Available online at www.sciencedirect.com
The syntax of predicate ellipsis in ItalianSign Language (LIS)
www.elsevier.com/locate/linguaLingua 166 (2015) 214--235
Carlo Cecchetto a, Alessandra Checchetto b, Carlo Geraci c,*,Mirko Santoro c, Sandro Zucchi d
aUniversity of Milan-Bicocca, ItalybCa Foscari University, Venice, Italy
cCNRS, Institut Jean-Nicod, Paris, FrancedUniversity of Milan, Italy
Received 5 February 2014; received in revised form 23 December 2014; accepted 29 December 2014Available online 30 January 2015
We analyze a hitherto undescribed case of ellipsis in Italian Sign Language (LIS) and show that it has common properties with VPellipsis in languages like English. For example, the ellipsis site can contain a wh-trace and semantic restrictions on the type of predicatethat can be omitted are only derivative. We thus propose a phonological deletion approach for the LIS construction. We also consider theissue of how the content of the ellipsis site is recovered from its linguistic antecedent. We present new arguments for a syntactic identitycondition, although a limited number of mismatches between the ellipsis site and its antecedent, notably including vehicle change cases,must be accommodated. 2015 Published by Elsevier B.V.
Keywords: Ellipsis; Predicate ellipsis; Italian Sign Language (LIS); Adverb incorporation; Vehicle change; Role shift
Because of the huge bias toward spoken languages in the existing linguistic literature, no systematic exploration ofpredicate ellipsis has ever been tried in any sign language, although scattered information on the phenomenon is present.For example, Jantunen (2013:30--31), in a paper mainly devoted to argument ellipsis in Finnish Sign Language, reportsthree sentences, which are candidate respectively for being cases of gapping, VP ellipsis and sluicing. Quer and Rossello(2013), in a paper on argument drop in Catalan Sign Language, address the issue of predicate ellipsis and Schlenker(2014) discusses possible semantic mismatches between the missing constituent and its antecedent.
In this paper we aim at filling this gap by systematically analyzing an Italian Sign Language (LIS) construction in which apredicate goes unuttered. We will show that this construction has common properties with VP ellipsis as attested inlanguages like English.
We will focus on two main questions through the paper. Does the ellipsis site contain internal structure? Does ellipsisrequire an antecedent with the same form or an antecedent with the same meaning? Our account of LIS, which is spelledout in the Principles and Parameters framework, will lead us to answer yes to the first question, while, as far as the secondquestion is concerned, we will favor the identity in form hypothesis but recognize that it raises some non-trivial issues.
* Corresponding author. Tel.: +33144322679.E-mail address: [email protected] (C. Geraci).
http://dx.doi.org/10.1016/j.lingua.2014.12.0110024-3841/ 2015 Published by Elsevier B.V.
C. Cecchetto et al. / Lingua 166 (2015) 214--235 215
Taken together, the answers to the two questions provide additional evidence from languages in the visual modality infavor of a phonological deletion account of VP ellipsis.
This paper is organized as follows. In section 2 we present some background information about LIS grammar, whichwill be required later in the paper. In section 3, we explain how we collected the data and give information about ourinformants. Section 4 introduces the basic properties of predicate ellipsis in LIS. In section 5 we show that genuine casesof VP ellipsis can be found in LIS (in addition to stripping) and in section 6 we argue that two arguments from the spokenlanguage literature can be reproduced in LIS and support a phonological deletion analysis for VP ellipsis. In section 7 weinvestigate the problem of the recoverability of the content of the ellipsis site and, based on arguments which specificallyexploit the simultaneous nature of morphology in sign languages, we conclude that identity of form with a linguisticantecedent is required. Section 8 presents some further thoughts on the issue of recoverability conditions on ellipsis.
2. Essential background information about LIS grammar
In recent years, several aspects of LIS grammar have been investigated fairly extensively, although many areas arestill unexplored. In this section we do not aim at offering a comprehensive review of this literature but we give backgroundinformation on LIS which is required for the discussion about ellipsis.
Although in LIS, like in other sign languages, word order is not rigid, our informants agree that the unmarked word orderin simple declarative sentences is S(ubject)--O(bject)--V(erb):1
1 See A2 In the
informatio3 The s
GIANNI MARIA LOVE
Gianni loves Maria
Functional signs like the modal verb MUST (Fig. 1), the perfective marker DONE (Fig. 2, cf. Zucchi, 2009; Zucchi et al.,2010) and the auxiliary for future (Fig. 3)2 are postverbal. Manual negation is also found after the verb in LIS (Geraci, 2006), asshown in (2)--(5). When both negation and a modal sign are used, the modal precedes the negative sign as in (6).
GIANNI APPLY MUST
Gianni must apply
GIANNI HOUSE BUY DONE
Gianni bought a house
(4)GIANNI HOUSE BUY FUT3
Gianni will buy a house
(5) GIANNI MARIA LOVE N OT
'Gi anni doesn't lov e Maria'____ neg
(6) GIANNI CONTRACT SIGN CAN N OT
Fig. 3. FUTURE marker.ig. 1. MUST. Fig. 2. DONE.
ppendix for conventions used in the glosses of LIS sentences. variety of LIS under investigation, tense is not marked by a morpheme on the verb and temporal information is inferred from contextualn when no time adverb or FUT auxiliary is present (see Zucchi, 2009 for a study on tense in another variety of LIS).ign glossed here as FUT has been glossed as MUST by Zucchi (2009). The reason for this difference in glossing is that this sign was
a modal verb of necessity and it is still used like that by some LIS signers. However, at least for our informants, now it is used only as anfor future. Our consultants use a different sign (glossed here as MUST) for the modal of necessity.
C. Cecchetto et al. / Lingua 166 (2015) 214--235216
Wh-phrases are found at the right periphery of the sentence (cf. (7) and (8)), cf. Cecchetto et al. (2009). Although it is veryrare in spoken languages, systematic clause-final placement is not uncommon in sign languages (cf. Cecchetto, 2012 foran overview of the literature):
_____ wh(7) GIANNI BUY WHAT
'What did Gianni buy?'
(8) HOUSE BUY WHO
'Who bought a house?'
There is clear evidence that wh-phrases are more peripheral than modal verbs, negation, DONE and FUT:
(9) CAKE EAT CAN WHO 'Who ca n eat the cake?'
(10) CAKE EAT N OT WHO
'Who did not eat the cake?'
(11) HOUSE BUILD DONE WHO
'Who built the house?'
(12) HOUSE BUILD FUT WHO
'Who will build the house?'
Research on topicalization in LIS is very limited but two points are clear: (i) topic phrases are found in the left periphery ofthe clause (unlike wh-phrases), (ii) they occur with a specific non-manual-marking (roughly, raised eyebrows). This isillustrated by the example in (13)
_____ raised eyebrows(13) VASE GIANNI BREAK
'Gi anni broke a vase'
Also yes/no questions occur with a raised eyebrows non-manual-marking. Indeed, declaratives and yes/no questionsare distinguished in LIS only by the occurrence of this non-manual-marking, not by a change in word order.
The last aspect of LIS we would like to briefly introduce here concerns locative phrases. Locative phrases will be usedin this paper as a way to create a contrast between the antecedent clause and the elliptical clause, an expedient to makeellipsis constructions fully acceptable. When no topicalized element is present, the most natural position of locativephrases is the sentence initial position and the locative phrase co-occurs with raised eyebrows, as in (14a). When both atopic and a locative phrase are present, the natural position of the locative phrase is after the main verb (as in (14b)),although it is also possible to have a topic and a fronted locative phrase. When the latter happens, then the topicalizedphrase must precede the locative element, as shown by the contrast in (15)
______________ raised eyebrows(14) a. GARDE N IX - LOC GIANNI VASE BREAK
_____ raised eyebrowsb. VASE GIANNI BREAK GARDEN IX-LOC
'Gi anni broke a vase i n the gard en'
_____raised eyebrows_______________ raised eyebrows
(15) a. ? VASE GARDE N IX- LOC GIANNI BREAK
'Gi anni broke a vase i n the gard en'
b. * GARDE N IX- LOC VASE GIANNI BREAK
Following Cecchetto et al. (2006), we assume that LIS is a head-final language, at least in the clausal domain, since theverb follows the object and the functional heads that host aspectual markers, negation, the auxiliary for future and modalsfollow the verb. Following Cecchetto et al. (2009), we assume that wh-phrases sit in Spec, CP but this position is linearizedto the right. Following Geraci et al. (2008), we accommodate topicalized