Rutgers University - LSRL 36

Linguistic Symposium on Romance Languages (LSRL 36)

ABSTRACTS

Rutgers University New Brunswick, New Jersey March 31-April 2 Organizing committee: José Camacho, Nydia Flores, Viviane Déprez and Liliana Sánchez With the support of:

The Vice President for Academic Affairs; the Dean of Rutgers College, Department of Spanish and Portuguese, Linguistics Department and

Welcome to the 36th edition of the Linguistic Symposium on Romance Languages! We would like to thank the many people have contributed to the organization of this conference, among them Dan Althshuler, Dan Zachary, Carlos Fasola, Slavica Kochovska and Jessica Rett. María José Cabrera and Jennifer Flaherty carried much of the load of the organization and Ezra Van Everbroeck, from UCSD, put in many hours developing and worked out the glitches in the abstract submission and review system (pasha), locally maintained by John Amodeo. This year, we greatly expanded the number of abstract reviewers, who we would like to thank for their thorough, professional and timely work: Aafke Hulk, Alessandra Giorgi, Ana Perez-Leroux, Ana Roca, Andrea Calabrese, Anna Cardinaletti, Anna Gavarró, Anne-Marie DiSciullo, Antonella Sorace, Aurora Bel, Barbara Bullock, Barbara Vance, Bernard Tranel, Bruce Tesar, Carlos-Eduardo Piñeros, Carlos Piera, Carmen Dobrovie-Sorin, Carmen Silva-Corvalán, Cecilia Poletto, Christina Tortora, Claudida Parodi, Contxita Lleo, Cristina Schmitt, David Heap, Diane Massam, Dominique Sportiche, Donca Steriade, Elena Benedicto, Enric Vallduvi, Esthela Treviño, Esther Torrego, Eulalia Bonet, Fernando Martinez-Gil, Francisco Ocampo, Francisco Ordóñez, Gemma Rigau, Gillian Sankoff, Giuseppe Longobardi, Gorka Elordieta, Grant Goodall, Jane Grimshaw, Javier Gutierrez-Rexach, Haike Jacobs, Heles Contreras, Henrietta Cedergren, J. Clancy Clements J-C. Smith J.M. Brucart, Jacqueline Toribio, Jairo Nunes, Jean-Pierre Montreuil, Jennifer Austin, Joan Mascaró, Johanne Paradis, John Lipski, Jon Franco, José Alberto Elias, José Camacho, Jose-Ignacio Hualde, Juana M. Liceras, Judy Bernstein, Julia Herschensohn, Julie Auger, Karen Zagona, Jurgen Klausenburger, Laurent Dekydtspotter, Liliana Paredes, Liliana Sánchez, Lisa Reed, , Lori Repetti, Louise McNally, Luis López, Lydia White, Magdalena Romera, Manuel Español-Echevarria Marc Authier, Marcel den Dikken, Mark Baker, Mary Kato, Marcello Modesto, Maria-Luisa Zubizarreta, María Luisa Hernanz, Marie Labelle, Marie-Therese Vinet, Mario Montalbetti, Michel DeGraff, Mónica Cabrera, Naomi Nagy, Nina Hyams, Nuria Sagarra, Nydia Flores, Paola Dussias, Pascual Masullo, Paul Hirschbühler Paula Kempchinsky, Rafael Nunez-Cedeno, Rafaella Zanuttini, Randall Gess, Richard Cameron, Richard Morris, Rodrigo Gutierrez Bravo, Roger Schwarzschild, Shana Poplack, Silvina Montrul Sonia Colina, Sonia Frota, Timothy Face, Veneeta Dayal, Virginia Motapanyane. Viviane Déprez, Yves Roberge Finally, our thanks to our financial sponsors: the Vice President for Academic Affairs, the Dean of Rutgers College, the Department of Spanish and Portuguese, the Linguistics Department, and the Instituto Cervantes, New York, whose contribution made possible prof. Vallduví’s invitation.

José Camacho, Viviane Déprez, Nydia Flores and Liliana Sánchez The Organizing Committee

FRIDAY MORNING (Rutgers Student Center MPR) 8:15-8:45 8:45-9:00

Registration/coffee Opening 1. Syntax (Rutgers Student Center MPR A)

The Syntactic Operator se in Spanish: A Contemporary Account

Interpretational deficit: evidence from the future tense in L2 Spanish

Eva Juarros (University of New York at Buffalo)

Elena Valenzuela, Joyce Bruhn de Garavito (University of Western Ontario)

Parenthetical null topic constructions

Processing of Grammatical Gender and Agreement in L2 Spanish

Daniel Altshuler and Viviane Déprez (Rutgers University)

Nuria Sagarra (Pennsylvania State University)

Topicalization in European and Brazilian Portuguese

Inversion in wh-questions in child Romance and child English: The roles of syntax and processing

9:00-10:30

10:30-11:00

11:00-12:30

12:30-1:45

2. Parasession on biling./ SLA (Rutgers Student Center MPR B)

Mary Kato (Univ. Estadual de Campinas, UNICAMP) and Eduardo Raposo (UC, Santa Barbara) Coffee 3. Sociolinguistics (Rutgers Student Center MPR A)

Grant Goodall (University of California, San Diego)

4. Parasession on biling./ SLA (syntax) (Rutgers Stud. Ctr MPR B)

A Variationist Approach to Verb Types and Subject Position in Spanish.

Specificity, learners know: On the acquisition of clitic drop in L2 Spanish

Roberto Mayoral Hernández (University of Southern California)

Joyce Bruhn-Garavito*, Claudia Borgonovo**, Pedro Guijarro-Fuentes***, Philippe Prévost**, Elena Valenzuela* (*U. Western Ontario, **U. Laval, ***U. of Plymouth)

Sociolinguistic Factors Conditioning the ordering of adverbial expressions in Spanish

When learners know more than linguists: direct object clitics are not pronouns

Roberto Mayoral Hernández and Asier Alcázar (University of Southern California)

Theres Grüter (McGill U.)

Performative verbs in Spanish monolingual and bilingual colonial court documents

Developing L2 morphosyntax in TP and DP: A case study of French

Anna María Escobar (University of Illinois, UrbanaChampaign) Lunch

Julia Herschensohn (U. of Washington) and Deborah Arteaga (U. of Nevada, Las Vegas)

FRIDAY AFTERNOON (Rutgers Student Center MPR) Context and grammar: Catalan rai Enric Vallduví (Universitat Pompeu Fabra)

1:45-2:45 2:45-3:00

Coffee 5. Syntax (Rutgers Student Center MPR A)

3:00-4:30

4:30-4:45

4:45-6:15

6. Semantics/Syntax (Rutgers Stud. Ctr MPR B)

Expletives and the Rise of Number

How do Usted in Spanish and Lei in Italian differ?

Monique Dufresne (Queen’s University)

Hyun-Jong Hahm (University of Texas, Austin)

Grammar vs usage in the study of linguistic change

Macro event types and aspect shift in Spanish

Nathalie Dion and Shana Poplack (University of Ottawa)

Joshua Rodríguez (Northern Illinois University)

Old French Quirkies

Tacit verbs and semi-direct speech in Iberian Spanish

Eric Mathieu (University of Ottawa) Break 7. Syntax/Semantics (Rutgers Student Center MPR A)

Ricardo Etxepare (CNRS) 8. Phonology (Rutgers Stud. Ctr MPR B)

Structure and Interpretation of Degree Correlatives

Rhythmic constraints on the distribution of schwa in French

Javier Gutiérrez-Rexach (Ohio State University)

Marie-Hélène Côté (Université d'Ottawa)

Discontinuous Wh-questions in Brazilian Portuguese.

Accent without stress in Spanish

Ana C.P. Bastos (University of Connecticut)

Marta Ortega-Llebaria and Stephen Meiners (University of Texas, Austin)

Syntax and Semantics of Split Questions

Karlos Arregi (University of Illinois Urbana-Champaign)

Investigating phrasing levels in French : Phonology and phonetics of prenuclear and nuclear accents Mariapaola D’Imperio*, Roxane Bertrand*, Albert Di Cristo** and Cristel Portes** (*Laboratoire Parole et Langage **Université de Provence & Laboratoire Parole et Langage)

SATURDAY MORNING (Murray Hall) 8:30-9:00

Registration 9. Syntax (Murray 211)

10. Parasession on biling./ SLA (Phonology) (Murray 204)

The classification of Spanish prepositions: Evidence from Media Lengua 9:00-10:00

Suzanne Dikker (New York University)

Aspect and the internal structure of Romance PLACE PPs

10:00-10:15

Christina Tortora (College of Staten Island) Coffee 12. Syntax (Murray 211)

Compositional Telicity Prepositional System

10:15-11:15

in

Asymmetric phonetic convergence in Attributive superlatives in bilingual speech Romanian Barbara E. Bullock, Amanda Dalola, Verónica González, Almeida Jacqueline Toribio (Pennsylvania State University)

Alexandra Teodorescu (University of Texas, Austin/McGill University)

The Acquisition of Spanish Onsets

The double negation puzzle in a strict negative concord language

Rafael Núñez-Cedeño (University of Illinois, Chicago)

Anamaria Falaus (Laboratoire de Linguistique de Nantes)

13. Phonology (Murray 204)

the Perceptual Properties of Palatalization in Romanian

Heather Burnett*, Katrina Petrik** and Mireille Tremblay* (*Queen’s University, **U. of Ottawa)

Laura Spinu (University of Delaware)

On the use of the definite article with prepositions in Romanian

Voicing assimilation in Ecuadoran Spanish: evidence for stratal OT

Alexandru Mardale (LLF, Paris 7 & U. of Sonia Colina (Arizona State University) Bucharest)

14. Syntax (Murray 212)

Sub-extraction from subjects Ángel J. Gallego (Univ. Autònoma de Barcelona) and Juan Uriagereka (Univ. of Maryland and U. del País Vasco)

A-movement, dislocated subjects and the syntax of the left periphery in Spanish Omar Beas (Univ. of Southern California)

Lexical conservatism and the encoding of declension class membership Donca Steriade (M.I.T) Murray, 211

11:15-12:15

12:15-1:45 1:15-1:45

11. Semantics/ Syntax (Murray 212)

Lunch Business meeting

SATURDAY AFTERNOON (Murray Hall) 15. Syntax (Murray 211)

1:45-2:45

2:45-3:15

3:15-4:15

4:15-4:45

17. Phonology (Murray 212)

Unaccausative Manner-of-Motion Verbs in Italian: A Compositional Analysis

Evidence for the continuum from Phonetic correlates of stress and narrow to peripheral syntax among accent in Catalan and Castilian bilinguals and monolinguals Spanish

María Luisa Zubizarreta (University of Southern California)

Gabriela Zapata (U. of Alberta), Almeida Jacqueline Toribio , Ana de Prada Pérez, and Eva-María SuárezBudenbender (Penn State University)

Pilar Prieto (Universitat Autònoma de Barcelona & ICREA) and Marta OrtegaLlebaria (University of Texas, Austin)

Past Participle Agreement in Abruzzese: Split Auxiliary Selection and the Null-subject Parameter

Verb movement – patterns of acquisition and triggering experience in EP and BP

On Majorcan Catalan pre-nuclear pitch accents

Fernanda Maria Ribeiro Gonçalves (Universidade de Evora)

Miquel Simonet (University of Illinois, Urbana-Champaign)

Roberta D’Alessandro and Ian Roberts (University of Cambridge) Coffee 18. Syntax (Murray 211)

19. Parasession (Syntax) (Murray 204)

20. Phonology (Murray 212)

Agree and Existential Constructions in Romance

Incomplete acquisition of prepositional Why neither Sympathy Theory nor relative clauses in Spanish-English Comparative Markedness are able early bilinguals to handle prosodic opacity: the case of Latin vowel deletion

Erik Schoorlemmer (LUCL, Leiden University)

Silvia Perpiñán (University of Illinois at Urbana-Champaign)

Verb Movement and PhaseSliding

Topicalizations and object drop in child language

Ángel J. Gallego (Universitat Autònoma de Barcelona)

Ana Teresa Pérez-Leroux, Mihaela Pirvulescu, Yves Roberge (Univ. of Toronto) Fernando Martínez-Gil (Ohio State U.)

Haike Jacobs (Radboud Univ. Nijmegen)

What consonant vocalization tells us about the syllabic affiliation of muta-cum-liquida clusters in ProtoSpanish and Western Romance

Break

Deviance and convergence in early child bilingualism Aafke Hulk (University of Amsterdam/ ACLC) Murray Hall, 211

5:00-6:00 7:00-9:00

16. Parasession (Syntax) (Murray 204)

Dinner, Zimmerli Art Museum

SUNDAY MORNING (Rutgers Student Center) 21. Syntax (Rutgers Student Center, 411ABC)

22. Phonology (main and parasession) Rutgers Student Center, 410AB

The role of Number within nominal arguments: the case of French pluralized Event Nominalizations

Excluding Extrametrical elements

Jasper Roodenburg (NWO / University of Venice)

José Sueli de Magalhães (Universidade Federal de Uberlândia)

Romanian- A PCC Language?

Integrating articulatory markedness into a theory of L2 phonological development

Oana Savescu Ciucivara (New York University)

Laura Colantoni and Jeffrey Steele (University of Toronto)

9:15-10:15

10:1511:45

Coffee

Subjects and Wh-Questions 10:4511:45 11:4512:00

Anna Cardenaletti (University of Venice) Rutgers Student Center, 411ABC Closing remarks

Friday, session 1

Parenthetical null topic constructions in Romance Daniel Altshuler and Viviane Déprez Rutgers University [email protected]; [email protected] The aim of this talk is to argue that there are null topic constructions in Romance which resemble the overt topic constructions of Clitic Left Dislocation (CLLD). These null topics are found in parenthetical constructions (PNT), c.f. (1), and manifest the following properties: (i) (ii) (iii) (iv)

A clitic necessarily resumes the null topic. The null topic can occur at the front of a subordinate clause. There is no theoretical limit to the number of null topics. The main clause subject that binds the null topic must be understood as definite or specific indefinite.

The properties in (i)-(iv) have been argued to be unique to CLLD (Cinque 1990, Zubizarreta 1998). Thus, in contrast to Stowell (2005), which analyzes the English counterpart of the parenthetical in (1) as a restrictive relative clause without an overt complementizer, we argue that this analysis makes the wrong predictions for Romance. We propose that the parenthetical clause has the structure of CLLD with a null topic in TopicP (in Rizzi’s (1997) exploded CP). Unlike CLLD, however, PNT manifest additional locality and contextual restrictions: (a) they must be local to the element their gap refers to in the clause they are dependent on and (b) they must occur in the context of a parenthetical adverbial or a particular speech/attitude report. To account for these additional properties, we argue that the null topic in PNT is subject to two conditions of recuperation: (a) there must be no other element intervening between it and the overt element that it is bound to (see (2)-(3)) and (b) the TopicP projection in which the null topic resides must be c-commanded by or be in the specifier of an EvidentialP projection (see (4)-(8)). Following Rooryck (2001) we take parenthetical adverbs to be evidentials; they head their own projection in the exploded CP structure. Following Simons (2005) we take speech/attitude verbs (e.g. “say/think”) as carrying evidential force; they license a covert element in the head of EvidentialP. There are (at least) four reasons to believe that PNT are not relative clauses: (i) (ii) (iii) (iv)

Negative quantifier phrases like “No student” can head a relative clause, but they are not possible in PNT (see (9)). An overt complementizer is impossible in PNT (see (10)). Extraposition is impossible. Relative clauses with a null C are not found in (main) Romance.

Moreover, as (11) illustrates, CLLD cannot apply to bare quantifier phrases (BQPs), i.e. BQPs never reside in TopicP. Both (12) and (13) contain a BQP but (13) differs from (12) in that there is a parenthetical after “algo.” Our proposed analysis explains the ungrammaticality of (13) as follows. There is a null Topic in the parenthetical clause analogous to an overt topic in CLLD and like in CLLD, the PNT topic cannot be a BQP.

1

A relative clause analysis could not explain this contrast since BQPs can head a relative clause. Data (1)

Algunos libros (Juan los leyó ayer, supuestamente) intrigan al prof. Alvarez. Some books (John CL read yesterday, reportedly) intigue prof. Alvarez.

(2)

Maria, a un hombrei (TOPi Bill loi conocio en Kansas, supuestamente) loi odia con pasión. Mary, a man (Bill CL met in Kansas, reportedly) CL hates with passion.

(3)

*A un hombrei, Maria (TOPi Bill loi conocio en Kansas, supuestamente) loi odia con pasión. A man, Mary (Bill CL met in Kansas, reportedly) hates CL with passion.

(4) (5) (6) (7) (8)

Algunos libros (Juan los leyó ayer, supuestamente) intrigan al prof. Alvarez. Algunos libros (supuestamente, Juan los leyó ayer) intrigan al prof. Alvarez. Supuestamente, algunos libros (Juan los leyó ayer) intrigan al prof. Alvarez. *Algunos libros (Juan los leyó ayer) intrigan al prof. Alvarez, supuestamente. *Algunos libros (Juan los leyó ayer) intrigan al prof. Alvarez.

(9)

*Ningún estudiante (Bill lo conocio en Kansas, supuestamente) se ha mudado al lado. No student (Bill CL met in Kansas, reportedly) CL has moved in next door.

(10)

*Algunos libros (a que Juan los leyó ayer, supuestamente) intrigan al prof. Alvarez. Some books (that John CL read yesterday, reportedly) intigue prof. Alvarez.

(11)

*Algo, Juan lo leyó ayer. Something, John CL read yesterday.

(12)

Algo intriga al prof. Alvarez. Something intigues prof. Alvarez.

(13)

*Algo (Juan lo leyó ayer, supuestamente) intriga al prof. Alvarez. Something (John read CL yesterday, reportedly) intigues prof. Alvarez.

References CINQUE, G., 1990. Types of A bar-dependencies. LI Monograph 17. MIT Press. RIZZI, L. 1997. The Fine Structure of the Left Periphery. In Liliane Haegeman, ed, Elements of Grammar. Kluwer, Dordrecht, pp. 281-337. VAN ROORYCK, J. 2001. Evidentiality Part I, Glot International 5: 3-11. SIMONS, M. 2005. Observations on embedding verbs, evidentiality, and presupposition, Ms., Carnegie-Mellon University. STOWELL, T. 2005. Appositive and parenthetical relative clauses, Ms., UCLA. ZUBIZARRETA, M.L. 1998. Prosody, Focus and Word Order. MIT Press.

2

Karlos Arregi (University of Illinois, Urbana-Champaign)

Friday, session 7

Syntax and Semantics of Split Questions Split questions in Spanish contain a wh-question part (up to the comma in (1)) and a tag (after the comma in (1); see Camacho 2002). (1)

¿Qué libro leyó Juan, El Quijote? what book read Juan, El Quijote ‘What book did Juan read, El Quijote?’

Split questions are interpreted as non-wh-questions, despite the fact that the first part has the syntax of a wh-question (Camacho 2002; see below for details.) I argue that split questions are the result of ellipsis; e.g. (1) has the same syntax and semantics as the following: (2)

¿Qué libro leyó Juan? ¿Leyó El Quijote? what book read Juan read El Quijote ‘What book did Juan read? Did he read El Quijote?’

The proposal has two ingredients, discussed below in more detail: (i) the wh-question part has the syntax of a wh-question, and (ii) the tag is a separate non-wh-question, where everything but the tag undergoes ellipsis. (Ellipsis is licensed in the same way as in fragment answers; see, a.o. Merchant 2004) I compare this analysis with Camacho’s (2002), where the tag forms a constituent with the wh-phrase and is stranded in its base position after wh-movement: (3)

[CP [DP1 what book ]i read Juan [DP1 ti [DP2 El Quijote ]] ]

1. Semantics. In the present analysis, the semantics of split questions is straightforward. The overall question type of the split question is determined by the tag, just as it is in the non-elided source: if the source of the tag is a yes/no-question, so is the split question. For instance, both (1) and its non-elided source (2) behave as a yes/no-question as a whole: both must be answered by yes or no; in both cases, answering only the wh-question part (e.g. with He read El Quijote/La Celestina) is not felicitous. If the tag contains a disjunction, the split question can be understood as an alternative question: (4)

¿Qué libro leyó Juan, El Quijote o La Celestina? what book read Juan El Quijote or La Celestina ‘What book did Juan read, El Quijote or La Celestina?’

In this case, the non-elided source of the tag is the alternative question Did he read El Quijote or La Celestina?. As expected, the split question must be answered in the same way as its non-elided source: El Quijote, La Celestina, or whatever other appropriate answer there is, but, crucially, not with yes or no. It is not clear how this semantics is accounted for in Camacho 2002. In particular, no account is given as to how the denotation of DP1 in (3) is obtained compositionally from the denotations of ti (or its wh-phrase antecedent) and DP2 (the tag). In the present analysis, the trace and the tag do not form a constituent, so the question does not arise. Furthermore, no account is given in Camacho 2002 as to why the split question can be interpreted as a yes/no or alternative question. 2. Syntax. As shown in Camacho 2002, the wh-question part involves wh-movement, as witnessed by the fact that extraction from a complement CP is grammatical, but extraction from an island is not: 1

(5)

a. ¿Qué libroi crees [ que compró Juan ti ], El Quijote? what booki you.think [ that bought Juan ti ] El Quijote b. *¿Qué libroi conoces [ al editor que publicó ti ], El Quijote? what booki you.know [ to.the editor that published ti ] El Quijote

Furthermore, the movement is subject to weak and strong crossover: (6)

*¿A quiéni cree { sui madre / eli } que quieres, a Juan? to who thinks { hisi mother / hei } that you.love to Juan

Camacho also argues that the wh-word is not a [+wh] operator, since, apparently, split questions cannot be embedded: (7)

*Se REFL

preguntan qué compraste, { ¿un libro? / si un libro } they.ask what you.bought { a book / whether a book }

Camacho argues the wh-phrase cannot satisfy the selectional requirements of the embedding verb ask. Although, at present, the ungrammaticality of this type of sentence is not understood, this explanation is not valid; if the tag is understood as an (embedded) alternative question, the embedded split question is grammatical, which shows that the wh-phrase can satisfy the relevant selectional requirement: (8)

Se

preguntan qué compraste, si un libro o una revista. REFL they.ask what you.bought whether a book or a magazine

To conclude, as in other wh-questions, the wh-question part in a split question is a normal whphrase. (Camacho also provides a theory-internal argument from Condition C that does not apply to the present analysis.) 3. The position of the tag. Since the wh-question and the tag are separate sentences, we expect the latter to always follow the former, even in cases where the trace of the wh-phrase is not final in the wh-question: (9)

¿A quién dijiste ti que estabas enfermo, a Juan? to who you.told ti that you.were sick to Juan

Camacho 2002 wrongly predicts that it should be at least possible to have the tag inside the whquestion (in the position of the trace): (10)

*¿A quién dijiste ti a Juan que estabas enfermo? to who you.told ti to Juan that you.were sick

A related point is that, in multiple questions, the whole tag must follow the wh-question; no part of it can be in the position of the in situ wh-phrase or the trace of the moved wh-phrase: (11)

a. ¿Quién vio a quién ayer, Juan a María? who saw to who yesterday Juan to María b. *¿Quiéni vio ti Juan a quién a María ayer?

4. Conclusion. The ellipsis analysis proposed here explains several semantic and syntactic properties of split questions: (i) they are interpreted as non-wh-questions, (ii) the wh-question part involves wh-movement, and (iii) the tag must follow the wh-question part. 2

Friday, session 7

Discontinuous Wh-constituents in Brazilian Portuguese Ana Bastos (University of Connecticut) Introduction: I study discontinuous wh-constituents in Brazilian Portuguese (BP) as in (1a), contrasting them with regular cases of Wh-fronting, in situ in non-echo questions (Mioto 1999, Grolla 2000 and Zocca 2004) and low inversion in (1b). Generalization: discontinuous Wh-questions are possible if and only if the PP that is left behind, for instance, [of-this picture] in (1a), is potentially extractable out of the DP. Data: This paper contributes with a description of the possibilities of extraction of PPs in the DP domain in BP comparing them to Spanish (Ticio 2003). I discussed two cases of extraction: a. when there is a single PP in the DP, only PPs headed by the preposition de ‘of’ are extractable out of the DP, and b. when there are multiple argumental PPs modifying the same noun, only the hierarchically higher PP can move out of the DP. By the application of tests with binding and quantifiers, I conclude that the hierarchy among PPs modifying a same noun in BP is the same found in Spanish: possessive > agent > theme. Given this, I extend to BP the analysis of Spanish DP by Ormazabal (1991) and Ticio (2003) that argumental PPs are right specifiers in the DP domain. These patterns of extraction of PPs show a full correlation with the patterns found in discontinuous Wh-questions, supporting the generalization above, i.e. the PP that is left behind in discontinuous Wh-questions has to be headed by the preposition de and it has to be hierarchically higher than any other PP in the same DP (cf. (2)-(3)). Analysis: I analyze discontinuous Wh-questions as remnant movement constructions, against the alternative possibility of left-branch extraction (LBE). I argue that these constructions in BP are a result of two independent steps of movement, as given in (4). The first step is the left movement of the PP out of the DP. The second step of movement is just a regular uncontroversial movement of a DP [+wh] to the left periphery. Against LBE: the first argument against an alternative analysis in terms of LBE comes from the fact the sequence WH-N in [DP WH N PP] does not form a constituent. For instance, the pronominalization test for constituency shows that the sequence WH-N can not be replaced by a pronoun (Cf. ex. (5)). Given that the sequence WH-N does not form a constituent, it can not move up as a unit leaving the PP behind. A second argument comes from the fact that standard cases of LBE discussed in the literature (Ross 1986, Bošković 2005) are not allowed in BP as shown in (6), which suggests that LBE is not available in BP. A third argument comes from the correlation discussed above between the potential of extracting a PP out of the DP and the acceptability of discontinuous Wh-questions. Under the LBE analysis, the PP never moves, therefore their potential of being extracted should be irrelevant for discontinuous Wh-questions. Under the remnant movement analysis, the answer comes straightforwardly: discontinuous Whquestions involve the movement of the PP out of the DP. Arguments for PP-movement: I analyze constructions with low inversion like (1b), in which the PP appears in front of the sequence WH-N, as evidence for the first step of movement represented in (4), i.e. (1b) exemplifies a case in that the PP moves out of the DP, but Whmovement does not apply. Note that BP allows Wh-in-situ constructions as full questions (Zocca 2004). Low inversion fully correlates with discontinuous Wh-questions in the patterns of extraction discussed above (Cf. (7)-(8)). The claim that the movement of PP is to the left is supported by tests with interaction of scope adverbs. Conclusion: Discontinuous Wh-questions in BP are better analyzed as cases of remnant movement, given their correlation with the patterns of extraction of PPs out of DPs in BP and the fact that the sequence WH-N does not form a constituent.

Examples: (1) a. Quantas reproduções o João destruiu, [dessa pintura]? How-many reproductions the John destroyed [of-this picture] b. O João destruiu [dessa pintura] quantas reproduções? the John destroyed [of-this picture] how-many reproductions Both: ‘How many reproductions of this picture did John destroy?’ (2) Discontinuous Wh-questions: a single PP Que livro o João destruiu [de Físicatheme ] / [*sobre Físicatheme]? which book the John destroyed [of Physicstheme] / [*about Physicstheme] ‘What is the topic such that John destroyed a book about it?’ (3) Discontinuous Wh-questions: multiple PPs: possessive > agent a * [Que livro tag [do João]poss] ela destruiu [de Albert Einstein]ag? [which book tag [of John]poss] she destroyed [by Albert Einstein]ag tposs] ela destruiu [do João]poss? b. [Que livro [de Albert Einstein]ag [which book [by Albert Einstein]ag tposs] she destroyed [of John]poss Both: ‘Which book by Albert Einstein of John’s did she destroy?’ (4) Step 1: [CP … [V PP [DP WH N tPP]] Step 2: [CP [DP WH N tPP] … [V PP tDP ] (5) O João destruiu [quantas reproduções / *elas dessa pintura]? the John destroyed [how-many reproductions / *they of this picture] ‘How many reproductions of this picture did John destroy?’ (6) a.*Quanto o João ganhou dinheiro? b. *Quantos o João conhece meninos? How-much the John earned money? How-many the John knows boys? ‘How much money did John earn?’ ‘How many boys does John know?’ (7) Low inversion: a single PP O João destruiu [de Físicatheme ] / [*sobre Físicatheme] que livro? The John destroyed [of Physicstheme] / [*about Physicstheme] which book ‘What is the topic such that John destroyed a book about it?’ (8) Low inversion: multiple PPs: possessive > agent a *Ela destruiu [de Albert Einstein]ag [que livro tag [do João]poss]? she destroyed [by Albert Einstein]ag [which book tag [of John]poss] b. Ela destruiu [do João]poss [que livro [de Albert Einstein]ag tposs] ? She destroyed [of John]poss [which book [by Albert Einstein]ag tposs] Both: ‘Which book by Albert Einstein of John’s did she destroy?’ References: Bošković, Željko. 2005. On the locality of left branch extraction. Studia Linguistica 59: 1-45. Grolla, E. B. 2000. A Aquisição da Periferia Esquerda da Sentença em Português Brasileiro. Dissertação de Mestrado, IEL/Unicamp, Campinas, SP. Mioto, Carlos. 2001. Sobre o sistema CP no português Brasileiro. Curitiba, Revista Letras, 56:97-140. Ticio, Emma. 2003. On the structure of DPs. Ms. Doctoral Dissertation, Uconn. Zocca, Cynthia. 2004. And the Wh- goes where? Ms. Uconn. (Term paper). Ross, J. R. 1967/1986. Infinite syntax. Norwood: Ablex Publishing.

Omar Beas (University of Southern California)

Saturday, session 14

A-MOVEMENT, DISLOCATED SUBJECTS AND THE SYNTAX OF THE LEFT PERIPHERY IN SPANISH One of the most controversial issues in Spanish syntax is the characterization of pre-verbal subjects. Under the standard analysis in the 90’s, pre-verbal subjects are described as leftdislocated or Clitic Left Dislocations (CLLD) that are generated as adjuncts to IP or CP and are resumed by subject agreement or pro (Olarrea, 1996; Ordóñez, 1997; Ordóñez & Treviño, 1999). From this point of view, Spanish pre-verbal subjects are non-configurational in parallel to Baker’s analysis of Mohawk (Baker, 1996) and Alexiadou & Anagnostopoulou (1998)’s proposal for Greek. In this paper, I show that this approach is incorrect for two reasons. (i) A sequence SVO is ambiguous with respect to the subject: it can be dislocated or non-dislocated. (ii) Two types of dislocated subjects with different properties must be distinguished. Given the logic of movement as a last resort, Alexiadou & Anagnostopoulou (1998)’s influential work on the syntax of subjects in pro-drop languages that display SVO/VSO alternation predicts that if verb movement in these languages checks the EPP feature, this language must resort to left dislocation to express its pre-verbal subjects. Despite the fact that Spanish fulfills the above mentioned requirements as a pro-drop language, I show that it does not respond uniformly to the tests Alexiadou & Anagnostopoulou suggest are indicators of the leftdislocated nature of subjects. The strongest piece of evidence to support this affirmation comes from the scope interactions between pre-verbal subjects and QP objects, modals and negation. On the one hand, the scope interaction between an indefinite pre-verbal subject and a QP object results in ambiguous interpretations (example 0), as does the relation between a quantificational subject and a modal (example (2)), unexpected outcomes under a left dislocated analysis. On the other hand, the obligatory wide scope interpretation of pre-verbal subjects with respect to assertion particles (affirmation/emphasis, negation and reportatives) suggests that they have properties associated with dislocation (example (3)). Therefore, one must discriminate between a dislocated pattern and a non-dislocated pattern. Further examination of the left periphery in Spanish uncovers an additional distinction in the dislocated pattern of subjects. Despite the fact they share a crucial cluster of properties with CLLD constructions (selective island sensitivity (example (4)) and absence of weak crossover effects (example (5)), subject dislocation also differs in displaying connectivity effects in idiom chunks, which are absent in CLLD (example (6)a versus (6)b). As a result, two types of dislocated subjects must be distinguished: one which is Ā-related and another which is non-Ārelated. The former intercepts Ā-movement operations, does not reconstruct, and does not license parasitic gaps whereas the latter does not intercept Ā-dependencies, does reconstruct, and does license parasitic gaps. Following Zubizarreta (1994), I analyze the non-dislocated pattern as generated by regular Amovement. In addition, the two dislocated patterns correspond to two separate analyses. The Ārelated dislocated subject is base generated a la Cinque above the projection that hosts wh- and focused constituents (Cinque, 1990). The non-Ā-related dislocated subject originates within VP as part of a Big DP headed by the resumptive pro, undergoing movement to the left periphery to a position below the functional projection containing wh- questions and focus constituents. Examples algún + every-NP AMBIGUOUS a. En este país algún lingüista estudió cada lengua nativa exhaustivamente In this country, some linguist studied every native language exhaustively “In this country, some linguist studied every native language exhaustively” b. un + every-NP AMBIGUOUS Finalmente, un lingüista ganó cada beca en su universidad finally a linguist won each scholarship in his university “Finally, a linguist won every scholarship in his university” (1)

(2)

Indefinites and modals Un lingüista debe leer libros de texto AMBIGUOUS a linguist must read books of text “A linguist must read textbooks”

Indefinites and negation a. Muchos lingüistas no leen libros de texto NOT AMBIGUOUS: *no > muchos; muchos > no Many linguists not read books of text “Many linguists do not read textbooks” b. Esta noche, un policía no vigiló cada banco NOT AMBIGUOUS: un > cada; *cada > un this evening, a policeman not guarded every bank “This evening, a policeman did not guard every bank” (3)

Selective Island Sensitivity a. Strong islands: Relative Clause Island *Dices que esa mujeri conociste al sintactista que proi quiere enamorar (you)-say that that woman met-you the syntactician that pro wants to enamor “You say that this woman you met the syntactician that she wants to enamor” (intended) b. Weak islands: wh-Island Me dices que esta chiquillai, te preguntabas cuándo proi se compró un auto como este. you me-tell that this little girl (you) you-wondered when pro self-bought a car like this. “You said to me that this little girl you were wondering when she bought herself a car like this” (intended) (4)

(5)

¿Quiéni a sui hermana sí (que) la vió? who ACC-his sister yes-(that) (pro) her-saw “Who did see her sister?”

SV idioms and assertion markers a. Hierba mala (sí o sí, sí que) nunca muere herb bad (yes or yes, yes-that) never dies “Bad seed never dies” (idiomatic) “Weed never dies” (literal) b. La toalla la tiró Cristina the towel it-throw Cristina “Cristina threw in the towel” (*idiomatic) “Cristina threw the towel” (literal) (6)

References ALEXIADOU, Artemis & Elena ANAGNOSTOPOULOU (1998) “Parametrizing AGR: Word order, Verb-movement and EPP-Checking”. NLLT, 16.3: 491-539. BAKER, Mark (1996) The Polysynthesis Parameter. Oxford University Press, New York. CINQUE Guglielmo (1990) Types of A-bar Dependencies, MIT Press. OLARREA, Antxon (1996). Pre and Postverbal Subjects in Spanish: A Minimalist Account. Unpublished Ph.D. dissertation, University of Washington. ORDÓÑEZ, Francisco (1997). Word Order and Clausal Structure in Spanish and Other Romance Languages. PhD dissertation, CUNY, Graduate Center. ORDÓÑEZ, Francisco & Esthela Treviño (1999). “Left Dislocated Subjects and the Pro-drop Parameter: A Case Study of Spanish”. In: Lingua 107, pp. 39-68. ZUBIZARRETA, María Luisa., (1994). "El orden de palabras en español y el caso nominativo". In: Nueva Revista de Filología Hispánica VI (ed. by Violeta Demonte) Colegio de México.

Friday, session 4 Joyce Bruhn de Garavito *, Claudia Borgonovo**, Pedro Guijarro-Fuentes***, Philippe Prévost**, Elena Valenzuela* (*U. Western Ontario, **U. Laval, ***U. of Plymouth)

Specificity, learners know: On the acquisition of clitic drop in L2 Spanish Two of the main issues regarding the nature of interlanguage (IL) representation are whether it diverges from the target language and if so, whether it is somehow defective. It was recently suggested that interfaces such as syntax/semantics are vulnerable in different types of acquisitions, including that of second languages (L2). According to Sorace, ‘interfaces are more complex than narrow syntax and more inherently difficult to acquire’; they may even remain ‘permanently indeterminate’ in the underlying grammar (Sorace 2004: 143). This may lead to variability and to the construction of IL systems that differ from first language (L1) grammars. Indeterminacy would certainly qualify as a representational deficit. In contrast, recent research shows that L2 learners are able to acquire the target representation even when the interpretation is difficult to come by (Borgonovo et al. 2005; Dekydtspotter & Sprouse 2001). In this paper, we further explore the L2 acquisition of interfaces, showing that adult L2 learners can acquire knowledge of the syntactic correlates of the semantic notion of specificity in constructions involving topicalization and null objects in Spanish. The learners are native speakers of Brazilian Portuguese (BP) where specificity does not constrain the two constructions tested. In both Spanish and BP it is possible to topicalize and drop objects. However, in Spanish specific topicalized DPs must be doubled by a clitic, while non-specific DPs may not (1a-b). Moreover, only non-specific objects can be dropped; specific objects cannot (2a-b). (1) a. Este libro, *(lo) he leído muchas veces. (Specific) this book, it have read many times b. Revistas, Ø leo a menudo. (Non-specific) magazines, read.1S often (2) a. - ¿Has comprado los billetes para el concierto? (Specific) has bought the tickets for the concert - Sí, *(los) he comprado. yes them have.1S bought b. -¿Habrá gente que sepa hablar español? (Non-specific) (there) will+be people who know speak spanish -Sí, habrá ∅. yes (there) will+be Following Uriagereka (1995), we assume a [+specific] feature in the clitic head. This accounts for the presence or absence of the clitic depending on the specific/non-specific interpretation of the topicalized or dropped object. In BP specificity is not a restriction on topicalization and object drop: both specific and non-specific objects can be topicalized without the presence of a clitic, and both specific and non-specific objects may be dropped (De Oliveira 2000; Negrão & Viotti 2000). In the latter case, recovery of the null object is achieved via the discourse. Since the [±specific] feature, which is necessary for interpretation in Spanish, is found in the clitic head, the type of DP (definite, indefinite or bare) will not give the learners any clues as to the specificity of the dropped/topicalized object. They will hear relevant constructions with and without clitics in the input, but will have to learn the subtle interpretive differences (specificity distinction) from the context/discourse. Transfer from BP will give them knowledge of clitics (e.g. clitic projections), but it will not give them how specificity is marked in L2 Spanish. Research so far has looked at knowledge of topicalization and object drop separately, and it has largely involved L1s with no clitics (e.g. Valenzuela 2005). It has also examined constraints on object drop other than specificity (Bruhn de Garavito & Guijarro-Fuentes 2002).

In our study, 26 participants were tested on both object drop and topicalization in Spanish. There were 12 Spanish native speakers and 14 BP-speaking learners of Spanish at the advanced level. The participants were administered a preference judgment task with 60 little scenarios each followed by 4 options. The first two options (a and b) displayed sentences targeting either object drop or topicalization. One contained an overt clitic whereas the other did not. The participants were asked to judge whether the sentences were appropriate given the context provided by the stories, which were meant to establish the specificity of the topicalized/dropped DP. The two other options were ‘both a and b’ and ‘neither a nor b’. There were 30 scenarios with specific DPs and 30 with non-specific ones. In each case, half of the scenarios targeted object drop and half topicalization. In addition, the DP was either definite, indefinite or bare. The stories and the sentences were presented orally to the participants who could also follow them on paper. Both the native speakers and the learners performed better on specific DPs than on nonspecific ones. They correctly accepted overt clitics with specific DPs in both topicalization and object drop contexts. In particular, the L2 learners rarely selected the no-clitic option, although it is acceptable in their L1. With non-specific DPs, the participants tended to overgeneralize overt clitics. For each group, the accuracy rate was significantly lower than with specific DPs (native speakers: 79% vs. 95%; learners: 46.7% vs. 82.2%). The difference is particularly striking for the learners who had a strong tendency to accept overt clitics in the wrong contexts. This behaviour obtains with both topicalization and object drop, and it affects the different DP types (definite, indefinite, and bare) relatively equally. As to the no-clitic option, it was accurately selected with non-specific DPs (natives: 97.4%; learners: 85%). So, it was much less overgeneralized than the overt clitic option. This suggests that the learners have acquired the specificity restriction on noclitic constructions, which does not exist in their L1. Finally, some learners scored within the range of native speakers on non-specific DPs, while others were very close. The results show that learners are able to go beyond the properties of their L1 as far as the acquisition of interface phenomena is concerned. They also suggest that native-like grammars are attainable in L2 acquisition. Taken together, the results argue against a strong version of vulnerability: (1) there is no optionality across the board; instead, errors tend to go in one direction, i.e. overgeneralization of overt clitics; and (2) optionality (i.e. the optional omission of overt clitics in non-specific contexts) can be overcome; it is not inevitable. Vulnerability may mean temporary underspecification of some feature(s) in IL grammars. The errors involving overt clitics may stem from the fact that these elements are underspecified with respect to specifity in IL systems, which would allow them to appear in contexts requiring either a [+specific] or a [-specific] element. Borgonovo, C., Bruhn de Garavito, J. & P. Prévost (2005). Acquisition of mood distinctions in L2 Spanish. In A. Burgos, M.R. Clark-Cotton and S. Ha (eds.) Proceedings of the 29th BUCLD (pp. 97-108). Somerville, MA: Cascadilla Press. Bruhn de Garavito, J. & P. Guijarro-Fuentes (2002). L2 acquisition of indefinite object drop in Spanish. In J. Costa and M. J. Freitas (eds.), GALA 2001 Proceedings (pp. 60-67). Lisbon: Associação Portuguesa de Linguística. Dekydtspotter, L. & R. A. Sprouse (2001). Mental design and (second) language epistemology: adjectival restrictions of wh-quantifiers and tense in English-French interlanguage. Second Language Research 17: 1-35. Negrão, E. V. & E. Viotti (2000). «Brazilian Portuguese as a discourse-oriented language». In Mary A. Kato and Esmeralda V. Negrão (eds.), Brazilian Portuguese and the Null Subject Parameter(pp. 105-125). Vervuert: Iberoamericana. Sorace, A. (2004). Native language attrition and developmental instability at the syntax-discourse interface: Data, interpretations and methods. Bilingualism: Language and Cognition 7: 143-145.

Saturday, session 10 Barbara E. Bullock, Amanda Dalola,Verónica González, Almeida Jacqueline Toribio (Pennsylvania State University)

Asymmetric phonetic convergence in bilingual speech Studies of the VOT values of voiceless stops in the speech of highly proficient Spanish-English and French-English bilinguals demonstrate that these speakers often produce VOT values that fall between the native-like norms for both languages (Flege 1987, 1991, Zampini & Green 2001). Sancier and Fowler (1997) demonstrate that the VOT values for both languages of a bilingual speaker “drift” in tandem according to a speaker’s L1 versus L2 language use. The effect of phonetic convergence in production can be tested directly in code-switching tasks where both languages are simultaneously activated. Grosjean & Miller (1994) contend that the alternation between the VOT values of one phonetic system and another in code-switching is immediate and complete, a finding at odds with the studies cited above. However, a later study by Toribio et. al (2004) probed the degree of phonetic convergence across language modes, measured by VOT of voiceless stops, as Spanishdominant and English-dominant bilinguals read sentences in Spanish, English, and code-switching. The VOT measures for target segments in the code-switched sentences revealed significantly enhanced convergence relative to the monolinguals utterances in both languages (see Fig.1). However, the Toribio et. al. study failed to control for language dominance. If phonetic categories are established early, it is possible that a speaker’s L1 may be less permeable to influence than the L2. The present study, then, tests the effect of the L1 on phonetic convergence in bilinguals’ monolingual and bilingual productions. The participants are 15 L1 Spanish speakers who acquired English in late childhood (AOA average = 12). The study comprised a language background and use questionnaire followed by a production task in which participants read sentences in monolingual English, monolingual Spanish and code-switched blocks as in (1). Sentences within blocks were randomized. (1)

a.

Monolingual English:

His camera was lost on campus.

b. Monolingual Spanish:

Compro pasteles para los niños. ‘I buy cakes for the children.’

c.

The university paga a los empleados muy poco. ‘The university pays its employees very little’

Code-switching (E→S):

d. Code-switching (S→E):

Los pasajeros packed their bags. ‘The passengers packed their bags.’ All tokens of voiceless word initials stops were extracted from the data, measured for VOT, and coded for whether they occurred in monolingual or bilingual contexts. Results show that the VOT measures for target segments at code-switched junctures (exemplified in 1c and 1d) revealed significantly enhanced convergence relative to the monolinguals utterances only in condition (1d). Thus, the results demonstrate that Spanish L1 near-native speakers of English manifest converged VOT values during code-switching but only in their English productions (that is, in their L2). The L1 Spanish VOT values did not converge toward English (see Fig. 2). The implication of this research is that there may be a “phonetic advantage” (see Knightley et al. 2004) to early childhood exposure (particularly when the L1 continues to be used during the life-span) that is not paralleled in other areas of the grammar. Specifically, the phonetic asymmetry observed in this study runs counter to that observed in lexical switching studies where it has been observed that there is a cost in terms of response times when switching from L2 → L1. The results suggest that there may be a phonetic “base language” in bilingual speech, which results from setting the phonetic values of the L1 early and that this may always affect a bilingual’s L2 phonetic productions.

References Flege, J. E. 1987. The Production of ‘New’ and ‘Similar’ Phones in a Foreign Language: Evidence for the Effect of Equivalence Classification. Journal of Phonetics 5, 47-65. Flege, J. E. 1991. Age of Learning Affects the Authenticity of Voice Onset Time (VOT) in Stop Consonants Produced in a Second Language. Journal of the Acoustical Society of America 89, 395-411. Grosjean, F. & Miller, J.L. 1994. Going in and out of languages: an example of bilingual flexibility. Psychological Sciences 5, 201-206. Meuter, R.F.I. & Allport, A. 1999. Bilingual language switching in naming: Asymmetrical costs of language selection. Journal of Memory and Language 40, 25-40. Sancier, M. L. & Fowler, C. 1997. Gestural drift in a bilingual speaker of Brazilian Portuguese and English. Journal of Phonetics, 25, 421-436. Toribio, A.J., Bullock, B.E., Botero, C. and Davis, K. A. 2004. Perseverative phonetic effects in bilingual code-switching. Forthcoming, in Experimental and Theoretical Approaches to Romance Linguistics, R. Gess and E. Rubin (eds.), Benjamins. Zampini, M. and K. Green. 2001. The Voicing Contrast in English and Spanish: The Relationship between Perception and Production. One Mind, Two Languages: Bilingual Language Processing, ed. by Janet Nicol, 23-48. Malden, Mass: Blackwell. VOT Cline (Abstracted from Toribio et al. 2004) 60

50

VOT in ms.

40

30

20

10

0 English

Spanish -English

English-Spanish

Spanish

Figure 1 Mean VOT for Spanish L1 speakers (early second language acquirers of English) 0.08 0.07 0.06

VOT

0.05 0.04 0.03 0.02 0.01 0 English

Span->Eng

Eng->Span Language

Figure 2

Spanish

Heather Burnett*, Katrina Petrik** and Mireille Tremblay*

Saturday, session 12

(*Queen’s University, **U. of Ottawa)

Compositional Telicity in the Prepositional System 0. Introduction This paper bears on the relationship between particles and prepositions in the history of French. We argue that there are no inherent distinctions between particles and prepositions and that the syntactic and semantic differences between the two result from the application of the Merge operation. Under the approach advocated here, transitivity is not a categorial or lexical property, and it cannot be attributed to differences in argument structure. This study is based on the Base de français médiéval (created by C. Marchello-Nizia (ENS-Lyon, France)). 1. The Problem As noted in Buridant 2000, a number of Old French locative or temporal prepositions (1a) could also be used as intransitive particles (1b). (1) a. Sus sa poitrine tenoit ses mains croisant On his chest kept his hands crossed. "He kept his hands crossed on his chest" b. Si corent sus au chevalier If (pro 3pp) run on (part.) to-the knight "They pursued the knight." Moreover, most particles had two interpretations: directional and aspectual. For example, the particle ariere "back" is interpreted as directional in (2a) and iterative in (2b). (2) a. il se traient adont un poi ariere directional they CL withdraw then a little back "Then they withdraw a little back" b. arriere les voit consillier iterative back CL see deliberate "He sees them deliberate again." These data raise a number of questions: 1- how can we account for the two uses of prepositional elements? 2- how do we account for the polysemy of particles? 3- how can we account for the systematic semantic differences between particles and prepositions? We argue that the systematicity of the syntactico-semantic relations calls for a unified analysis. 2. Analysis 2.1 Deriving the interpretation of particles At first, it would seem possible to argue that the directional and aspectual particles correspond to distinct lexical entries and that there exists a systematic homophony in the particle system. However, such an approach would not account for the fact that the two interpretations are in complementary distribution: the directional interpretation arises when the particle is used with a locative or motion verb, while the aspectual interpretation arises in all other cases. Table 1 below illustrates the complementary distribution of particles ariere "back" and avant "forward" in Old French , while (3) lists (in order of frequency) the verbs occurring with directional ariere (3a) and iterative ariere (3b). # of occur. Locative and Motion verbs Other verbs Ariere Directional 199(94%) 100% Iterative 12(6%) 100% Avant Directional 143(99%) 100% Durative 2(1%) 100% Table 1 Distribution of ariere and avant in 13th century prose and verse (3) a. revenir, traire, aller, retorner, torner, venir, remettre, mettre, enchacier, realler, repairier, reporter, mener(se), retraire, courir, suivre, assambler, porter, remener, verser,. revertir. b. consiller, chier, bouter, mander, remander, refuser, rebouter, s’abattre, armer If one can predict the interpretation of particles from the environment in which they occur, this suggests that particles are not marked for the directional/aspectual alternation in the lexicon. Rather, they acquire their specific interpretation when there are merged with the verb as a result of complex predicate formation. 2.2 Deriving Prepositions

2

As mentioned above, particles also seem to have prepositional uses. Again, one could argue that particles and prepositions correspond to distinct lexical entries and that there exists a systematic homophony in the prepositional system. However, such an approach appears suspicious if no account is provided for the systematicity of the phenomenon. More importantly, it would fail to account for the fact that categorial labels correspond to distinct syntactic environments: P elements with a direct object are labeled "Prepositions", and P elements with no direct object are labeled "Particles". Finally, these syntactic differences correspond to systematic semantic differences: P elements with an object are locative, temporal, etc. but never directional or aspectual. This correspondence yields us to believe that in Old French, transitivity is not an inherent property of a subclass of P elements (Prepositions), but rather is determined by the syntactic context in which P elements are inserted. 2.3 Telicity in the P system Many authors (such as Comrie 1976 and Verkuyl 1972) have linked the presence of a direct object with telicity. For example, the addition of an argument can change a non-telic event (an activity) into a telic event (an accomplishment). Extending this approach to the P system yields promising results. We've just argued that in Old French, the difference between particles and prepositions is structural and that this structural difference is responsible for the difference in interpretation between the two. In the V domain, the addition of a direct object provides an endpoint to the event. One could argue that in the P domain, the addition of a direct object also adds an endpoint. According to Jackendoff (1990) and Tenny (1994), prepositional elements encode the semantic feature PATH. If telicity is not restricted to the V domain or event structure, then the addition of an object transforms a non telic P into a telic P (i.e. a particle into a preposition). Under this view, a particle would be a PATH without an endpoint (4a), and a preposition would be a PATH with an endpoint (4b). (4) a. [PP P ] Particle (-telic) b. [PP P NP] Preposition (+telic) This approach provides a straightforward account for the semantic difference between particles and prepositions: directional particles are not telic and locative prepositions are. For example, a lexical item such as avant will be interpreted as directional (i.e. non telic) when not limited by an object, and interpreted as locative (i.e. telic) in presence of such an object. (5) a.Aller avant "To go forward"

[-telic P = Particule]

b.Aller avant le château [+telic P = Preposition] "To go in front of the castle"

3. Conclusion and consequences To conclude, we have argued that it is possible (and desirable) to provide a unified analysis of the semantic and configurational properties of V and P elements. In both cases, transitivity is not a categorial or lexical property. Prepositions are simply Particles with an endpoint: The fact that the interpretation of particles and prepositions can be derived is an example of the impoverished character of the lexicon and highlights the central role of the operation Merge in the construction of meaning. The relevance of the particle-preposition alternation in the debates concerning the role of the lexicon and transitivity has long been underestimated. This is partly due to the fact that prepositions have often been treated as functional elements (e.g. Abney 1987, Baker 2003). Our paper provides good reasons to consider prepositions and particles as lexical rather than functional categories. 4. References

Abney, Steven 1987 The English NP in its Sentential Aspect. Ph.D. Dissertation. MIT. Baker, Mark 2003 Lexical Categories: Verbs, Nouns and Adjectives. Cambridge University Press. Borer, Hagit. 2005a. Structuring Sense: In Name Only. Oxford: Oxford University Press Borer, Hagit. 2005b. Structuring Sense: The Normal Course of Events. Oxford: Oxford University Press. Buridant, Claude. 2000. Grammaire nouvelle de l'ancien français. Paris: Sedes. Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press. Jackendoff, Ray 1990 Semantic Structures. Cambridge, Mass.: The MIT Press. Tenny, Carol 1994 Aspectual Roles and the Syntax-Semantics Interface. Dordrecht/Boston/London: Kluwer Academic Publishers. Verkuyl, 1972 On the Compositional Nature of the Aspects., Dordrecht: Reidel.

Murray Hall, Sunday 10:45-11:45

Subjects and Wh-Questions Anna Cardinaletti University of Venice This paper addresses the long-standing question of the ungrammaticality of overt subjects occurring between a wh-phrase and the verb or between an auxiliary and the verb in Romance languages. Italian data are provided in (1) and (2). (1) (2)

* *

Chi Gianni ha invitato? Chi ha Gianni invitato? Whom did Gianni invite?

Contrary to current proposals, no verb-movement to C is assumed to take place in interrogatives in Romance languages. This explains the ungrammaticality of (2) with no further assumption. This also explains the ungrammaticality of (1). I claim that the distribution of overt subjects in wh-questions does not correlate with pro-drop, as currently assumed, but with verb movement. A strong subject blocks wh-movement unless verb movement applies, as happens in Germanic languages. Notice that only strongsubjects have this blocking effect on wh-movement: weak and clitic subjects do not, as is the case in French and Caribbean Spanish on the one hand and northern Italian dialects on the other.

Laura Colantoni and

Jeffrey Steele (University of Toronto)

Sunday, session 22

Integrating articulatory markedness into a theory of L2 phonological development A theory of L2 segmental development must be able to account for the nature of a learner’s grammar at a given point in time and grammatical development, including relative difficulty of acquisition. In predicting the latter, two theories are often used, namely Eckman’s (1977) Markedness Differential Hypothesis (MDH) and Flege’s (1995) Speech Learning Model (SLM). The MDH predicts that target language sounds present difficulty only when they differ from the L1 equivalent and are typologically more marked. The SLM, in contrast, predicts that new sounds following within the acoustic space of an existing L1 category will present greater difficulty than sounds being discerned as perceptually distinct. As such, the two theories make distinct, empirically testable hypotheses. In this paper, we present data from the study of the acquisition of French and Spanish by native English speakers relevant for testing both theories. We will demonstrate that neither theory alone can account for all of the developmental trends observed. Consequently, we propose that any model should integrate concepts of articulatory markedness. The rhotics of English, French and Spanish differ in realization. In English, is a voiced alveolar approximant [®]. In French, the rhotic is a voiced uvular fricative [“], which may be approximantized and vocalized in codas (Straka 1979); in Spanish, it is a voiced alveolar tap.1 Given the phonetic properties of both target language rhotics, the MDH predicts that Spanish should prove easier, given that tapped rhotics are less marked (i.e. are more frequent) than voiced fricative rhotics (Maddieson 1980). The SLM, in contrast, predicts that Spanish should be harder, given that the tap is highly similar to the flap allophone (e.g. butter [b√R´®]) in the learner’s L1, whereas French [“] is distinct from all L1 segments. To test these predictions, 40 English-speaking learners of French and Spanish (10 Intermediate and Advanced per language) as well as 20 controls were tested on their production of singleton in onsets and codas via reading tasks. This generated 900 tokens for French and 1040 for Spanish, which were analysed acoustically to measure voicing and manner.2 Results show that all French learners mastered frication in onsets. In contrast, non-target-like devoicing was observed, particularly with the Intermediates. In codas, voicing was achieved, however, via higher rates of vocalization than in the control data. The acquisition of Spanish was also variable. In onsets, was realized as an overly long tap. Moreover, even some advanced speakers failed to master voicing. In codas, learners were much less accurate both with manner and voicing. In summary, overall, none of the groups mastered all properties of the rhotic. If we consider manner versus voicing in both target languages, it would appear that only French frication is acquired. As such, both the MDH and SLM make only partially accurate predictions. In order to explain the developmental patterns, the markedness of voiced obstruents must be considered. Cross-linguistically, voiced obstruents are shorter than their voiced counterparts to facilitate voicing. For both languages, learners produced overly long rhotics which leads to devoicing. As such, the integration of articulatory markedness allows for the more accurate prediction and explanation of developmental trends.

1 2

In Spanish, we focus solely on the tap and not the trill. Acquiring place proved no difficulty for the learners and, thus, was not considered.

References Eckman, F. 1977. Markedness and the contrastive analysis hypothesis. Language Learning 27: 315-330. Flege, J. E. 1995. Second language speech learning: Theory, findings, and problems. In W. Strange (ed.), Speech Perception and Linguistic Experience: Issues in Cross-Language Research. Timonium, Maryland: York Press, pp. 233-277. Maddieson, I. 1980. A survey of liquids. UCLA Working Papers in Phonetics 50: 93-112. Straka, G. 1979. Contribution à l'histoire de la consonne R en français. In G. Straka (ed.), Les sons et les mots. Paris: Éditions Klincksieck, pp. 465-499.

Sonia Colina (Arizona State University)

Saturday, session 13

Voicing Assimilation in Ecuadoran Spanish: Evidence for Stratal OT Ecuadoran Spanish has a unique process of /s/-voicing assimilation that differs from other dialects of Spanish in that word-final /s/ is realized as [z] intervocalically (Lipski 1989, 1994). This paper shows that, while both general and Ecuadoran /s/-voicing can be easily accounted for by a serial model of phonology with ordered rules, Ecuadoran presents a problem for Classical Optimality Theory (OT). It is argued that a stratal model of OT (StrOT) is needed to account for the data at hand and that, once a stratal framework is adopted, the OT account appears clearly superior to a serial one. In most dialects of Spanish /s/ is realized as [z] when followed by a voiced consonant (1a, 1c); /s/ surfaces as [s] when followed by a voiceless consonant (1b, 1d) and in intervocalic position (1e, 1f). Ecuadoran Spanish, however, voices intervocalic /s/, albeit only in word-final position (2f, 2h), as word-internal and word-initial /s/ do not undergo voicing (2e, g)1. In lexicalphonological terms, it can be said that in most dialects of Spanish /s/-voicing is a postlexical process that affects coda /s/ when followed by a voiced segment, while the same process in Ecuadoran Spanish applies in word-final position. As a result, a derivational model can account for the data by ordering resyllabification before /s/-voicing in most dialects (thus effectively limiting voicing to postconsonantal position) and after /s/-voicing in Ecuadoran. It has been pointed out, however, that despite accounting for the data, this analysis is not explanatorily adequate as it “gives no insight into the interaction between voicing assimilation… and resyllabification (Lipski 1989: 53)”. In looking to OT for a more explanatory account, the theory runs into serious difficulties. Classical OT, as a non-derivational model of phonology with only one level of evaluation, cannot adequately account for forms such as [lo.zo.tros]. In order for assimilation to take place, (i) [s] in [los] must be in the coda; and (ii) [otros] must be part of the input (i.e, visible/available for evaluation) to provide the trigger for voicing. Yet, if [otros] is in the input, resyllabification must have occurred and [s] can no longer be in the coda. In sum, a paradox arises, namely, [s] must and must not be in the coda. In an OT analysis, [z] is the result of the domination of AGREE (voice) over *z (no voiced alveolar fricatives) and the domination of IDENTOns over A GREE (voice) explains coda voicing (3, 4). While this ranking explains (1), it fails to account for the contrast between (2f) [lo.zo.tros] and (2g) [a.si.Do] or between (2h) [a.zi.Do] and (2g) [a.si.Do], since [z] and [s] are both in the onset. Resorting to anti-allomorphy constraints will not produce the correct output either, in view of alternations such as (2c, 2d). By contrast, Stratal OT (StrOT) (Bermúdez Otero forthcoming, Kiparsky 2003) offers a way out of this impasse. In StrOT some constituents in the morphosyntactic structure of a linguistic expression define phonological domains; phonological domains may in turn invoke different rankings of CON which are thus said to belong to different levels (stem, word and phrase). Consequently, I argue that Ecuadoran differs from other dialects of Spanish in the phrase-level ranking AGREE (voice)>> IDENTOns, *z (5). Other Spanish dialects (word and phrase-level), as well as the Ecuadoran word-level, have the ranking IDENTOns >> AGREE (voice)>> *z (6). This ranking prevents prevocalic assimilation word-internally in Ecuadoran by ruling out candidates such as *[kaza]. In addition to the classic advantages of an OT account, the analysis proposed highlights the connections between s-voicing and other types of coda neutralizations (high ranking of IDENTOns, various rankings of AGREE) on the one hand, and intervocalic voicing (e.g. obstruent voicing), on the other (low ranking of IDENTOns). 1

Compounds are predicted to undergo voicing assimilation, [djo.se.ze.roes] ‘god-heroes’, yet no data are available at this point to confirm that prediction.

(1)

a. b. c. d. e. f. g. h.

/mismo/ /asta/ /las manos/ /las kasas/ /kasa/ /los otros/ /a sido/ /as ido/

[miz.mo] [as.ta] [laz.ma.nos] [las.ka.sas] [ka.sa] [lo.so.tros] [a.si.Do] [a.si.Do]

‘same’ ‘until’ ‘the hands’ ‘the houses’ ‘house’ ‘the others’ ‘has been’ ‘has gone’

(2)

a. b. c. d. e. f. g. h.

/mismo/ /asta/ /las manos/ /las kasas/ /kasa/ /los otros/ /a sido/ /as ido/

[miz.mo] [as.ta] [laz.ma.nos] [las.ka.sas] [ka.sa] [lo.zo.tros] [a.si.Do] [a.zi.Do]

‘same’ ‘until’ ‘the hands’ ‘the houses’ ‘house’ ‘the others’ ‘has been’ ‘has gone’

(3)

AGREE (voice): Adjacent segments agree in voicing *z: No voiced alveolar fricatives IDENTOns: Input features are identical to those of the output when parsed in the onset.

(4)

IDENTOns >> AGREE (voice)>> *z

(5)

Ecuadoran Phrase level: AGREE (voice)>> IDENTOns, *z

(6)

Other dialects: Word level: IDENTOns >> AGREE (voice)>> *z Phrase level: IDENTOns >> AGREE (voice)>> *z Ecuadoran Word level: IDENTOns >> AGREE (voice)>> *z

References: Bermúdez Otero, Ricardo (forthcoming). Stratal Optimality Theory. Oxford: Oxford University Press. Lipski, John. 1989. /s/- voicing in Ecuadoran Spanish. Lingua 79: 49-71. Lipski, John. 1994. Latin American Spanish. London and New York: Longman. Kiparksy, Paul. 2003. Syllables and moras in Arabic. In Caroline Féry & Ruben van der Vijer (eds). The syllable in Optimality Theory. Cambridge and New York: Cambridge University Press. 147-182.

Marie-Hélène Côté (Université d'Ottawa)

Friday, session 8

RHYTHMIC CONSTRAINTS ON THE DISTRIBUTION OF SCHWA IN FRENCH Facts. It has been noticed since at least Léon (1966) that schwa is more likely to surface in penultimate (1a) than in antepenultimate syllables (1b). This generalization, illustrated here with schwas at word boundaries (the same applies to clitic schwas), has been related to stress: schwa omission is disfavored in pre-stress position (Dauses 1973, Morin 1974, Dell 1985). (1) a. La secte part ‘the sect is leaving’ [lasE$ktœpaèr] ?[lasE$ktpaèr] b. La secte partait ‘the sect was leaving’ [lasE$kt(œ)partEè] Previous analyses. Mazzola (1992) has developed a formal analysis of these facts based on the avoidance of stress clash. The omission of schwa in (1a) would result in two adjacent stresses on each of the word-final syllables (?[lasE$ktpaèr]). This account, however, leaves unexplained the fact that the same effect based on the number of syllables following the schwa shows up in contexts where stress clash is not involved. First, in (2a), schwa omission is dispreferred compared with (2b), even though it does not result in adjacent stressed syllables. Schwa insertion here appears to be motivated by a constraint against monosyllabic phrases - a minimality condition distinct from stress clash avoidance. Second, schwa omission is also less likely when preceded by a monosyllabic rather than a disyllabic word, even though, again, no stress clash is involved (3). In addition, the stress clash analysis does not specify how the rhythmic effect interacts with segmental constraints. At word boundaries, schwa is systematically omitted after one consonant even in pre-stress position (compare with (1a)). (2) a. De l’or ‘some gold’ [dœlçèr] ??[dlçèr] b. De l’ortie ‘some nettle’ [d(œ)lçrtiè] (3) a. Just partait ‘J. was leaving’ [Zy$støpartEè] ?[Zy$stpartEè] b. Saint-Just partait ‘St.J. was leaving’ [sE)Zy$st(ø)partEè] (4) La bête part ‘the beast is leaving’ [lab´tpar] *[lab´tøpar] Proposal. I develop an analysis of the rhythmic effect on the distribution of schwa in terms of restrictions on prosodic phrasing interacting with constraints on segment perceptibility. I adopt a perceptual approach to the distribution of schwa (Côté 2000): consonants are most perceptible when adjacent to a vowel and schwa insertion serves to enhance the salience of consonants by providing them with the cues associated with vocalic transitions. The likelihood of schwa insertion depends on the quality of the perceptual cues available to the consonants. Schwa omission is the rule in contexts where the surrounding consonants are adjacent to a vowel, as in (4). Consonants that are not adjacent to any vowel trigger schwa insertion as a function of the strength of their non-vocalic cues. One relevant factor appears to be the position of consonants in the prosodic structure : consonants adjacent to edges of prosodic constituents are more salient than consonants in internal positions, due to the perceptual effects of lengthening and articulatory strengthening processes applying at edges (Wightman et al. 1992; Fougeron & Keating 1997). Schwa omission is therefore more likely at an edge (5b) than internally (5a). (5) a. c’est le triste sort ‘it is a sad fate’ [selœtristœsçr] ??[selœtristsçr] b. je suis triste, sors ! ‘I am sad, go out !’ [ZøsÁitrist(œ) sçr] The facts in (1)-(4) all follow from the fact that consonants that are not adjacent to a vowel preferably appear at the edge of a prosodic boundary, which interacts with a constraint against monosyllabic prosodic phrases. The distribution of schwa relies on the general constraint in (6a), which requires consonants to be adjacent to a vowel. This constraint can be relativized with respect to the strength of the perceptual cues available to C : the more salient C is, the lower ranked the corresponding C↔V constraint is. Specifically here, a C adjacent to a prosodic boundary more easily resists the absence of a flanking vowel, hence the constraint in (6b) and the fixed hierarchy in (6c). The two constraints in (6a-b) interact with the one in (6d), which prohibits monosyllabic prosodic phrases (see Di Cristo 1998), and DEP, which prohibits schwa

insertion (6d). The only language-specific ranking is C|Ø↔V >> DEP, the variability of schwa insertion resulting from the crucial non-ranking of the other constraints in (6) (e.g. Anttila 1997). (6) a. C↔V A consonant is adjacent to a vowel. b. C|Ø↔V A consonant not flanked by a prosodic boundary is adjacent to a vowel. c. C|Ø↔V >> C↔V d. *(s) Do not have a monosyllabic prosodic phrase. The behavior of schwa interacts with prosodic phrasing, the general idea being that the positioning of prosodic boundaries competes with schwa insertion to accomodate complex consonant sequences. The utterances in (2) comprise one phrase, but those in (1) and (3-4) may be parsed into one or two phrases, yielding four relevant candidates, as in (7-8) for (1a-1b). (7) a. (lasEktøpar) b. (lasEktpar) c. (lasEktø) (par) d. (lasEkt) (par) (8) a. (lasEktøpartE) b. (lasEktpartE) c. (lasEktø) (partE) d. (lasEkt) (partE) The candidates in (7-8b) violate C|Ø ↔V, since they contain a phrase-internal consonant that is not adjacent to a vowel, and they are less harmonic than (7-8a). The candidates in (7-8d) violate the lower-ranked C↔V since the interconsonantal [t] appears phrase-finally. The candidates in (7c-d) violate the minimality constraint in (6d). The analysis sketched above correctly predicts : obligatory schwa omission in (4), preference for schwa insertion in (1a) and (3a), a stronger preference for insertion in (2a), and no preference between insertion and omission in the (b) sentences in (1)-(3). In other words, schwa omission is more likely in three-consonant clusters if a prosodic boundary intervenes and the phrasing does not yield sub-minimal constituents, as in (8d). If the two-phrase parsing violates minimality, as in (7d), schwa insertion is favored. Advantages and consequences. This analysis covers a wider range of facts than previous analyses and it provides a unified account of the segmental and prosodic/rhythmic constraints on the distribution of schwa, which is motivated by consonant perceptibility. Perceptual salience constraints simply interact with a well-motivated minimality constraint on the size of prosodic constituents. This account also has consequences for analyses of prosodic phrasing, which have focused on the role of syntactic constituency and prosodic constraints in the location of prosodic boundaries (e.g. Truckenbrodt 1999, Prieto 2005). The schwa facts presented here add a new dimension by showing how prosodic phrasing may also be influenced by segmental constraints. References Anttila, A. (1997) "Deriving variation from grammar", in F. Hinskens, R. van Hout & W.L. Wetzels (réd.) Variation, change and phonological theory, Amsterdam : Benjamins, 35-68. Côté, M.-H. (2000) Consonant cluster phonotactics : a perceptual account, Ph.D., MIT. Dauses, A. (1973) Etudes sur l’e instable dans le français familier, Tübingen : Max Niemeyer. Dell, F. Les règles et les sons, 2nd ed., Paris : Hermann. Di Cristo, A. (1998) "Intonation in French", in D. Hirst & A. Di Cristo (eds.) Intonation systems, Cambridge : Cambridge University Press, 195-218. Fougeron, C. & P.A. Keating (1997) “Articulatory strengthening at edges of prosodic domains”, JASA 101 (6): 3728-3740. Léon, P. (1966) “Apparition, maintien et chute du e caduc”, Linguistique 2: 111-122. Mazzola, M.L. (1992) “Stress clash and segment deletion”, in C. Laeufer & T.A. Morgan (eds.) Theoretical analyses in Romance linguistics, Amsterdam : Benjamins, 81-97. Morin, Y.-C. (1974) “Règles phonologiques à domaine indéterminé: chute du cheva en français”, Cahier de linguistique 4: 69-88. Prieto, P. (2005) "Syntactic and eurhythmic constraints on phrasing decisions in Catalan", Studia linguistica 59 (2). Truckenbrodt, H. (1999) "On the relation between syntactic phrases and phonological phrases", Linguistic inquiry 30 : 219-255. Wightman, C.W., S. Shattuck-Hufnagel, M. Ostendorf & P.J. Price (1992) “Segmental durations in the vicinity of prosodic phrase boundaries”, JASA 91 (3): 1707-1717.

Past participle agreement in Abruzzese: split auxiliary selection Saturday, and the null-subject parameter

session 15

Roberta D’Alessandro and Ian Roberts (University of Cambridge)

1. Past participle (pp) agreement in Romance languages is usually thought to be governed by the mechanisms proposed by Kayne (1989) and Belletti (2002). According to Belletti, pp agreement always takes place with internal arguments. Here, we present counterevidence to this generalization, and propose a novel analysis of pp agreement, as well as person-driven auxiliary selection, using previously unknown data from a southern Italian dialect spoken in Abruzzo. 2. In Eastern Abruzzese (EA), pp agreement may take place either with the subject (1) or with the object (2) of transitive verbs, with the subject of unergative (3) and unaccusative verbs (4). The pp always agrees with a plural DP, be it the subject or the object; see (5). 3. EA has a “split-ergative” system of auxiliary selection (Manzini & Savoia (2005)). Many dialects in Central and Southern Italy show this type of auxiliary selection. In these varieties, the argument structure of the verb is not the conditioning factor, but rather the person-number specification of the subject. There is much variation, but EA represents probably the most common pattern, that where “be” appears with a 1st- or 2nd-person subject, and “have” with a 3rd-person subject is ((Cocchi (1995, Chapter 4), Kayne (2003, Chapter 7), Ledgeway (2000, Chapter 6), Loporcaro (1998), Tuttle (1986)). If the auxiliary is merged in v, this indicates that v is sensitive to the -features of the subject; it is in this sense that such systems are ergative (see Müller (2004)). 4. We provide an analysis of EA auxiliary selection and participle agreement following Chomsky’s (2001, 2004, 2005) recent proposals centring on the Agree relation, a head-head feature-checking relation subject to locality conditions. Our main proposal is that EA has two vPs, one of which licenses the (nominative) subject and the other the (accusative) object. We further assumes that singular agreement represents the underspecified value of the Number feature, and as such can be overridden by the plural value. We show how these assumptions give rise to an effect of past-participle with both plural subjects and plural objects. 5. An important consequence of our analysis is that Nominative case is assigned within vP, and not by T as is standardly assumed. Thus the role of T is somewhat reduced. We suggests that T does not license the subject in EA, but only the inflected verb. T thus can only trigger verb-movement, not subject raising. We conclude that varieties with the kind of auxiliary-selection and participle-agreement system described here must be null-subject languages (see Alexiadou & Anagnostopoulou (1998)). Support for this claim comes from the impossibility of inserting a subject immediately after a low complementizer in EA. 6. The correlation with the null-subject parameter explains a significant cross-linguistic generalisation. Although both argument-structure-driven auxiliary selection and the generalised use of “have” as perfect auxiliary are found in Germanic (German vs. English), the Central-Southern Italian person-driven system is not found. Our analysis entails that a person-driven system can only be found in a null-subject language. Since no Germanic language is a null-subject language, we understand the typological generalisation as a nontrivial case of parameter interaction. Examples: (1) Giuwanne e Mmarije a pittite nu rd John and Mary have-3 sg/pl painted-pp pl a-sg masc ‘John and Mary have painted a wall’ (2) Giuwanne a pittite ddu mure John has-3rd sg/pl painted-pp pl two walls ‘John has painted two walls’

mure wall-sg masc

(3) Nu seme tilifunite st we-1 pl are-1st pl called-pp pl ‘We called’ (4) Vu sete arrivite you-2nd pl are-2nd pl arrived-pp pl ‘You arrived’ (5) a. Giuwanne a pittate nu mure/pittite ddu mure rd John has-3 sg/pl painted-pp sg a wall / painted-pp pl two walls ‘John has painted a wall/painted two walls’ b. Giuwanne e Mmarije a pittite nu mure/pittite ddu mure rd John and Mary have-3 sg/pl painted-pp pl a wall /painted-pp pl two walls ‘John and Mary have painted a wall/painted two walls’

Suzanne Dikker (New York University)

Saturday, session 9

The classification of Spanish prepositions: Evidence from Media Lengua Media Lengua, spoken in Ecuador, is a bilingual mixed language (Thomason 2001) of Spanish and Quechua: Quechua provides the morpho-syntactic frame, within which Spanish lexical roots (+ 95%) are incorporated. As a consequence of this apparent perfect split between lexicon and grammar, the structure of Media Lengua can provide insight into the linguistic status of the elements of which it is composed. More specifically, it is argued that evidence from Media Lengua can tell us something about the lexical and syntactic properties of adpositions, a much debated category. A study of spontaneous Media Lengua production data reveals the following pattern: Media Lengua makes use of Spanish prepositions, but only prepositions whose semantic functions correspond to those expressed by spatio-temporal suffixes in Quechua (Fernández López 1999; Cole 1982), sometimes resulting in the double marking of a single function, as shown in (2): (1)

En año 1986-pi LOCSPA yearSPA 1986-LOCQUE “In the year 1986”

Spanish prepositions that do not form the direct realization of basic semantic functions (Mackenzie 2001), on the other hand, are not found. In contrast, spatio-temporal elements that are generally considered to be adverbial in nature do occur in Media Lengua, occupying the slots that in Quechua are filled by relational nouns: (2)

cama

adentro

-manta

bedSPA insideSPA - SOURCEQUE “From inside the bed”

It is argued that the reanalysis of these elements as relational nouns can be explained in view of their (restricted) nominal behavior in Spanish, for example by allowing possessive modification. In spite of the head-final character of Quechua, no preposition was found in postnominal position. Thus, the generalization that prepositions are never borrowed as postpositions (Campbell 1993) is supported. This finding has interesting implications for the nondetachability of subcategorization features from the phonological form of lexical entries. Tosunoda et al (1993) argue that the prepositional versus postpositional character of a given language is the best indicator of its typological word order properties in general (e.g., Hawkins 1983). In view of this, the presence of prepositions can be linked to the relatively high frequency of SVO word order in Media Lengua (Ocampo & Klee 1995). The present findings are interpreted as follows: 1. The Spanish class of prepositions can be subdivided into two categories, namely: a) prepositions that express basic spatio-temporal semantic functions and are stored in the lexicon as semantic operators, and b) prepositions that are proper predicates (Bakker & Siewierska 2002). This subdivision is supported by data from Media Lengua if one adopts Muysken’s (1996) generalization that Media Lengua conserves the parts-of-

speech system of Quechua: While the former correspond to semantic suffixes in Quechua, and can thus be incorporated into Media Lengua, there is no slot available for the latter type, since Quechua lacks a proper class of adpositional predicates. 2. Adpositional lexical entries carry features that encode the syntactic directionality of the relation they express. These features cannot be detached from their corresponding phonological form. Otherwise, we would have expected the Spanish prepositions to occur in postnominal position in Media Lengua. References: Bakker, D. & A. Siewierska (2002) ‘Adpositions, the Lexicon and Expression Rules’. In: R. Mairal Usón & M.J. Pérez Quintero (eds) New Perspectives on Argument Structure in Functional Grammar. Berlin: Mouton de Gruyter, pp. 125-178. Campbell, L. (1993) ‘On Proposed Universals of Grammatical Borrowing’ In: Aertsen, H. & R. Jeffers (eds): Historical Linguistics 1989: Papers from the 9th International Conference on Historical Linguistics. Amsterdam: John Benjamins, pp. 91-109 Cole, P. (1982) Imbabura Quechua. Amsterdam: North-Holland Publishing Company. Fernández López, M. (1999) Las Preposiciones en Español. Valores y Usos. Construcciones Preposicionales. Salamanca: Ediciones Colegio de España. Hawkins, J.A. (1983) Word Order Universals. New York/London: Academic Press. Mackenzie, J.L. (2001) ‘Adverbs and Adpositions: The Cinderella Categories of Functional Grammar’ Revista Canaria de Estudios Ingleses, 42, abril 2001, pp. 119135. Muysken, P. (1996) ‘Media Lengua’ In: S.G. Thomason (ed): Contact Languages: A Wider Perspective, Amsterdam: John Benjamins, pp. 365-426. Ocampo, F.A. & C.A.Klee (1995) ‘Spanish OV/VO Word-Order Variation in SpanishQuechua Bilingual Speakers’ In: C.Silva Corvalán (ed) Spanish in Four Continents: Studies in Language Contact and Bilingualism. Washington D.C.: Georgetown University Press, pp. 71-82. Thomason, S.G. (2001) Language Contact, an Introduction. Edinburgh: Edinburgh University Press. Tosunoda, T., S. Ueda & Y. Itoh (1995) ‘Adpositions in Word-Order Typology’ Linguistics 33, pp. 741-761.

Mariapaola D’Imperio*, Roxane Bertrand*, Albert Di Cristo** and Cristel Portes** (*Laboratoire Parole et Langage

**Université de Provence & Laboratoire Parole et Langage) Friday, session 8

Investigating phrasing levels in French : The phonology and phonetics of prenuclear and nuclear accents Traditional accounts of French prosody distinguish at least two constituency domains, such as the groupe accentuel and the groupe intonatif (Verluyten, 1983). While some of the criteria previously employed to define the size of these constituents were mainly phonetic, we question nowadays the phonological status of these domains within a phonological model of intonation. According to some autosegmental models of French intonation, the lower of these two phrasing levels is the domain of the primary accent (LH*) which corresponds to the “accentual phrase” (AP) described by Jun & Fougeron (2000, 2001). Within this model, the LH* pitch accent is in fact associated with the last strong syllable within the AP. The AP also roughly corresponds to the “prosodic word” of Martin (1977), the “phonological word” of Selkirk (1972), the “intonation group” of Mertens (1993), the “phonological phrase” of Post (2000), the “rhytmic unit” of Di Cristo & Hirst (1993) and the “rhythmic group” (RG) more recently proposed by Delais-Roussarie (1995). Higher in the French prosodic hierarchy (being also the highest level), we find the intonation phrase (IP). This level is described by Jun & Fougeron (2000, 2001) and Post (2000), matching, among others, Vaissière’s “breath group” (Vaissière 1997) and Hirst & Di Cristo’s “intonation unit” (Di Cristo 1998). In the British tradition (‘O Connor and Arnold 1973) this last phrasing level is the domain of the “nuclear accent”, which is positionally defined as being the last accent within this constituent, as well as being the most prominent one in the prosodic hierarchy. Any preceding accent in the intonation phrase is defined as “prenuclear”. Within the autosegmental-metrical model of Pierrehumbert and colleagues (Pierrehumbert 1980, Beckman and Pierrehumbert 1986) only the positional definition remains but it defines a lower level, the intermediate phrase. Specifically, the nuclear accent is the accent immediately preceding the phrase accent. It is also the case in the model of French intonation proposed by Jun and Fougeron (2000) that there is no formal distinction between nuclear and prenuclear accent apart from a strict positional definition. However, Di Cristo (1999), in line with Ladd (1986), claims that a phonological distinction between these two accents should be kept for French, thus supporting the difference between two phrasing levels. Hence, our goal here is to determine whether there is a formal difference between rising nuclear and prenuclear accents (IP-final vs. AP-final only) in French and at the same time verifying the existence of two phrasing levels in the French prosodic hierarchy (AP and IP). The analysis was conducted within the laboratory phonology framework, in that the experimental utterances were prosodically transcribed by two experts and then checked by a third one. Only later, we compared the different accentual and phrasal categories by analyzing, among the other things, the tonal and temporal characteristics of their tonal targets as well as durational characteristics of the target syllables. The hypothesis tested here is that nuclear accents differ from prenuclear ones in terms of formal characteristics of the contour which cannot be explained, for instance, by invoking the presence or absence of an upcoming tone, as proposed, for example, for English by Silverman and Pierrehumbert (1990). Our corpus consists of sentences in which the same syllable [mi] could occur either in nonfinal (1) or final (2) position within the IP, and was always potentially at least AP-final. Moreover, we varied the syllable structure between open and closed as well as the number of APs within the IP.

(1) {Oui mais [les amis]AP [de mamie]AP}IP, ils n’arrivent que demain. “Yes, but grandma’s friends, they only arrive tomorrow”. (2) {Oui mais [la famille]AP [des amis]AP}IP, elle n’arrive que demain.

“Yes, but the friends’ family, it only arrives tomorrow”. After the transcriptions were carried out, we calculated inter-transcriber agreement scores and carefully separated the different events so as to perform the acoustic-phonetic analysis in an informed way. We specifically measured the temporal alignment of both L and H targets within the same LH* pitch accent (while falling accents were excluded from the analysis), as well as their tonal scaling. Moreover, we measured the duration of the segments composing the target syllable, as well as its global duration. This last measure was intended to test whether possible target alignment differences between nuclear and prenuclear accents can be entirely predicted by the amount of preboundary lengthening (IP-final>AP-final). If not, the tonal category difference might compensate for the lack of a tonal marking which would be exclusive to the lower boundary domain (AP). The results bear upon the general issue of phrasing constituents and their marking in Romance languages, which has started to be tackled for languages such as Italian, European Portuguese, Catalan and Spanish (cf. D’Imperio, Elordieta, Frota, Prieto and Vigário, 2005; Frota, D’Imperio, Elordieta, Prieto and Vigário, to appear) and still awaits a clear treatment for French.

Nathalie Dion and Shana Poplack (University of Ottawa)

Friday, session 5

Grammar vs usage in the study of linguistic change This paper confronts the evolution of the expression of future temporal reference over 450 years of French grammatical tradition with speech data spanning a century and a half. With three competing variants, the synthetic future (SF), as in (1), the periphrastic future (PF), as in (2) and the futurate present (P), as in (3), this grammatical sector has long been a major site of inherent variability. (1) Il dit, "Je vous dirai (SF) ça demain matin, pensez à vos affaires cette nuit." (XIX/037/580) ‘He said, “I will tell you that tomorrow morning, think about your things tonight” (2) "Demain matin", il dit, "il va manger (PF) une volée", il dit, "il va manger (PF) une rince". (XIX/023/4292) ‘ “Tomorrow morning”, he said, “he is going to get smacked”, he said, “he is going to get a beating.” ’ (3) Arrive chez eux, il dit à sa femme, il dit, "Écoute", il dit, "demain matin, je tue (P) la vache." (XIX/038/2111) ‘Gets home, he says to his wife, he says, “ Listen”, he says, “tomorrow morning, I kill the cow.”’ The received wisdom is that the variant forms illustrated in (1) – (3) are selected according to distinctions in the way the speaker envisions the future eventuality or the semantic reading s/he wishes to convey. But in a large-scale study of over 3500 future temporal reference contexts in Canadian French, Poplack & Turpin (1999) found that forms were rarely selected by speakers in accordance with the values commonly attributed to them. This is because reference to future states or events is basically made by the periphrastic variant, which has ousted the synthetic future from virtually all contexts of productive usage but one. Is this an innovation and if so, when did it take place? To address this question, we make use of two novel sources of diachronic data, which, we argue, represent complementary historical benchmarks against which the facts of contemporary usage may be assessed. One is an audio corpus of 19th-century spoken Quebec French, with a time-depth of about a century and a half. The other is a meta-analysis of normative grammatical tradition, as instantiated by the Recueil historique des grammaires du français, a corpus of 163 grammars of French published between 1530 and 1998. Under the assumption that forms salient enough to have incited the opprobrium of grammarians were likely widespread in the speech of the time, we demonstrate how these materials may be used to 1) infer the existence of prior variability (from citations of alternate forms in like contexts), 2) trace the evolution of normative dictates associated with the variants, and most important, 3) discern hints of prior linguistic conditioning of variant selection. These are then operationalized as factors in a multivariate analysis and tested against the facts of usage. This method not only detects factors that are statistically significant in the prediction of variant choice, but also their magnitude and direction of effect. This information is expressed as

constraint hierarchies, which may be compared with counterparts at different points in time to establish and date change. Systematic comparison of grammatical treatments with actual speaker behaviour in apparent time shows virtually no correspondence between the motivations offered in the literature and those constraining actual variant choice. For well over four centuries, proximity of the realization of the future eventuality was offered as the predominant explanation; in contemporary spoken French this factor is not statistically significant. Instead, polarity, a factor that had not been previously acknowledged, emerged as the major determinant of variant choice. Future predications with negative polarity are overwhelmingly expressed with the synthetic variant, while their affirmative counterparts tend to co-occur with the periphrastic form, as in (4). (4) Dire que dans quatre cents ans d'ici bien, il va avoir (PF) encore des Fauteux puis ils vont encore parler (PF) français! Qu'ils parleront (SF) pas l'anglais. (XX/004/3611) ‘To think that in 400 years from now, there are still going to be Fauteux, and they’re still going to speak French. That they won’t speak English.’ In an effort to date this apparent change, we replicated the study on the 19th-century corpus. The synthetic future was found to be far more common (and the periphrastic future correspondingly less so) at that time, with the futurate present maintaining the same minor role. But the contemporary constraint hierarchies conditioning variant choice were already in place. In particular, negative polarity was already the strongest predictor, showing that this effect, despite being unattested, is by no means a recent innovation. Proximity, on the other hand, did have the expected effect in the 19th century, however tenuous: remote states and events tended to favor the synthetic future, while proximate ones were associated with the periphrastic variant. This constraint has since been lost. Thus, contrary to assumptions based on either the synchronic facts or the grammatical tradition alone, our method shows that any change that has taken place in the Canadian French future temporal reference sector involves loss of the proximity effect, and not innovation of the polarity effect, which turns out to be long-established. We explore why neither of these developments have been acknowledged by the grammatical tradition, and the implications of these findings for the use of grammarians’ observations as the data for linguistic analysis. REFERENCES: Poplack, Shana & Turpin, Danielle. 1999. Does the FUTUR have a future in (Canadian) French? Probus 11 (1):133-164.

Expletives and the Rise of Number

Friday, session 5

Monique Dufresne (Queen’s University)

1.

Introduction The French language has gone through many changes in the syntax of its impersonal constructions between Old French and the Renaissance period. In this paper, we examine the variation in verbal agreement and its subsequent loss at the end of the medieval period. We will show that the changes observed from Old French (OF) to Modern French (ModFr) are the result of a change in feature specifications, in particular, the specification of the NUM feature. It has always been the case that, in French, there is only one expletive, il. 2.

The facts of OF Both grammarians of Old French (Moignet, 1976; Buridant, 2000) and researchers working in the framework of Generative grammar (Roberts, 1993; Arteago & Herschenson, 2004) have noticed that there exists great variation in verbal agreement in impersonal structures. In the following examples, we can see that the verb can with an associate (the post-verbal subject), whether the subject is in nominative case (1a) or in “régime” case (1 b). 1. a. Il i vont ci (nom. plur) viel (nom. plur) prestre (nom. plur) et cil viel clop et cil manque There went those old priests and those injured persons and those one armed persons (Auc. p. 14) b. En cel pré avoit un ratelier ou il manjeoient cent rég. plur) cinquante toriaus (rég. plur) (Queste, p. XX) In that field was a rack where there ate hundred fifty bull

However, associate agreement is not obligatory, and, according to Buridant, structures of the type illustrated in (2) are not rare in OF. 2.

Eissi neia des nex plusors / Qui ne purent avoir secors Chron DN, 40541-42, in Buridant, p. 404) Thus sank PRO many (rég. plur) ships (rég. plur) / who could not have been saved.

Finally, the examples in (3) also highlight that, in OF, it was possible to use either il (i) a or estre in existential structures; whereas only il y a is permitted in ModFr. It is also noted that here, again, the associate can be marked for nominative or régime case. This fact is remarkable since, in constructions with estre, the case-marking is predominantly nominative. 3. a. Li (nom. sing) jacinctes (nom. sing.) clers (nom. sing) i est il BRENDAN, p. 75 the gem stones bright (loc) is there/ b. Et lors commença a fere si grand duel qu’il ne fust si dur (rég. sing) home (rég. sing) form Moignet p, 145 And so PRO began to make such a great duel that there was never such a strong man

It is important to highlight the fact that, in cases where the postverbal associate has a different case than the expletive, this associate is always singular. These facts raise two questions: 1) How can we account for such a variation in verbal agreement within a single language? 2) Why did French change from a language where existentials could be formed with estre to one where they are formed only with il y a? We propose to account for these facts by postulating a change in the marking of the feature NUM. Our analysis is based on three sets of data which will be elucidated in section 3. 3. Analysis 3.1.1 The loss of case and the rise of number There exist two cases in OF : nominative and complement. As in Latin, the case morphemes are of the porte-manteau type as they express gender, number and case. SING PLUR Sub Li chevaliers Li chevalier Obj Le chevalier Les chevaliers Table 1: Case/number marking on definite nouns

SING PLUR Sub Uns chevaliers Uns/φ chevalier Obj Un chevalier Uns/φ chevaliers Table 2: Case/number marking on indefinite noun

4. Li (nom sing) roys (nom sing) Artus vist (3ps) le (rég. sing) chevalierφ

(rég. sing)

“King Arthur saw the knight”

Tables 1 and 2 show that the OF morpheme -S was ambiguous: it could either be used for nominative or complement marking. By the end of the OF period (circa end of the 13th c.), and from the beginning of the Middle French (MidFr) period, an important change occurred in the language: a new system emerged, where case was no longer marked and where the morpheme –S was reanalyzed as a plural marker. (Buridant 2000, Vance 1997). 5. Les (subject plur) barons (subject plur) du pays conforterent (3pp) la dame et ses enfans The lords of the country comforted the lady and her children.

Melusine, p. 9

Moreover, during the MidFr period, the language underwent another important change: the loss of all its inflexional markers. All these changes meant that the determiners became the only elements that expressed the distinction between singular and plural. 3.1.2 The emergence of singular expletives Now turning to the pronominal system of OF, it is characterized by the fact that the 3rd person was only realized through one form, regardless of whether it expressed either the singular or the plural, as can be seen in (6): 6. Il mangent(3pp)

il mange(3ps)

However, the reanalysis of the morpheme -S as a plural marker brought forth a regularization in the expression of number within the pronoun system. Just as a singular/plural distinction could be seen in OF in the feminine pronouns el/els, in MidF there arose a similar difference in the masculine forms, with ils emerging as a new plural pronoun. As a consequence, the expletive il was reanalyzed as a only singular pronoun. 4.

Consequences Our analysis explains why agreement with the postverbal associate is no longer possible in ModFr, while it was possible in medieval French. Moreover, we can now understand why the existential structure with être is no longer part of the French grammar: Recall that during the Renaissance period, subject pronouns became compulsory, and that the OF estre required agreement between the expletive and the postverbal element (which is still the case in ModFr). Thus a sentence such as the one in (7) crashes in ModFr because of a number clash. Indeed, since il is singular, it does not bear the same feature as des hommes, which is plural. 7. *Il sont des hommes. sing. plur.

5

Bibliography

Arteaga, Deborah and Julia Herschensohn (2004) “Case, Agreement, and Expletives: A parametric difference in Old French and Modern French”. In Contemporary Approaches to Romance Linguistics, Auger, Julie, J. Clancy Clements and Barbara Vance (eds.), 1–15, Cambridge: Benjamins Bouchard, Denis. (2002) Adjectives, Number and Interface. Why Languages Vary. Amsterdam : Elsevier. Buridant, Claude. (2000) Grammaire nouvelle de l’ancien français. Paris : SEDES Chomsky, Noam. (2001) “Derivations by phase”. In Ken Hale: A life in language, Kenstowicz. M.( ed.), 1-52, MA: MIT Press. Cardinaletti, Anna. (1997) “Agreement and control in expletives constructions”. Linguistic Inquiry 28, 521-533. Moignet, Gérard. (1976) Grammaire de l’ancien français. Klincksieck: Paris. Travis, Lisa. (1984) Parameters and Effects on Word order Variation phd dissertation, MIT. Vance, B. (1997) Syntactic Changes in Mediaval French:Verb Second & Null Subjects. Boston: Kluwer

Anna María Escobar (University of Illinois, Urbana-Champaign)

friday, session 3

Performative verbs in Spanish monolingual and bilingual colonial court documents Following Austin's seminal work (1962), performative verbs are defined as verbs that "can be used not only to report on sayings but to have the force of a doing" (Traugott and Dasher 2005:190). Historically, performative verbs develop crosslinguistically from a small set of verbs (Traugott 1991). However, Traugott and Dasher characterize performative verbs as "deeply embedded in the social deixis of the Speaker/Writer - Addressee/Reader pair" (2005:191), making the definition of performativity dependant on the pragmatics of each language (Traugott and Dasher 2005:196). In this paper, I want to address this problem by hypothesizing that in monolingual and bilingual data of the same language, differences can also be found in the use of performative verbs. The study analyzes the use of performative verbs in historical court documents written in Spanish in the late 16th century and early 17th century. These documents are complaints for criminal, civil, or religious matters, where the claimants make a request to the court to address their problem. The claimants are either Spaniards (monolingual documents) or individuals who identify themselves in the documents as of Indian origin (bilingual documents). The study distinguishes the following subtypes of performative verbs: assertives (e.g. decir 'to say'), strong assertives (e.g. alegar 'to allege'), weak requests (e.g. pedir 'to ask for'), strong requests (e.g. suplicar 'to beg'), and commissives (e.g. prometer 'to promise') (following Fraser 1975; Traugott 1991; Traugott and Dasher 2005). Following Traugott 1991, I also include two subtypes of verbs which can be used with performative meaning: mental verbs (e.g. saber 'to know') and verbs of vision (e.g. ver 'to see'). The preliminary results suggest that the type of document favors different types of performative verbs. I find that while in the monolingual documents verbs of both types of assertives are found, in the bilingual documents assertives type 1 are favored. In the monolingual documents, I find that the use of mental and vision verbs are also favored; while in the bilingual documents, strong requests are favored, as well as strategies to hedge the illocutionary force of the performative verb. Since little research has been done on the historical development of performatives (cf. Traugott and Dasher 2005:195), the discussion addresses the similarities that exist between performatives, and deontic and epistemic modalities (cf. Traugott 1991; Traugott and Dasher 2005:192), as well as the use of these verbs in modern varieties of Spanish in the region. Cited Bibliography Austin, J. L. 1962. How to do things with words. Oxford: Oxford University Press. Fraser, Bruce. 1975. Hedged Performatives. In: Syntax and Semantics Volume 3: Speech Acts, edited by Peter Cole and Jerry L. Morgan, 187-210. San Diego: Academic Press. Traugott, Elizabeth Closs. 1991. English Speech Act Verbs: A Historical Perspective. In: New Vistas in Grammar: Invariance and Variation, edited by Linda Waugh and Stephen Rudy, 387-406. Amsterdam/Philadelphia: John Benjamins. Traugott, Elizabeth C. and Richard B. Dasher. 2005. Regularity in Semantic Change. Cambridge: Cambridge University Press.

Ricardo Etxepare (CNRS)

Friday, session 6

Tacit verbs and semi-direct speech in iberian Spanish In colloquial speech, main clauses in iberian Spanish can be headed by an overt complementizer (Spitzer, 1942; García, 1996; Etxepare, 2002, 2005): (1) a. Juan/oye, el Liverpool ha ganado la Champions Juan/hey, Liverpool has won the Champions League b. Juan/oye, que el Liverpool ha ganado la Champions Juan/hey that Liverpool has won the Champions League The apparent optionality of the complementizer masks an important semantic difference between the (a) and (b) sentences. As a typical declarative sentence, (1a) constitutes an assertion, whose propositional content is that a given soccer team (Liverpool) has won the Champions League. Compared to (1a), (1b), uttered with declarative intonation, contributes the additional meaning that someone else (who is not the speaker) said (1a), such that the (speaker’s) utterance of (1b) constitutes a report of what has been said. (1b) is thus reported speech (Coulmas, 1986). Etxepare (2005) develops the idea that such structures in Spanish involve an extra speech eventuality, syntactically represented as a (tacit) VP akin to quotative verbs in other languages (Guldemann, 2001, for a recent overview). As I will show, the presence of certain aspectual operators in the root VP licenses Context Shift, a shift in the domain in which indexical parameters are evaluated (Schlenker,2003;von Stechow, 2004). First note that root complementizer constructions in Spanish may involve overt thematic participants (2a), and allow adverbial modification of the underlying speech event, impossible without the overt complementizer (2b): (2) a. Si viene mi madre, tú a ella que el tabaco es tuyo if comes my mother you to her that the tobacco is yours “If my mother shows up, you tell her that the tobacco is yours” b. Si viene mi madre, tú *(que) el tabaco es tuyo, y educadamente/rápidamente if comes my mother you tell her that the tobacco is yours, and politely/quickly “If my mother comes, you tell her that the tobacco is yours, and you do it politely” The structures in (2) raise a question concerning the syntactic status of the tacit verbal structure. There are reasons to think that it is not just an ellided verb say (or any such other verb of speech). The missing verb shows some intriguing restrictions which do not arise in normal cases of ellipsis. Only universal quantifiers can quantify over the speech eventuality in quotative constructions. Any other quantifier is out: (3) a. Tu padre siempre que cuándo vamos a ir a visitarles your father always that when we-are-going to visit them “Your father is always saying: ‘when are you going to visit us?’” b. *Tu padre nunca/rara vez/alguna vez/de vez en cuando que cuándo vamos a ir allí your father never/rarely/sometimes/from time to time that when we-are-going there “Your father never/rarely/sometimes/from time to time says:when are you going there?” None of those restrictions are operative in cases of ellipsis: (4) Tu siempre dices que cuándo venimos pero nunca [dices] que cuándo nos vamos you always says that when we are coming, but you never say that when we are leaving “You always say: ‘when are you coming?’, but you never say:’when are you leaving?’” Furthermore, quotative constructions only admit a subset of the logical operators: they can be conjoined, but they don’t easily admit disjunction (3). (3) Oye, tu padre que no quiere venir y/??o tu madre que quiere venir más tarde hey, your father that neg wants to come and/or your mother that she wants to come later “Your father says:‘I don’t want to go’, and/*or your mother says: I want to go later on” Those restrictions are reminiscent of the ones holding of speech act operators (as argued by Krifka, 2001, 2003). According to Krifka, the illocutionary force of a sentence is

semantically represented by a speech act operator. Krifka shows that certain logical operations, such as disjunction or negation are hardly applicable to speech acts. This is so because the domain of speech acts does not constitute a boolean algebra, but at most a semi-lattice. In such a domain, some operations such as conjunction are well defined, whereas disjunction and negation are not. Universal quantifiers, unlike the rest of the quantifiers, are logically equivalent to generalized conjunctions (Keenan and Stavi, 1985). I therefore conclude that the relevant verbal structure missing in (2) is a tacit verb with an illocutionary feature, akin to speech act operators. Indirect evidence supporting this conclusion comes from selection: unlike normal verbs of saying, the tacit verb in (2) only selects what has been called “speech act” (Suñer, 93), or “quotative” (Lahiri, 2002) dependents. Consider thus (4): (4) a. Pedro dijo quién venía b. Pedro dijo que quién venía Pedro said who was-coming Pedro said that who was coming “Pedro said who was coming” “Pedro said:’who is coming?’ Only in (4b) is the dependent a true question. In (4a), the dependent has no interrogative force, and contributes a set of propositions of the form λx [x was coming]. The tacit verb, unlike normal verbs of saying (cf. decir “say”in (4)), only selects for speech acts: (5) Tu padre *(que) quién viene your father that who is coming (“Your father is saying: ‘who is coming?’”) The tacit verb can also select for finite dependents which look more complex than the speech act dependent in (5). Those finite dependents include the particle si. Si - formally identical to the interrogative complementizer and the conditional particle- introduces features of direct speech. Root forms which are impossible in subordinate clauses, such as the root-imperative form (6), or vocatives (7), become possible following si: (6) Tu siempre que *(si) trabaja más, ya estoy harto you always that si work-imp more, already I am sick-of-it “You are always telling me: ‘work more’. I am sick of it” (7) Éste siempre que *(si) Manolo porqué no me prestas dinero This guy always that si Manolo, why don’t you lend me money “This guy is always saying: Manolo, whay don’t you lend me money?’” The presence of si also licenses what Schlenker (1999, 2003) calls “kaplanian monsters”: indexicals whose antecedents are to be found not in the speech situation but in an embedded constituent (cf. also (7)). Consider for instance the following contrast: (8) a. Tu padre siempre que pobre de mí b. Tu padre siempre que si pobre de mí your father always that poor of me your father always that si poor of me “Your father1 is always saying: poor me2!’” “Y.F.1 is always saying: poor me1/2!’” Whereas in (8a) the only possible referent for the possessive 1st p. pronoun mí “me” is the speaker, in (8b) it can also be tu padre “your father”.Despite its formal similarity to several CP-related heads in Spanish, it can be shown that si in these constructions doesn’t belong in the CP domain: it forces an iterative reading of the speech eventuality (9), and is better analyzed as an aspectual operator in the tacit verbal structure: (9) Tu padre el otro día que si porqué no venís your father the other day that si why don’t you come “Your father was saying continuously the other day: “why don’t you come?” (9) suggests the following important amendment to Schlenker (1999, 2003): rather than taking propositional attitude verbs to be quantifiers over speech contexts, it is the aspectual layer of the heading tacit verb –represented by si- which contributes the relevant quantificational force. The order Comp-si will be shown to derive from head movement of que (plus the tacit verb) to the aspectual head.

Anamaria Falaus (Laboratoire de Linguistique de Nantes)

Saturday, session 11

The double negation puzzle in a strict negative concord language 1. N-words are negative quantifiers In this analysis we argue that n-words in Romanian are negative quantifiers. 2. Negative concord Romanian is a strict negative concord language (NC), as the sentential negative marker is always required in order to license an n-word, regardless of its position in the sentence (1): (1) a. Nimeni *(nu) a venit nobody neg has come

b. *(Nu) stiu nimic neg know.1sg nothing

[NC]

The following generalisation holds in all languages that exhibit strict negative concord: (i) the interpretation of a sentence with the sentential negative marker and an n-word contains a single semantic negation In order to derive (i),n-words have been analyzed as negative polarity items (NPIs) contributing a variable(Ladusaw 1992) or an existential quantifier(Laka 1990) to the semantic representation. 3. N-words are not NPIs However, the analysis of n-words as NPIs predicts they should have an existential interpretation in typical polarity contexts (questions, conditionals, scope of negative predicates, etc.) just like in (2a) for Italian. This prediction is not borne out in Romanian (2b): (2) a. E venuto nessuno ? Has come nobody

(Italian)

b. *A venit nimeni ? Has come nobody

(Romanian)

Moreover, the various diagnostics (scope properties, long-distance licensing of anaphora or topicalisation) that Giannakidou 2002 uses for Greek show that Romanian n-words cannot be analyzed as indefinites or existential quantifiers. Giannakidou concludes that n-words in strict negative concord languages are universal quantifiers without inherent negative meaning. She also makes the following prediction on the interpretation of n-words: (ii) a sentence with the sentential negation and several n-words always has an interpretation containing a single semantic negation We show that this prediction is not valid for Romanian. Contrary to what is generally argued for strict NC languages, I’ll show that in Romanian the double negation reading (DN) is not only possible with two or several n-words, but even obligatory in certain pragmatic (5) or syntactic contexts (8). 4. The double negation puzzle: two or more n-words Contrary to the prediction made by Giannakidou, the double negation reading is available in Romanian with two or several n-words arguments of the same predicate (3-4). Pragmatic factors may sometimes force the reading with double negation as shown in (5): (3) Nimeni nu iubeste pe nimeni. [NC/DN] Nobody neg loves ACC nobody (4) Nimeni nu spune nimic nimǎnui. [NC/DN] Nobody neg says nothing nobody.DAT (5) Nimeni nu moare niciodata [DN preferred] Nobody neg dies never

The facts of Romanian thus give rise to the following paradox: (6) (a) a sentence with sentential negation and an n-word is always interpreted as containing only one semantic negation (it never has a double negation reading ) (b) a sentence with two or more n-words arguments of the same predicate can have a double negation reading 5. Solution – a polyadic approach In order to account for the data, we analyze Romanian n-words as inherently negative quantifiers. The paradox in (6) is then solved by adopting an analysis of n-words in terms of polyadic quantifiers (Swart & Sag 2002). Two derivations are thus possible for a sentence with two or several n-words: iteration and resumption. Iteration consists in separately interpreting 1

each monadic quantifier (binding one variable), which in the case of two negative quantifiers gives rise to a double negation reading (7a). The negative concord reading is derived by an operation of resumption: a sequence of negative quantifiers is reinterpreted as a single polyadic quantifier binding several variables (7b): (7) a. NOx , NOy (LOVE (x,y))

[DN]

b. NOx,y LOVE (x,y)

[NC]

This analysis has two main advantages. The possibility of having two distinct derivations accounts for the ambiguity of a sentence with several negative quantifiers (3-4). Moreover, it accounts for the contrast between (6a) and (6b) in Romanian. Iteration and resumption are only possible with two quantifiers of the same type, more precisely with anti-additives. Sentential negation is analyzed as a negative scope marker. Since it is a propositional operator, it does not bind individual variables (unlike negative quantifiers), Consequently, its co-occurrence with an n-word cannot be interpreted as a double negation. Thus, sentences like those in (1) are correctly predicted to have only a NC reading. Moreover, Romanian offers interesting empirical arguments in favour of an approach which treats n-words as negative quantifiers interpreted separately by an operation of iteration. Intonation favours a double negation reading (9) (as shown in Corblin & Tovena 2003 for French), whereas the topicalisation forces this reading (8b): (8) a. Maria nu a scris nimǎnui nici o scrisoare. Maria neg has written nobody-DAT no letter. b. Nici o scrisoare, Maria nu a scris nimǎnui No letter Maria neg has written nobody-DAT (9) NIMENI nu vorbeste cu nimeni Nobody neg talks with nobody

[NC/DN] [DN only] [DN favoured]

Both in (8b) and in (9), the first n-word is separated (either syntactically or by a special intonation contour) from the rest of the sentence. Consequently, the two n-words are interpreted as a sequence of two monadic negative quantifiers, yielding a double negation reading. 6. Morphology and diachrony The availability of DN readings represents a serious challenge for any theory that takes n-words in strict NC languages to be non-negative. Our analysis of n-words as negative quantifiers is further supported by the negative morphology (the negative prefix ni- appears in all n-words) and the evolution in diachrony. The distribution of n-words in Old Romanian (10) is parallel to the one in contemporary Spanish or Italian (negative quantifiers in preverbal position only): b. sa nu spui nemunuia that neg say.2sg nobody-DAT ‘don’t tell to anyone’

(10) a. nimea amu sa se apropie nobody now that approaches ‘that nobody approaches now'

Romanian later became a strict negative concord language. The evolution of n-words in Romanian is thus the mirror image of that of Spanish, as described by Herburger 2001, although both come from Latin n-words which were clearly negative quantifiers. Thus, just like contemporary Romanian, former Spanish was a strict negative concord language (11). In contemporary Spanish, n-words are negative quantifiers in preverbal position and are also used as NPIs in polarity contexts with an existential reading: (11) que a myo Cid Ruy Diaz, que nadi no diessen posada. that with my Lord Ruy Diaz, that nobody neg gives lodging

(Old Spanish)

The crucial difference between languages like Spanish or Italian on the one hand and Romanian on the other hand is the availability of a non-negative existential reading. The fact that n-words never had an existential reading in Romanian is a strong argument in favour of our hypothesis on their inherent negative meaning. REFERENCES: CORBLIN, F. & TOVENA , L., 2003, «L’expression de la negation dans les langues romanes », D. Godard (éds), p. 281-343; GIANNAKIDOU, A., 2002, « N-words and Negative Concord », The Linguistics Companion, Blackwell, Oxford.; HERBURGER, E., 2001, « The Negative Concord Puzzle Revisited », Natural Language Semantics 9 : 289-333; LADUSAW, W., 1992, « Expressing Negation », Semantics and Linguistic Theory (SALT) II : 237-259, Cornell University, Ithaca; LAKA, I., 1990, Negation in Syntax : On the nature of functional categories and projections, PhD dissertation, MIT; De SWART, H. & SAG, F., 2002, « Negation and Negative Concord in Romance », Linguistics & Philosophy 25 : 373-417.

2

Saturday, session 18

Verb Movement and Phase-Sliding Ángel J. Gallego (Univ. Autònoma de Barcelona)

A central issue being investigated within the Minimalist Program is whether (and how) access to the Lexicon must be restricted in some manner. Under standard assumptions, this process is supposed to unfold phase by phase, with ‘small’ subarrays of Lexical Items placed in active memory (cf. [3],[4]). This view immediately raises the question of what the metrics to establish such sequential accesses are. Chomsky has provided both conceptual and interface/output motivations supporting the claim that v*P and CP constitute the strong phases, and extending conclusions from previous work, in Chomsky (2005), he claims that all the relevant features are generated in the phase heads. In the same breath, Chomsky (2005) argues that A-/A’-movements are now distinguished by the triggering head: only phase heads launch A’-movement by means of what Chomsky (2005) calls edge feature (presumably the current version of the “generalized EPP” of Chomsky (2000)); any others are A-movements, driven by φ-features, which appear in T and V derivatively, as a process of inheritance. Interestingly, only phase heads bear edge features, these being an optional device to yield surface-semantics (i.e., specificity, presupposition, etc.) or cyclicity through phase edges (i.e., the head and SPECs of C and v*; cf. [4]). The fact that edge features are optional in phase heads is explicitly pointed out in Chomsky (2000): (1) The head H of a [strong] phase Ph may be assigned an EPP-feature.

[from Chomsky (2000:109)]

This said, one might wonder whether inheritance can feed parametric variation. Let me elaborate. Chomsky (2005) actually considers two different scenarios: either inheritance of φ-features by T and V is a device of UG (perhaps parametrized) or else it constitutes an optimal solution to capture the A/A’ distinction. Chomsky (2005) follows the second route (the strongest one), suggesting that the SEM component may require distinguishing Case/Agreement systems from peripheral ones through inheritance. Be that as it may, it is tempting to explore the first possibility. To be specific, suppose that T can be endowed not only with φ-features, but also with edge features in the relevant languages. Consider, therefore, a parameter roughly as indicated in (2): (2) Parameter: T is endowed with edge features. Here I would like to argue that (2) is marked positively in Null Subject Languages (NSLs), a move that virtually turns T into a strong phase head in those languages (cf. (1)). However, given that such a possibility is conceptually problematic (only C and v* are the strong phase heads; cf. [3],[4],[5]), I propose that the phase effects manifested in T, though pervasive and robust, can be regarded as a side effect of v*-to-T movement: when internally merged to T, v* re-labels the whole structure, forcing a species of Reprojection (cf. [6],[7]), to which I will refer here as phase-sliding (metaphorically, it is as if v*’s movement pushed the v*P phase up to the TP level). The analysis clearly revamps Chomsky’s (1986) idea that V-T amalgamation can freed the VP of its “barrierhood”, and, if correct, refutes the empirical phenomena threatening the phase status of v* in NSLs, for all the operations that appear to take place at the T level, actually occur at an “extended v*” level. The proposal is interesting, at least in two respects: first, it trivially derives the hybrid A/A’ nature that SPEC-T has been said to have (cf. [10]); and, second, it still allows us to say that v* and C are the strong phase heads universally, for T acts as a phase just derivatively, due to the hybrid label it forms with v* (in other words, if v* does not merge with T, T cannot display phase-like properties). Consequently, notice that T seems to exist just as a feature holder (of both edge and φ-features) feeding parametric variation. This phase-sliding proposal makes immediate predictions. Consider first intervention effects in VOS configurations, taking the object to be in an outer-SPEC-v* (cf. [8]). If we assume that movement of the object to SPEC-v* and nominative case assignment to the subject by T operate in different phases (say, v*P and CP –Chomsky’s system), we predict that a sentence like (3) must show intervention effects. Intervention, however, does not obtain, and the sentence is fine. The conclusion, then, reinforces the hypothesis defended here: all operations (both object raising and nominative case assignment) take place at the first strong phase level (i.e., v*/TP): (3) [CP [TP Mercouz T[uφ] [v*P [o coche][iφ]k [v*P Xoán[iφ] [VP tz tk ]]]] Bought-3SG the car Xoán ‘Xoán bought the car’

(Galician)

A second prediction concerns subextraction. As Chomsky (2005) shows, in English, the Subject Condition affects SPEC-v* –a phase edge-, not SPEC-T (for otherwise the contrast in (4) would not be explained). If what I have argued for is right, the prediction is clear: in NSLs, subextraction should be fine from subjects in SPEC-v* (i.e., postverbal ones), since that position no longer qualifies as a phase edge after v*-to-T movement, SPEC-v*/T does instead (cf. (5)): (4) a. [CP Of which carj was [TP [the driver tj]z [vP awarded tz a prize]]]? b. *[CP Of which carj did [TP [the driver tj]z [v*P tz cause a scandal]]]? (5) a. ??[CP De qué universidadj dices que [TP [muchos estudiantes tj]z protestaroni [v*P tz ti ]]]? Of which university say-2SG that many students protested-3PL ‘Of which university do you say that many students protested?’ b. [CP De qué universidadj dices que [TP protestaroni [v*P [muchos estudiantes tj] ti]]]? Of which university say-2SG that protested-3PL many students ‘Of which university do you say that many students protested?’

(English) (English) (Spanish) (Spanish)

It is possible to draw a connection between this proposal and more remarkable facts of NSLs if we endorse Pesetsky & Torrego’s (2004) claim that ‘case’ is an uninterpretable tense feature [uT]. In Pesetsky & Torrego (2004), C is also endowed with [uT], which must be deleted by a pure T head (that, in (6a)) or the subject DP (cf. (6b)), crucially assuming that the case feature of such DP can participate in checking processes within the phase it has been “marked for deletion”, something which is feasible if ‘valuation of features’ eliminates the interpretable/uninterpretable distinction for a short period of time (specifically, until a strong phase level is reached; cf. [4]): (6) a. John said [CP that[iT]k C[uT, EPP] [TP Mary Tk called him ]] b. John said [CP Mary[uT]j C[uT, EPP] [TP tj called him ]]

(English) (English)

Importantly for our concerns, if v*/TP is a strong phase in NSLs, we expect that the [uT] of subject DPs will never remain ‘computationally alive’ to check C’s [uT]. The prediction is borne out, as these subjects never trigger that-trace effects (cf. (7)), and that-deletion is almost barred (cf. (8)): (7) [CP Chij credi[iT]k C[uT,EPP] [EF] Tk pro [CP tj chez C[uT,EPP] [TP tj Tz ha parlato]]]]? Who think-2SG that have-3SG talked ‘Who do you think has talked?’ (8) *En Joan diu [CP C[uT,EPP] [TP la Maria no ha vingut]] The Joan say-3SG the Maria not have-3SG come ‘Joan says Maria has not come’

(Italian) (Catalan)

Synthesizing, the logic of the proposal outlined here argues that parameters can be associated with whether or not (functional) heads are endowed with edge or φ-features (cf.[1],[4],[5]). Viewed that way, parametric variation boils down to feature assembling into Lexical Items, reconciling a longstanding insight about linguistic universals –and their parameters- with the worthy goal (embodied in the inquiry undertaken by the Minimalist Program) of accounting for properties of the Faculty of Language in a principled way (i.e., reducing them to interface conditions or conceptual necessities). References [1] Borer, H. (1984): Parametric Syntax: Case Studies in Semitic and Romance Languages, Dordrecht:Foris. [2] Chomsky, N. (1986): Barriers, Cambridge (Mass.):MIT Press. [3] Chomsky, N. (2000): “Minimalist Inquiries: The Framework”, in Martin, Michaels, & Uriagereka, Step by Step. Essays on Minimalist Syntax in Honour of H. Lasnik, Cambridge (Mass.):MIT Press,89-155. [4] Chomsky, N. (2001): “Derivation by Phase”, in Kenstowicz (ed.), Ken Hale: A Life in Language, Cambridge (Mass.):MIT Press,1-52. [5] Chomsky, N. (2005): “On Phases”, Ms.,MIT. [6] Donati, C. (2000): La sintassi della comparazione, Padova:Unipress. [7] Hornstein, N. & J. Uriagereka (2002): “Reprojections”, in Epstein & Seely (eds.), Derivation and Explanation in the Minimalist Program, Malden (Mass.):Blackwell,106-132. [8] Ordóñez, F. (1998): “Postverbal Asymmetries in Spanish”, NLLT,16:313-346. [9] Pesetsky, D. & E. Torrego (2004): “Tense, Case, and the Nature of Syntactic Categories”, in Guéron & Lecarme (eds.), The Syntax of Time, Cambridge (Mass.):MIT Press,495-537. [10] Uribe-Etxebarria, M. (1992): “On the Structural Positions of the Subject in Spanish, their Nature and their Consequences for Quantification”, in Lakarra & Ortiz de Urbina (eds.), Syntactic Theory and Basque Syntax, San Sebastian:ASJU,447-491.

Ángel J. Gallego (Univ. Autònoma de Barcelona) and Juan Uriagereka (Univ. of Maryland and U. del País Vasco)

Saturday, session 14

Sub-extraction from Subjects Since Huang’s (1982) seminal work, it is a well-known fact that subjects and adjuncts constitute opaque domains to sub-extraction (cf. (1)). Different proposals have tried to account for this fact, either capitalizing on PF-based constraints like Kayne’s (1994) LCA (cf. Uriagereka (1999)), or assuming a representational constraint along the lines of Chain Uniformity (cf. Boeckx (2003), Stepanov (2001), Takahashi (1994), i.a.). Each type of approach makes different predictions for the subject case: in particular, while for the first type the problem arises the minute a complex specifier is created (SPEC-v* being the first candidate), for the second one only sub-extraction from derived specifiers matters, since only those pose chain-uniformity concerns. Chomsky (2005) has reassessed this issue, arguing that sub-extraction from subjects is blocked in the base position (i.e., SPEC-v*, a strong phase edge), and not the surface position (i.e., SPEC-T), for otherwise the contrast in (2) would remain unexplained. As Chomsky himself points out, an immediate prediction is that sub-extraction from SPEC-C (another strong phase edge) should be as degraded as sub-extraction from SPEC-v*; that prediction is borne out, at least according to data of the sort in (3), from Lasnik & Saito (1992). Although Chomsky (2005) does not provide a straightforward explication of this state of affairs (he merely speculates that sub-extraction from strong phase edges raises locality problems), the parallelism is interesting in itself. That said, the minimal pair in (4), originally noted by Torrego in 1985, casts some doubt on the whole approach: in Spanish, it would appear, sub-extraction from SPEC-T is worse than sub-extraction from SPEC-C. To complicate things further, Uriagereka (1988) noted that subextraction from a postverbal subject (which we take to be in SPEC-v*) is much better than from a preverbal one –arguably in SPEC-T, as in Torrego’s (1985) examples- (cf. (5)). Here we want to argue that, despite initial appearances, Spanish data can be accommodated into Chomsky’s “phase theory”. Consider first the subject case; what we must find out is why sub-extraction from postverbal subjects is fine even if they constitute a phase edge. Our solution relies on v*-to-T movement: we propose that verb movement in languages like Spanish is a bona fide instance of syntactic movement (contra Chomsky (2001)) that forces a reprojection at the TP level, with both labels coexisting (cf. Chomsky (2005) and Hornstein & Uriagereka (2002)). Technically, we claim that the verb ‘pied-pipes its phasehood’ to TP so that v*P is not a phase domain any longer, a process we refer to as phase-sliding (the device is clearly equivalent to Chomsky’s (1986) idea that V-to-I movement, where it obtains, frees VP from its ‘barrierhood’). The analysis is reinforced by the simple possibility of having VOS sequences in Spanish, which, following Ordóñez (1998), we analyze as involving scrambling of the object to an outer-SPEC-v*: if v*P were a phase in Spanish, nominative case assignment should be blocked by the object (cf. (6)); if, however, v*/TP is, both operations (object raising and nominative case assignment) can take place in parallel within that phase boundary. The logic of the proposal predicts that extracting from the true phase edge in Spanish (under our account, SPEC-v*/T) should be deviant –and it is, as shown in (4a). Consider next sub-extraction from SPEC-C; although we agree with Torrego’s (1985) judgments, we argue that trying to subextract here is actually as bad in Spanish as it is in English: the key for sub-extraction from SPEC-C to be possible –we argue– is the presence of negation, which ‘rescues’ sub-extraction by forcing a presuppositional (de re) reading of the wh-phrase undergoing movement. Notice that, when negation is dropped, sub-extraction is out (7a), as expected. In sum, the sketched analysis argues that sub-extraction from what Chomsky (2005) calls phase edge runs into locality problems in all languages, with apparent counterexamples emerging from independent language-specific factors (i.e., parameters). One such parameter (the crucial one, for us) rests on v*-to-T movement, which, if our account is on track, should not be cornered to the PF component: it has non-trivial computational consequences. Recall, to conclude, that neither of the theories proposed to predict sub-extractions works for (2); while this is radically so for the Chain Uniformity approach, the PF-based theory could be made compatible with the data if it is assumed that linearization (hence freezing) becomes definitive only at phase edges.

(1) a. ?*[CP Whoi did [TP [a friend of ti ]j [v*P tj admire a picture of Aristotle]]]? Subject Condition b. ?*[CP Whoi did [TP Maryj [v*P tj cry [afterP after Peter hit ti ]]]]? Adjunct Condition [from Stepanov (2001)] (2) a. ?*[CP Of which cari did [TP [the driver ti ]j [v*P tj cause a scandal]]]? b. [CP Of which cari was [TP [the driver ti ]j [vP awarded tj a prize]]]? [from Chomsky (2005)] (3) a. ?*[CP Whoi do you wonder [CP [which picture of ti ]j [TP Maryz [v*P tz bought tj]]]]? b. ?*[CP Whoi do you wonder [CP [which picture of ti ]j [TP tj [vP is tj on sale]]]]? [from Lasnik & Saito (1992)] (4) Spanish a.??Esta es la autora [CP[de la que]i [TP[varias traducciones ti]j han ganado [v*P tj premios inter.]]] This is-3SG the author of the that several translations have-3PL won awards international ‘This is the author whom several translations by have won international awards’ b. [CP De qué autorai no sabes [CP [qué traducciones ti ]j [TP tj están [vP tj a la venta]]]? Of what author not know-2SG what translations are-3PL to the sale ‘Which author don’t you know which translations by are on sale?’ [from Torrego (1985)] (5) Spanish a. [CPDe qué conferenciantesi te parece que [TP me van a impresionar [v*P [las propuestas ti] ]]]? Of what speakers CL-to-you seems-3SG that CL-me go-3PL to to-impress the proposals ‘Which speakers does it seem to you that the proposals by will impress me?’ b.* [CPDe qué conferenciantesi te parece que [TP[las propuestas ti]j me van a impresionar [v*P tj]]]? Of what speakers CL-to-you seems-3SG that the proposals CL-me go-3PL to to-impress ‘Which speakers does it seem to you that the proposals by will impress me?’ [from Uriagereka (1988)] (6) [CP C [TP compróvT[uφ] [v*P el cochej [iφ] [v*P Isabel[iφ] [VP tv tj ]]]] (Spanish) _________________
____________↑

(7) Spanish a. [CP De qué directori no ha dicho María [CP [qué películas ti ] j tj han ganado un Óscar]]? Of what director not has-3SG said María what films have-3PL won an Oscar ‘Which director hasn’t María said which films by have won an Oscar?’ b. ??/* [CP De qué directori ha dicho María [CP [qué películas ti ]j tj han ganado un Óscar]]? Of what director has-3SG said María what films have-3PL won an Oscar ‘Which director has María said which films by have won an Oscar?’ References: Boeckx, C. (2003): Islands and Chains, Amsterdam: John Benjamins. Chomsky, N. (1986): Barriers, Cambridge (Mass.): MIT Press. Chomsky, N. (2001): “Derivation by Phase”, in Kenstowicz (ed.), Ken Hale: A Life in Language, Cambridge (Mass.): MIT Press, 1-52. Chomsky, N. (2005): “On Phases”, Ms. MIT. Hornstein, N. and J. Uriagereka (2002): “Reprojections”, in Epstein & Seely (eds.), Derivation and Explanation in the Minimalist Program, London: Blackwell, 106-132. Huang, J. (1982): Logical Relations in Chinese and the Theory of Grammar, Ph.D. dissertation, MIT. Kayne, R. (1994): The Antisymmetry of Syntax, Cambridge (Mass.): MIT Press. Lasnik, H. & M. Saito (1992): Move α, Cambridge (Mass.): MIT Press. Ordóñez, F. (1998): “Post-verbal asymmetries in Spanish”, NLLT, 16, 313-346. Stepanov, A. (2001): Cyclic Domains in Syntactic Theory, Ph.D. dissertation, U.Conn. Takahashi, D. (1994): Minimality of Movement, Ph.D. dissertation, U.Conn. Torrego, E. (1985): “On Empty Categories in Nominals”, Ms. U.Mass Boston. Uriagereka, J. (1988): On Government, Ph.D. dissertation, U.Conn. Uriagereka, J. (1999): “Multiple Spell-out”, in Epstein & Hornstein (eds.), Working Minimalism, Cambridge (Mass.): MIT Press.

Grant Goodall (University of California, San Diego)

Friday, session 2

Inversion in wh-questions in child Romance and child English: The roles of syntax and processing 0. The problem Wh-questions in both English and Romance generally require inversion, i.e. the verb (or auxiliary) must appear to the left of the subject: (1) What will John buy? (2) Qué compró Juan? (Spanish) ‘What did Juan buy?’ This similarity has often been captured by universally requiring T-to-C movement in wh-clauses (e.g. Rizzi 1996). This neat parallelism between the two languages breaks down, however, when one looks at child English and child Romance, where the following significant differences in inversion show up. English (Stromswold (1990))

Romance (Grinstead (2001), Guasti (2000), Serrat & Capdevila (2001), Soares (2003))

Very wide. Children overall correctly Individual invert in wh-questions 93% of the variation time, but individuals range from 60.1% to 99.3%.

None. All children correctly invert 100% of the time.

Many children show a long period of All children show adult pattern of Course of erratic use of inversion before settling inversion from the very beginning. development on adult pattern. Although these differences have been noted in the literature, there as yet exists no satisfying explanation. Here I will attempt to show that if we adopt for Romance the analysis of inversion in Goodall (2004), in which inversion results from both syntactic and processing factors, the above differences follow straightforwardly. I will focus here on Spanish, though the approach is extendable to other Romance languages. 1. Analysis of inversion The core idea of Goodall (2004) is that the syntax allows both (3a) and (b) (the subject may raise to SPEC/T (a) or stay in situ (b)), but that (3a) is excluded for processing reasons. (3)a. *Qué Juan compró? b. Qué compró Juan? ‘What did Juan buy?’ In both (a) and (b), the wh-phrase/filler must be held in working memory until a gap can be posited, which occurs when the subcategorizing verb (compró) is reached. This is trivially easy in (b) (compró follows qué directly), but in (a) the subject intervenes. This subject is preverbal and overt, and as such, it carries a marked discourse value (e.g. Byrne (1998), Casielles-Suárez (1999), Contreras (1976), Zubizarreta (1998)), which places greater demands on the processor than a postverbal or null subject would (Kluender (1998), Frazier & Clifton (2002), Warren & Gibson (2002)). These demands are great enough that they appear to interfere significantly with the processing of the filler-gap dependency, and the sentence is thus perceived as unacceptable. This model explains a number of key facts about inversion in Spanish. First, the verbal complex to the left of the subject must contain a lexical verb, not just an auxiliary (4) (only the lexical verb subcategorizes for the gap and thus resolves the filler-gap dependency). Second, the prohibition against an intervening subject is relaxed when the wh-phrase is D-linked (5) (D-linked fillers are known to tolerate a longer delay in finding the subcategorizing head (Kluender 1998)). Third, this prohibition is

also relaxed when the wh-phrase is why (6), since this type of filler has no corresponding gap (Rizzi 1996). Fourth, the contrast in (3) is found in both matrix and embedded questions (7) (the processing problem of pairing the filler with a subcategorizing head is the same in both). (4)a. *A quién había la madre de Juan visto? ‘Who had Juan’s mother seen?’ b. A quién había visto la madre de Juan? (Ordóñez 1997) (5) Cuáles de esos libros Juan compró? ‘Which of those books did Juan buy?’ (6) Por qué Juan trabaja tanto? ‘Why does Juan work so much?’ (7)a. *No sé [qué Juan compró] ‘I don’t know what John bought.’ b. No sé [qué compró Juan] Unlike in Spanish, overt preverbal subjects in English are the norm and do not have a special discourse value, so we do not expect them to unduly interfere in the processing of a filler-gap dependency. This is seen in (1), where John intervenes between the wh-phrase and the subcategorizing verb. The inversion effect in (1) must then be the result of some other mechanism, presumably the syntactic process of Tto-C movement. As expected, then, English shows none of the properties in (4)-(7). T-to-C movement must clearly be allowed by UG under this account (English uses it), but not required (Spanish does not use it). 2. Solving the problem Given this analysis, the child learning English must specifically acquire T-to-C movement. Since this is not mandated by UG, it is not surprising that children take time to learn to do this and show significant individual variation. In Romance, in contrast, there is no T-to-C movement to acquire. Instead, the child does not produce the non-inverted pattern (3a) for the same reason that the adult does not: the intervening preverbal subject makes it too difficult to process the filler-gap dependency. There is, in a sense, nothing to learn, so all children produce the inversion pattern consistently from the beginning. This analysis of the difference in inversion between child English and child Spanish rests on two important assumptions: (i) that a preverbal overt subject is similar in discourse status in both child and adult Spanish (and thus interfers with processing a filler-gap dependency), and (ii) that children are no better at processing filler-gap dependencies than adults (and thus find (3a) unacceptably difficult to process). Both assumptions are very plausible. With regard to (i), substantial evidence now suggests that children converge very early on the fact that Spanish is a null subject language and preverbal subjects are marked (Bel (2003), Grinstead (2004), Valian (1991)). With regard to (ii), children give no sign of being better than adults at processing and crucially appear to be worse at processing subjects (Kluender (2004), Bloom (1990, 1993)). 3. Conclusions By taking seriously the roles of both syntax and processing, we can thus begin to explain the otherwise mysterious differences between child English and child Spanish seen above. As time permits, extensions of the analysis to Romance languages beyond Spanish will be discussed. Selected references: Bel, Aurora (2003), “The syntax of subjects in the acquisition of Spanish and Catalan” Probus 15, 1-26. Goodall, Grant (2004) “On the Syntax and Processing of Wh-questions in Spanish” WCCFL 23 Proceedings. Grinstead, John (2001) Wh-movement in child Catalan. Issues in Applied Linguistics 12, 5-28. Guasti, M.T. (2000) “An excursion into interrogatives in early English and Italian.” In Friedemann and Rizzi (eds.), The Acquisition of Syntax. Stromswold, K. (1990) Learnability and the Acquisition of Auxiliaries. Unpublished Ph.D. dissertation, MIT.

Theres Grüter (McGill University)

Friday, session 4

When learners know more than linguists: direct object clitics are not pronouns It is the general view, in both traditional grammars of French as well as the majority of contemporary linguistic approaches, that direct object clitics (as in (1)) are pronouns (Grevisse 1993, Cardinaletti & Starke 1999, inter alia). Within the long-standing debate on clitic constructions, it is often argued that object clitics differ from, for example, English object pronouns with regard to morphological status (clitic/weak pronoun vs. strong pronoun/DP), and syntactic derivation. Yet it is also generally assumed – particularly in the field of language acquisition – that French object clitics and English object pronouns do not differ in terms of argument status; both are taken to represent the internal argument of the verb. (1)

Il le met dans le bain he it puts in the bath ‘He puts it in the bath.’

In this paper, I will argue that this last assumption is wrong for French (and perhaps Romance in general), drawing on evidence from first, second and atypical (S.L.I.) language acquisition. I will show that the developmental data are best explained under a syntactic approach which assumes a null pronominal (pro) in the complement of V bearing the relevant theta-role, and classifies the clitic as strictly inflectional material (e.g., Roberge 1990, Sportiche 1996). In consequence, I will argue that neither strict movement approaches (e.g., Kayne 1975, Belletti 1999) nor strict lexicalist accounts (Miller and Sag 1997) – both entailing the argumental status of the clitic – are capable of integrating the developmental facts observed in various learner populations. It has been noted at least since Clark (1985: 714) that object clitics are “a fairly late acquisition” in the Romance languages. Numerous proposals have been made concerning the source of this particular difficulty with the acquisition of clitic constructions, yet this lively debate is far from resolved today (cf. Hamann 2003). In this paper, I will argue that there is one crucial factor that has not been given enough weight in this debate. This is the empirical observation that in contexts where, by adult standards, an object clitic might be expected, learners (of French) regularly produce utterances without an overtly realized direct object, as illustrated in (2). (For L1 acquisition, see e.g., Jakubowicz et al. 1996, Pérez-Leroux et al. 2005; for L2 and SLI, see Paradis 2004, among many others.) (2)

a.

il met Ø dans le bain he puts in the bath ‘He is putting it in the bath.’

(Lou (2;5), Jakubowicz et al. 1996)

Analyses attributing argument status to the clitic must consider (2) ungrammatical due to illicit argument omission, i.e., violations of (at least) the Projection Principle and the Theta Criterion. The question that must be raised is why learners should ever produce utterances like this, violating fundamental principles of the grammar, especially given that a grammatical alternative – the use of a lexical DP object – would have been easily available. Moreover, the omission of obligatory internal arguments is not otherwise characteristic of language development, in any population or language. Thus an explanation based on difficulties with the realization of argument structure is doomed on crosslinguistic grounds, and the question remains open. Under syntactic analyses which posit a null pronominal (pro) as the bearer of the verb’s internal theta role in object clitic constructions, (2) violates neither the Projection Principle nor the Theta Criterion. Instead, (2) must be seen as ungrammatical only due to missing inflection. Missing functional and inflectional material is a hallmark of learner language, in any population

and context. Under a theoretical view of clitics as inflection, the phenomenon of clitic omission in French learner language can be seen as similar in nature to such well-attested developmental phenomena as missing tense and agreement marking (see e.g., Wexler 1994, Pierce 1992). Not only is this latter approach superior in aligning clitic omission with other characteristic properties of learner language, but it also provides a straightforward explanation for the categorial absence of an error type that may well be expected under an argumental analysis of clitics, namely clitic misplacement, as shown in (3). (3)

*Il met le dans le bain

(not attested)

Errors of this type are (to the best of my knowledge) not attested in L1 acquisition, and have been observed only occasionally in early stages of L2 acquisition. If the clitic is indeed basegenerated in this position (Kayne 1975, Belletti 1999), learners could be expected to produce utterances such as (3) as a result of an occasional failure to apply movement (analogous, for example, to the apparent optionality of verb raising observed in learner language). Yet utterances like (3) do not occur. Thus it appears that learners know that a clitic can at no stage occupy the position of the internal argument. More precisely, I argue, they know that clitics are simply not arguments – as proposed in the syntactic accounts of Roberge (1990) and Sportiche (1996). References Belletti, A. (1999). Italian/Romance clitics: Structure and derivation. In H. van Riemsdijk (Ed.), Clitics in the languages of Europe (pp. 543-579). Berlin: Mouton de Gruyter. Cardinaletti, A., & Starke, M. (1999). The typology of structural deficiency: A case study of the three classes of pronouns. In H. Van Riemsdijk (Ed.), Clitics in the languages of Europe (pp. 145-233). Berlin: Mouton de Gruyter. Clark, E. (1985). The acquisition of Romance with special reference to French. In D. I. Slobin (Ed.), The crosslinguistic study of language acquisition (pp. 688-782). Hillsdale: Erlbaum. Grevisse, M. (1993). Le bon usage – grammaire française (13th ed.). Paris: Duculot. Hamann, C. (2003). Phenomena in French normal and impaired language acquisition and their implications for hypotheses on language development. Probus, 15, 91-122. Jakubowicz, C., Müller, N., Kang, O.-K., Riemer, B., & Rigaut, C. (1996). On the acquisition of the pronominal system in French and German. In A. Stringfellow, D. Cahana-Amitay, E. Hughes & A. Zukowski (Eds.), Proceedings of the 20th BUCLD (pp. 374-385). Somerville, MA: Cascadilla Press. Kayne, R. (1975). French syntax: The transformational cycle. Cambridge, MA: MIT Press. Miller, P. H., & Sag, I. A. (1997). French clitic movement without clitics or movement. Natural Language and Linguistic Theory, 15, 573-639. Paradis, J. (2004). The relevance of specific language impairment in understanding the role of transfer in second language acquisition. Applied Psycholinguistics, 25, 67-82. Pérez-Leroux, A., Pirvulescu, M., & Roberge, Y. (2005). Early object omission in child French and English, Paper presented at LSRL 35. Austin, Texas. Pierce, A. (1992). Language acquisition and syntactic theory. Dordrecht: Kluwer. Roberge, Y. (1990). The syntactic recoverability of null arguments. Kingston & Montreal: McGill-Queen’s University Press. Sportiche, D. (1996). Clitic constructions. In J. Rooryck & L. Zaring (Eds.), Phrase structure and the lexicon (pp. 213-276). Dordrecht: Kluwer. Wexler, K. (1994). Optional infinitives, head movement and economy of derivation. In N. Hornstein & D. Lightfoot (Eds.), Verb movement (pp. 305-350). Cambridge: CUP.

Javier Gutiérrez-Rexach (Ohio State University)

Friday, session 7

STRUCTURE AND INTERPRETATION OF DEGREE CORRELATIVES It has been stated in the literature (Keenan 1985, Srivastav 1991, Bhatt 2003) that correlativization is a strategy mostly limited to Indo-Aryan languages, such as the Hindi example in (1), possibly related to head-finality. (1)

Hindi: Jo CD sale-par hai, Maya us CD-ko khari:d-egi (lit. ‘Which CD is on sale , Maya will buy that CD’)

In this paper it is argued that Romance languages, more concretely Spanish, also have correlative structures, restricted to relativization on degrees, illustrated in (2): (2)

a. Cuanto más lo miro, (tanto) más me gusta. ‘The more I look at it, the more I like it.’ b. Cuanto menos bebo, menos me enfado. ‘The less I drink, the less angry I get.’

The degree correlative construction can take several shapes: (i) normally a strict repetition of the comparative degree proform (más ...más, menos ... menos), but also (ii) a cross-polar correlation (más ... menos)(Kennedy 1997) or (iii) one with synthetic comparatives (peor): (3)

a. Cuanto menos trabajo, más disfruto. ‘The less I work, the more I enjoy myself.’ b. Cuanto menos trabajo, peor me siento. ‘The less I work, the worse I feel.’

Further evidence for the hypothesis that this structure is an instance of (single headed) correlativization is provided by a host of facts related to matching or identity of degrees: (i) the absence of the degree proform renders the structure ungrammatical (4); (ii) only the demonstrative tanto is allowed as a quantifier restricting the proform, whereas other degree quantifiers (tres veces, muchisimo más ) are blocked (5), given that they would introduce a non-matching correlation; etc.: (4) (5)

*Cuanto más lo miro, me gusta. ‘The more I look at it, the more I like it.’ a. *Cuanto más lo miro, tres veces más me gusta. ‘*The more I look at it, three times more I like it.’ b. *Cuanto menos trabajo, muchísimo más me canso. ‘The less I work, much-SUPERLAT I get tired.’

Following Srivastav (1991), degree relatives are syntactically treated as an instance of the structure in (6), where a free relative is coindexed with a pronoun. (6)

[ [ Free relative ] .... pronoun ...]

Several arguments are presented to elucidate between two competing analyses of correlatives, as applied to Spanish degree correlatives: a base-generated adjunction analysis (Srivastav 1991) and a movement analysis (Bhatt 2003, Pancheva and Bhatt 2004). It is argued that the latter option is preferable and that the degree proform raises to a DegP projection (Corver 1997, Neeleman et al. 2004) to check the degree feature and derive the matching requirement (Groos & van Riemsdijk 1979, Harbert 1983, Suñer 1984, etc). Evidence for the rasing analysis comes, among other facts, from positional (7) and island-violation (8) data.

(7)

(8)

a. Cuanto más leo, más aprendo. ‘The more I read, the more I learn.’ how-much more read-I more learn-I b. ??Cuanto más leo, aprendo más. how-much more read-I learn-I more c. ??Cuanto leo más, más aprendo how-much more read-I more learn-I a. Cuanto más leo, más creo que aprendo. ‘The more I read, the more I believe I learn.’ b. *Cuanto más leo, más reconozco el hecho de que aprendo. ‘*The more I read, the more I see the fact that I learn.’

From a semantic point of view, degree correlatives can be argued to be similar to comparative conditionals (McCawley 1988, Beck 1997) -- cf. The more you read the more you learn. The degree correlative behaves as a conditional tripartite structure headed by a quantifier on degrees. This quantifier is claimed to be an unselective binder, and simultaneously bind the degree variable in the free relative and the variable corresponding to the degree proform in the main clause. Thus, it is not possible for one of the variables to be bound by a comparative operator: (9)

*Cuanto más cansado está que Pedro, más grita. ‘*The more tired he is than Pedro, the more he shouts.’

Other semantic issues are also explored: the role of the proform and its relationship to conditional then (Iatridou 1991, Izvorski 1998); the emergence of several readings (generic, universal, and quasi-universal); and the role of other quantificational and scopal restrictions. Finally the absence in Romance of nominal correlativization and of multi-headed correlatives, common in South-Slavic and Indo-Aryan languages, is addressed in terms of an interface restriction on reconstruction (Fox 2000, Fox and Nissenbaum 2004), related to the raising analysis of relatives (Kayne 1994) and the semantics of degrees (Heim 1996).. References Beck, S. (1997) “On the semantics of comparative conditionals”, Linguistics & Philosophy 20, 1997. Bhatt, R. (2003) “Locality in correlatives”, Natural Language & Linguistic Theory. Corver, N. (1997) “The internal syntax of the Dutch extended adjectival prohection”, NLLT. Fox, D. (2000) Economy and Semantic Interpretation, MIT press. Fox, D & J Nissenbaum (2004) “Condition A and scope reconstruction”, Linguistic Inquiry Groos & van Riemsdijk (1978) “Matching effects in free relatives” Proceedings GLOW 4. Harbert, W. (1983) “On the nature of the matching parameter” The Linguistic Review. Heim, I. (1996) “On the logical syntax of degree operators”, ms., MIT. Iatridou, S. (1991) Topics in Conditionals, PhD diss, MIT Izvorsky, R. (1997) “Syntax and semantics of correlative proforms”, Proceedings NELS. Kayne, R. (1994) The Antisymmetry of Syntax, MIT Press. Keeenan, E. (1985) “Relative clauses”, in T. Shopen (ed.) Language Typology and Syntactic Description, Kennedy, C. (1997) Projecting the Adjective, PhD diss, UC Santa Cruz McCawley (1988) “The comparative conditional construction in English, German and Chinese”, Proc. BLS. Neeleman, A. et al. (2004) “Degree expressions” The Linguistic Review Pancheva, R. and R. Bhatt (2004) “Late merger of degree clauses”, Linguistic Inquiry. Srivastav, V. (1991) “The syntax and semantic of correlatives” Natural Language & Linguistic Theory. Suñer, M. (1984) “Free relatives and the matching parameter”, The Linguistic Review.

Friday, session 6

How do Usted in Spanish and Lei in Italian differ? Hyun-Jong Hahm (University of Texas, Austin)

The polite pronouns referring to the addressee, usted in Spanish and lei in Italian, trigger the third person agreement on the finite verbs as in (1b) and (2b), while the second informal pronoun in both languages triggers the second person agreement as in (1a) and (2a). When those polite pronouns are in the coordinated phrase, the verbs agree in the different person values in two languages – the third person in Spanish as in (3b) vs. the second person in Italian as in (4b). This paper analyzes the agreement patterns with the polite pronouns in two languages and explains why two languages result in the different person agreement with the coordinated phrasal subject. The coordinated phrase does not have a syntactic head and causes the semantic agreement (Pollard and Sag 1994). I propose that the pronouns usted and lei have the same person value, but they have the different semantic interpretations. The second person intimate pronouns and the second person polite pronouns have the different person values in both languages: 2nd vs. 3rd respectively. The person value of usted and lei explains the third person verb agreement with those pronoun subjects. The semantic interpretation of the pronouns and finite verbs in each language explains the different person resolution in the coordinated phrase. The pronouns and finite verbs form a Horn Scale (Horn 1989) in terms of semantic strength. In Peninsular Spanish, the second person intimate pronouns (tu[SG] and vosotro/as[PL]) and corresponding second person finite verbs (e.g. sois in (3)) have the meanings, related to the intimate addressee/hearer '{Hintimate}': its singleton set ('=') in the singular ones vs. its superset ('⊇') in the plural ones. The pronouns usted(es) refer to any hearer(s) '=/⊇ {H}'. In the example (3b), the coordinated phrasal subject does not include the pronoun, meaning {Hintimate}, and the specific 2nd person intimate verb sois cannot be chosen by the scalar implicature and more general 3rd person verb son is selected by the scalar implicature. In Italian, the meaning of the pronoun lei is identified by the form of verb: only when lei refers to the addressee, the verb agrees in 2nd person. The pronoun lei refers to any unspecified human ('={X}'), while the semantic interpretation of pronoun tu is the singleton set including just the intimate hearer ('={Hintimate}'). The second person plural verb siete has the meaning of a superset ‘⊇’ of {H, X} like the pronoun voi, while the third person plural verb sono has the most unmarked meaning (= {…}). Only when a conjunct lei is realized as the addressee in the context in the example (4b), the more specific verb siete has to be selected; otherwise the more general verb sono in the Horn Scale is chosen. This paper examines the syntactic and semantic properties of the pronouns in Spanish and Italian, focusing on the agreement with the coordinated phrasal subject. This comparative study 1

explains the different patterns in person resolution with the coordinated/non-headed phrasal subject as well as the same agreement patterns with the headed phrasal subject in these languages. In two languages, the polite pronouns for the addressee have the same person value, but each language has its own specific semantic interpretation of pronouns and finite verbs, which causes the different patterns of person resolution. (1) a. Tu veniste you came.2sg 'You (one informal addressee) came.' b. Usted vino / *veniste you.polite came.3sg / *came.2sg 'You (one formal addressee) came.'

[Peninsular Spanish]

(2) a. Tu sei venuto you be.2sg come.participle.M.sg 'You (one informal male addressee) came.' b. Lei é /*sei venuta. 3.F.sg be.3sg /*2sg come.participle.F.sg 'You (one formal female addressee) came.' or 'She came.'

[Italian]

(3) a. Tu hermana y tú sois honestos. 2sg.Gen sister and 2sg 2pl honest.M.pl 'You (one informal male addressee) and your sister are honest. b. Su hermana y Usted son /*sois honestos. 3Gen sister and 2sg.polite be.3pl /*2pl honest.M.pl 'Your sister and you (one formal male addressee) are honest.' [Peninsular Spanish] (4) a. Tu e tua sorella siete onesti 2sg and your sister be.2pl honest.M.pl 'You (one informal male addressee) and your sister are honest.' b. Lei e Sua sorella siete /sono oneste 3sg and his/her sister be.2pl /3pl honest.F.pl 'You (one formal female addressee) and your sister / She and her sister are honest.' [Italian]

Selected References: Horn, L. 1989. A Natural History of Negation. Chicago: Chicago University Press. Pollard, C. and Sag, I. 1994. Head-driven Phrase Structure Grammar. Stanford/Chicago: CSLI Publications and University of Chicago Press. Wechsler, S. 2004. Number as person. In Olivier Bonami and Patricia Cabredo Hofherr (eds.) Empirical Issues in Syntax and Semantics 5, pp. 255-274.

2

Friday, session 4 Deborah Arteaga (U. of Nevada, Las Vegas) and Julia Herschensohn (U. of Washington)

Developing L2 morphosyntax in TP and DP: A case study of French Introduction. Two UG accessible approaches to L2A propose opposing possibilities for parameter resetting, dependent on the capacity of interlanguage grammars to gain new values for uninterpretable functional features. The Representational Deficit hypothesis (RD, Hawkins & Chan 1997, Franceschina 2001) maintains that parameter settings are limited to L1 values, the cause of L2 morphological errors, while Missing Inflection (MI, Lardiere 2000, Prévost & White 2000) attributes morphological errors to PF mapping and interface difficulties. We examine accuracy of verb and noun morphology and syntax in production and grammaticality judgement (GJ) of “C” and “M,” advanced anglophone learners of L2 French, in terms of these hypotheses, first describing theoretical issues, then presenting the data and discussing it. We conclude that the results support parts of both hypotheses. Syntactic theory. In the bare phrase structure framework (Chomsky 2002), uninterpretable features (e.g., [Agr]) motivate Merge and Agree, with interpretable features matching and deleting uninterpretable ones. Main verbs in French are fully inflected, raised and checked for tense/person inflection in TP and AgrSP, while English main verbs do not raise and check only auxiliaries in TP (Lasnik 1999). The lexical verb raising (VR) parameter (Pollock 1989) attributes the difference between the two languages to overt morphology: “rich” French verbs raise above neg/adverb, but English lexical verbs do not (1). (1) a. Nous (n’)embrassons souvent / jamais Marie. b. We often / never kiss Mary. For DP (2), unlike English, French N raises (NR) to check interpretable number and gender on nouns against uninterpretable features of adjectives and determiners, once they are in close enough proximity (Carstens 2003). (2) a. le voile blanc / la voile blanche the-M-SG veil-M-SG white-M-SG / the-F-SG sail-F-SG white-F-SG b. the white veil / the white sail Gender is an idiosyncratic feature that must be learned for each lexical item; concord is a morphosyntactic rule that operates through Agree in the course of a syntactic derivation. L2A theory. For VR and NR, anglophone learners need to gain uninterpretable [Agr] on T and AgrS (deleted by the raised finite verb) and gender concord for L2 French. Two approaches to L2 morphosyntax predict opposite developments for the L2 French. RD holds that a post-Critical Period (CP) inability to gain L2 functional feature values restricts parameter settings to the native ones (in this case no reset T and D features, no VR or NR) and causes surface L2 morphological errors. MI maintains that L2 parameter values may eventually be gained (Schwartz & Sprouse 1996) and that inflectional errors are due not to syntactic deficits, but to difficulties in surface morphological realization. For anglophones learning French, RD predicts no parameter resetting for VR or NR and concomitant morphological errors; whereas MI predicts possible mastery of VR and NR and morphological performance errors. Data and discussion. Using surface L2 morphological errors as diagnostic, we compare production and GJ of two advanced post-CP anglophone learners of L2 French, C (Age of Onset 15) and M (AO 48). Here we examine the last of three interviews

conducted after a period abroad of 8-9 mos. Tokens of non-finite/finite verbs, finite accuracy, and accurate VR above neg/adverb are tallied in Table 1. Subject -finite Accur +finite error accur +VR* -VR GJ Accur M 58 100% 278 8 97% 30 0% 43/45 96% C 85 100% 372 3 99% 56 0% 43/45 96% Table 1 Production, tense and verb raising (*With diagnostics of neg/adv/quantifier) The 30/56 tokens of neg/adv/quant serve as clear diagnostics for VR. Although C and M produce numerous lexical and other grammatical errors, they use all persons of verbal inflection, and make no errors of non-finite, nominative subject, tense, negation, adverb and quantifier placement. On 45 GJ sentences they make two mistakes each with quantifiers. Tokens of lexical DPs, adjective use and errors are tallied in Tables 2 and 3. Subject Lex. DPs Gender Accuracy Lex. DPs Number Accuracy Vbl Gen Errors Vbl Num Errors M 155 6 96% 76 1 99% C 146 7 95% 86 0 100% Table 2 Production, lexical DPs: Number and Gender Subject DP+adj Concord Accuracy VR Lex DPs Det Accuracy Vbl Gen Errors accuracy Total Errors M 61 5 92% 100% 312 10 97% C 56 5 91% 100% 232 5 98% Table 3 Production, lexical DPs: Concord and Determiner Use Their consistently target-like L2 morphological inflection and word order indicate mastery of person/number/tense inflection for TP, supporting MI over RD. However, within DP, number accuracy trumps gender, suggesting some residual RD effect, possibly related to Age of Onset. Conclusion: Longitudinal data from advanced anglophone learners of L2 French supports MI’s proposal of eventual L2 parameter resetting (VR, NR) and morphological mapping difficulties. References Chomsky, N. 2002. On Nature and Language. Cambridge: Cambridge UP. Franceschina, F. 2001. Morphological or syntactic deficits in near-native speakers? An assessment of some current proposals. Second Language Research 17: 213-247. Carstens, V. 2000. 2003. Rethinking complementizer agreement: Agree with a case–checked goal. Linguistic Inquiry 34: 393–412. Hawkins, R. & C.Y-h. Chan. 1997. The partial availability of Universal Grammar in second language acquisition: the failed features hypothesis. Second Language Research 13: 187-226. Lardiere, D. 2000. Mapping features to forms in second language acquisition. In Archibald, Second Language Acquisition and Linguistic Theory, 102-129. Prévost, P. & L. White. 2000. Truncation and missing inflection in second language acquisition. In Friedmann & Rizzi, The Acquisition of Syntax, 202–235. Schwartz, B.D. & R. Sprouse. 1996. L2 cognitive states and the full transfer/full access model. Second Language Research 12: 40-72.

Deviance and convergence in early child bilingualism

1 of 1

file:///C:/Documents%20and%20Settings/jcamacho/My%20Documents... Saturday, 5:00-6:00

Deviance and convergence in early child bilingualism Aafke Hulk University of Amsterdam/ACLC

In the recent literature many hypotheses and claims have been made about the vulnerability of simultaneously emerging grammars in bilingual children. In this talk we will consider some of the internal and external factors mentioned to play a role in the manifestation of this vulnerability in production data. We will examine in what way these factors can contribute to the explanation of both deviance and convergence in different stages of development. To illustrate these general points, we will discuss spontaneous production data of dislocation constructions in Dutch/French and English/French bilingual children and we will present some new elicited production data by Dutch/French bilingual children, showing their problems with the acquisition of gender in the nominal determiner of both languages.

2/13/2006 4:56 PM

Saturday, session 20

Why neither Sympathy Theory nor Comparative Markedness are able to handle prosodic opacity: the case of Latin vowel deletion Haike Jacobs (Radboud University Nijmegen)

Vowel deletion in Latin has been well-studied throughout the history of Romance linguistics. It has received considerable attention in derivational metrical phonology (Mester 1992, among others) as well as in constraint-based Optimality Theory (OT) (Apoussidou and Boersma 2003, Jacobs 2004). The reason for devoting attention again to Latin vowel deletion is its relevance to the actual theoretical debate of how to handle opacity in OT. We will take the analysis of Jacobs (2004) as a starting point. After briefly reviewing it, we will concentrate on two current approaches to deal with opacity in OT: Sympathy Theory (the two versions of it presented in McCarthy 1999 and 2003) and Comparative Markedness (McCarthy 2002). We will demonstrate that both of them are in principle unable to deal with prosodic opacity. The essential fact for our purposes is the following: in pre-main stress sequences consisting of a heavy and a light syllable (HL) or of two light syllables (LL), vowel deletion deleted the second vowel. Some examples of syncope in HL are ārdēre ‘to burn’ for ārĭdēre and ārdōrem ‘fire’ for ārĭdōrem (cf. aridus). Examples of syncope in LL are calfacere for călĕfácere ‘to heat’ and caldarius for călĭdarius (cella caldaria) ‘room for hot baths’. The analysis by Jacobs (2004) accounts for all cases of syncope, but one: syncope of the second vowel in pretonic HL-sequences, such as, ārĭdēre > ārdēre ‘to burn’, as illustrated in tableau 1. Tableau 1: A problematic case of Latin vowel deletion /HLσ/ ārĭdōrem a. ) (H)L σ b. (HL) σ c. / (H) σ d. H L σ e. / H σ

*V IN (X.)

MAX-V

CLASH

W/L

PARSE-σ

*(HL)

* *!

* *! *!

* σ !σ σ

** *

Tableau 1 focuses on the pre-main stress part of the word: the main stressed syllable is indicated by underscoring, the secondary stressed syllable by boldface. Vowel deletion is indicated by braces. The relative ranking of the constraints CLASH (avoid two consecutive stressed syllables), W/L (the left-edge of a Prosodic word should be aligned with the left-edge of a foot, that is the word should start with a foot) and PARSE-σ (a syllable must be parsed into a foot) is motivated by the fact that a single remaining syllable before the main stressed syllable is not stressed, regardless of whether it is light, as in, for instance, pedéster ‘on foot’ or heavy, as in, for example, mercédem ‘salary, profit’. This means that CLASH has to dominate QS (heavy syllables must be stressed) and both have to dominate W/L (left wordedge starts with a foot) and PARSE-σ. The problem is that, rather than deleting the vowel in the second syllable, not parsing the second syllable in a foot is a far better solution given the ranking in tableau 1. Candidate (1a) has only one single violation: it violates PARSE-σ. Jacobs (2004) discusses a number of

alternatives. Ranking the constraints *V in (x.) and MAX-V below PARSE-σ is not helpful, because then the output candidates with a deleted vowel (1c) and (1e) will be ruled out by violating respectively CLASH and W/L. Ranking PARSE-σ above the constraints *V in (x.) and MAX-V is excluded as well, because it leads to problems in accounting for normal main stress (instead of pedéster ‘on foot’ ill-formed *pédester is produced). Jacobs claims, without however further substantiating that claim, that Sympathy Theory is not capable of accounting for the facts, and concludes by providing a derivational OT-account. The question that will be addressed in this paper, is whether Sympathy Theory or Comparative Markedness are capable of providing a fully parallel account. We will start by critically reviewing Jacobs’ analysis and firmly motivate the constraint ranking required for Latin. After that, we will show why Sympathy Theory, for principled reasons, is unable to deal with similar cases of prosodic opacity. The main reason is that the opacity in this case crucially depends on the prosody and not on faithfulness. In a nutshell, the wrong optimal output candidate (1a) should violate a sympathetic constraint in order to be eliminated. Given that this wrong optimal output candidate is a faithful candidate (not violating any faithfulness constraints), it will always be in the set of potential sympathetic candidates, and, being the optimal candidate in (1), it will necessarily be the optimal candidate in the subset of potential sympathetic candidates. Next, it will be demonstrated that the problematic case above can be accounted for by using Comparative Markedness by splitting up the constraint PARSE-σ in a new (NPARSE-σ) and an old (OPARSE-σ) constraint. High ranking of OPARSE-σ will select candidate (1a) as the FFC (Fully Faithful Candidate) and, in a full ranking, candidate (1e) as the optimal output candidate. However, the ranking necessary to deal with the cases of opaque deletion will be shown to lead to incorrect outputs in cases of transparent deletion. We will argue that no correct ranking of OPARSE-σ is possible which takes at the same time care of both opaque and transparent deletion. Finally, we will briefly discuss a derivational OT account for the deletion facts and provide some corroborating evidence for it by pointing out an essentially parallel case in contemporary Dutch. References Apoussidou, D. and P. Boersma. 2003. ‘The Learnability of Latin Stress’, Institute of Phonetic Sciences, Amsterdam. Proceedings 25, 101-148. Jacobs, H. 2004. ‘Rhythmic Vowel Deletion in OT: Syncope in Latin’ Probus 16, 63-90. McCarthy, John. 1999. ‘Sympathy and Phonological Opacity’, Phonology 16, 331-399. McCarthy, John. 2003. ‘Sympathy, Cumulativity, and the Duke-of-York Gambit’ C. Féry and R. van de Vijver (eds.) The Syllable in Optimality Theory. McCarthy, John. 2002. ‘Comparative Markedness’, A. Carpenter. A. Coetzee and P. de Lacy (eds.) Papers in Optimality Theory II. Amherst: GLSA. Mester, R. 1994. ‘The Quantitative Trochee in Latin,’ Natural Language and Linguistic Theory 12, 1-61.

Eva Juarros

(University of New York at Buffalo)

Friday, session 1

The Syntactic Operator se in Spanish: a Contemporary Account  There  is  a  long  history  of  literature  dedicated  to  the  Romance  clitic  se  (see  Sánchez  López  (2002)  for  a  recent  survey).  Accounts  fall  into  two  groups:  (a)  those  that  distinguish  the  different  functions  of  se  on  the  basis  of  inherent  properties  of  several  homonymous  morphemes  se  (Zubizarreta  (1987),  Dobrovie­Sorin  (1994),  Reinhart  and  Siloni  (2004),  etc.);  and  (b)  those  that  argue  that  the  surface  similarities  between  these  constructions  are  not  accidental,  and  hence  give  a  unified  account  of se  (e.g.,  Manzini  (1986), Postma (1993), Everett (1996)). The main criticism to the first group is that they  fail to express what is common to all the se­constructions, rendering the occurrence of se  in  different  constructions  as  largely  accidental,  a  simple  case  of  homonymy.  The  challenge  of  the  second  group  is  to  explain  how  a  simple  morpheme  can  give  rise  to  a  number of (seemingly unrelated) constructions, and how the typology is restricted.  In  my  paper  I  present  a  unified  account  of  the  properties  of  whay  I  call  the  syntactic  operator se in Spanish, which incorporates some recently developments in morphosyntactic  theories.  Impersonal,  reflexive  passive,  middle,  and  reflexive/reciprocal  constructions  are  examined. Basically, it is claimed that se has the following functions across the board:  a)  blocking the canonic syntactic realization of the external theta role  b)  blocking the features [case], [person] and [number] of Agr selectively  Property (a) is common to all constructions, and derives  from se’s uniform generation  in  head position of the functional node that would otherwise introduce the external argument  of the verb, and hence is in complementary distribution with such node (little v). This has  as a consequence that se is a diagnostic for unaccusativity in Spanish. It explains why se is  incompatible with unergative verbs when their external argument is also present (*Juan se  llora), since both the external argument and se are competing for the same position.  Property  (b)  belongs  to  the  syntax/morphology  interface.  Basically,  se  has  the  ability  to  randomly block phi­features in Agr, rendering them inaccessible for further checking (i.e.,  inactive); however, only the derivations that conform to morphological constraints surface,  and hence the different se constructions. Such constraints derive from the feature geometry  (Harley and Ritter 2002) of the Agr nodes in Spanish and the syntactic hierarchy of such  nodes. In Spanish, those constraints are *[case, ­phi­features], and *[person, ­number]. In  feature­geommetric  terms,  this  amounts  to  say  that  [case]  is  a  secondary  feature  to  [phi­  features],  and  [person]  is  a  secondary  feature  to  [number].  (Secondary  features  are  only  possible  in  the  presence  of  primary  features;  secondary  features  are  selected  by  primary  ones  in  syntax).  This  property  explains  the  existing  agreement  paradigms  in  se  constructions, and the impossibility of others.  My account of se has the advantage that it presents a totally unified treatment of se as a  syntactic  operator.  It  is  therefore  a  highly  economical  one.  It  is  also  explicit  in  the  functions  that  se  accomplishes,  including  its  syntactic  behavior  and  its  interaction  with  the hierarchically organized nodes of syntax and morphology. It also predicts which verbs  will never occur with se, by relating se’s function to the lexical structure of those verbs.  Finally, it presents se as one instance of a familiar class of syntactic operators, of which  the passive morphology is the clearest example (Baker, Johnson and Roberts (1989), see  Wherli  (1986)  for  a  similar  claim),  but  not  the  only  one  (cf.  Catalan  presentational  sentences,  Rigau  (1991)),  and  hence  it  relates  to  universal  grammar  in  a  natural  way,  without incorporating any new paraphernalia into the theory.

1 

Basic Data  1.  2.  3.  4.  5. 

Aquí se vive bien (Impersonal with intransitive)  En esta granja se mata [a seis mil pollos] cada día/ [se los mata] (Impersonal with transitive)  Se necesitan buenas ideas/[*se las necesitan] (Reflexive passive)  Las puertas se han cerrado de golpe/[*se las han cerrado de golpe] (Middle)  Marta y Juan se aman con locura (Reflexive/reciprocal) 

Diagrams of lexical entries:  a. transitive verb 

b. unergative verb 

vP 

c. unaccusative verb 

vP 

y 

y  v 

se  v 

VP 

VP 

VP  x 

x  V  romper, “break” 

V  llorar, “cry” 

V  desmayar, “faint” 

Summary table of syntactic features in se­constructions  External 

AgrS 

AgrS 

AgrS 

AgrO 

argument 

nom case 

Person 

Number 

acc case 

Impersonal 

­ 

­ 

­ 

­ 

+ 

Reflex Pass 

­ 

+ 

­ 

+ 

­ 

Middle 

­ 

+ 

­ 

+ 

­ 

Refl/Recipr 

­ 

+ 

+ 

+ 

­ 

Construction 

Syntactic tree for se constructions (partial) on Harley and Ritter’s terms  [Case] P  [Case]’  [Case] 

[Person] P  [Person]’  [Person] 

[Number] P  [Number]’  [Number] 

… 

Selected References  Everett,  D.  (1996):  Why  there  are  no  Clitics.  An  alternative  perspective  on  pronominal  allomorphy.  The  Summer Institute of Linguistics and the University of Texas at Arlington.  Harley,  Heidi  and  Elizabeth  Ritter  (2002):  Person  and  number  in  pronouns:  a  feature­geometric  analysis.  Language 78 (3), 482­527.  Manzini, Maria  Rita (1986): On  Italian  si. In  Syntax  and  Semantics  19:  The  syntax  of  pronominal  clitics,  edited by H. Borer. New York: Academic Press.  Postma,  G.  (1993):  On  reflexive  se,  passive  se  and  the  theory  of  person.  In  HIL  manuscripts  1.2.  Holland  Institute of Generative Linguistics, Leiden.  Reinhart, T.  and T.  Siloni  (2004):  in  Alexiadou,  Anagnostopoulou  and Everaert (eds.), The  Unaccusativity  Puzzle, Oxford University Press, 159­180.  Rigau,  G.  (1991):  On  the  functional  properties  of  Agr.  Catalan  Working  Papers  in  Linguistics,  UAB,  Bellaterra, 235­260.  Sánchez­López, C. (2002): Las construcciones con se. Visor, Madrid.

2 

Friday, session 1

Topicalization in European and Brazilian Portuguese Mary A.Kato (UNICAMP) & Eduardo Raposo (U.California,. Santa Bárbara) In European Portuguese (EP), the DP Topic and the associated clause hold a predicational relation through a clitic (Clitic Left dislocation, CLLD), a pronoun (Left Dislocation, LD), or a gap (Topicalization, Top), inside the comment clause. Brazilian Portuguese (BP) has no visible third person clitics, and construal is obtained through LD or Top: (1) a. Este livro, ninguém o resenhou ainda. EP=CLLD this book nobody 3pcl reviewed yet b. Este livro, ninguém resenhou ele ainda. BP=LD this book nobody reviewed it yet c. Este livro, ninguém resenhou ec ainda. EP & BP= Top this book nobody reviewed yet Though resumption of a direct object is realized through a clitic in EP (CLLD) and a nonclitic pronoun in BP (LD), both constructions behave alike regarding subjacency effects: they are immune to islands, being at odds with Cinque’s (1990) observation that island effects occur both with clitics and empty categories, but not with strong pronouns and epithets (2) a. Este livro, conheço [um jornalista que o resenhou]. EP this book (I) know the journalist who 3pcl reviewed b. Este livro, eu conheço um jornalista que resenhou ele BP this book, I know the journalist who reviewed it The traditional view on Topicalization also faces problems in the comparison of the two varieties of Portuguese, if we assume that movement that leaves a gap is constrained by subjacency. Although all examples involving topicalization with gaps in islands discussed in Raposo (1986) are starred in EP, when comparable data such as (3) were tested, EP speakers´ judgments were not consistent, while BP speakers consistently accepted such sentences. (3).Esse bolo [o rapaz que trouxe ec agora da pastelaria] era teu afilhado ?*EP aBP this cake the boy who brought ec now from the bakery was your godson Assuming that all construal relations derive from movement – even those with resumptives (Kayne, 2001;among others)–, we argue in this paper that this complex paradigm can be accounted for if subjacency effects are a function of the type of movement (with or without stranding), and the agreement relation between the Topic and the resumptive element. Following Kayne’s (2001) analysis for Spanish Clitic Doubling, we take the right or left dislocated element (4 a and b), and the pronoun/clitic to form one constituent (5), before the element in [Spec, DP] moves and leaves the pronoun/clitic stranded behind. See the derivation of sentences (2 a and b) along these lines in (6): (4) a.. Le doy un libro a Juan. (CLDOUBL) b.. A Juan, le doy un libro. (CLLD) (5) Doy un libro [Juan le] (6) a. eu só encontreiV [DP[DPeste CD] o/ele ] na FNAC b. [DPeste CD]i eu só encontrei [ ti [ o/ ele ] na FNAC Examples with an ec as (1c), on the other hand, are proposed to be derived from two sources: (a) full-argument extraction in EP (7) a. eu só encontreiV [DPeste CD] na FNAC b. [DPeste CD]i eu só encontrei ti na FNAC. (b)clitic left dislocation in BP (with stranding of a null clitic) (8) a. eu só encontrei [DP [DPeste CD] ∅-] na FNAC . b. [DPeste CD]i eu só encontrei [ ti [∅-]] na FNAC The derivations in (6) (CLLD and LD) and in (8) (CLLD with an invisible clitic) have stranding of a pronominal element, while the derivation in (7) the DP Top leaves only its copy, or its trace. According to our proposal, only full extraction(Top) is subject to islands: (9) a. Este livro, conheço um jornalista que o resenhou. EP

b. Este livro, eu conheço um jornalista que resenhou ele. BP c. Este livro, conheço um jornalista que ∅-resenhou. BP d. Este livroi, conheço um jornalista que resenhou. ti ?*EP “This book, I have met the journalist who reviewed it” (for a.b and c) Following Boeck (2001), we will also assume that case agreement between the A’-constituent and the resumptive, as in Spanish CLLD, turns the stranding type of movement also subject to subjacency. This is not the case of CLLD and LD in EP and BP. One type of Topicalization that is not sensitive to islands in either EP or BP is VPTopicalization: (10) a. Visitar os amigos, a Maria não conhece [ninguém que visite ec]aEPaBP visit the friends the Maria not knows [nobody that visits ec] In order to account for lack of subjacency effects in BP, Kato (2003) had proposed that Top in this variety resulted from Remnant Movement of VP Top (VPRT). (11) a. Os amigos, a Maria visita ec todos os anos the friends the Maria visits every year b.[VP visitarV os amigos]j a Mariai visitaV [vP ti [VP ___]j todos os anos. But, against Kato’s generalized claim that all Top constructions in BP derive from VPRT, speakers do not find VPT, in the non-remnant version, appropriate in all contexts where a DP appears as a Topic. Compare the answers to questions (12) and (14): (12) —Onde é que (vocês) encontraram este CD? where is it that you found this CD (13). a.Este CD, só encontramos ec na FNAC. aEP aBP this CD (we)only found- at-the FNAC b.Encontrar esse CD, só encontramos ec na FNAC. #EP #BP find this CD, (we) only found at the FNAC’ (14) --Vocês encontraram os CDs que (vocês) queriam? you found the CDs that you wanted ? (15) a. Os CDs nós encontramos, mas os livros que queríamos, não. aEP aBP the CDs we found , but the books that (we) wanted, not b. Encontrar os CDs, (nós) encontramos, mas os livros que queríamos, não aEP aBP find the CDs (we) found, but the books that we wanted, not We conclude that both EP and BP resort to VP-Topicalization or VPRemnant-Topicalization for contrastive Topics (14a and b). However, for non-contrastive Topicalization EP resorts to argument Topicalization, whereas BP resorts to Null-clitic left dislocation.. The fact that the PF output of non-constrastive Topicalization is identical in both dialects, despite their different derivations, is what underlies their apparent conflicting behavior. References Boeck,Cedric (2001) Mechanisms of Chain Formation: Ph.D.Dissertation. .U.Conn. Cinque, Guglielmo. (1990)Types of A'-Dependencies. MIT Press. Kato, Mary A. (2003).Null objects, null resumptives and VP-ellipsis in EP and BP. In: . J.Quer et alii (eds) Romance Languages and Linguistic Theory. John Benjamins. Kayne, Richard (2001) Pronouns and their antecedents. In: In S. Epstein and D. Seely eds.),Derivation and Explanation in the MinimalistProgram. Blackwell. Raposo, Eduardo (1986).The null object in European Portuguese”. Studies in Romance Linguistics. ed. by O.Jaeggli & Carmen Silva-Corvalán, Foris :

Friday, session 8 Marta Ortega-Llebaria and Stephen Meiners (University of Texas, Austin)

Accent without stress in Spanish. Stress is a structural linguistic property of a word that specifies which syllable is more prominent than any of the others. Accent, in its turn, is the greater prominence given to one or more words in the utterance according to pragmatic factors. When a word is accented, this prominence is realized primarily in the stressed syllable of the word. However, not all syllables with stress are accented in all discourse contexts: the presence or absence of an accent depends on the larger prosodic structure in which the lexical item is found. In this view, we have the four levels of syllabic prominence illustrated in Table 1. +stress + +accent (a) +stress, +accent -accent (b) +stress, -accent Table 1. Levels of syllabic prominence.

-stress (c) –stress, +accent (d) -stress, -accent

In the past decades, a great deal of research studied the phonetic correlates of stress. However, this earlier work examined stress in accented contexts (syllables ‘a’ and ‘d’ in Table 1), and therefore, it conflated the cues of stress with those of accent (for English see for instance, Bolinger, 1958; for Spanish, Navarro Tomás 1964, Contreras 1964, Quilis 1971, Solé 1984, Canellada & Kuhlman-Madsen 1987, Llisterri 2003, among others). Only in recent research, unaccented stressed syllables were also studied (context ‘b’)(for Dutch see Slujter & van Heuven 1996a,1997, for American English see Slujter & van Heuven 1996b, Campbell & Beckman 1997), and it was found that stressed syllables were cued by increased duration, lack of vowel reduction and a more balanced spectral tilt regardless of the presence or absence of an accent. Moreover, accent correlated with an increase in pitch and overall duration. They concluded that stress and accent were two different units both at the conceptual and at the physical level. However, these studies did not include context (c), i.e. [-stress, +accent] syllables. The study of this context is crucial to confirm the independence of stress from accent: when we have an accented syllable that is not stressed, will the cues of pitch and overall intensity – which have been related to accent – be sufficient to mark that accent? Or will we need duration and/or other intensity cues – which have been related to stress? If so, which cues pattern together and how do they interact? And more importantly, how should we characterize stress and accent: in terms of two different sets of cues – as it has been concluded in the cited studies – or, as a composite of ‘overlapping’ cues whose combination produces two different percepts, one for stress and one for accent? Fifteen native speakers of Mexican Spanish were asked to produce the sentences in the Appendix in order to obtain segmentally identical syllables with the four levels of prominence in Table 1. Preliminary results indicate the following trends: 1. As in previous studies, stressed syllables differ from unstressed syllables in terms of duration, vowel quality and spectral tilt. 2. Accented syllables differ from unaccented syllables in terms of pitch and overall intensity when they are stressed. When they are unstressed, they also differ in vowel quality, but not in duration. 3. Syllables that bear broad focus accent have shorter durations that those with narrow focus accents.

These tendencies seem to indicate that stress and accent are not realized with two different sets of cues, but with different combinations of similar cues and these different combinations relate with the different levels of prominence depicted in Table 1. References

Bolinger, D.L. (1958) A theory of pitch accent in English. Word 14, 109-149. Canellada, M and John Kuhlman-Madsen (1987). Pronunciación del español. Lengua hablada y literaria. Editorial Castalia: Madrid. Campbell, Nick and Mary Beckman (1997). Stress, prominence and spectral tilt, in Antonis Botinis, Georgios Kouroupetroglu, & George Carayiannis (eds.) Intonation: Theory, Models and Applications (Proceedings of ESCA Workshop, September 18-20, 1997, Athens, Greece). ESCA and University of Athens, Department of Informatics: 67-70. Llisterri, J., M. Machuca, M. de la Mora, M. Riera & A. Ríos (2003). The perception of lexical stress in Spanish. Proceedings of the 15th International Congress of Phonetic Sciences, 2023-2026. Barcelona, 3-9 August 2003. Navarro- Tomás, T. (1964) La medida de la intensidad. Boletín del Instituto de Filología de la Universidad de Chile 16, pp. 231-235. Quilis, A. (1971). Caracterización fonética del acento en español. Travaux de Linguistique et de Lítterature 9, 53-72. Quilis, A. (1981). Antonio Quilis , Fonética acústica de la lengua española. Biblioteca Románica Hispánica, Gredos, Madrid. Slujter, Agaath and Vincent van Heuven (1996a). Spectral balance as an acoustic correlate of linguistic stress. Journal of the Acoustical Society of America 100 (4). 2471-2485. Slujter, Agaath and Vincent van Heuven (1996b). Acoustic correlates of linguistic stress and accent in Dutch and American English. Proceedings of ICSLP, 96, 630-633. Slujter, Agaath and Vincent van Heuven (1997). Spectral balance as a cue in the perception of linguistic stress. Journal of the Acoustical Society of America 101 (1). 503-513. Solé, M.J. (1984). Experimentos sobre la percepción del acento. Estudios de Fonética Experimental I, Universidad de Barcelona 134-243.

Appendix DE [+stress] Lola y Eva le prestaron su ____ a Damián. Ahora, ellas quieren que Damián les devuelva la _________. ¿Qué quieren Lola y Eva? Quieren que Damián les dé la _______ ¿Quieren que PEDRO les dé la ________? No, quieren que DAMIÁN les dé la _______ ¿Quieren que Damián se QUEDE con la ________? No, quieren que Damián les DÉ la _________ ¿Qué preguntabas? ¿Quieren que Damián les dé la ________? DE [-stress] Damián tiene una ______ nueva. Está muy contento y quiere contárselo a sus amigas Lola y Eva. ¿Qué quiere Damián? Quiere platicarles de la ______. No lo entiendo. ¿Dices que quiere platicarles EN la _______? No, quiere platicarles DE la _________. ¿Damián quiere prestarles la _______ ? No, quiere PLATICARle de la _________. ¿Qué preguntabas? ¿Quiere platicarles de la ________?

In each blank, subjects insert 30 different words like cafetera ‘coffee maker’, lavadora ‘washer’, etc...

Saturday, session 12

On the use of the definite article with prepositions in Romanian Alexandru Mardale (LLF, Paris 7 & U. of Bucharest)

Goal: (i) examine the use of the definite article with prepositions in Romanian and Albanian; (ii) provide a morpho-syntactic analysis of article “drop”. The data: definite article omission In Romanian, language with a suffixal definite article, a non modified noun preceded by a preposition is necessarily used without the definite article (1). Conversely, when the noun is modified, it requires the definite article (2). Further evidence shows that this phenomenon is not sensitive to the position in which the PP occurs: preverbal subject in copular sentences (3), indirect object (4), the so-called “prepositional” direct object (5), complement of the noun (6), modifier (7). It is to be noted tha t, despite the omission of the article, these constructions have an obligatory referential and definite reading. The same phenomenon exists in Albanian (8), but not in other languages we have examined, e.g. French (9), Italian (10) or English (11). The suffixal article: necessary or sufficient condition? One question that arises when examining these data is whether the definite article omission takes place in all languages where it has a suffixal status. In other words, is suffixal status necessary and / or sufficient condition? A partial answer can be supplied if we look at Bulgarian (12), where the definite article, in spite of its suffixal status, always appears after the preposition. In fact, this can be seen as a necessary condition since the definite article does not fall in languages where it has a different status, i.e. proclitic (9)-(11) supra. An additional argument in favour of this view is that the Romanian indefinite article, which is proclitic, does not fall either (13). However, this does not seem to be a sufficient condition since we have seen that article omission is impossible with modification (2). To sum up: the definite article omission in Romanian and Albanian is subject to two constraints: (i) the article’s status and (ii) modification. Given the facts presented so far, a number of issues have to be addressed: what is the licensing mechanism for the lack of the definite article with prepositions? Why are these constructions allowed in argument positions? Why must the definite article appear when the noun combines with a modifier? Why is the definite reading always permitted? We suggest that the ability to appear in argument positions must be explained in terms of Case. This could be correlated with the fact that Romanian bare singulars cannot appear in argument positions (14). In order to be licensed in these positions, they must combine with an article (15). If we assume, following Grosu (1988) and Giusti (1993), that the article functions as Case (and implicitly that this is the licensing property for non prepositional arguments), on the one hand, and that the preposition is also a Case marker (Stowell (1981)), on the other hand, we are led to conclude that constructions we analyze here receive Case from two categories, whereas they are supposed to receive it only once (Chomsky (1981a)). Intuitively speaking, this is why the definite article “disappears” in P + NP constructions. On the contrary, when the construction is more complex (i.e. P + NP + modifier), the preposition is no longer able to assign Case to the hole construction, which is why the article must be overt. As for the definite reading of these constructions, even when the definite article is “dropped”, we will consider it as a special morphological property of the noun preceded by the preposition. In other words, in languages like Romanian, nouns can “absorb” the definite article in strict syntactic conditions without deleting its definite features.

Analysis: we will use the morpho-syntactic mechanism of m- merger (Matushansky (2005)) to formalize the article “drop” with prepositions. This mechanism is defined for two heads in a particular configuration (16), which means that they are subjects to strict locality (i.e. nothing may intervene between them). It consists in the following two operations: (i) movement of the attracted head to the attracting head and (ii) m- merger. The result of m- merger is a single syntactic head which contains the features of both initial heads. Since modification disrupts this very local relation between the two heads (i.e. adjacency), the attracted head can no longer incorporate and thus it must be overt. Exemples (1) (2) (3) (4) (5) (6) (7) (8)

(9) (10) (11) (12)

(13) (14) (15) (16)

Ma îndrept catre parc. me head towards park ‘I head towards the park’ Ma îndrept catre parc ul cu copaci mari. me head towards park-the with trees big -PL ‘I head towards the park with big trees’ Sub masa este un loc preferat de copii pentru a se acunde. under table is a place preferred by children for to SE hide ‘Under the table is children’s favourite place to hide’ Am optat pentru presedinte. have opt for president ‘I opted for the president’ L-am vazut pe profesor. him-have seen PE professor ‘I saw the professor’ (S-a procedat la) construirea de palate pentru rege. (SE-has proceed to) construction-the of palaces for king ‘(It was proceeded to) the construction of palaces for the king’ Comoara a fost îngropata în gradina. treasure-the has been buried in garden ‘The treasure has been buried in the garden’ a. Vuri librin mbi trapezë. put book-the on table ‘I put the book on the table’ b. Vuri librin mbi trapezën që bëri gjyshi. put book-the on table-the that made grandfather ‘I put the book on the table grandfather made’ Je me dirige vers le parc / vers le parc avec de grands arbres / vers le vieux parc. Mi dirigo verso il giardino / verso il giardino con fiori. We are heading towards the park / towards the park with big trees. a. Otpraviam se kam masata. head me towards table-the ‘I head towards the table’ b. Otpraviam se kam tservenata masa. head me towards red-the table ‘I head towards the red table’ Ma îndrept catre un parc. me head towards a park ‘I head towards a park’ *A venit profesor. / *Am citit carte. has came professor have red book A venit profesorul. / Am citit cartea. has came professor-the ‘The professor came’ have red book-the ‘We red the book’ a. XP b. XP X°

YP Y°

movement

Selected references

X° ZP

X°

Y°

YP Y°

ZP

m-merger

Chomsky 1981a, Lectures on Government and Binding; Giusti 1993, La sintassi dei determinanti; Grosu 1988, On the Distribution of Genitives Phrases in Romanian; Longobardi 1994, Reference and Proper Names; Matushansky 2005, Moving a-Head; Stowell 1981, Elements of Phrase Structure.

What consonant vocalization tells us about the syllabic affiliation of muta-cumliquida clusters in Proto-Spanish and Western-Romance Fernando Martínez-Gil, The Ohio State University [email protected] It is a well-known fact that stop-liquid clusters in modern Spanish are invariably syllabified as onsets (Hooper 1976, Harris 1983, 1989, Hualde 1991, 1999); the same holds for Classical Latin (Palmer 1954, Devine and Stevens 1977, Allen 1978), and the modern Romance languages. However, some Romance scholars have suggested that the prosodic affiliation of such clusters when located in the last syllable of proparoxytonic words must have shifted in early Vulgar Latin, becoming heterosyllabic, thus rendering the penult heavy, a scenario that, given the stress pattern of Classical Latin, neatly accounts for the rightward shift of stress from the antepenult to the penult in items such as ÍNTĔGRU, CÁTĔDRA > (MSp.) entéro, cadéra, etc. (Grandgent 1907, Lausberg 1965, Väänäänen 1968), although it crucially entails the preservation of phonemic vowel length distinctions at the relevant Vulgar Latin stage (Klausenburger 1975). Such an analysis, however, has been challenged by other scholars who argue that stop-liquid groups been invariably tautosyllabic throughout the known history of Latin and the Romance languages (Pulgram 1975, Steriade 1987). The primary purpose of this paper is to argue that certain stop-liquid clusters may have been syllabified as heterosyllabic (or, perhaps ambysyllabic) in Proto Western-Romance, lending further support to the hypothesis of a shift in the syllabific parsing of such sequences as the underlying cause of the early Vulgar Latin stress-shift phenomenon. Our analysis is based on two intimately related pieces of evidence. First, Hispano-Romance exhibited a phonotactic constraint prohibiting oral plosives syllable-finally (Malmberg 1965, Harris-Northall 1990); and second, such a constraint can be seen as responsible for one of the most characteristic developments in Western Romance, namely, the vocalization of velar stops in coda position (quite generally when the stop is followed by a coronal consonant in the onset of the following syllable): LACTE > *la[j].te/ > le[j].te ‘milk’, AXE > *a[j].se > e[j.š]e ‘axis’, LIGNA > *le[j].na > le.ña ‘wood’ (cf. Galician-Portuguese leite, eixe, leña). While stop vocalization naturally targetted the first member of such sequences, unquestionably syllable-final in Proto-Spanish, its is rather striking that the process also affected the first member of -KL- and -GL- clusters (Menéndez Pidal 1980: 159): OC(U)LU > *o[k].lo > *o[j].lo > o.[¥]o ‘eye’, VET(U)LU > *ve[k].lo > *ve[j].lo > ve.[¥]o ‘old-MASC.’, REG(U)LA > *reg.la > *re[j].la > re.[¥]a ‘plowshare’, FLAGRARE > *fla[j].rare > cheirare ‘to smell’ (cf. Galician-Portuguese o.[¥]o, ve.[¥]o, re.[¥]a, chei.rar). Analogous developments can be observed in other Ibero-Romance languages, as well as in Old French (e.g. LACRIMA > lairme; Lausberg 1965, Bourciez 1967), and in Provençal, in which the vocalization of the first member of a stop-liquid cluster is widespread. The fact that vocalization applied in these cases leads to the natural conclusion that here the stop was syllablefinal. Further support for this analysis is provided by the sporadic vocalization of a dental followed by a rhotic; thus, in addition to expected voicing of /t/ and spirantization of /d/ (cf. Spanish padre < PATRE, cua[D]ro < QUADRU), the historical record often shows instances of vocalization (and often subsequent deletion of the yod), as in Toledo Mozarabic frayre (< FRATRE), Peyrolu (from PETRU), Valencia and Mallorca Mozarabic pare (< PATRE), enrera (< IN RETRUM) (Galmés de Fuentes 1983), Old Catalan noyrir (< NUTRIRE), caire (< CUADRU), veure (< VIDRE; Badía Margarit 1951, Moll 1952), Old Aragonese fraire (< FRATREM), cuairón (< QUADRONe; Zamora Vicente 1979), GalicianPortuguese pai (< patrem), mai, mae (< MATREM) (Williams 1934, Açevedo Maia 1986), Gascon hray (< FRATRE; ), ma[j]r ( cho[w]ro, FABRU > fa[w]ro, PETRA > pe[j]ra; CAD(E)RE > chè[j]re; FLAGRARE > fla[j]ra; NIGRU > nè[j]r(e), etc. (Anglade 1921, Ronjat 1930).

References Açevedo Maia, Clarinda de. 1986. Historia do galego-português. Coimbra: INIC. Allen, W. 1978. Vox Latina: The Pronunciation of Classical Latin. New York: Cambridge Univ. Press. Anglade, J. 1921. Grammaire de l'ancien provençal. Paris: Klincksiek. Badía Margarit, A. 1951Gramática histórica catalana. Barcelona: Noguera. Bourciez, E. 1967. Phonetique française: étude historique. Paris: Klincksiek. Bustos Tovar, E. 1960. Estudios sobre asimilación y disimilación en el iberorrománico. Madrid: RFE, Anejo 70. Devine, A. M. and L. D. Stephens. 1977. Two Studies in Latin Phonology. Ann Arbor: Anma Libri. Galmés de Fuentes, A. 1983. Dialectología mozárabe. Madrid: Gredos. Grandgent, Ch. H. 1907. An Introduction to Vulgar Latin. Boston: Heath and Co. Harris, J. W. 1983. Syllable Structure and Stress in Spanish: A Non-Linear Approach. Cambridge, Mass.: MIT Press. Harris, J. W. 1989. Sonority and Syllabification in Spanish. Selected Papers from the Linguistic Symposium on Romance Languages 17. Philadelphia: John Benjamins, 139-153. Harris-Northall, R. 1990. Weakening Processes in the History of Spanish Consonants. London: Routledge. Hooper, J. 1976. An Introduction to Natural Generative Phonology. New York: Academic Press. Hualde, J. I. 1991. On Spanish syllabification. In Current Studies in Spanish Linguistics, ed. by H. Campos and F. Martínez-Gil, 475-493. Washington, DC: Georgetown University Press. Hualde, J. 1999. La silabificación en español. In Fonología de la lengua española contemporánea, ed. by R. Núñez-Cedeño and A. Morales-Front, Washington DC: Georgetown Univ. Press. Klausenburger, J. Latin vocalic quality to quantity: a pseudo problem? In Diachronic Studies in the Romance Linguistics, ed. by M. Saltarelli and D. Wanner. The Hague: Mouton, 107-117. Lausberg, H. 1965. Lingüïstica románica, vol. I. Madrid: Gredos. Menéndez Pidal, R. 1980. Manual de gramática histórica española. Madrid: Espasa-Calpe. Malmberg, B. 1965. Estudios de fonética hispánica. Madrid: CSIC. Moll, F de B. 1952. Gramática histórica catalana. Madrid: Gredos. Palmer, L. R. 1954. The Latin Language. Norman, Oklahoma: University of Oklahoma Press. Pulgram, E. 1975. Latin-Romance Phonology. Múnchen: Wilhem Fink Verlag. Rohlfs, G. 1970. Le gascon: études de philology pyrénée. Tübingen: Max Niedemeyer Verlag. Ronjat, J. 1930. Grammaire historique des parlers provençaux modernes, vol. I. Montpellier: Societé des Langues Romanes. Steriade, D. 1987. Gemination and the Proto-Romance syllable shift. In Advances in Romance Linguistics, ed. by D. Birdsong and J.-P. Montreil, 371-410. Dordrecht: Foris. Väänänen, V. 1968. Introducción al latín vulgar. Madrid: Gredos. Williams, E. 1938. From Latin to Portuguese. Philadelphia: University of Pennsylvania Press. Zamora Vicente, A. 1979. Dialectología española (2nd ed.). Madrid: Gredos.

Eric Mathieu (University of Ottawa)

Friday, session 5

OLD FRENCH QUIRKIES Quirky subjects are subjects surfacing with non-nominative case. They have long been observed in Icelandic (Andrews 1976, Zaenen, Maling & Thráinsson. 1985 and Sigurðsson 1989) and have also been claimed to exist in Old English (Lightfoot 1979, Fischer & Van der Leek 1983, Allen 1986, Kemenade 1997), as well as in older stages of Mainland Scandinavian languages (Barðhal 1997). As for Romance languages, Spanish is one language that has recently been the centre of much attention (Masullo 1993, Fernández-Soriano 1999, Rivero 2004). The aim of this paper is to show that Old French makes great use of quirky subjects. Dubbed ‘impersonal constructions’ in the traditional literature, such structures typically involve an empty subject position that a dative, accusative or genitive XP comes to fill. This, I show, very much resembles the operation called Stylistic Fronting (SF, henceforth) which is clearly available not only in Icelandic (Holmberg 2000), but also in Old French (Mathieu, to appear). More generally, a connection for Old French is made between the availability of quirky subjects, SF and V3 structures (see Fischer 2004 for the same generalization made with respect to Scandinavian languages, old and new). The OF example in (1) illustrates a prepositional (oblique) phrase XP appearing in the subject position while (2) involves a dative-marked XP. The example in (3) is the equivalent of (2), but with a clitic subject. This pattern has been traditionally treated as thematically based, i.e. the dative is an inherent case intrinsically linked to the role of Experiencer. The weak dative pronoun is often replaced by a strong (‘emphatic’) pronoun of the same case as illustrated in (4). Quirky subjects can also surface in the accusative form. However, the dative form comes to replace the accusative very often not only across texts, but also in some cases within the same texts as (5) and (6) illustrate (both examples are from Le Chevalier à la Charrette). This process is well-known from Icelandic and has been referred to as Dative Sickness (Eythórsson 2000) for that language. Finally, genitives in subject position are not common in OF, but are attested in Icelandic. One typical feature of quirky subjects is that, unlike other (nominative) subjects, they do not agree with the verb, a feature which is wellknown for Icelandic. In (7) the verb takes the 3rd person singular while the subject is 3rd person plural. In (8) the verb also appears with 3rd person singular whereas the subject is 1st person singular. First, I show that quirky subjects in Old French have all the properties of subjects (the case of Icelandic, Andrews 1976, Zaenen Maling & Thráinsson. 1985, Sigurðsson 1989), and not of so-called I-nominals (the case of German, Moore & Perlmutter 2000). The standard tests in the literature are: (i) reflexivisation; (ii) subject-verb inversion (in V1 and V2 environments); (iii) raising; (iv) control (i.e. being a controllee); (v) conjunction reduction; (vi) subject position in ECM infinitives. Although the search was not easy, I was able to find conclusive examples for (i)-(v), except for the remaining property. Then, I concentrate on the agreement patterns found in quirky subject constructions and on the way nominative case is assigned. It is shown that in OF it is Tense rather than agreement that is responsible for the licensing of nominative case (there is thus no need to claim that agreement has nevertheless taken place in the syntax, as has been popular of late). Next, I show that the operation SF is very productive in OF. Then, I make a connection between StylFronted elements and oblique subject XPs. Both types of elements target a special Topic position above TP dubbed TopP+. I argue that when a dative or accusative pronominal form is used rather than a full XP, the pronominal, which at first sits in Spec-TP, affixes to its host (via a process of lowering between Spell-Out and PF as in the framework of DM, Halle & Marantz 1993; Embick & Noyer 2001). Thus, the canonical subject position (Spec-TP) is free for an SFronted XP to proceed through that position to then raise further to Spec-Top+.

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

Et bien set qu’a sa mere plest that-to his mother please.3SG and well know.3SG Que rien a feire ne li lest that nothing to do.INF not him.DAT remain.3SG ‘And she knows well that it is her mother's will that she shall leave nothing undone for him.’(Le Chevalier au Lion 5437-5438) Et se Deu plaist, outre s’en passera and if God-DAT please.3SG others self-of.it go.FUT.3SG ‘If such is the will of God, he will force the passage.’ (Aliscans, 1099, in Buridant, 322) Si li plest, el l’amera please.3SG she him.ACC-love.FUT.3SG. if him.DAT ‘If it pleases him, she will love him.’ (Lais, Milun 28) Si lui plaist if him.DAT please.3SG ‘If it pleases him.’ (La Chanson de Roland 519, Le Chevalier à la Charrette 2585) Car desfandre le covenoit, thus defend.INF him.ACC necessitate.PAST.3SG ‘thus it was necessary for him to defend (himself).’ (Le Chevalier à la Charrette 1182) Qu’a remenoir li covenoit him.DAT necessitate.PAST.3SG that-to remain.INF ‘that he had to stay.’ (Le Chevalier à la Charrette 3760) Tous les estuet morir all them.ACC.3PL necessitate.3SG die.INF ‘It was necessary for all of them to die.’ (Joseph 630, in Pearce 1990:182) Kar mei meïsme estoet avant aler Since me.DAT.1SG myself.1SG necessitate.3SG ahead go.INF ‘Since I myself alone should go ahead.’ (La Chanson de Roland 2858)

Allen, C. 1986. Reconsidering the history of like. Journal of Linguistics 22, 375-409 | Andrews, A. 1976. The VP complement analysis in modern Icelandic. NELS 6, 1–21 | Embick, D. & Noyer, R. 2001. Movement operations after syntax. LI 32, 55–595 | Eythórsson, T. 2000. Dative vs. nominative: Changes in Quirky Subject in Icelandic. Leeds WPL 8,27–44 | Fernández-Soriano, O. 1999. Two types of impersonal sentences in Spanish: Locative and dative subjects. Syntax 2, 101–140 | Fischer, S. 2004. The diachronic relationship between quirky subjects and stylistic fronting. Non-nominative subjects,193–212. Amsterdam: Benjamins | Fischer, O. & Van der Leek, F. 1983. The demise of the Old English impersonal construction. Linguistics 19,337–368 | Halle, M. & Marantz, A. 1993. Distributed morphology and the pieces of inflection. The view from building 20,111–176. Cambridge, Mass.: MIT Press | Holmberg, A. 2000. Scandinavian stylistic fronting: how any category can become an expletive. LI 31, 445–483 | Kemenade, A. van. 1997. V2 and embedded topicalization in Old and Middle English. Parameters of morphosyntactic change, 326–352. Cambridge: CUP | Lightfoot, D. 1979. Principles of diachronic syntax. Cambridge: CUP | Masullo, P. J. 1993. Two types of quirky subjects: Spanish versus Icelandic. Proceedings of NELS 23, ed. by GLSA,303–317. Amherst: UMass. | Mathieu, E. to appear. Stylistic Fronting in Old French. Probus. | Moore, J. & Perlmutter, D. 2000. What does it take to be a dative subject? NLLT 18,373–416 | Rivero, M-L. 2004. Quirky subjects, person restrictions, and the Person-Case-Constraint. LI 35, 494–502 | Sigurðsson, H. 1989. Verbal syntax and case in Icelandic. PhD diss., Uni. of Iceland, Reykjavík | Zaenen, A., Maling, J. & Thráinsson, H. 1985. Case and grammatical functions: The Icelandic passive. NLLT 3, 441–48.

Friday, session 3 Roberto Mayoral Hernández and Asier Alcázar (University of Southern California)

Sociolinguistic factors conditioning the ordering of adverbial expressions in Spanish: A computer-aided corpus analysis 1. Problem Spanish frequency adverbials may occur before or after the verb with no change in meaning: (1) Frecuentemente Juan (frecuentemente) estudia (frecuentemente) en casa (frecuentemente) Frequently Juan (frequently) study.3.Sg (frequently) in house (frequently) ‘Juan often studies at home’ The syntactic distribution in (1) cannot be accommodated by syntactic approaches that assume different meanings for preverbal and postverbal positions (Cinque 1999, Kovacci 1999, Fernández & Anula 1995). In this paper we show that the position of adverbials is affected by sociolinguistic and stylistic factors: gender, country and source. The relevant sentences were taken from the online corpus Corpus de Referencia del Español Actual (CREA). We developed a suite of software tools to automatically extract, select and annotate the data. 2. Method: variationist approach with the aid of computational tools We coded a sample of 1600 sentences from CREA. CREA is a non-annotated corpus of modern Spanish that ranges in time from 1976 to the present day. It includes metadata corresponding to country, source, author name, topic, publisher, and date of publication. While gender required manual annotation, our software automatically generated the coding for country and source, and produced annotated files ready to be utilized by statistical software. Using the statistical tool SPSS, we applied cross-tabulations and the χ2 test to determine the significance of each independent variable. A value of p=.05 was adopted as the threshold for significance. The sentences were coded as nominal data. The addition of a binary logistic regression analysis ensures the fitness of the model. The dependent variable was the position of adverbials (preverbal, postverbal), and the independent variables were gender (male, female), country (Spain, Latin America) and source (book, press). 3. Results The analysis of the data shows that the distribution of frequency adverbials is influenced by source, country and gender. Source was determined to be relevant for Spain (p=.018), as press is associated with a higher percentage of postverbal occurrences (see table 1). However, in Latin America the distribution was almost even (p=.189). Regarding gender, the sample consists of 657 tokens, since CREA includes author information for books only. Table 2 indicates that Latin American women prefer postverbal positions, unlike men. In fact, gender is statistically significant in Latin America (p=.000). In contrast, table 3 demonstrates that gender is not significant in Spain (p=.088). Both Spanish men and women prefer postverbal positions, similarly to Latin American women. Our analysis thus restates the sociolinguistic relevance of gender (Silva-Corvalán 2001, Chamber et al. 2002, Coulmas 1997, Cameron 2005). 4. Conclusion Our research shows that gender, country and source influence the position of adverbials in Spanish, independently of the underlying syntactic structure assumed. The position of frequency adverbials has proven to be an interesting indicator of stylistic differences between Spain and Latin America, while pointing out the existing social distinctions related to gender. The text processing programs we have created will facilitate future corpus-based linguistic studies.

Table 1: Books vs. Press in Spain

Source Book

Position of Adverbials Total

Postverbal count % column Preverbal count % column count

Pearson Chi Square

Value 5.644b

Table 2: Gender distribution for books in Latin America Postverbal count Position of % column Adverbials Preverbal count % column count Total Pearson Chi Square

Value 24.988b

Table 3: Gender distribution for books in Spain Postverbal count Position of % column Adverbials Preverbal count % column count Total Pearson Chi Square

Value 2.913b

Press 233 55.9% 184 44.1% 417

279 63.8% 158 36.2% 437

Degrees of Freedom

Total 512 60.0% 342 40.0% 854

Significance 1

.018

Gender Female

Male

104 62.3% 63 37.7% 167

50 34.0% 97 66.0% 147

Degrees of Freedom

Total 154 49.0% 160 51.0% 314

Significance 1

.000

Gender Female

Male

91 62.8% 54 37.2% 145

106 53.5% 92 46.5% 198

Degrees of Freedom

Total 197 57.4% 146 42.6% 343

Significance 1

.088

References Cameron, Richard. (2005): Aging and gendering. Language in Society 34. 23-61. Chambers, J.K., et al. (2002): The handbook of language variation and change. Oxford: Blackwell. Cinque, Guglielmo (1999): Adverbs and functional heads: A crosslinguistic perspective. New York and Oxford: Oxford University Press. Corpus de Referencia del Español Actual (CREA), in www.rae.es Coulmas, Florian, ed. (1997): The handbook of sociolinguistics. Oxford: Blackwell. Fernández, Marina & Anula, Alberto (1995): Sintaxis y cognición. Introducción al conocimiento, el procesamiento y los déficits sintácticos. Madrid: Editorial Síntesis. Kovacci, Ofelia (1999): El adverbio. In Bosque and Demonte (eds.), Gramática Descriptiva de la Lengua Española. Madrid: Espasa. Silva-Corvalán, Carmen (2001): Sociolingüística y pragmática. Washington, D.C.: Georgetown University Press

Roberto Mayoral Hernández (University of Southern California)

Friday, session 3

A Variationist Approach to Verb Types and Subject Position in Spanish: Verbs of light and sound emission 1. Introduction There seems to be a striking, though predictable, relation between subject position and verb types in Spanish. Applying a quantitative analysis, this paper shows that transitive, unergative and unaccusative verbs can be successfully differentiated by looking at the position of their subject. By using the statistical tool SPSS and the online corpus CREA, I could determine the significance of this claim. This method was then utilized to check the status of a controversial subtype of verbs: Verbs of Light and Sound Emission (VLSEs), which were classified here as unaccusative (pro Perlmutter 1978) rather than unergative verbs (contra Mendikoetxea 1999). 2. Unaccusativity and verbs of light and sound emission Spanish lacks overt unaccusative morphology, unlike Dutch, French or Italian, and therefore it also lacks an explicit way to identify unaccusative verbs. However, Spanish subjects can either precede or follow the verb, and Suñer (1982) observed that unaccusative verbs show a higher tendency to appear with postverbal subjects. The fact that unaccusative verbs are underlying objects could account for the higher percentage of postverbal occurrences in Spanish. Although VLSEs have been classified both as unaccusative (Perlmutter 1978) and unergative (Mendikoetxea 1999), the subject position favors an unaccusative interpretation in Spanish. 3. Two experiments The first experiment was originally designed to determine if unaccusative verbs show indeed a higher percentage of postverbal subjects when compared to transitive and unergative verbs. Three prototypical verbs were chosen to represent the different types of verbs. Table 1 confirms the prediction about unaccusative verbs, i.e., that they show a higher percentage of postverbal subjects when compared to unergative and transitive verbs. However, it also proved that unergative and transitive verbs pattern differently in a statistically significant way. In fact, Pearson’s Chi-square shows a P value lower than 0.05, which is statistically significant. The second experiment was designed to determine if VLSEs behave as unergative, i.e., with a higher percentage of preverbal subjects, or as unaccusative verbs, i.e., with a higher percentage of postverbal subjects. Table 2 shows that VLSEs and unergative verbs behave differently when it comes to the position of their subject, P [péntə], /kwato/ > [káto], /anbre/ > [ámbə], and /blanka/ > [báka], respectively. Similar processes have been attested in the development of onsets of other languages. I thus hypothesize that this consonantal onset reduction is subject to the Sonority Sequencing Principle (SSP), which is impacted by a headship constraint, both to be defined and formalized in terms of OT . The former basically says that a consonant must not decrease its sonority in the onset, or increase it in the coda. The results of this study are at variance with those reported by Lleó and Prinz(1996) for German and Spanish, and, indirectly, by Gnanadesikan (2004), for Dutch. Lleó and Prinz have studied the speech of monolingual German and Spanish children and conclude that while German follow the normal developmental path in which the least sonorous consonants survive, Spanish favors the most sonorous, as in /globo/ that produces [lóβo]. In the face of mixed results in both studies, I further suggest that the variability that Spanish children exhibit when acquiring onsets may be explained by the interplay of two different acquisition strategies: some children opt for emphasizing the SSP in conjunction with headship of onset, while others are predisposed to focus on the relative perceptibility of adjacent consonants (Côté 2004). Additionally, data such as /tinbre/ > [tímbərə] and maestro > [maεstəra], among many others presented for the first time in Spanish acquisition literature, show that the occurrence of an epenthetic vowel in the child’s speech provides irrefutable evidence against the claim that the process is just unique to Portuguese (Freitas 2003). Following Freitas, I argue that the reason for inserting a

vowel to resolve complex onsets is because of the propensity to conform to the universal CV structure. Like Portuguese children, the child’s emerging speech provides many instances of onsetless syllables, which, in OT terminology, calls for ranking low the Onset constraint that requires syllables to contain onsets. The paper also includes a discussion on the relative ranking of the acquisition of liquids, where *Rhotic must be a high-ranked constraints as many tokens of the collected data show that /l/ precedes the flap. Selected references Beers, Mieke. 1996. Acquisition of Dutch phonological contrasts within the framework of Feature Geometry theory. In Proceedings of the UBC International Conference on Phonological Acquisition, Barbara Bernhardt, John Gilbert and David Ingram, eds., 28-41. Somerville, MA: Cascadilla Press. Côté, Marie Hèléne. 2004. Consonant cluster simplification in Québec French. Probus 16. 151-202. Fikkert, Paula. 1994. On the Acquisition of Prosodic Structure. Leiden: The Netherlands: University of Leiden dissertation. Freitas, M. Joao. 2003. The acquisition of onset clusters in European Portuguese. Probus 15:1.27-47. Gnanadesikan, Amalia. 2004. Markedness and faithfulness constraints in child phonology. In Constraints in Phonological Acquisition, René Kager, Joe Pater and Wim Zonneveld, eds.73-107. Cambridge: Cambridge University Press. Goad, Heather and Yvan Rose. 2004. Input elaboration, head faithfulness, and evidence for representation in the acquisition of left-edge clusters in West Germanic. In Constraints in Phonological Acquisition, , René Kager, Joe Pater and Wim Zonneveld, eds.774-109. Cambridge: Cambridge University Press. Jakobson, Roman. 1968. Child Language Aphasia and Phonological Universals. The Hague: Mouton. Lleó, Conxita and Michael Prinz. 1996. Consonant clusters in child phonology and the directionality of syllable structure assignment. Journal of Child Language 23.3156. Ohala, Diane K.1999. The influence of sonority on children’s cluster reductions. Journal of Communication Disorders 32:6.397-422. Stites, Jessica, Catherine Demuth, and Cecilia Kirk. 2004. Markedness vs. frequency effects in coda acquisition. In Alejna Brugos, Linnea Micciulla, & Christine E. Smith (eds.), pp. 565-576. Proceedings of the 28th Annual Boston University Conference on Language Development. Somerville, MA: Cascadilla Press

Saturday, session 19 Ana Teresa Pérez-Leroux, Mihaela Pirvulescu, Yves Roberge (Univ. of Toronto) Topicalizations and object drop in child language Recent work on child language has shown variability across languages on children’s developmental resolution of object omission in early utterances (Jakubowicz et al 1996, Schmitz et al. 2004, Wexler et al. 2002, Ud Deen 2003, Perez-Leroux et al. 2005, among others). We argue that such crosslinguistic differences should be examined in light of the complexity of the target grammar, including the fact that adult grammars allow for null objects in a variety of contexts. In this presentation, we report on a comparative study (French and English) on the semantic and syntactic conditions regulating young children’s object omissions in 3, 4 and 5 year olds. Recent work reveals that null object phenomena in French and English are complex both in terms of the narrow syntax operations and pragmatics involved in their interpretation; Groefsema (1995), Lambrecht & Lemoine (1996), Larjavaara (2000). Cummins & Roberge (2005) show that null objects are diverse, and that most of the observed variation can be attributed to the various modules involved in the recoverability/licensing of the object. Although verbs can be classified according to how freely they allow for null objects, argument structure alone is not sufficient to account for the facts. This is consistent with an approach to verbal transitivity that dissociates in part the availability of an internal direct argument position from lexical specification; see Borer (2005) among others. Consequently, null objects can be classified on the basis of the type of recoverability at work in a given expression. For instance, the reference and/or contents of a null object can be obtained: 1) internally, through material in IP (including lexical semantics as in Cette mauvaise habitude agace --. ‘This bad habit annoys --.’); this null object corresponds to the prototypical (or generic) object of a given verb and can be analyzed as a null cognate object in a non-individuated context. Despite the nonindividuated interpretation, pragmatic considerations make it possible to produce an inferred reference for this null object; 2) binding from the left periphery in cases of topicalizations, such as Les bananes, je mange --. ‘Bananas, I’ll eat --’ with a type (not token) interpretation; 3) discourse recoverability, as in Si on prenait Tigre et Dragon? Qui a vu --? (= Qui l’a vu?) ‘How about Crouching Tiger Hidden Dragon? *Who has seen --?’. This null object is semantically referential and thus an individuated one and can be replaced by a clitic. Descriptively this is a clitic-drop construction. Object omission in acquisition could be the result of lexical discontinuity (learning which verbs tend to require an object), of pragmatic discontinuity (relating to what informational demands are to be met in a successful exchange) and c) syntactic discontinuity. Predictions vary: lexical discontinuity should lead to gradual, conservative development (Ingham 1993/1994), and pragmatics to comparable development across languages. Parametric discontinuity is the only account that predicts radical cross-linguistic differences, as the developmental paths are expected to vary according to the complexity of the system to be acquired. Previous research has considered elicited production in contexts that introduce a specific, previously mentioned, individuated referent for the clitic (Jakubowicz et al. 1996; Schaeffer 2000; Schmitz et al. 2004; Wexler et al. 2003). Pérez-Leroux et al. (2005) found that French children’s preference for null objects went beyond the possibility of clitic drop (as it could be in (1a))and extended to non-individuated contexts, as set up by questions such as (1b): (1) a. Question: Qu’est-ce que la fille fait avec la fleur? “What is the girl doing with the flower?” Target answer: Elle la dessine. “She is drawing it.” b. Question: Qu’est-ce que la fille fait? “What is the girl doing?” Target answer: Elle dessine. “She is drawing.” Children acquiring French showed much higher rates of null objects in both contexts when compared to both French adults and children acquiring English. Pérez-Leroux et al. (2005) argued that it is the complexity of the contexts where null objects appear in French that lead to a delay in child French: the distribution of null cognate objects, discourse identified referents, and the combination with clitic drop present a formidable challenge for a child. However, variation in the

input can have two results: on one hand, children may omit clitics because they are aware of the clitic-drop construction and over-rely on it; on the other hand, they may have not fully resolved the proper parametric space for French null objects, overextending the initial null cognate object. In the present paper, we examine these two possibilities by testing children on a context that forces them to represent a syntactically referential null object, not simply a null object that can inherit pragmatically inferred reference. This case has different outcomes in the two languages: topicalization (English) vs. clitic left dislocation (French). In English, this is a movement construction. In French, when the dislocated element is interpreted as a token as opposed to a type, clitic drop is not allowed: this differs from contexts such as (1a). (2) Ce livre, je *(l’)ai acheté hier. This book, I bought yesterday. 37 French-speaking children and 27 English-speaking children participated in a comparative sentence completion study. The child task was to help a puppet answer questions about stories. The story and picture introduce two definite contrastive elements. When the puppet fails to finish his answer to the question, the child is invited to help him complete what he is saying. The target stimuli is a sentence with the target direct object as a left-dislocated topic, and interrupted after negation: (3) Experimenter: Non, ce n’est pas vrai. Allons aider Kermit. La fille construit un château mais la voiture, elle ne … No, it’s not true. Let’s help Kermit. The girl is building the castle, but the car, she is not…. Expected answer: la construit pas. building. Adults in both groups performed as expected: their data approximated 100% null objects for English and clitics in French. The children’s response patterns showed highly significant differences across languages. French children aged 3 and 4 had minimal rates of clitic production (less than 4%), and even at 5 rates were far from target (~40%). English children aged 3 showed some lexical and pronominal error (28% and 15%, respectively). By age 4, error rates had fallen to 10%, and children’s null object reached 80% target. Both French adults and children have the possibility of null objects—however, it is clear that the conditions that regulate their distribution in adult grammar are not (fully) operating in child grammar. French children uniformly prefer null objects across different domains, as opposed to adults and English children, and continue to do so even in context that prohibit clitic drop. So French children do not perform better in the clitic left-dislocation contexts where there is reduced adult variability. This result supports a unified account of null objects in child grammar. Assume that a null N is uniformly available across languages as the default possibility, made available by UG (consistent with a minimal default grammarview; Roeper 1999). Rather than develop separate accounts for topicalized, non-individuated, and individuated objects, we propose that French children retain the null cognate object as a default longer than Englishspeaking children because they are not yet able to grammatically interpret variation in null objects across domains. English children are never exposed to null objects in individuated, referential contexts, and therefore learn earlier to restrict the distribution of null objects. For French children, acquisition of the language specific clitic/pro associate relation restricts the contexts of use of the null cognate object default. Selected References: Groefsema (1995) “Understood arguments: A semantic/pragmatic approach”, Lingua 96; Ingham (1993/1994). “Input and learnability: direct object omissibility in English” Language Acquisition 3; Jakubowicz, Müller, Kang, Riemer & Rigaut (1996) “On the acquisition of the pronominal system in French and German” BUCLD 20; Pérez-Leroux, Pirvulescu & Roberge (2005) “Early Object Omission in Child French and English” LSRL 35; Schaeffer (2000) The Acquisition of Direct Object Scrambling and Clitic Placement; Wexler, Gavarro & Torrens (2003) “Object clitic omission in child Catalan and child Spanish”. Research report.

Saturday, session 19 Silvia Perpiñán (University of Illinois at Urbana-Champaign)

!"#$ !!%$' ()

* !!+$, $

-

!"%$

2 $

&&# 0 1 -

0 $

!"%

3

)

.

$ $ '' 5 :

) 2 9 .

( '- 0 4

$

3

3 .

'0!!" . !7" 0 6 ; 0=

A0 =

;

.

= . ,>= ,>= $

+&

"

;

?@ ''

A

( C" ; $B

)'

' '@ C

; C"

&

)

D

0

E F=

;

;

;

G

0H

E F=

;

;

G

0 HH

E

F=

0 I3

;

0 H3

;

;

G

J

F=

0 1

.

G

J

0

F=

= - = ( 0

' = . $#& ( B'$ C" @ C"

1

. *

.

; 0=

G

( B' ;

;

,>=

;

;

;

; D 06 ; !#K

*

''

$

'-

.

B

; ;

'' 0 1 /&K

!K 3

$

#+K $

)' .D; G

L . *

$

$ $

0 *

05

(

'' ,

G

&&/ 0 =

;

;

C

( D $;

)

$

( 0 $

*

$ 06 ;

$

0=

$

-

%

;

-

;

0 D $0 < 5

)6

, $90

$>0 !"# 0 $ -

D 1 .0 1 3 .

0

0$

!"7 0 2 . ) 0 /7D/"+)#&70 , $2 0=0 . $0 !"% 0 =O D, B 0 2 8 $80$2 9 $ 0 5 $>0 !!" 0 6 ; G P$ 7#D/&")/%#0 2 $-0 && DQ1 R P$! " # + $/!)%"0 NNNNNNNNN &&# D D> 5 0 ' () * !!+ DQ P$ #$ &+) /"0 : .$60 !7" $ # $ % " ! & $M $8 0

Pilar Prieto (Universitat Autonoma de Barcelona and ICREA) Marta Ortega-Llebaria (University of Texas, Austin)

Saturday, session 17

Phonetic correlates of stress and accent in Catalan and Castilian Spanish. In the past decades, a great deal of research has been dedicated to the acoustic realization of stress and its cues. However, a lot of this work has studied the correlates of stress in an accented context, thus suffering from covariation between the contribution of accent and stress (for Spanish see Navarro Tomás 1964, Quilis 1971, 1978; Llisterri 2003, among others). We agree with Beckman & Edwards (1994) and Slujter & van Heuven (1996:2472) that in order for results on the correlates of stress to be comparable across languages there should be a systematic control of the levels of prominence. The goal of this article is to disentangle the acoustic correlates of stress from those of accent in two languages, Catalan and Spanish. Following work by Slujters & van Heuven (1996, 1997) on Dutch and Campbell & Beckman (1997) on American English, we analyzed four acoustic correlates of stress (syllable duration, vowel quality, overall intensity, and spectral balance) in four conditions, namely, stressed and unstressed syllables in both accented and unaccented environments. This allows us to examine the relative strength of these correlates and see how they interact with the presence versus absence of a pitch accent. Given that Spanish and Catalan differ greatly in the way they use vowel reduction to mark stressed positions (Catalan, like English and Dutch, has a phonological rule of vowel reduction), the goal of this article is to test whether Spanish and Catalan will also differ in the way they use the other acoustic correlates (duration and overall intensity, and spectral balance) to signal the presence of stress and accent. We recorded 10 subjects, 5 native speakers of Catalan and 5 native speakers of Castilian Spanish, producing short sentences in their native language. These sentences contained four-syllable long verbs that ended in “-minar” or “-nimar”, and were cognates in Spanish and in Catalan, like determinar ‘to determine’, iluminar ‘to illuminate’, abominar ‘to detest’, desanimar ‘to discourage’. Each verb was conjugated in the present tense and in the preterit in order to obtain the endings “-mino/ -minó” for Spanish and “–mina/-minà” for Catalan. Sentences were produced within 2 different intonation contours in which the verb ending: (1) bore a pre-nuclear accent like in Me desanimó Catalina ‘Catalina discouraged me’, (2) did not have an accent like in parenthetic sentences such as Va a llover —me desanimó Catalina ‘It’s going to rain —Catalina discouraged me’. A total of 600 syllabic tokens were analyzed: 2 syllabic positions (ultimate and penultimate) x 2 utterance-types (declarative and parenthetic) x 15 verbs x 5 subjects x 2 languages (Spanish and Catalan). We ran a Repeated Measures ANOVA with stress (+/- stress) and intonation (declaratives/parenthetic sentences) as main factors on each vowel ([i] / [o]) for each set of measurements. The results showed no differences between the two languages’ use of the acoustic cues. Along with the findings of Slujter & collaborators and Campbell & Beckman (1997) on Dutch and English, Catalan and Spanish reveal systematic phonetic differences in the acoustic characterization along the accent and stress dimensions. Specifically, while syllable duration, vowel quality, and spectral tilt are reliable acoustic correlates of the stress difference in Catalan and in Spanish, accentual differences are acoustically marked by overall intensity cues.

Finally, Linear Discriminant Analyses (LDA) with duration, vowel quality, spectral tilt, and overall intensity as the predictor variables and stress or accent as the criterion variables reveal that duration is the most effective correlate of stress in both parenthetic and declarative sentences in both languages, followed by spectral tilt. References Beckman, Mary and Jan Edwards (1994). Articulatory evidence for differentiating stress categories, in Phonological Structure and Phonetic Form. Papers in laboratory phonology III, pp. 7-33. Ed. by P. A. Keating. Cambridge: Cambridge University Press.

Campbell, Nick and Mary Beckman (1997). Stress, prominence and spectral tilt, in Antonis Botinis, Georgios Kouroupetroglu, & George Carayiannis (eds.) Intonation: Theory, Models and Applications (Proceedings of ESCA Workshop, September 18-20, 1997, Athens, Greece). ESCA and University of Athens, Department of Informatics: 6770. Llisterri, J., M. Machuca, M. de la Mora, M. Riera & A. Ríos (2003). The perception of lexical stress in Spanish. Proceedings of the 15th International Congress of Phonetic Sciences, 2023-2026. Barcelona, 3-9 August 2003. Navarro- Tomás, T. (1964) La medida de la intensidad. Boletín del Instituto de Filología de la Universidad de Chile 16, pp. 231-235. Quilis, A. (1971). Caracterización fonética del acento en español. Travaux de Linguistique et de Lítterature 9, 53-72. Quilis, A. (1981). Antonio Quilis , Fonética acústica de la lengua española. Biblioteca Románica Hispánica, Gredos, Madrid. Slujter, Agaath and Vincent van Heuven (1996). Spectral balance as an acoustic correlate of linguistic stress. Journal of the Acoustical Society of America 100 (4). 2471-2485. Slujter, Agaath and Vincent van Heuven (1996). Spectral balance as a cue in the perception of linguistic stress. Journal of the Acoustical Society of America 101 (1). 503513.

Fernanda Maria Ribeiro Gonçalves (Universidade de Evora)

Saturday, session 16

The 36th Linguistic Symposium on Romance Languages Rutgers University, New Brunswick, NJ March 31 – April 2, 2006 Verb movement – patterns of acquisition and triggering experience in European and Brazilian Portuguese In this paper, partial results of a larger analysis on the acquisition process of Brazilian (BP) and European Portuguese (EP) are presented. The corpus consists of data (from Laboratório de Psicolinguística da Faculdade de Letras da Universidade de Lisboa and from Centro de Documentação Cultural Alexandre Eulálio, Instituto de Estudos da Linguagem, Unicamp) of four Portuguese and three Brazilian children, aged from 1;08.21 to 3;01.15, taped in sixty sessions, which include 15323 child verbal forms. The data were transcribed and codified according to the CHILDES rules (for more details, see Gonçalves (2004) and, about the CHILDES system, MacWhinney (2000)). With relation to the patterns of acquisition of verb movement, I intended to determine whether there was evidence to the existence of V-to-I in EP and/or BP at the relevant stage. To determine so, the following aspects were analyzed: relative position of adverbs (including markers of sentence negation) and verbs (as in (1) below); complexity of verbal clusters (as in (2) below); answers to global questions, assuming these are instances of VP ellipsis and, thus, of V-to-I (as in (3) below) and alterations of the basic word order (adult like or not) (as in (4) below): (1) LUI_P_02.cha": line 338. *LUI: o (L)uís tem um u(r)sinho aqui. %syn: 1suj 1v 3/3(ref1) pres 1obj the Luís has a little bear here (“Luís has a little bear here”). (2)

PAU_B_01.cha": line 3084. *CHI: [/] quer descer 0com papai. %syn: 0suj 1v 3/1 pres 1vinf want-3rdSing.Pres. to come down 0with daddy (“I want to come down with daddy”).

(3)

LUI_P_01.cha": line 973. *MAE: que(re)s qu(e) a mãe vá buscar um livro # pa(ra) lermos um livro aqui? “do you want mum to go fetch a book # to read a book here?” *LUI: que(r)o. %syn: 0suj 1v 1/1 pres 0obj want-1stSing.Pres. (“yes”)

(4)

PAU_B_01.cha": line 2999. *CHI: cacá@c guardei. %syn: 0suj 1obj 1v 1/1 pass (o) cacá@c kept-1stSingPast (“I’ve kept cacá@c”)

The results show that there are no major differences between EP and BP (although there are notorious interindividual distinctions) in what concerns verb movement, which exists, as an

1

option, since the first relevant productions. In theoretical terms, we are led to the conclusion that the inflectional node (I) is already available at the stage under analysis. With relation to the trigger for verb movement, under recent hypotheses it has been assumed that the acquisition of the inflectional paradigms was a necessary pre-requisite; this assumption can be seen as an instantiation of a larger principle: morphology drives syntax, which was referred to (but not supported) under the term “Rich Agreement Hypothesis” (RAH) in Bobaljik (2000) and explored in many works on acquisition (both L1 and L2), as reviewed in Herschensohn and Stevenson (2003). With the purpose of evaluating such hypothesis, I present data on the acquisition of the inflectional paradigms in EP and BP, which are then compared to the conclusions on verb movement referred to above. Those cross results lead to the conclusion that the acquisition of inflectional morphology and verb movement should be dissociated, as they are seen to evolve independently. In fact, evidence goes against the existence of a morphological trigger, since there is evidence of V-to-I before the inflectional paradigms are acquired. These conclusions go against the RAH, confirming the basic preliminary results assumed in Gonçalves (2001). Instead, the notion of “subspecification” is considered fundamental, been seen as associated with each specific category from the onset: it is argued that categories are thus available from the beginning, being gradually specified, which supports the Continuity Hypothesis. Finally, reference is made to similar conclusions about other Romance languages, as in Gaya (1998) for Castilian and Catalan and Ferdinand (1996) for French. References Bobaljik, Jonathan David (2000), The Rich Agreement Hypothesis in Review, ms., McGill University. Ferdinand, Astrid (1996), The Development of Functional Categories – the Acquisition of the Subject in French, The Hague, Holland Institute of Generative Linguistics. Gaya, Aurora (1998), Teoria Lingüistica i Adquisició del Llenguatge, Dissertação de Doutoramento, Universitat Autònoma de Barcelona. Gonçalves, Fernanda (2001), “Comparing Acquisition Processes in European Portuguese and Brazilian Portuguese - Additional Evidence for Morphology After Syntax”, Proceedings of GALA, 312-319. ___________________ (2004), Riqueza Morfológica e Aquisição da Sintaxe, PhD Diss., Universidade de Évora. Herschensohn, Julia and Jeffrey Stevenson (2003), “Failed Features or Missing Inflection? Child L2A of Spanish Morphology” in Proceedings of the 27th Annual Boston University Conference on Language Development, Volume 1, 299-310. MacWhinney, Brian (2000), The CHILDES Project: Tools for Analyzing Talk. Third Edition, 2 volumes, Lawrence Erlbaum Associates, Mahwah, New Jersey.

2

Joshua Rodríguez (Northern Illinois University)

Friday, session 6

Macro event types and aspect shift in Spanish The aspectually distinctive past tenses of the Romance languages have long been an area of interest to linguists and pedagogues. The core of this interest lies in the correlation between aspect and event type, namely imperfective with stative and perfective with non-stative event types. One area of particular relevance has been the so called aspectual shift frequently encountered in the interpretation of stative verbs in the simple preterit, which often imply a nonstative event type. One persistent, if not problematic hypothesis has been the treatment of aspectual shift as a case of aspectual coercion (de Swart 1998, 2000, Michaelis 2004). Coercion, which plays out as a syntactic intervention with semantic repercussions, is presumably activated to salvage an otherwise uninterpretable or ill-formed proposition. For coercion to function properly in this case, it is assumed that perfective and imperfective forms respectively require or select nonstative or stative event descriptions only. Therefore, in the case that a stative verb is found in the simple preterit, aspectual coercion is responsible to alter the type of event description inherently encoded by the verb to some other non-stative event type. This is illustrated in the following Spanish examples and presupposes that conocer is lexically encoded as stative (to know). 1. Conocía a todos los estudiantes que asistieron a la conferencia. I knew-IMP all the students that attended the conference. 2. Conocí a todos los estudiantes que asistieron a la conferencia. I met-PRET all the students that attended the conference. Instead of conceiving of the contrast between stative and non-stative interpretations as a “shift” from an inherently stative event type to some other “forced” non-stative event type, I propose to treat stative event types in a more fluid fashion, such that in many cases a verb may ambiguously refer to different phases of a single macro event type. A macro event type, in the case of conocer, would encode both the initial meeting phase as well as the prolonged resultative phase of established acquaintance. The advantage of this hypothesis over coercion is three-fold. First, because coercion relies on aspectual mismatch to be activated, it is incapable of explaining the distribution of interpretations in aspectually neutral verb forms (cf. Bonami 2002). Both stative and non-stative interpretations are possible (at times in slightly different contexts) in such verb forms. This is illustrated in the following examples. 3. Quiero que conozcas a mi hermano. I want you to meet- PRES SUBJ [literally know] my brother. [inchoative] 4. Espero que conozcas bien a mi hermano. I hope you know- PRES SUBJ /get to know my brother well. [ambiguous] 5. Mario insistió en que Yolanda conociera a mi hermano anoche. Mario insisted that Yolanda meet- PAST SUBJ my brother last night.[inchoative] 6. Mario dudaba que tú conocieras a mi hermano. Mario doubted that you knew/met- PAST SUBJ my brother. [ambiguous] 7. Mañana conocerás a mi hermano. Tomorrow you will-meet my brother. [inchoative] 8. Conocerás mejor a mi hermano después de hablar con él esta noche. You will-know my brother better after you talk to him tonight. [stative] Second, coercion only requires a shift in overall aspectual properties, but does not explicitly predict any particular sort of modification to the event type. Therefore, to the extent

that it accurately predicts the appropriate overall shifts in interpretation at all, it already implicitly assumes some sort of macro event structure, similar to what I’m proposing. That is, from a logical stance, why should the non-stative interpretation of conocer be inchoative instead of terminal (i.e., implying a forgetting)? However, if we assume a certain set of ordered subevents in the macro event encoded by conocer, then the alternatives are already limited to a known set. Thus, if context limits the interpretation of a verb to either a stative or non-stative interpretation, the particular event type must be chosen from corresponding sub-events of particular aspectual types made available by the macro event description. Third, because my proposal assumes that the context in which a verb appears (including but not limited to aspectually differential verb forms) serves as an aspectual filter, the semantic restrictions can more easily be stated in overtly semantic terms. This again contrasts with the usual treatment of aspectual coercion, which is fundamentally a syntactic phenomenon and sensitive only to formal rather than purely semantic considerations. This implies that I will be able to eliminate the use aspectual features (which are generally only vaguely defined) and rely directly on truth conditions, which may be more rigorously defined in model theoretic terms, as undertaken by Cipria and Roberts 2000. Furthermore, under these assumptions it will be shown that various other cases of supposed aspectual coercion may be naturally accounted for as lexical ambiguity, with macro event types being a special case of lexical ambiguity. It will also be shown that the interpretation of the progressive of stative verbs is similarly licensed and constrained in its interpretive potential. And finally, it will be argued that the “shift” from stative to non-stative interpretations is fundamentally different from the shift from non-stative to stative (e.g., progressive), which may more readily be accounted for by some sort of coercive or interpolative semantic process. References Bach, Emmon. 1986. The algebra of events. Linguistics and Philosophy 9.5–16. Bonami, Olivier. 2002. A syntax-semantics interface for tense and aspect in French. Proceedings of the International Conference on Head-Driven Phrase Structure Grammar, ed. by Frank Van Eynde, Lars Hellan, and Dorothee Beermann, 31-50. Stanford: CSLI Publications. Cipria, Alicia and Craige Roberts. 2000. Spanish imperfecto and pretérito: truth conditions and aktionsart effects in a situation semantics. Natural Language Semantics 8. 297-347. Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press. Dowty, David. 1979. Word meaning and Montague grammar: the semantics of verbs and times in generative semantics and in Montague's PTQ. Boston: D. Reidel Publishing Company. Michaelis, Laura. 2004. Type Shifting in Construction Grammar: An Integrated Approach to Aspectual Coercion. Cognitive Linguistics, 15, 1, 1-67. Swart, Henriëtta de. 1998. Aspect shift and coercion. Natural language and linguistic theory 16, 347-385. Swart, Henriëtta de. 2000. Tense, aspect and coercion in a cross-linguistic perspective. Proceedings of the Berkeley Formal Grammar conference University of California, Berkeley. CSLI Publications. Rodríguez, Joshua. 2004a. Interpreting the Spanish imperfecto: Issues of aspect, modality, tense, and sequence of tense. Ph.D. dissertation. The Ohio State University. Rodríguez, Joshua. 2004b. The Imperfecto Reference Time. WCCFL 23, 677-689. Vendler, Zeno. 1967. Linguistics in Philosophy. Ithaca, New York: Cornell University Press.

Jasper Roodenburg (NWO / University of Venice)

Sunday, session 21

The role of Number within nominal arguments: the case of French pluralized Event Nominalizations Introduction It is commonly assumed in recent syntactic literature that the functional domain of so-called Event Nominalizations (ENs) – like destruction, assassination, etc. – contains both verbal FPs (like AspP, vP, VP) and nominal FPs (like NumP, DP, NP). However, discussion on crosslinguistic variation of the internal syntactic structure of ENs mainly focusses on the verbal layer (see e.g. Van Hout & Roeper 1998, Alexiadou 2001, Borer 2003), while the nominal layer has been paid much less attention to in this respect, and is often supposed to be more uniform, in accordance with the strong claim of Grimshaw 1990 that ENs can never be pluralized, neither within nor across languages (compare (1) and (2)). (1) The destruction of the city by the soldiers. (2) *The destructions of the city by the soldiers. In this talk, I show that this claim is too strong from an empirical point of view on the basis of French, where the pluralization of ENs can argued to be productive and not to be subject to the same restrictions as often supposed for a language like English. In particular, I take a closer look at cross-linguistic differences within the nominal layer of ENs, and confront it to the idea often suggested in the the syntactic and semantic literature on so-called “Bare Nouns” that Number is necessary for the licensing of nominal arguments (cf. Delfitto & Schroten 1991, Bouchard 2002, Farkas & De Swart 2003, Déprez 2005, among others). More precisely, I relate the behaviour of pluralized ENs to less well understood differences between (existential) BNs in Germanic and Romance languages, and show that cross-linguistic variation can be accounted for by the more complex mechanism of syntactic Number licensing argued for by Roodenburg 2004ab, who has proposed a new classification of BNs on the basis of unexplored data from French (among others the behaviour of coordinated BNs). The results of the discussion are compared to other Romance and Germanic languages and are integrated into a framework that attempts to give more insight into the relation between syntactic structure and semantic interpretation (i.e. the syntax/semantics interface). On Pluralized ENs My discussion on Number touches on a central factor in the well know distinction between (Complex) ENs and Result Nominals (RNs) (Grimshaw 1990), which differ as for their denotation (event vs. object), the choice of their determiner (restricted to the or the lack of an overt determiner vs. non-restricted), the presence of complements (obligatory vs. optional), and their aspectual properties (compare (3) and (4)). (3) The/Ø/*a/*one (frequent) examinationEvent of papers for hours. (4) The/a/one examResult *for hours/was long. Theoretically, the described distinction has been argued to represent a crucial difference at the level of their internal syntactic structure (cf. Alexiadou 2001): while RNs have a functional domain containg exclusively nominal layers, the functional domain of (Complex) ENs also has verbal functional layers. The impossibility to pluralize ENs derived from e.g. transitive predicates, as in (5), could than by hypothesis be ascribed to the lack of NumP (a claim that we will reject however). Moreover, an example like (6) is only an apparent counterexample, because the plural mark is commonly considered as having triggered a shift from EN to RN. (5) *The (frequent) examinations of the papers. [Alexiadou 2001: 11] 1

(6) The assignments were long. [Grimshaw 1990: 54] This supposed shift can be made explicit in languages like German and Greek by a parallel interpretational shift of its of-complements from Theme to Possessor (Alexiadou 2001: 81): while the birds are the objects being observed in (7), they are interpreted as the observers in (8). (7) Die BeobachtungEvent von VögelnTheme. “The observation of birds.”

(8) Die BeobachtungenResult von VögelnPossessor. “The observations of birds.”

However, that languages do not present a uniform behaviour at this point is clearly illustrated by French. The EN status of the plural Nominalization in (9) is illustrated by the obligatory Theme interpretation of the complement (showing that no shift from EN to RN has been triggered by the pluralization), while a comparable observation can be made about (10), whose EN status is confirmed by the obligatory presence of a Theme in presence of a by-phrase, and its compatability with a verb like to film (favouring an event reading over a pure result reading). (9) Seules les observationsEvent d’animaux adultesTheme ont été prises en compte. “Only the observations of adult animals have been taken into account.”

(10)

Le général a filmé les désamorçagesEvent ??(de bombes lourdes) par les recrues. “The general filmed the dismantlements of heavy bombs by the young soldiers.”

Pluralized ENs and BNs: cross-linguistic variation and parallels The behaviour of ENs endowed with a plural mark reminds in an interesting way less well understood cross-linguistic differences between the “quantificational force” of (existential) BNs between Germanic and Romance languages. As observed by Delfitto & Schroten 1991, an object BN in a Germanic language like Dutch is capable, among others, to give an endpoint to durative predicates more easily than a BN in a Romance language like Italian (an observation apparently overlooked by Carlson 1977), as shown by its compatibility with an adverb in X time: (11) Gisteren heeft Jan vergelijkingen in twee uur opgelost. “Yesterday, John has solved equations in two hours.”

(12)

??Ieri Gianni ha risotto equazioni in due ore. “Yesterday, Gianni has solved equations in two hours.”

Interestingly, ENs endowed with a plural mark also behave differently in both groups of languages. As observed by Roeper 2000, object control is possible with singular discussion in (13) but blocked with plural discussions in (14): (13) The criminal needs discussion. (= discussion about him) (14) The criminal needs discussions. (= he discusses) In a language like Italian, however, object control is not blocked by the plural mark on an EN. (15) Questa persona richiede cura. (= taken care of him) “This person needs care.”

(16)

Questa persona richiede cure. ( = taken care of him (several times)) “This person needs cares.”

The same observation can be made for French des débats (the equivalent of Romance BNs) in: (17) Ce crime nécessite des débats. (= discussion about it (several times)) “This crime needs discussions.”

In my analysis of pluralized ENs and the contrasts mentioned, I show that cross-linguistic variation can be accounted for by the hypothesis put forward by Roodenburg 2004ab, who argues that the syntactic licensing of bare nominal arguments can involve various kinds of interaction between an underspecified NumP (containing the feature [αPl]) and a higher FP, tentatively labelled QP. 2

Friday, session 2

Processing of Grammatical Gender Agreement in L2 Spanish Nuria Sagarra (Pennsylvania State University)

Attention to input is necessary for second language (L2) development (e.g., Gass & Selinker, 2001) but is cognitively taxing, particularly at early stages of acquisition. The capacity to process and store incoming information—i.e., working memory (WM)—is limited and varies from person to person (e.g., Baddeley, 2003; Just & Carpenter, 1992). Whenever such capacity becomes insufficient to support the level of activation necessary to perform a specific task, processing will slow down and/or storage will decrease, resulting in worse morphosyntactic processing (see e.g., King & Just, 1991; Fiebach, Schlesewsky, & Friederici, 2002, for L1 evidence; and Ellis & Sinclair, 1996; Vos, Gunter, Schriefers, & Friederici, 2001, for L2 evidence). Because lack of proficiency imposes additional cognitive demands on WM, it has been suggested that beginning L2 learners tend to inhibit redundant information to release attentional resources that can be allocated more effectively (e.g., VanPatten, 1996, 2004). However, studies on this issue have ignored the role of WM and have employed methodological techniques that can only assess conscious knowledge about the target language (e.g., Lee, 1999, 2000; Rossomondo, 2005). To address this limitation, a subject-paced moving window experiment was conducted with native speakers of English who were beginning L2 learners of Spanish. To make the reading task more cognitively taxing and avoid regressions to the lexical items, when a word appeared on the screen the previous word disappeared automatically. Grammatical noun-adjective gender agreement was selected as the target form, because is formed by bound unstressed morphemes, has little communicative value and is acquired later than other linguistic forms. Results revealed no significant differences between the reading times of correct adjectives like roja in Carmen llevó la blusa de nylon roja a la tintorería and incorrect adjectives like rojo in Carmen llevó la blusa de nylon rojo a la tintorería. While these findings can be interpreted as evidence that participants ignored the redundant cue, the proximity of a masculine noun (nylon) and the participants’ L1 tendency to do low attachment may have influenced the reading times of the adjectives. To solve this obstacle, a second experiment was carried out. In this experiment, 311 English monolinguals in their third semester of study of Spanish completed two language proficiency tests, a pretest, a self-paced moving window task, and a WM test based on Waters, Caplan and Hildebrandt’s (1987) and Waters and Caplan’s (1996) reading span test. For the WM test, participants were asked to read 80 sentences in their native language (English) grouped into sets from 2 to 6 sentences, decide whether each sentence was plausible, and remember the final word of all sentence within a set. In order to obtain 1 point for the WM score, both the plausibility judgment and the sentence-final word had to be correct. In addition, the participants’ ability to suppress unwanted information was measured by adding the number of words that had been recalled incorrectly but that had appeared earlier (e.g., dentist would count as an intrusion in the sentence It was the dentist that the cavity extracted). For the moving window paradigm,

participants read 124 sentences word-by-word and answered a comprehension question after each sentence. Each participant read 24 experimental sentences divided into 6 sentences per condition: (1) Correct gender agreement within a phrase Carmen llevó la blusa roja a la tintorería (2) Incorrect gender agreement within a phrase Carmen llevó la blusa rojo a la tintorería (3) Correct gender agreement across clauses Carmen llevó la blusa que era roja a la tintorería (4) Incorrect gender agreement across clauses Carmen llevó la blusa que era rojo a la tintorería Results from a repeated-measures ANOVA suggest that beginners do not attend to morphological cues that encode the same morphosyntactic information than other cues previously encountered in visual stimuli when cues are located in different clauses (the head noun is in the main clause and the adjective is in a center-embedded subject relative clause). The ANCOVA conducted with WM as a covariate reveal that individual differences in the processing of redundant L2 morphosyntactic cues are not governed by working memory capacity. However, the ANCOVA carried out with intrusions as a covariate indicate that inhibitory control predicts processing of redundant cues in the input and can act as a compensatory device for lack of language proficiency.

Oana Savescu Ciucivara (New York University)

Sunday, session 21

Romanian-A PCC language?

The *me-lui or the Person Case Constraint (PCC) has been commonly taken to be a universal well – formedness condition, banning accusative clitics other than 3rd person when a dative 3rd person clitic is present (and similarly for agreement markers). The aim of this paper is two-fold: (i) I present evidence from Romanian pre- and postverbal clitic combinations that contradict the view that the PCC in Bonet’s (1991) formulation has universal validity; although the constraint has also been previously challenged by Haspelmath (2004), the data presented also contradict Haspelmath’s reporting of Romanian as a language in which the PCC is otherwise active; (ii) I provide a syntactic analysis of the combinatorial possibilities that clitics are subject to in Romanian. The following generalizations govern the distribution of clitics in Romanian: (i) clitic clusters are always found in the order dative>>accusative; (ii) a 3rd person dative clitic can freely combine with a 2nd person accusative clitic, contrary to what the PCC states (1); (iii) a 2nd person dative clitic cannot combine with a 1st person accusative clitic (2); (iv) 3rd person dative reflexive clitics cannot combine with 1st and 2nd person accusative clitics in Romanian (3); (v) interestingly, no combinatorial restrictions with respect to person occur in post verbal position, after gerunds and true imperatives, as can be seen in (4-6): (1) I te - am recomandat. 3rd dat 2nd acc have.I recommended I introduced you to him.. (3)

(2) *Ţi m - a prezentat Maria. 2nd dat 1st acc has presented Mary Mary has introduced me to you.

*Maria şi m-/ tea luat drept sclav. Mary refl 1st/2nd acc has taken as slave Mary has taken me/you to be her slave (for herself).

(4)

de nevastă, tata a câştigat mulţi bani. Dîndu -ţi -mă nd st giving 2 dat 1 acc of wife, father has gained much money Giving me to you in marriage, my father has gained a lot of money.

(5)

Luîndu – şi – mă/te drept martor, Petru a câştigat procesul. st nd Taking-refl.dat 1 /2 acc as witness, Peter has won trial.the Taking me as a witness for himself, Peter has won the trial.

(6)

drept martor, şi vei câştiga procesul! Ia - ţi - mă nd st nd trial.the Take.2sg 2 dat 1 acc as witness and will.2 win Take me as a witness (for yourself) and you will win the trial!

As they currently stand, current syntactic approaches to pronominal clitic combinations that fall under Bonet’s (1991) generalization (Anagnostopolou 2003, Bejar and Rezac 2003) cannot do justice to the Romanian facts. I suggest a different approach, namely that the co-occurrence restrictions that clitics are subject to in Romanian arise as the result of two requirements operating at the same time: a case requirement, according to which the dative clitic has to precede the accusative one, and a person requirement, which has the 1st person clitic always be the first one in the cluster. The two

hierarchies emerge as: (i) dative>>accusative; (ii) 1st person clitic >>3rd person clitic> 2nd person clitic >>3rd reflexive clitic. The rigidity that we notice with clitic sequences recalls the familiar strict ordering of functional projections in the phrase marker. I thus argue that the hierarchies above can be integrated into the functional sequence. I adopt a movement approach to cliticization in which object clitics are base-generated in the post verbal argument position and then target two different types of designated positions in the IP functional field: a pair of Kase projection (KP), of the familiar type, located immediately below TP, where the direct and indirect object clitic check uninterpretable Case features, and a set of Person projections, which are above Tense, and where clitics, being specified for Person features, are subsequently attracted to check uninterpretable Person features on the Person head. In the 'Person-field', I suggest that Romanian distinguishes several such projections: Person1P, where the 1st person clitic will end up, Person2P, hosting the 2nd person clitic, Person 3P, attracting the 3rd person clitic, and 3rd ReflexiveP, which will attract the 3rd person reflexive clitic (7): (7) Person1P>>Person 3P>>Person2P>>3ReflP>>TP>>KIO>>KDO>>…>>V The surface ordering of clitics is derived by feature checking driven movement, in a system in which the paths of the clitics to the relevant functional projections cross, rather than nest (in agreement with Bianchi 2005). We thus obtain an order preservation structure: the order of the clitics in the Case projections will necessarily be maintained even after the clitics have undergone subsequent movement to the Person projections, as can be seen in the derivation of the grammatical clitic combination in (8). The derivation in (9), which violates the prohibition against nesting, gives us the wrong output: (8) √ mi te (1st dat>>2nd acc) 9) *mă ţi (1st acc>>2nd dat) Person1P Person2P (TP) KPdat KPacc mi te 1 2

Person1P Person2P (TP) KPdat KPacc mă ţi 1 2

When clitics appear in post verbal position, after true imperatives and gerunds, I argue that the verb moves past the cliticization site, to a higher functional position (cf Kayne 1994, Laezlinger (1998)). Interestingly, person considerations no longer restrict the distribution of clitics within the cluster, as can be see in examples (4)-(6). However, the order dative>accusative is still maintained, so, in the light of the current proposal, the clitics move as far as KP and never reach PersonP. I suggest that with gerunds and true imperatives in Romanian, the PersonP is not merged at all, so the clitics are not attracted to this higher projection. Therefore, the person constraints due to nesting violations do not appear. The lack of person restrictions in these postverbal contexts is entirely unexpected under previous approaches to the PCC, but it is made sense of by the account proposed here, which attributes person and case effects, (when they hold), to interactions between the Person field and the Case field. I furthermore suggest that the absence of PersonP with true imperatives and gerunds may correlate with the absence of Tense in these environments (cf. Zanuttini 1997), as well as with the unavailability of subject-verb agreement. Selected References: Anagnostopolou, E. 2003. Strong and Weak Person Restrictions : A feature Checking Analysis,Benjamins, Bianchi, V. 2005. On the Syntax of Pronominal Arguments, to appear in Lingua, Bonet, E. 1991. Morphology after Syntax: Pronominal clitics in Romance. PhD Diss, MIT, Haspelmath, http://email.eva.mpg.de/~haspelmt/2004dpr.pdf, Zanuttini, R 1997 Negation and Clausal Structure. A Comparative Study of Romance Languages, OUP

Erik Schoorlemmer (LUCL, Leiden University)

Saturday, session 18

! # $#

"

%

# $#

&'(# '(#

% !

"

) * ) $# +

)

)

%

,,,

.

.'

& .

&

# '

/

'

# -

-

-

.

#

#

. -

-

-

0 0

!#&

#1

0&2

!

!

0&2 3

"

' !#& #

4

!

!

"""

' 5

-

6

'

7

.. )

.