Gapping as vp-coordination: an argument from French strict NPI licensing Anne Dagnac (Université Toulouse 2 / CLLE-ERSS) - PDF

Description
Gapping as vp-coordination: an argument from French strict NPI licensing Anne Dagnac (Université Toulouse 2 / CLLE-ERSS) le presbytère n' a rien perdu de son charme, ni le jardin

Please download to get full document.

View again

of 6
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information
Category:

Small Business & Entrepreneurship

Publish on:

Views: 17 | Pages: 6

Extension: PDF | Download: 0

Share
Transcript
Gapping as vp-coordination: an argument from French strict NPI licensing Anne Dagnac (Université Toulouse 2 / CLLE-ERSS) le presbytère n' a rien perdu de son charme, ni le jardin de son éclat (G. LEROUX, Le Parfum de la dame en noir, 13-14) Introduction Debates on the syntactic structure of gapped clauses such as (1) involve two major questions: (1) Ward can t eat caviar and Sue beans (a) What is the second conjunct? In particular, is it a (a.o. Hartmann 2000, Kim 2006, Gengel 2009), or a vp (a.o. Coppock 2001, Johnson 1996, 2009), or a special construction (Culicover & Jackendoff 2005, Abeillé & al. 2011)? (b) Does the gap in it proceed from the non-pronunciation of some constituent at PF (Sag 1976, Hartmann 2000, Gengel 2009, Coppock 2001) or from ATB movement (Johnson 1996, 2009)? B. vp coordination what is coordinated is NOT a but a vp (Coppock 2001, Johnson 1996 & sq) deletion, but the remnants adjoin to VP2 rather than at the sentence level (Coppock 2001) (2) John eats caviar, and [ VP Mary 1 [ VP beans 2 [VP t1 eats t2 ]]]]. there is no «ellipsis»(ie deletion) but movement of the common parts (Johnson s analysis) (4) Some will eat beans and other rice (from Johnson 2009 : 22) This talk focuses on the first question, and argues that data from the licensing of French negative coordinator ni nor bring further support to the vp-coordination view. I will first briefly present the three competing analyses, then the behavior of ni. In the last section, I will show the consequences of this behavior for an analysis of gapping. 1. Gapping : competing analyses A. Classical analysis : coordination & PF-deletion coordination at level the second conjunct is a full part of which gets unpronounced, presumably after topicalization of the remnants (a.o. Ross 1967, 1970, Sag 1976, Neijt 1979, Hartmann 2000, Coppock 2001, Merchant 2004, Chaves 2005, Konietzko & Winkler 2010, Molnár & Winkler 2010). (2) John eats caviar and Mary beans (A) (B) Variant : Munn (1992) 1 FP 1 & 1 ConjP John eats caviar Mary FP beans 2 t Mary likes eats t beans Main claim : gapped clauses have the same syntax / interpretation as ungapped ones (± focus) Cf. in particular extraction (3) De quel sujet i [Paul n a-t-il jamais parlé t i à son fils] ni [Léa n a-t elle jamais parlé t i à sa fille]? About which topic didn t Paul ever talk to his son nor did Léa ever talk to her daughter 1 For arguments against Kayne s (1994) proposal that and heads a ConjP whose specifier is 1, see. Borsley (2005), in particular for French. & 2 Main argument : semantic scope, in particular negation (plus DP quantifiers, binding, ): Negation can scope over the two conjuncts (4 ) Kim didn t play bingo and Sandy sit at home all evening [not [Kim play bingo and Sandy sit at home all evening]] [not [Kim play bingo]] and [not [Sandy sit at all all evening]] C. «Special construction» French : HPSG framework (Abeillé & al. 2011) - in line with Cullicover & Jackendoff (2005) coordination at level (adjunction in a flat structure) the second conjunct is not a full but a special construction Interpretation : maps a headless structure to a clausal meaning (à la Dalrymple & al. 1991) : (2) John eats caviar and Mary beans John and XP eats caviar Mary beans Main claim : - vs A : gapped and ungapped clauses do not have the same distribution (5) a. Paul a parlé à Marie et non pas Marie à Paul. b. *Paul a parlé à Marie et non pas Marie à parlé à Paul [gap ok / * full clause] Paul has talked to Mary and not Mary (*talked) to Paul (6) a. *Il a dit que Paul viendrait lundi et que Marie jeudi. b. Il a dit que Paul viendrait lundi et que Marie viendrait jeudi [full clause ok / *gap] He said that Paul would come on Monday and that Mary *(would come) on Thursday S S ConjP 1 A. Dagnac Ellipsis2012 Vigo November 2012, - vs B : makes wrong predictions wrt position of initial conjunctions (7) a. Et Paul ira à Paris, et Marie à Londres And Paul will go to Paris and Mary to London b. * Paul ira et à Paris et Marie à Londres Paul will go and to Paris and Mary to London 2. French ni I argue that French gapped clauses in ni - are a problem for A and C - are more compatible with some version of B A. Ni is a coordinating conjunction Ni, like et ( and ) and ou ( or ), is a coordinating conjunction, with a meaning comparable to nor. It can conjoin any kind of phrases ( constituent ni ) : (8) a. Je n ai jamais parlé [à Paul ni à Pierre] [PP] I have never talked [to Paul nor to Pierre] b. Je n ai jamais vu Jean ni parlé à Pierre [VP] I have never seen John nor talken to Peter It can conjoin clauses ( clausal ni ) : In this case, contrary to English nor, it does not trigger inversion. (9) a. Jamais Paul ne serait allé à Londres ni Marie ne serait restée à Paris Never Paul would have gone to London nor Mary would have stayed in Paris b. Jamais les envoyés du roi Jacques n'avaient vu autant de rivières se jeter les unes dans les autres ni autant de gorges et de vallées s'entrelacer. (J Lanzmann, La Horde d or, 346) It can enter a «double conjunction» construction with XPs (10) Je n ai parlé ni à Paul ni à Pierre I have talked neither to Paul nor to Peter With s (11) Ni l Allemagne ne triomphera de nous, ni nous ne triompherons de l Allemagne (Gide, in Grevisse 1033b) Neither Germany will win over us, nor we will win over Germany (12) Ni il travaille, ni il étudie, ni il braque! Alors quoi? Il baise. (V. Therama, Bastienne, 1985, pp ) NI he works, NI he studies, NI he steals! So, what? He fucks. It enters gapped constructions (13) Jamais Angiolina n'a été aussi amoureuse de moi, ni moi d'elle. (G. Matzneff, Ivre de vin perdu, 1981, p. 242) Never Angiolina NE has been as in love with me, nor me with her B. Ni is a strong/ strict NPI (De Swart : 2001, Mouret 2007) De Swart (2001) and Mouret (2007) : simple ni is a strong NPI (contra Doetjes 2005). In contemporary French (vs previous stages of the language) : It can only appear in downward entailing non-additive contexts; in the scope of a negation within the first conjunct in the scope of the averidical preposition sans ( without ) (14) a. *Il(n )a parlé à Marie ni à Pierre [no negation] He talked to Mary nor to Peter b. *Il voit rarement Paul ni Marie [not non-additive] He sees rarely Paul nor Mary c. *Jim ni Lea (n')aime(nt)(pas) le poisson [not in the surface scope of negation] Jim nor Lea like (not) fish (15) a. Il n a pas/jamais parlé à Marie ni à Pierre [in the scope of a negation] He didn t talk to Mary nor to Peter b. Personne n a parlé à Marie ni à Pierre Nobody talked to Mary nor Peter (16) Il est venu sans Marie ni Pierre [in the scope of without ] He came without Mary nor Peter Initial ni is a negative item, and it can license ni2 without the help of another negation (de Swärt 2001). (17) Je n ai vu ni neg Paul ni pol Marie (De Swärt) I haven t seen ni neg Paul ni pol Mary ni can be seen as a polar version of or in logical terms, a or in the scope of negation: A ni B corresponds to (A B), which is logically equivalent to ( A) ( B) ni must be in the scope of a negative licensor, as strict NPIs do. Problem : which scope? An expression a has direct scope over an expression b, if and only if b is in the semantic scope of a, and a c-commands b in the syntactic structure. (Szabolcsi 1997) The requirement on direct scope implies that syntax and semantics converge. Hoeksema (2000): NPI licensing does not require syntactic c-command it relies on semantic factors. What about French ni? 3. Licensing ni in clausal coordination Given the highly debatted status of French negation and negative concord, I assume the following basic assumption (cf. Godard 2004) : Negative expressions (quantifiers : personne nobody, rien nothing, aucun, nul no, adverbials jamais, never,nulle part nowhere, plus no longer, negative marker pas not ) - license a domain for de-np objects and de-np time adjuncts, and a ne-verb (18) Il n a pas/jamais bu de vin de sa vie He has not/never drunk of-wine of-his life - license a ne-verb : ne marks the syntactic scopal domain of the negation - dependancy ne is generally local : negation can t scope higher than the tensed ne-verb domain. a (strong) NPI must be c-commanded by ne-verb, by an N-expression or by both. 3 A. Dagnac Ellipsis2012 Vigo November 2012, Context (A): When ni conjoins two clauses, it must be c-commanded commanded by a negative licenser (19) Jamais Paul ne votera pour Romeny ni Marie ne s abstiendra Never Paul NE will vote for Romney nor Mary NE will abstain (20) Jamais Paul n est allé à Madrid ni Marie n est allée à Rome Never Paul NE is gone to Madrid nor Mary NE is gone to Rome (21) Nulle part Paul n a trouvé Lison ni Marie n a recueilli le moindre indice Nowhere Paul NE has found Lison nor Mary NE has collected the slightest piece of evidence (22) Pour rien au monde Jim n ira à Londres ni Marie ne restera à Rome For-nothing-in-the-world Jim NE will go to London nor Mary NE will stay in Rome Benincà & Poletto (2004) : preposed negative expressions stand in a functional projection inside the CP field ; alternatively : the negative adverbials are adjoined to. NEG c-commands ni. (23) Le problème s est réglé sans [que Jim n aille à Rome ni que Lea ne démissionne] The problem was solved without that Paul NE go-subj to Rome nor that Mary NE quit-subj NEG c-commands commands ni. The negation c-commands ni and scopes over the two s (whether it is merged in the left-periphery or moved ATB there is irrelevant here). Context (B): conversely, when the negation is inside 1, ni is ruled out : (24)*Paul ne votera pas/jamais pour Romney ni Marie ne s abstiendra Paul NE will not/never vote for Romney nor Mary NE will abstain (25)*Paul n est pas/jamais allé à Londres ni Marie n est allée à Rome Paul NE is not/never never gone to Madrid nor Mary NE is gone to Rome (26) *Paul n a trouvé Lison nulle part ni Marie n a recueilli le moindre indice For-nothing-in-the-world Jim NE will go to London nor Mary NE will stay in Rome (27)*Paul n ira pour rien au monde à Paris ni Marie ne restera à Rome Jim NE will go for-nothing-in-the-world to London nor Mary NE will stay in Rome 1 ni 2 1 ConjP ( NEG ) ( NEG ) ni 2 (A) (B) NEG does not c-command command ni : the sentence is ruled out Clausal ni must be in the syntactic scope of a negative licenser 4 Ni and gapping In contexts (A), as expected, gapping is ok : (28) Jamais Paul ne votera pour Romney ni Marie pour Obama Never Paul NE will vote for Romney nor Mary for Obama (29) Jamais Paul n est allé à Madrid ni Marie à Rome Never Paul NE is gone to Madrid nor Mary to Rome (30) Nulle part Paul n a trouvé Lison ni Marie le moindre indice Nowhere Paul NE has found Lison nor Mary the slightest piece of evidence (31) Pour rien au monde Jim n ira à Londres ni Marie à Rome For-nothing-in-the-world Jim NE will go to London nor Mary to Rome but in context (B), GAPPING IS OK TOO. (32) Paul ne votera pas/jamais pour Romney ni Marie pour Obama Paul NE will not/never vote for Romney nor Mary for Obama (33) Paul n est pas/jamais allé à Londres ni Marie à Rome Paul NE is not/never gone to Madrid nor Mary to Rome (34) Paul n a trouvé Lison nulle part ni Marie le moindre indice. Paul NE has found Lison nowhere nor Mary the slightest piece of evidence (35) Paul n ira pour rien au monde à Paris ni Marie à Rome Jim NE will go for-nothing-in-the-world to London nor Mary to Rome A. Consequences for the classical analysis of gapping An analysis by coordination plus deletion can t account for the contrast in (24-27) vs (32-35) : Take the grammatical (36) : (36) Marie n est pas allée à Londres ni Paul à Paris Mary is not gone to London nor Paul to Paris It should proceed from some version of the ungrammatical (37), with the bracketted part unpronounounced : (37) a. *Marie n est pas allée à Londres ni Paul n est allé à Paris b. *Marie n est pas allé à Londres ni Paul à Paris n est allé - in (pre-ellipsis) (37), ni is not properly licensed - since ni stands outside the unpronounced structure, there is no reason why the non-pronunciation of the vp should affect the licensing of ni in any way : no usual repair process, e.g. deletion of an illicit trace, can be at stake. gapped clauses headed by ni can t proceed from clause-coordination. B. Consequence for the «special construction» analysis In this framework, the gapped constituent does not have the syntax of a clause, but : - it is adjoined/conjoined to S (=) - it gets its clausal meaning through a special process. since it is adjoined to, ni is not licensed by a negation in 1 either. YET : [ni XP YP] is not syntactically a /CP when ni conjoins non-clausal constituents, it might be licensed differently (semantically in the spirit of Hoeksema 2000). this does not seem to be the case with non-gapped non-clausal constituents: none of the «semantically licensing contexts» reviewed by Hoeksema seems to license constituent-ni. 5 A. Dagnac Ellipsis2012 Vigo November 2012, E.g. Embedding of the NPI in the subject of some predicates among which passives, unaccusatives, and exist-type predicates. (38) [A doctor who knew anything to acupuncture] was not available/ didn't exist (39) *[Un médecin (qui connaisse l'acupuncture ni l'osteopathie)] n'était pas disponible A doctor who knew acupuncture nor osteopathy was not available [see Appendix for the full range of contexts and the special behavior of conjoined negative subjects] it is unclear to me how ni, being outside of the semantically reconstructed [XP YP], can enter the semantic reconstruction process. the gapped ConjP is held to be adjoined to S in these approaches. 2. Ni does not seem to be semantically licensed in other adjuncts to S unless they are to the right of V, where they are arguably adjoined to vp, or at any case within S : (40) *Avant le printemps ni (avant) l automne, Jean n ira pas à Rome Before spring nor (before) autumn, John will not go to Rome This option, though not unfeasible, needs further investigation and must be made more explicit. It probably requires for the [ni XP YP] constituent to be conjoin not to S but inside S C. Consequences for vp-coordination analysis If gapping relies on vp coordination, on the contrary, the facts above are predicted : the gapped structures in (32-35) do not proceed from - or CP- coordination, but from vp-coordination. In (41), based on Johnson s analysis, ni is c-commanded by a negative epression and a ne-v. (41) Paul n est pas allé à Paris ni Marie à Rome DP FP Paul NegCl ne T est Neg PredP pas/jamais VP PredP allé Pred vp When a negative quantifier is embedded in VP1, its correlate in VP2 must be either a negative quantifier or a NPI quantifier, just as with simple conjoined constituents 3 : (42) a. * Bunk n a trouvé aucun indice ni Kima une information [gapped constituent] Bunk NE has found no clue nor Kima a piece of information 2 Abeillé & al. (2011), though, rightly argue that adjunction wrongly predicts asymmetrical extraction from the first conjunct. 3 This holds for conjoined subjects, too. See appendix. vp1 ni vp2 t DP VP DP VP Marie t V PP V PP à Paris à Rome 7 A. Dagnac b. * Bunk n a trouvé aucun indice ni une information [simple conjunct] Bunk NE has found no clue nor a piece of information (43) a. Bunk n a trouvé aucun indice ni Kima aucune information [gapped constituent] Bunk NE has found no clue nor Kima no piece of information b. Bunk n a trouvé aucun indice ni aucune information [simple conjunct] Bunk NE has found no clue nor no piece of information (44) a. Bunk n a trouvé aucun indice ni Kima la moindre information [gapped constituent] Bunk NE has found no clue nor Kima the slightest piece of information b. Bunk n a trouvé aucun indice ni la moindre information [simple conjunct] Bunk NE has found no clue nor the slightest piece of information This suggests that while the c-command by a ne-verb licenses ni, ne must be symmetrically licensed by negative items in both conjuncts. Another potential arguments for vp-coordination : sans When ni conjoins clauses under sans, the repetition of the complementizer que is preferred (i.e. CPs, rather than s, are conjoined) : (45) a.?? Cela s est réglé sans que Paul aille à Paris ni Marie aille à Rome This was solved without that Paul go-subj to Paris nor Mary go-subj to Rome (46) b. Cela s est réglé sans que Paul aille à Paris ni que Marie aille à Rome This was solved without that Paul go-subj to Paris nor Mary go-subj to Rome When gapping is involved, on the contrary, the presence of que is banned : (47) a. Cela s est réglé sans que Paul aille (souvent) à Paris ni Marie (jamais) à Rome This was solved without that Paul go-subj (often) to Paris nor Marie (never) to Rome b. *Cela s est réglé sans que Paul aille (souvent) à Paris ni que Marie(jamais) à Rome This was solved without that Paul go-subj (often) to Paris nor that Mary (never) to Rome this is not due to a general ban against gapping in CPs : gapping in clauses headed by que is sometimes possible in French, in particular when the gapped constituent involves a polarity contrast : (48) Paul va souvent à Paris alors que Marie jamais Paul goes often to Paris whereas Mary never this may mirror the fact that only a vp is involved in ni-gapping the structure of gapping may depend on the conjunction involved (Repp 2009) among the problems raised by such analyses (cf. Johnson 1996 & sq for discussion), it is unclear whether the double coordination argument holds for ni. 4 (49)?? Ni Paul ne votera pour Romney ni Marie pour Obama Neither Paul will vote for Romney nor Mary for Obama Conclusion clausal-ni, and constituent-ni, must be in the scope of a negation. this is mirrored in overt syntax : ni must be c-commanded by a negation or sans. gapping in ni is possible even in the absence of a high ( ) negation 4 Abeillé & al. (2011) provide an example from a dictionary that I find very weird, or at least, far from natural. I found no such example in Frantext. (i) Ni le compromis ne me paraît justifié, ni l acceptation pure et simple nécessaire. (Dictionnaire du Français Contemporain) Lit. Neither a compromise NE seems to me justified, nor pure and simple acceptance necessary Ellipsis2012 Vigo November 2012, gapping in ni can t be analysed as a -level coordination. It argues for a vp-coordination analysis, where ni can be both in the semantic and overt syntactic scope of the clausal negation. this would besides be coherent with the availability of the various constructions conjoined by ni : while -coordination by ni is unfrequent in French (no case in a Frantext corpus of about 500 texts after 1965) 5, and rather marked, VP coordination by ni, as in (49), is common and unmarked just as gapped constructions in ni, in spite of the discourse constraints bearing over them (ratio = 1 gapped clause for 5 VP-coordination in the same corpus): (49) Il n'avait pas fait son devoir d'anglais, ni préparé les interrogations écrites. (G. Pérec, Je suis né) He hadn t done his English exercice, nor prepared the written assignments References Abeillé, Anne, Bilbiie, Gabriella, & François Mouret. (2011). A Romance Perspective on Gapping Constructions. In H. Boas & F. Gonzalvez Garcia (eds) to appear, Romance in Construction Grammar. Constructional Approaches to Language series. John Benjamins Publishing Company. Abeillé, Anne & François, Mouret. (2010). Quelques contraintes sur les coordinations elliptiques en français. Revue de sémantique et de pragmatique 24, Benincà, Paola & Cecilia Poletto (2004). Topic, Focus and V2: Defining the CP sublayers. In Rizzi L. (ed.), The structure of CP and IP, The Cartography of Syntactic Structures, Vol. 2, New York, Oxford University Press, Borsley, Robert. (2005a). Against ConjP, Lingua : Borsley, Robert. (2005b). Les coordinations relèvent-elles de la syntaxe X-barre? Langages, 160 : Bošković, Željko, and St
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks