Basque and Romance: Aligning Grammars 9004395385, 9789004395381

Citation preview

Basque and Romance

Grammars and Language Sketches of the World’s Languages

The titles published in this series are listed at brill.com/gswl

Basque and Romance Aligning Grammars

Edited by

Ane Berro Urrizelki Beatriz Fernández Jon Ortiz de Urbina

LEIDEN | BOSTON

Library of Congress Cataloging-in-Publication Data Names: Berro Urrizelki, Ane, editor. | Fernāandez, Beatriz, editor. | Ortiz de Urbina, Jon, editor. Title: Basque and Romance : aligning grammars / edited by Ane Berro Urrizelki, Beatriz Fernandez, Jon Ortiz de Urbina. Description: Leiden ; Boston : Brill, [2019] | Series: Grammars and language sketches of the world's languages | Includes bibliographical references and index. Identifiers: LCCN 2019008329 (print) | LCCN 2019009310 (ebook) | ISBN 9789004395398 (ebook) | ISBN 9789004395381 (hardback : alk. paper) Subjects: LCSH: Basque language–Grammar, Comparative–Spanish. | Spanish language–Grammar, Comparative–Basque. | Basque language–Grammar, Comparative–Romance. | Romance languages–Grammar, Comparative–Basque. Classification: LCC PH5023 (ebook) | LCC PH5023 .B298 2019 (print) | DDC 499/.925–dc23 LC record available at https://lccn.loc.gov/2019008329

Typeface for the Latin, Greek, and Cyrillic scripts: “Brill”. See and download: brill.com/brill‑typeface. ISSN 2352-9342 ISBN 978-90-04-39538-1 (hardback) ISBN 978-90-04-39539-8 (e-book) Copyright 2019 by Koninklijke Brill NV, Leiden, The Netherlands. Koninklijke Brill NV incorporates the imprints Brill, Brill Hes & De Graaf, Brill Nijhoff, Brill Rodopi, Brill Sense, Hotei Publishing, mentis Verlag, Verlag Ferdinand Schöningh and Wilhelm Fink Verlag. All rights reserved. No part of this publication may be reproduced, translated, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission from the publisher. Authorization to photocopy items for internal or personal use is granted by Koninklijke Brill NV provided that the appropriate fees are paid directly to The Copyright Clearance Center, 222 Rosewood Drive, Suite 910, Danvers, MA 01923, USA. Fees are subject to change. This book is printed on acid-free paper and produced in a sustainable manner.

Contents 1

Introduction 1 Ane Berro, Beatriz Fernández and Jon Ortiz de Urbina

2

Word Order 14 Jon Ortiz de Urbina

3

Tense, Aspect and Mood 59 José Ignacio Hualde and Céline Mounole

4

Non-verbal Participles in Basque and Spanish Ane Berro

5

(In)transitive Verbs: Unergatives and Unaccusatives Ane Berro

6

Light Verb Constructions in Basque and Romance Víctor Acedo-Matellán and Anna Pineda

7

On Non-selected Datives: Ethical Datives in Basque and Spanish Beatriz Fernández

8

Differential Object Marking in Basque and Spanish Dialects Ane Odria

9

Complex Causative Verbs and Causes in Basque (and Romance) Jon Ortiz de Urbina

10

Grammaticalization Processes in Causal Subordination José Ignacio Hualde and Manuel Pérez Saldanya Name Index 347 Subject Index 351

82

139

176

317

221

243

276

chapter 1

Introduction Ane Berro, Beatriz Fernández and Jon Ortiz de Urbina

1

Introduction

The book Basque and Romance: Aligning Grammars is a collection of articles describing and analyzing several of the most important morphosyntactic features for which the formal comparison between Basque and its surrounding Romance languages is relevant. In the context of a language virtually all of whose speakers are bilingual in either Spanish or French, the theoretically informed in-depth description offered in this volume focuses on the fine grain of linguistic structures from languages typologically quite apart but coexisting and probably interacting in the minds of speakers. The comparative approach taken here directs the attention not only to broad macroparameters but also to narrow empirical domains where microparametric differences show up in full complexity. The coexistence of Basque with two different Romance languages also allows us to contrast Basque varieties both from an internal and ‘external’ perspective. Some of the phenomena described in this book have a dialectal distribution which can be linked to similar patterns in the Romance language spoken in that area; this enables us to observe Basque internal phenomena which may be connected with contact with different external contacting languages. The aim of this book is therefore to produce a fine-grained comparison of some areas of Basque and Romance grammar, delineating similarities and differences, thus not only providing theoretical analyses of similar systems in apparently dissimilar languages but also feeding studies in language contact. We hope this grammar alignment, i.e., this close parallel examination of the grammars of bilingual speakers, can shed some light on the types of interactions between different systems and on the systems themselves. Moreover, we strive to provide detailed and in-depth descriptions on which contact studies could be reliably built as they address the analysis of the changes this type of grammar coexistence may bring about, favor or, perhaps, also hinder.

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004395398_002

2 2

berro, fernández and ortiz de urbina

Basque and Romance Grammars in Contact

Basque is a genetically isolated ergative language spoken by around 715,000 people in the Basque Country (5th Sociolinguistic Survey, Basque Government 2013), a territory divided into two different political and administrative regions, Spain and France. It is in permanent contact with Romance languages, as all Basque speakers are, nowadays, Spanish or French bilinguals. It used to be the majority language three or four centuries ago (Hualde 2003), but today, it is the native language of about 375,000 people, the 15 % of the total population (5th Sociolinguistic Survey, Basque Government 2013). Thanks to the official and semi-official status that Basque has in the Basque Autonomous Community and Navarre, both in the Spanish side of the border, many young speakers have learned the language through the educational system as a second language. As a consequence, almost half of the speakers of Basque are L2 speakers. On the other hand, Basque is also spoken in the French side of the border, in the historical provinces of Lapurdi, Low Navarre and Zuberoa. In this region, the language does not have official status and the number of speakers is decreasing rapidly. With virtual complete bilingualism, the contact between Basque and Romance has changed qualitatively in the 20th and 21st century, but, even if probably at a smaller scale, Basque has been in very close contact with Latin and Romance languages in the last two millennia, with interactions that have been identified in both directions. In the north, contact has been most intense with Gascon, and French itself only came into generalized contact with northern forms of Basque in modern times (Bayonne, the contemporary unofficial capital of the French Basque Country, was largely a Gascon speaking town). To the south of the present border, Basque shows greatest similarities with Pyrenean Aragonese dialects (Hualde 2016, Allières 1992) mostly extinct at present. Similarities with these Romance forms, and with early Castilian Spanish are based mostly on lexical, phonetic and phonological developments, and grammatical contact phenomena are difficult to trace given the scarcity of Basque language documents until the 16th century. But even so, they obviously existed. Imperfective markers derived from locative nominalizations are common enough not to require an external explanation, but as for perfective participial markers, the first isolated Basque glosses to Latin texts from the 10th or 11th century already show a -tu marker from early Romance origin which would eventually oust native -i in the course of the following centuries. Similarly, the verbal system of the first Basque long texts in the 15th and 16th century displays clear signs of an ongoing fairly thorough reorganization of the Tense, Aspect and Mood (TAM) system (Lafon 1943, Mounole 2011) leading to a system which by the 18th cen-

introduction

3

tury was more similar to that of neighboring Romance. As Hualde and Perez Saldanya (this volume) also show, grammaticalization patterns for causal subordinate clauses have followed similar paths. As a last instance, Basque shared with neighboring languages (even with English) important developments such as the deployment of second personal plural pronouns and verbal morphemes as markers of respectful second singular forms. All this indicates that shared developments and changes towards more Romance-like patterns have been taking place in Basque for a long period now. In the context of this centurieslong contact, the fact that we can still speak of Basque and Romance as being typologically ‘different’ languages may also require some explanation. In the last 150 years, however, factors such as the strengthening of French and Spanish state-level political organization, early Romance-exclusive schooling and contemporary bilingual schooling, sociological changes in population make-up and urbanization, the development of mass communication systems, among many others, have led to the extended bilingualism which may underlie the extensive changes Basque seems to be currently undergoing and which can be seen in this book. Obviously, language coexistence in space compounded with extensive language coexistence in the speakers’ brains triggers linguistic changes which make of Basque and Romance contact a data-rich field for linguistic studies from many different angles. The articles in this book aim at striking a balance in presenting as wide an overview of the most important phenomena as possible while providing at the same time detailed data and, in many cases, theoretical analyses.

3

Topics Covered in This Book

Research on Basque and Romance has made it clear that despite their typological differences (ergative vs. accusative marking, head final vs. head initial linearization, among others), Basque and Spanish/French show remarkable similarities. Thus, even if Basque can be descriptively characterized as a fairly typical OV language and both French and Spanish also display fairly standard VO features, information packaging mechanisms, however they are construed, produce in Basque clause-level orders which are (increasingly) similar to those found in Spanish and French (Ortiz de Urbina, this volume). Similarities are more evident in colloquial registers and in some dialectal contexts (schooling is mostly done in the standard variety). Some important similarities are therefore formally related to Romance phenomena (whether dialectal or not). For instance, Differential Object Marking (Bossong 1991, Lazard 2001, Aissen 2003) is attested in some South-western varieties in the Spanish-speaking area of the

4

berro, fernández and ortiz de urbina

Basque territory and shows great similarities with Spanish general a-marking (Torrego 2004), or even with the dialectal leísmo of the Basque Country (Fernández & Rezac 2016, Odria 2017 this volume). In the same vein, the use of the se clitic in French and Spanish is closely related with the absolutive marking of the subject and izan ‘be’ auxiliary selection in Basque intransitive verbs, especially in Romance-loan verbs (Alberdi 2003). When the verb in the original Romance language occurs with the se clitic, the borrowed predicate in Basque takes an absolutive subject and izan ‘be’ auxiliary, whereas when the verb in the original Romance language does not take se, then the predicate is generally used with an ergative subject and *edun ‘have’ auxiliary in Basque. As a matter of fact, when Spanish and French differ in the use of se in certain verbs, the loan verb in Basque shows dialectal variation, with a contrast between the Basque varieties in contact with Spanish and those in contact with French. The analysis of loan verbs and the comparison of counterpart intransitive verbs in Basque and Romance is of particular interest, bearing in mind the different case systems of the languages under study—Basque being ergative (Levin 1983, Ortiz de Urbina 1989, Oyharçabal 1992 among others) and Spanish/French being accusative—and auxiliary alternation. For instance, Spanish, unlike French, does not display auxiliary alternation, but, even so, the presence of the se clitic in Spanish goes hand in hand with the selection of the izan ‘be’ auxiliary in Basque loan verbs (Berro, this volume). Other similarities go beyond dialectally constrained structures, such as the nature of datives in general and non-argumental datives in particular. Although recent literature has focused on the high vs. low nature of applied datives especially after Pylkkänen 2008 [2002], even higher in the architecture of the sentence seem to be the so-called ethical datives, as analyzed by Jaeggli (1982, 1986) for Spanish and by Jouitteau & Rezac (2007) for French among others. The sole existence of ethical datives in Basque is not uncontroversial, as affected (experiencer) datives seem to be of an argumental type, and therefore an indepth analysis of affectedness in Basque and Spanish/Basque is important. Actually, Bosse, Bruening & Yamada (2012) provide with a fine-grained typology that includes Hebrew, French, Japanese among other languages but lacks both Basque and Spanish materials. Thus, an incursion into affected datives and even allocutives (Oyharçabal 1993) will give us the opportunity to review classical syntactic tests (as in Borer & Grodzinsky 1986) and more novel criteria proposed by Bosse et alia (2012) and Horn (2008, 2013) in order to see to what extent affected datives are argumental or not in both Basque and Spanish (Fernández, this volume). The interplay between goal, experiencer, ethical datives on one hand and causee datives in causative structures on the other is also described for Basque and Spanish in Ortiz de Urbina (this volume). The appar-

introduction

5

ent typological distance created by ergative marking and complex agreement patterns in Basque largely dissolves to reveal both expected fundamental similarities and also high degrees of convergence when dative clitics and a marking in Romance is considered. Another topic covered in this volume involves non-verbal participles in Basque and Spanish. Resultative participles in many languages have been reported to have hybrid category, as they are both verbs and adjectives at the same time. As a consequence, they have been the subject of great discussions, particularly regarding the study of the interface between syntax and the lexicon (Wasow 1977) and syntax and morphology (Marantz 2001 2007, Anagnostopoulou & Samioti 2014), as well as lexical categories and the projection of syntactic heads like v, Voice and Asp(ect). Spanish past participles have been studied in a number of works (among others Luján 1981, Demonte 1983, Bosque 1999 2014, Marín 1997 2000 2004ab 2009, Gehrke & Marco 2014 among others), but there are not many studies about Basque resultative participles (Ortiz de Urbina & Uribe-Etxebarria 1991, Krajewska 2012 2013, Berro 2017) and none comparing both Spanish and Basque. Both languages are similar in building resultative participles by means of a locative copula (egon in western and central Basque varieties and estar in Spanish) and an adjectival (or adverbial) participle. However, Basque resultative participles are morphologically more complex (showing an additional resultative morpheme attached to the bare participle) and seem to be acceptable in more syntactic contexts, since they allow event-related modification. Additionally, if the copula turns into izan ‘be’ (substituting egon), Basque resultatives are compatible even with non-restricted event-initiators, whereas in Spanish only a subset of initiators is acceptable (Gehrke & Marco 2014). These contexts have been used as tests in order to analyze the structural composition of resultative participles across languages (Gehrke 2011, Alexiadou, Gehrke & Anagnostopoulou 2014, Alexiadou, Anagnostopoulou & Schäfer 2015) and are also useful in order to compare Spanish and Basque, and make a contribution to the current discussion of the topic (Berro, this volume).

4

Overview of the Chapters

As mentioned in previous sections, Basque and the surrounding Romance languages display significant similarities as well as obvious differences both in their structural characteristics and in their diachronic development. The chapters in this book address and analyze in detail what may be considered to be the most important areas where the comparison between Basque and Romance

6

berro, fernández and ortiz de urbina

grammar and the analysis of phenomena which may be due to their interaction can have a wider linguistic interest. Throughout the chapters it contains, topics such as word order, inflection, non-verbal participles, intransitive predicates, light verb constructions, ethical datives, differential object marking, causative constructions and causal clauses will be presented. In what follows, we briefly outline the main contributions in each chapter. The labels OV/VO or head-first/head-last, at least as general characterizations of (some) central elements in (some) phrasal constituents, would seem to place Spanish and French on one side of this typological divide and Basque on the other. Jon Ortiz de Urbina’s chapter on ‘Word Order’ shows that while Basque and Romance often differ in the location of their morphological and syntactic heads, considerable similarities (sometimes even convergence) may be found at the clausal level. Thus, SOV and SVO may correspond to grammatically neutral, pragmatically bleached orders in both languages (even if, at least in the case of Basque, certainly not to the most common ordering). However, there are numerous information packaging mechanisms which account for the majority of actual clause-level orders, and some of these, old and new, produce similar orderings in the two languages. Even though Basque does not have the right peripheral information focus of Spanish, the latter does have as an option a preverbal focus akin to the general-purpose focalization strategy of Basque, and clause final corrective focus is also increasingly used in Basque. In the course of the presentation of phrasal order, the chapter also points at ‘non-harmonic’ orders within lexical phrases, appositions and compounds. Their diachronic development and the comparison with Romance suggests that, at least in some cases, their presence is not just a function of well-known OV inconsistency. With virtually universal bilingualism in either Spanish or French and, in the western area, an important mass of new speakers that make up a good portion of Basque speakers in heavily populated areas, it would be surprising if a certain degree of grammatical confluence did not take place. We seem to observe increasing use of orders which exploit possibilities, traditional or not, which are compatible with the Spanish/French grammar of the same users. The chapter ‘Tense, Aspect and Mood’ coauthored by José Ignacio and Céline Mounole explores the morphological shape of the Tense, Aspect and Mood (TAM) system in finite verbal forms. As previously noticed in the literature (Michelena 1981), Basque analytical verbal forms have Romance counterparts that suggest similar diachronic development. Some similarities between Basque and Romance are the distinction between indicative and subjunctive, along with conditional and imperative, as far as the mood is concerned, or the contrast between perfective and imperfective forms regarding aspect. Nevertheless, there are also certain differences. For instance, Basque shows

introduction

7

a be/have auxiliary alternation whereas Spanish has only have auxiliary and French selects mainly have and reserves be for a small class of intransitives. These and other aspects show that Basque has both a convergent and a divergent development with its Romance neighbors. Ane Berro’s chapter ‘Non-verbal Participles in Basque and Spanish’ compares and contrasts non-verbal participles in both languages, in light of the literature on adjectival participles (among others, Wasow 1977, Marantz 2001 2007, Kratzer 1994 2000, Anagnostopoulou 2003, Embick 2004). The author points out a number of differences, regarding the morphological shape of the non-verbal participles and their aspectual interpretation. In fact, participles take extra morphemes in Basque when they are adjectival or postpositional, whereas in Spanish both verbal and non-verbal ones look similar. As for the aspectual interpretation, Basque non-verbal participles can be used with both resultative and experiential interpretation, but Spanish non-verbal participles only trigger the resultative reading. Despite these differences, Spanish and Basque participles share a number of properties. On the one hand, they show a similar interaction between the lexical aspect of the verb embedded under the participle and the aspectual interpretation of the whole structure. On the other, non-verbal participles in both Basque and Spanish (at least, the Spanish spoken in the Basque Country) (cf. Gehrke & Marco 2014), allow spatial and temporal modification of the event underlying the participle. The chapter ‘(In)transitive Verbs: Unergatives and Unaccusatives’, also authored by Ane Berro, analyzes the unaccusative/unergative divide of intransitive predicates (Perlmutter 1978 1989, Burzio 1981, 1986, Belletti 1988 1999, Sorace 1993, Levin & Rappaport Hovav 1995 and subsequent works) in Basque and several Romance languages. The author focuses on the verb classes that are systematically aligned in the unergative way or, alternatively, in the unaccusative way, both in Basque and Romance, and also border-line verbs (like Romance loan verbs and some new stative verbs). As proposed by the author, intransitive stative verbs are the most variable verbs both cross-linguistically and in Basque, precisely because the only argument of these predicates can be introduced externally or internally without making a big semantic difference. Víctor Acedo-Matellán and Anna Pineda’s chapter ‘Light Verb Constructions in Basque and Romance’ analyzes the morphosyntactic and semantic properties of Light Verb Constructions (LVC) in Basque and compares them to their counterparts in Romance—mainly Spanish and French. They argue that the Light Verb (LV) is responsible for the event- and argument-properties of the construction, whereas the Non-Verbal Element (NVE) provides conceptual information of the event involved. As shown in Basque and also Spanish and French, the choice of the LV is meaningful, given that only DO and GIVE give

8

berro, fernández and ortiz de urbina

rise to synonymous LVC s in these languages. Regarding the morphosyntactic and semantic relation of the NVE and LV, the authors show that Basque LCV allows NVE to be headed by postpositional or adverbial suffixes—with certain aspectual interpretations—, contrary to Spanish and French, where the NVE must be nominal. This nominal can be either bare—in French—or, in the case of Spanish and Catalan, necessarily accompanied with a determiner. Finally, this chapter also addresses the syntactic cohesion existing between the LV and NVE and shows that there is variation among the different LVC s with respect to the degree of incorporation. Beatriz Fernández’s chapter entitled ‘On Non-selected Datives: Ethical Datives in Basque and Spanish’ deals with non-selected datives. Basing on Franco & Huidobro (2008), the chapter proposes a three-way typology of ethical datives: First of all, Class I, reminiscent of personal datives (Horn 2008 2013); secondly, Class II, which correspond to affected experiencers in Bosse, Bruening & Yamada (2012); and thirdly, Class III, ethical datives stricto sensu. All classes are attested in Spanish, whereas as the author shows, only Class II—affected experiencers—can be found in Basque. Both classical and new tests are used in order to support this claim, such as the existence of person restrictions, doubling, clitic clusters, and the presence of truth conditional and non-truth conditional meaning. Although no particular analysis is provided, the author extends Odria’s (2017) analysis of benefactives and external possessors to Class II ethical datives, and claims that these datives are merged in the specifier position of an Applicative head above v. The chapter on Differential Object Marking (DOM) by Ane Odria compares this phenomenon in both Basque and Spanish, particularly the variety of Spanish spoken in the Basque Country (Landa 1995). The chapter shows that DOM in these languages (or language varieties) is conditioned by animacy and specificity, and has the same morphological shape of indirect objects, that is, dative case and agreement. The author claims that DOM objects have the categorial status of DP s, like causee, experiencer and possessor datives. By contrast, datives of bivalent unergatives of the lagundu/ayudar ‘accompany’, ‘help’ and ditransitive predicates are PP s. In fact, the author shows that DP datives, DOM objects among others, can control depictive secondary predicates, whereas PP datives cannot. DOM-like phenomena are also discussed within the wider domain of causative formation in Jon Ortiz de Urbina’s chapter ‘Complex Causative Verbs and Causees in Basque (and Romance)’. These constructions present another case of a prima facie drastic typological difference between Basque morphological and Romance analytical complex verb formation mechanisms in causatives (Comrie 1989, Dixon 2000), which, when considered in detail, turn out to

introduction

9

present perhaps not-so-striking parallelisms. The Basque causative morpheme is fairly transparently a verb with an archaic causative affix, while some Romance fare/faire/hacer causatives with infinitival complement clauses have been treated, at least since Zubizarreta (1985) as monoclausal structures. The chapter also provides a fine-grained comparison of direct and indirect causation and impersonal causatives in Basque and Romance, resorting further to Basque mediopassive causatives as a term of comparison with Romance passive causatives. Causatives involve the presence of an extra, causer argument supplied by the causative verb, and case marking patterns for this argument are also discussed in detail in this chapter. Dative marking for the causee is standard in Basque for some unergative causee subjects and for transitive subject causees, while dialectal and substandard patterns increasingly resort to dative marking even in these contexts with animate causees, a DOM pattern particularly clear in dialects in contact with Spanish. This dative may in turn cooccur with a dative argument of the verb in the caused event, whether a goal, an experiencer or a non-argumental ethical dative, and the chapter presents a detailed description of the availability of dative doubling in these contexts, comparing it with dative clitic marking in Spanish. José Ignacio Hualde and Manuel Pérez Saldanya’s chapter ‘Grammaticalization Processes in Causal Subordination’ focuses on Basque and Romance causal clauses. As shown in the chapter, Basque exhibits more heterogeneity than Romance regarding the subordinators involved in these clauses, given that not only verbal suffixes but also prefixes and clause initial conjunctions are attested in Basque. The chapter explores in-depth the development of Basque and Romance causal structures in both their similarities and differences. From a functional perspective, in both Basque and Romance, there is a causal subordinator with a polyfunctional character and other subordinators appearing in more restricted contexts. The authors show that the grammaticalization paths of causal elements are similar in many respects, even though some Basque strategies such as the grammaticalization of the enclitic eta ‘and’ as a clausal connector lack a Romance counterpart.

References Aissen, Judith. 2003. Differential object marking: iconicity vs. economy. Natural Language and Linguistic Theory 21: 435–483. Alberdi, Xabier. 2003. Euskal aditz mailegatuen erregimena: hurbilketa. In Jesus Mari Makazaga & Bernard Oyharçabal (eds.), P. Lafitteren sortzearen mendemugako biltzarra. Gramatika gaiak. Iker 14 (I). Bilbao: Euskaltzaindia. 37–60.

10

berro, fernández and ortiz de urbina

Alexiadou, Artemis, Elena Anagnostopoulou & Florian Schäfer. 2015. External Arguments in Transitivity Altenations. A Layering Approach. Oxford: Oxford University Press. Allières, Jacques. 1992. Gascón y euskera: afinidades e interrelaciones lingüísticas. ASJU 26 (3): 801–812 Anagnostopoulou, Elena. 2003. Participles and Voice. In Artemis Alexiadou, Monika Rathert and Armin von Stechow (eds.), Perfect explorations. Berlin: Mouton de Gruyter. 1–36. Anagnostopoulou, Elena & Yota Samioti. 2014. Domains within words and their meanings: A case study. In Artemis Alexiadou, Hagit Borer and Florian Schäfer (eds.), The Syntax of Roots and the Roots of Syntax. Oxford: Oxford University Press. Belletti, Adriana. 1988. The case of unaccusatives, Linguistic Inquiry 19: 1–34. Belletti, Adriana. 1999. “Inversion” as focalization and related questions, Catalan Working Papers in Linguistics 7: 9–45. Berro, Ane. 2017. Resultative and non-resultative participles in Basque. Ms, University of the Basque Country & SFL (UMR 7023, CNRS/Paris 8). Borer, Hagit & Yosef Grodzinsky. 1986. Syntactic cliticization and lexical cliticization: The Case of Hebrew Dative Clitics. In Hagit Borer (ed.), Syntax and Semantics 19. New York: Academic Press. 175–215. Bosse, Solveig, Benjamin Bruening & Masahiro Yamada. 2012. Affected experiencers. Natural Language and Linguistic Theory (NLLT), 30(4): 1185–1230. Bossong, Georg. 1991. Differential Object Marking in Romance and beyond. In Dieter Wanner & Douglas A. Kibee (eds.), New Analyses in Romance Linguistics. Amsterdam / Philadelphia: John Benjamins. 143–170. Bosque, Ignacio. 1999. El sistema adjectival. Complementos y modificadores del adjetivo. Adjetivo y participio. In Ignacio Bosque & Violeta Demonte (eds.), Gramática descriptiva de la lengua española. vol. 1. Madrid: Espasa Calpe. 217–310. Bosque, Ignacio. 2014. On resultative past participles in Spanish. Catalan Journal of Linguistics 13: 41–77. Burzio, Luigi. 1981. Intransitive verbs and Italian auxiliaries. Ph.D. diss, MIT. Cambridge, MA. Burzio, Luigi. 1986. Italian Syntax. Dordrecht: Reidel. Comrie, Bernard. 1989. Language Universals and Linguistic Typology. Chicago: Chicago University Press. Demonte, Violeta. 1983. Pasivas léxicas y pasivas sintácticas en español. Serta Philologica F. Lázaro Carreter. Madrid: Cátedra. 141–157. Dixon, Robert M.W. 2000. A Typology of Causatives: Form, Meaning and Syntax. In Robert M.W. Dixon & Alexandra Y. Aikhenwald, eds. Changing valency. Case studies in transitivity, Cambridge: Cambridge University Press, 30–83. Embick, David. 2004. On the Structure of Resultative Participles in English. Linguistic Inquiry 35-3: 355–392.

introduction

11

Fernández, Beatriz & Milan Rezac. 2016. Differential Object Marking in Basque varieties. In Beatriz Fernández & Jon Ortiz de Urbina (eds.), Microparameters in the grammar of Basque. Amsterdam: John Benjamins. 93–139. Franco, Jon & Susana Huidobro. 2008. Ethical Datives, Clitic doubling and the Theory of pro. Joyce Brukes de Garavito & Elena Valenzuela (eds.), Selected Proceedings of the 10th Hispanic linguistics Symposium. Somerville: Cascadilla Proceedings Project. 215–224. Gehrke, Berit. 2011. Stative passives and event kinds. In I. Reich, E. Horch and D. Pauly (eds.), Proceedings of Sinn und Bedeutung 15. Saarbrücken: Universaar—Saarland University Press. 241–257. Gehrke, Berit & Cristina Marco. 2014. The role of by-phrases in adjectival passives. Lingua 149: 188–214. Horn, Laurence R. (2008). I love me some him: The landscape of non-argument datives. In Olivier Bonami & Patricia Cabredo Hofherr (eds.), Empirical Issues in Syntax and Semantics 7, CSSP, Paris, 169–192. [On-line publication, http://www.cssp.cnrs.fr/ eiss7]. Horn, Laurence R. (2013). I love me some datives: Expresive meaning, free datives, and F-implicature. In Daniel Gutzmann & Hans-Martin Gärtner (eds.), Beyond Expressives: Explorations in Use-Conditional Meaning. Leiden: Brill. 153–201. Hualde, José Ignacio. 2003. Introduction. In José Ignacio Hualde & Jon Ortiz de Urbina (eds.), A grammar of Basque. 1–14. Berlin: Mouton de Gruyter. Hualde, José Ignacio. 2016. On Basque dialects. In Beatriz Fernández & Jon Ortiz de Urbina (eds.), Microparameters in the Grammar of Basque. Amsterdam/Philadelphia: John Benjamins.15–35 Jaeggli, Osvaldo, A. 1982. Topics in Romance Syntax. Dordrecht: Foris. Jaeggli, Osvaldo, A. 1986. Three issues in the theory of clitics: case, doubled NP s and extraction. In Hagit Borer (ed.), Syntax and Semantics 19. New York: Academic Press. 15–42. Jouitteau, Mélanie & Milan Rezac (2007), The French ethical dative, 13 syntactic tests. Bucharest Working Papers in Linguistics, IX (1): 97–108. Krajewska, Dorota. 2012. Resultatives in Basque: a diachronic study. Lingua Posnaniensis LIV(2): 56–67. Krajewska, Dorota. 2013. The diachrony of resultative constructions in Basque. Master Thesis, UPV/EHU. Kratzer, Angelika. 1994. The Event Argument and the Semantics of Voice. Ms., University of Massachusetts at Amherst. Kratzer, Angelika. 2000. Building Statives. Berkeley Linguistic Society 26. 385–399. Lafon, R. 1980 [1943]. Le système du verbe basque au XVIe siècle. Zarautz: Elkar. Landa, Alazne. 1995. Conditions on Null Objects in Basque Spanish and their relation to Leísmo and Clitic Doubling. University of Southern California.

12

berro, fernández and ortiz de urbina

Lazard, Gilbert. 2001. Le marquage différential de l’object. In Martin Haspelmath, Ekkehard König, Wulf Oesterreicher & Wolfgang Raible (eds.), Language Typology and Language Universals. An International Handbook. 2 vol. Berlin / New York: Mouton de Gruyter. 873–885. Levin, Beth. 1983. On the Nature of Ergativity. Doctoral Dissertation, MIT. Levin, Beth & M. Rappaport Hovav. 1995. Unaccusativity. At the Syntax-Lexical Semantic Interface. Cambridge, MA: the MIT Press. Luján, Marta. 1981. The Spanish copulas as aspectual indicators. Lingua 54: 165–209. Marantz, Alec. 2001. Words. WCCFL XX Handout, USC, February 2001. Marantz, Alec. 2007. Phases and Words. In Sook-Hee Choe (ed.) Phases in the Theory of Grammar. Seoul: Dong In. 191–222 Marín, Rafael. 1997. Participios con aspecto de adjetivos: entre la diacronía y la morfología. Moenia 3: 365–376. Marín, Rafael. 2000. El componente aspectual de la predicación. Doctoral dissertation. Universitat Autònoma de Barcelona. Marín, Rafael. 2004a. Entre ser y estar. Madrid: Arco/Libros. Marín, Rafael. 2004b. Sobre pasivas adjetivales. Verba 31: 455–471. Marín, Rafael. 2009. Del adjetivo al participio. In Elena de Miguel et al. (eds.), Fronteras de un diccionario: las palabras en movimiento. San Millán de la Cogolla: Cilengua. 327–348. Michelena, Luis. 1981. Lengua común y dialectos vascos. ASJU XV: 291–303. Reprinted in: Luis Michelena. 2011. Obras completas [Supplements of ASJU VII]. ed. by Joseba A. Lakarra & Iñ igo Ruiz Arzallus, Donostia-San Sebastián & Vitoria-Gasteiz: Gipuzkoako Foru Aldundia & UPV/EHU. 517–544. Mounole, Céline. 2011. Le verbe basque ancien: étude philologique et diachronique. Doctoral dissertation, University of the Basque Country (UPV/EHU) and Université Michel de Montaigne-Bordeaux III. Odria, Ane. 2017. Differential Object Marking and datives in Basque syntax. Doctoral dissertation, UPV/EHU. Ortiz de Urbina, Jon. 1989. Parameters in the Grammar of Basque. Studies in Generative Grammar 33. Dordrecht: Foris Publications. Ortiz de Urbina, Jon & Myriam Uribe-Etxebarria. 1991. Participial predication in Basque. ASJU Gehigarriak 14-2. 993–1012. Oyharçabal, Bernard. 1992. Structural and Inherent Case-Marking; Ergaccusativity in Basque. In Joseba, A. Lakarra & Jon Ortiz de Urbina (eds), Syntactic Theory and Basque Syntax. Supplements of ASJU 27. Donostia: UPV/EHU and Gipuzkoako Foru Aldundia. 309–342. Oyharçabal, Bernard. 1993. Verb agreement with non-arguments: On allocutive agreement. In José Ignacio Hualde & Jon Ortiz de Urbina (eds.), Generative studies in Basque linguistics. Amsterdam: John Benjamins. 89–114.

introduction

13

Perlmutter, David, M. 1978. Impersonal Passives and the Unaccusative Hypothesis. Proceedings of the 4th Annual Meeting of the Berkeley Linguistics Society. 157–190. Perlmutter, David, M. 1989. ‘Multiattachment and the unaccusative hypothesis: the Perfect Auxiliary in Italian.’ Probus 1-1: 63–119. Pylkkänen, Liina. 2008. Introducing arguments. Cambridge / London: MIT Press. Sorace, Antonella. 1993. Unaccusativity and auxiliary choice in non-native grammars of Italian and French: asymmetries and predictable indeterminacy. Journal of French Language Studies 3: 71–93. Wasow, Thomas. 1977. Transformations and the lexicon. In Peter Culicover, Thomas Wasow & Adrian Akmajian (eds.), Formal Syntax. 327–360. New York: Academic Press. Zubizarreta, María Luisa. 1985. Romance accusatives. Linguistic Inquiry 16: 247–289.

chapter 2

Word Order Jon Ortiz de Urbina

Basque and Spanish differ rather drastically in the organization of constituents internal to major categories, the former being largely head-final while the latter is head-initial. On the other hand, word order at the clausal level is fairly flexible in both languages. Thus, while Basque is usually classified as an SOV language, this is largely a typological shorthand for word orders other than those found for the combination of subject, object and verb with each other. While SOV is often recognized as a ‘basic’ order, Basque is far from being a rigid verbfinal language, and heaviness as well as information packaging produce a wide array of possible orders of major clausal constituents. It may not be misleading to claim that Basque is a fairly typical OV/head-last language, while Spanish is a fairly typical VO/head-first language. For the purposes of this chapter, however, rather than discussing the validity of these macro-labels, it is more interesting to examine in more detail the situation across categories and the possible existence of changes which may increase convergence between the two languages.1

1

Morphology

Disregarding dvandva cocompounds, very productive in Basque but where there is an additive, symmetric relation between their components (Hualde 1998), NN compounding is systematically right-headed: itsasgizona ‘seaman’ (itsas(o) ‘sea’ + gizona ‘man’) is structurally akin to the English translation.2 This

1 The abbreviations used in glosses are as follows: aux ‘auxiliary’, dat ‘dative’ det ‘determiner’, comp ‘complementizer’, erg ‘ergative’, fut ‘future’, gen ‘genitive’, impf ‘imperfective’, nom ‘nominalizing affix’, part ‘partitive’, pl ‘plural’, q ‘question morpheme’. Glosses in a language like Basque can be very obstrusive, so for readability purposes they have been kept to a minimum, providing in most cases only the grammatical information for which the example is relevant. Thus, the determiner (and agreement) marker -a may be glossed as det, ‘the’ or simply left out depending on the relevance to the discussion. 2 Apparent exceptions such as erpuru ‘thumb’ or hiriburu ‘capital’, (eri ‘finger’ and hiri ‘town’ + buru ‘head’), designating types of fingers or towns may not be so when compared with the

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004395398_003

word order

15

type of compounding is also extremely productive in Basque, while it is much less so in Spanish; examples in the latter are systematically left-headed. Thus, hombre rana contrasts structurally with both its English and its Basque counterparts, frogman and igelgizona, respectively, and perro pastor ‘shepherd dog’ places the two elements in the opposite position with respect to English and Basque artzain txakurra. Spanish often resorts to syntactic phrases where the dependent is introduced by the preposition de or expressed as an adjective in cases where Basque displays NN compounds: copo de nieve ‘snowflake’, Basque elur maluta (elur ‘snow’, maluta ‘flake’); similarly, Spanish derechos humanos ‘human rights’ corresponds to Basque giza eskubideak (giza ‘human being’, eskubide ‘right’). NN compounds of the type briefly described in the previous lines are particularly interesting in that they present mergers of elements which are not usually combined in the (traditional view of) syntax. Other types of compounding of categories have clearer parallels in the syntax, and often mirror the ordering of their syntactic parallels. Thus, NV primary compounds are not very productive in Basque, but existing ones are again right-headed from the earlier known stages of the language. Thus Michelena (1977), Mounole (2009, 2011) identify itxaron ‘wait’ as hitzedun (hitz ‘word’ + edun ‘have’), and iguriki ‘wait’ as eguneduki (egun ‘day’ + eduki ‘hold’); berdin ‘same’ is an old compound with predicative complement ber- ‘same’ and copulative edin ‘be, become’. In a more productive and contemporary vein, bahuvrihi compounds are very common with what de Rijk (2008) identified as edun ‘have’ in the second element of words like euskaldun ‘Basque (speaker)’ (euskara ‘Basque’), haurdun ‘pregnant’ (haur ‘child’), etc. Other such compounds with egin ‘make’ as second conjunct are okin ‘baker’ (ogi ‘bread’) or iturgin ‘plumber’ (iturri ‘fountain’). Endocentric compounds with a second verbal element are also productive, both in traditional words and neologisms: itxuraldatu ‘transform’ (itxura ‘shape’ + aldatu ‘change’), antzeman ‘guess’ (antza ‘appearance’ + eman ‘give’), lurrartu ‘land’ (lur ‘earth’ + (h)artu ‘take’), etc. Compounds with verbs in Spanish conform to the basic VO pattern of the language, as in the very productive type forming

expected asteburu ‘weekend’ (aste ‘week’) or iturburu ‘spring’ (iturr(i) ‘fountain’). Notice that while the latter are interpreted as the head of the week and perhaps as the head of a fountain, the former are not the head of a finger or of a city. Rather, buru receives an interpretation actually close to the one represented in capital, originally an adjective (from the Latin word for head) in the collocation capital city, which the Basque word follows closely. Given that the categorial distinction between nouns and adjectives was not clear in older stages of the language, we may assume that buru in the first two examples is working as an adjective, and that these are N Adj compounds of the types described below.

16

ortiz de urbina

nouns with a third person verbal form as first term: saltamontes ‘grasshopper’ (salta ‘jumps’ + montes ‘mountains’), sacacorchos ‘corkscrew’ (saca ‘takes out’ + corchos ‘corks’). Basque NN secondary compounds (again just as productive in Basque as they are scarce in Spanish) also follow systematically an OV pattern, with the deverbal noun to the right as in ile apaingile ‘hairdresser’ (ile ‘hair’ + apaingile ‘dresser’, from apain- ‘beautify’ and -gile ‘-er’) or suhiltzaile ‘firefighter’ (su ‘fire’ + hiltzaile ‘killer’, from hil ‘kill’ and -tzaile ‘-er’).3 We also find word orders similar to those of syntactic phrases in N Adj compounds. These follow this order in both Spanish and Basque, as expected since adjectives usually follow nouns in the syntax of both languages; cf. Bsq. belarrimotz ‘small-eared, foreigner’ (belarri ‘ear’ + motz ‘short’), and Span. paticorto ‘short-legged’ (pata ‘leg’ + corto ‘short’). See 2.2.1.3 below for adjective ordering within nominal phrases. There are few compounds containing adpositions in either Basque or Spanish, and the few cases that do exist show the expected order between the adposition head and its nominal complement. Spanish sin ‘without’ is the most productive preposition found in compounds: sinsentido ‘nonsense’, sinsabor ‘setback’, sinvergüenza ‘shameless, scoundrel’, sinsorgo ‘silly’;4 its Basque postpositional counterpart is gabe ‘without’, also very productive, as in paregabe ‘matchless’, etengabe ‘continuous’, lit. ‘without break’, lotsagabe ‘shameless’ or beldurgabe ‘fearless’. Other prepositions are only occasionally found in Spanish: bajotecho ‘attic’, lit. ‘under roof’, sobrecama ‘bedspread’, lit. ‘over bed’, or sobrepeso ‘overweight’. There are even fewer examples in Basque, partly because many Spanish prepositions correspond to Basque ‘case’ endings, bound inflectional morphemes. There are, however, independent postpositions in Basque, of which gabe ‘without’ is a typical example. Most postpositions, however, are relational nouns inflected for a locative case. These combine with a noun in the genitive (or a verb in the participial form) or may be compounded with it;5 in either case, the expected order is main-

3 Itself from (e)gi(n) ‘do’ and -le ‘-er’. 4 All are nouns/nominalized expressions, with only sinvergüenza having a possible adjectival use. The predominance of negative meanings may be related to similar semantics in what Bosque (1989) calls evaluative nominalization. 5 The absence of a genitive marking is not necessarily an indication of compounding. Thus, etxe aurrean ‘in front of the house’ (lit. ‘at the house front’) is spelled as two independent words, capturing a phonetic and perhaps morphological intuition. Notice that the postposition still requires a locative case, indicating the phrase has not become an independent expression. There is a difference between ate aurrean ‘in front of the door’, with this type of genitiveless phrase, and the bona fide compound ataurre ‘threshold’ (ate ‘door’ + aurre ‘front’). The issue is irrelevant for ordering, however, so different types are used in the text as illustration.

word order

17

tained: etxe(aren) aurrean ‘in front of the house’, bazkalondo ‘chat after eating’ (cf. bazkaldu ondoren/ondoan ‘after having lunch’). Some relational nouns that form the basis of postpositions, however, can be found in prenominal position, only indirectly as a result of accommodation to Romance patterns. Thus, azpi ‘below’ is used as the equivalent of the Romance prefix sub- in neologisms like azpiegitura ‘substructure’ or azpimultzo ‘subset’. Basque has a wide array of derivational suffixes but extremely few prefixes,6 and this type of semantically transparent Latinate prefix is also transparently rendered by the noun azpi. This is then a case of accommodation to the prefixation possibilities of Romance, rather than to any kind of prepositional pattern. Notice that the prefixed element is the basic noun without the locational marker, and that there is no complement relationship between this noun and the one to its right: azpiegitura ‘substructure’ is not interpreted as ‘under the structure’ or ‘underside of the structure’, but as a structure which is located under some other. The locational noun acts as an adjunct, not as a head.7 Similar considerations apply to the use of aurre ‘before/front’ as equivalent to the prefix pre- in neologisms like aurrizki ‘prefix’, aurreritzi ‘prejudice’, lit. ‘before opinion’. The latter is interpreted not as what is held before an opinion, but as an opinion formed previous to contact with data and/or arguments.8 We have already briefly mentioned derivation while discussing relational nouns. As indicated above, Basque is extremely limited in terms of derivational prefixes. This in itself fits Greenberg’s findings on the relationship between affixation and phrasal ordering, since he observed a tendency for SOV languages to display suffixation exclusively, while SVO languages usually exhibit both. In principle, if the affix is the head of a derived word, determining its category, one

6 As pointed out by de Rijk (2008), only three: ber- ‘re-’ and the loaned prefixes arra- ‘re’ and des- ‘un-’. 7 This is also the use found in other words containing azpi- in the traditional vocabulary, such as azpijoko ‘foul play’, lit. ‘under play’, or azpizapo ‘traitor’, lit. ‘under toad’. 8 Some morpheme ordering relationships within the number system were already claimed to be linked to other phrasal relations in Lehman (1975). In particular, he claimed that in additive numbers (for instance, thirteen, three plus ten, as opposed to thirty, three times ten) the smaller number added to the bigger one (the decimal in this type of system) behaves like the verb in the O/V relation. Regardless of the overall validity of this correspondence, it does capture the relation between the Basque and Spanish systems. Few numbers are additive in these languages, 11 through 19 in Basque and 11 through 15 in Spanish, the remaining higher numbers being either factorial (berrogei ‘40’, lit. 2 20 in Basque and cuarenta in Spanish, with initial 4) or coordinations (berrogeitamar ‘50’, lit. 2 20 and 10 in Basque or dieciséis ‘16’, lit. 10 and 6 in Spanish). Basque additive numbers take the form ten + unit (hamairu ‘13’, lit. 10 3) and, more opaquely, unit + ten in Spanish (doce ‘12’, 2?).

18

ortiz de urbina

would expect that head-first languages would favor prefixation, whereas headfinal languages like Basque would favor suffixation; however, this is clearly not the case.9

2

Syntax

After examining morpheme ordering phenomena, let’s turn now to word order phenomena in a more strict sense, devoting special attention to interlinguistically or intralinguistically unexpected patterns, i.e., unexpected when compared to patterns typical of OV languages and/or incongruent with other ordering internal to Basque. We begin with appositions in 2.1, devoting 2.2 to word order in major lexical phrases and 2.3 to ordering at the clausal and sentence level. 2.1 Appositions The relative order between type and token in restrictive appositions correlates with other VO/OV characteristics, so that the type expression corresponds to the V position and the token phrase to the O position. In fact, this was the content of Greenberg’s Universal 23, where he showed a correlation between proper noun/common noun appositions and the order of genitive/noun pairs.10 Since the genitive/noun order patterns with the V/O order, this type of apposition would then show overall mirror image effects when comparing Basque and Spanish. The differences between the two languages are clear in this respect: (1) Agirre Jauna/ El señor Aguirre ‘Mr. Aguirre’ (2) Kantauri Itsasoa/Mar Cantábrico ‘Bay of Biscay’ (3) Ibaizabal ibaia/El río Ibaizabal ‘the river Ibaizabal’ (4) 2000 urtea/el año 2000 ‘the year 2000’ (5) Gernikako Arbola Plaza/ Plaza Gernikako Arbola ‘Gernikako Arbola Square’

9 10

Kayne (2017) for an antisymmetric approach to this aspect of morphology. The wording of this Universal was in fact mistaken, as noted in Bennet (1979); see Cinque (2013: 110).

word order

19

(6) El País Egunkaria/Diario El País ‘the newspaper El País’ (7) purpura kolorea/el color púrpura ‘the color purple’ As usual, it is possible to find exceptions, more in Basque than in Spanish,11 although contact with Spanish may be the main driving force behind them. Thus, the pressure of official designations, systematically in a Romance language, accounts for the presence of VO-like apposition patterns in Basque different from the general ones displayed above. It is perhaps not surprising that the names of saints are identical in both languages, with a Saint+Name order unexpected in Basque (San Juan). More interestingly, the same situation obtained in earlier, pre-Trento times, when saint denominations were not suspect of heresy and therefore were not subjected to uniformity constraints; the same is the case for toponyms (which, themselves, often double with an official designation). Earlier saint names usually contain the title Done (from Latin domine)¸ often preceded by the same (preposed) Jaun in (1) above in the case of male saints (Michelena & Yrigarai 1955): Jaun Done Petri ‘Saint Peter’, top. Doneztebe ‘Saint Stephen’, Jaona domne Jakue ‘Saint Jacques’ (12th century). Puristic tendencies at the end of the 19th century introduced the treatment Deun(a) for the then standard Spanish San(to/ta), initially at the same prenominal position but later in the expected OV position following the name: Agate Deuna ‘Saint Agatha’ (at first Deun Agate), Mikel Deuna ‘Saint Michael’, etc. The unexpectedly preposed title Jaun ‘Sir’ of traditional saint denominations can also be found preceding the token in other official contexts where the pressure of official Romance denominations would be easily felt. This was probably the case of high status positions in a traditional society, such as mayor, priest, doctor, governor, etc. Similarly, early texts also present prenominal Jaun with proper names, probably following the Romance model too, and this position has remained frequent in northern dialects. Appositions whose type nominal corresponds to a family relation noun are often built on a relation+name order. This is expected in the case of nouns like aita ‘father’ and ama ‘mother’, only used in religious contexts akin to those described in the previous paragraph, as in Aita Donostia ‘Father Donostia’ or

11

Cinque (2013) also notes counterexamples in both OV and VO languages, but the only Romance language mentioned in this respect, Italian, conforms to the common/proper order in all instances he checks. English displays more deviations (Church Street, Coney Island, Potomac River), perhaps connected to the head last organization of NN compounds.

20

ortiz de urbina

Ama Teresa Kalkutakoa ‘Mother Teresa of Calcutta’, or even Ama Lurra ‘Mother Earth’, but less so in the much more productive cases involving relations like izeko ‘aunt’, osaba ‘uncle’ or amona ‘grandmother’. Example (8) is a celebratory radio message for the Bizkaia Irratia radio station indicating the persons wishing their 6-year-old relative a happy birthday, while (9) displays two family relation names, each in a different order: (8) Bere aita Gaizkak, ama Isabelek eta neba Aritzek. Baita her father G. mother I. and brother A. also osaba Mikelek, aitona Jose Martin eta amona uncle M. grandfather J.M. and grandmother Mari Karmenek. M.K. ‘Her father Gaizka, mother Isabel and brother Aritz. Also uncle Mikel, grandfather Jose Martin and grandmother Mari Karmen.’ (9) Bere osaba Pedrok Guillerma amatxiren ganik ukan zuen etxe his uncle Pedro Guillerma grandm.gen from have aux house gotorrari. strong.dat ‘To the strong house that his uncle Pedro had (received) from grandmother Guillerma.’ (Oncle, Ardoy SFran 84) The Orotariko Euskal Hiztegia [Basque General Dictionary], from which the latter is taken, lists examples with preposed and postposed osaba, but the tendency seems to be towards preposing. 2.2 Order in Major Phrases Although the distinction is epiphenomenal, we will discuss first ordering relationships in the major lexical phrases (and associated functional elements), excluding verbal complements and modifiers, which will be considered in section 2.3. This enables us to discuss in the latter more traditional ‘clause level’ ordering relationships. 2.2.1 NP s Not unlike the situation described in 2.1, Basque behaves in general like an OV language, with some ‘anomalies’, some but not all of which can be attributed to contact phenomena with Spanish and Romance in general. We will describe determiners and quantifiers in 2.2.1.1, adposition phrase modification of nouns in 2.2.1.2, adjective modifiers in 2.2.1.3 and relative clauses in 2.2.1.4.

word order

21

2.1.1 Determiners and Quantifiers Let’s briefly review first of all order relations between grammatical elements such as determiner/quantifiers and nouns. One of Abney’s (1987) crucial arguments in favor of the existence of Determiner Phrases headed by articles, demonstratives and genitives in English was that if they are indeed heads taking NP complements, the expected order in an OV, head-last language would be noun-determiner. This is indeed the difference between Spanish and Basque: the three of them precede the noun in the former, while articles and demonstrative follow the noun in Basque (possessives are expressed by postpositions, as indicated above). In fact, while it is likely that the grammaticalization of a demonstrative as an article was taking place as an areal phenomenon, affecting Romance and Germanic languages as well as Basque at roughly the same early medieval times, the grammaticalization in Basque retained the final position of the original demonstrative, producing contemporary mirror image orders like este chico/mutil hau ‘this guy’ (lit. ‘guy this’) and el chico/mutila ‘the guy’ (lit. ‘guy the’). There is only one type of construction where the determiner apparently precedes the noun, Biscayan hori mutilori, where hori ‘that’ precedes the noun. But notice that this is possible only if there is yet another demonstrative in the expected position (the -ori form following the noun).12 There are in fact two phrases involved here, and case marking must then be doubled. Thus, the ergative form of this phrase would be horrek mutilorrek, with two -ek endings, akin to Hungarian forms like abban a házban ‘in that house’, literally ‘in that in the house’. The initial determiner is therefore not directly associated with the noun. In the case of Spanish postnominal demonstratives like el chico ese ‘that guy’ we do not find demonstrative doubling, since the final demonstrative coexists with the article el. The traditional analysis of ese as an adjective in this case, rather than as a determiner, captures the distinction between the determiner demonstrative and this one.13 It shares the postnominal position of adjectives, although its positional possibilities are more varied than for normal adjectives: el chico ese alto de gafas, el chico alto ese de gafas, el chico alto de gafas ese are

12 13

The sound [h] is not pronounced in this dialect, so there is no difference here between the initial and final determiners. The Biscayan pattern in the plural seems closer to the Spanish article+demonstrative pattern: hónek mutilok ‘those guys’, with the proximal article -ok instead of a doubled demonstrative. Since the proximal plural form -ok derives historically from the proximal plural demonstrative *hauk (rather than the distal *hak, which is the source of the plural article), the mismatch is actually smaller than seems to be the case at first sight. In any event, final -ok is still the determiner, unlike the final demonstrative in Spanish.

22

ortiz de urbina

all ‘that tall guy with glasses’. In contrast, the final demonstrative in the Basque examples is exclusively final, since it is a regular determiner after all. Quantifiers make up a semantic class sometimes behaving like determiners, and therefore incompatible with them. Thus, muchos in mis muchos amigos ‘my many friends’ is compatible with a determiner and, therefore, probably not a determiner itself, while bastantes ‘quite a few, several’ cannot cooccur with the genitive determiner, as shown in *mis bastantes amigos ‘my several friends’. There is nothing semantically odd about the latter, since the same meaning can be rendered in a fully grammatical way as bastantes amigos míos, with an adjectival genitive míos, crucially not a determiner. Strong Basque quantifiers like guzti ‘all’, den ‘all’, bakoitz ‘each’ or gehien ‘most’ must in fact cooccur with the definite determiner, while weak quantifiers like asko ‘many’, batzu(e)k ‘some’, gutxi ‘few’, zenbait ‘some’ and hainbat ‘some’ cannot coocur with them (see Etxeberria 2005). Thus, we find lagun guzti*(ak) ‘all friends’ with obligatory plural determiner but lagun gutxi(*ak) ‘few friends’ with a bare quantifier. From the point of view of word order, we find again the basic mirror image between Basque and Spanish, so that the set expression to express thanks contains initial muchas ‘many’ in Spanish but final asko ‘many’ in Basque (muchas gracias versus eskerrik asko ‘many thanks’). But, at the same time, we also find the occasional disharmonic orders in OV languages, so that even if most quantifiers are postnominal in Basque, there are also some prenominal ones. In particular, zenbait ‘some’ and hainbat ‘some’ precede the noun they quantify over: zenbait lagun ‘some friends’.14 These prenominal orders may be syntactically more complex than the postnominal ones, since the former contain operators which typically induce displacement in clauses and/or noun phrases: zenbait appears as zeinbeit in the earliest printed book in Basque, containing the wh-word zein-, and hainbat contains hain ‘so’, which, as discussed in 2.2.2.2 below also appears in a displaced position with respect to adjectives it combines with. If extra syntactic positions and processes affect these prenominal quantifiers, then, the overall mirror-image ordering of quantifiers and nouns in Spanish and Basque would be more thorough at a more abstract level of analysis.15

14

15

Zenbait may (rarely) follow the noun Nahiko(a) ‘enough, quite a few’ is more complex, since depending on the dialect it can precede or follow the noun it quantifies over; it can also occur as an external quantifier, as if floating out of the NP, taking then the determiner: a phrase like enough guys can be found across dialects as mutil-a nahiko-a, mutil nahiko-a, nahiko-a mutil, nahiko-a mutil-a, nahiko mutil-a and nahiko mutil. The form is also morphologically complex and may contain the genitive suffix -ko. For an analysis of Basque NP s, see Artiagoitia (2002).

word order

23

2.2.1.2 PP Dependents of Nouns Noun dependents and modifiers often take the form of PP s, which will then be postnominal in Spanish but prenominal in Basque: (10) etxearen jabea / el propietario de la casa house.gen owner.det the owner of the house ‘the owner of the house’ (11) etxeko jauna / el señor de la casa house.gen lord.det the lord of the house ‘the lord of the house’ (12) mugarik gabeko mundua / un mundo sin fronteras border.part without.gen world.det a world without borders ‘a world without borders’ (13) bihotzeko mina / una pena en el corazón heart.gen pain.det a pain in the heart ‘a pain in the heart’ (14) bihotz oneko pertsona / una persona de buen corazón heart good.gen person.det a person of good heart ‘a person of good heart’ The preferred adnominal preposition in Spanish is de, for both argumental and adjunct modifiers of the noun (as in (10) and (11)), as well as attributive modifiers (14). Other prepositions are also possible, as in (12) and (13). In all cases, the PP follows the head noun. As famously put by de Rijk (1993), Basque nouns are not very ‘hospitable’ to any type of phrasal dependents, which require the intervention of a genitive marker to be able to attach to the head noun. Argumental, thematic relations are typically marked with the genitive suffix -ren (as in (10)), while more peripheral adnominals are marked with -ko, added to an adverbial or a PP as in (12) or directly to the nominal in the case of locative/temporal relations like (11) or (13) and in attributive relations as in (14).16 Although the status of these ‘genitive’ markers is not uncontroversial (see de Rijk 1993), positionally they are always placed at the end of the phrase, and

16

De Rijk (1993) claims all cases of apparent bare nominals to which -ko is attached actually hide PP s, whereas genitive (r)en would attach to NP s.

24

ortiz de urbina

we can equate them with postpositions for word order considerations, forming PP s just like those headed by de or other prepositions in Spanish. Prepositional phrases are more ‘harmonic’ with head initial VP s and NP s, as in Spanish, while postpositional phrases are more typically found with head final phrases, as in the Basque examples above. Let us examine now deviations from these patterns. As usual, it is OV languages like Basque that exhibit some degree of deviation from the expected pattern: there are no prenominal PP s in Spanish, while postnominal PP s do exist in Basque, if in a limited way. First of all, dative adnominals constitute both a gap in the complementation pattern of nominal heads and a possible anomaly in the expected ordering relationships between head noun and complement. Thus, goal complements of nouns cannot be attached to the head by using either of the genitive markers discussed above, neither as nominals with -(r)en or as dative marked nominals with -ko: (15) a. entzulegoari eskaini / eskertu audience.dat offer / thank ‘to offer/thank the audience’ b. * entzulegoariko eskaintza / eskerrak audience.dat.gen offer / thanks ‘an offer/thanks to the audience’ c. entzulegoaren eskaintza/eskerrak audience.gen offer/thanks ‘the offer/thanks of the audience’ The dative verbal complement in (a) cannot be found as the complement to the noun in (b), expected if -ko only attaches to adjuncts and datives are complements. But if they are argumental complements, one would expect them to occur as prenominal elements displaying the -(r)en genitive marker, as in (c). However, as Kayne (1984) already noted, genitivized nominals cannot be construed as goal arguments of the modified noun, so that entzulegoaren ‘of the audience’ in (c) can only be interpreted as agent and/or theme. In order to express a goal relationship, it is possible to join the two phrases, but then the most common (although not exclusive) order is head-goal (see Fernández & Sarasola 2010):

word order

25

(16) a. omenaldia Xalbadorri / Xalbadorri omenaldia homage Xalbador.dat ‘homage to Xalbador’ b. abisua marinelei warning sailors.dat ‘a warning to sailors’ Thus, a song like Gracias a la vida ‘Thanks to life’ is usually rendered as Eskerrak bizitzari, literally ‘Thanks life to’, identical to the order in Spanish and English in the relevant respect. It is not clear, however, whether the two phrases are really integrated into a single expression or act with at least relative independence.17 Certainly, in the rare cases where the goal can be expressed with a dative, it occupies the pre-head position expected for an OV language. Compare for instance the previous example eskerrak bizitzari ‘thanks to life’ with an expression like berari esker ‘thanks to him’, where esker ‘thank’ has lost its nominal status and is used as a postposition requiring a dative complement. In any event, the possibility that the examples in (16) involve complements which are not really fully integrated with the head means the variation in word order, while still a piece of data that must be explained, is not a counterexample for the right-headed nature of noun+adnominal pairs. Real counterexamples would be expressions of the type N+XP-(r)en/-ko, where the genitive bearing nominal follows the head noun, and these do occur; however, most are old and found in lexicalized expressions of legal or religious origin where Romance influence is expected, as in aita pontekoa ‘godfather’ (lit. ‘father of the [baptismal] font’), aita besoetakoa ‘godfather’ (lit. ‘father of the arms’), aita zerukoa ‘father of heaven’, aingeru guardakoa ‘guardian angel’ (Span. ángel de la guarda), aita gurea ‘our father’ (as in the prayer, Span. Padre nuestro), etc. Similarly, corresponding to Spanish last names with a toponym preceded by de, earlier texts or forms fixed in songs often give sequences like Peru Abendañoko, Span. Pedro de Abendaño, Matxalen Busturiko ‘Magdalen of Busturia’; these, however, are not productive at present. While complement and adjunct PP s are prenominal in Basque and postnominal in Spanish, their position relative to each other is more flexible. The most neutral ordering is the expected one, namely adjunct-complement-head in

17

See Berro & Fernández (to appear). They show that these constructions may be found in titles or as complements of a predicative head, but not in the normal nominal functions such as subject, etc.

26

ortiz de urbina

Basque and its mirror image head-complement-adjunct in Spanish. Thus, in the absence of any contextual information, a phrase with a picture noun head like Spanish el retrato de Pedro de Miguel or Basque Mikelen Peruren irudia is interpreted as having Pedro/Peru as the person being portrayed and Miguel/Mikel as owner or portrayer (or any of the contextually relevant interpretations of genitive nouns). However, where function recoverability is contextually possible, other orders are also acceptable, if perhaps more marked. 2.2.1.3 Attributive Adjectives After examining PP (including genitive phrase) modification of nouns and determiners, let us turn now to the other major noun modifiers, namely, adjectives; relative clauses will be dealt with in the following subsection. As Dryer (1992) showed, adjectives do not enter in correlation pairs with nouns, whereas relative clauses do. Thus, the postnominal position of adjectives in Basque, identical to the position of restrictive adjectives in Spanish, should not be ‘unexpected’ from a typological perspective. If in Roman Aquitanian inscriptions like ummesahar ‘old child’, zahar ‘old’ is working as an adjective, as seems likely, the postnominal position for adjectives would be quite old, certainly not the result of recent convergence with Romance. Such order is found in N/Adj compounds too, in words like abelgorri ‘bovine cattle’ (lit. red animals), zorion ‘congratulation’ (lit. good bird), etc. Basque adjectives are much more robustly postnominal than Spanish ones, which, depending on the type and with semantic differences, may occur in the prenominal position: both especial interés and interés especial ‘special interest’ are possible, while only interes berezia (lit. interest special) is in Basque. In fact, according to Demonte (1999), only relational adjectives and adjectival perfective participles are found exclusively in a postnominal position (la distribución lingüística ‘the linguistic distribution’ but *la lingüística distribución, or la antena reparada ‘the repaired antenna’ but *la reparada antena). In contrast, very few adjectives can occur preposed in Basque (see Trask 2003: 138–139 for a list); most prominently, adjectives of origin with the suffix -(t)ar (bizkaitar abizena ‘Biscayne last name’, frantziar jendea ‘French people’, etc.) or adjectives formed with the suffix -dun (akin to English -ed, as bizardun gizona ‘bearded man’, euskaldun jendea ‘Basque (language speaker) people’). Even with these, the postposed position is far more common.18 The fact that preposabil18

Euskaltzaindia (1985: 243) points out that preposed adjectives can be found in literature, giving examples like justu manamenduak ‘fair commands’, venial faltak ‘venial sins’ or enganoso mundua ‘deceitful world’. The type of adjective and expression, hardly vernacular, may indicate this is/was a stylistic device for elevated register.

word order

27

ity hinges on these specific suffixes, rather than on a general semantic type, for instance, indicates the positional properties may stem from properties of the affixes themselves. Be that as it may, the restrictive versus non-restrictive interpretations obtained in well-known postnominal and prenominal orderings of adjectives in Spanish phrases like nieve blanca/blanca nieve ‘white snow’ or incas valientes/valientes incas ‘bold incas’ does not have any structural correlate in Basque, where both interpretations can be associated with the basic, postnominal position of the adjective: lagun maiteak ‘dear friends’ receives an unmarked non-restrictive interpretation (probably for pragmatic reasons), while lagun minak ‘close friends’ receives a restrictive interpretation. We will briefly dwell on the three types of adjectives which may be preposed in Basque in the remainder of this subsection. Beginning with -(t)ar derived items, de Rijk (2008: 127) explicitly treats them as nouns: “Nouns derived with this suffix freely combine with other animate nouns, acting as a first or as a second component of a bipartite compound”, exemplifying with erromatar soldadua/soldadu erromatarra ‘the Roman soldier’. Correspondingly, he translates mundutar ‘earthly’ as earthling, zerutar ‘celestial, heavenly’ as ‘celestial being’ or ezkertiar as ‘leftist, left-winger’. But at least some -(t)ar derivatives seem to be used as bona fide adjectives, sharing nonetheless the positional properties of better-behaved items: soñu eztitsu ta zerutarra ‘a sweet and celestial tune’ (Bilbao, Ipuin Barreka 158), zerutar ta jainkotiar gauzetaz ‘on celestial and devout things’ (Gazt MusIx 166).19 Even if most -(t)ar derivatives were actually nouns and we were dealing with compounds, however, we would probably be transferring the problem to morphology: erromatar soldadua ‘Roman soldier’ may be equivalent to Erroma soldadua ‘Rome soldier’, a right-headed compound or an apposition, but it is more difficult to analyze soldadu erromatarra, the alternative order, as headed by a noun erromatarra: this expression corresponds to Spanish el soldado romano, rather than something like appositive el romano soldado. It is of course possible that some of these derived words have become adjectives, but that only nouns can be found in prenominal position, so the basic line of de Rijk’s analysis may perhaps still be maintained. Basque grammars (Euskaltzaindia 1985: 244; Trask 2003: 138) often mention that loaned adjectives of origin like frantses ‘French’ share with the native

19

These are literary uses, where the derived item has a qualifying function; the preposed order here may be taking a free ride on the preposing capacity of -(t)ar forms, producing a pattern similar to preposed qualifying adjectives in Spanish. See below for a short discussion on qualifying and classifying uses of preposed adjectives.

28

ortiz de urbina

adjectives built on -(t)ar like frantziar the ability to appear preposed, so that both frantses/frantziar mutilak ‘French youths’ are acceptable, as well as postposed mutil frantsesak/frantziarrak. But, leaving aside the preference for postposed orders, this is far from a general fact: many adjectives do not sound felicitous in prenominal position (??aleman/*suizo/*espainol mutilak ‘German/ Swiss/Spanish youths’), and frantses is by far the most common adjective of this group. Like nafar ‘Navarrese’ (not a loanword), these are found in some lexicalized expressions like frantses belar ‘alfalfa’ (lit. French grass), nafar hizkuntza ‘Navarrese language’ (i.e., Basque), or nafar-ilar ‘Navarrese pea, garbanzo bean’,20 where it is not clear the first element is always an adjective, rather than a noun.21 Certainly, euskal ‘Basque’ as in euskal dantzak ‘Basque dances’ is a nominal bound form of euskara ‘Basque language’, and as such cannot even be found following the noun.22 Turning now to the case of words formed with the suffix -dun, the standard analysis identifies it as a reduced form of the finite relative verb duen ‘that has …’, so that a form like euskaldun jendea ‘Basque (speaking) people’ would actually be something like ‘people who have the Basque language’. The prenominal order would follow directly from the fact that relative clauses are prenominal in Basque. Postnominal orders would then have to be explained, probably as analogy with normal adjectives. As de Rijk (2008:503) points out, however, there is little evidence for a finite form doubling as a suffix, at least in Basque, while there is solid evidence for verbs as second elements of exocentric compounds being grammaticalized as derivational suffixes. This would be the case of -gin ‘-er’ ( Sp cantaré ‘I will sing’ (see, e.g., Fleischman 1982). The transitional phase can be observed in the following examples from El Cid: doblarvos he la soldada ‘I will double your salary’ (ModSp os doblaré el sueldo); El Campeador dexar las ha en vuestra mano ‘The Campeador will leave them in your hand’ (ModSp las dejará en vuestra mano) The same construction, but with the past imperfect of ‘have’ gave rise to the Romance conditional; e.g.: Lat cantāre habēbam ‘I had to sing’ > OSp cantar ia > Sp cantaría ‘I would sing’. With these Romance diachronic facts in mind, let us now consider the structure of verbal forms in Basque. 2.2 Basic Analytical Forms in Basque Leaving aside a handful of verbs that possess synthetic forms in a few tenses, conjugating a verb in Basque always involves combining a nonfinite form of the verb with an auxiliary. A respect in which Basque differs from Romance is the rarity of synthetic forms. Whereas, in Romance, like in English, tenses may be simple (e.g. Sp compro ‘I buy’, compré ‘I bought’) or compound (e.g. he comprado ‘I have bought’, había comprado ‘I had bought’), all tenses in Basque are compound tenses, again with the exception of a very small number of verbs, which have some synthetic forms in present-day Basque (see section 4).

62

hualde and mounole

table 3.1

Non-finite forms

Perfective part. Imperf. part. Prospective part. Radical

erosi erosten erosiko eros

erori erortzen eroriko eror

hartu hartzen hartuko har

eman ematen emango, emanen eman

The nonfinite forms used in the conjugation of verbs include the perfective, imperfective and prospective participles, in addition to the radical. The perfective participle is the most basic form of the verb in the sense that, knowing the perfective participle, it is possible to derive all other forms. The nonfinite forms of erosi ‘to buy’, erori ‘to fall’, eman ‘to give’ and hartu ‘to take’ are shown in Table 3.1. The prospective participle bears a suffix -ko that can be identified, at least diachronically, with the relational suffix, which, among other functions, serves to form the locative genitive as in mendi-ko ‘of the mountain’. In the standard language, this suffix -ko competes with -en for verbs whose perfective participle ends in -n. In northeastern dialects, however, the suffix -(r)en is also used with other verbs; e.g. harturen, erosiren instead of hartuko, erosiko. From a diachronic perspective, the latter is the same suffix that forms the regular genitive when attached to noun phrases, e.g. mutil-a-ren ‘of the boy’, mutil-en ‘of (the) boys’. Both suffixes -ko and -en may have benefactive or destinative value. Although these usages of these suffixes are restricted in the modern language (e.g. bihar-ko ‘for tomorrow’), they were prevalent in earlier centuries, which may explain their use in the prospective participle (Mounole 2011: 312–315). Consider the examples in (1), from B. Etxepare (1545):1 (1) a. amoretan plazer baten mila dira dolore love.loc pleasure on.gen 1000 are pain ‘In love, for one pleasure there are a thousand pains’ b. Guiçonaren prouechuco emaztea bethi da man.gen profit.rel wife always is ‘A wife is always for the benefit of man’ (Etxepare II, 10)

1 Abbreviations used in the glosses in this chapter: loc = locative, gen = genitive, rel = relational.

tense, aspect and mood table 3.2

63

Basic indicative tenses (transitive conjugation)

erosi dut erosi nuen erosten dut erosten nuen erosiko dut erosiko nuen

‘I have bought’ ‘I bought’ ‘I buy’ ‘I was buying; I used to buy’ ‘I will buy’ ‘I would buy’

The imperfective participle is diachronically an inflected form of the verbal noun, with the addition of the locative suffix -n (as has long been noticed, see Aldai 2002: 109–110). The verbal noun is created by adding -tze or -te to the radical; e.g. hartze ‘act of taking’. In the standard language, the allomorph -te is used with verbs whose radical ends in a sibilant or a nasal; in some dialects, this allomorph, which is diachronically prior (see Urgell 2007), has a wider distribution. In its use, the verbal noun corresponds, roughly, to the Romance infinitive; e.g. asko hartze-a ‘to take a lot’, cf. Sp ‘el tomar mucho’. A development in Western and Central dialects has been the replacement of the radical with the perfective participle in all contexts, so that in these dialects the radical is now only found in a few fixed expressions (see Mounole 2007 for a detailed diachronic study). The basic indicative forms of a verb not possessing synthetic forms arise from the combination of one of the participles in Table 3.1 with the present or past tense of an auxiliary. In Table 3.2 we illustrate with the first person singular indicative forms of erosi ‘to buy’, which include all possible combinations of the three participles with the auxiliaries dut ‘I have’ and nuen ‘I had’. As mentioned, unlike in present-day Spanish or English, but like in earlier stages of these languages and in modern Italian, French and German, there are two auxiliaries, ‘have’ and ‘be’. All transitive verbs take the ‘have’ auxiliary. On the other hand, some intransitive verbs take the ‘be’ auxiliary and others the ‘have’ auxiliary, a matter to which we will return. Table 3.3 shows the same forms as in Table 3.2, but for erori ‘to fall’. The forms of the auxiliary that are shown, when used alone, are the copulas naiz ‘I am’ and nintzen ‘I was’. This system of indicative forms is the result of relatively recent processes of grammaticalization. As Mounole (2011) argues, the earliest Basque texts, from the 16th century, appear to show the transition from a very different older system where synthetic forms were more prevalent, and these were opposed to analytical perfective forms based on the radical (which have now

64 table 3.3

hualde and mounole Basic indicative tenses (intransitive conjugation)

erori naiz erori nintzen erortzen naiz erortzen nintzen eroriko naiz eroriko nintzen

‘I have fallen’ ‘I fell’ ‘I fall’ ‘I was falling; I used to fall’ ‘I will fall’ ‘I would fall’

been confined to the subjunctive and other non-indicative moods) and other periphrases, now obsolete, which we will consider below (see section 5.4). Given this, it seems sensible to consider the possible role of language contact in the development of the system of forms illustrated in Tables 3.2 and 3.3. The present perfect form erosi dut ‘I have bought’ has the same structure as in the Romance and Germanic languages, as it combines a perfective participle with the present tense of the auxiliary. In both Peninsular Spanish and Basque, the present perfect, in addition to its perfect meaning, functions as a hodiernal or recent past; e.g. Esta mañana he visto a Mikel/ Gaur goizean Mikel ikusi dut ‘I saw Mikel this morning’ (vs Ayer vi a Mikel/Atzo Mikel ikusi nuen ‘Yesterday I saw Mikel’). In dialects spoken in France, this form can also be used as a non-hodiernal past. Although the latter development mirrors that in French, it appears that it is due to an independent evolution, as Mounole (2011) argues, since the widening of meaning took place in those Basque varieties in the 17th century, before it did in French, and, furthermore, the contact Romance language at the time was not French but Gascon, which did not undergo this change. On the other hand, notice that erosi nuen, which combines the perfective participle with the past of ‘have’ does not have a pluperfect meaning, ‘I had bought’ (Sp había comprado), but, rather it is a simple perfective nonhodiernal or remote past ‘I bought’. Both tenses with the perfective participle must have developed from resultative structures through the same grammaticalization path as we find in Romance and Germanic (see Aldai 2001; Mounole 2011). Nevertheless, some structural differences should be noted: in Basque there is no singular/plural agreement between participle and direct object, not even in the earliest texts (from the 15th–16th centuries), when the structure was clearly biclausal and had a resultative reading (Obviously, there cannot be gender agreement, as Basque lacks this morphological category). The explanation would be that, when the

tense, aspect and mood

65

structure arose, the use of the definite article had not yet been extended to predicative constructions (morphological plurality developed from the grammaticalization as articles of singular and plural demonstratives, Manterola 2015). In fact, in 16th century texts, the form we are exemplifying with erosi nuen could function both as a perfective past and as a pluperfect (Mounole 2011: 55). The change in meaning from pluperfect (‘I had bought’) to perfective past (‘I bought’) is thus a recent innovation, without a direct parallel in Romance, although along a well-attested grammaticalization path (Aldai 2002: 2017–218): pluperfect > pluperfect & perfective nonhodiernal past > perfective nonhodiernal past; cf. also the evolution of the Latin pluperfect indicative to past subjunctive in Spanish, e.g. Lat amāveram ‘I had loved’ > Sp amara ‘I loved, subjunctive’. In Old Basque we also find a structure erosi duket ‘I will have bought’, where the auxiliary bears the suffix -ke. This form has a future perfect interpretation and may also express probability (see section 3.4). Nowadays, to express a pluperfect meaning (i.e. a past of the past), as opposed to a remote past, Basque makes use of a “double-compound past perfect” (Aldai 2002), as in erosi izan nuen ‘I had bought’, where izan is the radical and past participle of ‘be’ (there is also a parallel double-compound present perfect, erosi izan dut ‘I have bought’, which usually expresses the experiential perfect). Double-compound forms have a formal parallel in the formes surcomposées of French and Occitan (Mounole 2011: 47). A meaning close to that of the pluperfect can also be expressed by means of a resultative construction, erosia nuen ~ erosita neukan ‘I had it bought’ vs erosia dut ~ erosita daukat ‘I have it (in the state of being) bought’, which also has a parallel in Spanish (e.g. las tengo compradas ‘I have bought them, fem.’). The forms with the prospective participle, erosiko dut ‘I will buy’, erosiko nuen ‘I would buy’, also appear to share a diachronic origin with the corresponding tenses in Spanish and most other Romance languages (Michelena 1981). The dialectal alternation between suffixes in Basque, e.g. erosiko dut ~ erosiren dut ‘I will buy’ supports this identification between the Romance and Basque structures to a certain extent. There is a more exact parallel with a construction of obligation in Spanish with the preposition de ‘of’, e.g. Sp he de comprar ‘I have to buy’ (although this Spanish construction retains a meaning of obligation and has not been grammaticalized as a future). Nevertheless, notice that the Basque prospective participle is built on the perfective participle. A literal Spanish translation of erosiko dut would be *he de comprado. The parallel, thus, is not perfect. It is more likely that the grammaticalization of the suffixes -ko and -en in the prospective participle actually arose from their use in phrases with benefactive or destinative value (Mounole 2011: 312–315). Struc-

66 table 3.4

hualde and mounole Modern Spanish simple and compound TAM forms and their Basque counterparts

Present Past imperf. Past perfective Future Conditional Present perfect Pluperfect

compro compraba compré compraré compraría he comprado había comprado (hube comprado) Future perfect habré comprado Conditional perf habría comprado

erosten dut erosten nuen erosi nuen erosiko dut erosiko nuen erosi dut erosi izan nuen

‘I buy’ ‘I bought’ ‘I bought’ ‘I will buy’ ‘I would buy’ ‘I (have) bought’ ‘I had bought’

erosi izango dut ‘I will have bought’ erosi izango nuen ‘I would have bought’

tures like, e.g., bihar-ko dut ‘I have (to do) it for tomorrow’ may have provided the original model for erosi-ko dut. The Basque forms with the imperfective participle do not have a Spanish parallel. The fact that the imperfective participle incorporates a locative suffix suggests a similarity with, e.g., Eng I am on buying > I’m a-buying. In this respect, we may note that Gascon has a similar progressive construction with ‘be’ and the infinitive preceded by the preposition a/en, e.g. Gasc èste a fer quaucòm ‘s/he is doing something’ (Haase 1992: 93), (cf. also Port estou a comprar ‘I am buying’). If the Basque present indicative form is the result of a similar process of grammaticalization starting from a progressive construction, it may originally have taken the ‘be’ auxiliary with all verbs, with a later shift in auxiliary for transitive and unergative verbs, following the choice of auxiliary in other tenses (erortzen naiz ‘I am falling, I fall’ and also *erosten naiz > erosten dut ‘I am buying, I buy’) (see Mounole 2008, 2011). Western Basque also has a periphrasis with egon ‘be, stay’ for all verbs (e.g. erosten nago ‘I am buying’) that more closely resembles the Spanish progressive (e.g. estoy comprando). In Table 3.4, a comparison of modern Spanish and Basque indicative forms is offered, arranged according to Spanish morphology, with simple forms before compound forms. As can be observed, in spite of the diachronic parallels noted above, in the present-day languages the only forms that show agreement in their morphology are those of the present perfect. Table 3.5 offers the same comparison between modern Basque and Spanish forms, but using the basic forms of the indicative in Basque (excluding doublecompound forms) as point of comparison and pointing out parallel evolutions. Forms that are identical in structure and meaning are indicated with an equal

tense, aspect and mood table 3.5

67

Modern Basque TAM forms and their Spanish formal counterparts

erosi dut ‘I (have) bought’ = he comprado erosi nuen ‘I bought’ (perfective past) ≈ había comprado ‘I had bought’ (pluperfect) erosiko dut ‘I will buy’ ≈ compraré ‘I will buy’ (< comparāre habeō) he de comprar ‘I have to buy’ erosiko nuen ‘I would buy’ ≈ compraría ‘I would buy’ (< comparāre habēbam) erosten dut ‘I buy’ No formal equivalent erosten nuen ‘I bought’ (imperfect) No formal equivalent

sign, whereas the “approximately” sign (≈) is used for less perfect equivalencies (with the caveats noted in the text above). To conclude this section, the most striking difference between Basque and Romance is the complete lack of synthetic forms in Basque for most verbs, so that even for the present tense a compound form must necessarily be used. Also strikingly different is the consistent choice in Basque between the ‘be’ and the ‘have’ auxiliary, depending on the semantics of the verb, across all compound tenses. When the contemporary languages are considered, a comparison between Basque and Romance indicative forms reveals an exact match only in the structure of the present perfect. When the diachronic evolution of the compound forms of Basque is also considered, other similarities with Romance developments become apparent, including in the form of the perfective past, whose structure mirrors the Romance pluperfect, and, to some extent, the future and the future of the past or conditional. Regarding the distinctions that are expressed morphologically, we notice the existence of present, past and future forms, with a further contrast in both Romance and Basque between perfective and imperfective in the past. This restriction of the contrast between perfective and imperfective to the past tense is cross-linguistically common (Dahl 1985; Bybee et al. 1994:83).

3

Other Moods

3.1 Subjunctive In both Basque and Romance, we find a contrast in mood between indicative and subjunctive in subordinate clauses. In Basque, the subjunctive is formed with the radical and a present or past form of a distinct auxiliary verb. As in the

68 table 3.6

hualde and mounole Subjunctive forms in Basque

Present subjunctive eros dezadan eror nadin Past subjunctive eros nezan Past subjunctive eror nendin

compre caiga comprara cayera

‘(that) I buy’ ‘(that) I fall’ ‘(that) I bought’ ‘(that) I fell’

indicative forms considered in the previous section, the auxiliary is different for transitive and unergative verbs, on the one hand, and unaccusative verbs, on the other. The auxiliaries used in the subjunctive do not have participial forms and are reconstructed as *edin (intransitive) and *ezan (transitive) (in Bizkaian or Western Basque, the non-indicative transitive auxiliary is egin ‘do’, instead). Examples are given in Table 3.6, together with their corresponding forms in Spanish for comparison. Whereas in the Romance languages, the existence of a separate subjunctive mood is inherited from Latin, in Basque, the subjunctive has arisen as an independent mood only in very recent historical times—between the 15th and 18th centuries—as a consequence of the restriction of forms containing the radical and *edin/*ezan to non-assertive contexts (Mounole 2011: 335). In 16th century texts, forms like eros nezan ‘I bought’, which are nowadays used only in certain types of subordinate clauses, appear also in main clauses with the value of a simple perfective past; e.g. jaigi zidi nagia, erra zizan uria (Refranes y Sentencias 1596: 294) ‘the loafer got up and burned the town’. Conversely, forms that nowadays only have an indicative value could be used in contexts from which they are now excluded, such as to express wishes, etc.; e.g. Jainkoak hala nahi duela ‘May God will it thus’ (Axular, 1643), cf. modern Bq Jainkoak hala nahi dezala. In the past subjunctive there is no perfective/imperfective contrast in either Romance or Basque. In Spanish, for instance, the contrast between María cantó muy bien ‘M. sang very well (perfective)’ and María cantaba muy bien ‘M. sang (imperfective)/was singing/used to sing very well’ is neutralized if the clause is embedded under a matrix verb requiring the subjunctive mood in the subordinate clause: No creo que María cantara muy bien ‘I don’t think that María sang (perfective or imperfective) very well’. In Basque, the fact that the subjunctive makes use of the radical precludes the possibility of an aspectual contrast. (Although, as mentioned, in Western and Central dialects the radical has been replaced by the perfective participle in all contexts). In Basque, the subjunctive is used less extensively than in Romance, as uninflected clauses with the verbal noun can be used as an alternative to the sub-

tense, aspect and mood

69

junctive in contexts where that is not an option in Romance, such as indirect commands and expressions of desire; e.g. Sp quiero que lo compres, Bq zuk eros dezazun nahi dut ~ zuk erostea nahi dut ‘I want you to buy it’. In addition, the subjunctive is not used to express doubt, unlike in Romance; e.g. Sp no creo que lo compre, Bq ez dut uste erosiko duenik ‘I don’t think s/he will buy it (present subjunctive in Sp vs future indicative in Bq)’. French and Spanish do not use the subjunctive in exactly the same contexts. For instance, in temporal subordinate clauses with future reference, Spanish uses the present subjunctive and French the future indicative, e.g. Sp cuando lo veas, Fr. quand tu le verras ‘when you see it’. In Basque the present indicative is the preferred option in this context, although the subjunctive is sometimes used in writing, Bq ikusten duzunean (~ ikus dezazunean) ‘when you see it’. In sum, the subjunctive is a very recently grammaticalized mood in Basque, with a more restricted usage than in Spanish or French. 3.2 Imperative, Exhortative and Jussive In exhortatives or first person plural commands, subjunctive forms are used in both Basque and Romance: Bq eros dezagun, Sp comprémoslo ‘let’s buy it!’; Bq ez gaitezen eror, Sp no caigamos ‘Let’s not fall!’ Third person commands or jussives are expressed in Basque either by means of subjunctive forms with the complementizer suffix -la ‘that’ or with forms obtained from the same auxiliary as in the subjunctive, but with the prefix b: eros dezala ~ eros beza ‘let him/her buy it’, eror dadila ~ eror bedi ‘let him/her fall’. In Spanish and French subjunctive forms are used for this purpose. True (second person) commands also show some affinity with the subjunctive in both Basque and Romance, but the details are not identical. In Basque there are distinct imperative forms, which are used in both positive and negative commands; e.g. eros ezazu! ‘buy it!’, ez ezazu eros! ‘do not buy it’, except that for verbs that have synthetic imperative forms these can only be used in positive commands. In Eastern and Central varieties, subjunctive forms are nevertheless preferred in negative commands, ez dezazula eros ‘do not buy it!’. In Spanish and Occitan (including Gascon), distinct imperative forms are only used in positive (informal) commands, whereas negative commands are expressed by the subjunctive: Sp compra ‘buy!’, no compres ‘do not buy!’, Gasc non tornes pas tà casa ‘Do not go back home!’ (Birabent & Salles-Loustau 1989: 86). 3.3 Conditional In Spanish and French the morphological conditional is used in the apodosis of hypothetical conditional sentences, whereas past subjunctive or past

70 table 3.7

hualde and mounole Basque potential forms

Present potential Past potential Hypothetical potential

eros dezaket eror naiteke eros nezakeen eror nintekeen eros nezake eror ninteke

puedo comprar puedo caer podía comprar podía caer podría comprar podría caer

I can buy I can fall I could buy I could fall I could buy I could fall

imperfective indicative forms are used in the protasis: Sp si lo encontrara, lo compraría / Fr si je le trouvais, je l’acheterai ‘If I found it, I would buy it’. The morphological conditional is also used as a future of the past, in indirect speech: Sp dije que lo compraría ‘I said I would buy it’. There is a corresponding compound form used in counterfactuals or failed conditions: Sp si lo hubiera encontrado, lo habría comprado ‘If I had found it, I would have bought it’. Basque, on the other hand, possesses special conditional forms in both the protasis and the apodosis of conditional clauses: ikusiko banu, erosiko nuke ‘If I saw it, I would buy it’ (or, in Northeastern varieties, ikusten banu, erosiko nuke). In principle this allows for a distinction with the future of the past: esan nuen erosiko nuela ‘I said I would buy it’ (non-conditional, reporting ‘I said: I will buy it’) vs esan nuen erosiko nukeela ‘I said I would buy it’ (conditional, if some condition were to be met). In practice, this difference is not always observed and the future of the past may appear in conditional sentences. Morphologically, the Basque conditional auxiliary forms are related to past indicative forms cf, e.g., nu-en ‘I had’, ba-nu ‘if I had’, nu-ke ‘I would have’; nintzen ‘I was’, ba-nintz ‘If I were’, nintza-teke ‘I would be’. 3.4 Potential An aspect in which the Basque TAM system differs markedly from Romance is in the existence of a set of potential forms. Potential forms bear a suffix -ke or -teke and are based on the same auxiliary as in the subjunctive. In Romance potential forms are periphrastic constructions with ‘be able’ and the infinitive. The suffix -(te)ke of potential forms is clearly the same suffix that is found in the conditional apodosis. In older texts and (rarely) in present-day literary usage, this suffix may also attach to the indicative auxiliary to indicate either future time or probability, e.g. ikusten dugu ‘we see’, ikusten dukegu ‘we may see, we will see’ (also ikusi dukegu).

tense, aspect and mood

71

Basque can also express the potential in periphrases with ahal ‘to be able’ and ezin ‘not to be able’; e.g. erosi ahal dut ~ erosten ahal dut ‘I can buy it’ (with the perfective or the imperfective participle depending on the dialect, see section 6).

4

Synthetic Forms in Basque

As has been mentioned, a handful of Basque verbs deviate from the patterns illustrated above in that, besides compound or analytical forms, they also possess synthetic forms. In present-day Basque, in addition to the two auxiliaries of the indicative, which also function as main verbs with the meanings of ‘have’ and ‘be’, the following verbs have synthetic forms for most speakers: a) With intransitive morphology: joan ‘go’, etorri ‘come’, ibili ‘walk’ b) With transitive morphology: eduki ‘have’, jakin ‘know’, ekarri ‘bring’, eraman ‘take’, *io(n) ‘give’ A few other verbs have a more restricted use as synthetic, limited sometimes to a few forms, e.g. erion ‘flow’, jardun ‘be engaged in’, iraun ‘last, endure’, etzan ‘lie’. These verbs possess a synthetic present and past indicative and, in some cases, also conditional, potential and imperative forms; e.g. etorri ‘come’: dator ‘s/he comes, is coming’, zetorren ‘s/he came, was coming’, baletor ‘if s/he came’, letorke ‘s/he would come’, betor ‘may s/he come’, zetorkeen ‘s/he could come’, zatoz ‘come!’; jakin ‘know’: dakit ‘I know’, nekien ‘I knew, used to know’, baneki ‘if I knew’, nekike ‘I would know’. Referring to the contrast between synthetic and compound forms, as in dator ‘s/he is coming (continuous aspect)’ vs etortzen da ‘s/he comes (generally or habitually’), Trask (1997:212) writes: “What is striking is that, in contrast to nearly all other European languages which have developed such an aspectual contrast, in Basque it is the synthetic form which is marked for continuous aspect […] and this in spite of the fact that the periphrastic form is etymologically transparent as involving the locative of a verbal noun”. The online Basque grammar Sareko Euskal Gramatika lists 26 verbs with synthetic forms, but most of the forms listed are not commonly used by a majority of speakers. In the oldest texts, from the 16th and 17th centuries, we find a much larger number of verbs with synthetic forms and also additional synthetic TAM forms (Lafon 1944; Mounole 2011). The progressive loss of synthetic forms does not seem to be directly attributable to contact with Romance. Influence from Romance should have resulted instead in a generalization of synthetic forms for all verbs. What we see in this development is, instead, the application of a universal tendency to replace synthetic forms with periphrases. This tendency

72

hualde and mounole

operated in Romance, for instance, in the loss of the Latin synthetic passive and the synthetic future, as well as the recent loss of the passé simple in spoken French. As other researchers have noticed, however, all verbs that are or have been synthetically conjugated in historical Basque belong to the class whose participial forms and radical take the prefix e- (Lakarra 2017:77), so that the extension of the synthetic conjugation to other classes of verbs could not easily follow an established pattern.

5

Less Grammaticalized Constructions (Periphrases)

5.1 Progressive Constructions In Spanish the simple present and imperfective past may be used to convey progressive aspect. In addition, there is a progressive structure with estar ‘be, stay’ and the present participle, very much like in English; e.g. estoy comprando ‘I am buying’, estaba comprando ‘I was buying’. French, on the other hand, lacks progressive tenses and uses instead a periphrasis with être en train ‘to be in the process of’; e.g. je suis/étais en train d’acheter ‘I am/was buying’. Western Basque (Bizkaian) has a progressive construction with egon ‘be, stay’ and the perfect participle that mirrors the Spanish construction. In this construction, the auxiliary is always egon. That is, we do not find a transitivityconditioned alternation in auxiliaries; e.g. erosten nago ‘I am buying’. Other dialects, instead, make use of a periphrasis with the perfective participle, the semi-auxiliary ari ‘to be engaged in’ and the ‘be’ auxiliary; e.g. erosten ari naiz ‘I am buying’. Aldai (2002: 135) suggests that the lack of auxiliary alternation in both of these Basque progressive constructions may be due to Romance influence. However, this may also be explained as a sign of the less grammaticalized character of these constructions. As mentioned above (and as suggested by Mounole 2011), the present indicative as well (erosten dut ‘I buy’, erortzen naiz ‘I fall’) may have started as a construction with ‘be’ for all verbs, with a later introduction of the auxiliary alternation that developed in the present perfect because of the originally resultative source of this tense. In fact, the possibility of having auxiliary alternation with ari ‘to be engaged in’ appears to have arisen in some dialects, starting in the second half of the 19th century, indicating a further step in the grammaticalization process (Mounole 2008, 2010).

tense, aspect and mood

73

5.2 Resultative Besides the compound present perfect, Spanish has developed a new resultative construction with the new ‘have’ verb tener (as mentioned, haber has been lost in Spanish as a main verb indicating possession and only functions as an auxiliary and in the ‘there be’ construction). In this resultative construction, the participle agrees with the direct object in gender and number, cf he escrito las cartas ‘I have written the letters’ vs tengo escritas las cartas ‘I have the letters written’ (see Harre 1991). Basque also has a parallel construction, which differs from the present perfect in that the participle takes the article -a(k), cf gutunak idatzi ditut ‘I have written the letter’ vs gutunak idatziak ditut ‘I have the letters written’. In addition, Basque possesses an experiential construction of identical structure, with an inflected perfective participle, but with the ‘be’ auxiliary. This experiential construction can be used with both intransitive and transitive verbs, e.g. Parisen egona naiz ‘I have been in Paris’, ‘I have had the experience of being in Paris’ (vs Parisen egon naiz ‘I have been in Paris’), gutun asko idatzia naiz ‘I have written many letters’ (e.g. ‘I have the experience of having written many letters’), gorriak ikusia naiz ‘I’ve had a hard time’, lit. ‘I am (having) seen them red’ (Villasante 1980: 152; Zabala 2003: 431). Sentences with an inflected participle are perhaps to be analyzed as biclausal constructions, with a larger domain of applicability, beyond their resultative and experiential functions (Ortiz de Urbina & Uribe-Etxebarria 1991). Basque surpasses Romance in the flexibility of participial predicates. With the same meaning and functions, Basque also has other structures where the past participle bears the suffix -ta (probably from the conjunction eta ‘and’, Krajewska 2013) or -rik (partitive, probably form an earlier ablative usage of this suffix, Mounole 2008: 291–292) and inflected forms of egon ‘be, stay’ and eduki ‘have’, e.g. nekatuta/nekaturik nago ‘I am tired’, ikusita/ikusirik daukat ‘I have seen it’. 5.3 Must, Need, Want To express the notions of ‘must’ or ‘need’, Basque has developed a construction involving the noun behar ‘need, obligation’ in its uninflected form and ‘have’; e.g. telefono bat behar dut ‘I need a telephone’. This construction can combine with a participial clause; e.g. telefono bat erosi behar dut ‘I need to/must buy a telephone’; etxera joan behar dut ‘I must go home’. As it is crosslinguistically common for obligation expressions, the construction with behar has also developed the meaning of an immediate future (Bybee et al. 1994: 258–259), e.g. Tokatzen baldin bazait aurten loteria, erosi beharko det Citroën berria (song by X. Lete) ‘If I win the lottery this year, I am going to buy a new Citroën’.

74

hualde and mounole

The closest Romance parallel is offered by the avoir besoin de ‘need, must, have need of’ construction in French: j’ai besoin d’un téléphone ‘I need a telephone’, j’ai besoin d’acheter un téléphone ‘I need to/must buy a telephone’ (or It avere bisogno). Notice, however, that in Basque the object of behar appears in the absolutive case, not in the genitive, showing a higher degree of grammaticalization as a verbal construction than in French. In fact, behar ‘need’ can also be inflected and be used as a regular noun, in which case it takes a genitive complement: telefono baten (erosteko) beharra dut ‘I have the need of (buying) a telephone’. In present-day Basque, then, we must distinguish between the verbal construction behar + ‘have’ and the noun behar ‘obligation, need, work’. Another indication of a high degree of grammaticalization of the verbal construction behar + ‘have’ is that in this construction the word behar can take the ending -ko of the prospective participle directly. As an alternative, however, the prospective participle of izan may also be added: telefonoa erosi beharko nuke ~ telefonoa erosi behar izango nuke ‘I should buy the telephone’. On the other hand, in this construction, behar cannot bear the imperfective suffix (irrelevantly, the denominal verb behartu ‘oblige, force’ conjugates like a regular verb). Both Spanish and French make use of deber/devoir as a modal verb with deontic (obligation) and epistemic (probability) value, like English must, should: Sp debes comprarlo, Fr tu dois l’acheter ‘you must buy’, Sp debe haber caído, Fr il doit être tombé ‘It must have fallen’. In Spanish, to express epistemic meaning this construction competes with the future (possibility in the present) and conditional (possibility in the past). As already mentioned, Basque also has this usage. To express obligation or need, French and Spanish make use of some additional constructions. French possesses a construction with falloir ‘be necessary’ used as an impersonal verb with a subjunctive complement, e.g. il faut que je l’achète ‘I must buy it’. This construction has no parallel in present-day Spanish or in Basque. On its part, Spanish has constructions with haber de + infinitive and tener que + infinitive (e.g. he de comprarlo ~ tengo que comprarlo ‘I must buy it’). Above we mentioned the similarity between the haber de construction of Spanish and the Basque future. An older Spanish construction with caler ‘need’ (lit. ‘to be hot’) is now obsolete in this language (but alive in JudeoSpanish and Catalan). With a parallel structure to the behar ‘need’ + ‘have’ construction, Basque possesses another construction with nahi expressing the notion of ‘want’. This construction has resulted from the grammaticalization of the uninflected noun nahi ‘desire’ with ‘have’. In some Western varieties gura (< Lat gula) is used instead of nahi. Whereas in standard Basque behar and nahi always take the ‘have’ auxiliary, a further step in the grammaticalization process, found in some

tense, aspect and mood

75

local varieties, is an alternation between ‘have’ and ‘be’ depending on the main verb; e.g. joan nahi ba duzu > joan nahi bazara ‘If you want to go’, etorri behar du > etorri behar da ‘s/he must come’ (see Mounole 2010). Here as well we see that in its tendency to expand the lexically-determined alternation between ‘have’ and ‘be’ across verbal forms, Basque has been moving in the opposite direction from Romance. 5.4 Other Periphrases Basque and Spanish share several other periphrases showing parallel grammaticalization patterns. One of them is the frequentative periphrasis with ‘walk’, Sp andar, Bq ibili: Sp Mikel anda aprendiendo inglés/ Bq Mikel ingelesa ikasten dabil ‘Mikel is engaged in learning English’. Another common, but more restricted, shared periphrasis employs ‘carry’ (Sp llevar, Bq eraman) with a time expression to indicate the duration of an activity: Sp Mikel lleva dos años aprendiendo inglés / Bq Mikelek bi urte daramatza ingelesa ikasten ‘Mikel has been learning English for two years’. In some Western Basque varieties, we find a habitual construction with jakin ‘to know’, e.g. Sanantolinétan nire lagúnak bádakídxe etórten Lekittora ‘By St Antolin’s day my friends usually come to Lekeitio’ (lit. ‘they know how to come’) (Hualde, Elordieta & Elordieta 1994: 145). A parallel construction is found in the Spanish of Ecuador and some other South American countries, but it is no longer found in Peninsular Spanish, cf., e.g. No sabe venir el gozo sin pensiones de pesares ‘Pleasure doesn’t usually come without a price in pain’ (Tirso de Molina 1629).2 In older Basque texts, we find another habitual construction with the perfective participle and joan ‘go’ or eraman ~ eroan ‘carry’, depending on the transitivity of the main verb (Lakarra 1996: 179–181, Mounole 2011), e.g. Oski gitxi urratu doa (Refranes y Sentencias 1596: 275) ‘few shoes tend to break apart’; Mundu honek anhitz jende enganatu darama ‘This world tends to deceive many people’. This construction, which in the Bizkaian area survived until the 19th century but seems to be now obsolete in Basque, has a possible Old Spanish parallel in expressions with ir ‘go’ and a past participle, although this Spanish construction has a resultative, rather than habitual, meaning: Sin Dios todo va perdido (González de Eslava 1569) ‘without God, everything is lost’; Sardina que gato lleva, galduda va (Marqués de Santillana 1454) ‘A sardine that is carried by a cat, is lost’; de manera que va vendido por junto el trigo a excesivos precios ‘So that wheat is sold together at excessive prices’ (Pedro de Valencia 1605); parte

2 Historical examples in Spanish are taken from the Royal Spanish Academy’s CORDE database.

76

hualde and mounole

del dicho pasadizo va cerrado con celosías (Luis Cabrera de Córdoba 1599–1644) ‘Part of that passageway is closed by lattices’. The less frequent corresponding Spanish expression with llevar ‘carry’ and a past participle also has a resultative meaning: como llevo declarado ‘As I have already declared’. The ‘go’ +infinitive construction, which, like in English, has given rise to a periphrasis with future value in Spanish, French and Portuguese (Sp voy a cantar = Fr je vais chanter = Port vou cantar ‘I am going to sing’) has also a parallel in Basque: kantatzera noa, where the verbal noun takes the allative suffix (but in some Basque varieties, kantatzen noa). Incidentally, this construction has been grammaticalized with perfective past value in Catalan, Cat vaig cantar ‘I sang’, and in some varieties of Gascon, Gasc son mari va revenir à l’ ostel ‘her husband went back to the hotel’ (Rohlfs 1977: 217–218).

6

Preverbal Particles in Basque and Romance Parallels

Basque uses a set of particles for modality and aspectual purposes. These particles are usually placed before the finite verb, i.e. before a synthetic verb or between main verb and auxiliary in analytical constructions. There is only one aspectual particle, ohi, which expresses habitual aspect and has been documented in all dialects since the earliest texts. In Western dialects it is used in the ‘perfective participle + izan/*edun construction’ and in Eastern dialects in the ‘imperfective participle + izan/*edun construction’, instead. The construction with ohi does not have the same status in both dialectal areas, due to different paths of grammaticalization. Whereas in Eastern dialects ohi is clearly an aspectual marker, in the Western area it behaves as a semiauxiliary (Ortiz de Urbina 2003: 321–323).3 The use of ohi is not necessary to express habituality. As mentioned above, Basque, like the Romance languages, can use imperfective forms to refer to habitual actions, e.g.: goizeko bederatzietan etortzen da/zen ~ goizeko bederatzietan etorri ohi da/zen ‘s-he usually comes/s-he used to come at nine in the morning’. The other preverbal particles of Basque belong to the sphere of modality. Among them, we find ahal ‘desiderative, epistemic modality, yes/no questions’, bide ‘apparently’, omen ‘hearsay information’ and ei ‘hearsay information’.

3 There are, for instance, syntactic differences in negative sentences; e.g. Eastern etortzen ohi da ‘s-he usually comes’→ ez ohi da etortzen ‘s-he usually does not come’, where ohi is fronted with the auxiliary and negative ez ‘not’, vs Western etorri ohi da → ez da etorri ohi.

tense, aspect and mood

77

Ahal is a desiderative and epistemic modality particle, e.g.: ez ahal da berandu etorriko ‘I hope s-he does not come late!’, gizon ederra egin ahal da ‘he must have become a handsome man’. As mentioned above, it is also used as a modal particle is potential forms, e.g.: etorri ahal da = etortzen ahal da = etorri (ahal) daiteke ‘s-he may/can come’. In addition it is used as an interrogative particle in Central dialects (spelled al), e.g.: etorriko al da ‘will s-he come?’. This last use is very recent, attested only since the 19th century, although it is easy to explain as a reinterpretation of the epistemic value. Bide expresses probability: etxean gelditu bide da ‘s-he may have stayed at home’, ez bide da etorriko ‘s-he may not come’. As was mentioned above, in the Romance languages epistemic modality is mainly expressed by a periphrasis of obligation (Fr devoir + infinitive: Il a dû bien grandir ‘he must have grown a lot’, Sp deber (de) + infinitive: Debe de haber crecido mucho) or by the future tense (Sp Estará enfermo ‘he may be sick’, Old Fr Il sera malade). In Basque varieties in contact with Spanish the future tense can be also be used to express probability in the present (and the future of the past to express probability in the past): gaixorik egongo/egonen da ‘s-he may be sick’. In varieties in contact with French, on the other hand, the periphrasis with behar izan (see section 5.3) has recently started to be used with this meaning: eri izan behar da ‘s-he may be sick’. Basque also displays particles for reporting hearsay information, omen in Central and Eastern dialects and ei in the Western area: eri omen da ‘s-he is apparently sick’, etorriko omen dira ‘it seems that they will come’. The Romance languages in contact do not have any grammatical elements with this function, although dizque (< dicen que ‘they say’) is used in some varieties of Latin American Spanish with a similar meaning, e.g.: dizque vendrán hoy ‘they are supposed to be coming today’, dizque lloverá mañana ‘apparently it will rain tomorrow’. In some varieties of French d’après may also have this function. As for the origin of the Basque preverbal particles, most of them clearly derive from nouns and are still used as such: bide ‘road’, ahal ‘power, ability’, omen ‘fame, reputation; rumor’. The origin of ei remains obscure. Gascon is the Romance language that employs the most preverbal particles and some authors have sought a historical connection between the Gascon and Basque particles (Allières 1992). However, the Gascon particles are far from having a modal function, as they are mainly “enunciative”. The best known if these particles is que, which is frequently employed preceding the verb in the main clause of affirmative sentences: que i a ua hont qui canta ‘there is a spring that sings’, qu’arriba ‘s/he is coming’ (Birabent & Salles-Loustau 1989: 72). This has undoubtedly originated from the conjunction que ‘that’ heading a subordinate clause introduced by a clause or an expression indicating certainty (‘I’m sure

78

hualde and mounole

that, the point is that; certainly, of course’). The main clause or the adverb must gradually have fallen out of use, and then this use of que was extended to all affirmative sentences (Rohlfs 1977: 206). In Spanish and Romanian as well, the particle que is very often employed for emphasizing the clause (Sp ¡que voy!), and for marking negation (Sp ¡que no quiero!, Rum ca no pot ‘I can’t’). Historical Basque has no particle of this kind. Nevertheless, Gómez & Sainz (1995) established a comparison between the 3rd person past prefix z- and the conjunction eze, proposing that originally it must have appeared as an enuntiative particle in narration, which was later reinterpreted as a past morpheme. Gascon displays a few more particles: be (exclamative particle, be parlas plan lo gascon! ‘you speak Gascon very well’), e (essentially interrogative particle and conjunction e sabes parlar gascon? ‘Do you speak Gascon?’) and ja (exclamative and affirmative particle, J’ac sabi ‘I know it’). The latter—also present in Spanish and Catalan—can be compared to the Basque prefix ba-. It is indeed the case that the Basque particle originates from bai ‘yes’. In Basque the use of this preverbal particle was extended and has become obligatory when the finite verb is located at the beginning of the clause (badakit nik, *dakit nik ‘I know (it)’, badator ‘s/he’s coming’) (Birabent & Salles-Loustau 1989: 73; Rohlfs 1977: 208–211).

7

Summary

In this paper we have undertaken a detailed comparison, from a diachronic perspective, of the Basque TAM system with that of the Romance languages with which it is in contact. We have pointed out that, whereas a number of relatively recent developments in Basque, including the development of a subjunctive mood and several periphrases, show convergence between Basque and its Romance neighbors, not every important development has been convergent. In particular, in Basque the diachronic tendency may have been towards strengthening the choice between transitive and intransitive auxiliary, depending on the semantics of the verb, while in Spanish an original ‘be’/ ‘have’ alternation in auxiliaries was later lost. Some apparent morphosyntactic similarities, such as the presence of preverbal particles in Basque and Gascon, and the morphological structure of the future tense, turn out to be less obvious when examined in some detail.

tense, aspect and mood

79

Acknowledgements The first author acknowledges the research funding for this project received from the Spanish Ministry of Economy and Competitiveness (FFI2014–51878P).

References Aldai, Gontzal. 2002. The grammaticalization of present and past in Basque. Ph.D. thesis, University of Southern California, Los Angeles, CA. Allières, Jacques. 1992. “Gascón y euskera: afinidades e interrelaciones lingüísticas”. ASJU 26(3). 801–912. Axular, Pedro. 1643. Gero. Bordeaux: G. Milanges. Edition and Spanish translation by L. Villasante, 1964, Barcelona: J. Flors; Critical edition by B. Urgell, 2015, Bilbao: Euskaltzaindia & Nafarroako Gobernua. Bec, Pierre. 1968. Les interférences linguistiques entre gascon et languedocien dans les parles du Comminges et du Couserans. Paris. Birabent Jean-Pierre, & Salles-Loustau Jean. 1989. Memento grammatical du gascon, Pau: Escòla Gaston Febus & Nosauts de Bigòrra. Bybee, Joan, Revere Parkins & William Pagliuca. 1994. The evolution of grammar: Tense, aspect and modality in the languages of the world. Chicago: University of Chicago Press. Creissels, Denis & Mounole, Céline. 2012. “Non-canonical valency patterns in Basque: variation and evolution”. 45th Annual Meeting of the Societas Linguistica Europaea 29 August–1 September 2012, Workshop Contrastive studies of verbal valency in European languages. Dahl, Östen. 1985. Tense and aspect systems. Oxford: Blackwell. Etxepare, Bernard. 1545. Lingue Vasconum Primitiae. Bordeaux. Facsimile, critical edition and translation by P. Altuna, 1995. Bilbao: Euskaltzaindia. Fleischman, Suzanne. 1982. The future in thought and language: Diachronic evidence from Romance. Cambridge: Cambridge Univ. Press. Felischman, Suzanne. 1983. “From pragmatics to grammar: Diachronic reflections on complex pasts and futures in Romance”. Lingua 60(2–3). 183–214 Gómez, Ricardo & Sainz Koldo. 1996. “On the origin of the Finite Forms of the Basque Verb”. In Hualde Jose Ignacio, Joseba A. Lakarra & Trask Larry (eds.), Towards a History of Basque. Amsterdam/Philadelphia: John Benjamins. 235–254. Haase, Martin. 1992. Sprachkontakt und Sprachwandel in Baskenland: Die Einflüsse der Gaskgonischen und Französischen auf das Baskische. Hamburg: Helmut Buske Verlag.

80

hualde and mounole

Harre, Catherine. 1991. Tener + Past Participle: A case study in linguistic description. London: Routledge. Hualde, José Ignacio, Gorka Elordieta & Arantzazu Elordieta. 1994. The Basque dialect of Lekeitio. [Supplements of ASJU XXXIV]. Bilbao/Donostia-San Sebastián: Univ. of the Basque Country Press/ Diputación Foral de Gipuzkoa. Krajewska, Dorota. 2013. The diachrony of resultative constructions in Basque. Master’s Thesis, UPV/EHU. Lakarra, Joseba A. 1996. Refranes y sentencias (1596): Ikerketak eta edizioa. Bilbo: Real Academia de la Lengua Vasca/Euskaltzaindia. Lakarra, Joseba A. 2017. “Basque and the reconstruction of isolated languages”. In Lyle Campbell (ed.), Language isolates. London: Routledge. 59–99. Ledgeway, Adam. 2011. “Syntactic and morphosyntactic typology and change”. In Martin Maiden, John Charless Smith & Adam Ledgeway (eds.), The Romance languages, vol. 1: Structures. Cambridge: Cambridge Univ. Press. 382–471. Manterola, Julen. 2015. Euskararen morfologiaren historiarako: artikuluak eta erakusleak/ Towards a history of Basque morphology: Articles and demonstratives. Doctoral dissertation, UPV/EHU, Vitoria-Gasteiz. Michelena, Luis. 1981. “Lengua común y dialectos vascos”. ASJU XV: 291–303. Reprinted in: Luis Michelena. 2011. Obras completas [Supplements of ASJU VII]. ed. by Joseba A. Lakarra & Iñigo Ruiz Arzallus, Donostia-San Sebastián & Vitoria-Gasteiz: Gipuzkoako Foru Aldundia & UPV/EHU. 517–544. Mounole, Céline. 2007. “Perifrasi zaharra mendebalde eta erdialdeko euskara zaharrean: Azterketa kuantifikatiboa eta proposamen berria”. ASJU 41(1): 67–138. Mounole, Céline. 2008. “Sintaxi diakronikoa eta aditz multzuaren garapena: inperfektibozko perifrasiaren sorreraz”. In Xabier Artiagoitia & Joseba A. Lakarra (eds.) Gramatika jaietan Patxi Goenagaren omenez [Supplements of ASJU LI]. 548–604. Mounole, Céline. 2010. “Alineazio aldaketak euskararen diakronian”, In Beatriz Fernández, Pablo Albizu & Ricardo Etxepare (eds.), Euskara eta euskarak: aldakortasun sintaktikoa aztergai [Supplements of ASJU LII]. 151–170. Mounole, Céline. 2011. Le verbe basque ancien: étude philologique et diachronique. Ph.D. thesis, UPV/EHU. ASJU 48 (2014 [2018]). Bilbao: UPV/EHU. Ortiz de Urbina, Jon. 2003. “Semiauxiliary verbs”. In José Ignacio Hualde & Jon Ortiz de Urbina, (eds.), A Grammar of Basque, 300–328. Berlin/New York: Mouton de Gruyter. Ortiz de Urbina, Jon & Myriam Uribe-Etxebarria. 1991. “Participial predication in Basque”. In Joseba A. Lakarra (ed.), Memoriae L. Mitxelena Magistri Sacrum, vol. 2, [Supplements of ASJU XIV]. Donostia-San Sebastián: Diputación de Gipuzkoa. 993– 1012. Real Academia Española. Corpus diacrónico del español (CORDE). [Available on the internet: http://www.rae.es]

tense, aspect and mood

81

Rohlfs, Gerhard. 1977. Le gascon: Etudes de philologie pyrénéenne. Tübingen: Max Niemeyer Verlag. Trask, Robert L. 1997. The history of Basque. London: Routledge. Urgell, Blanca. 2007. “Para la historia del sustantivo verbal en vasco”. In Joseba A. Lakarra & José Ignacio Hualde (eds.), Studies in Basque and historical linguistics in memory of R.L. Trask/R.L. Trasken oroitzapenetan ikerketak euskalaritzaz eta hizkuntzalaritza historikoaz [ASJU XL]. 921–948. Villasante, Luis. 1980. Sintaxis de la oración simple. Oñati: Editorial Franciscana Aránzazu. Vincent, Nigel. 1982. “The development of the auxiliaries habere and esse in Romance”. In Nigel Vincent & Martin Harris (eds.), Studies in the Romance verb. London: Croom Helm. 71–96. Zabala, Igone. 2003. “Nominal predication: copulative sentences and secondary predication”. In José Ignacio Hualde & Jon Ortiz de Urbina, (eds.), A grammar of Basque, Berlin/New York: Mouton de Gruyter. 426–448.

chapter 4

Non-verbal Participles in Basque and Spanish Ane Berro

1

Introduction

Adjectival participles have been argued to be mixed categories, since they show, at the same time, properties belonging to verbs and to adjectives. On the one hand, they are superficially adjectives: they denote a state predicated about an entity, and can occur in attributive (1) or predicative position (2). On the other hand, they can also be internally verbal, because, in some cases, participles also imply a previous event that has as a result the state denoted by the participle. (1) The broken tree. (2) The tree is broken. The different types of participles constitute an area of substantive study for the analysis of the interface between syntax and the lexicon (Wasow 1977) and syntax and morphology (Marantz 2001, 2007; Anagnostopoulou & Samioti 2014), as well as for analysis of lexical categories and the projection of syntactic heads like v, Voice and Aspect. Depending on whether an event is implied or not, and on the acceptability of different event-related and subject-oriented modifiers, adjectival participles have been argued to be of different types, with different structural compositions, particularly involving different verbal layers (e.g. Anagnostopoulou 2003; Embick 2004; Alexiadou & Anagnostopoulou 2008; Bruening 2014; Doron 2014). Those without an event implication have been considered to be built on top of an acategorial Root (Embick 2004). In contrast, others may involve, at different degrees, verbal projections like vP, VoiceP and even a perfect AspP (Alexiadou et al. 2014; Alexiadou et al. 2015, see also Sleeman 2011, 2014). In this chapter, Basque and Spanish adjectival (and adpositional) participles will be addressed, in the light of the literature on adjectival participles. As in most of the literature following Kratzer (1994, 2000), I will assume that there is a differentiation between adjectival and verbal participles: adjectival participles denote a state, whereas verbal participles describe an event. Additionally, I will also consider that adjectival participles can be phrasal and that

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004395398_005

non-verbal participles in basque and spanish

83

they can also have event implications. With this picture in mind, in this chapter, I will explore the properties of phrasal and non-phrasal non-verbal participles in Basque and Spanish, comparing them with verbal ones, and analyzing whether the presence/absence of these properties is subject to a classification of different types of non-verbal participles, such as resultant state / target state participles (Kratzer 2000) and resultative/stative participles (Embick 2004). Spanish past participles have been studied in a number of works (among many others, Luján 1981; Demonte 1983; Bosque 1999, 2014; Marín 1997, 2000, 2004ab, 2009; Gehrke & Marco 2014), but there are not many studies about Basque non-verbal participles (Ortiz de Urbina & Uribe-Etxebarria 1991; Krajewska 2012, 2013; Berro 2017) and none comparing both Spanish and Basque. Both languages, although being typologically distant, build and use non-verbal participles in a strikingly similar way: they are alike in building resultative participles by means of a stage-level copula (egon in western and central Basque varieties and estar in Spanish) and an adjectival (or adpositional) participle. Additionally, as will be shown, both languages show a similar interaction between lexical aspect and grammatical aspect, with adjectival (and adpositional) participles interpreted in a non-resultative way when the embedded event is non-dynamic. On the other hand, they also differ in a number of aspects. For example, Basque non-verbal participles are morphologically more complex (showing an additional resultative morpheme attached to the bare participle) and, depending on the suffix, they can be adjectival and adpositional. Moreover, they are, in principle, acceptable in more syntactic contexts, given that Basque non-verbal participles allow temporal and spatial modification of the event underlying the participle and, if the copula turns into izan ‘be’ (substituting egon), are compatible with all kinds of agent arguments. In Spanish, it has been argued that the insertion of por-phrases (by-phrases) in adjectival participles is restricted to those that modify an event kind (Gehrke & Marco 2014). However, as will be shown, at least, in the Spanish spoken in the Basque Country, temporal and spatial modification of the event is acceptable, and por-phrases with wide scope, strong determiners and discourse referents are also judged grammatical. These features have been taken as indicators for the presence of verbal projections in adjectival participles cross-linguistically (Gehrke 2011, 2015; Alexiadou et al. 2014; Alexiadou et al. 2015) and will be considered in this chapter in order to approach the structural composition of non-verbal participles in Basque and Spanish. Apart from the different degrees of eventiveness implied by the participles, in this chapter, I will also pay attention to the configuration in which the participle occurs, particularly, to whether the participle occurs in the attributive or

84

berro

predicative position, and regarding the latter, to whether the copula is intransitive or transitive. As will be seen, these aspects are sometimes related with the interpretation of the participle and also with the acceptability of event-related modifiers. The structure of the chapter is the following: in section 2, the form and meaning of non-verbal participles in Spanish and Basque will be analyzed, paying attention to their morphological and clausal composition, their combination with the copula, their category and their position in the clause. In section 3, the interaction between the lexical aspect and grammatical aspect is studied, following particularly Bosque (2014), and also addressing the experiential interpretation that Basque adjectival and adpositional participles can have (Krajewska 2012, 2013ab). In section 4, the acceptability of non-verbal participles with different types of event-oriented modifiers is explored, in the light of the literature on adjectival participles in other languages such as German and Greek (Kratzer 1994, 2000; Anagnostopoulou 2003; Alexiadou & Anagnostopoulou 2008; Gehrke 2011, 2015; Alexiadou et al. 2014; Bruening 2014; Alexiadou et al. 2015). In section 5, the restrictions that hold on event-modification are addressed, particularly following the event token and event kind distinction introduced in Gehrke (2011 and subsequent work) and discussing Spanish data that suggest that adjectival participles in Spanish involve event kinds, and not events that are instantiated (Gehrke & Marco 2014). I will argue that, in Basque adjectival and adpositional participles, there is an event token, and that that may also be the case in Spanish varieties spoken in the Basque Country.

2

Non-verbal Participles in Basque and Spanish

Like in English and German, adjectival participles in Spanish have the same form as verbal participles used in perfect clauses (4), with the participial ending attached to the verbal Root [+theme vowel]. Particularly, they look very similar to verbal passive participles, since adjectival participles show gender and number agreement with the entity they are predicated about (4), just like verbal passive participles (5).1

1 These are the abbreviations used in the glosses: abl = ablative; abs = absolutive; cl = clitic; dat = dative; erg = ergative; fem = feminine; imperf = imperfect; ine = inessive adposition; instr = instrumental; loc = locative (used in the locative copula); masc = masculine; part = partitive; pl = plural; prt = participle; pst = past; rel = relational adposition; res = resultative suffix; sg = singular.

non-verbal participles in basque and spanish

85

(3) He comprado un coche. have.1sg buy.prt a car ‘I have bought a car.’ (4) a. El coche está comprado. the car be.loc.3sg buy.prt.masc ‘The car is bought.’ b. Las manzanas están compradas. the apple.pl be.loc.3pl buy.prt.fem.pl ‘The apples are bought.’ (5) Las manzanas han sido compradas por mí. the apple.pl have.3pl be.prt buy.prt.fem.pl by me ‘The apples have been bought by me.’ In Basque, the participle used in perfect clauses is headed by the suffix -tu/i/-n/-ø, and, like in Spanish, it does not agree with the arguments of the verb. Differing from Spanish, adjectival participles in Basque are formed attaching an additional suffix to the participle used in perfect clauses (glossed here as Res(ultative), see section 2.1): -a. (6) Auto bat eros-i dut. car a buy-prt have.1sg.erg ‘I have bought a car.’ (7) a. Auto-a eros-i-a dago. car-the buy-prt-res be.loc.3sg.abs ‘The car is bought.’ b. Sagarr-a-k erosi-a-k daude. apple-the-pl buy-prt-res-pl be.loc.3pl.abs ‘The apples are bought.’ When the participle is headed by -a, it agrees in number with the entity is predicated about. As will be explained in the following subsection, Basque has other suffixes to form resultative participles, but unlike -a ending ones, those do not agree in number with the entity they are predicated about.

86

berro

2.1 Different Participial Suffixes in Basque Apart from the -a ending, resultative participles can also be built with the suffixes -ta [-da after a nasal or lateral] or -rik [-ik after a consonant]. (8) Mahai-a apur-tu-ta dago. table-the buy-prt-res be.loc.3sg.abs ‘The table is broken.’ (9) Gizon-a Amazonia-n gal-du-rik dago. man-the Amazonia-ine lose-prt-res be.loc.3sg.abs ‘The man is lost in Amazonia.’ -Ta and -rik are dialectally distributed: in western varieties, we find -ta headed participles, whereas in eastern dialects -rik participles are more prominent. In central varieties, both types of participles are found (Krajewska 2013). -A participles, on the other hand, are used in all dialects. Participles headed by -a are considered in the Basque linguistic tradition adjectival, because they can occur in attributive and predicative configurations and because they agree in number with the absolutive noun they are predicated about (7), just like individual level adjectives (10b). (10) a. Mendi altu-a-k mountain high-the-pl ‘(The) high mountains’ b. Mendi-a-k altu-a-k dira. Mountain-the-pl high-the-pl be.3pl.abs ‘Mountains are high.’ The suffix -a is homophonous with the determiner meaning ‘a’ or ‘the’, and has been analyzed as a resultative aspect marker (Oyharçabal 1992) or a predicative marker (Ortiz de Urbina & Uribe-Etxebarria 1991; Eguren 2012) in the context of participles. As shown in example (10), -a also occurs heading individual level adjectives like altu-a-k [tall-a-pl]. Eguren (2012), in this respect, claims that -a, in the latter context, is the exponent of a Pred(icative) head (Baker 2003). On the other hand, -ta and -rik headed participles are usually referred to as adverbial (Rebuschi 1984), given that they cannot appear in attributive position and they do not show number agreement with the absolutive argument. Their historical origin is adpositional. According to Krajewska (2013), -rik has evolved from the ablative adposition: it started to be used to mark converbs

non-verbal participles in basque and spanish

87

(a diachronic tendency that is attested cross-linguistically), then secondary participial predicates, and finally turned into a resultative suffix. As for -ta, it originated from the copulative conjunction eta ‘and’ (de Rijk 2008: 692), obtained the temporal meaning of after, and ultimately was reanalyzed as a suffix attached to the verb. Synchronically, -ta and -rik can be argued to be adpositional, and this way, we would be able to account for their lack of number agreement. Moreover, these suffixes behave like adpositions in not being acceptable in attributive position, unless they are additionally headed by the relational suffix -ko (Goenaga 1978; de Rijk 1998 [1993]; Elordieta 2001; Oyharçabal 2010). Adpositions cannot be adnominal modifiers unless they are headed by -ko, a distributional feature that is not shared by adjectives. (11) a. *Laser bide-z depilazio-a laser way-instr hair removal-the Intended: ‘hair removal through laser’, ‘laser hair removal’ b. Laser bide-z-ko depilazio-a laser way-instr-rel depilation-the ‘Laser hair removal’ (12) a. *gal-du-ta / *gal-du-rik gizon-a lose-prt-res / lose-prt-res man-the Intended: ‘The lost man’ b. galdu-ta-ko / galdu-ri-ko gizon-a lose-res-rel / lose-res-rel man-the ‘The lost man’ In this respect, -ta and -rik participles replicate the behavior of adpositions. They are good adnominal modifiers only in the presence of the relational morpheme. This adnominal use of adpositional resultatives is usually referred to as a relative form (Hualde et al. 1994). All in all, Basque seems to have adjectival (ending in -a) and adpositional (ending in -ta and -rik) participles, apart from the verbal participial form (ending in -tu/-i/-n) used in perfect clauses. 2.2 Stative and Resultative Participles in Basque and Spanish The suffixes -a/-ta/-rik in Basque can be used to form stative participles. Embick (2004) proposed that adjectival participles in English can be stative or resulta-

88

berro

tives. Stative participles denote a characteristic state, like a non-derived adjective, whereas the resultative denotes a state that is the result of a previous event, where this event is grammatically represented. In the former, the participle is syntactically structured with an Aspectual head (Asp) projected directly on top of an acategorial Root, whereas in the latter, the Root is verbalized (in a verbal phrase, vP) before being selected by Asp. There are different diagnostics to tell apart stative participles from resultatives ones. One involves the use of manner adverbs that modify the manner in which the event has taken place (Kratzer 1994). Stative participles (called lexical adjectival participles in Kratzer 1994) do not allow this kind of modification, whereas resultatives do (13). Another one is the ability to occur as complement of verbs of creation like build, create or make (Embick 2004) (14). Resultative participles are not acceptable in this environment, since they imply a previous event and this is incompatible with the presence of the verb of creation. Stative adjectival participles, in contrast, are perfectly grammatical. (13) a. The package remained carefully opened. b. *The package remained carefully open. (14) a. The door was built open. b. *The door was built opened. Thus, even if the same root (√open) is used in (13ab) and (14ab), the contrasts found suggest that it is being used in different configurations, and this is supported by the participial morphology (-ed) only found in the resultative and not in the stative adjective. In other cases, however, the same participial form is used in both (e.g. closed). (15) a. The package was carefully closed. b. The door was built closed. In Basque, the participial suffixes -a/-ta/-rik can be used to form resultative and stative participles. In the case of the Root zabal ‘open, wide’, the stative and the resultative participles are morphologically differentiated. Zabal-ik, consisting of the Root and the suffix -rik, is used as a stative adjective, whereas zabaldu-a/-ta/-rik is used in the resultative participle. In this case, the resultative involves the Root (zabal), the participial suffix -tu (-du after -n/-l) (the one used in perfect clauses) and the adjectival/adpositional suffixes -a/-ta/-rik.

non-verbal participles in basque and spanish

89

(16) a. Ate-a zabal-ik eraiki dute. door-the open-res build.prt have.3sg.abs.3pl.erg ‘They have built the door open’ b. ??Ate-a zabal-du-a/ta/rik eraiki dute. door-the open-prt-res build.prt have.3sg.abs.3pl.erg ‘*They have built the door opened’ (17) a. *Ate-a kontu handi-z zabal-ik dago. door-the care big-instr open-res be.loc.3sg.abs ‘*The door is open carefully’ b. Ate-a kontu handi-z zabal-du-a/ta/rik dago. door-the care big-instr open-prt-res be.loc.3sg.abs ‘The door is opened carefully’ Contrast like this one are attested for a number of other participles such as hutsik ‘empty’ vs. hus-tu-rik ‘emptied’, biluz-ik ‘naked’ vs. biluz-tu-rik ‘undressed’, poz-ik ‘happy’ vs. poz-tu-rik ‘delighted’ and bakarr-ik ‘alone’ vs. bakar-tu-rik ‘isolated’. The first member of the pairs is not usually considered participial, since it does not involve the participial suffix -tu that is used in perfect clauses. These stative elements are referred to as stage-level adjectives (Zabala 1993) or stative adverbs (de Rijk 2008). In the case of the Roots ireki ‘open’ and itxi ‘close’, in contrast, the same form is used as the stative and the resultative participles. It must be noted that, in these cases, the bare participle has the same form as the bare Root. (18) a. Ate-a ireki-ta eraiki dute. door-the open.prt-res build.prt have.3sg.abs.3pl.erg ‘They have built the door open.’ b. Ate-a kontu handi-z ireki-ta dago. door-the care big-instr open.prt-res be.loc.3sg.abs ‘*The door is open carefully.’ (19) a. Ate-a itxi-ta eraiki dute. door-the close.prt-res build.prt have.3sg.abs.3pl.erg ‘They have built the door open.’

90

berro

b. Ate-a kontu handi-z itxi-ta dago. door-the care big-instr close.prt-res be.loc.3sg.abs ‘*The door is open carefully.’ In Spanish, we find a similar scenario. Like in English and Basque, Spanish has also stative adjectives, called perfective adjectives (Bosque 1999, Fábregas & Marín 2015): lleno (vs. the participial llenado), seco (vs. the participial secado), limpio (vs. the participial limpiado), sucio (vs. the participial ensuciado), vacío (vs. vaciado) etc. They are defined as follows (Bosque 1999: 185; translation taken from Fábregas & Marín 2015). (20) [Perfective adjectives] denote states reached by the entities they are predicated of. Bosque (2014) argues that when participial forms have counterpart perfective adjectives of this sort, the post-nominal participial form can only have eventive (passive) meaning and not resultative, in Embick’s terms. (21) a. una piscina llenada a.fem swimming.pool filled.fem ‘A swimming pool filled up’ vs. b. Una piscina llena a.fem swimming.pool full.fem ‘A full swimming pool’ (22) a. la ropa secada the.fem clothes dried.fem ‘The clothes dried’ vs. b. La ropa seca the.fem clothes dry.fem ‘The dry clothes’ According to Bosque (2014), (21a) and (22a) have a passive eventive meaning, corresponding to their active counterpart (and do not denote the state reached

non-verbal participles in basque and spanish

91

by these actions). In his opinion, the resultative interpretation is blocked when the lexicon contains a lexical item for the adjective in the lower state of the causative structure. In relation to this, the adjectives in (21b) and (22b) seem to belong to the stative class proposed in Embick’s (2004), given that they are incompatible with a verb of creation. (23) a. Esta piscina se construyó llena. this.fem swimming.pool cl build.3sg.pst full.fem ‘This swimming pool was built full.’ b. *Esta piscina se construyó llenada. this.fem swimming.pool cl build.3sg.pst filled.fem ‘*This swimming pool was built filled up.’ Both the participle and the perfective adjective denote states. The difference is that in the perfective participle, an event has not necessarily taken place. Thus, in (23a), there is not an event of filling up, just like in stage-level adjectives headed by -rik in Basque. In relation to this, Fábregas & Marín (2015) defended the view that perfective adjectives are the exponents of a Result subevent (syntactically represented in a First Phase Syntax, Ramchand 2008) and an adjectival phrase (AP). Thus, according to these authors too, perfective adjectives correspond merely to a stative layer, with no dynamic event. They do not involve the Process subevent (responsible for providing dynamicity), so that the Result is simply interpreted as a stative subevent. They must be then paralleled to Embick’s stative participles, rather than to resultative ones. Fábregas & Marín (2015) support this analysis by means of several criteria, such as the following: (24) a. Juan dejó la casa limpia / *limpiada. John left the.fem house clean.fem / *cleaned.fem ‘John left the house clean/*cleaned.’ b. La casa quedo limpia / *limpiada. the.fem house stayed clean.fem / *cleaned.fem ‘The house stayed clean/*cleaned.’ The verbs dejar ‘to leave’ and its anticausative counterpart quedar ‘to stay’ focalize the result state and are only possible with telic verbs (García Fernández et al. 2006). According to Fábregas & Marín (2015), dejar and quedar lexicalize, subsequently, the Initiation + Process subevents and the Process subevent. The

92

berro

Result subevent is, in turn, spelled out by the perfective adjective. Adjectival participles like limpiada ‘cleaned’ are not acceptable in this context because they are blocked by the perfective participle. In the cases where the adjectival participle entails a change of state, they claim that they necessarily lexicalize Process. 2.3 Copular Verbs and the Biclausal Analysis In Spanish, adjectival and verbal participles (resultative and eventive participles in the terminology proposed by Embick 2004) are differentiated on the basis of the inflectional element accompanying the participle. Unlike in English, and more similar to German, the inflectional element used with verbal passive participles and adjectival resultative participles is different. In adjectival passives, the inflectional element is a copular verb, estar ‘(the stage level) be’, whereas in verbal passive participles, it is an auxiliary, namely ser ‘be’. (25) Las sillas están tiradas. the.fem.pl chair.pl be.loc.3pl throw.prt.fem.pl ‘The chairs are thrown.’ (26) Las sillas son tiradas (todas las mañanas). the.fem.pl chair.pl be.3pl throw.prt.fem.pl (every the mornings) ‘The chairs are thrown every morning.’ The configuration involving the verbal passive, with the auxiliary ser ‘be’ in the present tense, has a habitual interpretation, with the event of throwing the chairs repeated every morning. The structure with the adjectival participle, in contrast, only asserts that the state denoted by the participle holds at the moment of the reference time. Verbal passive participles occur in a monoclausal configuration, where the main predicate is the participle and the inflected element stands as an auxiliary, whereas resultative participles are bi-clausal: the main predicate is the copula (estar in Spanish), which takes the adjectival participle as its complement. Regarding Basque, the bi-clausal analysis of adjectival and adpositional participles is proposed by Hualde et al. (1994) and de Rijk (2008), and is particularly analyzed and defended in Ortiz de Urbina & Uribe-Etxebarria (1991). The monoclausal vs. bi-clausal analysis in Basque has been more controversial, though, precisely because adjectival participles in Basque are acceptable with ergative marked agent arguments, just like in monoclausal clauses, and because they can occur with the copular izan ‘be’, instead of with egon ‘(the stage level) be’.

non-verbal participles in basque and spanish

93

(27) Gutun-a Miren-ek idatz-i-a da. letter-the Mary-erg write-prt-res be.3sg.abs ‘The letter is written by Mary.’ Ortiz de Urbina & Uribe-Etxebaria (1991) argue in favor of a bi-clausal analysis of configurations like (27), which involve an adjectival participle, particularly on the basis of the absence of ergative agreement markers on the auxiliary and word-order restrictions that differentiate clearly configurations like (27) and perfect monoclausal clauses like (28). See section 8.3.3 for an overview of these facts. (28) (Miren-ek) gutun-a (Miren-ek) idatz-i du. (Mary-erg) letter-the (Mary-erg) write-prt have.3sg.abs.3sg.erg ‘The letter is written by Mary.’ In both Spanish and Basque, apart from using the intransitive copulas (izan ‘be’ and egon ‘(the stage level) be’ in Basque and estar ‘(the stage level) be’ in Spanish), adjectival (an adpositional) participles can occur with transitive copulas (edun and eduki ‘have’ in Basque and tener ‘have’ in Spanish). This configuration has been called possessive resultative (Nedjalkov & Jaxontov 1988) and it has been attested that, in the diachronic evolution of many languages, the perfect has evolved from earlier possessive constructions of this sort (among others, Detges 2000; Pinkster 1987). Actually, the perfect analytic configuration of modern Basque, with the participle ending in -tu/-i/-n, used to be employed as a resultative construction in previous stages of the language (Mounole 2011). In possessive resultatives, the state denoted by the participle is predicated about the object of the transitive copula, and the subject may or may not be interpreted as the agent of the event underlying the participle. For instance, in (29) and (30), the subject of ‘have’ can be interpreted merely as the possessor or holder of the state denoted by the object, but not as the agent initiating the event of cleaning. (29) a. Ni-k sagarr-a-k eros-i-ta dauzkat. I-erg apple-the-pl buy-prt-res have.3pl.abs.3sg.erg ‘I have the apples bought.’ b. Ni-k zapata-k garbi-tu-ta dauzkat. I-erg shoes.the-pl buy-prt-res have.3pl.abs.3sg.erg ‘I have the shoes cleaned.’

94

berro

(30) a. Tengo las manzanas compradas. have.1sg the.fem.pl apple.pl bought.fem.pl ‘I have the apples bought’ b. Tengo los zapatos limpiados. have.1sg the.masc.pl shoes.pl cleaned.masc.pl ‘I have the apples cleaned.’ Similar constructions can be found in many other languages, like English and French. In these cases, like in Spanish, the non-pronominal object of ‘have’ takes a pre-participial position, instead of its usual post-participial position, and the participle agrees in number and gender with the object in the case of French. In these configurations, the subject is interpreted as the possessor/holder but not necessarily as the agent of the event underlying the participle. (31) I have my shoes cleaned. (32) J’ai une lettre écrite. I have.1sg a.fem letter written.fem ‘I have a letter written.’ Iatridou (1996) analyzes similar configurations in Modern Greek, and proposes that, in these configurations, the object forms a passive small clause with the participle, and the small clause is the complement of ‘have’. (33) echo grameno ta grama. have.1sg written.neut.sg.acc the letter ‘I have the letter written.’ According to Iatridou (1996), in (33), the entity ‘the letter’ is not the object of ‘have’ since (33) does not assert a possessive relation between the subject (1st person singular) and ‘the letter’. This fact is explained considering that the whole passive small clause is the complement of ‘have’, and consequently, that there is no thematic relation between ‘have’ and the entity ‘the letter’, as argued in Iatridou (1996). Additionally, she argues that the agreement relation between ‘the letter’ and the participle in Modern Greek, and in similar configurations in French (32), must not be regarded as a case of object agreement, but that it should be analyzed as subject-predicate agreement inside a passive small clause.

non-verbal participles in basque and spanish

95

2.4

Categorial Status: Tests to Tell Verbal and Adjectival (and Adpositional) Participles Apart In the literature, several diagnostics have been identified to differentiate verbal and adjectival participles (Wasow 1977; Levin & Rappaport Hovav 1986; Embick 2004; Sleeman 2011; McIntyre 2013 etc.): prenominal position in attributive participles, un- prefixation, semantic presence/absence of an external argument, selection by adjective phrase taking verbs, incompatibility with double objects etc. As pointed out by Bosque (2014), though, the majority of these tests cannot be applied to Romance, and, actually, neither to Basque. As mentioned in section 3.2.3, adjectival participles in Spanish occur with the copular verbs estar ‘(the stage level) be’ or tener ‘have’, whereas verbal passive participles are combined with the auxiliary ser ‘be’ or haber ‘have’. This way, adjectival participles occurring in predicative position can be differentiated from verbal passive participles. (34) Esta mesa está rota. this.fem table be.loc.3sg broken.fem ‘This table is broken.’ (35) Esta mesa fue rota. this.fem table be.3sg.pst broken.fem ‘This table was broken.’ Number and gender agreement between the participle and the predicated theme cannot be taken as a diagnostic to differentiate adjectival participles from verbal ones. For instance, as shown in section 3.2, verbal passive participles also agree with the derived subject. Bosque (2014) considers that both participles depicted in (34) and (35) are verbal,2 and not adjectival, and points out that the ability of resultative participles to occur with secondary predicates (36) and infinitive complements in causative sentences (37a) or verbal set phrases (37b) support their verbal status. (36) a. Un acusado considerado culpable a.masc defendant considered.masc guilty ‘A defendant found guilty’

2 According to Bosque (2014), the difference between (34) and (35) is that, in (34), the participle is resultative, whereas in (35), it is eventive. Estar is only compatible with resultative participles.

96

berro

b. Un concejal elegido alcalde a.masc councilman elected.masc mayor ‘A councilman elected mayor’ (37) a. El lío de las pruebas hechas desaparecer the.masc mess of the proofs made.fem.pl disappear ‘The mess of the proofs that were made disappear’ (El País, 30/01/2012) b. Garbanzos puestos a remojo chickpeas put.masc.pl to soak ‘Chickpeas left to soak’ Nevertheless, their grammaticality in these contexts may be explained if we consider that they syntactically involve a vP layer, embedded under the adjectival layer. In Basque, adjectival participles in predicative position can occur either with the copula egon ‘(stage level) be’ or with the copula izan ‘be’. The latter can also be used as an auxiliary in intransitive analytic verbal configurations (perfect, perfective, imperfective and prospective). Another diagnostic that has been identified in the literature to differentiate adjectival and verbal participles is the ability to occur in the complement position of verbs like act, appear, become, look, remain, seem, smell and sound in English (Wasow 1997), which is restricted to adjectives. Both Spanish and Basque adjectival participles are licit in this context. (38) María parece cansada / triste. Mary looked tired.fem / sad ‘Mary looked tired/sad.’ (39) neka-tu-a / triste dirudizu. exhaust-prt-res / sad seem.2sg.erg ‘you look exhausted/sad’ Even Basque adpositional participles (headed by -ta and -rik) can occur as the complements of verbs like eman ‘to seem’ and of irudi ‘to seem, to look like’ (40). In contrast, bare participles in Basque—those headed by the participial ending -tu/-u/-n—are ungrammatical in all these contexts. They cannot be the complement of eman ‘to seem’ and of irudi ‘to seem, to look like’ and, moreover, they are incompatible with the copulas egon ‘(the stage level) be’ and eduki ‘have’.

non-verbal participles in basque and spanish

97

(40) neka-tu-ta/-rik dirudizu. exhaust-prt-res seem.2sg.erg ‘you look exhausted’ (41) *neka-tu dirudizu exhaust-prt seem.2sg.erg Intended: ‘you look exhausted’ (42) *Liburu-a-k apur-tu daude / book-the-pl break-prt be.loc.3pl.abs / dauzkat have(eduki).3pl.abs.1sg.erg Intended: ‘(The) books are broken / I have (the) books broken’ Another test that is usually referred to is the compatibility of adjectival participles with degree modifiers. As shown by Kratzer (2000), degree modifiers are only compatible with target state participles and ungrammatical with verbal resultant state participles (see also Anagnostopoulou 2003). The modifier very, in particular, selects for gradable adjectives that have a context-sensitive standard, in terms of Kennedy and McNally (1999). This means that they usually select for adjectives that denote open scales. In Spanish, adjectival participles denoting open scales, both in predicative position (43) and in attributive position (44), can be modified by muy ‘very’, unlike verbal passive participles (45). (43) a. Alex Mumbrú, que estuvo muy vigilado en el Alex Mumbrú, that be.loc.3sg.pst very guarded.masc in the poste bajo, … post low ‘Alex Mumbrú, who was very guarded in the low post, …’ (Deia, 12/05/ 2017) b. Los globos están muy inflados the.masc.pl ballon.pl be.loc.3pl very inflated.masc.pl ‘The ballons are very inflated’ c. Las casas están muy the.fem.pl house.pl be.loc.3pl very encarecidas become.expensive.prt.fem.pl ‘The houses have become very expensive’ Lit. ‘The houses are very become expensive’

98

berro

(44) a. Un edificio muy vigilado (por la policía) a.masc building very guarded.masc (by the police) b. Un político corrupto muy protegido por sus a.masc politician corrupt.masc very protected.masc by his colegas en los últimos años colleagues in the last years ‘A corrupt politician very much protected by his colleagues in the last years.’ (Bosque 2014: 54, footnote 12). (45) a. *Las casas fueron muy the.fem.pl house.pl be.3pl.pst very encarecidas become.expensive.prt.fem.pl b. *Los globos fueron muy inflados the.masc.pl ballon.pl be.3pl.pst very inflated.masc.pl Degree modifiers are also compatible with Basque adjectival and adpositional participles (46) (47), which, once again, contrast with bare participles used in perfect simple clauses (48). (46) a. Oso leku zain-du-a da. very place protect-prt-res be.3sg.abs ‘It is a very protected place.’ b. Estilo oso puz-tu-a du. style very boast-prt-res have.3sg.abs.3sg.erg ‘He/she has a very boasted style.’ (47) a. Mahai-a oso apur-tu-ta dago. table-the very break-prt-res be.loc.3sg.abs ‘The table is very broken.’ b. Puxika-k oso puz-tu-a-k / puz-tu-ta balloon.the-pl very inflate-prt-res / inflate-prt-res daude. be.loc.3pl.abs ‘The balloons are very inflated.’

non-verbal participles in basque and spanish

99

(48) a. *Mahai-a oso apur-tu da table-the very break-prt be.3sg.abs b. *Puxika-k oso puz-tu dira / balloon.the-pl very inflate-prt be.3pl.abs / ditut have.1sg.erg.3pl.abs As can be seen, in Spanish, adjectival participles that are modified by muy ‘very’ can also occur with certain por-phrases (44). Interestingly, the por-phrases occurring in this context modifies the state denoted by the participle, rather than the previous event (see Alexiadou et al. 2014 and Alexiadou et al. 2015 for similar cases in Greek). 2.5 Location in the Clause: Attributive vs. Predicative Participles The standard assumption in the literature is that verbal participles (eventive passives in Embick’s 2004 terminology) cannot occur in attributive position because only adjectives, and not verbs, can be prenominal modifiers (Embick 2004). Sleeman (2011 2014) proposes that attributive participles can be either adjectival (resultative) or verbal (eventive), showing that, in Dutch and English, prenominal passive participles are adjectival while post-nominal ones are verbal. In Basque, adjectival participles in attributive position are only possible in post-nominal position, but there exists a contrast that seems to be related to the prenominal vs. post-nominal position of (participial) adjectives in other languages like English. In Basque, adjectival participles can modify a bare noun or a DP, and these two options present a number of contrasting distributional features. As shown in section 2.3, adjectival participles in predicative position can license an agentive argument bearing ergative case marking, which unlike in monoclausal simple clauses, does not agree with the inflectional element and has a fixed position: it must remain between the absolutive theme and the adjectival participle (Ortiz de Urbina and Uribe-Etxebarria 1992, see also section 4.2.2 in this chapter). In attributive position, the introduction of an ergative argument is more restricted. It can only be introduced in adjectival participles that modify a DP nominal, and not in those that modify a bare noun. (49) Gutun-a guraso-ek sina-tu-a da. letter-the parents-erg firm-prt-res be.3sg.abs ‘The letter signed by the parents’

Predicative

100

berro

(50) a. Gutun-a sina-tu-a (ekarri behar dute.) Attributive with DP letter-the firm-prt-res (they have to bring) ‘(They have to bring) the letter firmed’ b. Gutun-a guraso-ek sina-tu-a (ekarri behar dute.) letter-the parents-erg firm-prt-res (they have to bring) ‘(They have to bring) the letter firmed by the parents’ (51) a. Gutun sina-tu-a (ekarri behar dute.) letter firm-prt-res (they have to bring) ‘(They have to bring) the firmed letter’

Attributive with bare noun

b. *Gutun guraso-ek sina-tu-a (ekarri behar dute.) letter parents-erg firm-prt-res (they have to bring) ‘*(They have to bring) the firmed letter by the parents’ The adjectival participle without the ergative argument can modify either a DP (50a) or a bare noun (51a). In the presence of the ergative argument, the participle can only modify a DP, cf. (50b) (51b). The same contrast arises with indirect dative arguments. A dative indirect object, participant of the event underlying the participle, can only be introduced when the adjectival participle modifies a DP. (52) a. Gutun-a bidal-i-a (ikusi nahi dut.) letter-the send-prt-res (I want to see) ‘(I want to see) the letter sent.’

Attributive with DP

b. Gutun-a Miren-i bidal-i-a (ikusi nahi dut.) letter-the Mary-dat send-prt-res (I want to see) ‘(I want to see) the letter sent to Mary.’ (53) a. Gutun bidal-i-a (ikusi nahi dut.) Attributive with bare noun letter send-prt-res (I want to see) ‘(I want to see) the sent letter.’ b. *Gutun Miren-i bidal-i-a (ikusi nahi dut.) letter Mary-dat send-prt-res (I want to see) ‘*(I want to see) the sent letter to Mary.’

non-verbal participles in basque and spanish

101

The neutral relative order of the adjective, noun and the determiner in Basque is depicted in (54). As can be seen, adjectives occur between the bare noun and the determiner. (54) Noun Adjective Det Auto gorri -a ‘The red car’ But there are also instances where the adjective can modify a DP (55b). Let us compare the example in (55a), with the adjective modifying the bare noun, and the example in (55b). There is a subtle meaning contrast between them, which seems to be similar to that obtained between prenominal and post-nominal adjectives in English. (55) a. Auto gorri-a nahi dut car red-the want have.1sg.abs.1sg.erg ‘I want the red car’ b. Auto-a gorri-a nahi dut car-the red-the want have.1sg.abs.1sg.erg ‘I want the car red’ In the example in (55a), the subject wants a specific car; it picks up an individual, which is distinguished from its red color. In the example (55b), in contrast, the subject expresses which property the car he/she wants must have. The inability of prenominal adjectival participles to take further modifiers finds a correlate in non-derived adjectives. Actually, as shown by Sadler and Arnold (1994), prenominal and post-nominal non-derived adjectives show a clear contrast with respect to their ability to take modifiers and complements. (56) a. a child grateful for the present b. *a grateful for the present child Sadler and Arnold (1994) point out several interpretative contrasts between prenominal and post-nominal adjectives: in prenominal position, the reading of the adjective is characteristic, timeless or defining property of the noun, whereas in post-nominal position, the interpretation is of a temporary quality or property. This is particularly observed in the following example:

102

berro

(57) a. We need to find a responsible person b. We need to find the person responsible As suggested by Sadler & Arnold, a responsible person is someone who has generally the property of being responsible, whereas the person responsible is the individual who has performed a particular action. This contrast can be easily related to the individual level vs. stage level distinction (Carlson 1977). In relation to this contrast, Svenonius (1993) argued that post-nominal position is restricted to stage-level adjectives, but as pointed out in Sadler and Arnold (1994), individual levels may also occur post-nominally. In any case, as claimed by Sadler and Arnold (1994: 8), “prenominal adjectives are better when they provide a ‘natural’ classification in combination with the noun” or when they “denote some plausibly characteristic property of the noun they modify”. A possible analysis differentiating prenominal and post-nominal adjectives is to consider that post-nominal adjectives are reduced relative clauses, so that they would be really predicative, involving some kind of silent copular verb. Actually, Sadler and Arnold (1994) list several aspects in which post-nominal adjectives and predicative adjectives behave alike in English. For instance, (i) adjectives that cannot occur in predicative position cannot occur in attributive post-nominal position (58), (ii) when an adjective in the prenominal and the post-nominal position has different senses, the post-nominal sense is the same as that obtained in predicative position (59), (iii) prenominal adjectives allow idiomatic interpretations in combination with the noun, whereas postnominal adjectives and predicative adjectives do not (60). (58) a. *Those policemen are former b. *policemen former c. former policemen (59) a. The present editors b. The editors present c. The editors are present (60) a. a white lie b. ??a lie white and without malice c. ??a lie which is white Nevertheless, Sandler and Arnold do not consider that the analysis of postpositional adjectives can be reduced to that of predicative ones, since, as they

non-verbal participles in basque and spanish

103

argue, the semantics of post-nominal adjectives is not identical to that of predicative adjectives. As a matter of fact, this analysis would not be able to explain the individual level vs. stage level dimension attributed to prenominal and post-nominal adjectives, given that predicative adjectives can be either individual level or stage level (Carlson 1977).3

3

Interaction between Lexical Aspect and Grammatical Aspect

In this section, I will analyze the lexical aspect of the verbs involved in resultative participles and how the lexical aspect interacts with the interpretation of the adjectival/adpositional participle. I will show that, even though verbs involving an internal argument are the most frequent verbs giving rise to adjectival/adpositional participles in Basque and Spanish (and generally, crosslinguistically), unergative verbs and also transitive verbs where the theme of predication is co-referent with the subject can also be found in this configuration in Basque (and sometimes also in Spanish). As will be explained, the occurrence of these verbs and interpretations is closely correlated with the aspectual meaning of the participle. 3.1 Classes of Verbs Involved in Non-verbal Participles 3.1.1 Unaccusative/Transitive Verbs vs. Unergative Verbs The standard analysis of resultatives assumes that adjectival participles are derived from verbs that have an internal argument, that is, unaccusative (61a) (62a) and transitive verbs (61b) (62b). The theme of predication, agreeing with the copula, corresponds to the internal argument of the verb underlying in the participle. (61) a. Aulki-a jaus-i-ta dago. chair-the fall-prt-res be.loc.3sg.abs ‘The chair is fallen.’

3 It must be highlighted that, although there are some semantic correspondences between attributive participles with bare NP s and attributive participles with DP s in Basque, on the one hand, and pre-nominal and post-nominal adjectives in English on the other, I do not want to state that they are syntactically parallel, given that attributive participles with DP s in Basque can be regarded as being really predicative, but post-nominal adjectives in English do not need to be so.

104

berro

b. Mahai-a apur-tu-ta dago. table-the break-prt-res be.loc.3sg.abs ‘The table is broken.’ (62) a. La silla está caída. the.fem chair be.loc.3sg fallen.fem ‘The chair is fallen.’ b. La mesa está rota. the.fem table be.loc.3sg broken.fem ‘The table is broken.’ Cross-linguistically, unergative verbs do not occur in this configuration, because unergative verbs do not have an internal argument. Nevertheless, in Basque, adjectival and adpositional participles built on unergative verbs are sometimes acceptable. As will be shown in section 3.2.2, in the majority of cases, the use of an unergative verb in this configuration goes hand in hand with the experiential interpretation of the participle. In some particular cases, however, adjectival and adpositional participles built on unergatives have stative interpretation, like in the following examples. (63) a. Bazkal-du dut. have.lunch-prt have.1sg.erg ‘I have eaten lunch’ b. Bazkal-du-ta nago have.lunch-prt-res be.loc.1sg.abs ‘I am had lunch’ (64) a. Hori sines-ten dut. that believe-imprf have.1sg.erg b. Hala sines-tu-ta nago. that.way believe-prt-res be.loc.1sg.abs ‘Like this I am believed’ In these examples, the theme of predication corresponds to the subject of the event embedded under the participle; a subject that is marked with ergative case in simple clauses (63a) (64a), like subjects of unergative and transitive verbs. There are also some cases in Spanish where the theme of predication

non-verbal participles in basque and spanish

105

is correferent with the subject of the unergative verb embedded in the adjectival participle. These uses involve consuming verbs similar to that in (63): beber ‘to drink’ and comer ‘to eat’. (65) Los niños están bebidos y comidos. the.masc.pl children be.loc.3pl drunk.masc.pl and eaten.masc.pl ‘The children are drunk and eaten.’ The verbs beber ‘to drink’ and comer ‘to eat’ underlying the participle in the example (65) do not accept direct objects in this configuration. For instance, the example in (66), which includes direct objects, is not acceptable. (66) *Los niños están bebidos agua y the.masc.pl children be.loc.3pl drunk.masc.pl water and comidos papilla eaten.masc.pl porridge ‘The children are drunk water and eaten porridge.’ Like in the Basque examples, in the Spanish example of (65), the theme of predication (los niños ‘children’) corresponds to the unergative subject of the verbs underlying the participles. Subjects of unergative verbs, as well as subjects of transitive verbs, are considered to belong not the verbal projection, but introduced by a Voice head projected on top of vP/VP (Kratzer 1996 and following works). This is why, in an analysis where adjectival participles do not involve external arguments (Levin & Rapapport 1986, Kratzer 1994 2000 among others), the fact that predicates of this type are found in resultative participle configurations is unexpected (see section 4.2 for a discussion on this aspect). On the other hand, the ability of these verbs to occur in this configuration could also be related to the fact that their subjects, apart from being interpreted as agentive or initiators, can also be argued to undergo a change (see actually Ramchand 2008, where it is argued that the subject of consuming verbs is both an initiator and an undergoer), and can, therefore, hold the consequent state denoted by the participle. 3.1.2 Different Aspectual Classes of Verbs Bosque (2014) observes for Spanish resultative participles that the participle can have different interpretations depending on the aspectual class of the underlying verb. He shows that, if the verb belongs to the class of change-ofstate verbs, manner verbs or incremental verbs, the resulting participle has a bounded interpretation, meaning that the participle denotes a state obtained

106

berro

as a consequence of a past event. This is the case of the participles derived from the verbs castigar ‘to punish’ and atrapar ‘to catch’. (67) a. Un niño castigado a.masc child punished.masc ‘A punished child’ b. Ladrones atrapados por la policía thieves caught.masc.pl by the police ‘Thieves caught by the police’ In contrast, when the participle is derived from certain activity verbs (e.g. vigilar ‘to guard’, buscar ‘to seek’, perseguir ‘to chase’), extent verbs (e.g. ocupar ‘to occupy’) or verbs expressing constituency and other similar physical properties (e.g. formar ‘to constitute’), the meaning of the participle is unbounded: the participle denotes a state that holds at the relevant evaluation time, but does not imply a past event. (68) a. Un edificio vigilado a.masc building guarded.masc ‘A guarded building’ b. Una ciudad rodeada de montañas a.fem city surrounded.fem of mountains ‘A city surrounded by mountains’ The state can have started sometime in the past, e.g. in (68a), but this is irrelevant for the meaning of the participle, particularly in (68b). The activity verbs giving rise to the so-called unbounded past participles in Bosque (2014) also include the so-called D(avidsonian)-states (Maienborn 2005 2007, Rothmayr 2009, Fábregas & Marín 2017, Berro 2015) (e.g. esperar ‘to wait’, gobernar ‘to govern’): verbs that share properties with both eventive and stative verbs. They are eventive and, thus, behave like eventive verbs in some tests (they can occur in the progressive, they can be infinitive complements of perception verbs, they can be located in time and space etc.), but non-dynamic, and therefore act like stative verbs in some other aspects (they give rise to an on-going, non-habitual interpretation in the simple present, they are bad with dynamic adverbs etc.). Extent verbs (Gawron 2009; Koontz-Garboden 2010) and verbs expressing constituency and other physical property are predicates that denote nondynamic eventualities, and which alternate with dynamic interpretations (e.g.

non-verbal participles in basque and spanish

107

obstruct in English, Kratzer 2000). For example, the verb ocupar ‘to occupy’ is ambiguous between a dynamic and a non-dynamic reading: (69) Los estudiantes ocuparon / ocupaban la the.masc.pl students occupy.3pl.pst / occupy.3pl.imperf the sala room ‘The students occupied the room.’ (70) La cama ocupaba toda la habitación the.fem bed occupy.3sg.imperf all the room ‘The bed occupied the entire room.’ The unbounded reading of participles is derived from the non-dynamic interpretation of the verbs, and these two interpretations can be thought to be interrelated. For instance, the events denoted by the verb vigilar ‘to guard’ or the non-dynamic interpretation of rodear ‘surround’ do not cause any change on the internal argument. Consequently, the state they hold after the event of guarding or surrounding has taken place is basically interpreted as a state that may have started sometime in the past, but not as a consequent state reached after an event has taken place. The fact that adjectival participles built on certain verbs can have unbounded interpretation has been addressed in Anagnostopoulou & Samioti (2014: 14). As explained in section 2.2, Embick (2004) proposes that adjectival participles can be either stative or resultative. Stative adjectival participles denote a simple state that is not the result of a previous event. In this sense, stative participles seem to correspond roughly to Bosque’s unbounded participles, because in both types, a past event that gives rise to the state denoted by the participle is not implied.4 Embick (2004) claims that resultative adjectival participles are derived from a verb (a vP), whereas stative ones are built on an acategorial Root. The presence of the semi-functional head v in the former type of participles is responsible for the eventive interpretation of resultatives. Anagnostopoulou & Samioti (2014) oppose this view, showing that, in Greek, adjectival participles denoting characteristic states (without event implications) can also be built on verbs. In Greek, -tos ending participles have been

4 Note, however, that Bosque (2014) argues that unbounded participles are a subtype of resultative participles (opposed to bounded ones). Embick (2004), in contrast, makes a distinction between resultative participles and stative participles.

108

berro

claimed to be built at the Root level and not to entail the existence of a prior event (Alexiadou & Anagnostopoulou 2008). However, certain -tos participles involve the morphological exponents of verbalizers (e.g. -iz, -on, -a etc.), showing that -tos participles can also be derived from verbs.

(71)

Verbs a. b. c. d. e.

axn-iz koudoun-iz-o vathoul-on-o if-en-o fit-ev-o

Corresponding participle ‘steam’ ‘ring (a bell)’ ‘hollow out’ ‘weave’ ‘plant’

axn-iz-tos koudoun-is-tos vathoul-o-tos if-an-tos fit-ef-tos

‘steaming hot’ ‘ringing’ ‘hollow’ ‘woven’ ‘planted’

Despite the presence of verbalizers, -tos participles do not have eventive meaning and cannot be modified by adverbs (72), contrasting with -menos ending participles (73). (72) *Ta fita ine fitefta me diaforetika regalia The plants are planted with different instruments (73) Ta malia ine atsala htenismena The hair are sloppily combed Spanish unbounded participles (Bosque 2014) derived from activity verbs can be modified by manner adverbs, a fact which shows that even though unbounded participles of this type do not imply a past event, they involve the relevant functional projection that licenses manner adverbs. (74) Un edificio vigilado cuidadosamente a.masc building guarded.masc carefully ‘A carefully guarded building’ In contrast, participles derived from extent verbs and verbs expressing constituency and other physical property behave differently. Bosque (2014) argues, building on Gawron (2009), that these classes of verbs should be considered eventive. Nevertheless, it seems that the participial forms derived from these predicates do not accept easily manner adverbs, unlike participles derived from non-dynamic activity verbs.

non-verbal participles in basque and spanish

109

(75) a. Una ciudad rodeada (*cuidadosamente) de montañas a.fem city surrounded.fem (*carefully) of mountains b. Un apartamento habitado (??silenciosamente) por an.masc apartment inhabited.masc (??silently) by recien-casados newly-weds c. Un equipo formado (*detalladamente) por tres a.masc team formed.masc (*fine.grained) by three jugadores players If these predicates involved an event, it is not clear why their participial forms do not license manner modifiers. Moreover, if they were eventive in their nondynamic interpretation, they would be grammatical in the progressive, contrary to fact: (76) a. Las montañas rodean la ciudad the.fem.pl mountains surround the.fem city b. #Las montañas están rodeando la ciudad #the.fem.pl mountains are surrounding the.fem city (77) a. Los reciencasados habitan el apartamento the.masc.pl newlyweds inhabit the.masc apartment b. *Los recién-casados están habitando el the.masc.pl newly-weds are inhabiting the.masc apartamento apartment (78) a. Tres jugadores forman el equipo three players form the.masc team b. #Tres jugadores están formando el equipo #three players are forming the.masc team

110

berro

3.1.3

Morphological Evidence of “Verbalness” in Stative Adjectival Participles in Basque In Basque, we also find morphological evidence of verbs in adjectival and adpositional participles that are interpreted in a non-resultative way. This happens in participles derived from extent verbs and verbs expressing constituency and physical properties, like in Spanish. (79) Hiri bat mendi-ek ingura-tu-a city a mountain-erg.pl surround-prt-res ‘A city surrounded by mountains’ (80) Talde bat lau jokalari-k osa-tu-a team a four players-erg form-prt-res ‘A team formed by four players’ As explained at the beginning of section 2, Basque resultative participles are formed attaching the suffixes -a/-ta or -rik to the bare participle used in simple perfect clauses. Bare participles are generally headed by -tu or -i, being -tu the only productive suffix in contemporary Basque. Berro (2015) argues that -tu is the exponent of an eventivizer, since it can be used very productively to form new predicates (see de Rijk 2008). These are some examples of -tu headed verbs derived from an adjectival root (81a), an adverbial root (81b), a nominal root (81c) and an adposition (81d). (81) a. gorri ‘red’ b. hurbil ‘near’ c. ama ‘mother’

gorri-tu ‘to redden’ hurbil-du ‘to approach, to get closer’ ama-tu ‘to become a mother’ ‘to turn sb into a mother’ d. etxe-ra ‘to home’ etxe-ra-tu ‘to go home’ ‘to carry sb home’

The -tu suffix only attaches to the Root when the Root is being used as a verb. In some cases, the Root changes its final vowel depending on its categorial environment. Artiagoitia (2004) addresses this phonological change affecting the final vowel of certain Roots in verbal environments (where those ending in /e/, /o/ or /u/ turn to /a/) and proposes that the /a/ ending variants have verbal category, that is to say, involve the projection of a verbal layer.5

5 This phonological change is similar to a wider process of derivative morphology mentioned in Hualde (1991), and termed Minor Apocope (MA) in De Rijk (2008). It is a lowering process,

111

non-verbal participles in basque and spanish

(82) -tu derived verbs (de Rijk 2008: 150–152; Artiagoitia 2004: 151) neke ‘effort, fatigue’ nekatu ‘to get tired’ aipu ‘citation’ aipatu ‘to cite’ zoro ‘crazy’ zoratu ‘to go crazy’ gorroto ‘hatred’ gorrotatu ‘to hate’ errespetu ‘respect’ errespetatu ‘to respect’ When they are headed by -tu, the Roots end in /a/, showing that they are being used in a verbal environment, namely, in complement position of a verbalizer. The two non-resultative (or unbounded, in terms of Bosque 2014) adjectival participles illustrated in (79) and (80) are also examples of this phonological change. The verb inguratu ‘to surround’ and osatu ‘to constitute’ are derived respectively from the nominal Root inguru ‘surrounding’ and the adjectival Root oso ‘whole, complete’. Both Roots change their final vowel to /a/ when they are used as verbs.

(83)

Root

Verb

Non-verbal participle

a. inguru ‘surrounding’ ingura-tu ‘to surround’ ingura-tu-a/ta/rik b. oso ‘whole, complete’ osa-tu ‘to form’ osa-tu-a/ta/rik

The adjectival participles in (79) and (80) involve the /a/ ending Roots, as well as the -tu suffix, showing that both participles are built on verbs, and not directly on acategorial Roots. In other words, these participles are derived from verbs, and not from Roots. Thus, these pieces of data show that stative participles (those without event implications) can be derived from verbs, supporting the claim made in Anagnostopoulou & Samioti (2014), and against the opposition between stative and resultative participles in terms of the absence or presence of v, as presented in Embick (2004). 3.2 Aspectual Interpretations of the Participle Adjectival (and adpositional) participles typically have the meaning of resultative perfects (Kratzer 2000): they are derived from telic verbs and they have a

affecting non-high vowels of bisyllabic Roots. In the context of verbalizing derivational morphemes (80), however, it can also affect trisyllabic Roots and Roots ending in the high vowel /u/, as can be seen in the examples.

112

berro

meaning where the result of the event holds at the relevant evaluation time. However, in some cases, adjectival (and adpositional) participles in Basque (and sometimes also in Spanish) can have a broader use, and cover meanings pertaining to the experiential perfect, as observed for Basque by Krajewska (2012 2013ab). The experiential perfect can be built on any type of verb. It asserts that the subject has a certain experience related to the fact that the predicate has held at one or more intervals in the relevant evaluation time (Comrie 1976; Iatridou, Anagnostopoulou & Izvorski 2001/2003 among others). Finally, as already advanced in section 3.1.2 and 3.1.3, adjectival (and adpositional) participles can also have a continuous interpretation (Bosque 2014), also called persistent or universal (Comrie 1976; Iatridou et al. 2001/2003; Pancheva 2003), where the predicate holds from a certain point in the past till the relevant evaluation time. In the literature, it has been observed that the universal perfect is obtained from unbounded predicates such as states and progressives. In configurations involving adjectival (adpositional) participles, I have shown that it arises when the participle denotes a non-resultative state—unbounded in terms of Bosque (2014) or stative in terms of Embick (2004) 3.2.1 Resultative Interpretation vs. Continuous Interpretation According to Kratzer (2000), resultant state participles involve an aspectual operator that maps properties of eventualities into properties of times. More specifically, the aspectual operator present in resultant state participles asserts that a property holds at a time interval that is preceded by the running time of a completed event. Thus, the meaning of resultant state participles is similar to the perfect of result. Gehrke (2011) (and also Bosque 2014) notes that the state asserted by the participle in resultative participles does not need to be the result of a change of state, but should be rather regarded as the consequent or subsequent state of a previous event. They are compatible also with activity verbs such as push or pet as long as they get a ‘job is done’ or ‘that’s over’ interpretation (Kratzer 2000). Thus, all telic (or telicized) verbs like accomplishments or verbs allowing an inchoative interpretation are also fine in resultative participles. (84) a. Gutun-a idatz-i-a / irakurr-i-a dago letter-the write-prt-res / read-prt-res be.loc.3sg.abs ‘The letter is written/read.’ b. Mutiko-a haserre-tu-a dago boy-the become.angry-prt-res be.loc.3sg.abs ‘The boy is angered’

non-verbal participles in basque and spanish

113

Kratzer (2000) identifies another type of adjectival participles, different from resultant state participles: target state participles. She borrows both terms from Parsons (1990) and uses them to differentiate between states that are nonreversible (resultant states) from those that are reversible (target states). As shown in Kratzer (2000), target state participles differ from resultant state ones in that they are compatible with the German adverb immer noch ‘still’. According to Kratzer (1998, 2000), target state participles are only compatible with verbs that are lexically specified with an eventive and a stative component. As can be seen in the examples, the adverb oraindik in Basque is only compatible with participles derived from change of state verbs like puztu ‘to inflate’ or ireki ‘to open’, but is non-acceptable with participles built on incremental verbs like irakurri ‘to read’. The same contrasts are observed in Spanish with the adverb todavía ‘still’. (85) a. Puxik-a-k oraindik puz-tu-a-k daude. balloon-the-pl still inflate-prt-res-pl be.loc.3pl.abs ‘The balloons are still inflated.’ b. Ate-a oraindik ireki-ta dago. balloon-the still open.prt-res be.loc.3pl.abs ‘The door is still open.’ (86) *Gutun-a oraindik irakurr-i-ta dago. letter-the still read-prt-res be.loc.3sg.abs ‘The letter is still read.’ In Kratzer’s account (see also Alexiadou et al. 2015), the participial morpheme is meaningless in target state participles; its only function is to categorize the syntactic object as an adjective and to existentially quantify the Davidsonian argument of a predicate that has additionally a target state. Embick (2004) subsumes Kratzer’s target state participles into stative participles—those that denote a characteristic state, without event implications. Nevertheless, Alexiadou et al. (2014) argue, following previous work like Anagnostopoulou (2003) and Alexiadou & Anagnostopoulou (2008), that target state participles can have event implications, and consequently, should be considered to be derived from vP. With respect to unbounded participles understood as in Bosque (2014)— those that are interpreted as underived states, without a causing previous event, see section 3.1.2—, their aspectual interpretation is not similar to the perfect of result, but rather seems to be closer to the continuous (Bosque

114

berro

2014), also called persistent (Comrie 1976) or universal perfect (Iatridou et al. 2001/2003; Pancheva 2003). The interpretation of the universal perfect arises when participles are derived from unbounded predicates like states and progressives, and especially when a durative temporal adverbial like since-or foradverbs are used. (87) I have lived here since 2016. The example in (87) asserts that the property expressed by the predicate (living here) holds since 2016. Thus, the property does not hold at a time following the completion of the event denoted by the predicate, as in the perfect of result, but holds at the same interval as the predicate. The interpretation obtained in adjectival participles derived from nondynamic eventualities like vigilar ‘to guard’ or perseguir ‘to chase’ in Spanish, or extent verbs and verbs expressing physical properties is similar. (88) Un criminal perseguido por la policía a.masc criminal chased.masc by the police ‘A criminal chased by the police’ (Bosque 2014: 51) As pointed out in Bosque (2014), the aspectual interpretation of (88) is continuous: the state predicated of the theme does not hold after the event denoted by the verb is completed, but the interval in which the state holds is simultaneous with the running time of the event. 3.2.2 Experiential Interpretation As noted at the beginning of section 3.2, Basque and Spanish adjectival participles can also have experiential interpretation. This interpretation is particularly frequent in Basque (Hualde et al. 1994; Krajewska 2012 2013ab), when the theme of predication is correferent with the subject initiating the event embedded in the participle. Nonetheless, Basque is not alone in this aspect since Spanish adjectival participles can also license experiential interpretation in particular circumstances, as pointed out in Bosque (2014). In Basque, in the majority of configurations where an adjectival or adpositional participle is used, the state denoted by the participle is predicated about the direct object or the intransitive subject of the verb underlying the participle.

non-verbal participles in basque and spanish

115

(89) Mahai-a apur-tu-ta dago. table-the break-prt-res be.loc.3sg.abs ‘The table is broken.’ (90) Aulki-a jaus-i-ta dago. chair-the fall-prt-res be.loc.3sg.abs ‘The chair is fallen.’ This interpretation is highly predominant when the verb embedded in the participle is a transitive or intransitive change of state verbs, like in (89) and (90). Actually, a participle derived from a change of state verb like hondatu ‘break’ is quite hard with a frequency adverb like askotan ‘many times’. (91) ??Auto-a askotan honda-tu-ta dago. car-the many.times break-prt-res be.loc.3sg.abs ‘The car is broken many times.’ In contrast, when the participle is predicated about the subject initiating the event, the experiential or perfect readings of the participle arises naturally. The experiential interpretation contrasts with the resultative one in that the outcome of the event does not necessarily hold at the relevant evaluation time. Rather than the consequent state, it is the event itself that is focused. These are some examples taken from Krajewska (2013): (92) Meza-ra? —ez, entzun-ik nago. mass.the-all? no, listen.prt-res be.loc.3sg.abs ‘To mass?—no, I have already listened’, lit. ‘I am already listened.’ (Irigaray) (93) —Ezagu-tzen zenuen hiri? —ez; ez —know-imprf have.2sg.erg.1sg.abs.pst city? No; no nintzen inoiz izan-a. be.1sg.abs.pst never be.prt-res ‘—Did you know the city?—no, I had never been there.’, lit. ‘I was never been.’ (Satrustegi) (94) Ni askotxo ikus-i-a nago. I.abs a.lot see-prt-res be.loc.1sg.abs ‘I have seen many things’, lit. ‘I am a lot seen.’ (Anabitarte)

116

berro

(95) Asko-tan ene baitan erran-ik nago: “Zenbat itz A.lot-ine my ine say.prt-res be.loc.1sg.abs: “how.many words deusik ez errateko”. nothing no to say” ‘I have said many times to myself: “How many words to say nothing”.’, lit. ‘I am a lot said many times to myself: …’ (Irigaray) In examples (92), (94) and (95), the theme of predication (the absolutive argument) corresponds to the subject of the previous event (of ‘listening’ in (92), of ‘seeing’ in (94) and of ‘saying’ in (95)), so that the action affects mainly the state held by the subject (Krajewska 2013). As a consequence of being the subject of the event (once or more times), the theme of predication accumulates an experience. The experiential interpretation is particularly salient when the copula is in the past tense, and moreover when the copula is *edun or eduki ‘have’ instead of izan ‘be’. The latter configuration corresponds to the possessive resultative commented in section 2.3. As mentioned, this configuration has been attested diachronically in many languages as the source of the perfect. For instance, in the past tense, the configuration has the meaning of a past perfect (Hualde et al. 1994: 145–146); a past event completed before some other past event. (96) Etxe-tik joan-a nintzen (zu iritsi home-abl go.prt-res be.1sg.abs.pst (you.abs arrive.prt zinenerako). be.2sg.abs.pst-by.the.time) ‘I was gone by the time you arrived.’ (97) Pelikula ikus-i-ta daukat. film.the see-prt-res have(eduki).3sg.abs.1sg.erg ‘I have the film seen.’ The fact that these participles in Basque can have experiential interpretation can suggest that these are really perfect constructions, where the participles would be verbal rather than adjectival or adpositional. Krajewska (2012 2013ab) actually claims that Basque -a/-ta and -rik ending participles are in the middle of the grammaticalization path towards a new perfect. In her diachronic study, Krajewska mentions that, in the historical evolution of the Basque language in the last centuries, the majority of verbs used in the periphrases involving -a/-ta and -rik participles are telic change of state verbs where the state is predicated about the object of the event or the subject of an intransitive event. Nevertheless, in the cases where other type of predicates are used—

non-verbal participles in basque and spanish

117

such as mental, perception and communication verbs, as well as stative and atelic verbs—perfect and experiential interpretations have become more common in the last centuries. According to her, these configurations have not yet become full-fledged perfects, as the frequency of the latter type of verb classes has not increased in comparison to change of state verbs, which are still highly predominant. Krajewska then concludes that the configuration involving -a/-ta and -rik participles is still not a perfect, but that it is in the middle of a grammaticalization path, and that it is acquiring properties usually attributed to the perfect. This issue raises a number of questions regarding the structural composition of -a/-ta and -rik participles. Are -a/-ta and -rik participles with resultative meaning structurally different from those with experiential meaning? In relation to this, is there a categorial difference between resultative and experiential participles? Do the former ones occur in bi-clausal configurations and the latter ones in monoclausal ones? As for the first question, my suggestion is that there has to be some structural and/or semantic difference between them, as they have different interpretations. Regarding the categorial or biclausal/monoclausal configuration, it seems that even if -a/-ta and -rik participles get experiential interpretation, they are still in a bi-clausal structure. The bi-clausal analysis of experiential participles is supported by two facts: (i) they can occur with the copulas egon ‘(the stage-level) be’ and eduki ‘have’, which are never used as auxiliaries, and (ii) they combine with intransitive copulas (izan ‘be’ and egon ‘(the stage-level) be’) also when the underlying verb of the participle is transitive or unergative (92) (94) (95), a fact that is not attested in perfect monoclausal configurations. Be it as it may, Basque -a/-ta and -rik participles are not alone in the ability to get experiential interpretation. As a matter of fact, Bosque (2014) shows that accompanied by frequency adverbs, attributive participles can have experiential interpretations. (98) Un coche averiado en muchas ocasiones no es una buena a.masc car broken.masc in many occasions no is a good inversión. investment ‘A car many times broken down in not a good investment.’ (Bosque 2014: 51) Participles occurring in predicative position with the copula estar ‘(the stagelevel) be’ can also have this interpretation in certain cases, although it is not a generalized tendency.

118

berro

(99) Este ordenador está arreglado varias veces. this.masc computer be.loc.3g fixed.masc several times ‘This computer is fixed several times.’ (100)

?? Este

coche está averiado varias veces. this.masc car be.loc.3g broken.masc several times ‘This computer is fixed several times.’

In the examples (98) and (99), the participle has experiential interpretation, rather than resultative, since a car cannot be in a state of being broken many times and a computer cannot be in a state of being fixed many times. Still, the experiential interpretation of participles in this context is not always available (100). It is clear that the experiential interpretation is not as easily obtained as in perfect clauses. For instance, in a perfect clause in Spanish, we can get an experiential interpretation when the participle is derived from a stative verb (e.g. estar enfermo ‘be sick’) (101) He estado enfermo antes. have.1sg been sick.masc before ‘I have been sick before.’ Participles occurring in predicative position accompanied by the copula estar cannot be built on stative verbs (unlike in Basque (93)), and thus, they cannot have an experiential interpretation. (102) *María está estada enferma Mary be.loc.3sg been.fem sick.fem (103) *Este asunto está conocido por todos this issue be.loc.3sg known.masc by all ‘This issue is known by everybody.’ As in Basque, if the participle combines with the transitive copula tener ‘have’, an experiential interpretation is much easier to obtain. (104) Tengo la película vista have.1sg the.fem film seen.fem ‘I have the film seen.’

non-verbal participles in basque and spanish

119

(105) Tengo el libro leído have.1sg the.masc book read.masc ‘I have the book read.’ McFadden & Alexiadou (2010) analyze the structural differences between experientials and resultatives, particularly in the context of German and older English have and be auxiliary alternation (see also Alexiadou 2015). They propose that, in resultative adjectival participles, the verb is headed by a resultative Asp(ectual) head that introduces a state that is the result of a prior event. AspP is, in turn, the complement of the copular verb -be-. In experientials, on the other hand, the inflected element -have- is an auxiliary located in an aspectual Perf(ect) head, which is below T, and which selects an Asp head. In their account, this configuration has the meaning of a perfect.

4

Event-Related Modification

In this section, I will deal with the modification of the event underlying the participle. The ability of adjectival participles to accept different types of eventrelated modifiers is used in the literature as a diagnostic to find out the structural composition of adjectival participles; e.g. to test the presence/absence of a verb phrase (vP), a Voice phrase and also an aspectual (Asp) phrase (see also section 5.2). The fact that there are different classes of adjectival participles has been noted from early on. Starting from Wasow (1977), deverbal -ed forms in English have been claimed to be formed at different levels of the language architecture. Wasow (1977) proposed that adjectival passives are created in the lexicon, whereas verbal passives are created in the syntax. Adding a further distinction to this two-way classification, Kratzer (1994) claimed that adjectival passives could be classified into two groups: phrasal passives and lexical passives. According to her, adjectival participles, which in German are distinguished by the use of the auxiliary sein ‘be’, are headed by an adjectival passive morpheme that may be merged into the structure at different positions. In phrasal passives, the adjectivizing head merges to the verbal phrase (VP in her account), whereas in lexical passives, it merges to the verb (V). As a consequence, the resulting participle has different syntactic properties, and in this way, she accounts for the contrasting behavior between the two, for example in allowing manner adverbials and the ability to undergo -un prefixation. On the one hand, phrasal adjectival participles allow modifying adverbs (106a)—unlike underived adjectives (106b)–.

120

berro

(106) a. Das Haar war ziemlich schlampig gekämmt. The hair was rather sloppily combed ‘The hair was rather sloppily combed.’ b. *Das Haar war ziemlich schlampig fettig The hair was rather sloppily greasy ‘*The hair was rather sloppily greasy’ On the other hand, lexical participles can be negated by the un- suffix and cannot be combined with manner adverbs. Thus, lexical participles pattern with underived adjectives and contrast with phrasal participles. (107) *Das Haar war hässlich ungekämmt The hair was ugly uncombed ‘*The hair was ugly uncombed’ Another test that is used to explore the structural composition of adjectival participles concerns the temporal and spatial location of the event embedded in the participle. If participles are compatible with temporal and spatial adverbs that locate the event at a time and place that are different from those of the state, that means that the event is actually instantiated (Gehrke 2011, 2015). A modifier like recently or a time-frame adverb like three days ago can be used with this purpose. Gehrke (2011: 246) shows that in German adjectival participles, recently can only modify the state rather than the event (108), and that the adverb three days ago is non-acceptable (109) (von Stechow 1998; Rapp 1996, 1997). (108) Die Tür war kürzlich geöffnet. the door was recently opened ‘The door was in the opened state recently, but probably no longer.’ (NOT: the door is in the opened state, the opening took place recently) (109) a. *Der Computer ist vor drei Tagen repariert the computer is before three days repaired Intended: ‘The computer is repaired three days ago’ b. Der Computer ist seit drei Tagen repariert. the computer is since three days repaired

non-verbal participles in basque and spanish

121

As can be seen, only a temporal adverb that modifies the state, like seit drei Tagen ‘since three days ago’, is compatible with German adjectival participles. In the next sections, I will argue that Basque and Spanish are more flexible than German adjectival participles in this aspect, as they are compatible with this kind of modification. 4.1 Event-Related Modification in Basque and Spanish Like in German, adjectival (and adpositional) participles in Basque are compatible with manner adverbs. As can be seen in the examples below, Basque resultative participles are compatible with modifiers like arretarik gabe ‘sloppily’, kontu handiz ‘carefully’ an eskuz ‘by hand’. (110) a. Auto-a kontu handi-z konpon-du-a/-ta/-rik car-the.abs care big-instr fix-prt-res dago. be.loc.3sg.abs ‘The car is fixed carefully’ b. Auto-a arreta-rik gabe konpon-du-ta car-the.abs attention-part without fix-prt-res dago. be.loc.3sg.abs ‘The car is fixed sloppily’ c. Horma hau edozein modu-tan margo-tu-ta dago. wall this.abs any way-ine fix-prt-res be.loc.3sg.abs ‘This wall is painted sloppily’ d. Estatua hori esku-z egin-a/da/ik dago. statu that.abs hand-instr do.prt-res be.loc.3sg.abs ‘This statu is done by hand’ The compatibility with manner modifiers show that adjectival and adpositional participles in Basque are actually phrasal, derived from a vP which denotes an event. Spanish behaves similarly, as can be observed in the examples below, where the adjectival participle combines with a manner adverb ending in -mente. (111) Un edificio concienzudamente vigilado a.masc building thoughfully guarded.masc (Bosque 2014: 54, ex. 21b)

122

berro

(112) (Se nota que) este árbol está cortado (it isnoted that) this.masc tree be.loc.3sg cut.masc rápidamente. rapidly ‘It is noted that this tree has been cut rapidly.’, lit., ‘this tree is cut rapidly’ (113) El pelo está peinado descuidadamente. the.masc hair be.loc.3sg combed.masc sloppily ‘The hair is combed sloppily,’ The manner modifiers used in the above examples can be argued to be describing some aspect of the event that is relevant in the state denoted by the participle. Actually, in example (112) the initial se nota que ‘it is noted that’ has been introduced in order to make the sentence more natural. However, in other cases, the adverb seems to be modifying the event itself, rather than the state. (114) Un niño injustamente castigado a.masc kid unfairly punished.masc (adapted from Bosque 2014: 54, ex. 21a) Contrasting with German, both Basque and Spanish allow temporal and spatial modification of the event embedded in the participle. (115) Aspaldi geldi-tu-ta gaude. long.ago meet-prt-res be.loc.1pl.abs ‘We did an appointment long ago.’, lit. ‘we are done an appointment long ago.’ (116) a. Pastel hori duela hiru egun egin-a/-da/-ik dago. cake this.abs ago three days do.prt-res be.loc.3sg.abs ‘That cake is done three days ago.’ b. Pastel hori atzo goiz-ean egin-a/-da-/ik cake this.abs yesterday morning-ine do.prt-res dago. be.loc.3sg.abs ‘That cake is done yesterday morning.’

non-verbal participles in basque and spanish

123

(117) a. Pastel hori Lourdes-en etxe-an egin-a/-da/-ik cake this.abs Lourdes-gen house-ine do.prt-res dago. be.loc.3sg.abs ‘This cake is done in Lourdes’ house.’ b. Makina hau Danobat-en egin-a/-da/-ik dago. machine this.abs Danobat-ine do.prt-res be.loc.3sg.abs ‘This machine is done in Danobat’ Temporal and spatial modification of the event is also possible in Spanish adjectival participles, both in predicative and attributive position. As can be seen in examples (119), (120) and (121), the adjectival participle occurs with the copula estar and it is modified by temporal and spatial adverbs that specify when and where the event (bringing the state asserted by the participle) has taken place. (118) El ordenador está arreglado recientemente. the.masc computer be.loc.3sg fixed.masc recently ‘The computer is fixed recently.’ (119) El ordenador está arreglado hace poco en la tienda de the computer be.loc.3sg fixed.masc recently in the shop of un amigo mío. a friend mine ‘The computer is fixed recently in the shop of a friend of mine.’ (120) Este edificio está construido en 1800. this.masc building be.loc.3sg built.masc in 1800 ‘This building is built in 1800.’ (121) Un artículo publicado ayer en la prensa de la an.masc article published.masc yesterday in the press of the tarde. evening ‘An article published yesterday in the evening press.’ (Bosque 2014: 48, ex. 10b) Thus, both Basque and Spanish allow temporal and spatial modification of the event. In this aspect, they patter with Greek, which, unlike German, also accepts temporal adverbs like three days ago and spatial modification.

124

berro

(122) a. To pc itan diorthomeno prin tris meres. the pc was repaired three days ago ‘The pc was repaired three days ago.’

Greek

b. To pedi itan htenismeno sto banio. the child was combed in.the bathroom ‘The child was combed in the bathroom.’ (123) a. *Der Computer ist vor drei Tagen repariert. the computer is before three days repaired ‘The computer is repaired three days ago.’ b.

German

??? Das

Kind war im Badezimmer gekämmt. the child was in.the bathroom combed ‘The child was combed in the bathroom.’

4.2 Presence of Voice In the study of adjectival participles, another aspect that has generated much discussion is related to whether adjectival participles do actually involve the projection of Voice, the head introducing the external argument (Kratzer 1994, 1996). Kratzer (1994) claimed that adjectival participles are different from verbal ones precisely in this aspect: unlike verbal passive participles, adjectival ones are not VoiceP-derived, and thus, an implicit external argument, e.g. an agent, is totally absent from them. Later works have, nevertheless, questioned this claim. Anagnostopoulou (2003) and Alexiadou & Anagnostopoulou (2008) showed that VoiceP can be projected in Greek adjectival participles. For instance, in Greek resultant state participles, an agent can be explicitly introduced in an apo-phrase, similar to a by-phrase in English. (124) To psari itan tiganismeno apo tin Maria the fish was fried by the Mary ‘The fish was fried by Mary’

Greek

Recent studies on English and German have argued that VoiceP can be involved also in English and German adjectival participles (e.g. Alexiadou et al. 2014; Alexiadou et al. 2015 for German; McIntyre 2013; Bruening 2014 and Alexiadou et al. 2014 for English). In the literature, the absence of the external argument in adjectival participles has been analyzed using several diagnostics, such as the inability to control into purpose clauses, and the absence of

non-verbal participles in basque and spanish

125

the disjoint reference effect. Nevertheless, the lack of the disjoint reference effect as a test to prove the absence of an implicit external argument in German (Kratzer 1994) has been called into question in McIntyre (2013), Alexiadou et al. (2014) and Alexiadou et al. (2015). Particularly, it seems that when nonreflexive verbs are used, adjectival participles give rise to the disjoint effect, just like verbal passive participles, so that this diagnostic does not prove that VoiceP is not projected in German adjectival participles. On the other hand, the inability of adjectival passives to control into purpose clauses in German and English has also been counter-evidenced in McIntyre (2013) and Alexiadou et al. (2014). In the following two subsections, I will provide several pieces of data that indicate than an (implicit) external argument is also present in adjectival (and adpositional) participles in Basque and Spanish. Particularly, I will show that adjectival (and adpositional) participles are compatible with modifiers that rely on the presence of an agent. 4.2.1 Subject-Oriented Modification and Control into Purpose Clauses In both Basque and Spanish, adjectival (and adpositional) participles are compatible with controlled purpose clauses, where the implicit external argument controls the verb in the purpose clause (125ab). Additionally, agent-related modifiers nahita ‘on purpose’ in Basque (126a) and a propósito ‘on purpose’ in Spanish (126b) are also naturally admitted. (125) Control into purpose clauses a. Horma-k beltz-ez margo-tu-ta daude wall.the.abs-pl black-instr paint-prt-res be.loc.3pl.abs gela ilunago bihurtze-ko. room.the.abs darker turn-dest ‘Walls are painted in black in order to make the room darker.’ b. Las paredes están pintadas de negro para the.fem.pl wall.pl be.loc.3pl painted.fem.pl of black for hacer la habitación más oscura. make the room more dark ‘The walls are painted in black in order to make the room darker.’ (126) Modification with on purpose a. Auto-a nahita konpon-du-ta dago. car-the.abs on.purpose fix-prt-res be.loc.3sg.abs ‘The car is fixed on purpose’

126

berro

b. Horma-k nahita daude beltze-z wall.the.abs-pl on.purpose be.loc.3pl.abs black-instr margo-tu-ta. paint-prt-res ‘Walls are painted in black on purpose’ c. El coche está arreglado a propósito. the.masc car be.loc.3sg fixed.masc to purpose ‘The car is fixed on purpose.’ d. Las paredes están pintadas de negro a the.fem.pl wall.pl be.loc.3pl painted.fem.pl of black to propósito. purpose ‘The walls are painted in black on purpose.’ 4.2.2 Acceptance of Ergative Subjects and por-phrases As advanced in section 2.3, in Basque adjectival participles, agent arguments can occur explicitly marked with ergative case, the same case assigned to transitive and unergative subjects in monoclausal configurations. The ergative subject occurring with adjectival participles has several restrictions, though, that clearly differentiate them from the ergative subjects of monoclausal configurations. (127) Gutun-a guraso-ek sina-tu-a da. letter-the parent-erg.pl firm-prt-res be.3sg.abs ‘The letter is firmed by the parents.’ (128) Puxika-k Ane-k puz-tu-a-k dira. balloon.the-pl Anne-erg inflate-prt-res-pl be.3pl.abs ‘The balloons are inflated by Anne.’ The acceptance of an ergative subject interpreted as the agent of the event embedded in the participle is favored when the copula selected is izan ‘be’ (instead of egon ‘(the stage level) be’) and the resultative suffix is the adjectival -a. In this case, the ergative subject does not agree with the copula and it cannot scramble to the front of the theme of predication or to the right of the copula (Ortiz de Urbina & Uribe-Etxebarria 1991).

non-verbal participles in basque and spanish

127

(129) Ziurtagiri-a zuzendari-a-k sina-tu-a da. certificate-the.abs director-the-erg sign-prt-res be.1sg.abs ‘The certificate is signed by the director’ (130) (*Zuzendari-a-k) ziurtagiri-a (zuzendari-a-k) (*director-the-erg) certificate-the.abs (director-the-erg) sina-tu-a da (*zuzendaria-a-k) sign-prt-res be1sg.abs (*director-the-erg) ‘The certificate is signed by the director’ An agent argument can also be introduced in Spanish with a por-phrase in participles with both a bounded and unbounded interpretation, in terms of Bosque (2014). As can be seen, por-phrases are acceptable in both attributive and predicative participles. With bounded interpretation, we find the following examples: (131) Ladrones atrapados por la policía thieves caught.masc.pl by the police (132) Aviones derrivados por el enemigo aircrafts shot.down.masc.pl by the enemy (133) Un niño castigado por su maestro a.masc child punished.masc by his teacher (134) Este niño está castigado por su maestro this.masc child be.loc.3sg punished.masc by his teacher (135) Este edificio parece diseñado cuidadosamente por un this.masc building looks designed.masc carefully by an architecto surrealista en una noche de insomnia architect surrealist in a night of insomnia ‘This building seems designed by a surrealist architect in a sleepless night.’ (Bosque 2014: 54, ex. 22) And these with unbounded interpretation: (136) Un criminal perseguido por la policía a.masc criminal chased.masc by the police

128

berro

(137) Un apartamento habitado por reciéncasados an.masc apartment inhabited.masc by newlyweds (138) El tejado está protegido por una gruesa capa the.masc roof be.loc.3sg protected.masc by a thick layer de paja of straw ‘The roof is protected by a thick layer of straw’ (Bosque 2014: 65, ex. 50a) The data provided clearly indicates that Voice is projected in both Basque and Spanish. Nevertheless, the acceptability of por-phrases in Spanish is apparently more restricted than the ergative subjects in Basque, as noted by Gehrke & Marco (2014). According to them, por-phrases are grammatical when they modify the state denoted by the participle, rather than the event (i.e. when the participle has unbounded interpretation) or, alternatively, when they modify an event-kind, rather than an event token (see section 5). These restrictions do not seem to hold in the Spanish variety spoken in the Basque Country, though.

5

Event-Related Modification: Restrictions

In section 4, I have shown that Basque and Spanish adjectival (and adpositional) participles are compatible with event-modifiers of different sorts. They accept manner modifiers that describe the event embedded under the participle (such as kontu handiz ‘carefully’ in Basque and cuidadosamente ‘carefully’ in Spanish), and allow temporal and spatial adverbs that locate the event at a different time and place from those of the state denoted by the participle. Additionally, both Basque and Spanish accept agent-oriented modifiers and are compatible with controlled purpose clauses, facts that point out that an implicit external argument is syntactically active in Basque and Spanish adjectival (and adpositional) participles. Finally, explicit ergative arguments that are interpreted as the agent of the event embedded in the participle are also allowed in Basque, although these arguments are much better accepted when the copula is izan ‘be’ and the participial ending is the adjectival -a (and not -ta or -rik). In Spanish, too, por-phrases are accepted in both attributive and predicative adjectival participles, a fact which indicates that Voice is projected within the participle. Nevertheless, as we are going to see in the next subsections, the occurrence of por-phrases is not apparently totally unrestricted.

non-verbal participles in basque and spanish

129

Gehrke (2011 and subsequent work) observes that by-phrases in German are possible as long as they belong to the consequent state, rather than to the underlying event, or if they have an impact on the underlying event that is still visible in the consequent state. She refers to the latter event as an event kind. In her analysis, adjectival participles denote consequent states of event kinds. Event kinds cannot be modified by temporal or spatial adverbs because event kinds lack spatiotemporal manifestation. Thus, the only modifiers that are accepted are those which describe the consequent state, or otherwise, are interpreted as kind-related; modifiers that create a new subkind of the event by narrowing the event kind, and thus, having an impact on the consequent state. The type of manner adverbs allowed in Spanish adjectival participles seem to point in this direction (section 4.1). Manner adverbs like cuidadosamente ‘carefully’, descuidadamente ‘sloppily’ are adverbs that, although they describe the event, shape also the form of the consequent state. In the case of rápidamente ‘quickly’ in (112), the adverb is better accepted with the introduction of se nota que ‘it is noted that’ at the beginning of the example, another piece of evidence which shows that the relevance of the event-manner in the consequent state is crucial for the acceptance of this adverb. Nevertheless, we have seen that Spanish, like Basque, allows spatiotemporal modification of the event, something that is unexpected if they allowed only event-kind modifiers. 5.1 Restricted por-phrase in Spanish: Gehrke & Marco (2014) Agentive por-phrases in Spanish seem to be better when they belong to the consequent state denoted by the participle, rather than the event, or when they modify an event kind. Gehrke & Marco (2014) present a corpus study on Spanish that precisely supports this claim. They show that there is a clear contrast between the type of por-phrases found in adjectival participles and the por-phrases of verbal participles. In adjectival participles, por-phrases involve weak or non-referential nominals such as indefinites, bare nominals, weak and generic nominals, whereas, in verbal participles, the nominals embedded in por-phrases are more strongly referential, e.g. proper names, pronouns, demonstratives, regular definites. Additionally, they argue that the complements of por-phrases in Spanish cannot have wide scope, cannot introduce discourse referents, cannot be modified by common intersective modification and do not involve strong determiners (Gehrke & Marco 2014: 13).

130

berro

(139) a. Todos los cuadros estaban all.masc.pl the.masc.pl paintings be.loc.3pl.imperf.pst pintados por un niño. painted by a child ‘All the paintings were painted by a child.’ (>1 child possible) NOT: ‘There was a particular child that painted all the paintings’ b. El cuadro estaba pintado por [un the.masc painting be.loc.3sg.impr.pst painted.masc by a niño]1.*pro1 era pelirrojo. child was red-haired Intended: ‘The painting was painted by a child. He had red hair’. c. *El cuadro estaba pintado por the.masc painting be.loc.3sg.imperf.pst painted.masc by un niñ o pelirrojo a child red-haired Intended: ‘The painting was painted by a red-haired child’. d. *El cuadro estaba pintado por the.masc painting be.loc.3sg.imperf.pst painted.masc by este niño this child Intended: ‘The painting was painted by this child’ Finally, the study also shows that the contrasts found between por-phrases that belong to the event in verbal and adjectival participles do not find a correlate in the por-phrases that pertain to the consequent state. Below, an example that shows that a por-phrase modifying a state can have a strong determiner. (140) Este edificio estaba vigilado por este this.masc building be.loc.3sg.imperf.pst guarded.masc by this hombre de chaqueta negra. man of jacket black ‘This building is guarded by this man with black jacket.’ 5.2 Structural Motivation for the Event Kind vs. Event Token Distinction Gehrke (2011) and following works (Gehrke & Marco 2014; Gehrke 2015; Alexiadou et al. 2014; Alexiadou et al. 2015) claim that adjectival participles in languages like Spanish and German are interpreted as consequent states of event

non-verbal participles in basque and spanish

131

kinds because in these languages, the verb does not get further embedded into verbal projections, but is directly adjectivized. They generalize to verbs the analysis of nominals (Carlson 1977; Zamparelli 1995; Chierchia 1998; Dayal 2004 among others), where it is considered that nouns start out like predicates of kinds and get instantiated when they are embedded under Number. In the verbal domain, Gehrke (2015) proposes that events also enter the derivation as event kinds and become instantiated when they are embedded under further verbal projections, like Aspect or T. At this level, they obtain the interpretation of event tokens, that is, events that take place in the actual world. Since in languages like German and Spanish, participles are not further embedded under Aspect or T, the interpretation of the event remains in the kind domain, and is not interpreted as an instantiated event. According to the authors, the restrictions on event-modification, for example in languages like Spanish, follow from the fact that the event remains as an event kind. State-modifiers of a state token are generally accepted, because the adjectivized participle is further embedded under T, and thus, the consequent state gets instantiated. 5.3 Event Tokens in Basque Non-verbal Participles In Basque, ergative subjects introduced in adjectival participles are not restricted, a fact that suggests that the type of event projected in Basque adjectival participles is not an event kind but an event token, i.e. it is actually instantiated. Additionally, we have shown that Basque adjectival and adpositional participles allow event-related temporal and spatial modification. Within Gehrke’s (2015) analysis, this would mean that, in Basque, participles are embedded under further verbal projections before being categorized as adjectives or adpositions. For Greek adjectival participles, which show a similar distribution of that of Basque, Alexiadou et al. (2014) have proposed that vP/VoiceP is selected by an aspectual operator (ASP PERF) and then by an adjectivizing PASS(ive) head.6 A similar analysis can be proposed for Basque (see Berro 2018), with the difference that in Basque the adjectivizing head does not absorb the external argument, given that it can be explicitly introduced in the clause. Still, another analysis is needed for the cases in which the adjectival or adpositional participle is interpreted as experiential.

6 In Alexiadou et al. (2015), the PASS head is substituted by an a(dj) head that has the categorizing function.

132

berro

5.4 Dialectal Variation in Spanish In relation to the facts in Basque, my perception looking at the Spanish examples in (139)—and confirmed by other speakers—is that the Spanish spoken in the Basque Country is more flexible in allowing a wider range of por-phrases. For instance, similar examples to those in (139bcd) but in the present tense (e.g. (141)) are grammatical for me and other consulted speakers (Spanish/Basque bilinguals living in the Basque Country). In the same way, in example (141a) both interpretations are obtained. Additionally, the fact that Spanish adjectival participles are fine with temporal and spatial modification also seems to suggest that events in adjectival participles in Spanish pattern more with Basque and Greek, rather than with German. This behavior of Spanish, that may be restricted to the Spanish spoken in Basque Country, may be affected by the contact between Basque and Spanish. In any case, the restrictions on manner modifiers still hold, a fact that would remain unexplained if we considered that, in the Spanish variety spoken in the Basque Country, events are instantiated in adjectival participles. As observed in section 4.1, manner modifiers in Spanish adjectival participles are better if they describe an event kind, in other words, if they are relevant for the consequent state. In principle, if events were instantiated in Spanish adjectival participles, as suggested in in this section, that restrictions would be unexpected. Further investigation is needed on this issue. Spanish spoken in the Basque Country: (141) a. Todos los cuadros están pintados por un all.masc.pl the.masc.pl paintings be.loc.3pl painted by a niño. child ‘All the paintings are painted by a child.’ (>1 child possible) ‘There was a particular child that painted all the paintings’ b. El cuadro está pintado por [un the.masc painting be.loc.3sg painted.masc by a niño]1.pro1 era pelirrojo. child was red-haired ‘The painting was painted by a child. He had red hair’.

non-verbal participles in basque and spanish

133

c. El cuadro está pintado por un niño the.masc painting be.loc.3sg painted.masc by a child pelirrojo. red-haired ‘The painting was painted by a red-haired child’. d. El cuadro está pintado por este niño. the.masc painting be.loc.3sg painted.masc by this child ‘The painting was painted by this child’

6

Conclusions

In this chapter, non-verbal participles in Basque and Spanish have been compared and contrasted, in light of the literature on adjectival participles. I have shown that non-verbal participles differ in these languages in their morphological make-up: while Spanish adjectival participles are formally similar to verbal passive participles, in Basque, an additional participial suffix (-a, -ta or -rik) is added. -A participles have been considered adjectival, following the Basque linguistic tradition, and -ta and -rik ones have been regarded as adpositional. Non-verbal participles in both languages also differ in that Basque non-verbal participles can give rise to experiential interpretations, which is why Krajewska (2012, 2013ab) claims that Basque -a/-ta/-rik “resultative” participles are in the middle of a grammaticalization path towards a new perfect. Additionally, Basque adjectival participles also contrast with Spanish ones in that Basque participles allow ergative subjects to be present, which denote the agent of the event embedded under the participle. Despite these differences, this chapter has made visible that Basque and Spanish non-verbal participles are similar in a number of other aspects. For instance, the same interaction between the lexical aspect of the event underlying the participle and the aspectual interpretation of the participle is observed in both languages, with non-dynamic events giving rise to a non-resultative interpretation of the participle. On the other hand, even though Spanish adjectival participles have been argued to admit only certain type or por-phrases, and to involve, consequently, an event kind (and not an event token), this chapter has shown that, at least in the Spanish spoken in the Basque Country, adjectival participles allow temporal and spatial modification of the event underlying the participle and por-phrases with wide scope, strong determiners and discourse referents. These data indicate that this language variety may be similar to Basque in having non-verbal participles with an event token interpretation.

134

berro

Acknowledgements I want to thank Ane Odria for sharing with me her linguistic intuitions and Beatriz Fernández for her comments on the manuscript. The research leading to these results has received funding from the European Union’s Seventh Framework Programme for research, technological development and demonstration under grant agreement no. 613465. Moreover, this study has been developed thanks to several projects funded by the Basque Government (the post-doctoral grant (POS_2015_1_0086 and POS_2016_2_0023), and the project IT665–13) and by the Spanish Ministry of Economy and Competitiveness (FFI2014–51878-P).

References Alexiadou, Artemis. 2015. “On the irrealis effect on auxiliary selection”. In Rolf Kailuweit & Malte Rosemeyer (eds.), Auxiliary Selection Revisited: Gradience and Gradualness. Berlin: de Gruyter. 123–144. Alexiadou, Artemis & Elena Anagnostopoulou. 2008. “Structuring participles”. In Charles Chang & Hannah Haynie. Somerville (eds.), Proceedings of the WCCFL 26. MA: Cascadilla Press. 33–41. Alexiadou, Artemis, Elena Anagnostopoulou & Florian Schäfer. 2015. External Arguments in Transitivity Altenations. A Layering Approach. Oxford: Oxford University Press. Anagnostopoulou, Elena. 2003. “Participles and Voice”. In Artemis Alexiadou, Monika Rathert & Arnim von Stechow (eds.), Perfect explorations. Berlin: Mouton de Gruyter. 1–36. Alexiadou, Artemis, Berit Gehrke & Florian Schäfer. 2014. “The argument structure of adjectival participles revisited”. Lingua 149: 118–138. Anagnostopoulou, Elena & Yota Samioti. 2014. “Domains within words and their meanings: A case study”. In Artemis Alexiadou, Hagit Borer & Florian Schäfer (eds.), The Syntax of Roots and the Roots of Syntax. Oxford: Oxford University Press. Artiagoitia, Xabier. 2004. “The case of an enlightening, provoking and admirable Basque derivational suffix with implications for the theory of argument structure”. Supplements of ASJU XLVI: 147–183. Baker, Mark C. 2003. Lexical Categories. Verbs, Nouns and Adjectives. Cambridge: Cambridge University Press. Berro, Ane. 2015. Breaking verbs. From event structure to syntactic categories in Basque. Ph.D. diss., UPV/EHU and Université Bordeaux Montaigne. Berro, Ane. 2017. “Resultative and non-resultative participles in Basque”. Ms, University of the Basque Country & SFL (UMR 7023, CNRS/Paris 8).

non-verbal participles in basque and spanish

135

Bosque, Ignacio. 1999. “El sistema adjectival. Complementos y modificadores del adjetivo. Adjetivo y participio”. In Ignacio Bosque & Violeta Demonte (eds.), Gramática descriptiva de la lengua española. vol. 1. Madrid: Espasa Calpe. 217–310. Bosque, Ignacio. 2014. “On resultative past participles in Spanish”. Catalan Journal of Linuistics 13: 41–77. Bruening, Benjamin. 2014. “Word formation is syntactic: adjectival passives in English”. Natural Language and Linguistic Theory 32: 363–422. Carlson, Greg. 1977. Reference to kinds in English. Ph.D. diss., University of Massachusetts at Amherst, published by Garland, New York. Chierchia, Gennaro. 1998. “Generics and concepts”. Natural Language Semantics 6: 339– 405. Comrie, Bernard. 1976. Aspect. An Introduction to the Study of Verbal Aspect and Related Problems. Cambridge: Cambridge University Press. Dayal, Veneeta. 2004. “Number marking and (in)definites in kind terms”. Linguistic and Philosophy 27: 393–450. Demonte, Violeta. 1983. “Pasivas léxicas y pasivas sintácticas en español”. Serta Philologica F. Lázaro Carreter. Madrid: Cátedra. 141–157. Detges, Ulrich. 2000. “Time and truth: the grammaticalization of resultatives and perfects within a theory of subjectification”. Studies in Language 24: 345–377. Doron, Edit. 2014. “The interaction of adjectival passive and voice”. In Artemis Alexiadou, Hagit Borer & Florian Schäfer (eds.), The Syntax of Roots and the Roots of Syntax. Oxford: Oxford University Press. Eguren, Luis. 2012. “Predication Markers in Basque”. In Urtzi Etxeberria, Ricardo Etxepare & Myriam Uribe-Etxebarria (eds.), Noun Phrases and Nominalization in Basque: Syntax and Semantics. John Benjamins Publishing Company. 243–266. Elordieta, Arantzazu. 2001. Verb Movement and Constituent Permutation in Basque. Utrecht: LOT. Embick, David. 2004. “On the Structure of Resultative Participles in English”. Linguistic Inquiry 35-3: 355–392. Fábregas, Antonio & Rafael Marín. 2015. “How insertion in non-terminal nodes explains two participle puzzles”. Paper presented in the 25th Coloquium on Generative Grammar. Bayonne, May 21–23, 2015. Fábregas, Antonio & Rafael Marín. 2017. “On non-dynamic eventive verbs in Spanish”. Linguistics 55(3). García Fernández, Luis, Ángeles Carrasco Gutiérrez, Bruno Camus Bergareche, María Martínez-Atienza & María Ángeles García-Serrano. 2006. Diccionario de perífrasis verbales. Madrid: Gredos. Gehrke, Berit. 2011. “Stative passives and event kinds”. In Ingo Reich, Eva Horch & Dennis Pauly (eds.), Proceedings of Sinn und Bedeutung 15. Saarbrücken: Universaar— Saarland University Press. 241–257.

136

berro

Gehrke, Berit. 2015. “Adjectival participles, event kind modification and pseudo-incorporation”. Natural Language and Linguistic Theory 33: 897–938. Gehrke, Berit & Cristina Marco. 2014. The role of by-phrases in adjectival passives. Lingua 149: 188–214. Goenaga, Patxi. 1978. Gramatika bideetan. Donostia: Erein. Hualde, José Ignacio. 1991. Basque Phonology. London and New York: Routledge. Hualde, José Ignacio, Arantzazu Elordieta & Gorka Elordieta. 1994. The Basque dialect of Lekeitio. Supplements of ASJU XXXIV. Bilbao/Donostia: UPV/EHU, Gipuzkoako Foru Aldundia. Iatridou, Sabine. 1996. “To have and to have not: on the deconstruction approach”. In José Camacho, Lina Choueiri & Maki Watanabe (eds.), Proceedings of the 14th West Coast Conference on Formal Linguistics (WCCFL) 14. Stanford: CSLI, 185–201. Iatridou, Sabine, Anagnostopoulou, Elena & Roumyana Izvorski. 2001. “Observations about the form and meaning of the perfect”. In Michael. J. Kenstowicz (ed.), Ken Hale: a life in language. Cambridge, MA: MIT. 189–238. Iatridou, Sabine, Anagnostopoulou, Elena & Roumyana Izvorski. 2003. “Observations about the form and meaning of the perfect”. In Artemis Alexiadou, Monika Rathert & Arnim von Stechow (eds.), Perfect explorations. Berlin/New York: Mouton de Gruyter. 153–204. Kennedy, Christopher & Louise McNally. 1999. “From Event Structure to Scale Structure: Degree Modification in Deverbal Adjectives”. In Tanya Mathews & Devon Strolovitch (eds.), The Proceedings of SALT IX, 1999. Ithaka, NY: CLC Publications. 163–180. Koontz-Garboden, Andrew. 2010. “The lexical semantics of derived statives”. Linguistic and Philosophy 33: 285–324. Krajewska, Dorota. 2012. “Resultatives in Basque: a diachronic study”. Lingua Posnaniensis LIV(2): 56–67. Krajewska, Dorota. 2013a. The diachrony of resultative constructions in Basque. Master Thesis, UPV/EHU. Krajewska 2013b. “Euskarazko egitura erresultatiboaren diakronia”. In Ricardo Gómez et al. (eds.), 3rd Conference of the Luis Michelena Chair. Vitoria-Gasteiz: UPV/EHU Editorial Services. 264–274. Kratzer, Angelika. 1994. “The Event Argument and the Semantics of Voice”. Ms., University of Massachusetts at Amherst. Kratzer, Angelika. 1996. “Severing the external argument from its verb”. In Johan Roorych & Laurie Zaring (eds.), Phrase Structure and the Lexicon. Studies in Natural Language and Linguistic Theory, 33. Dordrecht: Kluwer Academic Publishers. 109– 137. Kratzer, Angelika. 2000. “Building Statives”. Berkeley Linguistic Society 26. 385–399. Levin, Beth & Malka Rappaport Hovav. 1986. “The formation of adjectival passives”. Linguistic Inquiry 17(1): 623–661.

non-verbal participles in basque and spanish

137

Luján, Marta. 1981. “The Spanish copulas as aspectual indicators”. Lingua 54: 165–209. Maienborn, Claudia. 2005. “On the Limits of the Davidsonian Approach: The Case of Copula Sentences”. Theoretical Linguistics 31 (3): 275–316. Maienborn, Claudia. 2007. “On Davidsonian and Kimian states”. In Ileana Comorovski & Klaus von Heusinger (eds.), Existence: semantics and syntax. Springer. 107–130. Marín, Rafael. 1997. “Participios con aspecto de adjetivos: entre la diacronía y la morfología”. Moenia 3: 365–376. Marín, Rafael. 2000. El componente aspectual de la predicación. Doctoral dissertation. Universitat Autònoma de Barcelona. Marín, Rafael. 2004a. Entre ser y estar. Madrid: Arco/Libros. Marín, Rafael. 2004b. “Sobre pasivas adjetivales”. Verba 31: 455–471. Marín, Rafael. 2009. “Del adjetivo al participio”. In Elena de Miguel et al. (eds.), Fronteras de un diccionario: las palabras en movimiento. San Millán de la Cogolla: Cilengua. 327–348. McFadden, Thomas & Artemis Alexiadou. 2011. “Pieces of the be perfect in German and Older English”. In Donald Baumer, David Montero & Michael Scanlon (eds.), Proceedings of the 25th West Coast Conference on Formal Linguistics. Somerville, MA: Cascadilla Proceedings Project. 270–278. McIntyre, Andrew. 2013. “Adjectival passives and adjectival participles in English”. In Artemis Alexiadou & Florian Schäfer (eds.), Non-canonical Passives, Amsterdam: John Benjamins. 21–42. Mounole, Céline. 2011. Le verbe basque ancien: étude philologique et diachronique. Ph.D. diss., University of the Basque Country (UPV/EHU) and Université Michel de Montaigne-Bordeaux III. Nedjalkov, Vladimir P. and Sergej Je. Jaxontov. 1988. “The Typology of Resultative Constructions”. In Vladimir Nadjalkov (ed.), Typology of resultative constructions. Amsterdam: John Benjamins Publishing Company. 3–62. Ortiz de Urbina, Jon & Myriam Uribe-Etxebarria. 1991. “Participial predication in Basque”. Supplements of ASJU 14-2: 993–1012. Oyharçabal, Bernard. 1992. “Structural and Inherent Case-Marking; Ergaccusativity in Basque”. In Joseba A. Lakarra & Jon Ortiz de Urbina (eds.), Syntactic Theory and Basque Syntax. Supplements of ASJU XXVII. Donostia: UPV/EHU and Gipuzkoako Foru Aldundia. 309–342. Oyharçabal, Bernard. 2010. “Basque ditransitives”. In Maia Duguine, Susana Huidobro & Nerea Madariaga (eds.), Argument structure and syntactic relations. Amsterdam/Philadelphia: John Benjamins. 261–282. Pancheva, Roumyana. 2003. “The Aspectual Makeup of Perfect Participles and the Interpretation of the Perfect”. In Artemis Alexiadou, Monika Rathert & Arnim von Stechow (eds.), Perfect explorations. Berlin/New York: Mouton de Gruyter. 277–306. Pinkster, Harm. 1987. “The strategy and chronology of the development of future and

138

berro

the perfect tense auxiliaries in Latin”. In Martin Harris, & Paolo Ramat (eds.), Historical development of auxiliaries. Berlin/New York: Mouton de Gruyter. 193–223. Parsons, Terence. 1990. Events in the Semantics of English. Cambridge, MA: the MIT Press. Ramchand, Gillian. 2008. Verb Meaning and the Lexicon: A First Phase Syntax. Cambridge: Cambridge University Press. Rapp, Irene. 1996. “Zustand? Passiv? Überlegungen zum sogenannten “Zustandspassiv””. Zeitschrift für Sprachwissenschaft 15.2: 231–265. Rapp, Irene. 1997. “Partizipien und semantische Struktur: Zu passivischen Konstruktionen mit dem 3”. Status. Tübingen: Stauffenburg. Rebuschi, George. 1984. Structure de l’énoncé en basque. Paris: SELAF. de Rijk, Rudolf P.G. 1993. “Basque Hospitality and the suffix -ko”. In José Ignacio Hualde & Jon Ortiz de Urbina (eds.), Generative Studies in Basque Linguistics. Amsterdam: John Benjamins. 145–162. [reprinted in Rijk, Rudolf P.G. de. 1998. De Lingua Vasconum: Selected Writtings. Supplements of ASJU XLIII. UPV/EHU. 377–390] de Rijk, Rudolf P.G. 2008. Standard Basque. A Progressive Grammar. Cambridge/London: The MIT Press. Rothmayr, Antonia. 2009. The Structure of Stative Verbs. Amsterdam/Philadelphia: John Benjamins. Sadler, Louisa & Douglas J. Arnold. 1994. “Prenominal adjectives and the phrasal/lexical distinction”. Journal of Linguistics 30(1): 187–226. Sleeman, Petra. 2011. “Verbal and adjectival participles: Position and internal structure”. Lingua 121: 1569–1587. Sleeman, Petra. 2014. “From participle to adjective in Germanic and Romance”. In Petra Sleeman, Freek Van de Velde & Harry Perridon (eds.), Adjectives in Germanic and Romance. John Benjamins. 171–198. von Stechow, Arnim. 1998. “German participles II in Distributed Morphology”. Ms., University of Tübingen. Wasow, Thomas. 1977. “Transformations and the lexicon”. In Peter W. Culicover, Thomas Wasow & Adrian Akmajian (eds.), Formal Syntax. 327–360. New York: Academic Press. Gawron, Jean Marc. 2009. The lexical semantics of extent verbs. Ms., San Diego State University. Zabala, Igone. 1993. Predikazioaren teoriak gramatika sortzailean (euskararen kasua). Ph.D. thesis, UPV/EHU. Zamparelli, Roberto. 1995. Layers in the determiner phrase. Ph.D. thesis, University of Rochester.

chapter 5

(In)transitive Verbs: Unergatives and Unaccusatives Ane Berro

1

Introduction

The fact that intransitive verbs do not conform a homogeneous class is an accepted and well-studied matter in the linguistic literature (Perlmutter 1978, 1989; Burzio 1981, 1986; Belleti 1988, 1999; Sorace 1993; Levin & Rappaport Hovav 1995 and subsequent work). According to the Unaccusative Hypothesis (Perlmutter 1978, 1989), intransitive verbs can be classified in two groups, depending on the nature of the only argument of the verb. In some intransitive verbs, the only argument corresponds to the initial direct object, whereas in others, it corresponds to the initial subject. The former are known as unaccusative verbs, and the latter are called unergative. In Basque, unaccusative and unergative verbs are clearly differentiated morphologically (Levin 1983): unaccusatives occur with absolutive subjects and izan (be) auxiliary, whereas unergatives generally appear with ergative subjects and edun (have) auxiliary. Case systems that show this differentiation among intransitive verbs have been variously called: Split S (Dixon 1979, 1994), Split Intransitive (Merlan 1985), Extended Ergative (Dixon 1979; Ortiz de Urbina 1989), Active (Bittner and Hale 1996; see also Sapir 1917), Agentive (Mithun 1991) or Semantically Aligned (Donohue & Wichmann 2008). In Romance languages, be auxiliary selection has also been proposed as a diagnostic of unaccusativity, particularly for Italian (Burzio 1981, 1986; Perlmutter 1989; Rosen 1981). However, not all Romance languages have auxiliary alternation, and those displaying it do not show the same distribution across verb classes (e.g. Sorace 1993, 2000, 2004; Bentley & Eythórsson 2003; MacKenzie 2006; Kayne 2008). In this chapter, I am going to analyze the unaccusative/unergative distinction in Basque and some Romance languages, paying special attention to the verb classes that are aligned in the unaccusative or the unergative way in both Basque and Romance, and considering also border-line cases such as Romance loan verbs in Basque and some new stative unergative verbs. I will propose that intransitive stative verbs are the most variable group of verbs, both cross-linguistically (in terms of auxiliary alternation, Sorace 2000, 2004; Keller & Sorace 2003) and in Basque (regarding auxiliary and subject case). The reason for this variability comes from the fact that the

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004395398_006

140

berro

subject in stative verbs—a holder argument—can be introduced either as an external subject or an internal subject, without creating a significant semantic contrast in its interpretation and the interpretation of the event configuration. The structure of the chapter is the following. In section 2, I will present the basic event decomposition that I assume, which shares many aspects with the system developed in Mateu (2002 et seq.) and Ramchand (2008 et seq.). Section 3 and 4 will describe and analyze unaccusative and unergative verbs in Basque, attending to factors such as morphological derivation, the simplex and complex contrast in unergatives, and dialectal variation. In section 5, I will focus on the distribution of auxiliary selection across verbs classes in Romance languages like Italian and French (mentioning also Old Catalan and Old Spanish), and I compare it to the distribution in Basque. In section 6, two aspects related to the contact between Basque and Romance will be considered, namely, the use of Romance loan verbs in Basque, and the emergence of new stative uses of certain verbs. In section 7, I will provide a syntactic explanation for the great variability exhibited by stative verbs, and section 8 will present the conclusions.

2

Unergatives and Unaccusatives in Basque

In this chapter, I am assuming a constructionist model of the event configuration, where the meaning of verbs is derived from the syntactic structure (Hale & Keyser 1993, 2002, 2005; Harley 1995, 2005; Mateu 2002; Cuervo 2003; Folli & Harley 2005; Ramchand 2008 among others). This way, I propose that unaccusative and unergative verbs have different syntactic configurations, and that this contrast is responsible for their different meaning and morphological realization. In Basque unaccusative verbs, the subject has absolutive (zero) case and the auxiliary selected is izan (be). (1) Jon jausi da.1 John.abs fall be.3sg.abs ‘John fell’ (2) Mahai-a apurtu da. table-det.abs break be.3sg.abs ‘The table broke’ 1 Abbreviations: abs = absolutive case, cl = se clitic, det = determiner, erg = ergative case, gen = genitive, imprf = imperfective, ine = inessive, nom = nominative case, part = partitive, pl = plural, prog = progressive, prt = participle, pst = past.

(in)transitive verbs: unergatives and unaccusatives

141

Unergative verbs, in contrast, have an ergative subject and edun (have) auxiliary. (3) Jon-ek ondo dantza-tzen du. John-erg well dance-imprf have.3sg.erg ‘John dances well’ I propose that, in unergative verbs, the subject is introduced in the specifier of Voice, a functional head that is projected outside the verbal phrase (Kratzer 1994, 1996). (4)

The subject of unaccusatives, on the other hand, is introduced within the verbal phrase. 2.1 Unaccusative Verbs In unaccusative verbs, the sole argument of the verb is introduced internally to the verbal phrase. Semantically, it has been associated with the role of undergoer of a directed change or of holder of a state whose existence is asserted or denied (Levin & Rappaport Hovav 1995). Unlike in the majority of unergative verbs, in Basque, unaccusative verbs consist of a single phonological word (leaving the auxiliary verb aside). The verbs grouped within unaccusatives in Basque are the following: verbs of change of state or location—telic and atelic ones—, verbs of appearance or occurring, verbs of existence and (some) aspectual verbs. (5) Verbs of change of state apurtu ‘break’, gorritu ‘redden’, handitu ‘get bigger’, txikitu ‘get smaller’, zabaldu ‘open’, itxi ‘close’, zahartu ‘age’, gaztetu ‘rejuvenate’, amatu ‘become a mother’, mutildu ‘become a lad’ etc.

142

berro

(6) Verbs of change of location igo ‘go up’, jaitsi ‘go down’, etorri ‘come’, joan ‘go, leave’, heldu ‘arrive’, ailegatu ‘arrive’, hurbildu ‘approach, come closer’, urrundu ‘move away’, etxeratu ‘go home’, lehorreratu ‘land, go ashore’ etc. (7) Verbs of appearance or occurring agertu ‘appear’, desagertu ‘disappear’, jaio ‘be born’, bukatu ‘run out’, gertatu ‘happen’ (8) Verbs of existence existitu ‘exist’, egon ‘be, stay’, izan ‘be’ (9) Aspectual verbs hasi ‘begin’, ari ‘be engaged in’, jarraiki ‘continue’ The majority of the verbs of change of state and change of location are derived, in the sense that they are built on elements that exist outside of this verbal complex, like nouns (e.g. ama ‘mother’ and mutil ‘boy’), adjectives (gorri ‘red’, handi ‘big’, txiki ‘small’ etc.) in (5), adverbs (hurbil ‘near’, urrun ‘far’) and adpositional phrases (e.g. etxe-ra ‘to house’ and lehorr-era ‘to land’) in (6). Most of them, and particularly derived ones, can have a causative variant, in which case the causer argument is marked ergative and the auxiliary switches to have. The ability to enter the causative alternation is considered a diagnostic for unaccusativity in Levin & Rapport Hovav (1995). Another test that has been suggested to differentiate unaccusatives from unergatives verbs in Basque involves the use of the partitive case. Only the subjects of unaccusative verbs (as well as the objects of transitive verbs) can occur bearing partitive marking (10) (11), contrasting, this way, with the subjects of unergative verbs (and also the subjects of transitive verbs) (12) (13) (Levin 1983; Ortiz de Urbina 1989; Salaburu 1992). (10) a. Ez da haurr-a etorri. no be.3sg.abs child-det.abs come ‘The child has not come’ b. Ez da haurr-ik etorri. no be.3sg.abs child-part come ‘No child has come’

☑ Unaccusative

143

(in)transitive verbs: unergatives and unaccusatives

(11) a. Ez dut haurr-a ikusi. no have.3sg.abs.1sg.erg child-det.abs see ‘I have not seen the child’

☑ Transitive object

b. Ez dut haurr-ik ikusi. no have.3sg.abs.1sg.erg child-part see ‘I have not seen any child’ (12) a. Ez du haurr-a-k deitu. no have.3sg.erg child-det-erg call ‘The child has not telephoned’

☒ Unergative

b. *Ez du haurr-ik deitu no have.3sg.erg child-part call ‘No child has telephoned’ (13) a. Haurr-a-k ez du ogi-a jan. child-det-erg no have.3sg.erg bread-det.abs eat ‘The child has not eaten (the) bread’

☒ Transitive

subject

b. *Haurr-ik ez du ogi-a jan child-part no have.3sg.erg bread-det.abs eat Unaccusative subjects and transitive objects pattern alike: they can be marked with the partitive. In contrast, unergative subjects and transitive subjects also behave in a similar fashion: they cannot bear partitive marking. 2.2 Unergative Verbs The sole argument of unergative verbs is considered the initial (Perlmutter 1978) or deep subject (Burzio 1981, 1986) as well as the external argument in the sense of Kratzer (1994, 1996). As indicated in section 2, I assume that it is introduced in the specifier of Voice (Kratzer 1994, 1996). Semantically, it is usually related to internal causation, as in agentive verbs or in non-agentive verbs of emission or bodily processes (Levin & Rappaport Hovav 1995). Unergativity has also been related with lack of telicity (Tenny 1987). However, with respect to Basque, as shown in section 2.1, telicity cannot be considered a discriminating factor, since many unaccusative verbs are actually atelic (see the list in (5) and (6)). Unergative verbs in Basque—which have ergative subjects and select for edun (have) auxiliary—are mainly verbs expressing volitional acts, verbs of

144

berro

emission, involuntary bodily processes, some verbs of existence of a state and some aspectual verbs: (14) Volitional acts a. Manner of motion dantzatu ‘dance’, saltatu ‘jump’, paseatu ‘stroll’, biratu ‘turn’, nabigatu ‘navigate’, eskiatu ‘ski’, eskalatu ‘climb’ (and many other Romance loans). b. Animate activities jolastu ‘play’, jokatu ‘play’, borrokatu ‘fight’, entrenatu ‘train’, bidaiatu ‘travel’ etc. c. Dining bazkaldu ‘have lunch’, afaldu ‘have dinner’, gosaldu ‘have breakfast’, askaldu ‘have an afternoon snack’. (15) Non-agentive manner of motion irakin ‘boil’ (16) Non-animate activities Funtzionatu ‘work properly (e.g. a machine)’ (17) Emission Distiratu ‘shine’, dirdiratu ‘shine, sparkle’, argitu ‘light’, usaindu ‘smell’. (18) Involuntary bodily processes Zurrungatu ‘snore’, izerditu ‘sweat’. (19) Existence of a state Iraun ‘last’, biziraun ‘survive’, balio izan ‘cost, value’, pisatu ‘weigh’. (20) Aspectual verbs Bukatu ‘finish’, jarraitu ‘continue’ The verbs shown above are all simplex, consisting of a single phonological word. See, for example, the sentence in (21). Note, however, that many unergative verbs in Basque are morphologically complex, consisting of the light verb egin ‘do’ and a non-verbal element (22). (21) Jon-ek ondo dantza-tzen du. John-erg well dance-imprf have.3sg.erg ‘John dances well’

(in)transitive verbs: unergatives and unaccusatives

145

(22) Jon-ek ondo dantza egi-ten du. John-erg well dance do-imprf have.3sg.erg ‘John dances well’ In the following section, I am going to deal briefly with complex unergative verbs, and in section 4 with simplex ones. Regarding simplex unergatives, I will show that there is dialectal variation in the subject marking and auxiliary selection of some verbs, particularly, in verbs expressing volitional acts like those in (14) (Oyharçabal 1992; Fernández 1997; Etxepare 2003; Aldai 2006, 2009; Berro 2010, 2012). 2.3 Complex Unergative Verbs As observed by Sarasola (1977), Levin (1983), Uribe-Etxebarria (1989), Laka (1993a), Hale & Keyser (1993), Levin & Rappaport Hovav (1995), Etxepare (2003), Oyharçabal (2006) and Aldai (2009), most unergative predicates in Basque are complex, built on the basis of the light verb egin ‘do’ and a non verbal element, usually a bare noun. These unergative verbs belong to many different semantic classes (Etxepare 2003; Zabala 2004). Note that some verbs also have a simplex counterpart, either as an unergative verb or as a transitive one. (23) Emission a. Sound emission deiadar egin ‘scream’, oihu egin ‘scream’, barre egin ‘laugh’, negar egin ‘cry’, zaunka egin ‘bark’ etc. b. Light emission Dir-dir egin ‘shine’, diz-diz egin ‘glow, sparkle’, nir-nir egin ‘twinkle, flicker’. c. Verbal emission hitz egin ‘talk’, solas egin ‘talk, chat’, marmar egin ‘grunt’, dei egin ‘call’, otoitz egin ‘pray’, errieta egin ‘reprimand, scold’, burla egin ‘mock, make fun of’ etc. (24) Mental activities gogoeta egin ‘meditate’, duda egin ‘doubt’, amets egin ‘dream’ etc. (25) Behavioral activities planto egin ‘stop by refusing to follow a game’, paso egin ‘be uninterested’, uko egin ‘refuse, reject’.

146

berro

(26) Internal body motion dar-dar egin ‘to tremble’, bor-bor egin ‘to boil’ (27) Physical activities a. Actions against an object or an individual zizt egin ‘puncture’, putz egin ‘blow’, bultza egin ‘push’, laztan egin ‘caress’, tiro egin ‘shoot’, min egin ‘hurt’ etc. b. Motion verbs alde egin ‘leave’, ihes egin ‘flee’, salto egin ‘jump’, dantza egin ‘dance’, laprast egin ‘slip’, ospa egin ‘leave’. c. Bodily functions eztul egin ‘cough’, aharrausi egin ‘yawn’, kaka egin ‘shit’, botaka egin ‘vomit’, izerdi egin ‘sweat’. (28) Other types of activities lan egin ‘work’, huts egin ‘fail’. (29) Weather verbs euria egin ‘rain’, eguzkia egin ‘be sunny’, elurra egin ‘snow’, hotz egin ‘be cold’, beroa egin ‘be warm’.2 As can be seen, complex unergative verbs can be either volitional or nonvolitional, but they can all conform to the property of internal causation, as formulated in Levin & Rappaport Hovav (1995). Regarding their complex morphological realization, Hale & Keyser (1993) argued, on the basis of languages like Basque, that unergative verbs are universally transitive in nature. The morphological contrast between simplex and complex unergatives like in dantzatu ‘dance’ and dantza egin [do dance] ‘dance’ would reflect the absence vs. presence of incorporation of the nominal object onto the verb. For instance, some analyses (Uribe-Etxebarria 1989; Bobaljik 1993; Laka 1993a and Fernández 1997) have proposed that the complement of the light verb egin represents actually a non-incorporated object, which may incorporate after spell out (Fernández 1997). As Uribe-Etxebarria (1989) pointed out (see also Oyharçabal 2006), some complex unergatives allow having egin and its complement separated in some contexts (in interrogative sen2 Weather verbs are different from other complex unergatives in a number of aspects. For instance, in most of them the non-verbal (nominal) element is headed by the determiner -a and they can be modified by adjectives like handi ‘big/a lot’ as in bero handia egin [heat big do] ‘be very warm’. I want to thank Jon Ortiz de Urbina for bringing these facts to my attention.

(in)transitive verbs: unergatives and unaccusatives

147

tences and in focalized structures) and the complement may be marked partitive in negative sentences, as common direct objects. (30) a. Nor-k egin behar du lan? who-erg do need have.3sg.erg work ‘Who has to work?’ b. Nor-k egin du lan? who-erg do have.3sg.erg work ‘Who has worked?’ (31) Oso ondo egin duzu lan. very well do have.2sg.erg work ‘Very well you have worked’ (32) Ez dut lan-ik egin. no have.1sg.erg work-part do ‘I haven’t worked’ Furthermore, the bare nominal may also be quantified, like common direct objects, and egin can be silenced in sentences with negative contrastive focus (Oyharçabal 2006: 792–793): (33) a. Lo gutxi egin dut. sleep little do have.1sg.erg ‘I have slept a little’ b. Lan gehiegi egin dut. work too.much do have.1sg.erg ‘I have worked too much’ (34) a. Lan egin dut, ez lo. work do have.1sg.erg no sleep ‘I have worked, not slept’ b. Irri egin dut, ez oihu. laugh do have.1sg.erg no shout ‘I have laughed, not shouted’

148

berro

Note, however, that the bare noun of complex unergative predicates does not behave as a common direct object in other aspects: e.g. in its lack of determiner and its position with respect to manner adverbials. See Oyharçabal (2006) for a discussion on the (non)-incorporation of the noun, and the variation in dialects and predicates. Another aspect in favor of the transitivity of complex unergative verbs is that, unlike simplex verbs, they are incompatible with a direct object (other than the non-verbal element selected by egin ‘do’). (35) a. Mikel-ek (tango-a) dantzatu du. Michael-erg (tango-det.abs) dance have.3sg.abs.3sg.erg ‘Michael has danced (a tango)’ b. Mikel-ek (* tango-a) dantza egin du. Michael-erg ( tango-det.abs) dance do have.3sg.abs.3sg.erg Intended: ‘Michael has danced (a tango)’ Many works have accounted for this contrast claiming that in complex unergatives, the complement of egin ‘do’ is assigned absolutive case, so that another direct object cannot be case-licensed (e.g. Laka 1993; Fernández 1997). In Berro (2015a), I additionally claim that the non-verbal element in complex verbs (dantza in (35b)) occupies exactly the same position of the direct object in simplex verbs (tangoa in (35a)), accounting in this way for the measuring properties of the direct objects in simplex verbs.3

3

The Building Blocks of the Event Configuration

The system of event configuration that I am assuming in this paper shares many similarities with that of Mateu (2002 et seq.) and Ramchand (2008 et seq.). In my system, the verbal phrase consists of subevents—Events and States—and their non-subeventive complements, namely Roots. Event provides eventivity to the predicate, introducing an event argument (Davidson 1967). State is a central coincidence relation, which relates a Figure (its specifier) with a Ground

3 Simplex verbs can be bounded if the direct object denotes a bounded scale (e.g. 35a). In a similar way, the event denoted in complex verbs is also measured by the scalar (aspectual) properties of its non-verbal complement (the Root). See Harley (2005) and Berro (2015a) for accounts about the measuring properties of Roots (Berro 2015a about Basque in particular).

(in)transitive verbs: unergatives and unaccusatives

149

(its complement) centrally. The subject of unaccusative verbs is introduced in the specifier of a State subevent. This is the configuration of an eventive transitive verb, with both Voice and State projected: (36) Eventive transitive verb

The different flavors of the subevents, and the theta roles of the subjects are derived from the whole event configuration (Wood 2012). Event can have a cause flavor, like in (36), if it is selected by Voice and if it selects for State. It can also have a do flavor (37), if it is selected by Voice, and if it selects straightaway for an Event naming Root (see Harley 2005; Berro 2015a). This is the case of unergative eventive verbs. (37) Eventive unergative verb

Event can also have a transitional (or go, become) flavor, in the case that it selects for a State (like in (36)), but Voice is not projected above (38). In this case, also like in (36), the Root selected by State is a State naming Root. This is the configuration of unaccusative verbs.4

4 Following Mateu (2002 et seq.), I consider that both telic and atelic unaccusatives have the configuration depicted in (3).

150

berro

(38) Eventive unaccusative verb

The theta-role of the subjects also varies depending on the configuration. When Voice selects for Event, the subject introduced in its specifier has an originator theta role, which can be further specified as causer or agent. In a configuration like (36), the subject has a causer theta-role, and in (37), it may have an agent role, but only if the entity occupying this position is animate and if the encyclopedic content related to the Root allows for that. On the other hand, the subject in the specifier of State has a patient role (undergoer and resultee), because the State is selected by Event. In the case it is not (as in the stative verbs that I will present in section 6), it has a holder role. As will be shown, the subject in Voice can also be a holder, if Voice selects for a State instead of for an Event.

4

Variation in Simplex Unergative Verbs

A subset of simplex unergative verbs shows dialectal variation. In eastern dialects, certain simplex verbs occur with absolutive subjects and izan (be) auxiliary, just like unaccusative verbs. (39) a. Antton bazkaldu da. Antton.abs have.lunch be.3sg.abs ‘Antton has had lunch’ b. Maialen borrokatu da. Maialen.abs fight be.3sg.abs ‘Maialen has fought’ The verbs that are used in the unaccusative fashion in eastern dialects comprise volitional verbs expressing speech (mintzatu ‘speak’, solastatu ‘chat’), manner of motion (promenatu ‘stroll’, jauzi ‘jump’), dining (bazkaldu ‘have lunch’, afaldu ‘have dinner’), and animate activities (borrokatu ‘fight’, jokatu ‘play, act’, jolastu ‘play’). As I have mentioned above, from this group, verbs of manner of

(in)transitive verbs: unergatives and unaccusatives

151

motion, dining and animate activities occur with an ergative subject and edun ‘have’ auxiliary in western and central varieties. (40) a. Antton bazkaldu da. Antton.abs have.lunch be.3sg.abs ‘Antton has had lunch’ b. Antton-ek bazkaldu du. Antton-erg have.lunch have.3sg.erg ‘Antton has had lunch’

Eastern dialects

Western & central dialects

Aldai (2008 2009) suggests that the dialectal divide within Basque corresponds to the divide between two case-marking systems: one based on the ergative alignment of case, represented by eastern dialects, and a semantically aligned one, represented by western and central dialects. However, Aldai’s characterization of the eastern case-alignment is questioned by the behavior some verbs, such as inanimate manner of motion verbs (41ab), verbs directly expressing duration (41c) and emission verbs (41d), which take an ergative subject and have auxiliary in eastern dialects (see Berro & Etxepare 2017). (41) a. Ur-a-k kurri-tzen du. water-det-erg flow-imprf have.3sg.erg ‘The water flows/is flowing’

Eastern dialects

b. Honek ez du funtziona-tzen. this.erg no have.3sg.erg work-imprf ‘This does not work/is not working’ c. Beran-tzen zuen. delay-imprf have.3sg.erg.pst ‘He/she was taking long’ d. Izarr-a-k distira-tzen du. star-det-erg shine-imprf have.3sg.erg ‘The star shines/is shining’ Therefore, not all unergative verbs are aligned in the unaccusative fashion in eastern dialects, so that the distinction between eastern and western/central dialects in terms of a semantically based alignment versus an ergative alignment is not as clear as suggested in Aldai’s proposal.

152

berro

Another line of analysis that has been put forward is that the verbs that take absolutive subjects in eastern dialects are somehow related to an aspectual head of telicity or boundedness (Berro 2010 2012). This head would be responsible for the absolutive marking of the subject and would also make it possible for those unergative verbs to become telic or bounded by selecting an incremental theme. A third type of explanation, developed in Pineda & Berro (2018), is that these unergative verbs are just associated to different syntactic-eventive configurations in eastern and western/central dialects. It is a well-known fact that verbs having very close meaning can correspond to different event configurations across languages. For example, the verb ‘blush’ is a change of state in Italian (arrossire, lit. ‘become red’) and it is a verb of internal causation in Dutch (bloezen) (McClure 1990, apud Levin & Rappaport Hovav 1995). Since both languages have auxiliary alternation, in Italian arrossire combines with essere (be) auxiliary, whereas in Dutch, bloezen occurs with hebben (have), as expected. On the other hand, in a closer Romance language like Spanish, the verb sonreír ‘smile’ is subject to intra-linguistic variation: although ‘smile’ is usually considered an unergative predicate (also in Basque, irribarre egin, lit. ‘do smile’), in some Spanish varieties it occurs with se clitic in non-reciprocal contexts, like some unaccusative verbs.5 (42) El niño se ha sonreído. det child.nom cl have.3sg smile.prt ‘The child has smiled’ From this perspective, sonreír and sonreírse are the externalizations of two different eventive configurations. Sonreír is an internally caused verb, whereas sonreírse is a change of state verb, where the subject undergoes a transition to a state of having a smile on the face, in the same way as the subject of the Italian verb arrossire. In Basque, too, we find similar cases, for example with the verbs leher egin and lehertu, which both mean ‘explode’. Leher egin is an unergative verb of internal causation, and lehertu, in its intransitive variant, is an unaccusative verb of change of state. These two different eventive configurations are morphologically externalized: the former having an ergative subject and have auxiliary, and the latter, an absolutive subject and be auxiliary.

5 This dialectal alternation is also attested in other verbs, like in engordar/engordarse ‘fatten’ and adelgazar/adelgazarse ‘lose weigh’.

(in)transitive verbs: unergatives and unaccusatives

153

(43) a. Auto-a-k leher egin du. car-det-erg explode do have.3sg.erg ‘The car blew up’ b. Gurpil-a lehertu da. wheel-det explode be.3sg.abs ‘The wheel exploded’ I propose that in leher egin, the verbal root is merged to an Event head with an internal cause flavor (do), whereas in lehertu, the Event head has a transitional flavor (trans). (44) a. leher egin

b. lehertu

The verbs that take an absolutive subject in eastern varieties could also be involved in an alternation of this sort. Dining verbs like bazkaldu ‘have lunch’ may be associated to a transition to a state of being finished with lunch. For instance, some speakers judge these verbs as telic in the progressive-perfect entailment test. (45) Bazkal-tzen ari naiz. ⇏ Bazkaldu dut. have.lunch-imprf be.engaged be.3sg.abs have.lunch have.3sg.erg ‘I am having lunch’ ⇏ ‘I have had lunch’

154

berro

In telic verbs, the progressive does not entail the perfect, and this is actually what happens in the verb bazkaldu. Telicity could perhaps lead to the change of state configuration (with Event + state) of this verb in eastern dialects. On the other hand, other verbs like mintzatu ‘speak’ and borrokatu ‘fight’ could be considered similar to some pronominal verbs in Spanish and French (expresarse and pelearse in Spanish, and s’exprimer and se battre in French), which have a rather reflexive or reciprocal flavor. In French, particularly, these predicates occur with être (be) auxiliary, as French is an auxiliary alternation language. (46) a. Je me suis mal exprimé. I.nom cl be.1sg wrong express.prt ‘I have not expressed myself well’ lit. ‘I have expressed myself wrong’ b. Nous nous sommes battus pour cela. we.nom cl be.1pl fight.prt for that ‘We have fought for that’ Pineda & Berro (2018) propose that several verbs which show dialectal alternation in Basque and that are aligned in the unaccusative fashion in eastern dialects are unaccusative verbs in (Middle) French and Occitan. These languages are actually the Romance languages that are in contact with northeastern dialects. The unaccusativity of these verbs is supported by several diagnostics, such as the use of the se clitic in previous stages of the language, occurrence in a causative variant, lack of direct object and available resultative participles (see Pineda & Berro 2018 for further details). Similarly to what is found in (Middle) French and/or Occitan, the configuration of these verbs in eastern dialects may be unaccusative (see section 6.1 in relation to this). The fact that this behavior is not generalized to all “unergative verbs” but that it is, actually, quite restricted would also suggest that we are dealing with certain lexical exceptions.

5

Unergatives and Unaccusatives in Basque and Romance

In Romance languages, unergative and unaccusative verbs have been differentiated on the grounds of several linguistic phenomena, such as prenominal passive participles, ne-cliticization in Italian (Bruzio 1981 1986), post-verbal subject position (Torrego 1989, Belleti 1988 1999) and auxiliary selection. In this section, I will focus on auxiliary selection, paying close attention to cross-linguistic variation and to the position that Basque occupies in this picture.

(in)transitive verbs: unergatives and unaccusatives

155

5.1 Unaccusativity and Auxiliary Alternation in Romance All Romance languages have or have had at some point of their history auxiliary alternation with intransitive verbs (MacKenzie 2006), some intransitive verbs taking have auxiliary and others taking be. have auxiliary has displaced be in Spanish, Catalan, Portuguese and Galician (although it survives in some dialects of Catalan and Aragonese). On the contrary, alternation between have and be exists nowadays in French and it is very productive in Italian. In European French, there are fewer verbs that select auxiliary be in comparison to Italian (Kayne 2008), and even fewer in Canadian French (Sankoff & Thibault 1977; apud Bentley & Eythórsson 2003). Sorace (2000 2004) and Keller & Sorace (2003) present an Auxiliary Selection Hierarchy where classes of predicates are distributed according to their auxiliary selection pattern. (47) Auxiliary Selection Hierarchy (ASH) Change of location Selects be (least variation) Change of state Continuation of a pre-existing state Existence of state Uncontrolled process Controlled process (motional) Controlled process (non-motional) Selects have (least variation) With this hierarchy, Sorace (2000 2004) accounts for the delimited and systematic variation found in intransitive predicates with respect to the auxiliary selected. According to Sorace, some predicates are consistent in showing auxiliary alternation across languages like Italian, French, German or Dutch, and also in language varieties, while others consistently select for be or have, both synchronically and diachronically. The verbs at the top and at the bottom of the hierarchy are the ones having least variation and most consistent behavior (either selecting be or have), whereas the ones in the middle are the most variable. Regarding the distribution of auxiliaries in unaccusative verbs in French and Italian, three groups of verbs can be distinguished (Bentley & Eythórsson 2003: 450–451). Firstly, change of location verbs select be in both Italian and European French (some of these verbs take have in Canadian French).

156

berro

(48) a. Maria è venuta alla festa. Mary.nom be.3sg come.prt to.det party ‘Mary came to the party’

Italian

b. Marie est venue en retard. Mary.nom be.3sg come.prt in late ‘Mary arrived late’ (Sorace 2000: 863)

Eu. French

Sorace (2000) includes, in this group, verbs that are inherently telic. Nevertheless, in certain atelic change of location verbs like ‘descend’, French selects for avoir (have) auxiliary, whereas Italian selects essere (be). (49) l’ actif net de la Société a baissé sous le det assets net of det society have.3sg descend.prt under det minimum. minimum ‘The net assets of the society descended to the minimum’ On the other hand, change of state verbs show internal variation, as they can be classified in two groups: those selecting be in both Italian and European French, and those selecting be in Italian and have in European French. The verbs of the former group, that is to say, the ones that occur with be also in European French, are telic (50a) (51a). This means that, in French, atelic change of state verbs (e.g. grandir ‘increase, become bigger’, viellir ‘become older’) occur with avoir (have) auxiliary (51b). (50) a. La temperature è salita improvisamente. det temperature.nom be.3sg rise.prt suddenly ‘The temperature suddenly rose’ (Sorace 2000: 865)

Italian

b. Gianni è invecchiato. John.nom be.3sg become.older.prt ‘John has become older’ (Kayne 2008: 1) (51) a. L’enfant est devenu triste. det-child.nom be.3sg become.prt sad ‘The child became sad’

Eu. French

157

(in)transitive verbs: unergatives and unaccusatives

b. Cette inquietude a grandi en raison de la this worry.nom have.3sg increase.prt in reason of det circulation incontrôlée. circulation uncontrolled ‘This worry has increased due to the uncontrolled circulation’ c. Jean a vielli. John.nom have.3sg become.older.prt ‘John has become older’ (Kayne 2008: 1) Finally, verbs of continuation of a pre-existing state (e.g. ‘last’, ‘survive’ etc.) and existence of a state (e.g. ‘be’ ‘exist’, ‘belong’, ‘seem’ etc.) select generally be in Italian (although, according to Sorace 2000, judgments are weaker and have is not uncommon), and consistently have in European French. (52) a. La guerra è / ?ha durato a lungo. det war.nom be.3sg / have.3sg last.prt to long ‘The war lasted for long’ (Sorace 2000: 867)

Italian

b. I dinosaur sono esistiti / ??hanno esistito 65 det dinosaurs.nom be.3pl exist.prt have.3pl exist.prt 65 milioni di anni fa. million P years ago ‘The dinosaurs existed 65 million years ago’ (Sorace 2000: 869) (53) a. Mes parents ont survécu au my parents.nom have.3pl survive.prt to.det tremblement de terre. earthquake ‘My parents survived the earthquake’

Eu. French

b. Les dinosaures ont existé il y a 65 millions det dinosaurs.nom have.3pl exist.prt there is 65 million d’ans. of years ‘The dinosaurs existed 65 million years ago’ (Sorace 2000: 869) As can be seen, the replacement of be auxiliary by auxiliary have takes place gradually, more prominently in the groups of verbs that are in the middle of Sorace’s (2000 2004) hierarchy. Mateu (2009) shows that this tendency is also

158

berro

attested in the diachronic evolution of have replacement in Old Catalan and in Old Spanish. On the basis of the diachronic studies of Old Catalan done by Batlle (2002) and of Old Spanish done by Aranovich (2003) and Castillo (2003), Mateu points out that the verbs that are more variable in Italian are the ones that lost be auxiliary earlier in both Old Catalan and Old Spanish. According to Batlle (2002), in Old Catalan, the most innovative verbs—replacing ésser (be) with haver (have)—were verbs of appearance and existence (‘happen’, ‘occur’, ‘become’), and copulative verbs (‘stand’, ‘rest’, ‘remain’). Telic verbs of change of state or location were the most reluctant to appear with haver (have), and consistently chose ésser (be). As Mateu (2000) shows, the same gradual process took place in Spanish (Aranovich 2003; Castillo 2002). The process of replacement of ser by haber started affecting the same classes of verbs: verbs of appearance and existence, which are actually the verb classes in the middle of the hierarchy. Therefore, the hierarchy put forward by Sorace (2000 2004) also holds for diachronic variation, and for the gradual spread of have auxiliary over be across verb classes in languages that do not have auxiliary alternation these days. 5.2 Comparison with Basque Coming back to Basque, this language behaves more like Italian than like French in the use of the auxiliary in intransitive verbs (see section 3). In Basque verbs expressing change of location and change of state occur with izan (be) auxiliary, regardless of their inherent telicity. (54) a. Prezio-a-k nabarmen jaitsi dira azken price-det-pl.abs notoriously descend be.3pl.abs last egun-etan. day-ine.pl ‘Prices have notoriously descended in the last days’ b. Jon zahartu zen. John.abs become.older be.3sg.abs.pst ‘John became older’ Regarding the third group of unaccusative verbs outlined before—verbs expressing continuation of a pre-existing state and existence of a state—there is more variation in auxiliary selection and subject marking. Certain verbs of this group occur with edun have auxiliary and an ergative subject: iraun ‘last’, biziraun ‘survive’, irudi ‘seem, look like’, eman ‘seem’.

(in)transitive verbs: unergatives and unaccusatives

159

(55) a. Guda-k bost urte iraun zuen. war.det-erg five year last have.3sg.erg.pst ‘The war lasted five years’ b. Oso gaztea ema-ten duzu. very yound seem-imprf have.2sg.erg ‘You seem very young’ Other verbs of this group, in contrast, select izan (be) auxiliary, like the verb gustatu ‘like’ (55) and faltatu ‘lack’, and others such as existitu ‘to exist’ can be found either with izan (be) (57a) or with edun (have) (57b) (selecting edun (have) particularly in non-standard uses). (56) Jon-i makarroi-a-k gusta-tzen zaizkio. John-dat macaroni-det-pl like-imprf be.3pl.abs.3sg.dat ‘John likes macaroni’ (57) a. Munstro-a-k ez dira existi-tzen. monster-det-pl no be.3pl.abs exist-imprf b. Munstro-ek ez dute existi-tzen. monster-erg.pl no have.3pl.erg exist-imprf ‘Monsters do not exist’ As can be seen, Basque intransitive verbs also conform to the hierarchy proposed in Sorace (2000, 2004), as stative verbs are the ones showing more variation in the selection of have and be auxiliary. In the next section, I will continue analyzing the variation existing in stative verbs, particularly focusing on the role of Romance se clitic and its influence on loan verbs in Basque.

6

Romance Loan Verbs and Contact Driven Uses

Basque is in permanent contact with Spanish and French, as all Basque speakers are these days Spanish-Basque or French-Basque bilingual. In this section, two consequences of this contact will be analyzed, namely, the use of Romance loan verbs, particularly intransitive verbs, and the emergence of new uses of certain stative predicates. In both cases, we will see that the presence or absence of the se clitic in the original Romance verbs is closely related to the subject case marking and auxiliary selection of the verbs in Basque.

160

berro

6.1 Loan Intransitive Verbs Alberdi (2003) analyzes Romance loan verbs in Basque and concludes that there are two factors influencing the marking of the subject and auxiliary selection of these verbs in Basque: (i) the semantic notion of internal causation (Levin & Rappaport Hovav 1995), and (ii) the presence of the se clitic in the original Romance verb. Verbs like bidaiatu ‘travel’ and abortatu ‘abort’ occur with an ergative subject and with have auxiliary, because both verbs are internally caused. On the other hand, deskuidatu ‘get confused, lose the head’ and mutinatu ‘mutiny’ are used with an absolutive subject and be auxiliary, since the original verbs descuidarse and amotinarse take the se clitic in Spanish.6 Certain predicates are used differently in Spanish and French, and also show dialectal variation in Basque. For instance, the verb divorce is used with the se clitic in Spanish (i.e. divorciarse), and without se in French (i.e. divorcer). This contrast is not surprising, given that ‘divorce’ can be configured as a verb of change of state—become divorced—or as an internally caused verb—conduct a divorce. (58) a. Maria y Juan se han divorciado. Mary.nom and John.nom cl have.3pl divorce.prt ‘Mary and John have divorced.’ b. Marie et Jean ont divorcé. Mary.nom and John.nom have.3pl divorce.prt ‘Mary and John have divorced.’

Spanish

French

In relation to this contrast, in the dialects that are in contact with Spanish, dibortziatu ‘divorce’ is used in the “unaccusative” pattern, selecting an absolutive subject and be auxiliary (59a). In contrast, in the dialects that are in contact with French, dibortziatu takes an ergative subject and have auxiliary (59b). (59) a. Miren eta Jon dibortziatu dira. Basque (cont. with Sp.) Mary.abs and John.abs divorce be.3pl.abs b. Miren-ek eta Jon-ek dibortziatu dute. Basque (cont. with Fr.) Mary-erg and John-erg divorce have.3pl.erg ‘Mary and John have divorced.’ 6 It must be noted that se influences the alignment only in the case of intransitive loan verbs. Uses of se in transitive clauses (comerse la manzana [eat-se the apple]) do not seem to alter the alignment in Basque.

(in)transitive verbs: unergatives and unaccusatives

161

A reversed, but similar, pattern is attested with the verb ‘stroll’. In French, when it is intransitive, it is a pronominal verb (i.e. se promener), whereas in Spanish, it is non-pronominal (i.e. pasear).7 As expected, in the Basque dialects that are in contact with French, promenatu ‘stroll’ combines with an absolutive subject and be auxiliary, whereas in the dialects that are in contact with Spanish, paseatu occurs with an ergative subject and have auxiliary. (60) a. Xabier promenatu da. Xavier.abs stroll be.3sg.abs

Basque (cont. with Fr.)

b. Xabierr-ek paseatu du. Xavier-erg stroll have.3sg.erg ‘Xavier has strolled.’

Basque (cont. with Sp.)

Another verb that shows (dialectal) variation is entrenatu ‘train’. The verb ‘train’ can be used in the pronominal and non-pronominal form in both Spanish (i.e. entrenar and entrenarse) and French (s’entraîner). In the dialects that are in contact with French, speakers prefer to use the intransitive entrenatu in the unaccusative pattern, namely, with an absolutive subject and be auxiliary. In contrast, in other dialects (central and western ones), the intransitive entrenatu is used either in the unergative fashion—with an ergative subject and have auxiliary—or in the unaccusative one. As in the previous cases, I suggest that these differences in the subject marking and auxiliary selection are the externalizations of different event configurations. (61) a. entrenatu [do training]

7 There is also a variant pasearse in Spanish, with the se clitic, but this one has a slightly different meaning, namely, ‘stroll without any clear intention’.

162

berro

b. entrenatu [become trained]

In the unergative pattern, entrenatu would correspond to an internally caused verb (do training) (61a), whereas in the unergative one, entrenatu would be a change of state (become trained) (61b). 6.2 New Stative Verbs In Basque, there is a number of stative verbs that are not accepted or recommended in the standard language (Berro 2015b, in press): pisatu ‘weigh’, neurtu ‘be long’, erre ‘burn, be very hot’, labaindu/irristatu/lerratu ‘be slippery’, usaindu ‘smell’, pikatu ‘be spicy’ and kubritu ‘be deep’. Acceptance varies depending on the speakers, but young Basque-Spanish bilinguals or bilinguals with strong Spanish influence accept and use these verbs naturally. (62) Asko pisa-tzen du [ez dakit eraman ahal izango duzun]. a.lot weigh-imprf have.3sg.erg ‘It is very heavy, I do not know if you are going to be able to carry it.’ (A. Cano, Belarraren Ahoa: 63) (63) 27 kilometro neur-tzen ditu alde 27 kilometer measure-imprf have.3sg.erg.3pl.abs side zabal-ean […] wide-ine ‘It is 27 kilometers long in its wide side.’ (Berria, 2004-12-16) (64) Plater-a-k erre-tzen du. dish-det-erg burn-imprf have.3sgerg ‘The dish burns.’ (65) Lurr-a-k laban-tzen du. floor-det-erg slip-imprf have.3sg.erg ‘The floor is slippery.’

(in)transitive verbs: unergatives and unaccusatives

163

(66) Piper hauek asko pika-tzen dute. pepper these.erg a.lot burn-imprf have.3pl.erg ‘These peppers are very spicy.’ (67) Lantegi horre-tako ke-a-k oso gaizki usain-tzen factory that-gen smoke-det-erg very wrong smell-imprf du. have.3sg.erg ‘The smoke from that factory smells very bad.’ As can be seen in the examples, these verbs are used in the unergative pattern (with ergative subject and have auxiliary) and they are morphologically simplex (they do not consist of the light verb egin ‘do’). Remarkably, these verbs are stative: they denote a property of their sole argument—like its weigh, length, temperature etc.—and they behave like stative verbs in a number of linguistic contexts, such as in their incompatibility with process adverbs and the degree interpretation of the modifier apur bat ‘a little’ (see Berro 2015b, 2016). The ergative subject of these verbs is not an agent and causation is not involved, but the verbs are still aligned in the unergative pattern. Most of these verbs are used standardly in the language, but with eventive meanings. In order to express these stative meanings, predicative constructions are used instead: (68) Hogeita bost bat milimetro luze ziren. twenty five about millimeter long be.3pl.abs.pst ‘They were about twenty five millimeters long.’ (J. Gartzia, Sistema Periodikoa. Primo Levi: 190) (69) a. Piper hauek oso pikante-a-k dira. pepper these.abs very spicy-det-pl be.3pl.abs b. Piper hauek min-a-k dira. pepper these.abs spicy-det-pl be.3pl.abs ‘These peppers are very spicy.’ (70) Lurr-a laban-a dago. floor-det.abs slippery-det be.3sg.abs ‘The floor is slippery.’

164

berro

(71) Jertse-a-k usain txarr-a dauka. jumper-det-erg smell bad-det.abs have.3sg.erg.3sg.abs ‘The jumper smells bad’; lit.: ‘The jumper has bad smell’. The emergence of simplex unergative verbs in order to express these stative meanings seems to be related to the contact with Romance languages. For instance, the verbs quemar ‘burn’ and resbalar ‘slip’ are also polysemic between an eventive and a stative interpretation (cf. (72a) and (73a) with (72b) and (73b)) in Spanish. (72) a. Juan se resbala en el suelo. John.nom cl slip.3sg in det floor ‘John slips on the floor.’ b. El suelo resbala. det floor.nom slip.3sg ‘The floor is slippery.’ (73) a. La casa se quema. det house.nom cl burn.3sg ‘The house burns.’ b. La sopa quema. det soup.nom burn.3sg ‘The soup burns’ or ‘The soup is burning hot.’ As it can be observed, in the eventive intransitive use of the verbs, the se clitic is present. In Basque, in the eventive intransitive use, the verb gets realized in the unaccusative pattern. Compare the eventive unaccusative verbs in (74a) and (75a) with the stative unergative uses in (74b) and (75b). (74) a. Jon lurr-ean irrista-tzen da. John.abs floor-ine slip-imprf be.3sg.abs ‘John slips on the floor.’ b. Lurr-a-k irrista-tzen du. floor-det-erg slip-imprf have.3sg.erg ‘The floor is slippery.’

(in)transitive verbs: unergatives and unaccusatives

165

(75) a. Etxe-a erre-tzen da. house-det.abs burn-imprf be.3sg.abs ‘The house burns.’ b. Zopa-k erre-tzen du. soup.det-erg burn-imprf have.3sg.erg ‘The soup burns’ or ‘the soup is burning hot.’ Once again, the meaning that has the se clitic in Romance is related to the unaccusative pattern in Basque, and similarly, the se-less meaning in Romance is related to the unergative pattern. Actually, se is ungrammatical in the stative uses of the verbs in Spanish.8 (76) a. *El suelo se resbala det floor.nom cl slip.3sg Intended: ‘The floor is slippery’ b. #La sopa se quema. det soup.nom cl burn.3sg Intended: ‘The soup burns’ or ‘The soup is burning hot.’ I propose that, in these cases too, the difference in the subject marking and auxiliary selection is derived from different eventive configurations in the syntax. The unergative stative verbs have the structure in (75a), whereas the eventive unaccusative verb has the configuration in (75b).

8 In French, there is a dynamic and intransitive use of the verb gliser ‘slide’ which does not take the se clitic and which requires have auxiliary. (i) Pierre a glissé sur la glace (et est tombé) Pierre has slid on the ice (and is fallen) Pierre slid on the ice (and fell) Labelle & Doron (2010) claim that, in this case, the root has merged with a dynamic verbal head (v), instead of with V. As a consequence, the construction focuses on the process, instead of on the result.

166

berro

(77) a. irristatu [stative]

b. irristatu [eventive]

In the stative variant, Event is not projected, and the subject is introduced in the specifier of Voice. There, it is interpreted as a holder argument. Recall that the subject of Voice is only interpreted as an originator when Voice selects for Event. On the other hand, in the eventive variant, the subject is introduced in the specifier of State, and after Event is projected, it is interpreted as a patient (undergoer/resultee).

7

Stative Verbs as Border-Line Cases

In this section, I will propose that stative verbs are the most variable predicates both in Basque and cross-linguistically because their subject—a holder argument—can be introduced either in the specifier of Voice or in the specifier of State, without that creating an important semantic contrast. This happens precisely because, in states, the Event head is not projected. Additionally, I will argue that, in Romance loan verbs, the presence of the se clitic in the original Romance language forces an unaccusative configuration of the verb in Basque, because se-variants necessarily project Event + State (Cuervo 2003, 2014). In section 4.1 and 5.3, we saw that stative verbs are the most variable group of verbs in terms of auxiliary alternation (and subject case marking) both in

(in)transitive verbs: unergatives and unaccusatives

167

Basque and cross-linguistically. As explained in section 4.1, verbs of existence of state are in the middle of Sorace’s (2000, 2004) hierarchy, since these are the verbs showing more cross-linguistic differences and less consistent judgments. On the other hand, we have seen that, in Basque, there are unaccusative and unergative verbs within the group of stative verbs, and additionally, I have shown that there is a group of new stative verbs that are aligned in the unergative pattern, presumably, created under the influence of the Basque-Romance contact. In this section, I will offer a syntactic explanation for the great variability exhibited by stative verbs, based on the event configuration assumed in this chapter. As explained in section 2, subjects of intransitive verbs can be introduced either in the specifier of Voice or in the specifier of State. (78) a. [VoiceP Subj Voice [EventP Event [Root]]] Subj = originator, agent b. [EventP Event [StateP Subj State [Root]]] Subj = patient It can be observed in (78) that the subject introduced by Voice is interpreted as an originator or an agent (the agent role being subsumed under the originator role), and that the subject introduced by State is interpreted as a patient. However, as explained in section 2, these interpretations are derived from the whole event configuration. The subject in (78a) is interpreted as an originator only if Voice selects for Event, and the subject in (78b) is a patient only if State is selected by Event. If these conditions are not met, the subjects of both Voice and State are interpreted as holders. (79) a. [VoiceP Sub Voice [StateP State [Root]]] b. [StateP Subj State [Root]]

Subj = holder Subj = holder

In stative verbs, Event is not projected. Thus, the subject of a stative verb can be introduced either in the specifier of Voice (79a) or in the specifier of State (79b), without that creating a significant semantic difference. Note that in eventive predicates, having an external or internal subject really makes a difference: if it is external, it is interpreted as the entity originating the event, whereas if it is internal, it gets the role of undergoing the event and ending in a final state. Since there is no Event in state verbs, stative verbs can have external or internal subjects without much semantic contrast.9

9 There is a contrast that, in Berro (2015a), I claim to be one between containing vs. being included, or have vs. be In unergative stative verbs, the subject contains or possesses the

168

berro

Regarding the role of the se clitic in Romance and its relation with the Basque unaccusative pattern, I propose that se unaccusatives in Romance involve the projection of both Event and State, in the spirit of Cuervo (2003 2014) (see also Cuervo 2003 2014 for Spanish, Folli & Harley 2005 for Italian, and Labelle 1992, Labelle & Doron 2010 for French). In the literature, se has been related to telicity in different ways: it has been argued that se is involved in a configuration having or emphasizing a final state (Labelle 1992, Cuervo 2003 2014, Folli 2002, Folli & Harley 2005, Legendre & Smolensky 2009, Labelle & Doron 2010) or a bounded path (Basilico 2010). Cuervo (2003 2014) analyzes unaccusative verbs that have a se and a se-less variant (e.g. caer(se) ‘fall’, salir(se) ‘come out/off’, morir(se) ‘die’, ir(se) ‘go’ etc.). On the basis of several systematic interpretive and formal contrasts between the se and the se-less variant of these verbs (79), she proposes that these two variants have different eventive structures, i.e. a bi-eventive vs. a mono-eventive configuration. Depending on the structure, the Root is interpreted as Manner or as Result (Mateu & Acedo-Matellán 2012, see also Labelle & Doron 2010). The seless variants are verbs of change or motion (telic or atelic), whereas se variants are verbs of change of state (necessarily telic). This difference is observed in the availability of durative and frame modifiers. (80) a. El avión cayó durante tres minutos / * en tres det plane.nom fall.3sg.pst for three minutes / * in three minutos antes de estrellarse. minutes before crashing ‘The plane fell for three minutes /*in three minutes before crashing.’ b. El avión se cayó * durante tres minutos / en det plane.nom cl fall.3sg.pst * for three minutes / in tres minutos (# antes de estrellarse). three minutes (# before crashing) ‘The place fell down *for three minutes/ in three minutes (before crashing).’ According to Cuervo (2003, 2014), the se-less variant of these verbs is monoeventive, consisting of only vGO, whereas the se variant is bi-eventive, having both vGO and vBE.

property denoted by the state, whereas in unaccusative states, the subject is included within the set of things having the property denoted by the state.

(in)transitive verbs: unergatives and unaccusatives

169

(81) a. se-less variant Salieron muchos yuyos ‘There appeared many weeds’

b. se variant Se salieron tres clavos ‘Three nails came off’ (Cuervo (2014: 51–52))

The light verbs vGO and vBE can be paralleled to the Event and State heads assumed in this chapter.10 The basic idea is that se-less variants are verbs of change without entailing a final state, where the subject undergoes the process ‘naturally’ (as in the case of snow, rain or leaves in the verb caer ‘fall’). In contrast, se variants are composed of two subevents, a dynamic subevent and a result. The subject is the holder the final result. According to Cuervo (2014), the se clitic is the realization of the unvalued phi-features of vGO, which are valued with the subject seated in the specifier of vBE. Assuming Cuervo’s analysis that se variants are bi-eventive, having an Event and a State subevent, we can understand why originally Romance se-verbs are analyzed in Basque as unaccusatives (selecting be auxiliary and having an absolutive subject). In Basque, they are structured in a configuration where the subject is introduced in the specifier of State, like unaccusatives.

10

Go or become are derived flavors of Event in the assumed event configuration. For instance, Event is interpreted as trans(ition) if it selects for State and if it is not selected by Voice.

170

berro

(82) Romance loan verbs with the se clitic

It must be noted, though, that the relation between the unaccusative pattern in Basque and the presence of the se clitic in Romance is not bi-directional, particularly in the domain of non-loan verbs. Many verbs that show an unaccusative alignment in Basque do not have se clitic in their Romance counterparts, for example, heldu ‘arrive’, etorri ‘come’, handitu ‘increase’, txikitu ‘decrease’, hazi ‘grow’, jaio ‘be born’, gertatu ‘happen’, faltatu ‘lack’ etc. Actually, the contrast between mono-eventive and bi-eventive unaccusatives proposed in Cuervo (2003, 2014) does not have a phonological reflex in Basque, since both show the same alignment pattern (be auxiliary and absolutive subject). The conclusion that can be drawn is that, for loan verbs, having a se clitic forces an unaccusative alignment of predicates in Basque, but that, those verbs not having a se clitic may be subject to other discriminating factors, such as transition, eventivity, internal causation etc.

8

Conclusion

In this chapter, I have described and analyzed unaccusative and unergative verbs in Basque, comparing the distribution of these verbs in several Romance languages, such as French, Italian and Spanish. I have provided a syntactic explanation for verbs that show unaccusative/unergative alternation (e.g. dibortziatu ‘divorce’, entrenatu ‘train’), basically proposing that they are the realizations of different syntactic event-configurations. Regarding the great variability exhibited by stative verbs, both cross-linguistically and in Basque, I have argued that intransitive stative verbs are most variable because holder subjects can be introduced either in the specifier of Voice or in the specifier of State. Since Event is not projected in stative verbs, the specifier of both Voice and State gets interpreted as a holder, and not as an originator or a patient. Thus, stative verbs are more variable basically because two syntactic event configurations are compatible with a similar semantic interpretation.

(in)transitive verbs: unergatives and unaccusatives

171

Acknowledgements This work has been supported by the Basque Government (the research group IT665–13) and the Spanish Ministry of Economy and Competitiveness (FFI2014–51878-P). The research leading to these results has also received funding from the European Union’s Seventh Framework Program for research, technological development and demonstration under grant agreement no. 613465.

References Alberdi, Xabier. 2003. “Euskal aditz mailegatuen erregimena: hurbilketa”. In Jesus Mari Makazaga & Bernard Oyharçabal (eds.), P. Lafitteren sortzearen mendemugako biltzarra. Gramatika gaiak. Iker 14 (I). Bilbao: Euskaltzaindia. 37–60. Aldai, Gontzal. 2006. “Is Basque morphologically ergative? The semantic split-intransitive case marking system of Western Basque”. In Beatriz Fernandez & Itziar Laka (eds.), Andolin Gogoan. Essays in honour of Professor Eguzkitza. Leioa: UPV/EHU. 117–138. Aldai, Gontzal. 2008. “From ergative case marking to semantic case marking: The case of historical Basque”. In Mark Donohue & Søren Wichmann (eds.), The Typology of semantic alignment. Oxford: OUP. 197–218. Aldai, Gontzal. 2009. “Euskararen kasu markaketaren aldakortasun dialektala”. In Beatriz Fernández, Pablo Albizu & Ricardo Etxepare (eds.), Euskara eta euskarak: aldakortasun sintaktikoa aztergai [Supplements of ASJU LII]. Donostia: UPV/EHU and Gipuzkoako Foru Aldundia. 5–19. Aranovich, Raúl. 2003. “The semantics of auxiliary selection in Old Spanish”. Studies in Language 27-1: 1–37. Basilico, David. 2010. The se Clitic and its Relationship to Paths. Probus 22: 271–302. Batlle, M. 2002. L’expressió dels temps compostos en la veu mitjana i la passive pronominal. El process de subsitutció de l’auxiliar ‘ésser’ per ‘haver’. Barcelona: Institut d’Estudis Catalans / Publicacions de l’Abadia de Montserrat. Belleti, Adriana. 1988. “The case of unaccusatives”, Linguistic Inquiry 19: 1–34. Belleti, Adriana. 1999. ““Inversion” as focalization and related questions”, Catalan Working Papers in Linguistics 7: 9–45. Bentley, Delia & Thórhallur Eythórsson. 2003. “Auxiliary selection and the semantics of unaccusativity”. Lingua 114: 447–471. Berro, Ane. 2010. “Unergative Predicates in Basque Varieties: Consequences for the Ergative Case Assignment”. DEA research work. University of the Basque Country (UPV/EHU).

172

berro

Berro, Ane. 2012. “Three Levels of Root Insertion in Basque Intransitive Verbs”. In Ernestina Carrilho & Beatriz Fernández (eds.), Journal of Portuguese Linguistics. Special Issue, 11-1. Ediçoes Colibrí-Universidade de Lisboa. 7–22. Berro, Ane. 2015a. Breaking verbs. From event structure to syntactic categories in Basque. Ph.D. diss., University of the Basque Country (UPV/EHU) and Université Bordeaux Montaigne (UBM). Berro, Ane. 2015b. “Egoera predikatu berriak euskaraz”. In Irantzu Epelde (ed.), Euskal hizkera eta dialekotoak gaur. Special Issue of Lapurdum 3: 11–25. Berro, Ane & Ricardo Etxepare. 2017. “Ergativity in Basque”. In Jessic Coon, Lisa M. Travis & Diane Massam (eds.), The Oxford Handbook of Ergativity. Oxford University Press. Bittner, Maria & Ken Hale. 1996. “The structural determination of case and agreement”. Linguistic Inquiry 27: 1–68. Bobaljik, Jonathan D. 1993. “On Ergativity and Ergative Unergatives”. In Collin Phillips (ed.), MIT Working Papers in Linguistics 19: Papers on Case and Agreement II. Cambridge: MIT. 45–88. Burzio, Luigi. 1981. Intransitive verbs and Italian auxiliaries. Ph.D. diss., MIT. Cambridge, MA. Burzio, Luigi. 1986. Italian Syntax. Dordrecht: Reidel. Castillo, María Elena. 2002. Inacusatividad y aspecto léxico en los verbos de movimiento. Estudio diacrónico, Series Documenta Universitaria. Scripta I, Girona: Edicions a Petició SL: Cuervo, María Cristina. 2003. Datives at Large. Ph.D. diss., MIT. Cuervo, María Cristina. 2014. “Alternating unaccusatives and the distribution of roots”, Lingua 141: 48–70. Davidson, Donald. 1967. “The Logical Form of Action Sentences”. In Nicholas Rescher (ed.), The Logic of Decision and Action. Pittsburgh: University of Pittsburgh Press. [Reprinted in Davidson 2006.] Davidson, Donald. 2006. The Essential Davidson. Oxford: Oxford University Press. Dixon, Robert M.W. 1979. “Ergativity”, Language 55: 59–138. Dixon, Robert M.W. 1994. Ergativity. Cambridge: Cambridge University Press. Donohue, Mark & Søren Wichmann. 2008. The typology of semantic alignment. Oxford: OUP. Etxepare, Ricardo. 2003. “Valency and Argument Structure in the Verb”. In José Ignacio Hualde & Jon Ortiz de Urbina (eds.), A grammar of Basque. Berlin: Mouton de Gruyter. 369–465. Fernández, Beatriz. 1997. Egiturazko kasuaren erkaketa euskaraz. Ph.D. thesis, University of the Basque Country (UPV/EHU). Folli, Rafaella. 2002. Constructing telicity in English and Italian. Oxford: Oxford University Press.

(in)transitive verbs: unergatives and unaccusatives

173

Folli, Rafaella & Heidi Harley. 2005. “Flavors of v”. In Paula Kempchinsky & Roumaya Slabakova (eds.), Aspectual Inquiries. Dordrecht: Springer. 95–120. Hale, Ken & S. Jay Keyser. 1993. “On Argument Structure and the Lexical Expression of Syntactic Relations”. In Ken Hale and S. Jay Keyser (eds.), The View from Building 20. Cambridge, Massachusetts: The MIT press. 53–109. Hale, Ken & S. Jay Keyser. 2002. Prolegomenon to a Theory of Argument Structure. Cambridge, Massachusetts: The MIT Press. Hale, Ken & S. Jay Keyser. 2005. “Aspect and the Syntax of Argument Structure”. In Nomi Erteschik-Shir & Tova Rapoport (eds.), The Syntax of Aspect: Deriving Thematic and Aspectual Interpretation. New York: Oxford University Press. 11–42. Harley, Heidi. 1995. Subjects, Events and Licensing. Ph.D. diss., MIT. Harley, Heidi. 2002. “Possession and the Double Object Construction”. In Linguistic Variation Yearbook 2: 29–68. Amsterdam: John Benjamins. Harley, Heidi. 2005. “How Do Verbs Get Their Names? Denominal Verbs, Manner Incorporation and the Ontology of Roots in English”. In Nomi Erteschik-Shir & Tova Rapoport (eds.), The Syntax of Aspect. New York: Oxford University Press. 42– 64. Kayne, Richard S. 2008. “A note on auxiliary alternations and silent causation”. Ms., New York University, New York. Keller, Frank & Antonella Sorace. 2003. “Gradient auxiliary selection and impersonal passiviation in German: an experimental investigation”. Journal of Linguistics 39: 57–108. Kratzer, Angelika 1994. “The Event Argument and the Semantics of Voice”. Ms., University of Massachusetts at Amherst. Kratzer, Angelika. 1996. “Severing the External Argument from its Verb”. In Johan Roorych & Laurie Zaring (eds.), Phrase Structure and the Lexicon. Studies in natural language and linguistic theory, vol. 33. Dordrecht: Kluwer Academic Publishers. 109– 137. Labelle, Marie. 1992. “Change of state and valency”. Journal of Linguistics 28: 375–414. Labelle Marie & Edit Doron. 2010. Anticausative derivations (and other valency alternations) in French. Probus 22–2: 303–316. Laka, Itziar. 1993. “Unergatives that Assign Ergative, Unaccusatives that Assign Accusative”. In Jonathan D. Bobaljik & Collin Phillips (eds.), Papers on Case and Agreement I, MITWPL 18. Cambridge, MA: MITWPL. 149–172. Legendre, Géraldine & Paul Smolensky. 2009. “French inchoatives and the Unaccusativity Hypothesis”. In Donna B. Gerdts, John Moore & Maria Polinsky (eds.), Hypothesis A/Hypothesis B: Linguistic Explorations in Honor of David M. Perlmutter. MIT Press. Levin, Beth. 1983. On the Nature of Ergativity. Ph.D. diss., MIT. Levin, Beth & Malka Rappaport Hovav. 1995. Unaccusativity. At the Syntax-Lexical Semantic Interface. Cambridge, MA: the MIT Press.

174

berro

Mackenzie, Ian. 2006. Unaccusative verbs in Romance Languages. Hampshire/New York: Palgrave. Mateu, Jaume. 2002. Argument structure: relational construal at the syntax-semantic interface. Ph.D. diss., Universitat Autònoma de Barcelona. Mateu, Jaume. 2009. “Gradience and auxiliary selection in Old Catalan and Old Spanish”. In Paola Crisma & Giuseppe Longobardi (eds.), Historical syntax and linguistic theory. Oxford: Oxford University Press. 176–193. Mateu, Jaume & Víctor Acedo-Matellán. 2012. “The manner/result complementarity revisited: a syntactic approach”. In María Cristina Cuervo & Yves Roberge (eds.), The end of argument structure? Vol. 38. Syntax & Semantics. Bingley: Emerald. 209– 228. McClure, William. 1990. A lexical semantic explanation for unaccusative mismatches. In Katarzyna Dziwirek, Patrick Farrell & Errapel Mejias-Bikandi (eds.), Grammatical Relations. A cross-theoretical perspective. Stanford: CSLI Publications. Merlan, Francesca. 1985. “Split intransitivity: functional oppositions in intransitive inflection”. In Johanna Nichols & Anthony Woodbury (eds.), Grammar inside and outside the clause: some approaches to theory from the field. Cambridge: Cambridge University Press. 324–362. Ortiz de Urbina, Jon. 1989. Parameters in the Grammar of Basque. Studies in Generative Grammar 33. Dordrecht: Foris Publications. Oyharçabal, Bernard. 1992. Structural and Inherent Case-Marking; Ergaccusativity in Basque. In Joseba A. Lakarra & Jon Ortiz de Urbina (eds), Syntactic Theory and Basque Syntax. [Supplements of ASJU XXVII]. Donostia: UPV/EHU and Gipuzkoako Foru Aldundia. 309–342. Oyharçabal. Bernard. 2006. “Basque Light Verb Constructions”. ASJU XL: 787–806. Perlmutter, David M. 1978. Impersonal Passives and the Unaccusative Hypothesis. Proceedings of the 4th Annual Meeting of the Berkeley Linguistics Society. 157–190. Perlmutter, David M. 1989. “Multiattachment and the unaccusative hypothesis: the Perfect Auxiliary in Italian”. Probus 1–1: 63–119. Ramchand, Gillian. 2008. Verb Meaning and the Lexicon: A First Phase Syntax. Cambridge: Cambridge University Press. Rosen, Carol. 1981. The relational structure of reflexive clauses: evidence from Italian. Ph.D. diss., Harvard University, Cambridge, MA. Salaburu, Pello. 1992. “Euskara, hizkuntza ergatiboa ote da?”. In J.H. Duhalde (ed.), Luis Villasanteri omenaldia. Bilbao: Euskaltzaindia. 417–433. Sankoff, Gillian & Pierrette Thibault. 1977. “L’alternance entre les auxiliaries avoir et être en français parlé à Montréal”. Langue Français 34: 81–108. Sapir, Edward. 1917. Review of C.C. Uhlenbeck “Het passieve karakter van het verbum transitivum of van het verbum actionis in talen van Noord-Amerika”, International Journal of American Linguistics 1: 82–86.

(in)transitive verbs: unergatives and unaccusatives

175

Sarasola, Ibon 1977. “Sobre la bipartición inicial en el análisis en constituyentes”, ASJU XI: 51–90. Sorace, Antonella. 1993. “Unaccusativity and auxiliary choice in non-native grammars of Italian and French: asymmetries and predictable indeterminacy”. Journal of French Language Studies 3: 71–93. Sorace, Antonella. 2000. “Gradients in auxiliary selection with intransitive verbs”. Language 76: 859–890. Sorace, Antonella. 2004. “Gradience at the Lexicon-Syntax Interface: Evidence from Auxiliary Selection”. In Artemis Alexiadou, Elena Anagnostopoulou & Martin Everaert (eds.), The Unaccusativity Puzzle. Oxford: Oxford University Press. 243–268. Tenny, Carol Lee. 1987. Grammaticalizing aspect and affectedness. Ph.D. diss., MIT. Cambridge, MA. Torrego, Esther. 1989. “Unergative-unaccusative alternations in Spanish”. In Itziar Laka & Anoop Mahajan (eds.), Functional heads and clause structure: MIT working papers in linguistics, vol. 10. Cambridge, MA. 253–272. Uribe-Etxebarria, Myriam. 1989. “Some Notes on the Structure of IP in Basque”. Ms., UConn, Storrs. Zabala, Igone. 2004. “Los predicados complejos en vasco”. In Igone Zabala, Elixabete Pérez Gaztelu & Lluïsa Gràcia Sole (eds.), Las fronteras de la composición en lenguas románicas y en vasco. Bilbao: Deustuko Unibertsitatea, Argitalpen Zerbitzua. 445– 534.

chapter 6

Light Verb Constructions in Basque and Romance Víctor Acedo-Matellán and Anna Pineda

1

Introduction

1.1 What is a Light Verb Construction? When Virginia Woolf writes in her diary I took a walk by myself, instead of I walked by myself, she is choosing to linguistically construe a particular conceptual scene by means of a L(ight) V(erb) C(onstruction). In these constructions the verb—took—bears little semantic weight, this being encoded in a N(on-)V(erbal) E(lement)—a walk. LVC s, such as take a walk, have a bath, or give a cuddle, are certainly not a quirk of English. In fact, as research over the last 25 years has shown, they are abundant crosslinguistically and in some languages like Basque they are actually the unmarked way to encode certain types of events.1 While the format of LVC s varies both intra- and cross-linguistically, it displays certain constant features. As their name suggests, LVC s are headed by a verb that, at least within the construction, has a very abstract semantic content. In fact, the semantic contribution of the L(ight) V(erb) seems to be limited to functional notions related to argument and event structure. On the other hand, the NVE, whose lexical category may change across LVC s and also depending on the language, largely encapsulates the conceptual content of the construction, that is, the particular event evoked.2 To show how this is actually the division of labour between the LV and the NVE, it is useful to compare, on the one hand, two constructions sharing the same LV but differing in their NVE and, on the

1 See Butt (2010) for an overview, and for useful terminological precisions, notably the distinction between LVC and complex predicate. See also Bowern (2008). On the other hand, as pointed out by Lin (2001:22), the use of light verb as related to LVC s is not exactly equivalent to what is known as Chomskyan light verb (Chomsky 1995), that is, a functional verb that takes the VP as complement and that is responsible for the projection of the external argument and for the assignment of case to the object (see Larson 1988 and Hale & Keyser 1993 for seminal proposals). 2 Although we use NVE throughout this work, since it constitutes a good label considering the kind of LVC s that we encounter in Basque and Romance, it should be noted that in other languages like Urdu the element containing the conceptual portion of the LVC can be of verbal category (see Butt 1995, 2010).

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004395398_007

light verb constructions in basque and romance

177

other, two constructions sharing the same NVE but being different with respect to their LV. We can illustrate the former case with the minimal pair involving the Spanish LVC s dar miedo “give fear” ‘frighten’ and dar vergüenza “give embarrassment” ‘embarrass’. These LVC s share the same LV, dar ‘give’, and the abstract semantics of transferal inherent to this verb is preserved in these constructions. Thus, both LVC s are construed with a dative understood as the experiencer of the emotion:3 (1) a. A Marisa le dan miedo las películas de terror. dat Marisa dat.3sg give.3pl fear the.pl films.pl of horror ‘Horror films frighten Marisa.’ b. A Marisa le da vergüenza hablar en público. dat Marisa dat.3sg give.3sg embarrassment talk.inf in public ‘Marisa feels embarrassed to talk in public.’ The difference between the two LVC s is exclusively of conceptual nature, and corresponds to the difference between the emotions of fear and embarrassment, as encoded in their NVE s. On the other hand, to exemplify the case of two LVC s differing in their LV but not in their NVE we may keep dar miedo and compare it with pasar miedo “pass fear” ‘be afraid’:4 (2) a. A Marisa le dan miedo las películas de terror. dat Marisa dat.3sg give.3pl fear the.pl films.pl of horror ‘Horror films frighten Marisa.’ b. Marisa pasa miedo con las películas de terror. Marisa pass.3pl fear with the.pl films.pl of horror ‘Marisa feels fear with horror films.’ In this case the same emotion (fear) is involved, but the argument-structure properties are different. Thus, pasar miedo requires a subject experiencer and 3 The abbreviations used in glosses are the following ones: abs ‘absolutive’, adv ‘adverbial suffix’, aux ‘auxiliary’, compl ‘complementizer’, dat ‘dative’, det ‘determiner’, erg ‘ergative’, F ‘feminine’, ines ‘inessive’, inf ‘infinitive’, instr ‘instrumental’, loc ‘locative’, M ‘masculine’, pl ‘plural’, prs ‘present’, pst ‘past’, ptve ‘partitive’, sg ‘singular’. Also, note that although we generally provide a full-fledged gloss for Basque auxiliaries, in some instances we will just gloss aux for reasons of space. 4 See, however, section 2.1, where we provide an account of several pairs of synonymous LVC s sharing a NVE and differing in their LV s.

178

acedo-matellán and pineda

the stimulus can be expressed as a PP headed by con ‘with’, while dar miedo appears with a subject stimulus and a dative experiencer. This difference, as is evident, can only depend on the choice of LV. Interestingly, however, this choice is not completely free: research has shown that the semantic properties of the NVE may determine, at least to some extent, which LV it combines with (Alba-Salas 2002: 51, and references cited therein). At the extreme of semantic cohesion between the LV and the NVE lie, of course, the many LVC s with idiomatic meaning, that is, a meaning that cannot be computed straightforwardly from the sum of the meanings of the LV and the NVE, as that shown by the Basque LVC s iskin egin “corner do” ‘elude’ or hanka egin “leg do” ‘escape rapidly’. A prominent fact about LVC s is that they tend to correlate with a verb that expresses more or less the same meaning, as was shown above with the pair take a walk and walk, and can be further illustrated with the pair dar vergüenza “give embarrassment” and avergonzar ‘embarrass’ in Spanish. It is important to emphasize that the semantic equivalence between these pairs is not complete, as seminally pointed out by Wierzbicka (1982). In any case, LVC s do not always count with synthetic counterparts. In fact, in languages like Basque most of them do not (Martinez 2015: 30, Aldai 2009: 823). The generality of the analytic, LVC-based expression of a large set of events in Basque is, as we will see, one of the major differences when we compare this language with modern Romance. 1.2 A Brief State of the Art LVC s have come to occupy an important place in the agenda of linguistic descriptions and syntactic theory ever since Jespersen (1954) coined the term light verb.5 This label was taken quite literally in the first papers dedicated specifically to LVC s. Indeed, with some exceptions such as Wierzbicka’s (1982), who studies the semantic contribution of have in constructions such as have a bath, a leading idea in the 70’s and 80’s was that the LV is completely devoid

5 In this short introduction we cannot do justice to the vast literature on LVC s, which encompasses a considerable number of languages—Asian ones featuring very prominently. Works providing general remarks on LVC s, or comparing different languages include Butt & Geuder (2001), Butt (2003, 2010), Bowern (2008), Butt & Lahiri (2013) and Ramchand (2014). For particular languages, see, among others, Jespersen (1954), Jackendoff (1974), Wierzbicka (1982), Cattell (1984) and Kearns (1988), on English; Grimshaw & Mester (1988), on Japanese; Ahn (1991), on Korean; Mohanan (1994), on Hindi; Butt (1995), on Urdu; and Lin (2001), on Chinese. For references on LVC s in Basque and Romance (particularly on French and Spanish) see the following sections.

light verb constructions in basque and romance

179

of predicational force and only contributes verbhood, and that the argumentstructure properties of the LVC are inherited from the NVE through some mechanism. For instance, in an influential paper on Japanese constructions based on the LV suru ‘do’, Grimshaw & Mester (1988) propose that the LV does not have an argument structure of its own, and that it’s the nominal NVE that transfers some or all of its arguments to the LVC. In a similar vein, Cattell (1984) argues that English LV s are predicationally vacuous. In turn, the lightness of LV s have made some authors propose that they are the historic result of a process of semantic bleaching of fully predicational verbs, and consequently, that they can be further grammaticalized into auxiliary verbs and, finally, into inflectional markers (Hook 1991, Hopper and Traugott 1993, 2003). A different analysis is provided by Kearns (1988) within a lexicalist framework distinguishing between Lexical Conceptual Structure and Syntactic Argument Structure (Hale & Keyser 1986, Rappaport & Levin 1988, among others). An LV like give provides a Syntactic Argument Structure, with a particular number of argument slots in a particular configuration, and also an event argument, but since the verb is delinked from its Lexical Conceptual Structure, no theta roles are available. Complementarily, the NVE is associated with a Lexical Conceptual Structure and a thematic array, but it lacks a full Syntactic Argument Structure. In the LVC these two defective components converge to contribute their lexical properties, yielding a full predicate.6 In any case, Kearns’s (1988) work is best known for its contribution to differentiating between T(rue) LV s like those in Take a walk, Give a groan or Have a bite, and what she calls Vague Action Verbs, like those heading constructions such as Make an inspection, Give a demonstration or Do the ironing. The basic difference is that True Light Verbs combine with NVE s that are not argumental, since they do not pattern with arguments: they cannot become passive subjects, they cannot be extracted, they cannot be pronominalized, and they cannot be definite. By contrast, Vague Action Verbs combine with full-fledged argumental DP s. However, as we will see below, the range of variation within Basque LVC s (see section 2.2.) casts some doubts on Kearns’ distinction between the two types of verbs found in LVC s. More recently, some of the abovementioned ideas have been to a large extent disputed. First, the predicational lightness or emptiness of the LV has been qualified, and standard analyses take the LV to be a predicate and to contribute to a joint predication with the NVE through some mechanism. For example, in Butt’s (1995) Lexical-Functional-Grammar approach, an operation called Argu-

6 See Jackendoff (1974) for a pioneering analysis along similar lines.

180

acedo-matellán and pineda

ment Fusion combines the LV predicate (for instance, of “give” or “let” semantics) and the NVE predicate to yield a single predicate. The new predicational unit hosts, on the one hand, argument slots directly contributed by either of the two basic predicates (e.g., an Agent from the LV, and a Theme from the NVE) and, on the other, argument slots that are the result of the fusion of two arguments slots of the basic predicates (for instance, a Causee that subsumes a Goal of the LV and an Agent of the NVE). Even more recently, Ramchand (2014) has taken a step further in this direction. In her analysis, LV s possess argument and event structures, which are, moreover, identical to those of their non-LV counterparts. Thus the LV give of LVC s like John gave Mary a cuddle is an item that lexicalizes an event structure containing the three subeventive heads proposed in Ramchand (2008)—Init(iation), Proc(ess) and Res(ult)—plus a Poss(essive) head merged under Res and projecting a phrase whose specifier is the recipient (Mary) and whose complement is the NVE (a cuddle). This structure provides the skeletal meaning of caused possession and is actually the very same one proposed for the non-light use of give (see Ramchand 2008: 103 for details). The difference between light and non-light give boils down to the nature of the complement: if the complement involves a bare event noun like cuddle, the interpretation of the construction is that of an LVC. We witness a type of approach, therefore, in which the emphasis is put on the similarities, rather than the differences, between the light and the non-light versions of the same verb.7 Ramchand’s (2014) work on LV s in English, Bengali, and Persian actually aims at further exploring and validating an empirical claim made by Butt (2003) and Butt & Lahiri (2013) on the basis of diachronic evidence from IndoAryan languages. Ramchand (2014: 217) dubs this claim Butt’s Generalization: “Unlike auxiliaries, which may become grammaticalized over time to have a purely functional use, light verbs always have a diachronically stable corresponding full or “heavy” version in all the languages in which they are found.” From the strictly diachronic point of view, Butt & Lahiri (2013:7) state that they “depart from the received view that the existence of a light verb is due

7 Cf. Ramchand’s (2014:218) Semantics of Structure Conjecture on the Limits of Lightness: “The meaning of a light verb and its corresponding heavy alternant are in a subset-superset relation in their conceptual semantics, the light version being a proper subset of the heavy. Only non-syntactic or conceptual information is systematically negotiable within the “same” lexical item. Anything that is present in the heavy version but not in the light must therefore be a species of Type B meaning. At its most pared down, a light verb can only be as light as the structural semantics corresponding to the type A meaning of the pair.”

light verb constructions in basque and romance

181

to a historical process of semantic bleaching” (see also Bowern 2008). Their results—and those of Butt & Geuder’s (2001) and Ramchand’s (2014)—thus cast doubt on the position, represented by Hook (1991) and Hopper & Traugott’s (1993, 2003), that LV s may be an intermediary link in a grammaticalization cline, sitting between full-fledged verbs and auxiliaries. As we will see below, synchronic data from Basque and Romance supports Butt’s Generalization. 1.3 About This Chapter In this chapter we will describe LVC s in Basque and compare them to Romance LVC s, focusing on those Romance languages that are in direct contact with Basque: French and Spanish. The structure of the chapter is as follows. In section 2, we provide a descriptive comparison of LVC s in Basque and Romance: we first describe the range of verbs that can function as LV s in these languages, and, taking Butt’s Generalization and Ramchand’s (2014) remarks as background, we explore their argument- and event-structure properties. We first focus on the LV (section 2.1). In section 2.2, we turn to the NVE, paying attention to its categorial properties (lexical and phrase-structural category) and its relational properties (caserelated properties and the like). In section 3 we present a discussion of the (different degrees of) syntactic and semantic cohesion between the LV and the NVE, with a special focus on idiomaticity. Section 4 concludes and points out issues that deserve attention in future research. As mentioned above, the discussion is based on Basque, and Romance, mainly French and Spanish, is brought in for comparison.

2

A Descriptive Comparison of Light Verb Constructions in Basque and Romance

While Basque and Romance LVC s share a number of properties that are also common to LVC s crosslinguistically, they also show some differences, partly due to independent syntactic properties of the two linguistic systems, like case. In this section, we offer a description of the morphosyntactic and semantic properties of LVC s in Basque and compare them to Romance LVC s. We focus first on the LV and then on the NVE and its cohesion with the LV.

182

acedo-matellán and pineda

2.1 The Light Verb In Basque, most LVC s are formed with the LV egin ‘do’:8 (3) agur egin “do goodbye” ‘greet / say goodbye’, alde egin “do side / zone” ‘leave’, aharrausi egin “do yawn” ‘yawn’, aipu egin “do mention” ‘mention’, aitor egin “do confession” ‘confess’, amets egin “do dream” ‘dream’, argi egin “do light” ‘light up / enlighten’, arnas(a) egin “do breath” ‘breathe’, barre egin “do laugh” ‘laugh’, behaztopa egin “do impediment” ‘bump into / complicate’, bekatu egin “do sin” ‘sin’, berba egin “do word” ‘talk’, bultza egin “do push” ‘push’, dantza egin “do dance” ‘dance’, dei egin “do call” ‘call’, doministiku egin “do sneeze” ‘sneeze’, duda egin “do doubt” ‘doubt’, elurra egin “do snow” ‘snow’, erregu egin “do plea / prayer” ‘beg / pray’, euria egin “do rain” ‘rain’, eztul egin “do cough” ‘cough’, gogoeta egin “do reflection” ‘think, reflect’, hitz egin “do word” ‘talk’, huts egin “do mistake / shortage” ‘fail / miss’, ihes egin “do escape” ‘run away’, irain egin “do insult” ‘insult’, irribarre egin “do smile” ‘smile’, iruzur egin “do trick” ‘mislead, deceive’, jauzi egin “do jump” ‘jump’, kasu egin “do attention” ‘pay attention’, keinu egin “do gesture / sign” ‘gesture, sign’, kontu egin “do care” ‘look after, take care’, korrika egin “do running (adv.)” ‘run’, lan egin “do work” ‘work’, laztan egin “do cuddle / caress / kiss” ‘caress, kiss’, lo egin “do sleep” ‘sleep’, min egin “do pain” ‘hurt’, negar egin “do weep” ‘cry’, oihu egin “do shout, yell” ‘shout, yell’, ospa egin “do “get out!”” ‘leave’, salto egin “do jump” ‘jump’, tiro egin “do gunshot” ‘shoot’, topo egin “do stumble” ‘run into’, txalo egin “do applause” ‘applaud’, zin egin “do oath” ‘swear under oath’, among many others.9 However, Basque has also some LVC s with hartu ‘take’ (4), eman ‘give’ (5) and eduki ‘have’ (Martinez 2015:113–114, Zabala 2004:486–488):10 8

9

10

In the Basque tradition, LV s are referred to as egin inergatiboak ‘do unergatives’ or I[zen]+egin inergatiboak ‘do+N[oun] unergatives’. In the French linguistic tradition the term verbe support ‘support verb’ is the one most widely used, while in the Spanish tradition we find verbo soporte ‘support verb’, verbo de apoyo ‘verb of support’, and verbo liviano ‘light verb’. As seen, Basque LVC s describe daily, general activities (viewed as recognizable, unitary concepts), a well-described tendency among this kind of constructions cross-linguistically (Mithun 1984: 850, Martinez 2015:42, fn. 43). Rodríguez & García Murga (2003) add to this sample LVC s with jo ‘hit, touch, give with no delicacy’, such as adarra jo “give the horn” ‘tease’. Zabala (2004:461, 465, 488–493) provides an even wider range of “complex predicates” including verbs such as bete ‘fill’ as in hitza bete ‘keep one’s word’, hautsi ‘break’ as in hitza hautsi ‘not keep one’s word’, jan ‘eat’ as in hitza jan ‘fail to keep one’s word’ or burua jan ‘wash someone’s brain’, and so on. These expressions, combining a fully contentful verb and an absolutive case-marked

light verb constructions in basque and romance

183

(4) arnasa hartu “take breath” ‘breathe’, atseden hartu “take repose, rest” ‘rest’, atsegin hartu “take pleasure” ‘feel happy, enjoy’, damu hartu “take pity, regret” ‘feel sad, regret’, esku hartu “take hand” ‘participate, take part’, gogo hartu “take mind, will” ‘decide’, hitz hartu “take word” ‘compromise’, indar hartu “take strenght” ‘strengthen’, kontu hartu “take care” ‘take care of, keep in mind’, lur hartu “take land” ‘land’, min hartu “take pain” ‘get hurt’, parte hartu “take part”, pena hartu “take pity” ‘feel sad’, plazer hartu “take pleasure” ‘enjoy’,11 poz hartu “take happiness” ‘feel happy’, su hartu “take fire” ‘burn’, and few others. (5) amore eman “give love” ‘yield, give in, collapse’, argi eman “give light” ‘light up / enlighten’, arrazoi eman “give reason” ‘agree with’, aurpegi eman “give face” ‘face’, begi eman “give eye” ‘look at’, bihotz eman “give heart” ‘encourage’, bizkar eman “give one’s back” ‘turn one’s back on’, buru eman “give head” ‘finish, accomplish / face’, damu eman “give pity” ‘sadden’, errieta eman “give scolding” ‘tell off’, esker eman “give thanks” ‘thank’, fede eman “give faith” ‘trust, vouch for’, gogo eman “give mind, will” ‘look, pay attention / think, consider’, hats eman “give breath” ‘blow’, ikara/beldur eman “give fear” ‘frighten’, kontseilu eman “give advise” ‘advise’, kontu eman “give account” ‘give an explanation, give account’, laztan eman “give cuddle, hig” ‘hug, caress’, leku eman “give place” ‘allow’, lur eman “give land” ‘bury’, min eman “give pain” ‘hurt’, musu eman “give kiss” ‘kiss’, parte eman “give report” ‘report, inform’, pena eman “give pity” ‘make someone feel sad’, poz eman “give happiness” ‘make happy’, su eman “give fire” ‘set fire to’, and few others. (6) arrazoi eduki “have reason” ‘be right’, esku eduki “have hand” ‘help, support’, kontu eduki “have care” ‘be careful’, min eduki “have pain” ‘feel pain’, on eduki “have good (adjective)” ‘maintain, claim’, pena eduki “have pity” ‘feel sad’, and few others.

11

DP, clearly fall beyond the scope of this chapter, which is mainly devoted to LVC s formed by a LV and a bare noun (hitz egin ‘talk’)—or by a full verb and a bare noun (hitz eman ‘promise’). Moreover, since such cases feature absolutive-marked DP s (adarra ‘(the) horn’, hitza ‘(the) word’ …), it seems more appropriate to view them as the result of a metaphorical (idiomatic) interpretation of expressions which are not bona fide LVC s. This example and the two preceding ones correspond to psych predicates, which frequently showed a similar pattern in Old Romance languages, as in Old Catalan prendre delit, Old French prendre plaisir or Old Spanish tomar plazer “take pleasure”, ‘enjoy’. See Acedo-Matellán & Mateu (to appear) for a detailed description of analytic psych predicates in Old Catalan.

184

acedo-matellán and pineda

There are also some expressions involving eman ‘give’ and hartu ‘take’ that show an extreme degree of coalescence between the LV and the NVE, since they are actually not separable. They are arguably not LVC s anymore, but plain, synthetic verbs: (7) onartu “take good” ‘accept’, atzeman “give hoof” ‘catch, understand’, hauteman “give choice” ‘realise, notice’ / choose, hitzeman “give word” ‘promise’ For French and Spanish, Alba-Salas (2002: 636–637) provides the following respective lists, in addition to faire (French) and hacer (Spanish) ‘do’: (8) asséner ‘deal’ (asséner un coup ‘hit’), avoir ‘have’ (avoir une maladie ‘be sick’), coller ‘give’ (coller une claque ‘smack’), commetre ‘commit’ (commettre une aggression ‘commit an assault’), donner ‘give’ (donner un conseil ‘give some advice’), dire ‘say’ (dire des compliments ‘pay some compliments’), diriger ‘direct’ (diriger une offensive ‘attack’), effectuer ‘carry out’ (effectuer un paiement ‘make a payment’), entamer ‘initiate’ (entamer une discussion ‘initiate a discussion’), ficher ‘give’ ( ficher une giffle ‘slap’), flanquer ‘give’ ( flanquer un coup de pied ‘kick’), foutre ‘give’ ( foutre une claque ‘smack’), jeter ‘throw’ ( jeter un regard ‘cast a glance’), lancer ‘launch’ (lancer une attaque ‘launch an attack’), livrer ‘carry out’ (livrer une bataille ‘engage a battle’), passer ‘make’ (passer un coup de téléphone ‘make a phone call’), porter ‘carry’ (porter un coup ‘strike’), poser ‘put’ (poser une question ‘ask a question’), prendre ‘take’ (prendre une décision ‘make a decision’), prêter ‘lend’ (prêter attention ‘pay attention’), procéder (à) ‘proceed (to)’ (procéder a une lecture ‘proceed to a reading’), subir ‘undergo’ (subir une métamorphose ‘undergo a metamorphosis’) (9) asestar ‘give’ (asestar un golpe ‘hit’), coger ‘catch’ (coger envidia ‘get envious’), cometer ‘commit’ (cometer un asesinato ‘commit murder’), contraer ‘contract’ (contraer matrimonio ‘marry’), correr ‘run’ (correr peligro ‘be in danger’), dar ‘give’ (dar un paseo ‘take a walk’), decir ‘say’ (decir un cumplido ‘pay a compliment’), dirigir ‘direct’ (dirigir una crítica ‘criticize’), echar ‘throw’ (echar una mirada ‘give a look’), efectuar ‘carry out’ (efectuar un pago ‘make a payment’), ejercer ‘make’ (ejercer presión ‘put pressure’), experimentar ‘experience’ (experimentar una mejora ‘improve’), llevar a cabo ‘carry out’ (llevar a cabo una privatización ‘privatize’), meter ‘put’ (meter una bofetada ‘slap’), pegar ‘give’ (pegar una patada ‘kick’), perder ‘lose’ (perderle el respeto (a alguien) ‘lose respect (for someone)’), pillar ‘catch’ (pillar celos ‘get jealous’), presentar ‘present’ (presentar la dimi-

light verb constructions in basque and romance

185

sión ‘resign’), prestar ‘lend’ (prestar atención ‘pay attention’), proceder (a) ‘proceed (to)’ (proceder a una votación ‘proceed to a vote’), propinar ‘give’ (propinar una paliza ‘give a beating’), realizar ‘carry out’ (realizar una transacción ‘carry out a transaction’), soltar ‘let out’ (soltar un grito ‘shout’), sufrir ‘undergo’ (sufrir un accidente ‘have an accident’), tener ‘have’ (tener hambre ‘be hungry’), tomar ‘take’ (tomar represalias ‘take reprisals’) The above lists, composed in alphabetical order, mask crucial asymmetries among the verbs: notably, they put basic LV s like French donner ‘give’ or Spanish echar ‘throw’ at the same level as other “heavier” verbs like French diriger ‘direct’ or Spanish ejercer ‘exert’. Thus, for the more restricted LV s, there is frequently a true lexical, i.e., non-semantic relation between the LV and the NVE (see also Alonso-Ramos 1998). This is clearly the case with LV s like Spanish correr ‘run’, selected by the NVE peligro ‘danger’ in correr peligro ‘be in danger’. Other verbs can arguably be rendered as register-marked alternates of more basic LV s. For instance, Spanish ejercer presión ‘exert pressure’ constitutes a formal rendition of hacer presión ‘make pressure’. French and Spanish, with Basque, converge, therefore, with the rest of languages in what constitutes the basic repertoire of LV s: DO, GIVE, HAVE, TAKE, etc. Nonetheless, as we will see below, there is a difference between French and Spanish as regards the choice of their most basic activity verb, which is faire ‘do’ for the former and dar ‘give’ for the latter. Another important difference between, this time, French on the one hand and Basque and Spanish on the other, is that DO may head stative, attributive predicates in the former, but not in the latter: Ce monsieur fait vieux (Giry-Schneider 1986: 50) “This gentleman does old” ‘This gentleman has an old appearance’. This brings French faire closer to Catalan fer than to Spanish hacer or Basque egin. In section 1.1, we mentioned that the choice of a given LV is not completely free, but that the semantic properties of the NVE may determine the LV chosen. This can be seen when comparing, for example, English and Basque LVC s (see (10)): as Martinez (2015: 78) points out, most English LVC are also found in Basque. However, this is only a tendency, and English and Basque semantically equivalent LVC s may also bear different LV s—cf. English have and give vs Basque egin ‘do’ in (11)—without any relevant difference regarding event- and argumentstructure properties:

186

acedo-matellán and pineda

(10) English take a break take part give a choice give a kiss give a sight give (an) account have an intention do a trick

(11) English have a cry give a shout have a thought have a dream give a call

Basque atseden hartu parte hartu aukera eman musu eman begi eman kontu eman intentzio eduki tranpa egin

Basque negar egin “do weep” garrasi egin “do shout” hausnar egin “do meditation” amets egin “do dream” dei egin “do call”

This random comparison between English and Basque, and particularly the mismatches in (11), actually lead us to discuss another important issue that was introduced in section 1.1: the division of syntactic and semantic labour in LVC s. Thus, whereas the semantic contribution of the LV is limited to functional notions related to argument and event structure, the NVE encapsulates the conceptual content of the LVC (the particular event evoked). Interestingly, Basque offers some examples where the same NVE can combine with different LV s, the resulting LVC s showing, however, no difference in meaning, at least at the conceptual level: (12) a. b. c. d. e. f. g.

argi egin/eman “do/give light” ‘light up, enlighten’ arnasa egin/hartu “do/take breath” ‘breathe’ atsegin egin/eman “do/give pleasure” ‘please’ dei egin/eman “do/give call” ‘call’ esker egin/eman “do/give thanks” ‘thank’ huts egin/eman “do/give failure” ‘fail’ min egin/eman “do/give pain” ‘hurt’

187

light verb constructions in basque and romance

Moreover, in the Romance area, similar examples are found when comparing Spanish and French:

(13) Spanish

French

dar miedo “give fear” faire peur “do fear” dar vergüenza “give shame” faire honte “do shame” dar pena “give sorrow” faire de la peine “do of the sorrow”

Meaning ‘frighten’ ‘shame’ ‘provoke sorrow’

However, these facts do not necessarily constitute counterexamples to the former claim regarding the division of labour between the LV and the NVE. Note that in all these pairs from Basque and Romance the event and argument structure remains the same: for example, both dar miedo “give fear” and faire peur “do fear” require a source of the emotion (the stimulus) as well as a dativemarked argument designating the experiencer of the emotion. Thus, it seems that these exceptional cases can be explained via the semantic closeness of DO and GIVE, attested in many languages, both verbs being usually interchangeable in the context of LVC s.12 The particular case of (12)b, where the alternation is between egin ‘do’ and hartu ‘take’, can easily be accounted for if we consider that arnasa, when combined with the former, means ‘breathing’, and when combined with latter, means ‘exhaled air’. In other cases, though, egin ‘do’ and eman ‘give’, when combined with a given NVE, yield LVC s with very different meanings: (14) a. b. c. d. e. f.

12

damu egin ‘hurt’ ~ damu eman ‘cause regret’ errieta egin ‘argue / tell off’ ~ errieta eman ‘tell off’ hitz egin ‘talk’ ~ hitz eman ‘promise’ hots egin ‘call / sound’ ~ hots eman ‘guide / stimulate’ ikara egin ‘tremble’ ~ ikara eman ‘scare’ kontu egin ‘look after, take care / suppose’ ~ kontu eman ‘give an explanation, give account’

See, for example, the almost systematic equivalence between Spanish dar ‘give’ and Catalan fer ‘do’: dar un beso ~ fer un petó ‘kiss’, dar un abrazo ~ fer una abraçada ‘hug’, dar miedo ~ fer por ‘frighten’, dar pena ~ fer pena ‘provoke sorrow’, dar rabia ~ fer ràbia ‘anger’, dar lástima ~ fer llástima ‘provoke pity’, dar envidia ~ fer enveja ‘provoke jealousy/envy’, and so on. Moreover, both verbs turn out to be even more similar, as far as argument structure is

188

acedo-matellán and pineda

Actually, obtaining two LVC s with different meanings and different eventand argument-structure properties is what normally occurs in Basque when different LV s are combined with one same NVE (see also Martinez 2015: 114; Zabala 2004: 457–458, 483). And this is what one expects given the abovementioned division of labour between the two components of the LVC. Thus, to the examples in (14) for the pair egin ‘do’/eman ‘give’, we can add the following ones with other LV s, including hartu ‘take’ and eduki ‘have’: (15) a. b. c. d. e. f. g. h.

arrazoi ‘reason’ eman ‘agree with’ ~ arrazoi eduki ‘be right’ atsegin eman ‘please’ ~ atsegin hartu ‘enjoy, take pleasure in’ damu hartu ‘regret’ ~ damu eman ‘cause regret’ esku ‘hand’ hartu ‘participate, take part’ ~ esku eman ‘help’ hats hartu ‘breathe’ ~ hats eman ‘blow’ kontu hartu ‘take care of, keep in mind’ ~ kontu eduki ‘be careful’ min hartu ‘get hurt’ ~ min eman ‘cause pain’ pena hartu ‘get sad’ ~ pena eman ‘sadden’

The examples in (15), where verbs other than egin ‘do’ are at play, clearly support Ramchand’s (2014) approach to LVC s, in the sense that the argument and event structure of a given LV is identical to those of their non-LV counterpart. For example, min hartu ‘get hurt’ behaves like liburua hartu ‘get, receive the book’; and min eman ‘give, cause pain’ behaves like liburua eman ‘give the book’. In other words, light eman ‘give’ and hartu ‘take’ display a meaning very close to that of their non-light counterparts, and preserve the number of arguments and their categorial requirements, generally selecting for NP s.13 By contrast, as will be shown in section 2.2, egin ‘do’ allows a wide range of NVE s, including instrumental and locative case-marked DP s, adverbials, and even words not used outside the particular LVC (Zabala 2004: 481–482). Thus, it seems clear that there is a difference between instances with egin ‘do’, clearly a light verb, and instances with eman ‘give’ or hartu ‘take’, which, even when appearing in this type of expressions, have been considered verbs with full semantic content (Zabala 2014: 458, 461). Actually, Zabala’s difference between DO-verbs (light) and GIVE/TAKE-verbs (fully contentful) could account for the many intra- and

13

concerned, once we adopt the view that the dative, commonly appearing with GIVE, is an “added argument” in the sense of Pylkkänen (2002, 2008)—see McFadden (2006) for relevant argumentation in the context of Germanic. Thus, under the applicative hypothesis, it is not the case that GIVE has a slot for the dative object which DO lacks, but instead both verbs can combine with a dative-introducing head, the applicative. And occasionally DP s: bide(a) eman “give (the) way”, hats(a) eman “give (the) breath”, gogo(a) hartu “take (the) mind”.

light verb constructions in basque and romance

189

cross-linguistic equivalences found, as the ones in (12)–(13). An alternative way to deal with this difference is in terms of the greater or lesser amount of structure lexicalized by the verb (Ramchand 2008): egin ‘do’ is more abstract and thus probably lexicalizes only the eventive head v. This is in tune with the fact that egin ‘do’ instantiates the least marked option to form LVC s, given its wide range of meanings and its flexibility in taking different types of complements; on the contrary, the more marked options are those that resort to the verbs eman ‘give’ or hartu ‘take’, which also form LVC s but are indistinguishable from their “heavy” counterparts, along the lines of Ramchand (2014). Particular LV s thus impose their argument- and event-structure properties to the resulting LVC. The predicative nature of the LV is also reflected in the fact that it combines with a particular semantic type of complement: one denoting an event. According to Rodríguez & García Murga (2003), when the NVE combines with egin ‘do’, there is a semantic process a la Jackendoff (1990) that turns a noun referring to a “thing”, e.g. ‘smile’ or ‘word’, into a noun referring to an event: ‘smiling, talking’.14 Interestingly, as will be shown in section 2.2, LV s can combine with different categories, including DP s, instrumental case-marked DP s, inessive case-marked DP s or adverbials (16), to which, according to these authors, the very same semantic process applies yielding the final meaning of the LVC. In other words, the NVE is always interpreted as denoting an event, no matter what syntactic category it bears.15 The fact that the very same event can be expressed by different categories (17) seems to provide evidence for this: (16) a. b. c. d.

eztula egin “do the cough” ‘cough’ hega-z egin “do flight-instr” ‘fly’ dantza-n egin “do dance-ines” ‘dance’ lasterka egin “do running (adv)” ‘run’

(17) a. b. c. d.

eztul / eztula egin “do cough/the cough” ‘cough’ hega-z / hega-n egin “do flight-instr/flight-ines” ‘fly’ dantza / dantza-n egin “do dance.abs/dance-ines” ‘dance’ laster / lasterka egin “do run/running (adv)” ‘run’

14

15

We use the term thing following Rodríguez & García Murga’s (2003: 433) terminology: “Baina barre gauza bati egiten dio erreferentzia, ez egoera bati.” ‘But barre refers to a thing, not to a situation.’ It seems to us that this concept must be understood in a broad sense, beyond material objects. The “semantic coercion” of the NVE as denoting an event is reminiscent of Marantz’s (2005) account of creation predicates like Bake a cake, in which the DP a cake is also taken to denote an event. An updated view of this type of syntactically driven semantic coercion is that expounded by Wood & Marantz (2017) in terms of contextual allosemy.

190

acedo-matellán and pineda

Other data seem to contradict the above-mentioned division of labour between the LV (argument- and event-structure properties) and the NVE (conceptual content). Zabala (2004: 478–479), for instance, argues that the argument structure of the LVC largely depends on the particular NVE that the LV combines with. Thus, in line with Grimshaw & Mester’s (1988) claims for Japanese, she assumes that the LV is thematically incomplete and that the NVE thus transfers to it some or all its arguments, yielding the resulting LVC. An argument put forward in favour of this “fusion” of the argument structure of the NVE and the LV is that the NVE imposes restrictions on the type of external argument of the LVC (Zabala 2004: 479–480). In particular, with meteorological LVC s the subject is a quasi-argument that cannot be phonologically realised, and that is only reflected in the ergative agreement in the auxiliary: (18) a. Euri-a egin du. rain-det.abs do aux.prs.erg3sg.abs3sg ‘It has rained.’ b. *Horr-ek / *Zeru-a-k euri-a egin this-erg / sky-det-erg rain-det.abs do du. aux.prs.erg3sg.abs3sg ‘This/The sky has rained.’ Still, it is worth noting that examples such as the next one, while certainly marked, are possible: (19) “Nor-k egi-ten du euri? […]” egi-ten who-erg do-inf aux.prs.erg3sg.abs3sg rain do-inf dio galde Strepsiades zaharr-a-k aux.prs.erg3sg.abs3sg.dat3sg ask Strepsiades old-det-erg Sokrates-i, eta filosofo-a-k honela erantzu-ten: “Ez Socrates-dat and philosopher-det-erg thus answer-inf not behintzat Zeus-ek, hodei-ek baizik” at_least Zeus-erg cloud-det.erg.pl but “Who rains? […]”, asked old Strepsiades for Socrates, and the philosopher answered in this way: “Certainly not Zeus, but the clouds do.” (Dakiguna ikasten, Patziku Perurena, Alberdania, 2007, p. 132) What is worth noticing about examples like (19) is that, while the NVE certainly shapes the conceptual scene evoked (i.e., the appearance of rain is usually

light verb constructions in basque and romance

191

not conceived as a caused event), it cannot grammatically preclude an option licensed by the LV (i.e., egin allows the projection of an external argument). Moreover, the other contrasts (20)–(22) upon which Zabala bases her defence of such a “fusion” actually do not differ much from what is found in any transitive construction, where the verb “expresses a range of predicates depending on the choice of direct object” (Marantz 1984: 25). Thus, it is expected that the LV egin ‘do’ requires different arguments depending on the NVE that it combines with: in (20) egin combined with hitz ‘word’ admits a dative argument, in (21) egin combined with lo ‘sleep’ does not admit a dative argument, and finally in (22) egin combined with laztan ‘hug, kiss’ generally requires a dative argument. (20) a. Aduna-k hitz egin du. Aduna-erg word do aux.prs.erg3sg.abs3sg ‘Aduna has talked.’ b. Aduna-k gizon-a-ri hitz egin Aduna-erg man-det-dat word do dio. aux.prs.erg3sg.abs3sg.dat3sg ‘Aduna has talked to the man.’ (21) a. Aduna-k lo egin du. Aduna-erg sleep do aux.prs.erg3sg.abs3sg ‘Aduna has slept.’ b. *Aduna-k gizon-a-ri lo egin Aduna-erg man-det-dat sleep do dio. aux.prs.erg3sg.abs3sg.dat3sg ‘Aduna has slept to the man.’ (22) a. *Aduna-k laztan egin du. Aduna-erg hug do aux.prs.erg3sg.abs3sg ‘Aduna has hugged.’ b. Aduna-k laztan egin dio Aduna-erg hug do aux.prs.erg3sg.abs3sg.dat3sg gizon-a-ri. man-det-dat ‘Aduna has given the man a hug.’

192

acedo-matellán and pineda

Zabala’s (2004) position for Basque is similar to that defended by other authors dealing with LVC s in French and Spanish. For Spanish, Alonso-Ramos (1998) claims that the diathesis of LV s is special, since their syntactic valency changes according to the nominal NVE with which they combine—much as has been shown for Basque in the above examples. She illustrates with the contrast between tener miedo ‘have fear’ ‘be afraid’ vs tener esperanza ‘have hope’; while the former takes a dative encoding the stimulus, the latter can only take a genitive PP: (23) a. Rosa le tiene miedo a la oscuridad. Rosa dat has fear at the darkness ‘Rosa is afraid of darkness.’ b. Rosa tiene esperanza de un cambio. Rosa has hope of a change ‘Rosa hopes for a change.’ In a smilar vein, Rafel (2004: 404) also discusses how the valency of tener ‘have’ changes according to its use in a LVC, as has been shown above, or in a run-ofthe-mill transitive construction: (24) *Juan le tiene un coche (a su madre). Juan dat has a car at his mother Rafel points out that the licensing of the dative depends on the use of the NVE miedo ‘fear’, and that the dative is not licensed in a normal transitive predicate such as the one in (24). An extreme case in the position that the argument structure of the LVC basically depends on that of the NVE is that represented by Gross (1981) or Danlos (1992: 2), the latter considering that the function of the LV (“the support verb”) is “mainly to bear the tense and aspect values of the sentence”. There are several observations to be made in relation to the above facts, which we hope to show to conform with, rather than jeopardize, Ramchand’s (2014) conjecture on the division of labour between the LV (= argument and event-structure) and the NVE (= conceptual content), as adhered to in this paper. The first thing to notice is that, for instance, tener miedo may also surface with a genitive PP encoding the stimulus, undermining the more deterministic stance taken by Alonso-Ramos (1998): (25) Rosa tiene miedo de la oscuridad.

light verb constructions in basque and romance

193

What the alternation between this example and (23)a shows, we submit, is that, while the use of the NVE miedo ‘fear’, as opposed to esperanza ‘hope’, does indeed determine the presence of a stimulus in the conceptual scene evoked, that “argument” may find different ways of realization in the actual sentence, either as a genitive or as a dative. Perhaps more importantly, it can be dropped altogether, the grammaticality of the sentence remaining unaltered:16 (26) Rosa tiene miedo. Quite crucially, even in (26) the presence of the stimulus does not fail to be inferred: when we are afraid we are certainly always afraid of something. Grammatically, however, there is no need for the stimulus to be expressed in the LVC, and this, we note, must have to do exclusively with the syntactic properties of the construction, as headed by the LV.17 It is interesting, in this respect, to contrast the LVC at hand with the corresponding synthetic verb temer ‘fear’: (27) Rosa teme *(la oscuridad). 16

17

In some cases the complement of the NVE does seem obligatory: (i) Pepa hizo alusión *(a los problemas financieros). Pepa did hint at the problems financial ‘Pepa hinted at the financial problems.’ In these cases, however, we think that the NVE functions a complex event nominal in the sense of Grimshaw (1990), and its complement (a los problemas financieros ‘at the financial problems’) is a true structural complement that cannot, thefore, be dropped. This discussion is related to observations made by Gross (1976) and Giry-Schneider (1978) for French and taken up, among others, by Alonso-Ramos (1998) and Rafel (2004) for Spanish, namely, that the argument introduced by the NVE—cf. la oscuridad ‘darkness’ in (23)—can easily be reanalysed as a constituent of its own (Danlos 1992:10): (i) a. Jean a commis [une aggression contre Marie]. Jean has committed an aggression against Marie b. Jean a commis [une aggression] [contre Marie]. Constituency tests like pronominalization, cleft formation, or relativization indeed show that, e.g., contra Marie ‘against Marie’ above may be analysed as an independent constituent (see the mentioned works for details). However, this does not make contra Marie ‘against Marie’ a true argument, at the same level as une aggression ‘an aggression’, as proposed by Alba-Salas (2002: 13) and Rafel (2004: 424) for Spanish, since while the former can be dropped, the latter cannot. Moreover, “le faire”, i.e., do-so substitution, can strand the PP contre Marie, further revealing its adjunct status: (ii) Jean a commis une aggression et il l’ a fait contre Marie. Jean has committed an aggression and he it=has done against Marie All in all, while “secondary arguments” like contra Marie ‘against Marie’ are certainly participants of the conceptual scene, they are however not licensed as true structural arguments of the construction.

194

acedo-matellán and pineda

While tener miedo “have fear” ‘be afraid’ and temer ‘fear’ mean more or less the same, only the former can drop the constituent encoding the stimulus, suggesting that, indeed, argument structure qua argument realization is a prerogative of the head of the construction, i.e., the verb, either heavy or light, rather than of the NVE. With respect to (22), we observe that the dative is not completely out in predicates headed by non-light tener ‘have’: (28) La madrina les tiene regalos a sus ahijados. the godmother dat.pl has presents at her godchildren ‘The godmother has presents in store for her godchildren.’ This shows, again, that the structural licensing of the dative does not really depend on the complement of tener ‘have’, but on some independent configurational condition.18 It should have become clear, we hope, that we are making a crucial distinction between the arguments licensed structurally, by the configuration headed by the LV, and the participants of the event, encoded as a conceptual property of the NVE, and which are not obligatorily realized.19 2.2 The Non-verbal Element Basque LVC s are usually formed by the combination of a LV and a bare nominal element, as shown in all the examples above. Occasionally, the NVE can also be a postpositional phrase (29) or an adverbial expressing an activity (30) (Etxepare 2003: 402–403; Zabala 2004: 466–467; Martinez 2015: 114–126): (29) a. Hega-n/Hega-z egin wing-ines/wing-instr do ‘Fly’ b. Gain-ez egin top-instr do ‘Overfly’ 18 19

See Pineda (2016) for an implementation of the theory of applicatives (Pylkkänen 2002, 2008; Cuervo 2003) to the analysis of datives in Romance and Basque. We follow in this respect the different between structure and content arguments in Grimshaw’s (2005 [1993]) terms and structure and content participants in Rappaport Hovav & Levin’s (1998) terms. And see also, within Construction Grammar, Goldberg (1995). Notwithstanding our criticism of Alonso-Ramos’s (1998) take on tener miedo/esperanza “have fear/hope”, we acknowledge that the difference that we are pointing out roughly corresponds to her distinction between actant syntaxique and actant sémantique.

light verb constructions in basque and romance

195

c. Atze-ra egin back-adl do ‘Go back’ d. Borroka-n egin fight-ines do ‘Fight’ e. Aitzurr-ean egin hoe-det-ines do ‘Dig’ f. Ardatz-ean egin spindle-ines do ‘Spin’ g. Kart-etan egin cards-ines do ‘Play cards’ h. Futbol-ean egin football-ines do ‘Play football’ (30) a. Hagin-ka egin tooth-adv do ‘Bite’ b. Adar-ka egin horn-adv do ‘Gore’ c. Aitzur-ka egin hoe-adv do ‘Dig’ Some LVC s alternate between a version with a bare nominal NVE and a version with a locative or adverbial NVE, as the ones shown in (29) and (30) respectively—note that -ka is an adverbializing suffix. Unlike the versions involving a bare nominal NVE, those involving a non-nominal NVE show an iter-

196

acedo-matellán and pineda

ative meaning (Etxepare 2003: 402–403) or a nuance of a frequently repeated activity (Martinez 2015: 120–126). In addition, the versions with bare nominal NVE s are generally more frequent, except in the case of korrika egin “running (adv.) do”, according to Martinez’s corpus study: (31) a. Arrantza(-n) egin fishing(-ines) do ‘Fish’ b. Arraun(-ean) egin oar/rowing(-ines) do ‘Row’ c. Borroka(-n) egin fight(-ines) do ‘Fight’ d. Dantza(-n) egin dance(-ines) do ‘Dance’ e. Ehiza(-n) egin hunting(-ines) do ‘Hunt’ f. Errieta(-n) egin dispute(-ines) do ‘Dispute’ (32) a. Atzapar(-ka) egin claw(-adv) do ‘Scratch’ b. Borbor(-ka) egin Bubbling(-adv) do ‘Bubble’ c. Eztul(-ka) egin cough(-adv) do ‘Cough’

light verb constructions in basque and romance

197

d. Jauzi(ka)/Salto(ka) egin jump.adv do ‘Jump’ e. Korri(-ka) egin running(-adv) do ‘Run’ f. Laster(-ka) egin run(-adv) do ‘Run’ g. Musu(-ka) egin kiss(-adv) do ‘Kiss’ h. Oihu(-ka) egin scream-adv do ‘Scream’ i. Zotin(-ka) egin hiccup(-adv) do ‘Hiccup’ Additionally, the NVE can also be an adjective, as in the LVC on hartu ‘accept’ (lit. “good take”) and a particle, as in goiti egin ‘vomit’ (lit. “do up”), although these two options are “relatively rare and present a high degree of lexicalisation” (Oyharçabal 2006: 787).20 It is also worth mentioning that the NVE of meteorological LVC s bero egin “hot do” ‘be hot’ and hotz egin “cold do” ‘be cold’ can function as a noun or as an adjective (Martinez 2015: 54–56). Focusing now on nominal NVE s, they are inanimate nouns, and they are non-referential (Martinez 2015: 44; Zabala 2004: 470).21 Other than that, even though the range of nominals that can appear in Basque LVC is a closed set, there is no semantic property that can apply to all of them: nominals can be abstract (gogoeta egin “do meditation” ‘meditate’) or concrete (buru egin “do 20 21

Note that on hartu, actually written onartu ‘accept, admit’, is fully lexicalized and behaves as a transitive verb: legea onartu ‘accept the law’. A few NVE s, such as turrut (egin) ‘mock / fail’ or laprast (egin) ‘slip’, are not found in syntactic environments other than LVC s (Zabala 2004:449).

198

acedo-matellán and pineda

head” ‘face’), countable (hitz egin “do word” ‘talk’) or uncountable (min egin ‘do harm’ ‘hurt’), and eventive (arrantza egin ‘do fishing’) or resultative (laprast egin “do slip” ‘slip’) or with no argument or event structure at all (elurra egin “do snow” ‘snow’) (Rodríguez & García Murga 2003: 420). Thus Basque contrasts with Japanese, where, according to Grimshaw & Mester (1988), NVE s are always event nouns that transfer their argument structure to the LVC, as mentioned in section 1.2 (see Fernández 1997 for a comparison between Japanese and Basque LVC s). However, Grimshaw & Mester’s proposal can be maintained for Basque under a slightly different view: as argued above, regardless of the type of noun and the readings it may have in other syntactic environments, the NVE, in the context of the LVC, always expresses an event. Recall that considering all Basque NVE s as denoting events goes in line with Rodríguez & García Murga’s (2003: 432–434) semantic analysis, according to which the NVE may refer to a thing, but the final interpretation is achieved at the conceptual level by means of a semantic process (Jackendoff 1990) ensuring that the NVE no longer refers to a thing, but to an event. In addition, given that events inherently lack referenciality, the above-mentioned non-referenciality of Basque NVE s is accounted for (Zabala 2004: 470). The lack of referentiality is also a property of a small group of Basque NVE s that are not bare nominals but bear a determiner (Martinez 2015: 33–34, 184– 185). These LVC s mostly refer to meteorological phenomena and their NVE bears the definite article -a (the only determiner possible in these structures):22 22

Thus, other determiners such as bat ‘a, one’, the demonstratives hau/hori ‘this, that’ (which induce a necessarily specific interpretation) or the plural definite article -ak are not allowed in NVE s. Examples can be found where instead of -a the quantifier asko ‘very much, a lot’ is used, as in (i), but according to Martinez (2015:58, fn. 64) asko refers in this case to the whole meteorological situation, rather than quantifying the noun euri: (i) Euri asko egin du. rain a_lot do aux.prs.erg3sg.abs3sg ‘It has rained a lot.’ This contrasts with its use with common direct objects, such as in (ii), where asko generally triggers the plural absolutive agreement -it- in the verb (actually, there is dialectal variation in this point, with some dialects accepting (ii) with the singular auxiliary du): (ii) Liburu asko irakurri ditu. book a_lot read aux.prs.erg3sg.abs3pl ‘(S)he has read a lot of books.’ However, note that agreement does not constitute a reliable test here: even if asko quantified euri, we would not see the morphological effect on the verb, since euri is uncountable and thus never triggers plural agreement on the verb. Actually, in our view, the null hypothesis involves assuming that asko is quantifying directly over euri in Euri asko egin du. This scenario is reminiscent of what Bosque & Masullo (1998) argue is involved in Spanish synthetic predicates like sangrar mucho ‘to bleed a lot’. These authors propose that the degree

light verb constructions in basque and romance

199

eguzkia egin “do sun”, elurra egin “do snow”, euria egin “do rain”, haizea egin “do wind”, izotza egin “do ice”, trumoia egin “do thunder”, and so on (Martinez 2015: 56–60).23 In a few other LVC s, again mostly meteorological, the determiner is optional: hotz(a) egin “do cold” ‘be cold’ and bero(a) egin “do hot” ‘be hot’, eztul(a) egin “do cough” ‘cough’ and arnas(a) egin “do breath” ‘breath’. In any case, regardless of whether the NVE is determined or not, it is not referential, but has a generic interpretation (Zabala 2004: 470; Martinez 2015: 56– 60). Since the NVE in meteorological LVC s is usually determined but still nonreferential, non-referentiality cannot be directly related to the absence of the determiner. Another reason that prevents us from establishing this connection is that in common transitive structures a-marked DP s can be found that are not referential. If we compare the LVC lan egin “do work” and the common transitive structure lana egin (lana bearing a determiner -a), the latter has to be interpreted with a definite reference (“the job, that job”), but if we look at other common transitive structures such as etxea egin ‘build a/the house’ or ogia jan ‘eat (the) bread’, whose direct objects cannot appear in an LVC, then the determined direct objects can be either definite (‘the house, that house’, ‘the bread, that bread’) or indefinite (‘a house’, ‘bread’) (Martinez 2015: 63).24 In short, Basque arguments always need to be determined, even if they are generic. Thus, as Himmelmann (2001: 832) points out: “Definiteness, though undeniably of central importance to the grammar and typology of the articles, is only one of a number of meanings that can be conveyed by articles”. Under a wider conception of LVC s, one can find in Basque some constructions where the LV egin combines with a determined DP, such as bizarra egin “do the beard” ‘shave’, harrikoa egin “do the washing up”, erroak egin “do the roots” ‘settle down’, bakeak egin “do the peaces” ‘make peace’. We think that these few cases are the ones that can be considered constructions involving Vague Action Verbs in Basque, where the NVE seems to pattern with full-fledged argumental DP s—see Kearns’s (1988) distinction in section 1.2. In turn, these

23 24

modifier mucho ‘a lot’ quantifies over a nominal predicate SANGRE ‘blood’, embedded in the lexico-syntactic representation of the verbal phrase, involving the abstract LVC-like structure DO BLOOD (Hale & Keyser 1993, 2002). The difference between Spanish and Basque would lie in the synthetic vs analytic character of the predicate. On the quantification of the NVE in Basque LVC s, see examples (58) below and Oyharçabal (2006: 792). It is not surprising that meteorological LVC s are considered a particular class of Basque LVC s by several authors (Zabala 2004; De Rijk 2008; Martinez 2015). An anonymous reviewer points out that the contrast may need to be nuanced, as something like lan-a egin “work-a do” can be non-referential too in central dialects of Basque.

200

acedo-matellán and pineda

Vague Action Verbs constructions should be differentiated from others featuring egin as a non-light verb: ogia egin ‘make the bread’, lana egin ‘do the work’, ohea egin ‘make the bed’ (Rodríguez & García Murga 2003: 419, 424; Zabala 2004: 476, Martinez 2015: 115–126). These particular instances aside, and as mentioned in section 1.2, Basque LVC s are generally considered constructions headed by True Light Verbs. Thus, even if, as will be shown in section 3, in some LVC s the NVE may pattern with full-fledged argumental DP s in some respects (word order, partitive case, causative formation), NVE s and argumental DP s are fundamentally different. Most importantly, the former are overwhelmingly bare nominals, whereas the latter need to be determined to be licensed. Turning to Romance languages, we observe that some LVC s have a synthetic correspondent, whereas others do not. The lists below also show that, unlike Spanish (33), in French LVC s (34) the NVE is usually introduced by a determiner or a partitive (Martinez 2015: 77–78): (33) a. hacer fiesta “do holiday” ~ festejar ‘celebrate’, hacer mención “do mention” ~ mencionar ‘mention’, hacer juramento “do oath” ~ jurar ‘swear’, hacer elogio “do praise” ~ elogiar ‘praise’, hacer chantaje “do backmail” ~ chantajear ‘blackmail’, hacer daño “do damage” ‘hurt’ ~ dañar ‘damage’, hacer ayuno “do fast” ~ ayunar ‘fast’ b. hacer deporte ‘do sport’, hacer mérito “do merit” ‘mention’, hacer juerga “do fun” ‘have fun’, hacer luto “do mourning” ‘mourn’, hacer turismo ‘do tourism’, hacer dinero ‘make money’ (34) a. faire mention “do mention” ~ mentionner ‘mention’, faire (de) la cuisine “do (of) the cooking” ~ cuisiner ‘cook’, faire la recolte “do the harvest” ~ recolter ‘harvest’, faire du progrès “do of the progress” ~ progrésser ‘progress, advance’, faire le nettoyage “do the cleaning” ~ nettoyer ‘clean’, faire des études “do of the studies” ~ étudier ‘study’, faire de l’ ombre “do of the shadow” ~ ombrager ‘cast a shadow on’ b. faire du sport “do of the sport” ‘do sport’, faire du tourisme “do of the tourism” ‘do tourism’, faire la ronde “do the round” ‘go the rounds’, faire la revue “do the review” ‘review’, faire l’amour “do the love” ‘make love’, faire la guerre “do the war” ‘wage war’, faire fortune “do the fortune” ‘make one’s fortune’. The correspondence between LVC s and synthetic predicates in these languages is far from straightforward. Following Piera & Varela’s (1999) discussion on Spanish, we can distinguish the next four possible situations:

light verb constructions in basque and romance

201

1)

Morphological and semantic affinity: hacer transbordo “do transfer” ~ transbordar ‘transfer’; French: faire mention “do mention” ~ mentionner ‘mention’; some doublets exist whose members differ in register value, as in French faire la guèrre “do the war” ‘wage war’ (less formal) ~ guerroyer ‘wage war’ (more formal) 2) Morphological affinity without (complete) semantic equivalence: hacer fiesta “do holiday” ~ festejar ‘celebrate’, hacer daño “do damage” ‘hurt’ ~ dañar ‘damage’; French: faire le récit “do the account” ‘narrate, tell’ ~ réciter ‘recite’ (Giry-Schneider 1986) 3) Semantic equivalence without morphological affinity: hacer punto “do stitch” ‘knit’ ~ tricotar ‘knit’, dar asco “give disgust” ‘disgust’ ~ repugnar ‘disgust’; in some cases there is a semantic equivalence involving however a difference in register, as in French faire l’ amour ‘make love’ ~ baiser, niquer ‘fuck’. 4) LVC s that do not correspond to any synthetic verb, cf. (33)b and (34)b) Similarly to what has been pointed out for Basque above, we observe that many LVC s in French present an NVE that is truly bare, and may even not be a nominal at all: (35) avoir faim/ froid/envie/besoin “have hunger/fancy/need” ‘be hungry/cold, fancy, need’ (36) faire peur/honte/envie “do fear/shame/envy” ‘provoke fear/shame/provoke envy’ (37) prendre plaisir “take pleasure” ‘enjoy’ (38) prêter attention “lend attention” ‘pay attention’ As pointed out by Acedo-Matellán (2014) for the avoir cases, these NVE s are truly bare, since they are caseless (cf. failure of en-substitution) and cannot sustain nominal quantification (cf. availability of très ‘very’, but not of beaucoup de ‘a lot of’): (39) —Tu as faim / froid? —* Oui, j’en ai. —you has hunger / cold? — yes, I=ptve have.1sg (40) J’ai {très / *beaucoup de} faim / froid. I=have very / a lot of hunger / cold

202

acedo-matellán and pineda

(41) —Les serpents font peur à Magali? —* Non, ils ne lui en —the snakes do fear at Magali? — no they neg dat ptve font pas. do.3pl neg (42) Les serpents font {très / *beaucoup de} peur à Magali. the snakes do very / a lot of fear at Magali (43) —Thomas prend plaisir à marcher? —* Oui, il en y prend. —Thomas takes pleasure at walk.inf — yes he ptve loc takes In Spanish the corresponding LVC s do admit nominal quantification, as in Tengo mucha hambre “I have much.f hunger” ‘I am very hungry’. Even more conspicuously, and in spite of its proximity to French, the Catalan counterparts behave differently, in allowing both partitive case (as shown by ne-cliticization) and nominal quantification (cf. the use of molta ‘much.f’): (44) —Que tens gana / fred? —Sí, en tinc. —compl have.2sg hunger / cold? —yes, ptve have.1sg (45) Les serps li’n fan molta, de por, a l’Antoni. the snaked dat=ptve do.pl much.f.sg of fear at the=Antoni ‘Antoni is very much afraid of snakes.’ As also observed by Acedo-Matellán (2014), these bare nouns must occupy an argumental (object) position in the structure, a position that can, therefore, not be occupied by any other DP: (46) Thomas avait envie *(d’) une glace. Thomas had fancy of an ice-cream ‘Thomas fancied an ice-cream.’ (47) Anne-Catherine a faim *(de) fromage. Anne-Catherine has hunger of cheese ‘Anne-Catherine craves some cheese.’ From Acedo-Matellán’s (2014) perspective, these facts would be a problem for an analysis of such predicates in French as involving incorporation, i.e., fusion of the NVE and the LV at some level of representation (but see also section 3). We turn to incorporation analyses and other approaches to the relation between the LV and the NVE in the next section.

light verb constructions in basque and romance

3

203

The Syntactic Relationship between the LV and the NVE

In the generative tradition, Basque LVC s (48) have often been identified with syntactically unergative verbs, which show a synthetic correlate in other languages,25 such as Romance (49)–(50): (48) hitz egin ‘talk’, barre egin ‘laugh’, negar egin ‘cry’, lo egin ‘sleep’, amets egin ‘dream’, salto egin ‘jump’, dantza egin ‘dance’, lan egin ‘work’ (49) Sp. hablar ‘talk’, reír ‘laugh’, llorar ‘cry’, dormir ‘sleep’, soñar ‘dream’, saltar ‘jump’, bailar ‘dance’, trabajar ‘work’ (50) Fr. parler ‘talk’, rire ‘laugh’, pleurer ‘cry’, dormir ‘sleep’, rêver ‘dream’, sauter ‘jump’, danser ‘dance’, travailler ‘work’ Basing on facts like the analytic realization of unergatives in Basque (and other languages), Hale & Keyser (1993) postulated a universal transitive nature for unergative verbs: that is, unergative (i.e. activity) predicates take an internal argument that is not the subject. Thus, the only difference between the Basque analytic patterns in (48) and the Romance synthetic patterns in (49)–(50) is whether overt incorporation of the NVE into the LV takes place or not. Actually, even in Basque many complex unergatives have a simplex counterpart, as shown by dantza egin “do dance” and dantzatu, both meaning ‘dance’, and here too the difference between the two options has to do with whether nominal incorporation has taken place or not. If it does not take place (dantza egin ‘do dance’, ‘dance’), the derivation will parallel that of a transitive clause (zerbait egin ‘do something’). However, there is another dimension where syntactic incorporation becomes a controversial issue in the theoretical treatment of Basque and, to some extent,

25

See Martinez (2015: 21–32) for a very detailed semantic classification of Basque LVC s that takes into account several previous semantic classifications, especially De Rijk’s (2008), but also Rodríguez & García Murga’s (2003: 423), Etxepare’s (2003: 400–402) and Zabala’s (2004: 472–476) among others. As a matter of fact, Etxepare (2003) provides a rich sample of LVC s, distinguishing verbs of emission (sound emission oihu egin ‘yell’, light emission dir-dir egin ‘shine’, verbal emission errieta egin ‘argue / tell off’), internal body motion (dar-dar egin ‘tremble’), physical activities (actions against an object or an individual tiro egin ‘shoot’, motion verbs igeri egin ‘swim’, bodily functions izerdi egin ‘sweat’), mental activities (duda egin ‘doubt’) and behavioral verbs (axut egin ‘challenge’).

204

acedo-matellán and pineda

Romance LVC s.26 Indeed, beyond the obvious analytic/synthetic morphological dichotomy pointed out above, there is a debate as to whether the NVE in Basque LVC s (dantza egin, hitz egin) patterns with DO s in common transitive structures or it undergoes syntactic incorporation. The debate on this issue has been intense and has led to a variety of analyses, some emphasizing the parallelism of LVC s with regular transitive structures, and others focusing on the particularities of the NVE in LVC s and arguing instead that Basque LVC s are a subgroup of unergative verbs with noun incorporation (see Lafitte 1944; Sarasola 1977; Levin 1983; Ortiz de Urbina 1986; Uribe Etxebarria 1989; Hale & Keyser 1993, 1997; Bobaljik 1993; Laka 1993; Austin & López 1995; Fernández 1997; Etxepare 2003; Zabala 2004; Oyharçabal 2006; Martinez 2015, a.o.). The lack of a unified view with respect to this issue surely has to do with the non-homogenous behaviour of NVE s in Basque LVC s. Indeed, they pattern with absolutive objects in some cases, but differ from them in others. This is thus directly related to their case properties: some tests indicate that NVE s are case-marked (with absolutive case), but other tests point in the opposite direction, suggesting that the noun holds a connection with the LV that goes beyond that of a run-of-themill object, along the lines of noun incorporation. In what follows, some pieces of apparently contradictory evidence as regards this issue are presented.27

26

27

The term incorporation has been used to refer to certain phenomena found in a variety of languages in which verbs overtly host material arguably belonging to their complements, like prepositions (as in Chichewa, Baker 1988) or nouns (as in Siberian Koryak, Mithun 1984). Within generative approaches, and since the extremely influential work of Baker (1988), incorporation also refers to the syntactic mechanism proposed to account for these phenomena, to wit, the movement of the head of the verb complement to the immediately c-commanding verbal head (see Haugen 2015 for a recent overview and a new proposal on the nature of incorporation). Since incorporation, in the standard generative approaches, involves head movement, it is worth pointing out that this operation has come under considerable scrutiny at least since Chomsky (1995), who first pointed out the theoretical inconsistencies associated to this operation and proposed to relegate it to the purely morphophonological component of grammar. The debate is far from being settled, with proposals that actual movement (Matushansky 2006) or lack thereof (Brody 2000; Bye & Svenonius 2012) is involved in the phenomena traditionally dealt with under this rubric. For a recent overview, see Barrie & Mathieu (2014), and for interesting theoretical remarks, Haugen (2015). Due to space restrictions, we must refrain from dealing with some other (dis-)similarities between LVC s and common transitive structures, such as the position of NVE s with respect to manner adverbials such as ondo, ongi ‘well’, or gaizki ‘badly’ (Oyharçabal 2006: 794; Martinez 2015: 51, fn. 53), the (non-)acceptability of adjectives and other type of modifiers (García Murga 2003: 421; Zabala 2004: 452, Oyharçabal 2006: 796; Martinez 2015: 47–48) or the (im)possibility of ellipsis in different contexts (Oyharçabal 2006: 793; Martinez 2015: 48–49), among others.

light verb constructions in basque and romance

205

One first argument for the transitive analysis is the incompatibility of LVC s with a DO: if the NVE (dantza) bears absolutive case, a “second” direct object (tango) is expected not to be case-licensed (51)a, in contrast with the synthetic version of the predicate (51)b:28,29 (51) a. *Aduna-k tango-a dantza egin du. Aduna-erg tango-det.abs dance do aux.prs.erg3sg.abs3sg b. Adunak tango-a dantzatu du. Aduna-erg tango-det.abs dance aux.prs.erg3sg.abs3sg ‘Aduna has danced the tango.’ This is also the argument that Acedo-Matellán (2014) wields to defend that the case-less NVE in the French LVC s dealt with at the end of the previous section (e.g., avoir faim “have hunger” ‘be hungry’) occupies the object position, thus constituing an “unincorporated root”. That said, the incorporation analysis, whereby the NVE is caseless and undergoes incorporation, can also account for the contrast in (51). Indeed, invoking Chomsky’s (1981) Empty Category Principle, a “second DO” would fail to be licensed, since the trace left behind by the incorporated DO cannot be occupied by another argument. Following Martinez (2015: 177), the unfeasibility of licensing “second DO s” actually shows that, even for those cases in which there is actually phonological evidence of incorporation (such as hitz egin “do word”

28

29

Berro (2015) argues that the NVE occupies the same position as DO s in simplex verbs, basing on the aspectual properties of the predicate. Thus, just like simplex verbs can be aspectually bounded if the DO denotes a bounded scale, the event denoted in complex verbs is also measured by the scalar (aspectual) properties of its NVE (i.e., the root), thus indicating that they occupy the same position—see Harley (2005) for a seminal work on the parallelism between internal argument DP s and roots as regards their contribution to Aktionsart. A “second DO” is possible with LVC s expressing mental or communicative activities, when it is a subordinate clause (Martinez 2015: 74, 376–377; Zabala 2004:480): (i) Gau-ez, [bere armairuan borrokan zihardutela] amets egi-ten dream do-inf zuen. aux.pst.erg3sg.abs3sg. ‘At night, (s)he used to dream [that they were fighting inside his/her closed].’ However, these apparent double object constructions (which are also found with ditransitive verbs, such as abisatu ‘tell [do-abs me] [do subordinate clause]’) can easily be accounted for under the assumption that completive subordinate clauses don’t need case, or are self-sufficient as far as case is concerned (Pesetsky & Torrego 2001).

206

acedo-matellán and pineda

‘talk’, see below), it is a syntactic process, not a lexical one, unlike that argued to be involved in English laugh, formed in the lexical syntax out of do+laugh (Hale & Keyser 1993, 2002; Haugen 2015). In this latter kind of incorporation no trace is left behind allowing for a DO (a cognate object) to be projected—cf., e.g., laugh a good laugh.30 Both the proponents of the transitive analysis and those of the incorporation analysis have put forward many other arguments, based on a variety of syntactic tests (see below). However, since not all Basque LVC s show a homogenous behavior, most syntactic tests have actually been used to justify both views. In particular, the similarities and contrasts between NVE s and regular DO s arise when the focus is put on the degree of cohesion between the LV and the NVE. As we shall see, the apparently contradictory evidence in fact points out that while some Basque LVC s may be the result of an incorporation process, others cannot be. To makes matters worse, the non-homogenous syntactic behaviour of Basque LVC s does not only depend on the particular LVC, since differences across dialects and registers are also at play (see Oyharçabal 2006), thus mak-

30

Two LVC s in Basque are exceptional in that they allow a DO, namely huts egin “do failure/empty” ‘miss’ and hots egin “do noise/rumor” ‘call, propagate’ (Oyharçabal 2006: 791, fn. 6; Martinez 2015: 172, 377–378): (i) Ume-a-k eskol-ak huts egin ditu. kid-det.abs-erg class-abs.pl failure/empty do aux.prs.erg3sg.abs3pl ‘The kid missed the classes’ (ii) Gizon-a-k berrion-a hots egin man-det.abs-erg good_news-det.abs noise/rumor do du aux.prs.erg3sg.abs3sg ‘The man spread the good news’ In these cases a lexicalization process has taken place, so that the argument structure of the LVC includes a slot for a DO with the Theme-role. That a reanalysis has taken place in these cases is also evinced by phonology, huts egin and hots egin being generally pronounced as single words with a single accent: hutségin, hotségin. In other words, there is no longer any trace preventing the merger of a DO. Rather, there has been a reanalysis yielding a transitive verb able to take a full-fledged DP as complement that can check the structural absolutive case (Martinez 2015: 172, 377–378). The cases in (i) and (ii) could thus be seen as instances of Mithun’s (1984:857) case-manipulating type of incorporation, whereby the case left by the incorporated argument is available for another argument (until then an oblique), which is thus promoted. We think that huts egin—but not hots egin—is amenable to a different analysis, namely, that huts, which can also be an adjective meaning ‘empty’, is a secondary predicate rather than an argument. Specifically, huts would be a predicate of the DO, which would then be its external argument: trena huts egin “make the train empty, missed”, i.e., ‘miss the train’. If this analysis is on the right track, huts egin would pattern with the abovementioned onartu, whose NVE is also an adjective, on ‘good’.

light verb constructions in basque and romance

207

ing it difficult to find systematic patterns that allow to tip the scales towards an analysis or the other. It seems therefore that the most adequate way to proceed is by assuming the existence of syntactically different types of LVC s in Basque. One first type of Basque LVC involves constructions that behave as common transitive structures, formed by a transitive verb and its DO. Evidence of their run-of-the-mill transitive behaviour includes the separability of the LV and the NVE in interrogative sentences and focus structures (see (53)–(56)), and availability of partitive case marking on the NVE in negative sentences (57) (Etxepare 2003: 403): (52) Aduna-k dantza egin du. Aduna-erg dance do aux.prs.erg3sg.abs3sg ‘Aduna has danced.’ (53) Nor-k egin du dantza? who-erg do aux.prs.erg3sg.abs3sg dance ‘Who has danced?’ (54) ADUNA-K egin du dantza. Aduna-erg do aux.prs.erg3sg.abs3sg dance ‘ADUNA has danced.’ (55) DANTZA egin du Aduna-k. Dance do aux.prs.erg3sg.abs3sg Aduna-erg ‘Aduna DANCED.’ (56) Dantza, Adunak EGITEN du. ‘As for dancing, Aduna does dance’ (57) Aduna-k ez du dantz-ik egin. Aduna-erg not aux.prs.erg3sg.abs3sg dance-ptve do ‘Aduna has not danced.’ These LVC s pattern with regular transitive constructions also in that their NVE, qua incremental theme, can be quantified (58)a just like any incremental theme of a non-light verb (58)b (Oyharçabal 2006: 792): (58) a. Aduna-k lo gutxi egin du. Aduna-erg sleep little do aux.prs.erg3sg.abs3sg ‘Aduna has slept a little.’

208

acedo-matellán and pineda

b. Aduna-k ogi gutxi jan du. Aduna-erg bread little eat aux.prs.erg3sg.abs3sg ‘Aduna has eaten some bread.’ An additional piece of evidence for the transitive analysis comes from causative contexts (Ortiz de Urbina 1989: 47). The NVE behaves as a canonical argument in the formation of causatives, since the subject of the LVC (59) takes the dative case in the causative (60), just like the causee of a causativized transitive predicate (61): (59) Aduna-k dantza egin du. Aduna-erg dance do aux.prs.erg3sg.abs3sg ‘Aduna has danced’ (60) Aduna-ri dantza egin-arazi diote. Aduna-dat dance do-cause aux.prs.erg3pl.abs3sg.dat3sg ‘They have made Aduna dance’ (61) Aduna-ri afari-a antol-arazi Aduna-dat dinner-det.abs prepare-cause diote. aux.prs.erg3pl.abs3sg.dat3sg ‘They have made Aduna prepare the dinner’ However, not all Basque LVC s display the same behaviour in the battery of syntactic contexts just presented. For example, in contrast with (57), not all NVE s can be partitive-marked, as pointed out by Etxepare (2003: 404) and Zabala (2004: 451–452) and thoroughly assessed by Martinez (2015: 247–277): (62) a. *Aduna-k ez du laprast-ik egin Aduna-erg not aux.prs.erg3sg.abs3sg slip-ptve do b. *Aduna-k ez du alde-rik egin Aduna-erg not aux.prs.erg3sg.abs3sg side-ptve do (63) a. Aduna-k ez du laprast egin. Aduna-erg not aux.prs.erg3sg.abs3sg slip do ‘Aduna has not slipped.’

light verb constructions in basque and romance

209

b. Aduna-k ez du alde egin. Aduna-erg not aux.prs.erg3sg.abs3sg side do ‘Aduna has not left.’ Moreover, even those LVC s showing partitive case-marking do not exactly replicate the behaviour of a common transitive structure; indeed, for LVC s such case marking is not compulsory (64), whereas for regular transitive structures it is (65): (64) a. Ez dut lo egin. not aux.prs.erg1sg.abs3sg sleep do ‘I haven’t sleep.’ b. Ez dut lo-rik egin. not aux.prs.erg1sg.abs3sg sleep-ptve do ‘I haven’t sleep.’ (65) a. *Ez dut liburu irakurri. not aux.prs.erg1sg.abs3sg book read ‘I haven’t read a/any book.’ b. Ez dut liburu-rik irakurri. not aux.prs.erg1sg.abs3sg book-ptve read ‘I haven’t read a/any book.’ Thus, the existence of the case-less alternates above, whether general (63) or circumscribed to a particular dialect or register (Oyharçabal 2006) (64)a, favours an incorporation analysis.31 Such an incorporation analysis would also allow us to account for the differences regarding the adjacency between the NVE and the LV. As a matter of fact, in contrast to what was shown in (53)–(56) for dantza egin, some NVE s cannot be separated from the LV at all, unlike common DO s (Zabala 2004: 452– 453). Thus, the following examples, involving interrogative sentences (66) and focus structures (67)–(69), should be contrasted with (54)–(56). Also, another

31

There is no exhaustive dialectal survey on this topic. Oyharçabal (2006) observes some tendencies within dialects (and registers) but, in spite of some remarks on how some other dialects behave, he mainly focuses on Basque eastern varieties, i.e., those spoken in the French Basque Country (Iparralde).

210

acedo-matellán and pineda

sign of incorporation in the case of hitz egin (actually often written hitzegin) comes from its phonological pattern, bearing one single accent: hitzégin, to be contrasted, for instance, with that of negar egin, negár egín. (66) *Nor-k egin du hitz? who-erg do aux.prs.erg3sg.abs3sg word Intend. ‘Who has talked?’ (67) *ADUNA-K egin du hitz. Aduna-erg do aux.prs.erg3sg.abs3sg word Intend. ‘ADUNA has talked.’ (68) *HITZ Aduna-k egin du. word Aduna-erg do aux.prs.erg3sg.abs3sg Intend. ‘Talk, has Aduna done.’ (69) *Hitz, Aduna-k EGI-TEN du. word, Aduna-erg do-prs aux.prs.erg3sg.abs3sg Intend. ‘As for talking, Aduna does talk.’ Likewise, as regards quantification, LVC s and transitive structures do not necessarily behave alike, in spite of what we have shown in (58). Indeed, in LVC s the quantifier may occupy a different position, in such a way that the noun+quantifier sequence we saw in (58)a, repeated here as (70)a, alternates with (70)b, where the quantifier gutxi targets the whole VP instead of just the NVE, the latter option arguably signalling incorporation. Here too variation is dialectal or idiolectal, and it even may depend on the particular LVC. For example, with hitz egin (71) the quantifier+LVC sequence seems to be preferred all over the linguistic domain (Oyharçabal 2006: 802): (70) a. Aduna-k lo gutxi egin du. Aduna-erg sleep little do aux.prs.erg3sg.abs3sg b. Aduna-k gutxi lo egin du. Aduna-erg little sleep do aux.prs.erg3sg.abs3sg ‘Aduna has not slept much.’ (71) a. ??Aduna-k hitz gutxi egin du. Aduna-erg word little do aux.prs.erg3sg.abs3sg

light verb constructions in basque and romance

211

b. Aduna-k gutxi hitz egin du. Aduna-erg little word do aux.prs.erg3sg.abs3sg ‘Aduna has not talked much.’ It is interesting at this point to go back to the French LVC s involving a case-less NVE, like avoir faim ‘be hungry’ or faire peur ‘scare’, since they behave differently from Basque hitz egin ‘talk’ and pattern with lo egin in allowing for degree quantifiers—albeit non-nominal!—to intervene between the LV and the NVE: (72) Marjorie a très faim. Marjorie has very hunger ‘Marjorie is very hungry.’ (73) Les serpents font très peur à Magali. the snakes do very fear at Magali ‘Magali is very afraid of snakes.’ The position of manner adverbials such as ondo, ongi ‘well’ or gaizki ‘badly’ also reveal differences between LVC s and ordinary transitive structures. Whereas with the latter the unmarked order is S O Adv V (74)a, with the former two options are possible: in (75)a, available at least in Western dialects, the LVC is treated like a transitive structure, whereas in the more general (75)b (cf. Oyharçabal 2006: 794), the LVC is treated like an intransitive structure—thus, by hypothesis, signalling incorporation. Here again, we find variation depending on the particular LVC. For example, hitz egin is not grammatical if treated as a transitive structure (76)a, only the option involving incorporation is good (76)b: (74) a. Aduna-k liburu-a ongi irakurri du. Aduna-erg book-det.abs well read aux.prs.erg3sg.abs3sg ‘Aduna read the book well.’ b. *Aduna-k ongi liburu-a irakurri du. Aduna-erg well book-det.abs read aux.prs.erg3sg.abs3sg (75) a. Aduna-k lan ongi egin du. Aduna-erg work well do aux.prs.erg3sg.abs3sg b. Aduna-k ongi lan egin du. Aduna-erg well work do aux.prs.erg3sg.abs3sg ‘Aduna has worked well’

212

acedo-matellán and pineda

(76) a. *Aduna-k hitz ongi egin du. Aduna-erg word well do aux.prs.erg3sg.abs3sg b. Aduna-k ongi hitz egin du. Aduna-erg well word do aux.prs.erg3sg.abs3sg ‘Aduna has talked well.’ So far we have seen that variation depends on the particular LVC (with cases such as hitz egin favoring the incorporation analysis in general) but also on geographic variation. In particular, Oyharçabal (2006) finds quite a general tendency in Eastern dialects to avoid the treatment of the NVE as an object, thus preferring incorporation—which otherwise is also possible in other dialects or even idiolects. This observation nicely fits with Berro’s (2010:4) claim that “complex unergatives are much more common in southern varieties [that is, those in contact with Spanish: our clarification, VA&AP], whereas simplex are preferred in northern ones [that is, those in contact with French: our clarification, VA&AP].” The tendency shown by Eastern dialects can also be witnessed in the treatment of LVC s in causative contexts: in contrast to what we saw in (60), Eastern dialects do not always treat the causee of a LVC as the causee of a transitive verb (with dative) (77), since it may also appear with absolutive case (78), as the causee of an intransitive verb (thus indicating that incorporation has taken place) (Epelde & Oyharçabal 2012):32 32

Epelde & Oyharçabal (2012) propose a contact-based explanation: in French, with which Eastern dialects are in contact, the causee of unergative predicates is accusative (i); in turn, Western dialects are in contact with Spanish, and in particular with the Spanish spoken in the Basque Country, which is leísta and thus marks the causee of unergatives with dative case (Tu le haces reír, Yo le hago bailar). Actually, with simplex unergatives (ii) the dialectal divide is even more marked: whereas in Western dialects dative is more generally used (iii), in Eastern dialects absolutive is normally found (iv) (Ricardo Etxepare, p.c.; Oyharçabal & Epelde 2012). (i) Tu le fais rire, Je le fais danser you cl.3sg.m.acc make.prs.2sg laugh, I cl.3sg.m.acc make.prs.1sg dance ‘You make him laugh, I make him dance’ (ii) Aduna-k dantzatu du. Aduna-erg dance aux.prs.erg3pl.abs3sg ‘Aduna has danced.’ (iii) Aduna-ri dantz-arazi diote. Aduna-dat dance-cause aux.prs.erg3pl.abs3sg.dat3sg ‘They have made Aduna dance.’ (iv) Aduna dantz-arazi dute. Aduna dance-cause aux.prs.erg3pl.abs3sg ‘They have made Aduna dance.’

light verb constructions in basque and romance

213

(77) Aduna-ri dantza egin-arazi diote. Aduna-dat dance do-cause aux.prs.erg3pl.abs3sg.dat3sg ‘They have made Aduna dance.’ (78) Aduna dantza egin-arazi dute. Aduna dance do-cause aux.prs.erg3pl.abs3sg ‘They have made Aduna dance.’ Last, but not least, NVE s present a strong dissimilarity with DO s: whereas common DO s must bear a determiner (79) NVE s (generally) lack it (80) (Oyharçabal 2006: 793–794). (79) *Liburu irakurri dut. book read aux.prs.erg1sg.abs3sg ‘I have read book.’ (80) Barre / Lo / Dantza33 egin dut. laugh / sleep / dance do aux.prs.erg1sg.abs3sg ‘I have laughed, slept, danced.’ Thus, the lack of a determiner is argued to signal incorporation (Martinez 2015). Nonetheless, this difference can be accounted also in the transitive analysis, for instance by postulating that NVE s and plain DO s receive case differently (cf. Laka 1993). To summarize, NVE s in Basque LVC s pattern with absolutive objects in some cases, but differ from them in others. What one can conclude from all the previous tests and studies on Basque LVC s is that there is no single analysis for all of them; rather, whereas some LVC s pattern with transitive structures, some others show what can be taken as signals of syntactic incorporation. As Oyharçabal (2006) suggests, Basque LVC s constitute a non-homogeneous set and the degree of cohesion between the LVC and the NVE, and thus whether there is incorporation or not, is subject to dialectal and idiolectal variation, with some particular LVC s such as hitz egin “do word” ‘talk’ being a generalized case of incorporation across varieties. Basque LVC s can be thus seen as forming a continuum, in Martinez’s (2015: 350) terms, comprising the three groups of LVC s that we next describe. 33

The -a ending of dantza does not correspond to the determiner, but is part of the nominal root, making ‘dance’ and ‘the dance’ formally equal. Compare: hitz ‘word’ ~ hitz-a ‘the word’ ~ hitz bat ‘one word’ and dantza ‘dance’ ~ dantza ‘the dance’ ~ dantza bat ‘one dance’.

214

acedo-matellán and pineda

At the beginning of the continuum, we find transitive-like structures, such as dantza egin “do dance” ‘dance’, negar egin “do weep” ‘weep’, irribarre egin “do smile” ‘smile’ and so on. As in common transitive structures, in Basque LVC s the auxiliary is *edun, the subject bears ergative case, and the NVE functions as a DO, which can bear partitive case and can be separated from the verb. However, although this type of LVC s has a transitive syntactic structure, they show morphological, semantic and syntactic properties that distinguish them from common transitive structures (Martinez 2015: 337, 373). At the end of the continuum, we find the fossilized LVC s, such as alde egin (aldegin), hanka egin, ospa egin, hitz egin (hitzegin), hots egin and a few others. These can be syntactically accounted for as instances of incorporation and, as a result, they do not admit partitive case, they show adjacency to the verb, etc.— actually, these LVC s are generally phonologically treated as one single word, e.g. aldégin, hitzégin and so on. As expected, many LVC s with a very tight relation between their components have idiomatic readings, as clearly exemplified by alde egin “do side/zone” ‘leave’, hanka egin “do leg” ‘leave’ (Martinez 2015: 337, 347–348), and they contrast with LVC s whose word order can be altered and have a compositional reading (e.g. dantza egin “do dance” ‘dance’). Finally, there is a group of LVC s presenting a behaviour ambiguous between these two extremes (e.g. amets egin ‘dream’, barre egin ‘laugh’) (Martinez 2015: 337).

4

Conclusions

We have presented a description of LVC s in Basque, and we have compared them with LVC s in Romance, mainly French and Spanish, keeping in mind Ramchand’s (2014) position that there is no fundamental difference between LVC s and the constructions headed by the same verbs in their non-light use (cf., e.g., Basque lo egin “do sleep” ‘sleep’ vs ogia egin ‘make bread’). Rather, it is the nature of the NVE, as opposed to other DO s, that makes LVC s special. This hypothesis is paired with the assumption that in LVC s the LV, as head of the construction, is responsible for its event and argument-structure properties, while the NVE encodes the conceptual scene evoked. We have dealt with prominent discussions on the syntax and semantics of LVC s in Basque, and also French and Spanish. With respect to the nature of the LV, for instance, we have remarked that the choice of LV is generally meaningful, and that it is only DO and GIVE that seem to give rise to synonymous LVC s, at least at the conceptual level, both within Basque and when we compare French (favouring DO) and Spanish (favouring GIVE).

light verb constructions in basque and romance

215

We have also paid attention to the NVE and its categorial properties. Basque, unlike French and Spanish, often allows a type of LVC in which the nominal NVE appears in the inessive case, or endowed with an adverbial suffix, the resulting LVC displaying peculiar aspectual properties. As regards unmarked nominal NVE s, we have shown that they can be bare, i.e., determinerless and caseless, in Basque and French—unlike Spanish and Catalan. For Romance, we have also explored the different kinds of relations—semantic and morphological— between LVC s and their synthetic counterparts. Finally, a discussion has been dedicated to the degree of cohesion between the LV and the NVE. It has been shown that there is empirical evidence pointing in diverse directions as regards this issue: the NVE is in some cases quite syntactically independent of the LV, in other cases it seems to be somehow fused with it (i.e., incorporated into it), and, finally, there are cases that lie in between these two extremes. Dialectal and lexical factors have been pointed out as predictors of this variation in Basque. It should be clear from these conclusions that, much in the spirit of Butt’s (2010) title, Basque (and Romance) LVC s constitute a jungle, and we are still hacking away.

Acknowledgements This work has been supported by the postdoctoral research fellowship Beatriu de Pinós 2014 BP_A 00165 (Secretaria d’Universitats i Recerca del Departament d’Economia i Coneixement de la Generalitat de Catalunya) (second author), the postdoctoral research fellowship Juan de la Cierva-incorporación (IJCI2016–30474, Ministerio de Economía y Competitividad, Spanish Government) (second author) and the research project FFI2014–56968-C4–1-P (Ministerio de Economía y Competitividad, Spanish Government) (first and second author). The research leading to this work began at Queens’ College, University of Cambridge, in the case of the first author, and at CNRS-IKER, in the case of the second author. We are grateful to Jaume Mateu and an anonymous reviewer for very valuable comments. All errors remain our own.

References Acedo-Matellán, Víctor. 2014. “Can we dance without doing a dance? Two opposite views on the integration of roots in the syntactic structure of the vP”. In Asaf Bachrach, Isabelle Roy & Linnaea Stockall (eds.), Structuring the argument. Mul-

216

acedo-matellán and pineda

tidisciplinary research on verb argument structure. Amsterdam/Philadelphia: John Benjamins. 23–43. Acedo-Matellán, Víctor, & Jaume Mateu. To appear. “El verb i el sintagma verbal [‘The verb and the verb phrase’]”. In Josep Martines & Manuel Pérez-Saldanya (eds.), Gramàtica del català antic. Ahn, Hee Don. 1991. Light Verbs, VP-Movement, Negation and Clausal Architecture in Korean and English. PhD thesis. University of Wisconsin–Madison. Alba-Salas, Josep. 2002. Light verb constructions in Romance: A syntactic analysis. PhD thesis. Ithaka (NY): Cornell University. Aldai, Gontzal. 2009. “Is Basque morphologically ergative?: Western Basque vs. Eastern Basque”. Studies in Language 33(4): 783–831. Aldai, Gontzal. 2010. “Euskararen kasu-markaketaren aldakortasun dialektala”. In Beatriz Fernández, Pablo Albizu & Ricardo Etxepare (eds.), Euskara eta euskarak: Aldakortasun sintaktikoa aztergai, [Supplements of ASJU 52]. Bilbao: Universidad del País Vasco. 5–19. Alonso-Ramos, Margarita. 1998. Étude sémantico-syntaxique des constructions à verbe support. PhD thesis. Université de Montreál. Austin, Jennifer & Luis López. 1995. Nominative, absolutive, and dative languages. In Jill N. Beckman (ed.), Proceedings of the Twenty-fifth North Eastern Linguistic Society. Amherst, MA: GLSA. 1–16. Baker, Mark. 1988. Incorporation: a theory of grammatical function changing. Chicago: University Press of Chicago. Barrie, Michael & Éric Mathieu. 2014. “Head movement”. In Andrew Carnie, Yosuke Sato & Daniel Siddiqi (eds.), The Routledge Handbook of Syntax. Milton Park, Abingdon, Oxon; New York: Routledge. 133–149. Berro, Ane (this volume). (In)transitivity. In Ane Berro, Beatriz Fernández & Jon Ortiz de Urbina (eds.), Aligning grammars: Basque and Romance. Berro, Ane. 2015. Breaking verbs. From event structure to syntactic categories in Basque. Ph.D. thesis, University of the Basque Country (UPV/EHU) and Université Bordeaux Montaigne (UBM). Bobaljik, Jonathan David. 1993. “On ergativity and ergative unergatives”. In Collin Philips (ed.), MIT Working Papers in Linguistics 19: Papers on Case and Agreement II. Cambridge: MIT, 45–88. Bosque, Ignacio & Pascual Masullo. 1998. “On verbal quantification in Spanish”. In Francesc Roca and Olga Fullana (eds.), Studies on the syntax of central romance languages. Proceedings of the III Symposium on the Syntax of Central Romance Languages [held in Girona on the 15th and 16th of November 1996], 9–63. Girona: Universitat de Girona. Bowern, Claire. 2008. “The diachrony of complex predicates”. Diachronica 25(2), 161– 185.

light verb constructions in basque and romance

217

Brody, Michael 2000. “Mirror theory: Syntactic representation in perfect syntax”. Linguistic Inquiry 31: 29–56. Butt, Miriam. 1995. The structure of complex predicates in Urdu. Standord, California: CSLI Publications. Butt, Miriam. 2003. “The light verb jungle”. In Gulsat Aygen, Claire Bowern, Conor Quinn (eds.), Harvard Working Papers in Linguistics, vol. 9, 1–49. Papers from the GSAS/Dudley House Workshop on Light Verbs. Butt, Miriam. 2010. “The light verb jungle: Still hacking away”. In Mengistu Amberber, Brett Baker, Mark Harvey (eds.), Complex Predicates: Cross-linguistic Perspectives on Event Structure, 48–78. Cambridge: Cambridge University Press. Butt, Miriam, & Wilhelm Geuder. 2001. “On the (semi)lexical status of light verbs”. In Norbert Corver & Henk van Riemsdijk, H. (eds.), Semi-lexical Categories: On the Content of Function Words and the Function of Content Words. Berlin: Mouton de Gruyter. 323–370. Butt, Miriam & Aditi Lahiri. 2013. “Diachronic pertinacity of light verbs”. Lingua 135: 7–29. Bye, Patrick, & Peter Svenonius. 2012. “Non-concatenative morphology as epiphenomenon”. In Jochen Trommer (ed.), The Morphology and Phonology of Exponence. Oxford: Oxford University Press. 427–495. Cattell, Ray. 1994. Composite Predicates in English. Sydney: Academic Press Australia. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press. Cuervo, Mª Cristina. 2003. Datives at large. PhD thesis, MIT Danlos, Laurence. 1992. “Support verb constructions: linguistic properties, representation, translation”. French Language Studies 2: 1–32. De Rijk, Rudolf. 2008. Standard Basque: A Progressive Grammar. Cambridge, MA: MIT Press. Epelde, Irantzu & Beñat Oyharçabal. 2012. Resultados lingüísticos del contacto francéseuskera: el cambio dativo > absolutivo en algunas construcciones causativas. Sociedad Española de Lingüística. Universitat de València. Etxepare, Ricardo. 2003. “Valency and Argument Structure in the Verb”. In Jose Ignacio Hualde & Jon Ortiz de Urbina (eds.), A grammar of Basque. Berlin: Mouton de Gruyter. 369–465. Fernández, Beatriz. 1997. Egiturazko kasuaren erkaketa euskaraz. PhD thesis, Euskal Herriko Unibertsitatea / Universidad del País Vasco. Giry-Schneider, Jacqueline. 1978. “Les Nominalisations En Français: L’opérateur ‘faire’ Dans Le Lexique”. Genève: Droz. Giry-Schneider, Jacqueline. 1986. “Les noms construits avec “faire”: compléments ou prédicats?”. Langue française 69, 49–63. Goldberg, Adele. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago; London: University of Chicago.

218

acedo-matellán and pineda

Grimshaw, Jane. 1990. Argument Structure. Cambridge, Mass: MIT. Grimshaw, Jane. 2005 [1993]. “Semantic Structure and Semantic Content in Lexical Representation”. In Jane Grimshaw (ed.), Words and Structure. Stanford, CA: CSLI Publications. 75–89. Grimshaw, Jane & Armin Mester. 1988. “Light Verbs and θ-Marking”. Linguistic Inquiry 19(1): 205–232. Gross, Maurice. 1976. “Sur quelques groupes nominaux complexes”. In Chevalier, JeanClaude and Maurice Gross (eds.). Méthodes en grammaire française. Paris: Klincksieck, 97–119. Gross, Maurice. 1981. “Les bases empiriques de la notion de prédicat sémantique”. Langages 63: 7–52. Hale, Kenneth and Samuel Jay Keyser. 1986. “Some transitivity alternations in English”, Lexicon Project Working Papers #7. Cambridge, MA: Center for Cognitive Science, Massachusetts Institute of Technology. Hale, Kenneth & Samuel Jay Keyser. 1997. “On The Complex Nature of Simple Predicators”. In Alex Alsina Joan Bresnan & Peter Sells (eds.), Complex Predicates. Stanford, Calif.: CSLI Publications. 29–65. Hale, Kenneth & Samuel Jay Keyser. 1993. “On argument structure and the lexical expression of syntactic relations”. In Kenneth Hale and Samuel Jay Keyser (eds.), The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger. Cambridge, MA: MIT Press. 53–109. Haugen, Jason. 2015. “Incorporation”. In Peter Müller, Ingeborg Ohnheiser, Susan Olsen & Franz Rainer (eds.), Word Formation: An International Handbook of the Languages of Europe. Berlin: Mouton de Gruyter, 413–434. Himmelman, Nikolaus. 2001. “Articles”. In Martin Haspelmath et al. (eds.), Language Typology and Language Universals. Handbücher ders Sprach und Kommunikationswissenschaft, vol. 20, 1. Berlin, New York: Walter de Gruyter. 831–841. Hook, Peter E. 1991. “The Emergence of Perfective Aspect in Indo-Aryan Languages”. In Elizabeth Traugott and Bernd Heine (eds.), Approaches to Grammaticalization, vol. II: Types of grammatical marker. Amsterdam: Benjamins. 59–89. Hopper, Paul J. & Elizabeth C. Traugott. 1993. Grammaticalization. Cambridge: Cambridge University Press. Jackendoff, Ray. 1974. “A Deep Structure Projection Rule”. Linguistic Inquiry 5(4), 481– 505 Jespersen, Otto. 1954. A Modern English Grammar, vol. 4. New York: Barnes and Noble. Kearns, Kate. 1988/2002. Light Verbs in English. Ms., MIT. Lafitte, Pierre 1944. Grammaire Basque. Baiona: Elkar. Laka, Itziar. 1993. “Unergatives that Assign Ergative, Unaccusatives that Assign Accusative”. In Jonathan D. Bobaljik & Collin Phillips (eds.), Papers on Case and Agreement I, MITWPL 18. Cambridge, MA: MITWPL, 149–172.

light verb constructions in basque and romance

219

Larson, Richard K. 1988. “On the double object construction”. Linguistic Inquiry 19(3), 335–391. Levin, Beth & Malka Rappaport. Hovav. 1995. Unaccusativity: At the Syntax-lexical Semantics Interface. Cambridge, Mass.; London: MIT, 1995. Lin, Tzong-Hong. 2001. Light Verb Syntax and the Theory of Phrase Structure. PhD thesis, University of California, Irvine. Marantz, Alec. 2005. “Objects Out of the Lexicon: Objects as Events”. Ms., MIT. Martinez, Arantzazu. 2015. “[Izen + egin]a aditz-lokuzioak: inkorporazio mailak”. PhD thesis, Deustuko Uniberstitatea. Matushansky, Ora. 2006. “Head movement in linguistic theory”. Linguistic Inquiry 37, 69–109. McFadden, Thomas. 2006. “German inherent datives and argument structure”. In Daniel P. Hole, Abraham Werner & André Meinunger (eds.), Datives and Other Cases: Between Argument Structure and Event Structure. Amsterdam; Philadelphia: John Benjamins. 49–77. Mithun, Marianne. 1984. “The evolution of Noun Incorporation”. Language 60, 847–893. Mohanan, Tara. 1994. Argument Structure in Hindi. Stanford, California: CSLI Publications. Ortiz de Urbina, Jon. 1989. Parameters in the Grammar of Basque. Dordrecht: Foris. Oyharçabal. Beñat. 2006. “Basque Light Verb Constructions”. ASJU XL: 787–806. Pesetsky, David & Esther Torrego. 2001. “T-to-C Movement: Causes and Consequences”. In Michael Kenstowicz (ed.), Ken Hale: A Life on Language. Cambridge, MA: MIT Press. 355–426. Piera, Carlos & Soledad Varela. 1999. “Relaciones entre morfología y sintaxis”. In Violeta Demonte & Ignacio Bosque (ed.), Gramática descriptiva de la lengua española, vol. 3: Entre la oración y el discurso. Morfología. Madrid: Real Academia Española / Espasa-Calpe. 4367–4422. Pineda, Anna. 2016. Les fronteres de la (in)transitivitat. Estudi dels aplicatius en llengües romàniques i basc [published and revised version of the PhD dissertation]. Barcelona: Institut d’Estudis Món Juïc. Col·lecció Cum Laude, 6. Pylkkänen, Liina. 2002. “Introducing Arguments”. PhD thesis, MIT. Pylkkänen, Liina. 2008. “Introducing Arguments”. Cambridge, Mass.; London: MIT. Rafel, Joan. 2004. “Los predicados complejos en espanyol”. In Elixabete Pérez Gaztelu, Igone Zabala & Lluïsa Gràcia (eds.), Las fronteras de la composición en lenguas románicas y en vasco. Donostia: Deustuko Unibertsitatea. 393–443. Ramchand, Gillian C. 2008. Verb meaning and the lexicon: A first phase syntax. Cambridge: Cambridge University Press. Ramchand, Gillian C. 2014. “On structural meaning vs conceptual meaning in verb semantics”. Linguistic Analysis 39(1–2), 207–243. Rappaport, M. & Levin, B. 1988. “What to do with theta-roles”. In Wendy Wilkins (ed.) Syntax and Semantics 21: Thematic relations. New York: Academic Press, 7–36.

220

acedo-matellán and pineda

Rappaport Hovav, Malka & Beth Levin. 1998. “Building verb meanings”. In Miriam Butt & Wilhelm Geuder (ed), The Projection of Arguments: Lexical and Compositional Factors. Stanford, Calif.: CSLI Publications, 97–134. Rodríguez, Sonia & Fernando García Murga. 2003. “IZEN + EGIN predikatuak euskaraz”. In Jesús Makatzaga & Beñat Oyharçabal (eds.), Euskal gramatikari eta literaturari buruzko ikerketak XXI. mendearen atarian, Gramatika gaiak, Iker-14. Bilbo: Euskaltzaindia. 417–436. Sarasola, Ibon. 1977. “Sobre la bipartición inicial en el análisis en constituyentes”, ASJU XI: 51–90. Uribe-Etxebarria, Myriam. 1989. “Some Notes on the Structure of IP in Basque”. Ms., UConn, Storrs. Wierzbicka, Anna. 1982. “Why Can You Have a Drink When You Can’t *Have an Eat?” Language, Vol. 58, No. 4, 753–799. Wood, Jim & Alec Marantz. 2017. “The interpretation of external arguments”. In Roberta D’Alessandro, Irene Franco & Ángel Gallego (ed.), The verbal domain. Oxford: Oxford University Press. 255–278. Zabala, Igone. 2004. “Los predicados complejos en vasco”. In Elixabete Pérez Gaztelu, Igone Zabala & Lluïsa Gràcia (eds.), Las fronteras de la composición en lenguas románicas y en vasco. Donostia: Deustuko Unibertsitatea. 445–534.

chapter 7

On Non-selected Datives: Ethical Datives in Basque and Spanish Beatriz Fernández

1

Introduction

There is a long tradition of research dealing with datives in general and nonselected datives in particular. These non-selected datives include external possessors, benefactives and ethical datives, among others (see for example, Bosse, Bruening & Yamada 2012). In this chapter, I will analyze some of these datives, particularly, ethical datives. In doing so, one of the major difficulties to overcome is the extreme variation of designations and the wide range of properties (not necessarily coherent with each other) attributed to them. For instance, ethical datives have been considered to refer to the participants of the speech act and mostly to the speaker (sometimes also to the hearer). If this were the case, then ethical datives would be restricted to 1st (or 2nd) person and couldn’t be doubled by a DP, among other properties (see Jaeggli 1986 for Spanish). Nevertheless, not all the so-called ethical datives seem to be restricted to clitics or 1st and 2nd person. Hence, a more-fine grained typology seems to be necessary in order to clarify this puzzling scenario. In this chapter, I will follow Franco & Huidobro (2008) in distinguishing three types of ethical datives in Spanish. First of all, Class I, reflexive nonargumental clitics as in me comí la manzana ‘I ate the apple’, alternatively analyzed as aspectual by Miguel & Fernández Lagunilla 2000, and reminiscent of personal datives (Horn 2008, 2013). Second, Class II of the ‘non volitional se + dative clitic type’, as in se me caló el coche ‘the car stalled on me’. Ethical datives from Class II correspond to affected experiencers in Bosse, Bruening & Yamada’s (2012) typology or affected datives (dativos de afectación) in Maldonado (1994) for Spanish and in Fernández (2010) for Basque. Third, Class III, as in me le pegaron una paliza a mi primo ‘they beat my cousin up’ with a dative clitic cluster including the ethical dative me and the goal dative le. This Class III shows the most restricted version of ethical datives, as they seem to be necessarily 1st and 2nd person (me le/te le/*se le) whereas ethical datives from Class I and II do not. Besides, ethical datives from either Class III or Class I cannot be doubled (me/*a mí) whereas those from Class II can.

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004395398_008

222

fernández

On the other hand, unlike in Hispanic linguistics, ethical datives have been only briefly mentioned but not analyzed in depth in Basque. The sole existence of ethical datives in Basque is not uncontroversial, as they are neither restricted to 1st and 2nd person nor reduced to clitics (Fernández 2010). As a consequence, some of the properties attributed to ethical datives in general and Spanish ethical datives in particular are not met in Basque. In this chapter, I will review classical syntactic tests (as in Borer & Grodzinsky 1986 or Jouitteau & Rezac 2007) and discuss more novel criteria proposed by Bosse, Bruening & Yamada (2012) and Horn (2008, 2013). I will claim that the only ethical datives attested in Basque are those related to Spanish ethical datives from the Class II as neither Class I nor Class III is attested in Basque. I will also claim that only Spanish (and not Basque) exhibits ethical datives in their most restricted definition (Class III), i.e. 1st and 2nd person clitics referred to the participants of the speech act with no DP-doubling among other properties. As we will see, ethical datives of the Class II contribute truth conditional meaning—or at issue meaning in Potts’ (2005) designation—but contrary to what is argued by these authors, they do not contribute non-truth conditional meaning—or not-at-issue meaning—in either Basque or Spanish. Although no particular analysis will be provided here, I will claim that ethical datives of Class II are merged in the Specifier of an Applicative head above v, as argued in Odria (2017) for experiencers of psychological verbs and external possessors. Thus, the similarity in the behavior of ethical datives of Class II, external possessors and also benefactives follows. Besides, ethical datives from Class I and III need further explanation that go beyond the limits of this paper. This chapter will be structured as follows. First of all, I will briefly present non-selected datives in general and ethical datives in particular (section 2). Secondly, I will review and discuss some classical and novel tests on Basque and Spanish ethical datives of Class II (section 3). Section 4 will present the main conclusions and a brief discussion.

2

Ethical Datives

2.1 Some Clarifications before Starting Although the theoretical analysis of datives has been exponentially increased especially since the nineties and the first decade of this century, there are still some controversial issues in the literature of datives, particularly, the distinction between selected and non-selected ones. This distinction goes back to some classical papers such as Borer & Grodzinsky (1986) on Hebrew. In this

on non-selected datives: ethical datives in basque and spanish

223

paper, some of the most salient properties of non-selected datives are presented and discussed in depth, as far as external possessors and ethical datives are concerned. The basis of the typology of these non-selected datives and the properties attributed to each type are still studied as a classical benchmark test. Other studies such as Jouitteau & Rezac (2007) have explored the literature on ethical datives in great detail in order to characterize this particular type of non-selected datives in French. Paradoxically, one of most influential recent works on datives, i.e. Pylkkänen 2008 [2002] and her well-known high v low applicative distinction does not solve the issue, as non-selected datives such as ethical datives and benefactives can be analyzed as high applicatives, blurring the distinction between both of them. Besides, although external possessors are treated as low applicatives in her proposal, they behave similarly to ethical datives and benefactives (high applicatives) in many respects. An attempt to typify non-selected datives properly is due to Bosse, Bruening & Yamada (2012), an analysis that combines both syntactic and semantic properties of non-selected datives. In particular, these authors analyze affected experiencers which seem to correspond to a particular type of ethical datives in Spanish, i.e. Class II in Franco & Huidobro’s (2008) typology. Unfortunately, some of the properties attributed to affected experiencers, in particular their contribution to non-truth conditional meaning and sentientness, are not met in Spanish and Basque ethical datives of this class. In this chapter, I will concentrate on ethical datives and I will leave aside external possessors and benefactives, although I will refer to these non-selected datives in order to compare the properties of one another. Both external possessors and benefactives are also attested in the languages under discussion—see for instance Arregi (2003), Fernández (2010), Fernández & Ortiz de Urbina (2010) and Odria (2017) for Basque external possessors, and Maldonado (1994) for a discussion on the distinction between benefactives and ethical datives (named as datives of interest by the author) in Spanish.1 As we will see, the 1 Correspondence between designations attested in some of the papers mentioned in this chapter: ethical datives Class I (Franco & Huidobro 2008) = personal datives (Horn 2008, 2013) = subject co-referent datives (Bosse, Bruening & Yamada 2012); ethical datives Class II = affected datives (Maldonado 1994, Fernández 2010) = affected experiencers (Bosse, Bruening & Yamada 2012) = experiencer datives (Bosse & Bruening 2010); ethical datives Class III (Franco & Huidobro 2008) = datives of interest Maldonado (1994) = attitude holders and also affected experiencers (Bosse, Bruening & Yamada 2012). In referring to ethical datives, Strozer (1978) uses the term datives of interest. Although no particular designations are provided in her paper, the three dative types distinguished by this author can accommodate to Franco & Huidobro’s (2008) three-way typology assumed here.

224

fernández

label ethical datives involves quite different types of datives. Hence, a more finegrained typology of ethical datives is necessary in order to characterize them properly. 2.2

Ethical Datives in Basque and Spanish: Some Data and a Three-Way Typology Let us present some basic data on Basque and Spanish ethical datives. In principle, an ethical dative is a non-selected one, i.e. a dative not included in the argument structure of the verb. This ethical dative experiences the positive or negative effect of an event, thus, it seems to be affected by the event. For instance, Basque examples in (1) show a dative of this sort: (1) a. (Niri) haurra lokartu zait. I.dat kid (abs) fall asleep aux + 1sg.dat2 ‘The kid has fallen asleep.’ b. Haurrak (niri) liburua apurtu dit. kid.erg I.dat book (abs) break aux + 3sg.dat ‘The kid has broken a book.’ The sentences in (1) include a 1st person singular dative clitic -t in both the bivalent unaccusative form (zait) and in the ditransitive one (dit).3 The datives in these sentences are non-selected arguments, as their counterparts with no dative are equally available and grammatical in Basque: (2) a. Haurra lokartu da. kid (abs) fall asleep aux ‘The kid has fallen asleep.’

2 Abbreviations: abs absolutive, acc accusative, adl adlative, allo allocutive, aux auxiliary, cl clitic, dat dative, d determiner, ed ethical dative, ed-ii ethical dative of Class II, ed-iii ethical dative of Class III, erg ergative, fam familiar, fem feminine, mas masculine, nom nominative, pl plural, psr possessor, pst past, sg singular, 1, 2, 3 for first, second and third person. 3 Basque is a language with auxiliary alternation: BE is selected with unaccusative predicates and HAVE with unergative and (di)transitive predicates. Besides, Basque verbal forms include clitics/agreement markers for absolutive, ergative and dative arguments. For ease of exposition, I will gloss the verbal form as aux (auxiliary) + cl.dat (dative clitic) when it includes a dative clitic, and as aux, when it does not. For ease of understanding, further details will be omitted.

on non-selected datives: ethical datives in basque and spanish

225

b. Haurrak liburua apurtu du. kid.erg book (abs) break aux ‘The kid has broken a book.’ The datives in (1) experience the effect of the kid falling asleep (1a) and the kid breaking the book (1b). The examples in (1) are equivalent to Spanish examples in (3). (3) a. Se me ha dormido el niño. cl cl.1sg.dat aux fall asleep the kid (adapted from Maldonado 1994) b. El niño me ha roto el libro. the kid cl.1sg.dat aux break the book Ethical datives of the ‘(non volitional) se + dative clitic type’ type as the one in example (3a) are classified as ethical datives within Class II in Franco & Huidobro’s (2008) three-way typology. I will name them as Ethical Datives-Class II (ED-II). Other designations have been used for the very same datives, such as affected datives (Spanish dativos de afectación) in Maldonado (1994) for Spanish and Fernández (2010) for Basque. These ED-II seem to correspond to affected experiencers in Bosse, Bruening & Yamada (2012). The interpretation of the ethical dative in (1a) and (3a) can be both negative and positive, whereas it seems to be negative in (1b) and (3b). Thus, the designation affected experiencer is used instead of malefactive or adversative in some works, as explicitly said in Bosse & Bruening (2010). Nevertheless, the malefactive interpretation of datives involved in constructions similar to (1b) and (3b) is not obligatory. For instance, if instead of ‘break the book’, we say ‘pass the exam’, the interpretation of the dative will be, in principle, positive. Therefore, I will assume that both benefactive and malefactive interpretations are available for examples such as (1) and (3), an assumption also made by Maldonado (1994) and Franco & Huidobro (2008) for Spanish. It is worth pointing out that that the (di)transitive structure in (3b) is not included in Franco & Huidobro’s Class II, as it does not involve the clitic se. Leaving aside this clitic se, the main difference between both constructions resides in the fact that the former example (3a) is intransitive whereas the latter is transitive. As far as Basque is concerned, I see no reason to exclude transitive structures in this ED-Class II. Actually, Maldonado (1994) includes both intransitive and transitive structures when dealing with ED-Class II. More-

226

fernández

over, the second type of datives distinguished by Strozer (1978), as in me ha comido la manzana (a mí) ‘he/she has eaten the apple’ fits well within this EDClass II. The other two types of ethical datives distinguished by Franco & Huidobro (2008) (and also in Strozer 1978) are on the one hand, those including an aspectual or reflexive (non-argumental) clitic (Class I), as in example (4a), and on the other hand, dative clitics such as the first singular clitic in (4b) (Class III). (4) a. Yo me he comido toda la sopa para cenar. cl cl.1sg.dat aux eat.pfv all the soup to dine ‘I have eaten all the soup for dinner.’ (Franco & Huidobro 2008) b. El niño me le ha roto el libro. the kid cl.1sg.dat cl.3sg.dat aux break the book ‘The kid has broken a book.’ The dative in (4a) is reminiscent of the so-called personal datives attested in Southern and Appalachian U.S. English as in I love me some him (Horn 2008: 176, 2010)—see also Christian (1991) and Webelhuth & Dannenberg (2006) among others. I will not include the reflexive non-argumental dative in this chapter, as it has no dative counterpart in Basque. Besides, the clitic cluster me-le in example (4b) and any other are not attested in Basque—see section (3.3). Thus, if a construction with a clitic cluster is needed in order to typify ethical datives in Class III, then Basque shows no evidence of these datives. Nevertheless, Franco & Huidobro (2008) mentions another structure with a sole clitic in this very same Class III, exemplified by (5a). Its Basque counterpart is provided in (5b). (5) a. A mi primo, me lo llevaron preso. to my cousin cl.1sg.dat cl.3sg.acc take prisoner ‘My cousin was taken prisoner.’ (Franco & Huidobro 2008) b. Nire lehengusua, preso eraman zidaten. my cousin prisoner take aux + 1sg.dat The Spanish example (5a) shows some restrictions that lead Franco & Huidobro (2008) to typify it as ED-Class III. However, its Basque counterpart shows exactly the same behavior of ED-Class II. In this point Spanish and Basque split. I will come back to this discussion in section 3.2.

on non-selected datives: ethical datives in basque and spanish

227

2.2 Typifying Ethical Datives of Class II Now let us concentrate on (1) and (3). As I have said before (section 2.1), the datives in these examples correspond to ED-Class II in Franco & Huidobro (2008) along with some (di)transitive structures including an ethical dative. These ethical datives show the following properties: 1) it can be 1st, 2nd and 3rd person; 2) it can be doubled by a DP; 3) it cannot be combined with any dative other than ethical dative of Class I and III in Spanish or allocutive clitics in Basque; 4) it contributes truth conditional meaning, as it can be questioned and bind a variable in the truth-conditional tier; and finally, 5) it does not contribute non-truth conditional meaning, as the experience itself can be negated and the meaning of experiencers does not survive beyond yes/no questions. This ED-Class II contrasts with the ED-Class III attested as a first person singular clitic me in example (4b) repeated here as (6): (6) El niño me le ha roto el libro. the kid cl.1sg.dat cl.3sg.dat aux break the book ‘The kid has broken the book.’ Compared to ED-Class II, the ED-Class III exemplified in (6) shows the opposite properties: 1) it can only be 1st or 2nd person; 2) it cannot be doubled by a DP (it is a weak pronoun); 3) it can be freely combined with any other dative argument; 4) it does not contribute truth conditional meaning, as it cannot be questioned or bind a variable in the truth-conditional tier; and 5) it only contributes non-truth conditional meaning (an implicature). Although Basque allocutives (Oyharçabal 1993) fall outside the scope of this chapter, they share all the properties attributed here to by ED-Class III, with the particularity that allocutives can only be 2nd person, as in (7). This example includes an allocutive marker: a 2nd person singular morpheme that can be either masculine -k or feminine -n and corresponds to a familiar addressee. (7) Haurrak liburua irakurri ziok/n. I.erg book (abs) read aux + 3sg.dat + 2sg.allo.mas/fem ‘The kid has read a book (it affects him/her + I tell you)’ In what follows, I will analyze the properties attributed here to ED-Class II in comparison to other non-selected datives in general and ED-Class III in particular.

228 3

fernández

Ethical Datives of the Class II: Basque and Spanish in Comparison

In section 2, I have presented Franco & Huidobro’s (2008) three-way typology of Spanish ED. This typology is a good start to review some of the properties attributed to ED in general and ED-Class II in particular. In this section, I aim at revisiting some of these properties in order to characterize properly Spanish and Basque ED-Class II in comparison to each other. 3.1 Ethical Datives-Class II and Person One of the properties repeatedly mentioned in the literature on ethical datives has to do with person: ethical datives can only be 1st and 2nd person (see Jaeggli 1982; Cuervo 2003: 195; Jouitteau & Rezac 2007: 2.1 among many others). In a more refined typology of ethical datives, it is worth analyzing which particular class shows this restriction. As far as ED-Class II is concerned, there is no person restriction at all. (8) a. Se me/te/le ha dormido el niño. cl cl.1/2/3sg.dat aux fall asleep the kid ‘The kid has fallen asleep.’ b. Haurra lokartu zait/zu/o. kid fall asleep aux + 1/2/3sg.dat As discussed in Fernández (2010), Basque shows that either external possessors or ED-Class II can be 1st, 2nd or 3rd person. The same behavior has been observed by Maldonado (1994: 259) for Spanish ED-Class II.4 This pattern is also attested in transitive structures: (9) a. El niño me/te/le ha comido la manzana. kid.erg cl.1/2/3sg.dat aux eat the apple ‘The kid has eaten the apple.’ 4 The datives involved in Maldonado’s (1994) examples can be argued to be external possessors (and not ED-Class II), as there is a relationship of inalienable possession between the dative (the possessor) and a PP including a body part (the possessum), as in (i). (i) Se me/te/le murió en los brazos. cl cl.1/2/3SG.dat die in the arms ‘He/she died on me/you/him-her / He/she died in my/your/his-her arms.’ Nevertheless, similar examples including a possessor inside the PP would be also grammatical: en los brazos de Juan ‘in John’s arms’. Hence, the datives involved in (i) can be ED and not necessarily external possessors.

on non-selected datives: ethical datives in basque and spanish

229

b. Haurrak (niri) sagarra jan dit/zu/o. kid.erg I.dat apple eat aux + 1/2/3sg.dat This property is also shared by other non-selected datives, such as benefactives and ED-Class I (the class involving the reflexive non-argumental dative clitic, as in Spanish Yo me como la manzana or Southern and Appalachian English instances such as I love me some him). Thus, only one class remains: ED-Class III. By definition, if ethical datives correspond to the participants of the utterance, then the restriction to be 1st and 2nd person follows. Nevertheless, leaving a priori definitions aside, the impossibility for ED-Class III to be 3rd person seems to be real. Take for instance, one of the examples provided for EDClass III, repeated here as (10a). As we will see, the combination of a hypothetical 3rd person ED-Class III and a 3rd person argumental clitic is out, whereas 1st and 2nd person are fine (10a,b): (10) a. El niño me le / te le / *se le ha roto el libro. kid.erg ed-iii 1/2/3 + cl.3sg.dat aux break the book ‘The kid has broken a book.’ b. El niño me le / te le / *se le ha comido la manzana. kid.erg ed-iii 1/2/3 + cl.3sg.dat aux eat the apple ‘The kid has eaten the apple.’ Note that the clitic cluster se le would be grammatical in a context where the clitic se were co-referential with the subject, that is, ED-Class I (reflexive nonargumental dative clitic) combined with a ED-Class II. To the contrary, a se clitic not co-referential with the subject would be out. Other ED-Class III are not so easily explained, but seem to exhibit the same constriction: (11) A mi primo, me/te/*se lo llevaron preso. to my cousin cl.1/2/3sg.dat cl.3sg.acc take prisoner ‘My cousin was taken prisoner.’ The clitic se cannot be interpreted as an ED, although the same cluster se lo would be grammatical, assuming (11) to be an aspectual se construction (de Miguel 2000). Nevertheless, the ungrammaticality of *se lo in a construction other than aspectual seems to be a side-effect of a Me-lui Constraint (see Odria 2017 and references therein), as also pointed out by Franco & Huidobro (2008: footnote 4). Interestingly enough, Basque shows no person constraint in similar structures:

230

fernández

(12) Nire lehengusua, preso eraman zidaten/zizuten/zioten. my cousin prisoner take aux + 1/2/3sg.dat In Basque, nothing prevents the ED from being 3rd person. Hence, we can speculate with the idea that the inability for the ED in (11) to be 3rd person cannot be only due to its nature but to the morphological and syntactic constraints that govern particular clitic combinations. Be that as it may, Basque and Spanish seem to separate in this respect. Summing up, nothing prevents ED-Class II from being 3rd person. With regard to this, ED-Class II behaves in the same way external possessors, benefactives and any other argumental datives such as experiencers do. Being nonargumental, ED-Class I also aligns with ED-Class II in this respect. It remains an open question whether ED-Class III is restricted or not to 1st (and 2nd person), as frequently discussed in the literature. The examples provided in (10) point in that direction. 3.2 Ethical Datives-Class II and Clitic Doubling Another property mentioned in the literature in order to distinguish ethical datives from other datives (selected or not) has to do with their impossibility to be doubled—see Jaeggli (1986: 22–23), Borer & Grodzinsky (1986: 180), and Jouitteau & Rezac (2007: section 2.4). (13) a. Juan me le arruinó la vida a esa chica (* a John ed-iii cl.1/2/3sg.dat ruin the life to that girl ( on mí). me) ‘John ruined her life on me.’ (Perlmutter 1971) b. Te me ensuciaste el pantalón (* a mí). cl.2sg.dat ed-iii get dirty the trousers ( to me) ‘You got your trousers dirty (on me).’ (Jaeggli 1986: 22) As can be seen in Perlmutter and Jaeggli’s classical examples the 1st person singular dative clitic cannot be doubled (me … *a mí)—note that the 3rd person singular dative clitic le in Perlmutter’s example (11a) can be doubled, as the presence of the DP (a esa chica) reveals. This clitic corresponds to the EDClass III of the typology assumed in this chapter. To the contrary, ED-Class II does allow doubling dative clitics in Spanish:

on non-selected datives: ethical datives in basque and spanish

231

(14) a. Se le quedó dormido a su madre. cl cl.3sg.dat fall asleep to his/her mother ‘He/she felt asleep (and it affected his/her mother).’ b. Se le escapó al policía. cl cl.3sg.dat get away to the police ‘He/She got away from the police.’ (Maldonado 1994: 260) The same behavior can be observed in their Basque counterparts: doubling does not entail ungrammaticality. (15) a. Bere amari lokartu zitzaion. his/her mother.dat fall asleep aux + 3sg.dat b. Poliziari ihes egin zion. Police.dat get away aux + 3sg.dat This property is closely related to the dative being or not an argument. Thus, for instance, Franco & Huidobro (2010: 218, section 3) assume that the possibility of having a DP in a case position doubled by a clitic reflects the argumental status of that clitic. Being this the case, ED-Class II are argumental, as shown in (12), whereas ED-Class I and ED-Class III are not—(4a) and (4b) respectively repeated here as (14a) and (14b): (16) a. Yo me he comido toda la sopa para cenar (*a cl cl.1sg.dat aux eat.pfv all the soup to dine to mí mismo). myself ‘I have eaten all the soup for dinner.’ b. El niño me le ha roto el libro (*a mí). the kid cl.1sg.dat cl.3sg.dat aux break the book ‘The kid has broken a book.’ The proper approach to the argumental vs non argumental nature of ED is open to discussion and depends on the particular theory on non-selected datives. For instance, Franco & Huidobro (2008) assume that ED-Class II are arguments whereas ED-Class I and III are adjuncts. Other novel approaches such as Bosse (2015) take the (German) ED to be an applied argument with only not-at issue meaning (or non-truth conditional meaning). Actually, in the particular per-

232

fernández

spective taken by this author in her (2015) work and in Bosse, Bruening and Yamada (2012) among others, ED are non-selected arguments of particular heads. In this chapter, I will assume ED-Class II to be arguments in their more generalized meaning. In particular, I will propose that these ED-Class II behave exactly as benefactives, experiencers and external possessors. I will follow Odria (2017) in assuming the very same merging position for experiencers and external possessors, i.e. the Specifier of an Applicative (Appl) head above v. Moreover, I will claim that these ED-Class II merge in this position. Hence, the properties attributed here to ED-Class II follow directly. In particular, as observed by Franco & Huidobro (2008), it is worth noting that the presence of the dative clitic doubling the DP is not optional but obligatory in Spanish inversion predicate structures, a particular instance of ED-Class II. The designation refers to structures with a subject like dative and a nominative theme with an agreeing verbal form, as in examples (17). The absence of the dative clitic entails ungrammaticality: (17) a. A Juan *(le) falta dinero. to Juan cl.3sg.dat lack money ‘Juan is missing some money.’ b. A Juan *(le) gustan los libros. to Juan cl.3sg.dat like the books ‘Juan likes the books.’ (Franco & Huidobro 2008: 219) With regards to these forms, Spanish and Basque look alike. (18) a. Joni dirua falta zaio / *da. to Juan lack money lack aux + 3sg.dat / aux ‘Jon is missing some money.’ b. Joni liburuak gustatzen zaizkio / *dira. to Juan the books like aux + 3sg.dat / aux ‘Jon likes the books.’ Note that the Basque examples include an absolutive marked argument instead of the corresponding nominative theme in Spanish counterparts. In order to be grammatical, the presence of the dative clitic is obligatory, as pointed out by Fernández (2010) Fernández & Ortiz de Urbina (2010) and recently Odria (2017) among others. This is not surprising, as leaving aside north-eastern varieties

on non-selected datives: ethical datives in basque and spanish

233

of Basque, datives clitics are obligatory elsewhere. However, in north-eastern varieties of Basque where dative clitic dropping is attested, dative clitics are obligatory with experiencers of psychological verbs of the gustatu ‘like’-type as in (18b), and ED-Class II in particular (18a)—external possessors and benefactives behave alike. Nevertheless, the designation inversion predicate structure does not seem accurate in Basque. The word inversion implies that the word order in Spanish examples (17a,b) is an inverted and thus, derived dative-nominative word order, but their Basque dative-absolutive word order counterpart seems to be the canonical and underived order. If ED-Class II are introduced by an Appl head above v, as proposed here, then the dative-absolutive order does not seem to be consequence of inversion. Under the same hypothesis, even Spanish examples (17a,b) show the order expected, if an Appl head above v is assumed to be responsible for introducing the ED-Class II, as I claim here. 3.3 Ethical Datives-Class II and Clitic Clusters In order to distinguish ethical datives from other datives, their ability to be combined with any other dative clitic seems to be crucial (see Jouitteau & Rezac 2007 and references therein). In particular, ED-Class III can co-occur with selected dative goals (19a) and non-selected datives such as benefactives (19b), ED-Class I (19c), ED Class II (19d) and external possessors (19e)—see also Maldonado (1994: 254–255). Example (19a) is provided by Franco & Huidobro (2008: 218) and examples (19b–d) by Maldonado (1994): (19) a. A mi primo me le pegaron una paliza en una fiesta. to my cousin ed cl.3sg.dat hit a beating at the party Intended: ‘They beat my cousin up at the party (on me).’ (ED-Class III + Goal) b. Nacho me le hizo la tarea a Valeria. Nacho ed cl.3sg.dat do the homework to Valeria ‘Nacho did the homework for Valeria.’ (ED-Class III + Benefactive) c. Juan se me compró una moto. Juan cl ed buy a motorbike ‘Juan bought a motorbike.’ (ED-Class I + ED-Class III) d. Me les rompieron el libro a los niños. ed cl.3pl.dat break a book to the kids Intended: ‘They broke the book to the kids (on me).’ (ED-Class III + EDClass II)

234

fernández

e. Me le rompió la nariz. ed cl.3sg.dat read break the nose Intended: ‘He/she broke his/her nose (on me).’ (ED-Class III + External possessor) This is particularly interesting as the presence of two dative clitics shows that ED-Class III, whatever they are, need to be separated from the rest, specifically, from ED-Class II (19d). On the other hand, ED-Class II, benefactives and external possessors cannot be combined with one another. Assuming, as I do, that all these non-selected datives are merged in Spec, Appl, as Odria (2017) do for Basque experiencers, external possessors and causees,5 the inability of ED-Class II to be combined with all these non-selected datives follows. On the other hand, Basque can never combine two dative clitics, whether they be selected or not (Fernández 2010), as there is a sole slot for a dative clitic in the morphological configuration. Therefore, the Basque counterpart (20a) to Perlmutter’s example in (13a) is ungrammatical. Besides, the ungrammaticality affects both ditransitive auxiliary forms (20a) and intransitive auxiliary forms (20b): two dative clitics never combine with each other. In (20a) the morpheme -t can only be interpreted as an ergative and never as a dative. (20) a. *Juanek neska horri bizitza hondatu Juan.erg girl that.dat life ruin diot aux + 3sg.dat + 1sg.dat Intended: ‘Juan ruined that girl’s life (on me).’ b. *Neska horri bizitza hondatu zaiot girl that.dat life ruin aux + 3sg.dat + 1sg.dat Intended: ‘That girl’s life is ruined (on me).’ As a consequence, this test does not apply in Basque. Nevertheless, the examples in (20) indirectly show that there is nothing comparable to ED-Class III in Basque.

5 I am deliberately ignoring other double dative constructions, such as those including differentially marked objects and goals or any other including derived datives (mainly goals), exhaustively analyzed by Odria (2017). For an analysis of other double dative constructions including an experiencer and an applied dative without verb, see in Berro & Fernández (to appear).

on non-selected datives: ethical datives in basque and spanish

3.4

235

Ethical Datives-Class II Involving Truth-Conditional Meaning: Ability to be Questioned and Binding Contrary to the properties discussed so far, Bosse, Bruening & Yamada (2012) take an approach to non-selected datives that combines both syntactic and semantic aspects. Basing on Potts (2005) among others, they distinguish truthconditional meaning (at-issue meaning in Potts’ designation) and non-truth conditional meaning (not-at-issue meaning), which are represented on separate tiers of meaning. As they show, only material on the truth conditional tier can be questioned or bind a variable. Thus, in their typology of non-selected datives, external possessors and benefactives are entirely truth conditional whereas attitude holders and subject co-referential datives are entirely non-truth conditional. Besides, in their proposal, affected experiencers may or may not contribute truth-conditional meaning. Hence, this is a parameterized property with yes value for German and Albanian and no value for Hebrew, as argued by the authors. ED can fit in three categories in their proposal: they can be either subject co-referential, attitude holders or affected experiencers. As far as Basque and Spanish are concerned, ED-Class I correspond to subject co-referential datives; ED-Class II correspond, roughly speaking, to affected experiencers and EDClass III to attitude holders. Let us explore the ability of ED-Class II to be questioned in the two languages under discussion. The property of some non-selected datives, specifically external possessors, being able to be questioned by a Wh-question and extracted is also mentioned and discussed by Borer & Grodzinsky (1986: 1981–1982). Thus, the test is not a new one, and it has been also explored by Arregi (2003) and Fernández (2010) with regard to Basque external possessors. In Bosse, Bruening & Yamada (2012), if ED are part of the truth-conditional tier, then they can be questioned by a Wh-phrase and extracted. As far as Basque is concerned, ED-Class II seem to contribute truth-conditional meaning, as they can be questioned by Wh-questions and extracted. (21) a. Nori lokartu zaio haurra? who.dat fall asleep aux + 3sg.dat kid Lit. ‘To whom has the kid fallen asleep?’ b. Nori apurtu dio liburua? who.dat break aux + 3sg.dat book Lit. ‘To whom has he/she broken the book?’ Spanish ED-Class II show the very same behavior.

236

fernández

(22) a. ¿A quién se le ha dormido el niño? to who cl cl.3sg.dat aux fall asleep the kid b. ¿A quién le ha roto el libro? to who cl.3sg.dat aux break the book Notice that ED-Class III—exemplified in (4b) and repeated here as (23a)— show the opposite behavior, as they cannot be questioned. Hence, a hypothetical Wh-question on the ED-Class III would be ungrammatical (23b). (23) a. El niño me le ha roto el libro the kid cl.1sg.dat cl.3sg.dat aux break the book ‘The kid has broken a book.’ b. *¿A quién le ha roto el libro? to who cl.3sg.dat aux break the book Intended: To whom has he/she broken the book to someone else? Notice that (23b) would be grammatical when asking about the one affected by the book being broken, but not about the one involved in the book being broken to him/her. It is worth pointing out that (23b) would be perfect if the clitic le were coreferential with ‘to whom’. Besides, ED-Class II can be a quantifier that binds a truth conditional element as a variable. Thus, ED-Class II contribute truth conditional meaning with regard to binding too. Basque examples show the pattern: (24) a. Ama guztiei beren haurrak lokartu zaizkie. mother every.dat their kids fall asleep aux + 3pl.dat ‘Kids have fallen asleep to every mother.’ b. Haur guztiei apurtu diete beren liburua. child every.dat break pfv aux + 3pl.dat their book ‘They have broken the book to every children.’ Spanish shows no difference regarding binding. (25) a. A todas las madres se les han dormido sus niños. to every mother cl cl.3pl.dat aux fall asleep their kids

on non-selected datives: ethical datives in basque and spanish

237

b. A todos los niños les han roto su cuento. to every kid cl.3pl.dat aux break their book If this is on the right track, then the value for both Basque and Spanish is yes, ED-Class II contribute truth conditional meaning. 3.5

Ethical Datives-Class II and Non-truth Conditional Meaning: Negation Till now, I have shown, following Bosse, Bruening & Yamada (2012), that Basque and Spanish ED-Class II contribute truth conditional meaning. This is not surprising, as these datives are arguments (within Inflection Phrase) and behave as such. However, affected experiencers, equivalent to our ED-Class II, are also claimed to contribute non-truth conditional meaning, as their meaning projects beyond negation and yes-no questions and they cannot be directly negated. In this regard, Spanish and Basque show an unexpected pattern in these authors’ proposal, as ED-Class II do not contribute non-truth conditional meaning in either Spanish or Basque. Let us start with Basque and Spanish negative examples, similar to those in Bosse & Bruening (2010). (26) a. Joni ez zitzaion haurra lokartu (Aneri bai). Jon.dat not aux + 1sg.dat kid fell asleep Ane.dat yes ‘The kid didn’t fall asleep.’ b. A Jon no se le ha dormido el niño (A Ane sí). to Jon not cl cl.3sg.dat aux fell asleep the kid to Ane yes As argued by Bosse & Bruening (2010) and Bosse, Bruening & Yamada (2012), the experience itself cannot be negated: it is the whole verbal event, but not the experience, that is negated. However, at least in Basque and Spanish, EDClass II can be negated. Hence, the meaning of the experience does not go beyond negation (against what has been observed by these authors for languages such as Albanian and German among other languages). If this is so, ED-Class II do not contribute non-truth conditional meaning. Hence, it does not confirm the yes value attributed to affected experiencers by the authors with regard to this property. This unexpected value is also attested in ditransitive structures such as (27a) and (27b), negative counterparts of examples (1b) and (3b). Once again, it seems that not only the verbal event but also the experience itself can be negated:

238

fernández

(27) a. Haurrak ez dit liburua apurtu. kid.erg not aux + 1sg.dat book break ‘The kid hasn’t broken a book (but if s/he had, it would have affected me).’ b. El niño no me ha roto el libro. the kid not cl.1sg.dat aux break the book The reason for that might be that, contrary to Maldonado (1994) and this chapter, the datives in (1b) and (3b) and their corresponding negation in (26a) and (26b) do not show an affected experiencer but a benefactive—assuming that there is a distinction between affected experiencers and benefactives, something not so obvious attending to our results (see Table 7.1 below). If they were benefactives and not affected experiencers, then, as in other benefactives analyzed by Bosse & Bruening (2010), the benefactive relation itself could be negated, as benefactives are entirely truth conditional. (28) a. Haurrak liburua apurtu du, baina ez dit niri kid.erg book break aux but not aux + 1sg.dat me.dat apurtu. break ‘The kid has broken a book, but not for me.’ b. El niño ha the kid aux ha roto a aux break to

roto el libro, pero no me lo break the book but not cl.1sg.dat cl.3sg.acc mí. me

Nevertheless, I see no clear reason to treat these examples as not showing an ED. Although much research is needed in principle, the contribution to nontruth conditional meaning does not seem to apply in Basque and Spanish as far as ED-Class II are concerned.

4

Discussion and Conclusions

So far, I have presented and analyzed some significant properties of nonselected datives in general and ED in particular. These properties and their yes/no values are represented in Table 7.1. This table has been inspired by those in Bosse & Bruening (2010) and Bosse, Bruening & Yamada (2012), but has been

on non-selected datives: ethical datives in basque and spanish table 7.1

239

Typology of non-selected datives inspired by Bosse & Bruening (2010), extended and modified

Person restriction Doubling Clitic Cluster Truth conditional meaning Non-truth conditional meaning Languages

ED-Class I

ED-Class III ED-Class II (allocutives)

Benefactives External possessors

no no yes no

yes no yes no

no yes no yes

no yes no yes

no yes no yes

yes

yes

no

no

no

Basque, Spanish

Basque, Spanish

Basque, Spanish

Spanish

Spanish, (Basque) Other designations Subject co-referential, Attitude Personal datives holders

Affected experiencers, Affected datives

extended and modified mainly, but not only, on the basis on Franco & Huidobro (2008: 223). Besides, I have maintained the general designation along with its particular classes, ED-Class I, III and II, instead of subject co-referential, attitude holders and affected experiencers respectively. The table also includes allocutives, which share the values of the properties attributed to ED-Class II, although this only covers a part of the intricate nature of allocutives not necessarily involving dative-like clitics. One more property could be added to the table, that is, subject co-reference attested in ED-Class I with a yes value, and with a no value in the remaining cases. Nevertheless, person restriction distinguishes both ED-Class I and III. Thus, it has not been included in the table. Besides, there is a striking (and undesirable) scenario with regard to EDClass II, benefactives and external possessors: they show exactly the same properties and cannot be distinguished from one another. The property ‘NP must be sentient’ analyzed by Bosse, Bruening & Yamada (2012) could in principle distinguish ED-Class II and benefactives, being ED-Class II sentient and benefactives not necessarily. Unfortunately, ED-Class II do not seem to be necessarily sentient in either Basque or Spanish, as shown in the following examples: (29) a. Aitita zenari hilarria apurtu diote. grandfather late.dat gravestone break aux + 1sg.dat ‘They have broken the gravestone to the late grandfather.’

240

fernández

b. Le han roto la lápida al difunto abuelo. cl.3sg.dat aux break gravestone to the late grandfather As a consequence, ED-Class II and benefactives share exactly the same properties in Basque and Spanish, and cannot be distinguished from each other, since both choose the value no for the property NP must be sentient and the same no value for the property Non-truth conditional meaning (vs Bosse, Bruening & Yamada 2012). On the other hand, sharing the same properties discussed here, the claim that ED-Class II merge in [Spec Appl], as proposed in Odria (2017) for Basque experiencers and external possessors, seems to be on the right track. More research is needed in order to distinguish syntactically ED-Class II, benefactives and external possessors from one another. Their merging position can explain their uniform behavior with regard to the properties mentioned in this chapter, but still more syntactic tests are needed in order to explain their divergent nature. Bosse, Bruening & Yamada (2012) is a good attempt to do so, although unfortunately it does not solve some of the main issues, at least with regards to Basque and Spain.

Acknowledgements The author would like to thank Jon Ortiz de Urbina for his valuable comments on a previous version of this paper and to Solveig Bosse for some feedback. All errors remain my own. The author also acknowledges the research funding received from the Basque Government (IT665–13) and the Spanish Ministry of Economy and Competitiveness (FFI2014–51878-P). The research leading to these results has also received funding from the European Union’s Seventh Framework Program for research, technological development and demonstration under grant agreement no. 613465.

References Arregi, Eurídice. 2003. “On possessor raising effects in Basque”. Ms., UPV/EHU. Authier, Marc & Lisa Reed. 1992. “On the syntactic status of French affected datives”. The Linguistic Review 9: 295–311. Berro, Ane & Beatriz Fernández. To appear. “Applicatives without verbs”. Natural Language and Linguistic Theory (NLLT). Bosse, Solveig & Benjamin Bruening. 2010. “Benefactive versus experiencer datives”. In

on non-selected datives: ethical datives in basque and spanish

241

Mary B. Washburn, Katherine McKinney-Bock, Erika Varis, Ann Sawyer & Barbara Tomaszewicz. (eds.), Proceedings of 28th WCCFL. 69–77. Bosse, Solveig, Benjamin Bruening & Masahiro Yamada. 2012. “Affected experiencers”. Natural Language and Linguistic Theory (NLLT), 30(4): 1185–1230. Bosse, Solvig. 2015. Applicative arguments. Peter Lang. Christian, Donna. 1991. “The personal dative in Appalachian English”. In Peter Trudgill & J.K. Chambers (eds.), Dialects of English. London: Longman. 11–19. Borer, Hagit & Yosef Grodzinsky. 1986. “Syntactic cliticization and lexical cliticization: The Case of Hebrew Dative Clitics”. In Hagit Borer (ed.), Syntax and Semantics 19. New York: Academic Press. 175–215. Fernández, Beatriz. 2010. “Goi datiboak eta are goragokoak euskaraz: jabe datiboak eta datibo hunkituak vs. datibo etikoak”. Gogoa 10, 1–2. San Sebastián/Donostia: UPV/EHU. 1–20. Fernández, Beatriz & Jon Ortiz de Urbina. 2010. Datiboa hiztegian. Bilbo: UPV/EHU. Franco, Jon & Huidobro, Susana. 2008. “Ethical Datives, Clitic doubling and the Theory of pro” Joyce Bruhn de Garavito & Elena Valenzuela (eds.), Selected Proceedings of the 10th Hispanic linguistics Symposium. Somerville: Cascadilla Proceedings Project. 215–224. Gutierrez Ordoñez, Salvador. 1999. “Los dativos”. In Ignacio Bosque & Violeta Demonte (eds.), Gramática descriptiva de la lengua española II. Madrid: Real Academia Española / Espasa Calpe. 1855–1930. Horn, Laurence R. 2008. “I love me some him: The landscape of non-argument datives”. In Olivier Bonami & Patricia Cabredo Hofherr (eds.), Empirical Issues in Syntax and Semantics 7, CSSP, Paris, 169–192. [On-line publication, http://www.cssp.cnrs.fr/ eiss7]. Horn, Laurence R. 2013. “I love me some datives: Expresive meaning, free datives, and F-implicature”. In Daniel Gutzmann & Hans M. Gärtner (eds.), Beyond Expressives: Explorations in Use-Conditional Meaning. Leiden/Boston: Brill. 153–201. Jaeggli, Osvaldo A. 1982. Topics in Romance Syntax. Dordrecht: Foris. Jaeggli, Osvaldo A. 1986. “Three issues in the theory of clitics: case, doubled NP s and extraction”. In Hagit Borer (ed.), Syntax and Semantics 19. New York: Academic Press. 15–42. Jouitteau, Mélanie & Milan Rezac. 2007. “The French ethical dative, 13 syntactic tests”. Bucharest Working Papers in Linguistics, IX (1): 97–108. Maldonado, Ricardo. 1994. “Dativos de interés sin intereses”. Verbo y estructuras frásicas [Actas do IV Coloquio Internacional de Lingüística Hispánica]. Porto, Facultad de Letras. 241–264. Miguel, Elena de & Marina Fernández Lagunilla. 2000. “El operador aspectual se”. Revista Española de Lingüística 30 (1). 13–43. Odria, Ane. 2017. The syntax of Differential Object Marking and datives in Basque. Ph.D. thesis, UPV/EHU.

242

fernández

Oyharçabal, Beñat. 1993. “Verb agreement with non-arguments: On allocutive agreement”. In José Ignacio Hualde & Jon Ortiz de Urbina (eds.), Generative studies in Basque linguistics. Amsterdam: John Benjamins. 89–114. Perlmutter, David. 1971. Deep and surface constraints in syntax. New York: Holt, Reinhart and Winston. Potts, Chris. 2005. “The logic of conventional implicatures”. Oxford: Oxford University Press. Pylkkänen, Liina. 2008 [2002]. Introducing arguments. Cambridge/London: the MIT Press. Strozer, Judith. 1978. “On the so-called ‘dative of interest’ ”. Hispania 61: 117–123. Webelhuth, Gert & Clare Dannenberg. 2006. “Southern American personal datives: The theoretical significance of syntactic variation”. American Speech 81: 31–55.

chapter 8

Differential Object Marking in Basque and Spanish Dialects Ane Odria

1

Introduction

Many southwestern varieties of Basque display Differential Object Marking (DOM) (Bossong 1985, 1991; Aissen 2003). In these varieties, human and (generally) definite objects—especially first and second person pronouns—tend to bear dative marking, instead of the canonical absolutive expected in an ergative language like Basque (Fernández & Rezac 2010, 2016; Mounole 2012; Odria 2012, 2014, 2017, to appear; Rodríguez-Ordóñez 2013, 2016).1 Southwestern varieties of Basque are those which are in contact with Spanish, and thus, also with Spanish DOM—i.e., a-marking (Pensado 1995, Torrego 1998, Leonetti 2004, a.o.). The Basque varieties situated in the French speaking area are referred as northeastern varieties and, as happens in Standard Basque, they show no DOM, since only the canonical absolutive marking is available for them. This paper compares the syntactic behavior of Basque and Spanish DOM, paying special attention to the Spanish variety spoken in the Basque speaking area—i.e., Basque Spanish (Landa 1995). Additionally, the paper distinguishes the syntax of DOM from dative objects in bivalent unergative predicates. It argues that the distinction between these kinds of objects is not only configurational—as has been argued in previous work—, but also categorical. Basque and Spanish DOM share significant commonalities. The differential marking is determined by animacy and specificity in both of them, and it is morphologically identical to the dative marking of indirect objects. Besides, Basque Spanish is particular in showing DOM both in the nominal—i.e., amarking—and in the clitic system of the verbal complex—i.e., leísmo (Franco 1993, Landa 1995, Fernández-Ordóñez 1994, 1999). This makes Basque Spanish even closer to Basque DOM, as in Basque the differential marking is attested not only in the nominal, but also in the finite verbal form. 1 DOM is optional for many southwestern Basque speakers and its actual distribution is subject to dialectal as well as idiolectal variation. Besides, its appearance is highly reduced with third person objects.

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004395398_009

244

odria

In both Basque and Spanish, DOM objects show the same morphology as the dative objects of bivalent unergative predicates of the lagundu (Basque) / ayudar (Spanish) ‘accompany, help’ type: (i) nominal as well as verbal dative marking in Basque, (ii) a-marking in Spanish, and (iii) cliticization with le(s) in (Basque) Spanish. However, in the case of bivalent unergatives, the marking of the object happens to be independent from factors like animacy and specificity and, semantically speaking, the object patterns more akin to the goal dative of ditransitive predicates. As a consequence, it has been argued that, contrary to the direct object configuration of DOM objects, the dative objects in bivalent unergatives have an indirect object configuration—see Fernández & Ortiz de Urbina (2010, 2012) and Ortiz de Urbina & Fernández (2016) for Basque, Torrego (2010) and Fábregas (2013) for Spanish, and Pineda (2016) for Romance languages in general. This paper makes a further step in the syntactic distinction between Basque and Spanish DOM on the one hand, and dative objects in bivalent unergatives on the other. It claims that apart from their syntactic configuration, these objects are also distinguished by their categorical status. While DOM objects pattern with causee, experiencer and possessor datives in exhibiting a DP syntactic category, the datives in bivalent unergatives behave more akin to goal datives and thus show a PP-like behavior. This contrast is evidenced by the licensing of depictive secondary predication, which—along with the rest of DP s—is allowed with DOM datives, but—as happens with PP s and goal datives—is generally rejected with datives in bivalent unergatives (Odria 2015, 2017). The paper is structured as follows. In section 2, I lay out the main properties of the DOM found in Basque (section 2.1), in Spanish (section 2.2) and in Basque Spanish (section 2.3). Section 3 compares DOM objects with dative objects in bivalent unergatives in both Basque (section 3.3) and Spanish (section 3.4). Section 4 deepens on the syntactic distinction exhibited by these types of objects. Based on the licensing of depictive secondary predication, it argues that, while DOM objects are DP s categorically, the datives in bivalent unergatives show a PP-like—i.e., a dual DP/PP—character. This is again proved with both Basque (section 4.1) and Spanish (section 4.2) data. Finally, chapter 5 closes the paper by summing up the main conclusions.

differential object marking in basque and spanish dialects

2

245

DOM in Basque, Spanish and Basque Spanish

2.1 Basque DOM 2.1.1 Canonical vs. Non-canonical Object Marking In Standard as well as in other varieties of Basque, the subject of a transitive predicate is marked with ergative case (-k) and the object with absolutive case (-ø). Likewise, as illustrated in (1), the finite verbal form agrees with the two arguments by means of their respective markers: -zu for the second person ergative and n- for the first person singular absolutive. (1) Zu-k ni ikusi nauzu. you-e I.a see aux[1sg.a-2sg.e]2 ‘You have seen me.’ In contrast to the canonical configuration, southwestern Basque varieties displaying DOM tend to mark human and (generally) definite objects dative, as in (2). (2) Zu-k ni-ri ikusi didazu. you-e I-d see aux[1sg.d-2sg.e] ‘You have seen me.’ In (2), the subject is marked with ergative case (-k), but the object appears with dative case (-(r)i) rather than with the absolutive. Besides, the two arguments are coded by their respective markers in the finite verbal form: the second person ergative by -zu and the first person singular dative by -da-. Contrary to what happens in the DOM of widely studied accusative languages, in Basque the differential marking co-appears with the ergative marking of the transitive subject, leading thereby to a configuration with two arguments bearing a morphologically marked case: the ergative and the dative.3 This is an unexpected pattern from a typological point of view, because the canonical configuration in transitive clauses is supposed to bear a single

2 The following abbreviations will be used in the text: e (ergative), d (dative), a (absolutive), ine (inesive), acc (accusative), all (allative), dest (destinative), caus (causative), fut (future), prog (progressive), nom (nominalizer), comp (complementizer), m (masculine), f (feminine), aux (auxiliary verb), sg (singular), pl (plural), 1/2/3 (first, second and third person). 3 Basque is not an exception in this regard. Other languages like Hindi also display DOM together with an ergative subject (Mahajan 1990, Bhatt & Anagnostopoulou 1996).

246

odria

marked case, either in the subject—as happens in ergative languages—or in the object—as in accusative languages. In addition, as occurs in many DOM languages, in Basque the differential marking happens to be morphologically to the dative marking in indirect objects. This is a common pattern cross-linguistically (Bossong 1991: 154, 157– 158; Aissen 2003: 446), as other DOM languages governed by both animacy and specificity show the same behavior as well.4 The morphological identity shared by DOM and indirect objects can be observed by comparing the example in (2) with that in (3). In (3) we see that the indirect object of a ditransitive predicate is marked dative in Basque. (3) Zu-k ni-ri liburua eman didazu. you-e I-d book.a give aux[3sg.a-1sg.d-2sg.e] ‘You have given me the book.’ In (3), the subject is marked with ergative case (-k), the direct object with the absolutive (-ø) and the indirect object with the dative (-(r)i). Likewise, the three arguments are cross-referenced by the finite verbal form, which shows the same shape as that in (2): the second person ergative is coded by -zu and the first person singular dative by -da-.5 Despite being morphologically identical to indirect objects, DOM objects share the same thematic as well as argumental relationships with absolutive direct objects. Hence, it seems reasonable to assume that DOM objects generate in the same syntactic position as canonical absolutives—i.e., in the complement position of V. This is common ground in the literature on the topic, given that, in spite of the differential marking, DOM objects are considered to be direct objects configurationally—see, among others, Bárany (2018). 2.1.2 The Main Factors Conditioning Basque DOM As happens in many DOM languages, in Basque the dative marking of the object is favored by both animacy and specificity, as the object has to be human and (generally) definite in order to be differentially marked (Fernández & Rezac 2010, 2016; Mounole 2012; Odria 2012, 2014, 2017; Rodríguez-Ordóñez 2013, 2016).

4 With few exceptions, Romance and Semitic languages are clear examples of this typological tendency, and so are other modern Indo-Aryan languages like Hindi or Punjabi and Amerindian languages like Guaraní and Aymara (Bossong 1991: 157). 5 In Basque the third person absolutive is not overtly marked in the verbal form.

differential object marking in basque and spanish dialects

247

On the one hand, dative marking is only accepted with human-referring objects. This is illustrated in the examples in (4), where only the human object Jon (4a) is able to display dative marking.6 (4) a. Jon-i ikusi diot. Jon-d see aux[3sg.d-1sg.e] ‘I have seen Jon.’ b. *Telebista-ri ikusi diot. television-d see aux[3sg.d-1sg.e] ‘I have seen the television.’ c. Telebista ikusi dut. television.a see aux[3sg.a-1sg.e] ‘I have seen the television.’ Besides, even though the main cutting point between DOM and non-DOM objects is determined by humanness, person deserves special attention too. In fact, many DOM speakers make a further distinction within human objects and distinguish between first and second person objects on the one hand, and third person objects on the other (Fernández & Rezac 2010, 2016; Mounole 2012; Odria 2012, 2014, 2017; Rodríguez-Ordóñez 2016). As a consequence, it is easier to find dative marking with first and second person objects than with third person. This distinction is realized in different manners. For some speakers, only first and second person objects can carry dative marking with a verb like ikusi ‘see’. As shown in (5) and (6), this is the case of a speaker from Zumaia (Central Basque).7 (5) a. Kali-an ikusiyazu. street-ine see.aux[1sg.d-2sg.e] ‘You have seen me in the street.’

6 Some DOM speakers consider examples like (4a) quite marginal, and add that the dative marking of a third person object would be more natural with other verbs. This may be due to the social stigmatization of the DOM phenomenon with the verb ikusi ‘see’, as it is one of the most corrected errors at school as well as in Basque academies. 7 The dialectal data presented in this paper were collected in the fieldwork carried out by myself in previous work—see especially Odria (2017).

248

odria

b. Kali-an ikusizut. street-ine see.aux[2sg.d-1sg.e] ‘I have seen you in the street.’ (6) a. *Jon-i kali-an ikusiyot. Jon-d street-ine see.aux[3sg.d-1sg.e] ‘I have seen Jon in the street.’ b. Jon kali-an ikusiet. Jon.a street-ine see.aux[3sg.a-1sg.e] ‘I have seen Jon in the street.’ For other speakers, the distinction between first and second vs. third person objects is not so rigid. A speaker from Larrabetzu (Western Basque), for instance, finds DOM with first (7a) and second (7b) person very natural, but prefers the absolutive for the third one—although in present tense the dative is acceptable in this case too (8). (7) a. Ez dostezu ikusi? not aux[1sg.d-2sg.e] see ‘Haven’t you seen me?’ b. Ikusi dotsut kontzertu-en. see aux[2sg.d-1sg.e] concert-ine ‘I have seen you in the concert.’ (8) a. Leire-ri ikusi dotset kotxe barri-en. Leire-d see aux[3sg.d-1sg.e] car new-ine ‘I have seen Leire in the new car.’ b. Leire ikusi dot kotxe barri-en. Leire.a see aux[3sg.a-1sg.e] car new-ine ‘I have seen Leire in the new car.’ The different behavior attested between first and second vs. third person objects is also mentioned in some descriptions of dialectal varieties. Hualde, Elordieta & Elordieta (1994: 125–127), for instance, notice that in Lekeitio Basque (a Western variety), DOM occurs more frequently with first and second person than with the third. Likewise, Ibarra (1995: 427) reports that in Ultzama Basque (a Navarrese variety), first and second person objects are gen-

differential object marking in basque and spanish dialects

249

erally marked dative, while DOM with third person objects is just optional. Furthermore, it is important to note that, generally speaking, the DOM examples provided by dialectal grammars involve usually objects of first and second person. Along with animacy—or even person—, specificity is also an important conditioning of Basque DOM. As argued by Mounole (2012), only human objects with a referential interpretation are able to display the differential marking—see also Fernández & Rezac (2010, 2016) and Odria (2012, 2014, 2017). Mounole (2012: 368–369) shows that DOM is mostly incompatible with the indefinite determiner bat ‘a’, the indefinite quantifier asko ‘many’ or the indefinite personal pronoun inor ‘nobody’. This is illustrated by the examples in (9), (10) and (11), provided by a speaker from Itsasondo (Central Basque). This speaker admits DOM with first, second and—less commonly—third person objects. However, in order to bear dative marking with the verb ikusi ‘see’, those of third person have to be not only human, but also definite. As a consequence, as happens with many other speakers, human objects containing indefinite determiners like bat ‘a’ (9), asko ‘many’ (10) and inor ‘nobody’ (11) tend to bear absolutive marking. (9) a. Neska bat ikusi det. girl one.a see aux[3sg.a-1sg.e] ‘I have seen a girl.’ b. *Neska bat-i ikusi diot. girl one-d see aux[3sg.d-1sg.e] ‘I have seen a girl.’ (10) a. Neska asko ikusittut. girl many.a see.aux[3pl.a-1sg.e] ‘I have seen many girls.’ b. *Neska asko-i ikusi diet. girl many-d see aux[3pl.d-1sg.e] ‘I have seen many girls.’ (11) a. Ez det inor ikusi. not aux[3sg.a-1sg.e] anyone.a see ‘I haven’t seen anyone.’

250

odria

b. *Ez diot inor-ri ikusi. not aux[3sg.d-1sg.e] anyone-d see ‘I haven’t seen anyone.’ In her description of DOM in Tolosa Basque (a Central variety), Mounole (2012; 369) observes that DOM can additionally distinguish between specific and nonspecific interpretations, arguing that indefinite nominals bearing dative marking entail a specific interpretation when combined with the verb bilatu ‘look for’.8 However, I have found no speaker admitting the dative marking with an indefinite nominal in that context; an example like (12a) is considered ungrammatical by all my consultants. The indefinite object can only carry absolutive marking, as in (12b)—the same results are obtained by Arraztio (2010) too in Araitz-Betelu Basque (a Central variety).9 (12) a. *Idazkari bat-i bila-tzen diot. secretary one-d look for-prog aux[3sg.d-1sg.e] ‘I look for a secretary.’ b. Idazkari bat bila-tzen dut. secretary one.a look for-prog aux[3sg.a-1sg.e] ‘I look for a secretary.’ Contrary to Mounole (2012), I thus conclude that instead of specificity per se, the main factor conditioning the marking of the object along with animacy is in fact definiteness. Basque DOM is then more restrictive than Romance languages like Spanish, where specificity rather than definiteness seems to be relevant. All in all, despite its significant impact, it should be noted that—as happens in many Romance languages (Bossong 1991: 160–161)—, in Basque definite-

8 Instead of ‘specificity’, Mounole uses the term ‘referentiality’. However, for the sake of simplicity, I continue using the notion ‘specificity’, assuming that both share the same meaning. 9 At least in some southwestern varieties, the sentence in (12b) would sound more natural with the verb behar ‘need’, which—as happens with bilatu ‘look for’—requires absolutive marking with indefinite objects (i). In these varieties, bilatzen dut ‘look for.prog aux’ would occur in a verbal periphrasis formed with the verb ibili ‘be’ (ii): (i) Idazkari bat/*-i behar dut / *diot. secretary one.a/bat-d need aux[3sg.a-1sg.e] / aux[3sg.d-1sg.e] ‘I need a secretary.’ (ii) Idazkari bat-en bila nabil. secretary one-ine look for be.aux[1sg.a] ‘I’m looking for a secretary.’

differential object marking in basque and spanish dialects

251

ness does not condition the marking of the object as rigidly as animacy does. Whereas DOM is never attested with non-human or inanimate objects, in some cases indefinite objects can occasionally occur with the differential marking. This groups the DOM found in Basque with Spanish DOM, where animacy is also known to be more influential than specificity (Brugger & Brugé 1996, Leonetti 2004). Besides, although their influence is not as generalized as that caused by the referential properties of the object, clausal properties like tense and finiteness affect the marking of the object as well, given that DOM is sometimes reduced in present tense as well as non-finite contexts (Fernández & Rezac 2010, 2016; Odria 2017). This is a particular aspect of Basque DOM, as cross-linguistically clausal factors like tense and finiteness do rarely condition the distribution of DOM. Be that as it may, the influence of these factors is far from being a systematic pattern and seems to be related to morphological facts. Likewise, it should be noted that the actual distribution of Basque DOM is also determined by the nature of the verb, since the differential marking is more common with some verbs than with others (Mounole 2012, Rodriguez-Ordóñez 2016, Odria 2017). 2.2 Spanish a-marking In Spanish, animate and specific objects are marked by the preposition a, whereas inanimate and non-specific ones are left unmarked—see, among many others, Pensado (1995), Brugè & Brugger (1996), Torrego (1998), Leonetti (2004), Rodríguez-Mondoñedo (2007), Fábregas (2013) and Ormazabal & Romero (2013ab). This is illustrated in the examples in (13), provided by Ormazabal & Romero (2013a: 222). In (13a), the animate and specific object la niña ‘the child’ bears a-marking. On the contrary, although specific, the inanimate el libro ‘the book’ is unable to be preceded by a in (13b), and the same thing happens with the animate generic object niñas ‘children’ in (13c). (13) a. He encontrado *(a) la niña. have found.1sg dom the child.f ‘I have found the girl.’ b. He encontrado (*a) el libro. have found.1sg dom the book ‘I have found the book.’ c. He encontrado (*a) niñas. have found.1sg dom children.f ‘I have found girls.’

252

odria

As in Basque, Spanish DOM involves a marker that is morphologically identical to that found with indirect objects. This is shown in (14), where the indirect object is preceded by the same a-marker carried by the animate and specific direct object in (13a). (14) Le he comprado un libro a la niña. 3sg.d have bought.1sg one book to the child.f ‘I have bought the child a book.’ Despite their similarities, Spanish behaves different to Basque in four main aspects. First, being an accusative language, DOM leads to a typologically expected configuration, with only one of the two arguments of a transitive construction being overtly marked. Second, it has no person distinction when assigning the differential marking—i.e., in Spanish a-marking is equally available with first, second and third person objects. Third, the differential marking appears only in the noun phrase, and not in the clitic system of the verbal complex. And fourth, clausal properties like tense and finiteness do not affect the distribution of DOM, given that there is no contrast between present vs. past tenses and inflected vs. non-inflected clauses. 2.3 Basque Spanish leísmo In contrast to Standard as well as other varieties of Spanish, Basque Spanish exhibits a pattern which makes its DOM closer to that found in southwestern Basque, as the differential marking is not only attested in the nominal, but also in the clitic system of the verbal complex. Basque Spanish is a leísta dialect: the dative clitic le(s) is used instead of the accusative lo(s)/la(s) when referring to an animate object.10 This is exemplified in (15) and (16), where the inanimate object el coche ‘the car’ is coded by the accusative clitic lo (15), whereas the dative le pronominalizes both the animate masculine object el chico ‘the boy’ and the feminine la chica ‘the girl’ (16).11

10

11

Contrary to what happens in Basque Spanish, the leísmo in Standard Spanish refers to the coding of masculine, singular (usually personal) direct objects by the dative clitic le, substituting this way the accusative lo—and only exceptionally la (Fernández-Ordóñez 1994: 7, 1999: 1319; Landa 1995: 152). Consequently, the Basque Spanish leísmo is also referred as real leísmo (Fernández-Ordóñez 1999) or animated leísmo (Urrutia-Cárdenas 1995, 2003). Urrutia-Cárdenas (2003: 292) mentions that the use of le(s) can also be extended to inanimate objects. Nevertheless, I agree with Landa (1995: 8) in claiming that such pattern is hardly acceptable among Basque Spanish speakers.

differential object marking in basque and spanish dialects

253

(15) a. Vi el coche aparcado. saw.1sg the car parked ‘I saw the car parked.’ b. Lo / *le vi aparcado. 3sg.acc / 3sg.d saw.1sg parked ‘I saw it parked.’ (16) a. Vi al chico / a la chica contento/-a. saw.1sg dom the boy / dom the girl happy.m/-f ‘I saw the boy/girl happy.’ b. Le vi contento/-a. 3sg.d saw.1sg happy.m/-f ‘I saw him/her happy.’ Moreover, the leísmo in Basque Spanish is special in allowing clitic doubling with a-marked objects in their canonical object position (Franco 1993, Landa 1995, Fernández-Ordóñez 1994, 1999). Consider now the examples in (17).12 (17) a. Le he llevado al niño a casa. 3sg.d have carried.1sg dom the child.m to house ‘I have carried the child at home.’ b. Le conozco a Jon desde pequeño. 3sg.d know.1sg dom Jon since child ‘I know Jon since he was a child.’ As noted by Landa (1995: 117), in Basque Spanish clitic doubling depends on a-marking. That is, those objects that are doubled by the le(s) clitic must be a-marked, which means that the objects allowing clitic doubling are also animate and specific—or more precisely, animate and presuppositional, as argued by Franco (1993) and Landa (1995).13 Take the sentences in (18). In (18a), the

12 13

This is rejected in the Standard leísmo, where the same object would only be differentially marked in the nominal by means of a-marking. Landa (1995: 162) claims that, instead of specificity, in Basque Spanish clitic doubling with le(s) is constrained by presuppositionality and gives the example in (i) to support her hypothesis.

254

odria

object la niña ‘the child’ is a-marked, and the doubling with the dative clitic le is acceptable. On the contrary, in (18b), the object el libro ‘the book’ is inanimate, and thus, non-a-marked. As a consequence, doubling it with a dative or accusative clitic is ruled out. The same is true for the animate non-specific object camareros ‘waiters’ in (18c). (18) a. Le he visto a la niña. 3sg.d have seen.1sg dom the child.f ‘I have seen the child.’ b. (*Le / lo) he visto el libro. 3sg.d / 3sg.acc have seen.1sg the book ‘I have seen the book.’ c. (*Les / los) necesito camareros que sepan inglés. 3pl.d / 3pl.acc need.1sg waiters that know English ‘I need waiters that know English.’ Note that this is also the case in Basque DOM, as only those objects bearing dative case are able to be coded by dative marking in the agreement complex (19). (19) Zu-k ni-ri / *ni ikusi didazu. you-e I-d / I.a see aux[1sg.d-2sg.e] ‘You have seen me.’

(i)

Los burócratas le llegan a cansar a uno. the bureaucrats 3sg.d manage to tire.3pl dom one ‘Bureaucrats end up boring you.’ The object bearing clitic doubling in (i) cannot be interpreted as specific –uno is a generic animate pronoun, and the sentence is still acceptable in Basque Spanish. Based on this fact, Landa concludes that clitic doubling in Basque Spanish is only possible with animate and presuppositional objects. This would also explain the ungrammaticality of examples like (ii) and (iii), which contain non-presuppositional objects (Landa 1995: 165). (ii) ???Les vi a bastantes hombres 3pl.d saw.1sg dom enough men ‘I saw quite a few men.’ (iii) *Les necesito a seis hombres 3pl.d need.1sg dom six men ‘I need six men.’

differential object marking in basque and spanish dialects

255

Summing up, in spite of their differences, we see that Basque DOM and (Basque) Spanish DOM coincide in a great extent: (i) DOM is triggered by animacy and specificity, (ii) the differential marking is morphologically identical to the marking of the indirect object, and (iii) together with the noun phrase, the differential marking is also realized in the finite verbal form. These shared aspects are relevant because, as mentioned in section 1, DOM in Basque only exists in the dialects that are in contact with Spanish. In the literature on the topic, some authors have suggested that Basque influences the leísmo in Basque Spanish. As Basque makes no gender distinction on direct objects, authors like Fernández-Ordóñez (1994, 1999) and Landa (1995) have suggested that this could have reinforced the use of the dative le(s) instead of lo(s) for masculine objects and la(s) for feminine ones. Be that as it may, the so-called contact situation has also been analyzed in the opposite direction. Mounole (2012), for instance, attributes the spread of Basque DOM to the influence of Spanish, claiming that Basque DOM is at least reinforced by the contact with Spanish. Other authors like Austin (2006, 2015) or Rodríguez-Ordóñez (2013, 2016) go even further and claim that Basque DOM is not only reinforced, but also induced by the contact with Spanish—and more precisely, by the leísta Basque Spanish. This is, thus, an open issue.

3

DOM vs. Datives in Bivalent Unergative Predicates

3.1 Basque DOM vs. Datives in Bivalent Unergatives In addition to the transitive predicates exhibiting DOM, in Basque, dative objects can also occur with other predicates involving an ergative-dative configuration. As can be observed in (20), this is what happens with bivalent unergative predicates like lagundu ‘accompany, help’, jarraitu ‘follow’ or begiratu ‘look at’ among speakers that do not necessarily accept DOM, both in southwestern and northeastern Basque (Etxepare 2003, Fernández & Ortiz de Urbina 2010).14

14

Along with lagundu ‘accompany, help’, jarraitu ‘follow’ and begiratu ‘look at’ bivalent unergatives involve other verbs like abisatu ‘notify’, barkatu ‘forgive’, bultzatu ‘push’, deitu ‘call’, entzun ‘hear, listen to’, eskertu ‘thank’, heldu ‘hold’, obeditu ‘obey’ or ukitu ‘touch’— see Etxepare (2003) and Fernández & Ortiz de Urbina (2010) for a complete classification of this group of predicates.

256

odria

(20) Jon-i lagundu / jarraitu / begiratu diogu. Jon-d help / follow / look at aux[3sg.d-1pl.e] ‘We have helped/followed/looked at Jon.’ As happens with DOM objects, along with the ergative-dative configuration, in bivalent unergatives, the dative object alternates with the absolutive, forming thereby a canonical transitive configuration. This is illustrated in (21a) and (21b) with the verb deitu ‘call’. In (21a), the object is marked absolutive and a canonical ergative-absolutive alignment emerges. On the contrary, in (21b) the same object receives dative marking, giving rise to an ergative-dative frame.15 (21) a. Goiz-ean deitu zaitut. moring-ine call aux[2sg.a-1sg.e] ‘I have called you in the morning.’ b. Goiz-ean deitu dizut. morning-ine call aux[2sg.d-1sg.e] ‘I have called you in the morning.’ In spite of their similarities, robust evidence has been provided distinguishing the syntactic behavior of DOM objects on the one hand and the datives in bivalent unergatives on the other (Fernández & Ortiz de Urbina 2010, 2012; Fernández & Rezac 2010, 2016; Odria 2012, 2017; Ortiz de Urbina & Fernández 2016, Pineda 2016). While DOM objects are semantically similar to canonical absolutives, the datives in bivalent unergatives pattern more akin to indirect objects—i.e., goals of ditransitive predicates. This has lead Fernández & Ortiz de Urbina (2012), Ortiz de Urbina & Fernández (2016) and Pineda (2016) to propose that the ergative-dative frame with bivalent unergative predicates contains a silent direct object and that the sole dative argument is in fact an indirect object introduced by a Low Applicative head à la Pylkkänen (2008). In fact, as reported by these authors, many of these verbs are verbs of communication—i.e., abisatu ‘notify’ or deitu ‘call’—or obey verbs—i.e., obeditu ‘obey’. According to them, they involve a transfer of message or order, which is equivalent to a silent direct object.16

15

16

Generally speaking, in bivalent unergatives dative marking is preferred in southwestern dialects, while the absolutive is more frequent in northeastern ones (Fernández & Ortiz de Urbina 2010). Nevertheless, note that other bivalent unergatives correspond to verbs of relative mo-

differential object marking in basque and spanish dialects

257

In addition to their syntactic configuration, DOM and the datives in bivalent unergatives also differ with regards to the factors lying behind the dative marking. Contrary to what happens in DOM, animacy and specificity are irrelevant when marking the object dative in bivalent unergatives (Fernández & Ortiz de Urbina 2012, Ortiz de Urbina & Fernández 2016). The examples in (22), for instance, show that the object of verbs like jarraitu ‘follow’ (22a), itxaron/itxoin ‘wait for’ (22b) and begiratu ‘look at’ (22c) is assigned dative regardless of animacy, given that dative marking with inanimates is grammatical in all of them. (22) a. GPSa-ri jarraitu diot. GPS-d follow aux[3sg.d-1sg.e] ‘I followed the GPS.’ b. Hemen itxoin-go diogu trena-ri. here wait for-fut aux[3sg.d-1pl.e] train-d ‘We will wait here for the train.’ c. Koadroa-ri begira geratu naiz. painting-d look at stay aux[1sg.a] ‘I have stayed looking at the painting.’ It is true that in the case of some verbs animacy has a say in the dative marking of the object. As noted by Fernández & Ortiz de Urbina (2010: 164), this is the case of itxaron/itxoin ‘wait for’. These authors show that, with few exceptions, in the General Basque Dictionary Orotariko Euskal Hiztegia (OEH) (Mitxelena & Sarasola 1989–2005), the object of itxaron/itxoin ‘wait for’ is generally marked dative when animate (23a), and either dative (23b) or absolutive (23c) when inanimate.17 (23) a. Beste asko-ri etxeden ez ta ni-ri etxeden. other many-d wait for not and I-d wait ‘Not wait for many others and wait for me.’ (Cb Eg II 71)

17

tion—i.e., jarraitu ‘follow’ or lagundu ‘accompany, help’—and verbs of physical contact—i.e., heldu ‘hold’ or ukitu ‘touch’. Therefore, in those cases it would be harder to assume the indirect object hypothesis including a silent direct object. Of course, this does not mean that absolutive marking is rejected with animate objects. As

258

odria

b. Bihar joan-go naz Bilbora ontzia-ri itxaro-tera. tomorrow go-fut aux[1sg.a] Bilbao-all ship-d wait for-nomi ‘Tomorrow I will go to Bilbao to wait for the ship.’ (Echta Jos 929) c. Ikastetxe-ra sartzeko ordua itxaro-ten zaude-la. school-all to enter time wait for-prog aux[2sg.a]-comp ‘While you are waiting for the time to enter school.’ (Osk Kurl 73) However, in spite of its influence in verbs like itxaron/itxoin ‘wait for’, a closer look to the different types of bivalent unergatives indicates that animacy is not a conditioning trigger for the dative marking in these verbs. Its influence is only particular of certain verbs or even dialects/speakers, and it is not as systematic as in DOM. Likewise, person is neither determinant in marking the object dative, and the same thing happens with specificity. Consider now the examples in (24). (24) a. Lagun asko-ri deitu diet, baina ez dit friends many-d call aux[3pl.d-1sg.e] but not aux[1sg.d-3sg.e inor-k erantzun. anyone-e answer ‘I have called to many friends, but nobody has answered me.’ b. Ez zion inor-i begiratu. not aux[3sg.d-3sg.e] anyone-d look at ‘He/she didn’t look at anyone.’ c. Ikasle bat-i lagun-tzen geratu naiz. student one-d help-prog stay aux[1sg.a] ‘I have stayed helping to a student.’ In (24), we see that the same indefinite particles that are generally excluded with DOM are in fact acceptable with dative objects of bivalent unergatives. 3.2 Spanish a-marking vs. Datives in Bivalent Unergatives As happens in other Romance languages (Pineda 2016), Spanish also has some verbs whose objects bear a-marking regardless of specificity (Pensado 1995, To-

I have already pointed out, northeastern dialects show a preference to mark both animate and inanimate objects absolutive in bivalent unergative predicates.

differential object marking in basque and spanish dialects

259

rrego 1998, 2010; Leonetti 2004, Fábregas 2013). Similar to the Basque bivalent unergatives, in Spanish the animate object of certain verbs carries a-marking even with a non-specific reading. This is exemplified by Leonetti (2004: 85) with the verbs entrevistar ‘interview’ (25a) and admirar ‘admire’ (25b).18 (25) a. Cada estudiante entrevistará a un personaje conocido. each student will interview.3sg to a person well-known ‘Each student will interview a well-known person.’ b. Todas las niñas admiraban a algún cantante. every children.f admired.3pl to some singer ‘Every child admired some singer.’ Likewise, Fábregas (2013: 29) adds that a-marking can also be assigned to inanimate objects with other verbs. In (26a), for instance, it is possible for the inanimate object of acosar ‘harass’ to bear a-marking, and the marking is even obligatory for the inanimate object of ayudar ‘help’ in (26b) (Fábregas 2013: 29). (26) a. La policía acosa (a)l narcotráfico. the police harass.3sg to the drug-trafficking ‘The police harass drug-trafficking.’ b. La policía ayuda *(a)l narcotráfico. the police help.3sg to the drug-trafficking ‘The police help drug-trafficking.’ Similar to what happens in Basque, verbs like acosar ‘harass’ or ayudar ‘accompany, help’ show a parallelism with pure transitives involving DOM, as both are bivalent and both contain an a-marked object. Notwithstanding, contrary to the object of pure transitives, animacy and specificity do not necessarily hold in these verbs. Therefore, as in Basque, not all a-marked objects that occur in verbs entailing a bivalent argument structure involve DOM.

18

A more complete list of this group of verbs includes acusar ‘accuse’, admirar ‘admire’, afectar ‘affect’, ayudar ‘accompany, help’, castigar ‘punish’, entevistar ‘interview’, golpear ‘beat’, insultar ‘insult’, odiar ‘hate’, offender ‘offend’, saludar ‘greet’ and sobornar ‘bribe’ (Fábregas 2013: 27–28).

260

odria

3.3 Basque Spanish leísmo vs. Datives in Bivalent Unergatives In Basque Spanish, the apparent similarity shared by DOM and a-marked objects in verbs like ayudar ‘accompany, help’ is also visible in the clitic system of the verbal complex, since the dative clitic le(s) is used in both of them. Consider the examples in (27). (27a) corresponds to a DOM construction, as its object is semantically a theme and both a-marking and the dative clitic le(s) depend on the animacy and specificity of the object. In contrast, (27b) belongs to the class of verbs whose (usually) goal object receives a-marking regardless of animacy and specificity—i.e., bivalent unergatives. (27) a. Le llevé (al niño) a casa. 3sg.d brought.1sg dom the child.m to house ‘I brought the child home.’ b. Le ayudé (al niño) a casa. 3sg.d helped.1sg to the child.m to house ‘I helped/accompanied to the child home.’ As reported by Fernández-Ordóñez (1999: 123–1339), the dative clitic in sentences like (27b) should not be grouped with the leísmo attested in sentences like (27a). This author observes that in Spanish the dative clitic can code the object of verbs that tend to omit their direct object—i.e., tocar ‘touch’—as well as the object of verbs whose valency has been reinterpreted, as in ayudar ‘accompany, help’, avisar ‘notify’ or obedecer ‘obey’. In order to distinguish the two types of bivalent configurations involving le(s), Fernández-Ordóñez (1999) refers to the pattern in (27a) as real leísmo—see footnote 10—, while that in (27b) is considered as apparent leísmo. Besides, Torrego (2010), Fábregas (2013) and Pineda (2016) note that, in many cases, the configuration involving a-marking regardless of animacy and specificity corresponds to a ditransitive construction. In those cases, the verb happens to be related to a noun; such a noun would correspond to the direct object and the a-marked object to the indirect object, as illustrated in (28a) with golpear ‘beat’ and in (28b) with ayudar ‘help’ (Fábregas 2013: 29). (28) a. Golpear a uno ~ dar un golpe a uno beat to someone give a blow to someone b. Ayudar a uno ~ dar ayuda a uno help to someone give help to someone

differential object marking in basque and spanish dialects

261

Recall that this is also the main insight in Fernández & Ortiz de Urbina (2012), Ortiz de Urbina & Fernández (2016) and Pineda (2016). These authors claim that in Basque the dative object in bivalent unergatives is syntactically equivalent to the indirect object in ditransitive constructions involving a silent direct object. In fact, the Spanish correspondences in (28a) and (28b) are equally obtained in Basque bivalent unergatives. Consider, for instance, the ‘bivalent unergative/ditransitive’ correspondence between deitu/dei egin ‘call’ (29a), bultzatu/bultza egin ‘push’ (29b) or lagundu/laguntza eman ‘help’ (29c). (29) a. Norbait-i deitu ~ norbait-i dei egin someone-d call someone-d call do b. Norbait-i bultzatu ~ norbait-i bultza egin someone-d push someone-d push do c. Norbait-i lagundu ~ norbait-i laguntza eman someone-d help someone-d help give Hence, it seems reasonable to think that in both Basque and Spanish the single object that appears with dative morphology is in fact an indirect object in these cases. Having coupled the syntactic configuration of DOM objects with canonical absolutives and that of the datives in bivalent unergatives with indirect objects, the following section makes a further step in the syntactic distinction attested between these kinds of objects. It argues that the difference between DOM and dative objects in bivalent unergatives is basically reduced to their DP vs. PP-like—i.e., dual DP/PP—categorical status (Odria 2017). While DOM objects exhibit a DP category, the datives in bivalent unergatives behave more akin to the rest of PP-like goals. This is evidenced by the possibility or not to license depictive secondary predication.

4

The Licensing of Depictive Secondary Predication

4.1 The Dual DP vs. PP-like Category of Dative Arguments In Basque, some datives behave under certain circumstances more akin to PP s than to DP s, as they can occur without finite verbal agreement—see specially Albizu (2001) on this point. This is what happens with goal datives in northeastern Basque, which—contrary to experiencers and possessors—are able to appear as non-agreeing—see Masullo (1992), Demonte (1995), Romero

262

odria

(1997) and Cuervo (2003) for a similar pattern in Spanish.19 In northeastern Basque, causees are able to occur without dative markers too, but in this case the possibility to appear as non-agreeing does not seem to be as extended as with goals (Fernández, Ortiz de Urbina & Landa 2009).20 In southwestern Basque, non-agreeing datives are limited to specific configurations such as ditransitives affected by the PCC and double dative constructions involving a causee and a goal. Generally speaking, only goals are allowed to occur as nonagreeing in them, indicating that causees pattern more akin to experiencers and possessors in this regard. Based on these facts, in what follows I will refer to causee, experiencer and possessors as ‘DP datives’ and to both agreeing nonagreeing goals—as ‘PP-like datives’ (Odria 2017, to appear). The contrast between DP and PP-like datives has additionally been supported by the ability to occur as adnominals in headlines, which—along with the rest of PP s—is only possible for goal datives—see de Rijk (2008: 378), Albizu (2001: 63, 2009: 13), and specially, Fernández & Sarasola (2010) and Berro & Fernández (2018). As demonstrated by Fernández & Sarasola (2010), possessors and experiencers pattern with ergative and absolutive DP s in requiring the genitive marker -ren in order to function as adnominals. In this section, I claim that DP and PP-like datives can also be distinguished by their ability to license depictive secondary predication (Odria 2015, 2017, to appear). In Basque (Zabala 1993, 2003; Oyharçabal 2010) and Spanish (Demonte 1987, 1988; Romero 1997; Demonte & Masullo 1999), datives and PP s have been considered to be unable to control depictive secondary predication.21 Such a restriction has been illustrated by sentences like (30) and (31) 19

20

21

For non-agreeing datives in northeastern Basque, see Ortiz de Urbina (1995), Albizu (2001, 2009), Fernández (1997, 2010, 2014), Fernández & Landa (2009), Fernández et al. (2009), Fernández & Ortiz de Urbina (2010), Etxepare & Oyharçabal (2009ab, 2013), Etxepare (2014) and Ormazabal & Romero (2017). Fernández, et al. (2009: 214–215) report that non-agreeing datives in causative constructions should be analyzed more carefully, given that in this case the lack of dative markers could be influenced by external factors like indefiniteness, which is known to foster dative agreement drop (Ortiz de Urbina 1995). Besides, these authors note that, in their corpus—which is based on classical writings from the 19th and 20th centuries—some of the non-agreeing datives in causative constructions represent the goal of the embedded predicate, instead of the causee itself. Likewise, they add that the availability to have non-agreeing datives in causatives could be affected by idiolectal variation as well, as nonagreeing causees are found in the writings of some but not all north-eastern classical authors that make use of agreement drop when it comes to goals. In addition, as noted by Javier Ormazabal (p.c.), in northeastern Basque, dative agreement drop with causees appeared chronologically later than with goals. See Williams (1980) and Rothstein (1983) for the same pattern in English, McFadden (2004) for German and Koizumi (1994) for Japanese.

differential object marking in basque and spanish dialects

263

adapted from Zabala (1993: 255, 258)—, where neither the dative (30) nor the PP (31) is able to control the depictive secondary predicate. (30) a. *Miren-ii liburua urdurii oparitu diot. Miren-d book.a nervous gift aux[3sg.a-3sg.d-1sg.e] ‘I have gifted the book to Miren nervous.’ b. *A Maríai lei he regalado el libro nerviosai . to María 3sg.d have gifted.1sg the book nervous.m/-f ‘I have gifted the book to María nervous.’ (31) a. *Jostailu bat erosi dugu haurra-rentzati poziki . toy one.a buy aux[3sg.a-1pl.e] child-dest happy ‘We have bought a toy for the child happy.’ b. *Hemos comprado un juguete para el niñoi contentoi . have bought.1sg a toy for the child happy ‘We have bought a toy for the child happy.’ So far, the restriction in question has been mainly tested with goal datives. Notwithstanding, a closer look at the different nature of dative arguments reveals that not all kind of datives are equally reluctant when it comes to licensing depictive secondary predication. In fact, apart from the well-known exception of the causee, possessor and experiencer datives are also able to license this kind of predication. This indicates that, instead of the syntactic configuration—i.e., subject / direct object vs. indirect object—or case marking—i.e., ergative / absolutive vs. dative—, the restriction on secondary predication depends basically on the syntactic category of the controller, as DP—i.e., causee, experiencer and possessor—but not PP-like—i.e., goal—datives are able to do so. Let us first focus on the behavior of the causee. As illustrated in (32) (Zab.ala 1993: 269) and (33) (Demonte 1987: 154), causees are able to control depictive secondary predication.22

22

Comparing to intransitive and transitive causatives, secondary predication with ditransitive causatives seems to be harder to process both in Basque and Spanish, as apart from the subject—i.e., causer—and the direct object, they involve two dative arguments: the causee and the indirect object. For this reason, contrary to what I did in Odria (2014), I give no example of ditransitive causatives involving secondary predication.

264

odria

(32) a. Ama-k haurra-rii gaixoriki joan-arazi zion mother-e child-d ill go-caus aux[3sg.d-3sg.e] eskola-ra. school-all ‘The mother made the child go to school ill.’ b. Irakasle-ak Mikel-ii guzti-en aurrean bakarriki dantza-arazi teacher-e Mikel-d in front of everyone alone dance-caus dio. aux[3sg.d-3sg.e] ‘The teacher has made Mikel dance alone in front of everyone.’ c. Ama-k haurra-rii indabak gogorik gabei jan-arazi mother-e child-d beans.a without wanting eat-caus zizkion. aux[3pl.a-3sg.d-3sg.e] ‘The mother made the child eat the beans without wanting.’ (33) Juan (la)i hizo bailar a Maríai desnudai. Juan 3sg.acc made.3sg dance to María naked ‘Juan made María dance naked.’ Possessors show a similar behavior, as in spite of their dative marking, depictives are allowed to be hosted by them. Consider now the Basque examples in (34) and the Spanish examples in (35)—the examples in (35a) and (35b) are provided by Demonte & Masullo (1999: 2466) and that in (35c) by Hernanz (1988: 12).23 (34) a. (Ni-ri)i udaleku-etan ilea lokartutai moztu I-d summer camp-ine hair.a sleeping cut

23

Demonte & Masullo (1999: 2466–2467) explain that the licensing of depictive secondary predication by possessor datives is quite restricted in Spanish. On the one hand, they say that the depictives allowed in these contexts tend to be replaceable with gerundive verbal periphrases like estando dormida ‘being slept’ for dormida ‘slept’, or estando anestesiada ‘being anesthetized’ for anestesiada ‘anesthetized’. On the other hand, they note that in these constructions the possessor tends to be placed in a preverbal position and the depictive in a post-verbal one. Taking these and other facts into account, these authors conclude that in this case the secondary predicate appears in a position distinct to the predicates controlled by subjects and direct objects.

differential object marking in basque and spanish dialects

265

zidaten. aux[3sg.a-1sg.d-3pl.e] ‘In the summer camp, they cut me my hair sleeping.’ b. Jon-ii kistea anestesiatutai kendu zioten. Jon-d cyst.a anesthetized remove aux[3sg.a-3sg.d-3pl.e] ‘They removed the cyst to Jon anesthetized.’ (35) a. A Maríai, lei operaron el quiste dormidai. to María 3sg.d operated.3pl the cyst sleeping ‘María was operated the cyst sleeping.’ b. Lei extirparon el lunar a Consueloi anestesiadai. 3sg.d removed.3pl the mole to Consuelo anesthetized ‘They removed the mole to Consuelo anesthetized.’ c. A la enfermai, los cirujanos lei extirparon el quiste to the patient the surgeons 3sg.d removed.3pl the cyst anestesiadai. anesthetized ‘The surgeons removed the cyst to the patient anesthetized.’ Besides, the sentences in (36) illustrate that in Basque experiencers can license depictives too, a pattern that is again replicated in Spanish subject-like datives (37) (Fernández-Soriano 1999: 124).24 (36) a. (Ni-ri)i mozkortutai mundu guztia gustatzen zait, I-d drunk world whole.a like-prog aux[3sg.a-1sg.d] baina mozkortuta ez nago-en-ean ez but drunk not be.aux.[1sg.a]-comp-ine not dut inor aguanta-tzen. aux[3sg.a-1sg.e] no one stand-prog ‘When I’m drunk I like everyone, but when I’m not drunk I can’t stand anyone.’

24

The example in (37a) was gathered by Fernández-Soriano (1999: 124) in La Dorotea by Lope de Vega.

266

odria

b. (Ni-ri)i mozkortutai mundu guztia jaus-ten zait I-d drunk world whole.a get on-prog aux[3sg.a-1sg.d] ondo. well ‘When I’m drunk I get on well with everyone.’ c. (Ni-ri)i lotsatutai gaizki atera zitzaidan aurkezpena, beraz, I-d shy bad go aux[3sg.a-1sg.d] presentation hence hurrengoan lotsarik gabe joa-ten saiatu-ko naiz. next time with no shame go-comp try-fut aux[1sg.a] ‘The presentation went (to me) wrong shy, so next time I’ll try to do it with no shame.’ (37) a. Mal mei fue ausentei pero peor presentei. bad 1sg.d went.1sg absent but worse present ‘Things were bad for me while absent but worse when present.’ b. Nosi dieron las dos borrachosi. 1pl.d gave.3pl two o’clock drunk ‘It got as late as 2 o’clock on us and we were drunk.’ c. Lei ocurrió un accidente borrachai. 3sg.d happened.3sg an accident drunk ‘An accident happened to her while being drunk.’ Overall, contrary to goals and PP s, causee, experiencer and possessors seem to be compatible with depictive secondary predication, a pattern that is again attested not only in Basque, but also in Spanish. Taking this into account, the following sections examine the behavior of depictive secondary predication with the datives in DOM and in bivalent unergatives. Focusing on both Basque and Spanish, I show that only DOM objects allow the licensing of depictive secondary predication. Hence, given that depictives are only rejected by PP s and PP-like datives, this implies that, while datives in bivalent unergatives show a PP-like behavior, DOM objects behave categorically as the rest of canonical absolutives, that is, as DP s (Odria 2015, 2017, to appear). 4.2 DOM and Depictive Secondary Predication As has been demonstrated by Fernández & Rezac (2010, 2016) and Odria (2012, 2014, 2017, to appear), in Basque, DOM objects are able to control depictive sec-

differential object marking in basque and spanish dialects

267

ondary predication. This is illustrated in (39) with examples from Larrabetzu Basque (a Western variety). The examples in (38) belong to a speaker with no DOM (in present tense) and those in (39) to a speaker with DOM (in both present and past tenses). What is important for the purpose of this section is that the dative objects in (39) can control depictive secondary predication in the same way as the absolutive objects in (38). (38) a. (Zu-k) (ni)i ostondutei harrape nozu. you-e I.a hidden catch aux[1sg.a-2sg.e] ‘You have caught me hidden.’ b. (Ni-k) (zu)i tontotutei ikusi zaitut. I-e you.a silly see aux[2sg.a-1sg.e] ‘I have seen you silly.’ c. Lagun-ek Mireni mozkortutei grabe deure. friends-e Miren.a drunk record aux[3sg.a-3pl.e] ‘Her friends have recorded Miren drunk.’ (39) a. (Zu-k) (ni-ri)i ostondutei harrape dostezu. you-e I-d hidden catch aux[1sg.d-2sg.e] ‘You have caught me hidden.’ b. (Ni-k) (zu-ri)i tontotutei ikusi dotsut. I-e you-d silly see aux[2sg.d-1sg.e] ‘I have seen you silly.’ c. Lagun-ek Miren-erii mozkortutei grabe dotsie. friends-e Miren-d drunk record aux[3sg.d-3pl.e] ‘Her friends have recorded Miren drunk.’ Even though in Basque DOM exhibits a great deal of dialectal as well as idiolectal variation, the availability to control depictive secondary predication is a persistent pattern, given that the results obtained in Larrabetzu are generalized among the rest of the consulted speakers from other Basque varieties, such as Araitz-Betelu (Central Basque) (Arraztio 2010), Lekeitio and Dima (Western Basque) (Fernández & Rezac 2010, 2016), Elgoibar (Transitional CentralWestern Basque), Zumaia (Central Basque), Hondarribia (Transitional CentralNavarrese Basque), Oñati (Western Basque), Tolosa (Central Basque), Errenteria (Central Basque) and Itsasondo (Central Basque) (Odria 2014, 2017).

268

odria

As happens in Basque, in Spanish a-marked objects are equally able to license depictive secondary predication. As shown by Demonte (1987: 148), in (40a) the secondary predicate borracha ‘drunk’ cannot be controlled by the goal indirect object—i.e., Maria. Contrarily, the same author says that in (40b) the secondary predicate turns out to be controlled by the DOM object a María. (40) a. *Juan lei habló a Maríai borrachai Juan 3sg.d talked.3sg to María drunk ‘Juan talked to María drunk.’ b. Juan lai encontró a Maríai borrachai. Juan 3sg.acc found.3sg dom María drunk ‘Juan found María drunk.’ In order to justify the possibility for DOM objects to license depictive secondary predication, Demonte (1987) claims that, as in causee datives, a is not a true preposition in DOM objects. Instead, this author takes a in examples like (40b) to be a dummy preposition, a preposition that does not project a PP maximal projection and does not therefore impede c-command relations. This way, Demonte states that the DOM object a María in (40b) can control the depictive borracha ‘drunk’ because there is no P head that blocks the structural ccommand relation between the argument and the predicate. Interestingly, the possibility for DOM objects to license depictive secondary predication persists even when the object is coded—or clitic doubled—by the dative clitic le(s) in the leísta Basque Spanish. Consider, for instance, the examples in (41), which are the correspondent Spanish examples of those in (39).25 (41) a. Juan lei pilló (al niño)i escondidoi. Juan 3sg.d caught.3sg dom child.m hidden.m ‘Juan caught the child hidden.’ b. Juan lei vio (a Pedro)i atontadoi. Juan 3sg.d saw.3sg dom Pedro silly.m ‘Juan saw Pedro silly.’

25

In order to avoid interferences, in the Spanish examples in (41) the object bears the same gender as well as number specification as the subject. Likewise, so as to test the behavior of the dative clitic le(s), I have changed the objects in (41a) and (41b) to third person.

differential object marking in basque and spanish dialects

269

c. Su amiga lei grabó (a María)i borrachai. her friend 3sg.d recorded.3sg dom María drunk.f ‘Her friend recorded María drunk.’ Overall, assuming that depictives are only incompatible with PP s and PP-like datives, I conclude that both Basque and (Basque) Spanish DOM objects are DP s categorically. 4.3 Datives in Bivalent Unergatives and Depictive Secondary Predication Contrary to DOM objects, the datives in bivalent unergatives are generally unable to control depictive secondary predication (Fernández & Rezac 2010, 2016; Odria 2012, 2017; Ortiz de Urbina & Fernández 2016). As illustrated in (42), in this case, the only possible interpretation is that the secondary predicate tontotuta ‘silly’ (42a) and mozkortute ‘drunk’ (42b) modifies the subject Jon and not the dative object Mikel.26

26

Even being a bivalent unergative predicate, entzun ‘hear, listen to’ is exceptional allowing secondary predication (Fernández & Rezac 2010: 134, Odria 2017). This is illustrated in (i): (i) Ni-ki Mikel-ij mozkortutai/j entzun nion. I-e Mikel-d drunk hear/listen to aux[3sg.d-1sg.e] ‘I heard to Mikel drunk.’ I believe that the distinction between the two different meanings of entzun ‘hear, listen to’ might have an influence in this fact. Actually, when entzun means ‘listen to’, dative marking is generally attested in southwestern dialects, while the absolutive is preferred in northeastern ones (Fernández & Ortiz de Urbina 2010: 129, 134). On the contrary, when entzun means ‘hear’, the object can also be absolutive in southwestern varieties, behaving more akin to a canonical direct object. In fact, Ortiz de Urbina & Fernández (2016: 79– 80) mention that in a sentence like I heard the bossi angryi, the object is usually marked absolutive, and that dative marking makes the sentence slightly worse. Hence, if entzun in (i) means ‘hear’ instead of ‘listen to’, its dative object could be closer to a DOM object, and in that case, it would pattern similar to other perception verbs like ikusi ‘see’, which would explain its apparently exceptional behavior in (i). In addition, it is important to note that the dative object of lagundu ‘help, accompany’ can also control secondary predication when pragmatics forces to do so. This is the case in (ii), where along with the ergative subject, many speakers allow the depictive to modify the dative object. (ii)(Haiek)i (ni-ri)j etxe-ra mozkortutai/j lagundu zidaten they.e I-d house-all drunk help/accompany aux[1sg.d-3plE] ‘They helped/accompanied me home drunk.’ Crucially, some of the consultants add that in (ii) the possibility for the dative to control secondary predication is logically or pragmatically conditioned. Hence, I believe that the example in (ii) does not necessarily bring into question the fact that (under normal circumstances) datives in bivalent unergatives are unable to control secondary predication.

270

odria

(42) a. Jon-eki Mikel-ij tontotutai/*j begiratu zion. Jon-e Mikel-d silly look at aux[3sg.d-3sg.e] ‘Jon looked at Mikel silly.’ b. Jon-eki Mikel-ij mozkortutai/*j deitu zion. Jon-e Mikel-d drunk call aux[3sg.d-3sg.e] ‘Jon called Miren drunk.’ Spanish behaves once again like Basque, as it does not allow depictives to depict of the dative object in verbs like mirar ‘look at’ (43a) or llamar ‘call’ (43b).27 (43) a. Juani lej miró a Pedroj atontadoi/*j. Juan 3sg.d looked at.3sg to Pedro silly ‘Juan looked at Pedro silly.’ b. Juani lej llamó a Pedroj borrachoi/*j. Juan 3sg.d called.3sg to Pedro drunk ‘Juan called drunk to Pedro.’ All in all, the impossibility to license depictive secondary predication links the dative object of bivalent unergatives with PP-like goals of ditransitive constructions, and thus implies that these objects have a dual DP/PP categorical status.

5

Conclusions

In this paper I have compared the syntactic behavior of Basque and (Basque) Spanish DOM. I have highlighted the main similarities and differences attested between the two of them and I have distinguished them from the dative object in bivalent unergative predicates. Contrary to what happens in DOM, in bivalent unergatives the dative marking is independent from factors like animacy and specificity and the object patterns more akin to indirect objects of ditransitive predicates. Moreover, based on the licensing of depictive secondary predication, I have argued that these datives are additionally distinguished by their syntactic category. Contrary to DOM objects—which exhibit a DP categorical status—, datives in bivalent unergatives are generally incompatible with depictives, a pattern that groups them with the rest of PP-like goals. 27

In (43b), borracho ‘drunk’ can also be interpreted as a primary predicate referred to the dative Pedro. However, when behaving as a secondary predicate, borracho ‘drunk’ can only be controlled by the subject Juan.

differential object marking in basque and spanish dialects

271

On a broader perspective on dative arguments, the categorical distinction attested between DP and PP-like datives suggests that a different categorical origin should be assigned for the different kinds of datives. That is, a DP category for causee, experiencer, possessor and DOM datives, and a PP one for goal datives—this implies that the so-called restriction is related to the P head and holds at the level of argument structure. However, even being distinct in the first merge position, both DP and PP-like datives should share the same Agree/Case position, as both occur with dative markers in the finite verbal form and both pattern to a certain extent in the same way in different kinds of agreement restrictions (Odria 2017, to appear).

Acknowledgements This work has been supported by the Basque Government (the research group IT665–13) and the Spanish Ministry of Economy and Competitiveness (FFI2014–51878-P). The research leading to these results has also received funding from the European Union’s Seventh Framework Program for research, technological development and demonstration under grant agreement no. 613465. I am grateful to Ane Berro and Javier Ormazabal for their comments and suggestions. Needless to say, all errors are mine.

References Aissen, Judith. 2003. “Differential object marking: iconicity vs. economy”. Natural Language and Linguistic Theory 21: 435–483. Albizu, Pablo. 2001. “Datibo sintagmaren izaera sintaktikoaren inguruan: eztabaidarako oinarrizko zenbait datu”. In Beatriz Fernández & Pablo Albizu (eds.), Kasu eta komunztaduraren gainean. On Case and Agreement. Bilbo: UPV/EHU Editorial Services. 49–69. Albizu, Pablo. 2009. “Construcciones inacusativas con dativos posesivos y dativos de interés en vasco: un análisis derivacionalista”. Seminario de Lingüística Teórica, CSIC, Madrid, May 11. Arraztio, Kontxi. 2010. “Datibo osagarri bitxiak Araitz-Beteluko hizkeran”. In Beatriz Fernández, Pablo Albizu & Ricardo Etxepare (eds.), Euskara eta euskarak: aldakortasun sintaktikoa aztergai. [Supplements of ASJU LII]. Bilbo: UPV/EHU. 113–149. Austin, Jennifer. 2006. “Dative overmarking in Basque”. Euskalingua 9: 136–145. Austin, Jennifer. 2015. “Transfer and contact-induced variation in child Basque”. Frontiers in Psychology 5: 1–11.

272

odria

Bárany, András. 2018. DOM and dative case. Glossa: a journal of General Linguistics 3(1): 97, 1–40. Berro, Ane & Beatriz Fernández. 2018. “Applicatives without verbs”. Natural Language and Linguistic Theory (NLLT). Bhatt, Rajesh & Elena Anagnostopoulou. 1996. “Object Shift and Specificity: Evidence from ko-phrases in Hindi”. In L. Dobrin et al. (eds.), Proceedings of CLS 32. Chicago: Chicago Linguistics Society. Bossong, Georg. 1985. Empirische Universalienforschung. Differentielle Objekt Markierung in der neuiranischen Sphrachen. Tübingen: Narr. Bossong, Georg. 1991. “Differential Object Marking in Romance and beyond”. In Dieter Wanner & Douglas A. Kibee (eds.), New Analyses in Romance Linguistics. Amsterdam/Philadelphia: John Benjamins. 143–170. Brugè, Laura & Gerhard Brugger. 1996. “On the Accusative a in Spanish”. Probus 8(1). 1–51. Cuervo, M. Cristina. 2003. Datives at Large. Ph.D. thesis, MIT. Demonte, Violeta. 1987. “C-command, Preposition and Predication”. Linguistic Inquiry 18: 147–157. Demonte, Violeta. 1988. “Remarks on secondary predicates. C-command, extraction and reanalysis”. The Linguistic Review 6–1: 1–39. Demonte, Violeta. 1995. “Dative Alternation in Spanish”. Probus 7. 5–30. Demonte, Violeta & Pascual J. Masullo. 1999. “La predicación. Los complementos predicativos”. In Ignacio Bosque & Violeta Demonte (eds.), Gramática descriptiva de la lengua española. Madrid: Espasa Calpe, RAE. 2461–2524. Etxepare, Ricardo. 2003. “Valency and Argument Structure in the Verb”. In José Ignacio Hualde & Jon Ortiz de Urbina (eds.), A grammar of Basque. Berlin: Mouton de Gruyter. 369–465. Etxepare, Ricardo. 2014. “Contact and Change in a restrictive theory of parameters”. In Carme Picallo (ed.), Linguistic Variation in the Minimalist Framework. Oxford: Oxford University Press. 108–139. Etxepare, Ricardo & Bernard Oyharçabal. 2009a. “Bi datibo egitura ipar-ekialdeko zenbait hizkeratan”. Lapurdum XIII: 145–158. Etxepare, Ricardo & Bernard Oyharçabal. 2009b. “Datiboa ipar-ekialdeko zenbait hizkeratan”. Ms., IKER, April 25. Etxepare, Ricardo & Bernard Oyharçabal. 2013. “Datives and adpositions in northeastern Basque”. In Beatriz Fernández & Ricardo Etxepare (eds.), Variation in datives: a micro-comparative perspective. Oxford: Oxford University Press. 50–95. Fábregas, Antonio. 2013. “Differential Object Marking in Spanish”. Borealis. An International Journal of Spanish Linguistics 2–2: 1–80. Fernández, Beatriz. 1997. Egiturazko kasuaren erkaketa euskaraz. Bilbo: UPV/EHU Editorial Services.

differential object marking in basque and spanish dialects

273

Fernández, Beatriz. 2010. “Goi datiboak eta are goragokoak euskaraz: jabe datiboak eta datibo hunkituak vs. datibo etikoak”. Gogoa, 10: 1–2: 1–20. Fernández, Beatriz. 2014. “Goi-eta behe-datiboak eta Datibo-komunztaduraren murriztapena: ipar-ekialdeko hizkerak eta euskara estandarra”. Irantzu Epelde (ed.), Euskal dialektologia: lehena eta oraina [Supplements of ASJU LXIX]. UPV/EHU. 37–63. Fernández, Beatriz & Josu Landa. 2009. “Datibo komunztadura beti zaindu, inoiz zaindu ez eta batzuetan baino zaintzen ez denean. Hiru ahoko aldagaia, datu iturri bi, eta erreminta bat: Corsintax”. Lapurdum 13: 159–181. Fernández, Beatriz & Jon Ortiz de Urbina. 2010. Datiboa hiztegian. Bilbo: UPV/EHU Editorial Services. Fernández, B. & J. Ortiz de Urbina, J. 2012. “Dative (First) Complements in Basque”. In Ernestina Carrilho & Beatriz Fernández (eds.), Syntactic microvariation in Westmost European Languages. Journal of Portuguese Linguistics (Special Issue), 11-1. Edições Colibrí—Unviersidade de Lisboa. 83–98. Fernández, Beatriz, Jon Ortiz de Urbina & Josu Landa. 2009. “Komunztadurarik gabeko datiboen gakoez”. In Ricardo Etxepare, Ricardo Gómez & Joseba A. Lakarra (eds.), Beñat Oihartzabali gorazarre. ASJU XVIII: 352–380. Fernández, Beatriz & Milan Rezac. 2010. “Datibo osagarri bitxiak eta Datiboaren Lekualdatzea: ari nai diyot eta kanta egin nazu bidegurutzean”. In Beatriz Fernández, Pablo Albizu, & Ricardo Etxepare (eds.), Euskara eta euskarak: aldakortasun sintaktikoa aztergai. [Supplements of ASJU LII]. Bilbo: UPV/EHU. 113–149. Fernández, Beatriz & Milan Rezac. 2016. “Differential Object Marking in Basque varieties”. In Beatriz Fernández & Jon Ortiz de Urbina (eds.), Microparameters in the grammar of Basque. Amsterdam/Philadelphia: John Benjamins. 93–139. Fernández, Beatriz & Ibon Sarasola. 2010. “Marinelei abisua: izen ondoko datibo sintagmak izenburuen sintaxian”. Lapurdum XIV: 55–75. Fernández-Ordóñez, Inés. 1994. “Isoglosas internas del castellano. El sistema referencial del pronombre átono de tercera persona”. Revista de Filología Española 74-1/2: 71–125. Fernández-Ordoñez, Inés. 1999. “Leísmo, laísmo y loísmo”. In Ignacio Bosque & Violeta Demonte (eds.), Gramática descriptiva de la lengua española I. Madrid: Espasa Calpe. 1317–1394. Fernández-Soriano, Olga. 1999. “Two types of impersonal sentences in Spanish: locative and dative subjects”. Syntax 2–2: 101–140. Franco, Jon A. 1993. “Conditions on clitic doubling: the agreement hypothesis”. ASJU 27: 285–298. Hernanz, Mª Lluisa. 1988. “En torno a la sintaxis y la semántica de los complementos predicativos en español”. Estudis de Sintaxis, Estudi General, 8 (Colegi universitari deGirona) 7–29. Hualde, José Ignacio, Gorka Elordieta & Arantzazu Elordieta. 1994. The Basque dialect of

274

odria

Lequeitio [Supplements of ASJU XXXIV]. Bilbo & Donostia: UPV/EHU & Gipuzkoako Foru Aldundia. Ibarra, Orreaga. 1995. Ultzamako hizkera. Inguruko euskalkiekiko harremanak. Iruñea: Nafarroako Gobernua. Koizumi, Masatoshi. 1994. “Secondary predicates”. Journal of East Asian Linguistics 3: 25–79. Landa, Alazne. 1995. Conditions on Null Objects in Basque Spanish and their relation to Leísmo and Clitic Doubling. Ph.D. thesis, University of Southern California. Leonetti, Manuel. 2004. “Specificity and differential object marking in Spanish”. Catalan Journal of Linguistics 3: 75–114. Mahajan, Anoop Kumar. 1990. The A/A-bar distinction and movement theory. Ph.D. thesis, MIT. Masullo, Pascual J. 1992. Incorporation and case theory in Spanish. A crosslinguistic perspective. Ph.D. thesis, University of Washington. McFadden, Thomas. 2004. The position of morphological case in the derivation: a study on the syntax-morphology interface. Ph.D. thesis, University of Pennsylvania. Mounole, Céline. 2012. “Evolution of the transitive verbs in Basque and apparition of datively marked patients”. In Gilles Authier & Katharina Haude (eds.), Ergativity, Transitivity, and Voice. Berlin/New York: Mouton de Gruyter. 355–379. Odria, A. 2012. What lies behind differential Object Marking: a Basque dialect survey. Master’s thesis, UPV/EHU. In Pello Salaburu, Patxi Goenaga & Ibon Sarasola (eds.), Sareko Euskal Gramatika. Odria, Ane. 2014. “Differential Object Marking and the nature of dative case in Basque varieties”. Linguistic Variation, 14-2. Amsterdam/Philadelphia: John Benjamins. 289– 317. Odria, Ane. 2015. “Euskarazko bigarren mailako predikazioaren murriztapen sintaktikoez”. In Beatriz Fernández & Pello Salaburu (ed.), Ibon Sarasola gorazarre. Homenatge. Homenaje. Bilbo: UPV/EHU. 449–515. Odria, Ane. 2017. Differential Object Marking and datives in Basque syntax. Ph.D. thesis, UPV/EHU. Odria, Ane. to appear. “DOM and datives in Basque: not as homogeneous as they might seem”. In A. Pineda & M.A. Irimia (eds.), DOM and datives -a homogeneous class? Linguisticae Investigationes. Amsterdam/Philadelphia: John Benjamins. Ormazabal, Javier & Juan Romero. 2013a. “Differential Object Marking, Case and Agreement” Borealis. An International Journal of Hispanic Linguistics 2–2: 221–239. Ormazabal, Javier & Juan Romero. 2013b. “Non Accusative Objects”. Catalan Journal of Linguistics 12. 155–173. Ormazabal, Javier & Juan Romero. 2017. “Historical Changes in Basque Dative Alternations: evidence for a P-based (neo)derivational analysis”. Ms., UPV/EHU & University of Extremadura.

differential object marking in basque and spanish dialects

275

Ortiz de Urbina, Jon 1995. “Datibo komunztaduraren gainean”. In Ricardo Gómez & Joseba A. Lakarra (eds.), Euskal Dialektologiako Kongresua (Donostia, 1991ko Irailak 2–6) [Supplements of ASJU XXVIII]. Donostia: Gipuzkoako Foru Aldundia. 579–588. Ortiz de Urbina, Jon & Beatriz Fernández. 2016. “Datives in Basque bivalent unergatives”. In Beatriz Fernández & Jon Ortiz de Urbina (eds.), Microparameters in the grammar of Basque. Amsterdam/Philadelphia: John Benjamins. 67–93. Oyharçabal, Bernard. 2010. “Basque ditransitives”. In Maia Duguine, Susana Huidobro & Nerea Madariaga (eds.), Argument structure and syntactic relations: A cross-linguistic perspective. Amsterdam/Philadelphia: John Benjamins. 233–260. Pensado, Carmen. 1995. “El complemento directo preposicional: estado de la cuestión y bibliografía comentada”. In Carmen Pensado (ed.), El complemento directo preposicional. Madrid: Visor. 11–59. Pineda, Anna. 2016. Les fronteres de la (in)transitivitat. Estudi dels aplicatius en llengües romàniques i basc. Barcelona: Institut d’Estudis Món Juïc (Col·lecció Cum Laude, 6). Pylkkänen, Liina. 2008. Introducing arguments. Cambridge/London: MIT Press. de Rijk, Rudolf P. 2008. Standard Basque. A Progressive Grammar. Cambridge/London: The MIT Press. Rodríguez-Mondoñedo, Miguel. 2007. The Syntax of Objects: Agree and Differential Object Marking. Ph.D. thesis, University of Connecticut. Rodríguez-Ordóñez, Itsaso. 2013. “Contact-induced phenomena in Gernika Basque: the case of dative over-marking”. Selected Proceedings of the 15th Hispanic Linguistics Symposium, Athens, Georgia. 236–251. Rodríguez-Ordóñez, Itsaso. 2016. DOM in Basque: grammaticalization, attitudes and ideological representations. Ph.D. thesis, University of Illinois at Urbana-Champaign. Romero, Juan. 1997. Construcciones de doble objeto y gramática universal. Ph.D. thesis, Universidad Autónoma de Madrid. Rothstein, Susan D. 1983. The syntactic forms of predication. Ph.D. thesis, MIT. Torrego, Esther. 1998. The dependencies of objects. Cambridge, Mass.: MIT Press. Torrego, Esther. 2010. “Variability in the case pattern of causative formation in Romance and its implications”. Linguistic Inquiry 41: 445–470. Urrutia-Cárdenas, H. 1995. “Morphosyntactic features in the Spanish of the Basque Country”. In Carmen Silva-Corvalán (ed.), Spanish in four continents. Washington, D.C.: Georgetown University Press. 243–259. Urrutia-Cárdenas, Hernán. 2003. “Los clíticos de tercera persona en el español en el País Vasco”. Revista de Filología y su Didáctica 26: 517–538. Williams, Edwin S. 1980. “Predication”. Linguistic Inquiry 11: 203–237. Zabala, Igone. 1993. Predikazioaren teoriak Gramatika Sortzailean (Euskararen kasua). Ph.D. thesis, UPV/EHU. Zabala, Igone. 2003. Nominal Predication: copulative sentences and secondary predication. In José Ignacio Hualde & Jon Ortiz de Urbina (eds.), A Grammar of Basque. Berlin: Mouton de Gruyter. 426–446.

chapter 9

Complex Causative Verbs and Causees in Basque (and Romance) Jon Ortiz de Urbina

1

Introduction

Basque productive causatives are primarily morphological: a causative morpheme arazi (also erazi in some dialects, as well as eragin, erazo or arazo in western and especially Biscayan Basque) is added to a verbal base (the verbal root in standard Basque), forming a derived causative verb, as in (1b):1 (1) a. Semeak medizina hartu du. son medicine take aux ‘The son took the medicine.’ b. Jonek semeari medizina harrarazi dio. Jon.erg son.dat medicine take.cause aux ‘Jon made the son take the medicine.’ In principle, this initial description sets the Basque data apart from the Romance ones, if causatives in Spanish and French are of the analytic type (Comrie 1989; Dixon 2000). Although usually bound and without the ability to occur freely, however, the causative ‘morpheme’ in Basque is fairly transparently a verb. This brings the Basque causative construction closer to that with causative verbs like Spanish hacer or French faire ‘make’. The type of complex verb formed in all these languages and complementation types are addressed first in Section 2. Section 3 will be devoted to causee marking and agreement conflicts resulting from the presence of two datives in the same structure. As we will see, this is also an area where new patterns converging with Romance ones seem to be emerging or developing side by side with traditional ones. 1 The following abbreviations have been used in glosses: abs ‘absolutive’, aux ‘auxiliary’, comp ‘complementizer’, dat ‘dative’, det ‘determiner’, comp ‘complementizer’, erg ‘ergative’, fem ‘feminine’, fut ‘future’, gen ‘genitive’, inst ‘instrumental’, masc ‘masculine’, nom ‘nominalizing affix’, pl ‘plural’, q ‘question morpheme’, subj ‘subjunctive’.

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004395398_010

complex causative verbs and causees in basque (and romance)

2

277

Complex Verb Formation in Causatives

As indicated above, Basque causative verbs are complex words containing two verbal forms. There are reasons to treat this combination not as a result of morphological derivation, but as a syntactic one. Section 2.1 provides some justification for this analysis, which, in turn, allows for a closer comparison with Romance in general and Spanish in particular. The complex verb unit in the languages under discussion is examined in section 2.2, and section 2.3 focuses on direct and indirect causation in these structures. Impersonal causatives and/or faire par constructions are discussed in section 2.4. 2.1 Basque Causative Morphemes as Bound Morphemes In the standard language and in dialects where arazi/erazi ‘cause’ is used, this is a bound morpheme and must form a unit with the verbal root expressing the caused event, making this look like a standard case of morphological causative formation. However, all the Basque causative ‘morphemes’ mentioned so far are verbs, and more precisely, verbs diachronically formed by affixation of the archaic (probably morphological) causative marker -ra- to a verbal root (see de Rijk 2008: 375–377). Thus, although the root to which -ra- in arazi/erazi itself may have been attached is no longer extant as a verbal root, eragin ‘cause’ is clearly (to the linguist, not to the speaker) the archaic causative form of egin ‘do’. Synchronically, moreover, these causative forms exist as independent verbs, especially but not exclusively eragin and erazo. The General Basque Dictionary (OEH) gives independent entries for all of them, with one meaning ‘make do, cause, force’. This use is normal with eragin (2b) but definitely marginal in the case of arazi (2a): (2) a. Ene errenak ezion ikasinai biloaren motxtan baia my daughter.in.law neg.aux learn.want hair.gen cut but arazi naun.2 cause aux ‘My daughter in law did not want to learn how to cut hair, but I forced her to.’ b. Beharrak beharra eragiten du. need.erg need makes.do aux ‘Need makes one do what is necessary.’ 2 Example from a mid 20th century letter to Basque linguist Azkue, in Roncalese, included in the OEH under arazi.

278

ortiz de urbina

Eragin as an independent causative verb is standard in the case of causative of light verb constructions where egin ‘do’ combines with bare nominal complements in unergative-like expressions like lan egin ‘work, lit. do work’, negar egin ‘cry, lit. do cry’, lo egin ‘sleep, lit. do sleep’, etc. (see Levin 1983; Laka 1993 or Berro 2015): (3) a. Umeak negar egin du. child.erg cry do aux ‘The child cries. (lit. does crying)’ b. Umeari negar eragin diote. child.dat cry do.cause aux ‘They have made the child cry.’ Outside causative unergatives, though, eragin ‘cause’ as an independent verb, although clearly semantically linked to causation, has lost a transparent connection with egin ‘do’; as a result, a productive causative where arazi is attached to eragin is attested in the General Basque Dictionary, with the meaning of ‘cause to do, cause to move’ (4a), and in turn this has even spread to the light verb in substandard unergative causatives (4b): (4) a. egin deitzüdan gaizkiez oroz eta besteri eragin erazi do aux.that evil.acts all.for and other.to do.cause cause dütüdanez aux ‘for all the evil acts I performed and for those that I made others perform’ (UskLiB 101) b. ene tristeziak ihez eragiñ arazitzen baitarot gizonen my sadness.erg escape do.cause cause since.aux men konpaiñietarik company.from ‘since my sadness makes me flee from the company of men’ (Gç 47) The existence of causative morphemes as independent verbs, then, approximates these structures to Romance causatives more than if the causative morpheme were a regular derivational morpheme like, for instance, chi in Quechua or (t)at/(t)et in Hungarian. The order of the two verbs in Basque is, as expected for a head last language, event-cause (which would however be the case both if this is a morphological unit or a syntactic unit).

complex causative verbs and causees in basque (and romance)

279

The latter is the approach taken already in DHM (1989), following Baker (1988) and also in the spirit of Zubizarreta’s (1985) Complex Verb Hypothesis, although the choice is made there for theoretical reasons. Some evidence for this syntactic approach can be derived from the fact that VV compounding is not found in Basque morphology outside of these cases. Moreover, there is a correlation between the shape the event verb takes in causatives in any given dialect and the availability of the root in the syntax of that dialect. That is, while all dialects use the verbal root in morphological derivation, it is only found as a free form in the syntax of eastern dialects. Subjunctive and potential forms are then combined with the root in these dialects (har ‘take’ in 5a), but with the participial form (which happens to be the neutral, citation form of the verb in Basque) in western dialects (hartu in 5b): (5) a. … hori har dezan that take aux.comp ‘so s/he takes that’ b. … hori hartu dagian that take aux.comp ‘so s/he takes that’ Similarly, in western dialects that use eragin as the causative verb, this is combined not with the verbal root, as in the standard language, but with the perfective participle: we then find hartueragin rather than harrarazi ‘make take’, or ikusieragin rather than ikusarazi ‘make see’. The use of the ‘syntactic’ verbal base in causative formation, rather than the verbal root, reserved for morphology in western dialects, strongly suggests causatives involve a syntactic combinatory process rather than a morphological one.3 An important difference between causative verbs in Basque and Romance hacer/faire is that the former do not take tensed clausal complements, while the latter do (6b), generally conveying a more indirect type of causation than with tenseless ones:

3 According to the OEH, the easternmost use of eragin ‘cause’ can be found in the Labourdin writer Urte, where, as expected, it combines with the verbal root, productive in the syntax of that dialect. Central dialects also use this causative verb occasionally, and just like with their main causative verb arazi, it seems to combine with either the verbal root of the participial form depending on the syntactic productivity of the former.

280

ortiz de urbina

(6) a. Le hizo volver a Pedro. him made return to Pedro ‘S/he made Pedro return.’ b. Hizo que Pedro volviera. made that Pedro return.subj ‘S/he made Pedro return.’ Tensed complements with Basque causative arazi or eragin do not seem to exist. Very occasionally, examples with egin ‘do’ can be found, as in (7): (7) Egizu beraz guziak maita detzadan. make then all love aux.comp ‘Make [me] love them all.’ (Brtc 207) Similarly, in Sarasola’s (1985) dictionary, many entries for causative verbs with erazi are provided with definitions which involve the verb egin ‘do’ taking a subjunctive embedded clause, as ito erazi ‘cause to drown’, glossed as “make that something drown”, higi erazi ‘cause to move’, given as “make that something move” or iraun erazi ‘cause to last’ as “make that something last”. The glosses are as marked in Basque as their translations are in English, and look more like a metalinguistic device than actual examples of productive language.4 Distributionally equivalent to subjunctive tensed clauses in many respects, tenseless nominalized complement clauses are also very occasionally found as complements of a causative verb, instead of the general root or participial form. As before, this possibility does not exist with arazi itself but rather with other causative verbs, arazo ‘cause’ in (8a), eragin ‘cause’ in (8b) and eginarazi ‘cause to do’ in (8c): (8) a. Neuri be erazo egin daustazu emakumien aurka … esaten. I.dat too cause do aux women.gen against saying ‘You have also made me say [speak] … against women.’ (Otx 29)

4 This use is not registered in the very comprehensive General Basque Dictionary (OEH). B. Fernández (p.c.) points out, however, that it is found in the speech of young speakers of Ondarroa Basque.

complex causative verbs and causees in basque (and romance)

281

b. Irabazi nebanean poz handiagoa sorraraztea eragin win aux.when happiness bigger appear.cause.nom cause eban. aux ‘When I won, this made a bigger happiness appear.’ (Euskonews Gaztea, 94) c. Eginerazi zeitan funtsen partajatzia. do.cause aux fund.gen.pl share.nom.det ‘(S)he made me share the funds.’ (Etch 140) Leaving these isolated cases aside, this paper will exclusively focus on the general pattern, where the causative verb combines with a verb in its root or participial form, as in (1b). 2.2 Causatives as Complex Verbal Forms Already in DHM (1989) the combination of the two verbs in causative constructions is interpreted as a syntactic combination of two verbs, technically expressed as the result of head-to-head movement following Baker’s (1988) incorporation analysis. This analysis brings Basque causatives close to their Romance counterparts with infinitival complements of causative verbs. This initial parallelism between Basque and Romance is particularly clear with respect to Spanish constructions with adjacent verbs like (9a), as opposed to those like (9b) with the causee intervening between them. The first option, on the other hand, is the only one available in other Romance languages like Italian or French (9c,d): (9) a. Jon (le) hizo tomar la medicina a su hijo. Jon him made take the medicine to his son ‘Jon Made his son take the medicine.’ b. Jon (le) hizo a su hijo tomar la medicina. c. Jean a fait prendre le médicament à son enfant. Jon has made take the medicine to his son ‘Jon Made his son take the medicine.’ d. *Jean a fait à son enfant prendre le médicament.

282

ortiz de urbina

The connection between the two verbs in Basque, however, is more complete than what one finds in Romance. Thus, the type of complement taken by the causative verb makes it impossible for the complement verb to be negated (10b)5 or have aspectual modification or passive morphology, unlike Spanish (10c,d,e): (10) a. El alcohol te hace no caminar derecho. the alcohol you makes not walk straight ‘Alcohol causes you not to walk straight.’ b. *Edariek zuzen ez ibilerazten zaitu drinks.erg straight not walk.cause aux ‘(Intended) Drinks make you not walk straight.’ c. ??Su ceguera le hizo no haberse dado cuenta antes. his blindness him made not have given notice earlier ‘His blindness made him not notice earlier.’ d. Esta píldora te hará estar durmiendo muchas horas seguidas. this pill you make be sleeping many hours in.a.row ‘This pill will make you sleep for many hours in a row.’ e. ? El juez le hizo ser defendido por un abogado de oficio.6 the judge him made be defended by a lawyer of office ‘The judge made him be defended by a public defender.’ While perfective in (c) is very marginal (probably due to the temporal sequencing inherent in situations where someone is forced to do something), passive is

5 The presence of negation has the familiar effect of preventing ‘clause union’ effects, so that clitic climbing is not possible: (i) El alcohol te hizo entenderlo. El alcohol te hizo no entenderlo. the alcohol you made understand.it the alcohol you made not understand it (ii) El alcohol te lo hizo entender. *El alcohol te lo hizo no entender the alcohol you it made understand the alcohol you it made not understand As for Basque, negation blocks clause union in participial complements of nahi ‘want’ and behar ‘need’: compare ikusi nahi ‘want to see’, with clause union effects and ez ikustea nahi ‘want not to see’, with the nominalized verbal form ikustea instead. 6 Zubizarreta (1985) stars causatives with embedded passives as in this example. As also apparent from the discussion of matrix passives, there are important dialectal differences at work here.

complex causative verbs and causees in basque (and romance)

283

only slightly marked (e), while progressive seems perfectly acceptable (d). No Basque counterparts can be formed for (10). Some differences between Spanish and French causatives on one hand and Italian on the other are brought forth in Zubizarreta (1985), who accounts for them claiming Italian fare forms are monoclausal structures while Spanish and French causatives are associated “simultaneously with two structures: one biclausal, the other monoclausal” (1985: 280). In essence, the difference would imply a more close knit verbal union in the case of Italian as opposed to French/Spanish. The difference can be observed in the availability of passivization of the causative verb itself. Zubizarreta’s examples below contrast the grammatical Italian (11a) and ungrammatical Spanish (11b): (11) a. Quei brani furono fatti leggere da Giovanni. those passages were made read by Giovanni ‘Those passages were made to be read (by Giovanni).’ b. *La casa fue hecha construir (por Casimiro) the house was made build by Casimiro ‘The house was made to be built (by Casimiro).’ The contrast is not found in Peninsular Spanish, where sentences like (b) are acceptable. In the analysis presented by Zubizarreta (1985: 274), the causative verb is a morphophonological word but acts as a bound morpheme from a morphosyntactic perspective, which brings apparently syntactic causatives closer to morphosyntactic ones, just as the facts in 1.1 bring the apparently morphological causatives closer to the syntactic ones. In any event, we would expect Basque to behave like Italian if, from a descriptive perspective, a monoclausal analysis with a morpheme-like causative verb fare is assumed to account for the Italian facts. This cannot be checked directly with passives, very marginal in Basque, but parallel facts can be found in ‘long impersonal’ constructions, similar to Spanish ‘reflexive passives’ or English middles, where the object of a verb becomes the subject of an intransitivized version of that verb. As already pointed out in Aissen and Perlmutter (1986), this process can sometimes apply across clauses, as a result of their proposed clause union or reduction process. The examples below illustrate this monoclausal-like effect in both Basque and Spanish using long mediopassives based on impersonal (causee-less) causative structures, to which we will return below in section 1.4:

284

ortiz de urbina

(12) a. Liburu hauek eskola guztietan irakurrarazten dira. book these school all.in read.cause aux ‘These books are made to be read in all schools.’ b. Se hacen pulir los diamantes en Holanda. se make polish the diamonds in Holland ‘Diamonds are made to be polished in Holland.’ The logical objects of read (‘those books’) and of polish (‘diamonds’) are now subjects of the causative (or of the complex verb), as shown by the fact that they agree with the latter, even though, in principle, they are not its arguments. 2.3 Direct and Indirect Causation in Tenseless Complement Causatives As indicated, Romance allows for tensed subjunctive clauses as complements of causative verbs, in sentences like (6b). Causatives with embedded subjunctive complements convey a less direct type of causation than those with infinitival complements, in accordance with Comrie’s (1989) generalization. A consequence, according to Bordelois (1974), Treviño (1994) or Tubino (2011), is that structures without animate causees in the infinitival complement are deviant. This is exemplified below with expletive subjects (13a), passive infinitive complements (13b) and dative experiencer constructions with psych verbs (13c): (13) a. *El mago hizo empezar a llover. the magician made begin to rain ‘The magician made it begin to rain.’ b. *La policía hace (a) la ley ser obedecida. the police makes (to) the law be obeyed ‘The police makes the law be obeyed.’ c. *Su ruido hace a las discotecas molestar a Pedro. their noise makes to the discos bother to Pedro ‘Their noise makes discos bother Pedro.’ Pragmatic considerations seem to be involved. Thus, Tubino (2011) shows that an animate causee in some constructions is acceptable with inanimate causers but less so with animate ones as shown in (14):

complex causative verbs and causees in basque (and romance)

285

(14) a. Su falta de cuidado / ? María le hizo ser descubierto. his lack of care Maria him made be discovered ‘His lack of care/María made him be discovered.’ b. Tu ironía / ? Pedro le hizo enfadarse a Juan. your irony Pedro him made get.mad to Juan ‘You irony/Pedro made Juan get mad.’ Conversely, the embedded expletive in (13a) improves in an impersonal context like (15), which indicates that no such causer exists: (15) No se puede hacer llover cuando se quiere. not se can make rain when se wants ‘One cannot make it rain when he wants.’ Regardless of the actual constraints that play a role in the more or less deviant status of some of the previous sentences, what is clear is that they disappear when tensed complement clauses are used, so that (16), for instance, is acceptable: (16) El ruido hace que a Pedro le molesten las discotecas. the noise makes that to Pedro him bother the discos ‘Their noise makes discos bother Pedro.’ The Basque causative with arazi ‘cause’ also expresses a fairly direct type of causation, less felicitous where pragmatic considerations do not make this type of causation possible. In this, then, it clearly patterns with Romance cause+infinitive structures, rather than with tensed embedded patterns. This is shown in (17) with stative complements of the causative verb, which Oyharçabal (2003) identifies as excluded from causative structures. The stative predicate in the following examples is paired with clausal subjects for the causative verb, forcing a non-agentive interpretation: (17) a. ??Txikitan jogurta asko jateak Mikel(i) altua izanarazi youth.in yoghourt much eat Mikel(dat) tall be.cause du(dio). aux ‘Eating yoghourt as a child made Mikel be tall.’

286

ortiz de urbina

b. ? Comer yogur de pequeño le ha hecho ser alto a Mikel. eat yog. as small him has made be tall to Mikel ‘Eating yoghourt as a child made Mikel be tall.’ c. Comer yogur de pequeño ha hecho que Mikel sea alto. that Mikel be tall (18) a.

??/* Inguru

eleanitz batean bizi izateak hizkuntza asko environment multilingual one.in live be language many jakinarazi dit.7 know.cause aux ‘Living in a multilingual environment has made me know many languages.’

(19) b. ? Vivir en un entorno multilingüe me ha hecho saber muchas live in an env. multilingual me have made know many lenguas. languages ‘Living in a multilingual environment has made me know many languages.’ c. Vivir en un entorno multilingüe ha hecho que sepa muchas that I.know many lenguas. languages The corresponding Spanish (b) examples sound less deviant than the Basque (a) examples, while the (c) sentences with tensed complements are grammatical in Spanish. Similarly, weather expressions are formed in Basque with existential constructions which cannot be causativized; alternatives resort to egin ‘do’ for the root verb: (20) a. Euria / hotza da. rain / cold is ‘It is raining/cold.’

7 A verb like jakinarazi ‘cause to know’ is acceptable in the meaning ‘communicate’, shared with Spanish, but not in the stative reading in (18).

complex causative verbs and causees in basque (and romance)

287

b. Ipar haizeak euria / hotza *izanaraziko / eginaraziko du.8 north wind.erg rain / cold be.cause.fut / do.cause.fut aux ‘The north wind will make it rain/be cold.’ In fact, izan ‘be’ is particularly uncommon with a causative verb regardless of whether it denotes states or activities, as noted in Oyharçabal (2003), although they seem acceptable in Spanish. The online dictionary of Euskaltzaindia (Royal Academy of the Basque Language), contains a handful of examples for this entry, all of which except for one (given here as 23) actually correspond to ‘have’:9 (21) a. *Medizinak logurea izanarazi dit.10 medicine sleepyness be.cause aux ‘The medicine made me feel sleepy.’ b. La medicina me hizo tener sueño. the medicine me made have sleep ‘The medicine made me feel sleepy.’ (22) a. Irakasleak ??arduratsua / arduratsuagoa izanarazi ninduen. teacher careful / more.careful be.cause aux ‘The teacher made me be careful/more careful.’ b. *Aitak abokatua izanarazi ninduen. father.erg lawyer be.cause aux ‘My father forced me to be a lawyer.’

8

9

10

In the case of hotz ‘cold’, egin ‘do’ itself is commonly used instead of izan ‘be’, so the causative eginarazi is not totally unexpected. It is in the case of euria, though. Whether because it is a recent use of egin or because it is not a light verb here, unlike in expressions like negar egin ‘cry, lit. make cry’, the fact is that this egin is never causativized as eragin. There is no tenseless form of the transitive auxiliary; the intransitive auxiliary izan ‘be’ is used instead. For texts up to the 20th century, the OEH gives three examples, two from Bible translations and a third one in a modern philosophical book in the context of gods ‘making be’, i.e., creating, without causee. The Ereduzko Prosa Gaur online corpus gives 13 instances from year 2000 to the present, all of them in translations or, two of them, in a philosophical work. None in the written press. At best, the sentence may be interpreted as the causative of the verb used in the substandard expression logure(a) dut ‘I have sleep’, with a transitive ‘have’ probably patterned after Spanish tener sueño ‘be sleepy (lit. ‘have sleep’)’ rather than the standard Basque intransitive ‘be’ of logure izan ‘be sleepy’ (lit. ‘be sleep-wanting’).

288

ortiz de urbina

c. El profesor me hizo ser cuidadoso / más cuidadoso. the teacher me made be careful / more careful ‘The teacher made me be careful/more careful.’ (23) haiek atzarriago eta kartsuago izanarazteko they smarter and more.active be.cause.to ‘… so that they are smarter and more active’ The incremental scale in the comparatives in (22) and (23) has the effect of making the Basque causatives of izan ‘be’ more acceptable. As for ‘inversion’ predicates with experiencer datives as in (13c), they are often states and, in any event, one need not expect the theme subject of such predicates to act as causee, but rather the experiencer. The theme causee (example (24a,b), in the intended interpretation) with gustar is as unacceptable in Basque as in Spanish, and the experiencer causee is slightly better (24c,d). Again, the que clause is perfect (25): (24) a. *Bere etxeko giroak musika gustarazi zion Mikeli. his house.of atmosphere music like.cause aux Mikel.dat ‘His house atmosphere made music be likeable to Mikel.’ b. *Su ambiente familiar le hizo a la música gustarle a his atm. familiar him made to the music like.him to Mikel. Mikel ‘His family atmosphere made music be likeable to Mikel.’ c. ??Bere etxeko giroak Mikeli musika gustarazi zion. his house.of atmosphere Mikel.dat music like.cause aux ‘His house atmosphere made Mikel like music.’ d. ? Su ambiente familiar le hizo a Mikel gustarle la música. his atm. familiar him made to Mikel like the music ‘His family atmosphere made Mikel like music.’ (25) Su ambiente familiar hizo que a Mikel le gustara la música. that to Mikel him like the music In spite of apparent differences depending on the degree of control of nonagentive, cause-like subjects of the causative verb, it seems clear that Basque

complex causative verbs and causees in basque (and romance)

289

arazi ‘cause’ causatives can be directly connected with Spanish and perhaps Romance make+infinitive causatives. 2.4 Causative, Impersonal and Passive Within Romance, a further split separates passive-like causatives like (26) from the active-like ones in (9) above: (26) a. Jean a fait prendre le médicament par son enfant. Jean has made take the medicine by his son ‘Jean has made his son take the medicine.’ b. Juan hizo recoger el paquete (por uno de sus empleados). Juan made pick the package by one of his employees ‘John had the package picked up (by one of his employees).’ (Tubino 2011:220)11 Examples like (b) are quite marked in Spanish, and are found at best in very formal contexts.12 Notice that this so-called passive causative does not occur with passive morphology on the embedded verb; the latter, as indicated above, is possible with an animate causee: (27) a. El juez me hizo ser defendido por un abogado de oficio. the judge me made be defended by a lawyer of office ‘The judge made me be defended by a public defender.’ b. El médico le ha hecho a Pedro ser operado de urgencia. the doctor him has made to Pedro be operated of urgency ‘The doctor made Pedro be operated on urgently.’ In contrast, the examples in (26) can have a by-phrase which does not co-occur with passive morphology. The presence of a passive agent adjunct like par son 11

12

English does not admit ‘subjectless’ complements of make, so (i) is ungrammatical. This is why a translation with have is given: (i) *The City Hall made repair the house Passive complements of make are of course possible, with overt passive morphology and intervening NP; in those cases, however, make usually requires a to-infinitive in standard dialects: (ii) The City Hall made the house to be repaired In this they resemble ‘reflexive passive’ impersonals with overt agentive phrases like Se firmó la paz por los embajadores.

290

ortiz de urbina

enfant ‘by his son’ is only one aspect of the characterization of these structures, and in fact these structures stand out as a separate type even if there is no overt agent expressed. The reason is that, crucially, the causee forced to act by the causer is not expressed as a core argument of the faire+verb unit, so these constructions are impersonal, causee-less in that sense (not necessarily agentless), as discussed in Zubizarreta (1985). Thus, the following example would also qualify as a ‘passive’ causative of this type: (28) El ayuntamiento hizo reparar la casa. the city.hall made repair the house ‘The City Hall had the house repaired.’ Since what is left unexpressed is the external argument (logical subject) of the embedded verb, impersonal ( faire-par) causative constructions subsume impersonal causatives of embedded intransitive verbs, where no by-phrase is of course possible. Some examples are given below with both unaccusative and unergatives verbs. Unaccusative verbs, ungrammatical in this construction in French, do seem to be less acceptable than unergatives in Spanish, but not necessarily ungrammatical: (29) a. ? El hielo hacía resbalarse en las aceras. the ice made slip in the sidewalk ‘The ice made [people] slip in the sidewalks.’ b. Demasiadas veces, las caravanas hacen llegar tarde al trabajo. too.many times the traffic.jams make arrive late to work ‘Too often, traffic jams make [people] arrive late to work.’ c. ??Un contexto familiar pobre hace nacer con muchas a context family poor make be.born with many desventajas. disadvantages ‘A poor family context make [one] be born with many disadvantages.’ (30) a. En ese caso, el reglamento hace jugar en el campo contrario in this case the regulation makes play in the field opposite (*por el equipo). by the team ‘In this case, the regulations make [teams] play in the other team’s field.’

complex causative verbs and causees in basque (and romance)

291

b. Esta dirección hace trabajar hasta muy tarde (*por el this management makes work until very late by the personal). staff ‘This management team makes [staff] work until very late.’ The constraints on unaccusative faire-par causatives in (29) seem to be a subcase of the more general difficulty in forming impersonal constructions for verbs without external arguments, as discussed first by Perlmutter (1978) in connection to Dutch impersonals. There are no causative constructions in Basque where an agent/causee may be recoverable as an adjunct, so constructions like (26) are not found.13 However, we can still draw a parallelism if impersonal causatives like (28) actually correspond to the same type. Similar impersonal causatives constructions are also available with arazi ‘cause’ in Basque. This is exemplified with embedded transitive (31), unaccusative (32) and unergative (33) verbs: (31) a. Zuzendariak txosten berri bat idatzarazi zuen. director report new one write.cause aux ‘The director made [someone] write a new report.’ b. Haurrak katua hilarazi du. child.erg cat die.cause aux ‘The child made [someone] kill the cat.’ (Oyharçabal 2003, ex. (2b))14 (32) a. ? Sarritan, pilaketek lanera berandu helarazten dute. often jams.erg work.to late arrive.cause aux ‘Often, traffic jams make one arrive late to work.’

13

14

Ortiz de Urbina (2003: 604) brings forth the following example as a possible candidate for this construction: (i) Mutilez zain-arazten zitian bere arthalde handiak boys.inst care-cause aux his flock large ‘He had his large flocks looked after by boys’ (Eliss. P.A.: 80) The instrumental is used for passive agents in the dialect in question; it is possible, however, to regard this as a normal impersonal causative with an instrumental reading of mutilez. Oyharçabal’s example (b) is ambiguous between the intended impersonal interpretation and an irrelevant interpretation where causative arazi is added to an inchoative unaccusative hil ‘die’. The causative version of an inchoative is possible, though less common than the lexical causative pattern (meaning kill). See Berro (2015).

292

ortiz de urbina

b. ? Pobreziak bizitza negargarri baten ondoren hilarazten du. poverty.erg life deplorable one.gen after die.cause aux ‘Poverty makes one die after an awful life.’ c. ??Sekta horrek eritetxeetan gabe, etxean jaioarazten du. sect that.erg hospitals.in without home.at be.born.cause aux ‘That sect makes [babies] be born in hospitals rather than at home.’ d. Alkoholak ezbeharrez ahantzarazten du. alcohol.erg mishaps.about forget.cause aux ‘Alcohol makes one forget about mishaps.’ (33) a. ? Sarritan, ikastetxeetan futbolean jokarazten dute. often schools.in football.at play.cause aux ‘They often make [students] play football in schools.’ b. Armadetan norberaren sinesmen kontra ere borrokarazten armies.in one.gen beliefs.gen against also fight.cause dute. aux ‘In armies they make [people] fight even against one’s own beliefs.’ c. Ordutegia hobetzeko, zazpietan afalarazi beharko lukete. timetable improve.to seven.at dine.cause need.fut aux ‘To improve timetables, they should make people have dinner at seven.’ As the examples show, causativized impersonal unaccusative verbs are quite marked in these constructions. They seem most easily constructed in generic contexts, with the help of genericity activating phrases as in (32), but even so the results are quite deviant with them if compared with unergative predicates like those in (33). There are differences between regular infinitival causatives and impersonal/ passive causatives (see Folli & Harley 2007 for Italian, adapted in Tubino 2011 for Spanish), related to the semantic role of the subject on one hand and passivization on the other. Beginning with thematic roles, causes (as opposed to agents) are claimed to be ruled out as subjects of impersonal causatives, but are grammatical with regular infinitival causatives. The distinction in the former is shown by examples like the following (Tubino 2011: 227):

complex causative verbs and causees in basque (and romance)

293

(34) Juan / * su enfado hizo castigar al niño (por el profesor). John / * his rage made punish to.the child (by the teacher) ‘John/*his rage had the child punished (by the teacher).’ Other pragmatic factors may be involved, since causes are acceptable in other sentences. In the following sentence the animate noun phrase could be interpreted as an agent or perhaps also as a cause, without apparent effect on the grammaticality of the sentence; the inanimate subject would be a cause: (35) El vino / los buenos amigos hace(n) olvidar las penas. det wine / det good friends make forget det pains ‘Wine/good friends make one forget one’s pains.’ Similarly, although most of the Basque acceptable impersonal causatives in (31) through (33) do have agent subjects, this does not seem like a necessary condition:15 (36) Berak / bere presentziak pena guztiak ahantzarazten ditu. (S)he his / her presence pain all forget.cause aux ‘He/his presence makes one forget all pains.’ This sentence seems acceptable regardless of the cause or agent interpretation of the subject. This area, therefore, requires further research, at least in the case of Basque.16 The second difference mentioned in Folli & Harley (2007) relates to the possibility of passivization in impersonal/ fair-par causatives. We have already seen examples where the embedded infinitival complement is passivized (see example (27)). However, the differences emerge when passivization of the causative verb itself is considered. This is only possible in impersonal causatives (a), but not in ‘personal’ ones (b): (37) a. Il pacchetto fu fatto arrivare (da Gianni). the package was made arrive by Gianni ‘The package was made to arrive (by Gianni).’

15 16

The unaccusative verbs in (32) have a mixture of causes and agents; changing the former to the latter does not have any effect on the acceptability of these sentences. With respect to direct and indirect causation, Oyharçabal (2003) points out, quoting Danlos (2001), that “causality in the real world resembles a chain at the end of which it is always possible to attach a further link”.

294

ortiz de urbina

b. *Maria fu fatta mandare un pacchetto (da Gianni). Maria was made send a package (by Gianni) ‘Maria was made to send a package (by Gianni).’ Passives of causative hacer have an uncertain status in Spanish (Torrego 1998; Treviño 1994; Tubino 2011), but, as the latter claims, there seems to be a difference so that passives in impersonal causatives (38), uncommon though they are, seem more acceptable than those in regular infinitival complements (39): (38) a. El palacio fue hecho construir por Carlos V. the palace was made build by Carlos V ‘The palace was made to be built by Carlos V.’ b. El edicto fue hecho publicar en todos los diarios. the ruling was made publish in all the newspapers ‘The ruling was made to be published in all the newspapers.’ c. El himno fue hecho cantar en todas las escuelas. the anthem was made sing in all the schools ‘The anthem was made sing in all schools.’ (39) a. *El escuadrón fue hecho andar / correr / marchar varios the squadron was made walk / run / march several kilómetros. kilometers ‘The squadron was made to walk/run/march several kilometers.’ b. Juan fue hecho venir por su hijo. Juan was made come by his son ‘Juan was made to come by his son.’ (Trevino 1994; Tubino 2011) c. *Juan fue hecho llegar tarde. Juan was made arrive late ‘Juan was made to arrive late.’ d. *Los estudiantes fueron hechos cantar en todas las escuelas. the students were made sing in all the schools ‘The students were made to sing in all schools.’

complex causative verbs and causees in basque (and romance)

295

e. *Juan fue hecho reparar el coche. Juan was made repair the car. f. *El coche fue hecho reparar a los mecánicos. the car was made repair to the technicians ‘The car was made to be repaired by the technicians.’ According to Tubino (2011), only internal arguments of the impersonal/ fairepar construction can become matrix passive subjects. This excludes (39f), where the presence of the causee indicates this is not an impersonal causative. Since the faire-par construction is characterized as impersonal in the sense of having no external subject argument, unaccusatives like (39b) are included in this type and expected to be acceptable, as Tubino (2011) claims for this example. However, other unaccusative subjects do not seem to share this behavior (39c). These details aside, though, there does seem to be an asymmetry in the behavior of the two constructions with respect to matrix passivization. Examples like (12) above already show that even if passives are not productive in Basque, mediopassive impersonal structures can be formed by making the embedded object subject of the matrix verb. We give more examples in (40) and (41) below. As (41) shows, mediopassives are just as easily acceptable when the causee is expressed as in the impersonal counterparts in (40): (40) a. Pena zaharrak pena berriekin ahantzarazten dira. pain old pain new.with forget.cause aux ‘Old pains are made to be forgotten with new pains.’ b. Errefuxiatuentzako kanpamenduak hiri bazterretan eraikiarazi refugees.for camps town margins.in build.cause ziren. aux ‘Refugee camps were made to be built in the town margins.’ (41) a. Kanpamenduak hiri bazterretan eraikiarazi zitzaizkien camps town margins.in build.cause aux langileei. workers.dat ‘Camps were made to be built by workers.’

296

ortiz de urbina

b. Ikasleak garaiz helarazi beharko lirateke eskolara. students on.time arrive.cause need.fut aux school.to ‘Students should be made to arrive to school on time.’ c. Intxaurrak makilez jausiarazten dira. walnuts sticks.with fall.cause aux ‘Walnuts are made to fall with sticks.’ No contrast between the personal and impersonal causative constructions therefore emerges in Basque mediopassive structures. The complex causative verb behaves in them like a regular transitive verb, and the mediopassive eliminates the causer, leaving it as an implicit argument and detransitivizes the structure. This contrasts with Spanish, where mediopassives based on impersonal/ faire-par seem far better than those based on objects of regular causatives.17 Plural nominals are used throughout these examples to check easily whether the causative verb agrees with them in a (medio)passive structure; the sentences would all be acceptable with singular agreement on hacer ‘cause’; this is, however, a different, impersonal structure: (42) a. Se hacían construir puentes en todas partes. se made build bridges in all sides ‘Bridges were made to be built everywhere.’ b. Esos diamantes se hacen pulir en Holanda. those diamonds se make polish in Holland ‘Those diamonds are made to be polished in Holland.’ (43) a. ? Se hicieron construir puentes a los obreros. se made build bridges to the workers ‘Bridges were made to be built by workers.’ b. ??Se hacen publicar novelas románticas a los escritores. se make publish novels romantic to the writers ‘Romantic novels are made to be published by writers.’

17

Notice that in all cases the embedded verb is in an active form, unlike in some of the translations; the causee is given as a by phrase as a way to make the intended meaning understood.

complex causative verbs and causees in basque (and romance)

297

c. Se hacen caer las nueces con palos. se make fall the walnuts with sticks ‘Walnuts are made to fall with sticks.’ d. Se hicieron llegar refuerzos por mar. se made arrive reinforcements by sea ‘Reinforcements were made to arrive by sea.’ e. ??En el circo se hacen aparecen conejos en los sombreros. in the circus se make appear rabbits in the hats ‘In the circus, they make rabbits appear inside hats.’ The last three examples contain unaccusative verbs, and although judgements are not so clear for all examples/speakers, their inanimate causees, which would appear as direct objects in the personal causatives, seem to be able to become subjects of the mediopassive construction. In any event, leaving the status of Spanish causativized unaccusatives as faire-par or impersonal causatives aside, the contrast obtained by the presence of the causee in (a,b) is clear and may indicate a different status/position for causees in Basque and Spanish causatives. Summarizing, in spite of the apparent differences in causativization strategies, if the data described here are correct, Basque and Spanish (as well as Romance) causatives with tenseless complements share important fine-grain features, as well as some differences. Accounting for them goes beyond the limits of the modest descriptive goal of this paper, and the complexity of acceptability patterns makes the enterprise more difficult. However, it seems we might be able to gain insights about both languages and about causative structures in general from the close examination of their grammars. In the following section we turn to an area where contact may be having a more distinct impact on the grammar of Basque, namely causee marking, with particular emphasis on agreement conflicts created in Basque as a result of dative doubling.

3

Causee Marking in Basque (and Spanish) Causatives

Whether formed in a strictly morphological way by adding a bound morpheme to a verbal base, through verb/clause union processes from what at first look like biclausal structures or by selection of VP or phrase (or even Root) complements for a Cause head as in Pylkkänen (2008), complex verb formation in causatives involves adding an extra argument to those of the event predi-

298

ortiz de urbina

cate. This can be effected in widely different ways depending on one’s theoretical persuasion, but the general layout is fairly clear. As discussed in Comrie (1989) and Dixon & Aikhenwald (2000), the causer argument of the causative verb/morpheme/head is aligned with the subject function, and the causee cannot be a subject unless a fully biclausal structure is formed. If this is not the case, typologically the most widespread situation is for intransitive event subjects to occur as objects of the complex causative verb and for transitive subjects to be found as indirect objects. The following subsections will focus on causee case in Basque causatives. In section 3.1 the Standard Basque pattern in presented, focusing on causee marking and case doubling phenomena. DOM— (Differential Object Marking) related dialectal or substandard phenomena will be discussed in section 3.2. 3.1 Increasing Arguments by One: Causee Marking Standard Basque presents a fairly typical situation for valency increasing via causative formation. We examine causativization of basic intransitive and monotransitive predicates in section 3.1.1, concentrating mostly on the appearance of dative causees. Section 3.1.2 is devoted to causativization of argument configurations which already have a dative participant, focusing on the interaction between dative causees and the original datives of those configurations. 3.1.1 Causativizing Intransitive and Monotransitive Verbs Causativized verbs are always transitive, since even in the case of intransitive verbs, the basic intransitive argument will be increased by one, the causer. The ‘object’ like argument of the unaccusative verb will no longer surface as a subject, but will remain object, so the resulting complex verb resembles a typical monotransitive in terms of case marking and auxiliary configuration: (44) a. Jon garaiz etorri da. Jon.abs on.time arrive aux ‘Jon arrived on time.’ b. Edurnek Jon garaiz etorriarazi du. Edurne.erg Jon.abs on.time arrive.cause aux ‘Edurne made Jon arrive on time.’ Unergative predicates with agentive subjects are primarily constructed with egin ‘do’ as a light verb, taking a bare object indicating the action performed: lo egin ‘sleep, lit. do sleep’, lan egin ‘work, lit. do work’, etc. The connection between unergativity and Basque egin constructions was first discussed in

complex causative verbs and causees in basque (and romance)

299

Levin (1983), and since then many alternative analyses have been proposed (see Laka 1993; Etxepare 2003 or Berro 2015, among many others). Levin also noticed that some unergative predicates also (or exclusively) exist as simplex verbs, taking then ergative subjects and transitive auxiliaries: jolastu ‘play’, saltatu ‘jump’, distiratu ‘glitter’, iraun ‘last’, etc. There is more variation for the causative of these verbs. De Rijk (2008: 378–380), from where (45b) is taken, states than in northern dialects they tend to behave like unaccusatives (a), although occasionally now, and more often in the past, dative causees were also found. In the central dialects (Gipuzkoan mostly, since arazi ‘cause’ gives way to eragin, arazo ‘cause’ further to the west), unergative causees tend to be marked dative, behaving then in causatives just like the ergative subjects of transitive verbs (b): (45) a. Eguzkiak Sacre Coeur basilikaren kupula distirarazten zuen. sun.erg basilica.of dome glitter.cause aux ‘The sun made the Sacre Coeur basilica dome glitter.’ b. Norbaitek edo zerbaitek iraunarazi dio hizkuntzari. someone.erg or something.erg last.cause aux language.dat ‘Someone or something made the language last.’ As also pointed by de Rijk, the normative advice is to use datives with animate causees and absolutive with inanimate ones, introducing a DOM-like distinction to which we will return in 3.2. Both unaccusative and unergative verbs may take yet a second argument in the dative; these will be discussed in the following section, once dative causees are introduced for regular monotransitive verbs, to which we turn now. The external argument of a transitive verb appears as a dative causee in the causative version of that verb, as shown in (46a): (46) a. Irakasleak ikasleei liburu hori irakurrarazi die. teacher.erg students.dat book that read.cause aux ‘The teacher made the students read that book.’ b. Irakasleak ikaslei liburua irakurri die. teacher.erg students.dat book.the read aux ‘The teacher read the book to the students.’ This pattern is general. The embedded verb already includes a direct object and the logical subject of irakur- ‘read’ cannot appear as a direct object, given that

300

ortiz de urbina

direct object doubling is not found in Basque. The resulting configuration is then identical to that of regular ditransitive predicates like eman ‘give’ or the ditransitive use of irakurri ‘itself’ (46b). 3.1.2 Causativizing Verb Configurations Including Datives Each of the three previous complement configurations exists in another version including a dative. Ergative/dative configurations, absolutive/dative and, finally, ergative/absolutive/dative ditransitive patterns will be examined in turn. 3.1.2.1 Causativizing Basic Ergative/Dative Configurations Beginning with ergative/dative complementation types, some unergatives verbs take dative objects (Ortiz de Urbina & Fernández 2016). One such verb is begiratu taking complements in the dative and, occasionally for locative nouns, in the allative. As the following examples show, when causativized, the causee is also marked dative. The basic pattern is exemplified in (47a), while the causative version is found in (47b,c): (47) a. Jonek berari begiratu zion. Jon.erg he.dat look.at aux ‘Jon looked at him.’ b. Lurrera begirarazi zioten (berari). floor.to look.at.cause aux he.dat ‘(They) made him look down to the floor.’ c. Atzamarrez kokotsetik heldu eta niri begirarazi fingers.with chin.by grab and I.dat look.at.cause nion. 1erg.3dat.pst ‘I grabbed his chin and made him look at me.’ The (b) example illustrates the causative of the basic unergative configuration of this verb, with an allative complement; as described in the previous section, the causee appears in the dative. When the verb takes a dative complement, a dative doubling situation emerges, exemplified here in (c). The auxiliary gloss has been given in more detail to point out that when two datives vie for agreement, it is the causee that becomes cross-referenced; in this example dative agreement is with the unexpressed berari ‘to him’, while the goal niri ‘to me’ does not trigger any marker on the auxiliary. If we eliminate the causee by using

complex causative verbs and causees in basque (and romance)

301

an impersonal causative, the original dative is still unable to agree, so that in the following examples the dative auxiliary is flagged as ungrammatical. If used, the dative is interpreted as referencing an unexpressed causee: (48) Eskola honetan irakasleari begiratu / entzunarazten dute / school this.in teacher.dat look.at hear.cause aux / *diote. aux.3sg.dat ‘In this school they make [one] look at/listen to the teacher.’ (49) Hatsarre moral hauek eskubiderik gabeko jendeari lagunarazten principle moral these right without people.dat help.cause dute / *diote. aux / aux ‘These moral principles help people without rights.’ The dativeless auxiliary in sentences of this type is exceptional in Standard Basque, where datives must be cross-marked in the inflection, and these sentences are slightly dubious, perhaps the result of some repair strategy. However, the version with the dative auxiliary is clearly ungrammatical. Similar facts can be observed in Spanish, although in that case it is the position of clitics that is significant: (50) En esta escuela (*les) hacen llamar(les) a los padres in that school them.dat make look.at.them.dat to.the parents por teléfono. by phone ‘In this school they make (people, teachers) phone parents.’ As indicated, the clitic referring to the indirect object of llamar ‘call’ must remain attached to the lower verb, and may not climb to the matrix causative verb, as if ‘clause union’ had not taken place (Aissen & Perlmutter 1983). 3.1.2.2

Causativizing Absolutive/(Experiencer, Goal, Ethical) Dative Configurations. On top of the basic unaccusative configuration, with intransitive auxiliary and absolutive subject, some verbs display a bivalent pattern with an added dative participant. The dative argument may be both an optional goal with verbs like joan ‘go’, hurbildu ‘approach’ or an obligatory experiencer argument in an ‘inversion’ predicate like gustatu ‘like’, iruditu ‘seem’ or interesatu ‘interest’. Ethi-

302

ortiz de urbina

cal/interest datives may also be found with most verbal configurations and will be addressed briefly below. Beginning with inversion predicates, we already mentioned that the dative argument can only marginally be taken to be a causee, as in (24c), repeated here as (51a); attested examples are given in (51b,c): (51) a. ? Bere etxeko giroak Mikeli musika gustarazi zion. his house.of atmosphere Mikel.dat music like.cause aux ‘His family background made Mikel like music.’ b. Behar da [poesia] publikoari gustarazi. need aux poetry public.dat like.cause ‘One needs to make the public like poetry.’ c. Arren gustaraz diezadazu … kausitzen dena. please like.cause aux find aux.comp.det ‘Please, make me like what is found …’ (Arbill 43, III 15, 22) Dative is expected here for the causee, since there is already an absolutive element, and this case also fits its experiencer status. In order to check for dative doubling configurations, we would need an example where the theme is interpreted as a causee, a situation which, as indicated in 1.3 above, is usually unacceptable in Spanish with infinitival causatives. The type of interpretation sought is similar to that in English (52), where the intervening NP would correspond to the causee of the structures in question in Basque and Spanish: (52) They made the teacher be liked by the students. There is no thematic link between make and the following NP in English, while the causee does have one with the complex verb in both Spanish and Basque; these sentences are unacceptable:18 18

There are nonetheless occasional attested examples which may correspond to this interpretation, without an auxiliary: (i) [ikasleei] hizkuntza hurreratu behar zaie, gustarazi students.dat language draw need aux like.cause ‘language must be brought over to the students, made be likeable’ The interpretation seems to be that something must be done to make the language more appealing to students, rather than simply making the students like the language, and to the extent that this is a possible (and different) interpretation, it would perhaps be an (isolated) example of theme ‘causee’.

complex causative verbs and causees in basque (and romance)

(53) a.

303

??/*Hoy

en día al profesor le hacen gustarles a los today in day to.the teacher him make like.them to the estudiantes. students ‘Now a days they force the teacher to be liked by students.’

b. *Gaur egun irakaslea ikasleei gustarazten diete.19 today day teacher.abs students.dat like.cause aux.3pl.dat ‘Now a days, the force the teacher to be liked by students.’ The Spanish sentence is interpretable (with some effort) thanks to the two clitics and their placement. The single agreement marker in the Basque examples makes interpretability more complicated. The second type of ‘bivalent unaccusative’ is that where a goal dative is added to an intransitive verb (54a).20 While the directional complement is more commonly expressed in the allative, the dative exists as a formal register possibility. Given ergative marking, the absolutive intransitive subject remains absolutive, now as object of the transitive causative verb. The combination with the ergative causer and the dative goal would produce a pattern identical to that of ditransitive predicates (54b); however, the structure is considered deviant by speakers: (54) a. Jon lagunei hurbildu zaie. Jon.abs friends.dat approach aux ‘Jon approached his friends.’ b. *Beharrak Jon lagunei hurbilarazi die. need.erg Jon.abs friends.dat approach.cause aux.3pl.dat ‘Need made Jon approach his friends.’ c. *Mirenek sua Mikeli hurbilarazi zion. Miren.erg fire.abs Mikel.dat approach.cause aux.3sg.dat ‘Miren brought the fire close to Mikel.’ Since the basic non-causative pattern is rather literary, judgments may be uncertain; but agreement with the dative seems to be the problem here: not 19 20

Using a dativeless transitive auxiliary like dute does not improve the sentence. See Rezac (2009), who follows Ormazabal & Romero’s (2007) derivational approach for Double Object Constructions, deriving goal datives from PP s.

304

ortiz de urbina

all causees are marked dative in the standard language, but all datives seem to be interpreted as causees, and this is not the interpretation in (54b).21 In (54c) an inanimate causee is intended, with a goal dative; the sentence is only interpretable as the causative of a monotransitive use of hurbildu, again with a causee dative. An indication that unacceptability arises from the unintended connection between dative and causee is that if dative agreement is eliminated, the sentence is perfectly interpretable and acceptable for many speakers in the intended sense. The connection between dative and causee is also made evident by looking at impersonal, causee-less structures. Again, their dative is not interpreted as goal but as causee, so (55) is unacceptable in the intended interpretation: (55) Beharrak lagunei hurbilarazten *die / need.erg friends.dat approach.cause aux.3erg.3pl.dat / du. aux.3erg ‘Need makes [one] approach friends.’ More specifically, as discussed above, the agreement dative corresponds to the causee. The sentence becomes acceptable if dative agreement is eliminated (the auxiliary du only includes information about the subject and an unspecified object). As discussed in the following section, this sentence is acceptable in the western varieties that mark intransitive subject causees as dative. In those varieties, this sentence is the causative of a basic monoargumental unaccusative, with lagunei ‘to the friends’ interpreted either as a causee coreferential with the dative in the inflection (‘make the friend approach’) or as a goal. In this case, however, the dative in the inflection must be understood as marking an unexpressed causee. The connection may be relaxed for some speakers, so they find (56) acceptable: (56) Beharrak Mikel diru eske etorrarazi zidan. need.erg Mikel money asking approach.cause aux.1dat ‘Need made Mikel come to me asking for money.’ (Ortiz de Urbina 2003) The scant use of the basic goal dative configuration in non-formal contexts only adds to the complexity of factors which may help us understand speaker judgement variability in these cases. 21

This applies to argumental datives. Ethical datives can be found in causative structures without receiving a causee interpretation. We return to them below.

complex causative verbs and causees in basque (and romance)

305

To finish with causativized absolutive/dative basic configurations, it may be worthwhile considering, at least in passing, ethical datives, typically indicating external possession (alienable or inalienable) or interest relations.22 The most widespread type of ethical dative, geographically and diachronically, is that where the interest relation is predicated of an absolutive element, whether the intransitive subject (57a) or transitive object (57b); there is more variation in acceptability of a dative associated with an ergative (57c,d):23 (57) a. Umea erori zait. child fall aux.1dat ‘The child has fallen (on me).’ b. Umea eraman didate. child take.away aux.1dat ‘They have taken (my) child away.’ c. Umeak ondo borrokatu dit. son.erg well fight aux.1dat ‘My son has fought well (on me).’ d. Jonek indabak gorrotatzen dizkit. Jon.erg beans hate aux.1dat ‘(My) Jon hates beans.’ (Fernández & Ortiz de Urbina 2010: 181) Albizu (2009) reports dialectal variation as to the possibility of associating ethical datives with absolutive causees, as in (58a,b). Trying to associate ethical datives to the absolutive object of a transitive (58c) or the causee unergative subject of (58d) is impossible. Dative doubling results in these cases, with a

22

23

See Fernández (this volume) for a comparison between Spanish and Basque ethical and other non-selected datives. See also Fernández & Ortiz de Urbina (2010: 179) for Basque ethical and interest/possessive datives. Interest datives associated with ergatives are felt, right or wrongly, to show the influence of Spanish on Basque. The association with the absolutive would be expected in a treatment of ethical datives as low applicative heads, as in Pylkkänen (2008), where the dative introduces a source, goal, benefactive, or, in this case, interested argument related to the object. This approach is taken in Oyharçabal (2007) and Albizu (2009), among others. In the case of (d), even association with the absolutive object (‘my beans’) may be excluded in eastern dialects, as ethical datives are excluded from stative contexts (Oyharçabal 2007).

306

ortiz de urbina

priority for agreement with the cause excluding the ethical marking, in ways similar to what has been noted above for dative causees; we return to similar cases in the following section: (58) a. Solairuaren ezkoak umea eroriarazi dit. floor.of wax.erg kid.abs fall.cause aux.1dat ‘The floor wax made my kid fall.’ b. Ilargi beteak semea lehenago jaioarazi digu. moon full.erg son.abs earlier be.born.cause aux.1dat ‘The full moon has made our son to be born earlier.’ c. *Liburua Mikeli harrarazi didate. book Mikel.dat take.cause aux.1dat ‘They have made Mikel take (my) book.’ d. *Semeari / semea ondo borrokarazi didate. son.dat son.abs well fight.cause aux.1dat ‘They have made my son fight well.’ Although variability is rampant and the situation for each dialectal area must be checked, it seems that we can arrive at the generalization that the examples clearly ruled out are those where ethical dative agreement on the verb is given priority over agreement with a dative causee. This is the case in (58c) and (58d), in dialects where the unergative subject surfaces as a dative cause. Those examples which are acceptable for some speakers and dubious for others (58a,b and 58d with absolutive causee) are those where no such conflict arises; the variability in the acceptability of (58a,b) would then be unrelated to these agreement conflict considerations.24 3.1.2.3 Causativizing Ditransitive Verbs Let’s turn finally to causatives of ditransitive uses of verbs. Since the ditransitive verb also contains a dative argument, and a transitive subject is causativized as a dative causee, a dative doubling situation emerges. Such sentences are acceptable to most speakers provided the association of the dative nominals 24

Albizu (2009) provides a technical analysis in the line of Pylkkänen (2008) and Rezac (2009). In western dialects with DOM effects in causatives, all causees in (58) will be marked dative, as discussed in the following section. See footnote 28 for some comments on the dative doubling situation created by ethical datives.

complex causative verbs and causees in basque (and romance)

307

with a causee and a goal is possible; if not the result is probably more unacceptable because of the impossibility of disambiguation than ungrammatical: (59) Mikeli liburua emanarazi didate. Mikel.dat book give.cause aux.1dat ‘The made me give the book to Mikel.’ Disambiguation is achieved here through agreement, where the dative marker must correspond to the causee. Where the causee and goal have the same grammatical person (a), or in a tenseless context without agreement (b,c), acceptability decreases along with disambiguation ease: (60) a. Lagunek Mikeli Joni liburua emanarazi zioten. friends Mikel.dat Jon.dat book give.cause aux.3dat ‘Their friends made Mikel give a book to Jon.’ b. Niri Mikeli liburua emanaraztea ez da ideia ona. I.dat Mikel.dat book give.cause.nom no is idea good ‘Making me give a book to Mikel is not a good idea.’ c. Mikeli Joni liburua emanaraztea ez zen ideia ona. Mikel.dat Jon.dat book give.cause.nom no is idea good ‘Making Mikel give a book to Mikel is not a good idea.’ In a sufficiently clear context all these sentences would probably be judged acceptable; on top of contextual information, parsing seems easier if the first dative is interpreted as causee and the second one as a goal, especially if a pause separates them and the second one forms an intonational unit with the causative verb. The causee is also more salient in that most speakers prefer as a first interpretation of an example with focalized dative like (61), one in which the latter corresponds to the causee: (61) Mikeli emanarazi diote liburua Joni. Mikel.dat give.cause aux.3dat liburua Jon.dat ‘It is Mikel that they made give the book to Jon.’ Similar acceptability issues are found in the dative doubling examples (47c), (48) and (49) examined above, where the causee dative corresponds to the logical subject of an ergative verb.

308

ortiz de urbina

As discussed with regard to the (lack of) interaction between ethical datives and causee datives in unaccusative predicates, any other dative preempts the use of ethical datives. In a ditransitive sentence, in particular, the dative may not be interpreted as ethical. Thus, in the basic ditransitive (62) the dative may be interpreted as an ethical dative only if the verb eraman ‘take’ is used as a monotransitive one: (62) Mikeli semea eraman diote. Mikel.dat son take aux.3dat ‘They took the son to Mikel/They took away Mikel’s son.’ If we try to add an ethical dative, corresponding to an interpretation in which Mikel is the goal of taking and I am the interested participant or viceversa, the result is ungrammatical, as shown in (63): (63) *Semea Mikeli eraman didate. son Mikel.dat take aux.1dat ‘They have taken my son to Mikel/They have taken Mikel’s son to me.’ Notice, however, that even if speakers consider the sentence ungrammatical, they can assign it an interpretation only if the agreement corresponds to the ethical dative. Where no auxiliary is found, as in the nominalized clause in (64), ungrammaticality remains, as a result of the incompatibility between goal and ethical datives.25 (64) *niri semea Mikeli eramatea … I.dat son Mikel.dat take.nom.abs ‘taking my son to Mikel/taking Mikel’s son to me …’ We can now introduce dative causees into the picture. A sentence like (65) with a single overt dative and a verb like eraman ‘take’, which can have both monotransitive and ditransitive uses, is several ways ambiguous: (65) Mikeli semea eramanaraztea ez zen gauza ona. Mikel.dat son take.cause.nom.abs not was thing good ‘Making Mikel take the son/Making someone take the son to Mikel/Making someone take Mikel’s son was not a good thing.’ 25

Speakers can be made to understand the meaning of this clause if the first dative corresponds to the interested participant and the dative closer to the verb corresponds to the goal.

complex causative verbs and causees in basque (and romance)

309

The first and most prominent interpretation is one where the dative is interpreted as a causee, the agent of a monotransitive use of eraman ‘take’. A second interpretation, more formal, corresponds to an impersonal causative of ditransitive eraman ‘take’ where the dative is interpreted as a goal, i.e., making some unspecified causee take the son to Mikel. The third interpretation is one where the dative is interpreted as an ethical dative, and again, this is available only if there is no causee, i.e., making some unspecified agent take Mikel’s son. If we try to include two of these datives in the same sentence the result is quite complex: (66) Niri semea Mikeli eramanaraztea ez zen gauza ona. I.dat son Mikel.dat take.cause.nom.abs not was thing good ‘Making me take the son to Mikel was not a good thing.’ Again, this can be considered grammatical, if difficult to parse in the absence of inflection. The primary interpretation is for datives to be causee and goal, i.e., making me take the son to Mikel. The inverse interpretation (making Mikel take the son to me) is more difficult to obtain, as one of the interpretive strategies is to associate the dative closer to the verb with an internal argument of that verb (here goal). The causee and ethical interpretation for the datives is definitely more remote, whether it be making me take Mikel’s son or making Mikel take my son. Finally, interpreting the datives as ethical and goal is also remote, and obviously only feasible if the possibility of linking a dative with the causee is eliminated by giving the clause an impersonal causative interpretation. Speakers can be made to understand the meaning, but again only if the dative closer to the verb is interpreted as the goal (i.e. making someone take my son to Mikel). If the subject clause in (66) is turned into a tensed one with agreeing auxiliary, the agreement priorities mentioned above help rule out some of the meanings: (67) Niri semea Mikeli eramanarazi zidaten / zioten. I.dat son Mikel.dat take.cause aux.1dat / aux.3dat ‘They made me/Mikeli take the son.’ In both cases, the agreeing dative is interpreted as causee. The other dative may be interpreted as goal or ethical, perhaps with different likelihood depending on ordering alternatives which remain to be studied.

310

ortiz de urbina

3.2 DOM in Dialectal Patterns In western varieties, and to a lesser extent in central ones, eragin ‘cause’ and, less commonly, erazo ‘cause’ are the causative verbs, with some differences with respect to arazi. One difference has already been mentioned in section 1, namely, the fact that eragin combines with the participial form of the embedded verb, as opposed to arazi, which combines with the verbal root. Mixed patterns exist in central areas, combining arazi with the participle.26 A similarly distributed difference may be observed in causee marking. According to de Rijk (2008), eragin takes dative causees regardless of transitivity. This may have been the situation in earlier forms of the language, but at present, a differential object marking (DOM) pattern seems to have spread in the western and central varieties (roughly, those south of the France/Spain border). DOM in causatives may be a specific case of this general DOM pattern found in contemporary dialects south of the border (see for instance Fernández & Rezac 2016 or Odria this volume), less conspicuous and, perhaps as a result of that, less stigmatized. In the general pattern, similarly to Spanish, animate, specific direct objects, especially 1st and 2nd person for some speakers, are marked differently from other objects and identically to indirect objects; in the case of Spanish both bear the preposition a, while in Basque both are marked dative. As indicated, while the pattern is colloquial and stigmatized in the case of regular transitive constructions, it has gained more general ground in causatives, and can be observed both with eragin and with arazi in these dialects. We give some examples below with both; the first two examples contain dative animate causees, while the third one contains the absolutive inanimate one, identical to the standard: (68) a. Bide pendiz bat artu eragin eta goraiño joan eragin eutsan. path steep one take cause and up go cause aux.3dat ‘[S/he] made him take a steep path and go all the way up.’ (Aug. Zubikarai)

26

In one of the earliest attested occurrences of eragin, from the second half of the 16th century, this causative actually appears not only with a participial complement verb, but even in a tensed form, suggesting it is not fused with the embedded verb, in a structure similar to English make: (i) Azaiteau galdu lerait / ardura usatzakeak. nourishment lose cause often use.without.erg ‘Not using it often would make me lose this nourishment.’ (Lazarraga 1196v) The causee is dative, again possibly as an argument of the causative verb itself.

complex causative verbs and causees in basque (and romance)

311

b. Aurreko gaira itzuli eragin zion solaskideari. before.of topic.to return cause aux.3dat speaker.dat ‘[S/he] made his conversation partner return to the previous topic.’ (J. Landa, Odolbildua) c. Daukien bildurra desagertu eragin gura dogu. have.which fear.abs disappear cause nahi aux ‘We want to make the fear they have disappear.’ (Larrabetzu city info) (69) a. Eguraldi txarrak berriro portura itzularazi zion. weather bad.erg again port.to return.cause aux.3dat ‘The bad weather made him return to port again.’ (Wikipedia, Francis Drake) b. Komisaldegira joanarazi zieten bi gazteri. police.station.to go.cause aux.3pl.dat two youth.dat ‘[They] made the two young persons return to the police station.’ (Argia 3/1/2017) c. Txetxenian homosexualak hil eta desagerrarazi Chechnia.in gay kill and disappear.cause dituztela. aux.3pl.abs.that ‘… that in Chechnia gay people have been killed and made disappear.’ (Hiruka 4/7/2017) The (a) and (b) examples above contain animate causees which would have been marked absolutive in the standard language, as described in 3.1.2.1 above, marked dative in published texts. The (c) examples present the types of causees which, being inanimate or non-specific, would usually take absolutive case also in these dialects.27 This dative interacts and overlaps with other datives in ways which may but need not be similar to those described above for multiple dative situations, but, like DOM patterns in general, this is an understudied area in Basque linguistics.28 27

28

Fewer examples are given of the standard-like pattern precisely because it is the normative one, found in more monitored texts in the western and central area. The qualification ‘types of animate causees’ in the text is introduced because examples with inanimate absolutive causees should belong to the same author and, preferably, text, as those with animate causees, something which has not been attempted here. For instance, double datives will also emerge with unaccusative absolutive dative configurations:

312

ortiz de urbina

The development of DOM patterns in dialects of Basque in contact with Spanish may have been helped by the existence of dative causee patterns with unaccusative subjects with causative verbs like eragin. However, this could only account for the appearance of new dative causees, but not for the maintenance of absolutive causees with inanimate or indefinite causees. The correspondence of the DOM isogloss with the area of contact with Spanish, and the maintenance of the traditional pattern in the contact area with French, which does not display DOM effects, clearly points at convergence with the former; in fact, the pattern is stigmatized by the normative tradition and felt to be a Spanish calque. The relation with Spanish is, however, more complex than this indicates. This is so because, as Odria (this volume) points out, the Spanish variety of the Basque Country treats animate and inanimate objects differentially not only in the use of the preposition a, but also in the clitic system, as described by Landa (1995) and Fernández-Ordoñez (1999), for instance. While dative case can be associated with the prepositional marking, clitics are more connected with inflection, so this ‘objective conjugation’ enables us to see object marking in the verb in a way parallel to Basque agreement. In Basque Spanish, animate direct objects, whether masculine or feminine, are associated with the clitic le. This is etymologically the dative clitic, which in Castilian Spanish has been extended to (mostly singular) masculine personal direct objects, and in Basque Spanish has been further extended to feminine personal direct objects, as in (70): (70) a. Lei vi a Maríai / Pedroi en la calle. him/her saw to María / Pedro in the street ‘I saw Maria/Pedro in the street.’ b. La / lo vi en la calle. it.fem / it.masc saw in the street ‘I saw it in the street.’

(i)

Joni niri itzuli eragin eutsan. Jon.dat I.dat return cause aux.3dat ‘[S/he] made Jon return to me.’ As with doubling in bivalent unergative verbs, agreement must correspond to the causee rather than the goal. Causees also win out over ethical datives for agreement, so that an ethical dative would be ruled out in (ii): (ii) Semiari gerrara joan eragin eutsan / *eustan. son.dat war.to go cause aux.3dat / aux.1dat ‘S/he made my son to go war.’

complex causative verbs and causees in basque (and romance) table 9.1

313

Adapted from Fernández Ordóñez 1999

Masculine Feminine +Animate -Animate +Animate -Animate Direct object Indirect object

le le

lo le

le le

la le

The clitic le in (a) can refer both to a feminine and masculine direct object, while the standard feminine and masculine direct object clitics la/lo in (b) can only refer to inanimate objects with that gender. The resulting system is illustrated in Table 9.1, adapted from Fernández-Ordóñez (1999); for simplicity, only the singular forms are given. A result of this clitic/agreement system is that animate direct objects are marked in the same way as indirect objects,29 a distribution essentially identical to that of the preposition a with objects.30 The presence of DOM patterns in both preposition marking and clitic form in Spanish compounds the pressure of DOM marking on the Basque system, perhaps helping to account for the causative DOM patterns discussed above. There is a further aspect of DOM which highlights the degree of convergence with Spanish. As noted earlier in this section, DOM in Basque is more widespread and mainstream in causatives than in basic direct object marking, where it is largely stigmatized. Similarly, le as a feminine direct object clitic, although apparently spreading, is largely restricted to colloquial speech and lower socio-economic status (Fernández-Ordóñez 1999). However, DOM in Spanish causatives is common even in educated speech in the Spanish of the Basque Country and neighboring areas, so that the patterns in (71) are standard: (71) a. Lei hice venir a Maríai / Pedroi. him/her made come Maria / Pedro ‘I made María/Pedro come.’

29

30

In addition, as Odria (this volume) points out, DOM contexts are often sensitive to specificity, and both the appearance of the clitic and of the preposition a are more likely to occur with definite, known, direct objects. It is doubtful that the emergence of this clitic pattern in Basque Spanish may in turn be due to Basque influence. Animacy does play a role in the shape of locative cases, but these facts are quite different from DOM.

314

ortiz de urbina

b. La / lo hice caer. it.fem / it.masc made fall ‘I made it fall.’ The feminine and masculine direct object clitic in (b) are preferably interpreted as inanimate. Extensive bilingualism and the growing impact of ‘new speakers’, bilingual speakers with Spanish as the dominant language who have acquired/learned the language in a school environment (Ortega et al. 2015) are most likely having a deep impact on Basque. The complex interactions among the systems of different varieties of Basque and between Basque and the surrounding Romance languages, as also seen in other contributions to this volume, make of this contact a promising ground for linguistic research.

Acknowledgements Research for this paper was partially funded by the Spanish MINECO project FFI2014–51878-P. Special thanks to Ane Berro and Beatriz Fernández for their patience and comments on this paper. Examples taken from the General Basque Dictionary (OEH) have been left with the code given there identifying their source.

References Aissen, Judith & David Perlmutter. 1983 [1976]. “Clause Reduction in Spanish”. In Studies in Relational Grammar. Chicago: Chicago University Press. 360–403. Albizu, Pablo. 2009 “Construcciones inacusativas con dativos posesivos y dativos de interés en vasco: un análisis derivacionalista”. Talk presented at the Seminario de Lingüística Teórica, CSIC, Madrid, May 11, 2009. Baker, Mark. Incorporation: A Theory of Grammatical Function Changing. Chicago: University of Chicago Press. Berro, Ane. 2015. From Event Structure to Syntactic Categories in Basque. Ph.D. dissertation, University of the Basque Country-University Bordeaux Montaigne. Bordelois, Ivonne. 1974. The Grammar of Spanish Causative Complements. Ph.D. dissertation, M.I.T. Comrie, Bernard. 1989. Language Universals and Linguistic Typology. Chicago: Chicago University Press. De Rijk, Rudolf P.G. 2008. Standard Basque. A Progressive Grammar. Cambridge, MA: MIT Press.

complex causative verbs and causees in basque (and romance)

315

Dixon, Richard M.W. 2000. “A Typology of Causatives: Form, Meaning and Syntax”. In Richard M.W. Dixon & Alexandra Y. Aikhenwald (eds.), Changing valency. Case studies in transitivity. Cambridge: Cambridge University Press, 30–83. Dunlos, Laurence. 2001. “Event coreference in causal discourses”. In Pierrette Bouillon & Federica Busa (eds.), The Language of Word Meaning. Cambridge: Cambridge University Press. DHM. 1989. [Deustuko Hizkuntzalaritza Mintegia: Andolin Eguzkitza, Javier Ormazabal, Jon Ortiz de Urbina, Koldo Sainz & Myriam Uribeetxebarria] “Inkorporazioa perpaus kausatiboetan” [Incorporation in causative clauses]. In Pello Salaburu (ed.), Sintaxi Teoria eta Euskara. San Sebastián: EHU Argitarapen Zerbitzua. 87–108. Etxepare, Ricardo. 2003. “Valency and argument structure”. In José Ignacio Hualde & Jon Ortiz de Urbina (eds.), A Grammar of Basque. Berlin/New York: Mouton De Gruyter. 362–426. Fernández, Beatriz & Milan Rezac. 2016. “Differential Object Marking in Basque Varieties”. In Beatriz Fernández & Jon Ortiz de Urbina (eds.), Microparameters in the Grammar of Basque. Amsterdam: John Benjamins. 93–138. Fernández, Beatriz. This volume. “On non-selected datives: Ethical datives in Basque and Spanish”. Fernández-Ordóñez, Inés. 1999. “Leísmo, laísmo y loísmo”. In Ignacio Bosque & Violeta Demonte (eds.), Gramática Descriptiva de la Lengua Española. Madrid: Espasa Calpe, 1317–1397. Folli, Raffaella & Heidi Harley. 2007. “Causation, obligation, and argument structure: On the nature of little v”. Linguistic Inquiry 38(2). 197–238. Laka, Itziar. 1993. “Unergatives that Assign Ergative, Unaccusatives that Assign Accusative”. In Jonathan D. Bobaljik & Colin Phillips (eds.), Papers on Case and Agreement I, MITWPL 18. Cambridge, MA. 149–172. Landa, Alazne. 1995. Conditions on Null Objects in Basque Spanish and their Relation to leismo and Clitic Doubling. Ph.D. dissertation, Univesity of Southern California. Levin, Beth. 1983. On the Nature of Ergativity. Ph.D. dissertation, MIT Odria, Ane. This volume. “Differential Object Marking in Basque and Spanish Dialects”. OEH [Orotariko Euskal Hiztegia-General Basque Dictionary]. Available online at www .euskaltzaindia.eus/oeh. Ormazabal, Javier & Juan Romero. 2007. “The Object Agreement Constraint”. Natural Language and Linguistic Theory 25. 315–347. Ortega, Ane, Jacqueline Urla, Estibaliz Amorrortu, Jone Goirigolzarri & Belen Uranga. 2015. “Linguistic identity among new speakers of Basque”. International Journal of the Sociology of Language 231. 85–105. Ortiz de Urbina, Jon. 2003. “Causatives”. In José Ignacio Hualde & Jon Ortiz de Urbina (eds.), A Grammar of Basque. Berlin/New York: Mouton De Gruyter. 592–607. Ortiz de Urbina, Jon & Beatriz Fernández. 2016. “Datives in Basque Bivalent Unerga-

316

ortiz de urbina

tives”. In Beatriz Fernández & Jon Ortiz de Urbina (eds.), Microparameters in the Grammar of Basque. Amsterdam: John Benjamins. 67–92. Oyharçabal, Bernard. 2003. “Lexical Causatives and Causative Alternation in Basque”. ASJU Supplement XLVI. 223–253. Oyharçabal, Bernard. 2007. “Basque Ditransitives”. Ms, IKER-CNRS. Perlmutter, David. 1978. “Impersonal Passives and the Unaccusative Hypothesis”. In Fourth Annual Meeting of the Berkeley Linguistics Society, vol. 4. 157–190. Pylkkänen, Liina. 2008. Introducing Arguments. Cambridge, MA: The MIT Press. Rezac, Milan. 2009. “Person restrictions in Basque intransitives”. Lapurdum 2009, 13. 305–322. Treviño, Esthela. 1994. Las causativas del español con complemento infinitivo. México: El Colegio de México. Tubino, Mercedes. 2011. Causatives in Minimalism. Amsterdam/Philadelphia: John Benjamins. Zubizarreta, M.L. 1985. “The relation between morphophonology and morphosyntax: The case of Romance causatives”. Linguistic Inquiry, 16:2. 247–289.

chapter 10

Grammaticalization Processes in Causal Subordination José Ignacio Hualde and Manuel Pérez Saldanya

1

Introduction

Causal constructions express meanings related to the notion of “cause”, including, not only the reason why something happens, but also motivations, explanations and justifications for saying something or performing a specific speech act. Structurally, a causal construction may contain a finite clause (e.g. John got a fine [because he was speeding]), a nonfinite clause (e.g. John got a fine [ for going too fast]) or a noun phrase headed by a preposition (e.g. John was punished [ for his bad behavior]). In this chapter we will concentrate only on finite subordinate causal clauses. A well-known tripartite classification of causal clauses is as content causals (expressing the “real-world” cause of an event or state of things, e.g. The ground is wet [because it has rained]), epistemic causals (speaker’s inference or reason, e.g. It has rained, [because the ground is wet]) and speech-act causals (justification for a command or another speech act expressed in the main clause, e.g. Go home, [because it is about to rain]) (see Sweetser 1990: 76– 82). Structurally, causals may be internal, if integrated with the predicate of the main clause, or external, if they are sentence adjuncts, prosodically separated from the main clause by “comma intonation” (see Santos Río 1982, Goethals 2002, Pérez Saldanya 2014: §28.4.1, among others). External causals may appear at the left periphery (e.g. [Since it was about to rain], he decided to leave right away), at the right periphery (e.g. He decided to leave right away, [since it was about to rain]) or in a parenthetical position (e.g. He decided, [since it was about to rain], to leave right away). Internal causals are always in the content domain, whereas external causals may be in the content, epistemic or speech act domain. Regarding information structure, causals may be classified either as rhematic (expressing new information) or as thematic or presupposed (expressing given or assumed information). Internal causals are always rhematic, whereas external causals may be either rhematic or thematic/presupposed.

© koninklijke brill nv, leiden, 2019 | doi:10.1163/9789004395398_011

318

hualde and pérez saldanya

table 10.1 Typology of inflected causal clauses

Properties

Examples

internal causal rhematic content

The ground in wet [because it has rained]

external causal

Mary couldn’t see the road, [because it was so dark] epistemic

It has rained, [because the ground is wet]

speech-act Go home, [because it is about to rain] thematic content

[Since it was about to rain], he decided to leave right away

speech act [Since you cannot wait], do it yourself

Following the criteria just mentioned, it is possible to differentiate up to six types of causal clauses, as shown in Table 10.1. We will make use of all these classificatory features in our analysis and comparison of Basque and Romance causal clauses.

2

Types of Causal Subordinators in Basque and Romance

In the Romance languages, as in English, causal clauses, like other types of subordinate clauses, may be headed by a conjunction (e.g. Sp porque ‘because’, como ‘as’, que ‘that’, pues ‘as, so’) or by more complex expressions, which show different degrees of grammaticalization in this function (e.g. Sp ya que ‘since’, puesto que ‘since’, dado que ‘given that’). In Basque, on the other hand, there is more structural heterogeneity. Finite clauses with causal value in Basque may contain verbal suffixes, verbal prefixes and/or clause initial conjunctions, as shown in (1) with the example ikusi dugu ‘we have seen (it)’: (1) Causal subordination in Basque a. ikusi dugulako ‘because we have seen it’ b. ikusi dugunez ‘as we have seen’ (-n ‘subordinator’ + -ez ‘instrumental’) c. ikusi dugunez gero ‘since we have seen it’ (gero ‘later’)

grammaticalization processes in causal subordination

d. e. f. g. h. i. j.

319

ikusi dugu eta ‘since we have seen it’ (eta ‘and’) ikusi baitugu ‘since we have seen it’ (bait- ‘subordinator’) zeren ikusi dugun ‘because we have seen it’ (zeren ‘of/for what’) zeren (eta) ikusi baitugu ‘because we have seen it’ ezen ikusi dugu ‘since we have seen it’ (ezen ‘that, since’) zergatik ikusi dugu(n) ‘because we have seen it’ (zergatik ‘why’) nola ikusi baitugu ‘since we have seen it’ (nola ‘how’)

In spite of all the obvious differences between Basque and Romance causal elements, there are also many similarities, especially when we consider the historical origin of causal connectors. In Table 10.2, we make explicit some of these similarities by comparing Basque causal subordinators with etymologically equivalent expressions in the main Romance languages with which Basque has been in contact, Spanish, Gascon and French.1 Since contact with French is relatively recent, we indicate only the most usual forms for this language. (Another historically important contact language was Navarro-Aragonese Romance, which in all relevant respects coincides with Castilian Spanish). Subordinators that have become obsolete are signaled with “†”. As shown in Table 10.2, in both Basque and Romance we find causal subordinators that have developed historically from question words, from general complementizers (‘that’) and from time and modal expressions, as we also find in many other language (Kortmann 1997, 1998; Pérez Saldanya 2014: 3479–3483). On the other hand, Basque shows at least one grammaticalization phenomenon that has no parallels in Romance: the development of a causal construction involving the coordinating conjunction eta ‘and’ as a verb enclitic. We also find other differences in the specific meanings and uses of different causal structures. In modern Basque a direct answer to the question ‘why’ generally requires the construction with -(e)lako ‘because’ attached to the inflected verb. This suffix is thus the basic equivalent of Sp porque, Fr parce que or Eng because. However, unlike Sp porque, French parce que and Eng because, the Basque connector is usually only employed in internal causal clauses. Other elements such as preverbal bait- ‘since’, clause-initial zeren ‘since, because’ and postverbal eta have a larger set of causal usages in Basque.

1 We have obtained the information about Occitan from Alibèrt (1935–1937) and Fernández González (1985). For Gascon, we have consulted Palay (1974) and the Bearnais dictionaries in Lexilogos (2002–2017).

320

hualde and pérez saldanya

table 10.2 Causal subordinators in Basque and Romance classified by etymological origin Origin

Basque

Spanish

Old occitan and contemporary Gascon

Interrogative

(zer ‘what’ >) zeren … baitzeren … -(e)n ze (colloquial) baitezen -(e)lako

(Lat. quia >) †ca

(Lat. quāre >) †car car

(Lat. quid >) que

que

General complementizer ‘that’+ ‘of ’ (Bq) ‘for’ + ‘that’ (Rom) Locutions with nouns and ‘that’

French

que

porque

per çò que parce que perquè dela kausa/bide/medio a/por causa de que a causa que à cause que † por razón que per (a)mor/pramor que Time adverb -(e)nez gero (Lat.post >) pues (que) † pus (que), puix que puisque (or other temporal ya que items) puesto que vist que vu que del moment que †com/coma Modal adverb nola … bait(Lat. quomodo >) comme (& interrogative ‘how’) nola … -(e)n como -(e)nez Copulative conjunction (-la) eta n.a. n.a. n.a.

Connectors of modal origin (Bq -n-ez ‘subordinator+instrumental’, nola ‘how’> ‘as, since’; Sp como ‘how’> ‘as, since’) are essentially restricted in their use to thematic content clauses. On the other hand, regarding causal connectors related to time expressions, whereas Bq -n-ez gero seems to be used only in thematic causal clauses, Sp pues ‘since’ (from Lat post ‘after’), ya que ‘already that’> ‘since’, puesto que ‘given that’ are used both in thematic clauses and in external rhematic clauses. In the next sections we consider each of these types of causal constructions in more detail. We structure the discussion around Basque, pointing out similarities and differences with respect to Spanish and other neighboring Romance languages.

3

Direct Answers to the Question ‘Why?’ (Internal Causal Clauses)

In this section, we consider the basic connectors used in internal causal clauses in Basque. As mentioned, internal causal clauses are always rhematic and in

grammaticalization processes in causal subordination

321

the content domain. As such, they can be used to provide a direct answer to the question ‘why?’. With this function, modern English uses the conjunction because, Spanish has porque (por ‘for’+ que ‘that’) and French parce que (par ‘for’+ ce ‘this’ + que ‘which’). In contemporary standard Basque, the main subordinator used in this context is the suffix -(e)lako. Morphologically, -(e)lako can be analyzed as bimorphemic, with the relational suffix -ko attached to the subordinating suffix -la ‘that’ (cf., e.g., ikusi dugu-la ‘that we have seen it’, ikusi dugu-la-ko ‘because we have seen it’). Some examples illustrating the use of -(e)lako are given in (2) (adapted from EGLU 2005: 192) and (3) (from a journalistic interview). We provide Spanish translations as well for comparison:2 (2) —Zergatik eman diozu dirua? why give aux money.the ‘Why did you give him/her the money?’, Sp ¿Por qué le has dado el dinero? —Egunkaria erosi behar zuelako. newspaper.the buy must aux.because ‘Because s/he had to buy the newspaper.’, Sp Porque tenía que comprar el periódico. (3) —Derrigorrezko galdera: Zergatik? ‘Obligatory question: Why?’ —Gauzek ez zutelako lehen bezalako naturaltasunez funtzionatzen. ‘Because things were not working with the same naturalness as before.’ (http://www.argia.eus/argia‑astekaria/2560/napoka‑iria) Sp Pregunta obligatoria: ¿Por qué? Porque las cosas no funcionaban con la naturalidad de (como) antes. Unlike their Romance and English counterparts, causal clauses with -(e)lako often appear before the main clause as in (4) (adapted from Artiagoitia 2013: 715):

2 The following abbreviations are used in the glosses: abl ‘ablative’, aux ‘auxiliary’, comp ‘complementizer’, det ‘determiner’, erg ‘ergative’, fut ‘future’, gen ‘genitive’, imp ‘imperative’, impf ‘imperfective’, inst ‘instrumental’, loc ‘locative’, neg ‘negation’, part ‘partitive’, pl ‘plural’, rel ‘relative’, subor ‘subordinator’.

322

hualde and pérez saldanya

(4) Beste erremediorik ez nuelako egin nuen. other remedy.part neg I.had.because do aux ‘I did it because I did not have any other option.’ Sp Lo hice porque no tenía otro remedio. The reason for this difference between the languages is that in Basque focalized elements are generally placed in preverbal position (Artiagoitia 2003: 714, Ortiz de Urbina 2003: 454–455). In addition to its use as a causal connector, -(e)lako can also be used to link an inflected clausal complement to a following head noun (a function that can also be fulfilled by the suffix -(e)n). This is in fact the only use in which this verbal suffix is found in early Eastern authors (again, Spanish translations are provided for comparison): (5) Zenbait granoz arin delako seinalea da. some grain.inst light is.lako sign is lit. ‘It is a sign that he is light by a few grains.’ (i.e. ‘it is a sign that he is a little reckless’) (Axular, Gero: 270, 1643) Sp Es una señal de que es ligero por algunos granos. (6) erremediatuko zarelako esperantzarekin remedy.fut aux.lako hope.with ‘With the hope that you will reform yourself’ (Axular, Gero: 145, 1643) Sp con la esperanza de que te reformarás The bridging context between these two rather distinct usages of -(e)lako might be found in potentially ambiguous examples like (7), where the embedded clause can be understood as a complement of the noun poz ‘happiness’, but also as a causal complement of the main clause:3 (7) etorri zarelako ene poz handia come aux.lako my happiness big ‘my great happiness that/because you have come’ In (8) we give an example (from a contemporary novel) where both uses of -(e)lako can be observed:

3 In modern Eastern usage, there is a formal differentiation between -(e)lakotz, which is used as causal suffix, and -(e)lako, used with complements of nouns and adjectives.

grammaticalization processes in causal subordination

323

(8) Bakartia baldin banaiz, izan ere, inor ez molestatzeko da, solitary if if.I.am be too anyone not bother.for is ezin dudalako eraman inori traba edo kalte egiten not.be.able aux.because bear anyone difficulty or harm do diodalako irudipena edo kontzientzia, eta bakarrik egotea delako aux.of suspicion or awareness and alone being is.because hori saihesteko modu seguruena, bakarra ez bada. that avoid.for way surest only not if.is ‘If I am solitary, it is actually not to bother anybody, because I cannot bear the suspicion or the awareness that I am hurting or bugging anybody; and because being alone is the surest way to avoid that, if not the only way’. (J.L. Zabala, Ospa, 2017) As mentioned, the Basque verbal suffix -(e)lako is transparently analyzable as consisting of the complementizer -(e)la ‘that’ (cf. uste dut ikusi dugu-la ‘I think that we have seen it’) and the relational suffix -ko used to link complements to nouns (cf. mendi-ko etxea ‘the house of the mountain’, mendi-ranz-ko bidea ‘the road (of) towards the mountain’). As shown, in the Spanish translations provided in (5)–(6) above, the preposition+complementizer sequence de que in Spanish offers a direct parallel (in mirror image).4 It may be noted that Sp porque ‘because’ also has a somewhat similar structure, with the preposition por ‘for, by’ preceding the general complementatizer que ‘that’. Whereas Sp porque is morphologically identical to por qué ‘why’, however, Bq -(e)lako ‘because’ and zergatik ‘why’ are obviously dissimilar. Although Basque has a causative suffix -gatik, it is not used with inflected clauses, but only with noun phrases and nominalizations, e.g. euri-a-gatik ‘because of the rain’, Sp por la lluvia; Jon ikuste-a-gatik ‘because of seeing Jon’, Sp por ver a Jon; egia esate-a-gatik ‘for telling the truth’, Sp por decir la verdad. On the other hand, uninflected nominalized clauses with -ko (paralleling -(e)lako) have a final rather than causal interpretation, e.g. Jon ikusteko ‘in order to see Jon’, Sp para ver a Jon; egia esateko ‘in order to tell the truth’, Sp para decir la verdad. Also as already mentioned, unlike Sp porque or Eng because, which have a wider usage, the Basque suffix -(e)lako is restricted in its use essentially to internal causal clauses, at least in some dialects (cf. EGLU 2005: 169, 179). It is not commonly employed in epistemic causals (e.g. It has rained, because 4 Modal adverbs also take the suffix -(e)la (e.g. honela ‘thus, this way’, bestela ‘otherwise’). When these adverbs function as a complement of a noun we find the suffix sequence -la-ko (e.g. honelako galderak ‘this type of questions’, bestelako iritziak ‘different oppinions’, also with an empty noun head: bestelakoa da ‘it is different’).

324

hualde and pérez saldanya

the ground is wet) or speech-act causals (Go now, because it is about to rain), where other connectors are employed instead (see sections 4 and 5). Its (possible) origin in inflected complements of nouns and adjectives may account for the restriction of -(e)lako to the internal causal context. In both cases, the subordinate clause is an internal adjunct, and if -(e)lako causals derives from -(e)lako noun complements, the internal character of the former may be a conservative property of the latter. Nevertheless, nowadays -(e)lako is sometimes also used in external causal clauses, as in (9) and (10), both from the same novel: (9) Endredo bat zen enkargu hura, idazten ari nintzen nobelatik ateraraziko ninduelako. ‘That request was a complication, because it would take me out of the novel I was writing.’ (J.L. Zabala, Ospa, 2017) Sp Ese encargo era un enredo, porque me sacaría de la novela que estaba escribiendo. (10) jemezeak deituak, GMC markakoak zirelako ‘So called “ge-em-ces”, because they were of the GMC brand’ (J.L. Zabala, Ospa, 2017) Sp llamados jemecés, porque eran de la marca GMC This use of -(e)lako in external causal clauses seems to be relatively recent (EGLU 2005: 179) and may be due to convergence with Spanish. Internal content clauses in Basque may also be introduced by clause-initial conjunctions such as zeren, ze, ezen and zergatik, which also have a wider causal usage (see section 5). Although these clause-initial conjunctions are found since the first texts, they are often avoided in more recent writing, as they are viewed by some as calques from the Romance languages (Azkue 1923: 491).

4

External Causal Clauses

Leaving aside causal subordinators that have their origin in modal and time conjunctions (see sections 6 and 7), we find two additional main subordinators in external causal clauses in Basque, whether content, epistemic or speech-act: the verbal prefix bait- (attached to the inflected verb, sometimes with elision of t), as in (11), and enclitic -eta (usually enclitic to the inflected verb, but not exclusively, and frequently phonologically reduced to ta), as in (12).

grammaticalization processes in causal subordination

325

(11) Nik atzo hik baiño bide gehiago iragan nian inguratu bainuen I yest. you but way more pass aux surround since.aux mundu guztia. world all ‘Yesterday I walked more than you did, because I went around the whole world.’ (Axular, Gero: 21, 1642) (12) Ez barrerik egin, egia da ta. no laugh.part do truth is and ‘Do not laugh, because it’s true.’ (Azkue, Bein da betiko, apud EGLU 2005: 199) As in the two examples just mentioned, both subordinators are generally used in rhematic clauses, and are thus normally found in the right periphery of the sentence. Nevertheless, causals with bait- may also be thematic, located at the left periphery, as in (13). (13) ziren zirena baitzira, zuzaz pena dizit nik you.are.rel you.are.rel.det comp.you.are you.inst pain aux I.erg ‘Because you are who you are, I agonize over you.’ (Etxepare, Primitiae: VIII.10, 1545) Traditionally these two connectors had different geographical distributions, bait- being used in eastern varieties and -eta in western and central areas. Nevertheless, both are nowadays used in standard Basque. 4.1 The Subordinator baitThe verbal prefix bait- is employed in a wide variety of subordinate clauses in eastern dialects, including relative, completive, consecutive and causal clauses (Oyharçabal 2003). It is in complementary distribution with the subordinating suffixes -(e)n and -la (and those derived from them, such as -(e)nez, -(e)lako, etc.), in the sense that it cannot co-occur with any of them. This prefix is attested in early texts from both eastern and western regions, but it fell out of use in western and central dialects. The main use of bait- in the modern standard language is in causal clauses, with very much the same usages as postverbal causal -eta. Krajewska (2016) argues that bait- was initially a relative clause marker (which is its most common use in 16th century texts) and that it developed as a causal subordinator in contexts where there was ambiguity between a relative clause and a reason clause reading. In his study on causal constructions in P. Axular’s Gero (1643), Villasante (1982b: 364–380) provides several examples where a clause with bait- can be

326

hualde and pérez saldanya

interpreted as either an explicative relative clause or as an external causal clause. In his Spanish translation, Villasante chooses a causal interpretation for the example in (14) and a relative clause interpretation for (15), but it seems that both interpretations would be possible in both cases. (Notice that bait- and the verb may contract phonologically. In these examples, bait-da > baita and baitdira > baitira). (14) Iakin dezagun gauza bat, sentitzekoa baita. ‘Let’s keep one thing in mind, because it is worth noticing (or: which is worth noticing).’ (Axular, Gero: 75, 1964 [1643]) Sp Tengamos en cuenta una cosa, pues es muy digna de advertencia. (L. Villasante’s translation) (15) Utzten ditut plazer haren aitzingibelak, anarteraiñoko, eta ondoreko egitekoak, eta atsekabeak, handiak baitira. ‘I leave aside the pros and cons of that pleasure, the preceding ones, and those that follow, which are big (or: because they are big).’ (Axular, Gero: 251, 1643) Sp Dejo ahora a un lado las circunstancias de dicho placer, los quebraderos de cabeza y los disgustos que le preceden y le siguen, que son grandes. (L. Villasante’s translation) It is likely that the grammaticalization process that has led to bait- being used as a causal connector is the same that is responsible for a general subordinator being able to introduce external causal clauses, as is the case with Spanish que ‘that’. For instance, in a Spanish example such as (16a), literally, ‘I am leaving, that is late’ an inference of causality allows the general subordinator que ‘that’ to be interpreted as ‘because’. We may envision a similar process in the grammaticalization of Bq bait- as a causal connector, as in (16b). (16) a. Me voy, que es tarde. ‘I am leaving, because it is late.’ b. Banoa, berandu baita I.leave late that.it.is This explanation, however, does not account for why bait- has become specialized as a causal subordinator in modern Basque (whereas Sp que has not had this development). It also leaves unexplained the fact that bait- can be used in thematic causals, whereas Sp que is restricted to rhematic ones, especially in speech-act causals (Pérez Saldanya 2014: 3510–3514).

grammaticalization processes in causal subordination

327

4.2 The Subordinator -(e)ta One of the most productive clausal constructions in the present-day language of the western and central Basque-speaking area involves the conjunction eta ‘and’, usually cliticized to an inflected verb in sentence-final position, as in (17): (17) Horixe esango nuke, nahikoa ezkor aritu naiz-eta. that say.fut aux rather negative speak aux-and ‘I would say that, since I have been speaking rather negatively.’ (https:// barren.eus/albisteak/kultura/1806‑maialen‑liujanbio‑euskaratik‑eta‑ gure‑mundu‑erreferentzialetik‑egin‑behar‑dugu‑besteen‑ispiluetan‑ gure‑buruaren‑bila‑ibili‑gabe) This causal construction, which does not have a structural parallel in Romance, appears to have arisen from coordinated structures with some specific properties, as explained in EGLU (2005: 201) which provides the example in (18) to illustrate the process: (18) a. Mutil argia da eta berehala ikasiko du. boy smart is and immediately learn.fut aux ‘He is a smart boy and will learn quickly.’ b. Berehala ikasiko du, mutil argia da eta. ‘He will learn quickly, since he is a smart boy.’ In the coordinated structure in (18a) it is possible to make the inference that the reason for the proposition expressed in the main clause is that expressed in the second clause. By transposition of the two clauses, in coordinated clauses where this inference arises, we obtain (18b), with enclitization of the conjunction eta ‘and’: [S1]-eta [S2] → [S2] [S1]-eta. The reason why this grammaticalization phenomenon has taken place in Basque, but not in Romance (or in Germanic) may be a prosodic difference in the parsing of conjunctions. In Basque, a postpositional language, the conjunction eta is often prosodically cliticized to the preceding word, as a postposition. This is most evident in dialects that have certain phonological rules that apply only within word domains. One of these rules, for instance, is the raising of /a/ to /e/ after a high vowel, as in lagun-a → lagune ‘the friend’, which may also affect (e)ta ‘and’ in dialects with this rule: hiru ta lau → hiru-te lau ‘three and four’. Another such phonological rule is the palatalization of /t/ after /i/: mendi-tik → mendittik [mendicik] ‘from the mountain’ and also, e.g. bi ta hiru → bi-tte hiru ‘two and three’. In fact, encliticization of eta ‘and’ to a verbal form

328

hualde and pérez saldanya

has also resulted in other grammaticalizations, such as resultative constructions, e.g. ikusi eta ‘see and’→ ikusita (in some varieties, ikusitte) ‘having seen’ (Krajewska 2013). As also noted in EGLU (2005: 202), we have a parallel development with the coordinating conjunction baina ‘but’, which can also appear postposed to the second clause, giving rise to structures with adversative meaning. The causal construction with gero (see section 6) also has its origin in the clitization of an adverb to the right of a verb.5 In addition, the particle eta is suffixed to a copulative verb marked with -(e)la in causal constructions where the cause is taken to be a referential expression (a noun phrase), e.g. katuak direla-eta ‘because of the cats’, cf. also the grammaticalized expression zer dela-eta ‘because of what, why?’ (lit. what that-is-and). (EGLU 2005: 204–205). The enclitization of conjunctions and other elements to verbal forms, whether finite or participial, has thus been the source of a number of grammaticalized constructions in Basque. This path of development, of which the causal construction with -eta is an instance, does not have any parallels in the Romance languages.

5

Clause-Initial Causal Conjunctions

From the earliest texts in Basque we find causal clauses headed by conjunctions such as zeren (or zerren), zeren eta and zergatik. In these subordinate clauses, the inflected verb may take the prefix bait-, as in (19) and (20), the suffix -(e)n or even -(e)lako. It may also be unmarked as a subordinated verb, as in (21). (19) egiten dut ihes zeren ezagutzen baitut neure kondizinoa, do.impf aux flee because know.impf comp.aux my condition ez naizela iend’ arteko gai not I.am.comp people among.of able ‘I flee because I know my condition, that I am not capable in society’ (Axular, Gero: 192, 1643)

5 The conjunction edo ‘or’ is also used as a postposition; specifically it is postposed to measure phrases to convey the meaning ‘or so’, e.g. lauretan edo ‘at 4:00 or so’. In addition, both disjunctive conjuntions edo and ala are used as tags in questions, like German oder, cf. Eng or what, Sp o qué.

grammaticalization processes in causal subordination

329

(20) Zeren Jeinkoa egun oroz ongi ari baitzaigu guk ere hala because God day all.inst well act comp.aux we.erg too thus behar harzaz untsa orhitu. must 3.inst well remember ‘Since God rewards as every day, we too should remember Him in gratitude.’ (Etxepare, Primitiae: I.17, 1545) (21) Esarasu zurinean serren da berroago goxetaraco. pour.imp white.abl because is warmer morning.for ‘Pour some white [wine], because it is warmer for the morning.’ (Mikoleta, Modo breve: 208, 1653) (Sp translation in Mikoleta’s text: Echa de lo blanco, que es mas caliente para por la mañana.) Zeren and the nowadays stigmatized form zergatik are polyfunctional causal conjunctions, in the sense they may introduce internal clauses, as in (19) above, and external clauses, either thematic, as in (20), and rhematic, as in (21). On the other hand, zeren eta always introduces external clauses, mostly with epistemic value, as in (22). (22) Eta hutsuneak ere hor daude, bertzalde, zeren eta and lacunae also there they.are other.side because and memoriak bere zuloak eta bere haitzuloak baititu. memory.erg its holes and its caverns comp.aux ‘And the lacunae are there too, on the other hand, because memory has its holes and its caverns.’ (J.M. Irigoien, Lur bat haratago: 10, 2000) As has been already mentioned, zerga(i)tik is also the interrogative word ‘why’, so that there is an obvious parallelism with Sp por qué ‘why’ and porque ‘because’. The structure of this form is transparently zer ‘what’+ gai ‘matter’+ tik ‘through, from’. As for zer(r)en, it is the genitive form of zer ‘what’, which in earlier times was also used as benefactive; that is, it may be literally translated as ‘for what’. The use of zeren as ‘why’ is also found in early texts, as in (23)– (24): (23) Zeren, bada, erho gira gaixo bekhatariak? why thus crazy we.are poor sinners ‘Why, then, are we poor sinners so foolish?’ (Etxepare, Primitiae: 2.60, 1545)

330

hualde and pérez saldanya

(24) Ene Iainkoa, Ene Iainkoa, zeren abandonatu nauk? ‘My God, my God, why hast thou forsaken me?’ (Leizarraga, Testamentu, Matthew: 27:46, 1571) Question words are a common source of causal conjunctions crosslinguistically (Kortmann 1988: 514). In the Romance languages, in addition to the example provided by Sp por qué ‘why’, porque ‘because’, and cognates in other languages, French, Occitan and Catalan car ‘because’ has its origin in Latin quāre ‘by what means?’ (Baños 2011: 207, 2014: 54). The Latin causal conjunction quia ‘why’ also derives from a question word (Hofmann & Szantyr 1965: 585, ap. Baños 2011: 206–207). The basic mechanism seems to be an equation between question and answer particles, with a change in intonation. As argued by Pineda & Pérez-Saldanya 2017 (based on the diachronic analysis of car in Catalan), the evolutionary path would be as in (25): (25) question word > conjunction in internal causal clauses > polyfunctional causal conjunction > conjunction in external causal clauses Within the diachronic path in (25), zeren appears to be in the third stage, as it clearly has a polyfunctional use, as mentioned above. In this respect it is comparable in its range of functions to Spanish porque and to car in French, Occitan and Old Catalan (and English because). We may distinguish a number of causal contexts in which Bq zeren is employed. a) First of all, it is used in internal causal clauses, with a meaning similar to that of -(e)lako, as in (19) above, where the subordinate clause expresses the reason for the subject fleeing, and in (26), where it expresses the reason for Peter being grieved:6 (26) Triste zedin Pierris zeren erran baitziezon herenean, On dariztak niri? ‘Peter was grieved because he said unto him the third time, Lovest thou me?’ (Leizarraga, Testamentu: John, 21:17, 1571) As internal causal clauses, clauses headed by zeren can be used as a reply to questions with zergatik or zeren, as in (27), and can also be the focus of negation, as in (28):

6 All English translations from the Bible are from the King James version, which is roughly contemporary with Leizarraga’s Basque translation.

grammaticalization processes in causal subordination

331

(27) Zergatik ene lengoajea eztuzue aditzen? zeren ezin enzun baitirokezue ene hitza. ‘Why do ye not understand my speech? even because ye cannot hear my word’ (Leizarraga, Testamentu: John 8: 43, 1571) (28) Eta haur erraiten du hunela, ez zeren dantzatzea bera hain gauza gaixtoa den, baiña zeren dantzatik anhitz okhasino behar eztenik sortzen den. ‘And he says this this way, not because dancing is by itself such an evil thing, but because from the dance many inappropriate occasions arise.’ (Axular, Gero: 268, 1643) b) The conjunction zeren is also very commonly found in external causal clauses, especially those in the content domain: (29) eta bertan ilki zitezen, zeren ezpaitzuten lur barnerik. ‘and because they had no root, they withered away.’ (Leizarraga, Testamentu: Matthew 13:6, 1571) (30) Serren diñoe doctoreac goxeti xayguitea dala ona ossasunensaco. Porque dicen los medicos que para la salud es bueno el leuantar de mañana. ‘Because physicians say that to get up early is good for your health.’ (Mikoleta, Modo breve: 202, 1653) (31) Errege Daviti eman zerautzatenean Saulen harmak, baitziren ezin hobeagoak eta halakoak, guztiarekin ere, probetxu baiño kalte gehiago egiten zioten, trabatzen zuten. Zeren nola anarteraiñokoan bethi arzain ibili baitzen, ezpaitzen oraiño harmetan usatua. ‘When they gave King David Saul’s weapons, even though they were the best, they did him more harm than good; they bothered him. Because, since up to that moment he had always been a shepherd, he was not yet used to weapons’ (Axular, Gero: 61, 1643) c) It may also occur in external causals in the speech act domain, as in the following examples. In (32) the causal clause justifies a command, in (33) it justifies a question and in (34) it motivates an exclamation: (32) Boz eta alegera zaitezte, zeren zuen saria handi baita zeruetan ‘Rejoice, and be exceeding glad: for great is your reward in heaven’ (Leizarraga, Testamentu: Matthew 5:12, 1571)

332

hualde and pérez saldanya

(33) zenbat gehiago eztire bihotzez estekatuak, josiak beren tokiko mintzaiari, zeren erlisionea kasik orok haurdanik mintzai hartaz ikhasia dute, guk emen euskaraz bezala? ‘How many more are not tied and attached in their hearts to their native language, for almost all of them were taught religion in childhood in that language, just like we here in Basque?’ (H.U.Gonz 193, ap. Hualde & Ortiz de Urbina 2003:730) (34) Ai gure urrikalkizunak, galdu gara, egin du gureak. Zeren akhabatu hurran da eguna, beheititu da iguzkia, eta itzala ere luzatuago da arratsaldean. ‘Poor us, we are lost, our luck has finished. Because the day has almost ended, the sun has lowered and even the shade is longer in the evening.’ (Axular, Gero: 130, 1643) d) It is also found in thematic external causal clauses, especially those in the content domain, (35)–(36). This was also the case for Sp porque some centuries ago, but it is less so in contemporary Spanish usage (Pérez Saldanya 2014: 3519– 3524): (35) Gero iguzkia goratu eta, erre izan dirade, eta zeren ezpaitzuten errorik, eiarthu izan dirade. ‘And when the sun was up, they were scorched; and because they had no root, they withered away.’ (Leizarraga, Testamentu: Matthew: 13:6, 1571) (36) Zeren amak diren koleratsu, halatan dira umeak ere faltatsu. ‘Because the mothers are quick to anger, the children are also defective.’ (Axular, Gero: 200, 1643) e) Finally, zeren may introduce clauses that express both cause and finality, which is also possible in Spanish with porque: (37) Zeren unsa orhit ziten nizaz, ama eztia, / gogo honez erranen dut zuri Abe Maria. ‘So that you may remember me, sweet mother, with good intention I will say Hail Mary to you’ (Etxepare, Primitiae: 93, I.445, 1545) (38) Eçtoçu cerren berva oriec esan, Sirena ‘You have no reason to say those words, Sirena’ (Lazarraga, Eskuizkribua: 1151v, c. 1567)

grammaticalization processes in causal subordination

333

The conjunction zeren is thus a polyfunctional causal element which has a very wide range of causal usages, in some respects going beyond those of Sp porque or Eng because.

6

Causal Subordinators of Temporal Origin

Crosslinguistically, one of the most common lexical sources of causal subordinators are time expressions referring to a point in time after the point of reference, such as ‘after’, ‘since’, ‘already’ and even ‘when’ (in the meaning of ‘at that point’, ‘and right after that’). In the Romance languages we find several causal connectors with this origin. The most widespread ones are those deriving from Lat post ‘after’ or its comparative variant postiu ‘more after’, which had acquired a causal usage already in Late Latin: Sp pues (que), Port pois (que), Occ/Cat pus (que), puix (que), Fr puisque, It poiché. Causal connectors deriving from ‘already’ are also common in the Romance languages, but they arise later and are found in fewer languages: Sp ya que, Port já que, Cat (and some Occ varieties) ja que, It giacché (but not in French). Absolute participial constructions (‘given that’) can also be included in this group: Sp puesto que, dado que, Port visto que, Cat/Occ vist que, atès que, Fr vu que. In Basque, the adverb gero ‘later’ (e.g. gero ikusiko dugu ‘we will see it later’) has also developed a causal usage in postverbal position, even though, unlike the Romance connectors just mentioned, it has kept in addition its temporal interpretation (cf. Eng since). With temporal value, gero is a transitive adverb and may take a complement with the instrumental suffix -(e)z-, after the subordinator -n if the complement is a finite clause, as in (41). In this usage it is equivalent to después de in Spanish. In (39) gero’s complement is a noun phrase, in (40) a participial clause, and in (41) an inflected clause. Spanish translations are provided for comparison (examples from Elhuyar online dictionary): (39) Barbaroen inbasioaz gero, ez da sekula horrelakorik barbarian.gen.pl invasión.inst after not aux never thus.of.part ikusi. see ‘After the Barbarians’ invasion, nothing like that has ever been seen.’ Sp Después de la invasión de los Bárbaros, nunca se ha visto algo semejante.

334

hualde and pérez saldanya

(40) Aita hilez gero sortu zen. Father die.inst after be.born aux ‘S/he was born after his father had died’ Sp Nació después de morir su padre. (41) Hanka hautsi zuenez gero, ez da eskiatzera joan. leg break aux.inst after not aux ski.to go ‘After he broke his leg, he has not gone skiing.’ Sp Después de que se rompiera la pierna, no ha ido a esquiar. In addition to this use as a temporal connector, gero may also have a causal or conditional interpretation: (42) Eta gauden erne; zer gertatuko den ez and we.be.subor alert what happen.fut aux.subor not dakigunez gero. we.know.subor.inst after ‘And let’s be alert, since we do not know what will happen.’ (43) Gogor ikasiz gero, azterketa gaindituko nuke. Hard learn.inst after exam pass.fut aux ‘If I studied hard, I would pass the test.’ The word gero itself may take the instrumental suffix (gero-z) or both the instrumental and the ablative (gero-z-tik). We may note that a similar process of morphological reinforcement is found in the Romance languages in its translation equivalents. Thus, in Spanish when pues (< post) acquired a causal value, a new form después (des ‘from’+ pues) developed as a time adverb. In addition, there is a morphologically more complex nonstandard form endespués. In Gascon as well we find similar reinforced forms: despuix (des + puix), apuix (a +puix), d’après (de + après). By morphological reanalysis, the sequence -(e)z gero may appear as ezkero reflecting a pronunciation with devoicing of /g/ after a sibilant. There is also a historical variant azkero, as in (44), where the initial vowel is from reanalysis of the determiner sufix -a ‘the’. The reanalyzed form ezkero is alive in western dialects, but it is no longer used in writing. (44) a. aizkenengo konfesio azkero egin diraden pekatuak ‘the sins that have been committed after the last confession’ (Otxoa de Arin, Doktrina, 1713)

grammaticalization processes in causal subordination

335

b. konfesioaz gero > konfesio azkero confession.det.inst after All the forms that we have mentioned may have either temporal or nontemporal interpretation, but there is a tendency for inflected clauses with -(e)nez gero to be used as causal, for participial clauses with -(e)z gero to have conditional value and for -(e)nez geroztik to be used to express time, especially in the South (EGLU 2005: 191). To some extent, this specialization is parallel to that between pues and después in Spanish. The conditional and causal uses of gero are already found in the first texts from the 15th–16th centuries. It is thus not possible to trace the grammaticalization path of this element. Nevertheless, similar evolutions, from time expression to causal connector, have been well-studied for other languages. The standard view is that the causal meaning is the result of conventionalization of an inference: post hoc, ergo propter hoc (Geis & Zwicky 1971; Levinson 1995; Traugott & Dasher 2002: 16–17, 80; Pérez Saldanya 2014: 2536–2540, among others). By hypothesis, the causal value would arise in potentially ambiguous examples like (45), where precedence in time may also be interpreted as cause: (45) Adanek ausi eban ezkero / Xaunaren mandamendua / Miseriazko balle onetan / Echaku falta llantua ‘Since Adam broke the Lord’s commandment, we have not lacked tears in this valley of sorrows.’ (Barrutia, Gabonetarako ikuskizuna, 1715) The causal inference can only emerge in contexts like that in (45), in which the temporal clause is external to the predicate of the main clause and appears at the left margin of the sentence, in the thematic position (Gutiérrez Ordóñez 2000: 91; Goethals 2002: 111; Pérez Saldanya 2014: 2536–2540; Pérez Saldanya & Hualde 2017). The characteristics of the bridging context where the grammaticalization process is triggered tend to persist (Heine 2002: 84). Thus, Sp pues and ya que are only found in causal clauses external to the main predicate and this is also true of causal clauses with gero in Basque (cf. also causal clauses with since in English). For this reason, they cannot be used to answer the question ‘why’ in any of these languages. Moreover, at the initial stage in the grammaticalization process, causal conjunctions of temporal origin are used in causal clauses providing given or presupposed information, as in (46).

336

hualde and pérez saldanya

(46) Ene penea daucusun guero, arren, berba bat esazu ‘Since you see my pain, please, tell me a word’ (Lazarraga, Eskuizkribua: c. 1567) Subordinate clauses expressing given information are expected to occur preposed to the main clause (which is the position where the reinterpretation from time to cause is expected to take place), as in the example above. Nevertheless, in Basque we find instances of postposed clauses from the first texts showing that the grammaticalization took place some time before: (47) Suplicaetan nachaçu deguidaçula remedioa emun, medecinea çu çarean guero ‘I beg you that you give me the remedy, since you are the medicine’ (Lazarraga, Eskuizkribua: 1152v, c. 1567) At an early stage, causal connectors of temporal origin tend to be used in speech-act causals or in causals that justify some type of modality, mostly related to volition (Santos Río 1982: 242). This is what we find in the two examples just cited, where the subordinate clause justifies the request that the speaker makes by means of a command, as in (46) above, or through the predicate ‘to beg’, as in (47) above. At later stages along this evolutionary path, causal clauses with conjuntions of temporal origin may start being used in other contexts; in particular as rhematic clauses in the content or even the epistemic domain (Pérez Saldanya & Hualde 2017). This possibility is well documented in Spanish, where the conjunction pues is nowadays only used in rhematic causal clauses (e.g. No veía el camino, pues era de noche ‘I couldn’t see the road, since it was at nighttime’) and ya que has added this rhematic use to its original thematic value (Pérez Saldanya 2014: 3551–3556). In Basque as well we may find gero in rhematic content causal clauses in the content-domain, as in (48), and in the epistemic domain, as in (49) (examples from the newspaper Berria have been obtained from EPG): (48) Are handiagoa izango da gaurkoa, urteurrena mendeurren bihurtu denez gero. ‘And today’s will be even bigger, because the anniversary has become a centennial celebration.’ (Berria, 2004-06-16) (49) Bere garaiko Itsasondo badoa, baina jende gehiago ere etorriko da, etxeak eraikitzen ari direnez gero. ‘The Itsaondo of that time is going, but more people will come, because they are building more houses.’ (Berria, 2004-10-02)

grammaticalization processes in causal subordination

337

This appears to be a recent usage, perhaps due to contact with Spanish. Nevertheless, unlike what has been observed for Spanish pues and ya que, the extension in the use of gero in causal contexts has mostly been towards thematic clauses in the content domain, a context where in the Romance languages we may find conjunctions of modal origin (Sp como, Fr comme): (50) Baña aizen ezkeroz burutik maingutu artaraño, non nai dek nerekin jokatu ‘But since you are so short of wit, you want to play with me everywhere.’ (Iturriaga, Ipuinak, 1842) Sp ‘Pero como eres tan corto de mente, en todas partes quieres apostar conmigo.’ (51) desengañoa arturik zegoen ezkeroz eskatu zeban mesedez laga zegiotela edaten baso bat ur ‘As he was taken by disappointment, he asked them to please let him drink a glass of water.’ (Iztueta, Gipuzkoako probintziaren kondaira, 1847) Sp ‘como estaba desengañado, pidió que por favor le dejaran beber un vaso de agua’ (52) Baina Lotek hainbeste erregutu zienez gero, berarekin joan ziren etxera. ‘But, since Lot begged them so much, they went home with him.’ (Elizen Arteko Biblia: Hasiera [Genesis] 19:3, 2008) Sp ‘Pero como Lot les suplicó tanto, se fueron con él a casa.’ Basque causal clauses with gero have moved farther along the grammaticalization path than Fr puisque, which is used only in thematic clauses in the speech-act domain or expressing modality (Ducrot 1983; Zufferey 2012), but less than Sp pues and ya que, which have completely lost all temporal usages (Pérez Saldanya 2014). Two factors may have contributed to the extension of usages of gero in content thematic clauses. One of them may have been the development of adverbial clauses with -enez with thematic content value (e.g. ikusi dugunez ‘as we have seen’, section 7.1), which may have created an equivalence between -enez and -enez gero in this type of causal clause (Artiagoitia 2003: 717–719). The other factor could be the desire to avoid the use of clause-initial nola ‘how, as’ (section 7.2), since clause-initial subordinators have sometimes been regarded as calques from Romance syntax.

338 7

hualde and pérez saldanya

Causal Constructions of Modal Origin

Modal subordinators are also a common lexical source of causal subordinators (Kortmann 1977: 195–197). As is the case for causal subordinators of temporal origin, the causal meaning is the result of conventionalization of an inference that emerges in certain contexts; specifically, in contexts in which the modal clause is external to the predicate of the main clause and is placed in the left periphery of the sentence, in thematic position, as in, e.g., As you want it, we will do it (Pérez Saldanya 2014: 3576–3579). In the Romance languages, the modal conjunction deriving form the Latin adverb quomodo ‘as, like’ (> Sp/Port como, Cat com que, Occ coma, It come, Fr comme, Rum cum) has acquired a causal interpretation. In Basque, this grammaticalization is also the source of the causal use of the interrogative nola and the complex verbal suffix -enez. 7.1 How > Cause In Basque, the interrogative word nola ‘how’ can be used to introduce subordinate clauses with causal interpretation (‘since, as’), in addition to other uses of this subordinator in indirect interrogatives and comparatives. In these sentences, the finite verb of the subordinate clause bears either the prefix bait- or the suffix -en, both marking subordination. Causal constructions of modal origin tend to have different usages than those originating from time adverbials. In particular, their main use is as thematic causals that justify the propositional content of the main clause. This can be observed in the following 17th century examples of nola causals, from Axular’s Gero (Spanish translations from L. Villasante’s bilingual edition are also provided for comparison): (53) Eta nola bertze mundukoa baita geroko, erraiten dute, and how other world.of.det comp.is after.of say aux eztutela presenteko plazera, geroko esperantzagatik utzi not.aux.subor present.of pleasure after.of hope.for leave nahi want ‘And since what pertains to the other world is for later, they say that they do not want to leave the pleasure of the present for the hope of the future’ (Axular, Gero: 93, 1643) Sp ‘Y como lo del otro mundo es para después, dicen que no quieren dejar el placer presente por la esperanza futura’ (L. Villasante’s translation) (54) Zeren zu nola baitzaude deliberatua, eta bai batzutan because you how comp.you.are decided and also sometimes

grammaticalization processes in causal subordination

339

pasionatua eta itsutua ere, eztuzu den guztia ikhusten passioned and blinded too not.aux aux.subor all see ‘Because, since you are decided, and sometimes even passionate and blind, you do not see everything there is’ (Axular, Gero: 211, 1643) Sp ‘Porque, como tú estás decidido y a veces hasta enardecido y cegado, no ves toda la realidad’ (L. Villasante’s translation) (55) Eta nola baitzihoan dudatuaz eta fidantziaren galduaz, hala and how comp.he.went doubting and trust.gen losing thus zihoan itsasoan barrena ere sarthuaz, estaliaz eta hondatuaz he.went sea.loc through also entering covering and sinking ‘And, as/since he was doubting, and losing his confidence, he was thus sinking and drowning in the sea’ (Axular, Gero: 87, 1643) Sp ‘Y como iba dudando y perdiendo la confianza, en la misma medida se iba metiendo dentro del mar, se iba hundiendo’ (L. Villasante’s translation) The following 19th century example cited in Hualde & Ortiz de Urbina (2013: 834) shows the use of nola as a causal conjunction ‘since, as’ and as a comparative (‘like, as’): (56) nola sorho horiek ez baitute hainbertze lan nola how field those not comp.aux as.much work as laborantzek tillage.lands.erg ‘Since those fields do not require as much work as tillage lands’ (Duvoisin, Laborantzako liburua 1858: 79) cf. Sp como esos campos no requieren tanto trabajo como los de labrantío As can be seen in the Spanish translations of the examples above, the use of nola as a subordinator is very much like in Spanish, where interrogative cómo ‘how’ has also developed an unstressed variant como with these modal and causal values: (57) ¿Cómo lo has hecho? ‘How did you do it?’ (58) No sé cómo lo has hecho. ‘I do not know how you did it.’ (stressed cómo, indirect interrogative)

340

hualde and pérez saldanya

(59) Como lo has hecho, lo tienes que pagar. ‘Since you did it, you must pay for it.’ (causal) However, whereas Spanish unstressed como can also introduce purely modal clauses, in Basque a different construction is used in this case, with bezala ‘like, as’, moduan ‘in the manner’ or legez (Bizkaian, from lege ‘law’ in the instrumental case) ‘like, as’, postposed to a verb bearing the subordination marker -(e)n: (60) a. Debes hacerlo como te digo. ‘You must do as I tell you.’ (modal) b. Esaten dizudan bezala/moduan/legez egin behar duzu say.impf aux.subor do need aux ‘You must do as I tell you.’ (modal) c. *Nola esaten dizudan egin behar duzu / *Egin behar duzu nola esaten dizudan Nevertheless, in a relatively small western area the construction with legez ‘as, like’ has expanded its use to cover the same space as Spanish como, including subordinate clauses with both a modal and a causal interpretation (Zuazo 2017: 61). This seems to be a very recent development. Notice that the use of nola as both an interrogative ‘how’ and as a causal subordinator ‘as, since’ is parallel to that of zeren ‘why; because’. It can thus be explained in a similar fashion, as a possible syntactic calque from Romance.7 7.2 Instrumental > Cause Subordinate clauses in which an inflected verb bears the instrumental suffix -(e)z (following the subordination marker -(e)n) are essentially modal, as in (61) but they may also have a causal interpretation, as in the examples in (62)–(63) (examples from newspapers, extracted from EPG): (61) Jakin dugunez, beste bilera paralelo bat zegoen. know aux.inst other meeting parallel one there.was ‘As we have found out, there was another, parallel meeting.’

7 Modal expressions may also give rise to causal conjunctions, cf. Eng how come? In Sp we find cómo es que, cómo así and in Bq nolatan ‘why, how come’, with nola ‘how’ inflected in the locative case.

grammaticalization processes in causal subordination

341

(62) Euskal Herri osoko administraziorik ez dugunez, Basque Country whole.of administration.part not we.have.inst Udalbiltza ezinbesteko eragilea da. U. necessary agent is ‘Since we do not have an administration of the whole Basque Country, Udalbiltza (Federation of Town Governments) is a necessary agent.’ (63) Gimnasiorik ez dugunez, auzoko kiroldegira joan gym.part not we.have.inst neighborhood.of sports.center.to go behar izaten dugu. must be aux ‘Since we do not have a gym, we have to go to the neighborhood’s sports center.’ This construction is historically very recent, dating back only to the 18th century. It appears to have resulted from the omission of gero in the older construction -enez gero (§6). The first examples of -enez without gero have modal value and involve expressions such as dirudienez, antza denez ‘as it seems, apparently’, dioenez ‘as s/he says’ and the like. Clauses with -enez have a very similar distribution and interpretation as Spanish subordinate modal and causal sentences headed by unstressed como ‘as, like; since’. Thus, the Spanish translation of the subordinate clause in the three examples given in (61)–(63) would be, respectively, Como hemos sabido ‘As we have found out’, Como no tenemos una administración de todo el País Vasco ‘Since we do not have administration of the whole Basque Country’ and Como no tenemos gimnasio ‘Since we do not have a gym’. This thus seems to be a case of convergence with Spanish in recent Basque usage, even if the actual morphosyntactic mechanisms are quite different.

8

Conclusions

In this chapter, we have examined the main types of causal structures in Basque, establishing a comparison with Romance structures and developments. Structurally, Basque has more ways to indicate subordination than Romance, including verbal prefixes and suffixes, as well as clause-initial connectors, and all these elements are used in causal constructions. Clause-initial subordinators such as zeren ‘why; because’ and nola ‘how; as’ may have arisen under the influence of a Romance model, but they are found from our earliest texts.

342

hualde and pérez saldanya

table 10.3 Causal subordinators in Basque and Spanish

Type of causal

Subordinators

internal (rhematic)

(content)

Bq.: -(e)lako, zeren … bait-/-(e)n Sp.: porque

external rhematic

content

Bq: zeren bait-/-(e)n, bait-, -(e)ta Sp: pues, porque, ya que, puesto que, †ca, †que

epistemic

Bq: zeren bait-/-(e)n, bait-, (e)ta, zeren eta Sp: porque, pues, †ca

speech-act Bq: bait-, eta; ze Sp: que, porque, †ca thematic content (or presuposed)

Bq: -nez (gero), bait-, nola … bait-/-(e)n, zeren bait-/-(e)n Sp: como, † porque

speech-act Bq: -nez gero Sp: ya que, † pues, puesto que, dado que

In many respects, the grammaticalization paths of causal elements are parallel to those found in the Romance languages, even if the specific structures are very different in many cases, as with bait- ‘that’ and gero ‘later, since’. Other developments in Basque, however, lack a Romance counterpart. This is the case in the grammaticalization of enclitic eta ‘and’ as a causal connector. From a functional point of view, in Basque, like in Spanish and French, there is a causal subordinator that has a polyfunctional character and other causal subordinators with more restricted usages. In Table 10.3 we include the main causal subordinators in Basque and Spanish, classifying them according to the main types of causal clauses in which they tend to appear. (We indicate obsolete Spanish conjunctions with “†”) Basque zeren and Spanish porque are polyfunctional, since they may appear both in internal causal clauses and in different types of external clauses. The rest of the subordinators in Table 10.3 have a more restricted usage. Notice that whereas Basque has a specialized subordinator in internal clauses, -(e)lako, Spanish lacks a conjunction with these properties.

grammaticalization processes in causal subordination

343

The Basque subordinators bait- and -(e)ta have a certain functional parallelism with the Spanish conjunction que (and Old Spanish ca), since all of them can appear in external rhematic causal clauses. Nevertheless, Bq bait- may also be found in external thematic clauses. In both Basque and Spanish (and other Romance languages) we find causal subordinators of temporal origin: Bq gero, Sp pues (que), ya que. These subordinators, in a first stage, are mostly used in thematic causals in the speech-act domain. Later, they tend to broaden their usage to other external causal meanings. This evolution has been different in Basque in Spanish. In Basque the broadening of usage has been towards content thematic clauses. In Spanish, instead, it has been towards rhematic clauses, in particular, those in the content domain. Finally, in both languages we also find causal subordinators of modal origin, employed in content thematic causal clauses: Bq -nez, nola and Sp como.

Acknowledgements The first author acknowledges the research funding received from the Spanish Ministry of Economy and Competitiveness (FFI2014–51878-P).

References Alibèrt, Lois. 1935–1937. Gramatica occitana segons los parlars lengadocians. Montpelhièr: CEO. Artiagoitia, Xabier. 2003. “Adjunct subordination”. In Jose Ignacio Hualde & Jon Ortiz de Urbina (eds.). 2003. 710–762. Azkue, Resurrección María de. 1923–1925. Morfología Vasca. Published in fascicles in Euskera. [Repr. 1969, Bilbao: La Gran Enciclopedia Vasca.] Baños, José Miguel. 2011. “Causal clauses”. In Philip Baldi & Pierlougi Cuzzolin (eds.), New perspectives on the historical Latin syntax, 4: Complex sentences, grammaticalization, typology. Mouton de Gruyter: New York-Amsterdam. 195–234. Baños, José Miguel. 2014. Las oraciones causales en latín: su evolución diacrónica. Madrid: Escolar y Mayo editores. Ducrot, Oswal. 1983. “Puisque: essai de description polyphonique”. Revue Romane 24. 166–185. EGLU = Euskaltzaindia. 2005. Euskal gramatika lehen urratsak: Lehen urratsak-VI, (Mendeko perpausak-2, baldintzazkoak, denborazkoak, helburuzkoak, kausazkoak, kontseziozkoak eta moduzkoak). [Available online: www.euskaltzaindia.eus].

344

hualde and pérez saldanya

Elhuyar hiztegiak. Online: https://hiztegiak.elhuyar.eus EPG = Ibon Sarasola, Pello Salaburu, Josu Landa & Josu Zabaleta. 2009. Ereduzko Prosa Gaur [Model Prose Today]. Online corpus: http://www.ehu.eus/euskara‑orria/ euskara/ereduzkoa/ Fernández González, José Ramón. 1985. Gramática histórica provenzal. Oviedo: Universidad de Oviedo. Geis, Michael & Arnold Zwicky. 1971. “On invited inferences”. Linguistic Inquiry 2: 561– 566. Goethals, Patrick. 2002. Las conjunciones causales explicativas en castellano. Un estudio semiótico-lingüístico. Lovaina: Peeters. Goethals, Patrick. 2010. “A multi-layered approach to speech events. The case of Spanish justificational conjunctions”, Journal of Pragmatics 42. 2204–2218. Gutiérrez Ordóñez, Salvador. 2000. “Causales”, Boletín de la Real Academia Española 80. 47–159. Heine, Bernd. 2002. “On the role of context in grammaticalization”. In Ilse Wischer and Gabriele Diewald (eds.), New Reflections on Grammaticalization. Amsterdam / Philadelphia: Benjamins. 83–101. Hopper, Paul J. 1991. “On some principles of grammaticization”. In Elizabeth Closs Traugott & Bernd Heine (eds.), Approaches to Grammaticalization, vol. I. Amsterdam: John Benjamins, 1991. 17–36. Hualde, José Ignacio & Jon Ortiz de Urbina (eds). 2003. A Grammar of Basque, Berlin / New York: De Gruyter Mouton. Kortmann, Bernard. 1997. Adverbial Subordination. A Typology and History of Adverbial Subordinators Based on European Languages. Berlin-New York: Mouton de Gruyter. Kortmann, Bernard. 1998. “Adverbial subordinators in the languages of Europe”. In Johan van der Auwera (ed.), Adverbial Constructions in the Languages of Europe. Berlin and New York: Mouton de Gruyter. 457–561. Krajewska, Dorota. 2013. The diachrony of resultative constructions in Basque. Master’s Thesis, UPV/EHU. Krajewska, Dorota. 2016. “Basque subordinate clauses: a diachronic study”. Presented at 49th Annual Meeting of the Societas Linguistica Europaea 31 August–3 September 2016, University of Naples Federico II, Naples. Levinson, Stephen. 1995. “Three levels of meaning”. In Frank R. Palmer (ed.), Grammar and meaning: Essays in honor of Sir John Lyons. Cambridge: Cambridge University Press. 90–115. Lexilogos. 2002–2017. Dictionnaire béarnais. [Available online: http://www.lexilogos .com/bearnais_dictionnaire.htm] Mounole, Céline. forthcoming. Lazarragaren eskuizkribuko gramatika, UPV/EHU. OEH = Euskaltzaindia, Orotariko Euskal Hiztegia [Available online: http://www. euskaltzaindia.eus/index.php?option=com_content&view=article&id= 276&Itemid=413&lang=eu]

grammaticalization processes in causal subordination

345

Ortiz de Urbina, Jon 2003. “Word order”. In Jose Ignacio Hualde & Jon Ortiz de Urbina (eds.). 2003. 448–459. Oyharçabal, Bernard. 2003. “Relatives”. In Jose Ignacio Hualde & Jon Ortiz de Urbina (eds.). 2003. 762–823. Palay, Simin. 1974. Dictionnaire du béarnais et du gascon modernes (Bassin Aquitain): embrassant les dialectes du Béarn, de la Bigorre, du Gers, des Landes, et de la Gascogne maritime et garonnaise. Paris: Centre National de la Recherche Scientifique. Pérez Saldanya, Manuel. 2014. “Oraciones causales”. In Concepción Company (dir.), Sintaxis histórica de la lengua española. Tercera parte: Preposiciones, adverbios y conjunciones. Relaciones interoracionales, vol. 3. Ciudad de México: Universidad Nacional de México / Fondo de Cultura Económica. 3447–3610. Pérez Saldanya, Manuel and José Ignacio Hualde. 2017. “From theme to rheme: The evolution of causal conjunctions of temporal origin in Catalan”, Studies in Hispanic and Lusophone Linguistics 10. 319–348. Pineda, Anna & Manuel Pérez Saldanya. 2017. “Les construccions causals i finals”. In Josep Martines & Manuel Pérez Saldanya (eds) Gramàtica del català antic. In preparation. Santos Río, Luis. 1982. “Reflexiones sobre la expresión de la causa en español”, Studia Philológica Salmanticensia 6. 231–277. Sweetser, Eve. 1990. From Etymology to Pragmatics: Metaphorical and Cultural Aspects of Semantic Structure. Cambridge: Cambridge University Press. Traugott, Elisabeth C. & Richard B. Dasher. 2002. Regularity in Semantic Change. Cambridge: Cambridge University Press. Villasante, Luis. 1981a. “Las oraciones causales en Axular (I): La construcción de zeren con partícula que modifica el verbo”. Fontes linguae vasconum: Studia et documenta 37. 9–18. Villasante, Luis. 1981b. “Las oraciones causales en Axular (II): Causales subordinadas y Causales coordinadas”. Fontes linguae vasconum: Studia et documenta 38. 9–22 Villasante, Luis. 1982a. Las oraciones causales en Axular (III): Oraciones causales fronterizas. Fontes linguae vasconum: Studia et documenta 39. 9–20. Villasante, Luis. 1982b. “Las oraciones causales en Axular (IV): Oraciones causales construidas a base de bait solo”. Fontes linguae vasconum: Studia et documenta 40. 359– 386. Zuazo, Koldo. 2017. Mendebaleko euskara. Donostia: Elkar. Zufferey, Sandrine. 2012. “‘Car, parce que, puisque’ revisited: Three empirical studies on French causal connectives”. Journal of Pragmatics 44. 138–153.

Name Index Abney, Stephen 21 Acedo-Matellán, Víctor 7, 168, 183, 201–202, 205 Ahn, Hee Don 178 Aikhenwald, Alexandra Y. 298 Aissen, Judith 3, 243, 246, 301 Alba-Salas, Josep 178, 184, 193 Alberdi, Xabier 160 Albizu, Pablo 261–262, 305–306 Aldai, Gontzal 63–65, 72, 145, 151, 178 Alexiadou, Artemis 5, 41, 82–84, 99, 108, 113, 119, 124–125, 130–131 Alibèrt, Lois 319 Allières, Jacques 2, 77 Alonso-Ramos, Margarita 185, 192–194 Altube, Seber 42, 47, 49–50 Anagnostopoulou, Elena 5, 7, 41, 82, 84, 94, 107, 111–113, 124 Aranovich, Raúl 158 Arnold, Douglas J. 101–102 Arraztio, Kontxi 250, 267 Arregi, Karlos 44, 223, 235 Artiagoitia, Xabier 22, 110, 321–322, 337 Austin, Jennifer 204 Azkue, Resurrección María 324

Bosse, Solveig 4, 8, 221–223, 225, 231–232, 235, 237–240 Bossong, Georg 3, 243, 246, 250 Bowern, Claire 176, 178 Brody, Michael 204 Bruening, Benjamin 4, 8, 82, 84, 124, 221– 223, 225, 232, 235, 237–240 Brugè, Laura 251 Brugger, Gerhard 251 Burzio, Luigi 7, 139, 143, 154 Butt, Miriam 176, 178–180, 215 Bybee, Joan 60, 67, 73 Bye, Patrick 204

Baker, Mark 86, 204, 279, 281 Baños, José Miguel 330 Barrie, Michael 204 Basilico, David 168 Batlle, Mar 158 Belletti, Adriana 7, 139, 154 Bennet, Paul A. 18 Bentley, Delia 139, 155 Berro, Ane 4, 5, 7, 14, 25, 83, 106, 110, 131, 145, 148–149, 151–152, 154, 162–163, 167, 205, 212, 234, 262, 278, 291, 299 Bianchi, Valentina 44 Birabent, Jean-Pierre 69, 77–78 Bittner, Maria 139 Bobaljik, Jonathan D. 146, 204 Bordelois, Ivonne 284 Borer, Hagit 4, 222, 230, 235 Bosque, Ignacio 5, 16, 34, 83–84, 90, 95, 105–108, 111–114, 117, 122, 127, 198

Dahl, Östen 67 Danlos, Laurence 192–193, 293 Dannenberg, Clare 226 Dasher, Richard B. 335 Davidson, Donald 148 Dayal, Veneeta 131 de Miguel, Elena 221, 229 de Rijk, Rudolf P.G. 15, 17, 23, 27–31, 34, 36– 37, 39–40, 87, 89, 92, 110, 199, 203, 262, 277, 299, 310 Demonte, Violeta 5, 26, 83, 261–264, 268 Detges, Ulrich 93 DHM 279, 281 Dixon, Robert M.W. 8, 139, 276, 298 Doron, Edit 82, 165, 168 Dryer, Mathew 26, 29, 33 Ducrot, Oswal 337 Duguine, Maia 48

Carlson, Greg 102–103, 131 Castillo, María Elena 158 Cattell, Ray 178–179 Chierchia, Genaro 131 Chomsky, Noam 204–205 Cinque, Guglielmo 18–19, 29, 44 Comrie, Bernard 8, 112, 114, 276, 298 Contreras, Heles 41 Cruschina, Silvio 44 Cuervo, María Cristina 140, 166, 168–170, 194, 228, 262

348 EGLU 321, 323, 327–328, 335 Eguren, Luis 86 Elordieta, Arantzazu 42–44, 46, 48, 75, 87, 248 Elordieta, Gorka 44, 46, 75, 248 Embick, David 7, 82–83, 87–88, 90–92, 95, 99, 107, 111–113 Epelde, Irantzu 212 Escandell Vidal, Victoria 44–46 Etxeberria, Urtzi 22 Etxepare, Ricardo 42, 46, 145, 151, 194, 196, 203–204, 207–208, 212, 255, 261, 299 Euskaltzaindia 26–27, 37, 48, 50, 287 Eythórsson, Thórhallur 139, 155 Fábregas, Antonio 90–91, 106, 244, 251, 259– 260 Fernández, Beatriz 4, 8, 14, 24–25, 82, 145– 146, 148, 198, 204, 221–223, 225, 228, 232, 234–235, 243–244, 246–247, 249, 251, 255–257, 261–262, 266–267, 269, 280, 300, 305, 310 Fernández González, José Ramón 319 Fernández Lagunilla, Marina 221 Fernández Ordóñez, Inés 243, 252–253, 255, 260, 312–313 Fernández Soriano, Olga 265 Fleischman, Suzanne 60 Folli, Rafaella 140, 168, 292–293 Franco, Jon 8, 221, 223, 225–229, 231–233, 239, 243, 253 García Fernández, Luis 91 García Murga, Fernando 182, 189, 198, 200, 203–204 Gawron, Jean Marc 106, 108 Gehrke, Berit 5, 7, 83, 84, 112, 120, 128– 131 Geis, Michael 335 Geuder, Wilhelm 178, 181 Giry-Schneider, Jacqueline 185, 193, 201 Goenaga, Patxi 33, 87 Goethals, Patrick 317, 335 Goldberg, Adele 194 Gómez, Rikardo 59, 78 Grafmiller, Jason 49 Greenberg, Joseph 17–18 Grimshaw, Jane 178–190, 193–194, 198

name index Grodzinsky, Yosef 4, 222, 230, 235 Gross, Maurice 192–193 Gutiérrez Ordóñez, Salvador 335 Haase, Martin 66 Haddican, Bill 48 Hale, Kenneth 139–140, 145, 176, 179, 198, 203–204, 206 Harley, Heidi 140, 148–149, 168, 205, 292– 293 Harre, Catherine 73 Haspelmath, Martin 33 Haugen, Jason 204, 206 Heine, Bernd 335 Hernanz, Maria Lluïsa 264 Hidalgo, Bittor 40, 42 Hofmann, Johann Baptist 330 Hook, Peter E. 179, 181 Hopper, Paul J. 179, 181 Horn, Laurence R. 4, 8, 221–223 Hualde, José Ignacio 2–3, 6, 9, 14, 44, 46, 51, 75, 87, 92, 110, 114, 116, 248, 332, 335– 336, 339 Huidobro, Susana 8, 221, 223, 225–229, 231– 233, 239 Iatridou, Sabine 94, 112, 114 Ibarra, Orreaga 248 Irurtzun, Aritz 46–47 Izvorski, Roumyana 112 Jackendoff, Ray 178–179, 189, 198 Jaeggli, Osvaldo 4, 221, 228, 230 Jaxontov, Sergej 93 Jespersen, Otto 178 Jiménez Fernández, Ángel 44, 46 Jouitteau, Mélanie 4, 222, 228, 230, 233 Kayne, Richard S. 18, 24, 139, 155 Kearns, Kate 178–179 Keller, Frank 139, 155 Kennedy, Christopher 97 Keyser, Jay S. 140, 145, 176, 179, 198, 203–204, 206 Kiss, Katalin É. 46 Koizumi, Masatoshi 262 Koontz-Garboden, Andrew 106 Kortman, Bernard 319, 330, 338 Krajewska, Dorota 5, 73, 83–84, 86, 112, 114– 117, 133, 325

349

name index Kratzer, Angelika 7, 82–84, 88, 97, 105, 107, 111–113, 119, 124–125, 141, 143 Labelle, Marie 165, 168 Lafitte, Pierre 40, 47, 204 Lafon, René 2, 71 Lahiri, Aditi 178, 180 Laka, Itziar 43, 145–146, 148, 204, 213, 278, 299 Lakarra, Joseba 72, 75 Landa, Alazne 8, 243, 252–255, 312 Landa, Josu 261–262 Larson, Richard K. 176 Lazard, Gilbert 3 Ledgeway, Adam 60 Legrende, Géraldine 168 Lehman, Winfred P. 17 Leonetti, Manuel 44–46, 243, 251, 259 Levin, Beth 4, 7, 95, 105, 139, 141–143, 146, 152, 160, 179, 194, 204, 278, 299 Levinson, Stephen 335 Lin, Tzong Hong 176, 178 López, Luis 204 Luján, Marta 5, 83 MacKenzie, Ian 139, 155 Maienborn, Claudia 106 Maldonado, Ricardo 221, 223, 225, 228, 233, 237 Manterola, Julen 65 Marantz, Alec 5, 7, 82, 189, 191 Marco, Cristina 5, 7, 83–84, 128–130 Marín, Rafael 5, 83, 90–91, 106 Martinez, Arantzazu 178, 182, 185, 188, 194, 196–200, 203–206, 208, 213–214 Masullo, Pascual 198, 262, 264 Mateu, Jaume 140, 148–149, 157–158, 168, 183 Mathieu, Éric 204 Matushansky, Ora 204 McClure, William 152 McFadden, Thomas 119, 187, 262 McIntyre, Andrew 95, 124–125 McNally, Louise 97 Merlan, Francesca 139 Mester, Armin 178, 179–190, 198 Michelena, Luis [also Koldo Mitxelena] 15, 19, 42, 59, 65, 257 Mithun, Marianne 139, 182, 204, 206

Mohanan, Tara 178 Mounole, Céline 2, 6, 15, 63–66, 68, 71–73, 75, 93, 243, 246–247, 249–250 Nedjalkov, Vladimir 93 Odria, Ane 4, 8, 82, 222–223, 229, 232, 234, 240, 243–244, 246–247, 249, 251, 261– 263, 266–267, 269, 310, 312–313 Ordóñez, Francisco 41 Ormazabal, Javier 251, 261, 303 OEH 20, 34, 257, 276–280, 287 Ortega, Iván 44–45, 47 Ortiz de Urbina, Jon 3–5, 6, 8, 40, 42, 47, 73, 76, 83, 86, 92, 99, 126, 139, 142, 204, 208, 223, 232, 244, 255–257, 261–262, 269, 291, 300, 304–305, 322, 332, 339 Oyharçabal, Bernard [also Beñat Oihartzabal] 4, 29, 31, 87, 145–148, 197, 198, 204, 206, 207, 209, 210–213, 227, 261–262, 285, 287, 291, 293, 305, 325 Palay, Simin 319 Pancheva, Roumyana 112, 114 Parsons, Terence 113 Pastor, Alberto 38 Perez Gaztelu, Elixabete 28 Pérez Saldanya, Manuel 3, 9, 317, 319, 326, 330, 332, 335–338 Pensado, Carmen 243, 251, 258 Perlmutter, David 7, 139, 143, 230, 234, 283, 291, 301 Pesetsky, David 205 Piera, Carlos 200 Pineda, Anna 7, 152, 154, 194, 244, 258, 260– 261, 330 Pinkster, Harm 93 Potts, Chris 222, 235 Pylkkänen, Liina 4, 187, 194, 223, 232, 297, 305–306 Quer, Josep 44 Rafel, Joan 192, 193 Ramchand, Gillian 91, 105, 140, 148, 178, 180–181, 189, 192, 214 Rapp, Irene 120 Rappaport Hovav, Malka 7, 95, 105, 139, 141– 143, 146, 152, 160, 179, 194

350 Rebuschi, George 86 Rezac, Milan 4, 222, 228, 230, 233, 243, 246– 247, 249, 251, 266, 267, 269, 303, 306, 310 Rodríguez, Sonia 182, 189, 198, 200, 203 Rodríguez Mondoñedo, Miguel 251 Rodríguez Ordóñez, Itsaso 243, 246–247, 251, 255 Rohlfs, Gerhard 76, 78 Romero, Juan 251, 261–262, 303 Rosen, Carol 139 Rotaetxe, Karmele 40 Rothmayr, Antonia 106 Rothstein, Susan 262 Sadler, Louisa 101–102 Sainz, Koldo 59, 78 Sainz-Maza, Lorena 45 Salaburu, Pello 142 Salles-Loustau, Jean 69, 77–78 Saltarelli, Mario 37 Samioti, Yota 5, 82, 107, 111 Sankoff, Gillian 155 Santos Río, Luis 317, 336 Sapir, Edward 139 Sarasola, Ibon 24, 145, 204, 257, 262, 280 Schäfer, Florian 5 Shih, Stephanie 49 Siewierska, Anna 40 Sleeman, Petra 82, 95, 99 Smolensky, Paul 168 Sorace, Antonella 7, 139, 155–157, 159, 167 Strozer, Judith 226 Svenonius, Peter 102, 204 Szantyr, Anton 330

name index Tenny, Carol 143 Thibault, Pierrette 155 Torrego, Esther 4, 154, 205, 243–244, 251, 259–260, 294 Trask, Larry 26–27, 34, 59–60, 71 Traugott, Elizabeth 179, 181, 335 Treviño, Esthela 41, 284, 294 Tubino, Mercedes 284, 292, 294–295 Urgell, Blanca 63 Uribe-Etxebarria, Myriam 5, 73, 83, 86, 92, 99, 126, 145–146, 204 Urrutia-Cárdenas, Hernán 252 Varela, Soledad 200 Villasante, Luis 40, 73, 325, 338 Vincent, Nigel 60 von Stechow, Arnim 120 Wasow, Thomas 5, 7, 82, 95, 119 Webelhuth, Gert 226 Wichmann, Søren 139 Wierzbicka, Anna 178 Williams, Edwin S. 262 Yamada, Masahiro 4, 8, 221–223, 225, 232, 235, 237–240 Yrigarai, Ángel 19 Zabala, Igone 73, 89, 145, 182, 188, 190, 192, 194, 197–200, 203–205, 208–209, 263 Zamparelli, Roberto 131 Zuazo, Koldo 340 Zubizarreta, María Luisa 9, 44, 279, 282– 283, 290 Zufferey, Sandrine 337 Zwicky, Arnold 335

Subject Index Boldface numbers indicate (sub)sections where the subject is specifically addressed. absolutive 52, 114 adjectival participle 82, 154 (un)bounded interpretation 127 agent 124, 125, 126, 129, 132 agreement 85 and manner adverbs 121 and temporal modification 120 and verbal participle 82, 84, 95, 116, 119, 124 and Voice phrase 124 as complement of seem verbs 96 aspectual verb classes 105 attributive and predicative 83, 86 compatibility with degree modifiers 97 eventive vs. resultative 90 event-related modifiers 119 experiential interpretation 114 modifying bare N or DP 99 modifying external arguments 103 stative and resultative 107 temporal and locative modification 122 vs. perfective adjective 91 adjectives Adjective Phrase 32 attributive 26 classifying 29 complement and adjunct 33 complex attributive AdjP 35 extraposition 32 perfective 90, 91 pre- and postnominal 101–102 preposed 26, 27 qualifying 29 relational 26 stage-level 89 adverbs activity adverbial 194 modal 323 stative 89 affixation 17 Albanian 235, 237 allocutives 227, 239 animacy 244, 246, 252 a-objects in Sp. 251, 258, 260, 313

and secondary predication 268 as DP or PP 268 Applicative head 222, 232, 256, 305 appositions 18, 27, 31, 35 (non-)restrictive 31 Aragonese 2, 155 Argument Fusion 180 argument structure 180, 186, 190, 194, 224 arguments and participants 194 articles. See determiners aspect 59 (im)perfectivity in past 67 Aspect head 88 aspectual particle 76 Bsq double-compound perfective 65 frequentative periphrases 75 habitual periphrases 75 lexical and grammatical 83, 103 old Bsq perfective forms 63 perfective 282 present perfect 64 progressive 72, 283 progressive and locative marking 66 progressive grammaticalization 72 resultative periphrasis 75 auxiliary 180, 287 aux alternation 119, 154, 155, 224 aux alternation hierarchy 155 aux selection 59, 63, 66–68, 78, 139, 140 preposing 43, 47 semiauxiliary 54. See also modal particles Aymara 246 bare nominals 146, 148, 182, 195, 200–201 Basque Spanish 132, 243, 252, 260, 268, 312– 313 benefactive 62, 65 Bengali 180 Biscayan 21, 54, 68, 72 Case 148, 181, 204–205, 213 Catalan 74, 76, 78, 140, 155, 158, 183, 185, 187, 202, 330, 333, 338

352 causal clauses 317 and modal 337, 340 causal subordinators 318 content causals 317, 331 epistemic causals 317, 323 external 317, 323, 324, 331–332, 335 internal 317, 319, 320, 323, 330 rhematic 320, 325–326, 336 speech-act causals 317, 331, 336 thematic 320, 325–326, 332, 335 thematic and rhematic 317 causatives 95, 208, 212, 263, 276 affix as a verb 277 and infinitive adjacency 281 animate causee 311 archaic affix 277 aspect and passive complement 282 causee and ethical datives 305 causes and agents 292 clausal complement 279 direct and indirect 284 impersonal 283, 289, 301, 304, 309 indirect causation 279 mono- or biclausal 283 negated complement 282 nominalized complement 280 of abs/dat verbs 301 of bivalent unergative 300 of dativeless verbs 298 of ditransitive verbs 306 of light verbs 278 of monotransitive 299 of simplex unergatives 299 of unaccusative 312 passivization 283, 293 tensed complement 285 Chinese 178 clause union 282, 283 clitics 225, 312 aspectual 221 climbing 282, 301 Clitic Left Dislocation 41 cluster 226, 229 doubling 222, 227, 230, 232, 253 encliticization 327, 328 non volitional and dative clitic 221 comparatives 288 complementizers 77, 319, 324, 328, 337, 341 bait- 30, 325, 338

subject index causal 333 -en 322, 325, 333, 337–338, 340 -la 69, 321, 323, 325 que 323, 326 completive clauses 325 compounds 14–16, 26–28, 279 conditional clauses 49, 69, 334–335 conjunctions 324 adversative 328 and wh-words 323, 328–330, 338–340 causal 328 of modal origin 338 of temporal origin 333 coordinating 319, 324, 327 copulative 87 disjunctive 328 modal 337–338 consecutive clauses 325 converbs 86 copular verbs 92, 116–118, 126, 158 copular verb vs. auxiliary 92, 117 transitive copulas 93 with adjectival vs. passive participle 95 with resultative participle 73, 83 datives 4, 24–25, 187, 191–192, 194, 208, 221, 243 (non)argumental 231 adnominals 24, 262 affected 4 affected experiencer 221, 223–224, 235, 237–238 agreement dropping 232, 245, 261–262, 301 alternation with absolutive 256 and secondary predicates 262 as DP or PP 244, 261 aspectual 226, 229 attitude holder 235 benefactive 221–223, 229–230, 232–233, 235, 238, 239 binding a variable 236 causee 244, 262–263, 268, 276, 297, 309 combinations 227, 229, 233 double dative 234, 262, 300, 302, 305– 306, 311 ethical 4, 221, 305, 308, 311 negation 237 person restrictions 221, 227–229

353

subject index experiencer 177, 222, 230, 232, 240, 244, 261–262, 265, 284, 288, 301 external possessor 221–223, 228, 230, 232–233, 235, 239, 240, 244, 261, 262, 264 goal 233–234, 244, 262, 301, 303, 309 high and low 223 in bivalent unergative 269 in unergative and ditransitive 256, 260 malefactive 225 subject co-referential 235, 239 typology 239 degree words 32, 33–34, 36, 38, 98 degree adverbs and PP s 34 extraction 39 demonstratives. See determiners derivation 17, 28 destinative 62 determiners 21, 37, 198, 213 and predicative agreement 65 definiteness 38–39 demonstratives 21, 198 indefinite 32, 36 disjoint reference 125 DOM 243, 299, 306, 310 and dative marking 246 and secondary predication 266 and tense 251 animacy 250, 257 human objects 247 in causatives 310 marking 244 objects and goal IO 256 person 247, 252 position of DOM objects 246 specificity 249, 257 stigmatization 247, 310, 312 vs. dative in bivalent unergatives 255 Dutch 99, 152, 155, 291 English 60, 63, 76, 88, 94, 96, 99, 101, 102, 107, 119, 124, 178–180, 185, 206, 226, 229, 262, 283, 289, 302, 310, 318, 323, 328, 330 ergative 126, 131, 142, 190 ergativity and telicity 152 semantic split 139, 151

event 180, 189, 198, 237, 297 bare event noun 180 configuration and theta-role 150, 167 configurations 140, 148 event structure 186 events and states 148 flavor 149, 153, 169 in light verb constructions 176 kinds and tokens 129, 131 exclamatives 45 experiencer subject 177 extraposition 34, 37 final clauses 323, 333 focalization 42–44, 207, 209, 322, 330 contrastive 45, 48 corrective 42, 47, 52–53 focal stress 44, 46 Focus Fronting 44 Mirative Focus 44–46 presentational 52 Quantifier Fronting 44, 46 verb focalization 48 Verum Focus 44, 46 Galician 155 Gascon 2, 64, 66, 69, 76–78, 319, 334 genitive 21–24, 329 and future tense 62, 65 German 63, 84, 119–120, 122–124, 129–132, 155, 231, 235, 237, 262, 328 Gipuzkoan 34, 38 goal 24, 256, 261 grammaticalization 59–60, 63–65, 68, 72, 73–76, 116–117, 179–180, 317, 318–319, 326–327, 333, 335, 337–338 Greek 84, 94, 99, 107, 123, 124, 131–132 Guarani 246 heaviness 14, 49 Hebrew 222, 235 Hindi 178, 245 Hungarian 21, 278 imperatives 69 impersonal 296 incorporation 146, 202, 204, 213 and phonology 210 inference 317, 326–327, 335, 338

354

subject index

instrumental 333–334 interrogatives 42, 209, 235 multiple wh-Q 40 question words 319 wh-extractions 235 wh-words 43 yes/no 50 intonation 49, 317 inversion predicates 232, 288, 301 Italian 19, 140, 154–158, 281, 283, 292– 293

modification with manner adverbs 88 resultant vs. target state 83 resultative vs. stative 83 stative and resultative 87 stative participle vs. adjective 88 target state 113 temporal and spatial modification 83 noun complement clauses 322 noun modifiers 23

Japanese 178, 190, 198, 262

palatalization 327 participles 61 adjectival. See adjectival part. agreement 64, 86 aspectual interpretation 111 attributive and predicative 117 attributive vs. predicative 99 ending attached to V, not Root 110 experiential 104, 112 imperfective 2, 61, 63, 66 inflected experiential perfective 73 non-verbal. See non-verbal participle perfective 2, 26, 61, 64, 85 possessive resultative 93, 116 prospective 62, 65, 74 resultative 5, 60, 64–65, 73, 85, 111 copula with result. participle 84 suffixes 86 temporally (un)bounded 106–107, 113 resultative vs. continuous interpretation 112 particles 76 affirmative 78 epistemic modality 77 hearsay information 77 in Gascon 77 interrogative 77 modal 76 partitive 142, 147, 200, 207–208 passives 92, 283–284, 294 adjectival and verbal 119 in causatives 282 lexical and phrasal adjectival pass 119 passive agent 99, 289, 291. See also adjectival part. passive small clause 94

Korean 178 Latin 14, 59, 60–61, 65, 68, 72, 330, 333, 338 leísmo 243, 252, 260, 312 Lexical Conceptual Structure 179 locational nouns 17 low vowel raising 327 mediopassives 283, 289, 295–297 mood 59, 64 Basque indicative forms 63 exhortative and imperative 69 modality need, must, want 73 potential 70, 279 subjunctive 67, 279, 280, 284 and (im)perfective 68 diachrony in Basque 68 use in Bsq and Rom 68 morphological case 245 Navarro-Aragonese 319 negation 43 new speakers 53, 314 nominalization 323 nonverbal participles 84 adjectival. See adjectival participle adjectival and adpositional 83 adpositional 87 adverbial 86 and verbal projections 83 as complement of creation verbs 88 continuous interpretation 112 expressed agent 83

Occitan 65, 69, 154, 319, 330, 333, 338

355

subject index pattern convergence 17, 19, 25–26, 35, 42, 46, 53, 59, 64–65, 71, 77, 159, 162, 164, 167, 170, 185, 212, 243, 255, 276, 287, 305, 310, 312–313, 324, 337, 340–341 Persian 180 pluperfect 64–65 polarity 33, 37, 45, 48 Portuguese 76, 155, 333, 338 predicative agreement 34, 38 prefixation 17 prosody 327 Punjabi 246 purpose clauses 125 quantifiers 21, 22, 34, 38, 54, 198 Quechua 278 reanalysis 28, 206, 334 relational nouns 16 relational suffix 23–24, 62, 87, 321, 323 relatives 29, 40, 325, 326 appositive 31 extraposition 31–32 non-restrictive 30 participial 87 pre and postnominal 29 pronouns 30 reduced 28 Rumanian 78, 338 scrambling 48–49 secondary predication 95, 244, 261 Semitic 246 Siberian Koryak 204 specificity 244, 246, 253 and definiteness 249 State phrase 149, 166, 169 stimulus 178, 192–194 subordinator. See complementizer Syntactic Argument Structure 179 TAM Spanish and Basque chart 66–67 telic verbs 91, 156, 168 entailments 153 tense 59 compound tenses 61 future and epistemic modality 77 future periphrasis 61, 76

Romance diachrony 60 simple past 64 theta-roles 150, 179, 187, 292 topic 41 contrastive 41 topicalization 48 truth-conditional meaning 222–223, 227, 235 Urdu 176, 178 verb classes appearance and existence 158 causative alternation 142 change 169 change of location 142, 155 change of state 92, 112, 115–116, 141–142, 152, 156, 158, 162, 168 existence and continuation of state 157–158 internal causation 143, 146, 152, 160, 162 motion 168 psych verbs 183, 222, 232, 284 stative 159 verbs bivalent unergative 243–244, 255, 259, 260, 300 ditransitive 51, 224, 234, 237, 244, 246, 260–262, 270, 300, 303 infinitive 63 inversion. See inversion predicates light 169, 176. See also complex unergative and argument structure 177, 179 and heavy counterparts 180, 188–189 basic repertoire 185 causative 278 different verb, same NVE 186 fossilized 178, 214 list of egin LVC s in Basque 182 list with verbs other than egin 182 non-separable units 184 non-verbal element 194 noun as DO 205, 207 quantification 198, 201–202, 210 reanalysis 206 referentiality of nominal 197 V N adjacency 209 V N cohesion continuum 213

356 vs. synthetic 178, 200, 203 vs. transitive constructions 204 vs. Vague Action Verbs 179, 199 weather 190, 197–198 without faire and hacer 184 periphrases 75, 250. See also aspect radical of Bsq verbs 62–63, 67–68, 277, 279 stative 140, 167, 285 stative predicative constructions 163 statives and theta-role 167 synthetic 51, 71, 78 vs. compound forms 71 unaccusative 41, 103, 139, 142, 166, 290– 291, 295, 297–298, 308 argument role 141 case and auxiliary 140 Event and State phrases 168 telicity 168 verb classes 141

subject index unergative 104, 139, 203, 212, 290–291, 298 and telicity 143 and Voice phrase 141 built on light verbs 144 case and auxiliary 141 complex 145 causative 278 semantic classes 145 dialectal variation 150, 152, 160 initial arguments 143 stative 162 verb classes 143 weather verbs 146, 286 Voice phrase 141, 149, 166 word order 14, 233, 307–309 neutral 40