ROMANCE LINGUISTICS 2008
AMSTERDAM STUDIES IN THE THEORY AND HISTORY OF LINGUISTIC SCIENCE General Editor E.F.K. KOERNER Zentrum für Allgemeine Sprachwissenschaft, Typologie und Universalienforschung, Berlin
[email protected] Series IV – CURRENT ISSUES IN LINGUISTIC THEORY Advisory Editorial Board Lyle Campbell (Salt Lake City) Sheila Embleton (Toronto) Elly van Gelderen (Tempe, Ariz.) Brian D. Joseph (Columbus, Ohio) John E. Joseph (Edinburgh) Manfred Krifka (Berlin) Martin Maiden (Oxford) E. Wyn Roberts (Vancouver, B.C.) Joseph C. Salmons (Madison, Wis.)
Volume 313
Karlos Arregi, Zsuzsanna Fagyal, Silvina A. Montrul and Annie Tremblay (eds.) Romance Linguistics 2008. Interactions in Romance. Selected papers from the 38th Linguistic Symposium on Romance Languages (LSRL), Urbana-Champaign, April 2008
ROMANCE LINGUISTICS 2008 INTERACTIONS IN ROMANCE SELECTED PAPERS FROM THE 38TH LINGUISTIC SYMPOSIUM ON ROMANCE LANGUAGES (LSRL), URBANA-CHAMPAIGN, APRIL 2008
Edited by
KARLOS ARREGI University of Chicago
ZSUZSANNA FAGYAL SILVINA A. MONTRUL ANNIE TREMBLAY University of Illinois at Urbana-Champaign
JOHN BENJAMINS PUBLISHING COMPANY AMSTERDAM/PHILADELPHIA
8
TM
The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.
Library of Congress Cataloging-in-Publication Data Linguistic Symposium on Romance Languages (38th : 2008 : Urbana-Champaign, Ill.) Romance linguistics 2008 : interactions in romance : selected papers from the 38th linguistic symposium on romance languages (LSRL), Urbana-Champaign, April 2008 / edited by Karlos Arregi ... [et al.]. p. cm. (Amsterdam studies in the theory and history of linguistic science. Series IV, Current Issues in Linguistic Theory, issn 0304-0763 ; v. 313) “The papers selected for this volume were originally presented at the 38th Linguistic Symposium on Romance Languages held at the University of Illinois at Urbana Champaign on April 3-5, 2008. Linguists working on Romance languages from North America, South America and Europe met to discuss the latest developments in the field of Romance linguistics.” Includes bibliographical references and index. 1. Romance languages--Congresses. I. Arregi, Karlos. II. Title: Interactions in romance. PC11.L53â•…â•… 2010 440--dc22 2010022710 isbn 978 90 272 4831 2 (Hb ; alk. paper) isbn 978 90 272 8787 8 (Eb)
© 2010 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa
Table of contents Acknowledgments Editors’ introduction: Interactions in Romance
vii 1
part i.╇ Language contact and bilingualism Subject pronoun expression in bilinguals of two null subject languages Hilary Barnes
9
Where are hiatuses left? A comparative study of vocalic sequences in Argentine Spanish Laura Colantoni & Anna Limanni
23
Loanword adaptation in the French of Spanish-speaking immigrants in Montréal Michael L. Friesner
39
part 2.╇ Phonology and interfaces Morphology and phonology of word-final vowel deletion in spoken Tuscan Italian Luigia€Garrapa & Judith Meinschaefer
57
Relativization, intonational phrases and rich left peripheries in Romanian Simona Herdan
73
Stress domain effects in French phonology and phonological development Yvan Rose & Christophe dos Santos
89
part 3.╇ Syntax and morphophonology Syntactic realizations of plural in Romance and Germanic nominalizations 107 Artemis Alexiadou, Gianina Iordăchioaia & Elena Soare
Romance Linguistics 2008
The syntax of Spanish parecer and the status of little pro Adolfo Ausín
125
Two types of (apparently) ditransitive light verb constructions María Cristina Cuervo
139
Modal Ellipsis in French, Spanish and Italian: Evidence for a TP-deletion analysis Anne Dagnac Optional prepositions in Brazilian Portuguese Mary A. Kato
157
171
An apparent ‘number case constraint’ in Romanian: The role of syncretism 185 Andrew Ira Nevins & Oana Săvescu part 4.╇ Semantics and morphology Generic bare singulars in Brazilian Portuguese Carmen Dobrovie-Sorin & Roberta Pires de Oliveira
203
Aspect shift in stative verbs and their arguments Joshua Rodríguez
217
part 5.╇ Psycholinguistics Experimenting with wh-movement in Spanish Grant Goodall
233
How Spanish phonotactics informs psycholinguistic models of speech production Michael Shelton, Chip Gerfen & Nicolás Gutiérrez Palma
249
Index
265
Acknowledgments The papers selected for this volume were originally presented at the 38th Linguistic Symposium on Romance Languages held at the University of Illinois at Urbana Champaign on April 3–5, 2008. Linguists working on Romance languages from North America, South America, and Europe met to discuss the latest developments in the field of Romance linguistics. We gratefully acknowledge the Department of Linguistics, the Department of French, the Department of Spanish, Italian and Portuguese, the College of Liberal Arts and Sciences, the School of Languages, Cultures and Linguistics, the Center for Advanced Study, and the Beckman Institute at the University of Illinois, Urbana-Champaign for their generous financial support for this event. The selected papers went through a rigorous double-blind review process. We would like to express our gratitude to all the reviewers who contributed their time and expertise: Lourdes Aguilar, Daniel Altshuler, Julie Auger, Nicoleta Bateman, Dennis Bouchard, Travis Bradley, Joyce Bruhn de Garavito, José Camacho, Manuel DíazCampos, Elena Castroviejo Miró, Sonia Colina, Laura Colantoni, João Costa, Marie-Hélène Côté, Laurent Dekydstpotter, Mariapaola D’Imperio, Gorka Elordieta, Olga Fernández Soriano, Anastasia Giannakidou, Alessandra Giorgi, Grant Goodall, John Grinstead, Javier Gutiérrez-Rexach, David Heap, José Hualde, Tania Ionin, Haike Jacobs, Svetlana Kaminskaïa, Paula Kempchinsky, Peter Lasersohn, Juana Liceras, Luis López, Marta Luján, Pascual Masullo, Jaume Mateu, Jean-Pierre Montreuil, Francisco Ordóñez, Silvia Perpiñán, François Poiré, Cecilia Poletto, Lara Reglero, Morris Richard, Teresa Satterfield, Ryan Shosted, Roumyana Slabakova, Brian Smith, Cristina Schmitt, Esther Torrego, Irene Vogel, Thomas Werner, Malcah Yaeger-Dror, James Hye Suk Yoon, and María Luisa Zubizarreta. Finally, we thank the editor of the Current Issues in Linguistic Theory series E.F.K. Koerner for useful feedback and Anke de Looper of John Benjamins for assisting us with the publication process. All remaining errors are our own. Karlos Arregi, Zsuzsanna Fagyal, Silvina Montrul, Annie Tremblay Champaign-Urbana and Chicago, Illinois February 2010
Editors’ introduction Interactions in Romance
This volume consists of a selection of sixteen theoretical and methodological studies presented at the 38th Linguistic Symposium of the Romance Languages held in Champaign-Urbana, Illinois in the spring of 2008. The unifying theme of these papers is commonly referred to as linguistic interfaces, i.e., links between different grammar modules (lexicon, phonology, morphology, syntax, semantics), as well as between grammar and articulatory, perceptual, and cognitiveintentional systems (e.g., Chomsky 1995, Jackendoff 2002, Ramchand & Reiss 2007). Furthermore, in addition to focusing on linguistic theory, the papers explore connections with psycholinguistics, sociolinguistics, language-acquisition, and bilingualism. The selection of Interactions in Romance as the title of the volume also reflects the fact that, in addition to addressing interdisciplinary issues, several of the papers make significant typological-comparative contributions to the issue of the theoretical and practical dynamic connections between Romance and other languages. Interactions between, and interfaces within, the Romance languages themselves, as well as between Romance languages and other languages, are discussed in several papers, as all major Romance languages (Spanish, French, Italian, Romanian, and Portuguese), and many of the languages with which they are currently in contact are represented in the volume. In order to appreciate both the particular focus and the connections between the selected papers, we have organized them into five areas, as described below. 1. Language contact and bilingualism The papers in the first section of the volume shed light on linguistic interactions between phonology and morpho-syntax in stable societal bilingual situations. Each paper reports on a study of bilingual individuals or communities, and seeks to determine the effects of bilingualism on subject-pronoun expression, hiatusdiphthong realizations, and loanword adaptation. Barnes investigates variable patterns of subject-pronoun expression in the speech of Spanish-Veneto adult language learners, heritage speakers, and speakers
Romance Linguistics 2008
undergoing L1 attrition in Mexico. In these null subject varieties, variable patterns have previously been attributed to transfer/interference and loss of discourse-pragmatic constraints, among other factors. The study uses the variationist sociolinguistic toolkit to analyze naturalistic data quantitatively, and reveals a marked increase in overall rates of pronoun expression in Chipilo-contact Spanish relative to monolingual Mexican Spanish. The patterning of overt pronoun use is similar to that of monolingual varieties of Mexican Spanish, which is explained in terms of cognitive economy. In the second paper in this section, Colantoni and Limani examine intervocalic hiatus sequences in contact and non-contact Argentine Spanish, suggesting that dialect-specific coarticulation strategies rather than economy of cognitive representations account for the emergence of this sequence. The laboratory phonology-type comparison between a contact and a non-contact variety shows that while hiatuses are more frequent in the former, both varieties share the pan-Hispanic preference for diphthongs. Thus, fine-grained dialect-specific strategies essentially achieve the same percept. Contact and non-contact dialects of Argentine Spanish differ from each other and from Peninsular varieties in the conditioning environments where hiatuses are still realized. In the contact variety, few exceptional hiatuses are maintained and the etymological origin of the sequence is a relatively weak factor. In the last paper in this section, Friesner examines the degree of phonological integration of loanwords in Montreal French, claiming that different strategies borrowed from contacting languages result in variable language-specific outcomes. The adaptation of rhotics in loanwords in French by Hispanophone and Francophone Montrealers shows the mutual influence of native and nonnative speakers on the outcome of borrowing. Multivariate analysis reveals, among other things, that immigrants only acquire some of the constraints typical to natives, and second-generation immigrants share the grammar of the larger community but resemble other Hispanophones in terms of their rates of use of each variant at play. 2. Phonology and interfaces The second section examines linguistic interactions between phonology and other components of the grammar, as well as between different prosodic domains within phonology. The papers aim to explain phonological restrictions that depend on the morphological status of words, syntactic restrictions that depend on the phonological properties of phrases, and phonological restrictions within the syllable that depend on higher-level prosodic constituents such as the foot.
Editors’ introduction
Garrapa and Meinschaefer investigate vowel elision in function words followed by vowel-initial lexical words in spoken Tuscan Italian with the goal of characterizing the interplay between phonology and morphology in Italian elision. Through an analysis of a corpus of spoken Tuscan Italian and the collection of experimental data, vowel elision is shown to vary as a function of the morphological status of the function word: in determiners, it is more categorical and can be analyzed in part as allomorphy, whereas in pronouns it approximates free variation. These patterns of variation can be modeled by a gradual learning algorithm in an optimality-theoretic constraint hierarchy, but only if two different grammars are learned for elision, one for determiners and one for pronouns. The paper by Herdan focuses on the intonation-syntax interface. In Romanian, the choice of relativizer in relative clauses headed by bare quantifiers depends on the phonological properties of these phrases. Herdan argues that the location of the intonational phrase boundary at the level of the relative clause determines which relativizer can be used. Relativizer restrictions arise from violations of the requirement to phonetically mark the edge of an intonational phrase, suggesting that the structural position of the relativizer determines its grammaticality. The availability of articles in the nominal domain also has an effect on the relativizer choice. The third paper in this section examines the interaction between prosody and lexical representations in French. Rose and Santos show that stress domain effects arise in French when adult Québec French and child European French are compared. On the one hand, Québec French is known to have an allophonic process of closed-syllable laxing, which targets final vowels even though these vowels should be syllabified as open syllables in lexical representations. On the other hand, data from a first language learner of European French show an asymmetry in her production of CVC versus CVCV target forms, with the former displaying full preservation with concomitant manner harmony of both consonants, and the latter undergoing deletion of the initial syllable if the consonants are not manner-harmonic in the input. It is proposed that these distinct but related phonological patterns in French can be explained by drawing a formal distinction between phonological representation and phonetic implementation. 3. Syntax and morphophonology Current work on syntax directs a great deal of attention to the effect that syntactic structure has on the morphophonological realization of linguistic expressions. Several papers (see below) in this volume deal with phenomena in this area, including the relation between syntax, morphology, and argument structure, the
Romance Linguistics 2008
interaction of morphology and syntax in accounting for clitic combinations, and ellipsis and other types of null anaphora. All of them provide arguments for the important role that syntax plays in accounting for these phenomena. Two papers explore the relation between syntax and argument structure and their effect on morphology. The research reported by Alexiadou, Iordăchioaia, and Soare explores the ways in which plurality is related to argument structure in Romance and Germanic nominal expressions. They argue that aspectual properties like telicity and boundedness play a crucial role in this relation, thus advancing our understanding of the interplay between nominal and verbal properties in nominalizations. They provide evidence for crosslinguistic differences in this relation between number, argument structure, and aspect, showing how it is encoded in different categories such as verbal aspect, nominal number, and classifiers. Their evidence is drawn from Romanian nominalizations as well as infinitive/supine forms in this language, Spanish nominal infinitives, and English nominalizations. Argument structure is also the main topic of Cuervo’s paper on Spanish ditransitive light verb constructions with dar (‘give’). Cuervo provides evidence that these belong to two different types: some are double object constructions, while others are experiencer unaccusative structures. She shows that the morphological and syntactic properties of both types of constructions (including case and restrictions on bare nominals) are to be accounted for syntactically, not from lexical stipulation. The relation between syntax and morphology is also explored in the paper by Nevins and Săvescu, which discusses restrictions on Romanian postverbal clitic combinations. In particular, combinations involving first and second plural clitics are more restricted than their singular counterparts. The authors argue that this is related to the fact that these plural clitics display dative-accusative case syncretism, and propose an analysis in which this syncretism is due to a syntactic condition on plural clitics (inherently marked as [+animate]) which, together with an independently motivated analysis of the Person Case Constraint, also explains their combinatory restrictions. Three other papers explore the syntax of phonologically null elements, and provide evidence that their syntax is more complex than is apparent on the surface, in some cases arguing that it is very similar to their overt counterparts. Ausín discusses the structure of Spanish null subject sentences, arguing, contrary to several recent claims, that such sentences contain a null argument in preverbal position. His supporting evidence comes from the binding properties of these arguments, concentrating on their behavior in raising sentences with parecer ‘seem’. Dagnac argues for an ellipsis (PF deletion) analysis of gaps after root modals in French, Spanish, and Italian. She argues that, contrary to previous claims,
Editors’ introduction
these constructions allow for A’-movement from the gap, which shows that these gaps must have full syntactic structures, as predicted by the ellipsis analysis and not by one in terms of a null proform. Dagnac also discusses important differences between this type of ellipsis and the more familiar type of VP ellipsis in English, arguing that these differences have to do with the size of the elided constituent (TP in Romance, VP in English), and differences between the modals in these languages. Finally, Kato’s paper discusses the syntax of optional prepositions in Brazilian Portuguese, building on a previous analysis of a similar phenomenon in French by Denis Bouchard. She shows that the phenomenon is not restricted to relative clauses, and argues for an analysis in which the missing prepositions encode inherent case and are only licensed in A’-positions, where a DP can have a default case. 4. Semantics and morphology Two papers in this volume contain detailed discussion of the relation between semantics and certain types of morphology in both the nominal and the verbal domain in Romance languages. Dobrovie-Sorin and Pires de Oliveira discuss the generic readings of bare singulars in Brazilian Portuguese, arguing for an analysis in which bare singulars are kind-referring, and against a bound indefinite analysis. Their kind-referring account is based on the same down operator that accounts for the more familiar bare plurals, and explains differences between generic bare singulars and generic definite singulars, which are also kind-referring. Rodríguez discusses the semantic effect of aspect shift that different types of aspectual morphology (Spanish perfective/imperfective past tenses and the progressive) have on stative verbs. The account exploits lexical ambiguities to provide a rich set of VP Aktionsarten via normal compositional semantics. These different VP interpretations are filtered out by specific aspectual morphology. Thus, the analysis accounts for the alleged shifting effect compositionally from both the meaning of the VP and from aspectual morphology, providing a natural explanation for the absence of certain cases of aspect shift. 5. Psycholinguistics The fifth and final section of the volume examines linguistic interactions between syntactic and phonological representations and their implementation in real time. The papers, both on Spanish, use experimental psycholinguistic methods
Romance Linguistics 2008
with native speakers to test models and analyses proposed in theoretical syntax and in phonology. Goodall claims that native speakers’ judgments of wh-questions in Spanish are gradient and depend on the nature of the filler and of the intervening subject. On the basis of syntactic data elicited experimentally, Goodall proposes that this gradience results from the interplay between the properties of the syntax, including the type of wh-movement and the nature and position of the subject relative to the verb, and the properties of the processor, which has limited resources to process wh-dependencies and establish discourse referents. This analysis predicts a correlation between the rate of overt pronouns in any given variety of Spanish and the ability of a wh-dependency to tolerate an intervening subject, which is in fact confirmed in a comparison of Caribbean and mainland Latin-American Spanish. The influence of language-specific phonotactic patterns on speech production is investigated in Spanish in the last paper of the volume, by Shelton, Gerfen, and Gutiérrez-Palma. Data obtained from a delayed naming task eliciting nonwords that contained licit or proscribed patterns of stress and syllable structure provide support for a postlexical locus of syllable-level encoding, but also pose challenges for this model. The results point to the necessity of identifying more clearly the locus of and interaction between stress-level and syllable-level encoding, while suggesting that the self-monitoring system may track phonotactic patterns in addition to conceptual accuracy. References Chomsky, Noam. 1995. The Minimalist Program. Cambridge, Mass: MIT Press. Jackendoff, Ray. 2002. Foundations of Language. Oxford: Oxford University Press. Ramchand, Gillian & César Reiss (eds.) 2007. The Oxford Handbook of Linguistic Interfaces. Oxford: Oxford University Press.
part 1
Language contact and bilingualism
Subject pronoun expression in bilinguals of two null subject languages Hilary Barnes
Fayetteville State University This paper examines subject pronoun expression in the speech of SpanishVeneto bilinguals in central Mexico. Non-target subject expression has been found among adult language learners, heritage speakers, and speakers undergoing L1 attrition. Such patterns have been variously attributed to transfer/interference and loss of discourse-pragmatic constraints, among other factors. The situation discussed here is unique in that both languages are null subject languages in an environment of sustained bilingualism. Drawing on a quantitative analysis of naturalistic data, the present work reveals a marked increase in overall rates of pronoun expression in Chipilo-contact Spanish relative to monolingual Mexican Spanish; however, the patterning of overt pronoun use is similar to that of monolingual varieties of Mexican Spanish. The increase is explained in terms of cognitive economy.
1. Introduction This paper examines the structural consequences of language contact in Chipilo, a Veneto-Spanish bilingual community in central Mexico.1 Chipilo was established in the late 19th century when the Mexican government began recruiting Europeans to colonize the country and help spark the economy. Various groups of Italians came to Mexico from different Veneto-speaking regions in northern Italy, including Lombardy, Véneto, and Trentino. Chipilo was founded in 1882 by a fairly homogenous group of approximately 560 immigrants primarily from Veneto-speaking Segusino and neighboring towns in the northern Italian provinces of Treviso and Belluno. The founders of Chipilo were some of the last settlers and remain the only Italian community still thriving in Mexico (Romani 1992). As with other Italian immigrants in 1. I would like to thank Almeida Jacqueline Toribio, John Lipski, Ana de Prada Pérez, and the Chipileños for their help with this research project, and Teresa Satterfield and an anonymous reviewer for their comments.
Hilary Barnes
Mexico, the Chipileños were not immediately satisfied with the land they were assigned by the Mexican government (MacKay 1992; Romani 1992). The lands of the former Chipiloc and Tenamaxtla haciendas were sterile and unused. As a consequence, the first years were difficult and unproductive. However, by 1892, the dairy business supported and improved the town’s economy, and today, Chipilo is one of the most economically developed areas in the region and provides employment opportunities to neighboring communities. Studies conducted in the 1980s show that the overwhelming majority of Chipileños (93%) worked as agriculturists with only 2.04% of the population working in business, 2.02% in industry, and 1.05%, mostly younger people, working in offices, banks, and factories in Puebla (Romani 1992). There is a very small body of work examining the language situation of Chipilo, focusing largely on sociolinguistic issues. Romani’s (1992) study revealed nearly all Chipileños (99.2%) to be bilingual, with 95.31% active bilinguals and 3.90% passive bilinguals. There were few monolingual speakers of Spanish and no monolingual speakers of Veneto in the community. Romani reports that most Chipileños of Italian descent (and a small number of Mexican descent) learned Veneto in the home. Most children learned Spanish when they started school; however, children from mixed marriages were also observed to learn Spanish at home. She described the language situation as one of diglossia: Veneto is the language of the home, while Spanish is the language of the church and school. Moreover, she reports that Veneto and Spanish are used as markers of identity: in intraethnic relationships, the use of Veneto reflects solidarity with the Veneto group, whereas in inter-ethnic relationships, the use of Spanish signals identity with the national group.€Significantly, both Veneto and Spanish are held in high prestige. Thus it is not surprising that speakers demonstrate a strong degree of loyalty to both languages. Veneto speakers feel that their language serves as an important means of intra-group communication and they express desire to maintain their heritage language, pass it on to younger generations, and standardize it.2 At the same time, Chipileños recognize the importance and value of Spanish. Romani notes that “los dos idiomas no representan valores en lucha, sino que poseen un diferente significado social y una diferente utilidad que, en esta fase del contacto, no se enfrentan abiertamente” the languages do not represent battling values, but rather they possess a different social meaning and a different use that, in this phase of the contact, are not openly confronting each other (1992:€86). A different perspective is offered by MacKay (1992), who suggests that while the social networks in Chipilo are generally dense and strong, open social networks attributed to employment opportunities outside of the Chipilo community may be weakening 2. Efforts have been made to use social networking websites and email list-serves, created to exclusively write Veneto.
Subject pronoun expression in bilinguals of two null subject languages
the maintenance of Veneto. In addition, while membership in the community appears to entail knowledge of Veneto, the use of Veneto with peers is diminishing for some speakers and Mackay remarks that “the degree to which children are acquiring Veneto outside of the home is not clear” (1992:€143). She notes that codeswitching is becoming more common in younger generations. Thus, at the time of her writing, MacKay deemed the future of Veneto in Chipilo uncertain. In contrast, the fate of Spanish in Chipilo is assured. Still, there is little doubt that the language has been altered by its coexistence alongside Veneto. Most salient to the visitor are the lexical and prosodic characteristics of the local contact variety; of interest here are the syntactic properties that emerge in Chipilo Spanish. The syntax of contact varieties of Spanish have been studied in a number of contexts and populations, most commonly, in the United States among speakers of Spanish language heritage (Montrul 2002, 2004; Silva-Corvalán 1994/2000), and in Europe among children who acquire Spanish simultaneously alongside another language (Müller and Hulk 2001; Paradis and Navarro 2003) and among adults who achieve near-native proficiency in Spanish (Tsimpli et.al. 2004; Sorace 2005). In situations of language contact and bilingualism, some areas of the syntax are thought to be more stable, others more ‘permeable’ or susceptible to indeterminacy and variation. Among the stable properties are those that derive from abstract morpho-syntactic features of Case and Agreement, e.g., whether the language is VO or OV. But especially prone to interlingual effects are those grammatical properties that are regulated by discursive/pragmatic considerations. Both of these properties are exemplified with the omission and expression of subjects: null subjects are licensed in Spanish by Agreement features, but the distribution of null vs. overt subjects is regulated by discursive conditions such as topic and focus. While the expression and interpretation of null and overt subject pronouns has been found to be an area of indeterminacy in contact and bilingual grammars, such patterns have most often been found in contexts of advanced and near-native L2 acquisition and L1 attrition (Tsimpli, et.al. 2004; Sorace 2005). Most often, these studies examine the expression of null and overt subject pronouns in bilinguals of a null subject language (e.g., Spanish or Italian) in contact with a non-null subject language (e.g., English), and consequently often attribute the increase in rate of overt pronoun use to influence from the non-null subject language (Tsimpli et.al. 2004). However, since the distribution of null and overt pronouns is regulated by interface conditions (an area of instability), their nonnormative expression and interpretation might occur irrespective of the types of languages in contact. That is, non-target expression and interpretation might be found even among bilinguals of two null subject languages. The present study examines precisely this type of language contact: Spanish in contact with another null subject language, in this case, Veneto, a Northern Italian dialect.
Hilary Barnes
This study has two goals: (1) to present the sociohistorical and linguistic background of a unique contact situation in central Mexico, and (2) to examine the permeability of the expression of null and overt subject pronouns in a situation of sustained bilingualism between two null subject languages. This language contact situation presents a unique window for the study of contact Spanish for a number of reasons. First, the linguistic situation provides insight into a less studied area of language contact research – that between two typologically similar languages. In addition, Veneto has co-existed with Spanish for over 127 years and is the first language of many, if not all, of the bilinguals in the community, allowing for the examination of inter-lingual effects in a situation of sustained bilingualism. And unlike many other instances of language contact in Mexico, both community languages are ascribed high prestige, therefore slowing the process of language shift from Veneto to Spanish. This paper is organized as follows. Section€2 provides a theoretical background of the distribution of null and overt subject pronouns in Spanish and Veneto and further explains the permeability of the discourse-pragmatics interface. Section€3 describes the present study including participants and methodology. Section€4 includes the results and discussion. Section€5 discusses the conclusions. 2. Theoretical background 2.1
Distribution of null and overt subjects
Spanish and Veneto are null subject languages. As shown in (1) and (2), both employ strong pronouns, and both allow for their absence. (1) Ø/Él/Juan habla ‘He/Juan speaks’ (2) Ø/Ela/Maria *(la) parla ‘She/Maria speaks’ Thus, both Spanish and Veneto license null subjects. Unlike Spanish, however, the variety of Veneto spoken in Chipilo demonstrates obligatory subject clitic pronouns in the 2sg, 3sg, and 3pl forms (see Poletto 2000 and Goria 2004 for overviews of subject clitics in Northern Italian Dialects). (3)
Ela la parla She SCL speaks La parla SCL speaks
Subject pronoun expression in bilinguals of two null subject languages
*Parla Speaks *Ela parla She speaks ‘She speaks’
Spanish and Veneto also share similar conditions on the expression and interpretation of overt pronouns. As previously mentioned, the expression of overt subject pronouns is regulated by discourse-pragmatic factors. For example, overt pronouns are required when there is a switch in referent in discourse, as seen in the following discourse connected utterances (examples from Montrul 2004): (4) Pepe no vino hoy a trabajar. → *Pepe/?él/ Ø estará enfermo. ‘Pepe did not come to work today. → He must be sick.’ (5) Hoy no fui a trabajar. → Pepe/él/*Ø pensó que estaba enfermo. ‘Today I did not go to work. → Pepe/he thought I was sick.’ In example (4), the referent does not change and therefore an overt pronoun would be pragmatically infelicitous. In (5), there is a shift in topic and an overt subject pronoun is required. Overt subjects are also required when new information is introduced (6) or when a subject is focused or topicalized (7). (6) ¿Quién vino? → Él/Mario/*Ø vino ‘Who came? → He/Mario/*Ø came’ (7) El periodistai dijo que éli/ÉLi no había escrito ese reporte. ‘The journalist said that he (himself) had not written that report.’ 2.2
The syntax-discourse interface
As should be clear from the preceding examples, the distribution of null and overt subject pronouns in null subject languages is regulated by conditions at the interface of syntax and pragmatics. And as noted, recent research has shown that such properties are especially prone to interlingual influence. In many cases, the studies examine a null subject language in contact with a nonnull subject language. For instance, Tsimpli et. al. (2004) found that English advanced L2 Italian speakers use an overt subject pronoun in environments where a native Italian speaker would use a null pronoun, a finding attributed to L1 interference. However, as previously mentioned, since the distribution of null and overt pronouns is regulated by interface conditions (an area of instability), their non-normative expression and interpretation might occur irrespective of the types of languages in contact. Margaza and Bel (2006) examined
Hilary Barnes
the linguistic behaviors of learners of Spanish whose native language is Greek – both Spanish and Greek are null subject languages and share the same pragmatic conditions on their expression and interpretation – and found that intermediate level learners used overt pronouns in contexts where both Spanish and Greek would use a null pronoun. This shows that non-target subject expression may not be due to L1 interference but rather may be a result of bilingualism itself. The syntax-discourse/pragmatics interface is also the locus of optionality in bilingual L1 acquisition. Müller and Hulk (2001) note that there are two constraints on crosslinguistic influence in bilingual L1 acquisition: first, language A must allow for more than one possible structural analysis of some aspect of the grammar and language B must provide positive evidence for one of these analyses. Second, the area in question must require the interfacing between syntax and pragmatics (the ‘C-domain’). They propose that acquisition proceeds in two phases – before and after the instantiation of the C-domain – and that the locus of interlingual influence is the C-domain. Their hypothesis is supported in their analysis of pragmatically regulated object omission in monolingual German-, Dutch-, French-, and Italian-speaking children and Dutch-French and GermanItalian bilingual children. In the first phase, the bilingual children behave like monolingual children in that they use object drop, but they use it to a higher degree. Once the C-domain is integrated, object omission decreases to more adultlike levels but the transitional period lasts longer in the bilingual children. Similar results are also reported in Paradis and Navarro (2003) for a Spanish-English bilingual child who used more overt pronouns – and for a longer period of time – than her monolingual counterparts. Serratrice, Sorace, and Paoli (2004) expand on Müller and Hulk’s analysis and argue that even though the C-domain is in place, crosslinguistic influence should still be possible. They propose that even after the instantiation of the C-domain bilinguals will not make syntactically ungrammatical errors but rather pragmatically inappropriate ones. They suggest that to strengthen the argument for crosslinguistic influence in bilingual acquisition, bilingual English-Italian children (non-null subject and null subject languages) need to be compared with bilinguals of two null subject languages. They predict that bilinguals of SpanishItalian, for example, will make fewer errors than the English-Italian bilinguals. However, Bini (1993) found that Italian L2 learners of Spanish initially produce infelicitous overt subject pronouns and later converged on the monolingual norm. Therefore, as discussed above, interlingual influence may not be due to interference from the L1 (recall that Margaza and Bel (2006) found similar effects on the expression of overt subject pronouns in the Spanish interlanguage of Greek speakers), but to bilingualism itself. Sorace (2005) attributes this difference
Subject pronoun expression in bilinguals of two null subject languages
between bilinguals and monolinguals to processing abilities, stating that “these differences selectively concern the access to and the integration of the syntactic component and the coordination of multiple sources of knowledge…” noting that “speakers may resort to default strategies, such as the use of overt subject pronouns and the positioning of subjects in preverbal position” (76). The Veneto-Spanish bilingual situation in Chipilo provides an ideal environment to further pursue this line of inquiry given that many bilingual Chipileños are raised as simultaneous or early bilinguals of two null subject languages in an environment of sustained bilingualism. By examining two null subject languages, the factor of L1 interference is removed. If the expression of subject pronouns among Veneto-Spanish bilinguals is found to be native-like in Spanish (i.e., demonstrating patterns and rates of monolingual Mexican Spanish speakers), then the interface between syntax and pragmatics may be less permeable in situations of sustained bilingualism. (Of course, this would not provide evidence for or against L1 interference in general. Further study between two languages, such as English and Spanish, in a situation of sustained bilingualism would be able to examine the role of L1 interference in optionality.) In addition, while previous studies have focused on (bilingual children and) adults who have acquired a second language in an educational setting, and whose language has been examined in experimental settings, the proposed study of Chipilo will draw on a distinct population and multiple methodologies: adults in a bilingual community, studied via naturalistic means. 3. Present study Participants and methodology
3.1
This study includes the interview data of 16 bilingual participants, ranging in age from 18–96. Table€1 provides a breakdown of age groups and gender. All participants were interviewed by the author during two field work trips in 2006 and 2008. Each interview took place in the participant’s home or in a public place and lasted between 20 and 90 minutes. Interview topics focused on local customs and traditions, language use, and daily life within the community. Table 1.╇ Participants € women men
Age Group 1 (under 30)
Age Group 2 (30–60)
Age Group 3 (over 60)
2 3
3 3
3 2
Hilary Barnes
3.2
Analysis
For this study, the dependent variable was the use of null and overt subject pronouns. Pronouns were coded for the following independent linguistic and social variables: person, number, tense, topic (topic, topic shift, contrast), gender and age.3 All instances of possible usage of null and overt pronouns were included in the analysis, in what has been called the “envelope of variation” (Milroy and Gordon 2003; Otheguy et.al. 2007). Following Otheguy et.al. (2007), several instances of pronoun usage are considered “outside” of the envelope of variation in assessing pronoun use, such as impersonal verbs (llover) and set phrases (qué sé yo). Included in the analysis were all other instances of overt subject pronouns, including any repetitive, infelicitous uses of overt pronouns and infelicitous uses of null pronouns. These include environments where a speaker does not change referents but uses an overt pronoun with each finite verb and environments with switch-reference where a speaker does not use an overt pronoun. 4. Results and Discussion Initial VARBRUL analysis of 1178 tokens revealed that null pronouns are preferred overall except in the first person singular, imperfect tense, and in cases of contrast and shift where overt subject pronouns are favored. As shown in Table€2, the overall rate of expression of overt subject pronouns is 26%; this rate represents a marked increase when compared to the monolingual Mexican Spanish rate of 19% reported in Otheguy et al. (2007). However, the patterning of pronoun expression is in line with monolingual (non-contact) Mexican Spanish norms: overt pronouns are preferred only with the first singular (Figure 1), overt pronouns occur more with the imperfect tense (Figure 2), and there is a definite preference for null subjects with Topic and a preference for overt subjects in cases of shift and contrast (Figure 3). Table 2.╇ Overall rate of null and overt pronoun expression Pronouns Null Overt
Total #
%
864 314
73 26
3. Gender and age were included given that based on these factors a participant may have more or less exposure to Spanish. Women work primarily in the home and younger generations are more commonly finding employment and education opportunities outside of Chipilo.
Subject pronoun expression in bilinguals of two null subject languages 90 80 70 60 50 40 30 20
null
10
overt
0 1sg
1pl
3sg
3pl
Figure 1.╇ Overall Percentage of Null and Overt Subjects for Person and Number 80 70 60 50 40 30 20 null
10
overt
0 present
preterite
imperfect
Figure 2.╇ Overall Percentage of Null and Overt Subjects for Tense 100 80 60 40 20
null overt
0 topic
shift
contrast
Figure 3.╇ Overall Percentage of Null and Overt Subjects for Topic
Hilary Barnes
Examining the data more closely, we find instances of infelicitous pronouns, some uses mirror what is found in monolingual varieties of the language; others may be unique to Chipilo contact Spanish. The following sample provides some examples of infelicitous overt 1sg pronouns.
(8) ...y somos mexicanos, vaya, nada más con sangre de descendientes italianas, pero en realidad yo, cuando juega un partido de fútbol México Italia yo apago la televisión yo no quiero ver perder a ninguno de los dos, vaya, y creo yo que hasta me duele más cuando pierde México, vaya, me da más dolor, vaya, pero yo para evitar y no sufrir, yo, yo mejor apago la televisión y a volar, yo no quiero ver que se enfrenten dos, dos razas mías (Male, born 1958) ...and we are Mexicans, just with blood from Italian descendents, but in reality I, when Mexico and Italy play a soccer match I turn off the television I don’t want to see either team lose, and I think that it even hurts me more when Mexico loses, it causes more pain, but to avoid it and not suffer, I, I’m better off to turn off the television and fly, I don’t want to see them face each other, two of my races
(9) ...yo cuando era yo chiquito, me acuerdo que hacían la polenta aquí en esta esquina... Pero era yo, yo creo que dos años tenía yo, y me acuerdo (Male, born 1937) ...I when I was young, I remember that they made polenta here in this corner… but I was, I think that two years I was, and I remember.
A possible explanation for these redundant pronouns is priming, defined by Travis as “the process whereby the use of a certain structure in one utterance functions as a prime on a subsequent utterance, such that that same structure is repeated” (2007:€101).4 Another factor known to condition the use of overt pronouns is the semantics of the verb. Silva-Corvalán (1994) found that some verbs, what she calls estimative verbs, are commonly found with an expressed subject because they show an opposing opinion. In the following examples, the estimative verbs appear with overt 1sg pronouns (Silva-Corvalán 1994; Enríquez 1984). (10) Yo me considero mexicana, yo nací aquí, yo quiero mucho a México, estamos aquí, estamos contentos, yo sí estoy contenta vivir en México (Female, born 1938) 4. Also noteworthy is the postverbal position of the pronoun, which is also repeated. A reviewer pointed out that the postverbal placement of the overt pronoun in these examples may be a means of differentiating the two languages, along the lines of Seuren and Wekker (1986).
Subject pronoun expression in bilinguals of two null subject languages
I consider myself Mexican, I was born here, I love Mexico very much. We are here, we are happy, I am happy to live in Mexico. (11) pues yo creo que mexicano/chipileño porque la identidad es muy distinta de la, de un mexicano. Yo siento que es, tiene más identidad un chipileño que un mexicano (Male, born 1983) well I think that Mexican/Chipileño because the identity is very distinct from that of a, of a Mexican. I feel that it is, a Chipileño has more identity than a Mexican. The use of overt 1sg subject pronouns was a common trend among participants when asked about their identity as Chipileños and as Mexicans, a theme that aroused strong sentiments. Infelicitous pronouns also occur in the 3sg form. In the following passage, the participant establishes his mother as the topic early on, but the pronoun ella ‘she’ is repeated throughout. (12) Mi madre inclusive hablaba un español malo, no hablaba un buen español, con un acento muy raro y mezclaba mucho las palabras o sea, no que mezclaba las palabras. Ella cuando hablaba en español hablaba en español pero le daba la conjugación del italiano...pero lo que pasa es que ella pensaba en dialecto y traducía en español. Nunca pudo pensar en español mi mamá. Ella tenía que ped- inclusive si tú le hablabas en español ella no te respondía rápidamente ella. Ella pasaba uno o dos segundos para responder, se ve que traducía su pensamiento, eh, al español (Male, born 1958) My mother even spoke a bad Spanish. She didn’t speak a good Spanish, with a strange accent and she mixed words a lot, I mean she didn’t mix words. When she spoke Spanish, she spoke in Spanish, but she gave the Italian conjugation… but what happened is that she thought in dialect and translated to Spanish. She never could think in Spanish, my mother. She had to ask- even if you spoke to her in Spanish, she did not respond quickly. She took a second or two to respond, you could tell she translated her thoughts to Spanish. We could attribute the presence of these pronouns to transfer from Veneto to Spanish. Recall that Veneto has obligatory subject clitics that occur in the 2sg, 3sg, and 3pl verb forms. Transfer and reanalysis of these obligatory subject clitics could lead to overt subject pronouns in these forms in Spanish. However, we also find redundant use of the 1sg pronoun, where no subject clitic occurs in Veneto. As discussed above, these could instead be attributed either to a priming effect or to the semantic content of the verb, often an estimative verb denoting a participant’s personal connection to the discourse.
Hilary Barnes
Setting aside the examples of overt pronouns occasioned by priming and verb semantics (which, as noted, are found also in monolingual varieties of Spanish), the overall increased rates of overt pronominal subjects in Chipilo Spanish can be explained by reference to the situation of language contact and bilingualism itself. In this respect, the insights of Sorace (2004) with respect to the interface areas are especially relevant. In particular, Sorace suggests that “interfaces, because they are more complex than narrow syntax, are inherently more difficult to acquire” (2004:€144).5 In such case, speakers may draw on default strategies, reducing the syntactic option (null vs. overt). The result is an apparent loss of the pragmatic constraints on the distribution of null and overt forms.6 But the underlying mechanism is one of economy: the speaker lessens the cognitive load of bilingualism. 5. Conclusions This study has pursued two goals: (1) to present a unique contact situation of Spanish and Veneto in central Mexico and (2) to examine the syntax-discourse interface in a situation of sustained bilingualism by focusing on the expression of overt subject pronouns. The community of Chipilo has been shown to present a unique opportunity to study this feature given that both languages allow null subjects. The results of quantitative analysis have shown that the rate of overt pronoun expression is markedly higher than what has been reported in for monolingual Mexican Spanish, although their patterns of use are target-like. As in monolingual varieties, some cases of infelicitous overt pronouns occur; these may be attributed to syntactic priming or the semantic content of particular predicates. Of greater interest are the increased rates of overt pronouns in this situation of contact between two null subject languages. As discussed, the increased rate of pronouns could be a consequence of economy: in lessening the syntactic options available (even when these options coincide across languages), speakers lessen the cognitive burden of bilingualism. Ongoing research is pursuing an examination of the individual factors implicated in the attested syntactic patterns.
5. This would explain the results found by Bini (1993), discussed above, and by Lozano (2006), who finds that both native English and native Greek L2 learners of Spanish diverge with respect to word order patterns governed by discourse factors (where Greek and Spanish share word order properties). 6. Loss of pragmatic constraints as opposed to direct contact is also argued by Satterfield and Barrett (2004) in their discussion of varying word order in Sipakapense Maya in contact with Spanish.
Subject pronoun expression in bilinguals of two null subject languages
References Bini, Milena. 1993. “La adquisición del italiano: más allá de las propiedades sintácticas del parámetro pro-drop”. La linguistica y el análisis de los sistemas no nativos ed. by Juana Liceras, 126–139. Ottawa: Dovehouse. Enríquez, Emilia. 1984. El pronombre personal sujeto en la lengua española hablada en Madrid. Madrid: Consejo Superior de Investigacoines Científicas. Goria, Cecilia. 2004. Subject Clitics in the Northern Italian Dialects: A Comparative Study based on the Minimalist Program and Optimality Theory. Dordrecht: Kluwer. Lozano, Cristobal. 2006. “Focus and split-intransitivity: the acquisition of word order alternations in non-native Spanish”. Second Language Research 22.145–187. MacKay, Carolyn. 1992. “Language maintenance in Chipilo: a Veneto dialect in Mexico”. International Journal of the Sociology of Language 96.129–145. Margaza, Panagiota and Aurora Bel. 2006. “Null Subjects at the Syntax-Pragmatics Interface: Evidence from Spanish Interlanguage of Greek Speakers”. Proceedings of the 8th Generative Approaches to Second Language Acquisition Conference (GASLA 2006) ed. by Mary Grantham O’Brien, Christine Shea, and John Archibald, 88–97. Somerville, Mass.: Cascadilla. Milroy, Lesley and Matthew Gordon. 2003. Sociolinguistics: Method and Interpretation. Malden, Mass.: Blackwell. Montrul, Silvina. 2002. “Incomplete acquisition and attrition of Spanish Tense/Aspect distinctions in adult bilinguals”. Bilingualism: Language and Cognition 5.39–68. Montrul, Silvina. 2004. “Subject and object expression in Spanish heritage speakers: A case of morphosyntactic convergence”. Bilingualism: Language and Cognition 7.125–142. Müller, Natascha and Aafke Hulk. 2001. “Crosslinguistic influence in bilingual language acquisition: Italian and French as recipient languages”. Bilingualism: Language and Cognition 4.1–21. Otheguy, Ricardo, Zentella, Ana Celia and David Livert. 2007. “Language and dialect contact in Spanish in New York: Toward the formation of a speech community”. Language 83.770–802. Paradis, Johanne and Samuel Navarro. 2003. “Subject realization and crosslinguistic interference in the bilingual acquisition of Spanish and English: What is the role of the input?”. Journal of Child Language 30.1–23. Poletto, Cecilia. 2000. The higher functional field: Evidence from Northern Italian dialects. New York: Oxford University Press. Romani, Patrizia. 1992. Conservación del idioma en una comunidad italo-mexicana. México: Instituto Nacional de Antropología e Historia. Satterfield, Teresa and Rusty Barrett. 2004. “Generation Gap: Explaining new and emerging word-order phenomena in Mayan-Spanish bilinguals”. Proceedings of the First International Symposium about Bilingualism in Latin America. Buenos Aires: English Speaking Scholastic Association of the River Plate, 285–300. Serratrice, Ludovica; Sorace, Antonella; & Paoli, Sandra. 2004. “Crosslinguistic influence at the syntax–pragmatics interface: Subjects and objects in English–Italian bilingual and monolingual acquisition”. Bilingualism: Language and Cognition 7.183–205. Seuren, Pieter A.M. and Herman Wekker. 1986. “Semantic transparency as a factor in creole genesis”. Substrata Versus Universals in Creole Genesis ed. by Pieter Muysken and Norval Smith, 57–70. Amsterdam: John Benjamins.
Hilary Barnes Sorace, Antonella. 2004. “Native language attrition and developmental instability at the syntax-discourse interface: Data, interpretations and methods”. Bilingualism: Language and Cognition 7.143–145. Sorace, Antonella. 2005. “Selective optionality in language development”. Syntax and Variation: Reconciling the Biological and the Social ed. by Leonie Cornips and Karen Corrigan, 55–80. Amsterdam: John Benjamins. Silva-Corvalán, Carmen. 1994/2000. Language Contact and Change: Spanish in Los Angeles. Oxford: Oxford University Press. Travis, Catherine. 2007. “Genre effects on subject expression in Spanish: Priming in narrative and conversation”. Language Variation and Change 19.101–135. Tsimpli, Ianthi; Sorace, Antonella; Heycock, Caroline; & Filiaci, Francesca. 2004. “First language attrition and syntactic subjects: A study of Greek and Italian near-native speakers of English”. International Journal of Bilingualism 8. 257–277.
Where are hiatuses left? A comparative study of vocalic sequences in Argentine Spanish* Laura Colantoni & Anna Limanni University of Toronto
We investigate the realization of iV sequences in a contact (CV) and noncontact (NCV) variety of Argentine Spanish. We test the following hypotheses: (1) diphthongization is increasing in CV; (2) diphthongization is sensitive to the etymological origin of the sequence; (3) delateralization in CV is an attractor for diphthongization. Results only support (1) and (2). While hiatuses are more frequent in CV, both varieties share the pan-Hispanic preference for diphthongs. However, CV and NCV differ from each other and from Peninsular varieties in the environments where hiatuses are left: NCV maintains few exceptional hiatuses and CV is less sensitive to the etymological origin of the sequence. Additionally, CV and NCV differ in the realization of diphthongs and of the diphthong/hiatus contrast: CV diphthongs are more coarticulated with V2 and more clearly differentiated from hiatuses. Thus, we argue that different varieties exploit different coarticulatory strategies to achieve a similar percept.
1. Syllabification of vocalic sequences: Spanish and the Romance continuum The variable syllabification of high vowel-vowel sequences (iV henceforth) in Romance has been the object of both theoretical (e.g Harris & Kaisse 1999) and experimental studies (Aguilar 1999; Cabré & Prieto 2006; Chitoran 2002; Hualde & Prieto 2002). A recent comparative analysis (Chitoran & Hualde 2007 – henceforth C&H) shows that Romance varieties can be placed along a continuum ranging from a quasi-categorical preference for hiatuses in Portuguese to
* We thank J.I. Hualde and two anonymous reviewers for their comments on an earlier version of this manuscript. This research was supported by a SSHRC grant #410–2005–0079.
Laura Colantoni & Anna Limanni
quasi-categorical presence of diphthongs in French. Spanish, and probably Italian (Bertinetto & Loporcaro 2005), fall somewhere in between. Indeed Spanish displays the same continuum witnessed in Romance more generally. In Spanish, iV sequences belong to the same syllable (diphthongs), unless the high vowel is stressed (hiatus) or morphologically marked (exceptional hiatus – see (3) below). In addition to this general tendency, several Latin American varieties show a clear preference for diphthongs (Garrido 2007 for Colombian Spanish),1 which extends to the diphthongization of mid-vowels, illustrated in (1).
(1) pe.[ljan] > pe.[le.an] pelear ‘to fight’
On the opposite extreme, some contact varieties (CV), such as Guaraní-Spanish, show a preference for hiatuses (Abadía de Quant 2000; Vidal de Battini 1964), illustrated in (2). Indeed, it has been claimed that vocalic sequences in such varieties are interrupted by a glottal stop (Thon 1989):
(2) ca.[mi.�o].nes vs. ca.[mjo].nes
camiones ‘trucks’
Although this preference for hiatuses has been attributed to contact with Guaraní (Abadía de Quant 2000), the status of iV sequences in Guaraní is controversial. Indeed, Gregores & Suárez (1967) report that hiatuses are preferred in Guaraní but are variably realized as diphthongs in rapid speech (1967:€54). Finally, the intermediate stage in the Spanish diphthong-hiatus continuum is represented by Central and Northern Peninsular Spanish, where the general tendency towards diphthongization does not apply to a set of lexical items, called exceptional hiatuses (Hualde 1999; 2005), and exemplified in (3):
(3) [pi.a].no vs. [pja].no piano
In this paper we aim to contribute to the ongoing experimental research on the characterization of iV sequences in Romance, by comparing a contact (Corrientes, Misiones) and a non-contact variety (San Juan) of Argentine Spanish. We show that hiatuses in CV are less frequent than previously reported (Abadía de Quant 2000; Vidal de Battini 1964). However, the resulting diphthongs differ in their acoustic properties vis-à-vis those found in NCV. Additionally, we argue that diphthongization does not target all sequences equally, but is sensitive to the etymological origin of the sequence. More generally, we propose that both the syllabification of these sequences and their acoustic realization vary across Romance. Thus, we aim to contribute to a line of research showing that not all categories are equally variable (Stevens 1972), and that in the presence of sound variation and change, speakers may use different acoustic cues to realize a given category 1.
This is a general tendency, especially in non-standard varieties (Quilis 1993:€189).
Where are hiatuses left
(Colantoni 2006). In the rest of the paper, we summarize previous research (§2); introduce our hypotheses (§3) and methodology (§4); present and discuss our results (§5 and §6, respectively); and conclude briefly in §7. 2. The realization of vocalic sequences: experimental analyses Experimental studies have had two main concerns. Some have concentrated on the phonetic correlates of pre-established phonological categories, such as the distinction between hiatuses and diphthongs for Spanish (Aguilar 1999; Borzone de Manrique 1979; Hualde & Prieto 2002); glides and diphthongs for English (Lehiste & Peterson 1961) or diphthongs and pseudo-diphthongs for Dutch (Bond 1982; Collier et al. 1982). Others have analyzed the acoustic correlates of diphthongs alone (Lehiste & Peterson 1961; Gottfried et al. 1993). As for the acoustic parameters, there is a general consensus that diphthongs differ from hiatuses in both their frequential and temporal properties. Diphthongs have a less steep F2-slope than hiatuses (Aguilar 1999), as well as formant values that do not reach those reported for isolated vowels (Borzone de Manrique 1979). Additionally, it has been observed that F2 changes in the transition are smaller in diphthongs than in hiatuses (Chitoran 2002), although this parameter may not be relevant for Spanish (Borzone de Manrique 1979). As for temporal differences, hiatuses are longer than diphthongs (Aguilar 1999; Hualde & Prieto 2002), and have two clear targets (Aguilar 1999). Experimental studies have also indicated that there is considerable inter- and intra-speaker variation in the realization of these sequences, and that this variability is larger for diphthongs than for hiatuses (Collier et al. 1982; MacLeod 2007).2 Moreover, it has been suggested (MacLeod 2007) that Spanish varieties may select different acoustic parameters to realize the diphthong-hiatus contrast. Here, we explore this claim further, and show that varieties not only differ in the selection of the acoustic parameters but also in the realization of the phonetic contrast. 3. Hypotheses Three hypotheses are tested here. The first hypothesis is that CVs are following the same trend as NCVs, namely that diphthongization is spreading. This hypothesis will be confirmed if CV and NCV show similar rates of diphthongization. 2. Collier et al.€ (1982) only discuss the difference between diphthongs and pseudo-diphthongs in Dutch.
Laura Colantoni & Anna Limanni
Our second hypothesis is inspired by a prediction in C&H (41–42), namely that sequences originating from the diphthongization of Latin open mid-vowels act as attractors favoring the diphthongization of other sequences. Thus, we expect that diphthongization will be sensitive to the etymological origin of the sequence, with sequences originating from the diphthongization of Latin mid-vowels diphthongizing first in CV and exceptional hiatuses diphthongizing last (Table 1, §4.2). Our last hypothesis concerns a possible internal motivation for the change in CV. We hypothesize that there is an attractor (C&H: 42) for this process, namely the emergence of new iV sequences due to the loss of the palatal lateral, illustrated in (4) (Colantoni 2004):
(4) [kaÖˆan] > [kaÖjan] callar ‘to stop talking’
4. Methods 4.1
Corpus
To test our hypotheses, we analyze the speech of eight native speakers of two Argentine Spanish varieties. Three speakers represent a NCV spoken in San Juan (cities of Valle Fértil, San Juan and Ullúm). The remaining five speakers represent a CV spoken in Corrientes (cities of Bella Vista, Berón de Astrada, Caá Caatí and Mburucuyá) and Misiones (Apóstoles). These two varieties share the same vocalic inventories but differ in their consonantal inventories. In particular, CV maintains a contrast between a palatal lateral and a palatal affricate (see §5.3), which has been lost in other varieties, including NCV.3 All speakers are male and were born and raised in the location under study. None had a post-secondary education and their ages ranged from 25 to 50. Our data come from two-hour-long dialectological interviews recorded by the first author between 1994–1997 for the Linguistic Atlas of Argentina (Kovacci 1987). These interviews contain both semi-spontaneous and spontaneous speech where participants answered guided questions about local customs and important life events. The use of naturalistic data represents a departure from previous studies which rely on laboratory speech. This difference constitutes an important contribution because our data should reflect more closely the characteristics of the variety under study. However, in this type of study it is not possible to control for the same variables as in laboratory speech, and, as such, the comparison with those studies may prove more difficult. Finally, the use of 3. For details on both varieties, see Colantoni (2004); Fontanella de Weinberg (2000) and Vidal de Battini (1964).
Where are hiatuses left
semi-spontaneous speech may promote a higher rate of diphthongization. It is important to observe, however, that we compare, first, two groups that were tested using the same methodology. Second, in order to make hypotheses about the direction of the change, we compare our results to previous reports on the same varieties that used either dialectological (Vidal de Battini 1964) or sociolinguistic data (Abadía de Quant 2000; Thon 1989) 4.2
Variables
Target words containing sequences of two vowels where V1 = [i e] and V2 = [a e o u] were extracted from the interviews, and coded in Excel as diphthongs or hiatuses following standard syllabification rules. Exceptional hiatuses were coded as diphthongs, following the native-speaker intuitions of one of the authors.4 Realizations were determined auditorily by the researchers. All subsequent acoustic measurements and statistical analyses reflect the phonetic realizations. The independent variables include: (1) etymological origin of the sequence; (2) following vowel (V2); (3) preceding and following elements (consonant or pause); (4) stress; (5) position in the word; (6) variety (NCV or CV).5 Two separate codings were used for the etymological origin of the sequences (Table 1). Due to space constraints, we focus on ‘Coding II’, which allows us to compare our results against those reported in C&H. 4.3
Analysis
A total of 854 tokens (445 for CV and 409 for NCV) were analyzed acoustically using Praat 5.026 (Boersma & Weenink 2007). Measurements were taken using both waveforms and spectrograms. First, the duration of each vocalic sequence was determined by an increase in F1 and in intensity at the onset of the sequence and drop in F1 at its offset (Figure 1). Then, the duration of each part of the sequence (V1-Transition-V2) was established. V1 was measured from the beginning of the sequence to the first detectable change of at least 20 Hz in the F2 (Chitoran 2002). The transition was measured from the end of V1 to the beginning of the steady state of V2. Finally, V2 4. In the Argentine variety spoken by the first author all cases of exceptional hiatuses listed in Hualde (1999; 2005) are diphthongs. However, this coding will not change the analysis reported here. 5. We do not report results on the effects of stress, preceding and following segments or position in the word. These variables were not statistically significant overall but see §Â€5.2 for individual speakers.
Laura Colantoni & Anna Limanni
Table 1.╇ Origin of the sequence. Coding I: strict etymological coding. Coding II: Exceptional hiatuses included Phenomenon Diphthongization of Latin stressed mid-vowels
Coding Coding I II M
iV sequence in Classical and Vulgar Latin
L
Exceptional hiatuses
L
Latin VCV sequence
Other sequence (from language other than Latin)
S
O
Latin
Spanish
Gloss
M
PETRA
[pje].dra
piedra ‘stone’
L H
ITALIA DIES
Italia ‘Italy’ día ‘day’
E
DIABOLUS
E
DIARIUM
L H
LIMPIDUS RIVUS
E
PEDUCULUS
Ita.[lja] [di.a] [di.a].blo~ [dja].blo [di.a].rio~ [dja].rio lim.[pjo] [ri.o] [pi.o].jo~ [pjo].jo
O H E
N/A
a.lu.mi.[njo] lo.te.[ni.a] [pi.o].ne.ro~ [pjo].ne.ro
diablo ‘devil’ diario ‘daily’ limpio ‘clean’ río ‘river’ Piojo ‘louse’ aluminio ‘aluminum’ lotería ‘lottery’ pionero ‘pioneer’
was measured from the end of the transition to either a drop in F1 or a pause. Measurements of the three parts of each sequence were normalized as a percentage of the total duration of the entire sequence in milliseconds, following Lehiste & Peterson (1961).
Figure 1.╇ Waveform and spectrogram of the sequence [sjon], extracted from confección, ‘tailoring’, Caa Catí, Corrientes
Where are hiatuses left
Formant values (F1–F3, in Hertz) were taken at the stable portions of V1 and V2 and in the middle of the transition. F2 values were also taken at the beginning and end of the transition and were used to calculate the rate of change of F2 according to the formula in (5):
(5) Rate of change in F2 = F2 (beginning T) – F2(end T)/T duration
Results were evaluated using χ2 and repeated-measures ANOVAs with the statistical programs SAS 9.1 and SPSS15, with p level set at.05. Due to their small number, tokens where V1 = [e] and V2 = [u] were omitted from analysis. 5. Results 5.1
Hypothesis 1: Increasing rate of diphthongization in CV
Figure 2 shows the rate of diphthongization of iV sequences in CV and NCV. Crucially, CV exhibits a high rate of diphthongization, against what has been previously reported (Abadía de Quant 2000; Vidal de Battini 1964). Moreover, against previous observations (Thon 1989), no examples of glottal stops breaking up the sequences have been found. Although the rate of diphthongization between CV and NCV is still significantly different (χ2 = 7.27; p = 0.007), this difference mainly relies on the lowest rate of diphthongization of hiatuses in CV.6 Arguably, the testing of this hypothesis
% Diphthongization
Rate of diphthongization 100 90 80 70 60 50 40 30 20 10 0
Diphthong Hiatus CV
Variety
NCV
Figure 2.╇ Rate of diphthongization in CV and NCV
6. The difference between CV and NCV cannot be attributed to differences in speech rate since both groups were given the same interviews and both produced a similar style (see §Â€4.1)
Laura Colantoni & Anna Limanni
could finish here, since we have shown that CV exhibits a high rate of diphthongization. However, it is important to establish whether diphthongs in CV have the same acoustic properties as in NCV. Moreover, if we assume that CV is undergoing a sound change, we need to explain how this change proceeds, in order to verify hypotheses that treat diphthongization as a consequence of gestural overlap (C&H). 5.1.1 Acoustic characteristics of diphthongs Figure 3 shows that diphthongs in CV and NCV differ in the proportion taken by each part of the sequence. The former show significantly shorter V1 (F(2, 12) = 8.87; p = 0.004) and a significantly longer V2 (F(2, 12) = 26.7; p < 0.0001).7 In NCV, V1 and V2 are similar in duration and the longest portion is taken by the transition. As mentioned in §2, one of the characteristics of diphthongs is a relatively smooth F2 slope in the transition. The characteristics of the slope are also determined by the quality of V2; a steeper slope is expected with less fronted vowels. Apparently, this goes against the results reported in Figure 4, where we observe that NCV, overall, exhibits a significantly steeper slope (F(2, 12) = 9.93; p = 0.003) than CV. The differences in F2, however, can be explained by comparing the values of each part of the sequence in both varieties. CV is characterized by a near absence of movement in F2 (vowels in the sequence [ie], in particular, are highly coarticulated), whereas NCV exhibits clearer F2 targets for V1 and V2 (Figure 5). Overall durational differences
% Duration
45 40 35 30 25 20 15 10 5 0
CV NCV V1
T
V2
Portion of sequence
Figure 3.╇ Duration of each part of the diphthong (measured as a proportion of the whole sequence) in CV and NCV
7. Values reflect repeated-measures ANOVAs with Duration (V1, T, V2) as the within-group factor and Location (CV, NCV) as the between-group factor.
Where are hiatuses left
F2 Rate of change (Hz)
F2 rate of change 8 7 6 5 4 3 2 1 0
CV NCV [ia]
[ie]
[io]
Sequence
Figure 4.╇ F2 rate of change (Hz per ms) in the transition for CV and NCV
F2 values by sequence
F2 values (Hz)
2500 2000 1500 1000 500
CV NCV
0 V1
T [ia]
V2
V1
T [ie]
V2
V1
T
V2
[io]
Sequence
Figure 5.╇ F2 values for each part of the sequence by type of sequence
5.1.2 Acoustic correlates of the diphthong-hiatus contrast CV exhibits clear durational differences in the realization of diphthongs and hiatuses, whereas those differences are blurred in NCV.8 Diphthongs in CV have a short V1 and a long V2 (Figure 6); hiatuses have two clear targets and a short transition. In NCV, V1 and V2 are similar in duration in diphthongs and hiatuses. There were no significant within-group differences in F2 rate of change for hiatuses and diphthongs, suggesting that this is not the most important correlate of the contrast. 8. A comparison of the overall duration shows that hiatuses are longer than diphthongs for both groups. However, this difference is statistically significant only for CV, t(4) = –7.001, p = .002, but not for NCV (t(2) = –3.422, p = .076). This is additional evidence that the diphthong-hiatus contrast is being lost in NCV.
Laura Colantoni & Anna Limanni
% Sequence duration
Duration of Diphthongs and Hiatuses 50 40 30 20 10 0
CV NCV V1
T
V2
V1
T
D
V2
H Category
Figure 6.╇ Proportion taken by each part of the sequence in diphthongs and hiatuses
5.2
Hypothesis 2: Sensitivity to the origin of the sequence
To test the prediction that exceptional hiatuses diphthongize less than other sequences, we use Coding II (Table 1, §4.2).9 Figure 7 shows that this prediction is supported. In both CV and NCV, exceptional hiatuses (E) have the lowest rate of diphthongization of all categories of sequences (48.7% for CV and 73.8% for NCV). Although both groups diphthongize E the least, the between-group differences in rates of diphthongization (see §5.1.1) are maintained and are still statistically significant (χ2 = 5.287; p = 0.021). Diphthongization by origin of sequence % diphthongization
100 80 60 40 20 0
CV NCV M
L
O
E
Origin of sequence (II)
Figure 7.╇ Rate of diphthongization by origin of sequence10
9. See Hualde (1999); C&H; Cabré & Prieto (2006) for details on the morphological and prosodic characteristics of exceptional hiatuses. 10. See Table€1 for explanation of legend. The H category is excluded since both groups had 100% hiatus realization of these sequences.
Where are hiatuses left
The prediction that sequences originating from the diphthongization of Latin mid-vowels (M) should diphthongize the most is also largely supported. As expected, M had the highest rate of diphthongization for NCV at 98.6%, approaching the categorical diphthongization predicted by C&H (46). However, M exhibits some unexpected behavior in CV; diphthongization is not categorical, with the difference in rate between CV and NCV being significant (χ2 = 6.758; p = 0.009). Additionally, M diphthongizes at a lower rate than L (92.4% for M vs. 95% for L). A closer look at the data reveals that this outcome is largely due to the behavior of one CV speaker. Speaker BA tends to override the antihiatus tendencies of iV sequences in M, especially where these occur in word-initial position and/or in a stressed syllable (i.e. viento ‘wind’ is realized with a hiatus rather than a diphthong).11 We hypothesize that for this speaker, the influence of the origin of the sequence is less important than Initiality and Stress Conditions (Hualde 1999; C&H). When BA is removed from the analysis, M and L for CV have similar rates of diphthongization (95.3% for M; 96% for L), and the difference in rate of diphthongization for M between CV and NCV is no longer statistically significant (χ2 = 2.535; p = 0.111). These results suggest that for CV sensitivity to the origin of the sequence is not as fine-grained as for NCV. In light of this observation, we test whether the prediction that historical M diphthongs act as attractors to iV sequences from other categories (Hualde 1999; C&H) holds for both varieties. 5.2.1 Origin of the sequence and effects of the following vowel Table 2 illustrates the behavior of iV sequences for L, O and E across CV and NCV. For both varieties [ia] and [io] sequences in E diphthongize the least (Hualde 1999; C&H). In addition, [ia] and [io] have lower rates of diphthongization than [ie] in L. In fact, [ie] is categorically diphthongized in both varieties.12 Moreover, NCV Table 2.╇ Rate of diphthongization (% total N of sequence in category) by V2 [ia] Origin of sequence L O E
[ie]
Contact
Non contact
Contact
Non contact
93.4 81.8 45.8
â•⁄ 90.0 100.0 â•⁄ 70.0
100.0 â•⁄ 70.0 N/A
100.0 â•⁄ 87.5 100.0
[io] Contact 95.5 66.7 53.4
Non contact 96.6 86.7 66.7
11. This is the speaker with the highest rate of maintenance of the palatal lateral (see §Â€5.3). 12. Note that M is excluded from this analysis.
Laura Colantoni & Anna Limanni
Table 3.╇ Rates of diphthongization and delateralization in CV Diphthongization N % AP BA BV CC MB
54 65 71 60 68
87.3 83.5 88.8 93.3 95.6
Delateralization N % 29 â•⁄ 0 â•⁄ 2 â•⁄ 3 12
28.3 0 â•⁄ 7.4 â•⁄ 7.4 22.7
carries this categorical diphthongization of [ie] into E. Therefore, we find support for the proposal that [je] diphthongs in Spanish act as attractors (C&H: 64). 5.3
Hypothesis 3: Delateralization as an attractor in diphthongization
Finally, we examine a possible system-internal motivation for the increasing rate of diphthongization in CV; namely the emergence of new iV sequences as a consequence of the loss of the palatal lateral (see (4) above). We expect that speakers who have a higher rate of delateralization will also have a higher rate of diphthongization. As per Table€3, the speaker with the highest rate of maintenance of the palatal lateral (BA) has also the lowest rate of diphthongization, whereas speaker MB with one of the highest rates of delateralization also has a high rate of diphthongization.13 However, the correlation between the two processes is not significant (r = 0.37). Further testing of this hypothesis requires a larger number of speakers representing different stages in the delateralization and diphthongization process. Finally, delateralization may create triphthongs as well as diphthongs. 6. Discussion 6.1
Where are hiatuses left?
To answer the question in our title, we analyzed two varieties with different rates of diphthongization. If Spanish is moving towards the direction of French (C&H), then it is important to determine which sequences are more resistant to diphthongization and whether the conditions are identical in both varieties as well as in varieties previously studied.
13. For a description of delateralization see Colantoni (2004).
Where are hiatuses left
Our results showed that hiatuses are left in different environments in these two varieties. In NCV, there are very few hiatuses left, with [ie] exhibiting the highest rate of diphthongization. Indeed, this is consistent with both cross-linguistic (Sands 2004) and language-specific tendencies (C&H: 41). Moreover, the lack of a sharp phonetic contrast between diphthongs and hiatuses in this variety suggests that the distinction between diphthongs and hiatuses is probably being lost. In CV, the rate of diphthongization is higher than previously reported but still lower than in NCV. As such, CV differs from NCV in its lower rate of diphthongization; it also differs from other Spanish varieties, such as Central and Northern Peninsular Spanish, in its sensitivity to the origin of the sequence. 6.2
Motivations for diphthongization
We also addressed the possible motivations of this increasing rate of diphthongization in CV. We tested the hypothesis that delateralization acts as a possible attractor of diphthongization and we showed that our data did not strongly support this hypothesis. However, we analyzed only a small number of speakers. Moreover, the sequences that emerged as a result of delateralization may be perceptually different from those analyzed here. Thus, CV may have different types of ‘diphthongs’, as reported for Romanian (Chitoran 2002) or Dutch (Collier et al. 1982). Alternatively, diphthongization may be motivated by specific coarticulatory patterns (C&H). Our results indicated that diphthongization (with reference to CV only) involved a decrease in duration in V1 and a clear V2 target. This V1 is not only shorter than in NCV, it is also more coarticulated. If diphthongization were the consequence of articulatory overlap, this is exactly what we would expect (C&H: 67–68). However, this is not what we would expect on the basis of research on vowel-to-vowel coarticulation, which suggests that a V1, such as [i], should be more resistant to coarticulation (Recasens 1999:82). This research, however, is mainly concerned with perceptual and articulatory data (Beddor et al. 2002), and we analyzed only acoustic parameters. Thus, the variability observed in V1€may only be a reflection of the quantal nature of [i] (Stevens 1972). If diphthongization were the result of gestural overlap, we would also expect to see a similar degree of coarticulation in the existing diphthongs (C&H: 67). However, this was not the case. Indeed, CV showed more coarticulation with V2 than NCV. Two possible factors may account for this. First, we compared across varieties, not within varieties. Second, and in relation to this, coarticulatory patterns may differ between varieties and speakers and within the same speaker (Manuel 1990: 189). Thus, different patterns of overlap should be expected at least in the two varieties. Indeed, this is consistent with research that shows that there is variability in (1) the realization of the contrast; (2) in the parameters used to
Laura Colantoni & Anna Limanni
identify diphthongs (Bond 1982; Collier et al. 1982); (3) in the realization of diphthongs (Collier et al. 1982; MacLeod 2007). 6.3
The Romance continuum revisited
Our results showed that CV and NCV behaved similarly in their preference for diphthongs. These tendencies are consistent with those reported in other Latin American (Garrido 2007) and Peninsular varieties (Aguilar 1999; Cabré & Prieto 2006). If all Spanish varieties display a similar fundamental tendency but vary in where hiatuses are left, C&H’s continuum needs to be revised. Thus, varieties such as NCV behave almost like French; most of the expected hiatuses are realized as diphthongs, and the acoustic differences between diphthongs and hiatuses are being blurred. CV, which previously might have been placed on the other extreme of the continuum, is now somewhere closer to Peninsular Spanish. However, the remaining hiatuses in CV are not the same as those found in Peninsular Spanish. Additionally, Spanish varieties also seem to differ in the way diphthongs and the diphthongs/hiatus contrast is acoustically realized. 7. Conclusions The main goal of this paper was to expand our knowledge about the realization of vocalic sequences in Spanish. We showed that NCV exhibits a quasi-categorical preference for diphthongs, whereas CV has both more diphthongs than previously reported and hiatuses in different environments vis-à-vis Peninsular varieties. Thus, the situation described here is more complex than the one depicted in C&H; i.e. CV has hiatuses in different environments and NCV has few exceptional hiatuses. Additionally, varieties differ in the acoustic realization of the contrast and also in the way diphthongs are realized. If diphthongization is the result of gestural overlap (C&H: 67–68), what our results show is that varieties differ in their coarticulatory tendencies. Indeed, we observe that different articulatory strategies may be used in sound change to achieve a similar percept, which is consistent with both previous research on sound change (Colantoni 2006) and on first and second language acquisition (Nittrouer 2002; Colantoni & Steele 2008). References Abadía de Quant, Inés. 2000. “El español del Nordeste”. El español de la Argentina y sus variedades regionales ed. by María Beatriz Fontanella de Weinberg, 100–137. Buenos Aires: Edicial.
Where are hiatuses left Aguilar, Lourdes. 1999. “Hiatus and Diphthong: Acoustic Cues and Speech Situation Differences”. Speech communication 28.57–74. Beddor, Patrice, James Harnsberger & Stephanie Lindemann. 2002. “Language-specific Patterns of Vowel-to-Vowel Coarticulation: Acoustic Structures and their Perceptual Correlates”. Journal of Phonetics 30.591–627. Bertinetto, Pier Marco & Michele Loporcaro. 2005. “The Sound Pattern of Standard Italian, as Compared with the Varieties Spoken in Florence, Milan and Rome”. Journal of the International Phonetic Association 35.131–151. Boersma, Paul & David Weenink. 2007. Praat. http://www.fon.hum.uva.nl/praat/ Bond, Z. S. 1982. “Experiments with Synthetic Diphthongs”. Journal of Phonetics 10.259–264. Borzone de Manrique, Ana María. 1979. “Acoustic Analysis of the Spanish Diphthongs”. Phonetica 36.194–206. Cabré, Teresa & Pilar Prieto. 2006. “Exceptional Hiatuses in Spanish”. Optimality-theoretic Studies in Spanish Phonology ed. by Fernando Martínez-Gil & Sonia Colina, 205–238. Amsterdam: John Benjamins. Chitoran, Ioana. 2002. “A Perception-production Study of Romanian Diphthongs and GlideVowel Sequences”. Journal of the International Phonetic Association 32.203–222. Chitoran, Ioana & José Ignacio Hualde. 2007. “From Hiatus to Diphthongs: the Evolution of Vowel Sequences in Romance”. Phonology 24.37–55. Colantoni, Laura. 2004. “Reinterpreting the CV transition: an acoustic account of the emergence of the glide as an allophone of the palatal lateral in Argentine Spanish”. Contemporary approaches to Romance linguistics ed. by Julie Auger, J. Clancy Clements and Barbara Vance, 83–102. Amsterdam, John Benjamins. Colantoni, Laura. 2006. “Macro and Micro Sound Variation and Change in Argentine Spanish”. Selected proceedings of the 9th Hispanic Linguistic Symposium ed. by Nuria Sagarra and Jacqueline Toribio, 91–102. Somerville, Cascadilla Press. Colantoni, Laura & Jeffrey Steele. 2008. “Integrating Articulatory Constraints into Models of L2 Phonological Acquisition”. Applied Psycholinguistics 29.1–46. Collier, Renée, Fredericka Bell-Berti & Lawrence Raphael. 1982. “Some Acoustic and Phonological Observations on Diphthongs”. Language and Speech 25.305–323. Fontanella de Weinberg, Maria Beatriz. 2000. El español de la Argentina y sus variedades regionales. Buenos Aires: Edicial. Garrido, Marisol. 2007. “Diphthongization of Mid/Low Vowel Sequences in Colombian Spanish”. Selected Proceedings of the Third Workshop on Spanish Sociolinguistics ed. by Jonathan Holmquist, Augusto Lorenzino & Lotfi Sayahi, 30–37. Somerville: Cascadilla Press. Gottfried, M., J. Miller, & D. Meyer. 1993. “Three approaches to the classification of American English diphthongs”. Journal of Phonetics 21.205–229. Gregores, Ema & Jorge Suárez. 1967. A Description of Colloquial Guarani. Amsterdam: Mouton de Gruyter. Harris, James & Ellen Kaisse. 1999. “Palatal Vowels, Glides and Obstruents in Argentinian Spanish”. Phonology 16.117–190. Hualde, José Ignacio. 1999. “Hiatus with Unstressed High Vowels in Spanish”. Advances in Hispanic Linguistics ed. by Javier Gutiérrez-Rexach & Fernando Martínez-Gil, 182–197. Somerville: Cascadilla Press. Hualde, José Ignacio. 2005. “Quasi-phonemic contrasts”. WCCFL 23. West Coast Conference on Formal Linguistics ed. by Vineeta Chand, Ann Kelleher, Angelo Rodríguez & Benjamin Schmeiser, 374–398. Somerville: Cascadilla Press.
Laura Colantoni & Anna Limanni Hualde, José Ignacio & Mónica Prieto. 2002. “On the Diphthong/Hiatus Contrast in Spanish: Some Experimental Results”. Linguistics 40.217–234. Kovacci, Ofelia. 1987. Documentos del PREDAL Argentina 2. Atlas lingüístico-antropológico de la República Argentina. Buenos Aires: Ministerio de Educación y Justicia. Lehiste, Ilse & Gordon Peterson. 1961. “Transitions, Glides and Diphthongs”. Journal of the Acoustical Society of America 33.268–277. MacLeod, Beth. 2007. Spanish Dialects and Variation in Vocalic Sequences. MA thesis. University of Toronto. Manuel, Sharon. 1990. “The Role of Contrast in Limiting Vowel-to-Vowel Coarticulation in Different Languages”. Journal of the Acoustical Society of America 88.1286–1298. Nittrouer, Susan. 2002. “Learning to Perceive Speech: How Fricative Perception Changes, and how it Stays the Same”. Journal of the Acoustical Society of America 112.711–719. Quilis, Antonio. 1993. Tratado de fonética y fonología españolas. Madrid: Gredos. Recasens, Daniel. 1999. “Lingual Coarticulation”. Coarticulation: Theory, Data and Techniques ed. by William Hardcastle & Nigel Hewlett, 80–104. Cambridge: Cambridge University Press. Sands, Kathy. 2004. Patternings of Vocalic Sequences in the World’s Languages. Ph.D. dissertation. University of California at Santa Barbara. Stevens, Kenneth. 1972. “The Quantal Nature of Speech: Evidence from Articulatory-Acoustic Data”. Human Communication: a Unified View ed. by P.B. Denes & E. E. David, 51–66. New York: McGraw Hill. Thon, Sonia. 1989. “The Glottal Stop in the Spanish Spoken in Corrientes, Argentina”. Hispanic Linguistics 3.199–218. Vidal de Battini, Berta. 1964. El español de la Argentina. Buenos Aires: Consejo Nacional de Educación.
Loanword adaptation in the French of Spanish-speaking immigrants in Montréal Michael L. Friesner
University of Quebec at Montreal This paper examines the adaptation of rhotics in loanwords in French by Hispanophone and Francophone Montrealers, in order to investigate the linguistic integration of immigrant communities and the mutual influence of native and nonnative speakers on the outcome of borrowing. Multivariate analysis reveals that many factors govern the variation among the three possible variants. Immigrants have only acquired some of the constraints. Second-generation immigrants share the grammar of the larger community but resemble other Hispanophones in terms of the rates of use of each variant. The author concludes that both native and nonnative speakers influence the integration pattern of loanwords in particular ways.
1. Introduction Bilinguals are frequently reported to be the agents of borrowing, as both the introducers of loanwords and the individuals within the speech community who determine the adaptation patterns that yield the output forms of these words in the recipient language (cf.€Poplack & Sankoff 1984; Paradis & LaCharité 1997). Nonetheless, the treatment of loanwords by speakers who are natively bilingual, nonnatively bilingual, and monolingual is rarely compared directly. This paper analyzes data on loanword pronunciation in Montréal French from members of these different linguistic groups. There are several motivations for the present study. One of these is to consider how immigrants influence the language of the speech community that they join: how does the pronunciation of loanwords by nonnative speakers of French differ from that of native speakers within the community? This study also considers how nonnative speakers modify their treatment of loanwords as they become more integrated into the speech community.
Michael L. Friesner
1.1
The subjects
The data come from interviews with twelve Hispanophone Montrealers and twelve native Francophone Montrealers. The Hispanophones live in, and have spent most of their French-speaking years in, Montréal and are between the ages of 19 and 37. The twelve Francophones are a subset of the thirty-five speakers interviewed for a larger project on the phonetic, phonological, and social factors influencing loanword adaptation in Montréal (Friesner 2009). They were selected to match the social characteristics of the twelve Hispanophones as closely as possible, based on neighborhood, age, level of education, and level of English (low, intermediate, fluent, or native). The twelve Hispanophone subjects originate from various parts of Latin America. All were interviewed in French and have achieved fluency. Several of the first-generation immigrants benefited from Francisation courses (free or very inexpensive intensive courses in French language and Québec culture) offered by the Québec government. The subjects can be broken down as follows: – 6 L2 French speakers, who immigrated between the ages of 18 and 31; – 3 L1.5 French speakers, who immigrated between the ages of 2 and 10, thus before most estimates of the end of the critical period for language learning (cf.€Lenneberg 1967; Anderson & Lightfoot 2002); – 3 second-generation L1 French speakers, who have at least one immigrant parent, and who grew up in Montréal speaking Spanish at home (one of these speakers, Alicia1, is a natively trilingual speaker of Spanish, French, and English). More detailed demographic information can be found in Table€12. Information on the Francophone subjects is given in Table€2. These speakers vary in terms of age, sex, social class (upper middle class, lower middle class, or working class), type of neighborhood (predominantly French or having a substantial Anglophone population), level of English, and level of Spanish contact. Level of Spanish contact is akin to level of Spanish ability, except that one speaker, Magali, who does not speak Spanish, reported having some Spanish-speaking friends. 1.
All names used here are pseudonyms.
2. Calculation of social class is notoriously unreliable in the case of immigrants. The traditional scale which assigns social class based on education, profession, neighborhood, and social mobility, classifies most of the Hispanophone subjects as lower middle class, with the exception of two of the second-generation Hispanophones – Alicia and Laura – who are upper middle class, and one L2 immigrant – Patricia – who comes out as upper middle class. These figures should not be taken too seriously, however, in the case of the immigrants as with many of these speakers there are large discrepancies between level of education and profession and/or neighborhood of residence.
Loanword adaptation of immigrants in Montréal
Table 1.╇ The Hispanophone subjects Speaker
Age
Sex
25 27 24
F F F
â•⁄ 0 â•⁄ 0 â•⁄ 0
mother: Peru mother: Chile parents: Uruguay
NATIVE ENGLISH FLUENT FRENCH INT. FRENCH
L1.5 Cristobal David Karina
29 19 27
M M F
â•⁄ 5 10 â•⁄ 2
Peru Bolivia Cuba (father: Chile)
FLUENT MIXED INT. ENGLISH INT. ENGLISH
L2
Alberto Domingo
27 25
M M
24 21
INT. MIXED FLUENT FRENCH
Luis Miguel Patricia Teresa
23 29 33 29
M M F F
18 23 31 25
Mexico Mexico (parents: Argentina) Colombia Colombia Peru Mexico
L1
Alicia Jessica Laura
Age of Place of origin Arrival
English level
LOW LOW LOW INT.
Neighborhood
MIXED FRENCH FRENCH MIXED
Table 2.╇ The Francophone subjects (“N/R” = “none reported”) Speaker Chantal François Geneviève Magali Martin Mélanie Murielle Nathalie Nathan Nicolas Philippe Sébastien
1.2
Age
Sex
24 29 23 29 23 24 24 32 34 24 26 37
F M F F M F F F M M M M
Social class English level Neighborhood Spanish contact UMC UMC LMC WC LMC LMC UMC UMC LMC WC LMC LMC
INT. FLUENT FLUENT NATIVE INT. INT. INT. INT. LOW INT. INT. LOW
FRENCH FRENCH ENGLISH ENGLISH FRENCH FRENCH ENGLISH FRENCH FRENCH FRENCH FRENCH FRENCH
SOME N/R N/R LOW N/R N/R HIGH HIGH N/R N/R N/R N/R
The interview procedure
The interviews consisted of several parts. The first part was a traditional sociolinguistic interview (Labov 1984), including questions about life in Montréal, supplemented by modules designed to elicit loanwords in spontaneous speech. The
Michael L. Friesner
topics of these modules included food, sports, music, cinema, and television. The casual interviews were followed by a semantic differentials task (cf.€Labov 1989), where subjects were presented with two words and were asked to read them aloud and explain the difference in meaning between the two. This was designed to elicit many tokens of the loanwords included in the task, while closely mimicking casual speech. This was followed by a series of reading passages recited aloud and recorded, including the roster of the Montréal Canadiens hockey team.3 Finally, all subjects, regardless of their English ability, recorded a reading passage in English, and those who had knowledge of Spanish also recorded a passage in Spanish. The purpose of the reading passages was to compare the speakers’ pronunciation of /r/ in the source languages of many of the loans. The same interview format was administered to the Spanish speakers as that administered to the Francophones, with a couple of minor differences: more questions in the interview were focused on cultural and linguistic differences, and all Hispanophone subjects performed the reading passage in Spanish. The Spanish reading passage was particularly important in order to confirm that the Spanish spoken by the Hispanophones employed the apical variant of /r/. In fact, all of these subjects used an apical flapped or trilled /r/ in their Spanish. Several subjects did not distinguish reliably between flapped and trilled /r/, although the two realizations constitute separate phonemes in most varieties of Latin American Spanish. For this reason, and because no difference in adaptation pattern was observed between these two phonemes from Spanish, all instances of /r/ have been grouped together.4 1.3
The variable and coding procedure
The variation in the realization of the /r/ phoneme in Montréal French has been extensively studied (cf.€Clermont & Cedergren 1979; Tousignant 1987; Sankoff & Blondeau 2007; inter alia). This body of work has identified a number of variants
3. Proper names, which have been excluded from analysis in some studies (cf.€ Poplack, Sankoff, & Miller 1988), are included in the present data. Although these words exhibit different rates of each of the variants, the internal and external constraints that govern the variation seem to be consistent between common and proper nouns. 4. As noted earlier, all three major variants of /r/ – retroflex, posterior, and apical – are present in loanwords from all of the source language categories considered. In the case of the Hispanophones, it is possible that their pronunciation of English-origin borrowings is influenced by the presence of cognate loanwords in Spanish; however, even the Francophones exhibit some tokens of the apical variant in English-origin loanwords. Some examples of the loanwords from each of the source language categories are given in the appendix.
Loanword adaptation of immigrants in Montréal
of /r/ that are present in Montréal French. Two of these have been the focus of the study of variation and change in the Montréal French speech community: i. the apical variant ([r]), trilled or flapped, the dominant variant in Montréal French prior to the 1960’s; and ii. the posterior variant ([R]), a uvular or velar fricative or trill, the newer variant, as a result of change from above. Other variants are considered to be marginal, although they are present in the speech community, as well: iii. deletion: /r/ is generally deleted in word-final complex clusters (e.g., mettre, livre); iv. vocalized /r/: /r/ is often vocalized in coda position (e.g., faire, peur); this variant seems on initial examination to be less frequent in younger speakers today, but as it is exhibited only extremely rarely in the loanword data, it will not be discussed further here; and v. the retroflex or “English” /r/ ([p]): this variant is reported to be present in English-origin or English-associated words, but it is also sometimes noted, especially in older speakers, to surface in rare instances in codas of native words, notably after front rounded vowels (e.g., sœur, pur) (cf.€Dumas 1972). Sankoff & Blondeau (2007) found that the change from apical to posterior /r/ in the native vocabulary began between 1930 and 1950, and that almost all of those born after 1955 preferred the posterior variant. All but one speaker in their youngest cohort, born in the 1960’s, were posterior /r/ users. In the present interviews with Francophones, all with speakers born after 1968, the sporadic examples of apical /r/ in the native vocabulary (fewer than 10 tokens total) are instances in which a speaker deliberately aims to use an old-fashioned, countrified, or foreign manner of speaking. While in the native vocabulary, posterior /r/ certainly dominates and is found to be nearly invariable for the Francophone speakers, there is quite a bit of interesting variation in /r/ in loanwords. In my analysis, I focus on three of these variants: apical, posterior, and retroflex. All three of these variants are present in the loanword data. Paradis & Lebel (1994) discuss English borrowings into Québec French that contain /r/, based on data from Roy (1992). They report that English syllabic /r/ is frequently imported, so that the retroflex variant is used in words such as pattern and wiper, while elsewhere the /r/ is usually adapted to the posterior variant.5 None of the previous studies address directly the adaptation of loanwords from languages other 5. Since Roy’s (1992) data come from Québec City, where the apical variant has not been present in the native vocabulary, the lack of apical /r/ in these loanwords is unsurprising.
Michael L. Friesner
than English. One might assume that these would be adapted to the posterior variant, because of the different sociolinguistic meaning of apical /r/ in the speech community, mentioned above. Another possibility, however, is that apical /r/ could be reintroduced as a peripheral segment (i.e., a segment only allowed in loanwords, cf.€Paradis and LaCharité 1997) through borrowings from languages such as Spanish. 2. Results for /r/ All occurrences of /r/ in loanwords were coded as posterior, retroflex, apical, or vocalized. Words of foreign origin that appear to exhibit posterior /r/ invariably in the community were considered fully nativized and thus were excluded from the analysis. Invariability was determined by the lack of variation in the data set, with at least five speakers producing the word with posterior /R/. The excluded words are bar, Costa Rica, Régina, Toronto, tsar, and Vancouver. The data were inputted into Goldvarb for multivariate analysis (1332 tokens for the Francophones (an average of 111.0 tokens per speaker); 1261 tokens for the Hispanophones (an average of 105.1 tokens per speaker)), and separate runs were performed for each of the variants, for each of the speaker groups. The analysis discussed here does not include social factors such as age, education, etc., but rather focuses on the linguistic factors. It is important that almost all of these speakers use the retroflex variant 100% of the time in English reading passages (the lowest rate is Sébastien, with 94% retroflex). The Francophone speakers, most of whose parents are from the Montréal area, are likely to have had parents, grandparents, or other family friends and relatives who used apical /r/, the older variant, in the native vocabulary (cf.€ Sankoff & Blondeau 2007). Those speakers who were capable of doing the Spanish reading passages thus used the apical variant nearly categorically. These points suggest that the speakers theoretically have command of all three variants, an important prerequisite to studying the alternation between these variants. 2.1
The Francophones’ /r/
Although the posterior variant was virtually categorical in the spontaneous French of all Francophone subjects, the multivariate analysis for /r/ in loanwords revealed considerable variability. The following factors were selected in the Goldvarb analysis: i. syllabicity and syllable position (onset, coda, or syllabic /r/ from the source), ii. following phonological environment (consonant, vowel, or pause), iii. language of origin of the loanword (English, other Germanic, Slavic, Spanish/ Romance, or other),
Loanword adaptation of immigrants in Montréal
iv. word type (common6; proper, including proper names; and title/name7), and v. style (interview style, semantic differentials style, and reading/list style). Syllabic /r/ in the source language favors use of the retroflex variant. In terms of following phonological environment, the retroflex variant is disfavored before a vowel. Words of English origin, and to some extent Germanic, favor the retroflex variant, while words of Slavic or Spanish/Romance origin favor the apical variant. As for word type, the retroflex and apical variants are favored with titles/names, while common and proper nouns behave similarly to one another. Style was also selected as significant for the group as a whole, with the retroflex variant being favored in the most formal style, and the semantic differentials style, like interview style, favoring the posterior variant. However, in this case, analyses of subsets of speakers revealed that only some speakers exhibit this style-shifting effect. With respect to the other factors, the behavior of individual speakers is remarkably consistent. Table 3.╇ Rates of Francophone speakers for each variant of /r/ in loanwords* Speaker Geneviève François Philippe Nathan Nathalie Magali Chantal Mélanie Nicolas Murielle Martin Sébastien OVERALL
Posterior
Retroflex
Apical
n
15% 26% 35% 31% 36% 39% 42% 45% 53% 51% 65% 70% 42%
78% 73% 61% 59% 56% 55% 55% 49% 47% 39% 35% 30% 54%
5% 0% 4% 9% 7% 4% 3% 6% 0% 9% 0% 0% 4%
â•⁄ 110 â•⁄ 136 â•⁄ 134 â•⁄â•⁄ 98 â•⁄â•⁄ 95 â•⁄ 112 â•⁄ 106 â•⁄ 100 â•⁄ 110 â•⁄ 127 â•⁄â•⁄ 96 â•⁄ 108 1332
* The high rate of the retroflex variant cannot simply be due to the possible unnaturalness of the tasks from which some of these data are drawn. The average rate of the retroflex variant in common nouns in interview style only, and in which the r is not syllabic (possibly the most natural context and an environment that would be likely to disfavor the retroflex variant) is identical, at 54%, although with a wider range of values (Murielle and Sébastien being lowest at 13%; Magali and François highest at 100%).
6. The category “common” consists mostly of nouns, but it includes a small number of verbs (e.g., cruiser ‘to cruise’; pr[aj]mer ‘to prime’) and adjectives (e.g., rough and groovy), as well. 7. The category “title/name” includes names of restaurants, titles of movies and television programs, etc., such as Amores Perros, Burger King, and Heroes.
Michael L. Friesner
Table 4.╇ Results of three separate variable rule analyses by variant for Francophones’ /r/** /r/ Language of origin
Syllable position
Word Type
Style
Following environment
**
English Other Germanic Slavic Spanish/Romance Other Range Onset Coda Syllabic /r/ Range Common Proper Title/Name Range Interview Semantic differentials Reading/List Range Consonant Vowel Pause Range
posterior
retroflex
apical
0.32 0.64 0.87 0.76 0.93 61 0.63 0.48 0.22 41 0.51 0.58 0.19 39 0.58 0.61 0.45 16 [] [] [] N.S.
0.75 0.46 0.12 0.06 0.10 69 0.45 0.42 0.72 30 0.54 0.39 0.82 43 0.38 0.39 0.57 19 0.62 0.42 0.67 25
0.23 0% 0.85 0.98 0% 75 [] [] [0%] N.S. 0.38 0.52 0.71 33 [] [0%] [] N.S. [] [] [0%] N.S.
Italics indicate knock-out groups that were thus excluded from the Goldvarb analysis.
The speakers differ greatly in their rates of use of each of the variants. This result seems explainable by the social differences within the sample (speakers with higher levels of proficiency in English use more of the retroflex variant, for example). Nonetheless, while most of the speakers (8 out of 12) use the apical variant at least once, none of them has a rate of higher than 10% for this variant. The rates of each variant by speaker are shown in Table€3, while the results of the multivariate analysis are given in Table€4. 2.2
The Hispanophones’ /r/
It was initially necessary to determine which variant of /r/ the Hispanophones employed in French native vocabulary. Thus, at least 200 tokens of /r/ in the native vocabulary were coded for each of these speakers. Of the twelve speakers,
Loanword adaptation of immigrants in Montréal
only two used the posterior variant less than 96% of the time in the native vocabulary. These two speakers – Luis and Miguel, both L2 speakers of French – preferred the apical variant, using posterior /r/ only 17% and 10% of the time, respectively, in native French vocabulary. As can be seen in Table€5, even these speakers used the posterior variant at a lower rate in loanwords than they did in the native French vocabulary. However, because it cannot be assumed that these two speakers had full command of the three variants considered here, their results have been excluded from the statistical analysis. All of the speakers used the retroflex variant for the most part (although not categorically) in the English reading passage. Table€ 5 reveals that the Hispanophones’ rates of each variant differ widely in loanwords in French, but in the case of the posterior and retroflex variants, they are similar to the rates exhibited by the native Francophones. However, several of the speakers exhibit much higher rates of the apical variant than would be expected by Francophones in the community. Almost all use more than 10% of the apical variant in loanwords, and all but one of the L2 speakers has a rate above 20% for the apical variant.8 Table 5.╇ Rates for the Hispanophones for each variant of /r/ in native French words and in loanwords
L1
L1.5
L2
Speaker
% posterior in French words
Alicia Laura Jessica Cristobal Karina David Domingo Alberto Patricia Teresa Luis Miguel
100% 100% 100% 100% 100% 100% â•⁄ 99% â•⁄ 99% â•⁄ 96% â•⁄ 99% â•⁄ 17% â•⁄ 10%
% of each variant in loanwords Posterior Retroflex Apical â•⁄ 9% 37% 36% 17% 39% 50% 28% 20% 37% 43% â•⁄ 0% â•⁄ 9%
65% 53% 53% 71% 52% 43% 60% 56% 38% 18% 40% â•⁄ 9%
26% 10% 11% 12% â•⁄ 9% â•⁄ 7% 12% 24% 23% 39% 60% 82%
n 112 â•⁄ 92 102 118 â•⁄ 99 107 121 116 â•⁄ 95 100 â•⁄ 95 104
8. If only Spanish-origin loanwords are examined, the result is more dramatic, with the average rate of apical /r/ use in Spanish-origin loanwords by Hispanophones at 53% or 56% for the L2 speakers, as compared to an average rate of 7% for the Francophones. However, similar effects can be observed with words from all of the source languages, although in the case of English-origin words, even the L2 speakers use the apical variant in only 7% of cases.
Michael L. Friesner
The results of the multivariate analysis suggest that the speakers who are most integrated in the community employ more of the constraints that native speakers apply, while the newer immigrants use apical /r/ more frequently in the specific instances, so that they have incorporated this variant as a part of their grammar in a way that is not fully native-like. The L2 speakers (immigrants who came after the critical period), as shown in Table€6, have acquired some of the native constraints. They use the retroflex variant more in cases where the /r/ is syllabic, and also in words of English or other Germanic origin. However, their results also differ from those of the native Francophones. Specifically, word type is not selected as a significant factor in the realization of /r/, the distribution of apical /r/ is not wholly determined by language of origin of the loanword, and some of the constraints that govern the use of posterior /r/ among the Francophones are not present. While apical /r/ is favored in words of Spanish/Romance and Slavic origin, as with the native speakers, it is Table 6.╇ Results of three separate variable rule analyses by variant for the L2 speakers’ /r/ in loanwords /r/ Language of origin
Syllable position
Word Type
Style
Following environment
English Other Germanic Slavic Spanish/Romance Other Range Onset Coda Syllabic /r/ Range Common Proper Title/Name Range Interview Semantic differentials Reading/List Range Consonant Vowel Pause Range
posterior
retroflex
apical
[] [] [] [] [] N.S 0.60 0.40 0.30 30 [] [] [] N.S. [] [] [] N.S. [] [] [] N.S
0.76 0.89 0.48 0.04 0.32 85 0.44 0.48 0.70 26 [] [] [] N.S. 0.67 0.22 0.49 45 [] [] [] N.S.
0.28 0.26 0.76 0.86 0.66 60 0.51 0.66 0.27 39 [] [] [] N.S. [] [] [] N.S. [] [] [0% ] N.S.
Loanword adaptation of immigrants in Montréal
also found in words from “other” languages, a context in which the native Francophones never use this variant, and in words of Germanic origin.9 In addition, apical /r/ is favored in coda position, while it is disfavored in cases of syllabic /r/ from the source. Very few constraints govern these speakers’ use of posterior /r/, for which neither language of origin nor word type are selected as significant. The L1.5 speakers (those who immigrated before the end of the critical period), as shown in Table€7, exhibit behavior that is more native-like than that of the Table 7.╇ Results of three separate variable rule analyses by variant for the L1.5 speakers’ /r/ in loanwords /r/ Language of origin
Syllable position
Word Type
Style
Following environment
English Other Germanic Slavic Spanish/Romance Other Range Onset Coda Syllabic /r/ Range Common Proper Title/Name Range Interview Semantic differentials Reading/List Range Consonant Vowel Pause Range
posterior
retroflex
apical
0.27 0.87 0.82 0.76 0.93 66 [] [] [] N.S. 0.71 0.48 0.12 59 [] [] [] N.S. [] [] [] N.S
0.73 0.11 0.08 0% 0.07 66 0.40 0.50 0.76 36 0.33 0.52 0.81 48 [] [] [] 45 [] [] [] N.S.
0% 0% 0.27 0.76 0% 49 [] [] [] N.S. 0.17 0.64 0.71 54 [] [0% ] [] N.S. [] [] [0% ] N.S.
9. This result is presumably due to the existence of cognate loanwords in Spanish. However, because a large number of the loanwords in the data set are too recent to be included even in French dictionaries, this hypothesis is difficult to verify. In addition, in a multilingual contact situation, even loanwords that do not exist in the Hispanophone immigrants’ native dialects may be present in the local variety of Spanish, and thus, the form of these words may be influenced by the other languages present in the community.
Michael L. Friesner
L2 speakers. For these speakers, as with the native Francophones, syllable position is not a significant factor in the realization of apical /r/. Instead, the posterior /r/ constraints are active in the grammar of these speakers: posterior /r/ is favored with common nouns and disfavored with titles/names, and it is also favored for words from languages other than English, including Spanish/Romance. Like the native Francophones, these speakers do not use the apical variant in words of Germanic or “other” origin. The L1 speakers (second-generation Hispanophones) are essentially indistinguishable from the native Francophones in terms of constraints selected by multivariate analysis, as shown in Table€8. For these speakers, syllable position is selected as significant for the realization of the posterior variant, and the realization of the apical variant is only determined by language of origin of the loanword (almost exclusively Spanish/Romance and Slavic; also “other”, though there are Table 8.╇ Results of three separate variable rule analyses by variant for the L1 (secondgeneration) speakers’ /r/ /r/ Language of origin
Syllable position
Word Type
Style
Following environment
English Other Germanic Slavic Spanish/Romance Other Range Onset Coda Syllabic /r/ Range Common Proper Title/Name Range Interview Semantic differentials Reading/List Range Consonant Vowel Pause Range
posterior
retroflex
apical
0.35 0.64 0.77 0.65 0.86 51 0.61 0.65 0.08 57 0.40 0.59 0.21 38 [] [] [] N.S. [] [] [] N.S
0.71 0.46 0.05 0% 0.10 66 0.39 0.44 0.87 48 0.57 0.43 0.73 30 [] [] [] 45 [] [] [] N.S.
0.14 0% 0.95 0.98 0.83 84 [] [] [] N.S. [] [] [] N.S. [] [0% ] [] N.S. [] [] [0% ] N.S.
Loanword adaptation of immigrants in Montréal
very few relevant tokens). The only noticeable difference is that style is not selected as significant for these speakers, but, as mentioned earlier, not all native Francophones exhibit style shifting with this variable. It is perhaps unsurprising that the constraints selected for these speakers do not differ from those selected for the native Francophones since these speakers are also native French speakers. However, as can be seen by comparing Tables 3 and 5, there are differences in terms of the rates of use of each of the variants between these speakers and the native Francophones; notably, these speakers exhibit higher rates of use of the apical variant. This suggestion is further confirmed by a direct comparison of the factor weights by language of origin for apical /r/ for each of the groups, as shown in Figure 1. While the second-generation speakers look more like native Francophones in terms of the factor weights by language of origin, and thus they share Factor Weights for Apical /r/ 1 0.8 0.6 0.4
Francophones L2 (Immigrants) L1 (Second Generation)
0.2 0
Other Germanic
English
Slavic
Spanish/ Romance
Rates of Apical /r/ 0.60
0.40
0.20
Francophones
0.00
L2 (Immigrants) L1 (Second Generation) Other Germanic
English
Slavic
Spanish/ Romance
Figure 1.╇ A comparison of the factor weights vs. rates of use of apical /r/ in words of various origins by the different linguistic groups
Michael L. Friesner
the native community grammar, they have rates of the apical variant that more closely resemble those of the immigrants. This last finding thus relates to the question of how bilinguals and monolinguals within the speech community influence one another in loanword adaptation. While in terms of the grammar, it is the grammar of the wider community that prevails, in terms of the use of the variants that are present, the bilinguals may have higher rates of these variants, as long as their use is constrained within the grammatical system of the recipient language. Thus, if the ratio of bilinguals in the community increased, we might expect higher rates of the apical variant, but we would not expect a reversal of the grammatical constraints that govern this variation. 3. Conclusion This study, examining the linguistic behavior of Hispanophones with varying degrees of integration into the Francophone speech community of Montréal, as well as a sample of Francophones, explores directly the mutual influence of immigrants and native speakers on the phonology of loanwords. While the grammatical constraints governing the choice among possible variants of nonnative segments are primarily determined by the larger speech community, the rates of the different variants and the decision to import a nonnative segment rather than adapting it categorically appear to be influenced directly by the bilinguals within the community. The close examination of the adaptation of rhotics in loanwords demonstrates that increased contact between French and Spanish does not easily entail modification of the community grammar. The second-generation immigrants, while maintaining a higher rate of the apical variant, do not exhibit any influences of this increased rate in terms of their treatment of the three variants of /r/ within their grammar. Still, the surfacing of the apical variant in loanwords in a speech community where it was once present in the native vocabulary but has been steadily disappearing (cf.€Sankoff & Blondeau 2007), is somewhat surprising, and can only reasonably be ascribed to contact with immigrant languages such as Spanish. This paper has thus shown a mechanism for the introduction of imported segments, which has been shown to occur without giving rise to large modifications of the community grammar. This study thus contributes to a predictive model of when foreign segments are adapted, as opposed to being imported without modification, by different members of the speech community, both native and nonnative, in loanword adaptation.
Loanword adaptation of immigrants in Montréal
References Anderson, Stephen & David Lightfoot. 2002. The Language Organ: Linguistics as Cognitive Psychology. Cambridge, U.K.: Cambridge University Press. Clermont, Jean & Henrietta Cedergren. 1979. “Les ‘R’ de ma mère sont perdus dans l’air”. Le français parlé: Études sociolinguistiques ed. by Pierrette Thibault, 13–28. Edmonton, Alta.: Linguistic Research. Dumas, Denis. 1972. Le français populaire de Montréal: description phonologique. M.A. thesis, Université de Montréal. Friesner, Michael Lee. 2009. The Social and Linguistic Predictors of the Outcomes of Borrowing in the Speech Community of Montréal. Ph.D. dissertation, University of Pennsylvania. Labov, William. 1984. “Field Methods of the Project on Linguistic Change and Variation”. Language in Use: Readings in Sociolinguistics ed. by John Baugh and Joel Sherzer, 28–53. Englewood Cliffs: Prentice Hall. Labov, William. 1989. “Exact Description of the Speech Community: Short a in Philadelphia”. Language Change and Variation ed. by Ralph W. Fasold and Deborah Schiffrin, 1–57. Philadelphia: Benjamins. Lenneberg, Eric. 1967. Biological Foundations of Language. New York: Wiley. Paradis, Carole & Darlene LaCharité. 1997. “Preservation and Minimality in Loanword Adaptation”. Journal of Linguistics 33.379–430. Paradis, Carole & Caroline Lebel. 1994. “Contrasts from Segmental Parameter Settings in Loanwords: Core and Periphery in Quebec French”. Toronto Working Papers in Linguistics 13.75–95. Poplack, Shana & David Sankoff. 1984. “Borrowing: The Synchrony of Integration”. Linguistics 22.99–135. Poplack, Shana, David Sankoff, & Christopher Miller. 1988. “The Social Correlates and Linguistic Processes of Lexical Borrowing and Assimilation”. Linguistics 26.47–104. Roy, Marie-Josée. 1992. Le rôle des contraintes phonologiques dans l’adaptation d’emprunts anglais en français québécois. M.A. thesis, Université Laval. Sankoff, Gillian & Hélène Blondeau. 2007. “Language Change Across the Lifespan: /r/ in Montreal French”. Language 83:3.560–588. Tousignant, Claude. 1987. “Les variantes du /R/ montréalais: contextes phonologiques favorisant leur apparition”. Revue québécoise de linguistique théorique et appliquée 6:3.73–113.
Michael L. Friesner
5. Appendix The five most frequent words included in the analysis by language of origin, for the interview data alone, as well as for the data set as a whole, are included here. The words from the formal methods section come from recorded passages about food, dance, cinema, Canadian politics, and the names of players from the Montréal Canadiens hockey team. interview data
all data
English-origin words
rock rap (ham)burger(s) whatever New York
rap New York Rona Harper Brother
Spanish/Romance-origin words
tortillas Sala-Rosa Buenos Aires Monterrey merengue
merengue burrito(s) Buenos Aires daiquiri tierra-blanca
Other Germanic-origin words
Bernhard Petersen Mozart Wagner Zurich
Streit Englisberg Aebischer Friburg Bernhard
Slavic-origin words
pierogis Kusturica Emir
Andrei Sergei Brat Aleksandr Perezhogin
Other-origin words
Gibran karaté Kilimandjaro Himichal-Pradesh Dharamsala
Turku Raahe Gibran karaté Kilimandjaro
part 2
Phonology and interfaces
Morphology and phonology of word-final vowel deletion in spoken Tuscan Italian Luigia€Garrapa & Judith Meinschaefer
Universität Konstanz, Università del Salento & Universität Würzburg Italian elision consists in the optional deletion of the final vowel of a function word followed by a vowel-initial lexical word. The aim of the present study is to reach a better understanding of the interplay between phonology and morphology in Italian elision by analyzing a corpus of spoken Italian as well as by conducting an experimental study. The central claim is that two elision processes have to be distinguished, one that applies to determiners and is closer to categorical variation, to be analyzed in part as allomorphy, and another process that applies to pronouns and is closer to free variation. Italian elision is modeled in a simple optimality-theoretic constraint hierarchy: it is shown that the patterns of variation can be learned by the Gradual Learning Algorithm (Boersma & Hayes 2001), but only if two different grammars are learned for elision on determiners and on pronouns, respectively.
1. Introduction The process termed elisione ‘elision’ in Italian descriptive grammar consists in the deletion of the final vowel of a function word (henceforth Fnc) followed by a vowelinitial lexical word (henceforth Lex). In modern Italian, elision can apply to determiners (1) and to pronouns (2), as well as in other contexts. Here and in following examples, we use the apostrophe to indicate deletion of the word-final vowel. (1) a. questo esempio this example ‘this example’
→ quest’ esempio
b. quella idea that idea ‘that idea’
→ quell’ idea
Luigia€Garrapa & Judith Meinschaefer
(2) a. ci amava us he/she.loved ‘he/she loved us’
→ c’ amava
b. la odiava her he/she.hated ‘he/she hated her’
→ l’ odiava
Previous research has shown that elision on the masculine singular articles uno and lo as well as on the demonstrative quello is categorical, while elision on the remaining function words is subject to free variation (Agostiniani 1989; Nespor 1990; Rosati 2001; Vogel et al. 1983). While these studies acknowledge that morphology influences the application of elision, the exact nature of this influence on Italian elision is as yet unclear. The aim of the present study is to better understand this phonological phenomenon by describing the patterns of elision on determiners and pronouns in a corpus of Italian as spoken in Florence, as well as in an experimental study conducted with speakers of the same variety. The focus lies on the influence of morphology on vowel elision; prosodic factors will not be considered, with the exception of word stress, which has, however, little influence on the elision process studied here. After describing the methods employed and the data analyzed in Section€2, we will briefly present the results of the study in Section€3. Section€4 analyzes the morphological factors influencing elision, and presents a simple optimality-theoretic model of elision on determiners and pronouns. 2. Data and methods The variety under analysis is Italian as spoken in Florence (Tuscany), a variety in which final elision has been shown to be particularly productive (Agostiniani 1989; Rosati 2001). Our focus is on elision in sequences of vowel-final Fnc followed by vowel-initial Lex. The Fnc under analysis are presented in tables 1 and 2. Table 1.╇ Italian determiners studied here Sg
Definite article Indefinite article Demonstrative
Pl
Masc
Fem
Masc
Fem
lo uno questo
la una questa
gli
le
questi
queste
Morphology and phonology of final vowel deletion in Tuscan Italian
Table 2.╇ Italian pronouns studied here Sg Case Acc Dat
Pl
Person
Masc
Fem
Masc
Fem
3
lo
la
li
le
1 3
mi gli
ci le
The corpus study analyzes the C-Oral-Rom corpus (Cresti & Moneglia eds. 2005), a large corpus of spontaneous formal and informal speech, representing the language currently spoken in Florence. The corpus contains both the original audio files and their orthographic transcriptions. After having ensured that the transcription of vowel elision in the corpus is sufficiently accurate for our purpose, we analyzed all instances of elision and of vowel retention in the transcriptions for the Fnc under investigation (cf.€Garrapa 2009 for a more detailed description).1 In the experimental study conducted in Florence, we further explored the influence of two factors on elision: a) the morphological specifications of the Fnc and b) the presence vs. absence of a primary stress on the initial vowel of the following Lex. For this we created 134 sequences of determiners followed by nouns (containing 67 different nominal lexemes) and 133 sequences of pronouns followed by verbs (containing 22 different verbal lexemes). All Lex chosen were trisyllabic words; half of the words had penultimate stress, as in usáva ‘(s/he) used’ and half of the words had antepenultimate stress, as in úsano ‘(they) use’. Each of the function words under investigation was presented followed both by a stressed syllable and by an unstressed syllable. The stimuli consisted of a sequence of one of the Fnc given in tables 1 and 2, followed by a Lex. The sequences of Fnc and Lex were embedded in simple sentences of 5 to 7 words, as in the example (3). (3) Secondo me la ódiano profondamente. for me her they.hate deeply ‘I think they hate her deeply.’ We did not create sequences where the two relevant vowels are identical, as in la ámano ‘they love her’, which is difficult to distinguish from l’ámano. All informants (n = 9) were university students aged between 23 and 29 years and native speakers of Florentine Italian. The informants’ task consisted in 1. A more detailed description of the results of the corpus study may be found in Garrapa (2009). Garrapa (2008) presents a more detailed description of the experimental study.
Luigia€Garrapa & Judith Meinschaefer
producing the 267 test sentences. Each test sentence was shown on a computer screen for 3 seconds, spanning three lines, as in (4); gloss is given in (3). (4) Secondo me la odiano profondamente After a pause of one second, an acoustic signal indicated that the test sentence should be produced.2 The sentences produced by the informants were recorded digitally and were later analyzed auditorily by the first author in order to determine whether elision had applied or not.3 3. Results The results of the corpus study and of the experimental study are very similar. However, as expected, the overall frequency of application of elision is higher in the corpus, i.e. in spontaneous speech, and lower in the experimental study, i.e. in elicited speech (cf.€van Oostendorp 1997; Rosati 2001 for similar observations). The pooled results of the corpus and the experimental study are presented in Tables 3 and 4.4
2. The instruction given to the subjects was to repeat the sentence they had read on the screen, waiting, however, until they had heard the acoustic signal. 3. Visual presentation of non-elided stimuli may have biased the subjects’ responses towards retention, rather than elision of the final vowel. It should be noted, however, that such a bias does not affect the model of elision presented here, since this model relies entirely on elision rates for singular as compared to plural function words, for determiners as compared to pronouns and for pronouns followed by a stressed as compared to an unstressed syllable, not on overall elision rate. It is noteworthy that the masculine singular definite and indefinite articles, which undergo obligatory elision when followed by a vowel-initial word, have likewise been presented in their full forms, but were produced with elision by the subjects in all cases. Finally, we would like to add that in a pilot study we had subjects repeat sentences which had been presented acoustically without elision. We found that with acoustically presented stimuli, subjects made hardly any elision at all, possibly due to the influence of the auditory short term memory. Consequently, we chose to use visual presentation of the stimuli in the follow-up experiment (a procedure which may be compared to the cross-modal priming procedure that is widely used in psycholinguistic studies). 4. The test sentences produced for gli as a plural determiner and a dative pronoun were analyzed auditorily. It is difficult to decide reliably by ear whether the final [i] following the palatal lateral is elided or retained; an acoustic signal analysis would probably have been more successful. The data obtained for gli do not, however, materially affect the overall results.
Morphology and phonology of final vowel deletion in Tuscan Italian
Table 3.╇ Deletion rates for determiners Number
Function
Gender Masc
Article Fem
Sg Demonstr Article Pl
Masc Fem Masc
Demonstr Article
Fem
Form
% Deletion
lo uno la una questo questa
100 100 â•⁄ 99 â•⁄ 97 â•⁄ 86 â•⁄ 85
gli questi queste le
â•⁄ 16 â•⁄ 20 â•⁄ 21 â•⁄â•⁄ 8
Table 4.╇ Deletion rates for pronouns Number
Gender Fem
Sg
Pl
Masc
Case Acc Dat
Masc & Fem
Acc & Dat
Masc Fem
Acc
Sg
Dat
Form la lo gli mi ci li le le
% Deletion 47 39 46 26 43 â•⁄ 8 â•⁄ 5 â•⁄ 0
Tables 3 and 4 show that elision applies with higher frequency to determiners than to pronouns; it applies highly frequently or even obligatorily to singular determiners, but only infrequently to plural determiners. Likewise, the final vowels of singular accusative pronouns are elided more frequently than those of their plural counterparts. 4. Analysis The patterns of variation observed in Italian elision can be best analyzed in Optimality Theory (henceforth OT, cf.€Prince & Smolensky (2004 [1994]), a framework in which variation, gradience and exceptionality have received considerable attention in recent years (e.g., Anttila 2006; Nagy & Reynolds 1997; Boersma & Hayes
Luigia€Garrapa & Judith Meinschaefer
2001). Given that an OT grammar provides a metric to determine the optimal output from a set of candidates, an analysis of elision must first determine the constraint with respect to which elided forms are better than non-elided forms. Italian elision applies in the context given in (5a), where an open syllable is followed by an onsetless syllable. If the first vowel is elided, the sequence can be syllabified as a single CV syllable, thus avoiding the creation of an onsetless syllable. (5) a. CV1.V2 b. /ques.to1.#a 2.mi.co/ ‘this friend’
→ →
C’V2 [ques.t’#a2.mi.co]
Hence, the driving force behind elision is the constraint ONSET (Prince & Smolensky 2004 [1994]) (6).
(6) ONSET (ONS): “Syllables have an onset.”
Satisfying ONS, however, has a cost: vowel elision, which leads to a violation of MAX-VOWELfnc (cf.€McCarthy & Prince 1995) (7).
(7) MAX-VOWELfnc (MAX-Vfnc): “Do not delete input vowels of function words.”5
Hence, if elision applies, MAX-Vfnc is violated, but if elision does not apply, ONS is violated. Since forms with, as well as without, elision are attested, MAX-Vfnc and ONS cannot be strictly ranked. Before going further, however, to show how this variation can be represented in an OT grammar, let us identify morphological and phonological factors influencing the application of elision. 4.1
Morphological factors in elision
4.1.1 Elision on determiners As shown in Table€3 above, elision applies obligatorily to the masculine singular articles lo and uno, but only very infrequently to the feminine plural article le. The singular determiners la, una, questa, questo undergo elision very often, while the plural determiners gli, questi, queste are seldom elided. The generalization one can derive from the data is thus that a vowel which is an exponent of the feature [plural] resists deletion, while nothing inhibits deletion of the final vowel of a singular determiner. 5. Vowels of lexical words and vowels of function words do not seem to behave alike in elision contexts. In Italian a hiatus is tolerated more easily between two lexical words than between a function word and a lexical word. Therefore, we propose to distinguish (at least) two constraints on final vowel deletion, Max-Vfnc and Max-V lex, which are ranked differently in the grammar (with Max-V lex, but not Max-Vfnc outranking Ons).
Morphology and phonology of final vowel deletion in Tuscan Italian
To better understand why elision applies to singular forms, but not to plural forms, let us briefly explore the morphological structure of determiners. Italian determiners consist of a root followed by an inflectional affix realizing the morphological categories of number ([singular] or [plural]) and gender ([masculine] or [feminine]). Following standard assumptions about minimal specification (Embick & Noyer 2005), and given that [singular] and [masculine] are usually considered as the default values for the categories number and gender (Greenberg 1966:€92–96), we will consider the singular and the masculine forms as unspecified for these features (indicated as [ ]) and the plural and the feminine forms as morphologically specified, see Table€5. Following a proposal by Iscrulescu (2005), we assume that deletion of the plural exponent is blocked by the constraint REAL-NUM, see (8a). Since we assume that the singular forms are unspecified for number, this constraint is equivalent to (8b). (8) REALIZE-NUMBER (REAL-NUM) (Iscrulescu 2005) a. The category number must be realized by a morphological exponent. b. The feature [plural] must be realized by a morphological exponent. Real-Num as formulated in (8) is to be understood as a markedness constraint stating that iconic coding of number, where the feature plural is coded by additional morphological material as compared to the singular, is less marked universally (along the lines of Mayerthaler 1981 and subsequent work in Natural Morphology). As shown in tableau (9), REAL-NUM is violated only when an exponent of [plural] is elided (9d), but not when the final vowel of a singular form is elided (9b). Table 5.╇ Morphological specifications and deletion rates for determiners Determiners l-o un-o quest-o l-a un-a quest-a gl-i quest-i l-e quest-e
Morphological features expressed by final vowel Number Gender
[] []
[feminine]
[] [plural]
[feminine]
% Deletion 100 100 â•⁄ 86 â•⁄ 99 â•⁄ 97 â•⁄ 85 â•⁄ 16 â•⁄ 20 â•⁄ 21 â•⁄â•⁄ 8
Luigia€Garrapa & Judith Meinschaefer
(9) Input /questo amico/ ‘this friend’ /questi amici/ ‘these friends’
Candidates Realâ•‚Num a.╇ [ques.to.a.mi.co] b.╇ [ques.t’a.mi.co] c.╇ [ques.ti.a.mi.ci] d.╇ [ques.t’a.mi.ci] *
Ons * *
Max-Vfnc * *
Since all four forms given in (9a) to (9d) are attested in Italian, the constraints Real-Num, Ons and Max-Vfnc cannot be strictly ranked against each other. As noted above, elision always applies to the masculine singular articles lo and uno. What is more, the masculine singular definite article has not only two, but three alternants: l’ before vowel-initial words, lo before /s + C/, /dz/, /ts/, /ps/, /pn/, /ks/ and before a palatal nasal or glide, and il in all other cases. In contemporary Italian, the three alternants cannot be related to a single underlying form, at least not with a synchronically productive phonological rule. Furthermore, as to the masculine singular indefinite article, the elided form un (as opposed to uno) appears not only before vowel-initial words, but also before most consonantinitial words. The selection of un instead of uno is thus not determined by Ons; rather, un is a “dominant allomorph” in the sense of Mascaró (2007). For these reasons, we take it as more economical to analyze the alternation of the Italian masculine definite articles as allomorphy, i.e., as selection from a set of candidates in the input: the full allomorphs /lo/ and /uno/ and the elided allomorphs /l’/ and /un/ (Mascaró 1996; Perlmutter 1998), see (10).6 (10)
Input /{lo, l’} amico/ ‘the friend’
Candidates a.╇ [lo.a.mi.co] b.╇ [l’a.mi.co]
Realâ•‚Num
Ons *
Max-Vfnc
Thus, what looks like obligatory elision of the masculine singular articles lo and uno is a case of allomorphy. For the other articles, for demonstratives (and for pronouns), however, we shall need a phonological explanation for elision. 4.1.2 Elision on pronouns Table 4 above suggests that elision tends to apply more frequently when the elided vowel is the exponent of an unmarked feature; hence, frequently, to the singular accusative pronouns lo/la, but only exceptionally to their plural counterparts 6. For expository reasons, the alternant il has been omitted from the input in (10). To be sure, the lexicon and grammar proposed here need further elaboration to account for the distribution of il, lo, l’ and of uno and un. For lack of space, we will not go into this.
Morphology and phonology of final vowel deletion in Tuscan Italian
li/le. It applies frequently to the masculine dative pronoun gli, while its feminine counterpart le resists deletion. On the other hand, the plural pronoun ci is more frequently reduced than its singular counterpart mi. From a morphological point of view, the accusative pronouns consist of a (consonantal) root followed by an inflectional affix realizing the categories number and gender. Again, according to the principle of minimal specification (see § 4.1.1), we consider the singular and the masculine forms as unspecified, but the exponents of [plural] and of [feminine] as specified for these features; see Table€6. Table 6 shows that in the singular forms [a, o] are deleted more frequently than [i, e], realizing the feature [plural]. Here again, REAL-NUM can account for the lower deletion rate of plural pronouns as compared to singular pronouns. Now, let us turn to dative pronouns. As shown in Table€2 above, mi and ci function both as dative and accusative pronouns, whereas gli is only dative. Following Cardinaletti & Shlonsky (2004), we assume that the final [i] of these pronouns does not realize any morphological features, see Table€ 7. Since the final vowel of mi, ci, gli is not an exponent of number, REAL-NUM is not relevant. 4.1.3 Summary So far we have shown that elision affects determiners more frequently than pronouns and that morphology partially constrains elision. The feature gender is less Table 6.╇ Morphological specifications and deletion rates for accusative pronouns Morphological features expressed by final vowel Pronouns
Number
Gender
% Deletion
[]
[] [feminine]
39 47
[] [feminine]
â•⁄ 8 â•⁄ 5
l-o l-a l-i l-e
[plural]
Table 7.╇ Morphological specifications and deletion rates for mi, ci, gli Pronouns
Morphological features expressed by Root
m-i c-i gl-i
Case
Person
Acc
1
Dat
3
Number
Final vowel
% Deletion
Sg
26
Pl
43
Sg
46
Luigia€Garrapa & Judith Meinschaefer
relevant to the application of elision, while number is relevant: Exponents of [plural] are seldom elided, while the final vowels of singular forms are elided frequently. As to the masculine singular article, what lies behind the apparently ‘obligatory’ elision is an allomorphic alternation between the full and the elided variant, which has arisen from the lexicalization of the elided alternant. 4.2
Phonological factors in elision
In previous research it has repeatedly been observed that elision is sensitive to metrical structure. In particular, it has been shown that in a sequence of two vowels, deletion of the first vowel is less likely or is blocked altogether if the second vowel is stressed (Agostiniani 1989; Bisol 2003; Gili Fivela & Bertinetto 1999; Cabré & Prieto 2003; Nespor 1990; for various Romance languages). Previous research on the effect of metrical structure on elision has dealt mainly with final vowels of a Lex followed by another Lex, where the second Lex receives phrasal stress. In the following, we shall investigate whether stress also has an effect on elision in sequences of Fnc and Lex of the kind studied here. If stress does have a blocking effect, we expect that in a sequence like (11a), where the initial vowel of the second word bears main word stress, elision applies less often than in a sequence like (11b), where the second vowel is unstressed. Main word stress here and below is indicated by an orthographic accent. (11) a.
Second vowel bears main word stress lo úsano → l’ úsano it/him they.use ‘they use it/him’
b.
Second vowel is unstressed lo amáva → l’ amáva it/him he/she.loved ‘he/she loved him’
As can be seen in Table€8, the final vowel of a determiner elides with roughly the same frequency when followed by an unstressed or stressed vowel. For a subset of the pronouns under investigation, in contrast, deletion is less likely when the following vowel is stressed as compared to when it is unstressed. More precisely, the third person plural pronouns li, le and the first person pronouns mi, ci, when followed by an unstressed vowel, delete more often than when they are followed by a stressed vowel. Note that this does not hold for the singular pronouns lo and la, which undergo elision with the same frequency when followed by an unstressed or stressed vowel.
Morphology and phonology of final vowel deletion in Tuscan Italian
Table 8.╇ Elision on determiners and pronouns followed by an unstressed or by a stressed vowel % Deletion if following vowel is Number Determiners
Pronouns
Case
Forms
unstressed
stressed
Sg
lo, uno, questo, la, una, questa
96
97
Pl
gli, questi le, queste
26
28
lo, la
40
40
li, le
â•⁄ 9
â•⁄ 3
mi
24
10
ci
44
33
Sg
Acc
Pl Sg Pl
Acc & Dat
Finally, one might think that among the phonological factors influencing elision, vowel quality plays an important role, given that the dorsal vowels /a/ and /o/ (deletion rate ≈ 91 percent in determiners, cf.€Table€3) elide more often than the coronal vowels /i/ and /e/ (deletion rate ≈ 16 percent in determiners, cf.€Table€3). In fact, it is impossible to dissociate the influence of morphology from that of vowel quality for determiners and for third person pronouns, whose forms always end in a dorsal vowel in the singular and in a coronal vowel in the plural. There is, however, a set of pronouns whose forms end in a coronal vowel which cannot be considered an exponent of number: the first (singular mi ‘me’, plural ci ‘us’) and second person (singular ti ‘you’, plural vi ‘you’) pronouns. In the C-Oral-Rom corpus, these forms present deletion rates ranging from 52 percent (1st person singular mi) to 10 percent (2nd person plural vi), with an average deletion rate of 39 percent (cf.€Garrapa 2009). Hence, even though these pronouns do end in a coronal vowel, just like plural determiners and pronouns, they have a much higher deletion rate than these. The data thus do not support the conclusion that vowel quality influences deletion rates. 4.3
Two different elision phenomena
In the previous paragraphs, it has become clear that two different elision phenomena are to be distinguished. On the one hand, elision on determiners is characterized by quite high overall deletion rates, and it is irrelevant whether the following vowel is stressed or not. Among the determiners, the masculine articles (il with its alternants lo and l’ and uno with its alternant un) show allomorphic variation that is phonologically conditioned. On the other hand, overall deletion rates are lower for elision on pronouns, and stress on the following vowel has a
Luigia€Garrapa & Judith Meinschaefer
negative effect on deletion rates (with the exception of the third person singular pronouns lo and la). There is no reason to assume allomorphic variation. Relevant to both phenomena is the morphophonological constraint Real-Num, blocking deletion of vowels which are exponents of the feature [plural]. 4.4 Stochastic OT and the gradual learning algorithm In this section, we will briefly show that the variation found in deletion rates for the Italian Fnc under investigation can best be represented in a stochastic OT grammar if two different grammars are computed for elision on determiners and on pronouns. The two grammars proposed may be thought of as corresponding to different strata in a Stratal Optimality-Theoretic model along the lines of Bermúdez-Otero (1999, 2003, 2007) and Kiparsky (2000). In order to compute an optimality-theoretic grammar that models the frequencies of elided and full forms we observe in our data, we use the Gradual Learning Algorithm (henceforth GLA) developed by Boersma & Hayes (2001). The GLA is a constraint-ranking algorithm that can learn free variation so as to represent the relative frequencies of different alternants in the input data. As a stochastic constraint-ranking algorithm it uses a continuous scale of strictness of constraints, where constraints are not assigned a fixed numeric value indicating their strictness, but a range of possible values, such that constraints with higher values outrank those with lower values, as in conventional OT. However, each particular application of the evaluation algorithm raises or lowers the constraint values in a random fashion, so that over time each constraint value takes the form of a Gaussian distribution centered on the nominal value. The ranges of constraints with similar strictness values thus may overlap.€Whenever a given candidate set is evaluated, for each constraint a strictness value is chosen out of its value range. Hence, different rankings can occur (in different evaluation events) for two constraints with overlapping strictness value ranges. The GLA can thus not only represent free variation of alternants, but it can – to some degree at least – model the observed frequencies with which different alternants occur by selecting an appropriate strictness value for each relevant constraint. Tables 9 and 10 show the strictness values assigned by the GLA to each of the three constraints Real-Num, Ons, Max-Vfnc (based on 21,000 learning trials), as well as the frequencies with which the full and the elided forms are generated (based on 200,000 testing cycles) by the grammar that was learned to model the input frequencies given in Table€3 and 4 above (and repeated in tables 9 and 10). When testing whether the grammar correctly models the input frequencies of the full and the elided forms, the average error per candidate is around 6€percent for the
Morphology and phonology of final vowel deletion in Tuscan Italian
Table 9.╇ Inputs, candidate sets, constraint tableaux and frequencies for determiners Input
Candidates
[lo amico] [l’amico] /la oliva/ [la oliva] [l’oliva] /gli amici/ [gli amici] [gl’amici] /le olive/ [le olive] [l’olive] /uno + un amico/ [uno amico] [un amico] /una oliva/ [una oliva] [un’oliva] /questo amico/ [questo amico] [quest’amico] /questa oliva/ [questa oliva] [quest’oliva] /questi amici/ [questi amici] [quest’amici] /queste olive/ [queste olive] [quest’olive]
Constraints Real-Num Onset Max-Vfnc Observed Generated 106 102 98 frequencies frequencies
/lo + l’ amico/
* * * * *
* *
*
* * * * * * * * * *
*
* *
*
*
â•⁄â•⁄ 0 100 â•⁄â•⁄ 1 â•⁄ 99 â•⁄ 84 â•⁄ 16 â•⁄ 92 â•⁄â•⁄ 8 â•⁄â•⁄ 0 100 â•⁄â•⁄ 3 â•⁄ 97 â•⁄ 14 â•⁄ 86 â•⁄ 15 â•⁄ 85 â•⁄ 80 â•⁄ 20 â•⁄ 79 â•⁄ 21
â•⁄ 0 100 â•⁄ 8 â•⁄ 92 â•⁄ 92 â•⁄â•⁄ 8 92 â•⁄â•⁄ 8 â•⁄â•⁄ 0 100 â•⁄â•⁄ 8 â•⁄ 92 â•⁄â•⁄ 8 â•⁄ 92 â•⁄â•⁄ 8 â•⁄ 92 â•⁄ 92 â•⁄â•⁄ 8 â•⁄ 92 â•⁄â•⁄ 8
grammar for determiners and around 14€percent for the grammar for pronouns. Furthermore, when a single grammar is learned for both elision on determiners and pronouns, the average error per candidate goes up to around 32€percent. For determiners, the higher overall deletion rates are mirrored in a higher strictness value assigned to Ons as compared to Max-Vfnc. For pronouns, in contrast, the pattern of variation is closer to free variation between the full and the elided form. This is represented in the model by the (nearly) identical values of Max-Vfnc and Ons. The average error per candidate is higher for pronouns than for determiners, indicating that for pronouns the pattern of variation is noisier and less categorical. The two different grammars computed may thus be thought of as corresponding to two different strata in a Stratal OT model (integrating concepts of Lexical Phonology and Morphology, Kiparsky 1982, with Optimality Theory; cf.€Bermúdez-Otero 1999, 2003, 2007; Kiparsky 2000), where the grammar for determiners is located on a stratum which is closer to the lexical phonology and the grammar for pronouns is located on the postlexical stratum.
Luigia€Garrapa & Judith Meinschaefer
Table 10.╇ Inputs, candidate sets, constraint tableaux and frequencies for pronouns Constraints Input
Candidates
[lo usava] [l’usava] /la usava/ [la usava] [l’usava] /li usava/ [li usava] [l’usava] /le usava/, Acc Pl [le usava] [l’usava] /mi usava/ [mi usava] [m’usava] /ci usava/ [ci usava] [c’usava] /gli ometteva/ [gli ometteva] [gl’ometteva] /le ometteva/, Dat [le ometteva] [l’ometteva]
Real-Num Max-Vfnc Onset Observed Generated 106 100 100 frequencies frequencies
/lo usava/
* * * * * *
*
*
*
* * * * * * * * *
â•⁄ 61 â•⁄ 39 â•⁄ 53 â•⁄ 47 â•⁄ 92 â•⁄â•⁄ 8 â•⁄ 95 â•⁄â•⁄ 5 â•⁄ 74 â•⁄ 26 â•⁄ 57 â•⁄ 43 â•⁄ 54 â•⁄ 46 100 â•⁄â•⁄ 0
50 50 50 50 98 â•⁄ 2 98 â•⁄ 2 50 50 50 50 50 50 50 50
5. Conclusion This study has shown that Italian elision is not entirely optional; rather, it is governed by morphological factors and, to some degree at least, also by phonological factors. Furthermore, elision on determiners and on pronouns shows systematic differences: Elision on determiners is less variable than elision on pronouns; the former has partly developed into an allomorphic alternation, while the latter has not; the former is not sensitive to stress on the first vowel of the following Lex, while the latter is at least partially influenced by stress. On the one hand, an OT-grammar modeling the observed frequencies of the elided and full forms of determiners can relatively straightforwardly be learned by the GLA; on the other hand, elision on pronouns seems to be controlled by additional, possibly nongrammatical factors, such as speaking rate, and is less successfully learned by the GLA. Finally, the GLA is much less successful in learning a single, combined grammar for elision on both determiners and pronouns.
Morphology and phonology of final vowel deletion in Tuscan Italian
References Agostiniani, Luciano. 1989. “Fenomenologia dell’Elisione nel Parlato in Toscana”. Rivista di Dialettologia Italiana 13.7–46. Anttila, Arto. 2006. “Variation and Opacity”. Natural Language and Linguistic Theory 24.893–944. Bermúdez-Otero, Ricardo. 1999. Constraint interaction in language change: Quantity in English and Germanic. Unpublished PhD dissertation, University of Manchester & Universidad de Santiago de Compostela. Bermúdez-Otero, Ricardo. 2003. “The acquisition of phonological opacity”. Variation within Optimality Theory: Proceedings of the Stockholm Workshop on Variation within Optimality Theory ed. by Jennifer Spenader, Anders Eriksson & Östen Dahl, 25–36. Stockholm: Department of Linguistics, Stockholm University. Bermúdez-Otero, Ricardo. 2007. “Morphological structure and phonological domains in Spanish denominal derivation”. Optimality-theoretic studies in Spanish phonology ed. by Fernando Martínez-Gil & Sonia Colina, 278–311. Amsterdam & Philadelphia: John Benjamins. Bisol, Leda. 2003. “Sandhi in Brazilian Portuguese”. Probus 15.177–200. Boersma, Paul & Bruce Hayes. 2001. “Empirical Tests of the Gradual Learning Algorithm”. Linguistic Inquiry 32.45–86. Cabré, Maria Teresa & Pilar Prieto. 2003. “Vowel Contact Resolutions across Word Boundaries in Catalan”. Proceedings of the XVth International Congress of Phonetic Sciencies, Barcelona ed. by M. J. Solé et al., 1–4. Cardinaletti, Anna & Ur Shlonsky. 2004. “Clitic Positions and Restructuring in Italian”. Linguistic Inquiry 35.519–557. Cresti, Emanuela & Massimo Moneglia, eds. 2005. C-ORAL-ROM: Integrated Reference Corpora for Spoken Romance Languages. Amsterdam & Philadelphia: John Benjamins. Embick, David & Rolf Noyer. 2005. “Distributed Morphology and the Syntax/Morphology Interface”. The Oxford Handbook of Linguistic Interfaces ed. by Gillian Ramchand & Charles Reiss, 289–324. Oxford: OUP. Garrapa, Luigia. 2009.‘‘Vowel Elision in Spoken Italian’’. Romance Languages and Linguistic Theory 2006. Selected papers from ‘Going Romance’. Amsterdam, 7–9€December 2006 ed. by In Danièle Torck & W. Leo Wetzels, 73–87. Amsterdam: Benjamins. Garrapa, Luigia. 2008. “Contro l’impredicibilità dell’elisione vocalica applicata ai pronomi: il ruolo svolto da morfologia, fonologia, prosodia, frequenza e timbro vocalico.” Quaderni del Dipartimento di Linguistica dell’Università di Firenze 18.27–48. Gili Fivela, Barbara & Pier Marco Bertinetto. 1999. “Incontri Vocalici tra Prefisso e Radice (Iato o Dittongo?)”. Archivio glottologico italiano 84.129–172. Greenberg, Joseph. 1966. “Some Universals of Grammar. With Particular Reference to the Order of Meaningful Elements”. Universals of Language ed. by Joseph Greenberg, 73–113. Cambridge, Mass.: MIT Press. Iscrulescu, Christian. 2005. “Morpheme Realization and Morphological Coalescence”. Theoretical and Experimental Approaches to Romance Linguistics ed. by Randall Scott Gess & Edward Rubin, 145–160. Amsterdam & Philadelphia: John Benjamins. Kiparsky, Paul. 1982. “Lexical phonology and morphology”. Linguistics in the morning calm. Selected papers from SICOL-1981 ed. by In-Seok Yang, 1–91. Seoul: Hanshin Kiparsky, Paul. 2000. “Opacity and Cyclicity”. The Linguistic Review 17:351–366. Mascaró, Joan. 1996. “External Allomorphy and Contractions in Romance”. Probus 8.181–205.
Luigia€Garrapa & Judith Meinschaefer Mascaró, Joan. 2007. “External Allomorphy and Lexical Representation”. Linguistic Inquiry 38.715–735. Mayerthaler, Willi. 1981. Morphologische Natürlichkeit. Wiesbaden: Athenaion [English translation 1988. Morphological Naturalness. Ann Arbor: Karoma] McCarthy, John & Alan Prince. 1995. “Faithfulness and Reduplicative Identity”. University of Massachusetts Occasional Papers in Linguistics 18: Papers in Optimality Theory ed. by Jill Beckman, Suzanne Urbanczyk & Laura Walsh Dickey, 249–384. Amherst, Mass.: GSLA. Nagy, Naomi & Bill Reynolds. 1997. “Optimality Theory and Variable Word-final Deletion in Faetar.” Language Variation and Change 9.37–55. Nespor, Marina. 1990. “Vowel Deletion in Italian. The Organization of the Phonological Component”. The Linguistic Review 7.375–398. Perlmutter, David. 1998. “Interfaces: Explanation of Allomorphy and the Architecture of Grammars”. Morphology and its Relation to Phonology and Syntax ed. by Steven Lapointe, Diane Brentari & Patrick Farrell, 307–338. Stanford: CSLI. Prince, Alan & Paul Smolensky. 2004 [1994]. Optimality Theory. Constraint Interaction in Generative Grammar. Oxford: Blackwell. Rosati, Luca. 2001. “Variabilità della Pronuncia e Variabilità Sociale in Italia: I Fenomeni di Sandhi nelle Indagini Recenti”. Studi Italiani di Linguistica Teorica e Applicata 30.455–488. Van Oostendorp, Marc. 1997. “Style Levels in Conflict Resolution”. Variation, Change and Phonological Theory ed. by Frank Hinskens, Roeland Van Hout & Leo Wetzels, 207–229. Amsterdam & Philadelphia: John Benjamins. Vogel, Irene, Marina Drigo, Alessandro Moser & Irene Zannier. 1983. “La Cancellazione di Vocale in Italiano”. Studi di Grammatica Italiana 12.189–230.
Relativization, intonational phrases and rich left peripheries in Romanian Simona Herdan* This paper provides an analysis of relativizer restrictions in relatives headed by bare quantifiers. I argue that the restriction arises from violations of the requirement to phonetically mark the edge of an intonational phrase. Consequently, the structural position of the relativizer determines grammaticality. I also show that the availability of D in the nominal domain has an effect on the relativizer choice. Keywords: relativizer restrictions, intonational phrase boundaries, bare quantifiers, determiner phrases
1. Introduction This paper aims to account for some novel data involving the distribution of relativizers in relatives headed by quantifiers. Among the Slavic languages, SerboCroatian, Polish and Russian disallow the agreeing relativizer in relative clauses headed by a bare quantifier, while Bulgarian shows no such restriction. (1)
Našao sam sve što/*koje sam found.M.sg be.aux.1sg all that/which be.aux.1sg želeo. (Serbo-Croatian) wanted.m.sg I found all I wanted.
(2) Namerih vsichko koeto/deto go iskah. found.1sg all which/what it wanted.1sg
(Bulgarian)
Interestingly, Romanian shows mixed behavior, with the restriction affecting only the bare quantifier tot (all). I will therefore use Romanian to show that * I thank audiences at PLC32, LSRL38 and the University of Toronto Syntax Project for useful comments and discussion of this material. Many thanks also to Željko Bošković for many detailed discussions. My informants’ help is also gratefully acknowledged. All errors and omissions are my own.
Simona Herdan
nominal domain structure, in particular the availability of D, plays a role in this distinction. I will argue that the restriction itself results from the need to align intonational phrasing with the syntax. I will discuss the consequences of these facts for the syntax of the complementizer area of the clause and conclude that these restrictions provide evidence for locating the relevant relativizers, roughly corresponding to that and which in English, not just in different places in one phrase, but in different phrases altogether. The paper is organized as follows. Section€2 presents the paradigm to be accounted for, followed by an outline of the proposal. Section€3 presents evidence for the role of intonational phrasing in these relativizer restrictions and discusses the implications for the syntax of relativization. Finally, Section€4 addresses the derivation of the relativizer restrictions and discusses the role of nominal projections. Section€5 contains the conclusions. 2. The data and a proposal 2.1
The basic paradigm
Languages that possess more than one relativizer generally allow both in the vast majority of cases, as illustrated in (3) for English and in (4) for Serbo-Croatian.
(3) John got the present that/which/Ø he wanted.
(4) Zoran je dobio poklon što/koje je želeo. Zoran is got present that/which is wanted Zoran got the present he wanted. However, a closer look at relative clauses involving bare quantifiers reveals that some languages only allow in these cases the invariant relativizer, corresponding roughly to English that. This holds in several Slavic languages including SerboCroatian, Polish and Russian, illustrated in (5)-(7). (5) Serbo-Croatian a. Našao sam sve što/*koje sam želeo. found.m.sg be.aux.1sg all that/which be.aux.1sg wanted.m.sg I found all I wanted. b. Našao sam nešto što/*koje sam želeo. found.m.sg be.aux.1sg sth. that/which be.aux.1sg wanted.m.sg I found something I wanted. c. Nisam našao ništa što/*koje sam not-be.aux.1sg found.m.sg nothing that/which be.aux.1sg
Relativization, intonational phrases and rich left peripheries in Romanian
želeo. wanted.m.sg I didn’t find anything I wanted.
(6) Polish a. Jan włoył do kieszeni wszystko co/*które mógł Jan put.pst to pocket all that/which can.pst Jan put in his pocket everything that/*which he could. b. Jan przyniósł co co/*które mnie zaskiczyło. Jan brought something that/which me.dat surprised Jan brought something that/which surprised me. c. Jan nie przyniósł niczego co/*które mnie zaskoczyło. Jan neg brought nothing that/which me.dat surprised. Jan didn’t bring anything that/which surprised me. (7) Russian a. Ja kupil vsë čto/*kotoroe ty prosil. I bought all that/which you asked b. Ja zabyl čto-to čto/*kotoroe ty poprosil. I forgot something that/which you asked. c. Ja ničego ne kupil čto/*kotoroe ty prosil I nothing ne bought that/which you asked. At strst sight, this appears to be a property of the Slavic languages. However, at least two counter-examples stand out. Bulgarian, also a Slavic language, shows no restriction at all (8), and Romanian, a Romance language, shows the restriction (9), but only with the bare quantister tot (all/everything). (8) Bulgarian a. Namerih vsichko koeto/deto go iskah. found.1sg all which/what it wanted.1sg b. Namerih neshto koeto/deto go iskah. found.1sg something which/what it wanted.1sg c. Ne namerih nishto koeto/deto go iskah not found.1sg nothing which/what it wanted.1sg
(9) Romanian1
1. The relativizer ce (‘that’, literally ‘what’) is generally used in literary language, and may sound unnatural in some informal contexts. Surprisingly, however, it is the only natural option with the bare quantifier tot (‘all’):
Simona Herdan
a. Ion a pus în buzunar tot ce/*care a putut. Ion have.aux.3sg put.ptcp in pocket all that/which aux can.ptcp #John put in his pocket everything he could. b.
Nu am găsit nimic ce/care not have.aux.1sg found nothing that/which să-mi placă. sbjv-me.dat like.sbjv I didn’t find anything I might like.
c.
Am găsit ceva ce/care have.aux.1sg found something which mi-a plăcut mult. me.dat-have.aux.3sg liked.ptcp much I found something I liked a lot.
What Bulgarian and Romanian share is the use of articles. As I will show in Section€4, the availability of D rather than the article itself determines the relativizer options. I will argue that this restriction is triggered by failure to overtly mark the edge of an intonational phrase determined by a phase. 3. The role of intonational phrasing 3.1
Evidence from Romanian
Interestingly, the quantifier tot in Romanian can combine with nouns bearing the definite article or bear the article itself: (10) Tot zaharul/Totul e pe jos. All sugar-the/all-the is on floor All the sugar/Everything is on the floor. Surprisingly, the presence of the destnite article on tot precludes that of a relative clause, as in (11), regardless of relativizer choice, unlike in (12) where an overt noun bears the destnite article. (11) *Le place totul ce/care depăşeşte limita. them.dat like all-the that/which exceeds limit-the They like everything that is beyond the limit. (12) Le plac toate lucrurile ce/care depăşesc limita. them.dat like all things-the that/which exceed limit-the
Relativization, intonational phrases and rich left peripheries in Romanian
Since the Romanian destnite article undergoes phonological encliticization to the noun that moves to its left (Giusti, 1997, 2002), I take the above facts as evidence that there is an intonational phrase boundary (IPB) at the DP level. IPBs have long been known to prevent the application of certain phonological processes in many environments (see Cooper and Paccia-Cooper 1980, Selkirk 1984, 1986, Nespor and Vogel 1986, Schütze 1994, Bošković 2001, and An 2007). I assume that (11) is ungrammatical because the destnite article cannot encliticize across an IPB. To understand that, let us consider brie﬇y the syntax of relativization. I assume with Kayne (1994) that relative clauses are generated as a complement to D with the head noun raising from inside the relative clause.2 The noun moves strst to a position at the edge of the relative clause and then to D, via N-to-D movement (see Dimitrova-Vulchanova and Giusti 1998, Bouchard 2002, Dobrovie-Sorin 2000, and Ticio 2003 among others; for alternative views, see Dimitrova-Vulchanova 2003). In the case of (11), without a noun, encliticization of the destnite article on Q needs to take place across an IPB (marked with #), which triggers ungrammaticality. If a noun is present, as in (12) above, encliticization does not cross the DP phase/IPB; rather, the noun itself moves to D. (13)
QP
g Q' 3 Q
#DP
g
g
tot
D' 3 D
CP
g
g
-ul
C' 3 C
IP
g ce/care depășește limita 2. Carlson (1977), Sauerland (1998, 2003), and Bhatt (2002) argued that relatives are ambiguous between a raising and a matching analysis. In our case, nothing forces one choice over the other.
Simona Herdan
(14)
DP 3 toate
D' eo D
AgrP
g
3
lucruri1-le
t1
Agr' 3 Agr
CP 3 C' 3
t1
C
IP
g ce
depășesc limita t1
At this point, one may wonder why DP would determine an intonational phrase. This idea is actually natural given the assumption that intonational phrases correspond to phases, i.e. spellout domains, and the fact that DP has been argued to be a phase (see Bošković 2005 and Svenonius 2004). However, to make this account work, we must assume that intonational phrases correspond to full phases.3 3.2
Evidence from Serbo-Croatian
In the previous section I proposed that the inability of the Romanian quantister tot to be modisted by a relative clause when encliticization of the destnite article is necessary suggests that intonational phrasing may interfere with this process. I now provide further evidence for the existence of an IPB in relative clauses, this time at the CP level. Radanović-Kocić (1996) and Bošković (2001) argued convincingly that Serbo-Croatian clitics occur in the second position of their intonational phrase, a phonological rather than syntactic requirement. Compare the grammatical (15a) and the ungrammatical (15b), which differ only in the position of the auxiliary clitic sam. The ungrammaticality is not merely caused by the clitic being preverbal, as evidenced by the grammaticality of (15c). 3. It is standardly assumed that a CP boundary corresponds to an IPB. This would mean that IPBs correspond to phasal boundaries, not to complements of phasal heads. This, however, does not mean that, with respect to other phenomena, the complement of a phase head cannot play a special role, as in Chomsky’s work.
Relativization, intonational phrases and rich left peripheries in Romanian
(15) a. Našao sam sve što sam želeo. found.m.sg be.aux.1sg all that be.aux.1sg wanted.m.sg I found all I wanted. b. *Našao sam sve što želeo sam. found.m.sg be.aux.1sg all that wanted.m.sg be.aux.1sg c. Želeo sam to. wanted.m.sg be.aux.1.sg it I wanted it. If clitics are located in the second position of their intonational phrase, there must be an IPB between the quantifier sve and the relativizer što. Let us turn to explaining the basic contrast introduced in Section€2, partially repeated in (16). (16) Našao sam sve što/*koje sam želeo. found.m.sg be.aux.1sg all that/which be.aux.1sg wanted.m.sg I found all I wanted. We need to explain why only the invariant relativizer što is allowed in relatives with bare quantisters. My proposal builds on An (2007), who argues that the edge of an IPB (the specister or the head of the phase that is mapped into an IPB) needs to be overtly realized, as illustrated by (17). Subject clauses are only grammatical when either the spec or the head of the CP is overtly stlled. (17) a. [What Ø John likes] is apples. (Bošković 1997:€182) b. [That John likes apples] is widely known. c. *[Ø John likes apples] is widely known. Regarding our cases, I propose that relativizers not located in Cº (Serbo-Croatian koje, Polish które and Romanian care) cause ungrammaticality since the edge of the IPB at the CP-level is not properly marked. This interpretation of the relativizer restriction has consequences for the syntax of relativization in general. For Kayne (1994), relativizers differ only in category, both being located in the CP, which is incompatible with the present account since it would not predict ungrammaticality in the relevant cases. Bianchi (1999) proposes an alternative account, with different positions for the relativizers, shown below for SerboCroatian in (18) and (19). The relativizer što, which is located in Force0 (CP), properly marks the edge of the IPB at the relative clause level. On the other hand, koje, which is in SpecTopP4, leaves the ForceP phonetically unrealized, causing ungrammaticality. 4. An anonymous reviewer points out that multiple topics in Italian cannot precede the relative pronoun. However, I am not concerned here with the label of the projection targeted by the relative pronoun. The topics and the relative pronoun may well target different phrases. I leave this issue open for future research.
Simona Herdan
(18)
QP
g Q′ 3 Q
#ForceP 3
g
0
sve
Force′ 3 Force
TopP
g
g
što
Top′ 3 Top
IP sam Želeo
QP
(19)
g Q′ 3 Q
#ForceP
g
g Force' ei
sve
Force
TopP
g
ei
0
XP 2 X
NP
g
g
koje
N'
g N
g 0
Top' 2 Top
IP sam Želeo
Relativization, intonational phrases and rich left peripheries in Romanian
4. The distribution of the relativizer restrictions The question remains why ForceP is sometimes obligatorily parsed as an intonational phrase, which I argued is responsible for the relativizer restriction in general (as well as clitic placement in relative clauses). The explanation cannot be merely that all relative clauses are obligatorily parsed as separate intonational phrases. An (2007) argues based on cross-linguistic data that restrictive relative clauses may be, but need not be, parsed as separate intonational phrases, unlike the closely related noun complement clauses. An (2007), citing Richards (1999), points out that Tagalog features both afstxal (-ng) and non-afstxal (na) complementizers. (20) a. ang balita [na kinain ni Juan ang tambakol] (NC) news that ate Juan mackerel the news that Juan ate the mackerel b. *ang balita [-ng kinain ni Juan ang tambakol] (NC) news that ate Juan mackerel c. ang balita [-ng dinala ni Juan] (RC) news that brought Juan the news that Juan brought. The examples in (20) and (20c) illustrate a difference in the acceptability of these complementizers in noun complements and restrictive relatives. In the noun complement clause in (20b), the afstxal complementizer is ungrammatical, unlike in the relative clause in (20c). An’s account appeals to a difference in parsing. The afstxal complementizer needs the preceding nominal as a host, so it is only expected to be grammatical if there is no IPB between the afstx (-ng) and its host (balita). This is expected if the noun complement clause in (20b) is parsed as a separate intonational phrase, which makes the afstxal complementizer unacceptable; in the relative clause in (20c), the afstxal complementizer is acceptable, which suggests that the relative clause need not be parsed as an intonational phrase. The same distinction can be made with respect to English. The noun complement clause in (21a) cannot be introduced by a null complementizer – a null afstx according to Pesetsky (1992) (cf.€ Ormazabal (1995) and Bošković and Lasnik (2003)). This is not the case in the corresponding restrictive relative in (21b). (21) a. The claim that/*Ø Mary offended Bill is unsubstantiated. (NC) b. The claim that/Ø the defense rested on was supported by the witness. (RC) The data above are consistent with a view along the lines of Stowell (1981) who treats noun complements as appositives, which are always obligatorily parsed as
Simona Herdan
separate intonational phrases. Relative clauses, on the other hand, are only optionally parsed as intonational phrases, as evidenced by the grammaticality of the null complementizer in (21b). I propose that in the case of (21b), the noun above the CP plays a role in determining how the relative clause is parsed. The top layer of the relative clause can optionally be parsed together with the preceding noun. According to Bianchi (1999), the null C option for relative clauses (and declarative sentences) is a re﬇ection of the noun selecting for the phonetically null head Topic0. The resulting structure is shown in (22).5 If the IPB were located at the level of TopP, (21b) would be expected to be ungrammatical, contrary to fact. If the TopP layer is parsed together with the noun in SpecAgrP, no ungrammaticality is expected. This also explains why relative clauses with overt nouns, such as in Tagalog above, are only optionally parsed as separate intonational phrases. (22)
DP
g D′ ei D
AgrP
g
3
the
claim
Agr′ 3 Agr
TopP
g Top′ ei Top
IP
g 0
the defense rested on
5. In sentences with an overt noun, which under Bianchi’s analysis above is in SpecForceP, the IPB would fall between the specifier and the head of the ForceP (to locate the clitic in the second position of the intonational phrase in (i) (a)), an undesirable consequence. It is therefore necessary to adopt a slightly different version of these structures, also suggested by Bianchi (1999:€200), with the head noun of the relative clause in an AgrP projection above ForceP. (i) a. svi ljudi # što su došli all men that be.aux.cl come all men who came b. *svi ljudi # što došli su all men that come be.aux.cl This allows us to maintain the idea that the IPB is at the ForceP level.
Relativization, intonational phrases and rich left peripheries in Romanian
4.1
Relativizer restrictions in languages without articles
Let us now return to the issue of the relativizer restriction in Serbo-Croatian. In (18)-(19) there is an IPB at the ForceP level, which triggers the relativizer restriction, as a result of which što, but not koje, is possible in that context. However, Serbo-Croatian relative clauses with an overt noun do not exhibit this restriction. Interestingly also, relatives with the bare quantifier sve show an improved ability to combine with koje (which) when the quantifier agrees with a previously-mentioned noun: (23) Context: There are 15 girls in the choir, but 5 are home sick. Only 10 showed up for the show. (24)
Sve [devojke] koje su došle all.fem.pl [girls] which.fem.pl aux come.3fem.pl pevale su punim srcem. sung.3fem.pl aux full.instr heart.instr All [girls] who came sang whole-heartedly.
To understand why, consider relative clauses with overt relativized nouns, which evince no relativizer restrictions, as in (25) below. (25)
Sve devojke što/koje su došle pevale all.fem.pl girls that which.fem.pl aux come.3fem.pl sung.3fem.pl su punim srcem. aux full.instr heart.instr
The absence of this restriction may be at first attributed to the presence of the overt noun devojke in SpecCP, in line with Bianchi’s original proposal, which means that the IPB needn’t be marked by the relativizer, making koje acceptable. However, recall from the discussion in Section€3 that Serbo-Croatian clitics must be second in their intonational phrase. Therefore, the question arises why (25), represented in (26), is grammatical with koje. An IPB at the CP level would not be properly phonetically marked, which would trigger ungrammaticality. Moreover, optionally parsing the CP as an intonational phrase does not yield the right results since the clitic would be located in the fourth position in the whole relative clause.
Simona Herdan
(26)
AgrP ei sve
Agr′ 3 Agr
CP
g
g C′
devojke
3 C
TopP ei koje
Top′
2 Top
IP
5 su došle
I propose that the optionality of parsing the relative clause as a separate intonational phrase should not be interpreted to mean that the noun and the whole relative clause form an intonational phrase together, but rather that the CP/ForceP layer can be optionally parsed with the noun instead of with the rest of the relative clause.6 Consider again (25) when the relativizer koje is chosen. The two parsing options exist: (27) a. the CP is parsed as a separate intonational phrase: the clitic su is in 2nd position, but IPB is not properly marked ungrammaticality b. the noun devojke and the rest of the CP/ForceP layer are parsed separately; the relative clause starting with the TopP layer is parsed separately: the clitic su is in 2nd position and IPB is marked grammaticality When što is selected, the options are the same, but the strst option results in grammaticality: (28) a. the CP is parsed as a separate intonational phrase: the clitic su is in 2nd position and the IPB is properly marked by the presence of što grammaticality 6. This says that we are in principle allowing the IPB to appear at the CP/ForceP level or at the TopP level, which is a reasonable assumption. What leads to ungrammaticality in certain cases is the existence of phonological processes for which this IPB is relevant.
Relativization, intonational phrases and rich left peripheries in Romanian
b. the noun devojke and the rest of the CP/ForceP layer are parsed separately; the relative clause starting with the TopP layer is parsed separately: the clitic su is in 1st position ungrammaticality We can now explain the difference between relatives with nounless sve and relatives with agreeing sve. I propose that the distinction in the behavior of sve stems from its structural position. If sve is inserted into the structure without a noun, it occupies the head of QP and is not in an agreement constguration. I argue that only this sve triggers the relativizer restriction, since no noun is available for the CP layer to be parsed together with. On the other hand, when sve is generated with a noun it occupies SpecAgrP7 and is in an agreement constguration with the noun in Agr. The noun will be there when intonational phrasing takes place. In the cases where agreeing sve is not followed by an overt noun, the noun has undergone PF deletion following intonational phrasing. The remaining issue is the behavior of the Serbo-Croatian generalized quantisters nešto and ništa, which also trigger a relativizer restriction, as seen in (5b) and (5c) in Section€2. These quantisters are also invariant, so I assume that they behave essentially like invariant sve. I return to the relevance of the nominal structure, i.e the availability of D, when I discuss Romanian below. 4.2
Relativizer restrictions in languages with articles/D
Let us now turn to accounting for the relativizer restrictions in languages with articles, such as Bulgarian and Romanian. Recall from the previous discussion that Bulgarian shows no relativizer restrictions, while Romanian only shows a relativizer restriction in one case: when the universal quantifier tot is unaccompanied by a noun or a definite article.8 With respect to the universal quantifier tot, Romanian behaves just like Serbo-Croatian. We can therefore assume that Romanian also has two positions for the quantifier tot: the head of QP where the nounless quantifier appears, triggering a relativizer restriction, and SpecDP where the quantifier is always accompanied by a covert or overt noun with which it agrees. Recall that it is the presence of this noun that is responsible for the optionality in the location of the IPB of the
7. On Bianchi’s structure, the noun would occupy SpecAgrP. I adopt this slight modification following a suggestion by an anonymous John Benjamins reviewer that a difference related to structural distinctions is preferable to lexical ambiguity. 8. See (11), which shows that tot accompanied by a definite article cannot be modified by a relative clause. I have already given an account of these facts, which do not merely illustrate a relativizer restriction, since all relativizers are bad.
Simona Herdan
relative clause, as seen for Serbo-Croatian above. The CP layer has the option of being parsed with the noun, to the exclusion of material in TopP and below.9 On the other hand, the quantifiers ceva (something) and nimic (nothing) do not trigger a relativizer restriction, unlike in Serbo-Croatian, a fact to be accounted for. Essentially, relatives involving these quantifiers behave like those where a noun is projected. It is therefore possible that these quantifiers are the phonetic realization of a nominal projection incorporated into D (cf.€Bianchi’s AgrP), since they cannot combine with nouns, but only with adjectives, as shown in (31). (31) ceva bun/*creion something good/*pencil In fact, the English counterparts of ceva and nimic do overtly what Romanian may do covertly, i.e they incorporate a noun – thing. The Romanian ceva is also likely bimorphemic, though less transparently so. Va can combine with all the wh-forms, including ce (what), yielding indestnite quantisters: cineva (someone), careva (anyone), undeva (somewhere), etc. Here the va morpheme corresponds to a nominal while ce- is located in the DP, which also straightforwardly accounts for the data in (31), both the impossibility of combining with a noun and the word order with adjectives. Given what I have said above about the parsing effect of an overt nominal in the relative clause, I assume that the nominal part of these quantisters is responsible for optionality in the location of the IPB in this case as well. Before closing this section I return to Bulgarian, which by now is expected to behave like Romanian since it has a destnite article, a rare feature among the Slavic languages. Unlike in Romanian, the bare quantister vsičko always allows both relativizers – the invariant deto and the agreeing koeto. In light of our analysis of Romanian and Serbo-Croatian one possible explanation is that vsičko can never appear unaccompanied by a noun. For the other generalized quantisters (nešto and ništo), the morphological analysis presented for Romanian carries over straightforwardly.
9. An interesting question raised by an anonymous Benjamins reviewer is whether this analysis can be extended to cases where which is disallowed in English in structures without bare quantifiers. Regrettably, this does not seem likely, as the relevant part of the structure is identical in the following examples: (29) The books that/*which there were on the table. (30) The books that/which were on the table. Moreover, while the former example requires raising, there is no evidence to that effect for ACD relatives in English which show the same relativizer restriction (see Herdan (2008) for details).
Relativization, intonational phrases and rich left peripheries in Romanian
5. Summary and conclusions In this paper I presented an account of novel data involving the distribution of relativizers in some Slavic languages and in Romanian. I have argued that the observed relativizer restrictions arise from failure to satisfy the conditions for marking an IPB at the level of the relative clause. This approach has consequences for the syntax of relativization in that the two kinds of relativizers (invariant and agreeing) are located in different phrases at the left periphery of the clause (ForceP and TopP, respectively). Moreover, I show that the structure of the nominal domain, in particular the availability of D, plays a role in determining the relativizer options. References An, Duk-Ho. 2007. Syntax at the PF interface: Prosodic mapping, linear order, and deletion. Ph.D. Thesis, University of Connecticut, Storrs. Bhatt, Rajesh. 2002. The raising analysis of relative clauses: Evidence from adjectival modistcation. Natural Language Semantics 10:43–90. Bianchi, Valentina. 1999. Consequences of antisymmetry – headed relative clauses. Studies in Generative Grammar 46. Mouton de Gruyter. Bošković, Željko. 2001. On the nature of the syntax-phonology interface: Cliticization and related phenomena. Amsterdam: Elsevier Science. Bošković, Željko. 2005. On the locality of left branch extraction and the structure of NP. Studia Linguistica 59:1–45. Bošković, Željko, and Howard Lasnik. 2003. On the distribution of null complementizers. Linguistic Inquiry 34:527–546. Bouchard, Denis. 2002. Adjectives, number, and interfaces. Why languages vary?. Amsterdam: North Holland/Elsevier. Carlson, Greg. 1977. Amount relatives. Language 53:520–542. Cooper, William, and Jeanne Paccia-Cooper. 1980. Syntax and speech. Cambridge, Mass.: Harvard University Press. Dimitrova-Vulchanova, Mila. 2003. Modistcation in the Balkan nominal expression. An account of the (A)NA: AN(*A) order constraint. In From NP to DP: Volume€1: The syntax and semantics of noun phrases, ed. by Martine Coene and D’Hulst Yves, Linguistik Aktuell, 91–118. Amsterdam/Philadelphia: John Benjamins. Dimitrova-Vulchanova, Mila, and Giuliana Giusti. 1998. Fragments of Balkan nominal structure. In Topics in South Slavic syntax and semantics, ed. by A. Alexiadou and C. Wilder, 330–360. Amsterdam: John Benjamins. Dobrovie-Sorin, Carmen. 2000. (In)destniteness spread: from Romanian genitives to Hebrew construct state nominals. In Comparative studies in Romanian syntax, ed. by Virginia Motapanyane, 177–226. Elsevier/North Holland. Giusti, Giuliana. 1997. The categorial status of determiners. In The new comparative syntax, ed. by L. Haegeman, 95–123. London: Longman Linguistics Library.
Simona Herdan Giusti, Giuliana. 2002. The functional structure of noun phrases: A bare phrase structure approach. In Functional structure in the DP and the IP. The cartography of syntactic structures, ed. by G. Cinque, Volume€1, 54–90. Oxford University Press. Herdan, Simona. 2008. Degrees and amounts in relative clauses. Ph.D. Thesis, University of Connecticut, Storrs. Kayne, Richard. 1994. The antisymmetry of syntax. Linguistic Inquiry Monographs. Cambridge, Mass.: MIT Press. Nespor, Marina, and Irene Vogel. 1986. Prosodic phonology. Dordrecht: Foris. Ormazabal, Javier. 1995. The syntax of complementation: on the connection between syntactic structure and selection. Ph.D. Thesis, University of Connecticut, Storrs. Pesetsky, David. 1992. Zero syntax: Volume€ 2 – instnitives. Unpublished Ms., MIT, Cambridge, Mass. Radanović-Kocić, Vesna. 1996. Placement of Serbo-Croatian clitics: A prosodic approach. In Approaching second: Second position clitics and related phenomena, ed. by Aaron Halpern and Arnold Zwicky, 429–445. Stanford, Calif.: CSLI. Richards, Norvin. 1999. Complementizer cliticization in Tagalog and English. In Proceedings of the 6th meeting of the Austronesian Formal Linguistics Association: Toronto Working Papers in Linguistics 16. Sauerland, Uli. 1998. The meaning of chains. Ph.D. Thesis, MIT, Cambridge, Mass. Sauerland, Uli. 2003. Unpronounced heads in relative clauses. In The interfaces: Deriving and interpreting omitted structures, ed. by K. Schwabe and S. Winkler, 205–226. Amsterdam: John Benjamins. Schütze, Carson. 1994. Serbo-Croatian second position clitic placement and the phonologysyntax interface. In Papers on phonology and morphology.MIT Working Papers in Linguistics, Volume€21, 373–473. Cambridge, Mass.: MITWPL. Selkirk, Lisa. 1984. Phonology and syntax. the relation between sound and structure. Cambridge, Mass.: MIT Press. Selkirk, Lisa. 1986. On derived domains in sentence phonology. Phonology Yearbook 3:371–405. Stowell, Tim. 1981. Origins of phrase structure. Ph.D. Thesis, MIT, Cambridge, Mass. Svenonius, Peter. 2004. On the edge. In Peripheries: Syntactic edges and their effects, ed. by David Adger, Cécile de Cat, and George Tsoulas, 261–287. Dordrecht: Kluwer. Ticio, Emma. 2003. The structure of the Spanish DP. Ph.D. Thesis, University of Connecticut, Storrs.
Stress domain effects in French phonology and phonological development* Yvan Rose & Christophe dos Santos
Memorial University of Newfoundland & Radboud Universiteit Nijmegen In this paper, we discuss two distinct data sets. The first comes from the allophonic process of closed-syllable laxing in Québec French, which targets final (stressed) vowels even though these vowels should in theory be syllabified in open syllables in lexical representations (e.g.€lune [Öly.n(6)]; *[Öly.n(6)]). The second is found in the forms produced by a first language learner of European French, who displays an asymmetry in her production of CVC versus CVCV target (adult) forms. The former displays full preservation (with concomitant manner harmony) of both consonants (e.g.€passe → [Öpat]). The latter undergoes deletion of the initial syllable if the consonants are not manner-harmonic in the input (e.g.€appétit → [piÖti] versus tennis → [Öni]). We argue that both patterns can be explained in an approach that draws a formal distinction between phonological representation and phonetic implementation.
1. Introduction Within the field of modern phonological theory (e.g.€Chomsky & Halle 1968), the relationship between surface (phonetic) and deeper (phonological) levels of representation has constantly been subject to fierce controversy. While the 1980’s mainstream literature generally favored a clear separation of the two levels, the rise of output-driven approaches (e.g.€Prince & Smolensky 1993) contributed to a weakening of this trend. In this context, some phonologists have even proposed
* We wish to thank members of the audience at LSRL 38 for their useful questions and suggestions. This paper also benefited from independent discussions during a colloquium presentation at the Universidade de Lisboa by the first author in June 2008. We are also grateful to one anonymous reviewer for constructive feedback and illuminating comments. Of course, all errors or omissions remain our own.
Yvan Rose & Christophe dos Santos
versions of phonological theory that directly incorporate phonetics into phonology (e.g.€Steriade 1999). In this paper, we take a position against this integrated approach to phonetics and phonology; we argue for the maintenance of a formal distinction between phonological and phonetic levels of analysis. Our argument is based on data from two different sources. The first is the process of closed-syllable laxing in Québec French; the second relates to a number of asymmetries in the productions of a first language learner of European French. As we will see, both patterns, which have different expressions in surface forms, can share the same basic analysis if one considers both formal aspects of prosodic representations and general properties of domain-final (right-headed) stress systems. Because of space limitations, we will be primarily concerned with the formal approach needed to account for these phenomena, as opposed to their actual origin. Concerning the attestation of closed-syllable laxing in most dialects of Québec French versus its absence in other dialects, we would need to foray into the historical phonology of various dialects of French, an investigation that would extend far beyond the scope of this paper. Similarly, phonological patterns observed in child language development can arise from a number of (often interacting) factors. A close scrutiny of all of these factors would require extensive discussions that are not possible in the context of the current work. We refer the interested reader to Rose (2009) for a recent discussion on the emergence of phonological processes in child language phonology. 2. The data 2.1
Closed-syllable laxing in Québec French
As documented in much of the relevant literature, several dialects of Québec (or Canadian) French display an allophonic process referred to as closed-syllable laxing (e.g.€ Walker 1984; Charette 1991). As its name indicates, this process, which targets underlying high tense vowels, yields their laxing in phonetically closed unstressed and stressed syllables, as exemplified in (1).1
1. For additional observations about high vowel laxing in Québec French and related processes, see, for example Walker (1984), Déchaîne (1991), Poliquin (2007) and references therein.
Stress domain effects in French phonology and phonological development
(1) Closed-syllable laxing* Unstressed syllables a.╇ Open syllables
b.╇ Closed syllables
pitance lunette toucher pistache pulpeux rumba
[pi.Öts] [ly.Ön7t] [tu.Ö∫e] [pis.Öta∫] [pyl.Öpø] [t~m.Öba]
Stressed syllables petit vue beaucoup petite lune roucoule
[p6.Ötsi˜] [Övy˜] [bo.Öku˜] [p6.Ötsit] [Ölyn] [tu.Ök~l]
* Here and elsewhere, words are presented in citation forms, with stress on the final syllable; as we will see later, although it appears in citation forms, stress is not lexical but only occurs in phrase-final position, where vowels in open syllables or followed by voiced fricatives (see immediately below) display increased duration.
Closed-syllable laxing in stressed syllables however has a systematic exception to it: in stressed syllables, if the high vowel is followed by a voiced fricative ([v], [z], [Š], or[t]2), it undergoes lengthening instead of laxing, as in (2).
(2) Stressed high vowel lengthening before voiced fricatives
2.2
a.╇ Open σ: no process
b.╇ Closed σ (stressed): lengthening
virer gruger viser cuver
vire gruge vise cuve
[vi.Öte˜] [:ty.ÖŠe˜] [vi.Öze˜] [ky.Öve˜]
[Övi˜t] [Ö:ty˜Š] [Övi˜z] [Öky˜v]
désire refuge devise effluve
[de.Özi˜t] [t6.Öfy˜Š] [d6.Övi˜z] [7.Öfly˜v]
CVC versus CVCV forms in the acquisition of European French
The second data set comes from the longitudinal, naturalistic study of a child code-named Marilyn, a learner of European French, documented in dos Santos (2007). Of interest is a systematic pattern observed between the ages 1;10.17 and 2;00.25, a period during which Marilyn showed different behaviors in her productions of CVC versus CVCV target (adult) forms. First, as exemplified in (3), Marilyn preserved both consonants of CVC targets in her productions, which also display manner (continuancy) harmony.3 2. [t] is phonologically a rhotic but phonetically a fricative in Québec French. 3. We also observe place harmonies in some forms; these harmony processes, which are irrelevant for the present discussion, are discussed further in dos Santos (2007).
Yvan Rose & Christophe dos Santos
(3) CVC targets: Consonant preservation and harmony Orthography bottes berk coupe couettes passe case triche soupe
IPA Target
IPA Actual
Age
[Öbàt] [Öb7tk] [Ökup] [Ökw7t] [Öpas] [Ökaz] [Ötti∫] [Ösup]
[Öbàt] [Öb7k] [Ökup] [Ök7k] [Öpat] [Ökak] [Ötit] [Öpup]
2;00.12 1;11.13 2;00.25 1;11.28 1;11.13 1;10.17 1;11.13 2;00.25
As opposed to this, CVCV forms showed two distinct behaviors. On the one hand, target forms whose consonants have the same manner features, in (4a), were produced with full preservation of both consonants. On the other hand, target forms containing consonants with different manner features underwent deletion of the first syllable, in (4b).
(4) Multisyllabic target forms*
a. Manner-harmonic: full preservation of the two syllables appétit biquet escargot chaussure
[apeÖti] [biÖk7] [7skatÖ:o] [∫oÖsyt]
[piÖti] [beÖk7] [kaÖko] [lyÖly]
2;00.12 1;11.13 1;11.13 1;11.28
b. Manner-disharmonic: truncation of the initial syllable cassé tennis château Lego morceau jumeaux lunette
[kaÖse] [teÖnis] [∫aÖto] [leÖ:o] [màtÖso] [ŠyÖmo] [lyÖn7t]
[Öle] [Öni] [Öto] [Öko] [Ölo] [Ömo] [Ön7]
1;11.13 1;11.02 1;11.02 1;11.02 2;00.25 1;11.28 2;00.12
* Coronal fricatives were independently realized as laterals in onsets.
Stress domain effects in French phonology and phonological development
Further evidence for the relationship between the manner harmony and deletion processes observed above and word shape (CVC versus CVCV) comes from the observations that at no point in the longitudinal study does Marilyn present a stage during which she manner-harmonizes consonants in CVCV forms. In addition, syllable truncation in manner-disharmonic CVCV forms disappears at the same time as manner harmony disappears from CVC forms, at age 2;00.25 (see dos Santos 2007 for more detail). 2.3
Interim discussion
The processes observed above primarily refer to the location of the stressed syllable, although not entirely. As discussed in Section€2.1, the allophonic distribution of high vowels in Québec French final syllables is also dependent on whether these vowels are followed by voiced fricatives, an observation that suggests some interplay between phonology and phonetics. In a similar way, the different production patterns observed in Marilyn’s outputs can only be predicted if one considers the shape of the target (adult) word forms, an observation that points to conditioning at a more abstract, lexical level of representation (dos Santos 2007). Of course, the child’s immature motor control must be considered in any explanation of the patterns observed. It is also logical to think that the resolution of articulatory difficulties at the end of the period studied largely contributed to the disappearance of the processes observed. Learning speech articulations is similar in this respect to learning how to walk. It is however the prosodic conditioning observed in the developmental data that primarily concerns us here. Marilyn is in fact not a unique case; similar prosodic conditioning (i.e. distinctive behaviors of CVC versus CVCV forms) has been observed in the early productions of at least another French learner (child Clara in Rose 2000). Finally, formal issues about the syllabification of final consonants both in French phonology and in phonological development offer a series of analytical challenges, which we address in the next section. 3. Theoretical issues On the face of it, the patterns observed in Section€2 point to some basic syllabification and stress related processes in Québec French and in Marilyn’s phonology. However, hastily reaching this conclusion would elude important issues documented in the formal literature on the syllabification of word-final consonants in French as well as in phonological development across various languages.
Yvan Rose & Christophe dos Santos
3.1
Word-final consonants as onset of empty-headed syllables
Focusing first on French, several analyses proposed for syllabification in this language argue for the status of word-final consonants as onsets of empty-headed syllables (e.g.€Kaye, Lowenstamm & Vergnaud 1990; Charette 1991; Dell 1995). This hypothesis is supported by both distributional and phonetic arguments. Starting with distributional evidence, we first observe that the rhyme in French is maximally binary. For example, in Québec French, a dialect that maintains a lexical contrast of duration among non-high vowels (e.g.€faite [f7t] ~ fête [f7˜t]), with mid long vowels being more or less diphthongized across sub-dialects (e.g.€fête [f7˜t/fajt]), only short vowels can appear before (word-medial) codas, as exemplified in (5a). As opposed to this, we can see in (5b) that such long/diphthongized vowels are allowed before word-final consonants, similar to those lengthened high tense vowels before word-final voiced fricatives in (2b). In light of rhyme binarity, the examples in (2b) and (5b) strongly suggest syllabification of final consonants outside of the rhyme.
(5) Maximally binary rhymes: no branching nuclei before codas* a.╇ Before codas verte [v7t.t]/*[vajt.t] fierté [fj7t.te]/*[fjajt.te]
b.╇ Before word-final consonants vert [v7˜.t]/[vaj.t] fier [fj7˜.t]/[fjaj.t]
* For the sake of clarity and consistency, the syllabifications indicated in the examples here and below reflect the analysis maintained throughout the paper.
Another distributional argument comes from the fact that word-final consonant clusters can display falling, flat, or rising sonority profiles. Representative examples are provided in (6).
(6) Sonority profile of word-final consonant clusters a.╇ Falling sonority verte [v7t.t] carte [kat.t]
b.╇ Flat sonority veste [v7s.t] caste [kas.t]
c.╇ Rising sonority cadre [ka.dt] rectangle [t7k.t.:l]
An analysis of the falling sonority clusters in (6a) as branching codas would violate rhymal binarity. Furthermore, even though a branching coda analysis could accommodate the examples in (6a), any principled, sonority-based approach to syllabification would face problems with the flat-sonority clusters in (6b), and should preclude the attested rising sonority profiles in (6c) or else triconsonantal final clusters with various sonority profiles such as in words like
Stress domain effects in French phonology and phonological development
castre [kas.t]4 or arbre [at.bt]. As opposed to this, an analysis assuming rhymal binarity and allowing for the syllabification of word-final consonants as onsets of empty-headed syllables (with branching onsets also allowed in this position) readily accounts for all the attested data. Another distributional argument comes from the fact that except from [t], which is phonetically a voiced fricative but phonologically patterns as a liquid (similar to [l]), the truly phonologically voiced fricatives [v], [z] and [Š] do not appear in surface word-medial closed syllables unless these syllables are derived from the elision of an underlying lexical schwa (e.g.€ déguisement [de.:iz 6 .Öm˜]; mouvement [muv6 .Öm˜], where subscript [6] indicates the underlying location of the elided schwa). No such distributional restriction exists in word-final position (e.g.€words in (2b)) which, just like the word-medial onset position, displays a larger consonantal inventory, the onset being typically a better licenser of segmental contrasts than the coda (e.g.€ Itô 1986; Piggott 1999). Finally, another peculiarity of Québec French is that consonants that are part of a word-final consonant cluster tend to undergo reduction, sometimes with only the leftmost consonant of the final cluster being produced in the most extreme cases of erosion at the right edge (e.g.€ministre [minist]/[minist]/[minis]/*[mini]; quatre [kat]/[kat]/*[ka]). This fact is predicted by our analysis below, in which the consonant immediately following a stressed vowel phonologically interacts with this vowel to prosodically strengthen the stress domain and, as such, cannot be subject to deletion. Turning now to word-final consonants in child language, because we are interested in patterns from the acquisition of French, we are already in a position to argue that the child, attending to the distributional evidence of the language, should attain word-final onset syllabification during the course of development (e.g.€Goad & Rose 2004 for a similar view). This argument is further reinforced by the fact that several authors have argued in favor of an analysis of early consonants as onsets of empty-headed syllables across learners of different languages. First, final consonants generally emerge long before word-medial codas5 (this is the case for Marilyn, as documented in dos Santos 2007; see also Fikkert 1994 on Dutch; Freitas 1997 and Fikkert & Freitas 1997 on European Portuguese; Rose 2000 on Québec French; Barlow 2003 on disordered Spanish). Second, at the 4. The devoicing of the final liquid in this context matches that observed in branching onsets in other positions (e.g.€train [t�]). 5. Exceptions to this generalization come from languages such as Spanish or Japanese whose word-final consonants display clear coda profiles. In such languages, word-medial and final codas are typically acquired within the same developmental period (see Rose 2003 for further discussion).
Yvan Rose & Christophe dos Santos
phonetic level, these final consonants display onset-like behaviors, as they are often strongly aspirated or else released into epenthetic vocoids or nasals (e.g.€Goad & Brannen 2003). Finally, the same position has been argued for on structural grounds, as word-final onsets do not require a complex rhyme to license a consonantal position (Rose 2000), and may represent the default option in the face of typological evidence (e.g.€Piggott 1999; cf.€Rose 2003). 3.2
Representational implications
Following the arguments above, all final consonants in the data explored in Section€2 should be analyzed as onsets of empty-headed syllables and, as such, syllabified outside of the syllable where the stressed vowel is located. This conjecture is further complicated by another property of French, namely that it displays phrase-final stress (e.g.€Dell 1984; Walker 1984).6 Within prosodic phonology, the stress domain in French is often represented as a right-headed Foot constituent, as illustrated in (7).7 (7) Stress Foot in French Foot rg ... σ σ ]PPh When we combine domain-final stress with the analysis of final consonants as onsets of empty-headed syllables, one consequence is that these final consonants must be prosodified at a higher level, outside of the stress domain itself, as illustrated in (8).
6. Some dialects of French such as those generally spoken in southern France arguably display a different prosodic structure (e.g.€Durand 1976). This observation is however irrelevant to the current analysis, as both Québec French and Marilyn’s target dialect display phrase-final stress. 7. Following Prince (1985) and Kager (1993), we assume that the Foot is binary at the lexical level and that all syllables within the domain are incorporated into it through stray adjunction at the surface level (see also Charette 1991; Rose 2000; Goad & Buckley 2006 for further discussion). The issue of Foot binarity is however tangential to the current argument. What matters is the fact that the domain of stress assignment is right-headed.
Stress domain effects in French phonology and phonological development
(8) Prosodification of onsets of empty-headed syllables in French PhP
Foot σ
... C
V
σ C
Ø
Finally, apart from the high lax vowels derived through closed-syllable laxing (in (1)), which remain short in stressed syllables (except in emphatic contexts, where they remain laxed but are also stylistically lengthened), Québec French stressed vowels are up to twice as long as unstressed vowels (e.g.€ O’Shaughnessy 1981, Morasse 1995). The same observations apply to Marilyn’s data; measurements of 22 randomly-selected CVCV productions from the period studied show that her stressed vowels are systematically longer, with an average of 1.94 times the duration of the unstressed vowel. Final consonants are thus syllabified as onsets in both Québec French and Marilyn’s developing phonology. This conclusion however clashes with the behavior of final syllables as forming consistent domains for both closed-syllable laxing in Québec French, in (1b), and consonant preservation in Marilyn’s data, in (3). In the analysis proposed below, we argue that facts and theory can be reconciled if we take the position that while final consonants are truly onsets of empty-headed syllables phonologically, their behaviors with regard to closed-syllable laxing and in Marilyn’s data can be predicted from a general property of stress-final prosodic domains, whose head constituent typically undergoes strengthening, both crosslinguistically (e.g.€Hayes 1985, 1995) and, we argue, in phonological development. 3.3
Proposal
Our proposal is in line with a vast literature within the framework of Prosodic Phonology (e.g.€Selkirk 1984; McCarthy & Prince 1986). One of the tenets of Prosodic Phonology is that surface effects can be conditioned by deeper-level prosodic representations (see Harris 2004 for a summary of some of the relevant literature). As Jensen (2000) and Harris (2004) argue, neutralizing effects such as flapping or glottalization documented in the literature on consonant ambisyllabicity (e.g.€Kahn 1976, Giegerich 1992) can be explained from the fact that the
Yvan Rose & Christophe dos Santos
consonants that undergo neutralization are phonetically implemented in weak (dependent) prosodic positions. As opposed to this, in strong (head) positions, lexical contrasts are phonetically preserved and/or augmented (e.g.€through increased segmental duration or stronger aspiration of obstruent stops). While the literature on the topic primarily focuses on trochaic languages, that is, languages with strong-weak prosodic domains, similar effects are expected in languages with weak-strong prosodic domains. As argued for by, e.g.€Hayes (1985, 1995), languages of the latter type typically display heavy syllables in prosodic heads, even in languages which do not have vowel duration contrasts at the lexical level (Hayes 1995:€269). This phonological process, which we refer to as iambic strengthening, is defined in (9), where the syllable in the head of the prosodic domain is underlined. (9) Iambic strengthening (adapted from Hayes 1985) (σµσµ) → (σµσµµ) Still according to Hayes, across languages, phonological iambic strengthening can manifest itself in many ways at the phonetic level, through processes such as vowel lengthening, gemination of the following onset (with the first half of the geminate incorporated in the coda of the stressed syllable) or, alternatively, weakening of the preceding, unstressed syllable, which effectively provides relative prominence to the neighboring stressed syllable. Explicit to this analysis is a formal distinction between underlying and surface levels of representation. This distinction finds independent motivation in the formal literature on rhythmic and stress-related segmental phenomena (see, e.g.€Prince 1985 and Kager 1993 for further discussion). While iambic strengthening is typically discussed in the context of languages with alternating stress patterns, we hypothesize that it also applies to non-alternating stress systems such as that of French. We further hypothesize that all of the patterns discussed in Section€2 can be accounted for following the same logic: while word-final syllables are open in French (including Marilyn’s) lexical representations, these syllables undergo phonological strengthening, which can take one of two different forms on the surface: vowel lengthening or final consonant incorporation into the stressed syllable, the choice of which is driven by independent, principled considerations. 4. Analysis 4.1
Québec French
In Québec French, iambic strengthening yields different exponents at the surface level, depending on the segmental context. In vowel-final words, the stressed
Stress domain effects in French phonology and phonological development
vowel undergoes lengthening (as in (1a)), which effectively makes the syllable heavy, without any further complication. When stressed vowels are followed by consonants, one of three possibilities can occur. First, if the consonant is a true coda, itself followed by the onset of an empty-headed syllable, for example in words like caste [kas.t(6)] or perdre [p7t.dt(6)], then the stressed syllable is already heavy and does not undergo any further modification (except of course from laxing among high vowels; e.g.€piste [pis.t(6)]). Second, if the consonant is a voiced fricative syllabified in the onset of an empty-headed syllable, the vowel undergoes lengthening (and/or concomitant diphthongization, as in (5)), for example in pige [pi˜.Š], cave [ka˜.v] or faire [f7˜.t]/[faj.t]. This is illustrated in (10a). In this context, the final consonant remains outside of the stressed syllable. Third, if the consonant following the stressed vowel is not a voiced fricative, this consonant is incorporated into the coda of the stressed syllable, thereby yielding the context for closed-syllable laxing, and its manifestation, as illustrated in (10b). (10) Iambic strengthening before voiced versus voiceless obstruents a. buse: PhP PhP
Foot
Foot σ
σ
σ
σ
8 b b. butte:
y
z
Ø
b
y
PhP
PhP
Foot
Foot σ
σ
z
˜
Ø
σ
8 b
y
t
Ø
b
y
t
At this point in the analysis, one may wonder why iambic strengthening takes two different forms in the context of voiced fricatives versus other word-final consonants. In a nutshell, why are voiced fricatives so particular? While the full explanation to their patterning would require a consideration of the socio-historical
Yvan Rose & Christophe dos Santos
evolution of this dialect of French, we claim that it has foundations in acoustic properties of post-vocalic obstruents, a context where laryngeal contrasts are often weakened or neutralized across languages. As pointed out by Blevins (2004) for example, in languages where obstruents codas are allowed, voiceless codas are often favored over voiced codas. This typological tendency actually makes sense from the perspectives of both perceptual and articulatory phonetics. Indeed, post-vocalic voicing contrasts among obstruents are relatively difficult to perceive (e.g.€Steriade 2001). Also, when maintained in production, the implementation of such contrasts is often enhanced through a lengthening of the preceding vowel (e.g.€Borden, Harris & Lawrence 2003). We suggest that, in the context of iambic strengthening, vowel lengthening has been phonologized in Québec French as a means to preserve and enhance voicing contrasts among fricatives. Our explanation thus puts the burden on the phonetic implementation of a phonological contrast in post-vocalic positions, as opposed to particular representational aspects of voiced fricatives or their voiceless counterparts. Because vowel lengthening before voiced fricatives yields a heavy rhyme, no further iambic strengthening is needed in this context. Vowel lengthening thus effectively neutralizes closed-syllable laxing since high vowels followed by a voiced fricative remain in a phonologically open syllable. 4.2
Marilyn
Turning now to Marilyn’s data, we argue that phonological iambic strengthening also applies to her outputs, whose surface manifestation enables preservation of word-final consonants, as illustrated in (11). (11) Marilyn’s CVC forms: coda incorporation and harmony passe: PhP PhP
Foot
Foot σ
σ
σ
8 pi
a
sj
Ø
pi
a
ti (i = [–cont.]; j = [+cont.])
We further argue that word-final consonants undergo manner harmony because Marilyn can only maintain one continuancy feature per word in her outputs. Phonologically, this feature must be licensed by the prosodic head. This analysis
Stress domain effects in French phonology and phonological development
follows proposals by, e.g.€Rose (2000) and Goad (2001), who argue that harmony can emerge in child language as a surface exponent of prosodic licensing. This analysis is further supported by the contrast observed in Marilyn’s CVCV forms in (4), in which preservation of the first syllable is dependent on its onset being manner-harmonic with the onset of the stressed syllable, as illustrated in (12a) versus (12b). (12) Marilyn’s CVCV forms: preservation conditional on licensing a. biquet: PhP PhP
Foot
Foot σ
σ bi
i
b. château:
σ ∫j
a
ε
ki
ti
bi
e
PhP
PhP
Foot
Foot
σ
σ
o
σ
σ
ti
ki
ε
o
(i = [–cont.]; j = [+cont.])
This analysis also implies that harmony was bound within the head of the Foot and could not dynamically extend outside of the head of the Foot to yield preservation. Only CVCV forms that were manner-harmonized in the input could display full preservation of both syllables in the output during this developmental stage. 5. Discussion The analysis proposed above enables a unified account of the various patterns observed in both Québec French and Marilyn’s productions. Starting with Québec French, this analysis first reconciles the data on closed-syllable laxing (and
Yvan Rose & Christophe dos Santos
absence thereof) with the phonological analysis of word-final consonants as onsets of empty-headed syllables, thereby offering a solution to the long-standing controversy (at least since Charette 1991) as to why closed-syllable laxing should occur at all in the context of lexically open syllables. Under the current proposal, both closed-syllable laxing and its alter-ego, stressed vowel lengthening before word-final voiced fricatives, are triggered in this context by iambic strengthening, a general rule of phonological enhancement of the syllable located in the head of the stress domain. Our proposal also offers an explanation for the observation that laxing does not occur in the context of voiced ‘codas’ that are derived through phonetic schwa elision but that are onsets at the lexical level in French. Since surface ‘codas’ before elided schwas are not phonological codas, no closed-syllable laxing can occur in this context. Such observations would be extremely difficult to capture under approaches focusing solely on the shape of output forms. In a similar way, the current analysis offers a consistent account of Marilyn’s production patterns. Through iambic strengthening, onsets of empty-headed syllables are incorporated into the stressed syllable. This enables their preservation and manner harmony, analyzed here as a surface manifestation of prosodic licensing. In CVCV forms, however, word-initial consonants cannot be incorporated into the stressed syllable. Preservation in this context thus depends on a pre-condition, that of manner harmony between the two consonants in the lexical representation. Without a formal distinction between surface representations and deeper, lexical representations, it would be very difficult to explain such domain-related phenomena in a principled way while keeping with the larger literature on theoretical phonology and phonological development. References Barlow, Jessica. 2003. “Asymmetries in the Acquisition of Consonant Clusters in Spanish”. Canadian Journal of Linguistics 48:179–210. Blevins, Juliette. 2004. Evolutionary Phonology: The Emergence of Sound Patterns. Cambridge: Cambridge University Press. Borden, Gloria, Katherine Harris & Lawrence Raphael. 2003. Speech Science Primer. Philadelphia: Lippincott Williams & Wilkins. Charette, Monik. 1991. Conditions on Phonological Government. Cambridge, Mass.: Cambridge University Press. Chomsky, Noam & Morris Halle. 1968. The Sound Pattern of English. New York: Harper & Row. Déchaîne, Rose-Marie. 1991. “Stress in Québécois: Evidence from High Vowels”. Papers from the Regional Meetings, Chicago Linguistic Society 27:107–118.
Stress domain effects in French phonology and phonological development Dell, François. 1984. “L’accentuation des phrases en français”. Forme sonore du langage ed. by François Dell, Daniel Hirst & Jean-Roger Vergnaud. Paris: Hermann. 65–122. —–. 1995. “Consonant Clusters and Phonological Syllables in French”. Lingua 95:5–26. dos Santos, Christophe. 2007. Développement phonologique en français langue maternelle: Une étude de cas. Ph.D. dissertation. Université Lumière Lyon 2. Durand, Jacques. 1976. “Generative Phonology, Dependency Phonology and Southern French”. Lingua e Stile XI:3–23. Fikkert, Paula. 1994. On the Acquisition of Prosodic Structure. The Hague: Holland Academic Graphics. Fikkert, Paula & Maria João Freitas. 1997. “Acquisition of Syllable Structure Constraints: Evidence from Dutch and Portuguese”. Proceedings of GALA 1997 ed. by Antonella Sorace, Caroline Heycock & Richard Shillcock. Edinburgh: Edinburgh University Press. 217–222. Freitas, Maria João. 1997. Aquisição da Estrutura Silábica do Português Europeu. Ph.D. Thesis. University of Lisbon. Giegerich, Heinz. 1992. English Phonology: An Introduction. Cambridge: Cambridge University Press. Goad, Heather. 2001. “Assimilation Phenomena and Initial Constraint Ranking in Early Grammars”. Proceedings of the 25th Annual Boston University Conference on Language Development ed. by Anna€H.-J. Do, Laura Domínguez & Aimee Johansen. Somerville, Mass.: Cascadilla Press. 307–318. Goad, Heather & Kathleen Brannen. 2003. “Phonetic Evidence for Phonological Structure in Syllabification”. The Phonological Spectrum, Vol. 2 ed. by Jeroen van€de€Weijer, Vincent van Heuven & Harry van der Hulst. Amsterdam: John Benjamins. 3–30. Goad, Heather & Meaghen Buckley. 2006. “Prosodic Structure in Child French: Evidence for the Foot”. Catalan Journal of Linguistics 5:109–142. Goad, Heather & Yvan Rose. 2004. “Input Elaboration, Head Faithfulness and Evidence for Representation in the Acquisition of Left-edge Clusters in West Germanic”. Constraints in Phonological Acquisition ed. by René Kager, Joe Pater & Wim Zonneveld. Cambridge: Cambridge University Press. 109–157. Harris, John. 2004. “Release the Captive Coda: The Foot as a Domain of Phonetic Interpretation”. Phonetic interpretation: Papers in Laboratory Phonology 6 ed. by John Local, Richard Ogden & Rosalind Temple. Cambridge: Cambridge University Press. 103–129. Hayes, Bruce. 1985. “Iambic and Trochaic Rhythm in Stress Rules”. Proceedings of the Berkeley Linguistics Society 11: 429–446. —–. 1995. Metrical Stress Theory: Principles and Case Studies. Chicago: University of Chicago Press. Itô, Junko. 1986. Syllable Theory in Prosodic Phonology. Ph.D. dissertation. University of Massachusetts, Amherst. Jensen, John. 2000. “Against Ambisyllabicity”. Phonology 17:187–286. Kahn, Daniel. 1976. Syllable-based Generalizations in English Phonology. Ph.D. dissertation. MIT. Kager, René. 1993. “Alternatives to the Iambic-Trochaic Law”. Natural Language and Linguistic Theory 11:381–432. Kaye, Jonathan, Jean Lowenstamm & Jean-Roger Vergnaud. 1990. “Constituent Structure and Government Phonology”. Phonology 7:193–231. McCarthy, John & Alan Prince. 1986. Prosodic Morphology. Ms. University of Massachusetts, Amherst and Rutgers University.
Yvan Rose & Christophe dos Santos Morasse, Hélène. 1995. Variations intrinsèques et co-intrinsèques de durée vocalique en français québécois. M.A. Thesis. Université Laval. O’Shaughnessy, Douglas. 1981. “A Study of French Vowel and Consonant Durations”. Journal of Phonetics 9:385–406. Piggott, Glyne. 1999. “At the Right Edge of Words”. The Linguistic Review 16:143–185. Poliquin, Gabriel. 2007. Canadian French Vowel Harmony. Ph.D. dissertation. Harvard University. Prince, Alan. 1985. “Improving Tree Theory”. Proceedings of the Berkeley Linguistics Society 11: 471–490. Prince, Alan & Paul Smolensky. 1993. Optimality Theory: Constraint Interaction in Generative Grammar. Ms. Rutgers University and University of Colorado, Boulder. Rose, Yvan. 2000. Headedness and Prosodic Licensing in the L1 Acquisition of Phonology. Ph.D. dissertation. McGill University. —–. 2003. “Place Specification and Segmental Distribution in the Acquisition of Word-final Consonant Syllabification”. Canadian Journal of Linguistics 48:409–435. —–. 2009. “Internal and External Influences on Child Language Productions”. Approaches to Phonological Complexity ed. by Egidio Marsico & François Pellegrino. Berlin: Mouton de Gruyter. 329–352. Selkirk, Elizabeth. 1984. “On the Major Class Features and Syllable Theory”. Language and Sound Structure ed. by Mark Aronoff & Richard Oehrle. Cambridge, Mass.: MIT Press. 107–136. Steriade, Donca. 1999. “Phonetics in Phonology: The Case of Laryngeal Neutralization”. UCLA Working Papers in Linguistics 3:25–246. —–. 2001. “Directional Asymmetries in Place Assimilation: A Perceptual Account”. The Role of Speech Perception in Phonology ed. by Elizabeth Hume & Keith Johnson. New York: Academic Press. 219–250. Walker, Douglas. 1984. The Pronunciation of Canadian French. Ottawa: University of Ottawa Press.
part 3
Syntax and morphophonology
Syntactic realizations of plural in Romance and Germanic nominalizations Artemis Alexiadou, Gianina Iordăchioaia & Elena Soare* This paper offers a syntactic account for the various ways in which plural is expressed within Argument-Supporting Nominals (ASNs) in Romance and Germanic. We first show, starting from Romanian data, that pluralization is connected to aspectual properties like (a)telicity and (un)boundedness. These properties may be realized under the syntactic category of verbal Aspect, that of nominal Number, or encoded by a [±count] features on the nominal category Classifier. We then support our theory with a comparison between the Romanian Infinitive/Supine and Spanish Nominal Infinitives (SNIs). The latter are fully nominal, but lack plural and we relate this to a [–count] feature in their Classifier. The comparison with English also shows the relevance of inner aspect to nominalizations. The picture we get allows a better understanding of morphological and syntactic processes at work in nominalizations and of their nominal and verbal properties.
1. Introduction Grimshaw (1990) considers plural marking an important parameter distinguishing between ASNs and Referential Nominals (RNs): the former do not pluralize, while the latter do, cf.€examples in (1). (1) a. The assignments were long. b. *The assignments of the problems took a long time.
(RN) (ASN)
* We thank the editors of this volume, the reviewers and the participants of the 38th LSRL at the University of Urbana-Champaign for comments and suggestions. Alexiadou’s and Iordăchioaia’s research was supported by a DFG grant to the project B1: The formation and interpretation of derived nominals, as part of the Collaborative Research Center 732 Incremental Specification in Context at the University of Stuttgart.
Artemis Alexiadou, Gianina Iordăchioaia & Elena Soare
Roodenburg (2006) provides counterexamples to this generalization and claims that there is language variation with respect to plural marking in ASNs: in Romance (French and Italian) they may pluralize, whereas in Germanic languages like English, they cannot (see the contrast in (2)):
(2) a. De fréquentes destructions de quartiers populaires pour y ériger des tours staliniennes. ‘Frequent destructions of popular areas in order to construct Stalinist towers.’ b. *the destructions of the city by the soldiers The idea of a language parameter with respect to plural ASNs is contradicted by Romanian which displays both patterns in (2) with infinitive and supine ASNs (cf.€Iordăchioaia & Soare 2008). As shown in (3), the former admits plural, and the latter doesn’t. (3)
demolările/*demolaturile frecvente ale cartierelor vechi de către demolish-Inf-Pl/demolish-Sup-Pl frequent-Pl of quarters-Gen old by comunişti communists “The frequent demolitions of old quarters by the communists.”
On the basis of Romanian data, Iordăchioaia & Soare (2008) demonstrate that the ASNs which pluralize express bounded events and project Num(ber)P, while the ASNs which do not pluralize express unbounded events, and project Asp(ect) P[imperfective]. Here we compare ASNs in two Romance languages (Romanian and Spanish) and two Germanic languages (English and German). We argue that the pluralization of ASNs is not subject to language variation, since ASNs generally pluralize across languages. Pluralization is actually related to aspectual properties such as (a)telicity, (im)perfectivity and/or (un)boundedness. We further provide a syntactic account for the cases where Grismhaw’s generalization holds, as well as for the cases where it does not. More specifically, we show that plurality may be syntactically realized under the nominal category Num (Romanian infinitive) or under the verbal category Asp (Romanian supine and English verbal gerund), depending on the morphological and semantic properties of the ASN. An investigation of the aspectual properties of the English nominal gerund, in comparison to the Spanish nominal infinitive indicates that unbounded aspectual information can also block plural marking within a nominal structure without a verbal Asp projection. We propose to express this information under a feature [±count] on the nominal projection of Classifier (Picallo 2006) and thus
Syntactic realizations of plural in Romance and Germanic nominalizations
account for the variation with respect to plural marking in English nominal gerunds and German -ung ASNs. The paper is structured as follows: in Section€2, we summarize the Romanian data, documenting Number vs. Aspect projections in infinitive vs. supine nominals, respectively. In Sections 3 and 4, we look at German and English data and bring further confirmation for this line of analysis. The English nominal gerund also indicates the relevance of inner aspect (i.e. Aktionsart) in ASNs. Section€5 provides arguments from Spanish Nominalized Infinitives that show that inner aspect should be integrated in the nominal structure and not in a verbal one. We conclude in Section€6. 2. Number vs. Aspect in Romanian ASNs 2.1
Nominalization patterns
Infinitive and supine nominals are the most productive deverbal nouns in Romanian and they derive from the infinitive and the past participle stem, respectively. Examples in (4) show that the infinitive nominal admits plural marking, while the supine disallows it. (4) a.
The Infinitive ASN: a cînta: cînta–r –e/cîntă–r–i to sing: sing-Inf-F.Sg/sing-Inf-Pl a conduce: conduce–r–e/conduce–r–i to drive: drive –Inf-F.Sg/drive-Inf-Pl
b.
The Supine ASN: a cînta: cînta–t/#cînta–t–uri to sing: sing-Sup/sing-Sup-Pl a conduce: condu-s/#condu-s-uri to drive: drive-Sup/drive-Sup-Pl
Soare (2007) argues that the infinitive nominal involves a true nominalizer, -re, which always forms the nominal infinitive, while the supine is a syntactic nominalization based on a verbal stem marked with the verbal suffix -t/s which also builds the past participle. Here, this idea receives further empirical support.
Artemis Alexiadou, Gianina Iordăchioaia & Elena Soare
2.2
Plural marking and determiner selection
We just saw that the infinitive displays morphological plural; this holds both for the ASN infinitive and for the RN infinitive (5). Moreover, it combines with discrete quantifiers, while supine takes mass quantifiers (6): (5) a. interpretările acestui actor sînt memorabile interpret-Inf-Pl this-Gen actor are memorable
(RN)
b.
interpretările acestui rol de către diverşi actori i-au interpret-Inf-Pl this role-Gen by different actors Cl-have schimbat stilul (ASN) changed style
(6) a.
prea multe spălări/o spălare a(le) rufelor distrug(e) too many wash-Inf-Pl/one wash-Inf of clothes-Gen destroy(s) ţesătura fabric-the
b. prea mult/*un spălat al rufelor distruge ţesătura too much/one wash-Sup of clothes-Gen destroys fabric-the This contrast clearly shows that the infinitive resembles count nouns which project NumberP, whereas the supine resembles mass nouns which lack Number. 2.3
Nominal properties
We now turn to an investigation of the nominal properties gender and case declension in the infinitive and the supine Romanian nominals. A nominal structure involves the projection of a Classifier, and consequently the projection of nominal functional categories. We follow Picallo (2006) who argues that gender acts like a Class(ifier) and feeds Number. The Classifier projection hosts the class/ declension information of a noun; it triggers the projection of the specific nominal structure. Turning to Romanian ASNs, we find that infinitive ASNs are specified as [feminine] (the endings -e/-i in (4a) cf.€(7)), so they carry feminine gender features: (7) a. cîntar -e/cîntăr -i sing-Inf -F.Sg/sing-Inf –Pl b. floar -e/flor -i flower -F.Sg/flower -Pl Supine ASNs are traditionally considered neuter. If a plural is reconstructed by analogy with other participial nouns (since the supine event nominals do not
Syntactic realizations of plural in Romance and Germanic nominalizations
have one), they take the null ending for singular (like masculine nouns) and -uri in plural (like some feminine nouns). As Iordăchioaia & Soare (2008) argue, -uri in supine ASNs is a default gender, the neuter specification of gender is not the effect of neuter gender features, but a default gender marking. Further support comes from the observation in Iordăchioaia & Soare (2009) that the infinitive ASN successfully establishes anaphoric relations with the feminine demonstrative aceasta; the supine rejects the masculine/neuter syncretic form acesta and can only be referred to by the genderless form asta, the common anaphor for CPs in Romanian: (8) a. Că Ion a venit, asta ştiu. ‘That John came, I know it.’ b.
Am vorbit despre interpretarea rolului Hamlet în general. We spoke about the interpretation-Inf of Hamlet in general. Se pare ca aceasta/??asta îi consacră indubitabil pe actorii tineri. ‘Apparently, this-Fem/??it undoubtedly validates the young actors.’
c.
Am vorbit despre interpretatul rolului Hamlet în general. ‘We spoke about the interpretation-Sup of Hamlet in general. Se pare ca *acesta/asta îi atrage pe toţi actorii tineri. Apparently *this-M-N/it attracts all the young actors.’
The above suggests that the supine, as opposed to the infinitive, does not have a Classifier. In fact, the supine also has a defective case declension, since unlike the infinitive ASN it cannot realize the dative-genitive, but only the nominative-accusative (see Iordăchioaia & Soare (2009) for details). Alongside the nominal properties, adjectival modification is allowed with the infinitive and disallowed with the supine, as shown in (9). Note that despite the homonymy between the neuter form adjective and the adverb, in Romanian only adjectives can appear prenominally, so in (9b) we are dealing with adverbs and not with adjectives. (9) a. studierea atentă/atenta studiere a detaliilor study-Inf-the careful-Adj/careful-Adj study-Inf-the details-Gen b. studiatul atent/*atentul studiat al detaliilor study-Inf-the careful-Adj.Adv/careful-Adj study-Inf-the details-Gen We can therefore conclude that the infinitive ASN displays nominal properties that are missing in the supine. It projects ClassP and NumP. There is no evidence for these projections in the supine ASN. This structural difference straightforwardly explains the empirical facts observed above.
Artemis Alexiadou, Gianina Iordăchioaia & Elena Soare
2.4
Aspectual properties
The aspectual properties of ASNs are also related to event plurality. Investigating the aspectual differences between the two Romanian nominalizations at study here, Cornilescu (2001) shows that the infinitive ASN is telic, whereas the supine ASN is atelic. She supports this claim with three main arguments which we briefly note, referring the reader to Cornilescu (2001) for details. First, infinitive ASNs obligatorily project the theme argument which identifies the culmination of the event and thus indicates telicity. By contrast, the supine may project its external argument in the absence of the theme: (10) a. *citirea lui Ion read-Inf-the John-Gen
(only RN)
b. cititul lui Ion read-Sup-the John-Gen Second, only infinitives can develop R-readings, supines cannot since they do not have a resulting state. In (11), we take a test from Cornilescu (2001): locative prepositions modifying R-nominals contain the preposition de (‘of’); this kind of PP cannot modify a supine nominal: (11) a. cântatul lui Ion (*de) la baie sing-Sup-the John-Gen of in bathroom b. interpretarea de la Paris a operei Oedip perform-Inf-the of in Paris of opera-Gen Oedip Third, only the supine is compatible with unergative verbs known to express activities (cf.€some examples from Cornilescu’s list below)1. (12)
Unergative verb a locui (to live) a munci (to work) a rîde (to laugh)
Infinitive *locuire *muncire *rîdere
Supine locuit muncit rîs
Telicity and atelicity can be reframed in terms of [±b](oundedness) (Jackendoff 1991), since this cover term equally characterizes the nominal and the verbal domain. Telic and perfective aspect as well as singular count nouns are [+b]; atelic and imperfective aspect, plural and mass nouns are [–b]. Thus the infinitive ASN is [+b], the supine is [–b].
1.
Telic unergatives and unergative states will be addressed in the next section.
Syntactic realizations of plural in Romance and Germanic nominalizations
2.5
Aspect shift
The supine formation maps [+b] events into [–b] events and thus introduces aspect shift. This is supported by the interaction with verbal roots, as illustrated below. With achievements, which are [+b], the supine is ok and expresses habituality as shown in (13a). The supine is also grammatical with telic unergatives like verbs of birthing (Harley 2005). With the telic ‘in’-PP in (13b) the construction refers to a series of calving events each taking several hours, so the PP modifies the inner ([+b]) aspect of the verb. With the atelic ‘for’-PP the construction refers to several events of calving taking place over a time span of several hours, so the PP modifies the outer ([–b]) aspect. (13) a.
sositul lui Ion cu întîrziere la toate arrive-Inf-the John-Gen with delay at all întîlnirile importante meetings important `John’s (habit of) arriving late at all important meetings’
b. fătatul vacilor în cîteva ore/timp de cîteva ore calve-Sup-the cows-Gen in several hours/for several hours€ States are [–b], so the supine usually requires a bounding function like until, to suggest a plurality of bounded state eventualities and thus habituality. (14)
statul lui Ion la Maria/dormitul lui Ion *(pînă stay-Sup-the John-Gen at Mary’s/sleep-Sup-the John-Gen until după-amiaza tîrziu) afternoon late `John’s (habit of) staying at Mary’s/sleeping until late afternoon’
The supine is excluded with i-level predicates, because they are impossible to locate in space and time (Kratzer 1995), so they cannot become [+b]: (15)
*ştiutul/*cunoscutul limbilor străine/*descinsul know-Sup-the languages-Gen foreign/descend-Sup-the omului din maimuţă man-Gen from monkey
Accomplishments, punctual events and semelfactives pattern with achievements as [+b], and activities with states as [–b] (see Iordăchioaia & Soare 2008). The above patterns are analyzed by Iordăchioaia & Soare (2008, 2009) in terms of a pluractional operator (see Lasersohn 1995, Van Geenhoven 2004, Laca 2006) which pluralizes the individual event. The Asp projection in the supine nominal thus hosts the pluractional operator.
Artemis Alexiadou, Gianina Iordăchioaia & Elena Soare
2.6
The functional structure of Romanian ASNs
Drawing on the facts presented so far, we can now provide a structural configuration for the two nominal patterns documented here. There is first a nominal pattern, realized in the infinitive, as in (16). In this structure plural is realized morphologically under NumP. A verbal pattern characterizes the supine nominal as in (17): semantic plural is syntactically realized under AspP. (16)
DP 3 D
NumP 3 Num
ClassP 3 Class
N/nP 3
N/n -a [sing] -e[fem] -r (17)
VP citi
DP 3 D
AspP 3
Asp -(u)l PO [-b]
VP citit
The projection of AspP is also motivated in the case of the supine ASNs by the availability of aspectual adverbs like constantly (Cinque 1999); such adverbs are unavailable with infinitive ASNs: (18) a. cititul (constant) al ziarelor (constant) read-Sup-the constantly newspapers-Gen constantly ‘constantly reading newspapers’ b. omiterea (*constant) a unor informaţii (??constant) omit-Inf-the constantly of some information constantly
Syntactic realizations of plural in Romance and Germanic nominalizations
3. Germanic ASNs 3.1
Pluralization of ASNs in German
In German, –ung nominals are the only ones that can pluralize, infinitival ASNs do not (Ehrich 2002). Unlike Romanian infinitives, -ung nominals project either the theme or the agent of an activity verb (19a), so they are not necessarily telic. In this case, a plural under the R-reading is possible (19b). (19) a. die Beobachtung des Verdächtigen/der Polizei the observe-ung the-Gen suspect/the-Gen police b. die Beobachtungen der Polizei the observe-Ung-Pl the-Gen police
(R-plural)
Roodenburg (2006) claims that German ASNs do not exhibit plural, since the plural form does not maintain the argument structure: (20) a. die Beobachtung von VögelnTh the observing of birds b. die Beobachtungen von Vögeln*Th the observings of birds
(Alexiadou 2001)
Note that this situation is due to the aspectual ambiguity of -ung nominals – the examples above are atelic and atelic ASNs do not pluralize; the plural form is an R-nominal. However, with telic verbs, -ung nominals project the theme obligatorily, and plural ASNs can occur like in Romanian and English (see (21b)): (21) a. die Tötung des Feindes/des Verbrechers(Th/*Ag) the kill-ung the-Gen enemy-Gen/the-Gen criminal-Gen b. 3.2
die gezielten Tötungen der politischen Führer durch die the targeted kill-ung-Pl the-Gen political leaders via the Armee army
English verbal vs. nominal gerund
The two instances of the English gerund, the nominal and the verbal one, are distinguished by argument realization and the presence vs. absence of the determiner, as well as the adverbial vs. adjectival modification. The theme argument receives accusative case with the verbal gerund, and is realized as an of-PP in the nominal gerund (Alexiadou 2001):
Artemis Alexiadou, Gianina Iordăchioaia & Elena Soare
(22) a. Emma’s reading the poem. b. Emma’s reading of the poem. The subject DP cannot be replaced by any determiner in the case of the verbal gerund, while this is possible for the nominal gerund: (23) a. *That/*the criticising the book annoyed us. b. The jumping of the cows annoyed us. Verbal gerunds license adverbial modifiers and disallow adjectival modifiers (24a, 25b). Nominal gerunds disallow adverbial modifiers and license adjectival modifiers (24b, 25a): Pat disapproved of me/my quietly leaving the room before anyone noticed. b. *The carefully restoring of the painting took six months. (24) a.
(25) a. his prompt answering of the question b. *his prompt answering the question Under the standard assumption that adverbs are (at least) VP modifiers, while adjectives are noun modifiers, this contrast suggests that the verbal gerund contains a verbal internal structure (Abney 1987, Kratzer 1994, Borer 1993 and others), while the nominal gerund has a nominal internal structure. 3.3
Functional structure for English gerunds
As far as English nominal gerunds are concerned, we assume with Alexiadou (2001) that they contain NumberP, as opposed to verbal gerunds which are AspPs (Borer 2001, Alexiadou 2001, 2005) and lack Number (Alexiadou 2001, 2005). Contrary to what Roodenburg (2006) claimed, Alexiadou, Iordăchioaia & Soare (2010) show that some English and German telic ASNs pluralize, too, as (26) shows. In English, this is possible especially when no other deverbal affix is available for a telic verb. (26) a. the killing of the journalist/journalists b. the repeated killings of unarmed civilians DP We thus adopt the two structures below for the English gerund: 3 (27) a. D AspP 3 Asp ing
VP 4
verbal gerund
Syntactic realizations of plural in Romance and Germanic nominalizations
b.
nominal gerund
DP 3
D
NumP 3 Num
N/nP 3 N/n ing
VP 4
4. Inner aspect in English nominal gerunds Coming back to Romanian and English ASNs, on the view expressed above, verbal gerunds have a structure similar to that of Romanian supines, while nominal gerunds resemble Romanian infinitives. One would expect the two patterns to show a similar behavior in connection with Aktionsart. But as we will see below, the nominal gerund has a different inner aspect sensitivity in comparison to that of the Romanian infinitive. While the Romanian infinitive ASN disallows atelic roots (see (12)), Borer (2005:€239ff.) notes that the affix used for the formation of nominal gerunds – the nominalizer -ing – is ungrammatical with achievements (28) but OK with nonculminating events (activities and semelfactives) in (29). (28) a. *the arriving of the train b. *the erupting of the Vesuvius c. *the exploding of the balloon (29) a. the sinking of the ships b. the falling of the stock prices c. the jumping of the cows Other affixes, the verbal gerund as well as the progressive do accept achievements: (30) a. the arrival of the train b. The train arriving at 5 pm is unlikely. c. The train is arriving.
other affix verbal gerund progressive
An explanation of these facts would be that the affix for the progressive as well as the verbal gerund are oblivious to the Aktionsart of the predicate, they are equally fine with telics, atelics, and at least some statives. This means that the progressive and the verbal gerund operate on what has been labelled outer aspect (Verkuyl 1993). These affixes are located in AspP in (27a) and realize [–boundedness].
Artemis Alexiadou, Gianina Iordăchioaia & Elena Soare
The affix of the nominal gerund, however, is sensitive to the properties of the predicate and hence interacts with the inner aspect (following Verkuyl 1993, Borer 2005). We turn now to the locus of this sensitivity in the functional structure. 4.1
The relation between number and aspect
The layers relevant for introducing (un)boundedness (see also Engelhardt 2000, Rijkhoff 2002) are the morphological plural, realized as NumP, and outer, grammatical aspect (AspP). The English nominal gerund indicates that Aktionsart (inner aspect) is another such factor that should be represented in the syntax. Operations at the level of outer aspect do not alter the event type of the predicate, but do interfere with (un)boundedness: this is the case with the progressive, the verbal gerund, and the Romanian supine in (30b,c) (13–14). Operations at the level of inner aspect do affect the event type of the predicate: that happens in the nominal gerund (28–29) and the Romanian infinitive (12). Number indicates a nominal structure with a Classifier projection instead of inner aspect (see Fassi Fehri 2005 for the correlation between inner aspect and Classifier). The Classifier is unmarked for Number, it just indicates nominal character, and is by default singular (i.e. non-plural, see also Kratzer 2005), plural is added under NumP. 4.2
Do we need an inner aspect projection?
To begin with, if inner aspect is derived from the properties of the VP in nominalization, ideally we do not need such a projection in the syntax. Moreover, neither –re (in Romanian infinitives) nor –ing of (in English nominal gerunds) introduce anything about the aspect: they are just sensitive to a particular aspect: telic and atelic, respectively. They are not operators on outer aspect, so if there is aspect it is the inner aspect contributed by the VP. The nominalizer -re can also participate in atelic constructions with the argument expressed by de ‘of’ + bare plural; these constructions are atelic and cannot pluralize: (31) a. scrierea de scrisori write-Inf-the of letters b. *scrierile de scrisori write-Inf-the-Pl of letters
Syntactic realizations of plural in Romance and Germanic nominalizations
Does this mean that the structure of the Romanian infinitive ASN (with –re) should be ambiguous between projecting a Classifier (with Number) and an inner Aspect (without Number)? Here are some counterarguments to this idea. The atelicity effect of (31) does not have to do with the suffix –re but with the theme which is unquantized (a bare plural). Furthermore, the suffix –re maintains the same nominal properties (gender features, case inflection) which are hosted by the Classifier. In conclusion, the ClassifierP is still motivated for (31). The difference in telicity might motivate the difference between Romanian –re and English –ing, with the former projecting Classifier and the latter simply projecting VP inner Aspect. In English, we cannot test the Classifier properties of the suffix: there is no gender/case marking, although a non-overt Classifier is usually assumed (see Kratzer 2005 among others). However, a similar ASN in Spanish allows us to test nominal/Classifier properties better. 5. The Spanish nominal infinitive 5.1
Verbal vs. nominal infinitive
Miguel 1996 distinguishes between two patterns of infinitive ASNs in Spanish which primarily differ in the way they realize their external argument: with nominative case, or with de ‘of’:2 (32) a. el murmurar la gente the murmur-Inf the people-Nom ‘the murmuring of the people’
verbal infinitive
b. el murmurar de las fuentes the murmur-Inf of the fountains ‘the murmuring of the fountains’
nominal infinitive
2. Both the verbal (i) and the nominal infinitive (ii) assign accusative case to their internal argument. For the nominal one, the event should remain atelic though (see also (33) and Fábregas & Varela 2006). Otherwise, speakers prefer bare noun internal arguments with the nominal infinitive (Fábregas, p.c.): i. el comprar una casa Juan the buy-Inf a house-Acc Juan-Nom ii. el (constante) cuidar (los) niños de Juan the (constant) look-after-Inf (the) children-Acc of Juan€
Artemis Alexiadou, Gianina Iordăchioaia & Elena Soare
Miguel 1996, Demonte & Varela 1997, Fábregas & Varela 2006 have shown that Spanish Nominal Infinitives (SNIs) carry atelic/unbounded aspect, since they systematically reject telic roots. SNIs are ungrammatical with achievements and accomplishments: (33) a. *El llegar tardío de Juan nos preocupó a todos. the arrive-Inf late of Juan us worried to all b. *El comprar una casa de Juan nos alegró. the buy-Inf a house of Juan us made happy We concentrate on SNIs in (32b), since verbal infinitives in (32a) do not show any sensitivity to the inner aspect value (see Miguel 1996). 5.2
Nominal properties in SNIs
SNIs carry gender features which, although not visible in the suffix –r, become obvious in anaphoric contexts, where an SNI can only be referred to by the masculine pronoun él and not by the default neuter pronoun ello usually employed with non-nominal CP structures: (34)
Accostumbrado al dulce mirar de su amada, ya no podía used to-the sweet gaze-Inf of his beloved, now not could vivir sin él/*ello. live without it-M/it-N ‘Used to the sweet gaze of his loved one, he could no longer live without it.’
SNIs never combine with adverbs, only with adjectives; modifiers with aspectual value also appear as adjectives (35b): (35) a. El andar errabundo/ *errabundamente de Juan the go-about-Inf aimless/ *aimlessly of Juan b. El constante temer (*constantamente) de Juan the constant fear-Inf (constantly) of Juan Despite these nominal features, unlike the Romanian infinitives and similarly to the Romanian supine and the English verbal gerund, SNIs do not accept plural marking: (36) *los (constantes) construires casas de los albañiles the constant build-inf-pl houses-acc of the workers ‘the constant buildings of houses by the workers’ Both infinitive nominalizations – Spanish and Romanian – have clear nominal properties; we thus claim that they both involve a Classifier with Gender features
Syntactic realizations of plural in Romance and Germanic nominalizations
(see Iordăchioaia & Soare (2009) for further details of this analysis of SNIs). Something, however, blocks plural in SNIs. As shown above in (31), Romanian infinitive nominalizations also disallow plural in one context, namely when they combine with a bare plural theme introduced by the preposition de. This modifies the aspectual calculus of the construction, involving de-telicization. In this situation, the aspect of the event nominal becomes unbounded, a property which we specify on the Classifier projection as carrying a [–count] feature. In this case, we assume that the projection of NumP is blocked like with mass nouns. We propose to view the inner aspect specification in SNIs and Romanian infinitives, as a feature [±count] on the ClassifierP. We adopt the following structure for SNIs, first proposed in Iordăchioaia & Soare (2009): (37)
DP 3 D
ClassP 3 Class
NP 3
N el [-count] -r 5.3
VP construi-
The [±count] feature on ClassifierP
On the basis of the (nominal) similarities between SNIs and Romanian infinitive ASNs, Iordăchioaia and Soare (2009) proposed that SNIs project ClassifierP which accommodates the atelicity of the root under a [±count] feature specification. Alexiadou et al. (2010) propose the same kind of structure for the English nominal gerund. We summarize below the role of the [±count] feature in ASNs: – count nouns, Romanian infinitives (with genitive theme), and the few telic nominal gerunds in English project ClassP [+count], the input for NumP; – mass nouns, most English nominal gerunds, SNIs and Romanian infinitives with bare plural theme project Classifier [–count] which blocks NumP; – German –ung nominals are ambiguous between [+count] and [–count], depending on the inner aspect of the verb; only the former allow a NumP; – while Number gives information about the form of the noun/nominal (singular or plural), the [±count] specification indicates the semantic ‘number’;
Artemis Alexiadou, Gianina Iordăchioaia & Elena Soare
the morphological form of the noun under Classifier is always non-plural (Kratzer 2005); – [–count] means semantic plurality; [+count] means semantic singularity; these notions successfully accommodate the inner aspect specification, a verbal encoding of number; – Classifier[–count] has the same effects as (outer) Aspect[–bounded]/[imperfective]: the blocking of plural morphology/NumberP. 6. Conclusions We have shown in this paper that the occurrence of plural marking in ASNs is influenced by aspectual and categorial factors. Morphological plural appears with telic ASNs and correlates with a nominal functional structure projecting Classifier and Number. This appears in Romanian infinitives, English telic nominal gerunds, and German –ung ASNs. Plural is blocked by imperfective/atelic aspect. Number is thus replaced by AspP in more verbal ASNs (Romanian supine and English verbal gerund). A ClassifierP does not appear in this case because there are no nominal properties. Plural can also be blocked by atelic inner aspect (English atelic ing of, Romanian -re with de + bare plural, Spanish nominal infinitives). The inner aspect sensitivity and the nominal properties here suggest the presence of a ClassifierP hosting the nominal properties and carrying the specification of the inner aspect via the feature [–count] blocking the projection of NumberP. References Abney, Steve. 1987. The English Noun Phrase in its Sentential Aspect. Doctoral dissertation, MIT. Alexiadou, Artemis. 2001. Functional Structure in Nominals: Nominalization and Ergativity. Amsterdam: John Benjamins. Alexiadou, Artemis. 2005. “Gerund Types, the Present Participle and Patterns of Derivation”. Event Arguments in Syntax, Semantics and Discourse, ed. Claudia Maienborn & Angelika Wöllstein-Leisten. Tübingen: Niemeyer: 139–52. Alexiadou, Artemis, Gianina Iordăchioaia, Elena Soare. 2010. “Plural Marking in Argument Supporting Nominalizations”. Layers of Aspect, ed. by Patricia Cabredo Hofherr & Brenda Laca. Stanford: CSLI. Borer, Hagit. 2001. “The forming, the formation and the form of nominals”. Handout USC. Borer, Hagit. 2005. Structuring sense. Oxford University Press. Cinque, Guglielmo. 1999. Adverbs and Functional Heads: A Crosslinguistic Perspective. New York: Oxford University Press.
Syntactic realizations of plural in Romance and Germanic nominalizations Cornilescu, Alexandra. 2001. “Romanian Nominalizations: Case and Aspectual Structure”. Journal of Linguistics 37:3. 467–501. Demonte, Violeta & Soledad Varela. 1997. “Spanish Event Infinitives. From Lexical-Semantics to Syntax-Morphology”. Theoretical Issues at the Morphology-Syntax Interface, supplement of International Journal of Basque Linguistics and Philology XI, ed. by Amaya Mendikoetxea & Myriam Uribe-Etxebarria, 253–277. Ehrich, Veronika. 2002. On the verbal nature of certain nominal entities. More than words. A Festschrift for Dieter Wunderlich, ed. by Ingrid Kaufmann & Barbara Stiebels. Berlin: Akademie-Verlag. Engelhardt, Miriam. 2000. The Projection of Argument-Taking Nominals. Natural Language and Linguistic Theory 18: 41–88. Fábregas, Antonio & Soledad Varela. 2006. “Verb Classes with Eventive Infinitives in Spanish“. Selected Proceedings of the 9th Hispanic Linguistics Symposium, ed. by Nuria Sagarra and Almeida Jacqueline Toribio, 24–33. Somerville, MA: Cascadilla Proceedings Project. Fassi-Fehri, Abdelkader. 2005. “Verbal and Nominal Parallelisms”. Documents & Reports 8. Rabat: Publications IERA. Grimshaw, Jane. 1990. Argument structure. Cambridge, Mass: MIT Press. Harley, Heidi. 2005. “How Do Verbs Get Their Names? Denominal Verbs, Manner Incorporation€ and the Ontology of Verb Roots in English”. The Syntax of Aspect, ed. by Nomi Erteschik-Shir & Tova Rappaport, 42–64. Oxford: Oxford University Press. Iordăchioaia, Gianina & Elena Soare. 2008. “Two Kinds of Event Plurals: Evidence from Romanian Nominalizations”. Empirical Issues in Syntax and Semantics 7, ed. by Olivier Bonami & Patricia Cabredo-Hofherr, ISSN1769–7158. Iordăchioaia, Gianina & Elena Soare. 2009. Structural Patterns Blocking Plural in Romance Nominalizations. Romance languages and linguistic theory 2007, ed. by Enoch Aboh, Elisabeth van der Linden, Josep Quer & Petra Sleeman. Amsterdam: John Benjamins. Jackendoff, Ray. 1991. “Parts and Boundaries”. Cognition 41. 9–45. Kratzer, Angelika. 1994. “On External Arguments”. University of Massachusetts Occasional Papers 17: 103–30. Kratzer, Angelika. 1995. “Stage-level and Individual-level Predicates”. The Generic Book, ed. Gregory N. Carlson and Francis Jeffry Pelletier. Chicago: University of Chicago Press. Kratzer, Angelika 2005. “On the Plurality of Verbs”. Semantics Archive: http://semanticsarchive.net/ Archive/jI4YWRlO/. Laca, Brenda. 2006. “Indefinites, Quantifiers and Pluractionals: What Scope Effects Tell us about Event Pluralities”. Non-definiteness and Plurality, ed. by Liliane Tasmowski & Vogeleer, Svetlana. 191–217. Amsterdam: John Benjamins. Lasersohn, Peter. 1995. Plurality, Conjunction and Events. Dordrecht, Boston: Kluwer. Miguel, Elena de. 1996. “Nominal Infinitives in Spanish: An Aspectual Constraint”. Canadian Journal of Linguistics. 41:1. 29–53. Mourelatos, Alexander P. D. 1978. “Events, Processes and States”. Linguistics and Philosophy 2:3. 415–434. Picallo, M. Carme. 2006. “Some Notes on Grammatical Gender and l-Pronouns”. Proceedings of the workshop “Specificity and the evolution/ emergence of nominal determination systems in Romance”, ed. by Klaus von Heusinger, Georg A. Kaiser & Elisabeth Stark, Universität Konstanz, Arbeitspapier Nr. 119. 107–121. Rijkhoff, Jan. 2002. The Noun Phrase. Oxford: Oxford University Press.
Artemis Alexiadou, Gianina Iordăchioaia & Elena Soare Roodenburg, Jasper. 2006. “The role of Number within Nominal Arguments: the Case of French Pluralized Event Nominalizations”. Paper presented at the 36th Linguistic Symposium on Romance Languages. Rutgers, March-April 2006. Soare, Elena. 2007. “Morphosyntactic Mismatches Revisited: the Case of Romanian Supine”. Acta Linguistica Hungarica 54/ 2007 (Proceedings of the 12th International Morphology Meeting, Budapest, June 2005): 1–19. Van Geenhoven, Veerle. 2004. “For-adverbials, Frequentative Aspect, and Pluractionality”. Natural Language Semantics 12: 135–190. Verkuyl, Henk. 1993. A Theory of Aspectuality. The Interaction between Temporal and Atemporal Structure. Cambridge: Cambridge University Press.
The syntax of Spanish parecer and the status of little pro* Adolfo Ausín
Michigan State University The existence of a null argument in preverbal position in null subject sentences has been challenged in recent years. This paper presents new empirical evidence based on binding considerations in favor of a syntactically realized argument in preverbal position in null subject sentences. Using raising constructions such as those with the Spanish verb parecer ‘to seem’, I show that the null subject in raising constructions behaves as if in a preverbal position in the matrix sentence. The simplest explanation for this state of affairs is to assume that there is a pro in preverbal position. Keywords: pro, null subjects, preverbal subjects, raising constructions
1. Introduction In this paper, I present evidence for the existence of a null subject in preverbal position in sentences like (1): (1) a. Como muchas patatas. eat.1sg many potatoes ‘I eat many potatoes.’ b. Hablas de ti mismo. talk.2sg about yourself ‘You talk about yourself.’ Even though there is no overt subject in (1), it is clear that we need to have a syntactically represented subject. Otherwise the agent theta role could not be assigned in either sentence and the anaphor could not be bound in (1)b. Chomsky * I would like to thank the audience of LSRL and the two anonymous reviewers for their comments and suggestions, as well as Marcela Depiante, Francisco Fernández-Rubiera and Kazuko Hiramatsu. All errors are my own.
Adolfo Ausín
(1982:€81) was the first to propose the existence of a null pronominal element (pro) in subject position. This type of proposal was widely assumed within the LGB framework and has recently been defended within the minimalist program by Holmberg (2005), Kim (2006), and Gallego (2007) among others. However, there have been other proposals. It has been proposed that there is a subject but it is part of the verb. This was commonly found in traditional grammar. According to Fernández Soriano (1999:€1225) “la idea tradicional más extendida es que en estas oraciones la flexión ‘contiene en sí’ al sujeto”.1 It could be said that Manzini and Savoia (1997), Alexiadou and Anagnostopoulou (1998), and Platzack (2003) recast this traditional idea in minimalist terms. For instance, Platzack says: “I will claim that the agreement ending is merged into the structure as an ordinary argument, and that it, like clitics and weak pronouns, may be syntactically interpreted either as a head or a full phrase. However, ultimately agreement cannot live unsupported at PF but has to align to some other word.” (Platzack 2003:€332)
Another possibility is that there is a null subject but it does not appear in preverbal position. This is defended in Platzack (1996) where it is claimed that in (2) pro does not move to Spec,AgrSP until LF. (2) Verrà. come.3sg.fut ‘He will come.’ (3) a. SS: Verrà pro b. LF: pro Verrà On the other hand, Cardinaletti (1997) presented what she described as “definite arguments in favor of a preverbal position of pro in Italian” (35). One of her arguments is based on agreement phenomena. In the Central Italian dialect spoken in the area of Ancona, the agreement pattern found in sentences with null subjects is the same as the one found in sentences with preverbal subjects and different from the agreement pattern found in sentences with postverbal subjects. (4) a. b. c.
Questo, lo fa *Questo, i bambini lo fa Questo, i bambini lo fanno this, the children it do ‘This the children always do.’
sempre i bambini. sempre. sempre. always the children
1. “The most extended traditional idea is that in these sentences the inflection contains the subject in itself”.
The syntax of Spanish parecer and the status of little pro
The examples in (4) show that in this dialect, postverbal plural subjects do not require agreement with the verb, while preverbal subjects do. Crucially, when the subject is null as in (5), lack of agreement is not possible. (5) a. *Questo, lo fa sempre. (pro = 3rd pl) b. Questo, lo fanno sempre. The conclusion that Cardinaletti reaches is that pro must appear in preverbal position. If it could appear in postverbal position, we would expect (5)a to be possible in the same way as (4)a, contrary to facts.2 Similarly, Cecchetto (2000) considers certain facts in Italian that in his opinion can only be explained if we assume that there is a null pronominal in Spec,IP. First, he observes the contrast in (6): an R-expression inside of a Clitic Left Dislocated phrase can be coreferential with a postverbal subject, as in (6)a, but not with a preverbal subject as in (6)b. (6) a. L’opera prima di [uno scrittore]i la scrive sempre luii the work first of a writer it writes always he b. ??L’opera prima di [uno scrittore]i luii la scrive the work first of a writer he it writes sempre (volentieri) always with pleasure ‘The first work of a writer, he always writes with pleasure.’ Interestingly, if the subject in these sentences is replaced with a null subject as in (7), the sentence is ruled out. That is, (7) with pro patterns with (6)b with a preverbal subject. This parallelism also reinforces the idea that pro appears in a preverbal position. (7) L’opera prima di [uno scrittore]i pro*i la scrive sempre (volentieri) the work first of a writer (he) it writes always with pleasure3 In this paper I present empirical arguments from Spanish in favor of the appearance of pro in preverbal position. The evidence comes from raising structures in Spanish, mostly with the verb parecer ‘to seem’. In the first section, I introduce the type of empirical argument that I will use in this paper, as well as some background assumptions regarding raising constructions in Spanish. In Section€ 2, 2. See Cardinaletti (1997) and Barbosa (2009) for more arguments and counterarguments on the position/existence of pro. 3. Arregi (2006) considers similar facts in Spanish and concludes that the structural position of the (null) subject is not relevant at least for Spanish. For Arregi, the crucial difference is between null and lexical subjects.
Adolfo Ausín
I present new evidence for a preverbal position of pro. Finally, I present possible, alternative solutions which could handle the facts without a preverbal pro. 2. Background Binding arguments are commonly used when the position of an element in a sentence is at stake. Thus, in the controversy of whether the subject of an ECM infinitival complement raises to the object position of the matrix sentence or stays in the embedded infinitival, Lasnik (1999) presents a series of binding arguments in favor of raising to the matrix clause. Consider the sentences in (8) from Lasnik (1999). (8) a. Joan believes hei is a genius even more fervently than Bobi does. b. *Joan believes himi to be a genius even more fervently than Bobi does. According to Lasnik, the contrast between (8)a and (8)b can be easily explained if we assume that there is raising of the embedded subject to the matrix sentence in (8)b. After him raises to the object position in (8)b, it can bind Bob. This binding triggers a Condition C violation, which explains the ungrammaticality of (8)b. Similarly, Bošković (1997:€85) uses binding arguments to determine the position of the associate at LF. He considers sentences like the ones in (9):
(9) a. *There seems to himself to be someone in the garden. b. Someone seems to himself to be in the garden.
According to Bošković, the contrast in (9) can easily be explained if we assume that the associate stays in-situ and does not move to Spec,IP at LF. If the associate were to raise to the position occupied by there, we would not expect any difference between (9)a and (9)b. Both Lasnik and Bošković are using binding considerations to determine the position of a constituent. In the examples below, I will follow the same strategy. I will use evidence based on binding to determine the position of the null subject. However, we cannot use sentences like (1)a to determine the position of the null subject because there are no other elements which we could use to test the position of the null subject. (1)b could be more helpful since it contains an anaphor which must be bound. However, if the preverbal and postverbal positions are identified with Spec,IP and Spec,VP,4 the fact that the anaphor is bound in (1)b cannot helps us determine the position of the null subject, since the anaphor could 4. For expository purposes, I stay away from a VP shell representation, but nothing would change. Similarly, nothing changes if the landing position of the subject is somewhere other than Spec,IP.
The syntax of Spanish parecer and the status of little pro
be bound from either position. In other words, assuming that (10) is the correct representation of (1)b, the anaphor could be bound by pro in Spec,VP or Spec,IP. (10) [IP pro [I’ hablas [VP pro [V’ tV [PP de ti mismo]]]]] Movement from Spec,VP to Spec,IP doesn’t alter the binding possibilities of the subject. We need examples where raising to subject alters the binding possibilities of the subject. Raising constructions with parecer ‘to seem’ are a perfect candidate, since the relative position of the raising subject changes with respect to the experiencer. Unfortunately for the task at hand, raising over the experiencer in Spanish is controversial. Following Ausín and Depiante (2000) I assume that Spanish parecer in a sentence like (11) is a raising predicate that takes a small clause as complement and that the subject of the embedded clause raises over the experiencer.5 (11) a. Juan me parece inteligente. Juan to-me seems intelligent ‘Juan seems to me (to be) intelligent.’ b. Juan me parece [SC tJuan inteligente] In Ausín (2001) I offer evidence in favor of the raising analysis of examples like (11). The clearest type of evidence is examples like (12). (12) *María se parece inteligente. María to-herself seems intelligent ‘María seems to herself (to be) intelligent.’ (Torrego 1995:€231)6 5. Contrasts like the one illustrated in (i) are normally used to argue against the possibility of raising over the experiencer in Spanish. (i) Este taxista ‘This taxi driver
(* me) parece estar cansado. to-me seems to be tired.’ (Torrego 1996:€106)
Ausín and Depiante (2000) present an account of contrasts like this that does not rely on the experiencer blocking the raising of the subject. For alternative views, see Torrego (1996, 1998, 2002), Gallego (2007) among others. 6. Additional evidence that might be used to argue for the derived status of the subject in sentences like (11) comes from the following examples from Ausín (2001). i.
*El amigo de Juani lei pareció inteligente the friend of J. to-him seemed intelligent
ii. El amigo de sui madre lei pareció inteligente [a todo el mundo]i the friend of his mother to-him seemed intelligent to everybody iii. El amigo de sui madre no lei pareció inteligente a nadiei the friend of his mother not to-him seemed intelligent to nobody ese examples seem to indicate that the subject at some point of the derivation has been Th c-commanded by the experiencer.
Adolfo Ausín
I take the inability of the subject to bind the anaphor in (12) as an indication that we are dealing with a derived subject, and that María has raised over the experiencer.7 Now, consider the following example. (13) Me parece inteligente. to-me seems intelligent ‘He seems to me (to be) intelligent.’ The question we want to answer is where the subject is in (13). Assuming the raising analysis of parecer, the possible preverbal and postverbal positions for the null subject now have different binding capabilities. The null subject binds the experiencer from the preverbal position but not from the postverbal position. (14) pro me parece [SC pro inteligente] In the next section I will present three different types of evidence that seem to indicate that the null subject binds the experiencer. 3. The placement of the null subject The three types of evidence that I will present are based on: Condition C, Weak Cross Over, and anaphor binding. 3.1
Condition C
Consider the following sentence. (15) Le parece al hermano de Juan inteligente. him seems to-the brother of Juan intelligent ‘He seems to Juan’s brother (to be) intelligent.’8 The sentence in (15) becomes ungrammatical if Juan and the null subject are coreferential. The English counterpart of the ungrammatical reading would be: (16) *Hei seems to Juan’s i brother (to be) intelligent. The sentence in (16) is ruled out as a Condition C violation. It seems natural to seek the same type of explanation for the lack of the coreferential reading in (15), 7. A full explanation for the ungrammaticality of (12) and the contrasting grammaticality of (9)b falls beyond the scope of this article, but see Ausín (2001) for an attempt. 8. An anonymous reviewer points out that there are speakers who find this sentence completely out. I leave for future research the study of this possible dialectal variation.
The syntax of Spanish parecer and the status of little pro
and the easiest way to achieve this is to assume that there is a pro in the matrix subject position. hermano de [Juan]i [SC tproi inteligente] (17) proi le parece al pro him seems to-the brother of Juan intelligent The impossibility of a coreferential reading in (15) is accounted for as a Condition C violation in the same way as (16), but only because of the presence of pro in a preverbal position. If pro were to remain in-situ, as in (18), we could not account for the lack of the coreferential reading in (15). (18) le parece al hermano de [Juan]i [SC proi inteligente] him seems to-the brother of Juan pro intelligent 3.2
WCO
Consider the following sentences. (19) a. ?El representante de [cada futbolista]i lei traicionó the manager of every soccer.player him betrayed ‘Every soccer player’s manager betrayed him.’ b. *proi traicionó al representante de [cada futbolista]i betrayed to-the manager of every soccer.player ‘He betrayed every soccer player’s manager.’ The contrast in (19) shows that there is something like binding out of DP (Büring 2004) in Spanish. Even though the quantifier is embedded in the DP in (19)a, the interpretation of the pronoun can still depend on the interpretation of the quantifier, probably, among other things, because the DP containing the quantifier c-commands the pronoun. Although the sentence in (19)a is slightly marginal, it is much better than (19)b, where neither the quantifier nor the DP containing the quantifier c-commands pro. (19)b is a typical WCO violation. Now consider the sentences in (20). (20) a. ?Le parece al representante de cada futbolista que him seems to-the manager of every soccer.player that es incapaz de esforzarse. is unable to work-hard ‘It seems to every soccer player’s manager that he is unable to work hard.’ b. *Le parece al representante de cada futbolista incapaz him seems to-the manager of every soccer.player unable
Adolfo Ausín
de esforzarse. to work-hard ‘He seems to every soccer player’s manager unable to work hard.’
There is a clear contrast between (20)a and (20)b, that can be explained if we assume that the simplified structure of these sentences is as in (21): (21) a. ?Le parece [al representante de [cada futbolista]i] [CP que proi es incapaz de esforzarse] ‘It seems to every soccer player’s manager that he is unable to work hard.’ b. *proi le parece [al representante de [cada futbolista]i] [SC tÂ�i incapaz de esforzarse] ‘He seems to every soccer player’s manager unable to work hard.’ In (21)a, pro appears in the subject position of the embedded clause, and from there it can be c-commanded by the experiencer, the DP containing the quantifier, which allows the marginal bound interpretation of the pronoun. In (21)b, on the other hand, under the current proposal, pro appears in the subject position of parece, and from that position it cannot be c-commanded by the experiencer, making the bound interpretation of pro completely impossible. If pro were to appear in the subject position of the embedded small clause as in (22), we would not expect to find any differences between (21)a and (21)b.9 (22) *Le parece [al representante de [cada futbolista]i] [SC proi incapaz de esforzarse] ‘He seems to every soccer player’s manager unable to work hard.’ Again, the contrast that we find in (20)/(21) can be easily explained if we assume that pro appears in the subject position of null subject sentences.
9. Examples like (i) also come to mind if one is trying to test Cross Over violations in raising constructions. In (i) the quantifier cannot bind the null subject. One could interpret this as an argument for raising of pro to subject position. Once pro has raised, it cannot be bound by todo el mundo. We would have the typical Strong Cross Over configuration that appears in (ii). However, the lack of the bound reading interpretation in (i) can also be explained even if pro stays in-situ as in (iii), since (iii) can be ruled out as a condition B violation. i. Le parece a todo el mundo inteligente. him seems to everybody intelligent ‘He seems to everybody (to be) intelligent.’
ii. *proi le parece a [todo el mundo] i inteligente
iii. *le parece a [todo el mundo] i [proi inteligente]
The syntax of Spanish parecer and the status of little pro
3.3
Anaphor binding
Let us turn to anaphor binding. In principle, the sentences that should be considered are the Spanish counterpart of the English (23) with a null subject. (23) He seems to himself (to be) intelligent. Unfortunately, the Spanish counterpart of (23) is ungrammatical independently of the presence of a null or overt subject. (24) is ungrammatical but so is (12), (24) *Se parece inteligente. to-herself seems intelligent ‘She seems to herself (to be) intelligent.’ To be able to use anaphor binding as a test to detect the position of pro, we need to consider other raising constructions. For this purpose, let’s consider examples with declarar that appear in (25). This verb can take a finite embedded clause as in (25)a, or a small clause as in (25)b. (25)c shows that the subject of the small clause can be passivized. Furthermore, (25)d, although slightly marginal, shows that an anaphor can be bound by the passivized subject. (25) a. Juan declaró que María era incapaz de esforzarse. Juan declared that María was unable to make-an-effort ‘Juan declared that María was unable to make an effort.’ b. Juan declaró a María incapaz de esforzarse. ‘Juan declared María unable to make an effort.’ c. María fue declarada por Juan incapaz de esforzarse. ‘María was declared by Juan unable to make a effort.’ d. ?María fue declarada por sí misma incapaz de esforzarse. ‘María was declared by herself unable to make an effort.’ The simplified structure of (25)d would be (26) where the matrix subject has raised from the subject position of the embedded small clause. (26) María fue declarada por sí misma [SC t María incapaz de esforzarse] María was declared by herself unable to make-an-effort The exact grammatical status of (25)d is not very important for the present discussion. What is important is whether there is a grammaticality contrast between the following two sentences: (27) a. ?Fue declarada por sí misma incapaz de esforzarse. was declared.F by herself unable of make-an-effort ‘She was declared by herself unable to make an effort.’
Adolfo Ausín
b.
*Fue declarado por sí misma que era incapaz de was declared.M by herself that was.3sg unable to esforzarse. make-an-effort ‘It was declared by herself that she was unable to make an effort.’
There’s a clear contrast between the two sentences, (27)a being significantly better than (27)b. This is expected if it is assumed that the simplified structure of these sentences is: (28) a. ?proi fue declarada por [sí misma]i [SC tpro incapaz de was declared.F by herself unable of esforzarse] make-an-effort b. *(pro-expl) fue declarado por sí misma [CP que proi era was declared.M by herself that was incapaz de esforzarse] unable of make-an-effort In (28)a there is a pro in matrix subject position that can bind the anaphor. In (28) b, on the other hand, pro in the embedded sentence cannot bind the anaphor in the matrix clause and a clear Condition A violation results. Again, we have found a clear contrast that can be easily explained if we assume that there is a pro in preverbal position. 3.4
Null subjects in non-raising contexts
So far, I have presented empirical evidence from Condition C, pronoun binding and Condition A in support of the idea that pro needs to appear in a preverbal position high enough to bind the experiencer/passive by-phrase. The pronoun and anaphor binding facts simply require that pro may appear in a preverbal position. However, the Condition C facts require that pro must raise to that position. If raising of pro could be optional then there would be a derivation of (15) in which pro would remain in argument position (namely, (18)) and no Condition C would arise. Now the question that needs to be answered is whether the conclusion that pro must raise to subject position can be extended to all types of sentences including non-raising sentences like those in (1). Unless there is evidence to believe that raising to subject position is different in raising and non-raising contexts, the null hypothesis is that the conclusion reached after examination of raising contexts should be extended to non-raising contexts. In other words, if
The syntax of Spanish parecer and the status of little pro
pro has to appear in preverbal position in (15), then it also has to appear in preverbal position in (1). An anonymous reviewer claims that the contrast in (29) shows that the raising constructions that I have been discussing in this paper differ from non-raising constructions in that the subject cannot appear postverbally. (29) a. María le parece a Juan inteligente. María him seem to Juan intelligent ‘María seems to him (to be) intelligent.’ b. *Le parece a Juan María inteligente. If so, the conclusion that I have reached in this paper regarding the preverbal position of pro in raising constructions could not be extended to non-raising contexts, since in non-raising constructions a subject can stay in a postverbal position. Interestingly, (29)b seems to improve with a yes/no question intonation, and becomes perfect if a Juan is dropped. Further research is needed to investigate the contrast illustrated in (29) before it can be used as an argument against my analysis. 4. Alternative analyses without pro One could think of alternative ways of handling the evidence presented in this paper that do not require pro in preverbal position. If we assume with Platzack (2003) that agreement is a clitic or an incorporated pronoun, one could argue that it is the agreement morpheme itself that is causing the condition C violation in (15)/(17), making the appearance of pro unnecessary. Under such an approach, the representation of (15) would be: hermano de [Juan]i [SC t+ei inteligente] (30) le parec+ei al him seems to-the brother of Juan intelligent If so, (30) could be ruled out because the agreement morpheme is triggering a Condition C violation. However, such an approach does not seem sound mainly for two reasons. First, it would require that elements smaller than a word could be involved in binding, which is problematic as already pointed out by Chomsky (1970) using examples like (31). (31)a doesn’t mean the same thing as (31)b, as one would expect if self were a real anaphor that could be bound. (31) a. This is clearly a self-inflicted wound. b. This is a wound inflicted by the wound.
Adolfo Ausín
Also, if it were true that agreement morphemes could participate in binding relations based on their clitic-like status, then we would expect normal clitics to participate in binding relations too. However, Ausín (2007) presents some evidence that seems to indicate that clitics do not bind. Consider the following example from Zubizarreta (1998:€113): envió la madre de Juani varias cartas. (32) El mes pasado lei last month 3sg.dat sent the mother of Juan several letters ‘Last month, Juan’s mother sent him several letters.’ The grammaticality of (32) seems to indicate that the clitic le is not binding Juan. If le does not bind Juan and does not trigger a Condition C violation in (32), it is at least counterintuitive to assume that the agreement morpheme in (30) can trigger a Condition C violation.10,11 A different explanation without pro could be proposed using ideas sketched in Chomsky (2007), who builds on Reuland (2005).12 Chomsky considers the following situation: (33) [T…XP…R] Chomsky (2007:€13) claims that in this situation “XP binds R – indirectly via the common probe C-T”. If we assume that in (27)a, repeated below, pro remains insitu in the subject position of the embedded small clause, (27)a could be an example of this type of situation. (27) a. ?Fue declarada por sí misma incapaz de esforzarse. was declared.F by herself unable of make-an-effort T … … R … XP Following Chomsky, we could say that pro binds the anaphor via the common probe, which would be T in fue. If so, pro would still be needed but it would not have to appear in preverbal position. However, as with the previous alternative, 10. One could propose a different implementation of Platzack’s proposal and assume that the agreement morpheme moves to Spec,IP before merging with the verb as in: (i) [IP [t-ei] [I’ le parec+ei al hermano de [Juan]i [SC t+ei inteligente]]] Agr him-seems-Agr to-the brother of Juan Agr intelligent But this could be considered as a notational variant of a theory with pro. 11. As an anonymous reviewer points out, the argument based on (32) weakens if it turns out that le in (32) and dative clitics in general are not pronominal elements but applicative morphemes (see Demonte (1995), Cuervo (2003)). However, there would still be the question of why this morpheme’s phi features (person and number in all dialects, plus gender in laísta dialects), which can identify null arguments, cannot bind. 12. See also Uriagereka and Gallego (2006).
The syntax of Spanish parecer and the status of little pro
the problem with such an approach is that, if correct, one would expect clitics to be able to bind, and this does not seem to be the case. 5. Conclusion In this paper I have presented new empirical evidence that indicates that in sentences with null subjects there is a null pronominal element high enough to bind into the rest of the sentence. The need for pro has been somewhat controversial. If the empirical arguments presented in this paper are correct, then the idea that pro is necessary in preverbal position would be reinforced. References Alexiadou, Artemis & Elena Anagnostopoulou. 1998. “Parametrizing AGR, Word Order, V-Movement, and EPP-Checking”. Natural Language & Linguistic Theory 16.491–539. Arregi, Karlos. 2006. “Reconstruction and Condition C in Spanish: A NonÂ�structural Account”. Selected Proceedings of the 9th Hispanic Linguistics Symposium ed. by Nuria Sagarra and Almeida Jacqueline Toribio, 1–12. Somerville, Mass.: Cascadilla Proceedings Project. Ausín, Adolfo. 2001. On A-Movement. Ph.D. dissertation, University of Connecticut. Ausín, Adolfo. 2007. “Spanish Clitics, Binding and UG”. Paper presented at the Georgetown University Roundtable on Languages and Linguistics (GURT) 2007, Georgetown University, Washington D.C. Ausín, Adolfo & Marcela Depiante. 2000. “On the Syntax of parecer with and without Experiencer”. Hispanic Linguistics at the Turn of the Millennium ed. by Héctor Campos, Elena Herburger, Alfonso Morales-Front, and Thomas J. Walsh, 150–177. Somerville, Mass.: Cascadilla. Barbosa, Pilar. 2009. “Two Kinds of Subject pro”. Studia Linguistica 63.2–58. € Bošković, Željko. 1997. The Syntax of Nonfinite Complementation: An Economy Approach. Cambridge, Mass.: MIT Press. Büring, Daniel. 2004. “Crossover Situations”. Natural Language Semantics 12.23–62. Cardinaletti, Anna. 1997. “Subjects and Clause Structure”. The New Comparative Syntax ed. by Liliane Haegeman, 28–40. London: Longman. Cecchetto, Carlo. 2000. “Doubling Structures and Reconstruction”. Probus 12:1.93–126. Chomsky, Noam. 1970. “Remarks on Nominalization”. Readings in Transformational Grammar ed. by Roderick A. Jacobs & Peter S. Rosenbaum, 184–221. Waltham, Mass.: Blaisdell. [Reprinted in Chomsky, Noam. 1972. Studies on Semantics in Generative Grammar 11–61. The Hague: Mouton.] Chomsky, Noam. 1982. Some Aspects and Consequences of the Theory of Government and Binding. Cambridge, Mass.: MIT Press. Chomsky, Noam. 2007. “Approaching UG from Below”. Interfaces + Recursion = Language? ed. by Uli Sauerland & Hans-Martin Gärtner, 1–29. Berlin: Mouton de Gruyter. Cuervo, María Cristina. 2003. Datives at Large. Ph.D. dissertation, MIT. Demonte, Violeta. 1995. “Dative Alternation in Spanish”. Probus 7:1.5–30.
Adolfo Ausín Fernández Soriano, Olga. 1999. “El Pronombre Personal. Formas y Distribuciones. Pronombres Átonos y Tónicos.” Gramática Descriptiva de la Lengua Española. Real Academia Española ed. by Ignacio Bosque y Violeta Demonte, 1209–1274. Madrid: Espasa. Gallego, Ángel J. 2007. Phase Theory and Parametric Variation. Ph.D. dissertation, Universitat Autònoma de Barcelona. Holmberg, Anders. 2005. “Is There a Little pro? Evidence from Finnish”. Linguistic Inquiry 36:4.533–564. Kim, Jun-Han. 2006. La Teoría del pro y el Sujeto Pre/postverbal. Madrid Ediciones Universidad Autónoma de Madrid. Lasnik, Howard. 1999. Minimalist Analysis. Oxford: Blackwell. Manzini, Rita M. & Leonardo M. Savoia. 1997. “Null Subjects without pro”. UCL Working Papers in Linguistics 9.303–313. Platzack, Christer. 1996. “Null Subjects, Weak Agr and Syntactic Differences in Scandinavian”. Studies in Comparative Germanic Syntax ed. by Höskuldur Thráinsson, Samuel D. Epstein, & Steve Peter, vol. II:180–196. Dordrecht: Kluwer. Platzack, Christer. 2003. “Agreement and Null Subjects”. Nordlyd: Tromsø University Working Papers on Language and Linguistics, Proceedings of the 19th Scandinavian Conference of Linguistics ed. by Anne Dahl, Kristine Bentzen, and Peter Svenonius, vol. 31, 326–355. Reuland, Eric. 2005. “Agreeing to Bind”. Organizing Grammar: Linguistic Studies in Honor of Henk van Riemsdijk ed. by Hans Broekhuis, Norbert Corver, Riny Huybregts, Ursula Kleinhenz, and Jan Koster, 505–13. Berlin: Walter de Gruyter. Torrego, Esther. 1995. “From Argumental to Non-Arguemental Reflexives”. Probus 7.221–241. Torrego, Esther. 1996. “Experiencers and Raising Verbs”. Current Issues in Comparative Grammar ed. by Robert Freidin, 101–120. Dordrecht: Kluwer. Torrego, Esther. 1998. The Dependencies of Objects. Cambridge, Mass.: MIT Press. Torrego, Esther. 2002. “Arguments for a Derivational Approach to Syntactic Relations Based on Clitics”. Derivation and Explanation in the Minimalist Program ed. by Samuel D. Epstein & T. Daniel Seely, 249–268. Oxford: Blackwell. Uriagereka, Juan & Ángel J. Gallego. 2006. “(Multiple) Agree as Local (Binding and) Obviation”. Paper presented at Going Romance XX, Vrije Universiteit Amsterdam, December 2006. Zubizarreta, Maria Luisa. 1998. Prosody, Focus, and Word Order. Cambridge, Mass.: MIT Press.
Two types of (apparently) ditransitive light verb constructions* María Cristina Cuervo An analysis of Spanish ditransitive constructions with light verb dar reveals that despite surface similarity these constructions belong to two groups corresponding to distinct argument structures. While expressions like dar permiso/ánimo ‘give permission/encouragement’ correspond to double-object constructions, experiencer constructions like dar miedo/envidia ‘give fear/ envy’ correspond to an unaccusative structure that patterns syntactically and semantically with configurations with psychological predicates of the piacere/ gustar ‘like’ type. On the present account, the morphosyntactic and semantic properties of the construction –including subject properties of the dative, case, and restrictions on bare nouns– derive directly from the way sentences are built in the syntax, not from stipulations in lexical entries or linking rules. The proposal is extended to cover predicates formed with other light verbs (e.g., ser ‘be’, parecer ‘seem’, resultar ‘be’, quedar ‘remain’).
1. Introduction The Spanish verb dar ‘give’ combines with many bare nouns into light verb expressions of the form ‘give something to somebody’: dar+ apoyo ‘support’, envidia ‘envy’, calor ‘heat’, vergüenza ‘shame’, miedo ‘fear’, muerte ‘death’, gusto ‘pleasure’, lástima ‘pity’, ganas ‘wishes’, permiso ‘permission’, asco ‘disgust’, etc. These constructions involve three arguments: one bears nominative case, another one dative case, and a bare noun appears as direct complement of the verb. (1) El plantel siempre le da apoyo al técnico. the team.nom always cl.dat gives support the coach.dat ‘The team always supports the coach.’
* I thank the audience of LSRL XXXVIII and two anonymous reviewers for useful comments and suggestions.
María Cristina Cuervo
(2) Al técnico le dan rabia las protestas. the coach.dat cl.dat give.pl fury the complaints.nom ‘The complaints make the coach furious.’ I will argue that these ‘(di)transitive’ dar expressions belong to two different groups, such that sentences (1) and (2) correspond to two systematically distinct argument structures, that is, they have different underlying representations. Specifically, I argue that sentence (1) is truly ditransitive and corresponds to a double-object construction (i.e., the dative DP is an internal argument). In contrast, sentence (2) –and psychological dar-expressions in general– corresponds to an unaccusative configuration in which the dative argument is merged outside the verbal phrase (i.e., it is not an internal argument). This claim is supported by systematic contrasts in syntactic properties. I further argue, building on Masullo (1992), that the structure of (2) patterns syntactically and semantically with configurations with psychological unaccusative verbs of the gustar ‘like’ type (3). (3) Al técnico no le gustan las protestas. the coach.dat neg cl.dat like.pl the complaints.nom ‘The coach doesn’t like complaints.’ The parallel between (2) and (3) can only be made and accounted for if we assume a particular structure for psych predicates of Belleti & Rizzi’s third type, specifically, a structure where the nominative argument is not a complement of the verb (contra Belleti & Rizzi 1988; Fernández Soriano 1999; Bruhn de Garavito 2002; etc.).1 The analysis can naturally be extended to account for psychological and non-psychological predicates with other stative light verbs (ser ‘be’, parecer ‘seem’, resultar ‘be’, quedar ‘remain’ + Adj/Adv). (4) Al técnico le quedan grandes esas zapatillas. the coach.dat cl.dat remain.pl big those trainers.nom ‘Those trainers are too big for the coach.’ This paper is organized as follows. Section 2 presents the properties of the constructions, highlighting both the similarities and the differences between them. In Section 3 I present an analysis of each construction which accounts for the properties described in the previous section. A parallelism between psychological dar expressions and psychological predicates of the gustar type is revealed, and its predictions are tested. In Section 4 the analysis is extended to other light verb expressions, both psychological and non-psychological. Conclusions are presented in Section 5. 1. The parallelism argued for by Masullo (1992) refers mainly to the status of the dative argument and the lack of external argument; the present analysis extends to cover the details of the internal structure of the verbal phrase.
Two types of (apparently) ditransitive light verb constructions
2. Morphosyntactic properties The three-argument dar-expressions appear in sentences that share several morphosyntactic features. In all cases, the light verb appears immediately followed by a non-referential bare noun. Two other argument DPs appear: one in nominative, one in dative case. The verb agrees in person and number with the nominative DP; the dative DP is doubled by an agreeing (in person and number) dative clitic. So far, there seems to be one class of dar-expressions. The data in this section, however, will present a series of morphosyntactic contrasts that strongly suggest that there are two different types of dar-expressions, which I will refer to as ditransitive dar-expressions (DitDar) and psychological dar-expressions (PsychDar). In particular, we shall see that the dative DP in DitDar behaves like a regular dative in a ditransitive construction, while the dative in PsychDar exhibits many subject properties. 2.1
Clitic doubling
All dative arguments in Spanish must be doubled by a clitic, with the only exception of arguments in ditransitive constructions that express transfer of possession (Strozer 1976, Masullo 1992, Demonte 1995, among others). Clitic doubling of the dative is optional for DitDar (5) but obligatory for PsychDar (6). (5) Dimos apoyo a tres candidates. we.gave support to three candidates ‘We supported three candidates.’ (6) *Las protestas dan miedo al técnico. the complaints.nom give.pl fear the coach.dat ‘The coach is afraid of the complaints.’2 2.2
Word order
Normal word order (that is, the word order for wide focus interpretation, with nuclear stress; see Zubizarreta 1998) is Nom-V-Dative for DitDar, but Dative-VNom for PsychDar, as illustrated in (1)-(2). This clearly argues for a difference in underlying structures. Reverse orders are ungrammatical with neutral stress, but possible if the preverbal DP receives contrastive stress. 2. Clitic doubling is also required in Dat-V-Nom, but since left-dislocated objects independently require doubling, it is important to see that with PsychDar clitic doubling is required independently of word order.
María Cristina Cuervo
2.3
Indefinite quantifiers
Spanish indefinite generalized quantifiers can appear in subject position, but not in left-dislocation (Masullo, 1992). According to this test, given the ungrammaticality of (7) with neutral stress, a preverbal dative DP is left-dislocated in the case of DitDar; the dative appears in subject position in PsychDar (8). (7) *A nadie le da apoyo el equipo. nobody.dat cl.dat gives support the team.nom ‘The team doesn’t support anybody.’ (8) A nadie le dan miedo las protestas. nobody.dat cl.dat give.pl fear the complaints.nom ‘Nobody is afraid of the complaints.’ 2.4
Binding
The nominative DP can bind a possessive in the dative in DitDar (9); it is the dative DP that binds the possessive in the nominative in PsychDar (10). Sentences b are ungrammatical or very degraded even if with contrastive stress on the first DP. dio bola a sui técnico. (9) a. El equipoi no le the team.nom neg cl.dat gave ball his coach.dat ‘The team didn’t pay attention to its coach.’ b. ??Al técnicoi no le dio bola sui equipo. the coach.dat neg cl.dat gave ball his team.nom *‘Its team didn’t pay attention to the coach.’ (10) a. Al técnicoi le dio lástima sui equipo. the coach.dat cl.dat gave pity his team.nom ‘The coach felt pity for his own team.’ dio lástima a sui técnico. b. ??El equipoi le the team.nom cl.dat gave pity its coach.dat *‘Its coach felt pity for the team.’ 2.5
Scope
In regular ditransitive predicates, inverse scope is impossible between a quantifier in the preverbal nominative and a quantifier in a postverbal dative; if order is
Two types of (apparently) ditransitive light verb constructions
inverted, reverse scope is possible (Cuervo 1999).3 In the order Dat>Nom, reverse scope obtains in DitDar (11b) but not in PsychDar (12). This suggests that the preverbal dative in PsychDar is not left-dislocated. (11) a. Un técnico le dio aliento a todo equipo. *∀ >> ∃ a coach.nom cl.dat gave encouragement every team.dat ‘A coach encouraged every team.’ b. A un equipo le dio aliento todo técnico. ∀ >> ∃ a team.dat cl.dat gave encouragement every coach.nom ‘Every coach encouraged a team.’ (12) A un técnico le dio lástima a coach.dat cl.dat gave pity ‘A coach felt pity for every team.’ 2.6
todo every
equipo. team.nom
*∀ >> ∃
Superiority
In Spanish, a multiple wh-question that violates superiority lacks a pair-list reading and can only be interpreted as an echo question (Cuervo (1999). (13)
¿A quién le gritó quién? who.dat cl.dat shouted who.nom ‘At whom shouted who?’ # … Mary shouted at Peter, Daniel shouted at Stephanie, etc.
In order to obtain a pair-list reading, DitDar requires Nom-V-Dative (14) while PsychDar requires Dative-V-Nom (15). This is expected if the nominative is merged higher than the dative in (14) but lower in (15). (14) ¿Quién le da aliento a quién? who.nom cl.dat gives encouragement who.dat ‘Who encourages whom?’ (15) ¿A quién le da lástima quién? who.dat cl.dat gives pity who.nom ‘Who pities whom?’
3. If the order of nominative-dative is reversed, as in (11b), the dative is left-dislocated, and the sentence must be uttered with special intonation to be acceptable. This requirement on intonation does not apply to (12).
María Cristina Cuervo
2.7
Raising with parecer ‘seem’
The argument that raises from a transitive clause embedded under parecer ‘seem’ is the nominative subject (16). In the case of PsychDar, in contrast, it is the dative that raises (17), as in the case of dative subjects with unaccusative predicates (Masullo, 1992). (16) El equipo parece haberle dado apoyo al técnico. the team.nom seems have.cl.dat given support the coach.dat ‘The team seems to have supported the coach.’ (17) a. Al técnico parecen haberle dado rabia las propuestas. the coach.dat seem have.cl.dat given anger the proposals.nom ‘The suggestions seem to have angered the coach.’ b. ??/*Las propuestas parecen haberle dado rabia al técnico. the proposals.nom seem have.cl.dat given anger the coach.dat ‘The suggestions seem to have angered the coach.’ 2.8
Accusative clitic
In certain contexts, an accusative clitic can refer back to the bare noun in DitDar (18). The fact that this is never possible with PsychDar (19) suggests that accusative case is not available with these predicates. (18) ¿Le dieron apoyoi al técnico? → Sí, se loi dieron cl.dat gave support the coach. dat Yes, cl cl.acc gave ‘Did they give support to the coach? Yes, they gave it to him’ (19) ¿Le dimos miedoi al técnico? → *Sí, se loi dimos cl.dat gave fear the coach.dat Yes, cl cl.acc gave ‘Did we frighten the coach? Yes, we did’ All these data strongly argue for a systematic difference between DitDar and PsychDar.4 Specifically, the data show that DitDar behaves like an ordinary ditransitive structure in which a nominative subject asymmetrically c-commands a dative DP. In contrast, PsychDar constructions parallel the behaviour of unaccusative predicates in which a preverbal dative DP asymmetrically c-commands the postverbal nominative.
4. M. Teresa Espinal (p.c.) points out that, interestingly, the two types of expressions I argue for are formed by different light verbs in Catalan: while DitDar expressions employ the Catalan light verb donar ‘give’, PsychDar expressions use either donar or fer ‘make’. See also Espinal 2004.
Two types of (apparently) ditransitive light verb constructions
3. Analysis 3.1
The ditransitive structure of DitDar
Data from binding, scope, superiority, raising and word order show that DitDar expressions such as dar apoyo ‘support’, dar permiso ‘allow’, dar aliento ‘encourage’, dar cuerda ‘wind up’, dar muerte ‘kill’, correspond to regular ditransitive structures. In other words, sentence (20) is a double-object construction. (20) El plantel siempre le da apoyo al técnico. (=1) the team.nom always cl.dat gives support the coach.dat ‘The team always supports the coach.’ I follow Cuervo’s (2003a) analysis of Spanish double-object constructions in terms of low applicatives (Pylkkänen 2008). (21)
VoiceP 3 DPSubj
3 vP Voice
3 vDO
3 ApplP Root
3 DPDat
3 Appl0
DPObj
Structure (21) expresses the direct relation between the direct and the indirect objects. The applicative head, Appl, takes an object DP as its complement and projects a specifier, in which the dative argument is licensed. The verbal root combines with the low applicative phrase and then merges with the verbal head. Dative case is inherent, expressed by the particle a in the DP and the dative clitic in Appl0. Apparent optionality of clitic doubling is expected if (20) is a double-object construction: that is the way the two variants of the dative alternation are expressed in Spanish (Masullo 1992, Demonte 1995, Cuervo 2003a). The difference between the double-object construction in (21) and that of DitDar reduces to the lack of verbal root (dar spells-out the head vDO) and the size of the complement of Appl.
María Cristina Cuervo
3.2
The unaccusative structure of PsychDar
Data in Section 2 show that, in contrast to DitDar, PsychDar expressions such as dar +lástima ‘pity’, +miedo ‘fear’, +asco ‘nausea’ +hambre/sed ‘hunger/thirst’, etc., are not regular ditransitive structures; rather, they have an underlying unaccusative structure in which the dative argument is higher than the nominative. Unlike in ditransitives, there is no direct (possessive) relation between the dative argument and the bare noun. In a sentence like (22), the dative DP is an experiencer related to a stative predicate. Its structure appears in (23). (22) Al técnico le dan rabia las protestas. (=2) the coach.dat cl.dat give.pl fury the complaints.nom ‘The complaints make the coach furious.’ (23)
ApplP 3 3 vP
DPDat
al técnico Appl le
3 DP
3
las protestas v dan
6 n + √rabia
In (23), dar is the expression of a stative v which combines with the noun (n+√), which expresses the experience, and projects a specifier, where the “stimulus” nominative DP is merged. Crucially, here the dative experiencer is licensed as the specifier of a High Applicative head, that is, an applicative that takes the vP as its complement, not a DP (Pylkkänen 2008). VoiceP is not projected: there is no accusative case or (agentive) external argument. 3.2.1 Accounting for morphosyntactic and semantic properties In the analysis of psychological dar constructions presented above, the semantics derives directly from the configuration; the analysis can also account for their morphosyntactic properties. Dative case of the experiencer is not stipulated as a property of the verb: it is inherent case assigned by Appl. This means that whenever an applicative is licensed, there will be a dative clitic, and licensing does not depend on particular verb types. This licensing and case-agreement relationship between the DP and the applicative head is expressed by the dative particle a in the experiencer DP and by the agreeing clitic as the spell-out of Appl.
Two types of (apparently) ditransitive light verb constructions
In (23), the dative experiencer is higher than the nominative argument. This hierarchy accounts for the asymmetric syntactic relation between the arguments, as expressed by binding, superiority, and scope possibilities of the dative DP over the nominative DP. The dative DP is not only relatively higher than the nominative but it also displays subject properties, related to word order, movement and predication, as presented in Section 2 (see Masullo 1992, Fernández Soriano 1999, Cuervo 1999, Bruhn de Garavito 2002 for discussion). These subject properties derive both from its higher merge position, and from its movement to the specifier of Tense (subject position) and satisfaction of EPP (sentences are ‘about’ the dative argument). The verb agrees, in person and number, with the postverbal DP. Verbal agreement proceeds in the same way as in other Spanish unaccusative or passive configurations, with no requirement of movement to specifier of Tense. In the present case, there is a split between EPP checking and agreement, one DP –the dative– satisfying EPP on T, while another –the nominative– checks case and triggers verbal agreement (see Béjar 2000, Cuervo 2003b, among others, for similar proposals of splits in checking features of one head). Except for dative case marking and verbal agreement, these properties contrast with the properties of constructions with ditransitive dar expressions, in which the nominative argument, an external argument, is higher than the dative DP, a VP-internal argument. 3.2.2 Parallelism with psychological predicates As noted by Masullo (1992), the configurations in which psychological dar expressions appear share several fundamental properties with constructions with piacere/gustar predicates –Belletti & Rizzi’s (1988, henceforth B&R) third kind. (24) Al técnico no le gustan las protestas. the coach.dat neg cl.dat like.pl the complaints.nom ‘The coach doesn’t like complaints.’ Syntactically, the dative DP exhibits subject properties in both constructions (except those related to nominative case and agreement); the nominative argument is merged lower. In terms of their semantics, the structures are similar in the interpretation of the dative and nominative arguments: The dative DP is interpreted as an experiencer (a “possessor” of a psychological state); the nominative DP is the “stimulus” (it generates the experience). For B&R, gustar-type predicates are a kind of “double-object construction with a non-thematic subject position” (B&R:293).
María Cristina Cuervo
(25)
S
qp NP
VP
qp V´
NP
3
ec
V #
NP #
piacere
Theme
Experiencer
(B&R: 335)
Structure (25) arises from the lexical information associated with the verb piacere/ gustar, which interacts with a general statement on thematic prominence that states that Experiencers project higher than Themes. Although (25) expresses the asymmetric relation between the DPs, their proposal for psych predicates cannot express the parallelism with PsychDar in structural terms: since the theme DP is the complement of the verb and the experiencer is VP-internal, there is no place for the N (nP) complement of dar to merge. The same problem arises with other approaches, such as Masullo’s (1992) and Bruhn de Garavito’s (2002), in which the dative is an internal argument. In contrast, Cuervo’s (2003a) analysis of dative experiencers provides a way to express the syntactic and semantic parallel: the dative DP is licensed outside the vP, as the specifier of a High Applicative. (26)
ApplP 3 3 vP
DPDat
al técnico Appl le
3 DP
3
las protestas BE
Root gust-
In (26) the dative and the nominative arguments have the same properties as in the structure of PsychDar (23). The main difference between these configurations is that in (23) the head v Â�–spelled out as dar– takes the bare noun phrase as its complement (dar+NP), rather than merging with a verbal root, as illustrated below (cf.€Masullo 1992, Kornfeld 2005). (27) [ApplP expDP [Appl’ Appl0-le [vP themeDP [v’ v-dar NP ] ] ] ]
Two types of (apparently) ditransitive light verb constructions
3.3
Predictions of the parallelism
A crucial property of my proposal lies on the way the theme DP is licensed. The nominative theme is licensed as the specifier of the stative vP. Thus, it is not a complement of the verb: it is its subject. There being an internal subject, there is a predication relation between the verb and the nominative argument. The dative DP, in contrast, is not an internal argument. If structure (26) is correct, we expect PsychDar configurations to exhibit the following properties, shared by gustar-type constructions: 1. The nominative argument cannot be a bare noun, irrespective of whether it appears post-verbally or preverbally. 2. The dative argument is ‘external’ to the predication relation between the verb and the nominative DP. 3.3.1 Bare NPs Subjects in Spanish cannot be bare nouns: they must appear with a determiner (or some kind of heavy modification). This notion of subject includes pre and postverbal agentive subjects, and preverbal subjects of unaccusatives and passives, as illustrated in (28-30). (28) *Chicos comieron manzanas. kids.nom ate apples.acc ‘Some kids ate apples.’ (29) *Invitados llegaron a las ocho. guests.nom arrived at the eight ‘Guests arrived at eight.’ (30) *Puentes fueron construidos en los 70. bridges.nom were built in the 70 ‘Bridges were built during the 70s.’ The restriction does not apply to direct objects (manzanas in (28)) nor to postverbal subjects of unaccusative verbs (i.e., arguments licensed as objects): (31) a. Llegaron invitados a las ocho. arrived.pl guests.nom at the eight ‘Guests arrived at eight.’ b. Al técnico le faltan documentos. coach.dat cl.dat lack.pl documents.nom ‘The coach is missing some documents.’
María Cristina Cuervo
Interestingly, the restriction does apply to the postverbal nominative argument of verbs like gustar. (32) *Al técnico no le interesan protestas. *the coach.dat not cl.dat interest.pl complaints.nom ‘The coach isn’t interested in complaints.’ The restriction on bare NPs in these disparate environments can be generalized in terms of predication, which covers both the licensing position of arguments and the (derived) subject position, SpecTP. (33) The Bare Noun Phrase Constraint Revised5 “An unmodified common noun cannot be the subject of a predicate under conditions of normal stress and intonation” The restriction on nominative arguments of gustar-verbs is naturally accounted for if these themes are licensed as subjects. It is mysterious if they are regular objects of the verb. By the same token, the theme in psychological dar configurations (34) is an inner subject, and therefore bare NPs should not be allowed in this position. This prediction is confirmed. (34) *Al técnico le dan rabia protestas. the coach.dat cl.dat give.pl fury complaints.nom ‘Complaints make the coach furious.’ 3.3.2 Sentences without experiencer It is possible for gustar-type predicates to appear in sentences without a dative experiencer. (35) Las protestas sin sentido no molestan /gustaron. the complaints without sense.nom neg bother.pl /liked.pl ‘Senseless complaints are not bothersome/were not liked.’ As in the case of (35), the configuration proposed here for PsychDar predicts that there is a predication relation between the theme and the predicate dar+NP to which the dative experiencer is external. Thus, it should be possible to express this predication without a dative argument.
5. Adapted from Suñer’s (1982) Naked Noun Phrase Constraint (Suñer 1982:209) (see also Torrego 1989): “An unmodified common noun in preverbal position cannot be the surface subject of a sentence under conditions of normal stress and intonation”
Two types of (apparently) ditransitive light verb constructions
(36) Las protestas sin sentido dan rabia. the complaints without sense.nom give.pl fury ‘Senseless complaints make you furious.’ Acceptability of sentences with no dative (36) confirms this prediction; these sentences express some property of the nominative argument (are ‘about’ the nominative argument) without restricting it to an experiencer.6 To sum up, the configuration presented in (23) can naturally account for the syntactic and semantic properties of psychological dar configurations. The proposal also expresses and accounts for the parallel between PsychDar and the structure of other dative psych-predicates in the language. Previous approaches to dative experiencer constructions cannot account for these properties nor for the parallel. 4. Other light verb constructions The idea that the theme of PsychDar expressions is an internal subject, and that the experiencer is external to the predication is further supported by their syntactic and semantic parallels with predicates built by the combination of a copular or quasi-copular verb + an adjective or adverb. The resulting predicates can be psychological (Section 4.1) or express a non-psychological state (Section 4.2). 4.1
Psychological predicates
Copular light verbs such as ser ‘be’, parecer ‘seem’ and resultar ‘turn out to be’ can combine with adjectives of psychological nature such as aburrido ‘boring’, importante ‘important’, indiferente ‘indifferent’, to form a predicate. Sentences (37) express a property of the books or the movies, and it is hardly controversial that there is a predication relation between the nominative argument and the predicate.7 Their structure appears in (38). (37) a. Esos libros son /parecen importantes. those books.nom are.pl /seem.pl important ‘Those books are/seem important.’ 6. As M. L. Zubizarreta (p.c.) points out, it might be that an experiencer is implied in cases like (35)-(36). For the purposes of this argument, however, what is crucial is the possibility of no syntactic expression of the experiencer, that is, the absence of an applicative. This is particularly acceptable in tenses -such as present- which do not force an eventive reading of the sentence. 7. I am dealing here only with parecer+predicate taking a DP argument, not with parecer taking a complement clause.
María Cristina Cuervo
b. Las películas japonesas resultaron aburridas. the movies Japanese.nom turned-out.pl boring ‘The Japanese movies turned out to be boring.’ (38)
vPBE 3 DP
3
los libros v
6
son a + importantes /parecen If an experiencer argument is added as an applied argument (39), we obtain the structure in (40). (39) A Vera le son /parecen importantes esos libros. Vera.dat cl.dat are.pl /seem.pl important those books.nom ‘Those books are/seem important to Vera.’ (40)
ApplP 3 3 vPBE
DPDat
a Vera Appl le
3 DP
los libros
3 v son
6 a + importantes
Structure (40) is identical to the structure of PsychDar expressions, except that there the predicate is formed by a verbalizing head and a noun rather than an adjective. As predicted, the dative argument in (40) is interpreted as an experiencer, and the postverbal nominative cannot be a bare noun. 4.2
Non-psychological predicates
Light verbs such as ser ‘be’ and estar ‘be’, resultar ‘turn out to be’ and quedar ‘remain’ can also combine with non-psychological adjectives, or adverbs. The subject of the predicate can be a DP (41a), or an infinitival control clause, as viajar ahora in (41b). The subject of the infinitival clause is arbitrary PRO.
Two types of (apparently) ditransitive light verb constructions
(41) a. Esos zapatos son/resultan incómodos. those shoes.nom are.pl /turn-out.pl uncomfortable ‘Those shoes are/ turn out to be uncomfortable.’ b. Está difícil viajar ahora. is difficult to travel now ‘It is difficult to travel nowadays.’ Dative arguments can also be applied to these constructions whose predicates are not psychological (42). In the case of an infinitival clause, its PRO subject is obligatorily controlled by the dative argument (42b). (42) a. Al técnico le resultan incómodos *(esos) zapatos. the coach.dat cl.dat are.pl uncomfortable those shoes.nom ‘The coach finds those shoes uncomfortable.’ b. Al técnico le está difícil viajar ahora. the coach.dat cl.dat is difficult to travel now ‘It is difficult for the coach to travel now.’ The stative verb servir ‘be useful’ is another example of a non-psych predicate that has the same structure as gustar and the light verb constructions discussed above. Its meaning could be expressed with a copular light verb construction (ser útil ‘be useful’). A dative argument is understood as the individual for whom an object is useful. (43) Al técnico no le sirven esas herramientas. the coach.dat neg cl.dat be-useful.pl those tools ‘Those tools are not useful for the coach.’ As the PsychDar and gustar class, servir can appear without a dative argument (44a), in which case, it expresses a general property of the nominative argument (i.e., not restricted to the sphere of an individual). (44) a. Esas herramientas no sirven (para nada). those tools.nom neg be-useful.pl for nothing ‘Those tools are not useful (at all).’ b. *A Vera no le sirven herramientas. *Vera.dat neg cl.dat be-useful.pl tools ‘Tools are not useful for Vera.’ The ungrammaticality of a bare nominative NP in postverbal position (44b) provides further support for the analysis of servir as belonging to the gustar class: stative unaccusative predicates in which the postverbal DP is licensed as an internal subject.
María Cristina Cuervo
We have seen in this section that the structure proposed for PsychDar constructions is very productive in the language. It not only parallels the structure of dative experiencer verbs, but it also extends to several other light verbs that take adjectives or adverbs to form a predicate. 5. Conclusions I started by showing that expressions with light verb dar belong to two different groups, ditransitive dar and psychological dar. The systematic syntactic and semantic differences between the two types of expressions are captured by differences in the basic argument structure of each, specifically in the way the nominative and dative arguments are licensed. By spelling out each piece with separate words, light verbs make the structure of predicates more apparent, allowing us to see the ‘pieces’ inside ‘lexical’ verbs. The approach developed here has provided the elements to discover and account for the morphosyntactic and semantic parallels between psychological dar and B&R’s third type of psych predicates. The analysis also naturally extends to other (non-)psychological predicates formed by a stative light verb other than dar and an adjective or adverb. The structure proposed –a high applicative dative which takes a stative vP as its complement– is very productive in Spanish. The existence of these light verb+adjective/adverb predicates further supports the idea that, despite appearances, PsychDar are not transitive constructions. Ultimately, the proposal implies that unaccusativity is not a uniform phenomenon; that is, it does not correspond to a single underlying structure. Under this approach, syntactic structures are expressed or ‘filled’ by different kinds of elements (e.g. verbs, light verbs+adjective, etc.) that could not be classified as belonging to one lexical class. The morphosyntactic and semantic properties of the construction derive directly from the syntactic heads available in the language and the way sentences are built in the syntax, not from stipulations in lexical entries or linking rules. This highlights the benefits of constructionalist approaches, in terms of economy, and empirical and explanatory coverage. As far as this analysis is able to capture the properties of the configurations under study, it constitutes an advancement in our understanding on how argument structure (understood as syntax) determines verbal meanings and the interpretation of arguments.
Two types of (apparently) ditransitive light verb constructions
References Béjar, Susana. 2000. “Economy, cyclicity and markedness in Georgian verbal agreement”. GLOW Newsletter 44:8–19. Belletti, Adriana & Luigi Rizzi. 1988. “Psych-verbs and Theta Theory”. Natural Language and Linguistic Theory 6:291–352. Bruhn de Garavito, Joyce. 2002. “La position syntaxique de thème des verbes à expérienceur datif”. Revue québécoise de linguistique 31.2.137–155. Cuervo, María Cristina. 1999. “Quirky but not eccentric: dative subjects in Spanish”. MIT Working Papers in Linguistic 34, Papers on Morphology and Syntax, 213–228. Cuervo, María Cristina. 2003a. Datives at Large. Ph.D. dissertation, Massachusetts Institute of Technology. Cuervo, María Cristina. 2003b. “A control vs. raising theory of dative experiencers”. Romance linguistics: theory and acquisition ed. by A. T. Pérez-Leroux & Y. Roberge, 111–130. Amsterdam: John Benjamins. Demonte, Violeta. 1995. “Dative alternation in Spanish”. Probus 7:5–30. Espinal, M. Teresa. 2004. “Lexicalization of Light Verb Structures and the Semantics of Nouns”. Catalan Journal of Linguistics 3.15–43. Fernández Soriano, Olga. 1999. “Two Types of Impersonal Sentences in Spanish: Locative and Dative Subjects”. Syntax 2.2:101–140. Kornfeld, Laura. 2005. Formación de palabras en la sintaxis desde la perspectiva de la Morfología Distribuida. Ph.D. dissertation, Universidad de Buenos Aires. Masullo, Pascual. 1992. Incorporation and Case Theory in Spanish. A crosslinguistic perspective. Ph.D. dissertation, University of Washington, Seattle. Pylkkänen, Liina. 2008. Introducing Arguments. Cambridge, Mass.: MIT Press. Strozer, Judith. 1976. Clitics in Spanish. Ph.D. dissertation, UCLA. Torrego, Esther. 1989. “Unergative-Unaccusative Alternations in Spanish”. MIT Working Papers in Linguistics, 10, 253- 272 Zubizarreta, María Luisa. 1998. Prosody, Focus and Word Order. Cambridge, Mass.: MIT Press.
Modal ellipsis in French, Spanish and Italian Evidence for a TP-deletion analysis Anne Dagnac French, Spanish and Italian, reputedly non-VP-ellipsis allowing languages, can display gaps after root modals. I argue that these gaps are instances of ellipsis, viewed as PF-deletion of a constituent. They indeed allow for A’-movement, which an alternative null proform analysis cannot capture. Yet, they differ from English VP-ellipsis, in particular wrt the kind of remnants they allow, their tolerance to voice mismatches, and by displaying a constraint on subjects in Antecedent Contained Deletion. I propose that these differences follow from the status of the modals in these languages and the size of the deleted structure, which is a TP in the three Romance languages vs a VP in English.1
1. Introduction French, Spanish and Italian, unlike English, are classically held (cf.€e.g. Lobeck 1995) not to allow for VP-ellipsis after auxiliaries, cf.€(1) vs (2): (1) a. *Tom a vu Lee mais Marie n’ a pas__. [Fr] b. *Tom ha visto Lee pero María no ha__. [Sp] c. *Tom ha visto a Lee ma Maria non ha__. [It] Tom has seen (to) Lee, but Mary Neg2 has__. ‘Tom saw Lee, but Mary didn’t’
1. Special thanks to Jason Merchant, for his 2006 lectures in Paris that gave this work its seminal impulse and for later patient answers. Thanks, too, to J.Runner, L.Aelbrecht, D.Sportiche, S.Lappin, H.Demirdache, K.Johnson, the audiences of various talks including LSRL38, and two anonymous reviewers, for helpful comments, questions and/or answers. All errors are of course mine. 2. The French discontinuous negation ne...pas involves a scope marker (SM) ne and a negative word pas, which I subsume under NEG in the glose when giving similar Italian/Spanish examples.
Anne Dagnac
(2) John has gone to Paris but Alyson hasn’t.
Yet, in these languages just as in English, root modals3 can be followed by a gap, interpreted in relation with an antecedent VP, cf.€(3): (3) a. b. c. d.
Tom a pu voir Lee, mais Marie n’a pas pu__. [Fr] Tom pudo ver a Lee, pero María no pudo__. [Sp] Tom ha potuto veder Lee, ma Maria non ha potuto__. [It] Tom can:pst see (to) Lee, but Mary neg can:pst ‘Tom could see Lee but Mary couldn’t__.’
In French, (3a) has been analyzed as an instance of VP-ellipsis (henceforth VPE) by Busquets and Denis (2001), who christen it ‘Modal Ellipsis’ and assume the PF-deletion analysis proposed for English (2/3d) (cf., with variants, Hankamer and Sag 1976, Lappin 1996, Lasnik 2001, Johnson 2001, Merchant 2001, inter alia): a syntactically realized VP fails to be spelled out (‘is deleted’) at Phonological Form under identity with an antecedent VP, as featured in (4):
(4) [...Modal [...[VP V [DP....]...]]
Conversely, Spanish and Italian sentences such as (3b-c) have been recently argued to involve null proforms (Depiante 2001, Cechetto and Percus 2006), i.e. unpronounced proforms devoid of inner syntactic structure, semantically interpreted like an overt pronoun, cf.€(5):
(5) [...Modal [∅]]
In Section 1, building on Busquets and Denis (2001), I show that French ‘Modal Ellipsis’ must indeed be analyzed as a syntactically structured gap. In Section 2, I provide empirical arguments that Spanish and Italian non-restructured root modals involve a similar ellipsis. In Section 3, I show that, nevertheless, French, Spanish and Italian ellipses depart from their English counterpart on several respects, which argue for the deletion of a higher constituent, namely a TP. In Section 4, I show how this proposal may also account for a constraint on their ACD constructions.
3. It is fully productive with ability and allowance modals, and possible, cf.€(i), though restricted, with obligation ones, a fact I have no explanation for. (i) J’ai lu tous les livres que je devais (‘I read all the books that I had to’) Besides, cf.€Depiante (2001), this possibility is restricted to ‘non-restructured’ instances of the modals. See note 9 for a tentative explanation.
Modal ellipsis in French, Spanish and Italian
2. Arguments for a modal ellipsis in French A major diagnostic to distinguish between null proforms and ellipses is the (im) possibility to extract sub-items from the gap. Null proforms, being devoid of inner structure, cannot provide extraction sites for sub-items. On the contrary, if ellipsis is a deleted but otherwise ‘normal’ XP, any classical extraction from within this XP occurring before Spell Out is predicted to be possible. In French, extraction facts clearly support the PF-deletion analysis: modal ellipsis allows for ACD, cf.€(6), as first pointed by Busquets and Denis, but also for free relatives, cf.€(7), and, under the same kind of contrast conditions found with English VPE (Park 2004, Merchant 2008b), for WH-questions, cf.€(8): (6) Léa lit tous les livres qu’elle peut
. Lea reads all the books that she can. (7) Il embrasse [WH quii] il peut <embrasser ti> He kisses [WH who(ever)i] he can ‘He kisses who(ever) he can’ (8)
Je sais quels livres Lea peut lire et je sais aussi I know which books Lea can read and I know also quels livresi Ben ne peut pas . which booksi Ben sm can not . ‘I know which books Lea can read and I also know which books Ben can’t’
Whether ACD involves a classical relative operator or a copy of the head (Sauerland 2000, 2004, Fox 2002) in a matching/head raising analysis (cf.€6’), it requires an item to be moved from a base position inside the gap that a proform could not provide: (6’) Léa lit tous [dp les livres[cp [spec,cp livres]1 qu’elle peut ]] And so do free relatives and WH-questions. I assume that (7) involves WH-movement of the animate direct object qui (Rooryck 1994). In (8), quels livres (‘which books’), must undergo movement from the direct object position of lire (‘read’) in the embedded clause. That these constructions involve movement is confirmed by their island-sensitivity, cf.€respectively (9), (10) and (11): (9)
*Bob a lu tous les livres que Léa partageait l’opinion qu’il ‘Bob read all the books that Lea shared the opinion that ne pouvait pas. he couldn’t’
Anne Dagnac
(10)
*Il finit toujours par embrasser qui on se demandait He ends-up a lways by kiss:inf who one refl wondered s’ il pourrait whether he could. ‘He always ends up kissing who one wondered whether he could’
(11)
*Je sais à qui je peux parler et je sais aussi à qui I know to whom I can talk, and I know also to whom Marie partage l’opinion que je ne peux pas. Mary shares the opinion that I sm can not. ‘I know to whom I can talk, and I also know to whom Mary shares the opinion that I can’t.’
That a null proform cannot license them is further supported by their unavailability with an overt proform hosting no base-position for the moved item either, cf.€(12–14): (12) *Léa lit tous les livres qu’elle peut le faire / qu’ elle le peut Lea reads all the books that she can it=do / that she it=can ‘Lea reads all the books that she can do it’ (13) *Il embrasse qui il peut le faire / qui il le peut He kisses who he can it=do / who he it=can ‘He kisses who(ever) he can do it’ (14)
*Je sais quels livres Marie peut lire et je sais aussi I know which books Mary can read and I also know quels livres elle ne peut pas le faire. which books she sm can not it=do’
A’-movement thus provides solid evidence that FME involves a fully structured though unpronounced XP. In the next section, I show that this conclusion extends to Spanish and Italian. 3. Modal ellipsis beyond French Spanish and Italian similar gaps have been argued by Depiante (2001) and Cechetto and Percus (2006) to disallow extraction and hence to involve null proforms. This proves to be empirically wrong.4 4. Many thanks to Jaime, Enrique, Luisa, Zazil, Lucia, Fabio, Roberta and Andrea for patient discussions and judgements. My Spanish informants are from Galicia, Madrid, Andalucía and Mexico City, my Italian informants from northern Italy and Sardinia.
Modal ellipsis in French, Spanish and Italian
3.1
ACD and the ‘same subject constraint’
Cechetto and Percus (2006) argue that Italian doesn’t display ACD with modals on the basis of examples such as (15) (their[29b]), which I extend to Spanish in (16): (15) *Potrei mangiare ogni pizza che Gianni potrebbe. Could:1sg eat each pizza that Gianni could:3sg ‘I could eat every pizza that Gianni could’ (16) *María lee todos los libros que Juan puede. Mary reads all the books that John can This argument misses an important fact, though: in these examples, the subject of the modal (John) and the subject of the matrix clause (I, Mary) are referentially disjunct. But if both subjects corefer, ACD is perfectly grammatical in these languages, cf.€(17) – a contrast also found in French, cf.€(17c) vs (17’):5 (17) a. b. c.
Maria legge tutti i libri che può [Italian] María lee todos los libros que puede [Spanish] Marie lit tous les livres qu’ elle peut [French] Maryi reads all the books that (she) can:3sg Mary reads all the books that she can’
(17’) *Marie lit tous les livres que Jean peut. Mary reads all the books that John can 3.2
Other A’-movements
Likewise, contra Depiante (2001), under the correct contrast conditions, WHquestions are perfectly grammatical, cf.€(18–19)6. (18) a. b.
Ahora, ya sé a quién puedo confiar mi hijo, Adesso, so a chi posso affidare mio figlio, Now, (well) know:1sg to whom can:1sg confide my son, pero todavía no sé a quién no puedo. [Sp] ma non so ancora a chi non posso. [It] but (still) not know:1sg (still) to who neg can1:sg ‘Now, I know to whom I can confide my son, but I still don’t know to whom I can’t’
5. See Section 4 for an account. 6. Some informants require further contrast, on the WH-word or the modal itself. I ignore this point, since, crucially, given some contrast, they all admit WH-questions.
Anne Dagnac
Free relatives are licensed, too, cf.€(19). (19) a. b.
Besa a quien puede. [Sp] Baccia chi può. [It] Kisses (to) who can:3sg ‘S/he kisses who s/he can’
All these sentences involve A’-movement from the gap. Hence, Spanish and Italian pattern with French: their Modal Ellipses involve a deleted constituent. In the next section, though, I argue that they differ from English VP-ellipsis. 4. Modal ellipsis is a TP-Ellipsis If French, Spanish and Italian modal ellipsis was indeed similar to English VPE, as proposed for French by Busquets and Denis, they should display parallel properties, which is partly not the case. I propose to link these differences to the status of their deontic modals. English modals are commonly viewed as auxiliaries selecting a vP/VP, cf.€(20), and, following Merchant (2008a), as deleting a VP7: (20) ...[can...[vP tsujet [VP....]]]]]] / [can [have...[vP tsujet [VP ....]]]]]] French, Spanish and Italian modals are subject-raising verbs argued to select a TP (see e.g. Wurmbrand 2001, and Ruwet 1972 for French), whith a general structure such as (21):8 (21) [VP modal [TP tsujet [...T... [AspP ...[vP tsujet [VP ....]]]]]] I propose that they target this TP for deletion.9
7. vs a vP (Johnson 2004, Merchant 2007) or a VoiceP (Baltin 2007). A reviewer points that it predicts wrongly that Floated Quantifiers can be remnants. In fact, as Sag (1978) already noted, the problem extends to high adverbs such as often or probably, and is not specific to ellipsis, since this ban extends to various pre-extraction positions. I follow Sag in assuming this is due to a phonological output constraint. 8. Other proposals all involve a complement bigger than VP, which would not crucially change the general argumentation. 9. According to e.g. Abeillé & Godard (2003), Cardinaletti & Shlonsky (2004), Romance restructuring modals (who ban ellipsis, cf.€note 3) are auxiliary-like and select a vP/VP instead of a TP. They are then likely to disallow ellipsis for the same reason other auxiliaries do. This reason may be that the category of potential licensees for ellipsis is parametrical and deletion in these languages can target TPs but not vPs/VPs.
Modal ellipsis in French, Spanish and Italian
4.1
Remnants
Several elements can indeed intervene between the modal and the elided verb, among which aspectual and voice auxiliaries of the lower verb. In English, as expected if only VP elides, they can be left over by VPE, as shown respectively in (22) and (23): (22) Paul managed to repair this stuff – Luke couldn’t have (23) The janitor should remove the trash whenever it is apparent that it needs to be On the contrary, in French, Spanish and Italian, all projections above the elided VP must disappear, as shown for the aspectual head in (24) and the voice head in (25). (24) a. b. c.
*Tom peut avoir fini en juin, et Lea peut aussi avoir. *Tom puede haber acabado en junio, y Lea también puede haber. *Tom può aver finito in giugno e anche Lea può avere. Tom can have finished in June and (also) Lea (too) can (also) have ‘Tom can have finished in June, and Lea can have, too’
(25) a. b. c.
*Paul peut être muté, et Kiko aussi peut être. *Paco puede ser trasladado, y Kiko también puede ser. *Paolo può essere trasferito, e anche Kiko può essere. Paul can be transfered, and (also) Kiko (also) can be ‘Paul can be transfered, and Kiko can be, too’
Assuming that ‘have’ is in AspP (Demirdache & Uribe-Etxebarria 2002) and passive ‘be’ dominates either a VoiceP above vP (Collins 2005) or a vP with a passive feature, this is expected if modal ellipsis deletes the full TP, cf.€(26): (26) a. Modal [TP... T...[AspP ‘have’ [passive BE [VP [VP ....]]]]] The case of negation is less straightforward. With English modals, the nontensed verb may marginally admit a negation, which doesn’t survive ellipsis with modals, though, cf.€(27). We assume that, like the negation following to in He tried to not win, it is an instance of constituent negation, contrary to the sentential negation that appears before to in He tried not to win and can escape ellipsis (Travis 2000), cf.€(28b). (27)€ a. She cannot resist temptation, and he cannot [VP NOT resist it.] b. *She cannot resist temptation, and he cannot . (28)€ a. *He wanted to yield, but he tried [TP to <[VP NOT yield]>] b. He wanted to yield, but he tried [TP NOT to ]
Anne Dagnac
Thus, the (higher) sentential negation in (28b) survives VP-ellipsis, whereas the constituent negation in (27b/28a), the only one available with modals, must disappear with the VP. The three Romance languages can display a sentential negation on the infinitive clause, which is in a NegP above T (Zanuttini 2001)10. If TP is deleted, cf.€(30), it is predicted not to survive ellipsis, which is borne out, cf.€(29) (29) a. b. c.
*Paul peut aller à Madrid et Lea peut[tp ne pas <>]. [French] *Paco puede ir a Madrid y Lea puede[tp no<>]. [Sp] *Paolo può andare a Madrid e Lea può [tp non<>]. [It] Paul can go to Madrid and Lea can not ‘Paul is allowed to go to Madrid and Lea is allowed not to’
The different sizes of the deleted material in English VPE vs ‘Modal Ellipsis’ thus capture the differences in the remnants allowed: (30) a. Modal [TP... [(NegP1) [T...[Asp [passive BE [[vp [VP ....]]]]]]11 b. Modal (Neg1) [Asp [passive BE [vP [VP (Neg) ....]]]] 4.2
Voice mismatches
Another difference between the two constructions is their tolerance to voice mismatches between the elided constituent and its antecedent.
10. For French, cf.€Zanuttini (2001): a NegP1 above T, hosting Spanish and Italian negations, also hosts the scope-marker ne. I leave open (and don’t note in (30)), the exact location of the lower NegP hosting French pas, since it is irrelevant here. I assume that the AgrS projection above NegP, necessary in tensed clauses to host the subject, is not projected in infinitive TPs following raising verbs. 11. A reviewer notes that this proposal correctly predicts (i), with a positive polarity particle on the infinitive, but not (ii) or (iii): i. *Juan puede no llegar tarde al trabajo, pero Pedro puede sí Juan can not arrive late at-the job, but Pedro can ‘yes’ ii. *Juan (no) puede no llegar tarde al trabajo, pero Pedro puede Juan (neg) can not arrive late at-the job, but Pedro can trabajo, pero Pedro sí puede iii. ?Juan puede no llegar tarde al Juan can not arrive late at-the job, but Pedro yes can I n fact, my proposal predicts (ii), since it violates the identity requirement on ellipsis: the two TPs differ ([TP1 no haber llegado tarde al trabajo] ≠ [TP2 haber llegado tarde al trabajo]). In (iii), sí, though on poder, must have low scope on the infinitive for the sentence (in particular pero) to make sense. This suggests it has raised to a focus position in the matrix (before poder), which both enables it to be outside the elided site, and its trace/copy inside the gap to escape the parallelism requirement via focus (on this, see Section 4).
Modal ellipsis in French, Spanish and Italian
Voice mismatches are possible with English VP-ellipsis, under the right discourse conditions (Kehler 2000, Frazier & Clifton 2006). In (31), the antecedent verb is active and the elided one passive; (32) illustrates the reverse case. (31) The janitor must remove the trash whenever it is apparent that it should be (Merchant 2008a, [2b]) (32) This problem was to have been looked into, but obviously nobody did (Merchant 2008a, [1a]) On the contrary, Modal Ellipsis appears to rule them out, even in the same discourse conditions, as exemplified respectively for French, Spanish and Italian in (35–36).12 (33) a. b. c.
*Il faut remplacer l’ampoulei de l’ escalier, mais ellei *Hay que cambiar la bombillai de la escalera, pero proi *Bisogna cambiare la lampadinai della scala, ma proi ‘It needs’ replace the bulb of the staircase, but it/proi ne peut pas <être remplacée> – ellei est coincée. no puede <ser cambiada> – proi se bloqueó. non può <essere cambiata> – proi è bloccata. neg can – it/proi is jammed /refl jam:pst:3sg ‘Someone should replace the bulb in the staircase but it can’t – it’s jammed’
(34) a. *Ce problème aurait dû être résolu, mais ma b. ??Questo problema avrebbe dovuto essere risolto, debería haber sido solucionado, pero c ??Este problema This problem must- be-pst-cond solved, but visiblement personne n’ a pu. evidentemente nessuno ha potuto. parece que nadie ha podido. obviously/appears that nobody (SM) can:pst. ‘This problem should have been solved, but obviously nobody could.’ If Modal Ellipsis deletes more structure than VP-ellipsis, these facts may be given an explanation. Merchant (2008a) indeed notes that, cross-linguistically, VP-ellipsis allows for voice mismatches, whereas pseudo-gapping and sluicing rule them out. He proposes that in VP-ellipsis, only a VP being deleted, the v-head hosting the diverging voice feature remains outside of the ellipsis site. The material inside the elided VP thus complies to the usual identity requirement on 12. Some speakers seem to marginally admit the passive antecedent/active ellipsis cases. In any case, voice data need investigating as thoroughly as in English.
Anne Dagnac
ellipsis. With pseudo-gapping and sluicing, he argues, more material is deleted (respectively vP and TP). Hence, the diverging v-heads are inside the ellipsis site and violate the identity condition. If both his analysis and the data in (33–34) are right, the different tolerance to voice mismatches found in Modal Ellipsis and English VPE is straightfully accounted for by my proposal: in the latter, the deleted TP includes the diverging voice feature and thus violates the identity condition on ellipsis. 5. Accounting for the ‘Same Subject Constraint’ Another puzzling difference is that ACD constructions are submitted to the ‘Same Subject Constraint’ (cf.€Section 2) in the three Romance languages, but not in English – cf.€(35) vs (36), for the same context. (35) [Tom is married to Lea, and he is blind: he cannot read. So, Lea reads aloud for him. In fact] Lea reads every book that Tom can’t (J.Merchant, p.c.) (36) a. b. c.
*Lea lit tous les livres que Tom ne peut pas. *Lea lee todos los libros que Tom no puede. *Lea legge tutti i libri che Tom non può. Lea reads all the books that Tom neg can
The TP-deletion hypothesis offers a way to account for this difference, too, as I shall show here for French. As is well-known, ellipses must comply to an identity condition. Its nature has been much debated, but a current formulation is that the content of the elided constituent must be semantically identical to its antecedent, modulo Focus-marked elements (for a technical implementation, see e.g. Merchant 2001). Within recent accounts of ACD (Sauerland 2000, 2004, Fox 2002), this requirement extends to traces, viewed as copies. Practically, this means that diverging elements (here the subjects) must either be outside the ellipsis site or be focussed. Now, with an elided VP, such as in English (35), the trace of the subject is outside the elided site, cf.€(37): it always evades the identity requirement. (37) *Lea reads every book [cpbook2 that Tom1 can’t [vP Tom1 [VP read the book2] On the contrary, with a raising verb and a TP-ellipsis, as in (36) roughly detailled in (38), a copy of the subject is inside the ellipsis site. It must then either corefer with its antecedent, or be focussed. The SSC is then just a specific instanciation of the identity condition on ellipsis: if the infinitve subject corefers with the
Modal ellipsis in French, Spanish and Italian
antecedent subject, cf.€(39) the ellipsis is well-formed. If it doesn’t, as in (36/38), it violates the identity condition and is then ruled out. (38) *Lea lit les livres [cp livres2 que Tom1 ne peut pas [tp [vP Tom1 [VP lire les livres2]] (39) *Tom1 lit tous les livres [cp livres s2 qu’il1 peut [tp... [vP il1 [VP lire les livres2]] Tom reads all the books that he can ‘Tom reads all the books that he can’
he read the books
Its only way to avoid ill-formedness in this case is to be F-marked. This, we claim, happens in conjuncts, cf.€(40), where all F-marking devices (De Cat 2004), in particular left-dislocation and narrow focus intonation on DPs, are available to mark subjects: (40) (Context: Which neighbors can come?) Paul peut venir mais Marie ne peut pas. mais Marie)RG L%]IntP ne peut pas) RG L%] IntP. Paul can come but Marie sm can not So, why can’t subjects also resort to F-marking in ACD constructions? Because, we claim, restrictive relatives are not on the same information level as their main clauses: be they elided or not, the proper ways to F-mark subjects are not available for them. This is the case with contrastive topics, marked by left dislocated subjects, which require parallel information structures (Gergel et al. 2007):13 (41) *Lea lit tous les livres que Tom, il ne peut pas lire. Lea reads all the books that Tom, he sm can not read ‘Lea reads all the books that tom can’t read’ (42) *Lea lit tous les livres que Tom, il ne peut pas. Lea reads all the books that Tom, he sm can not ‘Lea read all the books that tom can’t’ DP subjects with a narrow focus intonation cf.€(43), or with a focus-sensitive particle as in (44), are also impossible in non-elided restrictive relatives, as the subject can’t correspond to the WH-part of a related question, contrary to what happens to the conjunct in (40) – only Lea, the main subject, can: (43) Lea lit tous les livres que Tom ne peut pas lire a. #Tom)RG L%]IntP ne peut pas lire) RG L%] IntP b. Tom) RG n’a pas pu lire) RG L%] IntP 13. Imponing such a focus structure on the relative may be possible but turns it into a nonrestrictive one, irrelevant here.
Anne Dagnac
(44) *Lea lit tous les livres que Tom aussi peut lire *Lea reads all the books that Tom also can read. So, in the elided relative in (36/38), the subject has no way out: the narrow focus intonation that would F-mark the disjunct subject (cf.€45a) is ruled out just as in its non-elided version in (43a). Only the broad focus intonation found in (43b) is possible, but it doesn’t F-mark the subject, and (45b), the elided version of (43b), violates the identity condition. (45) *Lea lit tous les livres que Tom ne peut pas. a. *Tom)RG L%]IntP ne peut pas) RG L%] IntP b. *Tom) RG ne peut pas) RG L%] IntP No device to (properly) F-mark subjects being available in restrictive relatives, ACD subjects can only corefer to the antecedent subject not to violate the general identity condition on ellipsis.14 Since this unavailability stems from the infornational status of restrictive relatives, Spanish and Italian are then expected to show such a restriction. 6. Conclusion French, Spanish and Italian pattern together in licensing, after modals, gaps that display a full range of island-sensitive extractions, in particular relativization and WH-movement, as long as the relevant semantic constraints are met. This casts strong doubt over a null proform analysis, that would be unable to provide a base position for the moved items. But they also differ from English VP-ellipsis: they cannot strand any item higher than VP, they display the ‘Same Subject Constraint’ 14. To my knowledge, the constraint on subjects in ACD modal ellipsis had gone unnoticed so far. The data and this first proposal may need refining. Some speakers seem to unevenly accept some cases of disjunction. One reviewer finds (i), marginally acceptable.
(i) ?A Juan le gusta comer todo lo que Pedro no puede Its French counterpart (ii), with an infinitive in subject position, is also marginally acceptable, and some French speakers accept (iii), where the relative subject is a contrastive pronoun with a somewhat underdocumented status:
(ii) ?Manger tout ce que Lea ne peut pas plaît beaucoup à Tom
(iii) %Je porterai tous les cartons que lui ne peut pas. These cases may suggest that the status of the subject/antecedent matters for the proper formulation of identity conditions, or that the claim that subjects can’t escape identity via focus in restrictive relatives is too strong: some kind of F-marking (but not all) may be available to contrast (some types of) subjects in ACD constructions. If our proposal can be maintained, this constitutes a field of investigation per se.
Modal ellipsis in French, Spanish and Italian
in ACD and they rule out voice mismatches. I propose that all these differences stem from the fact that ‘Modal Ellipsis’ deletes a TP while English VPE deletes a mere VP, a direct consequence of the different status of the modals in these languages. The SSC consists in the extension of the usual identity condition on elided material to the copy of the raised subject. Within this view, Lobeck’s (1995) claim that these Romance languages don’t display VPE remains fully grounded: they only allow for TPE. But the existence of an instance of TP-ellipsis with (nonauxiliary) deontic modals brings new material to the on-going debate on how ellipsis is licensed. References Abeillé, Anne & Danielle Godard. 2003.“Les Prédicats complexes dans les langues romanes”. Les Langues romanes: problèmes de la phrase simple. 125–184. Paris: CNRS Editions. Baltin, Mark.2007. “Deletion Versus Pro-Forms: A False Dichotomy?”. Ms., New York University. Busquets, Joan & Pascal Denis. 2001. “L’Ellipse modale en français: le cas de pouvoir et devoir”. Cahiers de Grammaire26. 55–74. Cardinaletti, Anna & Ur Shlonsky. 2004. “Clitic Positions and Restructuring in Italian”. Linguistic Inquiry 35:4. 519–557. Cecchetto, Carlo. & Orin Percus. 2006.“When we do that and When we don’t: a contrastive analysis of VP-ellipsis and VP-anaphora”. Phases of Interpretation, ed. by M. Frascarelli. 71–105. Berlin/New York: Mouton de Gruyter. Collins, Chris. 2005. “A Smuggling Approach to the Passive in English”. Syntax 8:2. 81–120. DeCat, Cécile. 2004. “Early Pragmatic Competence and its implication regarding the null subject phenomenon”. Romance Languages and Linguistic Theory 2002 ed. by R.Bok-Bennema et al. 17–32. Amsterdam/Philadelphia: John Benjamins. Demirdache, Hamida & Miriam Uribe-Etxebarria. 2002. “La Grammaire des prédicats spatiotemporels: temps, aspect et adverbes de temps”. Temps et Aspect ed. by Brenda Laca. 125–176. Paris: Presses Universitaires de Vincennes. Depiante, Marcela. 2001. “On Null Complement Anaphora in Spanish and Italian”. Probus13. 193–221 Fox, Danny. 2002. “Antecedent Contained Deletion and the Copy Theory of Movement”. Linguistic Inquiry 33: 63–96. Frazier, Lynn & C. Clifton. 2006. Ellipsis and Discourse Coherence. Linguistics & Philosophy 29: 315–346. Gergel, Remus et al. 2007. “Ellipsis and Inversion: a feature-based approach”.On Information Structure, Meaning and Form, ed. by Kerstin Schwabe & Susanne Winkler. Amsterdam/ Philadelphia: John Benjamins. Hankamer, Jorge & Ivan Sag. 1976. “Deep and Surface Anaphora”. Linguistic Inquiry 7:3. 391–428. Johnson, Kyle. 2001. “What VP Ellipsis Can Do, and What it Can’t, but not Why”. The Handbook of Contemporary Syntactic Theory ed. by Mark Baltin and Chris Collins. 439–479. Blackwell Publishing. Johnson, Kyle. 2004. “How to be Quiet”. Ms, University of Massachusets.
Anne Dagnac Kehler, Andrew. 2000. “Coherence and the Resolution of Ellipsis”. Linguistics and philosophy23: 533–575. Lappin, Shalom. 1996. “The Interpretation of Ellipsis”. The Handbook of Contempory Semantic Theory,ed. by Shalom Lappin. 145–175. Blackwell Publishing. Lobeck, Anne. 1995. Ellipsis: Functional Heads, Licensing and Identification. New York/Oxford: Oxford University Press. Merchant, Jason. 2001. The Syntax of Silence. New York/Oxford: Oxford University Press. Merchant, Jason. 2007. “Voice and Ellipsis”. Ms, University of Chicago. Merchant, Jason. 2008a. “An asymmetry in voice mismatches in VP-ellipsis and pseudogapping”. Linguistic Inquiry 39:1. 169–179. Merchant, Jason. 2008b. “Variable Island Repair under Ellipsis”. Topics in Ellipsis, ed. by Kyle Johnson. 132–153. Cambridge/New York: Cambridge University Press. Park, Myung-Kwan. 2004. “Phases, cyclicity, and extraction out of ellipsis in English”. Studies in Generative Grammar 14: 4. 469–497. Rooryck, Johann. 1994. “Generalized Transformations and the WH-cyle: free relatives as bare WH-CPs”. Minimalism and Kayne’s Asymmetry Hypothesis, ed. by Jan-Wouter Zwart. Groningen: University of Groningen. 195–208. Ruwet, Nicolas. 1972. Théorie syntaxique et syntaxe du français, Paris: Seuil. Sag, Ivan. 1978. “Floated Quantifiers, Adverbs, and Extraction Sites”. Linguistic Inquiry9:2. 146–150. Sauerland, Uli. 2000. “Obligatory reconstruction and the meaning of traces”. Arbeitspapiere des SFB 340, Bericht Nr153. 1–28. Sauerland, Uli. 2004. “The Interpretation of traces”. Natural Language Semantics 12. 63–127. Wurmbrand, Susanne. 2001. Infinitives: Restructuring and Clause Structure. Berlin/New York: Mouton De Gruyter. Zanuttini, Raffaella. 2001. “Sentential Negation”. The Handbook of Contemporary Syntactic Theory ed. by Mark Baltin and Chris Collins. 511–535. Blackwell Publishing.
Optional prepositions in Brazilian Portuguese* Mary A. Kato This paper discusses the phenomenon of preposition optionallity in Brazilian Portuguese (BP), starting from Bouchard’s (1981) observation that the preposition of a strictly subcategorized PP complement in French can be absent in the “chopping” type of relativization, though the same sort of “deletion” is ruled out in wh-questions, a contrast that leads him to propose that movement is absent in such relatives. My aim in this paper is (a) to show that this phenomenon is not restricted to relative clauses, (b) to propose a uniform analysis of preposition optionallity in several domains, (c) to argue that prepositions which encode inherent case are optional in the numeration, and (d) to claim that absence of the preposition involves only A’-positions, where the DP can have a “default” case. The paper ends up with a discussion on contrastive topicalization, assumed to be derived from VP-topicalization.
1. Introduction The phenomenon of optional prepositions in Brazilian Portuguese (BP) has been extensively studied in Tarallo’s (1983) work on relative clauses, inspired by Bouchard’s (1981) paper on a similar phenomenon in French. The aim of the present study is to show that this phenomenon is not restricted to relative clauses, and to propose a uniform analysis of this optionallity in several domains: topic positions, relative clauses, and cleft constuctions. The research questions are the following: a. what sort of prepositions can be optional in BP? b. what positions license optionallity of prepositions in BP and why? * This paper had the support of CNPq grant 301219/2008 and FAPESP 2006/00965–2. I thank the valuable contributions of Juanito Avelar, Heloisa Salles, Jairo Nunes, Marcello Marcelino and Paulo Medeiros. I also owe an enormous debt to Bouchard’s comments on the first version of the paper. The usual disclaimers apply.
Mary A. Kato
The paper is organized as follows: in Section€2, Bouchard’s theory of “prepositions that delete” will be presented; in Section€3, I present similar constructions in BP, and show that in this language the phenomenon is not restricted to relative clauses. In Section€ 4, I will outline my analysis, answering questions (a) and (b) above. In Section€5, following Kato and Raposo (2007), I will show that PP topicalization can result not from PP-complement topicalization, but from VP remnant topicalization, with the copy of the verb erased, in which case the preposition is necessarily present as it is merged in an A-position. 2. “Prepositions that delete” in French Bouchard (1981) observed that the preposition of a PP complement in French can be absent in certain types of relativization (cf.€(1)). (1) Le gars que je te parle, il est correct. the guy that I cl.2ps talked he is ok ‘The guy that I talked to is ok.’ He claims that only elements that are semantically related to the verb can have the preposition “deleted”. Thus, in (2a), the verb couper ‘cut’ subcategorizes the instrumental PP while in (2b) the verb manger ‘eat’ does not.1 (2) a. Le couteau que je coupe the knife that I cut ‘The knife that I cut with…’ b. *La cuillere que je mange est trop grosse. the spoon that I eat is too big But, as he shows, for (3a) the only reading is (3b) and not (3c). With the object expressed (le fromage) in (3d), on the other hand, the only interpretation available is that of an instrumental for fil, with the preposition avec deleted. His conclusion (p.€220) is that “although nonstrictly subcategorized preposition are not recoverable like strictly subcategorized preposition by the lexicon, they may be deleted
1. Bouchard shows that (2b) would become well formed if instead of preposition “drop”, we have stranding, allowed with certain prepositions.
La cuileere que je mange avec est trop grosse
P allows stranding only with sem (without) and contra, a minor phenomenon compared to B lack of prepositions. In this analysis we do not include the sporadic cases of stranding. But I subscribe to Bouchard’s analysis that in these types of preposition stranding there is a pro.
Optional prepositions in Brazilian Portuguese
and are recoverable if they have a semantic relation to the verb”, a notion that I update here as Thematic. (3) a. Le fil que je coupe b. Le fil que je coupe [e] “The wire that I cut [e]”
(direct object)
c. Le fil que je coupe avec “The wire that I cut with” d. Le fil que je coupe le fromage Comparing constructions such as (1) and (2) with wh-questions such as (4), Bouchard proposes that this type of relative clauses does not involve wh-movement and that wh-questions cannot have the preposition deleted, exactly because it involves wh-movement. (4) a. *Qu’est-ce que tu coupes le fromage? what is it that you cut the cheese b. Avec quoi est-ce que tu coupes le fromage? with what is it that you cut the cheese ‘What do you cut the cheese with?’ 3. Prepositions that “delete2” in Brazilian Portuguese (BP) In consonance with Bouchard’s analysis, Tarallo (1983) argues that preposition-less relatives in BP are possible with subcategorized PPs (cf.€ (5)), and do not occur throughout wh-movement (cf.€(6)). Therefore, he concludes that preposition-less relatives, or chopping relatives, in Brazilian Portuguese, do not involve wh-movement. (5) a. A faca que eu cortei o queijo the knife that I cut the cheese ‘The knife that I cut the cheese with’ b. *A faca que eu comi o queijo the knife that I ate the cheese ‘The knife that I ate the cheese with’ (6) a. *O que que você cortou o queijo? what that you cut the cheese
2. Barbaud (1998) calls the process “preposition absorption”, a consequence of lack of preposition stranding.
Mary A. Kato
b. Com o que que você cortou o queijo? with what that you cut the cheese ‘What did you cut the cheese with?’ There are two major problems with pairing up BP preposition-less relatives to the French ones. First, it is not the case that BP allows all subcategorized, or semantically related PPs, to have its preposition “deleted” in the process of relativization, as the examples in (7) show3. (7) a. A peça *(em) que eu pensei era muito boa. the play (prep) which I thought was very good ‘The play that I thought of was very good.’ b. O fundo *(de) que eu dependo não foi liberado. the fund (prep) that I depend not was released ‘The fund that I depend on was ot released.’ Second, lack of wh-movement cannot account for the possibility of prepositions absence, since in BP they can be absent in D-linked wh-questions, as shown in (8). (8) a. (De) que livro você vai precisar? (prep) which book you will need ‘Which book you will need?‘ b. (Com) que faca você cortou o queijo? (prep) which knife you cut the cheese ‘Which knife did you cut the cheese with?’ However, with verbs that do not license the absence of prepositions in relatives, D-linked wh-questions are also unacceptable without pied piping: (9) a. *(Em) que peça você pensou? (prep) which play you thought ‘Which play did you think of? b. *(Com) que moça o Pedro casou? (prep) which girl the Pedro married ‘Which girl has Peter married?’ Interestingly, the same set of verbs that allow absence of prepositions in relative clauses are the ones that license absence of prepositions in topic constructions, as can be seen in (10): 3. Some dialects of Brazilian Portuguese show up a wider set of verbs that allow “prepositions to delete” than the São Paulo dialect, which is the one considered in this paper. European Portuguese has an even smaller set of verbs, according to Duarte (1987).
Optional prepositions in Brazilian Portuguese
(10) a. (D’) essa música, nem todo mundo gosta. (prep) this music not everybody likes ‘This music, not everybody likes.’ b. A música (de) que nem todo mundo gosta the music (prep) that not everybody likes ‘The music that not everybody is fond of (11) a. (Com) aquela moça eu conversei ontem na festa. (prep) that girl I talked yesterday at the party ‘ That girl, I talked with her yesterday at the party.’ b. A moça (com) que(m) eu conversei na festa the girl (prep) that I talked at the party ‘The girl that I talked with at the party’ (12) a. *(Em)aquela peça, eu pensei todos os dias. (prep) that play I thought every the days “About that play, I thought everyday.” b. A peça *(em) que eu pensei todo dia the play prep which I thought everyday ‘The play which I though about everyday’ (13) a. *(Com) essa moça, o Pedro casou este ano. (prep) this girl the Peter married this year ‘To this girl, Peter got married this year.’ b. A moça *(com) que(m)4 o Pedro casou este ano. the girl (prep) who the Peter married this year ‘The girl to whom Peter got married this year’ In contrast with topicalization, Left Dislocation (LD) must have the preposition absent in Topic position independently of the verb type: (14) a. Aquela moça, eu conversei com ela na festa. that girl I talked with her at the party ‘That girl, I talked with her at the party.’ b. Aquele cara, eu ri dele na festa. that guy I laughed at him at the party ‘That guy, I laughed at him at the party.’
4. Brazilian Portuguese has a tendency to neutralize the relative pronoun quem and qual to que, in the presence of prepositions. This shows that relative pronouns are being neutralized in most functions, except in the locative, time and manner functions.
Mary A. Kato
In addition to topicalization, relativization and D-linked wh-questions, BP also exhibits the phenomenon of optional prepositions in cleft sentences5: (15) a. Foi (com) essa professora que eu conversei ontem. was (prep) this teacher that I talked yesterday ‘It was with this teacher that I talked yesterday.’ b. É (d)este livro que eu vou precisar para o teste. is (prep) this book that I will need for the test ‘It is this book that I will need for the test.’ 3. Proposals 3.1
The prepositions that can be optional
We assume with Bouchard (1981) that PPs that are not subcategorized by the verb cannot often have the preposition optionally realized in the constructions analyzed above. Bouchard gives an example with the preposition avec, which can “delete” if the PP is subcategorized (like with couper), but not when the PP is an adjunct. The same seems to happen with the preposition para, which can be absent if it belongs to an indirect object, but not when it is part of a benefactive: (16) a. A garota que o Pedro deu um doce the girl that the Peter gave a cookie ‘The girl that Peter gave a cookie’ b. *O mendigo que o Pedro comprou um sanduíche the beggar that the Peter bought a sandwich “The beggar to whom Peter bought a sandwich” Just as Bouchard, we argue that strict subcategorization is a favourable condition, but that´s not the whole story. Tarallo did not find any natural class of verbs that licence prepositions “to delete” in BP; he simply stated the list of verbs that are common in chopping relatives. Considering that licensing of an optional preposition is a lexical idiosyncratic matter, associated to particular θ-roles, Kato and Nunes (2008) propose, in
5. Barbaud (1998) shows that the phenomenon is present in advanced French: Lê traffic dês stupedifiants, cést ce qu’on (SKE) voulait mettre fin aujourd’hui C’est ça que j’aurais que tu me parles
Optional prepositions in Brazilian Portuguese
their analysis of relative clauses, that prepositions can be optional when they are inherent case assigners6. In fact, other contexts where the absence of the preposition can be noted is exactly where inherent case has been proposed (see Chomsky 1986): in PPs that are genitive complements (cf.€(17)), and in PPs that are the indirect objects (cf.€(18)): (17) a. Eu conheço o pai desse menino desde o Natal. I know the father of-this boy since the Xmas ‘I know this boy’s father since Xmas.’ b. (D) Esse menino, eu conheço o pai desde o Natal. this boy I know the father since the Xmas ‘This boy, I know his father since Xmas.’ c. (De) que menino que você conhece o pai? (prep) which boy that you know the father ‘Which boy do you know the father of?’ d. É este o menino (de) que eu conheço o pai. is this boy that I know the father ‘It is this boy whose father I know.’ (18) a. Todo mundo deu donativos para a unicef. everybody gave donations to the unicef b. (Para) a unicef, todo mundo deu donativos. (prep) the unicef everybody gave donations ‘Everybody made donations to unicef’ c. (Para) que entidade que todo mundo deu donativos? (to) what entity that everybody gave donations ‘What entity did everybody donate money to?’ d. A entidade (para) que todo mundo deu donativos the entity (prep) that everybody gave donations ‘The entity that everybody donated money to’ With oblique complements, which Chomsky also proposes to have inherent case, the situation seems to be more complicated, as observed not only cross-linguistically (French vs BP), but also among dialects of Portuguese. In short, for Chomsky (p.€193) the distinction between structural case and inherent case is that while the former is assigned at S-structure, the latter is 6. If we assume a hierarchy of θ-roles (theme > goal > locative > instrumental > benefactive), it may be the case that inherent case splits the hierarchy in different places according to speaker or dialect. Thus the goal would be easier to have prepositions absent in the numeration than benefactives. I thank Cristina Ximenes for this suggestion.
Mary A. Kato
associated with θ-marking, at D-structure. With the disappearance of such levels, by the advent of Minimalist concerns, we find ourselves in need to propose some other place where the distinction between the two types of case might be established. Positions where prepositions can be optional
3.2
3.2.1 A and A’-positions I propose that non-inherent, or ‘real’ prepositions, are obligatory in the numeration, but ‘inherent’ case marking prepositions (ICP) are optional. For both cases, the presence of the preposition in the numeration makes its presence obligatory in both A and A’-positions. Thus, if the preposition is present in the numeration and is merged with a DP, the resulting PP may remain in A-position, or be moved to an A’-position as a PP (cf.€(19)). (19) a. A Maria precisa d[esse dinheiro]inherent case para hoje. the Maria needs of-this money for today ‘Maria needs this money for today.’ b. D[esse dinheiro]i inherent case, a Maria precisa [PPti] para hoje prep-this money the Maria needs for today ‘This money, Maria needs it for today.’ By contrast, if the ICP is absent in the numeration, the relevant DP remains with its case unvalued (cf.€Chomsky 2001), and the derivation crashes (cf.€(20a)), unless the DP is moved to an A’-position (cf.€(20b)), where it can be assigned a ‘default’ case (cf.€Kato, 1999; Schütze, 2001) – which in BP is assumed to be nominative (Kato, 1999)7. (20) a. *A Maria precisa [esse dinheiro]? case para amanhã. the Maria needs this money for tomorrow b. [Esse dinheiro]i.default.nom a Maria precisa [DPti] para amanhã. This money the Maria needs for tomorrow ‘This money, Maria needs it for tomorrow.’ A different strategy is Left Dislocation, analyzed in the old tradition as a construction where the DP-topic is merged in-situ, also obtaining the ‘default’ case. 7. Schutze (2001) introduces the concept of “default” case using Germanic languages and morphological case. Kato (1999) uses strong pronouns, due to the fact that in Romance case is morphological only in pronouns. She shows that the “default “ case is nominative in Null Subject languages, but can be different in other languages, like the dative in French and the accusative in English.
i.
Me, I like bear.
ii. Moi, j’aime de bière. iii. YO, me gusta la cerveza.
Optional prepositions in Brazilian Portuguese
The resumptive is merged in an A-position, and, therefore, exhibits the preposition (21b). We will assume, moreover, that – when the complement is null – it is the result of some sort of ellipsis, represented here as Ø, as PPs have no corresponding proforms (see Kato 1993). This yields an output which is identical to topicalization (22b). In contexts of islands, however, the structure involved can only be obtained by Left Dislocation (23b), with a PP or a null complement: (21) a. Esse dinheiro, eu vou precisar dele para o teste. this money I will need of-it for the test ‘This money, I will need it for the test.’ b. [TopP Esse dinheiro+nom”default”]i eu vou precisar dele+inherent case (22) a. Esse dinheiro, eu vou precisar Ø para o teste b. [TopP Esse dinheiro+nom”default”]i eu vou precisar desse dinheiro+inherent case para o teste. (23) a. Esse dinheiro, eu encontrei o amigo que vai precisar dele. b. Esse dinheiro, eu encontrei o amigo que vai precisar Ø. 3.2.2 Questions with and without pied-piping D-linked questions can be preposition-less (cf (24a) while non D-linked questions cannot (see (24b)) because the D-linked expression merges as a Left Dislocated constituent (LD), after which it moves to CP8, just as shown in (25). In the lower LD position, the D-linked wh-constituent can get “default” nominative case. (24) a. (Com) que faca (que) você cortou o queijo? with which knife that you cut the cheese b. *(Com) o que (que) você cortou o queijo? with what that you cut the cheese (25) a. [LD [que faca +F.default case]i[IP você cortou o queijo com que faca]] b. [CP [que faca+F, +”default case]i que+F [LD ti [IP você cortou o queijo com que faca]] As non-D-linked expressions are not LD positioned, they require case to be assigned in A-position by the preposition in order to have wh-movement to CP directly, triggered by que+F: (26) a. [IP você cortou o queijo com que+F. inherent case] b. [CP Com que+F. inherent case [queF [IP você cortou o queijo] 8. I thank one of the reviewers who suggested that the topic would be merged in-situ, as LD, and not moved from an A-position.
Mary A. Kato
3.2.3 PP-focus and DP-focus in clefts Cleft sentences are also claimed to involve movement of the focalized DP to an A’-position (cf.€Kato & Ribeiro 2007), and: a. verbs that select prepositions optionally in the numeration are the ones that license focus with or without preposition in cleft structures (cf.€27a vs. 27b); (27) a. Foi (com) essa moça que eu conversei ontem. was with this girl that I talked yesterday ‘It was with this girl that I talked yesterday.’ b. Foi *(d) esta piada que todo mundo riu. was (prep) this joke that everybody laughed. ‘It was this joke that everybody laughed’. b. the position of the focalized element in cleft sentences is assumed to be the FocP of the lower clause (cf.€Belletti, 2001), which functions as the complement of the copula: (28) copula [FocP PP/DP [que+F [IP……….. c. if the preposition has been merged with the DP, the focalized PP can move directly to FocP, with a head que containing a +F feature, (cf 29a); d. if the preposition is not selected in the numeration, the DP to be clefted, with its case unvalued, moves to the FocP position, where it can be assigned the default nominative case and where it can check the +F feature against the +F complementizer que (cf.€29b). e. In both cases, the resulting FP is merged with the copula. (29) a. Foi [FocPcom [essa moça]i. inherent case que+F [TPeu falei [PPti]....] b. Foi [FocP[essa moça]i. default nom que+F [TPeu falei [DP ti]…]] f. The fact that the cleft element results from movement finds evidence in the impossibility of extraction from islands: (30) *Foi [(d)esse livro]i que eu falei com o aluno que precisava [ti] was of-this book that I talked with the student who needed ‘It was this book that I talked with the student who needed it” 3.2.4 The derivation of relative clauses BP has three types of relative clauses: i) the standard relatives ii) the resumptive relatives and iii) the chopping relatives (cf.€Tarallo 1983). Verbs that select optional prepositions in the numeration can have the three types of relatives. If the preposition is selected in the numeration, two outputs can
Optional prepositions in Brazilian Portuguese
result. The preposition is merged with the wh-element inside IP. The wh-element is extracted by relativization with obligatory pied-piping of the preposition (see 32a). In the second alternative, a resumptive pronoun is present in the numeration and the preposition merges with the pronominal complement. A co-referent DP merges at LD. Extraction takes place from this position (see 32b). If the preposition is not chosen, the DP complement has to move to TopP, where it gets nominative by “default”. From this position relative extraction takes place9. (31) a. O autor *(com) quem falei (standard relative) the author prep who I talked ‘The author who I talked with’ b. O autor que eu falei com ele (resumptive relative) the author who I talked prep him ‘The author that I talked with’ c. O autor que eu falei (chopping relative) the author who I talked ‘The author that I talked with’ (32) a. [DP o autor [CP com quemi [IP eu falei ti]]] b. [DP o autor [CP quei [LD ti. default nom [eu falei com elei] c. [DP o autor [CP quei [TopP ti. default nom [eu falei [DP ti] Verbs that select complements with obligatory prepositions can occur in standard relatives, with the preposition pied-piped governing the relative pronoun, or in resumptive relatives. However, chopping relatives are out as they require absence of the preposition in the numeration10. (33) a. A moça com quem ele vai casar “The girl with whom he will marry” b. A moça que ele vai casar com ela the girl that he will marry with her “The girl that he will marry c. *A moça que ele vai casar Ø the girl that he will marry “The girl that he will marry with”
9. When derivation starts from islands, the only possibility is with resumptive relatives or with disguised chopping relatives with an elliptical site in the PP position. 10. This shows that chopping relatives cannot be always considered a resumptive relative with an elliptical PP complement.
Mary A. Kato
In Bouchard’s analysis for French and Tarallo’s (1983) for Brazilian Portuguese the que in preposition-less chopping relatives is considered a complementizer. In the present analysis, I assume with Kato (1993) that que is a nominative relative pronoun and that chopping and resumptive relatives are derived as wh-relatives. Moreover, absence of the preposition would not be due to its “deletion” in the process of relativization, but would be due to the possibility of its absence in the site of extraction, namely, at TopP or LD. 4. Contrastive topicalization cannot have optional prepositions Kato & Raposo (2007) have observed that topicalization can be contrastive or non-contrastive, depending on the context question. Their claim is that, while argument topicalization yields non-contrastive topicalization, VP-topicalization11, in both its “remnant and non-remnant versions, yield contrastive topicalization12 . (34) a. –Onde é que (vocês) encontraram este cd? “Where is it that (you-pl) found this cd” b. –Este cd, (nós) só encontramos ec na fnac. ‘This cd we only found (it) at-the fnac’
(non-contrastive)
(39) a. –Vocês encontraram os cds e os livros que (vocês) queriam? you found the cds and the books that you wanted? ‘Have you found the cds and the books that you wanted?’ b. –(Encontrar) os cds, (nós) encontramos ec, mas não os livros (find) the cds we found but not the books que nós queríamos (contrastive) that we wanted ‘(Find) the cds we did, but not the books that we wanted’ As in VP-topicalization the complement merges in A-position obtaining inherent case, the preposition has to be pied-piped in topicalization, whether the verb is expressed or erased.
11. Bastos (2001) had already noticed that VP-topicalization was a contrastive construction, but she did not consider VP-remnant topicalization. 12. Kato and Raposo also demonstrate that argument topicalization shows islands effects, while VP-topicalization does not. That VP-topicalization exhibits no subjacency effects had already been observed in Matos (1992).
Optional prepositions in Brazilian Portuguese
(40) a. Eu preciso d- este livro para o teste, mas não daquele. I need (prep)- this book for the test, but not the other–one. ‘ I need this book for the test, but not he other one.’ (41) a.
(Precisar) *(d)este livro, eu preciso para o teste, mas não need (prep)-this book I need for this test, but not daquele. (VP-topicalization) that-one ‘ Need this book, I will for the test.’.
If we had a perfect contrast, PP with contrastive topicalization and DP with noncontrastive topicalization, we would not need the argument of optionallity of preposition in the numeration. However, when we have non-contrastive topicalization, we can have both PP and DP, while in contrastive topicalization we can only have PP. 5. Conclusions This paper has shown that an ICP can be optional in the numeration, and that, if it is selected, deletion can never occur, either during the derivation or post-syntactically, at PF. If, however, it is not selected, the derivation can attain well-formed outputs provided that the caseless DP moves to the TopP or FocP position, where it can get the “default” nominative case. This Topic can undergo further movements, if it is endowed with features to be checked, like the +wh feature of wh-elements. The paper has also claimed that with contrastive topics the preposition is predicted to be always present as it results from VP-topicalization, in which the complement is merged as a PP, but the copy of V may appear erased. References Barbaud, P. 1998. “Dissidence du français et evolution dialectale”. Revue Québecoise de Linguistique, 26(2). 107–128. Bastos, A.C. 2001. Fazer, eu faço! Topicalização de constituintes verbais em português brasileiro. Ms., Master’s thesis, Universidade de Campinas. Belletti, A. 2001. “Inversion as focalization”. Subject Inversion in Romance and the Theory of Universal Grammar. Ed. by A.Hulk & J.Y. Pollock. 60–90. New York: Oxford University Press. Bouchard, D. 1981. “An alternative to wh-movement in French relative clauses”. In: W.W. Cressey & D.J.Napoli (eds) Linguistic Symposium on Romance Languages: 9. 216–226. Washington, DC: Georgetown University Press. Chomsky, N. 1986. Knowledge of Language: Its Nature, Origin, and Use. Praeger, New York.
Mary A. Kato Chomsky, N. 2001. “Derivation by Phase”. Ken Hale: a Life in Language. Ed. by Michael Kenstowski, 1–52. The MIT Press, Cambridge, Mass.: The MIT Press. Duarte, Inês. 1987. A Construção de Topicalização na Gramática do Português. Regência, Ligação e Condições sobre Movimento. Ph.D. Dissertation. University of Lisbon. Kato, M. A. 1993. “Recontando a história das relativas”. In: Português Brasileiro: Uma Viagem Diacrônica. Ed. by I. Roberts and M. A. Kato, pp.€223–261. Campinas: Editora da UNICAMP. Kato M.A. 1999. “Strong pronouns, weak pronominals and the null subject parameter”. PROBUS, 11 (1), 1–37. Kato, M.A. & Nunes, J. 2009 “A uniform raising analysis for standard and non-standard relative clauses in Brazilian Portuguese”. In: Minimalist Essays on Brazilian Portuguese Syntax. Ed. by J. Nunes. 93-120. Amsterdam: John Benjamins. Kato, M.A. & Ribeiro, I. 2007. “A evolução das estruturas clivadas no português”. In: Para a História do Português Brasileiro. A experiência dos grupos de estudo. Tomo I. v. 6. Ed. by T.Lobo; I.Ribeiro; Z. Carneiro; N. Almeida.: 165–182. Salvador: EDUFBA. Kayne, R. 1994. The Antisymmetry of Syntax. Cambridge, Mass.: The MIT Press. Matos, M. Gabriela. 1992. Construções de Elipse de Predicado em Português: SV Nulo e Despojamento. Universidade de Lisboa: Doctoral Dissertation. Rizzi, L. 1997. “The Fine Structure of the Left Periphery”. In: Elements of Grammar: Handbook of Generative Syntax. Ed. by L. Haegeman, 281–337. Dordrecht: Kluwer Schütze. C. 2001. “On the nature of default case”. Syntax, 4:3. 205–238. Tarallo, F. 1983. Relativization Strategies in Brazilian Portuguese. Ph.D. dissertation. University of Pennsylvania.
An apparent ‘Number Case Constraint’ in Romanian The role of syncretism* Andrew Ira Nevins & Oana Săvescu Harvard University & NYU
Romanian pronominal plural clitics differ from their singular clitic counterparts in that they exhibit dative-accusative case syncretism. This contrast correlates with an asymmetry in the combinatorial possibilities of plural vs singular clitics: namely, plural clitics in direct object position in Romanian affect the acceptability of clitic clusters, as confirmed in quantitative acceptability judgements with multiple informants. Rather than invoking a new ‘Number Case Constraint’ governing the distribution of clitics, we link the Romanian data to familiar facts from Leísta dialects of Spanish, which manifest case syncretism between dative and animate accusative 3rd person clitics. We implement the fact that 1st and 2nd person plural clitics in Romanian are case-syncretic by implicationally marking them as inherently [+animate] in the syntax. The severe degradation in acceptability of clusters with direct object plural clitics is accounted for by following aspects of Adger & Harbour’s (2007) proposal for the connection between syncretism and the Person Case Constraint.
1. Introduction In this paper we examine the role of a paradigmatic property, namely the existence of clitic case syncretism, on a syntagmatic property, namely clitic co-ocurrence restrictions. We focus our discussion on Romanian, which is unique among modern Romance languages in that its singular pronominal clitics manifest a dative-accusative case distinction: the 1st person singular clitic, for instance, is * This paper was presented at LSRL 38 in Urbana-Champaign, April 2008. We would like to thank the following people for discussion: Peter Arkadiev, Karlos Arregi, Cristina Cuervo, Carmen Dobrovie-Sorin, Peter Graff, Daniel Harbour, and Pascual Masullo.
Andrew Ira Nevins & Oana Săvescu
mi in the dative case and mă in the accusative. Plural clitics, however, are casesyncretic (ne is 1st person plural DAT/ACC, vă is 2nd person plural DAT/ACC), thus yielding a singular-plural asymmetry in the Romanian clitic paradigm.1 We show that this asymmetry actually correlates with a contrast in the combinatorial possibilities of singular and plural clitics and we present evidence for a novel discovery in Romanian: the existence of co-occurrence restrictions on clusters with accusative clitics that are plural. The paper is structured as follows: Section€2 presents a brief overview of the combinatorial possibilities of singular clitics, which are contrasted in Section€3 with the behavior of plural clitic clusters. The latter seem to speak in favor of a new constraint governing the distribution of clitics in Romanian, namely a `Number Case Constraint’ (modeled along the lines of the Person Case Constraint familiar from studies of Romance clitic clusters). The fact that plural clitics in the direct object position significantly affect the grammaticality of clitic clusters is substantiated in Section€4 by a quantitative analysis of judgments of 17 native speakers of Romanian. In Section€5 we show that the Romanian plural data can be assimilated to familiar facts from Leísta dialects of Spanish, and therefore should not be understood in terms of a Number Case Constraint. Rather, we take the morphological syncretism of Romanian 1st and 2nd person plural clitics and Leísta 3rd person clitics to be evidence for the language learner of an underlying featural identity (animacy) which conditions a shared vocabulary item. The severe degradation in acceptability of clusters with direct object plural clitics in Romanian is then accounted for by following aspects of the syntax of Applicatives, interacting with selectional properties of the syntax-semantics interface (Adger and Harbour, 2007) thereby deriving the Number Case Constraint in Romanian and animacy co-occurrence restrictions in Leísta Spanish through the interaction of independently motivated grammatical principles (Sections€6 and 7). Section€8 presents a set of general conclusions on the nature of clitic cooccurence restrictions and the role of syncretism. 2. Combinatorial restrictions of Romanian singular clitics In clusters, Romanian pronominal clitics unambiguously exhibit the order dative accusative, both in preverbal and in post-verbal position. However, even when the dative accusative order is respected in Romanian, not all combinations of clitics 1. Romanian clitics are independently subject to a vowel-height change in clitic clusters, the same morphological process at work in Italian; cf.€ ci vs. ce lo (Cinque, 1994, Cardinaletti, 2008), and thereby one must look at Dative clitics in isolation to diagnose syncretism.
An apparent ‘Number Case Constraint’ in Romanian
with respect to person are possible. The data in (1)-(2) show that Romanian is subject to the Me-First combinatorial restriction, which disallows a 2nd person dative clitic in the presence of a 1st person accusative clitic: (1) *Ţi m- a prezentat Ion la petrecere. 2sg.dat 1sg.acc has introduced John at party ‘John has introduced me to you at the party’.
(2sg.dat > 1sg.acc)
(2) Mi te- a prezentat Ion la petrecere. 2sg.dat 1sg.acc has introduced John at party. ‘John has introduced you to me at the party’.
(1sg.dat > 2sg.acc)
One exceptional property of Romanian singular clitic clusters is that no person restrictions govern their distribution in post-verbal position, when clitics appear after true imperatives and gerunds (Săvescu, 2007). The ungrammatical example in (2) above thus contrasts sharply with the grammatical examples in (3 a–b), in which the cluster is post-verbal.2 (3) a. Dându- ţi- mă de nevastă, tata a câştigat mulţi bani. giving 2sg.dat 1sg.acc of wife, father has gained much money ‘Giving me to you in marriage, my father has gained a lot of money’. (2sg.dat > 1sg.acc) b. Ia- ţi- mă drept martor, şi vei câştiga procesul! take 2sg.dat 1sg.acc as witness, and will.2nd win trial.the ‘Take me as a witness (for yourself) and you will win the trial!’ (2sg.dat > 1sg.acc) In what follows we will restrict our attention to post-verbal clitic clusters, which are thus less restrictive with respect to the combinatorial possibilities they allow. We also restrict our attention to clusters with combinations of 1st and 2nd person clitics, as clusters with a 3rd person dative clitic are uniformly unacceptable, regardless of number, and clusters with a 3rd person accusative clitic are uniformally acceptable, regardless of number. 3. An apparent number case constraint in Romanian When we consider combinations of plural clitics, a different picture emerges. While a singular 2nd person dative clitic can freely combine in nonfinite 2. Nevins (2007) argues that the Romanian *2>1 restriction is due to an Agreement intervention for a marked [+author] feature. Gerunds and Imperatives lack Tense and hence do not agree for a Person feature.
Andrew Ira Nevins & Oana Săvescu
environments with a singular 1st person accusative clitic, changing the number of clitics to the plural results in severe degradation of acceptability: (4) ??/*Dându- vi- ne in grijă, tata s- a simţit uşurat. giving 2pl.dat 1pl.acc in care, father se has felt relieved ‘Entrusting us to y’all, my father felt relieved.’ (5) ??/*Dându- ţi- ne in grijă, tata s- a simţit uşurat. giving 2sg.dat 1pl.acc in care, father se has felt relieved ‘Entrusting us to you, my father felt relieved.’ The fact that clitics should be subject to restrictions of case and person is now a familiar fact in clitic languages, as previous accounts dealing with the Person Case Constraint have revealed (Perlmutter, 1971, Kayne, 1975, Bonet, 1991, Anagnostopoulou, 2003)). The data in (4)-(5) are therefore even more interesting, because they seem to suggest that Romanian exhibits a new type of restriction on its clitic clusters, namely one that is sensitive to number. The existence of a so-called `Number Case Constraint’ has actually not been documented before in the literature outside of Romanian, where it received a brief treatment in terms of empathy in Farkas and Kazazis (1980), who note the ungrammaticality of plural 2Dat1Acc combinations in pre-verbal position.3 In view of the already-existing person restrictions on pre-verbal singular clusters (cf.€(1) above), the data documented by Farkas and Kazazis (1980) are not as surprising as the fact that the plurality of the clitics should still have an effect on the grammaticality of a post-verbal cluster, given that in such non-finite environments singular clitics typically show significantly fewer co-occurrence restrictions. Moreover, as we show in a more finegrained analysis of the pattern of quantitative acceptability judgements in the next section, in post-verbal position, the effect of number in the second clitic position is independent of the person specifications on each clitic. 4. Quantitative acceptability judgments Since these are new observations and also pragmatically infrequent combinations in non-finite clauses, we attempted to confirm the judgements in (4) and (5) on a
3. Heap (1998) discusses the role of plural number with respect to clitic ordering, demonstrating that singular but not plural 1st/2nd person clitics can precede the reflexive clitic in non-standard orderings in Spanish. We leave it open whether these phenomena are related.
An apparent ‘Number Case Constraint’ in Romanian
larger scale.4 We tested the following 6 combinations, with 17 speakers, who rated the acceptability of the forms from 1-–5 (5 highest). (6)
Combinations tested in acceptability judgment study: dându mi te 1sg > 2sg dându ţi mă 2sg > 1sg dându ni vă 1pl > 2pl dându vi ne 2pl > 1pl dându mi vă 1sg > 2pl dându ţi ne 2sg > 1pl
The mean ratings and standard deviations are shown in the table below. The worst combinations were pl-pl.
(7) Summary of acceptability judgments (5 = completely acceptable, 1 = completely unacceptable): Number Person Mean rating St.Dev sg-sg 1>2 4.91 0.36 sg-sg 2>1 4.09 1.73 sg-pl 1>2 3.65 1.91 sg-pl 2>1 3.44 2.10 pl-pl 1>2 2.97 2.12 pl-pl 2>1 2.18 2.02
Statistical effects conducted by ANOVAs revealed a highly significant effect of Number: F(2,96) = 9.546, p < 0.0001. ANOVAs also revealed a non-significant effect for Person: F(1,96) = 2.850, p>0.09 (n.s.), as well as a non-significant effect for Number×Person Interaction: F(2,96) = 0.312, p>0.74 (n.s.). From these results, one can observe that the presence of plural 1st or 2nd person clitic, in particular in DO position, is the strongest factor affecting acceptability of Postverbal Clusters. In the following sections we will propose an account for the ungrammaticality of 1st/2nd person plural clitics in the direct object position of ditransitive constructions, without actually relying upon the postulation of a wholly new Number Case Constraint as a component of universal grammar. As a first step towards the analysis, we show in the following section that the behavior of Romanian plural clitics is reminiscent of already familiar facts from Leísta dialects of Spanish.
4. We did not include 2pl > 1sg or 1pl > 2sg in our original quantitative judgment survey, but future work should directly compare these, along with filler conditions. The second author’s judgement is that pl > sg clusters are much better than sg>pl.
Andrew Ira Nevins & Oana Săvescu
5. A tangent: Leísta Spanish Leísta dialects of Spanish (Landa, 1995, Bleam, 1999, Ormazabal and Romero, 2007) exhibit syncretism between dative and accusative animate 3rd person clitics: they are both le. (8) Le dí un reloj. 3.Dat.Anim gave-1sg a watch. “I gave him a watch” (9) Le vi. 3.Acc.Anim saw-1sg “I saw him”. Compare thus (9) with (10) below, in which the accusative clitic is inanimate: (10) Lo vi. 3.Acc.Inan saw-1sg “I saw it”. We propose that the syncretism exemplified in (8) and (9) provides evidence for the Leísta language learner of an underlying featural identity; this allows them to ‘map backwards’ what the phonological exponent is realizing. Syncretism of A & B when A & B have different case features (and different syntactic functions/patterns) can be taken by learners to mean that A & B share some other feature that phonological realization is eventually sensitive to.5 The syncretism of the two uses of le is due to a shared feature: [+animate], giving rise to the Leísta Spanish Spellout Rule below: (11) Leísta Spanish spellout Rule: [+anim] 3rd clitic → /le/ The result of (11) applies only upon completion of the syntax, when terminal nodes are supplied with phonological content (Halle and Marantz, 1993). Clitic nodes with different case, if they are specified for [+animate], will be ultimately realized with the same phonological form.
5. In cases of syncretism where categories A &B share no obvious feature at all, underspecification is a mechanism that can yield syncretism. Alternatively, when A & B stand in a clear superset-subset relationship with respect to morphosyntactic features, impoverishment of the superset category can yield syncretism. However, given dative/accusative syncretism for animate arguments, the more direct `shared feature’ approach to syncretism, as posited in the text, is arguably the preferred analytic option for the learner.
An apparent ‘Number Case Constraint’ in Romanian
5.1
Romanian clitic case syncretism
Returning to Romanian, recall that plural clitics, unlike singular ones, are casesyncretic. The form of the 1st and 2nd person plural clitics (e.g. 1st person plural ni in the examples in (6)) can be misleading due to the effect of a morphophonological rule in clusters (see Footnote 1). Syncretism with plural clitics must then be diagnosed in isolation, as illustrated in (12)–(13) below for plural 1st person clitics: (12) Ne vede. 1pl.acc sees ‘He sees us’ (13) Ne dă bomboane. 1pl.dat give candy ‘He is giving us candy’. This syncretism occurs when the Dat/Acc clitics are [+participant, +plural], where [+participant] is a feature grouping 1st and 2nd person. Learners who attempt an analysis for this syncretism will posit a featural basis for it in order to “map backwards” what the phonological exponent is realizing. (In other words, we assume that learners strive to find a “shared form implies shared features” basis for syncretism). Romanian has syncretism of two distinct uses of ne and vă, which differ in their case features, and this means that the Dat and Acc 1pl and 2pl must share some other feature: [+animate]. Learners who posit a featural basis for syncretism must do so only based on the universal inventory of phi-features, and [+animate] is available among them, and moreover an interpretively “grounded” feature present among 1st and 2nd person clitics, due to the nature of 1st and 2nd person. Overt morphological syncretism among the plural [+participant] clitics is evidence for the language learner of an underlying featural identity, implemented by the implicational statement in (14). (14) Implicational Statement for Romanian clitics: [+participant, +plural] → [+animate] This feature co-occurrence requirement is part of the clitic terminal nodes that enter the syntax. All 1st and 2nd Plural clitics must be syntactically [+animate]. Singular 1st and 2nd clitics are also interpretively animate, of course, but are not syntactically required to bear this feature. It is the syncretism among the [+participant] plurals, coupled with the learning of syncretism as due to an interpretively grounded available phi-feature, that compels only these clitics to obligatorily bear [+animate].
Andrew Ira Nevins & Oana Săvescu
The contribution of this syncretism and its feature-theoretic implementation to the apparent Number Case Constraint of Romanian clitics is that there is nothing special about “plural number” in Romanian, per se, and no need to invoke a reasoning in which plural is for some reason especially marked in Romanian. The only thing special about plural clitics is that they are syncretic between dative and accusative. It is the syncretism of these particular clitics (and the more general effect of syncretism on clitic clusters), and not their value for [±plural] per se that is responsible for the effect. While (14) requires that all clitic nodes bearing the features [+Part, +Pl] must also bear [+animate], the syntax itself sees only the result of this statement (i.e. the feature [+animate]), and not the eventual fact that post-syntactically, this shared feature will condition syncretism. In other words, since the syntactic computation does not see the phonological form of clitics, the effect of syncretism within the syntax is indirectly mediated by morphosyntactic features. 6. The syntax of ditransitives, specifically ApplP We turn to the syntax of double object constructions and the role of clausal structure, agreement, and economy conditions in deriving the clitic combination restrictions observed in Romanian and in Leísta Spanish. There are many proposals for the internal structure of double object constructions, and we adopt the specific proposal that double object constructions are introduced by the functional head ApplP (Pylkkänen 2002, p.19; Cuervo 2003), in which the IO is higher than the DO. Pronominal clitics are generated in argumental positions:
vP 3 Ag
v′ 3 v
VP
3 V
ApplP
3 IO
Appl′
2 Appl
DO
An apparent ‘Number Case Constraint’ in Romanian
The projection ApplP takes the indirect object in its specifier and the direct object in its complement. Following the spirit of Adger and Harbour (2007), we adopt this selectional requirement on the specifier of Appl: (15) Selectional Requirement: The Applicative head has a selectional requirement that its specifier be [+animate], implemented as a feature-checking process. The existence of (15) is responsible for the animacy requirements on the goal in double object constructions (Bresnan, 1978, Mazurkewich and White, 1984, Pesetsky, 1995):6, responsible for the contrasts in (16): (16) a. *I sent the border a package (cf.€I sent the boarder a package) b. *I showed the door you (cf.€I showed you to the door) Finally, while [+animate] is a necessary feature on certain DPs due to lexical and syntactic requirements, and can be optionally specified on other DPs upon their entry into the syntax, the logical form (LF) component can deal with underspecification, and can still interpret an element that is underspecified as [±animate]. The effects of various animacy requirements in distinct components of the lexicon, the syntax, and logical form are summarized in (17): (17) Summary of animacy conditions in Romanian: a. Due to Implicational Statement (14), all 1st.pl/2nd.pl clitics must be syntactically [+animate] b. Due to Selectional Requirement (15) the specifier of Appl must be syntactically [+animate] c. Underspecification okay at LF: if an argument lacks a syntactic [+animate] feature marking, this does not entail that the argument is inanimate Having presented the requirements of [+animate] marking in the lexicon and syntax, we turn to the details of case assignment in ditransitive constructions. We adopt the following syntactic requirement on Case assignment: (18) Agree with DO: Appl must establish an Agree relation with the DO in its complement domain in order to value its Accusative case Finally, we adopt the following economy principle (Adger & Harbour 2007:€26, see also Abels (2003)): 6. We put aside cases of benefactive scenarios involving tables and cakes, and assume that such cases involve coercion of animacy via a notion of affectedness. Placement of inanimates into goal position forces assignment of [+animate] to them, yielding the interpretive effect of affectedness.
Andrew Ira Nevins & Oana Săvescu
(19) Spec-Comp Distinctness: The features which a functional head requires its specifier to bear cannot be used as probes in the head’s complement domain. This economy principle dictates that a feature already checked within the complement domain of a functional head cannot be used again in the case of interaction with the same head’s specifier. In other words, if the feature [+animate] is already probed within the complement of Appl, then it cannot be used as a selectional feature in order to assign a thematic interpretation to the specifier of Appl. 7. Deriving the “Number Case Constraint” of Romanian The ungrammatical (4) is repeated below, with a tree representing the ApplP domain below it. (20) ??/*Dându- vi- ne in grijă, tata s- a simţit uşurat. giving 2pl.dat 1pl.acc in care, father se has felt relieved ‘Entrusting us to y’all, my father felt relieved.’
ApplP
IO
Appl′
always[+anim], by (15) Appl
DO 1/2Pl → [+anim], by (14)
The DO above must enter the syntax as [+animate], due to the syncretism-yielding implicational statement of (14). The IO is required to be [+animate], due to the selectional requirement of (15). However, as the DO is Agreed-with by Appl first (18), Appl cannot unify its selectional requirement with its specifier (19). The interaction of each of these independent principles (many of coming from different modules of the grammar, e.g. the syntax-semantics interface (15), the morphosyntax-phonology interface (14)), and purely syntactic computational principles (19)) thus derives the apparent “Number Case Constraint” of Romanian
An apparent ‘Number Case Constraint’ in Romanian
with no direct reference to where Number and Case co-occur.7 The seemingly idiosyncratic effects of plural number on clitic cluster acceptability fall out from the interaction of universal grammatical principles with the distribution of syncretism in the Romanian clitic paradigm. 7.1
Returning to Leísta Spanish
In this subsection we show how the account developed above, in part motivated by the role of dative-accusative syncretism among [+animate] clitics in Leísta Spanish, can be extended immediately to explain an interesting restriction in Leísta Spanish. As noted by Ormazabal and Romero (2007), in Leísta Spanish, the one place that le cannot be used for a [+animate] accusative is in the presence of dative clitics (21), with the corresponding ApplP shown below. (21) *Te le dí. 2.dat 3.Acc.Anim gave.1sg ‘I gave him to you’.
ApplP
IO
Appl′
always[+anim], by (15) Appl
DO [+anim] ⇒ /le/, by (11)
The unacceptability of (21) can be derived by a now-familar set of interactions. The choice of a DO specified as [+animate] (leading to phonological spellout as le, by (11)), in combination with the selectional requirement of Appl for a [+animate] specifier, will lead to a crash due to Spec-Comp distinctness. Interestingly, a ditransitive with two interpretively animate objects can be saved if the second one is not syntactically [+animate]. In other words, if the 7. The ditransitive clitic co-occurrence restrictions on two [+animate] clitics that these principles generate do not subsume all PCC effects, such as preverbal *2sg > 1sg in Romanian, or the weak vs. strong PCC. See Nevins (2007) for an approach to PCC microvariation based on sensitivity to marked and contrastive Person features.
Andrew Ira Nevins & Oana Săvescu
second DO is underspecified for [±animate] (and no well-formedness requirement forces its specification), ApplP can successfully check the selectional requirement of its specifier without interference: (22) Te lo dí. 2.dat 3.Acc gave.1sg ‘I gave it/him to you’.
ApplP
IO
Appl′
always[+anim], by (15) Appl
DO [?: anim] ⇒ /lo/
In (22), the DO clitic node is underspecified for [±animate], passes through the syntax without violating Spec-Comp distinctness, and is ultimately spelled out as lo. LF can provide the interpretation of animacy of the direct object without requiring syntactic specification for this feature, and the interaction of syntactic underspecification with LF interpretation and PF use of a lessspecific clitic form ultimately renders this configuration acceptable in place of the intended (21). The fact that the same ban on [+animate] direct objects is found in Leísta Spanish and Romanian is an encouraging confirmation of the generality of (15) and (19). In fact, the same derivation for (21) can be extended to KiRimi, in which “definite animate NPs cannot appear as O2” [Hualde 1989 p.180] – in other words, in double object constructions co-indexed by pronominal clitics on the verb, the theme argument cannot be [+animate, +definite], presumably due to the same sort of “competition” for agreement with Appl. The role of Spec-Comp distinctness for [+animate] may more generally relate to Ormazabal & Romero’s (2007) Object Agreement Constraint, which holds in what Kittilä (1989) calls “animacy-prominent” languages such as Lango, in which the computation of Agree of Appl looks first at whether the lower DP is syntactically animate or not. Jeong (2004) notes that for some speakers, in case of multiple accusative case checking in Korean, the accusative possessor NP cannot be [+animate] if the accusative possessee object NP is [+animate]:
An apparent ‘Number Case Constraint’ in Romanian
(23) John-i Swuni-lul meli-lul cal-ass-ta John-Nom Swuni-Gen/Acc hair-Acc cut-Pst-Dcl ‘John cut Swuni’s hair’ (24) *?John-i Swuni-lul yetongsayng-ul ttayly-ess-ta John-Nom Swuni-Acc sister-Acc hit-Pst-Decl ‘John hit Swuni’s sister’ Jeong (2004, 107) proposes a ‘General PCC’: Two NPs cannot be [+animate] if they check that feature against the same functional head. This formulation and the Korean data point to a more general basis for an interpretation of the Leísta facts, and the Romanian pattern in turn, as part of a series of grammatical interactions banning two [+animate] arguments checking this feature against the same head, argued here to be the functional node Appl, responsible for the thematic and case properties of ditransitive and possessor-raising constructions. 8. General conclusions The sudden emergence of a syntagmatic PCC effect in Romanian with postverbal plural DOs but not postverbal singular DOs is related to the fact that the former are inherently specified [+animate] in the syntax, a paradigmatic property independently available to the learner by virtue of its syncretism in IO and DO contexts. In this respect, this study reinforces a more general characteristic of morphology and perhaps human language more generally: the effect of a paradigmatic property of an item on that item’s syntagmatic possibilities in combinations. The contribution of Romanian, in which singular [+participant] clitics are not syncretic, constrains the space of previous PCC accounts invoking syncretism (e.g. Laenzlinger 1993): it’s specifically syncretism in the DO position that matters for Person-Case type effects. Importantly, in the implementation developed here, there is no phonological content within the syntax, i.e. the syntax does not “see the syncretism” directly; rather, the existence of a shared phonological form for seemingly disparate syntactic functions is accomplished by an implicational well-formedness statement on the features that must co-occur on clitic nodes that enter the syntax. Returning to the scare quotes in the title of our paper, despite initial appearances, there is no actual “Number Case Constraint” in Romanian (and nothing special about the morphosyntax of plural in Romanian per se; only the morphophonological fact of plural clitics’ syncretism). The ban on [+participant,
Andrew Ira Nevins & Oana Săvescu
+plural] clitics in Romanian ditransitives is instead derived from the way that [+animate] (arguably a subtype of Person feature) is checked in ApplP, through the interaction of complement-based Agree and feature economy, parallel to independent restrictions in Leísta Spanish and beyond. References Abels, K. (2003). The syntax of adposition stranding. PhD thesis, University of Connecticut. Adger, D. and Harbour, D. (2007). Syntax and Syncretisms of the Person Case Constraint. Syntax, 10.1:2–37. Anagnostopoulou, E. (2003). The Syntax of Ditransitives. Mouton de Gruyter. Bleam, T. (1999). Leísta Spanish and the Syntax of Clitic Doubling. PhD thesis, University of Delaware. Bonet, E. (1991). Morphology after Syntax: Pronominal Clitics in Romance. PhD thesis, MIT. Bresnan, J. (1978). MIT Class notes. Cardinaletti, A. (2008). On different types of clitic clusters. In De Cat, C. and Demuth, K., editors, The Bantu-Romance Connection: A comparative investigation of verbal agreement, DPs and Information Structure, page 41–83. Amsterdam: John Benjamins. Cinque, G. (1994). Italian Syntax and Universal Grammar. Cambridge: Cambridge University Press. Cuervo, C. (2003). Datives at Large. PhD thesis, Massachusetts Institute of Technology. Farkas, D. and Kazazis, K. (1980). Clitic Pronouns and Topicality in Rumanian. In Chicago Linguistic Society 16, pages 75–82. Chicago. Halle, M. and Marantz, A. (1993). Distributed Morphology and the Pieces of Inflection. In The View from Building 20, pages 111–176. Cambridge, MA: MITWPL. Heap, D. (1998). Optimizing Iberian clitic sequences. In Lema, J. and Treviño E., editors, Theoretical Analyses on Romance Languages, pages 227–248. Amsterdam: John Benjamins. Hualde, J. I. (1989). Double Object Constructions in KiRimi. In Newman, P. and Botne, R., editors, Current Approaches to African Linguistics 5, pages 179–190. Dordrecht: Foris. Jeong, Y. (2004). Asymmetries in multiple case assignment. In Schmeiser, B., Chand, V., Kelleher, A., and Rodriguez, A., editors, WCCFL 23 Proceedings, pages 101–114. Somerville, MA: Cascadilla Press. Kayne, R. (1975). French Syntax. Cambridge, MA: MIT Press. Kittilä, S. (1989). The woman showed the baby to her sister: On resolving humanness-driven ambiguity in ditransitives. In Kulikov, L., Malchukov, A., and de Swart, P., editors, Case, Valency and Transitivity, pages 291–308. Amsterdam: John Benjamins. Laenzlinger, C. (1993). A syntactic view of Romance pronominal sequences. Probus, 5.3:242–270. Landa, A. (1995). Conditions on Null Objects in Basque Spanish and Their Relation to Leismo and Clitic Doubling. PhD thesis, USC, Los Angeles. Mazurkewich, I. and White, L. (1984). The acquisition of the dative alternation: unlearning overgeneralizations. Cognition, 16:261–283. Nevins, A. (2007). The Representation of Third Person and its Consequences for Person-Case Effects. Natural Language and Linguistic Theory, 25.2:273–313. Ormazabal, J. and Romero, J. (2007). The Object Agreement Constraint. Natural
An apparent ‘Number Case Constraint’ in Romanian Language and Linguistic Theory, 25.2:315–347. Perlmutter, D. (1971). Deep and surface structure constraints in syntax. Holt, Rinehart, and Winston, New York. Pesetsky, D. (1995). Zero Syntax. Cambridge, MA: MIT Press. Pylkkänen, L. (2002). Introducing Arguments. PhD thesis, Massachusetts Institute of Technology. Săvescu, O. (2007). Challenging the Person Case Constraint: Evidence from Romanian. In Camacho, J., Flores-Ferrán, N., Sánchez, L, Deprez, V., Cabrera, M., J., editors, Romance Linguistics 2006. Selected Papers from LSRL 36 pages 261–275. Amsterdam: John Benjamins.
part 4
Semantics and morphology
Generic bare singulars in Brazilian Portuguese Carmen Dobrovie-Sorin & Roberta Pires de Oliveira The analysis of the generic readings of Bare Singulars in Brazilian Portuguese is controversial: for Munn & Schmitt (1999, 2005) and Schmitt & Munn (2002) generic Bare Singulars are names of kinds, whereas for Müller (2002) they are indefinites bound by a generic operator. This paper provides two arguments against Müller: (i) relying on a detailed corpus study by Pires de Oliveira et al. (2007), we argue that Müller’s judgments are not shared by all speakers of Brazilian Portuguese; (ii) we provide theoretical and crosslinguistic evidence against analyzing generic Bare Singulars as indefinites bound by a generic operator. We make explicit Munn & Schmitt’s analysis of kind-referring Bare Singulars in Brazilion Portuguese by proposing that they rely on Chierchia’s Down operator (1998). The contrast between generic Bare Singulars and generic Definite Singulars, both of which are kind-referring, is explained based on Dobrovie-Sorin & Pires de Oliveira (2007).
1. Introduction The analysis of Bare Singulars (BSs) in Brazilian Portuguese (BrP) is crucial for a better understanding of the complex relations between kind-denotation, generic quantification, indefiniteness and number neutrality. After re-examining the data, we conclude that generic BSs in BrP should be analyzed as kind-referring rather than as indefinites, supporting Munn & Schmitt (1999, 2005) and Schmitt & Munn (2002). Generic BSs can be kind-referring (they denote intensional maximal sums) because they lack (singular) Number. We propose that the Down operator, introduced by Chierchia (1998) in his analysis of English Bare Plurals (BPs), also applies to number-neutral domains. Furthermore, we argue that generic BSs cannot be analyzed as indefinites bound by a generic operator (such an analysis is ruled out by Dobrovie-Sorin & Mari’s (2006) constraint against quantifying over sums). Section 2 reviews the evidence that existential BSs in BrP are number-neutral. Section€ 3 shows that generic BSs pass the tests for kind-reference and that kind-denotation can be derived via Chierchia’s down operator. Section€4 shows that there is no evidence in favor of generic BSs being indefinites bound by a
Carmen Dobrovie-Sorin & Roberta Pires de Oliveira
generic quantifier and that such an analysis is ruled out by a constraint against quantifying over sums; in Section€5, we explain the contrasts between generic BSs and generic Definite Singulars ((DSs) in BrP, which replicate the contrasts between generic BPs and generic DSs in English. 2. Existential BSs are number-neutral It is uncontroversial (Munn & Schmitt (1999, 2005), Schmitt & Munn (2002), Müller (2002)) that BSs in BrP lack (the functional category of) Number and correlatively yield number-neutral interpretations on their existential reading.1 In support of this generalization, Schmitt & Munn (2002) observe that BSs may be resumed with both singular and plural pronouns: (1)
Eu vi criança na sala. I saw child in-the room E ela estava/elas estavam ouvindo And she was/they were listening. ‘I saw a child/children in the room. And she was/they were listening.’ (Schmitt & Munn, 2002:€24, example (43))
The number-neutrality of BSs is confirmed by its “aspectual” contribution: (2) Eu escrevi carta por duas horas. I wrote letters for two hours. (3) a. #Eu escrevi carta em duas horas. I wrote letters in two hours. b. Eu escrevi uma carta em duas horas. I wrote a letter in two hours. (Schmitt & Munn 2002:€25, examples (48a) and (48c), respectivelly) The BS forces non-durative readings of telic verbs, whereas the singular indefinite imposes a terminative interpretation. Munn & Schmitt (2005:€827) conclude that “the bare singular is not quantized, despite the fact that it is [superficially] 1. Although they agree that existential BSs are number-neutral, Munn & Schmitt and Müller disagree on the analysis of existential BSs: for Müller, they are indefinites bound by an existential operator; for Munn & Schmitt they are names of kinds that yield an existential reading due to Chierchia’s (1998) rule of Derived Kind Predication. Existential BSs are beyond the scope of this paper, but Dobrovie-Sorin & Pires de Oliveira (in preparation) adopt (in line with Müller (2002)) a Diesing-like analysis, according to which existential BSs are indefinite-like expressions that get bound by VP-level existential closure (Diesing (1992), Kratzer (1988, 1995)).
Generic bare singulars in Brazilian Portuguese
morphologically singular”. This is predicted by the hypothesis that existential BSs in BrP do not denote atomic individuals but rather are entities that are not specified as either singular or plural. 3. Generic BSs in Brazilian Portuguese denote intensional maximal sums The controversy between Munn & Schmitt (1999, 2005), Schmitt & Munn (2002) versus Müller (2002), discussed in Section€3.1, is due to divergent grammaticality judgments. Section€3.2 is devoted to the theoretical aspect of the problem; our contribution is to make clear that the generic readings of BSs result from an application of the Down operator. 3.1
The data
The most reliable test for kind-reference is to evaluate whether the nominal phrase can combine with a kind predicate. Pires de Oliveira et al. (2007) empirically verified the grammaticality of BSs with kind predicates. They investigated two oral corpora and written language, and found evidence of BSs with kind predicates. We review just the results of the experiment on speakers’ judgment of acceptability. It consisted of 36 target sentences (pairs of sentences which only differ with respect to the BS versus the DS) and 24 filler sentences, randomly arranged. The test was applied to 200 adult speakers, undergraduate students (none from Letters/Linguistics) at Universidade Federal de Santa Catarina (UFSC). Each subject read at most 18 sentences. The subjects, after reading the sentence, marked his/her judgment: (4)
Dinossauro foi extinto por um cometa. Dinosaur was extinct by a comet. ‘Dinosaurs were extinct by a comet’. .....................................
From the total occurrences of BS sentences (600 sentences) across all the subjects, 58% were considered plainly acceptable. Another test for kind-referring phrases is to show that they can have generic readings when occurring in episodic contexts: (5) a.
Batata foi introduzida na Europa em meados Potato was introduced in-the Europe in the middle do século XVIII. of-the 18 century. ‘Potatoes were introduced in Europe in the middle of 18 century’.
Carmen Dobrovie-Sorin & Roberta Pires de Oliveira
Pires de Oliveira et al. (2007) evaluated 6 pairs of episodic sentences (one sentence with the bare singular, as in (5), and itscounterpart with a DS). The experiment showed that BSs are acceptable, since their rate of acceptability is around 60%. In sum, the grammaticality judgments reported in Müller (2002) for some crucial examples testing the name-of-kind status of BSs in BrP are not shared by all speakers of Brazilian: for an important proportion of speakers, BSs can refer to kinds.2 3.2
Number neutrality and kind reference
In this section we show that kind reference is theoretically unproblematic for BSs in BrP: determiner phrases lacking Number yield intensional maximal sums via the application of Chierchia’s Down operator. Introduced by Chierchia (1998) in his analysis of kind-referring BPs in English, the Down operator is an intensional maximality operator that applies to a property of pluralities (type <s, <e, t>>) and yields a kind (type e), or more precisely the largest member of the extension (in each given world/situation) of that property:
(6) For any property P, world/situation s [and set of kinds K] = λs ι Ps, if λs ι Ps is in K Undefined otherwise Where Ps is the extension of P in s. (Chierchia (1998:€351))
∩P
Chierchia’s Down operator cannot apply to singular properties. The generic readings of singular indefinites and singular definites must therefore be analyzed without resorting to it. Although BSs in BrP are morphologically singular, they are number-neutral expressions and therefore, the Down operator may freely apply to them: since this operator picks up the maximal sum in a given domain, it is irrelevant whether the domain contains atoms, in addition to pluralities. Similarly, if the Down operator applies to domains that contain amounts of matter (see bare mass nouns) it yields the maximal intensional amount in the domain. Thus, mass, plural and numberneutral BNs show a three-way distinction for their existential readings, but are alike in generic contexts, where they yield names of kinds in exactly the same way, as the result of the application of the Down operator. This analysis directly explains why generic BSs in BrP behave on a par with generic BPs in English: just like BPs, they denote intensional maximal sums.
2. As pointed out by a reviewer, there may be two dialects, but the issue is not discussed in Pires de Oliveira et al.€(2007).
Generic bare singulars in Brazilian Portuguese
Although this analysis seems straightforward, it was not proposed by Chierchia himself.3 To the best of our knowledge, Cheng & Sybesma (1999) is the only place where the analysis proposed here was suggested in passing: “As for how the kind interpretation arises, we simply assume, following Chierchia (1998) that the “down” function – that is, the ∩ operator (the “nominalization”-of-the predicate operator) –yields kinds when plural predicates (common nouns) are involved. Since Chinese common nouns are neutralized in terms of number, the “down” function can always apply, yielding kind interpretation.” (Cheng & Sybesma (1999:€533)) Munn & Schmitt (2005) insist on the relation between number-neutrality and kind-denotation, but they do not explicitly assume the Down operator: “A bare singular, a bare plural or a mass denotation will therefore be able to function as a kind since they are all semantically non-singular, provided a weak D is available” (Schmitt & Munn 2002). One could understand their proposal as implicitly following Chierchia in postulating some kind of direct correspondence between number neutrality (and mass denotation) and kind-reference. 4. Generic BSs are not indefinites bound by a generic operator We have so far shown that (i) at least for a significant proportion of Brazilian speakers, generic BSs must be analyzed as names of kinds, and (ii) kind-denotation is predicted to exist for number neutral expressions if we adopt Chierchia’s Down operator. The existence of kind-referring BSs is compatible with the analysis of generic BSs as indefinites bound by a generic operator. This type of coexisting analyses is indeed currently assumed for generic BPs in English: because they pass the tests for kind-reference, all theorists agree that English BPs are kind-referring in at least part of their distribution; the kind-analysis is acknowledged even by those who believe that generic BPs can also be analyzed (in some other contexts) as indefinites bound by a generic operator (Heim (1982), Kratzer (1988, 1995), Diesing (1990), among others). We must then ask ourselves whether at least in certain contexts, generic BSs in BrP are indefinites bound by a quantifier, as in Müller (2002). Section€4.1 shows that there is no positive evidence in favor of this analysis in BrP. Section€4.2 argues, based on cross-linguistic evidence and theoretical considerations, that this 3. Chierchia wrongly assimilates number-neutrality and mass. For arguments against this view see Doetjes (1997) and Cheng & Sybesma (1999) for Chinese; Munn & Schmitt (2005) and Müller & Paraguassu (2007) for BrP.
Carmen Dobrovie-Sorin & Roberta Pires de Oliveira
analysis is ruled out by a constraint on quantification in natural languages. We are thus led to conclude that generic BSs in BrP (and BPs in English) cannot be analyzed as indefinites bound by a generic operator. 4.1
Testing the indefiniteness of generic BSs
Among the evidence adduced by Müller in favor of the non-kind reference of BSs are examples of the type shown in (7)-(8): (7) a. Garrafa de Coca tem gargalo estreito. Bottle of Coke has neck narrow. ‘Coke bottles have narrow neck.’ b. A garrafa de Coca tem gargalo estreito. The bottle of Coke has neck narrow. ‘The Coke bottle has narrow neck.’ (8) a. Garrafa verde tem gargalo estreito. Bottle green has neck narrow. ‘Green bottles have narrow neck.’ b. # A garrafa verde tem gargalo estreito.(Existential reading only) The bottle green has neck narrow. ‘The green bottle has narrow neck.’ (Müller 2002:€288 examples (32) and (35)) The observation is that definite singulars cannot refer to kinds that are not ‘wellestablished’ (8b), whereas BSs are indifferent to this constraint. Furthermore, Müller claims that BPs and singular indefinites behave on a par with BSs: (9) a. Garrafas verdes têm gargalo estreito. ‘Green bottles have narrow neck.’ b. Uma garrafa verde tem gargalo estreito. ‘A green bottle has a narrow neck.’ (Müller 2002:€288 examples (34) and (36)) From the identical behavior of BSs, BPs and singular indefinites, Müller concludes that [in examples where the nominal expression refers to non-well-established kinds] “we have generic quantification instead of kind-referring nominal expressions”. (2002:€282) The generic reading of sentence (9b) is, however, very marginal, whereas (8a) and (9a) are fully acceptable, which goes against treating generic BSs on a par with generic indefinites. Moreover, Müller’s conclusion is not supported by the reviewed data: the test relying on ‘well-established’ kinds distinguishes types of kind-referring expressions, e.g., BPs vs DSs in English (Krifka
Generic bare singulars in Brazilian Portuguese
et al (1995)), plural definites vs singular definites in French (Beyssade (2005)), BPs vs BSs in Hindi (Dayal (2004)). The fact that BPs in English and BSs in BrP are not sensitive to this test does not mean that they are not kind-referring but rather that, if they are kind-referring, they do not refer to a well-established kind (Section 5 below). Müller (2002, footnote 11) herself observes that there exists at least a class of examples that are problematic for her analysis. Indeed, dispositional predicates such as amar ‘to love’ allow the generic reading of BSs (also BPs in English) in object position, whereas the generic indefinite singular is disallowed: (10) Eu amo pedreiro. I love bricklayer. ‘I love bricklayers’. (11) #Eu amo um pedreiro. #I love a bricklayer.
(Existential reading only)
Generic objects constitute a well-known problem for the quantificational analysis of generic BPs in English (Kratzer (1988, 1995) Diesing (1990)) – see the contrast in the translations of (10) and (11): by virtue of Diesing’s Mapping Algorithm, indefinites acquire generic readings if and only if they are mapped into the restriction of the generic operator, but DPs in object positions are mapped into the scope of the generic quantifier, rather than in the restriction, and as a consequence they get caught by existential closure. Kratzer’s suggestion is that dispositional predicates allow the scrambling of their object at LF, placing it in the restriction of the generic operator. This suggestion does not explain why singular indefinites in the object positions of the same predicates disallow the generic reading4, in clear contrast with BPs (and BSs in BrP). Moreover, the scrambling solution does not account for the intuitive reading of the relevant examples: if the generic object is in the restriction of GEN (Generic operator) at LF, the expected reading is a generalization over pairs of me and bricklayers; but the observable reading expresses a generalization about the subject: love of bricklayers are said to be a stable property of mine. Regardless of the precise analysis, the conclusion is that in examples of this type, generic BPs in English, and generic BSs in BrP, cannot be analyzed as quantificational indefinites. The only option is therefore kind-reference. Further evidence for the kind analysis of generic BSs in BrP is related to the ‘nomicity’ constraint (Lawler (1973), Burton-Roberts (1977), among others). The
4. Contrastive stress on indefinite objects makes their generic reading possible, but BPs and BSs objects do not need contrastive stress to be interpreted generically.
Carmen Dobrovie-Sorin & Roberta Pires de Oliveira
examples below5 show that the generic reading of indefinites is allowed if the predicate expresses an essential/nomic property of the subject, (12a), but blocked with non-essential properties, (12b): (12) a. ??Um samba é polifônico. A samba is polyphonic. ‘A samba is polyphonic.’ b. #Um samba é popular. A samba is popular. ‘A samba is popular.’
(Existential reading only)
This constraint does not affect BSs: (13) a. Samba é polifônico. Samba is polyphonic. ‘Sambas are polyphonic.’ b. Samba é popular. Samba is popular. ‘Sambas are popular.’ According to Dobrovie-Sorin & Laca (1999), characterizing sentences built with generic BPs in English do not rely on generic quantification; rather, they involve a kind-referring BP in the subject position of a predicate that distributes down to the individual entities in the set denoted by the BP in a way similar to what happens with other collective or plural nouns. Examples like (13b) – also their English counterparts – are fully acceptable because distributive predication is not subject to the nomic constraint. Because GEN is subject to this constraint and because indefinite DPs necessarily rely on generic quantification, examples of the type in (12b) are unacceptable.6 The marginal acceptability of (12a), which obeys the nomicity constraint (being polyphonic is an essential/definitional property of sambas), must be attributed to a more general constraint on generic indefinites, which seems to be stronger in Romance languages than in English: they are fully acceptable only in modal contexts. To conclude, we have shown that generic BSs behave unlike generic indefinites with respect to all the current tests. 5. The examples in (12)-(13) are adaptations of the well-known contrast between singular indefinites and BPs in English, see the translations. 6. The account summarized here is controversial.€In particular, Greenberg (2007) assumes a quantificational analysis not only for indefinites, but also for BPs. But her analysis cannot explain why it is indefinites – and not BPs – that are subject to the nomicity constraint illustrated in (12b).
Generic bare singulars in Brazilian Portuguese
Arguments against the indefiniteness of generic BSs in BrP
4.2
The French examples in (14)7 show that plural indefinites cannot be bound by adverbial quantifiers (and in particular by GEN): (14) *Des chats sont en général intelligents. des Cats are usually intelligent. The same impossibility is observed for bare plurals in Romanian (also in Spanish and Italian): (15) *Pisici sunt rareori inteligente. ‘Cats are rarely intelligent’. The ungrammaticality of these examples was explained by Dobrovie-Sorin & Mari (2006) and Dobrovie-Sorin (2008) as being due to a violation of the following constraint on quantification: (16) A sum-variable cannot be bound by a quantifier.8 The constraint in (16) can be viewed as the consequence of an individuability condition that says that domains of quantification can only contain ‘strongly individualizable’ entities, i.e. every entity is disjoint from all the other entities of the domain.9 This constraint is satisfied by atomic individuals, but not by sums of 7.
Examples of this type should be distinguished from two other classes of examples:
i.
Des droites convergentes ont un point en commun. des lines convergent have a point in common ‘Convergent lines have a point in common.’
ii. Méfie-toi, des guêpes énervées sont un danger terrible. watch out, des wasps excited are a danger terrible. ‘Watch out, excited wasps are a terrible danger.’ e grammaticality of (i), originally pointed out by Corblin (1987) and Carlier (1989), was Th analyzed by Dobrovie-Sorin & Mari (2006) as being due to the fact that plural indefinites such as droites convergentes ‘convergent lines’ can denote groups; compare the plural indefinites in (14), which can only denote sums, i.e., entities that are related to each other by part-whole relations. Examples as (ii) involve quantification over events combined with the indirect binding of the plural indefinites (Longobardi (2001), Dobrovie-Sorin & Mari (2006)). 8. The constraint in (16) does not say anything about existential indefinites, which rely on existential closure (see the DRT literature stemming from Heim (1982) and Kamp (1981)) rather than on existential quantification. 9. A similar constraint on quantification is found in Landman (1989), and Kratzer (1995:€169): “Quite generally, any sort of quantification seems to require that the domain of quantification is set up in such a way that its elements are truly distinct’, where ‘truly distinct’ means ‘not related to each other by part-whole relations’”.
Carmen Dobrovie-Sorin & Roberta Pires de Oliveira
individuals, which are related to one another by part-whole relations. The examples in (14)-(15) violate (16) because indefinite plurals supply sum-variables. The generalization illustrated in (14)-(15) sheds doubt on one of the most influential analysis of generic BPs in English, according to which they are plural indefinites (Burton-Roberts 1977, Diesing (1990), Kratzer (1988, 1995)) bound by a Q-adverb: (17) Cats are usually intelligent. Given the unacceptability of the plural indefinite subjects in the French examples in (14), the English BPs in (17) cannot be indefinite-like expressions. Instead, they could be analyzed as names of kinds, since they can function as such in other contexts. Compare Romanian, Spanish, and Italian BPs, which cannot function as names of kinds (Delfitto (2002), Longobardi (2001)). BSs in BrP are comparable to BPs in English insofar as they denote sums rather than individualizable entities. Therefore, given (16), generic BSs occurring in (18) cannot be indefinites bound by an adverbial quantifier. The only option therefore is to analyze them as kind-referring entities: (18) Gato geralmente é inteligente. Cat generally is intelligent. Summarizing, generic BSs in BrP (and BPs in English) are kind-referring not only when combined with kind-predicates or when occurring in episodic sentences, but also in those contexts in which they are bound by an adverbial quantifier. Which means that all generic readings of BSs in BrP (and BPs in English) rely on kind-reference. 4.3
Kind-reference combined with generic quantification
The analysis of BSs as intensional maximal sums is compatible with adverbial quantification. According to a widely assumed view, kinds/maximal sums can be shifted to the set of atomic individuals that realize the kind (Chierchia (1998) among others) and the Q-adverb quantifies over those individuals; if GEN is the translation of ‘usually’, (17) has the LF in (19): (19) GEN x (x is a cat) [x is intelligent] The same analysis applies to generic BSs in BrP: as a result of the application of the Down operator, gato denotes a maximal sum, which is then shifted to the corresponding set of atomic individuals. Thus, generic BSs in BrP have exactly the same LF representations as generic BPs in English: the LF in (19) corresponds not only to (17) but also to (18).
Generic bare singulars in Brazilian Portuguese
5. On the contrast between generic BSs and generic DSs Since generic BSs in BrP have exactly the same LF representation as generic BPs in English and since it is uncontroversial that generic DSs have the same representation in the two languages, we expect kind-denoting DSs to contrast with kind-denoting BSs in BrP in the same way that they contrast with kind-denoting BPs in English. Krifka et al (1995) observed that in English, generic DSs can only refer to well-established kinds, whereas generic BPs are insensitive to this restriction. The BrP examples in (7)-(8) above, trivial adaptations of often quoted English examples, show that the same contrast is observed in BrP: sentences with DSs are generically interpreted only if the DS refers to a well established kind.10 This constraint does not hold for BPs in English, nor for BSs in BrP. Generic BSs in BrP also parallel English BPs in that they cannot occur in the object position of invent-type predicates, whereas DSs in both languages can. The following examples are adapted from examples built with BPs brought up in Krifka et al. (1995:€70): (20) #Babbage inventou computador. Babbage invented computer. (21) Babbage inventou o computador. ‘Babbage invented the computer.’ Dobrovie-Sorin & Pires de Oliveira (2007) explained the unacceptability of (20) by adapting Beyssade’s (2005) explanation of the corresponding French example built with a definite plural: the BS in BrP (as the English BP and the French definite plural) denotes the intensional maximal sum of computers, and it is not possible to invent a sum of individuals, but only the prototype (an atomic individual), hence the acceptability of DSs, which may refer to different sorts of individuals (particulars, taxonomic kinds, prototypes).11 Dayal (2004) pointed out that (but did not explain why) the passive version of examples built with invent-verbs and BPs are acceptable in English. The same is true for BSs in BrP: 10. Because the present paper is concerned with generic BSs and not with generic DSs, we will not comment further on the notion of ‘well-established’ kind, but see Dayal (2004) and Dobrovie-Sorin & Pires (2007). 11. Kester & Schmitt (2007) argue that in Papiamentu BSs have the same distribution as BSs in BrP, but they are acceptable as complements of invent. Our guess is that since Papiamentu does not have a definite article, BSs in this language may be analyzed as counterparts of kindreferring DSs in BrP or English rather than as counterparts of kind-referring BSs (see Dayal’s (2004) analysis of generic BSs in Hindi).
Carmen Dobrovie-Sorin & Roberta Pires de Oliveira
(22) Computador foi inventado por Babbage. Computer was invented by Babbage. According to Dobrovie-Sorin & Pires de Oliveira (2007), examples of the type in (22) are fine because DPs occurring in preverbal subject positions can function as Topics: (22) is a characterizing sentence that attributes the stable property of having-been-invented-by-Babbage to the kind/intensional maximal sum of computers and the sentence is acceptable because this property is relevant for the history of the kind. Compare the example in (20): because the BS occupies the direct object position, it cannot function as Topic, and therefore, this example cannot be analyzed as a characterizing sentence about the kind Computer. 6. Conclusion Based on Pires de Oliveira et al. (2007) and Dobrovie-Sorin & Pires (2007), the paper argues that generic BSs in BrP are kind-referring expressions, supporting Munn & Schmitt (1999, 2005) and Schmitt & Munn (2002). Given that BSs in BrP lack Number (Munn & Schmitt (1999, 2005)), we propose that the Down operator (Chierchia (1998) may apply to BSs, yielding an intensional maximal sum. The contrasts between generic BSs and generic DSs are due to the fact that the latter does not refer to maximal sums, but rather to atomic kinds, obtained via an application of the Iota operator to a property of kinds (Dayal (2004), Dobrovie-Sorin & Pires de Oliveira (2007)). Generic BSs cannot be analyzed as indefinites bound by a generic operator (as in Müller (2002)) because such an analysis would violate a constraint on quantifying over sums (Dobrovie-Sorin & Mari (2006), Dobrovie-Sorin (2008)). This proposal leaves open several issues. In particular, one may wonder whether the kind analysis extends to the existential readings of BSs, as in Carlson (1977) and Chierchia (1998) for BPs in English. Dobrovie-Sorin & Pires de Oliveira (in preparation) argue against the kind-referring analysis of existential BSs, which leads them to assume that BSs in BrP are underspecified: the null Determiner can be interpreted either as a Down operator or as a choice-function operator, respectively yielding the generic and the existential readings. Another crucial point is whether there is any difference between the BS in BrP and the BP in English. Our conclusion is that on generic uses there are none; nevertheless, they do differ in existential contexts: because they lack Number, existential BSs in BrP are number-neutral, whereas existential BPs in English are valued as plural, and as such they are interpreted as plural. We disagree with both Müller (2002) and Munn & Schmitt (1999, 2005), since both approaches assume a unified analysis of BSs in BrP. We, on the other hand,
Generic bare singulars in Brazilian Portuguese
propose an underspecified analysis. Thus, we agree with Munn &Schmitt (1999, 2005) on the analysis of generic BSs, and with Müller (2002) on the analysis of existential BSs. References Beyssade, Claire. 2005. “Les defines génériques en français: noms d’espèces ou sommes maximales”. Noms nus et généricité, org. by Carmen Dobrovie-Sorin, 33–64. Press Universitaires de Vincennes: Vincennes. Burton-Roberts, Noel. 1977. “Generic sentences and analyticity”. Studies in Language 1.155–196. Carlier, Anne. 1989. “Généricité du syntagme nominal sujet et modalités”. Généricité, spécificité et aspect. Travaux de linguistique, ed. by Marc Wilmet, 19, 33–56. Paris: Duculot. Cheng, Lisa Lai & Rint Sybesma. 1999. “Bare and not-so-bare nouns and the structure of NP.” Linguistic Inquiry 30. 509–542. Carlson, Gregory. 1977. Reference to Kinds in English. Ph.D. dissertation, University of Massachusetts, Amherst. Chierchia, Gennaro. 1998. “Reference to kinds across languages”. Natural Language Semantics 6:4. 339–405. Corblin, Francis. 1987. Indéfini, défini et démonstratif. Constructions linguistiques de la référence. Genève: Droz Dayal, Veneeta. 2004. “Number marking and (in)definiteness in kind terms”. Linguistics and Philosophy 27. 393–405. Delfitto, Denis. 2002. Genericity in Language. Issues of Syntax, Logical Form and Interpretation. Alessandria: Dell’Orso. Diesing, Molly. 1990. Indefinites. Cambridge: MIT Press. Dobrovie-Sorin, Carmen.1996. “Generic Plural Indefinites and (In)direct Binding”. A Handbook of French Semantics, ed. by Francis Corblin & Henriette de Swart. CSLI, Stanford. 55–71. —– 2005. “Introduction”. Noms nus et généricité, ed. by Carmen Dobrovie-Sorin. PUV. —– 2008. “Existential bare plurals: from properties back to entities”. Lingua. Doi:10.1016/j. lingua.2007.10.016 (in press) Dobrovie-Sorin, Carmen & Claire Beyssade. 2004. Définir les indéfinis. CNRS Editions, Collection Sciences du Langage. Dobrovie-Sorin, Carmen & Brenda Laca. 1999. “La généricité entre la référence à l’espèce et la quantification générique”, Patrick Sauzet ed. Langues et Grammaire III. Syntaxe. Communications présentées au Colloque Langues et Grammaire III (Paris, juin 1997), 163–177. Paris: Presses Universitaires de Vincennes. Dobrovie-Sorin, Carmen & Alda Mari. 2006. “Generic Plural Indefinites: Sums or Groups?”. Proceedings of NELS 2006. Dobrovie-Sorin, Carmen & Roberta Pires de Oliveira. 2008. “Reference to kinds in Brazilian Portuguese: Definite Singulars vs Bare singulars”. Proceedings of the SuB12, org. by Atle Grönn. Oslo: ILOS. Doetjes, Jenny. 1997. Quantifiers and Selection. On the distribution of quantifying expressions in French, Dutch and English. Ph.D. dissertation, Leiden University. La Hague: Holland Academic Graphics.
Carmen Dobrovie-Sorin & Roberta Pires de Oliveira Greenberg, Yael. 2007. “Exceptions to Generics: Where Vagueness, Context Dependence and Modality Interact”. Journal of Semantics 24. 131–167. Heim, Irene. 1982. The semantics of definite and indefinite NP’s. Ph.D. dissertation, University of Massachussetts. Kamp, Hans. 1981. “The theory of truth and discourse representation”. Formal Methods in the study of language, org. by Jeroen Groenendijk, Theo Janssen & Martim Stokhof, 277–322. Amsterdam: Mathematical Centre. Kratzer, Angelika. 1988. “Stage-Level and Individual-Level Predicates”. Genericity in Natural Language Proceedings of the 1988 Tübingen Conference, SNS-Bericht, ed. by Manfred Krifka, 88–42. —– 1995. “Stage-Level and Individual-Level Predicates”. The Generic Book, ed. by Gregory N. Carlson & Francis J. Pelletier, 125–175. Chicago: Chicago University Press. Kester, E-P and Cristina Schmitt. 2007. “Bare nominals in Papiamentu and Brazilian Portuguese”. Noun Phrases in Creole Languges. A multi-faceted approach, ed. by Marlyse Baptista & Jacqueline Guéron, 107–143. Amsterdam: John Benjamins. Krifka, Manfred, Francis J. Pelletier, Gregory N. Carlson, Alice ter Meulen, Gennaro Chierchia & Godehard Link. 1995. “Genericity: an Introduction”. The Generic Book, ed. by Gregory N. Carlson & Francis J.Pelletier, 1–124. Chicago: The University of Chicago Press. Landman, Fred. 1989. “Groups I”. Linguistics and Philosophy 12.559–605. Lawler, John. 1973. “Studies in English Generics”. University of Michigan Papers in Linguistics 1. Longobardi, Giuseppe. 2001. “How comparative is semantics? A unified parametric theory of bare nouns and proper names”. Natural Language Semantics 9. 335–369. Müller, Ana P. 2002. “The Semantics of Generic Quantification in Brazilian Portuguese”. PROBUS 14. 279–298. Müller, Ana & Nilse Paraguassu. 2007. “A distinção massivo – contável nas línguas naturais”. Paper presented at 8 Workshop on Formal Linguistics in Curitiba. Munn, Alan & Cristina Schmitt. 1999. “Against the Nominal Mapping Parameter: Bare nouns in Brazilian Portuguese”. Proceedings of NELS 29. —– 2005. “Number and indefinites”. Lingua 115: 821–855. Pires de Oliveira, Roberta, Josa Coelho & Mariana Rublescki Silveira Bressane. 2007. “O singular nu denota espécie. Uma investigação psicolingüística”. Paper presented at the 8 Workshop on Formal Linguistics in Curitiba. www.cce.ufsc.br/~pires. Schmitt, Cristina & Alan Munn. 2002. “The syntax and semantics of bare arguments in Brazilian Portuguese”. Linguistic Variation Yearbook 2. 253–269.
Aspect shift in stative verbs and their arguments Joshua Rodríguez This paper offers an alternative explanation for coercion effects or ‘aspect shift’ with stative verbs in combination with the progressive or the aspectually specific and distinct Spanish past tenses. The account follows Rodriguez (2007) in exploiting lexical ambiguities to provide a richer set of VP Aktionsarten via normal compositional semantics. Crucially it is the addition of aspectually specific morphology that functions to filter out those VP interpretations which have incompatible Aktionsarten. The purported shifting effect therefore derives from one or another VP meaning (potentially including alternating NP denotations as well) in association with corresponding morphology via aspectual compatibility. The analysis provides a natural means for explaining why certain cases of ‘aspect shift’ do not occur, namely on the basis of the lack of the underlying semantic potentials of the words involved. The proposal investigates three cases in depth, with implied applicability to a broader range of examples.
1. Introduction It is well known that the complement of a verb may directly impact the Aktionsart of the VP (e.g., Krifka 1992, Verkuyl 1993, 1999). Typical examples attesting to this fact include those in (1–2). (1) a. Mary wrote a letter ?for an hour/in an hour. b. Mary wrote letters for an hour/??in an hour. (2) a. John ran for 10 minutes/??in 10 minutes. b. John ran a mile ??for 10 minutes/in 10 minutes. In this investigation, I will explore a similar phenomenon involving verbs traditionally considered stative (i.e., verbs that head VPs which may be stative), such as have, know, consist, etc. and their Spanish counterparts. However, rather than only studying the effects of overtly varying the complement of the verb, I will also
Joshua Rodríguez
explore potential ambiguities that may arise in either the verb, its arguments, or both. Furthermore, instead of studying the compatibility with aspectually distinct temporal adverbs, I will use the acceptability of, and differences in interpretation associated with, aspectually specific morphology. I will show that polysemy ultimately impacts the possible compositionally derived Aktionsarten of the VP. Because of the covert nature of the variability involved, I will rely on morphologically marked grammatical aspect as a diagnostic for the possible Aktionsarten of the VP. These will include the progressive and the aspectually differentiated Spanish past tenses, which are each sensitive to a contrast in (non-)stativity for their interpretability or felicity (cf.€Rodriguez 2004a, 2007). I should also note that my approach, which permits a variable Aktionsart for the VP, differs from the implicit assumption in many coercion analyses (e.g., Smith 1991, de Swart 1998, Michaelis 2004, 2005) that the Aktionsart of the VP is in principle invariant based on its lexical contributors, and is manipulated covertly by more general grammatical principles as a means of correcting aspectually problematic combinations. By asserting this difference I do not suggest that coercion should be entirely eliminated from the grammar, only that once polysemy effects are considered, many apparent coercion effects disappear, greatly reducing the need for or the applicable breadth of aspectual coercion. 2. Polysemy and variable aktionsart First, and most obviously, the verb itself may be polysemous and therefore vary in its Aktionsart depending on the selection of interpretation. A typical example of ‘aspect shift’ in Spanish involves the ambiguity in the verbs of knowing: saber ‘know [facts]’ and conocer ‘know [people], be acquainted’.1 In their stative meaning, they imply a simple state of awareness, while in their non-stative sense they are inchoative, respectively ‘to learn/find out’ and ‘to meet/become acquainted’. Rodriguez (2007) argues that such polysemy is a general property of cognitive verbs, including also the English verbs understand and recognize. A similar argument might be made with regard to English have and its Spanish counterpart tener, which (despite some peripheral differences) both show a general polysemy between possession and experience, the former being stative and the latter nonstative (e.g., have [possess] a pencil vs. have [experience] a scary dream). 1. The term ‘aspect shift’ reflects the perspective that the Aktionsart of the VP is overtly modified or shifted away from its purportedly inherent Aktionsart. However, I use the term in its pre-theoretical sense as any contextually determined aspectual variant.
Aspect shift in stative verbs and their arguments
In addition, the internal temporal structure of the complement may also be ambiguous, and therefore interact in different ways with the verb. This may not be immediately evident, but with some background should become clear. Most nouns denote entities that have no inherent (non-trivial) temporal structure. In neo-Davidsonian terms (e.g., Bach 1986, Parsons 1990, Krifka 1989, 1992), nouns tend to denote sets of entities from the universe of objects or individuals, rather than events or times. Temporally speaking, such individuals normally exist in their entirety at every instant during the interval in which they exist at all – except perhaps where their ‘material constitution’ is mapped to temporally distributed subparts of an event of creation or consumption in an incremental fashion (Krifka 1989, 1992, cf.€Verkuyl’s 1993, 1999 notions [+add to] and [±sqa]). However, other nouns denote events or intervals of time directly, which therefore are internally distributed across time by definition. In more obvious cases, these respectively include such things as attempt (e.g., John’s final attempt) or Thursday. To this class of nouns, I propose to also include (at least polysemously) experiential nouns like dream or problem. In particular, there are multiple ambiguities associated with dreams and problems and the like which bear on the present analysis, and which are manifest in usage in both English and Spanish. A dream, for example, (call this dream1) may consist of a set of aspirations or desires, which despite potentially making reference to distinct moments of time, is not itself apprehended by the mind as a function of the passage of time. It cannot be ‘had’ in any other way than all at once and uniformly over time, just as an apple cannot in any obvious sense be held in one’s hand gradually, only in its entirety at every instant through time. This sense contrasts in its temporal constitution from a dream (call this dream2) as the dynamic product of the subconscious mind, which we experience through time. In this second sense, a dream has a beginning, middle, and end, each of which is produced by the mind at different moments in time. Thus, in contrast to have a dream1, have a dream2 specifically allows ‘having’ it incrementally. Furthermore, have itself might be similarly dichotomized (though not exhaustively) into have1 = possession and have2 = experience, where the former is temporally static, and the latter is by definition incremental. In other words, not only is having dream1 different from having dream2 in terms of what type of dream it is, but potentially also in the sense of what kind of having. Furthermore, have1 a dream2 essentially denotes remembering, as in (3a), whereas have 2 a dream1 is conceptually incongruent, attested by (3b), since have2 requires its complement to play out over time. (3) a. I have1 a nightmare I want to tell you about that I had2 last night. b. ??I’m having a dream to become a doctor someday.
Joshua Rodríguez
Similar effects could be found with other experiential nouns, such as problem, experience, success, etc. in both Spanish and English. 3. Grammatical aspect and aspectual filtering Using have with eventive and non-eventive complements as a test case, I will show in this section that the various possibilities in meaning of both verb and noun create an array of possible VP Aktionsarten, some of which are potentially filtered out by the addition of aspectually specific morphology. As background, I will adopt the intuition proposed by de Swart (1998) (and adopted by Michaelis 2004, 2005) that the Romance past tenses, as well as the progressive in English and Spanish, impose aspectually relevant selectional restrictions. However, as stated above, I do not assume that the VP should by assumption be restricted to a single possible aspectual classification whose meaning must be altered in every case of apparent aspectual mismatch. Instead, I will assume, following Rodriguez (2007), that aspectual incompatibility with one or the other morpheme generally filters out potential VP interpretations that are aspectually inconsistent. As shown there, when aspectually specific morphology (otherwise thought to induce aspect shift via coercion) is replaced by aspectually neutral morphology (e.g., the future tense or the subjunctive mood), two distinct Aktionsarten may become available, making the sentence ambiguous. Such an effect would not be predicted if one meaning were derived from the other via coercion (e.g., the inchoative from the stative) on the basis of aspectual incompatibility. (4) a.
Conociste mejor a mi hermano después de know-pfv-2sg better acc my brother after of tu viaje, ¿verdad? your trip, true “You came to know my brother better after your trip, right?”
b.
Conocías mejor a mi hermano después de know-impf-2sg better acc my brother after of tu viaje, ¿verdad? your trip, true “You were better acquainted with my brother after your trip, right?”
c.
Conocerás mejor a mi hermano después de tu viaje. know-fut-2sg better acc my brother after of your trip i.╇ “You will come to know my brother better after your trip.” ii.╇ “You will be better acquainted with my brother after your trip.”
Aspect shift in stative verbs and their arguments
d.
Espero que conozcas mejor a mi hope-1sg that know-sbj-2sg better acc my hermano después de tu viaje. brother after of your trip. i.╇ “I hope you come to know my brother better after your trip.” ii.╇ “I hope you are better acquainted with my brother after your trip.”
It is relevant that, all else being equal, the time of knowing begins later in (4a) than (4b). More specifically, the time of ‘getting to know better’ in the perfective version (4a) cannot logically occur before the end of the trip, whereas in the imperfective version (4b) there is a strong pragmatic implication that the ‘getting to know better’ occurred during the trip, such that a simple state of ‘knowing better’ (i.e., the state described by (4b)) persists from that time forward even after the trip (cf.€Rodriguez 2004b). This follows naturally if we assume that overtly imperfective morphology filters out non-stative interpretations, and that the reverse occurs (though in this case only pragmatically) with perfective morphology. Interestingly, sentences (4c) and (4d), which are not aspectually marked, are indeterminate in this regard, potentially describing either scenario. In addition, and following the tradition of Vendler (1967), Taylor (1977), and Dowty (1979), I assume that the progressive is routinely incompatible with stativity, being interpretable only in combination with non-stative VPs. However, I do not assume with the coercion tradition that felicitous combinations of purportedly stative verbs with the progressive are facilitated by covert corrective modifications, but instead by the polysemous lexical potentials inherent in the VP. Thus, I take the acceptability of the progressive as evidence for non-stative potential in the VP. Bearing in mind these restrictions imposed by the various aspectual morphemes, consider the following combinations of have and dream in English and Spanish. Observe that there is a corresponding difference in the ‘inner’ VP Aktionsart (see Verkuyl 1993) that arises between the simple present and progressive versions on the one hand, and between the aspectual contrast in the Spanish past tenses on the other. (5) a. Maria has a dream. María tiene un sueño. b. Maria is having a dream. María está teniendo un sueño. c. Maria had a dream. i. María tenía un sueño. Maria had-impf.3sg a dream ii. María tuvo un sueño. Maria had.pfv-3sg a dream
Joshua Rodríguez
In (5a), Maria possesses (or remembers) her dream (dream1 or dream2) all at once and uniformly over time, while in (5b) the having is a dynamic experience (have 2) that involves a dream as it plays out through time (dream2). Although in (5a) the dream may make reference to temporally ordered events, stative have1 does not ‘track’ the realization or playing out of those events. In (5b), however, the dream is ostensibly an experience that is played out as it is ‘had’. Of relevance for present purposes is the fact that (5c) is ambiguous in English between the two meanings described above, while the Spanish translations disambiguate these. The imperfective (5c.i) is stative, and thus translates have1 (leaving aside the possible habitual reading that may iterate the non-stative have2 dream2), whereas the perfective (5c.ii) regularly implies experiential have2 dream2 (though a bounded state reading of have1 is also in principle possible given sufficient context). The question that arises from these data is how to account for the restriction against *have2 dream1. Intuitively, the answer involves a harmonizing of Aktionsart between verb and noun. While have1 is stative and implies a static, non-incremental, or ‘all-at-once’ θ-relation to its complement, have2 implies an incremental or gradual θ-relation to its complement, whose internal structure must also be correspondingly distributed across time. Stated formally in the framework of Krifka (1989, 1992), a gradual θ-relation conforms to Uniqueness of Objects and Events, as well as Mapping to Objects and Events, as defined respectively in (6a-d).
(6) a. ∀R [UNI-O(R) ↔ ∀e, x, x’ [R(e, x) ∧ R(e, x’) → x = x’]]
b. ∀R [UNI-E(R) ↔ ∀e, e’, x [R(e, x) ∧ R(e’, x) → e = e’]] c. ∀R [MAP-O(R) ↔ ∀e, e’, x [R(e, x) ∧ e’ e → ∃x’ [x’ x ∧ R(e’, x’)]] d. ∀R [MAP-E(R) ↔ ∀e, x, x’ [R(e, x) → x’ x → ∃e’ [e’ e ∧ R(e’, x’)]] In addition to these restrictions, experiential verbs should further require that the subparts of the object be divided in time parallel to the structure of the primary event, and that their respective corresponding parts be simultaneous (i.e., you have each part of the dream as it happens). This can be captured by the following definition in (7), which in virtue of the temporal trace function τ (which maps events to the time they occupy), additionally requires the complement in question to be eventive if it is to be interpretable. I will henceforth refer to the combined properties of graduality and simultaneity as Temporally Linked Graduality (TLG).
(7) ∀R [SIMUL(R) ↔ ∀e, x [R(e, x) → τ (e) = τ (x)]
Aspect shift in stative verbs and their arguments
Conversely, a stative θ-relation is one that violates all of these. More generally, I assume that a stative VP must have a stative θ-relation (Stative Mapping to Object) and a stative event predicate (Stativity) as defined in (8).2
(8) a. ∀R [S-MAP(R) ↔ ∀e, e’, x [R(e, x) ∧ e’ e → R(e’, x)]]
b. ∀P [STV(P) ↔ ∀e, e’ [P(e) ∧ e’ e → P(e’)]] S-MAP requires that every subevent e’ relate to the same object x, while STV requires that every e’ qualify as the same event type P. Thus, every subevent of having is just another event of having, while each of the having subevents involves the same object in one’s possession, whether it be an apple or a dream. The practical effect of all of this is that if the object of have is not an event (and therefore does not have temporally distributed parts), have 2 cannot be inferred, and the VP will have only a stative (possessive) interpretation, and will necessarily be incompatible with the progressive. This is further corroborated in (9), which not only fails to allow the progressive, but its translation into the Spanish past tenses also lacks the gradual reading. Thus, (9c.ii) may be interpreted as inchoative (= Maria got $5), or given sufficient context, as a bounded state reading (= Maria had $5 once, but spent/lost it), but lacks an incremental reading. (9) a. Maria has $5. María tiene $5. b. *Maria is having $5. *María está teniendo $5. c. Maria had $5. i. María tenía $5. Maria had-impf.3sg $5. ii. #María tuvo $5. Maria had.pfv-3sg $5. Taken together, we see that the choice of direct object (or the choice among its various meanings) directly impacts the choice of meaning of the verb, and therefore the range of possible VP Aktionsarten, and consequently the range of possible aspectual morphemes which can convey that meaning. Stated the other way around, where there are multiple possible combinations of meaning, the choice of aspectually sensitive morphology may filter out certain verb meanings and concomitantly certain NP denotations in virtue of the same chain of compatibility restrictions. Which ever way it is stated, there is ultimately an inexorable link between the choice of morphology, the thematic relation to the object, and the 2. Space does not permit a formal proof to this effect.
Joshua Rodríguez
denotation of the object NP, which when clarified brings new insight to the term ‘aspect shift’. 4. Other cases of ‘aspect shift’ in Spanish In the remaining section, I will briefly discuss two additional cases of thematic polysemy between stativity and temporally linked graduality, and the concomitant interpretation of the complement of V as stative or eventive, and the corresponding compatibility between the VP and its aspectual morphology. In the first case, with consistir, I will look at the correlation between eventive vs. non-eventive NPs with this same set of aspectual morphemes. In the second, I will consider infinitival complements of saber, with potential vs. actualist interpretations which mimic the same (non-)stativity effects already discussed. 4.1
Consistir
VPs headed by the verb consistir may occur with stative (imperfective) morphology, and at times with non-stative (perfective/progressive) morphology. The fact that the imperfective (simple present or past) may yield a simple stative reading (i.e., not necessarily habitual) implies that consistir qualifies as a stative verb. However, it may also combine at times with the progressive or perfective, which additionally qualifies consistir as exhibiting ‘aspect shift’ in that the concomitant interpretations are aspectually different. In (10), the progressive is quite bad and the perfective odd, while the simple present and imperfective past are completely acceptable. This contrasts with (11), in which all forms are acceptable. (10)
Mario compró un suplemento alimenticio que {consiste / Mario buy-pfv.3sg a supplement nutritional that {consist.prs.3sg / consist-ía / *está consistiendo / ?consist-ió} consist-impf.3sg / be.prs.3sg consisting / consist.pfv-3sg} en varias vitaminas y minerales. in various vitamins and minerals. “Mario bought a nutritional supplement that {consists/ consisted/*was consisting/?consisted} of various vitamins and minerals.”
(11)
Esa fase d=el plan {consiste / consist-ía / That phase of=the plan {consist.prs.3sg / consist-impf.3sg / está consistiendo / consist-ió} en reemplazar todos be.prs.3sg consisting / consist.pfv-3sg} in replace.inf all
Aspect shift in stative verbs and their arguments
los ordenadores en el laboratorio. the computers in the lab. “That phase of the plan {consists/consisted/is consisting /consisted} of replacing all the computers in the lab.” If we look at the arguments involved in the two cases, we again find a difference in eventiveness. Because consistir describes the composition of its subject, it is sensitive to the substructure of the arguments involved. If what is described is a material object or substance, made up of other material objects or substances, then in virtue of their lack of temporal substructure, the only possible thematic interpretation will be stative, and so on down the line to imperfective and nonprogressive morphology. If, however, the arguments involved are eventive, meaning that they have a distributed temporal composition, there is an inherent possibility of tracking their constitution through time. Stated differently, if event x consists of various subevents y, then as subevents y transpire, they bring about the existence of x through time. It is precisely this sort of parallel temporal structure (TLG) that defies stative or static description in time, and therefore is predictably compatible with non-stative (perfective/progressive) morphology. The only remaining question, therefore, is how to interpret the imperfective examples in (11). To answer this question fully can be a bit tricky in light of the fact that the simple present and past imperfect may license a progressive type meaning as well (cf.€Cipria and Roberts (2000) and Rodriguez (2004a) who arguably treat this as coercion, though they overtly associate this effect directly with the imperfecto). In other words, it is possible to argue that the progressive version is superfluous.3 But even if we focus just on the imperfective/perfective contrast, we see that the former describes the subject in its potential or in the abstract, as if it were a plan rather than an action. Only the latter unambiguously represents the subject as a completed event or action that has been carried out by the accomplishment of its component subevents. We therefore arrive again at a distinction similar to the ambiguity of dream or problem in which the object may be viewed either atemporally constituted or as an event that transpires through time. 4.2
Saber
The final case for which I have space to discuss at any length is one that, to my knowledge, has not been previously mentioned in the literature on aspect shift. It 3. Some speakers find the progressive objectionable or difficult to interpret, though others find it completely acceptable. Those that accept it clearly see the eventive meaning: x is being brought about by y.
Joshua Rodríguez
again involves the verb saber, though this time in connection with its meaning of knowing how to do something. As stated above, saber may be classified as polysemous between describing a simple state of knowledge (saber1a) and an inchoative event of gaining knowledge (saber1b). Rodriguez (2007) presents evidence that the perfective and progressive of verbs like saber1 specifically manifest this inchoative meaning. However, saber is also polysemous between propositional knowledge (saber1 = ‘knowing that’) and procedural knowledge (saber 2 = ‘knowing how’). In this latter sense, it is used interchangeably between knowing how in potential (saber2a), and actually performing what one knows how to do (saber 2b). It is conceivable that the latter distinction is merely pragmatic, especially in light of the fact that purposefully accomplishing something entails having the knowledge to do it. Furthermore, in English there is no rigorous grammatical means of distinguishing these understandings. Again, the present tense requires stativity, thereby filtering out the performance reading, while the past allows either understanding, and gives pragmatic preference to the performance reading. (12) a. John knows how to get what he wants, so stay out of his way next the time he goes after something. b. John knew how to get what he wanted, i.╇ …and we were all amazed at how he did it. ii.╇ …but he lacked the courage to make it happen. In Spanish, on the other hand, the past tense requires aspectual disambiguation, which concomitantly establishes a clear distinction between what is respectively potential knowledge and what is actual accomplishment. (13) {Sab-ía / Sup-o} vencer a todos sus oponentes. {know-impf.3sg / know.pfv-3sg} defeat acc all his opponents. ‘He knew how to defeat all his opponents’. Furthermore, unlike English, which does not tolerate know-how-to sentences in the progressive at any register, these are not uncommon in conversational registers of Spanish (perhaps in some dialects more than others). As expected, the simple present implies knowledge purely as a potential, while the progressive version implies a temporal link between the ‘knowing’ event and the ‘performed’ event, so that the subject is also part way through defeating his opponents at the time of utterance. (14) {Sabe / Está sabiendo} vencer a {know-pres.3sg / be.pres.3sg knowing} defeat acc
Aspect shift in stative verbs and their arguments
todos sus oponentes. all his opponents. “He {knows/is knowing} how to defeat all his opponents.” I should reiterate at this point that what I argued in the previous section shows a clear chain of aspectual dependencies from nominal interpretation (the complement) to thematic interpretation (the verb) to morphological selection (the inflection), and vice versa. In this case, however, the complement is an infinitival IP. Though an IP does not vacillate between an object vs. event denotation (which was a crucial factor for nominal interpretation), the interpretive distinction correlated with stativity vs. temporally linked graduality is still relevant. For brevity, I will assume minimally that VP denotes an event predicate and that INFL denotes a temporal relation that binds the event argument of VP. That gives the following simplified translation for [IP PRO to defeat opponents], where i is the interval that locates the defeating event in time, and x translates PRO. (15) λi ∃e [defeat_opponents’(x)(e) ∧ τ(e) ⊆ i] Now for x to know in general how to do action Q implies logically that at any given moment i’ (within a relevant period), x has the potential to do Q (i.e., x does Q in at least one relevant world w at an interval i” of which i’ is a part). I therefore propose (16) to formalize the truth conditions of statements of the form x knows how to do Q. (16) For all intervals i’ (within the time of knowing), there is an interval i” such that i’ ⊆ i”, a possible world w (among those worlds that represent x’s knowledge), and [[λi ∃e [Q(x)(e) ∧ τ(e) ⊆ i](i”)]]w = 1. To properly capture this intensional relation, the complement of saber 2a ‘know how in general’ must be the intension of the denotation of the embedded IP. A translation for saber + IP is given in (17). (17) λQ λx λe [saber 2a(ˆλi ∃e’ [Q(x)(e’) ∧ τ(e’) ⊆ i])(x)(e) On the other hand, to have the knowledge to perform a specific event on a specific occasion need not make reference to potentials in other possible worlds just in case the action is actually performed in the real world. Thus, to characterize the performance reading of saber 2 we need only stipulate that the temporal argument i of the embedded IP be anchored by the time of the matrix event, as in (18). It should be further noted that those models that satisfy (18) also satisfy (16), though not vice versa, just as it should be. (18) [[saber 2b(λi ∃e’ [Q(x)(e’) ∧ τ(e’) ⊆ i])(x)(e)]] = 1 iff [[λi ∃e’ [Q(x)(e’) ∧ τ(e’) ⊆ i](τ(e))]] = 1
Joshua Rodríguez
Now even if we assume that this is a pragmatic distinction, the θ-relations associated with the corresponding morphology is still meaningful. Because of the universal quantification over times in (16), the θ-relation to the IP complement qualifies as stative (i.e., every subevent of knowing how refers to the same potential action), which in turn facilitates imperfective aspect, and vice versa. On the other hand, given the direct temporal link between the knowing event e and the performed event e’ described in (18), any subevent of knowing will be thematically related to a corresponding subpart of e’, which is the actual subevent to be accomplished in the real world at that moment. In other words, subevents of ‘knowing how’ translate into knowing how to accomplish each particular of the overall task as it transpires, which is therefore non-stative in virtue of violating (8a), which in turn facilitates perfective or progressive aspect, and vice versa. 5. Conclusion In this paper I have argued that there are interrelated aspectual dependencies between the verb and its arguments (which determine the Aktionsart of the VP in a compositional fashion), and between the VP Aktionsart and its inflectional morphology (which must agree for selectional requirements). I have argued that this type of arrangement greatly reduces the need for aspectual coercion in many instances, particularly with regard to perfective and progressive aspect. I have argued that what might appear to be a direct alteration of the interpretation of the verb or noun by the morphology that is applied, may turn out to be an indirect chain of compatibility effects that function to filter rather than modify meanings already inherent in the VP. Finally, to maximize the details of exposition, I have limited my discussion to only three cases, though examples could be multiplied. To mention a few in closing (though not all of which are equally used in both Spanish and English), have2 can be seen with other types of experiences and events (e.g., luck, fun, epiphany, difficulty, etc.) or intervals (e.g., Monday, time, etc.). Perfective and progressive examples of consistir are rampant in connection with events of diverse types (e.g., trabajo ‘job’, proceso ‘process’, etc.). Likewise, there are other modal verbs that behave like saber2, such as poder ‘be able’ and tener que ‘to have to’, which contrast potential vs. actual performance.
Aspect shift in stative verbs and their arguments
References Bach, Emmon. 1986. “The Algebra of Events”. Linguistics and Philosophy 9.5–16. 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. de Swart, Henriëtta. 1998. “Aspect Shift and Coercion”. Natural Language and Linguistic Theory 16.347–385. 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. Krifka, Manfred. 1989. “Nominal Reference, Temporal Constitution and Quantification in Event Semantics”. Semantics and Contextual Expressions ed. by Renate Bartsch, Johan van Benthem, and Peter van Emde Boas, 75–115. Dordrecht: Foris. Krifka, Manfred. 1992. “Thematic Relations as Links Between Nominal Reference and Temporal Constitution”. Lexical Matters ed. by Ivan A. Sag and Anna Szabolcsi, 29–53. Stanford, Calif.: CSLI Publications. Michaelis, Laura. 2004. “Type Shifting in Construction Grammar: An Integrated Approach to Aspectual Coercion”. Cognitive Linguistics 15.1–67. Michaelis, Laura. 2005. “Entity and Event Coercion in a Symbolic Theory of Syntax”. Construction Grammars: Cognitive Grounding and Theoretical Extensions ed. by Jan-Ola Ostman & Mirjam Fried, 45–88. Amsterdam & Philadelphia: John Benjamins. Parsons, Terence. 1990. Events in the Semantics of English: A Study in Subatomic Semantics. Cambridge, Mass.: MIT Press. Pulman, Stephen. 1997. “Aspectual Shift as Type Coercion”. Transactions of the Philosophical Society 95:2.279–317. 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”. West Coast Conference on Formal Linguistics 23, 677–689. Rodríguez, Joshua. 2007. “Macro Events and ‘Aspect Shift’ in Spanish”. Proceedings from the Linguistic Symposium on the Romance Languages 2006 ed. by José Camacho, Nydia Flores-Ferrán, Liliana Sánchez, Viviane Deprez & María José Cabrera, 233–246. Amsterdam & Philadelphia: John Benjamins. Smith, Carlota. 1991. The Parameter of Aspect. Dordrecht, Netherlands: Kluwer Academic Publishers. Taylor, Barry. 1977. “Tense and Continuity”. Linguistics and Philosophy 1:2.199–220. Vendler, Zeno. 1967. Linguistics in Philosophy. Ithaca, New York: Cornell University Press. Verkuyl, Henk J. 1993. A Theory of Aspectuality: The Interaction between Temporal and Atemporal Structure. Cambridge: Cambridge University Press. Verkuyl, Henk J. 1999. Aspectual Issues. Stanford, Calif.: CSLI Publications.
part 5
Psycholinguistics
Experimenting with wh-movement in Spanish Grant Goodall This paper provides evidence for an analysis of subject inversion in whquestions in Spanish and demonstrates that techniques of experimental syntax play an important role in developing such analyses. The techniques used show that there is gradience in judgments of wh-questions depending on the nature of the filler and of the intervening subject. The facts fall out from the interplay of straightforward properties of the syntax (e.g.€wh-movement, preverbal or postverbal placement of the subject) with straightforward properties of the processor (a common pool of limited resources to process wh-dependencies and establish discourse referents). The analysis predicts a correlation between the Overt Pronoun Rate in any given variety and the ability of a whdependency to tolerate an intervening subject, and the difference between Caribbean and mainland Latin American Spanish confirms this.
1. Introduction: Experimental syntax This paper has two interrelated goals: first, to provide evidence for a particular analysis of word order restrictions in wh-questions in Spanish, and second, to demonstrate that techniques of experimental syntax play a near essential role in allowing us to uncover this type of evidence.1 I use the term “experimental syntax” here to refer to the gathering of acceptability judgments with non-linguist subjects, clearly formulated instructions, a practice and/or training session, a factorial design for the construction of sentences, a counterbalanced and randomized sentence list, contextualization or other means to encourage uniform comprehension of the sentences across subjects, quantitative results, statistical analysis of these results, or any subset of these. As has been discussed in the literature (e.g.€ Cowart (1997), Myers (2009), Schütze (1996), Sprouse (2007)), there are a number of clear advantages to adopting at least some of these experimental techniques. They can give us more certainty 1. This paper is part of an ongoing study using experimental techniques to explore whquestions in Spanish. For earlier results from this study, see Goodall (2007, 2008, in press).
Grant Goodall
about the status of data where there have been disputes or doubts, as well as more precision in dealing with subtle contrasts among sentences. More generally, they allow the evidence used by syntacticians to be expressed in terms that the wider cognitive science community can understand and evaluate. The advantage of experimental techniques that I will focus on here is that they allow us to capture gradience in judgments in a precise and reliable way. I will show not only that such gradience exists in domains where it has not been fully recognized before, but also that attempting to account for it leads us to new ways of studying and conceiving of syntax. 2. Subject inversion in wh-questions My empirical point of departure is the inversion contrast in (1), a phenomenon well known in generative syntax since at least Torrego (1984). (1) a. *Qué Juan leyó en la biblioteca? what Juan read in the library b. Qué leyó Juan en la biblioteca? what read Juan in the library ‘What did Juan read in the library?’ What is not as well known is that the degree of unacceptability of the prohibited pattern in (1a) appears to vary depending on the nature of the intervening subject, as seen in Figure 1, which shows results from an experiment in which native speakers of Mexican Spanish judged the acceptability of such sentences on a fivepoint scale (N = 23, see Goodall (2008) for further details). 5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 none
tú
ellos
el niño
Figure 1.╇ Intervening subject in wh-question: separate circles indicate significant difference
Experimenting with wh-movement in Spanish
The most acceptable wh-question is clearly one with no intervening subject at all, but otherwise the hierarchy in (2) seems to hold, where 2nd-person pronouns are clearly better tolerated as interveners than lexical DPs, with 3rd-person pronouns appearing to have an intermediate status.
(2) 2p pronoun > 3p pronoun > lexical
These differences are very fine-grained and seem unlikely to be due to the syntax proper. One could reasonably speculate that they result from differences in ease of processing, since this factor is known to affect acceptability. Wh-dependencies place special demands on the processor, in that the filler must be held in working memory until the subcategorizing verb is processed, at which point a gap can be posited and linked to the filler (e.g.€Gibson 2000). Importantly, processing this dependency and processing the reference of the intervening subject appear to make use of the same limited pool of resources (Warren and Gibson 2002). In this way, we expect the nature of the intervening subject to affect the ease with which the wh-dependency itself can be processed. Since lexical DPs have been claimed to require more processing resources than 3rd-person pronouns, which in turn require more than 2nd-person pronouns (Warren and Gibson 2002), the hierarchy in (2) now follows, and the idea that processing factors account for these fine differences gains plausibility. Further evidence that processing effects play a role in the unacceptability of (1a) comes from the fact that as we manipulate the wh-dependency in ways that should change its ease of processing, the level of acceptability of the sentence changes accordingly. For instance, complex wh-phrases are believed to be able to survive in working memory at a higher activation level than bare wh-words, and this allows them to better tolerate intervening material (e.g.€ Kluender (1998), Hofmeister (2007)). As we would expect, then, sentences like (1a) improve significantly when the wh-phrase is complex, as seen in Figure 2, which shows results from an experiment of the same type as in Figure 1 (N = 26, further details in Goodall (2008)). Likewise, individual bare wh-words should vary in their ability to tolerate an intervening subject. Those that do not produce a filler-gap dependency at all, such as por qué ‘why’, should be oblivious to intervening material, and those that are similar to the intervening subject may be particularly affected by it, given the fact that similarity between filler and intervening material appears to lead to increased processing difficulty (e.g., Gordon, Hendrick and Johnson 2004). One might then reasonably expect that the nominal wh-words qué ‘what’ and quién ‘who’, which bear a clear category similarity to the subject DP, would tolerate an intervening subject less well than adverbial dónde ‘where’ and cuándo ‘when’, which do not bear this type of similarity to the subject. This expectation is supported by the results shown in Figure 3 (from the same experiment as in Figure 1; wh-words represented by their English glosses).
Grant Goodall 5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 Bare wh-word
Complex wh-phrase
Figure 2.╇ Type of wh-phrase in wh-questions with intervening subject: separate circles indicate significant difference 5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 what
who
where
when
why
Figure 3.╇ Effect of wh-word on wh-question with intervening subject: separate circles indicate significant difference
Note that inversion in Spanish is often said to exhibit an argument/adjunct asymmetry (e.g., Torrego 1984, Suñer 1994), but this characterization appears to be inadequate. Figure 3 shows that adjunct wh-words are better able to tolerate an intervening subject than are argumental wh-words, but such questions with where and when are still significantly degraded. Moreover, wh-words like where, though typically adjuncts, may also be arguments, yet speakers report to me that they do not find a contrast between these two uses with an intervening subject:
Experimenting with wh-movement in Spanish
(3) a. Dónde tú pusiste el libro? where you put the book ‘Where did you put the book?’ b. Dónde tú compraste el libro? where you bought the book ‘Where did you buy the book?’ We have seen at this point the following hierarchy in the ability of a wh-phrase to tolerate an intervening subject:2
(4) why > complex wh-phrase > how > where/when > what/who
Overall, then, as the ease of processing of the wh-dependency increases or decreases, depending on the nature of the filler or the nature of the intervening subject, the acceptability of the sentence increases or decreases correspondingly, just as we would expect. 3. Properties of subjects in Spanish The processing factors examined so far clearly predict the contrast seen at the outset in (1). In fact, we expect (1a) to be particularly bad, since the intervening subject is a lexical DP, which is the type that uses the most processing resources (see (2)), and the filler is qué ‘what’, which is of the type most easily impaired by intervening material (see (4)). (1b), on the other hand, should be particularly good, since its filler-gap dependency is resolved immediately. Moreover, intervening subjects in Spanish have specific properties that should make them even more of a drain on processing resources than their counterparts in a language like English. First, subjects in Spanish may be either overt or null. In the latter case, verbal inflection provides only person/number information about the intended referent, which must therefore be relatively salient and accessible in the discourse. Overt pronouns, in contrast, typically provide gender information also, and are used to refer to less accessible discourse entities, or for cases of emphasis, contrast or focus (Luján 1999). As we would then expect, overt pronouns appear to be more difficult to process than null ones in Spanish (Callahan et al. 2007), and as we saw in (2) (and Figure 1), lexical DPs are more difficult still. In line with this reasoning, we expect overt pronouns in Spanish, with their relatively inaccessible discourse referents, to be more difficult than their counterparts in English. 2. Manner cómo (‘how’) is provisionally placed here based on informant work.
Grant Goodall
Second, subjects in Spanish may be preverbal or postverbal, with complex implications for information structure in either case (e.g., Marandin 2003). Simplifying, preverbal subjects are associated with a categorical judgment type, i.e. one in which the subject is presented first as an individual and then the predicate is affirmed or denied of it. This is in contrast to the thetic judgment type, where the subject and the predicate are presented as a single event. What is relevant for our purposes is that highly individuated referents, such as occurs with a categorical judgment, appear to require relatively high processing costs (Warren and Gibson 2002). Given the model of processing we have been assuming, in which processing of intervening referents makes processing of the wh-dependency more difficult, both of the above properties of subjects in Spanish would be expected to play a role in the decreased acceptability of sentences like (1a). Other factors may be at work as well, however. It may be, for instance, that the information structure associated with preverbal subjects is at least partly incompatible with interrogative force. This view receives support from the claim in the literature that questions with preverbal subjects (e.g.€yes/no questions or wh-questions with por qué ‘why’) do not have a “true interrogative” interpretation (Escandell Vidal (1999), Gallego (2006)). If correct, this could be a contributing factor, in addition to the processing considerations already discussed, for the low acceptability of (1a).3 4. Accounting for variation At this point, we have taken the experimentally elicited facts from Figure 1 regarding the effect of the type of intervening subject on the (un)acceptability of the wh-question and explored the potential role of processing in explaining these subtle contrasts. We have seen that not only were known properties of processing able to do this, but they also appeared to offer an account of the much sharper contrast in (1), possibly together with information structure factors. The picture that emerges from this is appealing: Straightforward properties of the syntax (e.g.€wh-movement, placement of the subject in preverbal or postverbal position) interact with straightforward properties of the processor (a common pool of limited resources to process wh-dependencies and establish discourse referents), and
3. Such an analysis may also be able to shed light on extraction out of embedded clauses in Spanish, but space limitations preclude an exploration of that topic here. See Torrego (1984), Goodall (1993), and Baković (1998) for discussions of the phenomena.
Experimenting with wh-movement in Spanish
the effect known as inversion (as in (1)) results.4 In some ways, however, this picture is too appealing. It nicely accounts for the facts that we have seen, but it would seem to leave no way to account for cases where subjects do intervene in wh-dependencies. This phenomenon occurs both in Romance and beyond, but in fact we can profitably explore it even within Spanish. To begin, let us consider the use of overt pronouns across dialects of Spanish. For many years, there have been indications in the literature that the rate of use of overt pronouns varies substantially according to region, as shown in Table€1. Though this amount of variation does not seem implausible impressionistically, caution is nonetheless in order, since differences in the type of corpus and the method of calculation used across these studies could be accounting for much of the effect. These problems are overcome in a recent study by Otheguy, Zentella and Livert (2007), where a uniform interview format was used with 39 speakers from both Caribbean and mainland Latin American regions (all were residents of New York, but their length of residence was < 6 years and their age of arrival > 16). For each speaker, the number of overt pronominal subjects was calculated as a percentage of overall subject pronoun use (excluding lexical subjects), yielding that person’s “Overt Pronoun Rate.” Their results are summarized preliminarily in Figure 4. We saw earlier that overt (as opposed to null) pronouns in Spanish are used to refer to relatively inaccessible discourse referents, yet given the uniform interview format, it seems unlikely that the Caribbean speakers are systematically bringing up less accessible referents than their mainland counterparts. Rather, the threshold for how accessible a referent must be for the pronoun to be null must be higher in Caribbean than in mainland Latin American Spanish, resulting in less frequent null pronoun use among the former group. The higher Overt Pronoun Rate in Caribbean Spanish is generally taken to be correlated with a higher rate of use of preverbal subjects (e.g., Toribio 2000). If Table 1.╇ Overt subjects as percentage of total in five locations Rate of use 16% 20% 25% 38% 39%
Location and source Ciudad Juárez, Mexico (Strongman 1995) Caracas, Venezuela (Bentivoglio 1987) Madrid, Spain (Enríquez 1984) Santiago, Chile (Cifuentes 1980–1) Los Angeles, USA (Silva-Corvalán 1977)
4. For the analysis presented here, nothing hinges on the exact mechanisms underlying the syntactic properties mentioned.
Grant Goodall 40
36
30
24
20 10 0 Caribbean
Mainland
Figure 4.╇ Overt Pronoun Rate in two regions (Otheguy et al. (2007))
this observation is correct, we come to a similar type of conclusion: the threshold for how clearly categorical the judgment type must be in order for the subject to be preverbal must be lower in Caribbean Spanish than in mainland varieties, with the result that the referents of preverbal subjects are less individuated. Putting the rates for overt pronouns and for preverbal subjects together, we can conclude that overtness and preverbal position do not signal the same degree of discourse inaccessibility or individuation in Caribbean as they do in mainland Spanish. Given the processing model we discussed earlier, we now make a clear prediction: the effect of an intervening subject on the ease of processing a wh-dependency should be smaller in Caribbean than in mainland Spanish, and the corresponding sentences should thus have a higher level of acceptability. There are many indications in the literature that this prediction and the analysis behind it are correct. First, instances of intervening subjects in matrix wh-questions show up in spontaneous discourse in Caribbean Spanish in a way that is not attested in other varieties, as seen in (5) (from Toribio (2000)). (5) a. Qué yo les voy a mandar a esos muchachos? what I dat.3pl go to send to those boys ‘What am I going to send to those boys?’ b. Qué número tú anotaste? what number you write_down ‘What number did you write down?’ c. Qué ese letrero dice? what that sign say ‘What does that sign say? d. Cuánto un medico gana? how_much a doctor earn ‘How much does a doctor make?’
Experimenting with wh-movement in Spanish
Second, although there is considerable disagreement in the literature as to what type of intervening subject is possible, Ordóñez and Olarrea (2006) note that an interesting hierarchy emerges from surveying this literature:
(6) 2p pronoun > 3p pronoun > lexical
That is, all researchers report that intervening 2nd-person pronouns are possible, some report the same for 3rd-person pronouns, and only a subset of this last group reports the possibility of lexical DPs. This of course matches the hierarchy that we saw in (2) and is what we would expect. Increased processing difficulty should lead to a decreased level of acceptability, which should lead to less frequent reporting as “grammatical”. It may also be that the differences in what researchers report reflect true differences in the level of acceptability for speakers or dialects, which in turn reflect differences in the level of discourse inaccessibility and individuation signaled by overt preverbal subjects. This too would predict the hierarchy in (6). Third, Ordóñez and Olarrea (2006) find gradient acceptability in a survey of 65 speakers of Dominican Spanish, summarized in Figure 5. The striking result here is that the hierarchy of acceptability is the same as what we saw in (2) (and Figure 1) for non-Caribbean Spanish, but the level of acceptability is considerably higher here. It is difficult to be precise about this, since the subject task in the two experiments was different, but it seems very unlikely that the low acceptability seen in Figure 1 for intervening 2nd-person subjects, for instance, would obtain with these Dominican speakers, given their unanimous rating of this sentence type as “acceptable”.
Acceptability rate (%)
100 90 80 70 60 50 40 30 20 10 0 2nd person
3rd person
lexical
complex lexical
Figure 5.╇ % of Dominican speakers who rate intervening subject as acceptable in wh-question
Grant Goodall
Fourth, Ordóñez and Olarrea show that the acceptability of intervening subjects declines sharply for Dominican speakers when the subject is stressed, conjoined, or modified: (7) a. *Qué tú quieres? what you want ‘What do you want?’ b. *Qué tú y él comieron? what you and he ate ‘What did you and he eat?’ c. *Qué tú mismo comes? what you self eat ‘What do you yourself eat?’ In all three of these cases, we would expect that processing the subject would require more than the usual amount of resources, with a concomitant reduction in the processor’s ability to handle the wh-dependency easily. Finally, in a corpus study of Puerto Rican Spanish, Gutiérrez Bravo (2007) shows that only 10% of wh-questions with an overt subject have this subject in an intervening position. This is an important finding, because in the analysis developed here, we expect intervening subjects to be easier for the processor in Caribbean Spanish than in other varieties, but they should not be easy in any absolute sense. Having the subject in a non-intervening position should always be easier, for both Caribbean and mainland speakers, and the high rate of use of non-intervening subjects among Puerto Rican speakers in spontaneous discourse lends support to this idea. In short, the independent evidence seen earlier that overt preverbal subjects do not signal the same degree of discourse accessibility and individuation in Caribbean as in mainland Latin American Spanish predicts that Caribbean speakers will find wh-questions with such subjects relatively easy to process, compared to mainland speakers, and we have now seen several pieces of evidence that this prediction is correct. This question of how to account for the difference between Caribbean and mainland wh-questions has been a longstanding and well-known problem in Romance syntax, and it is thus significant that the analysis developed here appears to shed light on it. Interestingly, this analysis does not posit a syntactic difference between the two varieties, despite the ostensibly syntactic nature of the problem. Both have wh-movement, permit null subjects, and allow for placement of overt subjects in either preverbal or postverbal position. The difference stems, as we have seen, from the degree of accessibility and individuation ascribed to overt
Experimenting with wh-movement in Spanish
preverbal subjects. Since these factors affect the processing of subjects, and since the processing involved competes for limited resources with that of the wh-dependency, the result is that factors such as these have consequences that appear “syntactic”. If this analysis is correct, the Caribbean/mainland difference is not parametric, in the sense that it cannot be reduced to different settings of any putative parameter. In fact, unlike other treatments in the literature, the analysis here does not lead us to expect that the difference is in any way binary. The notions of “discourse accessibility” and “individuation” do not lend themselves to +/– values and should instead be subject to fine differences of degree from one region to another (or in fact, from one speaker to another). The study by Otheguy, Zentella and Livert (2007) discussed earlier actually corroborates this expectation. They found that the Overt Pronoun Rate differences surface not just with regard to the Caribbean and the mainland, but also when comparing smaller regions within these. This may be seen in Figure 6, a more fine-grained version of Figure 4. The same interpretation that we gave to Figure 4 applies here as well: The threshold for how accessible a referent must be for the pronoun to be null seems to differ from region to region, with the result that the frequency of overt pronouns varies.5 Once again, we clearly predict that this will have consequences for the placement of subjects in wh-questions, but this time the prediction is more fine-grained. Intervening subjects should be more acceptable for Dominican speakers than for Cubans, for instance, and for Ecuadorians compared to Mexicans. That is, we expect a correlation between the Overt Pronoun Rate and the degree of acceptability of wh-questions with intervening subjects. 45 40 35 30 25 20 15 10 5 0
41
Dom Rep
38
New York-born
35
Puerto Rico
33
Cuba
27
Ecuador
24
Colombia
19
Mexico
Figure 6.╇ Overt Pronoun Rate in seven regions (Otheguy et al. (2007)) 5. Presumably the frequency of use of preverbal subjects varies in a similar way, but this is not as well documented. Though it generally seems to correlate with the frequency of use of overt pronouns, it is not known whether this correlation always holds.
Grant Goodall
This is a very strong prediction and it is not known at this point whether it holds, but what data are available suggest that it is at least on the right track. Baković (1998) shows that there is considerable variation in the acceptability of intervening subjects, much more than the traditional Caribbean/mainland split suggests. He does not examine the level of acceptability of such subjects per se, but rather the set of wh-words that tolerate these subjects in each variety. Recall that we saw in Figure 3 that wh-words vary in how well they can do this. Those that do not correspond to a gap (such as por qué ‘why’) are not affected by an intervening subject, while those that do have a gap are affected. Among this latter group, nominal wh-words (i.e. those that are most similar to the intervening subject) show a stronger effect than adverbial wh-words. Since there is no reason to believe that the meaning or basic syntactic properties of the individual whwords would vary from one region to another, we would expect this hierarchy to be stable across varieties. In contrast, the degree to which intervening subjects divert resources away from processing the wh-dependency is expected to vary by region, as we have seen. Putting together this variability in the properties of the subject with the lack of variability in the properties of the wh-word, we expect that the level of acceptability of a given wh-question with an intervening subject will vary, but the hierarchy that we saw in Figure 3 and (4) will always be respected. For instance, it should be impossible for a speaker to accept an intervening subject with what but reject it with where. More generally, we expect to find the patterns of acceptability in (8) (where a starred form indicates that this wh-word does not tolerate an intervening subject), but not, for example, those in (9).
(8)
a. b. c. d. e.
*why *how *when/where *what/who why *how *when/where *what/who why how *when/where *what/who why how when/where *what/who why how when/where what/who
(9) a. why how *when/where what/who b. *why *how *when/where what/who Strikingly, these are exactly the results that Baković obtained: All five varieties in (8) are attested, whereas varieties as in (9) appear not to exist. We thus have preliminary evidence that dialects of Spanish show exactly the range of variation that we predict. We also predict that this variation in wh-questions will correlate with the Overt Pronoun Rate (e.g.€among the varieties in (8), we expect (8a) to have the lowest Overt Pronoun Rate and (8e) to have the highest), although this prediction has not yet been tested.
Experimenting with wh-movement in Spanish
As we have now seen, the study of Spanish dialects suggests that much of the variation in the ability of a wh-dependency to tolerate an intervening subject depends on the degree of discourse accessibility or individuation associated with overt preverbal subjects in a particular dialect. In principle, of course, there could be syntactic differences as well, although we have not needed to invoke these in this case. In other cases, though, syntactic differences could play a role. Spanish does not appear to make use of T-to-C movement in wh-questions (e.g.€Suñer 1994), for instance, but if some language variety that was otherwise just like Spanish did, the effects in Figures 1–3 would presumably disappear, since we would expect T-to-C movement to apply uniformly regardless of the nature of the wh-phrase or the intervening subject (see Goodall (2007, 2008) for discussion).6 5. Conclusion As stated at the outset, this paper has the twin goals of arguing both for a particular analysis and for experimental techniques in the study of syntax. As for the first, we have provided evidence for an analysis of the position of subjects in whquestions in Spanish which crucially relies on the interplay of syntactic properties of the language and processing considerations, especially the idea that processing a wh-dependency and a discourse referent both make use of the same set of limited shared resources. The resulting analysis yields appealing solutions to some traditional problems in this area, and also opens up intriguing new research questions to be explored. As for the second goal, we have seen that the analysis proposed here relies crucially on the fine-grained distinctions in acceptability that can be reliably captured by experimental techniques. Perhaps even more interestingly, some of the predictions that the analysis makes, such as the correlation between the Overt Pronoun Rate and the ability of wh-dependencies to tolerate intervening subjects, require a level of precision in the data that would be difficult, if not impossible, to obtain using only traditional techniques.
6. English has T-to-C movement and is also unlike Spanish in that it does not have null/ overt or preverbal/postverbal contrasts in its subjects. Overt preverbal subjects in English thus do not have the discourse value that they do in Spanish and thus do not appear to impose as heavy a processing burden, with the result that intervening subjects are allowed in whdependencies.
Grant Goodall
References Bakovic, Eric. 1998. “Optimality and inversion in Spanish”. Is the Best Good Enough? ed. by Pilar Barbosa et al.,35–58. Cambridge: MIT Press. Bentivoglio, Paola. 1987. “La posición del sujeto en el español de Caracas: Un análisis de los factores lingüísticos y extralingüísticos”. Studies in Caribbean Spanish Dialectology ed. by R.M. Hammond and and M. Resnick, 13–23. Washington: Georgetown University Press. Callahan, Sarah, Janet Nicol, Tracy Love, & David Swinney. 2007. “All pronouns are not created equal: The processing and interpretation of null and overt pronouns in Spanish”. Poster presented at 20th Annual CUNY Sentence Processing Conference, La Jolla, Calif. Cifuentes, Hugo. 1980–1. “Presencia y ausencia del pronombre personal sujeto en el habla culta de Santiago de Chile”. Boletín de Filología XXXI. 743–752. Cowart, Wayne. 1997. Experimental Syntax: Applying objective methods to sentence judgments. Thousand Oaks: Sage Publications. Enríquez, Emilia V. 1984. El pronombre personal sujeto en la lengua española hablada en Madrid. Madrid: Consejo Superior de Investigaciones Científicas. Escandell-Vidal, M. Victoria. 1999. “Los enunciados interrogativos. Aspectos semánticos y pragmáticos”, Gramática Descriptiva de la Lengua Española ed. by Ignacio Bosque & Violeta Demonte, Vol 3, cap.€61. pp.€3929–3991. Madrid: Espasa-Calpe. Gallego, Angel J. 2006. “Phase effects in Iberian Romance”. Selected Proceedings of the 9th Hispanic Linguistics Symposium, ed. by Nuria Sagarra and Almeida Jacqueline Toribio, 43–55. Somerville: Cascadilla Proceedings Project. Gibson, Edward. 2000. “The dependency locality theory: a distance-based theory of linguistic complexity”. Image, Language, Brain, ed. by Y. Miyashita, A. Marantz and W. O’Neil 95–126, Cambridge: MIT Press. Goodall, Grant. 1993. “SPEC of IP and SPEC of CP in Spanish wh-questions.” Linguistic Perspectives on the Romance Languages, ed. by M. Mithun, G. Perissinotto and E. Raposo 199–209. Amsterdam/Philadelphia: John Benjamins. Goodall, Grant. 2007. “Inversion in wh-questions in child Romance and child English.” Romance Linguistics 2006: Selected papers from the 36th Linguistic Symposium on Romance Languages (LSRL), ed. by María José Cabrera, José Camacho, Viviane Déprez, Nydia Flores and Liliana Sánchez 169-182. Amsterdam/Philadelphia: John Benjamins. Goodall, Grant. 2008. “The limits of syntax in inversion”. Proceedings of the 41st Annual Meeting of the Chicago Linguistic Society 161–174. Goodall, Grant. In press. “Syntactic satiation and the inversion effect in English and Spanish wh-questions”. Syntax 14. Gordon, Peter C., Randall Hendrick, and Marcus Johnson. 2004. “Effects of noun phrase type on sentence complexity”. Journal of Memory and Language 51, 97–114. Gutiérrez Bravo, Rodrigo. 2008. “Topicalization and preverbal subjects in Spanish wh-interrogatives”. Selected Proceedings of the 10th Hispanic Linguistics Symposium, ed. by Joyce Bruhn de Garavito and Elena Valenzuela, 225–236. Somerville, MA: Cascadilla Proceedings Project. Hofmeister, Philip.€ 2007. Representational Complexity and Memory Retrieval in Language Comprehension. Ph.D. dissertation, Stanford University. Kluender, Robert. 1998. “On the distinction between strong and weak islands: A processing perspective”. Syntax and Semantics 29: The Limits of Syntax, 241–279.
Experimenting with wh-movement in Spanish Luján, Marta. 1999. “Expresión y omisión del pronombre personal”. Gramática descriptiva de la lengua española, ed. by Ignacio Bosque and Violeta Demonte, vol. I, 1275–1315. Madrid: Espasa-Calpe. Ordóñez, Francisco and Antxon Olarrea. 2006. “Microvariation in Caribbean/non Caribbean Spanish Interrogatives”. Probus 18, 59–96. Marandin, Jean-Marie. 2003. “Inversion du sujet et discours dans les langues romanes”. Langues romanes: problèmes de la phrase simple, ed. by D. Godard, 345–392 Paris: éditions du CNRS. Myers, James. 2009. “Syntactic judgment experiments”. Language and Linguistics Compass 3, 406–423. Otheguy, Ricardo, Ana Celia Zentella, and David Livert. 2007. “Language and dialect contact in Spanish in New York: Toward the formation of a speech community”. Language 83, 770–802. Schütze, Carson. 1996. The Empirical Base of Linguistics: Grammaticality judgments and linguistic methodology. Chicago: University of Chicago Press. Silva-Corvalán, Carmen. 1977. A Discourse Study of Some Aspects of Word Order in the Spanish Spoken by Mexican-Americans in West Los Angeles. Ph.D. dissertation, UCLA. Sprouse, Jon. 2007. A program for experimental syntax: Finding the relationship between acceptability and grammatical knowledge. Ph.D. dissertation, University of Maryland. Strongman, Roberto. 1995. Is Border Spanish becoming a non-pro-drop language? M.A. thesis, University of Texas at El Paso. Suñer, Margarita. 1994. “V-movement and the licensing of argumental wh-phrases in Spanish”. Natural Language and Linguistic Theory 12, 335–372. Toribio, Jacqueline Almeida. 2000. “Setting parametric limits on dialectic variation in Spanish”. Lingua 110, 315–341 Torrego, Esther. 1984. “On inversion in Spanish and some of its effects”. Linguistic Inquiry 15, 103–129. Warren, Tessa and Edward Gibson. 2002. “The influence of referential processing on sentence complexity”. Cognition 85, 79–112.
How Spanish phonotactics informs psycholinguistic models of speech production Michael Shelton, Chip Gerfen & Nicolás Gutiérrez Palma
Occidental College, The Pennsylvania State University & Universidad de Jaén This study emphasizes the importance of close examination of languagespecific phonotactic patterns when testing models of language processing. The results of a delayed naming task testing native speakers’ reactions to nonwords containing licit vs. proscribed patterns of stress and syllable structure offer empirical data to inform models of speech production. While the findings generally support a postlexical locus of syllable-level encoding (Levelt et al. 1999), the data also pose challenges for this model. It is argued that the model must identify more clearly the locus of and the interaction between stress-level and syllable-level encoding. The findings also suggest that the self-monitoring system may track phonotactic patterns in addition to conceptual accuracy.
1. Introduction Psycholinguists modeling speech production and linguists modeling grammatical competence often view language from distinct perspectives. Speech production models, for example, examine the cognitive processes involved in speaking words, from conceptual initiation to lexical activation to phonological encoding to articulation. Theoretical linguistics, by contrast, often concerns itself with the systemic characterization of licit versus proscribed structures (syntactic, morphological, phonological, etc.) in particular languages within the context of putatively universal structural principles or interacting constraints. Given these diverging concerns, the finely-grained grammatical detail at the center of linguistic theorizing is rarely a focus of psycholinguistic models of production, nor is the relevance of such detail immediately obvious. The present study argues that close examination of language-specific phonotactic patterns in Spanish can pose important challenges for psycholinguistic models of speech production. The remainder of this paper is organized as follows: in §1 we consider one prominent model of speech production; §2 provides a brief overview of our language-specific
Michael Shelton, Chip Gerfen & Nicolás Gutiérrez Palma
data source, the interaction of Spanish stress and syllable structure; §3 describes the experimental design; and in §4 and §5 we discuss the results of the study and the implications for speech production models. 2. A specific model to test Psycholinguists modeling production examine the cognitive processes involved in speaking. Most models agree on three basic levels of processing. Activation feeds from a conceptual level, to a “word” level, to a “form” level, which feeds to articulation. While the exact nature of the flow of activation between levels is highly disputed in the literature (cf.€ Dell 1986, 1988; Shattuck-Hufnagel 1987, 1992; Stemberger 1985; Levelt 1989; Levelt et al. 1999; Garrett 1980) the general top-down nature of processing is a base assumption. To illustrate, we consider one prominent speech production model. Levelt, Roelofs, & Meyer’s (1999) model is serial in nature, meaning that activation begins at the highest level and spreads downward in a discrete fashion. Take for example the word sparrow. When an English speaker intends to speak this word, associated semantic concepts are activated (e.g.€“flies”, “lives in trees”, etc.). The conceptual nodes spread their activation to the lemma level, where lexical items associated with the concepts become activated. At this stage, the most highly activated lexical entry is selected. In our example, this would be the entry for sparrow. Stored with this entry is morphosyntactic encoding as well as certain phonological information. For sparrow, the lexical entry would include information such as “singular noun” and “trochee”. Once the phonological word is prepared, activation spreads to postlexical levels of processing, where phonetic encoding and gestural scores are prepared for output to articulation. Levelt et al.’s (1999) model is unique in that it posits a postlexical locus of syllable encoding. The so-called “syllabary” houses articulatory gestures for highly frequent syllables. Corpus studies (e.g.€Schiller et al. 1996) show that speakers of languages such as English and Dutch do most of their speaking with approximately 500 syllables, despite having over 10,000 possible different syllables in each language. Thus the storage of the most frequent gestural scores eliminates the need to recompute them every time they are required for production. Empirical studies (e.g.€Levelt & Wheeldon 1994) support a separate locus of syllable encoding by revealing that syllable frequency effects may be isolated from lexical frequency effects. In addition to explaining the processes involved in producing the words we intend to speak, models of speech production must also account for mistakes in speech. Speech errors in Levelt et al.’s model are explained by means of a monitor. When we speak, we monitor the phonological word (i.e. segments and
Spanish phonotactics and models of speech production
prosodification) before articulation. This monitor compares the form in preparation with the intended item via a connection to the conceptual level of the system. This mechanism makes intuitive sense, as most people have experienced moments in which they are about to speak the wrong word, catch the error, and prepare the appropriate word before articulation. This theory is supported by research (e.g.€Baars et al. 1975) showing that lexical bias effects disappear when all stimuli are nonwords, because nonwords lack lexical entries. According to this logic, if the monitor checks the output phonology against the conceptual level of processing, differences of a phonotactic nature among production errors in nonwords should not be attributed to the monitor, as these are not represented in the lexicon. However, there is evidence from Spanish phonotactics suggesting that the monitor may track more information than conceptual accuracy. Figure 1 renders a simplified illustration of Level et al.’s (1999) model of speech production. A close examination of language-specific patterns offers a rich source of data to evaluate this model. In particular, the interaction of stress and syllable structure in Spanish provides a testing ground for examining the nature of phonological and phonetic encoding within the model, as well as the nature of the monitoring system. In the next section, we discuss Spanish stress and how it poses questions for the model proposed by Levelt and colleagues.
–Conceptual Level– Semantic Activation
–Monitor–
–Lemma Level– Lexical access with morphosyntactic and prosodic encoding
–Form Level– Phonetic encoding including syllabary and gestural scores
Figure 1.╇ Simplified serial discrete speech production model with monitor
Michael Shelton, Chip Gerfen & Nicolás Gutiérrez Palma
3. A language-specific data source One aspect of Spanish phonotactics that poses important questions for modelers of speech production is the “three-syllable window” on stress placement. With the exception of verbs, all words in Spanish are stressed on one of the last three syllables (Harris 1983, 1992, 1995; Roca 1991; Contreras 1977; Hooper & Terrell 1976). Spanish exhibits proparoxytones such as plátano [plá.ta.no] ‘banana’, paroxytones such as tomate [to.má.te] ‘tomato’, and oxytones such as corazón [ko.na.θón] ‘heart’. Paroxytones comprise the dominant pattern in Spanish, followed by the oxytones. Proparoxytones are much less common, a fact reflected orthographically in that the stress in proparoxytones is always marked diacritically (e.g.€plátano ‘banana’). The lower number of words with antepenultimate stress in Spanish is due in part to syllable-based constraints that shrink the three-syllable window to two syllables. Of interest here is that both rising and falling diphthongs in the penultimate syllable reduce the Spanish stress window. Thus, Spanish has words such as cazuela [ka.θwé.la] ‘casserole’ and arcaico [an.káj.ko] ‘archaic’, with stess on the penult, but the language lacks words of the shape *[ká.θwe.la] or *[án.kaj.ko] in which stress falls to the left of a diphthong in the penultimate syllable. Linguistic treatments of Spanish stress, from numerous perspectives, account for this generalization by arguing that Spanish is sensitive to syllable weight (cf.€Harris 1983; Dunlap 1991; Face 2000, 2004; Bárkányi 2002; Eddington 2000; Roca 1988, 1991). If Spanish is quantity-sensitive, a heavy penult containing a diphthong, as in cazuela ‘casserole’ or arcaico ‘archaic’, attracts stress and thus blocks the possibility of antepenultimate stress.1 Close examination of the behavior of rising and falling diphthongs across the lexicon, however, reveals a significant observation. While both diphthongs block proparoxytone stress, rising diphthongs do not systematically attract stress. The Alameda & Cuetos (1995) database lists 803 paroxytones with rising diphthongs in the final syllable. Words such as familia [fa.mí.lja] ‘family’ or sitio [sí.tjo] ‘place’ illustrate the numerous cases in which rising diphthongs fail to attract stress. One interpretation of this behavior is that prevocalic glides are syllabified with the onset, thus adding no additional weight to the nucleus. This would allow for stress to be assigned to the left of a rising, but not falling, diphthong. However, this view becomes problematic when we consider the permissibility in Spanish of proparoxytones with complex onsets in the penult, such as biógrafo ‘biographer’ and república ‘republic’. If proparoxytones of this type are permitted, we lack an explanation for the absence of 1. While we outline a traditional quantity-sensitivity account here, we are aware that other perspectives do exist. Important to note is that our experimental results do not hinge on any particular analysis of Spanish stress. We rather seek to examine the processing implications of putatively proscribed phonotactic patterns in Spanish and how they can inform a general model of word production.
Spanish phonotactics and models of speech production
words of the shape *cázuela, as discussed previously. In contrast, final syllables containing falling diphthongs in Spanish words are systematically stressed. Consequently, we find words such as caray [ka.náj] ‘darn’ but not words like *[ká.naj]. The interaction of stress placement and syllable structure observed in the differential patterning of rising versus falling diphthong poses an interesting point of inquiry for models of speech production: At what point in the speech production chain are stress and syllable structures encoded? Here, we present an experiment that tests native speakers’ processing of licit and proscribed combinations of stress placement and syllable structure in order to probe for the locus of stress and syllable-level encoding. 4. The experiment To investigate how native Spanish speakers process stress assignment and syllable encoding in preparation for speech, we employed a nonword naming task which presented participants with controlled stimuli designed to test putative phonoÂ� tactic restrictions on stress placement in Spanish. Participants 55 functionally monolingual 18–22 year-old speakers of Spanish participated in this study. All were undergraduate students studying education or psychology at the University of Jaén, Spain. The stimuli Critical stimuli consisted of 56 three-syllable nonwords with a diphthong in the penultimate syllable. Half contained a rising diphthong (e.g.€dóviana [dó.βja.na] or fáriena [fá.nje.na]), and half contained a falling diphthong (e.g.€góvaino [gó.βaj. no] or fáteiga [fá.tej.γa]). For control purposes, all glides were palatal. The diphthong stimuli were presented with 56 monophthong nonwords segmentally controlled to be very similar to their diphthong stimuli counterparts (e.g.€ fárena, comparable to fáriena). All onset segments appearing in stimuli containing a diphthong in the penultimate syllable also appeared in stimuli containing a monophthongal penult, allowing for a controlled comparison across word types in which the only variable is the quality of the nucleus of the penultimate syllable: CV.CGV.CV – CV.CVG.CV – CV.CV.CV. A stress-type condition was designed by presenting all stimuli both with and without a diacritic accent mark over the first syllable. In Spanish orthography, a
Michael Shelton, Chip Gerfen & Nicolás Gutiérrez Palma
Table 1.╇ Summary of critical stimuli Critical Items
Proparoxytone
Paroxytone
Rising Diphthongs Falling Diphthongs Monophthong Controls
fáriena (28) fáteiga (28) fárena (56)
fariena (28) fateiga (28) farena (56)
three-syllable word with an accent mark on the first syllable is always assigned proparoxytone stress (e.g.€ fárena). If a three-syllable word ends in a vowel and carries no accent marks, it will always be interpreted as a paroxytone (e.g.€farena). To avoid any effects arising from the presentation of repeated identical segment sequences, stimuli pairs were counterbalanced so that no participant saw both forms. Thus, if participant 1 saw fárena, then participant 2 saw farena, and so forth. This allows for a controlled comparison between stress types in which the segmental makeup of the stimuli is identical, and the only variable is stress assignment: fárena – farena, fáriena – fariena, fáteiga – fateiga. Critical stimuli were presented with 168 fillers, consisting of 28 additional unrelated nonwords and 140 real words. Table€1 summarizes the critical stimuli for the experiment. Important to note are the boldface items. These are the theoretically proscribed patterns. If diphthongs shrink the three-syllable window to one of the two final syllables, then stressing the first syllable of a three-syllable word containing a penultimate diphthong should be illicit, as the first syllable is outside of the reduced window. Non-boldface print items are theoretically licit forms. Therefore, all stimuli in the paroxytone stress condition are permissible structures. Also, all monophthong controls are possible word shapes in Spanish. As monophthong vowels in open syllables are considered light, any of the three syllables in our control stimuli could carry stress. This design yields a clear prediction. If diphthongs in the penultimate syllable do, in fact, block antepenultimate stress, then participants should commit significantly more errors when naming these items than their phonologically licit counterparts. The task The 140 nonce forms and 140 real words described above were presented in a delayed naming task using E-Prime. Participants were tested in a quiet room, seated in front of a computer screen and microphone. The microphone was connected to a digital recorder (Marantz PMD600 sampling at 48 kHz) that was connected to a PST Serial Response Box. The button box was connected to the computer and served as a voice key to cue the presentation of the next stimulus when the
Spanish phonotactics and models of speech production
participants spoke. The digital recorder was used to record the experimental sessions for subsequent coding of response accuracy. In the delayed naming task, participants see words one at a time on a computer screen and are instructed to produce each word aloud as quickly and accurately as possible when cued by a tone to speak. The tones sounded either immediately as the stimulus appeared on the screen (0 ms delay), 600 ms after stimulus presentation, or 1200 ms after the stimulus. The task is useful in that delaying the presentation of the cue to speak for varying intervals offers controlled amounts of time for phonological processing (Laganaro & Alario 2006). Previous research (e.g.€ Forster & Chambers 1973; Monsell et al. 1989) has argued that delays in the naming task give speakers enough time to prepare speech so that any remaining effects are considered part of the articulation process, i.e. not a higher-level lexical effect. For our stimuli, we predict that a high error rate will arise from the phonological proscription on antepenultimate stress when there is a diphthong in the penultimate syllable. This would be driven by the modulation of stress placement by syllable structure. If the effect surfaces after lengthy delays, we may arguably attribute it to postlexical encoding. Such a finding would support Levelt et al.’s (1999) postlexical locus for syllable-level encoding (i.e. the syllabary, cf.€Figure 1). No effect would be challenging for the model. Summarizing, we make three predictions. First, if Spanish is sensitive to differences in syllable weight, diphthongs in the penult of a three-syllable proparoxytone should provoke higher error rates than controls containing a penultimate monophthong. Second, if Spanish speakers are sensitive to the differential behavior of rising diphthongs, then rising diphthongs in the penultimate syllable of a proparoxytone should provoke fewer errors than comparable words containing falling penultimate diphthongs. Third, if syllables are processed separately after access to the lexicon, the predicted error effect should persist with lengthy delays. In the following section we discuss our results. 5. Results Native or near-native listeners coded naming errors manually using the digital recordings of the experimental sessions. Production errors consisted of instances in which participants misspoke the stimulus, such as stress shifts to another syllable, changes in the segmental makeup of items, or disfluencies in which items were not finished or contained pauses. Errors were submitted to repeated measures ANOVAs by participants (F1) and by items (F2). Subsequent Tukey’s HSD post-hoc tests were run to identify independent effects across the experimental categories and conditions. Below we discuss general results followed by statistical analysis.
Michael Shelton, Chip Gerfen & Nicolás Gutiérrez Palma
Our first comparison examines error rates for the diphthong and monophthong stimuli in all three delay conditions. Errors - 0 Delay
Percentage of Error
100 80 60 40
47.5 *
20 0
14.2
14.6
22.1
Diphthongs Controls
Proparoxytone
Paroxytone Stress Type
Errors - 600 Delay Percentage of Error
100 80 60 40
38.2
20 0
* 10.9
9.7
12.9
Diphthongs Controls
Proparoxytone
Stress Type
Paroxytone
Errors - 1200 Delay Percentage of Error
100 80 60 40 20 0
28.4
* 6.2
Proparoxytone
4.9 Stress Type
11.4
Paroxytone
Figure 2.╇ Diphthongs (pooled) vs. controls by delay
Diphthongs Controls
Spanish phonotactics and models of speech production
Figure 2 reveals a clear pattern across three delay conditions. The grey bars represent the rising and falling diphthongs pooled together. The white bars provide the monophthong control stimuli. In the proparoxytone condition (when the first syllable was marked diacritically for stress, e.g.€fáriena/fáteiga – fárena) diphthong stimuli provoked significantly more production errors than their monophthong counterparts (p < .01). This is not the case in the paroxytone condition, where the difference between word types was not significant in any of the three delays. These results suggest that diphthongs and monophthongs are processed similarly when stress is assigned to the penultimate syllable. However, when stress is assigned to the antepenult, the diphthong stimuli are harder to process, provoking higher error rates. Figure 3 shows that the diphthong stimuli also behaved differently across the delay conditions. In the proparoxytone condition, where diphthong stimuli are theoretically proscribed, we observe a significant improvement over time. The differences among control stimuli are not statistically significant. These results support our first prediction that Spanish speakers should evidence sensitivity to differences in syllable weight. Proparoxytones such as fáteiga and fáriena provoked significantly more errors than items such as fárena. This pattern did not emerge in the paroxytone condition, e.g.€fateiga – fariena – farena. In Table€2 we list all main effects and interactions of ANOVAs comparing diphthong and monophthong stimuli. The most consequential test for our purposes is the interaction of the word, stress, and delay conditions. We are not concerned with differences within these categories alone, but rather we want to assess whether certain types of words are treated differently depending on the stress condition across the delays. Errors - Proparoxytones
Percentage of Error
100 80 60 40
47.5 *
38.2 *
28.4
20 0
14.2 Diphthongs
9.7
6.2
Controls Word Type
Figure 3.╇ Diphthong (pooled) vs. monophthong proparoxytones across delays
0 600 1200
Michael Shelton, Chip Gerfen & Nicolás Gutiérrez Palma
Table 2.╇ ANOVAs of error rates comparing diphthongs to monophthong controls Main effects
Comparison
Results
stress type
proparoxytone/paroxytone
word type
diphthong/monophthong
delay
0ms/600ms/1200ms
F1(1, 54) = 34.64, MSE = 2.08, p < .001 F2(1, 110) = 91.05, MSE = 2.08, p < .001 F1(1, 54) = 67.84, MSE = 2.13, p < .001 F2(1, 110) = 61.96, MSE = 2.00, p < .001 F1(2, 108) = 42.68, MSE = 0.78, p < .001 F2(2, 220) = 32.83, MSE = 0.66, p < .001
Interactions
Results
stress x word
F1(1, 54) = 90.24, MSE = 4.57, p < .001 F2(2, 220) = 187.88, MSE = 4.29, p < .001 F1(2, 108) = 1.20, MSE = 0.02, p > .30
stress x delay
F2(2, 220) = 3.16, MSE = 0.05, p < .05 word x delay F1(2, 108) = 4.08, MSE = 0.05, p < .02 F2(2, 220) = 5.44, MSE = 0.11, p < .01 stress x word x delay F1(2, 108) = 3.98, MSE = 0.05, p = .02 F2(2, 220) = 2.88, MSE = 0.05, p = .058
Results reveal a significant interaction of the word and stress conditions. Participants treated diphthongs and monophthongs differently when presented with proparoxytone and paroxytone stress. This interaction is critical, because it reveals a difference in treatment of the diphthong stimuli between the theoretically licit and proscribed stress conditions. Naming of the two word types also varied by delay. This interaction is important for identifying the locus of syllable-level and stress encoding in the speech production chain. To examine whether our participants evidenced sensitivity to subtler patterns within the lexicon, we compare the error rates of the two diphthong types. Figure 4 shows the patterning of the rising and falling diphthongs in both stress conditions in each delay. As predicted, in the paroxytone condition, where both rising and falling diphthongs are licit in the penult, we see no significant differences in any delay condition. Seemingly contrary to our second prediction, however, participants only appear to have reacted differently to stimuli with rising and falling diphthongs in the 600 ms condition (p < .01). At first glance, it is difficult to explain why an observed difference between rising and falling diphthongs does not appear in the 0 ms delay, when participants had the smallest amount of time for phonological preparation, appears in the 600 ms delay, and then disappears again after 1200 ms. However, if we track the behavior of each diphthong type across the delay conditions, the picture becomes clearer.
Spanish phonotactics and models of speech production Errors - 0 Delay
Percentage of Error
100 80 60 40
43.3
51
20 0
12.8
16.3
Rising Diphthongs Falling Diphthongs
Proparoxytone
Paroxytone Stress Type
Errors - 600 Delay
Percentage of Error
100 80 60 40
* 46.6 29.8
20 0
7.7 Proparoxytone
12.7
Rising Diphthongs Falling Diphthongs
Paroxytone Stress Type
Errors - 1200 Delay
Percentage of Error
100 80 60 40
26.5
32.1
20 0
4.5 Proparoxytone
7
Paroxytone Stress Type
Figure 4.╇ Rising vs. falling diphthongs by delay
Rising Diphthongs Falling Diphthongs
Michael Shelton, Chip Gerfen & Nicolás Gutiérrez Palma Errors - Proparoxytones
Percentage of Error
100 80 60 40 20 0
51
43.3 * 29.8
32.1
26.5
Rising Diphthongs
46.6 * 0 600 1200
Falling Diphthongs
Diphthong Type
Figure 5.╇ Rising vs. falling diphthongs across delays
Figure 5 compares rising and falling diphthongs in the proparoxytone condition across the three delays. Rising diphthongs evidence significant improvement in accuracy with 600 ms of preparation time (0 ms – 600 ms, p < .02). Falling diphthongs, in contrast, require additional preparation time, as evidenced by the later significant difference in accuracy after 1200 ms of delay (600 ms – 1200 ms, p < .01). Tracking the behavior of each diphthong type in this way suggests that rising and falling diphthongs do indeed pattern differently.2 This finding supports our second prediction. While we did not see lower error rates for rising diphthongs in each delay condition, we did find facilitation (i.e. significant improvement in accuracy) of rising diphthongs in earlier delay conditions than for comparable falling diphthong stimuli. Statistical analyses are listed in Table€3.
2. An anonymous reviewer offered two alternative interpretations to these data. One possible interpretation for the facilitation of rising diphthongs is the linear sequencing of the glide within the stimuli. We suspect that the order of the glide/vowel combination is not what drives our results, because differences between diphthong types would then be predicted to surface in the paroxytone condition as well. Additionally, the additional phoneme in our diphthong stimuli does not trigger significantly more errors than the monophthong controls in the paroxytone condition. A second interpretation concerns the possibility of syllabifying the prevocalic glide as part of the onset rather than the rhyme. This potential syllabification fails to provide an explanation for the restriction on antepenultimate stress in our data. This interpretation would also render a light penult, which arguably should pattern with our monophthong controls. In our data this is not the case. Nonetheless, the status of the prevocalic glide is an intriguing issue that we plan to pursue in further work.
Spanish phonotactics and models of speech production
Table 3.╇ ANOVAs of error rates comparing rising and falling diphthongs Main effects
Comparison
Results
stress type
proparoxytone/paroxytone
diphthong type
rising/falling
delay
0ms/600ms/1200ms
F1(1, 54) = 73.57, MSE = 12.99, p < .001 F2(1, 54) = 496.28, MSE = 6.63, p < .001 F1(1, 54) = 15.15, MSE = 0.78, p < .001 F2(1, 54) = 22.72, MSE = 0.54, p < .001 F1(2, 108) = 28.19, MSE = 0.98, p < .001 F2(2, 108) = 34.27, MSE = 0.65, p < .001
Interactions
Results
stress x diphthong
F1(1, 54) = 3.42, MSE = 0.17, p > .06 F2(1, 54) = 16.86, MSE = 0.23, p < .001 F1(2, 108) = 4.23, MSE = 0.11, p < .02 F2(2, 108) = 7.65, MSE = 0.13, p < .001 F1(2, 108) = 2.53, MSE = 0.07, p > .08 F2(2, 208) = 2.67, MSE = 0.05, p > .07 F1(2, 108) = 0.89, MSE = 0.03, p > .41 F2(2, 108) = 2.39, MSE = 0.04, p > .09
stress x delay diphthong x delay stress x diphthong x delay
€
Statistical results reveals a significant main effect of diphthong-type, indicating that Spanish speakers exhibit sensitivity to a difference between rising and falling diphthongs in general. The interaction between stress and diphthong type approaches significance, but the interaction of all three conditions is not significant. Prediction 3 stated that a differential patterning of diphthongs vs. monophthongs across delay conditions would support a postlexical locus of syllable-level encoding. This issue is discussed in more detail below. 6. Implications for psycholinguistic models of speech production Our findings have important implications for psycholinguistic models of speech production. Of particular interest to supporters of the model proposed by Levelt and colleagues is the locus of syllable-level encoding and processing. In their model, syllables are encoded via access to a bank of gestural scores for high-frequency syllables in the language. This level, the syllabary, is located in the form level of processing, after access to the lexicon. As discussed in §1, effects associated with access to the lexicon should disappear after lengthy delays, while postlexical effects remain. Given that our diphthong effect surfaces after 600 and 1200 ms of delay, our data arguably offer empirical support for a postlexical locus of syllabic processing.
Michael Shelton, Chip Gerfen & Nicolás Gutiérrez Palma
Nevertheless, the data also challenge Levelt et al.’s (1999) model. The model posits a lexical locus for the encoding of stress. This is a logical choice, given that many languages contain lexically minimal pairs differentiated only in stress placement, such as sábana ‘sheet’ and sabana ‘grassland’ in Spanish. If the only difference between these items is stress placement, it is reasonable to conclude that metrical information is stored in the lexicon. Experimental evidence also supports an earlier processing of stress assignment. For example, studies employing the implicit priming paradigm have shown that participants cannot prepare for the first syllable of a word without already knowing the word’s metrical frame (Roelofs & Meyer 1998). By contrast, linguists, who generally model the phonological word by building syllables out of segments, and words out of syllables, might find this organization troublesome. Taking our data from Spanish into account, it seems clear that it is syllable structure, i.e. the quantity of the nucleus of the penultimate syllable, that is modulating stress, not the other way around. Within the Levelt et al. (1999) model, a lower level of encoding, such as processing within the form level, cannot influence higher levels of encoding, such as the lemma level, given that activation within the system feeds unidirectionally downward. Once access to the lexicon is made, a word is activated, including information about stress placement. Subsequently, activation is spread down to later levels, such as the syllabary, for syllable-level encoding. It is non-trivial to understand how a later postlexical process, syllable-level encoding, could constrain stress placement, if stress is encoded earlier during lexical processing. A final point of interest that arises from our data concerns the monitor. As seen in Figure 1, Levelt and colleagues have claimed that we control speech errors via a self-monitoring system, noting “the person to whom we listen most is ourself” (Levelt et al. 1999:€6). The monitor functions such that speakers check formlevel processing against the activated conceptual nodes to ensure that the word to be spoken matches the intended idea. Our data suggest that this conception of the monitor may be too simple. As our stimuli consist of nonwords, they cannot be assumed to be associated with any lexical entries. Following this logic, production errors in nonwords should not be attributed to the monitor. The data presented here from Spanish phonotactics suggest that the monitor tracks more finely-grained information than conceptual accuracy. Descriptively, we know that rising diphthongs do not systematically block stress placement on syllables to their left. Falling diphthongs in contrast do. Consequently, the modulation of stress placement by rising diphthongs is arguably weaker than that imposed by falling diphthongs. Given this observation, we suggest that proparoxytones with rising diphthongs in the penultimate syllable register as “more word-like” and that, therefore, they are more
Spanish phonotactics and models of speech production
prone to slip past the monitoring process. In short, our results indicate that the monitor may be sensitive not only to mismatches at the conceptual level but also to gradiently weighted violations of a phonotactic nature. The results of this study indicate that an appropriately designed behavioral task is a useful tool to test the cognitive processing consequences of phonotactic constraints. The delayed naming task allows us to examine the processing of phonotactic knowledge over time. Broadly, our data support the view of a postlexical locus for syllable-level encoding, as proposed by Levelt et al. (1999). However, the results challenge Levelt et al.’s model in two ways: 1) The interaction of stress placement and syllable phonotactics is difficult to interpret within a model that encodes stress and syllables at different levels of processing; and 2) the conception of the monitor arguably needs to be refined to include sensitivity to sublexical phonotactic violations as well as conceptual accuracy. Our future research in this area will explore further the locus of stress-level encoding and the possible range of sensitivity attributed to the monitor. These findings illustrate the need for the finely-grained consideration of language-specific data sources when investigating language processing. The close examination of the phonological system of one language challenges modelers to refine their representation of speech processing as a whole. References Alameda, J. R., & F. Cuetos. 1995. Diccionario de frecuencias de las unidades lingüísticas del castellano. Oviedo: Servicio de Publicaciones de la Universidad de Oviedo. Baars, B. J., J. T. Motley, & D. MacKay. 1975. “Output editing for lexical status from artificially elicited slips of tongue”. Journal of Verbal Learning and Verbal Behavior 14.382–391. Bárkányi, Z. 2002. “A fresh look at quantity sensitivity in Spanish”. Linguistics 40:2.375–394. Contreras, H. 1977. “Spanish epenthesis and stress”. Working Papers in Linguistics 3.9–33. Seattle, Wash.: University of Washington Press. Dell, G. S. 1986. “A spreading activation theory of retrieval in sentence production”. Psychological Review 93.283–321. Dell, G. S. 1988. “The retrieval of phonological forms in production: Tests of predictions from a connectionist model”. Journal of Memory and Language 27.124–142. Dunlap, E. R. 1991. Issues in the Moraic Structure of Spanish. Ph.D. dissertation, MIT. Eddington, D. 2000. “Spanish stress assignment within the analogical modeling of language”. Language 76:1.92–109. Face, T. L. 2000. “The role of syllable weight in the perception of Spanish stress”. Hispanic linguistics at the turn of the millennium ed. by H. Campos, E. Herberger, A. MoralesFront & T. J. Walsh, 1–13. Somerville, Mass.: Cascadilla Press. Face, T. L. 2004. “Perceiving what isn’t there: Non-acoustic cues for perceiving Spanish stress”. Laboratory approaches to Spanish phonology ed. by T. L. Face, 117–141. Berlin: Mouton de Gruyter.
Michael Shelton, Chip Gerfen & Nicolás Gutiérrez Palma Forster, K. I., & S. M. Chambers. 1973. “Lexical access and naming time”. Journal of Verbal Learning and Verbal Behavior 12.627–635. Garrett, M. F. 1980. “Levels of processing in sentence production”. Language production ed. by B. Butterworth, 177–220. New York: Academic Press. Harris, J. W. 1983. Syllable structure and stress in Spanish: A nonlinear analysis. Cambridge, Mass.: MIT Press. Harris, J. W. 1992. Spanish stress: The extrametricality issue. Bloomington, Ind.: Indiana University Linguistics Club. Harris, J. W. 1995. “Projection and edge marking in the computation of stress in Spanish”. The handbook of phonological theory ed. by J. Goldsmith, 867–887. Cambridge, Mass.: Blackwell. Hooper, J., & T. Terrell. 1976. “Stress assignment in Spanish: A natural generative analysis.” Glossa 10.64–110. Laganaro, M. & F. Alario. 2006. “On the locus of the syllable frequency effect in speech production”. Journal of Memory and Language 55.178–196. Levelt, W. J. M. 1989. Speaking: From intention to articulation. Cambridge, MA: MIT Press. Levelt, W. J. M., A. Roelofs, & A. S. Meyer. 1999. “A theory of lexical access in speech production”. Behavioral and Brain Sciences 22.1–38. Levelt, W. J. M., & L. Wheeldon. 1994. “Do speakers have access to a mental syllabary?”. Cognition 50.239–269. Monsell, S., M. C. Doyle. & P. N. Haggard. 1989. “Effects of frequency on visual word recognition tasks: Where are they?”. Journal of Experimental Psychology: General 118.43–71. Roca, I. 1988. “Theoretical implications of Spanish word stress”. Linguistic Inquiry 19.393–424. Roca, I. 1991. “Stress and syllables in Spanish”. Current studies in Spanish linguistics ed. by H. Campos & F. Martínez-Gil, 599–635. Washington, D.C.: Georgetown University Press. Roelofs, A. & A. S. Meyer. 1998. “Metrical structure in planning the production of spoken words”. Journal of Experimental Psychology: Learning, Memory and Cognition 24:4.922–939. Schiller, N. O., A. S. Meyer, R. H. Baayen, & W. J. M. Levelt. 1996. “A comparison of lexeme and speech syllables in Dutch”. Journal of Quantitative Linguistics 3:1.8–28. Shattuck-Hafnagel, S. 1987. “The role of word-onset consonants in speech production planning: New evidence from speech error patterns”. Motor and sensory processes of language ed. by E. Keller & M. Gopnik, 17–51. Hillsdale, NJ: Erlbaum. Shattuck-Hufnagel, S. 1992. “The role of word structure in segmental serial ordering. Cognition 42.213–259. Stemberger, J. P. 1985. “An interactive activation model of language production”. Progress in the psychology of language (Vol. 1) ed. by A. W. Ellis, 143–186. Hillsdale, NJ: Erlbaum.
Index A ACD see Antecedent Contained Deletion Aktionsart╇ 117, 118, 217–223 animacy╇ 186, 193, 196 Antecedent Contained Deletion╇ 157–159, 161, 166–169 applicative╇ 186, 193 argument/adjunct asymmetry╇ 236 argument structure╇ 115, 154 aspect╇ 109, 112, 113, 117–122, 218, 220, 228 aspect shift╇ 112, 217, 218, 220, 224, 225 B bare plurals (BPs)╇ 203–204, 206–214 bilingualism╇ 11, 12, 14, 15, 20 boundedness╇ 108, 117, 118 C case case syncretism╇ 185, 191 inherent Case╇ 146, 177–180, 182 default Case╇ 179 cleft sentence╇ 176, 180 clitic cluster╇ 185–188, 192 closed-syllable laxing╇ 90, 91, 97, 99–102 coercion╇ 193, 217, 218, 220, 221, 225, 228 complementizer╇ 74, 81, 82, 180, 182 contrastive topicalization╇ 182, 183 D dative╇ 60, 65, 111, 136,╇ 139–154, 178, 185–187, 190, 192, 195 delayed naming task╇ 254, 255, 263
deletion╇ 43, 57, 61–63, 65–69, 85, 89, 92, 93, 95, 157–159, 162, 166, 171, 182, 183 deontic modal╇ 162, 169 diglossia╇ 10 diphthong╇ 23–25, 27, 30–36, 252–262 see also vocalic sequences ditransitive╇ 139–142, 144–147, 154, 189, 193, 195, 197 see also double object construction double object construction╇ 192, 193, 196 see also ditransitive down operator╇ 203, 205–207, 212, 214 E elision╇ 57–70, 95, 102 ellipsis╇ 157–160, 162–169, 179 TP-ellipsis╇ 162, 166, 169 VP-ellipsis (VPE)╇ 157–159, 162, 164–166, 168, 169 see also Antecedent Contained Deletion English╇ 11, 15, 25, 40–51, 74, 81, 86, 130, 133, 157–159, 162, 163, 178, 218–222, 226, 228, 237, 245, 250 English bare plurals see bare plurals English gerund see gerund English-origin╇ 42–43, 47, 54 English VP-ellipsis (VPE) see VP-ellipsis (VPE) experimental syntax╇ 233 F focus╇ 11, 13, 141, 164, 166–168, 180, 237 French╇ 24, 36, 47, 93–98, 100–102, 108, 157–166, 168, 171–173, 176, 178, 182, 209, 211, 213
European French╇ 89–91 Montréal French╇ 39, 42–44 Québec French╇ 43, 89–98, 100–102 G gradience╇ 6, 61, 233–234 gerund╇ 107–109, 115–122, 187 H hiatus╇ 23–29, 31–36 see also vocalic sequences I inchoative╇ 218, 220, 223, 226 infinitive╇ 111, 112, 114, 115, 117, 118, 164, 168 Spanish Nominal(ized) Infinitives╇ 107–109, 119, 120, 122 information structure(s)╇ 167, 168, 238 intonational phrase╇ 73, 76–79, 81–84 island╇ 159 Italian╇ 9, 18, 19, 24, 57, 58, 63, 64, 68, 79, 108, 126, 127, 157, 158, 160–165, 168, 186, 211, 212 Italian bilinguals see also Spanish-Italian Italian elision see elision English-Italian╇ 14 Florentine Italian╇ 59 Northern Italian Dialects╇ 11, 12 L language contact see bilingualism language processing see processing left dislocation╇ 175, 178, 179 left periphery╇ 87 light verb╇ 139–141, 144, 151, 153, 154
Romance Linguistics 2008 loanword adaptation╇ 39, 40, 52 M morphology╇ 57, 58, 63, 65, 67, 69, 122, 197, 218, 220, 221, 223–225, 228 movement╇ 77, 129, 147, 157, 160, 162, 180, 245 wh-movement╇ 159, 168, 173, 174, 179, 233, 238, 242, see also wh-question N null complement╇ 179 number neutrality╇ 203, 206, 207 O Optimality Theory╇ 61, 69 overt pronoun╇ 13, 14, 16, 18, 20, 237, 239, 240, 243 Overt Pronoun Rate╇ 239, 240, 243, 244 P parecer╇ 125, 127, 129, 130, 140, 144, 151 see also raising construction Person Case Constraint╇ 188 phonological development╇ 93, 97, 102 phonology╇ 52, 57, 69, 89, 90, 93, 96, 97, 102, 194, 251 pied-piping╇ 174, 179, 181 plural marking╇ 107–109, 120, 122 preposition optional preposition╇ 176 subcategorized preposition╇ 172 see also pied-piping pro╇ 125–137, 172 see also null subject under subject processing╇ 15, 235, 237, 238, 240– 245, 250–253, 255, 261–263 progressive╇ 118, 217, 218, 220, 221, 223–226, 228 prosodic licensing╇ 101, 102
psychological predicate╇ 140, 147, 151, 152, 154 R raising construction╇ 127, 129, 132, 133, 135 see also parecer relative clause chopping relative╇ 181 resumptive relative╇ 181 standard relative╇ 181 relativizer restriction╇ 81, 83, 85, 87 rhotics╇ 39, 52 Romanian╇ 3, 4, 35, 37, 73, 75–79, 85–87, 107–111, 113, 115, 117–124, 185–189, 191–199, 211, 212 S sound change╇ 30, 36 Spanish╇ 95, 125, 127, 129, 131, 133, 141–143, 145, 147, 149, 154, 157, 158, 160–165, 168, 188, 211, 212, 217–224, 226, 228, 236–238, 244, 245, 249, 251–255, 257, 261, 262 Argentine Spanish╇ 23–36 Caribbean Spanish╇ 239–242 Chipilo (contact) Spanish╇ 11, 18–20 Dominican Spanish╇ 241 Guaraní-Spanish╇ 24 Latin-American Spanish╇ 42, 233, 239, 242 Leísta (dialects of) Spanish╇ 185, 186, 189, 190, 192, 195, 196, 198 Mexican Spanish╇ 9, 15, 16, 20, 234 Peninsular Spanish╇ 24, 35, 36 Puerto Rican Spanish╇ 242 Spanish ditransitive (constructions) see ditransitive Spanish-English,╇ 14 Spanish-Italian,╇ 14 Spanish Nominal(ized) Infinitives see infinitive
Spanish-speaking (immigrants)╇ 39–53 Spanish stress see stress Spanish-Veneto/VenetoSpanish╇ 9–16, speech production╇ 249–253, 258, 261 stative verb╇ 217, 221 stress(ed)╇ 24, 27, 28, 33, 58–60, 66, 67, 70, 89–91, 93, 95–99, 101, 102, 141, 142, 150, 209, 242, 249–255, 257, 258, 260–263 Spanish stress╇ 251, 252 subject╇ 9, 11–16, 18, 20, 125–136, 141, 142, 144, 147, 149–153, 161, 162, 164, 166–169, 209, 233–245 null subject╇ 11, 12, 16, 20, 125, 126, 134, 137, 242 see also pro postverbal subject╇ 126, 149 preverbal subject╇ 126, 127, 149, 238–243, 245 subject clitic╇ 12, 19 subject inversion╇ 234 subject pronoun╇ 11–16, 19, 20 syllable encoding╇ 250 T telicity╇ 108, 112, 119 V variation╇ 11, 16, 24, 25, 39, 42–44, 52, 58, 61, 62, 108, 130, 238, 239, 244, 245 allomorphic variation╇ 67, 68 free variation╇ 3, 57, 68, 69 vocalic sequences╇ 23–25, 36 see also diphthongs and hiatus voice mismatch╇ 164–166, 169 VP remnant topicalization╇ 172 VP-topicalization╇ 182, 183 W wh-question╇ 143, 234–236, 238, 241, 244 see also wh-movement (under movement)
CURRENT ISSUES IN LINGUISTIC THEORY
E. F. K. Koerner, Editor
Zentrum für Allgemeine Sprachwissenschaft, Typologie und Universalienforschung, Berlin [email protected] Current Issues in Linguistic Theory (CILT) is a theory-oriented series which welcomes contributions from scholars who have significant proposals to make towards the advancement of our understanding of language, its structure, functioning and development. CILT has been established in order to provide a forum for the presentation and discussion of linguistic opinions of scholars who do not necessarily accept the prevailing mode of thought in linguistic science. It offers an outlet for meaningful contributions to the current linguistic debate, and furnishes the diversity of opinion which a healthy discipline must have. A complete list of titles in this series can be found on the publishers’ website, www.benjamins.com 315 Colina, Sonia, Antxon Olarrea and Ana Maria Carvalho (eds.): Romance Linguistics 2009. Selected papers from the 39th Linguistic Symposium on Romance Languages (LSRL), Tucson, Arizona, March 2009. xiv, 421 pp. + index. Expected November 2010 314 Lenker, Ursula, Judith Huber and Robert Mailhammer (eds.): English Historical Linguistics 2008. Selected papers from the fifteenth International Conference on English Historical Linguistics (ICEHL 15), Munich, 24-30 August 2008.. Volume I: The history of English verbal and nominal constructions. vii, 278 pp. + index. Expected October 2010 313 Arregi, Karlos, Zsuzsanna Fagyal, Silvina A. Montrul and Annie Tremblay (eds.): Romance Linguistics 2008. Interactions in Romance. Selected papers from the 38th Linguistic Symposium on Romance Languages (LSRL), Urbana-Champaign, April 2008. 2010. vii, 262 pp. 312 Operstein, Natalie: Consonant Structure and Prevocalization. 2010. x, 234 pp. 311 Scalise, Sergio and Irene Vogel (eds.): Cross-Disciplinary Issues in Compounding. 2010. viii, 382 pp. 310 Rainer, Franz, Wolfgang U. Dressler, Dieter Kastovsky and Hans Christian Luschützky (eds.): Variation and Change in Morphology. Selected papers from the 13th International Morphology Meeting, Vienna, February 2008. With the assistance of Elisabeth Peters. 2010. vii, 249 pp. 309 Nicolov, Nicolas, Galia Angelova and Ruslan Mitkov (eds.): Recent Advances in Natural Language Processing V. Selected papers from RANLP 2007. 2009. x, 338 pp. 308 Dufresne, Monique, Fernande Dupuis and Etleva Vocaj (eds.): Historical Linguistics 2007. Selected papers from the 18th International Conference on Historical Linguistics, Montreal, 6–11 August 2007. 2009. x, 311 pp. 307 Calabrese, Andrea and W. Leo Wetzels (eds.): Loan Phonology. 2009. vii, 273 pp. 306 Vigário, Marina, Sónia Frota and M. João Freitas (eds.): Phonetics and Phonology. Interactions and interrelations. 2009. vi, 290 pp. 305 Bubenik, Vit, John Hewson and Sarah Rose (eds.): Grammatical Change in Indo-European Languages. Papers presented at the workshop on Indo-European Linguistics at the XVIIIth International Conference on Historical Linguistics, Montreal, 2007. 2009. xx, 262 pp. 304 Masullo, Pascual José, Erin O'Rourke and Chia-Hui Huang (eds.): Romance Linguistics 2007. Selected papers from the 37th Linguistic Symposium on Romance Languages (LSRL), Pittsburgh, 15–18 March 2007. 2009. vii, 361 pp. 303 Torck, Danièle and W. Leo Wetzels (eds.): Romance Languages and Linguistic Theory 2006. Selected papers from ‘Going Romance’, Amsterdam, 7–9 December 2006. 2009. viii, 262 pp. 302 Ferraresi, Gisella and Maria Goldbach (eds.): Principles of Syntactic Reconstruction. 2008. xvii, 219 pp. 301 Parkinson, Dilworth B. (ed.): Perspectives on Arabic Linguistics. Papers from the annual symposium on Arabic linguistics. Volume XXI: Provo, Utah, March 2007. 2008. x, 206 pp. 300 Vajda, Edward J. (ed.): Subordination and Coordination Strategies in North Asian Languages. 2008. xii, 218 pp. 299 González-Díaz, Victorina: English Adjective Comparison. A historical perspective. 2008. xix, 252 pp. 298 Bowern, Claire, Bethwyn Evans and Luisa Miceli (eds.): Morphology and Language History. In honour of Harold Koch. 2008. x, 364 pp. 297 Dossena, Marina, Richard Dury and Maurizio Gotti (eds.): English Historical Linguistics 2006. Selected papers from the fourteenth International Conference on English Historical Linguistics (ICEHL 14), Bergamo, 21–25 August 2006. Volume III: Geo-Historical Variation in English. 2008. xiii, 197 pp.
296 Dury, Richard, Maurizio Gotti and Marina Dossena (eds.): English Historical Linguistics 2006. Selected papers from the fourteenth International Conference on English Historical Linguistics (ICEHL 14), Bergamo, 21–25 August 2006. Volume II: Lexical and Semantic Change. 2008. xiii, 264 pp. 295 Gotti, Maurizio, Marina Dossena and Richard Dury (eds.): English Historical Linguistics 2006. Selected papers from the fourteenth International Conference on English Historical Linguistics (ICEHL 14), Bergamo, 21–25 August 2006. Volume I: Syntax and Morphology. 2008. xiv, 259 pp. 294 Frellesvig, Bjarke and John Whitman (eds.): Proto-Japanese. Issues and Prospects. 2008. vii, 229 pp. 293 Detges, Ulrich and Richard Waltereit (eds.): The Paradox of Grammatical Change. Perspectives from Romance. 2008. vi, 252 pp. 292 Nicolov, Nicolas, Kalina Bontcheva, Galia Angelova and Ruslan Mitkov (eds.): Recent Advances in Natural Language Processing IV. Selected papers from RANLP 2005. 2007. xii, 307 pp. 291 Baauw, Sergio, Frank Drijkoningen and Manuela Pinto (eds.): Romance Languages and Linguistic Theory 2005. Selected papers from ‘Going Romance’, Utrecht, 8–10 December 2005. 2007. viii, 338 pp. 290 Mughazy, Mustafa A. (ed.): Perspectives on Arabic Linguistics. Papers from the annual symposium on Arabic linguistics. Volume XX: Kalamazoo, Michigan, March 2006. 2007. xii, 247 pp. 289 Benmamoun, Elabbas (ed.): Perspectives on Arabic Linguistics. Papers from the annual symposium on Arabic Linguistics. Volume XIX: Urbana, Illinois, April 2005. 2007. xiv, 304 pp. 288 Toivonen, Ida and Diane Nelson (eds.): Saami Linguistics. 2007. viii, 321 pp. 287 Camacho, José, Nydia Flores-Ferrán, Liliana Sánchez, Viviane Déprez and María José Cabrera (eds.): Romance Linguistics 2006. Selected papers from the 36th Linguistic Symposium on Romance Languages (LSRL), New Brunswick, March-April 2006. 2007. viii, 340 pp. 286 Weijer, Jeroen van de and Erik Jan van der Torre (eds.): Voicing in Dutch. (De)voicing – phonology, phonetics, and psycholinguistics. 2007. x, 186 pp. 285 Sackmann, Robin (ed.): Explorations in Integrational Linguistics. Four essays on German, French, and Guaraní. 2008. ix, 239 pp. 284 Salmons, Joseph C. and Shannon Dubenion-Smith (eds.): Historical Linguistics 2005. Selected papers from the 17th International Conference on Historical Linguistics, Madison, Wisconsin, 31 July - 5 August 2005. 2007. viii, 413 pp. 283 Lenker, Ursula and Anneli Meurman-Solin (eds.): Connectives in the History of English. 2007. viii, 318 pp. 282 Prieto, Pilar, Joan Mascaró and Maria-Josep Solé (eds.): Segmental and prosodic issues in Romance phonology. 2007. xvi, 262 pp. 281 Vermeerbergen, Myriam, Lorraine Leeson and Onno Crasborn (eds.): Simultaneity in Signed Languages. Form and function. 2007. viii, 360 pp. (incl. CD-Rom). 280 Hewson, John and Vit Bubenik: From Case to Adposition. The development of configurational syntax in Indo-European languages. 2006. xxx, 420 pp. 279 Nedergaard Thomsen, Ole (ed.): Competing Models of Linguistic Change. Evolution and beyond. 2006. vi, 344 pp. 278 Doetjes, Jenny and Paz González (eds.): Romance Languages and Linguistic Theory 2004. Selected papers from ‘Going Romance’, Leiden, 9–11 December 2004. 2006. viii, 320 pp. 277 Helasvuo, Marja-Liisa and Lyle Campbell (eds.): Grammar from the Human Perspective. Case, space and person in Finnish. 2006. x, 280 pp. 276 Montreuil, Jean-Pierre Y. (ed.): New Perspectives on Romance Linguistics. Vol. II: Phonetics, Phonology and Dialectology. Selected papers from the 35th Linguistic Symposium on Romance Languages (LSRL), Austin, Texas, February 2005. 2006. x, 213 pp. 275 Nishida, Chiyo and Jean-Pierre Y. Montreuil (eds.): New Perspectives on Romance Linguistics. Vol. I: Morphology, Syntax, Semantics, and Pragmatics. Selected papers from the 35th Linguistic Symposium on Romance Languages (LSRL), Austin, Texas, February 2005. 2006. xiv, 288 pp. 274 Gess, Randall S. and Deborah Arteaga (eds.): Historical Romance Linguistics. Retrospective and perspectives. 2006. viii, 393 pp. 273 Filppula, Markku, Juhani Klemola, Marjatta Palander and Esa Penttilä (eds.): Dialects Across Borders. Selected papers from the 11th International Conference on Methods in Dialectology (Methods XI), Joensuu, August 2002. 2005. xii, 291 pp. 272 Gess, Randall S. and Edward J. Rubin (eds.): Theoretical and Experimental Approaches to Romance Linguistics. Selected papers from the 34th Linguistic Symposium on Romance Languages (LSRL), Salt Lake City, March 2004. 2005. viii, 367 pp.
271 Branner, David Prager (ed.): The Chinese Rime Tables. Linguistic philosophy and historicalcomparative phonology. 2006. viii, 358 pp. 270 Geerts, Twan, Ivo van Ginneken and Haike Jacobs (eds.): Romance Languages and Linguistic Theory 2003. Selected papers from ‘Going Romance’ 2003, Nijmegen, 20–22 November. 2005. viii, 369 pp. 269 Hargus, Sharon and Keren Rice (eds.): Athabaskan Prosody. 2005. xii, 432 pp. 268 Cravens, Thomas D. (ed.): Variation and Reconstruction. 2006. viii, 223 pp. 267 Alhawary, Mohammad T. and Elabbas Benmamoun (eds.): Perspectives on Arabic Linguistics. Papers from the annual symposium on Arabic linguistics. Volume XVII–XVIII: Alexandria, 2003 and Norman, Oklahoma 2004. 2005. xvi, 315 pp. 266 Boudelaa, Sami (ed.): Perspectives on Arabic Linguistics. Papers from the annual symposium on Arabic linguistics. Volume XVI: , Cambridge, March 2002. 2006. xii, 181 pp. 265 Cornips, Leonie and Karen P. Corrigan (eds.): Syntax and Variation. Reconciling the Biological and the Social. 2005. vi, 312 pp. 264 Dressler, Wolfgang U., Dieter Kastovsky, Oskar E. Pfeiffer and Franz Rainer (eds.): Morphology and its demarcations. Selected papers from the 11th Morphology meeting, Vienna, February 2004. With the assistance of Francesco Gardani and Markus A. Pöchtrager. 2005. xiv, 320 pp. 263 Branco, António, Tony McEnery and Ruslan Mitkov (eds.): Anaphora Processing. Linguistic, cognitive and computational modelling. 2005. x, 449 pp. 262 Vajda, Edward J. (ed.): Languages and Prehistory of Central Siberia. 2004. x, 275 pp. 261 Kay, Christian J. and Jeremy J. Smith (eds.): Categorization in the History of English. 2004. viii, 268 pp. 260 Nicolov, Nicolas, Kalina Bontcheva, Galia Angelova and Ruslan Mitkov (eds.): Recent Advances in Natural Language Processing III. Selected papers from RANLP 2003. 2004. xii, 402 pp. 259 Carr, Philip, Jacques Durand and Colin J. Ewen (eds.): Headhood, Elements, Specification and Contrastivity. Phonological papers in honour of John Anderson. 2005. xxviii, 405 pp. 258 Auger, Julie, J. Clancy Clements and Barbara Vance (eds.): Contemporary Approaches to Romance Linguistics. Selected Papers from the 33rd Linguistic Symposium on Romance Languages (LSRL), Bloomington, Indiana, April 2003. With the assistance of Rachel T. Anderson. 2004. viii, 404 pp. 257 Fortescue, Michael, Eva Skafte Jensen, Jens Erik Mogensen and Lene Schøsler (eds.): Historical Linguistics 2003. Selected papers from the 16th International Conference on Historical Linguistics, Copenhagen, 11–15 August 2003. 2005. x, 312 pp. 256 Bok-Bennema, Reineke, Bart Hollebrandse, Brigitte Kampers-Manhe and Petra Sleeman (eds.): Romance Languages and Linguistic Theory 2002. Selected papers from ‘Going Romance’, Groningen, 28–30 November 2002. 2004. viii, 273 pp. 255 Meulen, Alice ter and Werner Abraham (eds.): The Composition of Meaning. From lexeme to discourse. 2004. vi, 232 pp. 254 Baldi, Philip and Pietro U. Dini (eds.): Studies in Baltic and Indo-European Linguistics. In honor of William R. Schmalstieg. 2004. xlvi, 302 pp. 253 Caffarel, Alice, J.R. Martin and Christian M.I.M. Matthiessen (eds.): Language Typology. A functional perspective. 2004. xiv, 702 pp. 252 Kay, Christian J., Carole Hough and Irené Wotherspoon (eds.): New Perspectives on English Historical Linguistics. Selected papers from 12 ICEHL, Glasgow, 21–26 August 2002. Volume II: Lexis and Transmission. 2004. xii, 273 pp. 251 Kay, Christian J., Simon Horobin and Jeremy J. Smith (eds.): New Perspectives on English Historical Linguistics. Selected papers from 12 ICEHL, Glasgow, 21–26 August 2002. Volume I: Syntax and Morphology. 2004. x, 264 pp. 250 Jensen, John T.: Principles of Generative Phonology. An introduction. 2004. xii, 324 pp. 249 Bowern, Claire and Harold Koch (eds.): Australian Languages. Classification and the comparative method. 2004. xii, 377 pp. (incl. CD-Rom). 248 Weigand, Edda (ed.): Emotion in Dialogic Interaction. Advances in the complex. 2004. xii, 284 pp. 247 Parkinson, Dilworth B. and Samira Farwaneh (eds.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume XV: Salt Lake City 2001. 2003. x, 214 pp. 246 Holisky, Dee Ann and Kevin Tuite (eds.): Current Trends in Caucasian, East European and Inner Asian Linguistics. Papers in honor of Howard I. Aronson. 2003. xxviii, 426 pp. 245 Quer, Josep, Jan Schroten, Mauro Scorretti, Petra Sleeman and Els Verheugd (eds.): Romance Languages and Linguistic Theory 2001. Selected papers from 'Going Romance', Amsterdam, 6–8 December 2001. 2003. viii, 355 pp. 244 Pérez-Leroux, Ana Teresa and Yves Roberge (eds.): Romance Linguistics. Theory and Acquisition. Selected papers from the 32nd Linguistic Symposium on Romance Languages (LSRL), Toronto, April 2002. 2003. viii, 388 pp.
243 Cuyckens, Hubert, Thomas Berg, René Dirven and Klaus-Uwe Panther (eds.): Motivation in Language. Studies in honor of Günter Radden. 2003. xxvi, 403 pp. 242 Seuren, Pieter A.M. and Gerard Kempen (eds.): Verb Constructions in German and Dutch. 2003. vi, 316 pp. 241 Lecarme, Jacqueline (ed.): Research in Afroasiatic Grammar II. Selected papers from the Fifth Conference on Afroasiatic Languages, Paris, 2000. 2003. viii, 550 pp. 240 Janse, Mark and Sijmen Tol (eds.): Language Death and Language Maintenance. Theoretical, practical and descriptive approaches. With the assistance of Vincent Hendriks. 2003. xviii, 244 pp. 239 Andersen, Henning (ed.): Language Contacts in Prehistory. Studies in Stratigraphy. Papers from the Workshop on Linguistic Stratigraphy and Prehistory at the Fifteenth International Conference on Historical Linguistics, Melbourne, 17 August 2001. 2003. viii, 292 pp. 238 Núñez-Cedeño, Rafael, Luis López and Richard Cameron (eds.): A Romance Perspective on Language Knowledge and Use. Selected papers from the 31st Linguistic Symposium on Romance Languages (LSRL), Chicago, 19–22 April 2001. 2003. xvi, 386 pp. 237 Blake, Barry J. and Kate Burridge (eds.): Historical Linguistics 2001. Selected papers from the 15th International Conference on Historical Linguistics, Melbourne, 13–17 August 2001. Editorial assistance Jo Taylor. 2003. x, 444 pp. 236 Simon-Vandenbergen, Anne-Marie, Miriam Taverniers and Louise J. Ravelli (eds.): Grammatical Metaphor. Views from systemic functional linguistics. 2003. vi, 453 pp. 235 Linn, Andrew R. and Nicola McLelland (eds.): Standardization. Studies from the Germanic languages. 2002. xii, 258 pp. 234 Weijer, Jeroen van de, Vincent J. van Heuven and Harry van der Hulst (eds.): The Phonological Spectrum. Volume II: Suprasegmental structure. 2003. x, 264 pp. 233 Weijer, Jeroen van de, Vincent J. van Heuven and Harry van der Hulst (eds.): The Phonological Spectrum. Volume I: Segmental structure. 2003. x, 308 pp. 232 Beyssade, Claire, Reineke Bok-Bennema, Frank Drijkoningen and Paola Monachesi (eds.): Romance Languages and Linguistic Theory 2000. Selected papers from ‘Going Romance’ 2000, Utrecht, 30 November–2 December. 2002. viii, 354 pp. 231 Cravens, Thomas D.: Comparative Historical Dialectology. Italo-Romance clues to Ibero-Romance sound change. 2002. xii, 163 pp. 230 Parkinson, Dilworth B. and Elabbas Benmamoun (eds.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume XIII-XIV: Stanford, 1999 and Berkeley, California 2000. 2002. xiv, 250 pp. 229 Nevin, Bruce E. and Stephen B. Johnson (eds.): The Legacy of Zellig Harris. Language and information into the 21st century. Volume 2: Mathematics and computability of language. 2002. xx, 312 pp. 228 Nevin, Bruce E. (ed.): The Legacy of Zellig Harris. Language and information into the 21st century. Volume 1: Philosophy of science, syntax and semantics. 2002. xxxvi, 323 pp. 227 Fava, Elisabetta (ed.): Clinical Linguistics. Theory and applications in speech pathology and therapy. 2002. xxiv, 353 pp. 226 Levin, Saul: Semitic and Indo-European. Volume II: Comparative morphology, syntax and phonetics. 2002. xviii, 592 pp. 225 Shahin, Kimary N.: Postvelar Harmony. 2003. viii, 344 pp. 224 Fanego, Teresa, Belén Méndez-Naya and Elena Seoane (eds.): Sounds, Words, Texts and Change. Selected papers from 11 ICEHL, Santiago de Compostela, 7–11 September 2000. Volume 2. 2002. x, 310 pp. 223 Fanego, Teresa, Javier Pérez-Guerra and María José López-Couso (eds.): English Historical Syntax and Morphology. Selected papers from 11 ICEHL, Santiago de Compostela, 7–11 September 2000. Volume 1. 2002. x, 306 pp. 222 Herschensohn, Julia, Enrique Mallén and Karen Zagona (eds.): Features and Interfaces in Romance. Essays in honor of Heles Contreras. 2001. xiv, 302 pp. 221 D’hulst, Yves, Johan Rooryck and Jan Schroten (eds.): Romance Languages and Linguistic Theory 1999. Selected papers from ‘Going Romance’ 1999, Leiden, 9–11 December 1999. 2001. viii, 406 pp. 220 Satterfield, Teresa, Christina Tortora and Diana Cresti (eds.): Current Issues in Romance Languages. Selected papers from the 29th Linguistic Symposium on Romance Languages (LSRL), Ann Arbor, 8–11 April 1999. 2002. viii, 412 pp. 219 Andersen, Henning (ed.): Actualization. Linguistic Change in Progress. Papers from a workshop held at the 14th International Conference on Historical Linguistics, Vancouver, B.C., 14 August 1999. 2001. vii, 250 pp.
218 Bendjaballah, Sabrina, Wolfgang U. Dressler, Oskar E. Pfeiffer and Maria D. Voeikova (eds.): Morphology 2000. Selected papers from the 9th Morphology Meeting, Vienna, 24–28 February 2000. 2002. viii, 317 pp. 217 Wiltshire, Caroline R. and Joaquim Camps (eds.): Romance Phonology and Variation. Selected papers from the 30th Linguistic Symposium on Romance Languages, Gainesville, Florida, February 2000. 2002. xii, 238 pp. 216 Camps, Joaquim and Caroline R. Wiltshire (eds.): Romance Syntax, Semantics and L2 Acquisition. Selected papers from the 30th Linguistic Symposium on Romance Languages, Gainesville, Florida, February 2000. 2001. xii, 246 pp. 215 Brinton, Laurel J. (ed.): Historical Linguistics 1999. Selected papers from the 14th International Conference on Historical Linguistics, Vancouver, 9–13 August 1999. 2001. xii, 398 pp. 214 Weigand, Edda and Marcelo Dascal (eds.): Negotiation and Power in Dialogic Interaction. 2001. viii, 303 pp. 213 Sornicola, Rosanna, Erich Poppe and Ariel Shisha-Halevy (eds.): Stability, Variation and Change of Word-Order Patterns over Time. With the assistance of Paola Como. 2000. xxxii, 323 pp. 212 Repetti, Lori (ed.): Phonological Theory and the Dialects of Italy. 2000. x, 301 pp. 211 Elšík, Viktor and Yaron Matras (eds.): Grammatical Relations in Romani. The Noun Phrase. with a Foreword by Frans Plank (Universität Konstanz). 2000. x, 244 pp. 210 Dworkin, Steven N. and Dieter Wanner (eds.): New Approaches to Old Problems. Issues in Romance historical linguistics. 2000. xiv, 235 pp. 209 King, Ruth: The Lexical Basis of Grammatical Borrowing. A Prince Edward Island French case study. 2000. xvi, 241 pp. 208 Robinson, Orrin W.: Whose German? The ach/ich alternation and related phenomena in ‘standard’ and ‘colloquial’. 2001. xii, 178 pp. 207 Sanz, Montserrat: Events and Predication. A new approach to syntactic processing in English and Spanish. 2000. xiv, 219 pp. 206 Fawcett, Robin P.: A Theory of Syntax for Systemic Functional Linguistics. 2000. xxviii, 360 pp. 205 Dirven, René, Roslyn M. Frank and Cornelia Ilie (eds.): Language and Ideology. Volume 2: descriptive cognitive approaches. 2001. vi, 264 pp. 204 Dirven, René, Bruce Hawkins and Esra Sandikcioglu (eds.): Language and Ideology. Volume 1: theoretical cognitive approaches. 2001. vi, 301 pp. 203 Norrick, Neal R.: Conversational Narrative. Storytelling in everyday talk. 2000. xiv, 233 pp. 202 Lecarme, Jacqueline, Jean Lowenstamm and Ur Shlonsky (eds.): Research in Afroasiatic Grammar. Papers from the Third conference on Afroasiatic Languages, Sophia Antipolis, 1996. 2000. vi, 386 pp. 201 Dressler, Wolfgang U., Oskar E. Pfeiffer, Markus A. Pöchtrager and John R. Rennison (eds.): Morphological Analysis in Comparison. 2000. x, 261 pp. 200 Anttila, Raimo: Greek and Indo-European Etymology in Action. Proto-Indo-European *aǵ-. 2000. xii, 314 pp. 199 Pütz, Martin and Marjolijn H. Verspoor (eds.): Explorations in Linguistic Relativity. 2000. xvi, 369 pp. 198 Niemeier, Susanne and René Dirven (eds.): Evidence for Linguistic Relativity. 2000. xxii, 240 pp. 197 Coopmans, Peter, Martin Everaert and Jane Grimshaw (eds.): Lexical Specification and Insertion. 2000. xviii, 476 pp. 196 Hannahs, S.J. and Mike Davenport (eds.): Issues in Phonological Structure. Papers from an International Workshop. 1999. xii, 268 pp. 195 Herring, Susan C., Pieter van Reenen and Lene Schøsler (eds.): Textual Parameters in Older Languages. 2001. x, 448 pp. 194 Coleman, Julie and Christian J. Kay (eds.): Lexicology, Semantics and Lexicography. Selected papers from the Fourth G. L. Brook Symposium, Manchester, August 1998. 2000. xiv, 257 pp. 193 Klausenburger, Jurgen: Grammaticalization. Studies in Latin and Romance morphosyntax. 2000. xiv, 184 pp. 192 Alexandrova, Galina M. and Olga Arnaudova (eds.): The Minimalist Parameter. Selected papers from the Open Linguistics Forum, Ottawa, 21–23 March 1997. 2001. x, 360 pp. 191 Sihler, Andrew L.: Language History. An introduction. 2000. xvi, 298 pp. 190 Benmamoun, Elabbas (ed.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume XII: Urbana-Champaign, Illinois, 1998. 1999. viii, 204 pp. 189 Nicolov, Nicolas and Ruslan Mitkov (eds.): Recent Advances in Natural Language Processing II. Selected papers from RANLP ’97. 2000. xi, 422 pp.
188 Simmons, Richard VanNess: Chinese Dialect Classification. A comparative approach to Harngjou, Old Jintarn, and Common Northern Wu. 1999. xviii, 317 pp. 187 Franco, Jon A., Alazne Landa and Juan Martín (eds.): Grammatical Analyses in Basque and Romance Linguistics. Papers in honor of Mario Saltarelli. 1999. viii, 306 pp. 186 Mišeska Tomić, Olga and Milorad Radovanović (eds.): History and Perspectives of Language Study. Papers in honor of Ranko Bugarski. 2000. xxii, 314 pp. 185 Authier, Jean-Marc, Barbara E. Bullock and Lisa A. Reed (eds.): Formal Perspectives on Romance Linguistics. Selected papers from the 28th Linguistic Symposium on Romance Languages (LSRL XXVIII), University Park, 16–19 April 1998. 1999. xii, 334 pp. 184 Sagart, Laurent: The Roots of Old Chinese. 1999. xii, 272 pp. 183 Contini-Morava, Ellen and Yishai Tobin (eds.): Between Grammar and Lexicon. 2000. xxxii, 365 pp. 182 Kenesei, István (ed.): Crossing Boundaries. Advances in the theory of Central and Eastern European languages. 1999. viii, 302 pp. 181 Mohammad, Mohammad A.: Word Order, Agreement and Pronominalization in Standard and Palestinian Arabic. 2000. xvi, 197 pp. 180 Mereu, Lunella (ed.): Boundaries of Morphology and Syntax. 1999. viii, 314 pp. 179 Rini, Joel: Exploring the Role of Morphology in the Evolution of Spanish. 1999. xvi, 187 pp. 178 Foolen, Ad and Frederike van der Leek (eds.): Constructions in Cognitive Linguistics. Selected papers from the Fifth International Cognitive Linguistics Conference, Amsterdam, 1997. 2000. xvi, 338 pp. 177 Cuyckens, Hubert and Britta E. Zawada (eds.): Polysemy in Cognitive Linguistics. Selected papers from the International Cognitive Linguistics Conference, Amsterdam, 1997. 2001. xxviii, 296 pp. 176 Van Hoek, Karen, Andrej A. Kibrik and Leo Noordman (eds.): Discourse Studies in Cognitive Linguistics. Selected papers from the 5th International Cognitive Linguistics Conference, Amsterdam, July 1997. 1999. vi, 187 pp. 175 Gibbs, Jr., Raymond W. and Gerard J. Steen (eds.): Metaphor in Cognitive Linguistics. Selected papers from the 5th International Cognitive Linguistics Conference, Amsterdam, 1997. 1999. viii, 226 pp. 174 Hall, T. Alan and Ursula Kleinhenz (eds.): Studies on the Phonological Word. 1999. viii, 298 pp. 173 Treviño, Esthela and José Lema (eds.): Semantic Issues in Romance Syntax. 1999. viii, 309 pp. 172 Dimitrova-Vulchanova, Mila and Lars Hellan (eds.): Topics in South Slavic Syntax and Semantics. 1999. xxviii, 263 pp. 171 Weigand, Edda (ed.): Contrastive Lexical Semantics. 1998. x, 270 pp. 170 Lamb, Sydney M.: Pathways of the Brain. The neurocognitive basis of language. 1999. xii, 418 pp. 169 Ghadessy, Mohsen (ed.): Text and Context in Functional Linguistics. 1999. xviii, 340 pp. 168 Ratcliffe, Robert R.: The “Broken” Plural Problem in Arabic and Comparative Semitic. Allomorphy and analogy in non-concatenative morphology. 1998. xii, 261 pp. 167 Benmamoun, Elabbas, Mushira Eid and Niloofar Haeri (eds.): Perspectives on Arabic Linguistics. Papers from the Annual Symposium on Arabic Linguistics. Volume XI: Atlanta, Georgia, 1997. 1998. viii, 231 pp. 166 Lemmens, Maarten: Lexical Perspectives on Transitivity and Ergativity. Causative constructions in English. 1998. xii, 268 pp. 165 Bubenik, Vit: A Historical Syntax of Late Middle Indo-Aryan (Apabhraṃśa). 1998. xxiv, 265 pp. 164 Schmid, Monika S., Jennifer R. Austin and Dieter Stein (eds.): Historical Linguistics 1997. Selected papers from the 13th International Conference on Historical Linguistics, Düsseldorf, 10–17 August 1997. 1998. x, 409 pp. 163 Lockwood, David G., Peter H. Fries and James E. Copeland (eds.): Functional Approaches to Language, Culture and Cognition. Papers in honor of Sydney M. Lamb. 2000. xxxiv, 656 pp. 162 Hogg, Richard M. and Linda van Bergen (eds.): Historical Linguistics 1995. Volume 2: Germanic linguistics.. Selected papers from the 12th International Conference on Historical Linguistics, Manchester, August 1995. 1998. x, 365 pp. 161 Smith, John Charles and Delia Bentley (eds.): Historical Linguistics 1995. Volume 1: General issues and non-Germanic Languages.. Selected papers from the 12th International Conference on Historical Linguistics, Manchester, August 1995. 2000. xii, 438 pp. 160 Schwegler, Armin, Bernard Tranel and Myriam Uribe-Etxebarria (eds.): Romance Linguistics: Theoretical Perspectives. Selected papers from the 27th Linguistic Symposium on Romance Languages (LSRL XXVII), Irvine, 20–22 February, 1997. 1998. vi, 349 pp. + index. 159 Joseph, Brian D., Geoffrey C. Horrocks and Irene Philippaki-Warburton (eds.): Themes in Greek Linguistics II. 1998. x, 335 pp.
