Verbs Observed: A Corpus-drivenPedagogic Grammar1 *SUSAN HUNSTON and GILL FRANCIS
University of Birmingham
This paper describes the outcome of a project to code the complementationpatterns of all the verbs in Collins COBUILD English Dictionary (1995) COBUILDstands for 'Collins and Birmingham University International LanguageDatabase' The coding is based on the Bank of English corpus at COBUILD anduses a simple notation based on words and word-classes rather than traditionalfunctional categories The result of this exercise is a list of verb patterns, with acomplete list of all the verbs in a corpus of 250 million words that have eachpattern It is found that the verbs that share a pattern fall into groups based onmeaning This grammar is the first pedagogic grammar to integrate syntax andlexis using corpus data The grammar is used to explore traditional grammaticalcategones such as Object, Complement, etc These are found to be inadequateto account for the actual behaviour of verbs Finally, the paper explores thepossibility of using a pattern grammar to analyse naturally-occurring discourse
INTRODUCTION
Current developments in English language teaching methodology andsyllabus design present a challenge to the writers of pedagogic grammarsBecause of the more widespread use of electronically-stored corpora amongteachers and researchers alike, there is a growing expectation that descriptionsof language will be based on quantities of authentic data rather than on acourse writer's intuitions and/or language prejudices Because of the rapid-processing requirements of task-based methodologies (J Willis 1996), there isa need for a grammar that addresses fluency as well as accuracy Because ofthe emphasis on lexis as the crucial element in language learning, there is aneed for a grammar that is lexically based and that can be integrated fully intoa lexical syllabus This paper proposes a pedagogic grammar of English that isderived from substantial quantities of naturally-occurnng data, is lexically-based, and that lends itself to exploitation by text-based and concordance-based consciousness-raising techniques
It is by now very well documented that large, electronically-stored corporaprovide an opportunity to study and describe language in ways that arequantitatively and qualitatively different from the models and descriptionsthat have previously been available In particular, grammarians using corporacan offer more comprehensive and authontative descriptions of the languagethan has hitherto been possible, because more data is available for
46 VERBS OBSERVED A CORPUS-DRIVEN PEDAGOGIC GRAMMAR
examination This in itself offers new insights into, for example, the relativedistributions of certain grammatical features (see, for example, Sinclair 1991,Halhday 1993, Biber 1988) In addition, corpora and the analytical tool ofconcordancing have prompted some grammarians to view grammar as asystematic collection of observations about the way words behave rather thanas a set of abstractions into which words may be fitted {see, for example,Sinclair 1991, Francis 1993, 1994, Hunston 1995, and the debate betweenOwen 1993 and Francis and Sinclair 1994)
This paper reports on a project to describe comprehensively the verbs thatare defined in the Collins COBUILD English Dictionary (1995) (hereafterCCED) In turn, this is part of a larger project to descnbe words belonging toall word-classes in a similar way The corpus used in the project was the 250million word Bank of English at COBUILD 2 At the time that the research wasdone, about 75% of the corpus comprised written data, 25% was spoken,71% of the data was from Britain, 25% from the United States and 4% fromAustralia
The first stage of the project was to descnbe the complementation patternsof each verb occurring frequently enough in the corpus to be defined inCCED 3 The descriptions themselves will be demonstrated, in some detailbelow, the aim was to provide a descriptive apparatus that was as transparentas possible, and which was flexible enough to cope with verb behaviours thathad not previously been descnbed This part of the work was carried outduring the compilation of CCED The second stage was to make general-izations about each of those patterns In particular, it was noted that verbsshanng a pattern also shared aspects of meaning, this enabled us to providelists of verbs that are useful, in that the verbs in each list share not only apattern of behaviour but also a meaning The result is a pedagogic grammar ofthe verbs of English as exemplified in our corpus (Francis et al 1996)
Observing patternsThe basic concept in our pedagogic grammar is the 'pattern' To demonstratethe meaning of this, we shall use the concordance lines below, showing asample of 30 instances of the verb consider (this is the lemma consider,compnsing the forms 'consider', 'considers', 'considered' and 'considering')The lines are arranged according to the type of group,'phrase or clause thatfollows the verb in each concordance line, that is, according to the behaviourof the verb
SUSAN HUNSTON and GILL FRANCIS 47
1 economy-minded Congress, no longer consider a Philippine base to be a2 it clearly unconstitutional and we consider it to be a flagrant violatio3 rraless foods, many researchers now consider them to be potentially harmf4 that to provide condoms could be considered to be promoting homosexual
5 indecently flattered He does not consider himself a celeb, retains the6 encourages the grief process by considering it a normal part of human7 arts and sciences It is valid to consider memory the oldest mental'ski8 To desecrate a holy spring is considered profanity Severe chastise
9 levous by premeditation many laws consider a premeditated crime more se10 history when few achievements were considered greater than writing a bea11 tterly. Howadays, of course, he is considered strange for questioning Am
12 gave these lectures in Aberdeen, I consider Aberdeen as an example Aber13 s long tradition Approaches which consider gender as the central oppre14 why the region should be considered as a special case foe extr
15 there is no scriptural reason to consider women as inferior Created i16 ranging species, it may be best to consider them as beneficial and only
17 nder-estimated anyway but when you consider it's concentrated on the str16 the social costs If we do not consider that the costs are worth pay19 UN High Commissioner for Refugees, considered that moat asylum seekers 1
20 of the victims, who they will consider a private prosecution of the21 says her government will consider a proposal from West Germany22 European Community countries are considering a proposal that the organ23 Act, the government would have to consider carefully the law which woul24 general's office says the study considered everything from malaria to
25 s birthday, we thought we'd consider what the Revolutionary War m26 Morris Was that a factor in considering whether or not to bring t27 with smaller peps should seriously consider whether to remain in the
28 say privately that they are considering asking to join the 59-dea29 Faldo claimed he had not even considered missing the halfway cut fo30 spread over into Austria, they are considering providing free electricit
In lines 1-3, the verb is followed by a noun group (realized by a pronoun inlines 2 and 3) and a to-infinitive clause (that is, a clause introduced by aninfinitive preceded by to) Line 4 shows the passive pattern an auxiliary andpast participle of the verb consider, followed by a to-infinitive clause For thesake of economy, we use the symbols 'V for verb group (capitalized becausethe verb is the word under scrutiny here), 'n' for noun group, and 'to-inf forto-infinitive clause, and we allow the sequence of symbols to represent thesequence of the groups and clauses themselves For lines 1-3, then, thesequence 'V n to-inf represents the pattern, and we say that the verb considerhas the pattern 'V n to-inf For line 4, it is necessary to show the passiveformed by the verb be and a past participle We therefore use the sequence 'beV-ed' For line 4 the complete pattern sequence is 'be V-ed to-inf
In lines 5-7, the verb is followed by two noun groups, one after the other 4
This is represented by the pattern sequence 'V n n' Line 8 is an example ofthe passive a passive verb followed by one noun group, or 'be V-ed n' In lines9-10, the verb is followed by a noun group and then an adjective group Thispattern is 'V n adj' Line 11 is the passive 'be V-ed adj'
In lines 12-13, the verb is followed by a noun group and then by aprepositional phrase beginning with as In other words, the verb is followed by
48 VERBS OBSERVED A CORPUS-DRIVEN PEDAGOGIC GRAMMAR
a noun group, the preposition as and another noun group This is representedby 'V n as n' (The as is italicized showing that it represents an actual lexicalitem, not a class of words ) Line 14 shows the passive 'be V-ed as n' Lines 15-16 show a similar pattern, except that as is followed by an adjective groupinstead of a noun group The pattern is 'V n as adj'
In lines 17-19, the verb is followed either by a clause beginning with theword that or by a simple declarative clause These two patterns could betreated separately, but because almost no verbs use one pattern but not theother we subsume both in the pattern 'V that'
Lines 20-24 show the verb followed by a noun group (the pattern 'V n')while lines 25-27 show the verb followed by a wh-clause, that is, a clauseintroduced by a wh-word (the pattern 'V wh') The final group of concordancelines, lines 28-30, show the verb followed by a clause beginning with apresent participle The pattern is 'V -ing'
CCED distinguishes three main senses of consider the first sense may besummanzed as 'to have an opinion about', the second sense is briefly 'to thinkabout', and the third sense is 'possibly intend to do something' (A fourthsense deals with the phrase all things considered ) In the concordance linesgiven above, lines 1-19 illustrate sense 1, lines 20-27 illustrate sense 2, andlines 28-30 illustrate sense 3 In the dictionary, the codes for the grammarpatterns associated with each sense are shown alongside the block ofexamples (see Figure 1) For reasons of space in the dictionary entry, thepassive patterns are not given Also, the two patterns 'V n n' and 'V n adj' areshortened to 'V n n/adj', and the two patterns 'V n as n' and 'V n as adj' areshortened to 'V n as adj/n' (the adjective group pattern being more frequentthan the noun group one in this case)
consider /ksnsidy/ consider*, considering,consideredI If you consider a person or thing to be something, you have the opinion that this is what theyire We don t consider our customers to be mereconsumers, we consider them to be our friends. Ihad always considered myself a strong, competentwoman. The paper does not explain why fontgnownership should be considered bad. I consideractivities such as jogging and weightlifting as un-
natural. Barbara considers that pet shops whichsell customers these birds are very unfair2 If you consider something, you think about it VEMcarefully The government Is being asked to consid ^*era plan to fix the date of the Easter break. You do v*ihave to consider the feelings of those around you.Consider how much you can afford to pay for acourse, and what is your upper limit3 If you arc considering doing something, you in VEMtendtodolt.buthavenotyetmadeaftnaldeclsion •"***«whether to do It I had seriously considered telling v-**the story from the point of view of the wives v n
Waterspora enthusiasts should consider htnng aweuuitaswetlasatijejacket They are consideringtht launch of their own political party4 You say all things considered to indicate that MUSE.you are making a judgement after taking til thefacts into account Ml things considered, I thinkyou haim behaved marvellously in coming here.5 Sec also considered comldertrm
Figure I
SUSAN HUNSTON and GILL FRANCIS 49
Here are similar concordance lines for another verb, insist
1 it impossible to trust. Laurence insisted he was being punished for Geor2 t, to avoid a boycott, the group insisted he take a stand to overturn3 have to go The troubled Premier insists he is not ready to step down4 ago in New York The ambassador insisted it was the first time ever tha5 in Crewe, Cheshire, bosses insisted no more redundancies were plan6 esterday (Sunday) decided not to insist that its members must support MrI lapse Mr Kravchuk, however, has insisted that he maintained constant8 as leader. The paper says they insisted that despite poor results m9 ead Surrounded by reporters, he insists that the only reason he's lunch10 s friend the prime minister, who insisted that the heritage secretary hII e asked to leave - The pair, who insist they will make no deal with the12 Brooke remained silent but aides insisted they were ready to deal with h
13 was smart and demanding, he had insisted on clear thinking and had litt14 en abandoned and that they would insist on a parliamentary vote on whe15 uldn't give him a game, but he'd insist on a battle and then they'd set16 treasurer, Kathleen Brown, to insist on ]ust that, and is atill rated
17 a and the other furniture they'd insisted on taking and destroying He sIB and, despite the lack of funds, insisted on dressing him in clothes sen
19 ch upheld the council's right to insist on housing applicants producing20 horses are his babies he always insists on them being treated very21 King of Corinth in Euripides, he insists on Jason marrying his daughter
22 to our potential ' Illingworth insisted 'I want us to have an England23 it shouldn't be successful,' he insists There's a serious side to Clar
24 on? Oh you here Pete? OK, if you insist Come on, let's get this show on25 ed to talk him out of it, but he insisted They were unahppy when they g
Lines 1-12 show the verb followed by a clause with or without the word that(the pattern 'V that') In lines 13-16, the verb is followed by the prepositionon and a noun group {the pattern 'V on n'), while in lines 17-18 thepreposition is followed by a clause beginning with a present participle {thepattern 'V on -ing') Lines 19-21 show another variation on this theme thepreposition on is followed by a noun group and then by a present participle(the pattern 'V on n -ing') Lines 22-23 show insist used with a direct quote(the pattern 'V with quote') The final pattern for this verb in the corpus isshown in lines 24-25, where insist is used by itself {the pattern 'V')
The above examples demonstrate our pnnciples in denoting patterns Ouraim is to give a sequence of symbols that will adequately represent the surfacepatterning of groups of similar concordance lines The minimum of analysis isused—there is no indication, for example, as to whether a noun groupfollowing a verb is an Object or a Complement Groups and clauses areidentified, however, and are generalized into noun group, adjective group,that-clause, wh-clause, '-ing' clause and so on Abbreviations are used torepresent these groups and clauses, such as 'n' for 'noun group', 'wh' for 'wh-clause' and so on Specific lexical items are shown in italics, and the wordwhich is the focus for the pattern (the verb, in most of the examples given inthis paper) is represented by a capital letter (V)
The following kinds of patterns have been identified (Francis et al 1996)
50 VERBS OBSERVED A CORPUS-DRIVEN PEDAGOGIC GRAMMAR
• The verb stands by itself V• The verb is followed by a group of a particular kind V n, V adj, V adv• The verb is followed by a clause of a particular kind, or by a particular form
of a verb V that, V wh, V wh to-inf, V to-inf, V -ing, V inf, V -ed• The verb is followed by a prepositional phrase beginning with a named
preposition V about n, V as n, V at n, e tc , or by a prepositional phrasebeginning with one of several prepositions V prep
• The verb is followed by a noun group and then by another group of aparticular kind V n n, V n adj, V n adv
• The verb is followed by a noun group and then by a clause of a particularkind, or by a particular form of a verb V n that, V n wh, V n wh to-inf, V nto-inf, V n -ing, V n inf, V n -ed
• The verb is followed by a noun group and then by a prepositional phrasebeginning with a named preposition V n about n, V n as n, V n at n, etc , orby a prepositional phrase beginning with one of several prepositions V nprep
• The verb has the word it as part of the pattern, e g it V that, V it adj that,etc
• The verb has the word there as Subject there V n, there V prep/adv
More about patternsIn CCED, every sense of every word is annotated with the set of patterns thatthe sense commonly occurs with, as in Figure 1 The lack of traditionalanalysis in this annotation is an advantage, as no artificial-hmit is placed uponthe constituent elements of a pattern (See below for further discussion oftraditional grammatical analysis ) A pattern may be composed of quite anumber of elements, such as 'V on n ing' (e g he insists on Jason marrying hisdaughter), or 'V it adj to-inf (e g She had not felt it necessary to tell him), or itmight contain elements expressed by lexical items, such as 'V way prep' {e g
he talked his way into the post of chief costume designer) Many patterns,however, are more simple, and consist of only two elements, such as 'V that'(eg We should insist that European aid funds focus more on helping people develop)or 'V -ing' (e g I'm considering taking you on)
Some patterns have always been part of a traditional system of grammaticalanalysis (see, for example, Hornby 1954, Quirk et al 1985) For example, thepatterns 'V to-inf and 'V -ing' ( e g / started to follow him up the stairs and Snowbegan falling again) have long been part of every English teacher's repertoireThe very simple patterns 'V and 'V n' (e g His parachute collapsed and Twoyoung girls were collecting firewood) correspond to the traditional distinctionbetween intransitive and transitive verbs The traditional analyses of Object-Object Complement ( e g / had always considered myself a strong, competentwomen) and Indirect Object-Object (e g They gave us T-shirts and stickers) areboth expressed by the pattern 'V n n'
Other patterns identified m CCED, however, are less familiar to the language
SUSAN HUNSTON and GILL FRANCIS 51
teaching syllabus This is particularly true of patterns involving specificprepositions, such as 'V about n' (eg / heard about the acadent) and 'V in n'(e g The books abound in social comedy), and it is even more true of patterns inwhich the verb is followed by a noun group and then by a prepositional phrase,such as 'V n against W <e g You have to weigh the responsibilities against the rewards)and 'V n into -ing' (e g His elder brother forced him into cutting his hair) Suchprepositions tend to be taught in an ad hoc way, if at all, with particular verbsand prepositions being associated fairly randomly We shall argue below thattreating the preposition as part of a pattern allows for a more systematic andhelpful approach to the grammar of the verb
As we saw above with the patterns of the verbs consider and insist, somesenses of some verbs have several patterns Sense 1 of consider, for example,has six patterns, not counting the passive ones As Sinclair (1991 53) pointsout, however, it is also true that sense and pattern tend to be associated witheach other, such that a particular sense of a verb may be identified by itspattern The verb recover has two main senses 'to get better' following anillness or period of unhappiness, and 'to get back' something that was lost Thefirst of these senses has the patterns 'V from n' (e g He is recovering from a kneeinjury) and 'V (e g // took her three days to recover), whilst the second has thepattern "V n' (e g Police recovered stolen goods) CCED divides each of thesetwo senses into three, covering metaphorical as well as literal meanings (seeFigure 2), but the pattern annotation shows where the major sense divisionlies Pattern and meaning are strongly associated
r v c o w r /nkAvr/ recovers, recovering, ra-covered1 When you recover from an illness or an Injury, veuyou become well again He is recovering from a vtamknee injury A policeman was recovering in hospi- Jtal last night after being slabbed. Heisfulfyrecovend from the virus2 If you recover from an unhappy or unpleatani VEMexperience, you itop being upset by It a tragedy v*wnfrom which hi never fully recovered. Her plane y
broke down and it was 18 hours before the got thereIt took her three dayt to recover& If something recover* from • period of weakneu VEMor difficulty, it Improve! or gen stronger again He ««w>recovered from a 4-2 deficit to reach the auarUr- v
finals The ttockmarktt Index fell by 80% before Itbegan to recover4 If you recover something that hai been lost or VEMstolen, you find It or get it back. Police raided five ^ w *houses In touth-east London and recovered stolen "goods. Rescue teami recovered more bodies fromthe rubble5Ifyouf«eov»ramentalorphrslcalstate,ftcomei VEMbackagaJn Forexample,ifyour«cowrconsdoui •™»lln
ness, you become conscious again Foramlnutehe vnlooked uncertain, and then recovered hit compasun She had a severe attack of asthma and it tookan hour to recover her breath. She never recoveredconsciousness61fyourccommoneythatyouhaveipent.lnvest- VEMed, or lent to someone, you get the same amount •f*oue
back Legal action is being taken to try to recover the vnmoney The British market alone was not largeenough to recover their costs of production
Figure 2
52 VERBS OBSERVED A CORPUS-DRIVEN PEDAGOGIC GRAMMAR
More about pattern and meaning
The close association between pattern and meaning is further demonstrated ifa large group of the verbs coded with a particular pattern in CCED are listedFor example,'here is an alphabetical list of 48 of the most common verbs,including phrasal verbs, which have the pattern 'V as n' {that is, the verb, orverb and particle, is followed by a prepositional phrase beginning with as)
Even the most cursory glance indicates that this is not a random group ofverbs Many of them are close to synonyms, for example enlist and volunteer,resign, stand down and step down, begin and start, end, end up, finish and finish up,pass and qualify, masquerade and pose, act, function and serve It is possible tohighlight the similarities in meaning by dividing the verbs into 'meaninggroups' Here are six meaning groups derived from the list of verbs givenabove {see Francis et al 1996 159-62) 5
Group 1 verbs concerned with working at a jobtraining for a job qualify, trainapplying for a job enlist, put up, run, sign up, stand, volunteerdoing a job (in a particular way) freelance, moonlight, practise, serve, shape up,workleaving a job resign, stand down, step down
I can see it [an auditing course] coming in so useful when I'm practising as a doctor toknow what it's all about [British spoken]Clinton returned to Little Rock In the next few days, he is expected to resign asgovernor and announce the first of his Cabinet appointments [US radiol
'SUSAN HUNSTON and GILL FRANCIS 53
Group 2 verbs concerned with having a role or function
Fleming emphasized that the departments within City Hall no longer function asseparate entities [US newspaper]
Group 3 verbs concerned with beginning and ending
beginning begin, originate, start, start off, start outcontinuing continue, remainending end, end up, finish, finish up
In most of these sports, Chinese athletes start as favourites [BBC radio)He ended up as one of the people in Oxford living rough [British spoken]
Group 4 verbs concerned with being perceived in a particular way
come across, come over, count, emerge, go down, pass, qualify, rank, rate, shape up
But I would never pass as a Norwegian fisherman [Bntish magazine]Your concern may come across as anger, and this won't help the situation [US non-fiction book]
Group 5 verbs concerned with trying to seem to be something you arenot
masquerade, parade, pose, dress up
So far as it was possible for a successful politician. Smith refused to pose as somethinghe was not [US non-fiction book]We think it's terribly sweet if the three-year-olds dress up as a bnde and groom[British spoken]
Group 6 verbs with other meaningscome, come out
Still, the news came as a shock [US newspaper]/ came out as a lesbian when I was still in my teens [British magazine]
Note that some verbs, i e qualify, serve and shape up, appear in more than onemeaning group In these cases, a different sense of the verb appears in eachgroup
In all the patterns we have identified (Francis et al 1996), a similar divisioninto meaning groups can be made Here is another example—the pattern 'V nfrom n', which has four mam uses
(a) It is used with some verbs concerned with movement, either physical ormetaphorical, as in these examples As soon as a leaf is large enough, remove^it from the plant [British magazine] He noticed a figure detach itself from the
54 VERBS OBSERVED A CORPUS-DRIVEN PEDAGOGIC GRAMMAR
shadows of the building and make its way towards him [British fiction) Buthe was unable to extract any information from the three men [US fiction]
(b) It is used with some verbs concerned with the prevention of an activityThe pattern 'V n from -ing' is also used with these verbs Here are someexamples It is in the interests of alt nations to prevent any state from marchinginto another and taking it over [BBC radio] Religion these days seems to justdivert people from what religion is about [Bntish spoken]
(c) It is used with some verbs concerned with keeping or making somethingsafe or hidden, as in these examples The body's immune system protects usfrom both infection and cancer [British non-fiction book) / never did concealanything from the Congress [US radio] A schoolgirl who helped rescue twotiny sisters from their blazing home was hailed a heroine yesterday [Britishnewspaper]
(d) It is used with some verbs concerned with making something, as in thisexample Boxgrove was a sandy beach on which Stone Age humansmanufactured cutting tools from flints [British magazine]
If we look at a complete list of all the verbs in CCED which have this pattern,we can put them into groups which correspond to each of these main uses
Note that absent and tear away are always used with a reflexive pronoun (atleast in our corpus) you absent yourself from a place, or tear yourself awayfrom something
3 Verbs concerned with transferring someone or something from oneplace to anotherdivert, forward, pass, promote, relay, send, transfer, transplant, transpose
In addition, there is a small group of verbs which are concerned with sayingthat someone is not guilty of a cnme or misdemeanour
11 absolve, exonerateThere are also a few verbs which have meanings not covered by any of thesegroups
12 deflect, dispossess, draw, omit, rouse, take awayThe division into meaning groups has been done intuitively, with the aim ofmaking each group as sensible as possible in order to help the learner associatemeaning with pattern and so remember which verbs have which patternsThe groups are, however, always open to reinterpretation by otherresearchers It would be possible to argue, for example, that in groups 1-4of 'V n from n', the verbs indicating physical movement should be separatedfrom those indicating metaphorical movement However, although the actualdivisions in the meaning groups are arguable, and therefore not trulyrephcable, it is also true that, given the list of verbs with a particular pattern
56 VERBS OBSERVED A CORPUS-DRIVEN PEDAGOGIC GRAMMAR
as a whole, any observer would be able to group the verbs in some sensibleway This supports our basic tenet, that patterns are closely associated withmeaning and that lexis and syntax are not separate as areas of language studyWe will return to this point below
Combinations of patternsLevin (1993) has noted that verbs which share a meaning may share not justone pattern, but two or more For example, many of the verbs meaning 'thinkor talk about someone or something as having particular qualities' have boththe pattern 'V n as adj' and the pattern 'V n as n' ( described the death asunfortunate describe their employment as a job) (Francis et al 1996 592-3)Another large group of verbs have the patterns 'V n as n' and 'V n n' (Franciseial 1996 593-4, cf Levin 1993 78) Some of these have the meaning'thinkor talk about someone or something' (branded the jobless as layaboutsbranded her an opportunist), others mean 'put into a job or role' (elected him asprime minister elect me President) A much smaller group of verbs have allthree patterns 'V n as adj', 'V n as n', and 'V n n' These are brand, consider,designate, label, rate An even smaller group has these patterns plus 'V n adj'brand, consider, label, rate (labels herself as fat label him as a psychopathlabelled her a saint labelled his work naive)
Figure 3 shows the distribution of patterns for a selection of the verbsinvolved
Pattern and transitivityAlthough the emphasis on verb pattern found in both Levin's work and ourown represents a break with tradition, our attempt to analyse verbcomplementation does not Traditional valency grammar has always cat-egorized verbs according to whether they are intransitive, monotransitive,ditransitive, and so on (see, for example. Quirk et al 1985 1168 ff), usingfunctional categones such as Object, Complement, Adjunct, and ObjectComplement In some cases, such an analysis has much to offer For example,we have already mentioned that the pattern 'V n n' has two distinctfunctional analyses the two noun groups may both be Objects {one of them issometimes, misleadingly we think, termed the 'Indirect' Object6), as withverbs such as give, bring, and tell, e g My parents gave me a microscope kit,alternatively, the first noun group may be an Object and the second an ObjectComplement, as with verbs such as call (in one sense), elect and adjudge, e gMy children called him Uncle Frankie Such an analysis makes very clear thedifference in the relation between the two noun groups, relations which aredependent on the verb used It is also helpful to distinguish between a Verbwith Object and a Verb with Complement, as in The children ate the biscuits andShe became a full-time singer, both of which have the pattern 'V n'
We have already mentioned that we found it unsatisfactory to take the
SUSAN HUNSTON and GILL FRANCIS 57
accept
acclaim
brand
characterize
certify
consider
count
declare
descnbe
designate
elect
label'
pronounce
proclaim
rate
view
V n adj
yes
yes
yes
yes
yes
yes
yes
yes
V n m adj
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
Vn n
* yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
yes
Vn«m
yes
yes
• yes
yes
yes
i yes
yes
• yes
yes
yes
yes
yes
yes
yes
Figure 3
traditional functional categories as the primary means of coding verbs, assuch a pnori categones were too restnctive to account for all the patterns wefound in our corpus However, given the usefulness of distinctions such asthose discussed above, we have attempted to map all our patterns on totraditional functional categones {Francis et at 1996) We did not aim toreplicate any one system of analysis, but to devise one that was internallyconsistent, rephcable, and which seemed to accord with common sense Theproblems we encountered in so doing convinced us that these traditionalcategones can give an adequate picture of only a small proportion of theverbs used in English For most of the patterns, and most of the verbs in ourcorpus, it is extremely difficult to devise coherent cntena that will lead to asatisfactory analysis, and even an analysis that was theoretically satisfactorywould not offer the user any more useful information than a coding of thesurface pattern does It is difficult to imagine, for example, how an analysisof I'd love her to go into politics as 'Verb + Object' helps the learner to know
58 VERBS OBSERVED A CORPUS-DRIVEN PEDAGOGIC GRAMMAR
how to use the verb love in this way more than the pattern description of 'Vn to-inf does
To corroborate this assertion, we will present two of the mam problems thatwe encountered in mapping our verb patterns onto functional categoriesThere is insufficient space here to do justice to the many problems we havemet (see also, for example, Francis 1994), but we will present a very smallsample, in each case giving the cntena that we used to make often difficultdecisions
Prepositional phrases
When a pattern consists of a verb followed by a prepositional phrase, there is aproblem as to how that prepositional phrase is to be analysed in functionalterms For example, in the clauses Let's separate into smaller groups or Bobbumped into him or Theodora Adams dug into her purse (all with the pattern V inton), what functional category/categories do into smaller groups, into him and intoher purse realize7 One possible solution .is to say that every prepositionalphrase is an Adjunct, but this presupposes a one-to-one relation betweenform and functional category which is not the way English typically behavesIn addition, the noun group a normal, healthy child 'feels as though' it playsmuch the same role in the clause he grew into a normal, healthy child as it doesin the invented equivalent he became a normal, healthy child (although grow intoand become are not interchangeable in context, of course) Similarly, a digestivebiscuit 'feels as though' it plays the same role both in Weatherby bit into adigestive biscuit and in the invented Weatherby ate a digestive biscuit Using this,admittedly highly subjective, criterion, a normal, healthy child would be aComplement in both of the clauses cited, and a digestive biscuit would be anObject However, to use this terminology would be to mask the work that thepreposition into is undoubtedly doing We have therefore used the term'prepositional Complement' to mean 'has the role of a Complement, but isintroduced by a preposition' and the term 'prepositional Object' to mean 'hasthe role of an Object, but is introduced by a preposition' This leaves thepossibility of some prepositional phrases being Adjuncts, but not all
We can then apply the same criterion to identify a prepositionalComplement as is used to identify a Complement if the person or thingconstrued by the noun group realizing the Subject is or becomes the person orthing construed by the noun group following the preposition, then theprepositional phrase is a prepositional Complement, as m these V into nexamples He grew into a normal, healthy child The downturn has turned into acollapse The oil tanker has broken into several pieces
Distinguishing between a prepositional Object and an Adjunct is muchmore difficult Some cases are clear Weatherby bit into a digestive biscuit ^eemsto be a kind of Object, whereas The guests crowded into the ornate gothic rooms isequally obviously an Adjunct But what about She burst into tears or Neurons inthe brain spark into life7 Passivization criteria are of little help there are no
SUSAN HUNSTON and GILL FRANCIS 59
examples of full passive forms such as a biscuit was bitten into and the rooms werecrowded into in the corpus, and we became increasingly uncomfortable withrelying on intuition regarding possible transformations In the absence ofabsolute criteria, it is necessary to operate a default (Sinclair 1986 44), that is,if one of the categones can be satisfactorily identified, everything that remainsis assumed to belong to the other category We have used the criterion of'place or circumstance' to identify Adjuncts That is, to qualify as an Adjunct,the prepositional phrase must indicate a place or circumstance, typicallyanswering the question 'where7 ' or 'how7 ' (For the use of question criteria inlanguage analysis, see Winter 1982 7-9 ) Everything else is a prepositional-Object Thus, the prepositional phrases in The guests crowded into the ornategothic rooms and Theodora Adams dug into her purse are Adjuncts (answenng thequestion where?), whilst those in Weatherby bit into a digestive biscuit and Sheburst into tears are prepositional Objects There remains the problem ofmetaphorical places or circumstances, such as into life in Neurons in the brainspark into life We have chosen to interpret examples such as this as Adjuncts,but we would admit that this is open to debate This approach is defensible onthe grounds that it allows us to maintain a consistent methodology
The problems that have faced us in making these analyses illustrate thechallenges which arise when, because a large corpus of naturally-occurringlanguage is used, and because examples are not selected to illustrate a pnontheones of transitivity, the analyst is faced with the full range ofcomplementation patterns verbs have, and the full range of verbs that havethose patterns As always when looking at large amounts of real data,theoretical categories that appear neat and watertight dissolve before one'seyes The point is not whether questions of functional analysis are difficult oreasy to solve, it is whether they are, ultimately, worth solving What examplessuch as bit into a biscuit, crowded into a room, dug into her purse, burst into tearsand spark into life have in common is their pattern, 'V into n' Where theydiffer is in the meaning group that the verb in each case belongs to We wouldargue that this information alone is necessary and sufficient for a grammaticalunderstanding of the clauses
Phase
'Phase' is the term used to refer to two verbs that constitute a single verbgroup Certain patterns place two verbs in sequence with each other, mostnotably 'V to-inf and 'V -ing' Verbs that have these patterns may beconsidered candidates for being in phase with the other verb in the patternFor example, many verbs such as begin, like, and promise all have the pattern'V to-inf, occurring in two-verb sequences such as began to laugh, likes toentertain, and promised to take It is possible that the two verbs in each sequenceare in phase and form a single verb group Similarly, the verbs finish, detest,and regret have the pattern 'V -ing' and give nse to two-verb sequences suchas finished eating, detested being photographed, and regret having revealed, which
60 VERBS OBSERVED A CORPUS-DRIVEN PEDAGOGIC GRAMMAR
again are candidates for a phase analysis (It is of course well known thatmany verbs have both patterns, either with a similar meaning, e g like, start,or with a different meaning, e g remember, try )
The scope of 'phase' is variously defined The COBUILD English Grammar(1990 184ff) considers any two verbs in sequence, that is, any two verbsoccurring in the patterns 'V to-mf or 'V -ing' to be in phase, whatever themeaning relation between the two verbs In other words, there is nopossibility of two verbs occurring in either of those two patterns and notbeing in phase The term 'phase', then, is being used as a simple shorthand forthe two patterns, rather than as an analytical category This alternative runsinto problems only when further functional analysis of the clause isattempted When two verbs are in phase, any group, phrase, or clause thatoccurs as the complementation pattern of the second verb in the verb group,and that is analysed as belonging to a particular functional category withrelation to that verb, must also belong to that category with relation to thewhole verb group For example, in the clause She ate an apple, the noun groupan apple is part of the complementation pattern of the verb eat—the pattern is'V n'—and in functional terms is analysed as the Object If another verb isused in front of eat, and if the two verbs are analysed as being in phase, theapple must then be the Object of the whole verb group For example,-if theclause is She began to eat an apple, and began to eat is considered to be in phase,then the apple is the Object of began to eat If the clause is She promised to eat anapple, and promised to eat is considered to be in phase, then the apple is theObject of promised to eat It seems to us that whereas the first of these examplesyields a satisfactory analysis, the second does not In the first example, themeaning of the verb group is essentially eat, with began to having a modal-likefunction The question which would elicit the clause is 'What did she begin to.do7 ' , not 'What did she begin7 ' In the second example, the meaning of theverb group is essentially promise, and it is not known whether the apple wasultimately eaten or not The question eliciting the clause may be either 'Whatdid she promise to do7 ' or 'What did she promise7 ' The COBUILD,EnglishGrammar ignores the distinction between begin and promise As we shall arguebelow, we agree in essence with this approach, but only if functionalcategones themselves, including the notion of phase, are dispensed withMeanwhile, we turn again to the distinction between begin and promise
Downing and Locke {1992 328-31), who use the term 'verbal groupcomplex' consider phase to apply to a more restricted set of relationshipsThey describe seven types of phased relationship which occur with a to-mfinitive initiation (It began to rain), continuation (It continued to snow),termination (/ have ceased to mind the harsh climate), appeanng or becoming real(The sky seemed to get darker), attempting or not attempting (He tried to learnArabic), manner or attitude (/ regret to inform you), and modality (She happenedto see the accident) The first four of these also occur with an -ing form, as in Itbegan raining We have followed what we see to be the unifying factor behindDowning and Locke's meaning categones, which is that when two verbs are
SUSAN HUNSTON and GILL FRANCIS 61
in phase, the main purpose of the first verb is to indicate either that the actionrealized by the second verb is done (and how it is done) or that it is not doneOur meaning groups are similar to theirs except that we put into one groupverbs of beginning, continuing or stopping an action, and we sub-divide the'modality' group into several groups We also add a group including verbssuch as fail, forbear, neglect, and omit, which indicate that the action indicatedby the second verb is not done
We therefore analyse as in phase only those sequences of verbs which entaileither the doing or the not doing of the activity indicated by the second verbThus, these examples are in phase
/ began to think that it was all a total waste of time [British spoken]The diagnosis turned out to be her worst nightmare (British magazine]What he learned served to improve the managerial skills needed in his presentemployment [US non-fiction book]He omitted to mention one or two things that might have helped him [Britishfiction]/ regret to tell you that very many American lives have been lost [US radio]
feet tend to swell if you sit still for too long [British non-fiction book]
These examples are not in phase
/ had planned to stay longer, but something came up [British fiction]The police asked to use Keith's video as evidence [British magazine]We expect to see her back on the screen in the autumn [British newspaper]It worries me I like to feel secure [British spoken]'He claims to have had no inside knowledge [British magazine]Everyone was waiting to get off the ferry [Bntish spoken]
Some problems, however, remain Often a verb is used in different ways, thatis, in phase with another verb or not, depending on the context For example,like to feel in the example above is not in phase just because the speaker likesto feel secure does not mean she does feel so (or that she does not feel so) Onthe other hand, in the following example we might say that the speaker doesentertain and does so willingly Alternatively we might argue that, as with /like to feel secure, the speaker is simply expressing a preference
She likes to entertain, shop and go to the theatre
The problems with phase do not end here If omit to mention is analysed as inphase, because it entails do not mention, then surely the sequence of verbs inrefrain from naming should also be analysed as in phase, as it entails do notname Using this argument, we have analysed many of the patterns involvinga verb followed by a preposition and the present participle of a second verb asphase Examples include 'V about -ing' (did not bother about digging), 'V from -ing' {shrink from making), 'V in -ing' (participate in making), 'V on -ing' (insistedon giving), 'V with -ing' (proceed with overhauling) The problem is that, as aconsequence of this, very similar clauses end up with very different analyses.
62 VERBS OBSERVED A CORPUS-DRIVEN PEDAGOGIC GRAMMAR
For example, in Sometimes we shrink from making decisions, the sequence ofverbs shrink from making is in phase, but in He never shrank from a fight, theequivalent verb and noun group shrank from a fight cannot be in phase,because no two verbs are involved
We must emphasize again at this point that, although we have arrived atanalyses which we believe to be reasonably replicable and internallyconsistent, we do not consider those analyses to be particularly satisfactoryWe do not really wish to make sharp distinctions between begin to eat, want toeat, and promise to eat, for example, but we believe that these false distinctionsare forced upon us if we wish to allocate functional category labels It ispreferable by far not to use these labels and to use only the pattern as analysis
Patterns and pedagogic grammarFunctional categories of analysis anse out of a view of language that sees lexisand grammar (or syntax) as separate features of language organization At itsworst, this is the 'slot and filler' approach—a language consists of a set ofstructures into which any lexical items can be dropped Such an extreme viewwould be unlikely to be espoused currently by any linguist, but it is still thecase that most studies of syntax, while making some indication of lexicalrestnctions on acceptability, do not explore in any comprehensive way theinterdependence of lexis and syntax
Unfortunately, the traditional study of syntax as a discrete system has formany years influenced pedagogic grammars, that is, descriptions of English forEnglish Language Teaching Kerr {1996 95), for example, reports that intheview of many trainee EFL teachers 'Grammar exists independently of other,aspects of language such as vocabulary and phonology' Like Lewis (1993),Willis D and Willis J (1996 64) make a plea for 'perspectives on grammarwhich take account of the interface between grammar and lexis'
This interface is precisely captured by the notion of pattern Our study ofverb patterns has suggested that it is impossible to descnbe verb comple-mentation without focusing on lexis Any complementation pattern isdescnbable only in terms of the particular lexical items that have thatpattern, and the meaning groups that they fall into Without a reasonablycomprehensive list of the verbs that have a given pattern, no account of thatpattern can be given Similarly, if we take lexis as a starting point, informationabout any lexical item must include the patterns it has It is, after all, a truismof language teaching that a learner does not really 'know' an item ofvocabulary until he or she knows how it is used Most learners' dictionariesacknowledge this fact and indicate syntactic information in some way in theway the definition is phrased, in the examples, by explicit coding, or througha mixture of these methods
Thus, because it is impossible to describe syntax without recourse to lexis orlexis without recourse to syntax, we argue that syntax and lexis arecompletely interdependent In fact, we would prefer ultimately not to use
SUSAN HUNSTON and GILL FRANCIS 63
the terms 'syntax' and 'lexis' at all, and to reserve the term 'grammar' for asingle descriptive system comprising words and patterns In other words, wewould prefer to speak of one system, rather than of two interdependentsystems However, for historical reasons, it is impossible to argue for such anapproach without using the notion of two systems as a starting point
In recent years, the dearth of innovation in pedagogic grammars has beencompensated for by a richness of research in lexis, notably, an awareness ofthe centrahty of lexis to the language learning task {see, for example. Carterand McCarthy 1988, Lewis 1993, D Willis 1990), an awareness of theimportance of collocation to language learning, and an awareness of multi-word lexical 'chunks' that are used by native speakers of a language withoutthe need for 'internal processing' (Skehan 1996 41, Pawley and Syder 1983,Nattinger and DeCamco 1992) Patterns, we argue, bridge the gap betweenlexicahzations and rules. Some of the word sequences which can be explainedin terms of patterns may also be regarded as lexical chunks Examples includeIt occurred to me/him/her that {'lit V ton that'), is remembered as ('V n as n'), do alot for ('V amount for n'), put (some) money into ('V n into n'), talk oneself out of('V pron-refl out of n'), blast something open, cram something full, drive someonemad {all 'V n adj') Other sequences which we descnbe in terms of theirpattern also have 'rule-governed' explanations These include the pattern 'V nabout n' with verbs such as dislike, feel, hate, like, love, and think, which usuallyoccur in questions or cleft constructions beginning with What, such as What Iloved about Gloria was her talent and her independence {Francis et a! 1996 340)
This suggests that the two types of system are not dichotomous, but arealternative ways of viewing the same phenomenon Highly frequentcollocations, which appear as 'fixed phrases', are in fact simply extremecases of patterning, where the lexis is particularly restncted Utterances thatappear to have been produced by the operation of rules, such as yes/noquestions, may be seen simply as patterns which accept a relatively widerange of lexical items
Patterns as analytic grammarWe have termed the description of English exemplified above a 'pedagogicgrammar' That is, it shows learners how the language they are learningworks Analytic, or theoretical grammars, on the other hand, perform adifferent function They are tools for analysing bits of a language one alreadyknows with the aim of saying something useful about that bit of language Inthis final section we will consider what the consequences are of treating apattern grammar as an analytic grammar
As an example, here is a sample of spontaneous dialogue taken from thespoken corpus in the Bank of English {Both speakers are male Numbersindicate parts of the tramcnpt that will be discussed below )
64 VERBS OBSERVED A CORPUS-DRIVEN PEDAGOGIC GRAMMAR
A- [l]Now this is an interesting point [2]I wonder if you found this here'cos I did at Sheffield
B [3]MmA [4]There was a tension between halls of residence and union peopleB. [5]MmA [6]Erm we tended to think of ourselves as not having to worry about
those freaks down there [7]Was there anything like that7
B [8]Oh yesA [laughs]B [9]Yes [lOlYes I mean we were seen really as erm not quite the oddities
but you know [11]I think there was an unspoken jealousy that therewere certain people in hall and others who were not
A [12]Yes [pausel MmB [13] And I suspect er although I'm not conscious of this we probably
treated the others a bit as second-class citizensA [14]Yes That's rightB [laughs]A [15] YeahB [16]Totally unjustified
Companng each utterance in this dialogue with the whole of the Bank ofEnglish, we can show that although the conversation itself is unique, almostnone of the component parts of it is Nearly every utterance in thisconversation exemplifies the way one or more of the words in it oftenbehaves
Multi-word items
To begin with, there are several multi-word lexical items in this dialogue, suchas halls of residence [4] and second-class citizens [13], common phrases such as ohyes [8], that's right [14], I mean [10], you know [10], I think [11), and frequentcollocations, such as interesting point [1], totally unjustified [16] (In the Bank ofEnglish there are 14,357 instances of adjective+/WMf and 284 instances ofinteresting point, giving a t-score of 16 7813 There are 596 instances ofunjustified and 46 instances of totally unjustified, giving a t-score of 6 7775 A t-score of 2 or more is considered to be significant)
Verb patterns
Turning to the focus of this paper, the verbs, we can show the pattern usedwith each verb in the dialogue In every case except one, the pattern isshown in CCED as part of the typical behaviour of the relevant sense of theverb concerned The pattern 'there V n prep', which occurs with were(utterance 11) is not shown in CCED, but the omission is rectified in Franciset al 1996
V thatbe V-ed as n (V n as n)V thatthere V nthere V n prepV prepV thatVadjV n as n
Other patterns
Other words besides verbs have patterns, of course Here are the otherpatterns in this dialogue, also coded in CCED
[4] tension N between pl-n[13] conscious ADJ ofn
In addition, there are some commonly-occurring patterns that do not attachto individual words but to whole phrases There are two examples in thisdialogue / wonder 'cos and not quite but
[2] I wonder 'cos
The combination / wonder and its complementation pattern is often followedby a clause beginning with because The because clause gives a justification forthe indirect question This pattern occurs a minimum of 34 times out of 4,718instances of / wonder in the Bank of English 7 These concordance lines showsome examples
1 here we were and so on I wonder where Agnes and I were Because she's qu2 rnor Clinton there, and I wonder about that because actually the economy3 ervaluing his own' And I wonder who the writer of this is because it gui4 pretty good really erm I wonder where she is 'cos she's <F01> I know it'5 e when I apply for jobs I wonder if there is now 'cos they've got so many6 d me forward. <F01> Cos I wonder whether you got the encouragement 'cos l7 d cope Virginia says I wonder how you'd cope? Because it sounds to me8 French and Wilson then I wonder if we saw the same film, because to me B
66 VERBS OBSERVED A CORPUS-DRIVEN PEDAGOGIC GRAMMAR
This co-occurrence of / wonder and because is not a complementation pattern ofthe verb wonder, but might be termed a discourse pattern belonging to thephrase / wonder
[10] not quite but
This is another possible discourse pattern It occurs a minimum of 162 timesout of a total of 6,229 instances of not quite in the Bank of English Theconcordance lines below show some examples
1 Mary Winga. They're not quite a cast of thousands, but you'll hear a lo2 ar the Colorado line, not quite a ghost town, but a place beginning to3 frais you will have, not quite a calorie-free dressing, but almost4 she had a nice voice Not quite Aled Jones, but certainly musical The fi5 more clearly defined, not quite Cambridge, but certainly not Albion, whoa6 y, twenty-four? Voice not quite flint-hard yet but ]ust this side of it
What this commentary suggests is that relative frequency of co-occurrence,which is said to ease the psychological task of language processing, is not aphenomenon that is restricted to certain lexical phrases, however numerousthese might be, but might be extended to patterns in general In other words,the'idiom principle' (Sinclair 1991 110) might be said to imply, not only thatcertain sequences of lexical items commonly occur (interesting point, halls ofresidence, I mean, I think, second-class citizens) but that certain sequences of wordand word-type—patterns—also commonly occur (think of reflexive pronounas -ing clause, be negative conscious o/noun group) Moreover, this second typeof co-occurrence plays a major role in organizing each utterance Unsurpris-ingly, there is no part of the analysed dialogue that is not accounted for interms of being part of the pattern of one or more of the lexical items in it
Atypical patterning
Comparing our dialogue with a corpus, we can show what is infrequent aswell as what is frequent In utterance [11)/ think there was an unspoken jealousythat there were certain people in the hall and others who were not, the use of jealousyin the pattern 'N that' is uncommon, in that out of 2,473 occurrences ofjealousy in the Bank of English, only 9 have the noun followed by appositivethat {Francis 1993 lists some of the many hundreds of nouns that occur muchmore frequently in this pattern, such as announcement, assertion, observation,recommendation, suggestion, assumption, conviction, hypothesis, knowledge, opinion,view, assessment, conclusion, inference, speculation, anger, astonishment, fear, sorrow,and surprise)
The ranty of jealousy that suggests that in our text the speaker is using apattern creatively Creativity in the association of words with patterns ispresumably one of the ways in which language change takes place It ispossible that the chances of an unusual usage being understood and judged tobe unremarkable may be increased if the word is used as part of a longer.
SUSAN HUNSTON and GILL FRANCIS 67
commonly-occurring sequence In this case jealousy that occurs as part of asequence that involves the introductory there was and the adjective unspokenThis lengthy pattern sequence ('there + form of be + noun group + that-clause') does not occur with any great frequency, but it does have a noticeable24 occurrences tn the Bank of English Here are all the concordance lines
he United States — there was an unspoken Agreement amongst the women thatike to say so publicly, there is unspoken agreement that big banks would nd with hia parents, there was an unspoken agreement that he and Viv would '
villager said 'There was an unspoken agreement in the village that wethe poor of London, there is an unspoken assumption that those who are
recedent There is, however, the unspoken assumption that only a man canother people*do not, there is an unspoken assumption, if the child choosesas can women There was also an unspoken assumption among other family
metaphor There is a substantial unspoken belief about AIDS, similar tothe question there is always an unspoken but never justified implication
ationship suffered There was an unspoken conflict that there shouldn't bet and irritable and there was an unspoken fear that this revival wasn'tgo Officials said there was an 'unspoken feeling of disquiet' that Franceactively fear him, there was an unspoken feeling among everyone m thet you know I think there was an unspoken jealousy that there were certainf King's fantasies, there was an unspoken promise that these diffusee sexual instincts, there was an unspoken rule that nothing ever went beyonstories like that, there was an unspoken understanding that they should no
o moved away There is always an unspoken understanding that someor ingratiating There is an unspoken understanding among women thatbat an eyelid ' There was an unspoken understanding between brother and
Here we find a range of nouns used, the most frequent being agreement,assumption, and understanding Jealousy that is still rare, but we would speculatethat it increases its acceptability by being part of the longer pattern
Pattern frequencyThis short piece of naturally-occurring dialogue illustrates a variety of verbpatterns, such as 'V adj', 'V that', 'V wh', 'V to-inf, 'V n as n', 'there V n', aswell as introducing noun patterns (N between pl-n), adjective patterns (ADJ ofn), and the more nebulous concept of discourse pattern (/ wonder because,not quite but) One of the questions that arises from considenng the role ofpattern in text is how frequent certain patterns are, and in particular howfrequently one can expect to encounter patterns other than the very familiar'V, 'V n', 'V that' and 'V to-inf This question cannot yet be answered, but itis noticeable that this dialogue does contain no fewer than three examples ofverb patterns with as ('V ofn as -ing', 'be V-ed as n', 'V n as n') These are
Speaker A think of ourselves as not having to worrySpeaker B we were seen as not quite the odditiesSpeaker B treated the others as second-class citizens
These examples of these patterns all illustrate the general meaning of 'havingan opinion about someone' or 'doing something which indicates that youhave an opinion about someone' It is perfectly possible to indicate thismeaning with other patterns, however, notably 'V that' (We thought we were
68 VERBS OBSERVED A CORPUS-DRIVEN PEDAGOGIC GRAMMAR
oddities), 'V as if (We behaved as if they were second-class citizens) and 'V n n' (Theyconsidered us oddities) Future research might investigate to what extent thecoincidence of pattern use is significant It is worth hypothesizing thatspeakers who are accommodating to one another (Giles and Coupland 199160 ff) may 'copy' each other's patterns, even if the lexis they use (think, see,treat) is different
Pattern flow
Although a descriptive grammar must deal with patterns in isolation, whenthat grammar is applied to analysis, what is striking is the way that thepatterns flow into one another In other words, each utterance moves fromone pattern to the next As an example, consider utterance [6] in the sampledialogue
We tended to think of ourselves as not having to worry about those freaks down there
This utterance begins with the pattern 'V to-inf, based on the verb tend Thenthe second verb in that pattern, think, begins the pattern 'V ofn as -ing' Thefinal element of that pattern, having to worry begins yet another pattern—'Vabout n'—based on the verb worry Whenever a word that is part of a patternhas a pattern of its own, this phenomenon of 'pattern flow' occurs As a result,there are two ways to analyse the patterns in this utterance one thatemphasizes the paniculate nature of the patterns, and another thatemphasizes their wave-like nature (Pike 1959, Halhday 1982) The particulateanalysis looks something like this
V
tended
V
think
V
not having
to-inf
to think of ourselves as not having to worry about those freaksdown there
ofn
of ourselves
to worry
as -ing - , /
as not having to worry about those freaks down there
•
about n
about those freaks down there
Alternatively, an analysis which drew attention to the 'flow' of patterns mightlook something like this
SUSAN HUNSTON and GILL FRANCIS 69
tended
V
to think
to-inf
V
of
of
ourselves
n
as
as
not having
-ing
to worry about
V about
those freaks down
-
n
there
Another example of a 'flow' analysis is this representation of utterance [10]
I mean
V
we were
that
be
seen
V-ed
as
as
not quite
not quite
the
n
oddities but
but
you
•
know
The flow analysis shows a form of language that is, in Halhday's terms,choreographic rather than crystalline (Halhday 1987) Unlike Halhday'sanalyses, however, this is based on the pattern rather than the clause, andin fact the importance of the clause to grammatical analysis seems todisappear In Sinclair's {personal communication) terms, what may matteris not where a unit of analysis finishes, but only where it starts
CONCLUSION
In this paper we have outlined an approach to the descnption of language thatunites lexis and grammar We have shown that lexical items have descnbablepatterns, and that words shanng the same patterns tend to fall into groupsbased on shared aspects of meaning This in turn suggests that the patternsthemselves can be said to have meanings, and there is some evidence that theuse of a lexical item with a pattern that it does not commonly have is aresource for language creativity and, possibly, for language change
This approach to grammar is, we believe, helpful to language learnersbecause it focuses on the surface behaviour of individual words The coding ofthe surface behaviour, with a minimum of interpretation into categories, isintended to be transparent and therefore accessible to learners It must be saidthat learners will need help initially to decipher codings such as 'V n from n/-ing', but the principle, once grasped, is straightforward enough, and it seemsto us that a learner is unlikely to find a coding such as 'V+O' or 'v trans' anymore helpful than our preferred 'V n' The focus on individual lexical items,and the attempt to provide exhaustive lists of verbs with a particular pattern,is intended to enable the learner to have confidence in whether a particularusage is correct or not Our lexical approach to grammar answers two
70 VERBS OBSERVED A CORPUS-DRIVEN PEDAGOGIC GRAMMAR
learners' questions 'How is this word used7 ' and 'What other words are usedin the same way7 '
A grammar of this type is, however, a reference book rather than acoursebook, and as such is as hkely to be used by teachers as by learners Wewould argue that a pattern-based approach to word behaviour is likely to beuseful to teachers devising consciousness-raising activities, particularly thosebased on authentic written or spoken texts The pattern is, we suggest, avaluable focus for the learner's attention In addition, the availability of lists ofwords which would otherwise be difficult for the teacher to retrieve fromintuition enables the teacher to impose on the chaos of individual wordbehaviours an order based on meaning and, more specifically, to use thepattern as a basis for vocabulary building
Although we have presented our argument mainly in terms of thebehaviour of one particular word class—verbs—the notion of patterns is notrestncted to verbs Words of all classes, especially nouns, adjectives, andadverbs, are also descnbable in terms of their patterns Ultimately, the divisioninto word classes may be seen as a matter simply of administrativeconvenience, rather than a reflection of the language
Our work suggests that a pattern grammar provides learners with adescription of English that is useful for the development of both accuracyand fluency It should aid accuracy because it provides detailed informationabout the behaviour of individual lexical items, moreover this information isnot random but is organized by the meaning groups that words having aparticular pattern form It should aid fluency because the patterns effectivelyconstitute learnable 'chunks', each pattern flowing into the next
(Revised version received April 1997)
NOTES1 The authors would like to dedicate this paper lish, although we are aware that this
to the memory of Eugene Gatt Winter, who assumption is not uncontroversialwas an enthusiastic and supportive consul- 3 The dictionary defines all verbs which occurtant of the project, as well as a dear fnend in the corpus at least 15 timesHis death in December 1996 was a great loss 4 In line 5 the first noun group is realized byto the world of Applied Linguistics a reflexive pronoun If reflexive pronouns
2 The Bank of English is constantly growing are always or very frequently used with aand statements about its size and propor- particular verb, the coding reflects this Fortions quickly become inaccurate At the example, the sense of exert exemplified intime that the research for this paper was Try not to exert yourself while working isdone, it was approximately 250 million coded V pron-refl' In the case of consider,words At the time of writing (May 1996), however, the reflexive pronoun is notit is approximately 290 million words, and judged to be significantly frequent and isthe proportion of spoken data and data from not codedAustralia has increased In this paper we 5 It has been pointed out to us that in the casewill treat the Bank of English as reasonably of most of these verbs, the noun grouprepresentative of certain varieties of Eng- indicates something that has not been
SUSAN HUNSTON and GILL FRANCIS 71
mentioned before and is therefore notusually 2 personal pronoun
6 Our dislike of the term 'Indirect Objectstems from the fan that in an examplesuch as My parents gave me a microscope kitme and a microscope kit are affected equallythough in different ways The distinctionbetween Direct Object and indirect Objectdoes noi capture these relationships Afterall, in answer the question and answer your
father, both the question and your father aretraditionally analysed as Direct Objects,even though the relations between verband noun in the two cases are by no meansthe same
7 We use the phrase 'a minimum of becausewe counted only those lines where becauseor cos occurred within 40 characters Thefull total may be more than 34 but will notbe less
REFERENCESBiber, D 1988 Variation Across Speech and
Writing Cambridge Cambridge UniversityPress
Carter, R and M McCarthy 1988 Vocabularyand Language Teaching London Longman
Collins COBUILD English Grammar 1990 LondonHarperCollins
Collins COBUILD English Dictionary 1995London HarperCollins
Downing, A and P Locke 1992 A UniversityCourse in English Grammar London PrenticeHall International
Francis, G 1993 'A corpus-driven approach togrammar principles methods and examplesin M Baker et al (eds ) 1983 Text andTechnology In Honour of John Sinclair Amster-dam Benjamins
Francis, G 1994 Grammar teaching inschools What should teachers be awareat7' Language Awareness 3 221-36
Francis, G, S Hunston, and E Manning1996 Collins Cobutld Grammar Patterns IVerbs London HarperCollins
Francis, G and J Sinclair 1994 '"I bet hedrinks Carling Black Label A riposte toOwen on corpus grammar Applied Linguist-ics 15 190-200
Giles, H and N Coupland 1991 LanguageContexts and Consequences Milton KeynesOpen University Press
Ha) lid ay, M A K 1982 How is a text like aclause7 in S Allen (ed ) Text ProcessingProceedings of Nobel Symposium 5/ StockholmAlmquist and Wiksell International
Halliday, M A K 1987 Language and theorder of nature' in N Fabb et al (eds ) 1987The Linguistics of Writing Manchester Man-chester University Press
Halliday, M A K 1993 'Quantitative studiesand probabilities in grammar" m M Hoey
(ed ) 1993 Data Description Discourse Paperson the English Language in Honour of John McHSinclair London HarperCollins
Homby, A S 1954 Guide to Patterns and Usagein English Oxford Oxford University Press
Hunston, S 1995 Grammar in teacher educa-tion The role of a corpus ' Language Awareness 4 1 5-32
Kerr, P 1996 Grammar for trainee teachers'in J Willis and D Willis
Levin, B 1993 English Verb Classes and Alter-nation A Preliminary Investigation ChicagoUniversity of Chicago Press
Lewis, M 1993 The Lexical Approach LondonLTP
Nattinger, J R and J S DeCarnco 1992Lexical Phrases and Language Teaching OxfordOxford University Press
Owen, C 1993 Corpus-based grammar andthe Heineken effect Lexico-grammaticaldescription for language learners AppliedLinguistics 14 167-87
Pawley, A and F Syder 1983 'Two puzzlesfor linguistic theory Nativelike selection andnativehke fluency' in J C Richards andR Schmidt {eds ) 1983 Language and Commumcation London Longman
Pike, K L 1959 'Language as panicle waveand field ' The Texas Quarterly 2 37-54
Quirk, R, S Greenbaum, G Leech, and JSvartvik 1985 A Comprehensive Grammar ofthe English Language London Longman
Sinclair, J 1991 Corpus Concordance, Colloca-tion Oxford Oxford University Press
Skchan, P 1996 'A framework for theimplementation of task-based instruction 'Applied Linguistics 17 38-62
72 VERBS OBSERVED A CORPUS-DRIVEN PEDAGOGIC GRAMMAR
Willis, D 1990 The Lexical Syllabus London Willis, J and D Willis (eds) 1996 ChallengeHarperCollins and Change in Language Teaching Oxford
Willis, D a n d J Willis 1996 'Consciousness- Heinemannraising activities in the language classroom', Winter, E 1982 Towards a Contextual GrammarinJ Willis and D Willis of English London Allen and Unwin
Willis, J 1996 'A flexible framework for task-based learning' in J Willis and D Willis
