I’m really sorry”.(Male, 28, health administration manager)

(11) And I think one of them punched my friend’s dad.(Female, 18, student)

Also excluded were ambiguous instances where past participle morphology was indeterminate, as in the highlighted verb phrases in (12) and (13).

(12) And like two o’clock in the morning this motorbike come down the street.(Male, 22, electrician)

(13) So I’ve- I’ve took Tim down the back.(Female, 24, administrative assistant)

The statistical modelling of variation necessitates that speakers display variable behaviour: “contexts that do not vary but are categorically encoded with one or other variant are not included in the analysis of variation” (Tagliamonte 2012: 10). Given this, only speakers evidencing variability in their use of tenses in the complication action portion of their narratives were considered in the statistical analysis presented in what follows.⁸

The variationist paradigm (cf. Tagliamonte 2006, 2012) is ideal to document variation in the grammar through an analysis of the sociolinguistic factors con-straining tense variation. Its unique toolkit of methodological and analytical deci-

8 Categorical use of the simple past was attested in the narratives of seven different speakers in the sample, four males and three females aged 15, 22, 23, 24, 26, 55 and 64 from professional (N = 5) and non-professional (N = 2) backgrounds.


sions, including the circumscription of the variable context and data extraction following the principle of accountability (Labov 1969, 1972b, 1978) allows us to examine narrative PP usage within the larger system of choices available to speakers, rather than focusing on individual innovative uses per se.

A key issue raised in previous research (e.g. Engel and Ritz 2000; Levey 2006; Ritz 2007) concerns the relationship of the narrative PP to other variants in the system. For example, Ritz and Engel (2008: 141) provide overall usage percent-ages for the SP, the PP (differentiating between standard and narrative uses of the PP; see Table 3) and the past perfect. Levey (2006) and Rodríguez Louro and Ritz (2014) carry out variationist analyses of tense variation in the complication action portion of their narratives in an attempt to document the sociolinguistic constraints governing this variation. We also adopt a holistic view and analyse all relevant alternations in the tense system in which the narrative PP can occur rather than focus on isolated examples. How the variable context or envelope of variation (Labov 1972b; Poplack and Tagliamonte 1989: 60) is defined is a key methodological decision in variationist research and one that has serious reper-cussions for data extraction, coding and interpretation of the findings. We discuss these issues in Section 6.

5 ResultsFigure 1 shows the overall distribution of the various tense variants in the dataset. As expected, the SP is pervasive in the sample (69.3 %, 2046/2954), in line with previous research (Fleischman 1990; Rodríguez Louro and Ritz 2014). The his-torical present follows suit, with a usage percentage of 20.1 % (595/2954), while predicates headed by an elided verb (i.e. zeroes) occur at 6.3 % (187/2954) and the narrative PP features at 4.3 % (126/2954).

Zero is categorically used in quotative constructions (see example (9)) and because of this cannot be included in the statistical analysis offered below; however, the overall distribution presented in Figure 1 shows that zero accounts for 6.3 % of the variability found in AusE narratives. This non-negligible use of the zero quotative is testament to the performed nature of the stories in our corpus. As Palacios-Martínez (2013: 457) argues, the more involved with the story, the more likely narrators will resort to the zero form. In a similar fashion, the more performed the narrative, the more likely tense switching will take place (Schiffrin 1981: 47; Wolfson 1982: 53). Figure 1 shows a marked propensity for tense varia-tion in our corpus.


69.3%

20.1%

6.3% 4.3%

0

10

20

30

40

50

60

70

80

90

100

Simple Past HistoricalPresent

Zero Narrative PP

Figure 1: Overall distribution of tenses (including elided verbs) in our AusE narrative corpus.

To uncover the constraints on tense variation in the complicating action portion of the 220 Labovian narratives extracted we resort to Goldvarb X (Sankoff, Tagli-amonte, and Smith 2005). This statistical tool assesses the relative strength and significance of the independent variables when these are considered simultane-ously (cf. Tagliamonte 2006).

Table 1 shows the results of two consistent variable rule analyses of the contri-bution of factors selected as significant to the probability of narrative PP usage in the Labovian complicating action clauses analysed. To control for potential inter-actions, we model linguistic and social factors separately. The statistical analysis is based on variability between the SP, the historical present and the narrative PP, and the latter is our application value. Three lines of evidence (Poplack and Tagliamonte 2001: 92; Tagliamonte 2003: 731) were used to interpret the results of the multivariate analyses: statistical significance (at the p = 0.05 level) of the dif-ferent factor groups, effect size (which factor groups feature the largest/smallest range) and constraint hierarchy, “the hierarchy from more to less of the categories within a factor group” (Tagliamonte 2006: 237). The factor weights (FW in Table 1) indicate how likely the narrative PP is to feature in a specific linguistic or social context (figures above .50 favour, figures below .50 disfavour its occurrence).

A few explanations are in order before introducing the results of the multivar-iate analyses. These notes should clarify disparities in the total Ns in some of the factor groups (specifically, tense in previous complicating action clause, gram-


matical person, presence of quotative verb, and age/decade of birth). Cross-tab-ulation across all factor groups indicated that further exclusions were needed before modelling the variability, as follows:(i) Quotative think tokens (e.g. And I thought, “Oh there’s a bit of brownie there,

I’ll just have a bit”) were excluded as they only featured with the SP (N = 2).(ii) Quotative be like tokens (e.g. I was like, “Look, I’m really sorry”) were

excluded: be like never occurred with the narrative PP.(iii) Speakers born in the 2000s were excluded because their generation was

poorly represented (N = 2) and, in line with Rodríguez Louro and Ritz (2014: 556), they did not use the narrative PP – instead displaying variability between the SP and the historical present.

(iv) Following Guy (1988: 131), near-categorical factors within factor groups were disregarded. These included say tokens encoded in the narrative PP (N = 1), and speakers born in the 1950s and 1990s whose use of the narrative PP was nearly negligible (respectively, N = 4 and N = 2).

The narrative PP in our AusE dataset is constrained by a suite of linguistic and social factors. Tense used in the previous complicating action clause, grammati-cal person, position in the complication, whether the head of the verb phrase in question is a quotative and the social variables of age/decade of birth, profession and sex all significantly favour narrative PP usage, albeit to different extents.

Chief amongst the linguistic conditioning of the narrative PP is the role of priming or persistence (cf. Tamminga and Ecay 2013), operationalized here as the tense used in the previous complicating action clause (range = 54). The narrative PP is highly likely to feature in storytelling when preceded by another narrative PP (.95). This finding is in turn related to the position of the narrative PP in the stories: the middle of the complication is the most hospitable environment for the narrative PP (.53).

Grammatical person is a statistically significant constraint on the narrative PP (range = 21), with the third person favouring its use. Rodríguez Louro and Ritz (2014: 559) draw on “basic discourse organisation” principles to explain their findings for the historical present: first person subjects present given information (see also Chafe 1994: 87); third person subjects introduce new information. The favouring effect of the third person on the use of the narrative PP in our corpus also lends further support to Ritz’s (2010: 3400) contention that the narrative PP is used to introduce new or unexpected information in AusE. Importantly, the grammatical person constraints documented for the narrative PP are diametri-cally opposed to those for the standard PP – which is predicted with first person subjects (Elsness 1997: 342).


Table 1: Variable rule analyses of the contribution of factors selected as significant to the proba-bility of narrative present perfect usage (highest factor weights in bold).

input 0.014

total N 2767

fw % n

Professional BackgroundNon-professional .92 11.8 1,014Professional .19 0.3 1,753range 73

Tense in Previous CAC⁹Narrative PP .95 44.9 98None .68 4.1 221Historical present .55 4.1 510SP .41 2.3 1,800range 54

SexMales .63 7.2 1,337Females .37 2.1 1,430range 26

Decade of BirthOlder (1940, 1960) .63 9.3 690Younger (1970, 1980) .40 5.2 1,070range 23

Grammatical PersonThird .54 5.1 1,094First .52 4.9 1,234Unexpressed .33 2.3 436range 21

Position in ComplicationMiddle .53 4.8 2,328Beginning .33 3.9 230End .32 2.4 209range 21

Quotative VerbGo .60 10.6 141Non-quotative .49 5.7 1,923range 11

9 CAC stands for “complicating action clause”. Total Ns for the Tense in Previous Complicating Action Clause factor group do not add up to the total N of 2,767 due to the exclusions listed in i–iv above.


Whether the complicating action clause is headed by a quotative verb significantly impacts narrative PP usage (range = 11). In line with Engel and Ritz’s (2000: 136) findings, the narrative PP collocates to a statistically significant extent with quo-tative go, with a factor weight of .60 (cf. example (8), line b). As per our exclusions listed in i–iv above, the narrative PP never occurs with quotative be like. Although featuring a relatively low factor weight (.49), the narrative PP occurs non-negligi-bly outside quotative go-headed predicates. A closer look at the data shows that, in fact, the narrative PP collocates with a range of activity verbs (Vendler 1967). The most frequent of these non-quotative verbs include go as in So she’s gone in there (N = 20), walk as in And I’ve walked into the ladies’ toilet (N = 9), come as in And they’ve come running out (N = 7), and look as in And he’s looked at me (N = 6).

The narrative PP is significantly constrained by various social factors. Fore-most amongst these is profession (range = 73): speakers of a non-professional background overwhelmingly prefer the narrative PP (.92), while professionals resolutely move in the opposite direction (.19). Sex is the second most influential social constraint on its use (range = 26), with males markedly preferring the nar-rative PP (.63) over their female counterparts (.37). Age/decade of birth is a salient factor significantly constraining the narrative PP (range = 23). Older speakers, born in the 1940s and the 1960s, favour the narrative PP (.63), while the younger speakers, born in the 1970s and 1980s, disfavour it (.40), a pattern in line with Rodríguez Louro and Ritz’s (2014: 556) observation that the narrative PP is pre-ferred by the older speakers (aged 54–62) in their sample (born in the 1940s and the 1950s).

The sociolinguistic conditioning of the AusE narrative PP is summarised in Table 2.

Table 2: Significant constraints on the Australian English narrative PP.

Factor Narrative present perfect favoured when…

Tense in previous CAC previous narrative clause headed by a narrative PPGrammatical person subject of narrative clause is in the third personPosition in complication located in the middle of the complicationQuotative verb heading the narrative clause is quotative goOccupation speaker is of non-professional backgroundSex speaker is maleAge speaker is older (born 1940s and 1960s)


6 Discussion and conclusionsOur quantitative sociolinguistic analysis of the narrative PP in an original corpus of strictly defined Labovian complicating action clauses reveals a system tightly constrained by a variety of linguistic and social factors (cf. Table 2). The ensuing sections discuss the implications of these findings.

The priming effect noted for the narrative PP aligns with previous findings. The historical present, for example, is significantly primed by other historical present tokens in Rodríguez Louro and Ritz (2014: 559) and this effect is also attested beyond tense variation in the clustering of pronominal forms in dis-course (see Travis 2007: 131), including the priming of unexpressed subjects by English speakers (Torres Cacoullos and Travis 2015). As shown in Table 1, priming accounts for 44.9 % of the narrative PP tokens modelled. Given this, is it fair to assume that speakers become subjectively involved with the story attempting to signpost narrative portions as “experientially important” (Fludernik 1991: 374) every time the choice of a particular narrative tense comes to the fore? Or do speakers imbue their narratives with devices that demonstrate subjective involve-ment once and a cognitively motivated priming effect takes care of the subse-quent occurrences? Our results suggest the latter is more likely to describe tense variation in narrative complication.

Our finding that the narrative PP is significantly favoured by third person sub-jects is also in line with a large volume of existing research noting that, in com-plicating action clauses, tenses other than the SP are predicted with third person subjects. Levey (2006: 146), for example, finds that while the SP is favoured with first person subjects (with a factor weight of .61 in his research), the historical present occurs most readily with the third person (.57). In a similar vein, Rodrí-guez Louro and Ritz (2014: 558) note that the SP is predicted with first person subjects (.63) while the historical present is favoured with the third person (.64). This divergence is explained drawing on Chafe’s (1994: 87) suggestion that while first person subjects encode given information, third person subjects are reserved for new information in the discourse.

The narrative PP has customarily been described as occurring “at the heart” of AusE narratives (e.g. Engel and Ritz 2000: 133) and this contention is supported in our finding that the narrative PP is predicted in the middle of the complicat-ing action (see example (14)). This priviledged position at the core of the story is reflective of its value as a foregrounding narrative tense as well as a result of the clustering effect noted above where the narrative PP is most likely to occur imme-diately following another narrative PP.


(14) And I got up to go to the toilet and um I’ve opened the door to the men’s toilet and there was about twenty-five ladies queued up to use the only stall in the boys’ one because the girls’ one as I remember walking to the toilet was queued out the door. And I remember opening the door like that. And that- these twenty odd ladies have all looked to their right, “Ah it’s alright love, come in. Don’t mind us, we’re just waiting to use the loo down here”. And I’ve just gone, “Ah, ok”.(Male, 46, process operator)

As noted by Engel and Ritz (2000: 136), a lexical effect is in place in our dataset such that the narrative PP is predicted in the presence of quotative go (see example (8), line b). Importantly, the narrative PP is virtually non-existent or wholly unattested with other quotative verbs, including say (N = 1), think and be like (cf. Rodríguez Louro and Ritz 2014: 560 on the favouring effect of be like, rather than go, on the historical present). To a lesser degree, the narrative PP also occurs with non-quotative verbs (e.g. go, come, walk) and closer inspection of the data shows that, in line with Engel and Ritz (2000: 135), all of these predicates are headed by activity verbs. It seems, then, that the narrative PP is well suited to encode situations which are presented as dynamic, in line with the climactic nature of the complicating action portion they stem from. At the outset of the chapter we referred to the inability of the standard English PP to move the narra-tive forward. The activity verbs collocating with the narrative PP make it possible for this generally unfit tense to progress movement in time by injecting the story with action (cf. the role of activity verbs in lexical, grammatical and social change in andar + gerund constructions in Mexican Spanish; Torres Cacoullos 2001).

We are now left with the task of explaining the impact of social factors on the narrative PP in AusE. Our findings show that the narrative PP is significantly favoured by non-professional males. The weighting of the various significant factors points to an indexical link between the narrative PP and non-professional masculinity. This link is strengthened when the narrative PP occurs with quota-tive go, which has been described as used by “uneducated, lower class males” or “men like Rocky” (Blyth, Recktenwald, and Wang 1990: 224). Our findings also indicate that the narrative PP is restricted to the older cohorts. This, in turn, sug-gests that the narrative PP does not constitute a far-reaching change in progress but is instead both grammatically and socially constrained.

A recurring question in research concerned with language variation is that of what to count. Studies of the present perfect in English (e.g. Davydova 2011) and languages other than English (e.g. Schwenter and Torres Cacoullos 2008; Rodrí-guez Louro 2009; Howe 2013) have, for example, examined variability between the present perfect and the simple past/preterite and, using a formal defi-


nition of their variable context, they have provided overall usage percentages for the present perfect and the simple past/preterite independently of discourse type (see also Fuchs, this volume; Van Rooy, this volume). This approach is unten-able in our research, as storytelling is vital to the narrative PP and our decision to study tense variation within the complicating action of strictly defined Labo-vian narratives is of consequence. Additionally, in research where the variants under study are strictly formally defined (such as corpus linguistics) the focus is on quantifying usage independently of other variants in the system. Conversely, variationists – whether adopting a form- or function-based perspective – calcu-late proportional usage within a subset of variants occurring in a well-defined variable context (see also Bowie and Wallis, this volume).¹⁰

As D’Arcy (forthcoming) argues, these decisions have important ramifica-tions and can impact on our understanding of how a linguistic feature is used in context. The approaches used to count linguistic variants for analysis differ vastly across studies. For example, in addition to the SP, the historical present and the narrative PP, Ritz and Engel (2008) also take stock of the standard PP and the past perfect. Ritz and Engel (2008) are unconcerned with explicitly defining a variable context (their research is not variationist), and their exploration of nar-rative data includes verb phrases heading clauses at various levels of Labovian narrative structure.

Conversely, our approach is to analyse tense variation in the complicating action section of the narratives in our corpus, and this decision is reflected in the overall findings and the tenses that are empirically established to sit outside the variable context. Table 3 compares our findings (with usage percentages recalcu-lated to reflect variability between the SP, the historical present and the narrative PP, excluding zeroes; cf. Figure 1) to those of Ritz and Engel (2008). Although Ritz and Engel’s (2008) work is not strictly concerned with variation between the narrative PP and other tenses, they offer “the relative proportion of tenses used to relate past events in narratives” (Ritz and Engel 2008: 141), which we draw on here. Also apparent in Table 3 is the higher usage rate for the narrative PP, a con-sequence of how the data were selected, as we discussed at the outset.

10 For example, Grieve-Smith (2007: 21) proposes that corpus studies of register and genre vari-ation could benefit from the variationist notion of “variable context” or “envelope of variation” so that variables can be “targeted with more precise algorithms”.


Table 3: Overall usage frequency of tense variation in our study and Ritz and Engel (2008).

Tense Ritz and Engel (2008) This research

% N % N

SP 58 838 69 2,046PP 6 83 Outside variable contextVivid present perfect/narrative PP¹¹ 22 315 4 126Historical present 12 183 20 595Past perfect 2 33 Outside variable context

Total 100 1,452 100 2,767

Our analysis and findings evidence a tightly constrained grammar, one in line with previous proposals for the narrative PP across Englishes. Heeding calls for an in-depth sociolinguistic analysis, we have underscored the significance of non-linguistic factors in the use of the narrative PP, offering an encompassing picture of how tense variation– and more specifically the narrative PP – play out when macro social categories like occupation, sex and age are brought to bear. Our choice of methods – with special focus on the way the variable context is defined – has made it possible to model variability across a large speaker sample and on an original corpus of naturalistic talk-in-interaction data. The patterns that emerge have shown that the narrative PP does crucial albeit specific work within a sub-section of the AusE narrative genre and that it is indexical of older, non-professional, male speech.

AcknowledgementsWe are grateful to the editors and two anonymous reviewers for their helpful comments and suggestions. All errors remain our responsibility. Sophie Richard is indebted to the Australian Government and The University of Western Australia for their funding of her PhD research through an Interna-tional Postgraduate Research Scholarship (IPRS), an Australian Postgraduate Award (APA) and a UWA Safety Net Top-Up Scholarship.

11 We provide both labels here as Ritz and Engel (2008: 141) examine all portions of their nar-ratives (rather than just the complication as we do here), and, because of this, they label the narrative PP “vivid PP”.


Corpora Richard, Sophie (under construction). UWA Narrative Corpus. Discipline of Linguistics.

University of Western Australia. Rodríguez Louro, Celeste (under construction). UWA Corpus of English in Australia. Discipline of

Linguistics. University of Western Australia.

ReferencesAsh, Sharon. 2003. Social class. In Jack K. Chambers, Peter Trudgill & Natalie Schilling-Estes

(eds.), The handbook of language variation and change, 402–422. Oxford: Blackwell.B auer, Laurie. 1989. The verb have in New Zealand English. English World-Wide 10(1). 69–83.Blum Kulka, Shoshana. 1993. ‘You gotta know how to tell a story’: Telling, tales, and teller in

American and Israeli narrative events at dinner. Language in Society 22(3). 361–402.B lyth, Carl, Sigrid Recktenwald & Jenny Wang. 1990. I’m like, ‘Say what?!’: A new quotative in

American oral narrative. American Speech 65(3). 215–227.B ybee, Joan, Revere Perkins & William Pagliuca. 1994. The evolution of grammar: Tense, aspect

and modality in the languages of the world. Chicago: University of Chicago Press.Chafe, Wallace. 1994. Discourse, consciousness and time: The flow and displacement of

conscious experience in speaking and writing. Chicago: University of Chicago Press.Cohen, Jacob. 1960. A coefficient of agreement for nominal scales. Educational and Psycho-

logical Measurement 20(1). 37–46.C omrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press.C omrie, Bernard. 1985. Tense. Cambridge: Cambridge University Press.Couper-Kuhlen, Elizabeth. 1987. Temporal relations and reference time in narrative discourse.

In Alfred Schopf (ed.), Essays on tensing in English, 7–25. Tübingen: Niemeyer.Couper-Kuhlen, Elizabeth. 1995. On the foregrounded progressive in American conversational

narrative: A new development? In Wolfgang Riehle & Hugo Keiper (eds.), Anglistentag 1994 Graz: Proceedings, 229–245. Tübingen: Niemeyer.

Cox, Rebecca. 2005. Preterite uses of the present perfect in New Zealand English narratives: A case study. Christchurch: University of Canterbury MA thesis.

Dah l, Östen. 1985. Tense and aspect systems. New York: Blackwell.D’Arcy, Alexandra. Forthcoming. Discourse-pragmatic variation in context. Amsterdam:

Benjamins.Davydova, Julia. 2011. The present perfect in non-native Englishes: A corpus-based study of

variation. Berlin: Mouton de Gruyter.Docherty, Gerry, Jennifer Hay & Abby Walker. 2006. Sociophonetic patterning of phrase-final

/t/ in New Zealand English. In Paul Warren & Catherine I. Watson (eds.), Proceedings of the 11th Australasian International Conference on Speech Science & Technology, 378–383. Canberra: Australasian Speech Science and Technology Association Inc.

Elli s, Kate. 2012. “That’s the way they’re used to speaking”: Attitudes towards innovative uses of the present perfect in Australian narratives and police media reports. Perth: University of Western Australia Honours thesis.


Elsn ess, Johan. 1997. The perfect and the preterite in contemporary and earlier English. Berlin: Mouton de Gruyter.

Elsness, Johan. 2009. The perfect and the preterite in Australian and New Zealand English. In Pam Peters, Peter Collins & Adam Smith (eds.), Comparative studies in Australian and New Zealand English: Grammar and beyond, 89–114. Amsterdam: Benjamins.

Enge l, Dulcie & Marie-Eve Ritz. 2000. The use of the present perfect in Australian English. Australian Journal of Linguistics 20(2). 119–140.

Fleischman, Suzanne. 1990. Tense and narrativity: From medieval performance to modern fiction. London: Routledge.

Fludernik, Monika. 1991. The historical present tense yet again: Tense switching and narrative dynamics in oral and quasi-oral storytelling. Text 11. 365–398.

Grieve-Smith, Angus B. 2007. The envelope of variation in multidimensional register and genre analyses. In Eileen Fitzpatrick (ed.), Corpus linguistics beyond the word: Corpus research from phrase to discourse, 21–42. Amsterdam: Rodopi.

Guy, Gregory. 1988. Advanced Varbrul analysis. In Kathleen Ferrara, Becky Brown, Keith Walters & John Baugh (eds.), Linguistic change and contact, 124–136. Austin: University of Texas at Austin.

Hopp er, Paul. 1979. Aspect and foregrounding in discourse. In Talmy Givón (ed.), Discourse and syntax, 213–242. New York: Academic Press.

Howe, Chad. 2013. The Spanish perfects: Pathways of emergent meaning. Basingstoke: Palgrave Macmillan.


Hymes, Dell. 1975. Breakthrough into performance. In Dan Ben-Amos & Kenneth S. Goldstein (eds.), Folklore: Performance and communication, 11–74. The Hague: Mouton.

Kamp, Hans & Uwe Reyle. 1993. From discourse to logic: An introduction to model-theoretic semantics of natural language, formal logic and discourse representation theory. Dordrecht: Kluwer.

Kortmann, Bernd & Kerstin Lunkenheimer (eds.). 2013. The electronic world atlas of varieties of English. Leipzig: Max Planck Institute for Evolutionary Anthropology. http://www.ewave-atlas.org (accessed 12 December 2015).

Labov , William. 1969. Contraction, deletion, and inherent variability of the English copula. Language 45(4). 715–762.

Labov, William. 1972a. Language in the inner city: Studies in the Black English vernacular. Philadelphia: University of Pennsylvania Press.

Labov, William. 1972b. Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press.Labov , William. 1978. Where does the sociolinguistic variable stop? A response to Beatriz

Lavandera. Texas Working Papers in Sociolinguistics 44. 1–17.Labov, William. 1984. Field methods of the project on linguistic change and variation. In John

Baugh & Joel Sherzer (eds.), Language in use: Readings in sociolinguistics, 28–54. Englewood Cliffs: Prentice Hall.

Labov, William. 1997. Some further steps in narrative analysis. Journal of Narrative and Life History 7. 395–415.

Labov, William. 2003. Uncovering the event structure of narrative. In Deborah Tannen & James E. Alatis (eds.), Georgetown University round table on languages and linguistics 2001, 63–83. Washington: Georgetown University Press.


Labov, William. 2004. Ordinary events. In Carmen Fought (ed.), Sociolinguistic variation: Critical reflections, 31–43. New York: Oxford University Press.

Labov, William & Joshua Waletzky. 1967. Narrative analysis: Oral versions of personal experience. In June Helm (ed.), Essays in the verbal and visual arts, 12–44. Seattle: University of Washington Press.

Leve y, Stephen. 2006. Tense variation in preadolescent narratives. Journal of English Linguistics 34(2). 126–152.

Maranhão, Tullio. 1984. The force of reportive narratives. Paper in Linguistics 17(3). 235–265.Mathis, Terrie & George Yule. 1994. Zero quotatives. Discourse Processes 18(1). 63–76.Mesthrie, Rajend. 2013. Transfer and contact in migrant and multiethnic communities: The

conversational historical be + -ing present in South African Indian English. In Marianne Hundt & Daniel Schreier (eds.), English as a contact language, 242–257. Cambridge: Cambridge University Press.

Milroy, Lesley. 1980. Language and social networks. Baltimore: University Park Press.Norrick, Neal R. 2005. The dark side of tellability. Narrative Inquiry 15(2). 323–343.Palacios-Martínez, Ignacio M. 2013. Zero quoting in the speech of British and Spanish

teenagers: A contrastive corpus-based study. Discourse Studies 15(4). 439–462.Partee, Barbara H. 1984. Nominal and temporal anaphora. Linguistics and Philosophy 7(3).

243–286.Poplack, Shana & Sali Tagliamonte. 1989. There’s no tense like the present: Verbal -s inflection

in early Black English. Language Variation and Change 1(1). 47–84.Poplack, Shana & Sali Tagliamonte. 2001. African American English in the diaspora. Malden:

Blackwell.Ritz, Marie-Eve. 2007. Perfect change: Synchrony meets diachrony. In Joseph Salmons &

Shannon Dubenion-Smith (eds.), Historical linguistics 2005: Selected papers from the 17th International Conference on Historical Linguistics, Madison, Wisconsin, 31 July–5 August 2005, 133–147. Amsterdam: Benjamins.

Ritz, Marie-Eve. 2010. The perfect crime: A study of illicit adverbial combinations with the present perfect in Australian police media releases. Journal of Pragmatics 42(12). 3400–3417.

Ritz, Marie-Eve & Dulcie Engel. 2008. Vivid narrative use and the meaning of the present perfect in spoken Australian English. Linguistics 46(1). 131–160.

Rodríguez Louro, Celeste. 2009. Perfect evolution and change: A sociolinguistic study of preterit and present perfect usage in contemporary and earlier Argentina. Melbourne: University of Melbourne dissertation.

Rodríguez Louro, Celeste. 2013. Quotatives down under: Be like in cross-generational Australian English speech. English World-Wide 34(1). 48–76.

Rodríguez Louro, Celeste. 2015. The evolution of epistemic marking in West Australian English. In Peter Collins (ed.), Grammatical change in English world-wide, 205–220. Amsterdam: Benjamins.

Rodríguez Louro, Celeste & Thomas Harris. 2013. Evolution with an attitude: The grammat-icalisation of epistemic/evidential verbs in Australian English. English Language and Linguistics 17(3). 415–443.

Rodríguez Louro, Celeste & Marie-Eve Ritz. 2014. Stories down under: Tense variation at the heart of Australian English narratives. Australian Journal of Linguistics 34(4). 549–565.

Romano, Manuela, María Dolores Porto & Clara Molina. 2013. The structure of emotion discourse: From Labovian to socio-cognitive models. Text & Talk 33(1). 71–93.


Sankoff, David, Sali Tagliamonte & Eric Smith. 2005. Goldvarb X: A variable rule application for Macintosh and Windows. Toronto: University of Toronto. https://www.jiscmail.ac.uk/cgi-bin/webadmin?A3=ind1201&L=VAR-L&E=quoted-printable&P=240131&B=--B_3409627595_4788199&T=text%2Fhtml;%20charset=ISO-8859-1 (accessed 05 July 2015).

Schiffrin, Deborah. 1981. Tense variation in narrative. Language 57(1). 45–62.Schwenter, Scott & Rena Torres Cacoullos. 2008. Defaults and indeterminacy in temporal

grammaticalization: The ‘perfect’ road to perfective. Language Variation and Change 20(1). 1–39.

Seoane, Elen a & Cristina Suárez-Gómez. 2013. The expression of the perfect in East and South-East Asian Englishes. English World-Wide 34(1). 1–25.

Silva-Corvalán, Carmen. 1983. Tense and aspect in oral Spanish narrative. Language 59(4). 760–780.

Smith, Barbara Herrnstein. 1981. Narrative versions, narrative theories. In William John Thomas Mitchell (ed.), On narrative, 217. Chicago: University of Chicago Press.

Tagliamonte, Sali. 2003. Comparative sociolinguistics. In Jack K. Chambers, Peter Trudgill & Natalie Schilling-Estes (eds.), The handbook of language variation and change, 729–763. Malden: Blackwell.

Tagliamonte, Sali. 2006. Analysing sociolinguistic variation. Cambridge: Cambridge University Press.

Tagliamonte, Sali. 2012. Variationist sociolinguistics: Change, observation, interpretation. Malden: Wiley-Blackwell.

Tamminga, Meredith & Aaron Ecay. 2013. Persistence as a diagnostic of grammatical status: The case of Middle English negation. Paper presented at Diachronic Generative Syntax (DiGS) 15, University of Ottawa, 1–3 August.

Torres Cacoullos, Rena. 2001. From lexical to grammatical to social meaning. Language in Society 30. 443–478.

Torres Cacoullos, Rena & Catherine Travis. 2015. Gauging convergence on the ground: Code-switching in the community. International Journal of Bilingualism 19(4). 365–386.

Travis, Catherine. 2007. Genre effects on subject expression in Spanish: Priming in narrative and conversation. Language Variation and Change 19. 101–135.

Trudgill, Peter. 1974. The social differentiation of English in Norwich. Cambridge: Cambridge University Press.

Vendler, Zeno. 1967. Linguistics in philosophy. Ithaca: Cornell University Press.Walker, Jim. 2008a. The footballer’s perfect: Are footballers leading the way? In Eva Lavric,

Gerhard Pisek, Andrew Skinner & Wolfgang Stadler (eds.), The linguistics of football, 24–38. Tübingen: Narr.

Walker, Jim. 2008b. The strange phenomenon of the footballer’s perfect. In Geneviève Girard-Gillet (ed.), Etrange/Etranger: Etudes de linguistique anglaise [Strange/Stranger: English linguistic studies], 21–32. Saint-Etienne: Publications de l’Université de Saint-Etienne.

Walker, Jim. 2011. The emergence of the narrative present perfect in British English: Reality or illusion? Groninger Arbeiten zur germanistischen Linguistik 53(2). 71–87.

Werner, Valentin. 2 013. Temporal adverbials and the present perfect/past alternation. English World-Wide 34(2). 202–240.

Winter, Joanne. 2002. Discourse quotatives in Australian English: Adolescents performing voices. Australian Journal of Linguistics 22(1). 5–21.


Wolfson, Nessa. 1978. A feature of performed narrative: The conversational historical present. Language in Society 7(2). 215–237.

Wolfson, Nessa. 1982. CHP: The conversational historical present in American English narrative. Dordrecht: Foris Publications.


Yao, Xinyue. 2015. The present perfect and the preterite in Australian English: A diachronic perspective. In Peter Collins (ed.), Grammatical change in English world-wide, 247–268. Amsterdam: Benjamins.

Part II: Perfects across varieties of English

Bertus van Rooy5 Present perfect and past tense in Black

South African EnglishAbstract: The direction of change in the use of the present perfect form across varieties of English is towards a lower frequency of the perfect and a higher fre-quency of past tense form use. In Black South African English, the present perfect in the news register shows a decline over time in frequency relative to the past tense, while synchronic data indicates that news is more advanced than conver-sation and student writing as far as this change is concerned. There is limited evi-dence for the past tense form encroaching on the semantic space of the present perfect. This study finds that a normative constraint may also support, rather than dampen, a change in the newspaper register, while production pressure among less proficient users may favour the more explicit, present perfect, form in student writing. Support is also found for the explanatory value of progression along the cycle of Schneider’s Dynamic Model.

1 IntroductionResearch on the present perfect has focussed mainly on differences between British and American English (henceforth BrE and AmE) until recently, yield-ing the insight that AmE is more prone to use the simple past tense form where BrE uses the present perfect form (Elsness 1997, 2009; Biber et al. 1999: 462–463; Hundt and Smith 2009). The scope of investigation has now broadened to other varieties (see, e.g., Van Rooy 2009; Davydova 2011; Yao and Collins 2012; Seoane and Suárez-Gómez 2013; Suárez-Gómez and Seoane 2013; Werner 2013, 2014; Roy 2014; Yao 2014; see also Werner, Seoane, and Suárez-Gómez, this volume). Many of these varieties can be plotted on a continuum of more perfect-friendly or more past-tense friendly (Yao and Collins 2012; Werner 2013), showing traces of the preference of the original parent dialect, but also evidence of the degree to which varieties have progressed along the phases of Schneider’s (2003, 2007) Dynamic Model. Research also shows that the differences between varieties are not of a similar kind across registers, with spoken conversation showing more encroach-ment of the past tense form on the semantic space of the present perfect than edited written registers (Suárez-Gómez and Seoane 2013: 177).

Bertus van Rooy, North-West University

150 Bertus van Rooy

Suárez-Gómez and Seoane (2013: 175) attribute the finding that the use of past tense forms in the semantic space associated with the present perfect is much more prevalent in (unplanned) spoken that written data to the weaker strength of normative constraints on the selection of forms in the spoken dialogues. Further-more, Werner (2014: 314) finds that the spoken dialogue registers of the varieties he investigates (native and non-native) are relatively more similar in the use of the present perfect than the written registers. In part, differences in kind of anal-ysis undertaken can account for these differences in findings, while differences in the selection of varieties for analysis may also be responsible for differences in conclusions. In addition, more research has been done on Asian than African varieties of English, which restricts the degree to which current findings can be expanded to generalisations about non-native varieties more widely.

The use of the present perfect has not yet been studied systematically in corpora of Black South African English (BSAfE), a non-native variety of English used by speakers of African languages. Mesthrie (2008: 626) observes that the past tense form is used in contexts where the present perfect is preferred in other varieties, which suggests a similarity with past-tense friendly varieties. De Klerk (2003: 234–235) notes that the unmarked form of the verb is occasionally found with the adverb already. On the basis of the informant questionnaires of the WAVE project, Huber (2012: 819) notes that the levelling of the distinction between the past tense and present perfect is widely attested in non-native vari-eties of English across the world, but such that the present perfect encroaches on the space of the simple past – which is the opposite of the trend observed for AmE and BrE (Elsness 1997, 2009). At the same time, Huber (2012) notes that African varieties of English are an exception to the trend observed for other non-native varieties.

In view of the reports in the literature that BSAfE moves in the opposite direc-tion to the general trend for non-native varieties by using the past tense form in contexts where other varieties would use the present perfect, but also provided the limited availability of data from African varieties of English, this chapter investigates the use of the present perfect and past tense forms in BSAfE. In the process, a number of specific methodological controls are exercised in an attempt to resolve some of the potential differences in reports about the degree to which spoken language either shows more similarity across varieties or more divergence from the patterns established for written language. Furthermore, given the fact that development and change over time is implicit in the Dynamic Model (Schnei-der 2003, 2007), and used to account for the findings by some researchers, this chapter examines diachronic data to determine if there has been any change over the course of a century in BSAfE. In the process, the chapter intends to contribute to the overall base of evidence on the development of the present perfect in vari-

Present perfect and past tense in Black South African English 151

eties of English. At the same time, it addresses specific methodological concerns that may underpin differences in findings and conclusions of previous studies. A closer examination of the differences between previous studies is presented in the next section of the chapter, followed by the methodology (Section 3), the results (Section 4) and the conclusions (Section 5).

2 Research on the perfect in varieties of EnglishThe perfect form, with the auxiliary have and the past participle of the main verb, only stabilised in the Early Modern English period, although its earliest attesta-tions are present in Old English (Elsness 2009: 229–230; see also Johannsen, this volume). The choice of the auxiliary have remained in competition with be until the nineteenth century for a group of intransitive verbs of motion (Anderwald 2014: 15–18; see also Werner, this volume). More importantly, the English present perfect construction competes with the simple past tense to convey events that either took place completely or at least started prior to the moment of speaking (Biber et al. 1999: 467). Over time, the present perfect had shown a slow but steady increase from Old English, through Middle English, until the eighteenth century, but since then, the trend reversed and the past tense form has regained some of the semantic space it ceded to the present perfect (Elsness 2009: 231, 235–239). AmE has advanced further along this road than BrE, as shown by various corpus studies (Biber et al. 1999: 462–463; Elsness 2009; Hundt and Smith 2009; Yao and Collins 2012), and American informants prefer the simple past tense in a number of contexts where British informants prefer the present perfect (Elsness 2009: 236–239). However, Hundt and Smith (2009), using the Brown family of corpora (and others) as their data, find that very little change took place from the 1960s to the 1990s.

Corpus studies of varieties of English other than BrE and AmE are not as extensive (Werner 2014: 1–2), nor do findings from these studies present a clear-cut picture, as subsets of native and non-native varieties are found at both ends of the frequency spectrum. Yao and Collins (2012), in a study of selected compo-nents and registers of the International Corpus of English (ICE), report that native varieties range from AmE with the lowest frequency of present perfect forms to BrE with the highest, while non-native varieties cover a similar frequency range, from Philippine English (PhiE) with the lowest frequency to Indian English (IndE) with the highest. Werner (2014) studies a different selection of ten varieties, using all the ICE-registers and not just a selection, but reports a similar finding: New Zealand English, Australian English and BrE among the native varieties, and

152 Bertus van Rooy

Hong Kong (HKE) and IndE among the non-native varieties show a relatively higher frequency of the present perfect form, while Irish (IrE) and Canadian English (CanE) join PhiE and Jamaican English (JamE) with lower average fre-quencies. In his study, Singaporean English (SinE) is the variety with the lowest frequency (Werner 2014: 266).

On pure frequency grounds, a generalisation that non-native varieties in Africa use the present perfect in the semantic space that native varieties assign to the simple past tense form (Huber 2012: 819) is not supported. Further evi-dence comes from a closer semantic analysis of the competition between the two forms, but only a small number of studies undertake this fine-grained analysis. Van Rooy (2009) analyses perfect forms from face-to-face dialogues and student writing in three ICE corpora (for BrE, HKE and East African English [EAfE]), and proposes a range of senses that the form conveys in the data. Only one of these senses, labelled “past fact”, and occurring in a small minority (less than 5 % per variety) of instances, encroaches on the semantic territory of the past tense.

In two related studies, Suárez-Gómez and Seoane (2013) and Seoane and Suárez-Gómez (2013) examine the choice between the two forms in the environ-ments where a typical perfect meaning is expressed. They find that BrE is more conservative in choosing the perfect form, while four Asian varieties of English – SinE, HKE, PhiE and IndE – use the past tense form with a significantly higher fre-quency in these contexts. They adopt two different analysis strategies. In Seoane and Suárez-Gómez (2013), they select the ten most frequent verbs (excluding the verbs be and do), and examine their use in the private dialogue section of the corpus. They manually identify contexts that convey a perfect meaning in their data, using the standard literature on the semantics of the perfect (Seoane and Suárez-Gómez 2013: 4), and then establish which tense and aspect form is selected to encode the perfect meaning in context. For these ten verbs, it is then established that BrE conversation only uses the simple past tense form in 16.4 % of the cases, while the four Asian varieties use it in 33 % of all cases; conversely, the perfect is selected in 80.8 % of the cases in BrE but only 60 % of the time in the four Asian varieties, a difference which is shown to be statistically significant (Seoane and Suárez-Gómez 2013: 9–10). In the other analysis (Suárez-Gómez and Seoane 2013) they take a different route when they select adverbials that may potentially cue the perfect meaning, and then determine the relative proportion of present perfect versus simple past tense forms used in those contexts. They find that the Asian varieties use the past tense form more often than BrE, but that IndE is relatively close to BrE in its choices.

Werner (2013) adopts a similar strategy to Suárez-Gómez and Seoane (2013). He analyses associations between temporal adverbials historically associated with the perfect and the form of verbs within their scope. He includes a wider range


of ICE components of both native and non-native varieties in his analysis, going beyond Asia. This wider scope of data leads him to a similar conclusion as studies on the frequencies of the perfect (e.g. Yao and Collins 2012; Werner 2014), namely that native and non-native varieties alike include varieties with a stronger prefer-ence for the past tense or the perfect forms in ambiguous contexts, in other words, that the native and non-native varieties do not form two discrete sets, with native varieties on one side of the spectrum and non-native varieties on the other side.

A number of explanations are offered for the findings about the levelling of the present perfect and simple past, as well as the decrease in frequency of the present perfect. Elsness (2009: 243) argues that the formal similarity of the present perfect and simple past is one reason why English is an exception to the trend observed for a number of European languages, such as German and French (where the present perfect forms encroach on the space of the simple past tense), in the sense that the past tense and past participle form of the verb are often identical, and in speech the auxiliary often undergoes cliticisation and reduction. He adds that BrE appears to be more sensitive than AmE to normative guidelines about the combination of the present perfect with a number of adverbials (2009: 243–244). In the case of non-native varieties, it is not clear whether the same type of explanation will be valid.

Mesthrie’s (2006) anti-deletion thesis for BSAfE, that it would rather make use of an overt form (in the case at hand, then, the auxiliary have) than omit it in cases where the two possibilities are relatively similar in meaning, predicts the use of the present perfect rather than simple past in overlapping contexts. While Mesthrie (2006) specifically restricts his claim to BSAfE, it is extended to African varieties more generally by Mesthrie and Bhatt (2008: 91–92). The basic argument is that the tendency in substrate languages not to have deletion rules supports anti-deletion, whereas in many Asian varieties, there is a stronger tendency to delete elements, in line with different pressures from those substrates. However, Mesthrie and Bhatt (2008: 72) also draw attention to other regularisation pres-sures which also result in the choice for more transparent forms in non-native varieties. Kruger and Van Rooy (2016) incorporate anti-deletion as part of a general trend towards explicitation in non-native language production, in terms of which non-native speakers select the more explicit form where two variants are available. This trend is supported by a production constraint that privileges the more analytic form on psycholinguistic grounds, as well as a normative con-straint that privileges the more formal choice among two variants.

Yao and Collins (2012: 400) add the possibility that colloquialisation across registers in the South-East Asian varieties further explains the clustering among the non-native varieties. Their explanation appears to contradict the explicitation thesis, but if one considers that colloquialisation is more likely found in countries

154 Bertus van Rooy

where English is used more widely in the society, then the greater degree of usage may overcome explicitation.

The developmental trajectory along the five phases of Schneider’s (2003, 2007) Dynamic Model is used by some researchers to explain the data. Schneider argues that native and non-native varieties of English gradually move from an exornomative phase to an endonormative phase, which would result in greater difference from the colonial parent of a variety. Applied to the decrease in fre-quency of the present perfect form, researchers such as Yao and Collins (2012), Seoane and Suárez-Gómez (2013), and Werner (2013) argue that this is why the native varieties outside Britain, which have all reached phase 5, have progressed furthest along this route, and why SinE, which has advanced the furthest among the non-native varieties, also shows the lowest frequency of present perfects among this group. However, there is an inherent historical implication in such an explanation, which is inferred from differences in synchronic linguistic data and external socio-historical data, without diachronic linguistic data available so far. If the necessary historical data can be examined, the force of the explanation would be strengthened.

Transfer from substrate languages is another factor that is considered, in that patterns of preference that are transferred from the home languages of non-native users of English. Seoane and Suárez-Gómez (2013: 20–21), however, find that in particular when comparing HKE and SinE, with typologically similar substrates, the developmental cycle offers a better account of the differences. They also con-sider processing constraints that give rise to simplification, and thus the choice of the simpler form among the two (2013: 20), although this explanation might be open to challenge if the more advanced non-native varieties also resemble native varieties other than BrE, where the processing constraint associated with language contact does not operate.

Kruger and Van Rooy (2016) propose that non-native varieties (alongside other forms of language production that take place under conditions of height-ened strain) can be approached fruitfully from the perspective of various con-straints that interact. They show that non-native varieties are subject to increased strain during production, which may lead to the selection of more overt, ana-lytical forms since they are easier to produce, while forms that make meaning more explicit for the receiver of the communication are also preferred. Often, the processing and explicitation constraints reinforce each other. Furthermore, non-native varieties are sometimes conservative in preferring more normative choices, in part due to the context of acquisition in the education system and more extensive exposure to written input compared to relatively less frequent use of informal spoken language. These constraints tie in with some of the expla-nations offered for the use of the present perfect form, as pointed out during the


course of this section, and will be employed as part of the framework for inter-pretation of the data.

3 MethodologyQuantitative and qualitative analyses of the use of the present perfect in BSAfE are reported in this chapter. The quantitative analysis focuses on the frequency of the present perfect and past tense form in three registers from a corpus of BSAfE: face-to-face conversation, student writing and news reportage. These registers are selected because they represent different kinds of text production circumstances: conversation is unmonitored, spontaneous language production, student writing represents a typical form of text production in Outer Circle con-texts but the texts are not edited by another party, while news reportage usually goes through an independent editing phase. The processing constraint is hypoth-esised to be stronger in the spoken data and the student writing, whereas the nor-mative constraint would be more in evidence in the news reportage. To determine whether there are changes over time, a quantitative analysis of newspaper data from earlier periods of Black South African English is undertaken.¹

The contemporary and historical corpora are the same as those used in recent studies of BSAfE (see, e.g., Van Rooy and Piotrowska 2015). Contemporary data for face-to-face conversation is taken from De Klerk’s (2003) Xhosa-English Corpus; student writing is taken from the Tswana Learner English Corpus (Van Rooy 2006); and news reportage from a selection of texts sampled according to ICE sampling principles.² The sampling period for all contemporary data is 2000–2007. The historical newspaper data, like the corresponding contemporary data, are taken from corpora of newspapers that were staffed largely by black South Africans, and not merely written by black authors, to minimise the effect that editing by native speakers might have on the outcome of the process. The newspaper data is divided into two periods: the older period is the late nineteenth to early twentieth century, and the middle period is the mid-twentieth century (1944–1952). Table 1 presents the sizes of the various components of the corpora used.

1 Data from published fiction is also available for contemporary and earlier BSAfE, but fiction generally (and also in the case at hand) makes extensive use of the past tense and very infrequent use of the present perfect, so that it does not contribute much new data.2 Referencing to the corpora is as follows: the conversation files start with the code “dpc”, as originally used by De Klerk (2006); the student writing files start with TS; and the news reportage are labelled W2C, as they would be in ICE.

156 Bertus van Rooy

Table 1: Corpora and size.

Corpus Word count

Contemporary Sample from Xhosa-English conversation 113,440Sample from Tswana Learner English Corpus 32,417News (2007) 41,810

Historical Older period news (1884–1918) 143,832Middle period news (1944–1952) 156,585

The corpora were tagged with the C7 tagset of the CLAWS tagger (Garside and Smith 1997), but given the challenges posed by especially unedited written and spoken non-native data, the samples from the conversation and student writing were submitted to a process of manual tag correction. The newspaper data were tagged, but not manually corrected. For the quantitative part of the study, the tagged corpora were used, and data were extracted using WordSmith Tools 6 (Scott 2014). To determine the number of simple past tense forms, all past tense tags (VBDZ, VBDR, VDD, VHD, VVD)³ were retrieved, and then all instances of the past progressive or past perfect removed manually from the concordances. To determine the number of present perfect forms, all instances of the present tense (finite) forms of the verb have (tags VH0 and VHZ)⁴ were extracted, and all present perfect forms manually selected, using the tags as guidance. The form have got and all present perfect progressive forms were excluded as was done by most other researchers. Absolute numbers, as well as normalised frequencies per one million words, are reported throughout.⁵

The qualitative analysis examined a random selection of 100 instances of the present perfect form and a further 200 instances of the simple past tense form from each of the three contemporary registers of BSAfE, to establish the degree to which the two forms are used in meanings that are typically associated with the other form.⁶

3 VBDZ = verb be, past tense, singular; VBDR = verb be, past tense, plural; VDD = verb do, past tense; VHD = verb have, past tense; VVD = any other verb, past tense.4 VH0 = verb have, present tense, non-singular; VHZ = verb have, present tense, singular.5 Given the corpus sizes, a base smaller than a million words, such as instances per 1,000 words, would seem more appropriate. In light of the fact that previous researchers either use corpora like ICE or the Brown family, or otherwise also choose to normalise smaller numbers to instances per million words, this choice is made to ensure comparability during the discussion of findings.6 The frequency of perfect-like and other non-past uses of the past tense form was quite low, and therefore a larger number of examples were analysed to safeguard against the effects of sampling bias due to small numbers.


The framework of analysis for the present perfect builds on recent work by Seoane and Suárez-Gómez (2013) and Yao (2014), which in turn is based on the description of the present perfect in the grammatical literature of English, espe-cially Huddleston (2002). The semantic categories that are distinguished are the resultative, continuative, recent past and experiential perfect.

The continuative reading of the perfect denotes an event that started prior to the utterance time, but continues to hold at the time of the utterance. It is proto-typically associated with stative predicates (Givón 1993: 166), as illustrated by (1), where the father started being an inspiration for the speaker since he was a child, but also continues to be an inspiration up to and beyond the time of utterance.⁷

(1) [...] mostly my father i can say that in fact ’m allowed to say that my father he’s been a source of inspiration since i was a kid(dpc080)

In principle, dynamic verbs that receive a habitual reading in context can also be used continuatively (Yao 2014: 90–91), but in practice, this was very rare in the data.

While a sense of current relevance is central to the use of the present perfect in any sense, the resultative profiles the result state following the event of the verb, as illustrated by (2), where the state of shattered dreams continues to obtain after the event that caused it, which occurred sometime prior to the time of utter-ance.

(2) [...] people who have graduated from technikons & University and most of them are unemployed. Their dreams have been shattered due to lack of knowledge about Science & technology.(TSNO1485)

Yao (2014: 99) sets as criterion that only verbs denoting events that are goal-ori-ented (accomplishments and achievements in terms of Vendler 1957) can be used in the resultative sense, while states and activities that do not have natural end points cannot. This is a stricter interpretation of the resultative than Seoane and Suárez-Gómez (2013: 13). Yao (2014: 108) gives a useful criterion to draw the dis-tinction between the resultative and experiential readings, when she argues that the resultative reading is excluded by contexts where the focus is on the occur-

7 The target verb and auxiliary are in bold print in all examples.

158 Bertus van Rooy

rence of the situation in the past, as opposed to its consequences for the refer-ence time.

The recent past reading (including the “hot news perfect”), following Seoane and Suárez-Gómez (2013: 13), usually requires adverbial support to indicate its recency, although information in the wider context may also be indicative of this. In example (3), the recent loss of the mother is evident from the fact that her burial is yet to take place at the time of utterance.

(3) Moropodi has lost his mother and she will be buried this coming Saturday.(W2C-004)

The experiential perfect is a default category that lacks the more specific associ-ations of the previous three (Yao 2014: 106–107). Any perfect form that does not have an obvious tense-like reading, but retains a discernible connection of rele-vance between the time of utterance and the prior event, while not correspond-ing to one of the more specific other readings, was classified as an experiential perfect. This included reference to events that happened at some indefinite time during a period leading up to the time of utterance (Seoane and Suárez-Gómez 2013: 13; Yao 2014: 105–106), as exemplified by (4), where families have lost their loved ones at various points in time in the period leading to “today”, during which AIDS has increased. The experiential perfect also includes speech acts or mental processes that do not specifically result in an end-state at the time of utterance (Van Rooy 2009: 321; Yao 2014: 107–108), exemplified by (5), where the speaker refers to an earlier point he made, but in choosing the present perfect form, he signals that the point is still relevant to the current argument.

(4) Aids is a veneral disease which is a concern to African Countries and it is increasing at an alarming rate that today many people have lost their life. More families have lost their loved ones the children have become orphans and others are also affected by this disease.(TSNO1277)

(5) SH: so that’s what i uh i’ve proved my point that yizo yizo it’s so wrong that is why because people they take negative side SS: so through yizo yizo, the crime and the corruption will increase rapidly you you have you have said that SH: yes i’ve said that(dpc183)


When analysing the present perfect-like uses of the past tense form, the same cat-egories were employed in determining whether the instance represented a perfect meaning. Beyond the conventional present perfect meanings, a number of remote past uses were also observed in the data, both among the present perfect and the simple past tense forms, as illustrated by (6), where an event is situated in a past time frame further back than some already established reference point in the past. These are typically expressed by the past perfect form in native varieties of English.

(6) According to Ntusi, the problem started after Ajax Cape Town coach left the team [...] (W2C-004)

The conventional past tense meanings were not in focus in this study, only the extent to which past tense forms were used in meanings typically conveyed by the present perfect form. Therefore, no finer semantic classification of past tense forms was attempted. In general, under the broad rubric of conventional uses of the past tense form, the following were included: events at some definite point in the past, as illustrated by (7); events within a past time frame without an overt adverbial, as illustrated by (8); and subjunctive and other conditional/irrealis uses,⁸ exemplified by (9).

(7) There will be no schooling in Khutsong until the incorporation dispute is resolved, SACP deputy chairman Jomo Mogale said yesterday.(W2C-014)

(8) and so we were an extended family i living with my eight aunts my uncle my uncles [...] because like my my my my grand my great grandmother was a very nice person(dpc228)

(9) It was not practical for me to sell it unless I owned a production company.(W2C-014)

8 Given the formal similarity of the subjunctive and the past tense, no attempt was made to draw a fine distinction between the subjunctive in the narrow sense, and forms that denoted a condition within the context of a past time-frame. Crucially, of course, these forms do not usually denote a specific event at a definite point in time, but are nevertheless regarded as conventional context for the use of the simple past tense form.

160 Bertus van Rooy

4 ResultsThe overall quantitative findings for the three contemporary registers are pre-sented in Table 2. Using the results of the corresponding registers in Yao and Collins (2012: 292) as basis for comparison, the results for BSAfE are in many ways similar to the results for the Asian varieties of English that have advanced to a moderate degree along the path towards past tense-friendliness, PhiE and HKE in particular. The frequency for news reportage is also similar to NigE (reported as 5,500 pmw by Werner and Fuchs 2014), although the frequency for conversation is much higher than NigE (1,100 pmw; Werner and Fuchs 2014), but again similar to PhiE and HKE (both 2,700 pmw; Yao and Collins 2012). The numbers are con-siderably lower than Kenyan English (KenE)/EAfE and IndE (both 4,500 pmw for conversation, and 6,200 and 9,800 pmw respectively for news; Yao and Collins 2012), and also lower than those reported for BrE (4,200 pmw for conversation, and 8,900 pmw for news; Yao and Collins 2012).

Table 2: Absolute and normalised frequencies (per million words) of past tense and present perfect forms in the contemporary data, together with the present perfect/simple past ratio.

Present perfect Past tense Ratio

N Norm. N Norm.

Conversation 321 2,880 1,925 17,290 0.17Student writing 115 3,550 277 8,540 0.42News 235 5,620 1,837 43,940 0.13

Results for the distribution of present perfect forms in student writing in differ-ent varieties of English have not been reported separately by many researchers. Van Rooy (2009: 316) reports values for the ICE-corpora from BrE (7,100 pmw), EAfE (5,400 pmw) and HKE (2,400 pmw), while Werner and Fuchs (2014) report a similar value for BrE, and a frequency of 2,700 pmw for NigE. The BSAfE fre-quency is lower than BrE, similar to EAfE, and higher than NigE and HKE. When the ratio of present perfects to simple past tenses is considered, a slightly differ-ent perspective emerges, in that BSAfE also uses the past tense very infrequently in student writing, compared to all other varieties for which results have been reported, which results in a very high proportion of present perfects. It is only in NigE news reportage that a similar ratio (approximately 0.50) is observed (Werner and Fuchs 2014).

The historical data in Table 3 confirm the overall picture known from pre-vious research: there is a gradual decrease in the proportion of present perfect forms relative to past tense forms over time in BSAfE news reportage, from a high


of 0.26 at the end of the nineteenth/beginning of the twentieth century to half that, at 0.13, by the beginning of the twenty-first century. Taken together, the data confirm that in BSAfE, like in the native varieties for which data are available, the present perfect form has gradually ceded its semantic space to the simple past tense when referring to events prior to the moment of utterance. This interpreta-tion is premised on the assumption that across such a relatively short time-span (judged against the time-scale of human evolution), the construal of the semantic space prior to utterance time has not changed dramatically, and the change is in the selection of the forms to convey the various construals of the time before the moment of speaking, where the past tense forms carry an increased functional load.

Table 3: Absolute and normalised frequencies (per million words) of past tense and present perfect distribution in the news data over time, together with the present perfect/simple past ratio.

Present perfect Past tense Ratio

N Norm. N Norm.

Earlier (1884–1918) 980 6,810 3,706 25,770 0.26Middle (1944–1952) 867 5,540 4,820 30,780 0.18Contemporary (2007) 235 5,620 1,837 43,940 0.13

BSAfE, which Schneider (2007) classifies as an indigenous (IDG) strand within South African English, is in the process of reaching phase 4 within the Dynamic Model (Schneider 2003, 2007: 185–188), and as such clusters with other varieties that are about the reach phase 4. It lags behind SinE, which has reached phase 4 already, but is more advanced along this route than KenE and IndE, which are still regarded as being in phase 3 (Schneider 2007). The explanation offered by Seoane and Suárez-Gómez (2013: 20–21), Werner (2013: 233) and Yao and Collins (2012: 399–400), that varieties that have moved further along the developmen-tal cycle of the Dynamic Model have moved further towards past-friendliness, is therefore supported, but also strengthened by the evidence of the historical data, which confirm the gradual decline in frequency of the present perfect relative to the simple past.

However, the exceptional behaviour of student writing and the fact that con-versation has a higher relative proportion of present perfect forms than news reportage in BSAfE are trends that depart from what is reported for other varieties. In particular, the possibility that some kind of colloquialisation takes place in a way that favours the use of the simple past above the present perfect (Yao and Collins 2012: 400) does not extend to BSAfE, since the spoken conversation regis-

162 Bertus van Rooy

ter is not the register with the lowest proportion of present perfect relative to the simple past tense forms. Before examining this finding in more detail, alongside findings about the influence of register by Werner (2013) and Suárez-Gómez and Seoane (2013), it is necessary to consider the findings of the semantic analysis first.

The basic findings for the two samples submitted to semantic analysis are presented in tables 4 and 5. Table 4 shows that past-tense like uses of the present perfect form occur in only a small number of instances, whereas the various con-ventional present perfect uses are in the majority by far, similar to the finding of Van Rooy (2009).

Table 4: Readings of the present perfect form (absolute frequencies).

Conversation News Student writing

Continuative 10 9 6Experiential 62 59 58Recent past 7 5 2Resultative 10 14 23Present tense-like 3 – 6Past tense-like 5 5 4Remote past 3 8 1

Total 100 100 100

There are, however, also a number of instances judged to represent typical present tense and remote past uses across the registers. Example (10) illustrates a typical present tense-like use of the present perfect form, where the context of a general truth present appears to be the intended meaning, and the event of using “our knowledge” is simultaneous, rather than prior to the acts of depositing or withdrawing money.

(10) In every day of our lives we use technology, even when we are not aware of this. When we go to the automatic bank machiens (A.T.M.) to deposit or to withdraw money, we have used our knowledge of technology to operate that machine (A.T.M.)(TSNO1434)

A remote past use, specifically backshifting (Huddleston 2002: 151–155) is illus-trated in (11), where the reporting verb “confirmed” establishes a time frame in the past, while the complement clause introduces the discussions that had occurred prior to the event of the main clause subject “Mohau” confirming those discussions.


(11) Mohau confirmed that he has had “discussions” with Nzimande about some statements he has issued.(W2C-016)

There is thus limited evidence in the corpus that the present perfect form encroaches on the semantic space of the simple past tense. To the extent that the opposite happens, that is, that the past tense form increases the scope of its use into the semantic space of the present perfect, as reported by Seoane and Suárez-Gómez (2013), one would predict a higher number of past tense forms used in meanings associated with the present perfect. However, as Table 5 shows, this is generally not the case in BSAfE.

Table 5: Readings of the past tense form (absolute frequencies).

Conversation News Student writing

Past 190 192 153Perfect-like 5 4 7Remote past – 4 5Present-like 5 – 35

Total 200 200 200

Apart from an unexpectedly high proportion of present-tense like uses of the simple past tense in student writing, the past tense is used in conventional ways in the three registers. Example (6) above has already illustrated the remote past use of the simple past tense form, while the use of the simple past form to evoke a perfect-like meaning is illustrated in (12). The time frame of a general truth is established in this passage and the preceding text, and then the fact that chil-dren are without parents leads to a consequence, the state of being poor. This is a resultative meaning, and can therefore potentially be expressed by a present perfect.

(12) The economy is going down because of the HIV/AIDS. Children are left with no parents and that caused them to be poor.(TSNO1095)

The most interesting innovation emerging from the data analysis is the use of the simple past form in a meaning that one would typically associate with the present tense, as exemplified in (13).

164 Bertus van Rooy

(13) My point of view is that poverty is the cause of problems leading to high per-centage of Hiv/Aids. Poverty sometimes had an influence on people to do something that was not their intention to do it for instance you will try by all means to get the money(TSNO1139)

The influence that poverty has on people is clearly a consequence that holds true generally and is not bounded to some period of time. The influence is actualised every time people try to do anything to get money.

In summary, the present perfect is used with a relatively lower frequency in BSAfE compared to other African varieties of English as well as BrE, which is its historical input variety. It is similar to PhiE and HKE in Asia, and more per-fect-friendly than IndE, but not as past-friendly as SinE. Diachronic data from the news register furthermore indicates that the proportion of present perfects has declined during the past century.

The conversation register is slightly more perfect-friendly than news report-age, which differs from other varieties for which results are available. According to available information (De Klerk 2006), the speakers who contributed to the spoken corpus are likely to be of lower education levels than the journalists who write for newspapers, which may account for some of the differences, but then it is also likely that a proper cross-section of the English using community in many Outer Circle contexts are likely to be of lower levels of education than journalists there. Both these registers use the present perfect and simple past tense in largely conventional senses, with few instances where the one is used in the semantic space conventionally associated with the other. Student writing is considerably more perfect-friendly than the other registers, and while the meanings typically conveyed by the perfect are largely conventional when judged from the literature on the perfect, the past tense shows a large number of exceptional usages where the inflected past tense form conveys a general truth meaning conventionally associated with the simple present tense form.

5 ConclusionThe progression of varieties of English along the developmental cycle of the Dynamic Model (Schneider 2003, 2007) is taken to be an important part of the explanation of the data from non-native varieties. As indicated earlier, most pre-vious investigations of data from the New English/Outer Circle varieties find that those countries that have proceeded further along the path towards endonorma-


tivity (phase 4) are more past tense-friendly, with lower proportions of present perfect usage compared to the simple past, while varieties that are still located in the nativization phase (phase 3) are more perfect-friendly with higher pro-portions of the present perfect (Yao and Collins 2012; Seoane and Suárez-Gómez 2013; Werner 2013).

When the register distribution of the BSAfE data is taken into account, there appears to be evidence for a normative constraint, in that the most edited register, news reportage, leads the change towards past-friendliness, with conversation lagging slightly behind news. There is a gradually decreasing frequency of the present perfect during the course of the twentieth century, with the more edited register leading the change. While it is known that news is a progressive register among the written registers (Hundt and Mair 1999), it is not a typical finding that a change in written language leads spoken language, except in the case of deliber-ate changes, such as non-sexist language reform or the conscious dissemination of prestige variants (Curzan 2014: 43, 129). The nature of the normative constraint in data such as this requires more investigation, to determine what potential role editorial intervention plays in the dissemination of language change, particu-larly when seen against the backdrop of the much more perfect-friendly nature of student writing.

For student writing, a processing constraint provides a better account, as these language learners who are still on their way to the kind of proficiency already achieved by newspaper writers, seem to find it useful to use an overt, analytical form more frequently. The inflected past tense is even to a degree under pressure to convey a present-like meaning, which is not a regular association with the other two registers in the data. The time-pressure of the production condi-tions of the student writing, together with generally lower proficiency in all like-lihood accounts for the ambiguity of the past tense forms, particularly when read together with reports that past-time meanings are sometimes not given overt mor-phological expression and the simple present conveys such past-time meanings (Minow 2010: 111–125; Makalela 2013: 101–102). The BSAfE data do not confirm the kind of morphological simplification that Seoane and Suárez-Gómez (2013) propose, since the simplification (and even uncertainty) that does occur relates to morphological inflection only, whereas the analytic perfect auxiliary appears to be more resistant to deletion in the registers where the cognitive constraint of processing difficulty is expected to exert a stronger influence. The present finding is in line, however, with the general characterisation of non-native varieties as more analytic, which is a different kind of simplicity that prefers overt forms and one-to-one correspondences between form and meaning (see Szmrecsanyi and Kortmann 2009).

166 Bertus van Rooy

Colloquialisation does not contribute to an explanation of data across the reg-isters of BSAfE. The direction of the change is towards the form in news reportage, which is the least colloquial of the registers. It may be that as varieties proceed further along the cycle of the Dynamic Model, and English is used even more widely and more proficiently, that colloquialisation potentially becomes a stron-ger force, but at the stage where BSAfE currently is, that has not yet happened.

AcknowledgementsThe financial support of the National Research Foundation (NRF) is gratefully acknowledged, although the views and opinions are those of the author and should not be attributed to the NRF. I would also look to acknowledge the valuable criticism of Valentin Werner and two anonymous reviewers, as well as assistance with data collection by Caroline Piotrowska, and discussions about the various constraints with Haidee Kruger. All remaining shortcomings are for my account.

ReferencesAnderwald, Lieselotte. 2014. The decline of the be-perfect, linguistic relativity, and grammar

writing in the nineteenth century. In Marianne Hundt (ed.), Late Modern English syntax, 13–37. Cambridge: Cambridge University Press.

Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad & Edward Finegan. 1999. Longman grammar of spoken and written English. Harlow: Longman.

Curzan, Anne. 2014. Fixing English: Prescriptivism and language history. Cambridge: Cambridge University Press.

Davydova, Julia. 2011. The present perfect in non-native Englishes: A corpus-based study of variation. Berlin: Mouton de Gruyter.

De Klerk, Vivian. 2003. Xhosa English as an institutionalised variety of English: In search of evidence. English World-Wide 24(2). 221–243.

De Klerk, Vivian. 2006. Corpus linguistics and World Englishes: An analysis of Xhosa English. London: Continuum.

Elsness, Johan. 1997. The present perfect and preterite in contemporary and earlier English. Berlin: Mouton de Gruyter.

Elsness, Johan. 2009. The present perfect and the preterite. In Günter Rohdenburg & Julia Schlüter (eds.), One language, two grammars? Differences between British and American English, 228–245. Cambridge: Cambridge University Press.

Garside, Roger & Nicholas Smith. 1997. A hybrid grammatical tagger: CLAWS4. In Roger Garside, Geoffrey Leech & Tony McEnery (eds.), Corpus annotation: Linguistic information from computer text corpora, 102–121. London: Longman.


Givón, Talmy. 1993. English grammar: A function-based introduction, Vol. 1. Amsterdam: Benjamins.

Huber, Magnus. 2012. Regional profile: Africa. In Bernd Kortmann & Kerstin Lunkenheimer (eds.), The Mouton world atlas of variation in English, 807–823. Berlin: Mouton de Gruyter.

Huddleston, Rodney. 2002. The verb. In Rodney Huddleston & Geoffrey K. Pullum (eds.), The Cambridge grammar of the English language, 71–212. Cambridge: Cambridge University Press.

Hundt, Marianne & Christian Mair. 1999. “Agile” and “uptight” genres: The corpus-based approach to language change in progress. International Journal of Corpus Linguistics 4(2). 221–242.


Kruger, Haidee & Bertus Van Rooy. 2016. Constrained language: A multidimensional analysis of translated English and non-native indigenised varieties of English. English World-Wide 37(1). 26–57.

Makalela, Leketi. 2013. Black South African English on the radio. World Englishes 32(1). 93–107.Mesthrie, Rajend. 2006. Anti-deletions in an L2 grammar: A study of Black South African

English mesolect. English World-Wide 27(2). 111–145.Mesthrie, Rajend. 2008. Synopsis: Morphological and syntactic variation in Africa and South

and Southeast Asia. In Rajend Mesthrie (ed.), Varieties of English, vol. 4: Africa, South and Southeast Asia, 624–635. Berlin: Mouton de Gruyter.

Mesthrie, Rajend & Rakesh M. Bhatt. 2008. World Englishes: The study of new linguistic varieties. Cambridge: Cambridge University Press.

Minow, Verena. 2010. Variation in the grammar of Black South African English. Frankfurt: Peter Lang.


Schneider, Edgar W. 2003. The dynamics of New Englishes: From identity construction to dialect birth. Language 79(2). 233–281.

Schneider, Edgar W. 2007. Post-colonial English: Varieties around the world. Cambridge: Cambridge University Press.

Scott, Mike. 2014. WordSmith Tools. Liverpool: Lexical Analysis Software.Seoane, Elena & Cristina Suárez-Gómez. 2013. The expression of the perfect in East and

South-East Asian Englishes. English World-Wide 34(1). 1–25.Suárez-Gómez, Cristina & Elena Seoane. 2013. They have published a new cultural policy that

just come out: Competing forms in spoken and written New Englishes. In Gisle Andersen & Kristin Bech (eds.), English corpus linguistics: Variation in time, space and genre, 163–182. Amsterdam: Rodopi.

Szmrecsanyi, Benedikt & Bernd Kortmann. 2009. Vernacular universals and angloversals in a typological perspective. In Markku Filppula, Juhani Klemola & Heli Paulasto (eds.), Vernacular universals and language contact: Evidence from varieties of English and beyond, 33–53. London: Routledge.

Van Rooy, Bertus. 2006. The extension of the progressive aspect in Black South African English. World Englishes 25(1). 37–64.

168 Bertus van Rooy


Van Rooy, Bertus & Caroline Piotrowska. 2015. The development of an extended time period meaning of the progressive in Black South African English. In Peter Collins (ed.), Grammatical change in English world-wide, 465–483. Amsterdam: Benjamins.

Vendler, Zeno. 1957. Verbs and times. The Philosophical Review 66. 143–160.Werner, Valentin. 2013. Temporal adverbials and the present perfect/past tense alternation.

English World-Wide 34(2). 202–240.Werner, Valentin. 2014. The present perfect in World Englishes: Charting unity and diversity.

Bamberg: University of Bamberg Press.Werner, Valentin & Robert Fuchs. 2014. The present perfect in Nigerian English. Paper

presented at the 20th International Association of World Englishes conference, Delhi, 18–20 December.

Yao, Xinyue. 2014. The present perfect in English: Meaning, interpretation and use. Sydney: University of New South Wales dissertation.

Yao, Xinyue & Peter Collins. 2012. The present perfect in world Englishes. World Englishes 31(3). 386–403.

Julia Davydova 6 The present perfect in New Englishes:

Common patterns in situations of language contact

Abstract: The main goal of this study is to pinpoint commonalities underlying the use of the present perfect in New Englishes. Drawing on spoken data obtained from the Indian, Singaporean and East African components of the International Corpus of English (ICE) as well as the London-Lund Corpus (LLC), which yields data for a reference variety in this study, I show that the use of the present perfect is constrained by the patterns which are remarkably consistent across the board. These patterns are semantic-pragmatic context and time adverbial specification. I argue that the variable realisation of the present perfect in indigenised English is best understood in terms of the diffusion mechanism fuelled by the classroom environment.

1 IntroductionThe major goal of this chapter is to investigate common patterns underlying the use of the present perfect, also known as the have-perfect (HP), in spoken forms of New Englishes. In contrast to many traditional L1 vernaculars, these varieties of English emerged as a result of the expansion of the British Empire. They evolved in multilingual settings involving high levels of bilingualism and adult second-language acquisition. This sociolinguistic situation has inevita-bly led to contact- induced changes on various levels of linguistic structure. It is then perhaps not surprising that the foregoing research has largely focused on the investigation of contact-induced, non-standard properties of New Englishes (Kortmann et al. 2004). Thus, IndE has been described as a pro-drop variety allowing for copula be omission, as in (1) and (2).

(1) A: You got tickets?B: No, pro sold pro already.(Bhatt 2004: 1025)

Julia Davydova, University of Mannheim

170 Julia Davydova

(2) I born in 1904.(Hosali 2008: 1038)

Singapore English (henceforth, SinE) is famous for its substrate-induced dis-course markers, as in (3) and (4).

(3) Come with us lah.(Wee 2004: 1068)

(4) A: What do you have to buy at the market?B: Fish lor, vegetables lor, curry powder lor.(Wee 2004: 1070)

To be sure, such contact-induced local innovations provide “some of the most spectacular features of a newly emerging variety” (Davydova 2015: 298), and are perhaps best explained in terms of the selection mechanism (Schneider 2000), which assumes the existence of a pool of individual features competing for the expression of a specific linguistic function in a new contact variety (see also Mufwene 2001: 4–6). The selection mechanism nicely captures the diverging pat-terns of New Englishes, showing how these ultimately develop into self-contained forms of English, quite distinct from their parent varieties. That said, the question arises: What is the common core underlying all these unique forms of English? To put it slightly differently, which aspects of language-internal organisation render these diverse varieties unmistakably English?

In this paper, I will explore this issue relying on the category of the present perfect. The present perfect seems to be a good testing ground since it is an essen-tial part of school/learner grammars and, thus, is a language feature that almost all learners of English are likely to be confronted with at some point in their lives irrespective of the type of indigenised environment and inputs from other lan-guages (Fuchs, Götz, and Werner, this volume). Looking for commonalties in the patterns of (acquisition and) use of the English present perfect might help to spot the converging area of L2 English grammar and thus provide empirical sub-stantiation for the descriptive claim that all varieties of English must share some “common core” (Schneider 2000: 209).

In what follows, I will look at the language-internal mechanism underly-ing the use of the present perfect in order to establish which patterns of use are common to all varieties studied here and which patterns are variety-specific. In this approach, similar or identical patterns are good candidates for “truly English” patterns, i.e. patterns transmitted from parent varieties to daughter varieties, and

The present perfect in New Englishes 171

idiosyncratic patterns are the patterns that require an explanation (compare also Collins 2015: 1–3).

This chapter is organised as follows. After providing an overview of the fore-going research (Section 2), I present the model of selection and diffusion (Schnei-der 2000) and motivate its usefulness for the study of the newly emerging L2 ver-naculars (Section 3). I then introduce the method (Section 4), present the results (Section 5) and discuss the main findings against the backdrop of the theoretical model explaining the co-existence of conservative and innovative features in New Englishes (Section 6).

2 Previous researchThe present perfect is perhaps one of the most well-studied features of World Englishes. Jespersen (1924) and Zandvoort (1932) count among the earliest descriptive accounts. Portrayals of the historical development of the have + par-ticiple construction are found in Brinton (1988) and Denison (1993). The present perfect and its functional equivalents received much attention in the study of L1 vernaculars (see, for instance, Harris 1984; Winford 1993; Tagliamonte 1996, 1997, 2000; Filppula 1997; Siemund 2004; Pietsch 2005, 2007, 2009; Van Herk 2008) and mainstream native-speaker varieties (Elsness 1997, 2009; Hundt and Smith 2009). With time, the study of the construction consisting of auxiliary have and the past participle of the main verb has been expanded to cover non-native variet-ies as well, both L2 and Learner Englishes (see, for instance, Davydova 2011, 2012; Seoane and Suárez-Gómez 2013; Suárez-Gómez and Seoane 2013). Other studies contrast the use of the present perfect in main regional varieties with that in post-colonial Englishes (Yao and Collins 2012; Werner 2014, 2016).

The foregoing research shows that both L1 and L2 Englishes attest vari-ety-specific features employed to convey the meaning obligatorily expressed with the help of the HP in standard British English (henceforth, BrE).¹ The electronic World Atlas of Varieties of English (eWAVE; Kortmann and Lunkenheimer 2013) reports that after-perfect and medial object constructions are recruited in some L1 vernaculars, notably IrE, instead of the standard HP (see Filppula, this volume).

1 Following Holmes (2013: 79), I define “standard variety” as a form of language that has under-gone codification through grammars and dictionaries. The prototypical “present perfect mean-ings” associated with BrE include resultative, extended-now, experiential and recent past mean-ings. See the section dealing with the data and method for more detail.

172 Julia Davydova

Bare verb stems can be used in place of the traditional present perfect in Singa-pore English (henceforth, SinE). These instances are illustrated in (5) through (7).

(5) Sheʼs after selling the boat. ‘She has just sold the boat’(Kortmann and Szmrecsanyi 2004, cited in eWAVE)

(6) And you eat nothing till you have the stations made.(Filppula 2008, cited in eWAVE)

(7) Maybe she increase the price already.(ICE-SIN S1A-006)

However, there is one variant that is invariably used as a main functional compet-itor of the present perfect across various forms of native and non-native Englishes. This variant is the simple past tense, also known as the preterite. The eWAVE cat-alogue demonstrates that the simple past tense can be used to convey the present perfect meaning in 59 % of all varieties targeted in the survey. These include tra-ditional and high-contact L1 vernaculars, English-based Creoles as well as indi-genised L2 varieties. Davydova (2011), a broad-scope study of variation in the domain of the present perfect marking, draws a similar conclusion. The analysis of nine datasets comprising six indigenised and two Learner Englishes as well as one reference variety shows that the HP alternates (to various degrees) with the simple past tense in all these forms of English.

(8) A lot of Germans I’ve met in India are very sweet Germans. And one I met in Sri Lanka.(HCNVE IE04)

(9) Whereas I think globalisation has made us sort of adjust to other style, the other countries […] and especially made India be understood to other cultures and countries.(HCNVE IE04)

Against this backdrop, the current study will present models of variation that essentially account for the competition of the two competing variants, the HP and the preterite, in the domain of present perfect marking.


3 Theoretical backgroundResearchers concerned with the study of New Englishes have relied on a variety of different frameworks in an attempt to explain linguistic outcomes characteris-ing New Englishes. Williams (1987) is a unique contribution assessing similarities across postcolonial Englishes in terms of psycholinguistic mechanisms involved in the learning of a second language. Mesthrie (1992) describes postcolonial varieties in terms of the developmental processes underlying second-language acquisition (henceforth, SLA). Other studies attempt to ascertain the relative role played by the indigenous languages (Bao 1995, 2005; Sharma 2005) and histor-ical English inputs, also known as superstrates, in the formation of the newly emerging varieties. Relying on Meyerhoff’s (2009) model for contact-induced language change, Davydova (2013) pinpoints identical constraints underlying the occurrence of the present perfect in SinE and IrE, a form of English that has been present in the Singaporean classroom. A considerable bulk of research has assumed a typological perspective on the study of New Englishes (see Siemund and Davydova 2014 for an overview).

This study draws on the model of selection and diffusion (Schneider 2000), the theoretical framework in which the subsequent discussion of findings is embed-ded. In this model, diffusion is a process whereby linguistic features are trans-mitted from the parent variety to the daughter variety with minimal modifica-tions and language-internal restructuring. In contrast, selection is “the outcome of a process of individual features competing with each other in language-con-tact situations, with one of the options available ultimately being “selected” into the newly-emerging variety, and other(s) usually being discarded” (Schneider 2000: 205). The model thus suggests that some features and their patterns of use are likely to be diffused, that is, transmitted from the parent to daughter variety without major modifications, whereas others are likely to result in highly diver-sified variation in the contact variety, yielding a cohort of variants competing for the expression of a specific function. In this paper, I argue that the variable real-isation of the present perfect meaning (the “perfect space”; Seoane, this volume) in indigenised English is best understood in terms of the diffusion mechanism.

Schneider’s (2000) model gives rise to the following question: Can we predict which linguistic features and their patterns of use would be transmitted from the L1 to the L2 variety through diffusion and which are more likely to result in pools of competing variants? We can begin approaching this issue by noting that English is acquired and put to use in two fundamentally different contexts in indi-genised settings. It is used as the language of instruction in the classroom and it is also used for spontaneous interactions in naturalistic settings (Mesthrie and Bhatt 2008: 5, 156). The former context tends to be heavily dominated by prescrip-

174 Julia Davydova

tivist language ideologies and orientation toward the norms of a target variety (see also Schneider 2007). The latter, in turn, lacks any such connotations, espe-cially in the environment of competing multilingual inputs. Thus, the classroom context arguably inhibits language variation and, as a result, linguistic innova-tions, whereas naturalistic contexts promote it. The implication of these observa-tions and the ensuing hypothesis is that features that are acquired in the postco-lonial classroom get transmitted through instruction with minimal modifications and are therefore best accounted for in terms of the mechanism of diffusion. In contrast, language features that are not addressed through explicit instruction but rather evolve in spontaneous interactions are more likely to give rise to the pool of competing variants that come about as a result of multilingual inputs, L1 transfer, and speakers’ internal drive for creativity (Matras 2009: 79–86). These features can be described in terms of the selection processes. While discussing the main findings of the study, I contrast present perfect marking with the use of quotative markers, that is, verbal strategies that introduce speech, thought and non-verbal material in the discourse. In so doing, I hope to be able to demonstrate how both processes are at work in New Englishes.

4 Data and methodWhile embarking on the study of the present perfect, or any other tense and aspect feature for that matter, it is crucial that the analyst should be aware that, as a domain of inquiry, this category has been subject to a considerable amount of controversy. The main difficulty concerns the method with which to study the present perfect. Because contexts in which (equivalents of) the present perfect can occur differ from language to language, and, in some instances, from variety to variety, a decision needs to be taken regarding the locus of variation.

That said, studies on the present perfect can be subdivided into three main cohorts as summarised in Van Herk (2008). There are (i) form-only studies that look at one application value only, namely the HP, and examine proportions of use across different linguistic categories; there are (ii) two-form studies which look at the past time reference system in its entirety, localise the functional niche of the present perfect and show how this relates to the functional niches of other categories with past-time reference, notably the preterite. And finally, there are (iii) function-based studies which spotlight variation of verb forms in semantic-pragmatic environments reported to be fundamental to the category of the present perfect (Comrie 1985). These contexts are illustrated in (10) through


(13) and include a resultative context, an extended-now context, an experiential context and a context of recent past.

Resultative context:

(10) I have broken my arm.

Extended-now contexts:

(11) It has been raining since morning.

Experiential contexts:

(12) I have never been to Brazil.

Contexts of recent past:

(13) The Prime Minister has been assassinated.

What all these contexts have in common is the semantic-pragmatic link of a past situation to the moment of utterance. Resultative contexts imply that a past action causes some change in the current situation or the state of affairs. Extend-ed-now contexts suggest that a situation that began in the past still persists into the present moment. Experiential contexts report on a series of past events that are likely to be repeated in the future. Finally, contexts of recent past highlight a past event as current news.

As Van Herk (2008) points out, “the choice of any one of these approaches potentially affects the eventual findings” (Van Herk 2008: 56). Following meth-odology employed in Winford (1993) and Tagliamonte (2000), I adopt the func-tion-based approach because present perfect contexts are the environments pri-marily linked to the category of the BrE present perfect (Huddleston and Pullum 2005: 49) and these are arguably the contexts which second-language speakers acquire as appropriate uses of the HP through classroom instruction. Thus, the English syllabus provided by the Ministry of Education in Singapore stipulates that the meanings of the present perfect are to be taught at the upper primary and secondary level. More specifically, pupils learn that the present perfect should be used for (i) “actions in the past that still affect the present”, (ii) “actions that have begun in the past that have continued to the present and possibly continuing into the future” and (iii) “for actions occurring at an un-specified time” (English Language Syllabus 2010: 88). They also learn that the simple past tense should,

176 Julia Davydova

in contrast, be used for completed actions or events in the past and for regular actions in the past.

The study of the present perfect marking in these contexts allows the analyst to ascertain the extent to which L2 speakers learn to approximate the target norms in an indigenised setting. The major drawback of this approach is that it by and large does not account for instances of overgeneralisations of the present perfect to narrative passages (see also Davydova 2011: 140).

To be sure, one of the main challenges imposed by this approach concerns the method with which tokens are to be included into the analysis. With this in mind, I explain in what follows how the variable context was determined and how the data was extracted. Given limitations of space on the one hand and the complexity of the issue on the other, I will only be able to provide the main rules of thumb here (see Davydova 2011: 119–131 for an exhaustive account).

Because I employed a function-based approach, all the tokens were extracted manually from the corpus data. In order to circumscribe present perfect contexts, the analyst needs a semantic-pragmatic approach that would allow her to differ-entiate between (i) events that are clearly anchored in the past and have no direct relation to speech time and (ii) those that are inherently oriented toward the moment of utterance (see also Winford 1993: 155). In order to identify the former, I employed the method of elimination, that is to say, I excluded all the contexts featuring definite past time adverbials such as long ago, five years ago, once, yes-terday, last month, etc. I furthermore excluded sentences containing subordinate clauses or adverbials specifying the exact time of the past event (e.g. He was never dated when he was a teenager vs. He was never dated as a teenager). Finally, I excluded all narrative passages from the analysis of data as these are inextricably linked to the definite past time reference and are, by this token, the functional domain of the simple past tense.

The contexts of the second type were primarily elicited through time adverbi-als reportedly linked to the category of the English present perfect. These include: for X time, since X time, so far, yet, always, in my whole life, ever, never, just, lately and recently. Finally and perhaps most importantly, contexts not featuring any time adverbial specification were by default included as part of the analysis as temporally underspecified contexts are inherently linked to the category of the present perfect (see, for instance, Van Herk 2008).

The study draws on data obtained from the Indian, Singaporean and East African components of the International Corpus of English (ICE). The IndE and SinE data was obtained from 100 texts for each variety. The texts are documen-tations of L2 spontaneous speech as reflected through direct conversations (S1A-001 to S1A-090) and telephone calls (S1A-091 to S1A-100). The data for East African English (henceforth, EAfE) stems from direct conversations and class-


room discussions (S1A-001K to S1A-030K and S1A-018T to S1A-020T).² All mate-rials studied here represent educated upper-mesolectal English as it is attested in indigenised settings (see also Davydova 2011). The speakers are “adults (18 or over) who have received formal education through the medium of English to the completion of secondary school” (Greenbaum 1996: 6).

I also introduce the data from the London-Lund Corpus (LLC) as a reference point allowing for cross-varietal comparisons. The LLC is a collection of recorded conversations, some of which were surreptitiously recorded. The corpus taps into educated L1 English, the form of English that oftentimes provides a baseline for English learning in a formal L2 setting such as classroom instruction. To illus-trate this point, speakers of IndE, SinE and EAfE orient toward BrE as opposed to American English (henceforth, AmE) as their target variety due to their colonial histories. Ooi (2001: x) claims that “exornomative standards continue to define the study of English in the classrooms” in Singapore. My own fieldwork in the Jawaharlal Nehru University, New Delhi, from 2007 to 2014 suggests that IndE speakers still hold BrE in high regard and consider it a socially prestigious form of language offering opportunities for prosperity and social advancement. “At least theoretically, British English and Received Pronunciation are still upheld as the target forms of language education” in Kenya and Tanzania, reports Schneider (2007: 194). Overall then, BrE continues to influence emerging L2 varieties in the classroom and is thus used as a yardstick of comparison here.

Assuming that L2 learners mainly orient themselves toward written language as their L1 benchmark during the acquisition of tense and aspect features, I decided to sample spoken native English that would also be highly representative of BrE written norms. The LLC corpus suited this purpose very well.

While analysing my data, I make use of the variationist methodology. This means that I explore the overall distribution of variants in present perfect con-texts as a first step of the analysis. In the next step, I construct fixed-effects models of variation using statistical software package Goldvarb 2001 (Robinson, Lawrence, and Tagliamonte 2001).³ These consider the simultaneous workings of different language-internal factors hypothesized to constrain the use of the present perfect.

2 I included classroom discussions as part of analysis because of the limited number of direct conversations.3 See Paolillo (2013) for the most recent discussion of the main advantages offered by fixed-ef-fects models.

178 Julia Davydova

5 ResultsLet us consider results of the distributional analyses first. Table 1 shows distri-bution of the verb forms in present perfect contexts in BrE and contrasts it with those attested for indigenised English.

Table 1: Overall distribution of variants in the present perfect contexts.⁴

BrE (N/%) IndE (N/%) SinE (N/%) EAfE (N/%)

have-perfect 1,812/90 715/54 532/56 247/59preterite 197/10 471/36 350/37 159/38present 4/0 60/5 10/1 4/0be-perfect 2/0 16/1 8/0 2/0past perfect NA 40/3 5/0 4/0bare verb stem NA 9/0 26/3 5/1lone past p⁵ NA 7/0 16/2 1/0lone present p NA 6/0 1/0 NA

Total 2,016 1,324 948 422

The analyses show that the HP is used at the rate 90 % in the standard variety, whereas is it used at the overall rate of 54 % to 59 % in the non-native speaker varieties. Notice also that the only other variant that consistently competes for the expression of the present perfect meaning across the board is the simple past tense. It is used at the overall rate of 10 % in the reference variety and its rates of use are also consistent, although much higher, in all L2 Englishes studied here. To be sure, these findings testify to the general fluidity between the two catego-ries which is perhaps best explained in terms of the overlap in their cognitive meanings.

All other functional equivalents, that is, innovative contact-induced features, are either fairly or extremely rare. These findings clearly demonstrate that the domain of perfect marking does not yield “an abundance of grammatical appara-tus” (Tagliamonte 1996: 351) in educated L2 English. They are, however, in sharp contrast to the type of variation produced in basilectal L2 English. Davydova (2011: 206) reports a highly heterogeneous system of present perfect marking for IndE speakers from rural areas who acquired English unsystematically and over the years, with a minimum of secondary school classroom input. When at univer-

4 The percentages sometimes add up to 99 % or 98 % because of rounding.5 In Table 1 “p” stands for “participle”, e.g. lone past p: You done homework yet (ICE-SIN S1A-086); lone present p: What have you been doing recently? – I doing my research work (HCNVE IE05).


sity, these speakers were all of a sudden presented with the task of making them-selves understood in the lingua franca English and had to learn it from scratch. Table 2 reports results for the use of the present perfect in basilectal IndE.

Table 2: Composition of variants in present perfect contexts in basilectal IndE (from Davydova 2011: 206).⁶

N %

have-perfect 4 5preterite 35 47present 5 6be-perfect 10 13bare verb stem 16 21lone present participle 2 2modal verb + infinitive 1 1

Total 73 100

The latter finding shows that limited instruction and naturalistic contexts of acquisition may lead to the emergence of highly heterogeneous variable systems in indigenised grammar, perhaps best understood in terms of the selection mech-anism. Furthermore, the type of variation described in Table 2 is also attested for discourse-pragmatic features such as quotative markers, features that primarily develop in spontaneous interactions (Davydova 2015: 307). I address this issue in the discussion of findings (Section 5).

How come that the ICE-datasets exhibit lower rates of the HP use in com-parison to the LLC data? The finding can be adduced in favour of the contention that the HP is a complex category, particularly in terms of its semantics, present-ing second-language learners with difficulties. This functional complexity of the English present perfect accounts for the lower rates attested in the present perfect contexts in L2 varieties (see also Davydova 2011: 296–297). Alternatively, L2 speak-ers may have been picking up on the general tendency toward a lower frequency of the present perfect use characterising varieties of English world-wide (see Fuchs; Van Rooy, this volume).

But what is the language-internal system underlying the use of the HP in L2 English varieties? To what extent do the similarities between native and non-na-tive English hold regarding the mechanism that produces the patterns of the HP use? In order to address this issue, I perform logistic regression analyses of

6 This data stems from the Hamburg Corpus of the Non-Native Varieties of English (Davydova 2011).

180 Julia Davydova

the data, in which I include five language-internal factors elucidated in previ-ous research as independent variables.⁷ These factors are (i) semantic-pragmatic context (Winford 1993; Elsness 1997; Tagliamonte 2000), (ii) time adverbial spec-ification (Tagliamonte 2000; Van Herk 2008), (iii) negation or polarity (Elsness 1997; Van Herk 2008), (iv) lexical verb semantics or Aktionsart (Shirai and Ander-sen 1995; Housen 2002; Odlin and Alonso-Vázquez 2006) and (iv) transitivity (Davydova 2008, 2011, 2013).⁸ These language-internal constraints, including their nominal values, are reproduced in Table 3 for convenience.

Table 3: A summary of language-internal constraints claimed to trigger the use of the English HP (from Davydova 2013: 206–207).

Factor group Relevant factor Effect on the HP Examples (bold = affected verb; italics = factor)

Semantic- pragmatic context

resultative favours But they have both left their families here (HCIE⁹ McCance08).

extended-now favours Alex has been in London some time (HCIE BrownW06).

experiential in contrast to resul-tative and extended- now contexts, favours less strongly or disfavours

But there is and always was more said about Australia than ever (HCIE Wyly_02).

recent past disfavours I write to let you know that I received your letter this day (HCIE SprouM09).

Time adverbial specification¹⁰

unspecified favours James Fife has Got Married to Margaret McCafrey two Miles Beyond Enniskellen on the Forencecourt road (HCIE Fife_01).

time adverbial disfavours But I am Grateful that ever there came an opportunity of yous going out to that Country Fathy in particular (HCIE Fife_01).

7 The application value is the HP. The multivariate analyses establish the probability of use of the HP variant against its major competitor, i.e. the simple past tense (see Davydova 2011: 141–142).8 These factors were arrived at by studying previous literature on the present perfect and by my own research documented in Davydova (2011).9 The Hamburg Corpus of Irish English (Pietsch 2007).10 Note that some time adverbials such as yet, now, up until now, etc. exhibit a close association with the HP. Yet these distinctions are not accounted for in this study.


Table 3: (continued).

Factor group Relevant factor Effect on the HP Examples (bold = affected verb; italics = factor)

Polarity Not-negation favours I have not ceased to pray for you night and morning since I bid you farewell (HCIE Fife_02).

No-negation (constituent negation: no, none, nothing, never, etc.)

disfavours We had no rain since last May. (HCIE Dunne_08)

Aktionsart stative disfavours But you never thought that you have twins lah (ICE-SIN S1A-048)

dynamic disfavours You travelled so often (ICE-SIN S1A-040).

accomplishment favours But have you actually run such a distance before? (ICE-SIN S1A-063)

achievement favours What has happened to this guy? (ICE-SIN S1A-087 )

Transitivity transitive favours I mean I’ve studied them you see (ICE-SIN S1A-088)

intransitive favours He has sent for a young woman (HCIE Hammon05).

mutative disfavours Ah your niece came back already leh (ICE-SIN S1A-088).The Cromack family are well but the old woman is much failed, she can scarcely walk (HCIE BrownW07)

Which of these factors will condition the use of the present perfect in the ICE-data-sets and the reference variety? Table 4 reports the results of the four independent multivariate analyses of data.

The findings reported in Table 4 show that two major constraints are oper-ative in the datasets explored here. These are (i) semantic-pragmatic context, which universally underlies the occurrence of the HP in the ICE and LLC data and also, to a large extent, (ii) time adverbial specification (EAfE is the only exception to the pattern). Furthermore, constraint hierarchies within the factor groups are remarkably consistent.

182 Julia Davydova

Tabl

e 4:

Mul

tivar

iate

ana

lyse

s of t

he co

ntrib

utio

n of

lang

uage

-inte

rnal

fact

ors t

o th

e pr

obab

ility

of t

he H

P us

e.¹¹

HP

SinE

(ICE

)In

dE (I

CE)

EAfE

(ICE

)Br

E (L

LC)

Log

likel

ihoo

d–4

78.5

1–7

45.3

7–2

57.9

5–5

64.5

8FW

%N

FW%

NFW

%N

FW%

NIn

put v

alue

.63

60 %

532/

882

.60

60 %

715/

1,18

6.6

060

247/

406

.90

901,

812/

2,00

9

1. S

eman

tic-p

ragm

atic

cont

ext

resu

ltativ

e.6

877

%21

8/28

2.6

976

%33

5/44

0.5

465

%88

/134

.69

96 %

788/

813

exte

nded

-now

.69

83 %

113/

135

.49

60 %

103/

169

.74

82 %

43/5

2.6

395

%39

4/41

1ex

perie

ntia

l.4

050

%12

9/25

5.3

751

%18

2/35

4.5

062

%71

/114

.27

81 %

430/

525

rece

nt p

ast

.25

34 %

72/2

10.3

242

%95

/223

3142

%45

/106

.19

76 %

200/

260

rang

e44

3743

50

2. Tr

ansi

tivity

trans

itive

.47

56 %

307/

539

[.52]

61 %

404/

660

[.52]

60 %

147/

242

[.51]

90 %

1,15

3/1,

277

intra

nsiti

ve.5

064

%12

4/19

1[.4

7]59

%31

1/52

4[.4

6]60

%10

0/16

4[.4

7]90

%65

9/73

2co

pula

be

.81

85 %

67/7

8m

utat

ive

.29

46 %

34/7

3ra

nge

51

11 S

tatis

tical

ly n

on-s

igni

fican

t fac

tor g

roup

s are

in sq

uare

bra

cket

s. G

rey

shad

ing

high

light

s sta

tistic

ally

sign

ifica

nt fa

ctor

gro

ups t

hat i

ndic

ate

com

mon

pa

ttern

s of t

he H

P us

e.


3. N

egat

ion

not-n

egat

ion

.80

88 %

113/

128

[.56]

65 %

91/1

39[.4

8]59

%31

/52

[.54]

89 %

292/

328

affir

mat

ive

.45

57 %

411/

715

[.49]

59 %

624/

1,04

7[.5

0]61

%21

6/35

4[.4

9]90

%1,

520/

1,68

1no

-neg

atio

n.2

020

%8/

39[.5

6]65

%91

/139

[.48]

59 %

31/5

2[.5

4]89

%29

2/32

8ra

nge

60

4. T

ime

adve

rbia

l spe

c.

unsp

ecifi

ed.5

365

%36

7/56

4.5

162

%57

4/91

8[.4

9]60

%19

1/31

7.5

492

%1,

340/

1,44

6tim

e ad

verb

ial

.43

51 %

165/

318

.39

49 %

42/8

5[.5

0]62

%56

/89

.39

83 %

472/

563

rang

e10

1215

5. A

ktio

nsar

t

stat

ive

[.53]

62 %

147/

234

.57

59 %

158/

266

[.51]

65 %

60/9

1[.4

8]87

%57

8/65

9dy

nam

ic[.5

4]61

%19

7/31

9.5

360

%23

2/38

4[.4

4]55

%79

/143

[.53]

91 %

645/

702

acco

mpl

ishm

ent

[.46]

56 %

37/

65.4

460

%32

5/53

6[.5

4]62

%10

8/17

2[.4

8]90

%58

9/64

8ac

hiev

emen

t[.4

2]57

%15

1/26

4ra

nge

13

184 Julia Davydova

The HP is strongly associated with resultative contexts and extended-now con-texts, and least so with experiential contexts and contexts of recent past. The pattern recurs in all varieties studied here. Moreover, the HP is favoured in con-texts that do not feature any time adverbial specification. That the HP should be favoured in resultative and extended-now contexts as opposed to, for instance, experiential contexts is perhaps best explained by the semantics of these con-texts which are arguably more explicitly oriented towards the moment of utter-ance (see Davydova 2011: 301–305 for further discussion).

Indefinite past time reference linked to the category of the English present perfect explains why the have variant is preferred in temporally unspecified con-texts. Indefinite past time reference does not presuppose the occurrence of time adverbials and this explains why the HP is preferred without temporal specifica-tion.

Before proceeding to discuss the major findings of this study, I would like to comment on several variety-specific patterns that emerge from the data. The results for SinE show that both transitivity and negation are operative constraints in this L2 variety. More specifically, the HP is associated with intransitive pred-icates, including copula be, and not-negation in this variety. A similar pattern could be detected in a historical variety of IrE, which served as an input variety in the formation process of SinE (Davydova 2013: 210–211). Relying on these pieces of evidence I argue that the patterns of use underlying the variable realisation of the SinE HP are at least partially related to the IrE superstrate: There is evidence suggesting that the Irish element of the Catholic missionaries was fairly strong in nineteenth-century Singapore. Many of these taught English in Singaporean schools to earn their living. Some primary and secondary schools are still run by the Catholics and are held in high regard by many Singaporeans (see Davydova 2013: 196–198 for further details).

Another notable finding is that verbal lexical aspect, or Aktionsart, is a sig-nificant constraint in IndE, whereas this constraint is not operative in other vari-eties studied here. The studies in second-language acquisition, however, reveal that while acquiring the semantics of the tense and aspect system, L2 learners rely heavily on verbal lexical semantics (Shirai and Andersen 1995; Housen 2002). This pattern has been described as a universal mechanism guiding the acquisi-tion of the L2 grammar. The variable grammar of the IndE present perfect echoes this universal tendency in that the realisation of this tense is still contingent upon the semantics of the main verb, in addition to semantic-pragmatic context, in this variety.


6 DiscussionOne of the main findings of the study is that present perfect marking in educated L2 English lacks in diversity of variants: there are two forms, the HP and the simple past tense, competing for the expression of the present perfect meaning, although the former variant is used at a lower rate than that attested for the refer-ence variety. This is surprising given that we are exploring forms of English that have emerged in multilingual contexts. How can we explain this finding? I argue that the key to understanding this phenomenon is related to the context in which a specific language feature is acquired. Relying on data obtained from basilectal IndE, I have demonstrated that uninstructed L2 acquisition of tense and aspect may result in a robustly variable system. This allows me to hypothesize that a naturalistic context functions as a trigger of linguistic diversity in the nativised setting, whereas a classroom context, in contrast, inhibits heterogeneity within language as L2 learners are under pressure to conform to the norms presented by the target variety. In what follows, I would like to argue for this hypothesis, while contrasting the results obtained for present perfect marking in educated L2 English with those obtained for quotative markers in IndE.

Quotative markers are strategies used by speakers to introduce direct speech, thought as well as non-lexicalised sounds and gestures in order to re-enact dia-logue in narratives (see, for instance, D’Arcy 2012: 347; Davydova and Buchstaller 2015). These are exemplified in (14) through (16).

(14) She said, “I will never see you again.”

(15) And then he goes, “Terrific!”

(16) I thought, “This cannot be happening!”

Relying on data obtained from educated IndE speech, Davydova (2015) shows that this language domain¹² undergoes robust contact-induced change featuring as many as 14 different token types, which can be further subdivided into individual tokens. The token types are illustrated in Table 5. The data was obtained through sociolinguistic interviews conducted at Jawaharlal Nehru University (JNU), New Delhi, in 2007 and 2011.¹³ The discourse genre is sufficiently analogous with the

12 “Language domain” is used here in the sense of “linguistic variable”.13 The first data collection trip was sponsored by the DFG-funded project “Varieties of English in the EFL classroom” headed by Peter Siemund. The second data collection trip was funded by the Landesexzellenzcluster “Linguistic Diversity Management in Urban Areas” (LiMA).

186 Julia Davydova

ICE-data in that, similar to the ICE-data, it taps into dialogic speech (see also D’Arcy 2012: 347 for more on issues regarding corpora comparability).

Language variation in this language domain is in stark contrast with that demonstrated for present perfect contexts in the ICE-data. In Davydova (2015: 307, 327), I argue that this tremendous diversity of linguistic variants emerges as a result of multilingual speakers’ drive for creativity (Matras 2009: 79–86), coupled with the native strategies for introducing quotation as attested in Hindi, the main

Table 5: Quotative marking in IndE (from Davydova 2015: 333–334).

Token type Example

1. verbs of reportingsay So I say, “Datsha, you should go.” (HCNVE IE52)ask And I asked my professor, “Is it the same movie that we have in India in

Hindi?” (HCNVE IE33)tell […] daughter would tell, “I want to be a DJ.” (HCNVE IE03)call […] she called me, “No, no, no, I cannot come. Please, you know I am so

shy, I cannot converse in English.” (HCNVE IE52)speak […] in Hindi we speak, “(unclear) Okay, okay [ind] thik hai [/ind]. It’s

okay. It’s okay.” (HCNVE IE31)discuss […] we were having pizza and we were discussing, “Oh, it’s so expen-

sive! And it can be better so let’s you know try preparing it as?” (HCNVE IE52)

2. zero quotative I notice earrings Ø, “Oh, she has good, she is having good earrings. I should get them.” (HCNVE IE51)

3. be like She is like, “That’s the pronunciation that I want”. (HCNVE IE35)

4. okay (fine) […] they do not make it a point that okay (,) fine,”‘This is American English. So we are teaching you. This is British English. We are teaching you.” (HCNVE IE35)

5. verbs of mental activity and perceptionfeel […] on Monday you feel, “Five more days to go!” (HCNVE IE03)hear It’s not like you heard, “Okay, now we are together, so we’ll start chat-

ting.” (HCNVE IE36)see […] he will see, “I have checked that.” (HCNVE IE37)think […] they are think, “We are fighting for our Islam.” (HCNVE IE14)consider They consider it, “Okay, Hindi is a language. It’s my mother tongue.

[…]” (HCNVE IE31)

6. verb + that So my father always dreamt that, “My children should read a lot, they should be really educated so that they do not face the problems that we have faced in our life.” (HCNVE’: IE52)

7. copula verb So for them it’ll be, “[ind] Accha [/ind] then!” (HCNVE IE35)


L1 language of the region. Yet under what circumstances is speakers’ creativity licensed in a multilingual setting? This seems to be happening in contexts where language features are put to use spontaneously through regular linguistic prac-tices within the community. In other words, this seems to be happening in con-texts when the pressure to conform to the target-oriented norm is off. Similar to other discourse-pragmatic features (Sankoff et al. 1997), quotative markers evolve through such communicative practices. In contrast to core morphosyntactic fea-tures, quotative markers, especially vernacular forms such as be like, go, be all, etc., do not receive much focused attention in the L2 classroom, if they receive it at all (Pramod Pandey, personal communication, 31 March 2015; Jacob Leim-gruber, personal communication, 03 April 2015; Josef Schmied, personal com-

Table 5: (continued).

Token type Example

8. dynamic verbs and verbs of motiongo Every time I would speak in English, my parents would go, “Oh, yeah!”

(HCNVE IE13)come And teacher comes with a stick, “No, you have to speak in English.”

(HCNVE IE52)struggle […] was he was still struggling with “What am I supposed to do with my

life?” (HCNVE IE35)

9. verbs of achievementrefute […] people refute, “No, it doesn’t work!” (HCNVE IE31)recommend […] if my friend recommend, “It’s a nice movie. Must watch it.” (HCNVE

IE54)specify Not only engineer, they would specify, “We want you to be electrical

engineer, mechanical engineer. […]” (HCNVE IE03)claim Delhi doesn’t have any population, which can claim, “We are originally

from Delhi!” (,) you know (,) no (.) (HCNVE IE03)

10. (that +) discourse marker

[…] or you know, “You should go down, take a walk, do things like that”. (HCNVE IE04)

11. ki (Hindi for ‘that’) You (,) it is supposedly (,) said [ind] ki [/ind], “You’ll give, you’ll get best coachings in Delhi only”. (HCNVE IE51)

12. prep./noun phrase + that

So she was very particular about it that, “My kids should go into a public school.” (,) which is English-medium strictly (.) (HCNVE IE35)

13. verb + noun + that […] they make it a point that, “You should pronounce it as a British English ehm, thing”. (HCNVE IE35)

14. (be) all Because in academics when I speak the kind of English (,) they Ø all, “Oh! Very serious. Okay, he is very serious.” (HCNVE IE31)

188 Julia Davydova

munication, 03 April 2015). These are the kinds of features that speakers tend to pick up naturally as they learn to interact in English. This is in line with the assumptions concerning the acquisition of these features by English as a Foreign Language (EFL) learners (Davydova and Buchstaller 2015).

The emergence of nativised quotative markers is in clear contrast to the genesis of the present perfect as well as other tense and aspect features, all of which receive long-term continuing support in the classroom throughout the schooling years. To illustrate this point, the English syllabus provided by the Singaporean Ministry of Education requires that the present perfect as well as other tense and aspect features be introduced and taught at the upper primary level (classes 4 to 6) and then revisited throughout secondary education and preferably, “reinforced and taught at increasing levels of difficulty, until pupils have mastery of it” (English Language Syllabus 2010: 88). The Kenyan secondary school curriculum requires that “perfective and progressive aspect” is introduced and explicitly taught in Form III.¹⁴ The present perfect is part of the teaching mate-rials covered in the New Oxford English Course textbooks in Tanzania.¹⁵ My eth-nographic work in the community of Jawaharlal Nehru University suggests that the English present perfect is repeatedly addressed in classroom-based language environments in India. These observations allow the conclusion that the category of the present perfect is a product of the classroom input, while quotative markers by and large evolve in naturalistic settings. The fundamental differences in con-texts that guide the evolution of quotative and present perfect marking in edu-cated L2 English account for diverging results – a robust pool of variants attested in one domain and its lack in the other.

The relative prominence given to the English HP in the L2 classroom may fur-thermore account for the fact that the constraint hierarchies regarding seman-tic-pragmatic context and time adverbial specification are remarkably consis-tent across the board and parallel those attested for the reference variety. What this effectively means is that the semantics of the present perfect is essentially acquired by the L2 speakers.¹⁶

Such an exact replication of constraints underlying the variable use of a lin-guistic feature in a contact variety is by no means a given (see Schleef, Meyer-hoff, and Clark 2011; Meyerhoff and Schleef 2012). Research into the acquisition

14 http://www.elimu.net/Secondary/Kenya/KCSE_Student/English/Form2/Grammar/Parts%20of%20Speech/Parts_of_Speech.htm (accessed 3 April 2015).15 http://www.kiliproject.org/newsletters-and-documents/Literacy-English-Primary-Sylla-bus-final.pdf (accessed 3 April 2015).16 Time adverbial specification arguably contributes to the semantics of the tense and aspect categories in general and to the semantics of the English present perfect in particular.


of L2 variation suggests that while borrowing the pattern from the donor variety, speakers tend to reorder the constraints in some way, the phenomenon called “transformation under transfer” (Meyerhoff 2009: 313). Another finding is that some patterns do not re-emerge in the contact vernacular at all. This is particu-larly true of semantic-pragmatic distinctions (such as, for instance, the content of the quote, i.e. reproduction of speech vs. reproduction of internal dialogue) within the variable grammar. The reconstruction of such constraints arguably presents non-native speakers with difficulties (Davydova 2015: 329–331).

And yet, the independent factors “semantic-pragmatic contexts” and “time adverbial specification” are operative patterns in the indigenised grammar of New Englishes. What is more, they are aligned with the constraints reported for the reference variety. This finding calls for an explanation which seems to be grounded in the fact that the feature under study is fostered through prolonged and repeated instruction in the countries of South and Southeast Asia as well as in East Africa. Instructed input apparently ensures a successful acquisition of the variable realisation of the present perfect regarding its semantics.

What do these findings reveal about the mechanism of diffusion and selec-tion? Diffusion is the process describing a relatively seamless transfer of patterns underlying the use of a given linguistic category. What ensures such a “regular, internal and largely stable transmission of linguistic features” (Schneider 2000: 205) and their patterns of use? Here, I argue that the structural properties of language that receive much attention in a classroom context seem to get trans-ferred from the parent variety to daughter varieties with minimal readjustments, manifested (i) through lower rates of use of the linguistic variants under study, and (ii) emergence of novel constraints not attested in the reference variety. The novel constraints can be related to (i) other superstratal inputs, and/or (ii) uni-versal processes underlying SLA.¹⁷ The former seem to have played a role in the emergence of the patterns related to transitivity and negation in SinE (see Davy-dova 2013 for a full discussion). The latter might account for the fact that verb semantics is a significant constraint underlying the use of the HP in IndE (see Davydova 2011: 298–299). It is, however, remarkable that those properties of vari-able linguistic structure that get learners’ focused attention through instructed input, semantics of the English present perfect being a case in point, eventually get replicated in the minute detail in the resulting variable grammar. It is then not unlikely that, while offering the learner an opportunity to scrutinize a specific lin-guistic item and its patterns of occurrence, the classroom setting might indeed be the driving engine ensuring the proper functioning of the diffusion mechanism

17 L1 transfer, another important variable underlying the formation of New Englishes, is not accounted for in this study.

190 Julia Davydova

in New Englishes. In contrast, those linguistic domains that thrive through con-stant linguistic practices in naturalistic settings are subject to further negotiation and re-analysis, resulting in a vigorous competition of functional equivalents. Spontaneous interactions then seem to lie at the heart of the selection process in Schneider’s (2000) model.

7 ConclusionNon-native Englishes and a reference variety reported on here attest quite remark-able commonalities in patterns of use of the HP. These commonalities are argu-ably best explained in terms of the diffusion process, which predicts that English varieties must share some core despite their obvious idiosyncratic forms and pat-terns. I hope to have demonstrated that under some circumstances, speakers of indigenised Englishes seem to follow the path trodden by the parent varieties, yet otherwise they can be increasingly creative. More specifically, I have argued that when a linguistic variable evolves through informal interactions it is more likely to yield a greater repertoire of linguistic variants or innovations. The category of the English present perfect is an area of grammar that is given considerable attention in the L2 classroom. This leads to a fairly uniform composition of the linguistic variants constituting the category and also ensures successful acquisi-tion of the semantics of the HP.

AcknowledgementsI would like to express my gratitude to the two anonymous reviewers for their valuable comments on the earlier version of this chapter. The general quality of this contribution significantly improved as a result of their very thoughtful feed-back. All remaining shortcomings are entirely my own.


ReferencesBhatt, Rakesh M. 2004. Indian English: Syntax. In Bernd Kortmann, Kate Burridge, Rajend

Mesthrie, Edgar W. Schneider & Clive Upton (eds.), A handbook of varieties of English. Vol. 2: Morphology and syntax, 1016–1030. Berlin: Mouton de Gruyter.

Bao, Zhiming. 1995. Already in Singapore English. World Englishes 14(2). 181–188.Bao, Zhiming. 2005. The aspectual system of Singapore English and the systemic substrate

explanation. Journal of Linguistics 41. 237–267.Brinton, Laurel J. 1988. The development of English aspectual systems. Cambridge: Cambridge

University Press.Collins, Peter. 2015. Introduction. In Peter Collins (ed.), Grammatical change in English

world-wide, 1–11. Amsterdam: Benjamins.Comrie, Bernard. 1985. Tense. Cambridge: Cambridge University Press.D’Arcy, Alexandra. 2012. The diachrony of quotation: Evidence from New Zealand English.

Language Variation and Change 24(3). 343–371.Davydova, Julia. 2008. Preterite and present perfect in Irish English: Determinants of variation.

Hamburg: University of Hamburg.Davydova, Julia. 2011. The present perfect in non-native Englishes: A corpus-based study of

variation. Berlin: Mouton de Gruyter. Davydova, Julia. 2012. Englishes in the Outer and Expanding Circles: A comparative study.

World Englishes 31(3). 366–385.Davydova, Julia. 2013. Detecting historical continuity in a linguistically diverse area: The

present perfect in modern Singapore English. In Joana Duarte & Ingrid Gogolin (eds.), Linguistic super-diversity in urban areas: Research approaches, 193–225. Amsterdam: Benjamins.

Davydova, Julia 2015. Linguistic change in a multilingual setting: A case study of quotatives in Indian English. In Peter Collins (ed.), Grammatical change in English world-wide, 297–334. Amsterdam: Benjamins.

Davydova, Julia & Isabelle Buchstaller. 2015. Expanding the circle to Learner English: Investigating quotative marking in a German student community. American Speech 90(4). 441–478.

Denison, David. 1993. English historical syntax: Verbal constructions. Harlow: Longman.Elsness, Johan. 1997. The perfect and preterite in contemporary and earlier English. Berlin:

Mouton de Gruyter.Elsness, Johan. 2009. The present perfect and preterite. In Günter Rohdenburg & Julia Schlüter

(eds.), One language, two grammars? Differences between British and American English, 228–245. Cambridge: Cambridge University Press.

English Language Syllabus. 2010. Primary & Secondary (Express/Normal [Academic]). http://www.moe.gov.sg/education/syllabuses/english-language-and-literature/files/english-primary-secondary-express-normal-academic.pdf (accessed 8 August 2015).

Filppula, Markku. 1997. The influence of Irish on perfect marking in Hiberno-English: The case of the “extended-now” perfect. In Jeffrey L. Kallen (ed.), Focus on Ireland, 51–71. Amsterdam: Benjamins.

Filppula, Markku. 2008. Irish English: Morphology and syntax. In Bernd Kortmann & Clive Upton (eds.), Varieties of English. Vol. 1: The British Isles, 328–359. Berlin: Mouton de Gruyter.

192 Julia Davydova

Greenbaum, Sidney. 1996. Introducing ICE. In Sidney Greenbaum (ed.), Comparing English worldwide: The International Corpus of English, 3–12. Oxford: Clarendon Press.

Harris, John. 1984. Syntactic variation and dialect divergence. Journal of Linguistics 20. 303–327.

Holmes, Janet. 2013. An introduction to sociolinguistics. London: Routledge.Hosali, Priya. 2008. Butler English: Morphology and syntax. In Rajend Mesthrie (ed.), Varieties

of English. Vol. 4: Africa, South and Southeast Asia, 563–578. Berlin: Mouton de Gruyter.Housen, Alex. 2002. The development of tense-aspect in English as second language and the

variable influence of inherent aspect. In Rafael Salaberry & Yasuhiro Shirai (eds.), The L2 acquisition of tense-aspect morphology, 155–197. Amsterdam: Benjamins.

Huddlesston, Rodney & Geoffrey K. Pullum. 2005. A student’s introduction to English grammar. Cambridge: Cambridge University Press.

Hundt, Marianne & Nicolas Smith. 2009. The present perfect in British and American English: Has there been any change, recently? ICAME Journal 33. 45–63.

Jespersen, Otto H. 1924. The philosophy of grammar. London: Allen.Kortmann, Bernd & Benedikt Szmrecsanyi. 2004. Global synopsis: Morphological and syntactic

variation in English. In Bernd Kortmann, Edgar W. Schneider, Clive Upton, Kate Burridge & Rajend Mesthrie (eds.), A handbook of varieties of English. Vol. 2: Morphology and syntax, 1142–1202. Berlin: Mouton de Gruyter.

Kortmann, Bernd, Kate Burridge, Rajend Mesthrie, Edgar W. Schneider & Clive Upton (eds.) 2004. A handbook of varieties of English. Vol. 2: Morphology and syntax. Berlin: Mouton de Gruyter.

Kortmann, Bernd & Kerstin Lunkenheimer (eds.). 2013. The electronic world atlas of varieties of English. Leipzig: Max Planck Institute for Evolutionary Anthropology. http://ewave-atlas.org (accessed 30 March 2015).

Matras, Yaron. 2009. Language contact. Cambridge: Cambridge University Press.Mesthrie, Rajend. 1992. English in language shift: The history, structure and sociolinguistics of

South African Indian English. Cambridge: Cambridge University Press.Mesthrie, Rajend & Rakesh M. Bhatt. 2008. World Englishes: The study of new linguistic

varieties. Cambridge: Cambridge University Press.Meyerhoff, Miriam. 2009. Replication, transfer, and calquing: Using variation as a tool in the

study of language contact. Language Variation and Change 21(3). 297–317.Meyerhoff, Miriam & Erik Schleef. 2012. Variation, contact and social indexicality in the

acquisition of (ing) by teenage migrants. Journal of Sociolinguistics 16(3). 398–416.Mufwene, Salikoko. 2001. The ecology of language evolution. Cambridge: Cambridge University

Press.Odlin, Terence & Cristina Alonso-Vázquez. 2006. Meanings in search of the perfect form: A look

at interlanguage verb phrases. Rivista di Psicolinguistica Applicata 6. 53–63.Ooi, Vincent B. Y. 2001. Introduction. In Vincent B. Y. Ooi (ed.), Evolving identities: The English

language in Singapore and Malaysia, ix–xvii. Singapore: Times Academic Press.Paolillo, John C. 2013. Individual effects in variation analysis: Model, software and research

design. Language Variation and Change 25(1). 89–118.Pietsch, Lukas. 2005. Re-inventing the ‘perfect’ wheel: Grammaticalisation and the Hiberno-

English medial object perfects. Hamburg: University of Hamburg.Pietsch, Lukas. 2007. The Irish English ‘after perfect’ in context: Borrowing and syntactic

productivity. Hamburg: University of Hamburg.


Pietsch, Lukas. 2009. Hiberno-Irish medial-object perfects reconsidered: A case of contact-induced grammaticalisation. Studies in Language 33(3). 528–568.

Robinson, John, Helen Lawrence & Sali A. Tagliamonte. 2001. Goldvarb 2001: A multivariate analysis application for Windows. User’s manual. http://www.romanistik.uni-freiburg.de/pusch/Download/variacionismo/Golvarb2001_User_manual.pdf (accessed 25 June 2015).

Sankoff, Gillian, Pierrette Thibault, Naomi Nagy, Hélène Blondeau, Marie-Odile Fonollosa & Lucie Gagnon. 1997. Variation in the use of discourse markers in a language contact situation. Language Variation and Change 9(2). 191–217.

Schleef, Erik, Miriam Meyerhoff & Lynn Clark. 2011. Teenagers’ acquisition of variation: A comparison of locally-born and migrant teens’ realization of English (ing) in Edinburg and London. English World-Wide 32(2). 206–236.

Schneider, Edgar W. 2000. Feature diffusion vs. contact effects in the evolution of New Englishes: A typological case study of negation patterns. English World-Wide 21(2). 201–230.


Seoane, Elena & Cristina Suárez-Gómez. 2013. The expression of the perfect in East and South-East Asian Englishes. English World-Wide 34(1). 1–25.

Sharma, Devyani. 2005. Language transfer and discourse universals in Indian English article use. Studies in Second Language Acquisition 27(4). 535–566.

Shirai, Yasuhiro & Roger W. Andersen. 1995. The acquisition of tense-aspect morphology: A prototype account. Language 71(4). 743–762.


Siemund, Peter & Julia Davydova. 2014. World Englishes and the study of typology and universals. In Markku Filppula, Juhani Klemola & Devyani Sharma (eds.), The Oxford handbook of World Englishes. Oxford: Oxford University Press. http://www.oxfordhandbooks.com/view/10.1093/oxfordhb/9780199777716.001.0001/oxfordhb-9780199777716-e-022 (accessed 25 June 2015).


Tagliamonte, Sali. 1996. Has it ever been ‘perfect’? Uncovering the grammar of early Black English. York Papers in Linguistics 17. 351–396.

Tagliamonte, Sali. 1997. Obsolescence in the English perfect? Evidence from Samaná English. American Speech 72(1). 33–68.

Tagliamonte, Sali. 2000. The grammaticalization of the present perfect in English. In Olga Fischer, Anette Rosenbach & Dieter Stein (eds.), Pathways of change: Grammaticalization in English, 329–354. Amsterdam: Benjamins.

Van Herk, Gerard. 2008. Letter perfect: The present perfect in early African American correspondence. English World-Wide 29(1). 45–69.

Wee, Lionel. 2004. Singapore English: Morphology and syntax. In Bernd Kortmann, Kate Burridge, Rajend Mesthrie, Edgar W. Schneider & Clive Upton (eds.), A handbook of varieties of English. Vol. 2: Morphology and syntax, 1058–1073. Berlin: Mouton de Gruyter.

194 Julia Davydova


Werner, Valentin. 2016. Overlap and divergence – aspects of the present perfect in World Englishes. In Elena Seoane & Cristina Suárez-Gómez (eds.), World Englishes: New theoretical and methodological considerations, 113–142. Amsterdam: Benjamins.

Williams, Jessica. 1987. Non-native varieties of English: A special case of language acquisition. English World-Wide 8(2). 161–199.

Winford, Donald. 1993. Variability in the use of perfect have in Trinidadian English: A problem of categorical and semantic mismatch. Language Variation and Change 5(2). 141–188.


Zandvoort, Reinard Willem. 1932. On the perfect of experience. English Studies 14(1). 11–20.

Elena Seoane7 The perfect space in creole-related

varieties of English: The case of Jamaican English

Abstract: Previous work on six different non-native varieties of English has shown that Jamaican English is the only variety in which the simple past outnumbers the have + past participle periphrasis to express perfect meaning. Also remarkable in Jamaican English is the relatively high incidence of isolated participles and bare forms in this function. Interestingly, it is the only creole-based variety in the group analysed, which raises the question of whether it is contact with Jamai-can Creole that triggers differences from other varieties. Through the study of ten high-frequency verbs in perfect environments in the Jamaican component of the International Corpus of English (ICE), this paper aims to confirm these tendencies and investigate the potential determinants of the use and distribution of the vari-ants found. As expected, the data point towards the influence of Jamaican Creole together with cognitive effects derived from language contact situations as the most significant factors. The syntax-lexicon interface also seems to play a part, since variants depend strongly on the lexical verb in question.

1 IntroductionPrevious studies have compared the expression of perfect meaning in a number of non-native varieties of English as represented in the International Corpus of English (ICE), in particular the varieties of East Africa (Kenya and Tanzania), Hong Kong, Singapore, India, Jamaica, and the Philippines (see Seoane and Suárez-Gó-mez 2012; 2013a). They examine all occurrences of ten high-frequency verbs in the spoken component of the corpora in order to identify perfect meaning envi-ronments, that is, the contexts where standard varieties would be most likely to select a have + past participle periphrasis.¹ Findings show that Jamaican English (JamE) differs from all other postcolonial varieties of English in a number of

1 This methodology precludes a strict comparison between the findings here and those of other studies, such as Werner (2014: 235–251), in which all and only instances of have + past participle are analysed.

Elena Seoane, University of Vigo

196 Elena Seoane

respects: in JamE, the simple past outnumbers the have + past participle periph-rasis to express perfect meaning; the isolated participle (she gone), the bare form (she go) and be-perfects (she is gone) also occur relatively more frequently to express perfect meaning, leaving the have-periphrasis with a frequency of only 35 % of the total. It is perhaps significant that, in the group analysed, JamE is the only variety that could be categorized as “English as a second dialect” (ESD; Görlach 1991: 12), that is, a variety which coexists with a creole language (Jamai-can Creole) whose lexicon is predominantly derived from English but which is grammatically different. Figure 1 below shows the use (in normalized frequencies per 10,000 words) of have + past participle (‘perfect’) compared to other forms (‘non-perfect’) to express perfect meaning. This study, based on data from ICE, includes all texts from the written component of the corpus (200 texts; 400,000 words) but only private dialogues (face-to-face conversations and phone calls, S1A texts) from the spoken component (100 texts; 200,000 words). We selected private dialogues as the kind of spoken language where the greatest deal of vari-ation is normally registered and thus as the most likely locus of change (Miller 2006: 689; see also Van Rooy, this volume).

0

10

20

30

40

50

60

70

80

90

BrE HKE IndE PhilE SingE EAfE JamE

Perfect

Non-perfect

Figure 1: Frequencies per 10,000 words of ‘perfect’ vs. ‘non-perfect’ forms to express perfect meaning in seven varieties of English (adapted from Suárez-Gómez and Seoane 2013b).

The perfect space in creole-related varieties of English 197

As can be seen in Figure 1, when written and private spoken texts are examined in British English and a number of postcolonial Englishes, JamE is the only one in which the non-perfect forms outnumber the have + past participle periphrasis to express perfect meaning.

In the current paper, I expand the corpus used to cover the whole of the Jamai-can component of ICE. Through this analysis I intend to (i) seek confirmation for the abovementioned data, especially the predominance of the simple past over the have + past participle periphrasis and the high frequency of alternative forms (participles, bare forms and be-perfects) in this function; (ii) compare spoken and written JamE, to assess the impact of mode in the expression of perfect meaning and also to gauge the degree of consolidation of the innovative forms in this variety of English; (iii) explore the patterns of distribution of the different variants according to the subtypes of perfect meaning (resultative, experiential, recent past and persistent situation) in order to identify potential specializations of meaning of particular variants; (iv) consider the role of adverbial support in the frequency and distribution of variants; (v) analyse the syntax-lexicon interface in the variation between the forms; and (vi) investigate the potential determinants of the use and distribution of the variants found.

For the last of these, I will be looking primarily at the fact that JamE is a variety of English in contact with (and therefore possibly influenced by) Jamaican Creole (JamC). Other issues examined are, firstly, the possible effects of cogni-tive constraints characteristic of language contact situations, which often lead to morphological simplification and more explicit isomorphism; and secondly, the implications of the fact that the input language is an earlier spoken variety of English and dates from a period when grammaticalization of the functional dis-tinction between the simple past and the have + past participle periphrasis had not taken place. Examination of these factors should yield a clear picture of the determinants of the expression of perfect meaning in creole-related JamE.

The paper is organized as follows. Section 2 provides a brief outline of the his-torical and linguistic background in Jamaica. Section 3 offers a description of the corpus and the methodology used and discusses the results of the study. Finally, Section 4 presents the conclusions of the study.

2 BackgroundEnglish has been the official language of Jamaica since it was colonized in the seventeenth century, and Jamaica remained a British colony until it regained independence in 1962. As a result, British English is the most obvious superstrate

198 Elena Seoane

language, confined to formal and official domains and imposed through the edu-cation system. Nevertheless, the greater exposure to American English (geograph-ically, culturally, socially and economically) also influences the language, espe-cially at the level of lexis and spelling, and, to a lesser extent, grammar (Hickey 2004: 14). As Mair and Sand (1998: 189) point out, “[t]he emergence of a local norm of educated English usage […] cannot merely be described in terms of an ‘indigenization’ of a colonial norm, because American English with its increasing prestige represents a powerful third player.” As a consequence, the local acrolect of the West Indies pertaining to the Commonwealth, just as that of the US Virgin Islands, is normally said to be oriented towards American English. Hackert and Deuber (2015), on the other hand, show that Caribbean varieties persistently seem more British- than American-oriented at the level of morphosyntax.

Hence, although British English is still taken as the reference standard variety, it is widely agreed that the local standard language is turning progres-sively away from this (Christie 1989: 247; Mair and Sand 1998: 188–189; Sand 1999: 70–71), and an emerging Jamaican standard variety is being recognized (cf. Devonish and Thomas 2012). Like all other varieties, standard JamE is character-ized by linguistic creativity at the levels of lexis (Sand 1999: 77–110, 2011: 165–172), phonology (Irvine 2004) and morphosyntax (Mair 1992: 11–24; Sand 1999: 111–150, 2011: 172–179, 2012). Following Schneider’s approach, JamE has already evolved to phase 4 (endonormative stabilization) and seems to be moving towards phase 5 (differentiation), as the presence of some divergent varieties is attested (Schnei-der 2007: 238).

Unlike JamE, which is spoken only by a minority for ‘high’ functions, JamC (or Patois), represents the vernacular of the majority of the population. JamC has arisen out of contact between speakers of English and several West African lan-guages, brought in the seventeenth century by the slaves imported to work on the sugar and coffee plantations. It exists largely at the spoken level, but in recent decades it has spread to semi-formal and formal contexts, such as its use by pol-iticians as a means of connecting with people, and its presence in the media. At the written level it has also started to appear in newspapers and in literary works (Sand 1999: 73; Schneider 2007: 235–237; Deuber 2014: 33–34).

In fact, the functional changes reflected in this increasing expansion of the Creole to domains formerly confined to standard English has led, in the current century, to educators, political activists and artists fighting for the recognition of JamC as the official language of Jamaica. Similarly, there are ongoing lan-guage-planning efforts to achieve partially bilingual education (Bilingual Educa-tion Project of the Jamaican Language Unit, as outlined in Carpenter and Devonish 2010).


This complex linguistic context in Jamaica has been described, since DeCamp (1971), as a situation of creole continuum, and this view has been defended more recently by Sand (1999, 2011, 2012), Patrick (1999, 2004) and Deuber (2014), among others. Following the creole continuum approach, rather than considering the varieties as independent and discrete languages, they are placed on a continuum which distinguishes basilectal speakers of JamC at one end from acrolectal speak-ers of standard JamE at the other, and a wide array of intermediate varieties. In DeCamp’s (1971: 350) words, “[m]any Jamaicans persist in the myth that there are only two varieties: the patois and the standard. But one speaker’s attempt at the broad patois may be closer to the standard end of the spectrum than is another speaker’s attempt at the standard.” The difficulty of distinguishing between the two dominant varieties (standard JamE and JamC) reinforces the idea of the continuum, to the extent that “it will be difficult to decide sometimes, whether a certain feature is due to the nativization of English in Jamaica or a result of minimal code-switching” (Sand 1999: 19). As with all varieties of a given language, the ones on the Jamaican continuum have different sociolinguistic connotations and functions, and “[e]ach Jamaican speaker commands a span of this contin-uum, the breadth of the span depending on the breadth of his social contacts” (DeCamp 1971: 350). As for the social significance of the varieties at both ends of the continuum, JamE represents the formal, educated variety. JamC, however, is also prestigious in that it has become a symbol of national identity and indepen-dence and an indicator of peer group solidarity, at times even an expression of opposition to the dominant social and political order (Sand 1999: 60).

In order to detect potential influence of JamC on the expression of perfect meaning in JamE, we need to bear in mind that for the expression of temporal ref-erence, JamC relies on the combination of preverbal particles and bare verb forms (Patrick 1999: Chapter 6, 2004: 411; Schneider 2011: 105). However, preverbal tense markers tend to be elided if there is an overt adverbial, and thus redundant elements are avoided (Patrick 2004: 414). For the expression of past time there are no distinct forms for the simple past and the present perfect, since JamC lacks the primary auxiliaries be, have and do: “There is no distinction between simple past and present perfect verb forms in JamC (e.g. iit ‘eat, eaten’), and neither of them requires an auxiliary or pre-verbal marker” (Patrick 2004: 416). In fact, Sand (2011: 177) observes that in order to express present perfect meaning JamE, like other World Englishes, often uses the present tense in combination with an adverbial marker such as since, which makes the time frame clear (see further Section 3.6), as in (1).

(1) which is being widely-used since 1987.(example from Sand 1999: 118)

200 Elena Seoane

Following Sand (1999: 118–120; see also Mair 1992: 85 and eWAVE feature 99 from Kortmann and Lunkenheimer 2013), simple tenses, such as the present or past tense, are preferred to the have + past participle periphrasis in JamE, since “the correct [i.e. according to the British standard, ES] use of the present perfective is one of the last features to be acquired in the decreolization process” (Sand 1999: 119), and is therefore characteristic of acrolectal speakers exclusively.

Both the lack of primary auxiliaries and the avoidance of have + past partici-ple periphrases would lead to a frequent usage of synthetic forms for the expres-sion of perfect meaning (i.e. the simple past). By contrast, Sand’s findings from the analysis of emerging Jamaican English in journalistic and radio texts also show the reverse phenomenon: a high frequency of have auxiliary derived from a process of hypercorrection for reasons of prestige (see eWAVE feature 100), as in (2).

(2) The police has detained […] the man was released.(example from Sand 1999: 118)

The presence of these hypercorrect forms might balance out the number of syn-thetic forms (simple pasts) and forms without an auxiliary (isolated participles and bare forms) which are to be expected from the influence of the substrate.

Another relevant feature for the analysis of perfect meaning in JamE is the final consonant cluster reduction characteristic of JamC: Basilectal JamC “does […] [not] allow the final consonant clusters of the type /-Ct/ or /-Cd/ that are produced in standard English when -ed is added to verb stems ending in con-sonant other than /t/ or /d/” (Sand 1999: 120; see also Mair 1992: 86). This could have an influence on the frequency of bare forms in JamE.

3 Methodology and results

3.1 Corpus and methodology

This paper investigates the expression of perfect meaning in the Jamaican compo-nent of ICE (ICE-JA), which contains one million words, 60 % of which are spoken and 40 % written (ice-corpora.net/ice/design.htm). Unlike previous studies, which looked only at written texts, the data will also include spoken texts: public dialogues, unscripted monologues, and scripted monologues (200 texts; 400,000 words).


In order to identify perfect environments, data were extracted and analysed on all occurrences of ten high-frequency verbs, excluding be, do and have (i.e. come, finish, get, give, go, hear, see, say, tell, and think). This selection emanates from the list of verbs with perfect meaning provided in Suárez-Gómez and Seoane (2013a). In this paper, all instances of just, ever, never and yet were retrieved from various ICE components. These instances were then analysed in context to identify whether or not they expressed perfect meaning. The above-mentioned verbs were the most frequent with perfect meaning in the ICE corpora.

In this paper, I analyse all the forms of these verbs by reading the context in which they appear. Examples of perfect meaning were identified by examining the time frame in which the action takes place, which is at times marked by an adverbial and some other times from the general linguistic context, as illustrated in the examples below (see sections 3.2 to 3.5). The classification of examples into the different semantic categories (see Section 3.5) also relied heavily on the linguistic context. However, the semantic categorization was not always straight-forward and some borderline cases were found. In such cases, expert colleagues were asked to code the examples separately and then we compared results and discussed differences in codification.² Even so, I am aware that some examples lend themselves to several interpretations and the instances included here show just that; in the most ambiguous cases, examples are embedded into a large context to facilitate temporal-aspectual interpretations.

The concordance program AntConc (Anthony 2014) retrieved 18,666 verbal forms, which were filtered manually to select relevant examples, that is, exam-ples which according to the contexts express perfect meaning. Only 1,077 of these were found to express perfect meaning, and are distributed as shown in Table 1.

Table 1: Number of examples expressing perfect meaning in ICE-JA (number in brackets indi-cates size of relevant corpus section).

Spoken [600,000 words] 854 Written [400,000 words] 223

Total 1,077

2 I am deeply grateful to Cristina Suárez-Gómez for her help with the analysis of many difficult examples.

202 Elena Seoane

3.2 Results: distribution of variants

Table 2 and examples (3) to (14) illustrate the variants expressing perfect meaning found in ICE-JA.

Table 2: Distribution of forms expressing perfect meaning in ICE-JA (reference to examples in brackets).

have + past participle (3) 589 (54.7 %)Simple past (4) 302 (28.0 %)Bare forms (bare form (5); present/participle/bare form (6), present/bare form (7))

69 (6.4 %)

Participle (8) 62 (5.8 %)be-periphrases (+ participle (9), + ’s participle (10)) 50 (4.7 %)Other (present (11), have + bare form (12), have + preterite (13), ain’t + participle (14))

5 (0.5 %)

Total 1,077 (100.0 %)

Table 2 shows that, with the inclusion of all the spoken registers in ICE-JA, the canonical form have + past participle is the most frequent form to express perfect meaning, but it is used on less than 55 % of all occasions. In Section 3.4 I examine differences according to text type, which should help to clarify why including spoken registers sees an increase in the proportion of have + past participle examples. As expected (see Kortmann and Szmrecsanyi 2004; Elsness 2009; Hundt and Smith 2009; Van Rooy 2009) the second most frequent form is the simple past, and in Section 3.5 I analyse the distribution of simple past forms by subtype of perfect meaning, since it has been shown that the preference for the simple past is normally strong for the expression of recent past but not for other very frequent perfect meanings (Seoane and Suárez-Gómez 2013a; Suárez-Gó-mez and Seoane 2013a). More interesting in comparison with other varieties of English, both native and non-native, is the high incidence of alternative forms, such as the bare form (6.4 %), the participle (5.8 %) and be-periphrases (4.7 %).

(3) Politics has been my life. I have given up an academic career to go into poli-tics, so the next step in terms of completing that career and helping the party that I believe in and love, is to contest a seat.(ICE-JA W2C-010)


(4) C: YeahA: And then we had to give we had to give to the uhm to the poor. Father would

come every year and say <&> imitation </&> Oh Holy Childhood gave more than you, George’s beat you this year, Immaculate gave more money than you <&> end-of-imitation </&>

(ICE-JA S1A-013)

(5) “I see nothing up to now which would cause me to entertain any doubt as to the sovereignty of this country being eroded or interfered with by any foreign government,” Mr. Andrade said. (ICE-JA W2C-019)

(6) B: I’m looking for a card for Robert birthdayA: For whoB: Robert birthday A: Robert Oh How old is he Hm Robert How old he gonna beB: Robert Right Huh Nineteen A: He goes oh He’s still youngB: Yeah I want find anything you know really nice and thoughtful Actually I

came in here to buy a phone cardA: Hm Mhm Mhm So when is his birthdayB: The sixth of juneA: Mhm That no farB: No two weeks and a couple dayA: Yeah but we come in here for buy one card that no kind of early you move

off you move offB: HuhA: You’re moving off hall(ICE-JA S1A-025)

(7) Suddenly the bony hands gripped his neck and the pair of bony legs folded around his waist. The overseer began struggling to free himself. With the increased weight and the struggle above him, the horse began galloping. ‘Let me go!’ cried the overseer. ‘Me say let me go!’ ‘When you picked me up, you picked up your troubles,’ said Dry Bones. The horse was galloping faster now, and in their struggling they lost their balance and fell to the ground.(ICE-JA W2F-002)

204 Elena Seoane

(8) A: Good afternoon SirD: What I wanted to say gentleman about the sound system or whateverA: The night noisesD: MhmA: Go on SirD: And the government do something about the people with businesses in

commercially in uhm private resident area residential areaA: Sir the horse gone through the gate taking both the gate and the road

because we have had that discussion about uh remember <unclear> words </unclear> said to Mr Patterson a nice man but I remember one of the things he said once that I was going to ask the town planning people to step up the thing about the zoning and they have managed to defeat him Hear what I’m saying because weren’t they talking about zoning and how which areas you could have commercial activity I don’t know

D: Wait wait you see this is going to be the problem Mrs GloudonA: Not going to be Sir it isD: Of a uh if you have a club set up in the heart of a residential area the people

will not sleepA: Yes Sir(ICE-JA S1A-095)

(9) I don’t think so in terms of having no friends my good friends are gone(ICE-JA S1A-033)

(10) I think that’s finished for probably finished(ICE-JA S1A-025)

(11) Dagmar I’m so glad you came because all this that comes out I never knew this about this chap(ICE-JA S1A-002)

(12) Right I haven’t tell him yeah(ICE-JA S1A-094)

(13) their ‘opium’ as Karl Marx describes it, that is good, we have came a long way.(ICE-JA W1A-018)

(14) Uh many people Americans on the North Coast that ain’t ever seen no white Jamaican before(ICE-JA S2A-040)


At first glance we are inclined to think that the presence of alternative forms might be related to contact with JamC since, as Sand points out, “a considerable amount of variation is inherent to all creoles due to their historical development” (Sand 1999: 52). In addition, according to Patrick (1999: 225), in JamC “nearly every speaker shows variable marking in nearly every verb class.” It is not far-fetched to suggest that intense contact with this variable creole might be partly responsible for the high frequency of alternative forms. However, these need to be analysed individually, which is done below.

Starting with the bare form, contact with JamC is indeed the prime suspect because of its relatively high frequency, as compared to other native and non-na-tive varieties.³ On the one hand, as Sand (1999: 120) notes, basilectal JamC “does not inflect verbs for past tense or past participle with the morpheme -ed […]”, and on the other hand, it is also characterized by reduction of word-final consonant clusters, which involves the pronunciation of the -ed endings after a consonant, as in talked or danced (see Section 2). Not surprisingly, the only truly regular verb in the database, finish, shows the highest proportion of bare forms, with 18.2 % of the total (see Table 9). This is in accordance with previous research (Suárez-Gó-mez and Seoane 2013a: 171), showing that regular verbs are more prone to appear in the base form in perfect contexts in non-native varieties of English (see also Gut 2009 for this phenomenon in Singapore English).

Earlier research on the use of bare forms in JamC found that, of all linguis-tic factors which model the variation between inflection and non-marking (bare forms), the morphological category of the verb is the strongest one (Patrick 1999: 226–227). Our verbs fall into five of the ten categories that Patrick (1999) recog-nizes: categories consisting of single verbs, such as (i) and (ii) below, irregular verbs (iii), semi-weak verbs with both ablaut and /-t, -d/ affixation, as in (iv), and consonant-final regular non-syllabic verbs (v): (i) go (ii) say (iii) irregular come, get, give, see, tell, and think (iv) semi-weak hear (v) consonant-final regular non-syllabic finish.

3 Previous studies (Suárez-Gómez and Seoane 2013b) compare the number of bare forms in the private spoken component of several ICE components. The proportions are: 13.5 % of all forms expressing perfect meaning in ICE-JA, 2.5 % in four Asian Englishes taken together (Hong Kong, India, The Philippines and Singapore), and 4.1 % in East African English as determined through ICE-EA. JamE is clearly exceptional in terms of the high rate of bare forms it exhibits.

206 Elena Seoane

Table 3 presents a comparison of Patrick’s (1999: 231) results for verbs expressing past time according to verb class in JamC with our results for JamE.

Table 3: Inflection rates by morphological category, compared with Patrick (1999).

ICE-JA Perfect meaning JamC (Patrick 1999) Past time

% N % N

(i) Go 100 167/167 51 76/150(ii) Say 95 142/150 18 35/198(iii) Irregular verbs⁴ 91 526/578 31 196/624(iv) Semi-weak verbs 97 134/138 44 44/100(v) Regular consonant-final 77 34/44 19 74/380

As expected, the rate of inflection in JamE is much higher than that in JamC. Of interest, however, is the comparatively low count for finish (see (v)), which is inflected in only 77.3 % of cases in the database. As already mentioned, finish is the only true regular verb in the data and adding the -ed ending results in a consonant cluster, which tends to be avoided in JamC and, through influence from JamC, in JamE. Moreover, given that it is a regular verb, its preterite and past participle forms are predictable from the bare form and may thus be deemed dis-pensable. This is not surprising in a creole language, where inflectional endings tend to be avoided, especially if they are redundant; that is, if there is adverbial support or the time frame is made conspicuous in some other way. In example (5), which illustrates the use of the bare form, the adverbial up to now makes the per-fect-time reference very clear (starting in the past and leading up to the present). In examples (6) and (7), which are ambiguous between a bare form and a present and/or participle form, it is the context that would make us expect a have + past participle form.

As we know, language contact situations tend to result in the operation of cognitive constraints, which can be seen in processes of morphological simpli-fication and the optimization and elaboration of syntax to express grammatical meaning (see Kortmann and Szmrecsanyi 2004: 1191, 2009: 274; Schneider 2007: 82; Mesthrie and Bhatt 2008: 58–59). In fact, the bare form is a vernacular uni-versal (eWAVE feature 177; see also Kortmann and Szmrecsanyi 2004; Kortmann 2006: 609) and as such is also frequently attested in contact varieties whose substrate does not favour morphological simplification. In Indian English, for example, the dominant substrate languages concerned are inflectional (e.g.

4 Results for individual irregular verbs: come 85.4 % (117/137), get 96.0 % (73/76), give 97.0 % (99/102), see 87.7 % (143/163), tell 93.6 % (73/78), and think 87.5 % (21/22).


Hindi) and agglutinating languages (Dravidian languages) and bare forms occur only in slightly lower proportions than in other Asian varieties which do have isolating languages as the substrate (Hong Kong English and Singapore English; see Seoane and Suárez-Gómez 2013a: 11).

As for participle forms (without auxiliaries), these account for 5.8 % of all forms expressing perfect meaning in ICE-JA. However, of the 62 examples found, 59 are occurrences of go (see example (8) above). If we look more closely at the way in which perfect meaning is expressed here (see Table 9), we see that go selects the have + past participle periphrasis in 37.7 % of the cases (63 tokens), the participle almost equals this proportion with 35.3 % (59 instances), and we also find be-periphrases, to be discussed below, with 18.0 % (30 tokens). In other words, there is a strong tendency in go to select participles and be-periphrases to express perfect meaning (together they amount to 53.3 %). Arguably, these two kinds of examples are related since participles seem to be cases of copula dele-tion (eWAVE feature 177), typical of Caribbean and Pacific varieties (Christie 2003; Kortmann and Schneider 2004: 1180; Schneider 2011: 105).⁵ As Deuber (2014: 23), among others, observes, “zero copula is the normal form in JamC with adjectival predicates.” Given that JamC does not have primary auxiliaries, it makes sense to find copula deletion and, consequently, to attribute the use of participles in JamE to the influence of JamC. The other three examples have the verb see, as in (15).

(15) never Never seen a diva so upset.(ICE-JA W1B-009)

Finally, and in contrast with bare forms and past participles, the fairly frequent occurrence of be-periphrases to express perfect meaning cannot be attributed to substrate influence. As shown in example (9) above and (16) below, most be-periphrases are examples of go and intransitive finish.

(16) Have they finished the work No they are not finished but there’s a(ICE-JA S1A-097)

In the previous paragraphs I mentioned be-periphrases in connection with the high incidence of participle gone to express perfect meaning, since, in such exam-ples, I interpret sequences such as he gone with perfect meaning as instances of

5 An anonymous reviewer rightly points out that, at least in some Caribbean English-lexifier cre-oles, gone functions in numerous contexts where a copula interpretation is precluded. However, in the examples included in our database the copula-deletion interpretation is not precluded and perfect meaning seems to be clear.

208 Elena Seoane

copula deletion: he is gone > he Ø gone. As for the frequency of the periphrasis be gone itself, the fact that it occurs predominantly with the intransitive verb of movement go, as was the case in the history of English, seems to indicate that this is a replica of the pattern found in earlier periods of the language, in which both have and be were used as perfect auxiliaries depending on the verb concerned. Thus, have would occur with transitive verbs and be with mutative intransitive ones (see Werner, this volume). At the time when the colonization of Jamaica took place, there was still variation between have and be in such a way that have was normally selected with transitive verbs and also intransitive ones if non-mu-tative. Mutative intransitive verbs such as go alternate between have and be in this period (see Seoane and Suárez-Gómez 2012: 625), which is precisely what our data show. The occurrence of be gone in such examples could thus be a case of diffusion from the input language (in Schneider’s 2000 sense), a spoken vernacu-lar variety of the seventeenth century, when the grammaticalization of the use of have for all verbs had not yet taken place and be was still frequently found with mutative intransitives.

As noted above, be-periphrases in ICE-JA are also frequent with intransitive finish (15 examples, 34.1 % of all instances of finish; see example (16) above). Though instances of be finished + direct object are attested in other varieties, such as in some US dialects and in Canadian English, as in I am finished my homework (Yerastov 2015: 1; see also Werner, this volume), in our corpus of JamE only intran-sitive finish is found. The periphrasis be + intransitive finish is frequent in both native and non-native varieties of English. Quirk et al. (1985: 170) refer to it as a pseudo-passive construction with a verb of completion having a meaning that is nearly synonymous to the perfect.

Finally, examples (17) to (20) illustrate the use of a particular variant for the expression of the perfect, already reported in Sand (1999: 115) and not included in the database because the verbs concerned are do, look and grow, which do not form part of the present study. The form been + Ving is registered four times out of a total of 333 examples of have been + Ving in ICE-JA. Three of them, (17) to (19) below, have perfect meaning.

(17) Well all I want to tell Annotto Bay uh uhm community is that they must con-tinue supporting us at the hospital continue do as much as they can to improve on the infrastructure you know and you know and support us and protect the institution protect the uh the the workers you know cos they been doing a a very good job(ICE-JA S1B-038)


(18) We would like to speak to you because we been getting a lot of callers who are suggesting by implication that Mister Anderson might have had some unfair advantage in the system by virtue of being a Member of Parliament(ICE-JA S1B-050)

(19) So we started spreading in nineteen well whe when when the Junior Centre was formed but in nineteen sixty-five sixty-one we moved out of Kingston because the the government thought it was important to to spread the culture a little and in sixty-five again we got another space and we been growing like that now we’re trying to go to Montego Bay and we’ll have a little museum showing not only uhm ethnography and history which I deal with but uhm natural history(ICE-JA S2A-053)

3.3 Distribution by mode

In this section, I present findings on the distribution of forms expressing perfect meaning by mode, spoken vs. written. The underlying question of this analysis is the degree of consolidation of the alternative forms found in JamE. An even dis-tribution of alternative forms in both spoken and written discourse would entail a considerable degree of integration of the forms concerned. In other words, they would be used not only in spontaneous speech, the typical locus of innovation, and change and also the setting “where shifts into more mesolectal styles are frequent” (Mair and Sand 1998: 190), but they would be frequent and acceptable enough to have permeated the written mode. Table 4 sets out the results of this comparison.

Table 4: Distribution of forms (absolute and normalized frequencies per 10,000 words as well as percentages) by mode.

Spoken Written TOTAL

N Norm. % N Norm. % N %

have+past participle 475 7.9 55.6 114 2.9 51.1 589 54.7Simple past 225 3.8 26.3 77 1.9 34.5 302 26.7Bare form 53 0.9 6.2 16 0.4 7.2 69 6.4Participle 61 1.0 7.1 1 0.03 0.4 62 5.8be-periphrasis 36 0.6 4.2 14 0.4 6.3 50 4.6Other 4 0.1 0.5 1 0.03 0.4 5 0.5

Total 854 223 1,077

210 Elena Seoane

In Table 4 I have included the normalized frequencies (per 10,000 words), since the total number of words is different in the spoken and written components of ICE-JA (see Table 1, Section 3.1). The differences in frequency between written and spoken modes are statistically significant, having excluded the category ‘other’ from the statistical test due to the low number of occurrences (χ2 = 47.98, p < 0.0001). In both spoken and written JamE, have + past participle is the most commonly used form to express perfect meaning. This finding differs from previ-ous studies, which did not include public dialogues, unscripted monologues, and scripted monologues (see Section 1 and Seoane and Suárez-Gómez 2013a). For this reason, a comparison of frequencies by text type is in order here. It is set out in Table 5 (Section 3.4), which subsumes all non-perfect forms, that is, all forms other than canonical have + past participle.

Starting with have + past participle, this is the most frequent form in both modes, and it features more prominently in the spoken one (55.6 % in speech vs. 51.1 % in writing). The simple past, however, is considerably more common in writing, where it is selected on 34.5 % of the occasions, as against 26.3 % in speech. A likely explanation for the higher frequency of the simple past in the written mode is the fact that writing often deals with the reporting of facts, which includes frequent past reference. In any case, the frequency observed in the use of the simple past to express perfect meaning in writing seems to indicate a good degree of consolidation of this form in JamE.

Of the remaining forms, the be-periphrasis is fairly frequent in both spoken and written JamE, which is plausible since its use has been put down to diffusion from the original vernacular input (see Section 3.2), and as this variant is also attested in standard varieties. As for the bare form, it also seems fairly stabilized in JamE, since it occurs in both modes with similar proportions (6.2 % in the spoken mode and 7.2 % in writing). If we consider normalized frequencies, however, the bare form is clearly more frequent in the spoken mode (0.9 % vs. 0.4 %). Finally the participle, derived from substrate influence, shows a pronounced preference for the spoken mode (7.1 % vs. only 0.4 % in writing). Judging by its mode distri-bution, the participle would not readily be accepted in written JamE and would still be considered innovative, typical of spontaneous speech, and thus be edited out of the written language.

3.4 Text type distribution

In Figure 1 I provided frequencies for the variants under investigation, looking at the written component and only one part of the spoken component (the private dialogue texts) of the ICE-JA corpus. In this figure, have + past participle


(‘perfect’) is not the most frequently selected form to express perfect meaning. In Section 3.2, however, the corpus in its entirety is included and the general distri-bution changes, with have + past participle being the most common option in both modes. This finding calls for further analyses of text type frequencies, which are set out in Table 5.

Table 5: Distribution of perfect vs. non-perfect forms by text type (each text comprises approxi-mately 2,000 words; relative frequencies in bold if non-perfect forms in majority).

Perfect Non-perfect Total

Spoken private dialogue (S1A, 100 texts) 117 (35.1 %) 216 (64.9 %) 333Spoken public dialogue (S1B, 80 texts) 158 (74.9 %) 53 (25.1 %) 211Unscripted monologue (S2A, 70 texts) 144 (65.5 %) 76 (34.5 %) 220Scripted monologue (S2B, 50 texts) 56 (62.2 %) 34 (37.8 %) 90Non-printed letters (W1B, 30 texts) 43 (40.6 %) 63 (59.4 %) 106Printed reportage (W2C, 20 texts) 13 (41.9 %) 18 (58.1 %) 31

Non-printed student writing (W1A, 20 texts) 10 (83.3 %) 2 (16.7 %) 12Printed academic writing (W2A, 40 texts) 9 (81.8 %) 2 (18.2 %) 11Printed popular writing (W2B, 40 texts) 14 (87.5 %) 2 (12.5 %) 16Printed instructional writing (W2D, 20 texts) 4 (57.1 %) 3 (42.9 %) 7Printed persuasive writing (W2E, 10 texts) 17 (94.4 %) 1 (0.6 %) 18Printed creative writing (W2F, 20 texts) 4 (18.2 %) 18 (81.8 %) 22

Total 589 (54.7 %) 488 (45.3 %) 1,077

As Table 5 shows, some of the text types produce low numbers of examples, either because they contain fewer samples (for instance, persuasive writing contains only 10 texts, that is, 20,000 words) or because their writing style is not prone to talking about past time (for example, instructional writing). Even though they may be in line with trends in other text types, small text categories may more easily have extreme values, which I have disregarded since they are based on just a few tokens. For this reason, text types producing less than ten tokens in the perfect and/or non-perfect column are not used here to identify general trends, and are shaded grey in the table.

The first six rows, with numeric values higher than ten, show that the pre-ferred realm of the have + past participle form is spoken public dialogue (S1B), unscripted monologue (S2A) and scripted monologue (S2B). The other three text types, where the non-perfect forms are more numerous, are the more spontaneous spoken private dialogue (S1A) and two written texts: non-printed letters (W1B) and printed reportage (W2C). The differences between perfect and non-perfect forms just described are statistically significant (χ2 = 107.86, p < 0.0001). If we identify more openness toward alternative variants to express perfect meaning with a low

212 Elena Seoane

level of formality and elaboration, we see it as only natural to find more alternative forms in spoken private dialogue and in non-printed letters, which tend to adopt a more familiar, spontaneous and speech-like style. On the other hand, both in non-printed letters and in printed reportage, the reporting or narrating of recent past events plays a very important role. Traditionally, the standard would require have + past participle to express recent past with present relevance. However, recent past events (as one type of perfect meaning) are precisely the context in which have + past participle and simple past are in variation in standard varieties, espe-cially in American English (Leech 2004: 43). If we count the number of simple pasts within the alternative forms, we see that in printed reportage (W2C) simple pasts amount to 13 out of the 18 alternative forms (72.2 %) and in non-printed letters (W1B) simple pasts are 50 out of the 63 alternative forms (79.3 %).⁶

The preference of the have + past participle periphrasis for public dialogue and monologue, both scripted and unscripted, invalidates a potential divide between written and spoken modes as far as the distribution of perfect-meaning forms is concerned.

3.5 Semantics

In order to find potential specializations of meaning of particular variants I studied the patterns of distribution of the different variants according to the sub-types of perfect meaning (as established in Comrie 1976 and as used in Dahl 1985: Chapter 5, 1999; Dahl and Hedin 2000: 385–388; Huddleston and Pullum 2002: 143–145; Miller 2000: 327–331, 2004: 230): resultative, experiential, recent past and persistent situation.

The distribution of perfect and non-perfect forms is fairly even in all types of perfect meaning, with the exception of experiential meaning, where have + past participle, as in example (20), clearly predominates over other forms (see example (21)), with 67.5 % of all occurrences (see Table 6).

(20) And what what to me is amazing is that I have never seen these people ever again and that I don’t know what ever happened to them(ICE-JA S2A-036)

6 I am grateful to an anonymous reviewer for drawing my attention to several aspects in this analysis. One of the issues raised is that creative writing behaves like conversations, non-printed letters, and news reports again. Arguably, this is exactly for the reason just outlined: frequent narrative, i.e., frequent occurrence of past events being reported (see also Richard & Rodriguez Louro, this volume).


Table 6: Distribution of perfect (have + past participle) and non-perfect forms by type of perfect meaning.⁷

Perfect Non-perfect TOTAL

Resultative 294 (50.6 %) 287 (49.4 %) 581Experiential 191 (67.5 %) 92 (32.5 %) 283Recent past 95 (48.5 %) 101 (51.5 %) 196Persistent situation 9 (52.9 %) 8 (47.1 %) 17

Total 589 (54.7 %) 488 (45.3 %) 1,077

(21) He helped David tuck his shirt in his pants, and zipped up his pants. He noticed that David didn’t have his belt. “Where yu belt?” David ran into the living room and brought back the belt. “What I tell yu bout leavin yu belt inside the livin room? Yu want people to tink is pure boogoyaga live in dis house?” “I won’ do it again, sah.”(ICE-JA W2F-020)

In contrast, the expression of recent-past perfect meaning disfavours the use of the have + past participle and tends to select alternative variants. Previous findings show a fairly strong tendency of the recent past meaning to single out the simple past in this function. In ICE-JA this is also true, as can be seen in Table 7, which displays the semantic subtypes of perfect meaning expressed by the simple past in the corpus. It shows that the highest proportion of simple past forms (43.4 %) is seen in examples of the expression of recent past.

Table 7: Distribution of simple past forms by semantic subtype.

Total N examples Simple past

Resultative 581 147 (25.3 %)Experiential 283 64 (22.6 %)Recent past 196 85 (43.4 %)Persistent situation 17 6 (35.3 %)

Finally, example (22) illustrates the expression of persistent situation, which nor-mally occurs with the adverbial always.

7 Statistical tests show that the differences between perfect and non-perfect forms are signifi-cant (χ2 = 25.71, p < 0.0001).

214 Elena Seoane

(22) I’ve always said divide that up by h c, equals B divided by h c so th this now converting it from joules into the wave numbers(ICE-JA S1B-020)

3.6 Adverbial support

Presence of adverbial support indicating time reference could at times render the explicit have + past participle form redundant and thus foster the use of alterna-tive forms. This mainly applies to the items ever (eWAVE feature 108) and already (eWAVE feature 109), but may in principle hold for others such as never, since, just and for. On a related note, Miller (2000: 335) suggests that the classical inter-pretation of the meaning of the perfect does not always reside in the perfect have + past participle construction but in the adverbials. Table 8 shows the rate of examples with adverbial support, specifying the number of examples with an adverbial, the total number of examples in each semantic category, and the rela-tive percentage of examples with adverbial in brackets.

Table 8: Semantic distribution and rate of adverbial support of perfect (have + past participle) and non-perfect forms (% indicates proportion of adverbial support).

Perfect Non-perfect Total

N % N % N

Resultative 35/294 11.9 24/287 8.4 581Experiential 74/191 38.7 42/92 45.7 283Recent past 32/95 33.7 58/101 57.4 196Persistent situation 8/9 88.9 5/8 62.5 17

Total 149/589 129/488 1,077

Table 8 shows that adverbial support is lowest (11.9 % and 8.4 %) when the perfect meaning is most prototypical and frequent, that is, with resultative meaning (Miller 2000, 2004), which constitutes more than 50 % of all examples in the database (581, 53.9 %). As the type of perfect meaning becomes more and more infrequent in the corpus, and hence semantically marked, adverbial support increases, reaching the highest level with the most exceptional meaning, that of persistent situation (see Werner 2014: 370). Examples (23) to (30) illustrate the use of perfect (have + past participle) and non-perfect with adverbial support for the expression of the different common present perfect meanings.


Resultative

(23) The industry has now come of age, keeping the best of that era, plus new ideas.(ICE-JA W2B-037)

(24) Bruce wants to know if you guys got the papers yet and what is the next move?(ICE-JA W1B-004)

Experiential

(25) I don’t know if you have ever heard of such a committee(ICE-JA S1A-098)

(26) Uh yes that can work Why we never think of that before(ICE-JA S1A-018)

Recent past

(27) I’ve just finished the assignments I had due this month, now I’m getting ready for my exams(ICE-JA W1B-010)

(28) Being friendly and helpful to the new customer who just came in? Planning next week’s promotion?(ICE-JA W2D-020)

Persistent situation

(29) I’ve always said divide that up by h c, right equals B divided by h c so this now converting it from joules into the wave numbers right(ICE-JA S1B-020)

(30) Jesus and I always said to him him say until <indig> una</indig> get big live with somebody(ICE-JA S1A-045)

If we compare the rate of adverbial support in perfect and non-perfect forms, the tendency is to find the presence of adverbials to be more frequent with non-per-fect forms. We could argue in these cases that the expression of perfect meaning

216 Elena Seoane

lies in the adverbial more than in the verb form, and this would make the use of the perfect have + past participle periphrasis redundant and unnecessary. This could well be the case in example (26), Why we never think of that before, where the adverb never conveys the experiential-perfect meaning and the have + past participle form is no longer strictly required for the expression of such meaning.

In this analysis, we need to make an exception with the expression of recent past situations, since not only in JamE but also in several Asian Englishes (Hong Kong, Singapore, Philippines, India), have the have + past participle periphra-sis and the simple past reached a kind of levelling that makes them fairly inter-changeable (Seoane and Suárez Gómez 2013b; see also Werner 2013; Davydova, this volume; Fuchs, this volume). In British and American English the simple past is not as frequent as in the varieties just mentioned, but its presence is fairly remarkable (see Quirk et al. 1985: 194; Leech 2004: 43). For this reason, in an example like (28), a customer who just came in, it is both the adverbial just and the simple past form that make clear that the time reference of the action is the recent past.

3.7 The syntax-lexicon interface

So far, I have sketched the role of mode, text type, semantics and adverbial support in the use of the different variants to express perfect meaning. In this section, I examine the syntax-lexicon interface to check whether particular verbs show a preference for the have + past participle periphrasis (‘perfect’) or whether, on the contrary, they favour non-perfect forms. Table 9 provides a breakdown of the results.

Table 9: Number and percentage of variants per verb.

Perfect Simple past be-periphrasis Bare Participle Other Total

come 86 (62.8 %) 26 (19.0 %) 3 (2.2 %) 20 (14.6 %) 0 2 (1.5 %) 137finish 13 (29.5 %) 8 (18.2 %) 15 (34.1 %) 8 (18.2 %) 0 0 44get 34 (44.7 %) 39 (51.3 %) 0 3 (3.9 %) 0 0 76give 77 (75.5 %) 23 (22.5 %) 1 (1.0 %) 1 (1.0 %) 0 0 102go 63 (37.7 %) 15 (9.0 %) 30 (18.0 %) 0 59 (35.3 %) 0 167hear 89 (64.5 %) 45 (32.6 %) 0 4 (2.9 %) 0 0 138say 76 (50.7 %) 65 (43.3 %) 1 (0.7 %) 8 (5.3 %) 0 0 150see 111 (68.1 %) 27 (16.6 %) 0 20 (12.3 %) 3 (1.8 %) 2 (1.2 %) 163tell 30 (38.5 %) 43 (55.1 %) 0 4 (5.1 %) 0 1 (1.3 %) 78think 10 (45.5 %) 11 (50.0 %) 0 1 (4.5 %) 0 0 22

Total 589 302 50 69 62 5 1,077


The findings in Table 9 are illustrative of a fairly strong specialization (higher than 60 %) of some verbs, as follows: (i) only four verbs show a clear preference for the present perfect: come, give, hear and see; (ii) a further four verbs illustrate level-ling of present perfect and simple past: get, say, tell and think; except for say, all of these favour the simple past over the present perfect; (iii) finish expresses perfect meaning predominantly with the auxiliary be in what Quirk et al. (1985: 170) call a pseudo-passive construction with perfect meaning (see Section 3.2 and examples (9) and (16) above); finally, (iv) go is very frequently used in the participle and with the be-periphrasis (53.3 % taken together) to express perfect meaning. Table 9 thus illustrates strong lexical influence on the choice of verbal form used to express perfect meaning. This finding might be connected with text type; for example, the verb say is frequently attested in printed reportage in the simple past serving the purpose introducing speech acts which take place in the recent past.

4 Conclusions This paper has identified all perfect contexts for ten high frequency verbs in ICE-JA, amounting to 1,077 instances. Most of these examples are cases of the canonical have + past participle form (54.7 %), but many alternative forms are also found. The most widespread of the non-perfect forms is the simple past (28.0 %), which is frequent in both modes; the relatively high frequency of the simple past in written discourse is taken as a sign of its advanced degree of con-solidation in JamE.

Other recurrent alternative forms attested are the bare form, the be + past participle periphrasis, and the participle. The bare form occurs mainly in spoken JamE and is interpreted as the result of influence from JamC, where verbs do not inflect for past tense or past participle and where final consonant reduction takes place. More specifically, it is the only regular verb in our database, finish, that shows the highest proportion of bare forms (22.7 %).

Be-periphrases in ICE-JA occur only with two verbs: (i) with go, whose occur-rence is put down to diffusion from the input language, a vernacular variety of early English, when the grammaticalization of have with mutative intransitive verbs had not yet taken place; (ii) with finish, in a construction that Quirk et al. (1985) call pseudo-passive and that has perfect meaning. Perfect periphrases with be as auxiliary occur frequently in both written and spoken texts, which indicates a fairly high degree of consolidation.

218 Elena Seoane

While previous studies show that be-periphrases are found in many varieties of English as a second language (ESL) and also in native varieties of English (see Seoane and Suárez-Gómez 2013a; Werner, this volume), the incidence of parti-ciples, like that of bare forms, is particularly high in spoken JamE. Participles account for 5.8 % of all the perfect forms in the corpus, in what I interpret to be the clear influence of JamC. Evidence for this comes from the fact that most par-ticiples found are gone (95.1 % of the total), a verb which frequently selects be as perfect auxiliary. It can speculated that in these examples be is elided due to influence from JamC and its characteristic copula deletion, which is also typical of other Caribbean and Pacific varieties.

Section 3.4 examined the distribution of variants per text type and shows that a written vs. spoken divide is not justified in terms of explaining the distribu-tion of perfect-meaning forms. What we find instead is a tendency for have + past participle forms to occur in formal written texts and also in spoken scripted texts, whereas alternative forms are favoured in spontaneous speech (private dia-logues) as well as in private letters, which tend to adopt a speech-like style. The simple past in particular is predominant in reportage since this register is partic-ularly concerned with the reporting of recent past events, which is precisely the semantic context in which the simple past is more common.

As is also the case in other native and non-native varieties of English, the most frequently found perfect meaning is resultative (almost 54 % of all the exam-ples here have this meaning), followed by experiential and recent-past meaning. Perfect (have + past participle) and non-perfect forms are evenly distributed to express resultative meaning, whereas experiential meaning is clearly associated with the have + past participle form. Interestingly, the simple past, as in other varieties of English, is seen to specialize in the expression of recent-past meaning. Analyses of perfect meaning and use of adverbial support show a correlation between high semantic prototypicality and low adverbial support, with resulta-tive examples showing adverbial support less frequently than experiential and recent-past examples. As for the rate of adverbial support in perfect and non-per-fect forms, the tendency in JamE is for higher rates with non-perfect forms, and I hypothesize that in these examples perfect meaning is carried by the adverbial, and therefore the have + past participle periphrasis is deemed redundant. The apparent exception here is the expression of recent past, where both the simple past and the adverbial can express perfect meaning.

This paper has also considered the syntax-lexicon interface to see whether the choice of verbal form is determined by the verb itself. Findings show a very strong lexical influence, with some verbs selecting the have + past participle form predominantly, others showing levelling between have + past participle and the simple past, and two verbs showing a special tendency: go frequently


selects the participle to express perfect meaning and finish shows a strong pref-erence for the be-perfect. The strength of the syntax-lexicon interface might inter-act with the effect of other factors (text type, semantics and adverbial support), and further research is needed here.

AcknowledgementsI am grateful to Cristina Suárez-Gómez, my “perfect” colleague and friend, for her help with earlier versions of this paper. I am also indebted to two generous but anonymous referees for their thorough and clever comments. For generous financial support, I am grateful to the Spanish Ministry of Economy and Compet-itiveness (grants FFI FFI2014-53930-P and FFI2014-51873-REDT).

ReferencesAnthony, Laurence. 2014. AntConc (Version 3.4.3). Tokyo: Waseda University. http://www.

laurenceanthony.net/ (accessed 15 December 2014).Carpenter, Karen & Hubert Devonish. 2010. Swimming against the tide: Jamaican Creole

in education. In Bettina Migge, Isabelle Léglise & Angela Bartens (eds.), Creoles in education: An appraisal of current programs and projects, 167–181. Amsterdam: Benjamins.

Christie, Pauline. 1989. Questions of standards and intra-regional differences in Caribbean examinations. In Ofelia García & Ricardo Otheguy (eds.), English across cultures, cultures across English: A reader in cross-cultural communication, 243–262. Berlin: Mouton de Gruyter.

Christie, Pauline. 2003. Language in Jamaica. Kingston: Arawak.Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press.Dahl, Östen. 1985. Tense and aspect systems. Oxford: Blackwell. Dahl, Östen. 1999. Perfect. In Keith Brown & Jim Miller (eds.), Concise encyclopaedia of

grammatical categories, 30–37. Oxford: Elsevier.Dahl, Östen & Eva Hedin. 2000. Current relevance and event reference. In Östen Dahl (ed.),

Tense and aspect in the languages of Europe, 385–401. Berlin: Mouton de Gruyter.DeCamp, David. 1971. Towards a generative analysis of a post-creole speech continuum. In

Dell Hymes (ed.), Pidginization and creolization of languages, 349–370. Cambridge: Cambridge University Press.

Deuber, Dagmar. 2014. English in the Caribbean: Variation, style and standards in Jamaica and Trinidad. Cambridge: Cambridge University Press.

Devonish, Hubert & Ewart A. C. Thomas. 2012. Standards of English in the Caribbean. In Raymond Hickey (ed.), Standards of English: Codified varieties around the world, 179–197. Cambridge: Cambridge University Press.

220 Elena Seoane

Elsness, Johan. 2009. The present perfect and the preterite. In Günter Rohdenburg & Julia Schlüter (eds.), One language, two grammars? Differences between British and American English, 228–246. Cambridge: Cambridge University Press.

Görlach, Manfred. 1991. English as a world language: The state of the art. In Manfred Görlach (ed.), Englishes: Studies in varieties of English 1984–1988, 21–48. Amsterdam: Benjamins.

Gut, Ulrike. 2009. Past tense marking in Singapore English verbs. English World-Wide 30(3). 262–277.

Hackert, Stephanie & Dagmar Deuber. 2015. American influence on written Caribbean English: A diachronic analysis of newspaper reportage in the Bahamas and in Trinidad and Tobago. In Peter Collins (ed.), Grammatical change in English world-wide, 389–410. Amsterdam: Benjamins.

Hickey, Raymond. 2004. Introduction. In Raymond Hickey (ed.), Legacies of colonial English: Studies in transported dialects, 1–30. Cambridge: Cambridge University Press.



Irvine, Alison. 2004. A good command of the English language: Phonological variation in the Jamaican acrolect. Journal of Pidgin and Creole Languages 19. 41–76.

Kortmann, Bernd. 2006. Syntactic variation in English. In Bas Aarts & April McMahon (eds.), The handbook of English linguistics, 603–624. Oxford: Blackwell.

Kortmann, Bernd & Kerstin Lunkenheimer (eds.). 2013. The electronic world atlas of varieties of English. Leipzig: Max Planck Institute for Evolutionary Anthropology. http://ewave-atlas.org (accessed 06 April 2015).

Kortmann, Bernd & Benedikt Szmrecsanyi. 2004. Global synopsis: Morphological and syntactic features in English. In Bernd Kortmann, Kate Burridge, Rajend Mesthrie, Edgar W. Schneider & Clive Upton (eds.), A handbook of varieties of English. Vol. 2: Morphology and syntax, 1142–1202. Berlin: Mouton de Gruyter.

Kortmann, Bernd & Benedikt Szmrecsanyi. 2009. World Englishes between simplification and complexification. In Thomas Hoffmann & Lucia Siebers (eds.), World Englishes – Problems, properties and prospects, 265–285. Amsterdam: Benjamins.

Leech, Geoffrey. 2004. Meaning and the English verb. London: Longman.Mair, Christian. 1992. Problems in the compilation of a corpus of Standard Caribbean English:

A pilot study. In Gerhard Leitner (ed.), New directions in English language corpora: Methodology, results, software developments, 75–96. Berlin: Mouton de Gruyter.

Mair, Christian & Andrea Sand. 1998. Caribbean English: Structure and status of an emerging variety. In Raimund Borgmeier, Herbert Grabes & Andreas H. Jucker (eds.), Anglistentag 1997 Giessen: Proceedings, 187–198. Trier: WVT.

Mesthrie, Rajend & Rakesh M. Bhatt. 2008. World Englishes: The study of new linguistic varieties. Cambridge: Cambridge University Press.

Miller, Jim. 2000. The perfect in spoken and written English. Transactions of the Philological Society 98. 323–352.

Miller, Jim. 2004. Perfect and resultative constructions in spoken and non-standard English. In Olga Fischer, Muriel Norde & Harry Perridon (eds.), Up and down the cline: The nature of grammaticalization, 229–246. Amsterdam: Benjamins.

Miller, Jim. 2006. Spoken and written English. In Bas Aarts & April McMahon (eds.), The handbook of English linguistics, 670–691. Oxford: Blackwell.


Patrick, Peter L. 1999. Urban Jamaican Creole: Variation in the mesolect. Amsterdam: Benjamins.

Patrick, Peter L. 2004. Jamaican Creole: Morphology and syntax. In Bernd Kortmann & Edgar W. Schneider (eds.), A handbook of varieties of English. Vol. 2: Morphology and syntax, 407–438. Berlin: Mouton de Gruyter.


Sand, Andrea. 1999. Linguistic variation in Jamaica: A corpus-based study of radio and newspaper usage. Tübingen: Narr.

Sand, Andrea. 2011. Language attitudes and linguistic awareness in Jamaican English. In Lars Hinrichs & Joseph T. Farquharson (eds.), Variation in the Caribbean: From creole continua to individual agency, 163–187. Amsterdam: Benjamins.

Sand, Andrea. 2012. Jamaican English. In Bernd Kortmann & Kerstin Lunkenheimer (eds.), The Mouton world atlas of variation in English, 210–221. Berlin: Mouton de Gruyter.

Schneider, Edgar W. 2000. Feature diffusion vs. contact effects in the evolution of New Englishes: A typological case study of negation patterns. English World-Wide 21(2). 201–230.


Schneider, Edgar W. 2011. English around the world: An introduction. Cambridge: Cambridge University Press.

Seoane, Elena & Cristina Suárez-Gómez. 2012. The impact of mode on morphosyntactic variation in Asian New Englishes. In Sara Martín Alegre, Melissa Moyer, Elisabet Pladevall & Susagna Tubau (eds.), At a time of crisis: English and American studies in Spain. Works from the 35th AEDEAN conference, 391–398. Barcelona: Universitat Autònoma de Barcelona.

Seoane, Elena & Cristina Suárez-Gómez. 2013a. The expression of the perfect in East and South-East Asian Englishes. English World-Wide 34(1). 1–25.

Seoane, Elena & Suárez-Gómez, Cristina. 2013b. Re-examining the levelling of present perfect and simple past in World Englishes. Paper presented at ICAME 34, University of Santiago de Compostela, 22–26 May.

Suárez-Gómez, Cristina & Elena Seoane. 2013a. They have published a new cultural policy that just come out: Competing forms in spoken and written New Englishes. In Gisle Andersen & Kristin Bech (eds.), English corpus linguistics: Variation in time, space and genre, 163–182. Amsterdam: Rodopi.

Suárez-Gómez, Cristina & Elena Seoane. 2013b. Grammatical variation in written and spoken World Englishes: A comparative analysis of the ICE corpora. Paper presented at the 46th Annual Meeting of the Societas Linguistica Europaea, Split University, 18–21 September.





Robert Fuchs 8 The frequency of the present perfect in

varieties of English around the worldAbstract: This paper investigates the frequency of the present perfect (PP) in 20 national varieties of English with data from the 1.9 billion word Corpus of Global Web-based English (GloWbE). Linear regression models were used to determine which factors can account for differences in PP frequency across varieties. The results revealed that a major factor is geographic proximity: Varieties spoken in the same region tend to be similar in PP frequency. Other factors such as degree of formality or the classification in Kachru’s Circle Model and Schneider’s Dynamic Model of Postcolonial English appear to be relatively unimportant. The paper also discusses other factors that might be influential, such as substrate influence, the heterogeneous superstrate, national identities, and the degree of cultural contact between varieties. Finally, the discussion explores implications for teach-ing English as a local and as an international language, arguing that norms of English language teaching should depend on the learners’ communicative needs.

Note: This chapter is accompanied by an interactive website that allows readers to explore different visualisations of the results: http://wp.me/p6rHfI-3 (descrip-tion) and https://english-linguistics.shinyapps.io/pp_we_shinyapp/ (direct link to dynamic world map)

1 IntroductionThe English language provides two main ways of referring to past events. One is the simple past (SP), and the other the present perfect (PP), consisting of a form of have and a past participle. This split between a synthetic (SP) and an analytical (PP) form of referring to the past is common to a number of Indo-Euro-pean languages, among them French and German (Anderson 1982: 243). In these languages, the analytical form has continued to gain ground; for example, the French passé composé is now preferred over the passé simple in all but the most formal registers. English, however, has defied this diachronic typological trend. After increasing in frequency until the second half of the eighteenth century, the English PP has been losing ground ever since (Elsness 1997), although this trend may have slowed down recently (Hundt and Smith 2009).

Robert Fuchs, Hong Kong Baptist University

224 Robert Fuchs

The decrease in the frequency of the PP appears to have taken place at dif-ferent rates in different varieties of English. For example, while it is decreasing in frequency in both British English (BrE) and American English (AmE), the PP continues to be more frequent in BrE than in AmE. The higher frequency of the PP in BrE goes hand in hand with its use in a higher number of contexts than in AmE (Elsness 1997, 2009a, 2009b; Hundt and Smith 2009).

The frequency of the PP is also known to vary across other varieties of English, such as East- and South-East Asian Englishes (Seoane and Suárez-Gómez 2013). This raises the question of how these differences can be explained. For example, the lower frequency of the PP in AmE is sometimes explained with the greater col-loquiality (or lesser formality) of AmE compared to BrE. Colloquialisation might also be a factor explaining the varying frequencies of the PP in a number of other varieties of English, but further factors such as ancestry and geographic proxim-ity might also play a role (Yao and Collins 2012: 400). Another factor influencing its frequency may be register, as in the use of the so-called ‘hot news perfect’ in journalistic language (Carey 2005).

Moreover, in a number of varieties alternative perfect structures are used, such as the after-perfect, the medial object perfect, perfective done, be + done + past participle, the perfect marker slam and the perfect marker already (see Table 1 for examples; Kortmann and Lunkenheimer 2013).

Table 1: Alternative perfect structures, examples, and feature number in The electronic World Atlas of Varieties of English (eWAVE; Kortmann and Lunkenheimer 2013).

Alternative perfect structure Example eWAVE feature

After-perfect She’s after selling the boat 98Medial object perfect And you eat nothing till you have the stations

made 97

Perfective done He done go fishing 104be + done + past participle He is done gone 105Perfect marker slam I slam told you not to mess up 107Perfect marker already We did move here a week already 109

The present paper aims to contribute to the ongoing debate about the distribution and frequency of the PP in varieties of English by studying a larger number of varieties with a larger dataset than in previous studies. On this basis, the validity of different factors that might explain differences in the frequency of the PP will be tested. Section 2 provides an overview of previous research on the frequency of the PP in varieties of English, concluding with the research questions for the present study. Section 3 describes the data and methods, and Section 4 the results of the study, followed by a discussion in Section 5 and the conclusion in Section 6.

The frequency of the present perfect in varieties of English around the world 225

2 Previous researchPrevious research on the frequency of the PP in varieties of English has applied different methods, and has invoked various factors to explain the differences in the frequency of the PP. Broadly, two approaches can be identified: the func-tion-based and the form-based approach.

2.1 The function-based approach

The function-based approach starts out by defining contexts that express past actions of current relevance, in which the PP can be used in English (referred to as “PP contexts” from here on). All verb phrases in the corpus data are then examined to determine whether they occur in such PP contexts. Davydova (2011, 2012, this volume) applied this method to data from the International Corpus of English (ICE) and other corpora. She found that in BrE the PP was used in 90 % and the SP in 10 % of all PP contexts, while in acrolectal Indian English (IndE) the PP was used in 74 % and the SP in 22 % of all contexts. In mesolectal IndE, the PP accounted for 45 % and the SP for 42 % of all contexts. Since Hindi (the most widely spoken Indian language) has a structure that is similar to the English PP, substrate influence cannot account for the lower frequency of the PP in IndE. Instead, Davydova argued that the PP is a complex category that is difficult to learn in the context of second language acquisition, even for speakers with L1s that include a structure similar to the English PP. This is also supported by pre-vious research on the second language acquisition of English tense and aspect, which found that the PP is acquired later than the SP (Klein 1995; Bardovi-Harlig 2000: 169–175; Fuchs, Götz, and Werner, this volume).

A special case of the function-based approach relies exclusively on the occur-rence of time adverbials that express current relevance (or its absence), and constitute prototypical PP (or SP) contexts (at least in BrE). Adverbials express-ing current relevance are, for example, for three days, since December, just and so far, while three days ago, in December and once imply no current relevance. Werner (2013, 2014) applied this approach to ICE data from a number of varieties of English, and found the highest rate of PP usage in BrE, followed by varieties that are relatively close to BrE (IndE, East African English [EAfE], and Hong Kong English [HKE]), and others with a much lower PP frequency (Philippine English [PhiE], Jamaican English [JamE] and Singapore English [SinE]).

Werner explained these differences by arguing (based on Schneider’s Dynamic Model of Postcolonial Varieties of English; Schneider 2003, 2007; see Section 2.4 below for details) that the first group continues to be oriented

226 Robert Fuchs

towards BrE, while the second group has already progressed to a phase where the varieties rely on their own norms. A phylogenetic analysis based on the occurrence of the PP with the time adverbials in question further confirmed the validity of the two groups, and also identified a third group, consisting of Inner Circle varieties (American English [AmE], Irish English [IrE], Australian English [AusE], New Zealand English [NZE]). Further explanations for the low frequency of the PP in PhiE, SinE and JamE were contact through geographic proximity, and the influence of AmE, where the PP is also relatively infrequent. However, there was little evidence of overall differences in PP usage between Outer Circle varieties (such as PhiE) and Inner Circle varieties (such as AmE; see Section 2.4 below for details).

2.2 The form-based approach

While the function-based approach determines contexts where the PP is proto-typically used, the form-based approach starts by defining the form of the PP (i.e. a variant of the auxiliary have + past participle). Then all cases in the data where this form is used are identified. Based on this, the frequency of the PP can be measured per million words, or as a percentage of past time reference by dividing the frequency of the PP by the sum of PP and SP frequency (see also Bowie and Wallis, this volume).

Using the form-based approach, Hundt and Smith (2009) found that the PP occurs more often in BrE than in AmE. Based on ICE data, Yao and Collins (2012) found that these two varieties form the poles of a continuum along which a number of other national varieties they investigated (NZE, AusE, Canadian English [CanE], PhiE, SinE, HKE, Kenyan English [KenE]) can be placed. The exception to this is IndE, which had an even higher PP frequency than BrE. More recently, Werner and Fuchs (2016) analysed data from the Nigerian component of ICE (which was not yet available at the time of Yao and Collins’ study) and showed that NigE is another variety where the PP is less frequent than in BrE. In terms of the factors that influence the frequency of the PP, Yao and Collins (i) documented a tendency for the PP to occur more often with a contracted auxiliary in those varieties where it is comparatively infrequent (i.e. the PP is less frequent in varieties that are relatively informal), and (ii) found evidence of areal clusters of varieties of English with similar frequencies of the PP.¹

1 Hundt and Biewer (2007) and Elsness (2009a) also investigated varieties beyond BrE and AmE, but, as Yao and Collins (2012) noted, the interpretation of their findings is made difficult by the methods they used. Hundt and Biewer (2007) used relatively small data sets, and Elsness (2009a)


2.3 Conclusions from previous research

The function- and form-based approaches used in previous research clearly complement each other. While the former takes a more fine-grained perspective, analysing particular contexts where the PP might occur, the latter focuses on the broader picture, allowing wide-ranging comparisons of many varieties at a time. The present study will therefore take the form-based approach to study variation in PP frequency across varieties of English.

While previous research has studied the frequency of the PP in a number of varieties of English, there are many more countries and varieties with large numbers of English speakers (e.g. Pakistani English [PakE], Bangladesh English [BanE] and Malaysian English [MalE]) which have not been accounted for in pre-vious research, most likely because of a lack of sufficient data. With the avail-ability of the Corpus of Global Web-based English (GloWbE), it is now possible to include these varieties in the study of the PP in varieties of English world-wide. With this data, it will be possible to test in how far the factors identified by pre-vious research can account for variation in the frequency of the PP in varieties of English.

Previous research has also highlighted that the PP is a complex category that is challenging for many (foreign) language learners (see also Fuchs, Götz, and Werner, this volume). After investigating the extent of variation in the frequency of the PP, the analysis will also explore which, if any, conclusions can be drawn for teaching English as a medium of local and international communication.

2.4 Explanatory factors

In summary, previous research has used a variety of methods, and has thus uncovered several factors that may influence the frequency of the PP in varieties of English:

– Formality: Since the PP tends to be favoured in formal contexts, varieties that are more formal than others might have higher PP frequencies.

– Geographic proximity: Varieties from the same region might be more similar to each other than to varieties in other regions. This follows from the notion that language contact leads to the diffusion of linguistic innovations and patterns of usage. Everything else being equal, geographic distance makes contact between speakers of different varieties less likely. Consequently, vari-

investigated how frequently the PP co-occurs with twelve verbs, most of which are not among the most frequent verbs, which might have skewed the results.

228 Robert Fuchs

eties from different regions might differ more from each other in PP frequency than varieties from the same region.

– Evolutionary development: Varieties that have progressed further in their development according to the Dynamic Model of Postcolonial English (Schneider 2003, 2007) might differ more from BrE than varieties in an earlier phase. The Dynamic Model posits a five-phase developmental process for postcolonial varieties of English. After phase 1 (foundation), where a new variety is born as the first English speakers take residence in the country, phase 2 (exonormative stabilisation) typically involves an increase in the number of settlers and indigenous speakers of English, who both continue to orient themselves towards the norms of the colonial mother country. During phase 3 (nativisation), the consequences of the prolonged contact between English and indigenous languages becomes apparent as the new colonial variety diverges from its ancestor variety. With phase 4 (endonormative sta-bilisation) these linguistic innovations become increasingly accepted, and by phase 5 (differentiation) local dialectal variation emerges. As more and more linguistic innovations emerge and become more and more accepted as a postcolonial variety progresses through the developmental phases of the Dynamic Model, it becomes increasingly different from its ancestor variety (BrE in most cases). For the use of the PP this means that varieties in a later phase in the Dynamic Model might differ to a greater extent from BrE in PP frequency than varieties in an earlier phase.

– Status in the Circle model: Varieties of English can be broadly classified as belonging to the Inner Circle, where the descendants of the settler commu-nity predominate and English is mostly spoken as a native language, and the Outer Circle, where usually the descendants of the indigenous popula-tion predominate and English is spoken as a second language (in addition, in Expanding Circle varieties, not covered in this paper, English is used for international communication only; Kachru 1985). During the development of Outer Circle (but not that of Inner Circle) varieties, second language acqui-sition has typically played an important role. Outer Circle varieties might therefore have lower PP frequencies than Inner Circle varieties because, as a complex category, the PP tends to be acquired late or incompletely in second language acquisition. This might also play a role in Outer Circle varieties that are increasingly spoken as a native language (such as SinE, Gupta 1992) because the consequences of second language acquisition might have taken effect at an earlier point during the development of the variety, as long as it was still spoken as a second language.


The following section explains how the influence of these factors on PP frequency in varieties of English was tested.

3 MethodsBased on GloWbE (Davies and Fuchs 2015), which comprises 1.9 billion words, this paper documents the frequency of the PP in 20 national varieties of English, many of which have so far rarely been studied from a corpus-linguistic perspec-tive. While ICE consists of a carefully balanced selection of different registers of spoken and written language, GloWbE was derived from internet websites. It con-sists of a general section, which is a representative sample of how the English language is used in each of the countries, and a blogs section, that is, texts pub-lished online by individuals to share their personal experiences and opinions. The general section is a representative sample because it consists of texts from quasi-randomly selected websites. Another difference between ICE and GloWbE is that while the national subcorpora of ICE were compiled at different times (starting in the 1990s for some varieties, but still ongoing for others), the national subcorpora of GloWbE were all compiled at the same point in time (December 2012).

3.1 Measuring the frequency of the PP

In the present study, the full text version of GloWbE was searched with regular expressions realised in Python scripts (available from the author upon request). To search for the PP, a regular expression (see Appendix) was applied that searches for have followed by a past participle, and allows intervening adverbials and noun phrases. This search routine was used previously by Hundt and Smith (2009) and Yao and Collins (2012). The former found it to be very reliable, which was confirmed for the present data by spot checks. The search routine includes complex forms of the PP such as the PP passive (1) and the PP progressive (2).

(1) 500,000–700,000 people have been displaced due to the floods. (GloWbE BD, emphasis added)

(2) Tekmira has been working in the field of nucleic acid delivery for over a decade.(GloWbE CA)

230 Robert Fuchs

Furthermore, the frequency of the SP was determined, including complex forms such as the SP passive (3) and the SP progressive (4).

(3) This software was checked for viruses three months ago. (GloWbE IN)

(4) I was waiting patiently for my laptop to arrive from Seattle. (GloWbE NZ)

Since the search routine for the SP inadvertently includes all past perfects, the frequency of the SP was computed by subtracting the frequency of the past perfect (following Yao and Collins 2012: fn. 1). Finally, the relative PP frequency was determined (i) by dividing the absolute PP frequency by the number of words in the respective sections of the corpus and multiplying it with 1 million, yielding the relative frequency of the PP per million words (pmw), and (ii) by dividing the absolute PP frequency by the sum of absolute PP and SP frequency multiplied by 100 (relative PP frequency as a percentage of all past references).

Alternative perfect structures (see Section 1 above) were for the most part not taken into account because they are known or were found to be very rare, com-pared to the PP, and are therefore unlikely to influence the results. The after-per-fect was found to reach its highest frequency in the general section of the IrE subcorpus, with a frequency of 1.7 pmw (uncorrected frequency, including false positives). Although this does not include examples with intervening nouns or adverbials, it still shows that its frequency is negligible compared to the PP (see below). Similarly, Werner (this volume) found the be-perfect to occur with a fre-quency of less than 2.5 pmw in any national subcorpus of GloWbE. The perfect markers done and slam (eWAVE features 104, 105 and 107) are said to be, respec-tively, “pervasive” and “neither pervasive nor rare” only in a number of creoles spoken in some of the countries covered in this paper, as well as in Bahamian English (not covered in this paper). The perfect marker already combines with the SP, and is therefore appropriately covered by the query for the SP. Finally, the medial object perfect is covered by the search routine used here, because it allows intervening noun phrases between the auxiliary and the participle.


3.2 Measuring formality and accounting for the proportion of journalistic texts

As the discussion in Section 2 showed, the PP might be particularly frequent in formal and in journalistic language. In order to account for how formal the national subcorpora of GloWbE are and how much journalistic language they contain, the present study relied on Xiao’s (2009) multidimensional analysis, which is itself an application of Biber’s (1988) model to varieties of English. Spe-cifically, in order to measure formality, Xiao’s factor 1 (informative elaborate dis-course vs. interactive casual discourse; i.e. formality) was measured. Moreover, in order to control for possible differences in the proportion of news-related texts in the national subcorpora of GloWbE (where the PP might be particularly fre-quent, and thus skew the results), Xiao’s factor 9 (attention to reported speech) was measured.² In the multidimensional analysis, each factor is associated with a number of linguistic features. Each linguistic feature has a factor weight that specifies how strongly it contributes to the factor. By measuring the frequency of these features and computing the factor score for each variety, we can compare the national subcorpora of GloWbE with each other with regard to these factors.

For example, factor 9 was determined by measuring the relative frequency (pmw) of two representative linguistic features with high factor weights asso-ciated with this factor in all subcorpora and sections,³ dividing the relative fre-quency by the mean of all subcorpora, and multiplying it with the weight asso-ciated with this feature (see Table 2). For example, public verbs occur with a frequency of 3,714 pmw in the general section of the AusE subcorpus, and the frequency of the SP is 31,224 pmw. Dividing these numbers by the mean relative frequencies for these features across all subcorpora yields an index of 1.06 and 3.83, respectively. Next, multiplying these numbers by their factor weights (taken from Xiao 2009) yields an overall score for attention to reported speech for AusE of 1.06*0.81 + 3.83*0.48 = 2.70. This is a relatively average value, and higher values indicate more attention to reported speech than lower values.

Xiao’s factor 1 (informative elaborate vs. interactive casual discourse) was computed in a similar way to measure the degree of formality/colloquiality in the

2 It should be noted here that the factor attention to reported speech does not account directly for the proportion of journalistic language in the subcorpora. Nevertheless, reported speech is particularly frequent in journalistic language, and this factor permits at least an estimate of the proportion of journalistic language in the subcorpora.3 Only linguistic features with particularly high (positive or negative) factor loadings were used because they contribute more to the factors than linguistic features with a small factor loading. Since Xiao’s factor 9 is associated with fewer linguistic features than factor 1, the analysis in the present paper was restricted to two linguistic features for factor 9, and seven for factor 1.

232 Robert Fuchs

data, relying on seven different features. Higher values indicate relatively interac-tive casual (i.e. informal/colloquial) discourse, lower or negative values indicate relatively informative elaborate (i.e. formal) discourse.

Table 2: Discourse factors and selected features from Xiao (2009) used in the analysis.

Factor Feature Corpus query Factor weight

Factor 1: Informative elaborate discourse vs. interactive casual discourse

Nouns All nouns –0.82Prepositions All prepositions –0.86Private verbs Hope, think, believe, imagine

(and inflected forms)+0.90

Non-past tense verbs All verbs not in the simple past +0.87Contractions Contracted are, is, not +0.84Personal pronouns: first person

I, we +0.83

Personal pronouns: second person

You +0.84

Factor 9: Attention to reported speech

Public verbs Announce, declare, confirm, tell, say (and inflected forms)

+0.81

Past tense All verbs in the past tense +0.48

3.3 The Circle Model and the Dynamic Model of Postcolonial English

The position of each variety in Kachru’s (1985) Circle Model was in most cases straightforward (see Appendix B), except for the following varieties: South African English (SAfE) was classified as being part of the Outer Circle due to its significant number of ESL speakers (see Van Rooy, this volume). However, there are also many ENL speakers of SAfE, so that it could possibly also be said to be in the Inner Circle. Information on which phase each variety has reached in the Dynamic Model was in most cases taken from Schneider (2003, 2007), and in a few cases from other sources (see Appendix B). Although BrE, since it is not a postcolonial variety, can technically not be classified in the Dynamic Model, it seemed desirable to include it in the regression analysis, which meant assigning a phase in the Dynamic Model to it; otherwise it would have had to be removed from the whole regression anal-ysis, which would have meant removing the variety from which most of the other varieties covered in the analysis are derived. For the analysis in Section 4.7, it was therefore classified as being in phase 5, the latest phase, on par with varieties such as AmE and AusE. However, in Section 4.3 BrE was excluded, as this section focuses only on the Dynamic Model, and not on interactions with other factors.


3.4 Geographic proximity

Geographic proximity was accounted for by classifying all varieties according to region: Africa, America, Europe, Oceania, South Asia, and Southeast Asia.

3.5 Linear regression models

Finally, in order to determine which factors influence the frequency of the PP across varieties of English, four linear regression models were computed with the statistical software package R: two models for the general and two for the blogs section of the corpus, with the dependent variables PP frequency pmw, and as a percentage of all past references, respectively. The following independent variables (regressors) were used: region, formality, attention to reported speech, position in the Circle Model, and phase in the Dynamic Model.

4 Results

4.1 Overview

In the general section of the corpus (see Figure 1), the PP is most frequent in KenE (4,815 pmw), followed by BrE (4,774 pmw) and PakE (4,724 pmw). The variety with the lowest frequency is PhiE (3,333 pmw), followed by MalE (3,745 pmw) and SinE (3,752 pmw). The variety with the highest frequency (KenE) has 44 % more PPs than the variety with the lowest frequency (PhiE). AmE has a PP frequency that is neither particularly high or low, but in the lower middle of the spectrum (4,005 pmw).

The results differ to a certain degree between the two sections of the corpora and depending on whether the frequency of the PP is normalised by number of words or as a percentage of all past references (see Figure 2). For example, IndE (12.9 %), and not BrE (12.5 %), is the variety with the second-highest PP frequency, when measured as a percentage of all past references in the general section of the corpus. However, KenE is still the variety with the highest PP frequency (13.1 %) and PhiE the variety with the lowest PP frequency (9.0 %). But overall, results for the different sections of the corpus and the different frequency measures are strongly correlated (correlation between PP pmw and PP percent: Pearson’s r = 0.91 and 0.85 in the general and the blogs section, respectively, p < 0.001 in both cases; correlation between general and blogs section: r = 0.88 for PP pmw, r = 0.80 for PP percent, p < 0.001 in both cases).

234 Robert Fuchs

Afric

aAm

eric

aEu

rope

Ocea

nia

Sout

h-Ea

st A

sia

Sout

h As

ia

0

1000

2000

3000

4000

5000 0

1000

2000

3000

4000

5000

general blogs

KenE

GhaE

NigE

SafE

TanE

Jam

EAm

ECa

nEBr

EIrE

AusE

NZE

HKE

Mal

ESi

nEPh

iEPa

kEIn

dESL

EBa

nE

frequency (pmw)

Figu

re 1:

Rel

ativ

e fre

quen

cy o

f the

PP

(pm

w) b

y reg

ion

in th

e ge

nera

l and

blo

gs se

ctio

n of

the

corp

us.


−50

−250255075

−100

010

020

0

long

itude

latitudePP

pm

w 3600

4000

4400

4800

10111213Pe

rcen

t PP

Figu

re 2

: Rel

ativ

e fre

quen

cy o

f the

PP

pmw

(repr

esen

ted

by si

ze) a

nd a

s a p

erce

ntag

e of

all

past

refe

renc

es (r

epre

sent

ed b

y col

our,

from

lowe

st fr

e-qu

ency

– ye

llow

– to

hig

hest

freq

uenc

y – p

ink)

in th

e ge

nera

l sec

tion

of th

e co

rpus

.

236 Robert Fuchs

4.2 Region/geographic proximity

The factor of region accounts quite well for the differences between varieties. For example, there is a minimal difference in PP frequency between AusE (4,389 pmw; all frequencies in this paragraph refer to the general part of the corpus) and NZE (4,341 pmw), which together make up the region Oceania. This region also has the smallest standard deviation in PP frequency of all regions (34 pmw, com-pared to 397 pmw for all varieties together). America also has a very small stan-dard deviation of only 99 pmw. Furthermore, the factor region separates very well the South-East Asian varieties with their rather low frequencies (PP frequency ranges from HKE with 3,876 pmw to PhiE with 3,333 pmw; standard deviation 237 pmw) from the rest of the field. An exception to this is Europe, where there is a large difference in frequency between BrE (4,774 pmw) and IrE (4,052 pmw). Europe is also the only region with a standard deviation higher than the standard deviation of all varieties together (511 vs. 397 pmw).

4.3 Developmental phase in the Dynamic Model of Postcolonial English

Schneider’s Dynamic Model of Postcolonial English (2003, 2007) posits a five-phase developmental sequence for varieties of English (see Section 2.4 above for details). The varieties considered here are classified as being in phases 3 to 5. Overall, this factor does not account well for differences in PP frequency (see Figure 3). If the developmental phase influences PP frequency, we could expect varieties that are in the same phase to be more similar in PP frequency than those that are in different phases. However, this is not the case. While for phase 5 vari-eties (Standard Deviation [SD] = 200 pmw) and phase 4 varieties (SD = 339 pmw), the standard deviation is still decidedly lower than for all varieties together (SD = 397 pmw), for phase 3 varieties it is much higher (SD = 517 pmw; all frequen-cies refer to the general section of the corpus). In fact, phase 3 includes both the variety with the highest and that with lowest PP frequency, that is, KenE and PhiE, respectively.

Alternatively, if we expect that varieties in later phases have come to rely less and less on BrE norms, but varieties in earlier phases still rely on BrE norms, then varieties in later phases might have PP frequencies that diverge more from BrE than those in earlier phases. However, the data does not support this hypothesis. Varieties in phase 3 differ on average by 624 pmw from BrE, varieties in phase 4 by 494 pmw, and varieties in phase 5 by 625 pmw (all pairwise comparisons not significant [n.s.]).


3 (n

ativ

isat

ion)

4 (e

ndon

orm

ativ

e st

abili

satio

n)5

(diff

eren

tiatio

n)

0

1000

2000

3000

4000

5000 0

1000

2000

3000

4000

5000

general blogs

KenE

GhaE

NigE

TanE

BanE

HKE

Mal

EPh

iEPa

kEIn

dESL

ESa

fEJa

mE

SinE

AusE

NZE

IrEAm

ECa

nE

frequency (pmw)

Figu

re 3

: Rel

ativ

e fre

quen

cy o

f the

PP

(pm

w) b

y dev

elop

men

tal p

hase

in S

chne

ider

’s (2

007)

Dyn

amic

Mod

el o

f Pos

tcol

onia

l Eng

lish

in th

e ge

nera

l and

bl

ogs s

ectio

n of

the

corp

us.

238 Robert Fuchs

Oute

r Circ

leIn

ner C

ircle

0

1000

2000

3000

4000

5000 0

1000

2000

3000

4000

5000

general blogs

KenE

GhaE

PakE

NigE

IndE

SLE

SafE

Jam

ETa

nEBa

nEHK

EM

alE

SinE

PhiE

BrE

AusE

NZE

IrEAm

ECa

nE

frequency (pmw)

Figu

re 4

: Rel

ativ

e fre

quen

cy o

f the

PP

(pm

w) b

y var

iety

stat

us a

ccor

ding

to K

achr

u (1

985)

in th

e ge

nera

l and

blo

gs se

ctio

n of

the

corp

us.


4.4 Classification in Kachru’s Circle Model

Another model of varieties of English is Kachru’s distinction between Inner Circle and Outer Circle varieties (see Figure 4; for details on Kachru’s model, see Section 2.4 above). Mean PP frequency is slightly higher for Inner Circle (4,420 pmw) than for Outer Circle varieties (4,247 pmw). Nevertheless, both groups include varieties with rather high as well as some with rather low PP fre-quencies. The Outer Circle includes KenE and PhiE, the two varieties with the highest and the lowest PP frequency. Also, neither Outer Circle nor Inner Circle varieties are very homogeneous in PP frequency, with rather high standard devia-tions of 448 and 311 pmw (in the general part of the corpus), respectively.

AusE

BanECanE

BrEGhaE

HKEIrE

IndE

JamE

KenE

SLE

MalE

NigE

NZE

PhiE

PakE

SinE

TanE AmESafE

AusE

BanE

CanE

BrEGhaE

HKE

IrEIndEJamE

KenE

SLE

MalE

NigE NZE

PhiE

PakE

SinE

TanE AmESafE

3500

4000

4500

5000

3500

4000

4500

5000

generalblogs

2 3 4informative elaborate vs. interactive casual discourse

frequ

ency

(pm

w)

Figure 5: Relative frequency of the PP (pmw) by Xiao’s factor 1 (informative elaborate vs. interac-tive casual discourse, i.e., formal vs. informal discourse) in the general and blogs section of the corpus. Higher values indicate more informal discourse.

240 Robert Fuchs

4.5 Formality

Apart from these classifications, there are also discourse-related factors that have been suggested to influence PP frequency. Figure 5 shows the relationship between PP frequency and Xiao’s (2009) factor 1 (informative elaborate vs. inter-active casual discourse). The higher the score on factor 1, the more interactive casual or informal a variety is. Overall, there appears to be a small negative correlation between factor 1 and PP frequency (pmw) in the general section of the corpus, suggesting that informal varieties tend to have lower PP frequency. However, the correlation is not significant (r = –0.36, p = 0.12) for the general part of the corpus, nor is it significant in the blogs sections of the corpus, or if PP fre-quency is measured as a percentage of all past references.

AusE

BanECanE

BrE GhaE

HKEIrE

IndE

JamE

KenE

SLE

MalE

NigE

NZE

PhiE

PakE

SinE

TanE AmESafE

AusE

BanE

CanE

BrE GhaE

HKE

IrEIndE JamE

KenE

SLE

MalE

NigENZE

PhiE

PakE

SinE

TanEAmESafE

3500

4000

4500

5000

3500

4000

4500

5000

generalblogs

1.2 1.4 1.6concern with reported speech

frequ

ency

(pm

w)

Figure 6: Relative frequency of the PP (pmw) by Xiao’s factor 9 (concern with reported speech) in the general and blogs section of the corpus. Higher values indicate more concern with reported speech.


4.6 Concern with reported speech

Finally, concern with reported speech (Xiao’s factor 9) might be related to PP fre-quency because national subcorpora with a higher amount of journalistic lan-guage might have higher PP frequencies. There is a weak positive correlation with this factor in the general section of the corpus when measuring PP frequency pmw, which suggests that varieties with greater concern for reported speech have higher PP frequencies (see Figure 6). However, the correlation is not significant (r = 0.39, p = 0.09). In the blogs section, there is a moderately positive and signif-icant correlation (r = 0.49, p < 0.05). However, when measuring PP frequency as a percentage of past references, the correlations with concern with reported speech are very small and insignificant (general section: r = 0.16, p = 0.51, blogs section r = 0.01, p = 0.97).

4.7 Comparison of the influence of all factors

The regression analysis showed that all factors together account for at least half of the variation in PP frequency between all varieties. A greater part of the vari-ation can be accounted for when PP frequency is measured pmw (67 % in the general part of the corpus) than when it is measured as a percentage of all past references (52 % in the general part of the corpus). Furthermore, a greater part of the variation can be explained in the blogs section than in the general section of the corpus (84 % for pmw, 72 % for percentage of all past references).⁴

Not all factors contribute equally to the explanation of the variation in PP fre-quency. Region accounts for 72–89 % of the total variance explained by all factors in three out of four models, only in the blogs section of the corpus and when PP frequency is at the same time measured pmw can it account for only 63 % of the total variance explained. Region is also the only factor that is significant (or close to significance) in all four regression models. As for the other factors, concern with reported speech (“repspeech” in Figure 7) accounts for 18 % (blogs section) and 11 % (general section) of the variance in PP frequency when measured pmw, but only 1 % when it is measured as a percentage of all past references. Formal-ity explains 18 % of the variance in the general part of the corpus when PP fre-quency is measured pmw, but less than 12 % otherwise. Which phase a variety has reached in the Dynamic Model explains 11 % of the variation in PP frequency

4 When only significant factors are included in the models, the percentage of the variation they can explain is somewhat lower (general section pmw 64 %, percent 50 %, blogs section pmw 75 %, percent 57 %).

242 Robert Fuchs

pmw

perc

ent

* **

*

(*) *

0255075100 0255075100

general blogs

regi

onci

rcle

sm

odel

form

ality

reps

peec

hre

gion

circ

les

mod

elfo

rmal

ityre

pspe

ech

fact

ors

% of R2 (variance) explained

Figu

re 7

: Rel

ativ

e im

porta

nce

(per

cent

of t

otal

varia

nce

R² e

xpla

ined

) of a

ll fa

ctor

s in

expl

aini

ng th

e di

ffere

nces

in P

P fre

quen

cy (m

easu

red

pmw

or a

s pe

rcen

tage

of a

ll pa

st re

fere

nces

) bet

ween

all

varie

ties i

n th

e ge

nera

l and

blo

gs se

ctio

ns o

f the

corp

us. A

ster

isks

indi

cate

sign

ifica

nce

of fa

ctor

s (*

* p <

0.01

, * p

< 0.

05, (

*) p

= 0

.15)

.


when it is measured as a percentage of all past references, but only in the blogs section of the corpus. In all other cases, the remaining factors each account for less than 10 % of the variation. Furthermore, apart from region, only the factor attention to reported speech was significant, and only in the blogs section and when PP frequency is at the same time measured as occurrences pmw.

5 Discussion

5.1 Frequency of the PP across varieties

This study has investigated the frequency of the PP across 20 national varieties of English based on data from the internet-derived GloWbE corpus. Two metrics were used to measure PP frequency, (i) the frequency of the PP pmw and (ii) the percentage of the PP compared to all past references. Results indicate that the variety with the highest frequency, KenE, has 44 % more PPs than the variety with the lowest frequency, PhiE. BrE was among the varieties with the highest PP frequencies, while AmE was not among the varieties with the lowest frequencies, but rather in the lower middle range.

The present study provides evidence on the frequency of the PP in varieties which have so far not been studied in this respect. Specifically, MalE was shown to be very close to other Southeast Asian varieties in PP frequency, and JamE to other American varieties. On the other hand, BanE and Tanzanian English (TanE) had a slightly lower PP frequency than all the other varieties in the South Asian and African groups, respectively.

The results confirm in certain respects previous research by Yao and Collins (2012), for example as to the high frequency of the PP in BrE, IndE and KenE, and its low frequency in PhiE. Yet there are also a number of differences. Most notable is perhaps that AmE is not among the varieties at the lower end of the spectrum, which would suggest that it is not “leading the way in the decline” of the PP, as Yao and Collins (2012: 391) concluded, and that BrE and AmE do not form the two extremes of a continuum of PP frequency along which the other vari-eties can be placed. Nevertheless, the presumably increasing American influence on many other varieties of English might have contributed to the comparatively low frequency of the PP in some of these varieties, especially in countries with historically closer ties with the United States, such as the Philippines and South-east Asia in general (Hiang and Gupta 1992: 148–149; Hänsel and Deuber 2013: 350–351; Collins, Borlongan, and Yao 2014: 69).

244 Robert Fuchs

Two methodological differences between the studies might perhaps account for this: First of all, the corpora differ in their composition. The ICE corpus con-sists of carefully balanced subcorpora that contain the same amount of language drawn from various registers. The general section of GloWbE, on the other hand, is a representative sample of internet-based English in the respective countries. Since most Outer Circle countries are multilingual, and English tends to be pre-ferred in the discussion of more formal topics and the vernaculars for less formal topics, the Outer Circle varieties might have higher PP frequencies than Inner Circle varieties in the GloWbE corpus.

However, this is unlikely to explain the differences between Yao and Collins’ (2012) results and the present results: (i) the Outer Circle varieties in the present study did not have higher PP values than the Inner Circle varieties; (ii) the present results differ from Yao and Collins (2012) only for specific Outer Circle varieties, not all of them. Where they differ substantially, for example for PhiE, the Outer Circle varieties have a proportionally lower PP frequency in the present study compared to Yao and Collins (2012), while from the different compositions of ICE and GloWbE the opposite would be expected. And finally, the blogs section of GloWbE provides data from a single register only, and therefore controls for this confounding variable, just as ICE does. Since the results of the present study do not differ substantially between the general and the blogs section of GloWbE, the differences in how GloWbE and ICE are compiled with regard to register variation cannot explain the differences between Yao and Collins (2012) and those of the present study.

A further difference between the corpora is that ICE contains only language produced by educated speakers, that is those who are in possession of a uni-versity degree or currently studying for one. This means that ICE represents the (emerging) local standards of English usage for the respective varieties. GloWbE, on the other hand, does not control for education background. This suggests that GloWbE might include language from less educated language users. In conse-quence, the differences between the present results for GloWbE and Yao and Collins (2012) for ICE might be due to the fact that GloWbE contains both stan-dard-like and non-standard (or less standard-like) language. The lower frequency (compared to Yao and Collins 2012) of the PP in Southeast Asian varieties, and in particular in PhiE, in the GloWbE data would then suggest that non-standard Southeast Asian Englishes have particularly low PP frequencies.


5.2 Explanatory factors

5.2.1 Overview

A regression analysis was used to test in how far various factors mentioned in previous research can account for variation in PP frequencies across varieties. Results showed that region is the only factor that is significant and able to explain the major part of the variation, independently of which PP frequency metric is used and which part of the corpus is considered. This confirms Yao and Collins (2012), who also observed areal clusters of varieties in their data. However, there is scant evidence in the present data for a relationship of formality with PP fre-quency (higher frequency in more formal varieties), also found by Yao and Collins (2012). In the present data, there was a weak general tendency for formal varieties to have higher PP frequencies. However, formality never turned out as a signifi-cant factor in the regression analysis. It accounted for only 13 % of the variation in one regression model, and less than 10 % in the other three. The reason for this difference between the present results and those of Yao and Collins might lie in the fact that the latter used the frequency of contracted auxiliaries in the PP as a measure of formality, while in the present study several linguistic means of expressing (in)formality were used to measure formality (following Xiao 2009: 429–431). In addition, the present study also tested the influence of formality in conjunction with other explanatory factors, which showed that formality con-tributes little to explaining variation in PP frequency across varieties of English. Nevertheless, although the present results suggest that formality may offer little explanatory power if measured at the level of the varieties, future research should explore whether formality can account for variation in PP frequency at the level of individual texts.

Other factors contributed even less to the explanation of variation in PP fre-quency. This is the case for Kachru’s Circle Model (suggested as an explanatory factor by Werner 2014: 359–361 and Yao and Collins 2012: 399–400) and Schnei-der’s Dynamic Model (Werner 2013: 213, 2014: 359–361 and Yao and Collins 2012: 399–400), whose explanatory power was revealed to be negligible for the present data. In certain cases the classification used here might be controversial. For example, SAfE might be classified as an Inner Circle instead of an Outer Circle variety, or IndE might be said to be in phase 3 (Schneider 2007) instead of phase 4 (Mukherjee 2007) of the Dynamic Model. But these modifications are unlikely to make the respective groups more homogeneous in PP frequency, and hence also unlikely to increase the explanatory power of these factors in this respect.

The finding that Kachru’s Circle Model and Schneider’s Dynamic Model account only to a very limited extent, if at all, for variation in PP frequency stands

246 Robert Fuchs

in contrast to previous results by Werner (2013: 213, 2014: 359–361). This might be explained by differences in methodology. While in the present study PP fre-quency was considered, Werner (2013, 2014) took a more fine-grained approach, for example also looking at the co-occurrence of the PP with time adverbials. On the other hand, the result that Kachru’s ENL-ESL distinction fails to account for variation in PP frequency mirrors similar results by Hundt (2009) for the progres-sive passive, and Hundt and Vogel (2011) for the progressive.

Finally, the analysis also revealed concern with reported speech (which was used to measure the amount of journalistic language in the corpus) to be of limited importance in explaining variation in PP frequency. This factor was expected to be important because the amount of newspaper and media language (where the PP is particularly frequent) that is included might differ between the national components of GloWbE. Concern with reported speech was revealed as a signif-icant factor for PP frequency (measured pmw) in the blogs section of the corpus. Given that this was the only case where attention to reported speech was revealed as significant, the evidence suggests that it is not an influential factor. Specif-ically, (i) the significant correlation in the blogs section may be due to the fact that varieties that scored highly on concern with reported speech simply contain more past references. This analysis is corroborated by the fact that concern with reported speech was not correlated with PP frequency if measured as the percent-age of past references. Moreover, (ii) no significant correlations involving concern with reported speech were found in the general section of the corpus, and (iii) the regression analysis showed that concern with reported speech is not a significant factor if it is considered together with all other factors. However, as pointed out above for formality, it is conceivable that taking into account variation in concern with reported speech between individual texts instead of just between varieties might have led to different results.

5.2.2 Geographic proximity

The importance of geographic proximity in explaining variation in PP frequency, revealed by the present analysis, is reminiscent of areoversals, that is linguistic features shared by varieties in geographic proximity (Szmrecsanyi and Kortmann 2009; Hickey 2012; Lunkenheimer 2012). In the case of the PP it is not the feature itself that is shared by a group of varieties in geographic proximity, but the fre-quency with which it is used.

The underlying cause is probably that speakers of varieties in the same region are more likely to come into contact, and migration from one country to another is also more likely if these countries are geographically close. For example, the


late nineteenth and early twentieth centuries witnessed significant migration from Canada to the US (Ramirez 2001). And later, during the Vietnam War, up to 40,000 Americans moved to Canada to avoid having to fight in the war (Knowles 2000). In addition to these population movements, the cultural influence of (mainly) AmE on CanE (which in close geographic proximity is likely to be greater than otherwise) might have contributed to the similarities in PP usage between these varieties. Population movement, and cultural and dialect contact are also likely to account for the similarities between AusE and NZE (Carmichael 1993). In other cases, such as AmE and JamE, or NigE and Ghanain English (GhaE) no sizeable historical population movements are known, and cultural and dialect contact is a probable cause for linguistic similarities. Also, in cases such as AmE and JamE, and NigE and GhaE, linguistic influence is likely to be (predominantly but not exclusively) a one-way street, due to the larger number of speakers of the former varieties, as well as the economic weight of their countries’ economies (Mair 2013).

South Asia may be a special case, since IndE, PakE and BanE have all de-scended relatively recently from the same variety, as all three countries belonged to British India until they were split after independence in 1947. Since then, there has been close contact and migration between India, Bangladesh and Sri Lanka. India has a larger economy and population than the other countries, and previ-ous studies have provided evidence for the linguistic influence of IndE on other South Asian and Southeast Asian varieties (Hoffmann, Hundt, and Mukherjee 2011; Parviainen 2012, 2016; Bernaisch, Gries, and Mukherjee 2014; Fuchs 2015a; Fuchs and Gut 2016; Gries and Bernaisch 2016; Parviainen and Fuchs 2016).

The analysis showed that geographic proximity can explain a major part of cross-varietal variation in PP frequency, which nevertheless left a smaller but still sizeable portion of variation unexplained. While it may be futile to expect any analysis of linguistic variation to explain all differences, there appear to be clear cases of varieties in close geographic proximity that diverge in PP frequency. For example, KenE and TanE have the highest and lowest PP frequency, respectively, of all African varieties considered here, although both countries are direct neigh-bours. There is also a sizeable difference between BrE and IrE, in spite of consider-able migration and cultural contact during the history of these countries. Further-more, in South Asia, PakE and BanE also differ to a certain degree in PP frequency.

One explanation for the remaining variation might be that the system of mac-ro-regions adopted for the present analysis is too crude to account for all areal effects. For example, Pakistan and Bangladesh are the two South Asian coun-tries separated by a larger distance than any of the other varieties considered here. Using geographic distance rather than macro-regions might improve the accuracy of the analysis. However, even this probably does not fully account

248 Robert Fuchs

for the strength or weakness of cultural and dialect contact. For example, this contact might have historically been, and perhaps still is, greater between AmE and CanE than between KenE and TanE. Such variation in the strength of cultural and dialect contact between countries might in the end be difficult to account for in a quantitative analysis.

5.2.3 National identities

Another important factor might consist of national identities and how they potentially influence how likely it is that dialect contact leads to the diffusion of linguistic features. In Schneider’s (2007) dynamic approach, the strength of emerging national identities is said to explain how emerging dialect features become accepted and reinforced through (national and ideological) indepen-dence from the colonial mother country. In a similar way, local identities might also strengthen diverging linguistic patterns. This might be the case for BanE and PakE. After independence, Bangladesh and Pakistan were a single country (then called East and West Pakistan, respectively), and Bangladesh only became inde-pendent from Pakistan after a civil war. In a related case, Ireland only became independent from the United Kingdom after decades of bitter and violent conflict. This historical heritage might partly account for the difference in PP frequency between IrE and BrE. Future research should seek to identify methods on how geographic proximity, migration, and local identities can be integrated into a quantitative model that will allow a refinement of the analysis provided here.

5.2.4 Substrate transfer

Yet another factor that might explain some of the remaining differences between varieties is substrate influence. Transfer from local languages that lack a struc-ture that is similar to the PP might be responsible for lower PP frequencies in these varieties. For example, Filppula (1997) argued that transfer from Irish led to the emergence of the “extended-now” perfect (i.e. be + past participle) in IrE, thus offering an alternative to the PP when talking about events that extend to the present. Since Werner (this volume) showed that the “extended-now” perfect occurs very rarely in the Irish subcorpus of GloWbE, this alternative perfect structure cannot explain why the frequency of the PP is lower in IrE than in BrE. However, it is conceivable that, for example, the present tense is used more fre-quently in IrE in contexts where in other varieties the PP would be used.


By comparison, transfer from local languages that have a structure similar to the English PP might explain why some varieties have a high PP frequency. IndE might be such a case, and Davydova (2011) found a certain amount of evidence for substrate transfer from Hindi that might explain why IndE is indeed among the varieties with a high PP frequency in the present study.

Because of a lack of systematic analyses of the substrate languages of all the varieties investigated here, substrate transfer could not be considered in the quantitative analysis of the present paper. Future research will hopefully provide comprehensive analyses on which varieties of English are spoken in countries with L1s that have structures that are similar to the PP. Such studies will also have to take into account that most Outer Circle varieties are spoken in multi-lingual countries. For example, Hindi is spoken natively by less than half of the Indian population, and the twelve most widely spoken Indian languages account together for slightly less than 90 % of the Indian population (Fuchs 2016: 12–14). Not all of these languages might, like Hindi, have a structure similar to the PP, and future analyses of substrate transfer need to take this into account. Another crucial point when considering substrate transfer is that while a certain language might have a structure that is similar to the PP, it might not be used in all contexts where the English PP occurs (see Davydova 2011).

5.2.5 Influence from a heterogeneous superstrate

In addition to influence from the substrate, influence from the superstrate might also be a source of the variation in PP frequency that we find across varieties of English today. According to the “founder principle” (Mufwene 1996: 83; see also Trudgill 2004), the first sizeable English-speaking population (at least several thousand people) of a new variety has a greater influence on the structure of this new variety than speakers that arrived later. Since the frequency of the PP in BrE has changed over time (Elsness 1997), the input to new varieties founded at differ-ent points in time is likely to have differed. More research is required to determine when exactly the first sizeable nucleus of immigrants arrived in the respective countries, but it seems safe to assume that there are considerable differences. In the case of IndE, this period is likely to have taken place at the end of the eight-teenth and beginning of the nineteenth century (Fuchs 2015b). In AmE, it is likely to have taken place at a much earlier point.

Overall we would expect varieties that were established earlier to have retained a higher PP frequency than varieties that were established later. From the present data, this is not readily observable: For example, AmE has a lower PP frequency than IndE. Nevertheless, it is still conceivable that when other factors

250 Robert Fuchs

(geographic proximity, substrate influence, etc.) are taken into account, variation in the time when varieties were founded might emerge as a significant factor.

Another cause of variation might be that not all settlers spoke BrE. An obvious case in point is PhiE, which had hardly any input from BrE, but from AmE. The analysis did indeed show that PhiE has a very low PP frequency. In fact, it is even lower than in AmE, so that input from AmE cannot have been the only, but is likely to be an important factor. Many former colonies have also received large numbers of Irish immigrants. Input from IrE, with a PP frequency lower than in BrE, might then have contributed to a lower PP frequency in the new varieties, in turn. This is the case for example for AmE (Crozier 1984), SinE (Davyodva 2013) as well as AusE and NZE (Engel and Ritz 2000; Yao and Collins 2012). Before the influence of a heterogeneous superstrate can be taken into account in a quan-titative analysis as in the present paper, estimates of the number of speakers of different input varieties that arrived during the foundation period are required.

5.3 Implications

The differences in PP usage revealed by the analysis have implications for the use of English as a Lingua Franca across the world and for teaching English as a second or other language. First of all, it seems unlikely that variation in the use of the PP as revealed by this study has an adverse impact on the international com-prehensibility of English and its use as a lingua franca (on the difference between intelligibility and comprehensibility, see Smith and Nelson 1985). Using the PP where a listener expects the SP (or vice versa) might lead to the impression that the utterance is somewhat ‘odd’, but a misinterpretation of its truth-conditional meaning seems very unlikely.

A related question is in how far variation in the use of the PP should be taken into account in English language teaching. The answer will depend on who is taught English and for what purpose.

Those who are taught English mainly for communication within their country (as might be the case for many learners in Inner and Outer Circle countries) would be best served by a focus on the local norms of English usage. Thus, English lan-guage teaching in Singapore and the Philippines, where the PP is used compar-atively rarely, should focus on local norms of English usage, which tend towards a low frequency of the PP. By contrast, English language teaching in Britain and Kenya should put a greater focus on teaching the use of the PP, as it is used par-ticularly frequently in these varieties.

Second, those who learn English exclusively for international communica-tion, such as learners of English in Japan or Germany (countries where English


is taught as a foreign language, see Section 2.4 above and Kachru 1985) will benefit from an approach that teaches the use of the PP in prototypical contexts (for prototypical contexts, see for example Werner 2013: 211–213, 2014: 365–374). However, beyond the use of the PP in such contexts it seems unnecessary to teach towards the norms of varieties with a moderately high (such as AmE) or a very high PP frequency (such as BrE). For learners of English as a foreign language, the PP is a complex and particularly challenging category (Klein 1995: 44–46; Bardovi-Harlig 2000: 169–175; Davydova 2011, 2012; Fuchs, Götz, and Werner, this volume). Given that the time and attention of language learners is limited, it seems wasteful to use this time to teach towards norms (such as in AmE or BrE) that go beyond the use of the PP in prototypical contexts and are very difficult to achieve for many learners. Once learners have reached a stage where they use the PP in prototypical contexts, teachers would also benefit from an approach that accepts that some learners may use the PP more often than other learners.

Third, those who want to use English both for communication within their country and internationally would be best served by (i) learning local norms of PP usage and (ii) simultaneously developing an awareness of worldwide variation in PP usage. In particular, this awareness might help speakers anticipate whether and when their own norms of English usage might seem odd (in the sense of the word used above) to their interlocutors, and if a certain convergence with the language norms of their interlocutor is a viable and appropriate communicative strategy for them.

In summary, how the PP is approached in English language teaching will depend on the needs of the learners. Local norms should be taught for local com-munication, and an international minimal standard of prototypical PP usage for international communication.

6 ConclusionDrawing on data from the 1.9 billion word GloWbE corpus, this paper presented evidence on the frequency of the PP across 20 national varieties of English. Geo-graphic proximity emerged as the most important factor, and explained a major part of the variation. Other factors such as the developmental phase of the vari-eties or the degree of formality appeared to be of little or no importance. Future research should attempt to gather quantitative information on the degree of cul-tural contact between varieties, the structure of the input during the foundation period and potential substrate transfer to account for the remaining variation.

252 Robert Fuchs

AcknowledgementsI would like to thank two anonymous reviewers and the audience at the ICAME 35 pre-conference workshop “Perfect and perfectivity re-assessed through corpus studies” for their valuable comments on earlier versions of this paper.

ReferencesAnderson, Lloyd B. 1982. The ‘perfect’ as a universal and as a language-specific category.

In Paul J. Hopper (ed.), Tense-aspect: Between semantics and pragmatics, 227–264. Amsterdam: Benjamins.

Bardovi-Harlig, Kathleen. 2000. Tense and aspect in second language acquisition: Form, meaning and use. Oxford: Wiley-Blackwell.

Bernaisch, Tobias, Stefan Th. Gries & Joybrato Mukherjee. 2014. The dative alternation in South Asian English(es): Modelling predictors and predicting prototypes. English World-Wide 35(1). 7–31.

Biber, Douglas. 1988. Variation across speech and writing. Cambridge: Cambridge University Press.

Carey, Kathleen. 2005. Subjectification and the development of the English perfect. In Dieter Stein & Susan Wright (eds.), Subjectivity and subjectivisation: Linguistic perspectives, 83–102. Cambridge: Cambridge University Press.

Carmichael, Gordon Alexander (ed.). 1993. Trans-Tasman migration: Trends, causes and consequences. Canberra: Australian Government Public Service.

Collins, Peter, Ariane M. Borlongan & Xinuye Yao. 2014. Modality in Philippine English: A diachronic study. Journal of English Linguistics 42(1). 68–88.

Crozier, Alan. 1984. The Scotch-Irish influence on American English. American Speech 95(4). 310–331.

Davies, Mark & Robert Fuchs. 2015. Expanding horizons in the study of World Englishes with the 1.9 billion word Global Web-Based English Corpus (GloWbE). English World-Wide 36(1). 1–28.


Davydova, Julia. 2012. Englishes in the Outer and Expanding Circles: A comparative study. World Englishes 31(3). 366–385.

Davydova, Julia. 2013. Detecting historical continuity in a linguistically diverse area: The present perfect in modern Singapore English. In Joana Duarte & Ingrid Gogolin (eds.), Linguistic super-diversity in urban areas: Research approaches, 193–225. Amsterdam: Benjamins.

Engel, Dulcie M. & Marie-Eve A. Ritz. 2000. The use of the present perfect in Australian English. Australian Journal of Linguistics 20(2). 119–140.



Elsness, Johan. 2009a. The present perfect and the preterit. In Günter Rohdenburg and Julia Schlüter (eds.), One language, two grammars? Differences between British and American English, 228–245. Cambridge: Cambridge University Press.

Elsness, Johan. 2009b. The present perfect and the preterite in Australian and New Zealand English. In Pam Peters, Peter Collins & Adam Smith (eds.), Comparative studies in Australian and New Zealand English, 89–114. Amsterdam: Benjamins.

Filppula, Markku. 1997. The influence of Irish on perfect marking in Hiberno-English. In Jeffrey L. Kallen (ed.), Focus on Ireland, 51–71. Amsterdam: Benjamins.

Fuchs, Robert. 2015a. Colonial lag or feature retention in postcolonial varieties of English: Evidence from South Asian Englishes and beyond. Unpublished manuscript, University of Münster.

Fuchs, Robert. 2015b. The acoustic realisation of light and dark laterals in Educated Indian English. Unpublished manuscript, University of Münster.

Fuchs, Robert. 2016. Speech rhythm in varieties of English: Evidence from educated Indian English and British English. Berlin: Springer.

Fuchs, Robert & Ulrike Gut. 2016. Register variation in intensifier usage across Asian Englishes. In Heike Pichler (ed.), Discourse-pragmatic variation and change: Insights from English, 185–210. Cambridge: Cambridge University Press.

Gries, Stefan Th. & Tobias Bernaisch. 2016. Exploring epicentres empirically: Focus on South Asian Englishes. English World-Wide 37(1). 1–25.

Gupta, Anthea F. 1992. The pragmatic particles of Singapore colloquial English. Journal of Pragmatics 18(1). 31–57.

Hänsel, Eva Canan & Dagmar Deuber. 2013. Globalization, postcolonial Englishes, and the English language press in Kenya, Singapore, and Trinidad and Tobago. World Englishes 32(3). 338–357.

Hiang, Tan Chor & Anthea F. Gupta. 1992. Post-Vocalic /r/ in Singapore English. York Papers in Linguistics 16. 139–152.

Hickey, Raymond (ed.). 2012. Areal features of the Anglophone world. Berlin: Mouton de Gruyter.

Hoffmann, Sebastian, Marianne Hundt & Joybrato Mukherjee. 2011. Indian English – an emerging epicentre? A pilot study on light-verbs in web-derived corpora of South Asian Englishes. Anglia 12. 258–280.

Huber, Magnus. 2014. Stylistic and sociolinguistic variation in Schneider’s Nativization Phase. In Sarah Buschfeld, Thomas Hoffmann, Magnus Huber & Alexander Kautzsch (eds.), The evolution of Englishes: The Dynamic Model and beyond, 86–106. Amsterdam: Benjamins.

Hundt, Marianne. 2009. Global feature – local norms? A case study on the progressive passive. In Thomas Hoffmann & Lucia Siebers (eds.), World Englishes – Problems, properties and prospects, 287–308. Amsterdam: Benjamins.

Hundt, Marianne & Carolin Biewer. 2007. The dynamics of inner and outer circle varieties in the South Pacific and East Asia. In Marianne Hundt, Nadja Nesselhauf & Carolin Biewer (eds.), Corpus linguistics and the web, 249–269. Amsterdam: Rodopi.

Hundt, Marianne & Katrin Vogel. 2011. Overuse of the progressive in ESL and learner Englishes – fact or fiction? In Joybrato Mukherjee & Marianne Hundt (eds.), Exploring second-language varieties of English and learner Englishes: Bridging a paradigm gap, 145–165. Amsterdam: Benjamins.


254 Robert Fuchs

Kachru, Braj B. 1985. Standards, codification and sociolinguistic realism: The English language in the outer circle. In Randolph Quirk & Henry Widdowson (eds.), English in the world: Teaching and learning the language and literatures, 11–30. Cambridge: Cambridge University Press.

Klein, Wolfgang. 1995. The acquisition of English. In Rainer Dietrich, Wolfgang Klein & Colette Noyau (eds.), The new handbook of second language acquisition, 502–522. Bingley: Emerald Group.

Knowles, Valerie. 2000. Forging our legacy: Canadian citizenship and immigration, 1900–1977. Government of Canada. http://www.cic.gc.ca/English/resources/publications/legacy/index.asp (accessed 08 March 2015).

Kortmann, Bernd & Kerstin Lunkenheimer (eds). 2013. The electronic world atlas of varieties of English. Leipzig: Max Planck Institute for Evolutionary Anthropology. http://www.ewave-atlas.org (accessed 10 March 2015).

Logghe, Sanne. 2014. Language variation and change: A closer look at varieties of English and the use of regularized irregular verbs. Gent: University of Gent MA thesis.

Lunkenheimer, Kerstin. 2012. Tense and aspect. In Raymond Hickey (ed.), Areal features of the Anglophone world, 329–353. Berlin: Mouton de Gruyter.

Mahmood, Muhammad A. 2009. A corpus based analysis of Pakistani English. Multan: Bahauddin Zakariya University dissertation.

Mair, Christian. 2013. The world system of Englishes: Accounting for the transnational importance of mobile and mediated vernaculars. English World-Wide 34(3). 253–278.

Mufwene, Salikoko S. 1996. The founder principle in creole genesis. Diachronica 13(1). 83–134.Mukherjee, Joybrato. 2007. Steady states in the evolution of new Englishes: Present-day Indian

English as an equilibrium. Journal of English Linguistics 35(2). 157–187. Mukherjee, Joybrato. 2008. Sri Lankan English: Evolutionary status and epicentral influence

from Indian English. In Klaus Stierstorfer (ed.), Anglistentag 2007 Münster: Proceedings, 359–368. Trier: WVT.

Parviainen, Hanna. 2012. Focus particles in Indian English and other varieties. World Englishes 31(2). 226–247.

Parviainen, Hanna. 2016. Use of the invariant tag isn’t it in Asian Englishes. World Englishes 35(1). 98–117.

Parviainen, Hanna & Robert Fuchs. 2016. Indian English as a super-central variety: Diffusion of clause-final focus particles in Asian Englishes. Unpublished manuscript, University of Tampere.

Ramirez, Bruno. 2001. Crossing the 49th parallel: Migration from Canada to the United States, 1900–1930. Ithaca: Cornell University Press.

Schneider, Edgar W. 2003. The dynamics of New Englishes: From identity construction to dialect birth. Language 79(2). 233–281.



Smith, Larry E. & Cecil L. Nelson. 1985. International intelligibility of English: Directions and resources. World Englishes 4(3). 333–342.

Szmrecsanyi, Benedikt & Bernd Kortmann. 2009. Vernacular universals and angloversals in a typological perspective. In Markku Filppula, Juhani Klemola & Heli Paulasto (eds.),


Vernacular universals and language contacts: Evidence from varieties of English and beyond, 33–56. London: Routledge.

Trudgill, Peter. 2004. New-dialect formation: The inevitability of colonial Englishes. Edinburgh: Edinburgh University Press.




Xiao, Richard. 2009. Multidimensional analysis and the study of World Englishes. World Englishes 28(4). 421–450.


Appendix A: Corpus queries (applying the CLAWS C7 tagset)(Format: One line per word form, with word form, lemma, and parts of speech tag separated by tabs)

Linguistic Feature Query (Regular Expression)

Time reference forms

Present perfect (vh0|vhz)(_\w+)?\n(\w+\t\w+\t(r\w+|md|xx)(_\w+)?\n){0,4}(\w+\t\w+\t(d\w+|at|appge)(_\w+)?\n){0,1}(\w+\t\w+\t(j\w+|n\w+\nn1)(_\w+)?\n){0,1}(\w+\t\w+\t(pp\w+|d\w+|n\w+)(_\w+)?\n){0,2}(\w+\t\w+\t(r\w+|md|xx)(_\w+)?\n){0,4}(\w+\t\w+\t(vbn|vhn|vvn|vvnk))The first part of this query ("vh0|vhz") targets a form of HAVE, and the last part a participle ("vbn|vhn|vvn|vvnk"). The intervening search terms allow for a limited number of adverbials and noun phrases between the auxiliary and the participle.

Simple past v.d.?Past perfect vhd(_\w+)?\n(\w+\t\w+\t(r\w+|md|xx)(_\w+)?\n){0,4}(\w+\t\w+\t(d\

w+|at|appge)(_\w+)?\n){0,1}(\w+\t\w+\t(j\w+|n\w+\nn1)(_\w+)?\n){0,1}(\w+\t\w+\t(pp\w+|d\w+|n\w+)(_\w+)?\n){0,2}(\w+\t\w+\t(r\w+|md|xx)(_\w+)?\n){0,4}(\w+\t\w+\t(vbn|vhn|vvn|vvnk))This query is identical to the one for the PP except that the first part targets had.

After-perfect be\t\w+(_\w+)?\nafter\tafter\t\w+(_\w+)?\n\w+ingThis query searches for a form of BE plus after plus a word ending in ing.

256 Robert Fuchs


Multi-dimensional analysis

Nouns \n[a-z,A-Z,0-9,-,']+\t[a-z,A-Z,0-9,-,']+\tn[a-z,A-Z,0-9,-,']+Verbs \n[a-z,A-Z,0-9,-,']+\t[a-z,A-Z,0-9,-,']+\tv[a-z,A-Z,0-9,-,']+Private verbs (know|think|want|believe|like|love|feel|mean)\tvPublic verbs (announce|declare|confirm|tell|say)\tvPrepositions \n[a-z,A-Z,0-9,-,']+\t[a-z,A-Z,0-9,-,']+\ti[a-z,A-Z,0-9,-,']+Non-past tense verbs v?0|(a|A)m\tbe|(a|A)re\tbe|(i|I)s\tbeContractions 's\tbe|n't\tn't|'re\tbePersonal pronouns: First person

I\ti|(w|W)e\twe

Personal pronouns: second person

(y|Y)ou\tyou

Appendix B: Independent and dependent variablesVariety Abbreviation Region Circle (Kachru) Model (Schneider)

American E. AmE America IC 5 (Schneider 2007)Canadian E. CanE America IC 5 (Schneider 2007)British E. BrE Europe IC 5Irish E. IrE Europe IC 5 (Werner 2014: 135)Australian E. AusE Oceania IC 5 (Schneider 2003)New Zealand E. NZE Oceania IC 5 (Schneider 2003)Indian E. IndE South Asia OC 4 (Mukherjee 2007)Sri Lankan E. SLE South Asia OC 4 (Mukherjee 2008)Pakistani E. PakE South Asia OC 4 (Mahmood 2009)Bangladesh E. BanE South Asia OC 3 (Logghe 2014: 23)Singapore E. SinE South-East Asia OC 4 (Schneider 2003)Malaysian E. MalE South-East Asia OC 3 (Schneider 2003)Philippine E. PhiE South-East Asia OC 3 (Schneider 2003)Hong Kong E. HKE South-East Asia OC 3 (Schneider 2003)South African E. SAfE Africa OC 4 (Schneider 2007)Nigerian E. NigE Africa OC 3 (Schneider 2007)Ghanaian E. GhaE Africa OC 3 (Huber 2014)Kenyan E. KenE Africa OC 3 (Schneider 2007)Tanzanian E. TanE Africa OC 3 (Schneider 2007)Jamican E. JamE America OC 4 (Schneider 2007)


Appendix C: Frequencies of PP, SP and other linguistic features in GloWbE (full text version)General section

Varie

tyW

ords

PP (abs

)PP (p

mw)

SP (abs

)SP (p

mw)

PP

(per

cent

)Af

ter-p

erfe

ct

(abs

)(In

)For

mal

ityCo

ncer

n fo

r re

porte

d sp

eech

AusE

1047

1636

645

9600

4389

3269

713

3122

412

.377

3.1

1.33

BanE

2870

0158

1184

9841

2989

2813

3110

811

.723

1.6

1.23

CanE

9084

6732

3599

2939

6229

5595

032

538

10.9

562.

61.

23Br

E25

5672

390

1220

640

4774

8532

871

3337

412

.520

83.

01.

35Gh

aE27

6447

2113

0363

4716

9113

8332

968

12.5

121.

61.

42HK

E27

9068

7910

8154

3876

8488

5830

418

11.3

151.

71.

06IrE

8053

0794

3262

7540

5227

0569

933

598

10.8

143

2.5

1.18

IndE

6803

2551

3092

3545

4520

8853

030

699

12.9

571.

91.

34Ja

mE

2850

5416

1183

2041

5110

2268

135

877

10.4

231.

91.

38Ke

nE28

5529

2013

7495

4815

9096

9831

860

13.1

381.

61.

26SL

E33

7937

7214

6511

4335

1163

668

3443

411

.241

1.2

1.28

Mal

E29

0268

9610

8706

3745

8768

0230

207

11.0

332.

81.

30Ni

gE30

6227

3814

0631

4592

1008

758

3294

112

.232

2.7

1.42

NZE

5869

8828

2547

8443

4118

3562

631

272

12.2

452.

71.

17Ph

iE29

7584

4699

188

3333

1001

747

3366

39.

023

2.9

1.43

PakE

3800

5985

1795

3947

2412

5015

032

894

12.6

381.

61.

65Si

nE29

2291

8610

9691

3753

8656

8029

617

11.2

303.

31.

14Ta

nE24

8838

4010

1379

4074

8037

3332

299

11.2

151.

51.

15Am

E25

3536

242

1015

425

4005

8233

181

3247

311

.017

73.

61.

33SA

fE31

6832

8613

2162

4171

9950

5231

406

11.7

222.

21.

20

258 Robert Fuchs

Blogs sectionVa

riety

Wor

dsPP

(a

bs)

PP

(pm

w)SP

(a

bs)

SP

(pm

w)PP

(p

erce

nt)

Afte

r-per

fect

(a

bs)

(In)F

orm

ality

Conc

ern

for

repo

rted

spee

ch

AusE

4339

0501

2061

4047

5113

4492

730

996

13.3

523.

81.

23Ba

nE10

9228

6943

195

3955

3273

1629

966

11.7

221.

91.

13Ca

nE43

8148

2718

5943

4244

1408

713

3215

211

.725

3.7

1.25

BrE

1316

7100

265

9417

5008

4113

844

3124

313

.812

13.

61.

19Gh

aE11

0881

6054

668

4930

3822

1934

471

12.5

111.

91.

57HK

E12

5087

9646

841

3745

3689

3329

494

11.3

72.

61.

08IrE

2041

0027

9066

344

4266

4913

3257

812

.025

2.9

1.16

IndE

2831

0511

1228

0943

3879

5746

2810

813

.416

2.5

1.11

Jam

E11

1242

7347

833

4300

4017

7036

117

10.6

132.

71.

58Ke

nE12

4807

7761

454

4924

4072

1932

628

13.1

192.

61.

35SL

E12

7607

2657

852

4534

4438

0234

779

11.5

71.

71.

25M

alE

1335

7745

5124

938

3741

2108

3085

211

.114

3.2

1.28

NigE

1199

6583

5718

247

6741

3963

3450

712

.16

2.9

1.61

NZE

2262

5584

1072

0047

3872

1681

3189

712

.915

3.3

1.33

PhiE

1345

7087

4697

934

9141

6434

3094

510

.114

3.6

1.17

PakE

1333

2245

6222

646

6741

7520

3131

713

.09

2.0

1.42

SinE

1371

1412

5190

037

8542

1709

3075

611

.019

3.7

1.15

TanE

1025

3840

4371

942

6434

7284

3386

911

.214

1.9

1.33

AmE

1330

6109

357

6262

4331

4133

449

3106

412

.298

4.5

1.36

SAfE

1364

5623

6048

744

3340

8947

2996

912

.913

3.1

1.20

Valentin Werner9 Rise of the undead? be-perfects in World

EnglishesAbstract: While the be-perfect (BEP) is found in a number of (particularly Euro-pean) languages, in English it has almost exclusively been considered from a dia-chronic perspective, and is commonly seen as a dying structure in present-day varieties. In contrast, previous acceptability and corpus studies have indicated that the BEP persists as a formal variant. Against the backdrop of this apparently conflicting evidence, in this paper I take a closer look at this “zombie” structure. With the help of data from the International Corpus of English (ICE) and the Corpus of Global Web-based English (GloWbE), I aim at updating and sharpening the syn-chronic perspective on the BEP. In particular, I (re-)address whether the BEP has really “died” or rather still represents a productive pattern in present-day variet-ies of English. In this connection, lexical restrictions are considered as important indicators, and the role of different factors favoring the BEP as well as its poten-tial status as a vernacular universal are discussed.

1 Introductionbe-perfects (BEPs) are commonly found in many European languages, both Ger-manic (e.g. Icelandic, German and Swedish) and Romance (e.g. Italian, French and Spanish). In contrast, English occupies a typologically special position, as this “once robust competitor” (Davydova 2011: 146; cf. Anderson 1982: 228) of the have-perfect (HP) seems to have sunk into oblivion in present-day usage (Kytö 1997: 33). This is also evident from the fact that the BEP in English in linguistic analyses has almost exclusively been considered from a diachronic perspective, and, from a contemporary viewpoint, has variously been labelled “receding” (Tagliamonte 2000: 347), “rare” (Schneider 2013: 159), “almost dead” (Ander-son 1982: 232), “marginal” (Yerastov 2012: 427), “sporadic” (Davydova 2011: 142), “archaic” (Denison 1993: 359), “being in decline” (Brinton 1994: 159), a “frozen reflex” (Yerastov 2015: 165) or a “relic” (Anderson 1982: 232). This assessment of the status of the BEP is further reflected in the electronic World Atlas of Varieties of English (eWAVE; feature 102), where high (that is, A or B) pervasiveness ratings

Valentin Werner, University of Bamberg

260 Valentin Werner

are observable for just 9 out of the 76 varieties covered (Kortmann and Lunken-heimer 2013).

This apparent consensus has been challenged by recent acceptability and corpus studies. These have indicated that the BEP persists as a formal variant in a number of varieties of English (see also Werner 2013, 2014), such as Irish (Siemund 2004) and Canadian English (Yerastov 2012, 2015), in addition to Asian Englishes (Davydova 2011; Suárez-Gómez and Seoane 2013) as well as smaller regional dialects (see, e.g., Tagliamonte 2000: 331).

Against the backdrop of this apparently conflicting evidence, my aim is to take a closer look at the abovementioned “zombie”. To this end, I use data from the International Corpus of English (ICE) and the Corpus of Global Web-based English (GloWbE; Davies 2013; Davies and Fuchs 2015). My main aim is to update and sharpen the perspective on the BEP from a synchronic point of view, and, in particular, to (re-)address these points:

– Has the BEP really “died” or does it still represent a productive pattern in present-day varieties of English?

– To what extent is the BEP lexically restricted and indistinguishable from genuine passives, as has been stated in earlier literature (e.g. Quirk et al. 1985: 170)?

– What role do substrate and learner effects play in the preservation of this form or should it rather be seen as a vernacular universal?

To provide a general framework, I recapitulate historical and typological aspects with regard to the BEP (Section 2), followed by an outline of findings from pre-vious analyses of the BEP in present-day varieties of English (Section 3). Next, I explain the data and methodology used (Section 4). In the core part of the study, I present quantitative and qualitative results emerging from the corpus analysis with a special focus on the topic of lexical restriction (Section 5), before pro-ceeding to a discussion of potential motivations for the persistence of the BEP (Section 6). In the final part (Section 7), I summarize and contextualize the find-ings, and establish some avenues for future research.

2 Historical and typological aspectsAs the BEP in English linguistics has largely been viewed from a diachronic per-spective (which is also reflected in the amount of historical literature available), I present a synopsis of the findings emerging from relevant studies. This also paves the way for the corpus analysis of the present investigation. To this end,

Rise of the undead? be-perfects in World Englishes 261

I offer a summary overview of the main aspects, as more detailed descriptions are available elsewhere (see, e.g., Wright 1986: 113–165; Faiß 1989: 298–300; Kytö 1997; McFadden and Alexiadou 2010: 394–406). In addition, I outline a number of typological aspects to highlight the particular position of the English BEP com-pared to other languages.

As to the typological perspective, it is important to note that BEPs (as well as HPs) are widely found in European languages, where perfect constructions involving auxiliary/copular verbs are pervasive (Dahl 1995: 20). Drinka (2003: 3–6) has further claimed a geographical division between Eastern and Central Europe (predominantly BEP) and Western Europe (predominantly HP). In any case, in many of the European languages, these perfects have developed into general pasts. There are also languages with a mixed system (e.g. Danish, French, Italian, German), where BEPs are basically restricted to use with mutative intran-sitive verbs, that is, those verbs that imply motion or a change of state (Denison 1993: 344; Kytö 1997: 17; Siemund 2004: 422; Dahl and Velupillai 2011).

Diachronic studies suggest that this functional split also applies to a large extent for older stages of English.¹ Commentators agree that intransitive muta-tives associated almost exclusively with the BEP, and that competition between the HP and the BEP was marked as both constructions had grammaticalized into anterior time reference markers in the Middle English period (Faiß 1989: 298; Denison 1993: 359; Shannon 1995: 147–150; Tagliamonte 2000: 335–336; Halas 2012: 224; Yerastov 2012: 430–431). The process of the HP supplanting the BEP is thought to have begun in the Middle English period already, and lasted until the end of the nineteenth century (Kytö 1997: 19; Halas 2012: 228).

In this development, frequency effects also played a part. It has been argued that the high type frequency of the HP was responsible for the extension of this paradigm to intransitive mutatives, the original domain of the BEP. At the same time, frequency effects were in place in terms of a high token frequency that led to the persistence of the BEP pattern with particular items (e.g. of the type he is retired, the sun is set, she is come, they’re gone; Yerastov 2012: 429–430, 2015: 175).

The alternative account presented by McFadden and Alexiadou (2006, 2010; cf. Roy 2014: 51–54; Yerastov 2015: 169) establishes that the HP and the BEP should not merely be viewed as variants of a general category “perfect” (where one retreats due to the advancement of the other), but rather as two separate constructions. In other words, only the HP is a true anterior, while the BEP is

1 There is limited evidence for the BEP with transitive verbs in Middle English (Yerastov 2012: 427) and Scots (McCafferty 2014: 338–339; Yerastov 2015: 174). Note that further auxiliaries be-sides have and be can be found with the perfect when we expand the time frame to Old Saxon (see Watts 2001: 125).

262 Valentin Werner

“a copular construction built around a stative resultative participle” (McFadden and Alexiadou 2006: 275). Their line of argumentation is the following: Both the HP and the BEP originally derived from resultative constructions (see also Johannsen, this volume). While the BEP has remained restricted to a resultative interpretation (and to intransitives), the HP has acquired “additional functions” (McFadden and Alexiadou 2006: 277) in the course of time. One of these func-tions is the “experiential perfect” interpretation, which was in competition with the simple past (but not the BEP). This may be one explanation for why the HP extended its territory to all kinds of verbs, eventually also affecting the ones usually associated with the BEP. The diachronic corpus data used in their studies support this view as BEP frequencies remained relatively stable until the end of the Early Modern period (MacFadden and Alexiadou 2010: 422). According to this construal, the generalization and expansion of the HP is only an indirect cause of the retreat of the BEP and took effect later (starting c. 1700).²

However, what unites all accounts is the outcome in the sense of a periph-eral position of the BEP in present-day English, namely in its use with a severely limited set of either intransitives or transitives. This allegedly marginal status is underscored by the practice of labeling relevant examples as “pretty much archaic” (Denison 1993: 344) or “pseudo-passives” (Quirk et al. 1985: 169). The latter label also reflects the finding that the BEP was reanalyzed as be + adjec-tive (< be + past participle), which clearly suggests a resultative state reading and illustrates the change from the original mutative association (Denison 1993: 361–363; Brinton 1994: 160).

2 Brinton (1994: 159–160) provides a comprehensive overview of further individual linguistic and sociolinguistic factors that may have motivated the replacement of the BEP through the HP, a major one among them being the functional overload that be (as an auxiliary used both in passive and perfect constructions) had to carry (see also Kilpiö 1997: 115; Halas 2012: 230; cf. Section 6.1.2). Censure of the BEP by prescriptive grammarians of the eighteenth and nineteenth century may have also played a part (see Faiß 1989: 298 and Anderwald 2014 for an in-depth account). Another prominent factor is the leveling between is and has (>’s), which may have led to reanalysis as a reduced from of has in all environments (Yerastov 2012: 433). However, this scenario only applies for this particular pair of realizations (and maybe was) but not for other forms of be (am, are, were). See Section 6.1.1 for further discussion.


3 Previous studies on the BEP in varietiesIn spite of the purportedly weak status of the BEP (see Section 2), previous re-search has suggested that the BEP persists in varieties of English, particularly in a number of non-standard ones (see, e.g., Faiß 1989: 300; Kytö 1997: 19; Taglia-monte 2000: 338; Smith and Durham 2011: 208–209).³ This indicates that the view of the BEP as a mere historical variant is too narrow. In what follows, I offer an overview of relevant studies to flesh out this general claim.

The different surface realizations available in perfect contexts in Irish English (IrE) have received extensive scholarly attention (see, e.g., Kallen 1989; Filppula 1999, this volume; Siemund 2004; Davydova 2008; McCafferty 2014). Although the origin of the BEP in this variety has its origin in older stages of English, it has been hypothesized that its persistence is supported by Irish substrate influence (Siemund 2004: 410–411). It has been found that the BEP in contemporary IrE reflects the pattern outlined above (Section 2), as it characteristically applies for intransitive verbs and carries a resultative meaning, as in example (1).

(1) Is Ron gone away(ICE-IRE S1A-087)

However, as strong lexical restrictions apply, the BEP is not seen as a productive structure of IrE any more (Filppula 1999: 120), a finding that has been confirmed by other corpus studies (e.g. Schneider 2013: 159).

Similar restrictions (again reflecting the historical pattern) are in place in other varieties. In Samaná English (a variety spoken by black immigrants from the US in the eponymous peninsula of the Dominican Republic), for instance, the BEP persists with mutative and stative verbs (Tagliamonte 2000: 344). Even more restrictions apply in Canadian English (CanE) and a number of related American dialects. One special case is the construction be + done/finished/started + NP. Although this structure, which arguably is a reflex of the productive BEP of the Scots dialects of Shetland and Orkney, as shown in (2), is a transitive construc-tion, it shares the resultative reading with the intransitive BEP variant (Yerastov 2015: 174).

(2) I’m seen it(Orkney and Shetland English; ewave-atlas.org/sentences/4102)

3 For remarks on the anterior marker be(n)/bin/been in English-based pidgins and creoles, which are not treated in this paper, see Patrick (1999: 171) and eWAVE feature 111 (Kortmann and Lunkenheimer 2013) for its regional distribution, for instance.

264 Valentin Werner

However, lexical choice is severely limited as only three options (done/finished/started) are available, one of them (started) being only marginally accepted. In addition, animate subjects are clearly preferred over inanimate ones, and further constraints apply for the object. Among these constraints is a clear preference for definite determiners and high frequency nouns, and the semantic fields from where the object nouns may derive are largely limited to everyday duties and household work (Yerastov 2015: 158), as illustrated in (3).

(3) I am finished my homework(CanE; from Yerastov 2015: 157)

Due to these constraints, Yerastov rejects the view of the transitive construction as found in CanE as a mere variant of the (present) perfect (which is fully sche-matic; see also Section 2), and rather proposes a “semantic/syntactic niche” (Yerastov 2012: 446) that is covered by the construction. He further concludes that it is best analyzed as a “prefab” or “idiosyncratic construction” where some parts are mandatory (be + done/finished/(started)), while others (subject, object) are variable (Yerastov 2012: 428, 2015: 176). Taking into consideration this analy-sis and Roy’s (2014: 54) finding that no BEP occurs in his modern corpus data, it is evident that the productivity of the BEP and related structures is restricted in contemporary CanE.

Turning to non-native varieties of English, the picture distinctly changes. Recent studies have found variation between the HP and the BEP (and further forms) in both L2 and learner varieties (Davydova 2011: 91; Suárez-Gómez and Seoane 2013: 167; Werner 2014: 327–328; Seoane, this volume). In contrast to the varieties described above, it is striking that the BEP is not restricted to intransitive and mutative verbs, as shown in (4) and (5).

(4) After 1947 we do everything, we are free, we are defined our liberty and equal-ity, so it’s different.(basilectal IndE; from Davydova 2011: 207)

(5) And he’s come up with this check list of things he is given to his students(ICE-HK S1A-053)

It is evident that different explanations are needed to account for the presence of these forms. Two major factors play a part: On the one hand, they could be seen as remnants of the colonizers’ dialects (Werner 2014: 328) that were, for instance, introduced through the language used in missionary schools (Davydova 2013: 197). On the other hand, substrate influence and learner effects may have


an impact (Davydova 2013: 195). This is particularly likely when a perfect struc-ture that uses an equivalent of be as an auxiliary is transferred (as may happen particularly with learners of English with a Germanic language background; see Section 2), or when a substrate language with an equivalent of be that functions at least as a marker of time reference is present (as is the case with Hindi as a poten-tial substrate of Indian English, for instance; Davydova 2011: 208, 308; Seoane 2014: 231–233). The latter factor may reinforce the former, which also provides a motivation for the fact that the BEP is extended to other contexts beyond resulta-tive mutatives in present-day varieties. Thus, also given that the BEP may occur in writing (Suárez-Gómez and Seoane 2013: 176–177), the main difference with respect to the L1 varieties is that it represents a productive pattern (see further Section 6.2).

In sum, although the status of the BEP as an overall productive pattern has been questioned, as illustrated above, there is converging evidence for the claims that the BEP is still used as an alternative to the HP. This applies in a range of ver-naculars “of contemporary English in North America, in the United Kingdom and undoubtedly elsewhere” (Tagliamonte 2000: 338). As variety type appears to be a decisive factor, in the following, I cast the net wider to focus on the “elsewhere”, and broaden the perspective to worldwide varieties of English.

4 Data and methodologyThe present study relies on two corpus resources, which are quite different in their structure and content. One is the long-established International Corpus of English (ICE), which has been used for a large number of comparative and variety-specific studies in the field of World Englishes. The other corpus is the recently compiled Corpus of Global Web-based English (GloWbE).

ICE represents a family of regional corpus components of both L1 and L2 varieties of English with a size of approximately one million words each (60 % spoken, 40 % written material). The intention of the ICE project is to have as many synchronic components as possible that are maximally comparable as regards text types and sociolinguistic status of the informants. To this end, there is a fixed scheme for the text types included, and all informants are adults with at least a completed formal English-medium secondary-school education (see Nelson 1996: 28–33 and ice-corpora.net/ice/design.htm). In its current shape, ICE comprises twelve completed components (Australia, Canada, East Africa [Kenya and Tanza-nia], Great Britain, Hong Kong, India, Ireland, Jamaica, New Zealand, Nigeria, the Philippines, Singapore), and two components for which only the written sections

266 Valentin Werner

are available (Sri Lanka and the US). A dozen other components are being com-piled (see ice-corpora.net/ice/index.htm).

In terms of size, GloWbE is much larger as it comprises approximately 1.9 billion words. As its name implies, it is global as it contains material from 20 dif-ferent countries where L1 or L2 English is spoken.⁴ Moreover, it is fully web-based; that is, the corpus data all stem from randomly selected, publicly available elec-tronic communication of various kinds. GloWbE also maps genre distinctions in that it includes informal blogs (60 %) as well as “general” material (40 %), such as texts from company, newspaper and magazine websites (Davies and Fuchs 2015: 4; see also Fuchs, this volume). Due to its size, it is particularly suitable for searching for low-frequency items and structures, such as the BEP.

As GloWbE is a fairly recent addition to the resources available for the study of worldwide varieties of English, some general methodological considerations should be borne in mind (see also Mair et al. 2015). While the compilation of ICE involves a considerable amount of fieldwork and requires decisions on which texts and informants to include to ensure the comparability among its compo-nents, GloWbE exploits the opportunities offered by the wide availability of texts on the internet. Arguably, compared to ICE, GloWbE also presents data in an unfiltered and thus more “direct” way, which may allow additional insights into properties of (particularly informal) electronic communication.

Notwithstanding the great care that has been taken to ensure the correct selection and categorization of the websites in terms of both regional prove-nance and genre (see Davies and Fuchs 2015: 3–5), a note of caution applies, as the “large and dirty” approach used for GloWbE unavoidably leads to some inac-curacies that may influence results. The following emerged in the course of the searches for the present study (on which see further below):

– The method developed for determining regional provenance (see Davies and Fuchs 2015: 4–5) may fail at times.

– Although the regional provenance of a website as such may be assigned cor-rectly, there is no way of controlling the provenance of people that contrib-ute, for instance in a comment section or a blog.⁵

4 The countries are Australia, Bangladesh, Canada, Ghana, Great Britain, Hong Kong, India, Ireland, Jamaica, Kenya, Malaysia, New Zealand, Nigeria, Pakistan, Singapore, South Africa, Sri Lanka, Tanzania, the Philippines, and the US. In contrast to ICE, the word count for the indi-vidual national domains is not equal, but rather maps the proportions of actual usage of the individual varieties on the web (Davies and Fuchs 2015: 5).5 Oddly enough, in one case a miscategorization of a freely available file from ICE India (S1A-015) that became part of GloWbE as Jamaican occurred: So even there the difference is there but uh they are identified themselves with American language (GloWbE JAM G; ice-corpora.net/ice/downloads/S1A-015.TXT).


– As noted by the compiler himself, it was difficult to exclude blogs from general searches (Davies and Fuchs 2015: 4). Although the URL clearly sug-gests otherwise, blogs may appear as “general”.

– Occasionally, examples may appear both as “blog” and “general”. – Due to copy-and-paste, examples may appear for two (or more) countries,

and it may be hard to determine which one is the original. – A minor point is the fact that among the corpus findings there may occur

didactic EFL material that deliberately contains mistakes, for instance in dis-tractors in a multiple choice task.

This list suggests that the information coming with the examples should not always be taken at face value and that the researcher has to be aware of poten-tial limitations of the available corpus material, which for practical reasons was obtained through a largely automatic process.

The following aspects were relevant for the actual corpus searches. One user-friendly feature of GloWbE is that searches for part-of-speech (POS) tags are pos-sible through its online interface (corpus.byu.edu/glowbe), which was used for the present study. The corpus material was POS-tagged with the help of the C7 tagset of the CLAWS tagger (ucrel.lancs.ac.uk/claws/). For the ICE data, which commonly are available in plain text format, POS-tagging was necessary as an intermediate step (also using CLAWS), before searches with WordSmith Tools (Scott 2011) were possible. Following Kytö (1997: 28), I looked for tokens which carry a “perfectivity” reading. I took a stepwise approach and first queried ICE for combinations of be and past participle (*_VB* + *_V*N within one slot to the right).⁶ It goes without saying that due to the formal congruence between BEP and passive forms many false positives appeared, and extensive manual disambigu-ation was necessary.⁷ Eventually, 67 verb types remained (including the stative resultative “relics” come and go and related phrasal and prepositional verbs, see (6) and (7)), and a further 13 verbs that co-occur with the BEP as mentioned in the previous literature were added for the search in GloWbE (see the Appendix).

(6) […] not one civil servant is come back though(ICE-HK S1B-049)

6 As the following slot to the right was not specified this also returned passive forms of the type he is been given credit, which were included.7 Forms where ’s appeared were excluded both for the ICE and GloWbE searches as the POS-tag-ging was unreliable and the ambiguity of the contracted form between an interpretation as either is or has could not always be resolved.

268 Valentin Werner

(7) […] which is gone down by forty-two percent(ICE-IND S1B-054)

Next, I conducted a POS-based search in GloWbE ([vb*] + past participle), with the 80 verb types already identified in the past participle slot. Due to the large size of GloWbE, there were often too many hits which would have required manual disambiguation. Where this was the case, the built-in sampling function of the web interface was applied. A second query used was a modification of the search string ([vb*] + past participle + [at*]|[nn*]|[j*]), in order to restrict the results to transitive uses, which are of particular interest for some aspects treated in the present study (see Section 5.2). Still, for the same reason as noted above, manual disambiguation was indispensable and the vast majority of the returned tokens had to be discarded. In addition to the exclusion of unwanted passive forms, occurrences of the following type had to be avoided: First, tokens where the syn-tactic structure produced the false appearance of a BEP, as in (8).

(8) This anti-gravity force was completely new to science, but again what it actu-ally was remained a mystery.(GloWbE USA G)

Occasionally, typos led to tagging errors and the identification of false positives, as exemplified in (9) to (12).

(9) Around 10 o’clock, thirteen people, all of them Rastafarians, got on his patio that was looked [locked] and protected with a nest. (GloWbE JAM G)

(10) I am read [ready] today to fulfill that promise!(GloWbE BAN B)

(11) The fact that is [it] happened on our own doorstep is the most upsetting thing and there will probably more of this to come. (GloWbE IRE G)

(12) This is the principal behind was [what] happened right at the beginning of the universe.(GloWbE GHA G)

Sometimes, the tagging of adjectives (ending in -ed) as past participles led to the identification of false positives that had to be excluded, as in (13) and (14).


(13) Is that because we expect panopticons everywhere so soon that they’re settled, ancient, unquestioned facts by 2512?(GloWbE USA G)

(14) It’s obvious that you’re failed, miserable, pathetic little people.(GloWbE USA G)

Another category to be excluded were citations exemplifying historical usage, such as the archaic transitive use of suffer [‘allow’], as in example (15).

(15) the poor lie in the streets upon pallets of straw… and are suffered to die in the streets like dogs or beast without any mercy or compassion(GloWbE GB B)

I also excluded tokens where potentially some kind of self-correction or false start (and missing deletion of the unintended variant) took place in the process of text composition (as in (16)), a feature that is more expected in spoken discourse.

(16) Through reading a book a boy in our world goes on a journey with a warrior in the story who is has been given the task to destroy the nothing which is con-suming Fantasia

(GloWbE NZ G)⁸

Note that the search at times returned occurrences which stem from pages with apparently randomly generated textual content, which is sometimes used to test the actual outcome of a particular style sheet (see, e.g., blogs.rediff.com/ aliplyfer1970/2012/07/02/computer-it-does-exciting-users-that-it-is-otherwise- potential-towards-the-case/ and bestbrandshere21.inube.com/blog/1454212/application-the-document-of-the-repair-is-that-the-pay-of-channelizing-a/), which also could not be used in the analysis.

As a final step, all relevant tokens were stored in spreadsheets along with information on their provenance (variety, register). Each of the 1,468 examples (114 from ICE, 1,354 from GloWbE) was then manually coded for voice (active, passive), verb semantics, transitivity, subject animacy, object definiteness, and whether the form could be interpreted as invariant in the sense of being poten-tially used like a V-ing form (see further Section 7), as in (17).

8 See also It is has been given to us by grace (GloWbE NZ G) and More than 615,000 customers were remained without power Tuesday morning (GloWbE USA B). Examples of this type exemplify the (in these cases incomplete) monitoring in this discourse type.

270 Valentin Werner

(17) they’re looked so stunning and elegant as you have them in you.(GloWbE IND B)

5 ResultsIn the following, I present the inferences that can be drawn from the ICE and GloWbE corpus data. The review of previous studies of the BEP (see Section 3) in essence leads to the following hypotheses, which serve as the backdrop against which the findings of the corpus data are read:

– H1: The BEP no longer represents a productive pattern of English (at least in L1 varieties and the majority of traditional regional dialects).

– H2: If relics of the BEP can be found in present-day data, for historical reasons, they only occur with a small set of mutative intransitive verbs.

– H3: The BEP hardly ever occurs with transitive verbs (where it does, this is only with a highly restricted set of verbs or in L2 and learner varieties of English).

H1 and H2 suggest that the BEP is lexically restricted, while H3 in addition raises the question whether the BEP with transitive verbs should be viewed as an inno-vation or a continuation of an older BEP pattern. These hypotheses as well as pos-sible motivations are tested as I proceed. I start with a few quantitative aspects to provide an overview of the structure of the data, and then discuss the important issue of lexical restriction, addressing the three hypotheses as I go along.

5.1 Quantitative aspects

It is clear that the BEP represents a rare structure in quantitative terms. However, a look at the BEP frequencies across the set of 20 varieties contained in GloWbE, as shown in Figure 1, reveals a diversified picture. Due to the varying size of the sections for the individual varieties in GloWbE, normalized frequencies are used.

Above all, the BEP is a structure that appears in all the varieties covered to some extent. There is considerable variability in the frequencies, with the BEP being seven times as frequent in BanE (21.2 per 10 million words) as in SAfE (3.1 per 10 million words), the variety where the BEP is least commonly found. Another general trend that emerges is that the vast majority of the L1 varieties, with IrE as an exception, can be found toward the lower end of the frequency


scale (around 5 per 10 million words or less). The special position of IrE can be explained by the fact that BEP use in this variety has been much more pervasive than in others (see Section 3; see also eWAVE feature 102). IrE can be grouped with a number of L2 varieties (GhaE, PhiE, JamE) that cover the middle ground (around 7–10 per 10 million words). A similar trend, though less robust, can be established for AusE, the L1 variety with the second highest BEP frequency. According to eWAVE (feature 102), the BEP enjoys some currency at least in vernacular AusE, so a slightly increased frequency could be expected. In contrast, a set of other L2 varieties (TanE, KenE, SinE, SAfE), most of them African, associate with the L1s. The most striking finding emerging from Figure 1, though, is the group of seven L2 varieties with comparatively high BEP frequencies (BanE, PakE, NigE, IndE, HKE, SLE, MalE), which are at least triple the value as for the L1 varieties (13.9–21.2 per 10 million words). The members of this cluster are all Asian varieties (NigE being an exception), so that an areal pattern appears likely. Moreover, this finding gives

0 5 10 15 20 25

SAfEAmEBrENZE

SinECanEKenEAusETanEJamEPhiE

GhaEIrE

MalESLEHKEIndENigEPakEBanE

Figure 1: Normalized BEP frequencies (per 10 million words) in varieties of English (based on GloWbE); grey coloring indicates L1 varieties.

272 Valentin Werner

quantitative substance to previous claims of the salience of the BEP in these vari-eties (see Section 3).⁹

Further general aspects that deserve attention are semantics and transitivity.

0%

10%

20%

30%

40%

50%

60%

70%

action result state

GloWbE

ICE

Figure 2: Semantic categories of the main verb in BEP constructions.

As to semantics, it is clear from Figure 2 that result and state readings, as in (18) and (19), are dominant. Note that verbs with a genuine action interpretation often carry a resultative connotation (see inform in (20)), though this is not always the case, as shown for explode in (21).

(18) I feel better in a way since she is made it clear and I have accepted that.(GloWbE USA G)

(19) I’m been an amateur astronomer for 10+ years built several fully operational telescopes the last being a computer guided 16’’ dob.(GloWbE AUS G)

9 As suggested by one of the reviewers, it may be worthwhile to draw a further subdistinction between South Asian and Southeast Asian varieties, and to recognize different waves of influ-ence from IndE – which arguably can be seen as a “super-central variety” (Mair 2013) – on the remaining varieties from the two areas. To be in a firm position to establish such connections, we would require diachronic data, or at least apparent-time data, however.


(20) However Shri Suvarna is informed the house that due to some personal problem Shri Gopal Mulgaon had gone out of station (ICE-IND S1B-078)

(21) Unless floors were been exploded in the collapse to help the collapse?? (GloWbE USA G)¹⁰

These findings suggest that resultative state still represents the prototypical semantic interpretation that is assigned to the BEP, and this is also what could be expected in the light of its historical development (see Section 2). Note that differences in the distributions of the categories between ICE and GloWbE are significant (χ2 = 8.18, df = 2, p < 0.05), but amount to a very small effect size only (φc = 0.08).¹¹ Therefore, a unified analysis is presented.

As regards transitivity, two apparently conflicting findings have to be recon-ciled. The “relic” view of the BEP presupposes that it is an intransitive structure. In contrast, transitive uses of the BEP have been found for CanE and a number of other varieties of English (Yerastov 2012, 2015).

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

GloWbE ICE

intransitive

transitive

Figure 3: Proportion of transitive and intransitive BEP tokens in GloWbE and ICE.

10 See Section 7 and Hundt (2016) for further discussion of be been constructions.11 All statistical calculations were conducted with R (R Core Team 2014). The function assoc-stats (package vcd) was employed to obtain Pearson’s chi-square and Cramer’s V (effect size φc) values.

274 Valentin Werner

As Figure 3 shows, in the present data, intransitive tokens are more frequent in both ICE (56.1 %) and GloWbE (52.3 %). The proportion of transitives is slightly higher in GloWbE, but the difference between the two corpora does not reach the level of statistical significance and the effect is small (φc = 0.05). However, what is striking against the backdrop of previous statements that the BEP is typically intransitive, is the finding that transitive uses, such as (22) or (23), amount to more than 40 % in both corpora, even approximating half of all tokens in GloWbE.

(22) And once again he he is started uh his apple orchard and raspberries and potatoes strawberries […] (ICE-CAN S1A-041)

(23) As a Masters student and with a full time job, I still look back at my undergrad-uate days and think – ‘I am made the right choices.’(GloWbE KEN B)

A more fine-grained look at the individual varieties yields considerable variation, as displayed in Figure 4.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

intransitivetransitive

GhaENigE IrE

JamENZE

TanECanE

KenE BrEAmE

AusEMalE

SAfE SLE HKEPhiE

IndEBanE

SinEPakE

Figure 4: Proportion of transitive and intransitive BEP tokens in varieties (GloWbE).


It is worth noting that a group of Asian varieties (MalE, SLE, PhiE, IndE, BanE, SinE, PakE) clusters toward the right hand side of Figure 4, which implies that the proportion of transitive BEPs is high in this group (54 % or more of all tokens). All L1 varieties and the remaining varieties have values below 50 % instead (SAfE, the variety with the lowest overall BEP frequency – see Figure 1 above – being an exception). The difference between the two larger groups is significant (χ2 = 103.11, df = 1, p = 0) and yields a moderate effect (φc = 0.27). Therefore, it is fair to say that the transitive pattern may be seen as an established feature of Asian varieties. At the same time, the data suggest that transitive uses – in addition to the ones already described – in general (including L1 varieties) are more frequent than indicated by previous analyses.

5.2 Lexical restriction

The low BEP frequencies for the L1 varieties found above may be interpreted as corroborating quantitative evidence for claims that the BEP is no longer produc-tive (see sections 2 and 3) in varieties of this type. This view is confirmed if the number of different verbs that occur with the BEP and the corpus size for the individual varieties in GloWbE are related.

Figure 5 shows that BEP range coefficients of the L1 varieties can be found toward the lower end of the range. SAfE, the variety with the overall lowest BEP frequency (see Figure 1), aligns. Note that the picture emerging from Figure 5 is neither just a mirror image of corpus sizes nor of the ranking of BEP frequencies (cf. Figure 1).¹² In other words, it could be supposed that a larger corpus size nat-urally leads to an increased number of verb types (and therefore to an increased verb type number for the BEP). This may be true in absolute terms,¹³ so it is nec-essary to control for corpus size (as is done for the BEP range coefficient). At the same time, it is evident that a qualitative analysis is indispensable to account for these absolute differences and to obtain a complete picture of productivity for the individual varieties. Above all, examples such as (24) or (25) suggest that a restric-tion of the BEP to the relic structure (BEP with come and go) falls short.

12 Tests for the Spearman rank correlation yield a moderate value for the comparison to the BEP frequencies (rs = 0.65) and even a negative correlation for the comparison with corpus sizes (rs = –0.38).13 For instance, the BrE part of GloWbE, which is the largest overall, has 38 different verb types, while TanE, the smallest, has 18 different verb types. In contrast, the IndE part, which ranks sixth in terms of absolute corpus size, is the part with the highest number of BEP verb types (41).

276 Valentin Werner

0 1 2 3 4 5 6 7 8 9

AmEBrE

CanEAusESAfENZE

IrESLE

PakEIndE

KenESinEJamEPhiE

GhaETanEBanEMalENigEHKE

Figure 5: BEP range coefficients ([number of verb types/corpus component size]*107) for GloWbE; grey coloring indicates L1 varieties.

(24) So, was it YouTube’s business profit which created the crucial situation that is happened today or some other technique?!? (GloWbE BAN G)

(25) a final guiding light is to be certain that you’re increased daily water con-sumption.(GloWbE HK B)

A closer look at the association of verb types with variety types reveals an inter-esting pattern.

Above all, Figure 6 shows that almost half of the verb types (n = 38) appear in both L1 and L2 varieties. While there is only a small set of verbs (n = 6) which exclusively appear in L1 varieties, the number of those verbs that occur mostly in either L1 or L2 is equal (n = 7). The most striking result, however, given the smaller word counts of the L2 parts of GloWbE, is the comparatively high number of verbs


that appear in L2 varieties only (n = 22). This result provides a first indication for a wider range of possible lexical choices in BEP contexts in L2 varieties (see the Appendix for the categorization of the individual items). It further widens the per-spective gained in Davydova (2011: 206–208), where a broader lexical range has been found for the BEP in IndE. Yet, the high number of verbs in the “both” cate-gory also suggests that the traditional view of the BEP being restricted to mutative intransitives (mainly come and go; see sections 2 and 3) may be too narrow for the present data. In what follows I offer a contrastive view of the situation in L1 and L2 varieties.

The high number of verb types that appear in L2 varieties is reflected in exam-ples (26) to (28).

(26) I hope he don’t know I’m turned out o’ the old place.(GloWbE PHI G)

(27) So even there the difference is there but uh they are identified themselves with American language(ICE-IND S1A-015)

(28) We are split the tasks of the entrepreneur into two or three tasks. (GloWbE GHA G)

L1 only mostly L1 both mostly L2 L2 only 0

5

10

15

20

25

30

35

40

Figure 6: Association of the 80 verb types with variety type (ICE+GloWbE); category “mostly” assigned if at least 75 % of occurrences of this verb type appeared with either L1 or L2.

278 Valentin Werner

The L1 data yield representative examples of the traditional pattern (cf. Davydova 2011: 146), such as (29) or (30), which are also prototypical in that they carry a state/resultative reading (see Section 5).

(29) I think they’re gone a bit out(ICE-IRE S1A-082)

(30) I am come here this day, and would not change my lot with the greatest in the world.

(GloWbE NZ G)¹⁴

Actually, in the L1 data, we also find verbs other than from the restricted set of mutatives, as (31) to (36) illustrate, while the state/resultative interpretation typ-ically persists.

(31) This is happened at a clinic that did not bulk bill.(GloWbE AUS G)

(32) In a unanimous decision, the Supreme Court of Canada are placed the con-ventional bifurcated approach of analyzing discrimination as being ‘direct’ or ‘adverse effect’ discrimination with a unified approach.(GloWbE CAN G)

(33) I love how you’re arranged the shelves -- thanks for inspiring me to empty my dresser and go for a totally new look.(GloWbE USA B)

(34) Under Barack Obama, the national Debt was exploded from $9 billion to now over $16 billion.(GloWbE IRE B)

(35) they’re been out a lot during the week(ICE-NZ S1A-013)

(36) I’m sorry I’m stopped reading after that, because that one line was the most pig headed thing I’ve ever read.(GloWbE USA B)

14 Although they appear in the web data, these biblical examples support the “relic” view of the BEP.


Furthermore, the quantitative analysis above implied that transitive uses of the BEP can be found to various degrees. While these uses are much more common in L2 varieties (see (37) to (39)), they are by no means restricted to this variety type, as (40) to (42) show.

(37) Cloud Computing Companies are changed the way of our businesses.(GloWbE IND B)

(38) All this because I’m found tat study on what we’re interested is much more easier than the subjects we unlike(GloWbE MAL G)

(39) Any one who was studied Buddhism (If I am not talking about prejudiced Hindu oriented scholars) can see that there is a major paradigm shift between Hinduism and Buddhism, in fact, between all other religious systems and Bud-dhism.(GloWbE BAN G)

(40) I’m made a lot of money across my career and I am no stranger to hard times, so I don’t write as someone who’s only lived one type of existence.(GloWbE USA G)

(41) They’re chock full of goodness and we’re made them for a long, long time.(GloWbE CAN B)

(42) Bill Gates was made the decision the team requires a new operating-system that may hand out billions and don’t miss it. (GloWbE GB B)

At this stage, a reconsideration of the related “Canadian” transitive BEP pattern (be + done/finished/(started) + NP), as described in Yerastov (2012, 2015; see Section 3) is in order. This is exemplified in (43) and (22) above.

(43) We are started a program with 25 devices (20 iTouch/5 iPads) at a school for children with special needs.(GloWbE CAN G)

Unlike the examples analyzed up to this point, this construction is restricted to these three specific lexical items, and Yerastov identifies “the conserving effect of high token frequency” (2012: 433) as a reason for this situation. When the find-

280 Valentin Werner

ings for the “Canadian” pattern are compared against the present data, a number of differences emerge (other restrictions being similar, such as the preference of the subject for animate referents, the preference for items from the semantic fields of education and household work in the object slot, or the validity of the “exhaustivity constraint”; cf. Yerastov 2015: 158).

Above all, it is noteworthy that comparable patterns emerge in other L1 vari-eties, for instance, IrE, AusE, and NZE, as illustrated in (44) to (46). BrE and AmE examples are rarer, but do occur, as (47) and (48) demonstrate.

(44) I’ve been reading it for nearly a year & I will miss it when I’m finished it.(ICE-IRE W1B-015)

(45) Umm when I’m finished uni he can get a dog.(GloWbE AUS G)

(46) I’m finished week 1 but I definitely saw a change.(GloWbE NZ B)

(47) closed means literature will be inaccessible once I’m done a PhD if I can’t immediately secure a long-term academic appointment (GloWbE GB G)

(48) With serialized entertainment, waiting is painful because you’re started a plotline that’s not gonna be finished for 6 months or more.(GloWbE USA G)

As was shown with examples (40) to (42) above, however, other transitive verbs in addition to do, finish, and start may also occur, so the lexical limitation does not seem to apply for the present L1 data. Surprisingly, similar patterns occasionally occur in L2 varieties (SAfE, PhiE, NigE, JamE, SinE, HKE, KenE, IndE), as shown in (49) to (51).

(49) hi, i am done mba in sales and marketing after b.com and right now i am working as a marketing manager kindly guide me whats next courses will i do.(GloWbE IND G)

(50) I am on Facebook multiple times a day, I Pinterest everything, and I am started Twitter when I made my blog, so still new to it. (GloWbE PHI B)


(51) So, the project is almost due, as I am finished the poster already. (GloWbE HK B)

As there is no reason to suppose that these specific instances represent examples where the source was miscategorized (in terms of a Canadian speaker appear-ing on an Indian webpage, for example), it seems hard to find a straightforward explanation for this. Provided that L2 varieties in general allow for many more transitive contexts (see Section 5) and therefore, for many more verb types to appear in these (see Figure 5), it seems likely that the extended lexical scope of the BEP in L2 varieties covers the “Canadian” uses, but goes well beyond this restricted set.

That this may indeed be the case can also be seen from the fact that some of the items that were viewed as unacceptable or only marginally acceptable (see Section 3; cf. Yerastov 2012: 448), such as (52) to (56), appear in the data (see also (38) above). This applies both in CanE ((52) is a structure of the type be done a Adj N) as well as in other L1 and L2 varieties.

(52) I’ll write more about the topic soon, once I’m done a brief bit of travelling(GloWbE CAN G)

(53) For me this is quite easy, he is made it clear he wants to wait 5years, for someone in his/her 30’s its a long time to wait i must say.(GloWbE NIG B)

(54) I’m made a playlist of the best videos that I found on YouTube called Running for Beginners.(GloWbE MAL G)

(55) It is true, unless you purport that millions of intelligent and diligent people that SOME scientists are made it all up and live in denial ever since.(GloWbE USA G)

(56) We’re reduced that dependence to ‘only’ 80 % of all our energy in the last 50 years.(GloWbE AUS B)

Consequently, if we extend the envelope of variation and take further varieties – both of the L1 and L2 type – into account, the evidence presented suggests that the schematicity of the pattern may not be as reduced and the effect of colloca-tional strength not as strong as asserted previously (cf. Yerastov 2012: 434, 448).

282 Valentin Werner

In addition, the fact that also non-telic verbs may appear in transitive BEP forms (see (37) and (41) above for non-telic/continuative instances) sheds some doubt on the claim that the construction as such “constrains the inherent lexical seman-tics of the verbs” (Yerastov 2015: 172) when viewed from the broader perspective. If this major constraint is relaxed, the question re-arises whether the treatment of the BEP (versus the HP) as a mere case of alternating auxiliaries (pace Yeras-tov 2015) is to be preferred over the one as a self-contained “semantic/syntactic niche” (Yerastov 2012: 446) construction.

Overall, the assessment of lexical (and other) constraints of the BEP indicates that the BEP is restricted in quantitative terms in L1 varieties (if corpus size for the individual varieties is considered). In fact, a qualitative analysis suggests that the BEP in the present data occurs with many more verbs than the “relic” forms (come and go). An even larger number of verb forms that are accepted in BEP contexts appear in L2 varieties. As this seems to hold universally for various L2 varieties, previous findings for individual varieties (e.g. BEP usage with verbs such as give, get or define in IndE; Davydova 2011: 206–208) are corroborated by the present analysis.

6 Discussion

6.1 Alternative scenarios

6.1.1 Phonological similarity

In the simplest of cases, a straightforward motivation for why BEP tokens exist in present-day data would be to explain them away as forms in which the ambiguity between phonologically similar auxiliary forms persists (see Section 2). If this was the case, BEP tokens should (almost) exclusively occur with the auxiliaries is and was, which are phonologically close to the HP forms has and ’s.

Indeed, Figure 7 shows that is and was, as in (57) and (58), are the most fre-quent auxiliary types.

(57) ETB is actually changed name for Takaful Nasional(GloWbE MAL B)

(58) It was cost the business a bit initially but the results will be for a long time.(GloWbE IND G)


0% 5% 10% 15% 20% 25% 30% 35%

am

're

were

'm

are

was

is

Figure 7: Relative frequencies of definite auxiliary types in BEP forms (ICE+GloWbE).

However, this line of argumentation is flawed for a number of reasons. First, it runs counter to the historical pattern, where a reinterpretation in the other direc-tion, that is, particularly of is > ’s-forms as abbreviated instances of has took place (Brinton 1994: 158). Second, it only applies to third-person forms and does not provide any genuine explanatory potential for other forms (other than by analogy). Most importantly, and this is closely related to the previous aspect, it fails in the light of the fact that also auxiliaries where the phonological similarity criterion does not operate appear in the present data. Although Figure 7 shows that these forms are less frequent in relative terms, are nearly reaches the pro-portion for was. Examples (59) to (63) illustrate instances of these auxiliary types.

(59) That said over the last 18 months Apple has been less innovative and made small steps, with other operators making bigger leaps forwards and then rather unfortunately everyone getting caught up in legal arguments that are cost them all a fortune!(GloWbE USA B)

(60) I’m adapted & become “cold-blooded”(ICE-IND W1B-008)

284 Valentin Werner

(61) We were thought that the slow and steady horse wins the race (GloWbE NIG B)

(62) As I’m sure you’re noticed, this trend eventually started finding it’s way into the Blogging world(GloWbE SA B)

(63) After the first general meeting and the sub-committee meetings about the first SAR chief executive and the setting up of the election committee concerned I am gathered some pa preliminary impression about the Preparatory Com-mittee(ICE-HK S2B-026)

6.1.2 Paradigm leveling

In her account of Bungi English (a rural Canadian dialect), Gold (2007: 7–8) dis-cusses paradigm leveling in the sense that be, mainly due to the extended scope of the progressive in this dialect, may have developed toward a quasi-universal auxiliary used in all periphrastic time-reference forms. However, the attribu-tion of the salience of the BEP in the present data to a similar frequency effect is implausible. Although there may be exceptions, the scenario would only work under the assumption that leveling toward be applies for all of the 20 varieties analyzed, for which there is no evidence. It becomes even more unlikely as the general historical development in English was toward an increased use of have in perfect contexts, as outlined above (see sections 2 and 6.1.1). Thus, this line of argumentation may apparently hold in the specific circumstances of one indi-vidual variety (such as Bungi English), but not as a blueprint for establishing a generalizable setup for the present data.

In sum, the alternative scenarios at best provide partial or additional moti-vations for the persistence of the BEP. Hence, different explanations need to be sought.

6.2 Continuation or innovation?

The crucial question remains whether the persistence of the BEP as supported by the present data should be viewed as a continuation of the historical BEP pattern or rather as an innovation. On a general note, the analyses have indicated a diver-


sified picture for the different variety types, for instance as to the higher propor-tion of transitives (Section 5.1) and the larger number of verb types used in BEP constructions (Section 5.2) applying in L2 varieties. Therefore, different factors have to be considered for different variety types, and in the following I present a multifactorial account as most likely scenario.

Principally, it is evident that a continuation of the historical BEP pattern does not fully account for the current presence of the BEP. Along the lines of previous analyses (such as Gold 2007: 6), this can mainly be attributed to the fact that the current BEP is not restricted to mutative intransitive verbs and thus has a much wider lexical and semantic scope compared to the historical pattern. This does not imply, however, that traces of the continuation of the historical pattern are fully absent from the data (see Section 5.2). Despite claims to the contrary (see, e.g., Görlach 1987), from a theoretical perspective this implies that the notion of continuation (more commonly featuring as “colonial lag” or “feature retention”) is suitable, even though its explanatory power is somewhat limited when taken as a sole factor.

A second continuation-related factor coming into play for the L2 varieties is the founder effect (see, e.g., Mufwene 1990; Kortmann and Szmrecsanyi 2009: 283). Here, the assumption is that both features of colonizers’ dialects as well as features of standard English of the time of the colonization are transferred. For the BEP, both developments seem plausible. Speakers of varieties where the BEP was an established feature (such as IrE) acted as missionaries and teachers, for instance in the Asian sphere (Davydova 2013: 197), and previous analyses more-over suggested that the BEP could be seen as a competitor of the HP in standard English as late as the end of the nineteenth century (see Section 2). However, this factor still does not fully account for the occurrence of the extended set of transi-tive patterns found in the data.¹⁵

Therefore, substrate and learner effects have to be considered (see Section 3). Although substrate influence has been viewed as a decisive factor in shaping time-reference systems in L2 speakers (Davydova 2013: 195), and may thus be seen as a reinforcing factor in individual varieties (such as IndE), it is unlikely that it can be universally applied to the present data for typological reasons. First, this is due to the areal distribution of BEPs worldwide (see Section 2). Second, it runs counter to the finding that the BEP occurs in L2 varieties from various areas with all kinds of different substrate languages, so that a parallel process of trans-fer seems highly improbable (cf. Van Rooy 2011: 191).

15 Note that although transitive BEPs may appear in IrE (in both its present-day and historical forms), occurrences are very sporadic only and the relevant set of verbs is severely restricted (McCafferty 2014: 343).

286 Valentin Werner

However, Van Rooy (2011: 191–192) has also drawn attention to the fact that “performance errors may occur, and some of these may eventually become accepted” (emphasis added). This especially (though not exclusively) holds for L2 varieties. Through this lens, two interesting parallels arise between the (tran-sitive) BEP and the innovative double modal structure can be able to (Van Rooy 2011: 197; see 195–204 for further related examples). Both constructions can be viewed as instances where a “latent pattern in the grammar” (Van Rooy 2011: 198) is revitalized. A fundamental difference, however, is that the BEP does not fill a “noticeable gap in the system” (Van Rooy 2011: 198), but rather illustrates “orna-mental complexity” (Kortmann and Szmrecsanyi 2009: 273) in that it renders the existing (standard) system more complex. Therefore, the present-day BEP seems to represent an instance of resistance against the general trend of regularization, which is particularly characteristic of L2 varieties (Van Rooy 2011: 195). A second parallel relates to the data on which assumptions are based, as instances of the kind discussed are hardly found in any of the standard corpora, although they are salient in web data (cf. Van Rooy 2011: 197). Electronically-mediated registers (especially the more informal ones such as blogs, forums, or comment sections) arguably represent the most direct and unedited (written) way of communicat-ing (Crystal 2006: 15), and, thus, are a likely origin and incubator for innovative uses (besides speech, from which the vast majority of the BEP examples from ICE originate). At the same time, this may imply that the “relic” view of the BEP is adequate in other written and more formal registers, where stronger monitoring processes or even formal editing operate, and where the acceptance of the BEP is highly restricted (but see Section 7).

Therefore, what I would like to suggest after the juxtaposition of potential factors for the persistence of the BEP is an account where continuation and inno-vation interact and where different motivations lie behind the similar surface structure of BEP constructions (Gilquin 2015: 117) in both L1 and L2 varieties. In addition, this account ties in with more general statements that see “both conser-vative and innovative usage patterns” (Hundt 2009: 289) in operation in L2 variet-ies, and potentially extends the scope of this claim when electronically-mediated registers are taken into account. Innovative uses (that is, transitive uses with an extended set of verbs) of the BEP may have begun as performance (or otherwise-in-duced) errors, but were (or are at least in the process of being) conventionalized to a certain extent (cf. Van Rooy 2011). The wide range of examples across all of the 20 varieties analyzed is testimony to this conventionalization. This development is supported by the presence of a historical template, even though this template is much more restricted in its communicative scope in the case of the BEP.


7 ConclusionIn this contribution, I have presented an analysis of the BEP in present-day vari-eties of English. This perspective may have seemed unconventional at first, as the BEP in previous studies was mostly viewed as a historical structure or a mere “relic” structure in modern English (Section 2), severely restricted in its commu-nicative scope. Findings of studies on contemporary non-standard varieties of English (Section 3) indicated, however, that the BEP has not died at all, both in its intransitive and transitive variant.

I was able to confirm this view with the help of the corpus data used for the present investigation, and the three hypotheses raised in Section 5 that stated the restrictions on the BEP could be refuted. With the help of both the ICE and partic-ularly the GloWbE data, it was shown that the BEP occurs to varying degrees in all 20 varieties. In addition, it was found that the BEP is especially vital in L2 vari-eties (those in Asia in particular; Section 5.1), where transitive uses that cannot be directly linked to the historical pattern are prominent, and where an increased lexical scope of the construction was in evidence (Section 5.2). However, transi-tive uses are not altogether absent from L1 varieties either, which is striking in the light of the traditional “relic” view of the BEP, and casts some doubt on the gen-eralizability of the analysis of the BEP as a “pseudo-passive” (Quirk et al. 1985: 169–170). Note that this position has also been criticized for its sole reliance on the surface similarity between the BEP and passive forms (e.g. by Faiß 1989: 299).

In the discussion of potential factors that account for the salience of the con-temporary BEP (Section 6.1), two scenarios (phonological similarity of the auxil-iary forms of the BEP and the HP, and paradigm leveling toward be as universal auxiliary) were dismissed as unlikely universal motivations. Instead (Section 6.2), it was argued that the contemporary BEP should best be assessed as a “revital-ized” structure that is both conservative and innovative. It remains to be seen whether the BEP (as was the case in the historical pattern) will be used in special-ized lexical and semantic contexts in the future, predominantly in L2 and other non-standard varieties.

A related aspect pertains to the issue of whether in its current status the BEP can be described as a vernacular universal. Earlier claims opposing this view base their stance on the finding that the BEP occurs only rarely in different ver-naculars of English (Gold 2007: 7; Yerastov 2012: 433). This position appears to be supported by quantitative typological evidence at first sight. Recall that the BEP (feature 102 in Kortmann and Lunkenheimer 2013) receives high (that is A or B) pervasiveness ratings in merely 9 out of the 76 varieties covered in eWAVE, is rated as rare (C) in further 12 varieties, and claimed to be absent (D) in 41 varieties. I do not have anything to say about varieties that should potentially be re-as-

288 Valentin Werner

signed to the A and B set, as the corpus data – large as they may be – also present no more than a snapshot of the actualities. Still, the ICE and GloWbE corpus data indicate that the inclusion of at least 12 varieties – (colloquial) AmE, (colloquial) SinE, GhaE, HKE, IndE, JamE, KenE, MalE, NZE, PakE, PhiE, SLE – under the absent (D) category may require revision. Some of these varieties even emerged among the ones with a particularly high BEP frequency in the present study (see Section 5.1). In sum, I propose that the consideration of electronic corpus data of the kind used here should increasingly be used to inform expert ratings of fea-tures for individual varieties, and consequently to further improve the validity of ratings in dynamic typological databases such as eWAVE.

It may be a truism of the study of variation that different types of data (expert ratings versus corpus data) yield different results. However, if we consider elec-tronically-mediated registers as one expanding form of communication that is potentially informal and vernacular-like,¹⁶ it would be wholly inadequate to treat this type of discourse as the odd one out, or even to ignore it altogether. With GloWbE, despite its inherent weaknesses (see Section 4 and Mair et al. 2015), we are now in the fortunate position to be able to widen the perspective beyond tra-ditional (corpus and other) data, and to exploit web-based data as a resource in a linguistically sound way. In the future, the consideration of this discourse type (in addition to other written and spoken registers) will become increasingly important for the study of variation.

If we take the analysis of web-based data as a particular kind of non-standard variety that is not regionally bound one step further (in the spirit of Mair 2013), we may be in a position to establish global “register versals” (Werner 2014: 356), for which the BEP seems a promising candidate. In this regard, the electronic medium may be seen as a facilitator that reinforces the BEP or, at least, slows down its recess (see further below). Despite the present results, no firm conclu-sions are possible as to the status of the BEP as a vernacular universal, though, in tendency, this view may be justified.

I conclude with a number of issues that have largely remained unaddressed in the present contribution. The first area that merits further attention is where the BEP yields overlap with other grammatical forms. This comprises instances

16 This view is supported by the fact that in the environment of BEP tokens, other non-standard features can be found on all levels of analysis. This may comprise elisions, missing congruence between subject and verb, lexical innovation (see verbal unlike ‘to not like’ in (38)), non-standard spellings (You’re mind is already been made up; GloWbE USA B), non-standard preposition use ([…] they were failed to reach at this […]; GloWbE BAN B), as well as relative freedom in morpho-logical marking of verbs (K. Asamoah is given them something that would have cost them millions to accomplished […]; GloWbE GHA G).


where ambiguity between a passive or a perfect interpretation, or even an adjecti-val reading is present (cf. Quirk et al. 1985: 170; Kytö 1997: 28; Yerastov 2012: 445), as in (64) or (65), and occurrences where in standard varieties a progressive form would be expected, as in (66) or (67), similar to (21) and (26) above.

(64) I’m retired for a couple of years from running a fair sized manufacturing company […](GloWbE USA G)

(65) Start-stop systems, in their current form, will shut down your car’s engine while you’re stopped at a red light and then start it again when you take your foot off the brake pedal.(GloWbE USA B)

(66) Little did I realise this was basically like been given a free transferable tattoo […](GloWbE HK G)

(67) It was looked and tasted incredibly as well as being technically masterful.(GloWbE SA B)

While examples such as (66) can in most cases be motivated as an idiosyncratic orthographical representation of a contracted progressive (-in’), this is impossi-ble for (67).¹⁷ In the same vein, Hundt (2016: 57), based on a study of a variety of corpora of native and non-native Englishes, has argued that constructions of this type are “too consistently attested to be written off as a mere performance error”. Therefore, it will be necessary to further determine how speakers of different vari-eties (including learner varieties; see Davydova 2011: 91) interpret instances of

17 An attribution to a spelling idiosyncrasy also appears less likely if an indefinite temporal adverbial typically found in perfect contexts (cf. Werner 2013) is present, as in […] a proposal that will see home-based resume camp ahead of their foreign counterparts is already been worked on (GloWbE GHA G). Note further that invariant been (< being/bein’) may occasionally occur in highly formal genres, such as academic journals: An analogy is been made with the case of Fujian Province in China, which has accomplished economic catch-up, mainly due to Taiwanese investments (GloWbE GB G) or It was initially been produced by the drug-manufacturing unit of the University College Hospital, Ibadan, but this was not enough for the country. (GloWbE NIG G). Suárez-Gómez and Seoane (2013: 176–177) have argued that if a (once vernacular) feature is trans-ferred to the written domain, this provides a strong indication for its conventionalization and productivity (see Section 6.2). Consider also a related discourse marker (that been said) occurring (n = 60) in the GloWbE data (see Davies and Fuchs 2015: 18–20).

290 Valentin Werner

this kind (possibly when contextual cues are present), and how acceptable they are in different types of written and spoken discourse.

A second avenue worth pursuing is an in-depth analysis of the etymology of the verbs occurring in BEP constructions. There is tentative evidence in the set of verbs used for the present investigation (see Section 5.2 and the Appendix) that BEP usage in L1 varieties is characterized by Germanic verbs (and their syn-onyms), while L2 varieties are more open toward later (Romance) loans (see also Bowie and Wallis, this volume). With an extended set of verbs, it would be possi-ble to determine whether this tendency can be universally validated.

On a general methodological note, I submit that GloWbE proved to be an excel-lent complement to the existing set of ICE components for the cross-variety investi-gation of the BEP as a low-frequency phenomenon, very much as intended by the compiler (see Davies and Fuchs 2015: 2). Given this comparatively low incidence of the BEP, we have to step back and put the findings in a quantitative perspective. While the present study confirmed that the BEP indeed is fairly rare (in L1 varieties even more so than in L2 varieties), it also showed that it is persistent across many L1 and L2 varieties of English. Therefore, and with regard to the discussion above (Section 6.2), I suggest that the view of the BEP as an “undead” structure that has never completely vanished, but seems to have acquired some new functionality, captures the overall situation quite well. A botanical image that may alternatively be used is that of old dying trees. Similar to the decline of the BEP, the process of dying may take dozens or even hundreds of years (and can, once started, not be inhibited). But when the relics of the old tree (or structure) eventually lie on the ground, new branches and leaves (or innovative uses) may spring from it.

AcknowledgementsI would like to thank Chris Lucas, Elena Seoane and two reviewers for comments on earlier versions of this paper. Particular thanks are also due to Gaby Knappe, who provided both insightful comments on diachronic aspects and suggestions on areas that needed to be explored further.


ReferencesAnderson, Lloyd B. 1982. The ‘perfect’ as a universal and as a language-specific category.

In Paul J. Hopper (ed.), Tense-aspect: Between semantics & pragmatics, 227–264. Amsterdam: Benjamins.

Anderwald, Lieselotte. 2014. The decline of the be-perfect, linguistic relativity, and grammar writing in the nineteenth century. In Marianne Hundt (ed.), Late Modern English syntax, 13–37. Cambridge: Cambridge University Press.

Brinton, Laurel J. 1994. The differentiation of statives and perfects in Early Modern English: The development of the conclusive perfect. In Dieter Stein & Ingrid Tieken-Boon van Ostade (eds.), Towards a standard English, 1600–1800, 135–170. Berlin: Mouton de Gruyter.

Crystal, David. 2006. Language and the internet. Cambridge: Cambridge University Press.Dahl, Östen. 1995. Areal tendencies in tense-aspect systems. In Pier M. Bertinetto, Valentina

Bianchi, Östen Dahl & Mario Squartini (eds.), Temporal reference, aspect and actionality. Vol. 2: Typological perspectives, 11–27. Torino: Rosenberg & Sellier.

Dahl, Östen & Viveka Velupillai. 2011. The perfect. In Matthew S. Dryer & Martin Haspelmath (eds.), The world atlas of language structures online. Munich: Max Planck Digital Library. http://wals.info/chapter/68 (accessed 08 December 2014).

Davies, Mark. 2013. Corpus of Global Web-based English: 1.9 billion words from speakers in 20 countries. http://corpus.byu.edu/glowbe (accessed 20 February 2015).

Davies, Mark & Robert Fuchs. 2015. Expanding horizons in the study of World Englishes with the 1.9 billion word Global Web-based English Corpus (GloWbE). English World-Wide 36(1). 1–28.

Davydova, Julia. 2008. Preterite and present perfect in Irish English: Determinants of variation. Hamburg: University of Hamburg.


Davydova, Julia. 2013. Detecting historical continuity in a linguistically diverse urban area: The present perfect in modern Singapore English. In Joana Duarte & Ingrid Gogolin (eds.), Linguistic superdiversity in urban areas: Research approaches, 193–225. Amsterdam: Benjamins.

Denison, David. 1993. English historical syntax. London: Longman.Drinka, Bridget. 2003. Areal factors in the development of the European periphrastic perfect.

Word 54(1). 1–38.Faiß, Klaus. 1989. Englische Sprachgeschichte [History of the English language]. Tübingen:

Francke.Filppula, Markku. 1999. The grammar of Irish English: Language in Hibernian style. London:

Routledge.Gilquin, Gaëtanelle. 2015. At the interface of contact linguistics and second language

acquisition research: New Englishes and Learner Englishes compared. English World-Wide 36(1). 91–124.

Gold, Elaine. 2007. Aspect in Bungi: Expanded progressives and be perfects. In Milica Radišić (ed.), Proceedings of the 2007 annual conference of the Canadian Linguistic Association. http://homes.chass.utoronto.ca/~cla-acl/actes2007/Gold.pdf (accessed 28 December 2014).

292 Valentin Werner

Görlach, Manfred. 1987. Colonial lag? The alleged conservative character of American English and other colonial varieties. English World-Wide 8(1). 41–60.

Halas, Ana. 2012. The change in the form of the present perfect in Middle English. In Vesna Lopicic & Biljana M. Ilic (eds.), Challenging change: Literary and linguistic responses, 223–234. Newcastle: Cambridge Scholars Publishing.

Hundt, Marianne. 2009. Global feature – local norms? A case study on the progressive passive. In Thomas Hoffmann & Lucia Siebers (eds.), World Englishes – Problems, properties and prospects, 287–308. Amsterdam: Benjamins.

Hundt, Marianne. 2016. Error, feature, (incipient) change – or something else altogether? In Elena Seoane & Cristina Suárez-Gómez (eds.), World Englishes: New theoretical and methodological considerations, 37–60. Amsterdam: Benjamins.

Kallen, Jeffrey L. 1989. Tense and aspect categories in Irish English. English World-Wide 10(1). 1–39.

Kilpiö, Matti. 1997. On the forms and functions of the verb be from Old to Modern English. In Matti Rissanen, Merja Kytö & Kirsi Heikkonen (eds.), English in transition: Corpus-based studies in linguistic variation and genre styles, 87–120. Berlin: Mouton de Gruyter.

Kortmann, Bernd & Kerstin Lunkenheimer (eds.). 2013. The electronic world atlas of varieties of English. Leipzig: Max Planck Institute for Evolutionary Anthropology. http://www.ewave-atlas.org (accessed 15 January 2015).

Kortmann, Bernd & Benedikt Szmrecsanyi. 2009. World Englishes between simplification and complexification. In Thomas Hoffmann & Lucia Siebers (eds.), World Englishes – Problems, properties and prospects, 265–285. Amsterdam: Benjamins.

Kytö, Merja. 1997. Be/have + past participle: The choice of the auxiliary with intransitives from Late Middle to Modern English. In Matti Rissanen, Merja Kytö & Kirsi Heikkonen (eds.), English in transition: Corpus-based studies in linguistic variation and genre styles, 17–86. Berlin: Mouton de Gruyter.

Mair, Christian. 2013. The world system of Englishes: Accounting for the transnational importance of mobile and mediated vernaculars. English World-Wide 34(3). 253–278.

Mair, Christian, Joybrato Mukherjee, Gerald Nelson & Pam Peters. 2015. Responses to Davies and Fuchs. English World-Wide 36(1). 29–44.

McCafferty, Kevin. 2014. ‘[W]ell are you not got over thinking about going to Ireland yet’: The be-perfect in eighteenth- and nineteenth-century Irish English. In Marianne Hundt (ed.), Late Modern English syntax, 333–351. Cambridge: Cambridge University Press.

McFadden, Thomas & Artemis Alexiadou. 2006. Pieces of the be perfect in German and older English. In Donald Baumer, David Montero & Michael Scanlon (eds.), Proceedings of the 25th West Coast conference on formal linguistics, 270–278. Somerville: Cascadilla Proceedings Project.

McFadden, Thomas & Artemis Alexiadou. 2010. Perfects, resultatives, and auxiliaries in earlier English. Linguistic Inquiry 41(2). 389–425.

Mufwene, Salikoko S. 1990. Transfer and the substrate hypothesis in creolistics. Studies in Second Language Acquisition 12(1). 1–23.

Nelson, Gerald. 1996. The design of the corpus. In Sidney Greenbaum (ed.), Comparing English worldwide: The International Corpus of English, 27–35. Oxford: Clarendon Press.

Patrick, Peter L. 1999. Urban Jamaican Creole. Amsterdam: Benjamins.Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A comprehensive

grammar of the English language. Harlow: Longman.


R Core Team. 2014. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing.


Schneider, Gerold. 2013. Describing Irish English with the ICE Ireland Corpus. Cahiers de l’ILSL 38. 139–162.

Scott, Mike. 2011. WordSmith Tools. Liverpool: Lexical Analysis Software.Seoane, Elena. 2014. Variation and change in New Englishes: Diachronic answers to synchronic

questions. In Alejandra Moreno-Álvarez & Irene Pérez-Fernández (eds.), New alleyways to significance: Interdisciplinary approaches to English studies, 209–240. Palma de Mallorca: Edicions UIB.

Shannon, Thomas F. 1995. Toward a cognitive explanation of perfect auxiliary variation: Some modal and aspectual effects in the history of Germanic. American Journal of Germanic Linguistics & Literatures 7(2). 129–163.


Smith, Jennifer & Mercedes Durham. 2011. A tipping point in dialect obsolescence? Change across the generations in Lerwick, Shetland. Journal of Sociolinguistics 15(2). 197–225.


Tagliamonte, Sali. 2000. The grammaticalization of the present perfect in English: Tracks of change and continuity in a linguistic enclave. In Olga Fischer, Anette Rosenbach & Dieter Stein (eds.), Pathways of change: Grammaticalization in English, 329–354. Amsterdam: Benjamins.

Van Rooy, Bertus. 2011. A principled distinction between error and conventionalized innovation in African Englishes. In Joybrato Mukherjee & Marianne Hundt (eds.), Exploring second-language varieties of English and learner Englishes: Bridging a paradigm gap, 189–207. Amsterdam: Benjamins.

Watts, Sheila. 2001. How to become an auxiliary: Progressive and perfect in Old Saxon. In Sheila Watts, Jonathan West & Hans-Joachim Solms (eds.), Zur Verbmorphologie germanischer Sprachen [On the verbal morphology of Germanic languages], 117–136. Tübingen: Niemeyer.



Wright, Susan. 1986. Tense, aspect and text: Processes of grammaticalization in the history of the English auxiliary. Cambridge: University of Cambridge dissertation.

Yerastov, Yuri. 2012. Transitive be perfect: An experimental study of Canadian English. Canadian Journal of Linguistics 57(3). 427–457.


294 Valentin Werner

Appendix – verb listvariety association

only L1 mostly L1 both mostly L2 only L2

verbs arrive, leave, let, retire, see, subju-gate

cut, do, finish, gather, read, stop, understand

appear, arrange, ask, attack, be, change, cost, describe, disap-pear, explode, fail, give, go, happen, inform, know, look, make, mean, place, plummet, reduce, relocate, remain, result, say, seem, send, settle, split, start, suffer, suppose, tell, think, turn, utilise/utilize, work

come, identify, investigate, plan, present, represent, serve

adapt, bandy, become, claim, conclude, contradict, deduct, find, increase, like, narrow, notice, open, originate, predict, prove, provide, renew, respect, shield, study

Part III: Building bridges

Robert Fuchs, Sandra Götz, and Valentin Werner10 The present perfect in learner Englishes:

A corpus-based case study on L1 German intermediate and advanced speech and writing

Abstract: Among the time-reference forms of English, the acquisition of the present perfect is regarded as the single most challenging task for non-native speakers, mainly due to the semantic peculiarities of this form in contrast to many other languages. In this chapter, we focus on variation between the present perfect and the simple past in German-speaking learners to assess the influence of learner proficiency level and mode on the use of the present perfect. Our results suggest that (i) the present perfect is more frequent in writing, and (ii) emerges very late in learner language, such that only the most advanced learners, and specifically those who started learning English very early in primary school, use it as frequently as native speakers. This finding tentatively supports proponents of early English language teaching. To explain the results, we explore potential reasons for the late emergence of the present perfect, and conclude with specific recommendations for English language teaching.

1 IntroductionThe present perfect (PP) is not just a form where variation is pervasive in L1 and L2 varieties of English (see, e.g., the other contributions to this volume, as well as Elsness 1997; Hundt and Smith 2009; Davydova 2011, 2012; Yao and Collins 2012; Seoane and Suárez-Gómez 2013; Werner 2013, 2014; Werner and Fuchs 2016; see further eWAVE features 99 and 100 in Kortmann and Lunkenheimer 2013). It is also a highly relevant topic for learners and instructors of English as a Foreign Language (EFL). On the one hand, the expression of temporal relations is central in all processes of communication (Housen 2002a: 78; Roberts and Liszka 2013: 414). On the other hand, it is a highly troublesome area for non-native speakers,

Robert Fuchs, Hong Kong Baptist UniversitySandra Götz, Justus Liebig University GiessenValentin Werner, University of Bamberg

298 Robert Fuchs, Sandra Götz, and Valentin Werner

and a number of causes have been identified why the PP is particularly “learn-er-hard” (see also van der Wurff 1999: 473; Davydova 2011: 289).¹

1.1 The present perfect as a “late” structure

Above all, the PP has been described as the most prominent “latecomer” in an (idealized) EFL acquisition sequence² from two perspectives.

In the broader perspective, which takes a concept-oriented view (how is temporality expressed?), studies have found converging evidence for a univer-sal three-stage model of how ways of expressing temporal reference develop in learners (see, e.g., Bardovi-Harlig 1999: 350, 2014: 130; Shirai 2009: 167; Zhang and Widyastuti 2010: 2–3). First, simple chronological ordering and turn-taking by the interlocutors appears to provide a temporal scaffold (pragmatic stage). Then, adverbials of place and time as well as conjunctions come into play (lexical stage). Lastly, morphological marking of tense-aspect distinctions emerges (mor-phological/grammatical stage).³

From the narrower perspective, which takes a form-oriented view (which forms are used to express temporality?) and thus focuses on the internal develop-ment of the morphological stage described in the foregoing, a universal sequence along the lines of simple past > present perfect > [present perfect progressive/past perfect] can be established (Klein 1995: 44–46; Bardovi-Harlig 2000: 169). It further emerges that forms approximating the PP, such as (1) to (3), already occur in early learner data.

1 Another factor to be considered in addition to the ones discussed in this contribution is the influence of learner grammars. Swan (1994: 50) discusses the “jungle of rules” surrounding the PP, which may lead to learners potentially avoiding the form altogether (on which see below). For further factors argued within a generative framework, see Liszka (2004: 213).2 The PP also emerges late in L1 acquisition (Gathercole 1986: 538; Bardovi-Harlig 1997: 376), a complex which cannot be explored in detail within the scope of this investigation (for relevant studies see, e.g., Antinucci and Miller 1976; Gathercole 1986; Slobin 1994; Klein 1995; Shirai and Andersen 1995; van der Wurff 1999; Wulff et al. 2009). It is evident, though, that some of the rel-evant factors overlap between L1 and L2 acquisition (see, e.g., van der Wurff 1999: 477–478), with the “obvious difference between L1 and L2 acquisition [being] the effect of L1” (Shirai 2009: 182).3 Some learners in an early/intermediate stage also use “boundary markers”, notably finish and start, to express temporal-aspectual distinctions (Klein 1995: 37), a finding that can be motivated by the typology of the respective first languages of the learners involved, which have a corre-sponding perfect marker (see Werner 2014: 11–12). As this feature is extremely rare, it will not be considered in the present chapter.

The present perfect in learner Englishes 299

(1) uh mister Neil have give me the uh la letter(from Housen 2002a: 91)

(2) yes I have understanding(from Housen 2002a: 91)

(3) I’m pass my test as well(from Klein 1995: 66)

However, in keeping with other learner-hard features such as consistent third person -s marking, target-like⁴ PP use on the surface can be observed at a highly advanced stage only (Housen 2002a: 79, 2002b: 158–162; Davydova 2011: 286), a fundamental prerequisite being that the simple past (SP) has been mastered beforehand (see above; Bardovi-Harlig 1997: 411; Collins 2002: 86). Note also that the functional distinction between PP and SP is delayed even further, “a general trend in the L2 acquisition of English” (Housen 2002b: 163; see also Klein 1995: 47; Odlin and Alonso Vázquez 2006: 54).

Strongly related to the latter point is the necessity on the part of the learner to take account of the “lack of iconicity” (Bardovi-Harlig 2000: 420) of the PP in terms of semantic overlap with the SP. Both forms encode anteriority, while the PP is commonly restricted to the additional property of signaling “current relevance” of the situation or event described (Bardovi-Harlig 1997: 376; Eriks-son 2008: 140–141). As this notion appears to be difficult to master, especially by learners with a native language where this specific semantic distinction does not exist, non-target-like occurrences, both in terms of over- and undergeneral-ization, can be found in learner data. Overgeneralization occurs when the PP is used in SP contexts, as exemplified in (4), where the temporal adverbial yesterday calls for the use of the SP. Undergeneralization occurs when the SP is used in PP contexts, such as (5), where the temporal adverbial already suggests the current relevance of the action, which calls for the use of the PP. Thus, what makes the PP additionally difficult is that learners do not merely have to struggle with its

4 Note that the notion of target-like can be defined in various ways. We use it in a traditional – that is mainly qualitative and absolute – sense for the time being. Thus, target-like signifies full (formal and functional) correspondence to native-speaker norms, an approach that applies best to the learners under investigation in the present study (see Section 2). However, as there is also variation in native-speaker use, we suggest that this absolute view does not do justice to actual linguistic realities and propose an alternative – quantitative – definition of target-like (or “na-tive-like”) below (see sections 3 to 5).


formal and functional properties, but also its “semantically close neighbors” (Bardovi-Harlig 1997: 376), the SP being the most important one.

(4) Today the sun is shining and yesterday I’ve finished all my work for school so far.(ICCI-AUT 0597)

(5) My parents think that 2000€ are too much but I already planned on what I want to spend the money.(ICCI-AUT 0450)

1.2 The present perfect and the German Perfekt

Notably, this incomplete form-function mapping in the learners’ conceptual-ization (Housen 2002a: 156, 2002b: 159; Odlin and Alonso Vázquez 2006: 54; Davydova 2011: 277; but cf. Liszka 2004: 225) also applies to languages where the devices for creating reference to anteriority formally resemble the English pattern. In German, for instance, both a simple form (Präteritum; see (6)) and a complex form (Perfekt, haben/sein + past participle; see (7)) exist.

(6) Martin ging (zum) Einkaufen.Martin go-pst-3sg (to) shopping.‘Martin went shopping’.

(7) Doris hat gestern einen Kuchen gebacken.Doris has yesterday a cake baked.‘Doris baked a cake yesterday’.

However, the Perfekt has extended its semantic scope and is now also used as a narrative tense (cf. also the French passé composé or the Italian passato prossimo; see Collins 2002). Hence, a definite temporal adverbial constraint does not apply, and unlike the English PP, the German Perfekt freely combines with adverbials referring to definite time points in the past, as in (7). Further, rather than covering different semantic domains, the German Perfekt and Präteritum can largely be used interchangeably (in the sense of truth-conditional equivalence). However, in spoken language, use of the Präteritum is stylistically marked and quite rare,


except for a limited number of verbs (differences in regional preferences notwith-standing; see König and Gast 2012: 87).⁵

If we now presuppose an incomplete form-meaning association in learners (see above), the contrastive analysis shows that the different and arguably wider semantic scope of the German structure (Perfekt) may, above all, lead to an over-generalization of the PP in German learner English due to its formal correspon-dence (see, e.g., Davydova 2011: 275). This hypothesis is backed up by psycho-linguistic evidence related to the general grammatical structure of the learner’s native language. An experimental study (Roberts and Liszka 2013: 413) has found that advanced German learners of English in an on-line reading task (in contrast to native speakers and learners with a language with aspectual distinctions, such as French) do not show any additional processing cost at all when reading con-structed English sentences where the definite temporal adverbial constraint is violated. This applies even though both the German and the French groups sim-ilarly take account of the constraint in explicit tasks, such as cloze tests and in terms of metalinguistic comments (Roberts and Liszka 2013: 426–427). We will return to this aspect below.

1.3 Factors in the acquisition of the present perfect

That the first language of the learners indeed may have an impact on the acquisi-tion of the PP can be seen from the results of various studies. Three broad trends, all ultimately pertaining to typological aspects, can be identified. They are listed here to provide a broader context and to be related to the German-speaking learner population where applicable:

(i) If there is no formal or functional category corresponding to the PP in the learner’s native language, PP use in English will be reduced (see, e.g., Bulut 2011: 225–226). At the same time, there is conflicting evidence on whether these learn-ers – in contrast to those where at least a formal analogy with the PP exists, as is the case for some Germanic (see, e.g., the contrastive analysis for German above) and Romance languages – will show a lower rate of PP overgeneralization (Sval-

5 For the sake of completeness we need to add that both the Perfekt and the PP can be used in resultative contexts, e.g. Sie hat sich den Arm gebrochen/She has broken her arm (König and Gast 2012: 88; see also Davydova 2011: 281). In addition, the present (Präsens) may be used in contin-uative contexts, e.g. Ich warte (schon) seit drei Stunden/I’ve been waiting for three hours (already) (a condensed contrastive overview is provided in König and Gast 2012: 92). A similar situation applies to other languages, such as Swedish (see Eriksson 2008: 146).


berg and Chuchu 1998: 27, 56; but cf. Collins 2002: 49; Odlin and Alonso Vázquez 2006: 55).

(ii) In contrast, other studies have found that the influence of the type of the learner’s first language on the “number of forms produced or emergent form-meaning associations en route to the target associations” (Bardovi-Harlig 2000: 182) is restricted (see also Klein 1995: 68; Klein, Dietrich, and Noyau 1995; Housen 2002a: 97). However, although generalizations may be drawn across learners and languages, both individual languages and the individual learner have to be considered as well, as rates of acquisition may vary, even within the same formalized instructional context (see, e.g., Bardovi-Harlig 1997: 385–386).

(iii) In addition, the inherent lexical aspect of individual verbs seems to play a role as non-target-like use of the PP is more pervasive with telic achievement verbs, that is, verbs with a defined end point such as build a house (Collins 2002: 85–86). Although the “Inherent Lexical Aspect Hypothesis” (see, e.g., Wulff et al. 2009) has been found valid in a number of acquisitional sequences of tense/aspect categories and for a range of language pairs, its explanatory potential for German seems to be limited for a number of reasons. First, its universal applica-bility has been doubted for situations where the native language of the learner (such as German, see above) does not recognize the grammatical category of aspect (Housen 2000: 251). More importantly, empirical findings of recent studies (Davydova 2011; Rogatcheva 2014) indicate that – contrary to what could be expected from a contrastive analysis of German and English – intermediate and advanced German learners produce reduced rates of the PP in relevant contexts and use the SP (or, less often, the simple present) instead. From a theoretical perspective, this is more in line with an alternative explanation, the “Default Past Tense Hypothesis”, originally established for English-speaking learners acquir-ing Romance languages. It predicts that “during the first stages of L2 develop-ment, learners will attempt to mark past tense distinctions rather than aspectual distinctions, and in so doing will initially rely on a single marker of past tense” (Salaberry and Ayoun 2007: 20, cited in Shirai 2009: 182) – the SP in the case of English. The main motivation for the choice of the SP as the default form is its formal and semantic simplicity compared to the PP (Davydova 2011: 288–289), which can potentially be viewed as a universal avoidance strategy on the part of the learners (Davydova 2011: 279–280; Rogatcheva 2014: 221; see also Svalberg and Chuchu 1998: 44; Housen 2002a: 95). The results of the relevant empirical studies thus indicate an extension of the Default Past Tense Hypothesis beyond the early stages of acquisition (and the English-Romance context) as submitted in its original version.

Another topic that has attracted specific scholarly interest is the role of tem-poral adverbials (see, e.g., Klein 1995: 38). As was established in the develop-


mental hierarchy for expressing temporal relations (see above), adverbials cover an intermediate position. Naturally, combinations of adverbials and morpho-syntactic marking can be used, but the additional usage of adverbials suppos-edly decreases the more advanced a learner is (Bardovi-Harlig 1999: 350).⁶ Even though more marginal PP forms (e.g. progressives and less common semantic readings such as continuatives) are temporally specified by a temporal adverbial in up to two thirds of all occurrences (Bardovi-Harlig 1997: 399; see also Werner 2014: 369–370), only a minority of the remaining PP forms (approximately one third; see Werner 2014: 136) in native varieties of English is actually specified.

Despite this low specification rate for the PP, studies have identified an over-emphasis on temporal adverbials in teaching the PP (see, e.g., Granger 1999: 200; Eriksson 2008: 143; Davydova 2011: 282, 287; Rogatcheva 2014: 241). This “signal word” approach may eventually give rise to an overuse of additional temporal specification in PP contexts also in learner production. Viewed differently, this association may be seen as a facilitator for the learners as regards the usage of the PP form in prototypical contexts (Bulut 2011: 223). Note in this regard that instances of indefinite temporal adverbials (commonly associated and quantita-tively preferred with PP use in native varieties, but allowing for the SP in princi-ple) have been established as a locus where variation between the PP and the SP can be explored well (see, e.g., Hundt and Smith 2009: 52–55; Yao and Collins 2012: 398–399; Werner 2013; Seoane, this volume). Therefore, we will follow this approach in the present analysis (see sections 3 and 4).

A final dimension, largely disregarded in extant studies, is the mode (written vs. spoken) of the learner language. From a developmental perspective, earlier work seems to show that learner PP usage can first be observed in written texts before spoken examples occur (Bardovi-Harlig 1997: 386).⁷ On a related note, it emerges that even though learners seem to master the PP quite well in monitored written production, this is less the case in spontaneous spoken production, where non-target-like usage occurs (including forms that deviate formally as in (8)) in up to one quarter of all potential PP contexts, with the SP being the most frequent

6 Shirai (2009: 169), following Bardovi-Harlig (2000), proposes an internal developmental scheme along the lines of adverbials of position (e.g. now, then), duration (e.g. all week) and frequency (e.g. twice, often) before those signifying contrast (already, yet). In the literature, we also find suggestions for internal developmental and instructional sequences along semantic lines (e.g. Schlüter 2002: 337–339).7 As regards PP frequencies, Bardovi-Harlig (1997: 386) adds that “many fewer instances of pres-ent perfect appeared in the oral sample. However, the learners produced a somewhat higher proportion of forms in the oral sample (105 in 175 texts) than in the written sample (502 in 1576 texts)”. Unfortunately, no token counts for average or total text lengths are provided, which im-pedes a quantitative assessment and comparison.


form (Bardovi-Harlig 1997: 414, 2000: 179). The latter finding also applies to Ger-man-speaking learners (Liszka 2004: 224).

(8) I’ve already went to a drivers-school, and my lessons will start in about two weeks, so I think you can imagine how excited I am!(ICCI-AUT 0575)

However, two notes of caution apply here. First, we have to consider that a nar-rative topic of the learner text may influence PP-SP ratios in favor of the latter (Eriksson 2008: 220). Second, we have to take account of experimental data that did not identify any significant register differences (Liszka 2004: 226). In sum, all this implies that the factor of mode (or register, in our case the difference between spoken and written language) further complicates the topic of learners’ PP usage, and, in consequence, that its effects have to be considered to obtain the full picture.

1.4 Research questions and chapter outline

As can be inferred from the previous paragraphs, we are not short of either theo-retical or empirical studies on EFL acquisition of the English PP. As yet, analyses have mainly relied on evidence gained through observation, elicitation tasks or interviews (see, e.g., Klein 1995; Bardovi-Harlig 1997; Davydova 2011; Roberts and Liszka 2013). With more and more data becoming available (see Section 2), we are now in a position to test the claims of previous works within the constantly evolving and already highly productive framework of learner corpus research. In addition, we can now complement the restricted body of corpus-based work focusing on how temporal reference is created by German-speaking learners of English (e.g. Davydova 2011; Rogatcheva 2014).⁸ Along with data from advanced learners at university level, we will also consider (early) intermediate learners at school level. This allows us both to track the learning process over several com-petence levels and to draw conclusions for a large group of learners, including

8 Davydova (2011) analyzes a small-scale corpus (26,000 words) of spontaneous speech from intermediate German learners gained through interviews (160 relevant tokens, 42 % PP, 58 % SP; Davydova 2011: 279). Rogatcheva (2014) contrasts written learner data from the German and Bul-garian components of the International Corpus of Learner English (see Section 2) with British and American native speaker data. In her study, the German learners comparatively show the lowest PP frequencies (4.9 per 1,000 words; see Rogatcheva 2014: 150). See also Liszka (2004) and Rob-erts and Liszka (2013) for non-corpus work on German-speaking learners.


the numerically dominant group of school age learners (considering that in Ger-man-speaking societies, as in most developed countries, English is obligatory at school). Moreover, we take into account variation between learner writing and speech, which has rarely been considered to date.

At the same time, it goes without saying that all of the findings presented in sections 1.1 to 1.3 have an impact on the interpretation of the quantitative results of the present study. We will therefore test the generalizability of the related hypotheses against the corpus data. In line with the general topic of this volume, we (re-)address the following issues in particular:(i) The PP as a “late” structure: Can the late emergence of the PP in learner lan-

guage be confirmed and do German-speaking learners tend to avoid the PP (as proposed, e.g., by Davydova 2011 and Rogatcheva 2014) and prefer the struc-turally simpler SP (along the lines of the Default Past Tense Hypothesis)?

(ii) German as a first language: Alternatively, does the influence of German (that is, the structural similarity between the PP and the German Perfekt) lead to an early use of the PP and an overgeneralization of the PP (respective to native-speaker frequencies)?

(iii) Temporal adverbials/variable contexts: Do less advanced learners rely on a larger number of temporal adverbials in PP contexts (according to the acqui-sitional hierarchy relating to the expression of temporality)? Which forms (PP or SP) are chosen in variable contexts that are temporally specified?

(iv) Register effects: Can the PP be viewed as a “written” form in learner lan-guage? How do learner speech – which has only scarcely been considered as yet – and writing differ?

We also aim to shed some light on the influence of additional factors such as years of instruction or a stay abroad. To this end, we employ a corpus-based approach, which will be further outlined in Section 2. We present the quantitative results in Section 3 and discuss these results in Section 4, before we further contextualize them in the concluding section 5.

2 Data and Methodology

2.1 Target norms, German learners of English and corpus data

Our study is based on the assumption that many (German-speaking) learners of English strive to emulate native norms. Consequently, our analysis relies on several German learner and native-speaker corpora of English. A desire to mirror


native norms is of course not universal. For many teachers, learners and research-ers, the overall goal in English language teaching is to prepare language learners to be able to engage in “successful communication” (Kennedy and Trofimovich 2008: 460) in the target language. In this approach, there is a clear focus on intelligibility and comprehensibility of learner speech, instead of speakers striv-ing for native-like target norms (e.g. Kenworthy 1987; Jenkins 2000). While this approach is applicable for many English language learners around the world, the present study takes a different stance and concentrates on English language learners who live in language communities formally striving for native-like target norms in teaching (i.e. Austria and Germany). There is evidence to support this approach, suggesting that the great majority of pupils and university students of English in Germany aim at speaking in a native-like manner and not according to a lingua franca norm (see a survey by Mukherjee and Rohrbach 2006). Con-sequently, it is helpful and relevant to provide findings that reveal significant deviances (e.g. over- or underuse) that contribute to a “foreign-soundingness” (Granger 2004: 132) of the learners’ output. In this vein, we take a corpus-based approach to perform a “contrastive interlanguage analysis” (Granger 1996, 2015) in order to describe significant differences between German-speaking language learners at different proficiency levels compared to each other and compared to native speakers of English.

The analysis of the spoken data is based on the Louvain International Data-base of Spoken English Interlanguage (LINDSEI; Gilquin, De Cock, and Granger 2010) and the native-speaker counterpart Louvain Corpus of Native English Con-versation (LOCNEC; De Cock 2004). The German subcorpus of LINDSEI (LIND-SEI-GE; Brand and Kämmerer 2006) comprises 50 orthographically transcribed interviews that were held with English majors in their third or fourth year of study at Justus Liebig University Giessen. The spoken native-speaker control corpus LOCNEC was specifically designed to be a reference corpus for work with LINDSEI (De Cock 2004). It is derived from 50 interviews with British English speakers who were linguistics students at Lancaster University at the time. The interviews were conducted according to the same criteria as in LINDSEI, which makes LOCNEC an ideal resource for a contrastive (interlanguage) analysis (Granger 1996, 2015).

The analysis of the written university student data is based on the German component of the International Corpus of Learner English (ICLE; Granger et al. 2009). GICLE consists of 300 timed and untimed argumentative and literary essays written by students of English majors between their first and fifth year of study at the Universities of Augsburg (Germany), Dresden (Germany), Basel (Switzerland), and Salzburg (Austria). The native control corpus is the Louvain Corpus of Native English Essays (LOCNESS) and consists of argumentative essays


composed of British pupils’ A-level essays, British university students’ essays and American university students’ essays.⁹

LINDSEI-GE and GICLE also provide additional information on the learners, for instance how long they learned English at school and whether they spent time in an English-speaking country. This information is available on all the learners in LINDSEI-GE and on 299 of the 300 in GICLE; information on the number of years of English learnt at university is available on all the learners in LINDSEI-GE and on 293 of the 300 learners in GICLE.

These data are compared to the German component of the International Corpus of Crosslinguistic Interlanguage (ICCI; Tono 2012), an international learner corpus of young learners of English across different proficiency levels and L1 backgrounds around the world. There are currently eight countries/regions of the corpus avail-able (Austria, China, Hong Kong, Israel, Japan, Poland, Singapore, Spain, and Taiwan). The L1 German subcorpus (ICCI-AUT)¹⁰ consists of 609 essays from pupils at secondary schools from years 5 to 11. The number of essays and their length vary somewhat between the years, partly because more advanced pupils produce longer essays. For example, 43 essays comprising 3,437 words are available for year 5, and 55 essays comprising 9,590 words for year 11. The authors of all essays in ICCI-AUT most likely received English language instruction at least from year 3 onwards (since this has been obligatory in Austrian schools for some time). Since the subcorpus was released in 2009, and the start of English language instruction was moved forward to year 1 in 2003/4 (de Cilia and Krumm 2010), it is conceivable that the year 5 and 6 students in the corpus started English language instruction as early as year 1 or 2. For the subcorpora of the university students, fortunately, information is available as to the duration of English instruction at school.

At the time of corpus compilation of LINDSEI-GE and LOCNEC, primary education was four years and secondary education was eight to nine years, so it is possible to distinguish between early starters (if they had learned English in

9 See https://www.uclouvain.be/cps/ucl/doc/cecl/documents/LOCNESSdescriptionall.pdf for details.10 Unfortunately, no ICCI (or other comparable) data have been collected in Germany. However, as differences between Standard German German and Standard Austrian German mostly per-tain to differences in phonology and in the lexicon, and as the school systems are by and large similar, the insights gained through the quasi-longitudinal perspective outweigh this apparent inconsistency in the choice of corpus material. Note further that a (selective) look at the English curricula for secondary schools reveals that both in Germany (see www.isb-gym8-lehrplan.de/contentserv/3.1.neu/g8.de/index.php?StoryID=26368) and in Austria (see www.ris.bka.gv.at/Gel-tendeFassung.wxe?Abfrage=Bundesnormen&Gesetzesnummer=10008568) the approximation to the standard form of the target language is the overarching goal of institutionalized foreign language instruction.


school for more than nine years) and late starters (if they indicated nine years or less). Table 1 gives the breakdown of the corpora used in the present study.

Table 1: Breakdown of native and non-native corpora used in the present study.

Native English (British and American English) Learner English (German Learner English)

c. 433,700 words total c. 400,700 words total

LOCNESS-BE LOCNESS-US LOCNEC ICCI-AUT GICLE LINDSEI-GE

Native writing of British school students (A-level exams)

Native writing of British university students

Native writing of American university students

Native English speech of British university students

Learner writing of school students

Learner writing of university students

Learner speech of university students

55,049 92,254 c. 168,400 c. 118,000 80,585 c. 234,000 c. 86,100

2.2 Data extraction and coding

First, all corpora were tagged with the C7 tagset of the CLAWS tagger (Garside and Smith 1997). All the PP tokens were counted with a regular expression that searched for a form of have + past participle and allowed for intervening adver-bials and noun phrases, and also found complex forms such as passive and progressive forms (see Appendix A). Forms of have got which are either modal or possessive in meaning were all automatically excluded afterwards. Next, the frequency of the SP was determined with a search query for past tense verbs. This search query invariably also included past perfect tokens. Consequently, we also searched for these and subtracted their number from the results for the SP search. The regular expressions used proved to be very reliable in previous research (adapted from Hundt and Smith 2009; cf. Yao and Collins 2012; Fuchs, this volume), which was confirmed for the present study via spot checks (less than 1 % false positives). In total, we analyzed 674 occurrences of the PP and 11,750 occurrences of the SP in the learner corpora, and 1,104 occurrences of the PP and 7,464 occurrences of the SP in the native corpora.

In a second step, selected temporal adverbials (since, already, yet, always, ever, never, recently, yet; cf. Werner 2013) were searched with AntConc (Anthony 2014). These adverbials were chosen because they are typically associated with the use of the PP (although the use of the SP is also possible) and are often used as “signal words” in EFL instruction (see Section 1.3). We then manually classi-


fied all instances for the time-reference form of the finite verb, that is, PP, SP, past perfect, present, and future (will-future and going to-future). In total, we analyzed 1,167 occurrences of these temporal adverbials in the learner corpora, and 613 occurrences in the native corpora (after excluding all undesired occurrences as described below).

We only included full verb phrases and temporal adverbials in variable order (i.e. subject + finite verb + temporal adverbial, or vice versa) in our analysis. Accordingly, we excluded all incomplete verb phrases, for instance due to refor-mulations as in (9), tokens without a subject as in (10), and without a verb, such as (11).

(9) well I’ve yeah cos … I’ve never … this is one of the most serious roles I(LOCNEC E04)

(10) often erm … quite derogatory … so and never call … for example …(LOCNEC E19)

(11) but never to Grand (sic!) Canaria(LOCNEC E25)

Repetitions were only excluded if the adverb or only a part of the verb phrase was immediately repeated, as in (12), but not if the complete verb phrase or clause was repeated, as in (13) or (14). In accordance with the guideline specified above, we deleted repetitions if they did not comprise a full verb phrase, as in (15), or if they were not intended by the speaker, as in file doublets.

(12) he never compl= he never completed his university degree(LOCNEC E21)

(13) but you you’re never told how old he is you’re never told(LOCNEC E54)

(14) there’s always there’s always a moral at the end(LOCNEC E54)

(15) I never I never went to university(LOCNEC E12)

We also excluded some further occurrences from our analysis: When two tempo-ral adverbials occurred together (e.g. ever since) we only coded them once accord-


ing to the following typology: Ever since and long since were counted as instances of since and never ever was counted as an instance of never, as in (16).

(16) So I will never ever forget this wonderful day!(ICCI-AUT 0338)

Again, to match the above criteria, we excluded all occurrences that were incom-plete verb phrases, or lacked a verb phrase, as in (17).

(17) no never no(LINDSEI-GE E29)

We also excluded all instances of temporal adverbials that occurred in idiomatic multiword-units, such as yet to come or never mind, see, for example (18), as well as constructions such as for ever (= forever), where ever (= wherever) or never the less (= nevertheless), as idiosyncratic spellings that do not match the temporal adverbials of interest here, as illustrated in (19).

(18) GE lines either which annoyed me but <laughs> but never mind yeah I’d I had I’d I(LOCNEC E09)

(19) HD got one then I would do an M A where ever I was offered (LOCNEC E42)

In addition, we excluded complex adverbials that cannot be subsumed under one of the simple usages, like just as. Even though it is used temporally, its meaning is different from plain just, as exemplified in (20).

(20) and she’s just sitting there and he’s painting just as as she .. sits there (LINDSEI-GE GE001)

We also excluded instances that only formally resembled temporal adverbials, but were used in a non-time referential meaning, for instance yet as conjunction, as in (21), just in the sense of ‘simply’, as in (22), or as a self-correction, as in (23), or when used as premodification, as in (24).

(21) being honest and knowing the truths, yet living freely and with choice.(LOCNESS BRSUR1_1)


(22) I came here and like I find Lancaster is just so unfriendly compared to Liver-pool(LOCNEC E10)

(23) I go there like virtually ever= every holiday(LOCNEC E55)

(24) I was as concentrated as never before(ICCI-AUT 0541)

Further, all participle and gerund constructions were ignored, for instance (25) due to the lack of a finite verb, as well as constructions including modal and semi-modal verbs (such as should, must and ought to, as in (26)), because most of them can only be used with the PP but not the SP; we also excluded imperative forms, catenative verbs (seem to + X, appear to X, keep Xing, etc.; cf. Quirk et al. 1985: 137), as well as constructions with implicit subjects, see (27).

(25) actually I actually enjoyed … it having never been to the States before (LOCNEC E51)

(26) so the only way I can ever actually eat again ever … is if I(LOCNEC E41)

(27) She is a loveable person, always in a good mood(ICCI-AUT 0598)

All forms of have got were excluded from the analysis because the construction have got to has deontic meaning (like many modal verbs) and have got has pos-sessive meaning (the difference between have got and got being one of style, and not parallel to the ordinary difference between PP and SP).

Finally, we also excluded clearly erroneous uses of adverbs, such as (28). If there was just a deviant formal realization of the verb form, we classified the instance according to the form of the main verb; for instance, in example (29), we coded the verb form as present.

(28) My whole body was waiting for since from my mom, and then finally I heard her key.(ICCI-AUT 0619)

(29) You’re right on time, because I just spend all my money on a festival. (ICCI-AUT 0669)


2.3 Statistical analysis

2.3.1 Frequency of the present perfect

In order to account for the frequency of the PP, we calculated how often the PP was used in past contexts (PP percent = PP/[PP+SP]*100). To determine whether differences between the various groups (such as written language as used by learners at university level compared to British native speakers at university level) are significant, we computed logistic regression models in R (R Core Team 2014), using the lme4 package (Bates et al. 2015).¹¹ Post-hoc Tukey tests were con-ducted with the glht function from the multcomp package (Hothorn, Bretz, and Westfall 2008).

In the logistic regression models, we used the choice between PP and SP as the dependent variable, and corpus as independent variable. In addition, the influence of the length of instruction at school and university, of a stay abroad and its length were considered as independent variables. Moreover, in contrast to previous studies on the PP in learner language (e.g. Eriksson 2008; Kämmerer 2012; Rogatcheva 2014), we used author/speaker as a random factor. This accounts for the fact that several PP/SP tokens occurring in the same text are not independent. To illustrate this with an example, consider two scenarios. In one we have 100 verbs with past reference produced by the same learner, and find that he uses the PP in 10 % of all cases. In the second case, we have 10 texts pro-duced by 10 learners, each with 10 verbs with past reference. The analysis shows that they use the PP in 10 % of all cases. Although in both scenarios we have 100 data points, the first has poor generalizability, since we only have data from a single learner. The second has better generalizability because data from 10 learn-ers are much more likely to provide an accurate estimate of the linguistic behavior of the whole learner population, although both are based on the same number of tokens. The difference between the first and second scenario is adequately cap-tured by using author/speaker as a random factor, so that variability across individuals, which can be quite substantial in learner language, is taken into account in statistical tests (see also Moulton 1986; Gries 2015).

11 Example function call: glmer(cbind(freq.x,freq.y)~corpus+(1|file_name),data=pp.all,family=binomial).


2.3.2 Use of the present perfect and simple past with temporal adverbials

In a broader analysis, we calculated how often each of the temporal adverbials under consideration (see Section 2.2 above) co-occurs with various time-reference forms and expressed this in percentages. In a narrower analysis, we then calcu-lated the frequency of PP forms as a percentage (PP ratio) of all past references (i.e. PP/[PP+SP]*100) in co-occurrences with the temporal adverbials. Differences between corpora were tested for statistical significance with Chi-square tests, and were Šidák-corrected to adjust for multiple comparisons (Šidák 1967).

Finally, we determined how often the learners and native control groups used the PP with any of the temporal adverbials and expressed that ratio as a percent-age of all PP tokens. Differences between the groups were also tested for signifi-cance with Šidák-corrected Chi-square tests.

3 Results

3.1 Frequency of the present perfect

Our first two research questions were whether the structural similarity between the German Perfekt and the English PP leads to an early use and possibly over-generalization of the PP in L1 German learner English. Alternatively, the PP may emerge later than the SP, as predicted by the morphological developmental hier-archy and the Default Past Tense Hypothesis (see Section 1). For written language, the analysis reveals the following (see Figure 1 and Appendix B): school students in the group from year 5 to 9 use the PP very rarely (1.7 % of all past references), which only increases by year 10 (6.0 %) and further in year 11 (11.9 %). Even this is still lower than the frequency of the PP in native university student essays, with a frequency of 14.2 % and 17.4 % in essays written by British and American univer-sity students, respectively. Surprisingly, essays written by native British school students have a still higher PP frequency (40.4 %), which may be motivated by various factors (see Section 4.1).

Learners in school year 5 to 9 differ significantly from year 10 and year 11 students (z = 4.5, p < 0.001; z = 3.4, p < 0.01). Year 10 and year 11 students do not differ significantly (z = 0.82, p = 0.96). Year 5 to 9 students also differ significantly from the native British and American university students and the British school students (z = 11.1; z = 7.1; z = 17.3; all p < 0.001). Year 10 students differ significantly from the native British but not from the American university students (z = 4.0, p < 0.001; z = 2.5, p = 0.11); they also differ significantly from the British school


students (z = 7.9, p < 0.001). Year 11 students do not differ significantly from the native British and American university students (z = 1.5, p = 0.67; z = 0.9, p = 0.95), but differ significantly from the native British school students (z = 3.6, p < 0.01).

spoken written

native US uni

native GB uni

native GB school

learners uni (early starters)

learners uni (late starters)

learners school (year 11)

learners school (year 10)

learners school (year 5−9)

0 10 20 30 40 0 10 20 30 40Percent PP

Figure 1: Frequency of the PP as a percentage of all past references in spoken and written language.

Focusing now on university students of English with L1 German, we can distin-guish between early and late starters. Students who received less than 10 years of English language instruction at school (“late starters”) use the PP for 8.9 % of all past references, while students who received ten or more years of English lan-guage instruction at school use the PP more than twice as often (19.3 %). However, this difference is not significant (z = –1.5, p = 0.55). The late starters differ signifi-cantly from the native British university students (z = 3.6, p < 0.001) but not from the native US university students (z = 1.5, p = 0.52). By contrast, the early starters do not differ significantly from either group (z = 0.03, p = 1; z = 0.47, p = 0.99).

For spoken English, only data from university students are available. We can again distinguish between late and early starters among the university students, who use the PP with a frequency of 4.7 % and 5.8 %, respectively. This is less than the PP frequency of the control group of native British university students (7.3 %). While the difference between the late starters and the control group is signifi-cant (z = 3.1, p < 0.01), the differences between early starters and the control group (z = 0.7, p = 0.78) and between early and late starters (z = –0.6, p = 0.84) are not significant.


Language experience other than being an early or late starter appears to have little or no effect on the PP frequency of the L1 German university students. In fact, length of instruction at university seems to have no effect: students who had received three years or more of instruction at university have an only marginally and insignificantly higher PP frequency in written English than those with less than three years of instruction (9.6 % vs. 7.2 %, z = 0.9, p = 0.98). Moreover, the results of the statistical test (p = 0.98) even show that the groups perform virtu-ally identically. In spoken language, those with three or more years of instruction at university yield a frequency below those with less than three years, contrary to what the additional instruction at university would suggest (4.5 % vs. 5.9 %, z = 0.7, p = 0.996; no data available for written language).

Results are similar for the effects of a stay abroad. Contrary to what might be expected, university students who had spent time in an English-speaking country use the PP marginally less frequently in spoken English than those without this experience (4.7 % vs. 5.3 %, z = 0.2, p > 0.999). Moreover, whether students spent time in Great Britain or the United States influenced the results only minimally (4.3 % vs. 5.1 %, z = 0.1, p = 1.0).¹² Consequently, the results for the present dataset indicate that only the number of years spent learning English at school, not the duration of instruction at university or time spent abroad, influence the frequency of the PP in the language of university students of English with L1 German.

Looking now at stylistic variation among the university students with L1 German, both the late and early starters use the PP more frequently in written than in spoken language, but this difference is only significant for the late start-ers (8.9 % vs. 4.7 %, z = 4.1 p < 0.001; 19.3 % vs. 5.8 %, z = 1.7, p = 0.47). The native British control group also uses the PP significantly more often in writing than in speech (14.2 % vs. 7.3 %, z = 4.2, p < 0.001).

In summary, the results reveal that learners of English with L1 German use the PP very rarely up until school year 9, with increases in year 10 and again in year 11, where PP frequency approaches the range of native usage. University stu-dents of English use the PP as frequently as native speakers only if they received 10 or more years of English instruction at school, but fall short of native usage rates if they received less instruction at school. Non-native university students also show stylistic variation, with higher PP frequencies in written than in spoken language. Overall, these results support the late emergence of the PP along the lines of the developmental pattern established above as well as according to the Default Past Tense Hypothesis (see Section 1). Moreover, additional input in the form of an early start of English language instruction increases the chances of

12 Interactions between the demographic variables could not be investigated because the num-ber of subjects is too low for such analyses.


reaching native-like rates of PP usage, but additional input at a later stage (in the form of a term abroad or length of instruction at university) seems to have no sig-nificant effect. Finally, the analysis also suggests that learners at university level reproduce native register differences in the sense that they use the PP more often in written than in spoken language.

3.2 Use of the present perfect and simple past with temporal adverbials

3.2.1 Present perfect and simple past in competition with other time-reference forms

While the previous section has shown that less advanced learners with L1 German use the PP less often than native speakers, it has not uncovered specific contexts where this happens. This is the aim of this section, where we will look at the usage of temporal adverbials together with the PP, SP and other time-reference forms (in learner grammars commonly referred to as “tenses”).

We will first look at what time-reference forms are used together with tem-poral adverbials in general, and then look at specific temporal adverbials (see Appendix C for percentages). The native speakers overall use the PP in almost half of all cases (46.5 %), followed by the present (26.3 %) and the simple past (16.2 %). The past perfect and future forms (will and going to) occur very rarely (6.3 % and 2.3 %, respectively). By contrast, the simple present (31.3 %) is the form that is used most frequently by the learners. They also use the PP much less often (28.8 %) and the SP much more frequently (28.7 %) than the native group. The past perfect (8.0 %) and the future (1.5 %) occur roughly as frequently as in native language.

Focusing now on which time-reference forms are used in conjunction with specific temporal adverbials (see Figure 2 and Appendix C),¹³ we find that in native spoken English (university) and native written English (school) the PP is the most frequent time-reference form for seven out of eight temporal adverbials. In native written English (university, British) it is the most frequent form for four temporal adverbials, and the second most frequent for two more. Moreover, the

13 In this part of the analysis, we exclude the native written English (US) data because very few of the temporal adverbials in question occurred in this corpus. For the learners, we also do not differentiate between different years for the school students, and early and late learners for the university students, since this would mean having less than ten tokens for some of the adverbials in these groups.


yet

just

sinc

eal

read

yal

way

sre

cent

lyne

ver

ever

lear

ner s

choo

l writ

ten

lear

ner u

ni s

poke

n

lear

ner u

ni w

ritte

n

nativ

e GB

sch

ool w

ritte

n

nativ

e GB

uni

spo

ken

nativ

e GB

uni

writ

ten

025

5075

100

025

5075

100

025

5075

100

025

5075

100

025

5075

100

025

5075

100

025

5075

100

025

5075

100

Perc

ent

Tim

e re

fere

nce

form

s

Pres

ent P

erfe

ct (P

P)Si

mpl

e Pa

st (S

P)Pa

st P

erfe

ctPr

esen

tFu

ture

Figu

re 2

: Sel

ecte

d te

mpo

ral a

dver

bial

s and

thei

r co-

occu

rrenc

e wi

th th

e PP

, SP

and

othe

r tim

e-re

fere

nce

form

s in

L1 G

erm

an le

arne

r Eng

lish

and

nativ

e En

glis

h (e

ach

bar s

hows

the

tota

l num

ber o

f occ

urre

nces

of a

tim

e ad

verb

ial i

n th

e re

spec

tive

corp

us, w

ith co

lors

indi

catin

g ho

w of

ten

the

time

adve

r-bi

al co

-occ

urre

d wi

th th

e va

rious

tim

e re

fere

nce

form

s).


present is the most frequent time-reference form for two temporal adverbials and the second most frequent one in four more cases.

In written learner English (university), the PP is the most frequently used form in five cases, and the second most frequently used in the remaining three. In spoken learner English (university), the PP is the most frequently used form in only one case, and second most frequently used one in five more cases. In written learner English (school), the PP is the most frequently used time-reference form in three cases, and never the second most frequent one, but the present is the most frequently used form in five cases.

Overall, these results imply that school-age learners use the present more often with these temporal adverbials than the native control group, a finding that converges with earlier studies (Bardovi-Harlig 1997: 403). Learners at uni-versity, by contrast, seem to have reached a stage where the distribution of the main time-reference forms appears to be within the scope of variation in native language. We will now move on to a comparison of the frequency of the PP and SP to check whether this closer analysis also supports this conclusion.

3.2.2 Present perfect and simple past in competition with each other

In this section, we compare how frequently the learners and the native control group use the PP and the SP with each of the temporal adverbials. For example, a PP ratio of 33 % for the learners (spoken, at university level) for just means that 33 % of all instances of the adverbial just that refer to the past are used with the PP, and the remaining 67 % with the SP. By contrast, the native control group has a PP ratio of 63 %, which means that 63 % of the occurrences of just referring to the past are used with the PP.

Overall, in spoken English at university level (see the top panel of Figure 3; Appendix D), the PP ratio of the learners is significantly lower than the PP ratio of the native British control group for the item always (χ² = 26.4, p < 0.0001, df = 1). Also, in five cases the PP ratio of the learners is lower than among the native speakers, but the difference is not significant (already, χ² = 5.5, not significant [n.s.]; ever, χ² = 1.7, n.s.; just, χ² = 1.3, n.s.; since, χ² = 1.2, n.s.; yet, χ² = 0.2, n.s.). Only never has a slightly (and insignificantly) higher PP ratio among the learn-ers than among the natives (χ² = 0.02, n.s.). The adverbial recently occurs only in native speech, but not in the learner corpus.

In written language (see the bottom panel of Figure 3), the learners at school level have lower PP ratios than the native British control group for six out of seven items, a difference that is significant for already, always and since (already, χ² = 7.8, p < 0.05; always, χ² = 7.6, p < 0.05; ever, χ² = 0, n.s.; just, χ² = 1.2, n.s.; never,


yet

just

sinc

eal

read

yal

way

sre

cent

lyne

ver

ever

lear

ners

uni

nativ

e GB

uni

spoken

025

5075

100

025

5075

100

025

5075

100

025

5075

100

025

5075

100

025

5075

100

025

5075

100

025

5075

100

Perc

ent P

P

yet

just

sinc

eal

read

yal

way

sre

cent

lyne

ver

ever

lear

ners

sch

ool

lear

ners

uni

nativ

e GB

sch

ool

nativ

e GB

uni

written

025

5075

100

025

5075

100

025

5075

100

025

5075

100

025

5075

100

025

5075

100

025

5075

100

025

5075

100

Perc

ent P

P

Figu

re 3

: PP

ratio

(per

cent

age

of th

e PP

ove

r all

PP a

nd S

P to

kens

) for

eig

ht te

mpo

ral a

dver

bial

s in

spok

en (t

op) a

nd w

ritte

n (b

otto

m) l

earn

er a

nd n

ativ

e la

ngua

ge.


χ² = 1.0, n.s.; since, χ² = 12.7, p < 0.001). Only yet is used as frequently with the PP by the learners as by the native speakers, and recently does not occur in the school learner corpus.

Learners at university level have lower PP ratios in written language than the native British control group for six out of eight temporal adverbials, and the dif-ference is significant for always and since (already, χ² = 2.5, n.s.; always, χ² = 7.4, p < 0.05; just, χ² = 0.2, n.s.; recently, χ² = 1, n.s.; since, χ² = 7.1, p < 0.05; yet, χ² = 0, n.s.). For ever, the learners have lower PP ratios than the native British control group (χ² = 2.1, n.s.). For never, the PP ratio is almost identical for the two groups.

The data also reveal a progression from lower to higher PP ratios in learner language from school to university level for four temporal adverbials, which, however, is only significant for since (already, χ² = 1.0, n.s.; always, χ² = 0.9, n.s.; just, χ² = 1.7, n.s.; since χ² = 7.6, p < 0.05). For ever and never there is virtually no development in PP frequency, for yet there is actually an insignificant decrease in the PP ratio, and recently is not used by the non-native school students.

Overall, we find that learners both at school and at university use the PP less often than native speakers in contexts where the temporal adverbials in question occur. This seems to be particularly the case for the temporal adverbials always, since and already, although it needs to be stressed that this list might be incom-plete; some of the differences between learners and native speakers may not be significant simply because some of the temporal adverbials are not very frequent in our data. Moreover, we also find that there is development from the level of the school learners to the university learners in the sense of a closer approximation to native usage rates. It has to be noted that there is also variability in native-speaker PP frequencies for the individual items (Werner 2013), and therefore what is con-sidered to be native-speaker norms or target-like usage ought to be reviewed.

3.3 Use of the present perfect with and without temporal adverbials

Our third research question was whether less advanced learners rely more often on the support of temporal adverbials when using the PP than more advanced learners. As Figure 4 shows, in spoken English, learners with L1 German at uni-versity level use the PP together with temporal adverbials slightly and insignifi-cantly less often (27.4 %) than the native British university students (30.1 %; n.s.).

Learners of school age use the PP very often in conjunction with temporal adverbials (46.1 %) in written language, and do so significantly more often than native British school students (15.4 %; χ² = 20.3, p < 0.0001; see also Appendix E). L1 German learners at university use the PP in 30.5 % of all cases with temporal


adverbials. This is slightly (but insignificantly) more often than native British uni-versity students (26.6 %; n.s.), and significantly more often than native US univer-sity students (8.8 %; χ² = 7.0, p < 0.05). Overall, then, this implies a development among the learners from a very frequent use of temporal adverbials in conjunc-tion with the PP at school level towards a lower percentage among learners at university. Although the difference between learners at school and at university is not significant, the lower percentage among learners at university falls within the native range, while the higher percentage among learners at school does not fall within the native range.

spoken written

learners school

learners uni

native GB school

native GB uni

native US uni

0 25 50 75 100 0 25 50 75 100

Percent PP with time adverbials

Figure 4: Percentage of the occurrences of the PP that are specified by any one of eight tempo-ral adverbials.

4 DiscussionThis study set out to determine PP usage rates of L1 German learners of English in contrast to native speakers, and whether the PP emerges later than the SP in learner language. Moreover, we wanted to determine whether learner variables such as the amount of previous instruction and time spent in an English-speaking country influence the frequency of the PP. The frequency of the PP was measured


as a percentage of all past references in general, and in conjunction with tem-poral adverbials that allow for a variable use of the PP and SP in native English. Finally, we also determined how much learners rely on the support of temporal adverbials in the use of the PP.

4.1 Frequency of the present perfect

The analysis revealed that L1 German school students in years 5 to 9 use the PP very rarely, and instead mostly use the SP to refer to past events and actions. The frequency of the PP increases substantially in years 10 and 11, and eventu-ally comes close to the range of native-like PP frequencies. University students of English with L1 German differed in the usage of the PP based on whether they are early starters (with 10 or more years of English instruction at school) or late starters (less than 10 years of English instruction at school). Only the early start-ers showed native-like PP frequencies. By contrast, other learner variables, such as whether students had spent an extended period of time in an English-speak-ing country or how long they had attended university, did not influence the fre-quency of the PP.

Our results support both the morphological acquisition sequence (SP before PP) developed in the form-oriented view and the Default Past Tense Hypothesis (see Section 1), which predicts that learners first acquire the SP and only later the PP, regardless of whether their L1 has a parallel structure. This is surprising, given that German has a structure (the Perfekt) that is formally identical to the English PP and partly shares its resultative semantic space. In German, the Perfekt is also used relatively more frequently than its competitor, the Präteritum, compared to the English PP and SP. Based on these similarities, and the comparatively more prominent role the Perfekt plays in German grammar, a facilitating role of the L1 and even an initial overuse of the PP could be expected in L1 German learner English. That this is not borne out by the data indicates that L1 influence does not play a major role in the acquisition of the PP, although this does not preclude that speakers of an L1 without a parallel structure to the PP acquire it even later than learners with L1 German.

However, a limitation of our study is that we did not take into consideration the proportion of accurate vs. erroneous uses of the SP and PP. This might, in turn, mirror similar structures in the L1, as proposed in previous research (e.g. Kämmerer 2012). Neither have we considered whether differences in the topics of the essays, which vary across school years in our data, influenced the results. Whereas the youngest students wrote about topics that do not directly prompt ref-erences to the past (such as how they would spend money won from the lottery),


students in the later years were set more complex topics, potentially necessitat-ing more references to past actions and events. However, it is not immediately clear how this could have influenced the frequency of the PP compared to the SP. A topic focusing on future actions (such as the lottery fortune) would make references to the past in general less likely, rather than de- or increasing the pro-portion of the PP compared to all references to the past. We thus conclude that variation in essay topic is theoretically a potential confound, but in practice may be of limited influence.

The present study confirms previous findings by Bardovi-Harlig (1997: 400) and Liszka (2004: 224), as well as corpus-based results by Davydova (2011: 277) and Rogatcheva (2014: 232), who also found underuse of the PP in (L1 German) learner language. Our analysis extends these results with a quasi-longitudinal perspective, showing that only the most advanced learners approach native-like usage in the frequency of the PP. In spite of this difference between early and late starters, our results indicate that both groups successfully reproduce native pat-terns to a certain extent. Both learners at university and native speakers use the PP more frequently in writing than in spoken language. Again, L1 influence seems to be restricted, since the German Perfekt is commonly used more frequently in speech than in writing.

The interpretation of our results is somewhat complicated by the fact that there is a great deal of variation in PP frequencies even in native varieties of English. The native control data showed only a relatively small difference between American and British usage, which, moreover, runs counter to previous results that show a higher frequency of the PP in British than American English (see, e.g., Yao and Collins 2012: 399; Werner forthcoming; see also Fuchs, this volume). More importantly, native British school-leaving exams showed an extremely high frequency of the PP, much higher than in British university student writing. Con-sidering previous results on the frequency of the PP in British English in general and in British English student exams and essays in particular (Werner 2014; Werner and Fuchs 2016), the extremely high frequency that we found in British school-leaving exams should be interpreted cautiously. It is conceivable that it is largely teaching-induced in terms of an emphasis on (implicitly) replicating what is conceived as a normative British standard that posits high PP usage, while atti-tudes at universities are apparently more relaxed. The essays may have also come from only a single or a few schools where a (prescriptive) focus on forms and formality in writing is prevalent, with an overemphasis on the PP as a form that is conceptualized as “typically written” in school grammars. It may also be spec-ulated that the essay topics predetermined increased PP rates.


4.2 Frequency of the present perfect and other time-reference forms in combination with temporal adverbials

Whereas native speakers by and large prefer the PP as the time-reference form to be used with temporal adverbials that allow variable use, the results showed that learners tend to use the present and SP much more frequently with these adverbials. The analysis of particular temporal adverbials also revealed that school-age learners use the SP rather than the PP much more frequently than the native control group. Learners at university level approximate, but still fall short of, native PP usage rates in both written and spoken language. Results were par-ticularly clear-cut for already, always and since. Ever, never and yet combine with the PP in learner language as often as in native English. For just and since, results were not significant, but suggestive of an underuse of the PP in learner English (both at school and university level).

These results show that, also in these contexts, the PP is overall used less frequently by learners at both school and university level, compared to native frequencies of PP use. Given the fact that English language instruction in Ger-man-speaking countries often relies on a signal word approach that teaches the use of the PP with many of the temporal adverbials in question (Davydova 2011: 282, 287; Rogatcheva 2014: 241), these results are surprising. We would have expected at least advanced learners at university level to show native-like or even exaggerated PP frequencies in combination with these temporal adverbials, while PP frequencies in general (whether or not in combination with temporal adverbials) lag behind.

4.3 Frequency of temporal adverbials in combination with the PP

Lastly, we considered how often the PP combines with temporal adverbials, and whether there is any development in this question between intermediate and advanced learners. The results show that learners at school level often rely on the support of temporal adverbials when using the PP, but that university level learners come within the range of native variation in this variable. This confirms the concept-oriented view (see Section 1) establishing that learners heavily rely on the support (“scaffolding”) of temporal adverbials in the expression of tempo-rality at an early-intermediate period in the language learning process (see, e.g., Bardovi-Harlig 2014: 130). Another, not mutually exclusive explanation is that the signal word approach (see Section 4.2) may lead to the more frequent combina-tion of temporal adverbials with the PP.


Our results are particularly striking given that we focused only on specific temporal adverbials that, in native speech, are more likely to combine with the PP. If we had extended our view to definite temporal adverbials usually com-bining with the SP (such as yesterday, a month ago, etc.), we also would have been able to describe combinations of these with the PP in learner English (recall example (4) above; see further, e.g., Bardovi-Harlig 1997: 403; Housen 2002a: 96; Roberts and Liszka 2013: 430; Rogatcheva 2014: 235).

4.4 Reasons for the late emergence of the present perfect in learner language

The long period of underuse of the PP in L1 German learner language begs the question of how it can be explained. While the Default Past Tense Hypothesis predicts late emergence of the PP in learner language, it does not offer any expla-nations as to why this might be the case.

4.4.1 L1 influence

As stated above, L1 German influence would rather be expected to encourage early emergence of the PP in learner English. The relative underuse of the PP then leads us to the conclusion that access to L1 structures is blocked or delayed in the case of the PP. To be consistent, we would then have to conclude that access to L1 structures is blocked or delayed in its totality, an unlikely interpretation, given the large body of literature on positive L1 transfer (e.g. Gass and Selinker 1992; Yu and Ren 2013). Alternatively, perhaps access to L1 structures might be blocked in the case of the PP only (for instance in view of the Default Past Tense Hypothesis), a contention for which there is little or no independent evidence.

Alternatively, the PP may be a structure that for some reason (see below) is extremely challenging to learners, and positive L1 transfer supports learners only at a later point. This explanation could be tested by comparing learners of English where past-time reference in the L1 is structurally and semantically similar to English (such as Spanish; see, e.g., Kämmerer 2012) with learners without such a structure in their L1. The latter would be expected to approximate native frequen-cies in PP usage even later than the former.


4.4.2 Complexity

The PP can be regarded as a structure that is more complex than its competitor, the SP (see Section 1). The PP consists of a combination of auxiliary and past participle, while the SP consists only of the inflected main verb, making it argu-ably less complex than the PP (Davydova 2011: 288–289). Learners focus on less complex structures in the early and intermediate stages of language learning (see Section 1.1), which would predispose them towards use of the SP rather than the PP. Again, this argument is in line with the Default Past Tense Hypothesis and becomes even more convincing when the PP is compared to other structures of the English language in terms of their complexity. In particular, the progres-sive also consists of a combination of an auxiliary and a deverbal form (V+ing), making it as complex as the PP. Previous research on the progressive in spoken (learner) language also revealed a highly significant underuse in LINDSEI-GE vs. LOCNEC (Dose-Heidelmayer and Götz 2016). However, these findings are different in (learner) writing, as the progressive has been shown to be highly overused in GICLE vs. LOCNESS (see, e.g., Wulff and Römer 2009). Thus, even though com-plexity seems to play a part in L1 acquisition (Bardovi-Harlig 1997: 412), it cannot fully account for the late emergence and potential avoidance of the PP, as forms that are structurally equally complex (such as the progressive as well as the past perfect) emerge even later.

4.4.3 Input frequency

Several theories of second language acquisition hold that input is crucial to the development of learner language (see, e.g., Slobin 1994: 120; Wulff et al. 2009: 355). Since the PP is much less frequent in English than the SP, this might prompt learners to conclude that past events and actions are described by default with the SP. Moreover, since the use of the SP instead of the PP does not lead to the loss of semantic content, there is little functional motivation for learners to acquire the restrictions on PP usage. This explanation could be tested with connection-ist networks and other machine learning approaches (see, e.g., Theijssen et al. 2013). In addition, comparisons with the acquisition of equally rare structures could shed light on the question of whether frequency in the input is indeed a viable explanation for the late emergence of the PP in learner language.


4.4.4 Summary

In summary, the present study was not able to fully reveal whether or not there is at least some positive L1 transfer from German in the acquisition of the PP, a ques-tion that should be tested in future research on various factors that have been shown to determine L1 influence (see Paquot 2013). In any case, explanations are needed for why L1 influence should be blocked in total or not be exerted until a late stage in the acquisition of the PP. Complexity may be a good candidate since the SP is less complex than the PP. Additionally, the low frequency of the PP in the input that learners receive is likely to contribute to explaining the late emer-gence of the PP. Nevertheless, it is unclear whether low frequency in the input can explain all of the facts, and future research should seek out further explanations.

5 Implications, conclusion and outlookThe present study revealed that intermediate and advanced German-speaking learners underuse the PP in both writing and speech, regardless of their formal proficiency level (i.e. school or university learners) and of a stay abroad in an English-speaking country. These results are in stark contrast to our initial hypoth-eses. The only learner group that comes close to the native target is the subgroup of university students who started English early at school. Thus, one first general conclusion is that the PP is a highly challenging feature for almost all subgroups of German learners of English investigated.

Our results on the influence of learner variables on the frequency of the PP among university students of English have important implications for both lan-guage pedagogy and policy. It needs to be highlighted that this learner group is likely, on average, to be more proficient at the beginning of their studies than the general population when they leave secondary school. Many English depart-ments do not admit students with low proficiency in English. Even where this is not the case, it is students with higher proficiency who are more likely to elect to specialize in the English language at the tertiary level than less proficient students. Against this background, it is surprising that a subgroup of university students (those with less than 10 years of English instruction at school) do not reach native-like values. Even more surprisingly, length of instruction at univer-sity (cf. Bardovi-Harlig 1997: 391) and time spent in an English-speaking country appear to have no statistically measurable effect at all. This suggests that English language instruction should focus more on the usage of the PP if the aim is to approximate frequency distributions comparable to those in established standard


models (particularly British usage), as is still common practice in German (and Austrian) schools and universities (Davydova 2011: 275). Alternatively, if we con-ceive of English as a pluricentric language with great differences in PP frequency across varieties (see, e.g., Fuchs, this volume), PP frequencies that diverge from British (and American) English might be permissible as a target in EFL instruc-tion. Moreover, the fact that the choice between PP and SP is often stylistic rather than semantic (in the sense of truth-conditional semantics) might motivate edu-cators to focus on structures of the English language that are more easily acquired and more important for effective communication. This naturally applies to other areas of grammar and all levels of linguistic description.

Whatever choices educators may make in this respect, the finding that uni-versity students who received EFL instruction prior to year 3 in school (“early starters”) reach native-like values in PP frequency has additional implications. Our results indicate that only an early start, but not further instruction at uni-versity, helps in approximating native-like usage in a (non-immersion) L2 learn-ing environment where learners typically receive a very limited amount of input (usually a maximum of five hours per week and often much less). This result concurs with previous research on the effect of early foreign language instruction in non-immersion L2 learning environments (Nikolov and Djigunović 2006; Dji-gunović, Nikolov, and Ottó 2008) and promotes the advocation of starting English instruction as early as possible in primary school (e.g. Cameron 2001; Nikolov 2009). It remains to be tested, however, whether the case is similar for other struc-tures in the English of L1 German learners.

Our results on the frequency of temporal adverbials in combination with the PP appear to support proponents of a signal word approach. Learners rely on the support of temporal adverbials in the use of the PP at the intermediate stages of English language learning (Schlüter 2002: 340). This implies that, without the pedagogic focus on temporal adverbials, PP usage rates might be even lower in early and intermediate learner English. However, suggestive as the results may be, they do not establish a causal relationship between pedagogic approach and lin-guistic pattern in learner language. Note further that in highly advanced stages, the focus should involve other contextual cues for the choice of time-reference forms (Eriksson 2008: 143), as native(-like) PP usage is in fact characterized by a widespread absence of additional temporal marking (see, e.g., Werner 2014: 136).

The present study mainly focused on the frequencies of use of the PP and the SP by different learner populations, in different modes and at different profi-ciency levels, and we were able to detect some general trends in the learners’ use of the PP that hold true across learner populations and across temporal adverbi-als. In future analyses, however, we will have to dig deeper in order to find expla-nations for the substantial underuse of PP in our data and see if this underuse


is also reflected across task types and individual learners (Bardovi-Harlig 1997: 385–386). On a related note, it would obviously be highly desirable to expand our perspective to other learner varieties in order to discriminate between German learner language on the one hand and foreign language acquisition universals on the other. This could also include potential priming effects in spoken language, where the learners appear to repeat, for example, a specific verb in the PP that the interviewer or conversation partner used in the preceding turn.

One promising road ahead would be a detailed error analysis to reveal if the learners show a frequent non-use in a required context of the PP compared to native speakers. At the same time, we could check if they even use the PP where a different form would normally be preferred by native speakers (see Granger 1999). This would help us see if avoidance strategies come into play. Individual error profiles would provide valuable insights, particularly into the case of learners who deviate most significantly from average native speaker usage. The ensuing more fine-grained results would allow us to draw better conclusions on PP use in intermediate and advanced German learners’ speech and writing. In a follow-up study, a more comprehensive qualitative error analysis would be very useful in determining which functions and contexts of the PP are the biggest “stumbling blocks” for German learners. This would include an identification of the verbs and temporal adverbials which make learners particularly error-prone. It would also be worth comparing the individual learners’ performances of the PP with the same learners’ performances of other challenging grammatical features, such as the distinction between the simple present and the present progressive, in order to see if there are learners whose overall grammatical performance is poor.

Rather than merely contrasting writing and speech, our analysis of PP fre-quencies could lead to a frequency and accuracy comparison in future research into the different task types in the corpora. For the spoken corpora, the types would be (i) talking about a set topic, (ii) discussion and (iii) picture descrip-tion; for the written corpora they would be (i) the distinction between argumen-tative essay topics and (ii) timed vs. untimed essays. Learner performance has previously been shown to be very mode- and task-sensitive (see, e.g., Foster and Skehan 1996; Tracy-Ventura and Myles 2015; Dose-Heidelmayer and Götz 2016) and the frequencies and learners’ accuracy levels for the different interview parts, essay types, and essay topics might turn out to deviate significantly across corpora.


AcknowledgementsWe would like to thank Fabian Vetter for support with regard to tagging the corpus data, two anonymous reviewers for comments on the draft version of this chapter, Sandra C. Deshors for bringing our attention to the issue of priming, and Rosemary Bock for proofreading.

ReferencesAnthony, Laurence. 2014. AntConc (Version 3.4.3). Tokyo: Waseda University. http://www.

laurenceanthony.net/ (accessed 10 January 2015)Antinucci, Francesco & Ruth Miller. 1976. How children talk about what happened. Journal of

Child Language 3(2). 167–189.Bardovi-Harlig, Kathleen. 1997. Another piece of the puzzle: The emergence of the present

perfect. Language Learning 47(3). 375–422.Bardovi-Harlig, Kathleen. 1999. From morpheme studies to temporal semantics: Tense-aspect

research in SLA. Studies in Second Language Acquisition 21(3). 341–382.Bardovi-Harlig, Kathleen. 2000. Tense and aspect in second language acquisition: Form,

meaning and use. Oxford: Blackwell.Bardovi-Harlig, Kathleen. 2014. Documenting interlanguage development. In ZhaoHong Han &

Elaine Tarone (eds.), Interlanguage: Forty years later, 127–146. Amsterdam: Benjamins.Bates, Douglas, Martin Mächler, Benjamin Bolker & Steven Walker. 2015. Fitting linear

mixed-effects models using lme4. Journal of Statistical Software 67(1). 1–48.Brand, Christiane & Susanne Kämmerer. 2006. The Louvain International Database of Spoken

English Interlanguage (LINDSEI): Compiling the German component. In Sabine Braun, Kurt Kohn & Joybrato Mukherjee (eds.), Corpus technology and language pedagogy: New resources, new tools, new methods, 127–140. Frankfurt: Peter Lang.

Bulut, Turkay. 2011. The licensing of the English present perfect tense by Turkish adults. International Journal of Humanities and Social Science 1(15). 221–227.

Cameron, Lynne. 2001. Teaching languages to young learners. Cambridge: Cambridge University Press.

de Cillia, Rudolf & Hans-Jürgen Krumm. 2010. Fremdsprachenunterricht in Österreich [Foreign language teaching in Austria]. Sociolinguistica 24. 153–169.

Collins, Laura. 2002. The roles of L1 influence and lexical aspect in the acquisition of temporal morphology. Language Learning 52(1). 43–94.


Davydova, Julia. 2012. Englishes in the Outer and Expanding Circles: A comparative study. World Englishes 31(3). 366–385.

De Cock, Sylvie. 2004. Preferred sequences of words in NS and NNS speech. Belgian Journal of English Language and Literatures 2. 225–246.

Djigunović, Jelena M., Marianne Nikolov & István Ottó. 2008. A comparative study of Croatian and Hungarian EFL students. Language Teaching Research 12(3). 433–452.


Dose-Heidelmayer, Stefanie & Sandra Götz. 2016. The progressive in advanced spoken learner language: A corpus-based analysis of use and misuse. International Review of Applied Linguistics in Language Teaching 54(3).


Eriksson, Andreas. 2008. Tense and aspect in advanced Swedish learners’ written English. Göteborg: University of Göteborg.

Foster, Pauline & Peter Skehan. 1996. The influence of planning and task type on second language performance. Studies in Second Language Acquisition 18(3). 299–323.

Gass, Susan M. & Larry Selinker (eds.). 1992. Language transfer in language learning. Amsterdam: Benjamins.

Garside, Roger & Nicolas Smith. 1997. A hybrid grammatical tagger: CLAWS4. In Roger Garside, Geoffrey Leech & Tony McEnery (eds.), Corpus annotation: Linguistic information from computer text corpora, 102–121. London: Longman.

Gathercole, Virginia C. 1986. The acquisition of the present perfect: Explaining differences in the speech of Scottish and American children. Journal of Child Language 13(3). 537–560.

Gilquin, Gaëtanelle, Sylvie De Cock & Sylviane Granger. 2010. The Louvain International Database of Spoken English Interlanguage. Louvain-la-Neuve: Presses Universitaires de Louvain.

Granger, Sylviane. 1996. From CA to CIA and back: An integrated approach to computerized bilingual and learner corpora. In Karin Aijmer, Bengt Altenberg & Matts Johansson (eds.), Languages in contrast: Text-based cross-linguistic studies, 37–51. Lund: Lund University Press.

Granger, Sylviane. 1999. Use of tenses by advanced EFL learners: Evidence from an error-tagged computer corpus. In Hilde Hasselgård & Signe Oksfjell (eds.), Out of corpora: Studies in honour of Stig Johansson, 191–202. Amsterdam: Rodopi.

Granger, Sylviane. 2004. Computer learner corpus research: Current state and future prospects. In Ulla Connor & Thomas A. Upton (eds.), Applied corpus linguistics: A multidimensional perspective, 123–145. Amsterdam: Rodopi.

Granger, Sylviane. 2015. Contrastive interlanguage analysis: A reappraisal. International Journal of Learner Corpus Research 1(1). 7–24.

Granger, Sylviane, Estelle Dagneaux, Fanny Meunier & Magali Paquot. 2009. The International Corpus of Learner English: Version 2. Louvain-la-Neuve: Presses Universitaires de Louvain.

Gries, Stefan Th. 2015. Some current quantitative problems in corpus linguistics and a sketch of some solutions. Language and Linguistics 16(1). 93–117.

Hothorn, Torsten, Frank Bretz & Peter Westfall. 2008. Simultaneous inference in general parametric models. Biometrical Journal 50(3). 346–363.

Housen, Alex. 2000. Verb semantics and the acquisition of tense-aspect in L2 English. Studia Linguistica 54(2). 249–259.

Housen, Alex. 2002a. A corpus-based study of the L2-acquisiton of the English verb system. In Sylviane Granger, Joseph Hung & Stephanie Petch-Tyson (eds.), Computer learner corpora, second language acquisition, and foreign language teaching, 77–116. Amsterdam: Benjamins.

Housen, Alex. 2002b. The development of tense-aspect in English as a second language and the variable influence of inherent aspect. In Rafael Salaberry & Yasuhiro Shirai (eds.), The L2 acquisition of tense-aspect morphology, 155–197. Amsterdam: Benjamins.



Jenkins, Jennifer. 2000. The phonology of English as an international language. Oxford: Oxford University Press.

Kämmerer, Susanne. 2012. Interference in advanced English interlanguage: Scope, detect-ability and dependency. In James Thomas & Alex Boulton (eds.), Input, process and product: Developments in teaching and language corpora, 284–297. Brno: Masaryk University Press.

Kennedy, Sara & Pavel Trofimovich. 2008. Intelligibility, comprehensibility, and accentedness of L2 speech: The role of listener experience and semantic context. The Canadian Modern Language Review 64(3). 459–489.

Kenworthy, Joanne. 1987. Teaching English pronunciation. New York: Longman.Klein, Wolfgang. 1995. The acquisition of English. In Rainer Dietrich, Wolfgang Klein & Colette

Noyau (eds.), The acquisition of temporality in a second language, 31–70. Amsterdam: Benjamins.

Klein, Wolfgang, Rainer Dietrich & Colette Noyau. 1995. Conclusions. In Rainer Dietrich, Wolfgang Klein & Colette Noyan (eds.), The acquisition of temporality in a second language, 261–280. Amsterdam: Benjamins.

König, Ekkehard & Volker Gast. 2012. Understanding English-German contrasts. Berlin: Schmidt.

Kortmann, Bernd & Kerstin Lunkenheimer (eds.). 2013. The electronic world atlas of varieties of English. Leipzig: Max Planck Institute for Evolutionary Anthropology. http://ewave-atlas.org (accessed 10 March 2015).

Liszka, Sarah Ann. 2004. Exploring the effects of first language influence on second language pragmatic processes from a syntactic deficit perspective. Second Language Research 20(3). 212–231.

Moulton, Brent R. 1986. Random group effects and the precision of regression estimates. Journal of Econometrics 32(3). 385–397.

Mukherjee, Joybrato & Jan-Mark Rohrbach. 2006. Rethinking applied corpus linguistics from a language-pedagogical perspective: New departures in learner corpus research. In Bernd Kettemann & Georg Marko (eds.), Planing, gluing and painting corpora: Inside the applied corpus linguist’s workshop, 205–232. Frankfurt: Peter Lang.

Nikolov, Marianne (ed.). 2009. The age factor and early language learning. Berlin: Mouton de Gruyter.

Nikolov, Marianne & Jelena Mihaljević Djigunović. 2006. Recent research on age, second language acquisition, and early foreign language learning. Annual Review of Applied Linguistics 26. 234–260.

Odlin, Terence & Cristina Alonso Vázquez. 2006. Meanings in search of the perfect form: A look at interlanguage verb phrases. Rivista di Psicolinguistica Applicata 6(3). 53–63.

Paquot, Magali. 2013. Lexical bundles and L1 transfer effects. International Journal of Corpus Linguistics 18(3). 391–417.


R Core Team. 2014. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing.


Roberts, Leah & Sarah Ann Liszka. 2013. Processing tense/aspect-agreement violations on-line in the second language: A self-paced reading study with French and German L2 learners of English. Second Language Research 29(4). 413–439.

Rogatcheva, Svetlomira. 2014. Aspect in learner writing: A corpus-based comparison of advanced Bulgarian and German learners’ written English. Giessen: University of Giessen dissertation.

Schlüter, Norbert. 2002. Present perfect: Eine korpuslinguistische Analyse des englischen Perfekts mit Vermittlungsvorschlägen für den Fremdsprachenunterricht [Present perfect: A corpus-linguistic analysis of the English perfect with suggestions for foreign language teaching]. Tübingen: Narr.


Shirai, Yasuhiro. 2009. Temporality in first and second language acquisition. In Wolfgang Klein & Ping Li (eds.), The expression of time, 167–193. Berlin: Mouton de Gruyter.

Shirai, Yasuhiro & Roger W. Andersen. 1995. The acquisition of tense-aspect morphology: A prototype account. Language 71(4). 743–762.

Šidák, Zbynek K. 1967. Rectangular confidence regions for the means of multivariate normal distributions. Journal of the American Statistical Association 62(318). 626–633.

Slobin, Dan I. 1994. Talking perfectly: Discourse origins of the present perfect. In William Pagliuca (ed.), Perspectives on grammaticalization, 119–133. Amsterdam: Benjamins.

Svalberg, Agneta M. & Hjh Fatimah bte hj awg Chuchu. 1998. Are English and Malay worlds apart? Typological distance and the learning of tense and aspect concepts. International Journal of Applied Linguistics 8(1). 27–60.

Swan, Michael. 1994. Design criteria for pedagogic language rules. In Martin Bygate, Alan Tonkyn & Eddie Williams (eds.), Grammar and the language teacher, 45–55. New York: Prentice Hall.

Theijssen, Daphne, Lois ten Bosch, Lou Boves, Bert Cranen & Hans van Halteren. 2013. Choosing alternatives: Using Bayesian networks and memory-based learning to study the dative alternation. Corpus Linguistics and Linguistic Theory 9(2). 227–262.

Tono, Yukio. 2012. International Corpus of Crosslinguistic Interlanguage: Project overview and a case study on the acquisition of new verb co-occurrence patterns. In Yukio Tono, Yuji Kawaguchi & Makoto Minegishi (eds.), Developmental and crosslinguistic perspectives in learner corpus research, 27–46. Amsterdam: Benjamins.

Tracy-Ventura, Nicole & Florence Myles. 2015. The importance of task variability in the design of learner corpora for SLA research. International Journal of Learner Corpus Research 1(1). 58–95.



Werner, Valentin. Forthcoming. The present perfect as a core feature of World Englishes. In Anna Mauranen, Juhani Klemola, Svetlana Vetchinnikova & Markku Filppula (eds.), Changing English: Global and local perspectives. Berlin: Mouton de Gruyter.



van der Wurff, Wim. 1999. Some observations on the present perfect puzzle in pedagogical grammars of English. In Guy A. J. Tops, Betty Devriendt & Steven Geukens (eds.), Thinking English grammar, 471–484. Leuven: Peeters.

Wulff, Stefanie, Nick C. Ellis, Ute Römer, Kathleen Bardovi-Harlig & Chelsea J. Leblanc. 2009. The acquisition of tense-aspect: Converging evidence from corpora and telicity ratings. The Modern Language Journal 95(3). 354–369.

Wulff, Stefanie & Ute Römer. 2009. Becoming a proficient academic writer: Shifting lexical preferences in the use of the progressive. Corpora 4(2). 115–133.


Yu, Hao & Chu Ren. 2013. The positive role of L1 in the acquisition of a second language. Studies in Literature and Language 7(2). 45–48.

Zhang, Yanyin & Ima Widyastuti. 2010. Acquisition of L2 English morphology: A family case study. Australian Review of Applied Linguistics 33(3). 1–17.

Appendix A: Corpus queries for CLAWS7 Tagset(Format: One line per word form, with word form, lemma, and parts of speech tag separated by tabs)


Time-reference formsPresent perfect \w+_(VH0|VHZ)( \w+_(XX|R(\w|\d)*?|MD|UH|FU)){0,4}( \w+_(AT(\

w|\d)*?|APPGE|UH|FU)){0,2}( \w+_(JJ(\w|\d)*?|N(\w|\d)*?|UH|FU)){0,2}( \w+_(PPH1|PP[IH]S(\w|\d)*?|PPY|NP(\w|\d)*?|D(\w|\d)*?|NN(\w|\d)*?|UH|FU)){0,2}( \w+_(XX|R(\w|\d)*?|MD|UH|FU)){0,4} \w+_(VBN|VHN|VVN|VVNK)The first part of this query (“VH0|VHZ”) targets a form of have, and the last part a participle (“VBN|VHN|VVN|VVNK”). The intervening search terms allow for a limited number of adverbials and noun phrases between the auxiliary and the participle. Note that this regular expression slightly improves on the ones used in Hundt and Smith (2009), Yao and Collins (2012), and Fuchs (this volume), in that hesitations (tag “UH”) and unclas-sified words (tag “FU”) are included. Also, in order to search for wildcards, “(\w|\d)*” was used instead of “.*”, which excludes some false positives.

Simple past v.d.?Past perfect \w+_VHD( \w+_(XX|R(\w|\d)*?|MD|UH|FU)){0,4}( \w+_(AT(\w|\d)*?|AP-

PGE|UH|FU)){0,2}( \w+_(JJ(\w|\d)*?|N(\w|\d)*?|UH|FU)){0,2}( \w+_(PPH1|P-P[IH]S(\w|\d)*?|PPY|NP(\w|\d)*?|D(\w|\d)*?|NN(\w|\d)*?|UH|FU)){0,2}( \w+_(XX|R(\w|\d)*?|MD|UH|FU)){0,4} \w+_(VBN|VHN|VVN|VVNK) This query is identical to the one for the PP except that the first part targets had.


Appendix B: Frequency of the PP and SP, and per-centage of the PP compared to all references to the past in the native and learner corporaCorpus Subcorpus PP tokens SP tokens % PP

ICCI-AUT Learners school written (year 5–9) 69 3,881 1.7ICCI-AUT Learners school written (year 10) 28 441 6.0ICCI-AUT Learners school written (year 11) 5 37 11.9LINDSEI-GE Learners uni spoken (early starters) 18 295 5.8LINDSEI-GE Learners uni spoken (late starters) 161 3,231 4.7GICLE Learners uni written (early starters) 23 96 19.3GICLE Learners uni written (late starters) 370 3,769 8.9LOCNESS-BE (A-level exams)

Native GB school written 410 604 40.4

LOCNEC Native GB uni spoken 415 5,256 7.3LOCNESS-BE (university)

Native GB uni written 222 1,334 14.3

LOCNESS-US Native US uni written 57 270 17.4

Appendix C: Frequency of co-occurrence of various time-reference forms (“tenses”) with selected temporal adverbialsCorpus Temporal adverbial PP SP Present Past perfect Future

ICCI-AUT yet 2 0 0 2 0ICCI-AUT just 1 4 2 2 0ICCI-AUT since 3 7 8 2 0ICCI-AUT already 14 14 14 5 0ICCI-AUT always 6 13 59 2 6ICCI-AUT recently 0 0 0 0 0ICCI-AUT never 8 6 20 4 11ICCI-AUT ever 13 6 4 1 2GICLE yet 16 7 14 0 0GICLE just 17 10 1 6 0GICLE since 36 9 16 3 0GICLE already 39 22 36 10 1GICLE always 27 30 138 6 14GICLE recently 6 14 1 2 0GICLE never 36 25 61 7 25


Corpus Temporal adverbial PP SP Present Past perfect Future

GICLE ever 24 11 14 4 3LINDSEI-GE yet 4 1 7 0 0LINDSEI-GE just 4 8 0 1 0LINDSEI-GE since 3 5 2 0 0LINDSEI-GE already 2 15 6 2 0LINDSEI-GE always 6 39 55 1 0LINDSEI-GE recently 0 0 1 0 0LINDSEI-GE never 23 29 5 0 1LINDSEI-GE ever 7 6 1 1 0LOCNEC yet 12 0 7 1 0LOCNEC just 10 6 1 3 0LOCNEC since 10 4 3 2 0LOCNEC already 9 7 4 3 0LOCNEC always 36 18 33 12 1LOCNEC recently 3 1 1 0 0LOCNEC never 31 44 21 21 4LOCNEC ever 14 3 4 4 1LOCNESS-BE (A-level exams) yet 3 0 5 0 0LOCNESS-BE (A-level exams) just 2 0 0 0 0LOCNESS-BE (A-level exams) since 17 0 1 0 0LOCNESS-BE (A-level exams) already 13 0 10 0 2LOCNESS-BE (A-level exams) always 10 1 12 0 6LOCNESS-BE (A-level exams) recently 10 4 0 0 0LOCNESS-BE (A-level exams) never 7 1 5 1 5LOCNESS-BE (A-level exams) ever 1 0 1 0 0LOCNESS-BE (university) yet 5 1 3 1 0LOCNESS-BE (university) just 1 0 0 1 0LOCNESS-BE (university) since 12 1 3 1 0LOCNESS-BE (university) already 16 5 17 1 0LOCNESS-BE (university) always 16 4 24 4 5LOCNESS-BE (university) recently 3 1 0 0 0LOCNESS-BE (university) never 5 5 8 0 10LOCNESS-BE (university) ever 1 4 3 0 1


Appendix D: Percentage of the PP compared to all references to the past in the native and learner corpora in co-occurrences with selected temporal adverbialsCorpus yet just since already always recently never ever

ICCI-AUT 100.0 20.0 30.0 50.0 31.6 – 57.1 68.4GICLE 69.6 63.0 80.0 63.9 47.4 30.0 59.0 68.6LINDSEI-GE 80.0 33.3 37.5 11.8 13.3 – 44.2 53.8LOCNEC 100.0 62.5 71.4 56.3 66.7 75.0 41.3 82.4LOCNESS-BE (A-level exams) 100.0 100.0 100.0 100.0 90.9 71.4 87.5 100.0LOCNESS-BE (university) 83.3 100.0 92.3 76.2 80.0 75.0 50.0 20.0

Appendix E: Co-occurrences of the PP with selected temporal adverbials and percentageCorpus Temporal adverbials

(tokens)PP (tokens) % PP with temporal adver-

bials

ICCI-AUT 47 102 46.1GICLE 201 659 30.5LINDSEI-GE 49 179 27.4LOCNEC 125 415 30.1LOCNESS-BE (A-level exams) 63 410 15.4LOCNESS-BE (university) 59 222 26.6LOCNESS-US (university) 5 57 8.8

Björn Rothstein11 Afterthought: Some brief remarks on

autonomous and speaker-centered linguistic approaches to the present perfect

Abstract: This paper discusses autonomous and speaker-centered approaches to the present perfect. The autonomous approach is a linguistic research tradition which focuses on the structural regularities of language(s), which are analysed without reference to their users, the speaker(s) of the society or the societies in which these languages are spoken. The speaker-centered approach focuses on the contexts in which the perfect occurs, to frequency and to various sociolinguis-tic factors governing its use. As for tense-aspect systems and the present perfect in particular, there have only been a few studies that have tried to combine a speaker-free and speaker-centered approach to the present perfect. The present paper discusses the two approaches and the reasons why there are hardly any combined approaches. The comparison is not in terms of better or worse. In fact, I will argue that both of them are necessary to get the full picture.

The type of analysis of the English present perfect that I have grown up with is in the spirit of Reichenbach (1947), Klein (1994) and Iatridou, Anagnostopoulou, and Pancheva (2001). These are time-relational approaches that distinguish at least between a speech time (time at which the utterance is made), an event time (time at which the eventuality obtains) and some sort of topic/reference time relative to which the event time is located. The first systematic time-relational approach goes back to Reichenbach (1947), who assumes the following meaning for the present perfect: according to Reichenbach, the present perfect localizes the event time (E) before the reference time (R) and the reference time as overlap-ping with the speech time (S), as shown in Figure 1.

E S,R

Figure 1: Reichenbach’s (1947: 290) analysis of the present perfect.

Björn Rothstein, University of Bochum

340 Björn Rothstein

Over the years, Reichenbach’s analysis has been modified and extended. However, the basic configuration of the speech time (S), the event time (E) and the reference time (R) has remained the same.

Another prominent time-relational approach comes from McCoard (1978). The ExtendedNow-interval (XN) is a time span whose right boundary (RB) ends in the case of the present perfect at the moment of speech (see Figure 2). The position of its left boundary (LB) is not specified or can be given by adverbials like since. Within XN, the event time is expressed by the present perfect.

XN

E S,R

Figure 2: The ExtendedNow analysis of the present perfect (McCoard 1978: 123).

The reader who has gone through the preceding chapters will have noted that at several points we argued the merits of an analysis of the perfect as the marker of prior events which are nevertheless included within the overall period of the present, the ExtendedNow, while the preterite marks events assigned to a past which is concluded and separate from the extended present (McCoard 1978: 123).

Advocates of this approach explain the ExtendedNow-interval by analysing adverbial modification of the present perfect: according to this view, since-ad-verbials require this additional interval, because otherwise modification of the present perfect would not be possible (McCoard 1978: 123).

Besides the time-relational approaches to the present perfect, there are also analyses in the terms of result states or current relevance. The current relevance analysis (see Figure 3) claims that the present perfect expresses that a past event (E) is relevant at the moment of speech (S).

current relevance

E S

Figure 3: The current relevance analysis of the present perfect (mostly found in grammar books; see, for instance, Huddleston and Pullum 2002: 143; see further Bowie and Wallis, this volume).

Afterthought 341

Within the result state analysis (see Figure 4), the present perfect result of a past event (E) obtains at the moment of speech (S).

result

E S

Figure 4: The result state analysis of the present perfect (Kamp and Reyle 1993: 559, 579–591).

All four approaches still have supporters and so far it has not been decided which approach is the most adequate. Following Newmeyer (1986), I will refer to this kind of analysis as the “autonomous” linguistic approach to language and, in particular, to the present perfect. This is a linguistic research tradition which focuses on the structural regularities of language(s), which are analysed without reference to their users, the speaker(s) of the society or the societies in which these languages are spoken. Therefore, the term speaker-free linguistics is also used instead of autonomous linguistics. My own work is much in this spirit. In Rothstein (2007, 2008), I argue for an integrative analysis combining the Reichen-bachian approach with the ExtendedNow analysis. This work focuses only on American and British English in comparison to German and Swedish.

However, there is also a wealth of investigations on the present perfect of a different kind. These include approaches with mostly text-linguistic and socio-linguistic analyses. A great deal of attention is paid to the contexts in which the perfect occurs, to frequency and to various sociolinguistic factors governing its use. Such approaches dominated the workshop “Perfect and perfectivity re-as-sessed through corpus studies” at the ICAME 35 conference in Nottingham (April 2014) organized by Elena Seoane, Cristina Suárez-Gómez, and Valentin Werner. This is what Lass (1980: 122) calls “speaker-centered linguistics”. With the excep-tion of my own paper, no talk referred to research normally found in the auton-omous linguistic approach and – I have to admit – I did not refer to the speak-er-centered approach myself. Therefore, I concluded that these talks and my talk seemed to represent quite parallel worlds without any link to each other. As far as I can see, all the papers within this volume are concerned with the speak-er-centered linguistic approach. The differences between speaker-centered and autonomous linguistics have been a concern of research from both sides (see the contributions in Åfarli and Mæhlum 2013). As for tense-aspect systems and the present perfect in particular, there have only been a few studies trying to combine a speaker-free and speaker-centered approach to the present perfect.


The existence of these two approaches leads to the following general ques-tions: (i) Why are there no links between the two approaches? (ii) How can this be? (iii) Having two approaches, isn’t one of them simply superfluous? (iv) Can they learn from each other, with the aim of arriving at a comprehensive

analysis accounting for all the data?

The following remarks are about the two approaches. In the end, I will argue that both of them are necessary to get the full picture.

The general idea of autonomous linguistics was, in its beginnings, closely related to Chomsky’s idealized speaker-hearer in a completely homogeneous speech community. In autonomous linguistics, variation in grammar started to become a more sophisticated research interest with the development of generative frameworks such as Government and Binding and the turn to syntactical micro-variation (cf. Brandner, Salzmann, and Schaden 2016). Contemporary genera-tive approaches investigate language variation and multilingualism (see Cornips 2015), a domain that is well suited for cooperation with sociolinguistics. However, an architecture integrating the structural linguistic rules from autonomous lin-guistics and speaker-oriented factors from the speaker-centered approach is still missing.

Within the autonomous linguistic approach, we find various analyses of the present perfect (also see above), which can be grouped as follows: according to the current relevance approach, the present perfect expresses that a past event is relevant at the moment of speech. Other approaches, however, claim that it is an aspect or a combined aspectual-temporal construction (see Rothstein 2008: 29–34 for discussion). While many scholars still follow the classical temporal Reichen-bachian tradition (e.g. Klein 1994), others propose an ExtendedNow approach (Portner 2003; Pancheva and von Stechow 2004) or a result state analysis (Kamp and Reyle 1993). The Reichenbachian school analyses the present perfect as an event time which is before the reference time, with the reference time being at the speech time (Reichenbach 1947; Klein 1994; Musan 2002). In the ExtendedNow approach the present perfect introduces a time interval, an extended now start-ing somewhere in the past and ending at the moment of speech (McCoard 1978; von Stechow 1999; Pancheva and von Stechow 2004; with modifications, Roth-stein 2008). The result state analysis assumes that the present perfect introduces a result state emerging immediately from the eventuality in the present perfect (Kamp and Reyle 1993).

There exist various formal approaches to the present perfect ranging from syntax (using a version of Government and Binding, such as Hornstein 1990, or

Afterthought 343

Distributed Morphology, such as Embick 2004) and semantics (using Discourse Representation Theory; see, e.g., Kamp and Reyle 1993; Rothstein 2008; Transpar-ent Logical Form, as in von Stechow 1999; or Event Semantics, as in Zybatow 2002) to pragmatics (Portner 2003). Thus, there is no single autonomous approach to the present perfect; there are many different frameworks within which the present perfect has been analysed so far (see also Werner, Seoane, and Suárez-Gómez, this volume).

In general, the autonomous linguistic approach focuses on the semantics and syntax of the present perfect. Mostly, it is interested in restrictions with temporal adverbials, and much time has been spent on explaining the “present perfect puzzle” (although the present perfect denotes some kind of anteriority, it does not combine with adverbials such as yesterday; see, e.g., Klein 1992; Pancheva and von Stechow 2004). Other research aspects concern its semantic composition (e.g. where anteriority is located in the present perfect; see, for instance, Musan 2002) and the semantics-pragmatics interface (is there one single meaning of the present perfect covering its different readings? see, e.g., Mittwoch 1988; Michae-lis 1994; Iatridou, Anagnostopoulou, and Pancheva 2001). Sociolinguistic factors such as usage and frequency play no role. This type of research mainly focuses on national standard varieties, for instance, American or British English, Stan-dard German and Standard French, although more recent work has also started to analyse variationist data (e.g. Brandner, Salzmann, and Schaden 2016).

Within the speaker-centered linguistic approach, we find analyses based on sociolinguistic and text-linguistic factors. Very often, these factors are combined. The present volume contains a number of such combinations. For instance, Seoane (this volume) provides an integrative sociolinguistic and corpus-lin-guistic view for Jamaican English: with the help of corpus-based methods, she approaches the sociolinguistic distribution of surface variants in the perfect space in this variety. Richard and Rodríguez Louro (this volume) present an analysis of the Australian narrative present perfect integrating aspects of its sociolinguis-tic and text-linguistic patterns. While I mention these two studies specifically for purposes of illustration, this applies similarly to all contributions to the present volume. All authors explore variation within the realm of the present perfect with the help of corpus data and consider the influence of various language-internal and language-external variables or factors.

From a broader perspective, there is a longstanding tradition of sociolin-guistic work on the present perfect. In German linguistics, for instance, many studies have investigated the loss of the German preterite and the extension of the present perfect. It has been claimed that this is due to language contact (Ternes 1988), phonological changes (Reis 1894), and speaker-based orientation to the present (Frei 1975). Differences between the varieties of English have recently


Table 1: Bibliographies from Meyer-Viol and Jones (2011) and Hundt and Smith (2009)

Autonomous linguistic approach Speaker-centered linguistic approach

Meyer-Viol, Wilfried P. M. & Howard S. Jones. 2011. Reference time and the English past tenses. Linguistics and Philosophy 34. 223–256.


Binnick, R. (1991). Time and the verb: A guide to tense and aspect. Oxford: Oxford University Press.Bryan, W. (1936). The preterite and the perfect tense in pres-ent-day English. Journal of English and Germanic Philology, 35, 363–338.Chomsky, N. (1970). Deep structure, surface structure, and semantic interpretation. In R. Jakobson & S. Kawamoto (Eds.), Studies in general and oriental linguistics (pp. 52–91). Tokyo: TEC Corporation.Comrie, B. (1976). Aspect. Cambridge: Cambridge University Press.Curme, G. O. (1935). A grammar of the English language, volume 2: Parts of Speech and Accidence. New York: D.C. Heath.Declerck, R. (1991). Tense in English: Its structure and use in discourse. London: Routledge.Dietrich, W. (1955). Erweiterte Form, Präteritum und Perfek-tum im Englischen. Eine Aspekt- und Tempusstudie. Munich: Hueber.Dowty, D. (1977). Toward a semantic analysis of verb aspect and the English continuative progressive. Linguistics and Philosophy, 1, 45–77.Dowty, D. (1979). Word meaning and Montague grammar: The semantics of verbs and times in generative semantics and in Montague’s PTQ. London: Reidel.Heim, I. (1982). The semantics of definite and indefinite noun phrases in English. PhD Thesis, University of Massachusetts, Amherst.Higginbotham, J. (2009). Tense, aspect and indexicality. Oxford: Oxford University Press.Iatridou, S., Anagnostopoulou, E., & Pancheva, R. (2001). Observations about the form and meaning of the perfect. In M. Kenstowicz (Ed.), Ken Hale: A life in language (pp. 189–238). Cambridge: MIT.Inoue, K. (1979). An analysis of the English present perfect. Linguistics, an Interdisciplinary Journal of the Language Sciences, 17, 561–589.Kamp, H., & Reyle, U. (1993). From discourse to logic. Dordrecht: Kluwer.Kiparsky, P. (2002). Event structure and the perfect. In D. I. Beaver, L. D. C. Martinez, B. Z. Clark, & S. Kaufmann (Eds.), The construction of meaning. Stanford: CSLI Publications.Klein, W. (1992). The present perfect puzzle. Language, 68 (3), 525–552.Klein, W. (2000). An analysis of the German Perfekt. Lan-guage, 76 (2), 358–382.

Bauer, Laurie. 1994. English in New Zealand. In Robert.W. Burchfield (ed.). The Cambridge history of the English language. Volume V: English in Britain and overseas. Origins and developments, 282–429. Cambridge: Cambridge University Press.Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad and Edward Finegan. 1999. Longman grammar of spoken and written English. Harlow: Longman.Burchfield, Robert W. 1996. The new Fowler’s modern English usage. Third edition. Oxford: Clarendon.Christ, Oliver. 1994. A modular and flexible architecture for an integrated corpus query system. Proceedings of COMPLEX’94: Third Conference on Computational Lexicogra-phy and Text Research (Budapest, July 7–10, 1994), 23–32. Budapest.Elsness, Johan. 1997. The perfect and the preterite in con-temporary and earlier English (Topics in English Linguistics 21). Berlin and New York: Mouton de Gruyter.Elsness, Johan. 2009. The present perfect and the preterite. In G. Rohdenburg and J. Schlüter (eds.). One language, two grammars? Differences between British and American English, 228–245. Cambridge: Cambridge University Press.Engel, Dulcie M. 1998. A perfect piece? The present perfect and passé compose in journalistic texts. Belgian Journal of Linguistics 12: 129–147.Engel, Dulcie M. and Marie-Eve Ritz. 2000. The use of the present perfect in Australian English. Australian Journal of Linguistics 20 (2): 119–140.Foster, Brian. 1968. The changing English language. London: Macmillan.Greenbaum, Sidney and Janet Whitcut. 1988. Longman guide to English usage. Harlow: Longman.Hughes, Arthur, Peter Trudgill and Dominic Watt. 2005. English accents and dialects. An introduction to social and regional varieties of English in the British Isles. London: Hodder Arnold.Kjellmer, Göran. 2003. On nonoccurring perfective have in Modern English. Studia Neophilologica 75: 11–20.Lass, Roger. 1987. The shape of English. Structure and history. London: Dent.Mair, Christian, Marianne Hundt, Geoffrey N. Leech and Nicholas Smith. 2002. Short term diachronic shifts in part-of-speech frequencies: A comparison of the tagged LOB and F-LOB corpora. International Journal of Corpus Linguistics 7 (2): 245–264.

Afterthought 345

Autonomous linguistic approach Speaker-centered linguistic approach

McCawley, J. D. (1971). Tense and time reference in English. In C. J. Fillmore & D. Terence Langendoen (Eds.), Studies in linguistic semantics (pp. 96–113). New York: Holt, Rinehart and Winston.McCoard, R. W. (1978). The English perfect: Tense-choice and pragmatic inferences. Amsterdam: North-Holland.Michaelis, L. (1994). The ambiguity of the English present perfect. Journal of Linguistics, 30, 111–157.Mittwoch, A. (1988). Aspects of English aspect: On the interaction of perfect, progressive and durational phrases. Linguistics and Philosophy, 11, 203–254.Moens, M., & Steedman. M. (1988). Temporal ontology and temporal reference. Computational Linguistics, 14 (2), 15–28.Musan, R. (1997). Tense, predicates, and lifetime effects. Natural Language Semantics, 5, 271–301.Musan, R. (2002). The German perfect. Its semantic composi-tion and its interaction with temporal adverbials. Dordrecht: Kluwer.Pancheva, R., & von Stechow, A. (2004). On the present perfect puzzle. In K. Moulton & M. Wolf (Eds.), Proceeding of the North East Linguistics Society (=NELS) 34. Amherst: GLSA.Partee, B. (1973). Some structural analogies between tenses and pronouns in English. The Journal of Philosophy, 70 (18), 601–609.Pickbourn, J. (1789). A dissertation on the English verb (Facsimile edition, 1968). Menston: Scolar Press.Portner, P. (2003). The (temporal) semantics and (modal) pragmatics of the perfect. Linguistics and Philosophy, 26, 459–510.Portner, P. (2011). Perfect and progressive. In C. Maienborn, K. von Heusinger, & P. Portner (Eds.), Semantics: An inter-national handbook of natural language meaning. Berlin: Mouton de Gruyter. Available at http:// semanticsarchive.net/Archive/jlmMTE0M/Portner-perfect-progressive-HSK-draft.pdf. Accessed 23 Nov 2011.Reichenbach, H. (1947). Elements of symbolic logic. London: Macmillan. Rothmayr, A. (2009). The structure of stative verbs. Amster-dam: Benjamins.Rothstein, B. (2008). The perfect time span: On the present perfect in German, Swedish, and English. Amsterdam: John Benjamins.Smith, C. (1997). The parameter of aspect. Dordrecht: Kluwer.Schaden, G. (2009). Present perfects compete. Linguistics and Philosophy, 32 (2), 115–141.Vendler, Z. (1967). Linguistics in philosophy. New York: Cornell University Press.von Stechow, A. (1999). Eine erweiterte ExtendedNow Theorie fur Perfekt und Futur. Zeitschrift für Literatur und Linguistik, 113, 86–118.

McCawley, John. 1971. Tense and time reference in English. In C.J. Fillmore and D.T. Langendoen (eds.). Studies in linguistic semantics, 96–113. New York: Holt, Rinehart and Winston.Meyer, Matthias. 1995. Past tense and present perfect in the Brown and the LOB corpora. In W. Riehle and H. Keiper (eds.). Anglistentag 1994 Graz. Proceedings, 201–228. Tübingen: Niemeyer.Miller, Jim. 2004. Perfect and resultative constructions in spoken and non-standard English. In O. Fischer, M. Norde and H. Perridon (eds.). Up and down the cline – the nature of grammaticalization, 229–246. Amsterdam: Benjamins.Powell, Enoch. 1990. Further thoughts: Grammar and syntax. In C. Ricks and L. Michaels (eds.). The state of the language, 483–485. London: Faber and Faber.Rastall, Paul. 1999. Observations on the present perfect in English. World Englishes 18 (1): 79–93.Sampson, Geoffrey. 2002. Regional variation in the English verb qualifier system. English Language and Linguistics 6 (1): 17–30.Schlüter, Norbert. 2006. How reliable are the results? Comparing Corpus -based studies of the present perfect. Zeitschrift für Anglistik und Amerikanistik 54 (2): 135–148.Strevens, Peter. 1972. British and American English. London: Macmillan.Tottie, Gunnel. 2002. Non-categorical differences between American and British English: Some corpus evidence. In M. Modiano (ed.). Proceedings from the Conference on Mid-At-lantic English, 37–58. Gävle: Gävle University College Press.Trudgill, Peter and Jean Hannah. 2002. International English. A guide to varieties of Standard English. London: Arnold. Trudgill, Peter. 1984. Language in the British Isles. Cam-bridge: Cambridge University Press.Vanneck, Gerard. 1958. The colloquial preterite in modern American English. Word 14: 237–242.Webster’s Dictionary of English Usage. 1989. Springfield, Mass.: Merriam-Webster.

Table 1: (continued)


been a major research interest in English linguistics (see, e.g., Davydova 2011 and Werner 2014 for an overview), including, of course, speaker-related use of the present perfect in these variants (see Hundt and Smith 2009).

Interestingly, in speaker-centered research on the present perfect, formal syntax and/or semantics such as the work by Pancheva and von Stechow (2004), Rothstein (2008), Schaden (2009) and Meyer-Viol and Jones (2011) is mostly neglected. This kind of “ignorance” is also typical of the autonomous linguistic approach, where speaker-centered linguistic work is rarely referred to. In general, almost no references to typical autonomous linguistic papers on the present perfect occur in speaker-centered linguistic studies and vice versa. As a case in point, a contrastive view of the bibliographies from two recent studies, Hundt and Smith (2009) and Meyer-Viol and Jones (2011), illustrates this nicely (see Table 1). We see that the quoted literature is completely different (with the exception of one publication, McCawley 1971, highlighted in Table 1).

In the present volume, there are some studies that have made it into both types of bibliographies. This is work by Bernard Comrie (1976, 1985) and Östen Dahl (1985). Both Comrie’s and Dahl’s work is typological and accounts for the data in a non-formalized syntactic or semantic manner. This might explain why

TP

D λxTWolfgangWolfgang λr.PERFP T

präs

QAλrλP t.IN(r)(t) & P (t) PERFP

Ø

PERFλt.PROGP XNP (r)

ist (r)PROG is

λe.VP λP e.t e & P(e)gewesen (t)

D beensein Papier λy . Vhis paper

am Vparticle schreiben (e) (y) (x)

write

Figure 5: ExtendedNow analysis of the German present perfect by von Stechow (1999: 105).

Afterthought 347

the studies by Reichenbach (1947), Klein (1994), Smith (1997), Pancheva and von Stechow (2004) and many others are not referred to.

Perhaps the formal language of many autonomous linguistic approaches turns off supporters of the speaker-centered linguistic analysis. Consider, for instance, the meaning of the German present perfect by von Stechow (1999: 105), as illustrated in Figure 5. Readers not familiar with the Transparent Logical Form may understandably not be able to grasp this meaning. Roughly speaking, von Stechow’s analysis of the present perfect locates its anterior meaning in the aux-iliary. The analysis is compositional and much in the spirit of the ExtendedNow approach. Von Stechow (1999) is interested in the semantics and syntax of the present perfect.

There exist various simplified semantic frameworks for tense that are more approachable for readers who are unfamiliar with formal semantics: these are the analyses using capital letters and simple semantic relations such as “prior to” and “after” (Reichenbach 1947; Klein 1994; Smith 1997). For a sentence such as Peter had left the party, Reichenbach proposes a time relational analysis con-taining three points in time: E < R < S. (E) is the time at which the event took place, (S) is the speech time and the reference time (R) is the time relative to which (E) is located (see above). In the pluperfect, the event time is before the reference time that is before the speech time. Thus, no reading skills in formal seman-tics and/or syntax are required. But very often, the speaker-centered linguistic approach does not consider these. At the same time, I must admit that I cannot come up with a convincing reason why supporters of the autonomous linguistic approach tend not to work with literature from the speaker-centered approach when dealing with the present perfect.

There are different ways to explain the missing link between the autonomous and the speaker-centered approach in research on the present perfect. As long as there is no need for a combined approach integrating both autonomous and speaker-centered studies, we could be satisfied by permitting two independent universes of research on the present perfect.

However, both approaches share a basic research question: all the analyses want to gain a deeper understanding of the present perfect. So why not work together? When looking beyond the present perfect, one discovers that, gen-erally speaking, linguistic theories have tended to avoid the interface between speaker-free linguistics and speaker-centered linguistics. There have been many attempts to analyse the interface between semantics and pragmatics (see the contributions in von Heusinger and Turner 2006 and, with special focus on the present perfect, Portner 2003), but there are almost no investigations on the semantics-sociolinguistics interface, at least not in the sense that a fully satis-fying architecture of language or grammar incorporating morphology, syntax,


semantics, pragmatics and sociolinguistics has been proposed (for recent discus-sion, see the contributions in Åfarli and Mæhlum 2013; Adli, García García, and Kaufmann 2015).

I believe that these parallel worlds are the result of the current lack of a lin-guistic model for integrating both autonomous and speaker-centered linguistic perspectives. Indeed, a number of recent studies on the present perfect argue for an approach integrating elements of the speaker-free and the speaker-centered approaches to the (German) present perfect. Especially sociolinguistic factors impact on the grammar of the present perfect. Jäger (2007) argues that speakers choose rather freely between the German preterite and the present perfect depend-ing on the choice their hearers have made themselves. He claims that the differ-ence between the German present perfect and the preterite is a social marker to express communicative adaptation to the hearer. However, no attempt is made to consider its structural restrictions, for example substitution between the German preterite and the present perfect is not possible in free indirect speech (Rothstein 2008: 26–27). Stark (2015) considers both sociolinguistic and structural restric-tions on the present perfect, but neglects to propose a linguistic model for these: the present perfect is more likely to be used in spoken German, while the past tense is favoured in written German. Schaden (2009) and Brandner, Salzmann, and Schaden (2016) propose microvariational analyses of the German present perfect, but topics such as dialect use are not integrated into their account.

The more traditional picture of integrating the speaker-centered and the speaker-free approach relies on the treatment of data: in this spirit, the speak-er-centered approach would provide both data and insights into the usage of the present perfect that could serve as verification or falsification of the speaker-free approach. However, this is not the integration I have in mind. As shown with the German present perfect, sociolinguistic factors impact on its grammar. For a better understanding of the present perfect, both in German and English as well as in other languages, I therefore believe that we should start to think about such an architecture. What needs to be known is whether sociolinguistic factors are somehow restricted by factors from phonology, semantics, syntax and mor-phology. According to the assumptions about the architecture, one could assume that a grammar-based meaning of the present perfect restricts the sociolinguistic choices a speaker has by using the present perfect. However, it could also be pos-sible that there are different sociolinguistic uses of the present perfect that restrict its grammar: for each use, a single “grammatical” meaning could be assumed.

Keeping the aforementioned issues in mind, I will leave for further research how a suitable architecture of language can model the relation between sociolin-guistic and autonomous grammatical factors that interact in meaning and use of the present perfect.

Afterthought 349

ReferencesÅfarli, Tor A. & Brit Mæhlum (eds.). 2013. The sociolinguistics of grammar. Amsterdam:

Benjamins. Brandner, Ellen, Martin Salzmann & Gerhard Schaden. 2016. Zur Syntax und Semantik des

doppelten Perfekts aus alemannischer Sicht [The syntax and semantics of double perfect constructions from a Germanic point of view]. In Alexandra Lenz & Franz Patocka (eds.), Syntaktische Variation: Areallinguistische Perspektiven [Syntactic variation: Areal- linguistic perspectives], 13–46. Vienna: Vienna University Press.

Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press.Comrie, Bernard. 1985. Tense. Cambridge: Cambridge University Press.Cornips, Leonie. 2015. The no man’s land between syntax and variationist sociolinguistics:

Idiolectal variability. In Aria Adli, Marco García García & Göz Kaufmann (eds.), Variation in language: System- and usage-based approaches, 147–172. Berlin: Mouton de Gruyter.

Dahl, Östen. 1985. Tense and aspect systems. Oxford: Blackwell.Davydova, Julia. 2011. The present perfect in non-native Englishes: A corpus-based study of

variation. Berlin: Mouton de Gruyter. Embick David. 2004. On the structure of resultative participles in English. Linguistic Inquiry 35.

355–392. Frei, Gertrud. 1975. Walserdeutsch in Saley [Walser-German in Saley]. Bern: Haupt.Hornstein, Norbert. 1990. As time goes by: Tense and universal grammar. Cambridge: MIT

Press.Hundt, Marianne & Nicolas Smith. 2009. The present perfect in British and American English:

Has there been any change, recently? ICAME Journal 33. 45–63.Iatridou, Sabine, Elena Anagnostopoulou & Roumyana Pancheva. 2001. Observations about the

form and meaning of the perfect. In Mark Kenstowicz (ed.), Ken Hale: A life in language, 189–238. Cambridge: MIT Press.

Jäger, Andreas. 2007. Der Gebrauch des Perfekt-Präteritum-Paradigmas in der gesprochenen deutschen Sprache [The usage of the present perfect/past tense paradigm in spoken German]. Baltmannsweiler: Schneider.

Kamp, Hans & Uwe Reyle. 1993. From discourse to logic. Dordrecht: Kluwer.Klein, Wolfgang. 1992. The present perfect puzzle. Language 68. 525–552.Klein, Wolfgang. 1994. Time in language. London: Routledge.Lass, Roger. 1980. On explaining language change. Cambridge: Cambridge University Press.McCawley, James D. 1971. Tense and time reference in English. In Charles J. Fillmore & Terence

Langendoen (eds.), Studies in linguistic semantics, 96–113. New York: Holt, Rinehart and Winston.

McCoard, Ronald W. 1978. The English perfect: Tense-choice and pragmatic inferences. Amsterdam: North-Holland.

Meyer-Viol, Wilfried P. M., & Howard S. Jones. 2011. Reference time and the English past tenses. Linguistics and Philosophy 34. 223–256.

Michaelis, Laura 1994. The ambiguity of the English present perfect. Journal of Linguistics 30. 111–157.

Mittwoch, Anita. 1988. Aspects of English aspect: On the interaction of perfect, progressive and durational phrases. Linguistics and Philosophy 11. 203–254.


Musan, Renate. 2002. The German perfect: Its semantic composition and its interaction with temporal adverbials. Dordrecht: Kluwer.

Newmeyer, Frederick J. 1986. The politics of linguistics. Chicago: University of Chicago Press.Pancheva, Roumyana & Arnim von Stechow. 2004. On the present perfect puzzle. In Keir

Moulton & Matthew Wolf (eds.), Proceedings of the North East Linguistics Society 34, 469–483. Amherst: University of Massachusetts.

Portner, Paul. 2003. The (temporal) and (modal) pragmatics of the perfect. Linguistics and Philosophy 26. 459–510.

Reichenbach, Hans. 1947. Elements of symbolic logic. London: Macmillan. Reis, Hans. 1894. Das Präteritum in den süddeutschen Mundarten [The past tense in the

dialects of Southern Germany]. Beiträge zur Geschichte der deutschen Sprache und Literatur 19. 334–337.

Rothstein, Björn. 2007. Tempus [Tense]. Heidelberg: Winter. Rothstein, Björn. 2008. The perfect time span: On the present perfect in German, Swedish and

English. Amsterdam: Benjamins. Schaden, Gerhard. 2009. Present perfects compete. Linguistics and Philosophy 32. 115–141.Smith, Carlotta. 1997. The parameter of aspect. Dordrecht: Kluwer.Stark, Linda. 2015. Vorlesesituationen und literale Lernmöglichkeiten am Beispiel des

deutschen Präteritum [Reading aloud and learning of literacy language: The case of the German past tense]. Bochum: University of Bochum dissertation.

Ternes, Elmar. 1988. Zur Typologie der Vergangenheitstempora in den Sprachen Europas [On the typology of past tenses in the European languages]. Zeitschrift für Dialektologie und Linguistik 55. 332–342.

von Heusinger, Klaus & Ken Turner (eds.). 2006. Where semantics meets pragmatics. Amsterdam: Elsevier.

von Stechow, Arnim. 1999. Eine erweiterte ExtendedNow-Theorie für Perfekt und Futur [A modified ExtendedNow theory for the perfect and the future]. Zeitschrift für Literatur und Linguistik 113. 86–118.


Zybatow, Tatjana. 2002. Das deutsche Perfekt: Ein kompositionaler Versuch [The German perfect: A compositional analysis]. In Reinhard Rapp (eds.), Sprachwissenschaft auf dem Weg in das dritte Jahrtausend: Akten des 34. Linguistischen Kolloquiums in Germersheim 1999 [Linguistics on its way to the third millenium: Proceedings of the 34th linguistic colloquium in Germersheim 1999], 441–452. Frankfurt: Peter Lang.

Subject indexadverbial support see temporal adverbialsAktionsart see lexical aspectAmericanization 6

be like see quotative markingbe-periphrasis see perfect – variant surface

realizations – be-perfect

catenative verb see modalitychange see language contactCricle Model see models of World Englishescolloquialization see formality and

informalitycommon core 170complexification 96–97, 112–114, 179comprehensibility see intelligibilityconstraints

– cognitive constraints see production constraints

– language-internal constraints 52, 109, 130, 136–137, 177, 180–181, 184, 188–189, 205, 224, 226, 231–232, 240, 264–265, 280, 282, 287, 301–303, 343, 348

– lexical constraints see lexical effects – normativity 80, 150, 153, 155, 165, 174, 262,

298, 323 – processing constraints 154–155, 165, 197 – production constraints 153, 197, 206 – sociolinguistic constraints 51, 96, 98,

100–101, 106, 114–115,122, 133–136, 138, 154, 224, 226–229, 236, 241–243, 245–251, 285, 303–305, 312, 327, 348

continuation see persistencecontrastive analysis 300–302, 306Corpus of Global Web-based English

(GloWbE) 35, 227, 229, 231, 244, 266–268, 287–290; see also electronic discourse

Corpus of Historical American English (COHA) 3, 35, 45–48, 88–91

default past tense hypothesis 302, 305, 313, 315, 322, 325–326

Dynamic Model see models of World Englishes

Early Modern English 106, 109, 111, 151, 262electronic discourse 229, 233–234, 237,

239–246, 266–269, 286, 288electronic World Atlas of Varieties of English

(eWAVE) 9, 102–104, 107–108, 110, 114, 121, 171–172, 200, 206–207, 214, 224, 230, 259–260, 263, 271, 287–288, 297

ELF see lingua francaevidentiality see verb phrase

factors see constraintsformality and informality 6–7, 104, 108,

153–156, 160–166, 198, 210–212, 218, 224, 227, 231–233, 239–246, 251, 265–266, 286, 288–289, 305, 323, 327, 329

Gaelic – Irish Gaelic 96–99, 101–102, 105–107, 109,

112, 114, 263 – Scottish Gaelic 96–99, 101, 106, 112

grammaticalization 23, 126, 197, 208, 217, 261

Hebridean English 95, 96, 100–114

innovation 120, 128, 138, 163, 178, 209–210, 228, 284–290

International Corpus of English (ICE) 4, 46, 55–60, 88, 114, 155, 160, 176–177, 196–197, 200–202, 229, 244, 265–267, 290

input 170, 173–174, 178, 188–189, 197, 249–251, 315

– input frequency 315–316, 326–328 – input variety 102, 164, 173, 184, 189, 208,

210, 250intelligibility 250, 306Irish see Irish GaelicIrish English 4, 95, 100–115, 180, 184, 248,

250, 260, 263

Jamaican Creole 195, 197–200, 205–207, 217–218

Jamaican English 195–218, 226, 243, 247, 271, 280, 288

352 Subject index

language acquisition 101, 154, 169–170, 173, 177–179, 184–185, 188–190, 225, 228, 297–305, 322, 326–329

language change see language contactlanguage contact 95–100, 106, 113–114, 154,

173, 195, 197–198, 205–206, 343 – contact-induced change 112, 169–170, 173,

178, 186, 205 – contact variety 100, 105–107, 170,

188–189, 206, 226–228 – high-contact variety 95–97, 113, 172,

246; see also simplification – low-contact variety 96–97; see also

complexification – dialect contact 95, 101, 107, 247–248

language instruction 98, 100, 173–179, 189, 302–308, 312–316, 321–324, 327–328

learner English 7, 106, 109–110, 114, 165, 170–172, 177, 179, 184–185, 188–189, 250–215, 264–265, 297–309, 312–329

lexical aspect 60, 180, 184, 282 – aspect hypothesis 184, 302

lexical effects 61–91, 138, 206, 217–218, 263–264, 270, 275–282, 287, 298

lexical variation see lexical effectslingua franca (English as a lingua franca) 179,

250, 306

meaning of the perfect see perfect – semantics

Middle English 109, 151, 261migration 98, 102, 105, 246–248modality 57, 61–63, 65–66, 75–79, 89, 179,

286, 311 – catenative verbs 52, 57–64, 66–67, 70–71,

80–81, 85–87, 89–90, 311mode see formality and informalitymodels of World Englishes

– Kachru’s Circle Model 100, 155, 164, 223, 226, 228, 232–233, 238–239, 244–245

– Schneider’s Dynamic Model 149–150, 154, 161, 164, 198, 228, 232, 236–237, 245

multidimensional analysis 231multilingualism 169, 174, 185–187, 244, 249,

342multivariate analysis 130, 133, 135, 180–183

naturalistic setting 100, 106, 127–128, 140, 173–174, 185, 188

Old English 23–34, 36–37, 98, 151

past tense see perfect – versus simple past tense

perfect – as possessive 25, 27–29, 31 – infinitival perfect 43–54, 56–90, 179 – narrative perfect 7, 119–140, 212, 300 – non-finite perfect see infinitival perfect – pragmatics 2–4, 7, 55, 126–127, 174–176,

180–184, 188–189, 298, 343, 347–348 – present perfect puzzle 2, 5–6, 8, 300–301,

343 – semantics

– continuative see extended-now – experiential 29, 96, 101, 103–107,

157–158, 175, 184, 212–216, 218, 262 – extended-now 29, 96, 101–102, 107–110,

114, 157, 162, 175, 184, 213–215, 248, 272, 340–342, 346–347

– hot news see recent past – indefinite anterior see experiential – persistent situation see extended-now – recent past 29, 69, 110, 113–114, 158,

162, 175, 184, 202, 212–218, 224 – resultative 23–25, 96, 114, 157, 163, 175,

184, 212–218, 262–267, 272–273, 322 – stative see extended-now

– variant surface realizations – after-perfect 102, 110–113, 171, 224, 230 – bare form 178–179, 196–200, 202,

205–207, 209–210, 217–218 – be-perfect 178–179, 196–197, 219,

259–290 – versus simple past tense 104, 121–122,

125–126, 130, 132–135, 137, 139–140, 149, 196–200, 202, 209–213, 216–218, 223–226, 230–231, 250, 298–300, 302–306, 308–320, 322, 324, 326–329

persistence 261–263, 270, 284–286, 290phonology 282–283, 348 polysemy 2, 31; see also perfect – semanticsproductivity 263–265, 275, 289

Subject index 353

prescriptivism see constraints – normativity preterite see perfect – versus simple past

tensepriming 130, 134, 137, 329

quotative marking 128, 132, 185–189 – be like 127, 130, 134, 136, 138 – quotative go 130, 136, 138 – zero quotative 128, 130, 132–133, 186

register see formality and informalityrevitalization 286–287

Scottish English see Hebridean Englishsecond-language acquisition (SLA) see

language acquisitionselection and diffusion, model of 170,

173–174, 179, 189semantic extension see semantic shiftsemantic shift 34, 121–122, 127, 152, 163,

285, 299–300semantics of the perfect see perfect –

semanticssimple past see perfect – versus simple past

tensesimplification 96–97, 112–114, 154, 165, 197,

206second-language instruction see language

instruction

sociolinguistic factors see constraints substrate 96, 100–102, 106, 109, 112,

114–115, 153–154, 170, 200, 206–207, 210, 248–251, 264–265, 285

temporal adverbial constraint see perfect – present perfect puzzle

temporal adverbials 2–8, 30, 33, 104, 106–107, 109, 114, 121–122, 127, 150, 152–153, 158–159, 176, 180–184, 188–189, 199, 206, 214–216, 218, 289, 298–303, 305, 308–313, 316–322, 324–325, 328–329, 340, 343

temporal specification see temporal adverbials

verb phrase – complementation 51, 86 – evidentiality 62–63, 89, 131 – mutative verbs 181–182, 208, 217,

261–265, 277–278, 285 – transitivity 151, 181–184, 189, 207–208,

217, 261–264, 268–275, 277, 279–282, 285–287

– verbs of motion 151, 187, 261versals 6, 114–115, 206, 246, 287–288

Date post:	15-Sep-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

Re-assessing the Present Perfect - WordPress.com · Collins, Johan Elsness, Andreas Eriksson,...

Documents