Information Structure in Indigenous Languages of the Americas: Syntactic Approaches 9783110228533, 9783110228526

Table of contents :
Frontmatter
Contents
Introduction
Language contact and pragmatic notions: Tariana in its multilingual context
Communicative structure in Lushootseed syntax: Thematicity and focalization
Null subjects in Shipibo switch-reference systems
Cartography of expanded CP in Kuikuro (Southern Carib, Brazil)
On word order and information structure in Yaqui
On the nature of word order in Yucatec Maya
The structure of CP in Karaja
Agree and the licensing of wh-words and polarity sensitive items in Southern Quechua
Backmatter

Citation preview

Information Structure in Indigenous Languages of the Americas

Trends in Linguistics Studies and Monographs 225

Editor

Volker Gast Advisory Editors

Walter Bisang Hans Henrich Hock Matthias Schlesewsky Founding Editor

Werner Winter

De Gruyter Mouton

Information Structure in Indigenous Languages of the Americas Syntactic Approaches

Edited by

Jose´ Camacho Rodrigo Gutie´rrez-Bravo Liliana Sa´nchez

De Gruyter Mouton

ISBN 978-3-11-022852-6 e-ISBN 978-3-11-022853-3 ISSN 1861-4302 Library of Congress Cataloging-in-Publication Data Information structure in indigenous languages of the Americas : syntactic approaches / edited by Jose´ Camacho, Rodrigo Gutie´rrez-Bravo, Liliana Sa´nchez. p. cm. ⫺ (Trends in linguistics. studies and monographs; 225) Includes bibliographical references and index. ISBN 978-3-11-022852-6 (hardcover : alk. paper) 1. Indians of South America ⫺ Languages ⫺ Grammar. 2. Indians of North America ⫺ Languages ⫺ Grammar. 3. Indians of South America ⫺ Languages. 4. Indians of North America ⫺ Languages. 5. Discourse analysis. 6. Language and culture. I. Camacho, Jose´. II. Gutie´rrez Bravo, Rodrigo, 1972⫺ III. Sa´nchez, Liliana. PM5008.I54 2010 498⫺dc22 2010022934

Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de. ” 2010 Walter de Gruyter GmbH & Co. KG, Berlin/New York Printing: Hubert & Co. GmbH & Co. KG, Göttingen ⬁ Printed on acid-free paper Printed in Germany www.degruyter.com

Contents

Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . José Camacho, Rodrigo Gutiérrez-Bravo and Liliana Sánchez

1

Language contact and pragmatic notion: Tariana in its multilingual context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Alexandra Y. Aikhenvald

17

Communicative structure in Lushootseed syntax: Thematicity and focalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . David Beck

39

Null subjects in Shipibo switch-reference systems . . . . . . . . . . . . . . . . . José Camacho and José Elías-Ulloa

65

Cartography of expanded CP in Kuikuro (Southern Carib, Brazil) . . . Bruna Franchetto and Mara Santos

87

On word order and information structure in Yaqui. . . . . . . . . . . . . . . . . Lilián Guerrero and Valeria A. Belloro

115

On the nature of word order in Yucatec Maya . . . . . . . . . . . . . . . . . . . . Rodrigo Gutiérrez-Bravo and Jorge Monforte y Madera

139

The structure of CP in Karaja . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 Marcus Maia Agree and the licensing of wh-words and polarity sensitive items in Southern Quechua. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 Liliana Sánchez Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . 213

Introduction José Camacho, Rodrigo Gutiérrez-Bravo and Liliana Sánchez

1. The study of Information Structure The study of the interface between syntax and information structure has attracted a great deal of attention since the publication of foundational works on this subject such as Vallduví (1992) and Lambrecht (1994), which together can be taken to be the pinnacle of continuous interest and research on information structure throughout the twentieth century. Furthermore these works and others provided characterizations of the interaction between discourse and syntax that were precise enough to allow the investigation of this phenomenon in a wide variety of theoretical frameworks. This in turn has lead the contemporary boom regarding the interest on the interaction between syntax and information structure. This book inserts itself in this contemporary interest on the interaction between discourse and syntax by providing a collection of articles on different aspects of the syntax-pragmatics interface based on data from indigenous languages of The Americas of diverse genetic affiliation. Before presenting these articles, we provide a brief introduction of the some of the basic descriptive issues addressed in them, and of some of the theoretical tools that have been developed to analyze them. The brief description that follows of the constructs of information structure and the way they interact with syntax is by no means intended to be exhaustive. For full details, the reader is referred to the specific references cited in the articles in this volume. 2. Basic descriptive notions Many theories of information structure consider that the primary distinction made in information structure is the distinction between topic and focus. Focus (also referred to as rheme in some works) is generally understood to be the part of a sentence or utterance that is not presupposed in the discourse. Although often characterized as “new information”, it is important to note that this does not imply that the referent of the focus be discourse-new.

2

José Camacho, Rodrigo Gutiérrez-Bravo and Liliana Sánchez

Discourse-old referents can function as foci as long as they are not presupposed. The standard diagnostic for identifying the focus of a clause is to take the focus to be the constituent in the answer to a wh-question that corresponds to the wh-operator in the question, as in (1). Henceforth, we represent the extension of the focus in square brackets and intonational prominence in small caps. (1)

a. Who screamed? b. [JOHN] screamed.

In many languages, focus is also closely correlated with intonation. As can be seen in the English example (1b) the constituent that corresponds to the focus is also the constituent that bears the nuclear pitch accent of the clause. In some languages like English, the need to signal the focus with the nuclear pitch accent brings with it a perturbation of the unmarked intonational pattern of the clause. In other languages, like Spanish, the intonational pattern of the clause remains undisturbed, and the focus must instead be displaced to the canonical position where the nuclear pitch accent is assigned. This is illustrated for Spanish in (2), where the otherwise preverbal transitive subject is displaced to the clause-final position where the nuclear pitch accent is assigned. We will see in what follows and throughout the works in this volume that syntactic displacement is a common property of both foci and topics, and hence that the study of word order is inherently linked with them. (2)

a. Quién compró los libros? ‘Who bought the books?’ b. Los compró [ UNA MUCHACHA]. ACC-CL bought a girl ‘A [GIRL] bought them.’

While focus can be understood as information that is not presupposed, a defining characteristic of topics is that they correspond to presupposed or old information. Topics can be classified in two different kinds, sentence topics and discourse topics. Sentence topics (also referred to as themes in some works) are full referential XPs that appear in the left edge of the clause. They typically establish a pragmatic relation of aboutness which holds between the topic and a proposition, such that a sentence topic can be understood as the syntactic constituent which the proposition is about. Hence in (3) the

Introduction

3

nominal expression the dogs functions as a sentence topic in that it is the referent that the rest of the proposition is about. (3)

The dogs, they really miss you.

Discourse topics, on the other hand, represent a completely different kind of phenomenon. Discourse topics can be understood as entities that have been introduced in the discourse and that continue to function as the topic of various clauses that together constitute a larger discourse unit. In sharp contrast with sentence topics, discourse topics are typically realized as weak pronouns, null pronominal elements such as pro, and in some instances they have no other realization other than verbal agreement morphology. This property of discourse topics can be illustrated with a pro-drop language like Spanish. In (4), (from Gutiérrez-Bravo 2008) once the referent Ana is introduced in the first clause, further reference to it cannot be established by means of full pronouns. Instead, pro must be used. (4)

Anai visitó la catedral esa mañana. Pasadas unas horas Ana visited the cathedral that morning after some hours øi/*ellai se aburrió y øi/*ellai fue a visitar a pro/she CL got.bored and pro/she went to visit ACC una amiga. a friend ‘Ana visited the cathedral that morning. After a few hours, she got bored and she went to visit a friend.’

In this way, contemporary research on discourse topicality has provided new insights into the pragmatic conditions that license pro and other null categories. A property of information structure that runs along the topic-focus articulation is the contrastive feature (Halliday 1967, Chafe 1976, Kiss 1998). Accordingly, a number of works have proposed the existence of contrastive topics and contrastive foci. In these works, contrastiveness is taken to be a primitive of information structure, albeit not an obligatory one. Kiss (1998), for instance, defines contrastive focus as a focus that expresses exhaustive identification. In the Hungarian example in (5), from Kiss (1998: 247), the indirect object MARINAK in (5) has been displaced from its canonical postverbal position to the position immediately to the left of the verb, a behavior characteristic of contrastive foci in this language and many others. This example expresses the exhaustive identification characteristic of

4

José Camacho, Rodrigo Gutiérrez-Bravo and Liliana Sánchez

contrastive focus, in that it means that of the set of individuals present in the domain of discourse, it was Mary and no one else that Peter was introduced to by the speaker. (5)

Tegnap este [MARINAK] mutattam be Pétert. last night Mary.DAT introduced-I PERF Peter.ACC ‘It was [TO MARY] that I introduced Peter last night.’

The notion of contrastive topic is considerably more obscure and still subject to much debate and investigation. It seems to us, however, that a particularly illustrative instantiation of this phenomenon is provided in Erteschik-Shir (2007) with the data in (6). In B’s answer the subject JOHN is identified as a topic and not a focus by virtue of being part of the presupposition, and syntactically, by being the subject of an individual level predicate (see Erteschik-Shir (2007) for further details). (6)

A: Tell me about your brothers John and Bill. B: JOHN is the smart one.

In contrast, other works like Vallduví (1992) and Lambrecht (1994) do not consider contrastiveness as an Information Structure category. Instead they consider possible contrastive readings as the result of specific inferences and conversational implicatures between speaker and hearer. It is perhaps worth mentioning, though, that this notion of contrastiveness is at odds with the fact that numerous languages do have grammatical means to code and distinguish contrastive topics and foci. Finally, a phenomenon that has only recently been fully incorporated in the study of the interaction between syntax and discourse is evidentiality, which is of particular relevance to some of the articles in this volume. Evidential markers are found in many Amerindian languages and they indicate the source of the evidence for the proposition (direct visual, or auditory evidence, or hearsay indirect evidence). The following example from Ecuadorian Quechua (Muysken 1977: 27) illustrates an instance of the grammatical expression of hearsay information that is commonly found in languages with evidentiality systems:

Introduction

(7)

5

Kaya shamu-nga-shi. Tomorrow come-3FUT-EVID (hearsay) ‘They say he’ll come tomorrow.’

In (7), verbal inflection is followed by a suffix that indicates the speaker’s perspective on the source of evidence for the propositional content of the sentence. As will be discussed in more detail in the following section, Cinque (1999) further notes that evidentiality markers are found crosslinguistically not only as verbal inflection suffixes. They can also be found as adverbs and as particles, with striking similarity regarding their relative ordering.

3. Information Structure and Syntactic Theory Different theories of syntax have different ways to incorporate the constructs of information structure discussed above. We now present a very brief summary of the main tendencies, with a specific focus on generative theories. Since the late 1980s, one of the concerns of Generative Grammar has been to explore the finer details of clausal structure. In order to do so, researchers have focused on three aspects: a) the position of adverbs relative to other syntactic elements (verbs, nominal arguments), b) the relative scope of certain items over others, and 3) the available grammatical ordering of constituents. A number of important works have related these aspects with the constructs of Information Structure sketched out in the preceding section. One proposal that explicitly addresses the displacement that topics and foci are subject to is Rizzi (1997). This influential proposal distinguishes three broad areas: the lower VP-area, which can be labeled the thematic domain, the intermediate IP-area (the tense-agreement domain), and the higher CP-area (the discourse-oriented domain). The CP-layer, which has been termed the left-periphery, contains topic, focus, and a wh-projections, but in addition, it also includes mood and modality, evidentiality, and a speaker-oriented projection. Specifically, Rizzi’s proposal for a more articulated version of the CP domain includes Force Phrase, Focus Phrase, a recursive Topic Phrase and Fin Phrase: (8)

[ForceP [TopicP [FocusP [TopicP [FinP [IP…]]]]]]

Force Phrase is the locus where the full clause is interpreted as a declarative, interrogative, exclamative, relative clause, etc. The specification

6

José Camacho, Rodrigo Gutiérrez-Bravo and Liliana Sánchez

of this higher projection can be selected by an external head: for example, wonder selects for questions. In (8), Topic is recursive, whereas Focus is not. Rizzi suggests a number of reasons for this distinction, among them that Focus is quantificational whereas Topic is not. In (9a), we see an example of two recursive topics around a focused element (from Rizzi 1997, ex. (23)). In (9b), on the other hand, we see that the focused constituent cannot precede a wh-word (located in ForceP). (9)

a. A Gianni, QUESTO, domani, gli dovrete dire. To Gianni this tomorrow CL should tell ‘Gianni, you should tell him [THIS], tomorrow.’ b. *A GIANNI che cosa hai detto (, non a Piero)? TO GIANNI what thing have tell (, not to Piero)

Finally, FinP is the projection that determines whether IP is finite. This distinction between finite and non-finite correlates with a number of properties across languages. For example, finite forms tend to manifest mood distinctions, agreement, and tense, whereas non-finite forms usually lack mood and agreement, and have a much more rudimentary tense system. Other research has suggested the existence of a lower focus position. For instance, Bosque (1999) has proposed a Focus position above VP for clefts in Caribbean Spanish (see also Camacho 2006), and Jayaseelan (1996, 2001) proposes IP internal topic and focus projections on the basis of evidence from the syntactic position of wh-words in Malayalam. Along these lines, Cozier (2006) suggests a similar position for clefts in Trinidad Dialectal English, and Belletti (2005) has argued for the existence lower focus and topic positions in Italian. In contrast with the cartographic approach of Rizzi (1997) (and of Cinque 1999, to be discussed below), other frameworks do not resort to specific projections to host these discourse categories. Rather, most nontransformational frameworks consider that the behavior of topics, foci and other discourse categories is regulated by well-formedness conditions that operate on output representations. For instance, in these frameworks the surface position of topics and foci is regulated by a set of properties that operate simultaneously on the clause as a whole, such as intonation, linear precedence, flexibility of Case assignment to the subject, etc. In Optimality Theory, for instance, the left-peripheral position of sentence topics is taken to be the result of the constraint in (10), from Costa (1998) and redefined in Gutiérrez-Bravo (2005).

Introduction

7

(10) TOPIC FIRST Topics are sentence initial: Violated once by every head Y or every ZP that c-commands a topic. The effects of the constraint in (10) are quite different from what is observed in the cartographic approach. Since the topic is merely required to c-command every head and every other maximal projection in the clause, there can be multiple different structures that satisfy this well-formedness condition. This is illustrated in the four structures in (11), where the sentence topic occupies a different position in each case. (11) a.

IP XP [topic] [Spec, I]

c.

b. I’

I

XP VP

CP XP [topic] [Spec, C]

IP

C

CP XP

IP

I’

[topic] [Spec, I]

d. C’

IP

[topic]

CP C’

[Spec, C]

C

IP

Analyses in OT exploit the flexibility of this kind of approach to account for the variable distribution that sentence topics can display not only crosslinguistically, but often also in any one specific language. Turning to other discourse categories and their interaction with syntax, another very influential paper in the generative tradition is Cinque (1999), which, just like Rizzi (1997), relies on a highly detailed clausal template. Cinque’s study reaches its conclusion based on two types of evidence: first, the relative ordering of morphemes in highly agglutinative languages, and second, the relative scope-ordering of different types of adverbs. Cinque assumes the underlying idea of Baker’s (1988) Mirror Principle, which suggests that the ordering of affixes in a verbal complex may provide evidence for a hierarchy of functional projections. Based on these two types

8

José Camacho, Rodrigo Gutiérrez-Bravo and Liliana Sánchez

of evidence, he proposes a hierarchy of functional projections that includes at least the following categories in the higher portion of the clausal tree: (12) Moodspeech act[Moodevaluative [Moodevidential [Modepistemic [Tpast [Tfuture [Moodirrealis [Modnecessity [Modpossibility [Asphabitual…]]]]]]]]]] In (13), we see an example from Korean (Cinque 1999: 71) showing the evaluative mood morpheme {-kwun} closer to the verb root than the speech act morpheme {-a}. In (14), we see an example of adverb ordering in English that matches the hierarchy in (12). (13) Ku say-ka cwuk-ess-keyss-kwun-a that bird-NOM die-ANT-EVALUAT-DECL ‘That bird must have died!’ (14) a. Frankly, John surprisingly doesn’t have a clue! b. *Surprisingly, John frankly doesn’t have a clue! Some of the projections included in Cinque’s template relate specifically to discourse-oriented content (evaluative mood, evidential mood, speech act), although he does not include focus and topic phrases in the general architecture of functional projections because these items offer some variability across languages. The complex structuring of the left periphery initially proposed by Rizzi and by Cinque has allowed researchers such as Speas (2004, 2008) to formalize, in terms of syntactic structure, previous categorizations of linguistic expressions initially thought to belong exclusively to the domain of pragmatics, such as evidentiality and logophoricity. Building on previous work by Willett (1988) on the basic categories of evidentiality found crosslinguistically, Speas (2004, 2008) has proposed that there is a crosslinguistic hierarchy of evidential categories that corresponds to a hierarchy of logophoric categories (Culy 1994). Speas proposes that these hierarchies can be linked to the left functional projections proposed by Cinque. The two hierarchies and their correspondence to Cinque’s proposed functional heads are shown in (15), based on Speas (2004: 264). (15) Cinque’s projections Speech act Evaluative Evidential Epistemological

Evidential Hierarchy say think know perceived

Logophoric Hierarchy hearsay less valuable evidence direct evidence unquestionable evidence

Introduction

9

Speas notes that the parallelism between the evidential and logophoric hierarchies suggests that these categories are more than linguistic expressions that correspond to a wide range of pragmatically-determined possibilities to express sources of evidence. They seem to constitute a limited set of primitives that enter into syntactic relations such as binding among themselves that indicate their relative position in the structure. On the basis of evidence form languages such as Tibetan, Sherpa and Maricopa, Speas shows that these categories also enter structural relations with other categories such as person, tense, and switch reference morphemes. These approaches have shown that the study of the interaction between what were once thought of as linguistic expressions related only to the pragmatic component of language and sentence-internal structural relations, is actually of crucial importance to the understanding of how information structure is expressed in syntax and morphology. The articles in this volume provide evidence from indigenous languages of the Americas that are rich in syntactic and morphological expressions of many of these categories, and thus put to test some of these theoretical approaches.

4. The contents of this volume The present volume addresses the variety of topics presented above. We have aimed for maximum diversity in putting this volume together, and so the articles contained here provide data and analysis from indigenous languages of the Americas of very diverse genetic affiliations. Furthermore, we have also aimed for diversity in the way in which the articles contained here address the phenomena sketched out in this introduction. Hence the reader will find articles that focus mostly on empirical issues, while others are mostly oriented to theoretical issues. The reader will also find diverse theoretical approaches in the latter articles, including Minimalism, Optimality-theoretic syntax, and Meaning-Text Theory. We conclude this introduction with short summaries of the articles contained herein. In “Pragmatic notions in Tariana (North-west Amazonia, Vaupés River basin),” Alexandra Aikhenvald addresses the different grammatical means that Tariana uses to encode pragmatic notions, and how these means have been changing through influence from Eastern Tucano. These include pronominal cross-referencing as a means to encode topical and focal constituents, plus clausal word order and differential case marking. In traditional Tariana, the post-predicate position correlates with contrastive focus, while position at the left periphery of a sentence is preferred for

10

José Camacho, Rodrigo Gutiérrez-Bravo and Liliana Sánchez

topics. For innovative speakers, flexible word order is replaced by rigid word order imported from Eastern Tucano, which makes the post-verbal position always right-dislocated (as in Brazilian Portuguese). The author further observes that in Tariana, tense/aspect/evidentiality serve to mark topicality by attaching to the topic, whereas in clauses without topic, these categories instead appear on the verb. In “Communicative Structure in Lushootseed syntax: Thematicity and focalization”, David Beck addresses the relation between clause structure and information structure in Lushootseed (Salishan, Washington State, US). The main conclusion in this paper is that clause structure in Lushootseed is primarily regulated by information structure and not by the projection of the lexical and syntactic properties of specific lexical items. Concretely, Beck argues that the selection of the clausal predicate in Lushootseed is governed by a condition which requires the main predicate to be rhematic, irrespective of its lexical category. Lushootseed, like other Salishan languages, displays a great flexibility with respect to which word classes can function as clausal predicates. It is shown in this paper that this property of Lushootseed, combined with the requirement that the main predicate be rhematic, results in a wide range of non-verbal predicates in this language. The author develops an account of this particular mechanism of predicate selection using Meaning-Text Theory (Mel’þuk 1974, 1981) combined with a ranking of constraints akin to that of Optimality Theory (Prince & Smolensky 2004). In essence, the author argues for the existence of a constraint REN, which requires the predicate to be rhematic. This constraint can sometimes conflict with a constraint VEN that requires the predicate to be lexicalized as a verb. In English, the constraint ranking VEN » REN produces an output where the main predicate is a verb even when it is not rhematic, whereas in Lushootseed the opposite ranking REN » VEN produces an output where the rheme surfaces as the main predicate even if it is not a verb. In “Null subjects in Shipibo switch-reference systems,” José Camacho and José Elías-Ulloa address the split distribution of null subjects in Shipibo, a Peruvian language of the Pano family. In this language, null subjects in 3rd person are possible, but not in 1st and 2nd person. The authors compare Shipibo with other mixed null subject languages such as Hebrew and Finnish, which show the mirror-image pattern (null in 1st/2nd, overt in 3rd). They argue that 1st and 2nd person subjects in Shipibo must be licensed in a Speech Act Participant projection, whereas 3rd persons are licensed as topics. The article further explores the minimal unit for computing null-subject domains: whenever a main clause and an adjunct (switch-reference) clause appear together, it is enough to have a single 1st/2nd person overt subject in

Introduction 11

either one. This follows from the fact that SR-morphology allows the null pronoun to be licensed in the higher clause. In “Cartography of Expanded CP in Kuikuro (Southern Carib, Brazil),” Bruna Franchetto and Mara Santos explore the syntactic structure of discourse-related functions in Kuikuro, an agglutinative and head final language. Kuikuro has a fixed word order, and its case assignment system requires merging between any head and its internal argument as a condition for the assignment of unmarked structural case (absolutive). Additionally, the ergative case of the external argument is lexical and realized by a postposition. Franchetto and Santos analyze the syntax of a number of leftperipheral functions, such as wh-questions, yes/no questions and focus constructions in light of Rizzi’s (1997) cartographic proposal. Specifically, the authors observe that the structure of questions varies depending on which argument is questioned, and they generally involve a wh-marker in CP, or a Q morpheme for yes/no questions. Wh-questions involve a cleft-like structure, with a nominalized verb for S and A, and a de-ergativized form of the verb for O interrogatives. In the case of adverbial adjuncts, the whole constituent is moved (pied-piped) to the beginning of the clause. In the case of focus, Kuikuro also employs a cleft, with the focused constituent at the beginning of the clause, followed by a deictic, a copular verb, and the rest of the clause. As a result of their analysis of Kuikuro, Franchetto and Santos find empirical support for Rizzi’s hierarchy of functional projections that includes: [ForceP [TopP [IntP [ FocP [FinP]]]]. In the paper “On word order and information structure in Yaqui”, Lilián Guerrero and Valeria Belloro describe and analyze a number of word order patterns observed in oral narratives in Yaqui (Uto-Aztecan, North-Western Mexico and Arizona). They further investigate the preferred means observed in this language for encoding the different kinds of foci proposed in Lambrecht (1994). Their results challenge two presumed properties of Yaqui word order; (i) the alleged flexibility of the ordering of its nominal arguments, and (ii) the rigidity of the distribution of its “second position” subject clitics. With respect to the fixed position of “second position” subject clitics, Guerrero and Belloro show that subject clitics in Yaqui cannot be described as second position clitics under any standard definition. The authors show that these clitics appear in clause-initial position almost as frequently as they do in second position. They further show that there are numerous instances where these subject clitics are preceded by two phrasal constituents, thus making the clitic the third phrasal constituent in the clause. With respect to the word order of nominal arguments in this language, Guerrero and Belloro conclude that Yaqui is a robust SOV language where

12

José Camacho, Rodrigo Gutiérrez-Bravo and Liliana Sánchez

word order is mostly determined by the grammatical relations of the verb’s arguments. In contrast, they show that the OSV order (often listed as common in the literature) is highly dispreferred, if not thoroughly ungrammatical. However, the authors show that Yaqui does allow for perturbations of its unmarked SOV order, mostly through right-dislocation of the subject or the object. Right-dislocation thus derives the OVS and SVO orders from the basic SOV order. Guerrero and Belloro note that the postverbal subjects and objects in OVS and SVO must be full, definite, nominal expressions. From this and from their distribution in narratives they conclude that postverbal subjects and objects in Yaqui are antitopics; discourse-old, topical elements with a low degree of information prominence (Vallduví 1992, Lambrecht 1994, 2001). In the article “On the nature of word order in Yucatec Maya”, Rodrigo Gutiérrez-Bravo and Jorge Monforte y Madera address the issue of whether the SVO order in Yucatec Maya (Mayan, Yucatan Peninsula, Mexico) is the result of topicalization of the subject. The authors present evidence from elicitation, corpus data, and a comparison with Tzeltal Maya to show that SVO in Yucatec Maya is not the result of subject topicalization, and that instead this order is the unmarked word order of transitive clauses in the language, contrary to what is often assumed in the literature. However, the authors show that subject initial constructions are only unmarked when both the subject and the object are simultaneously realized as full nominal expressions. In every other instance (intransitive clauses, clauses with a prodropped object, clauses where the complement of the verb is not nominal), the subject does surface in a post-verbal position. This makes Yucatec Maya an almost systematic verb-initial language, but presents the challenge of simultaneously accounting for the subject-initial nature of its transitive constructions. The authors make use of Optimality Theory to account for this split in word order. They show that the standard analysis of SVO languages in OT, derived by the ranking EPP » STAY, does not give the correct result in Yucatec Maya. This is because such a ranking does derive the SVO order of transitive clauses, but also incorrectly gives as a result preverbal subjects in other constructions. Instead, the authors propose that Yucatec Maya has the opposite constraint ranking STAY » EPP, which accounts for the mostly verb-initial character of the language. In their analysis, SVO is instead derived by a constraint DISTINCT, which forbids the simultaneous occurrence inside the VP/VP domain of two phonetically overt phrases with the same category feature. When the DISTINCT constraint is ranked over STAY, the optimal structure is one where the subject DP is jettisoned out of the VP and into Spec-IP in order to comply with the requirements of DISTINCT. This

Introduction 13

derives the SVO order of transitive clauses while deriving at the same time the verb-initial constructions observed in every other context. In ‘The structure of CP in Karaja’, Marcus Maia’ focuses on how the left periphery is structured in Karaja, a Macro-Je stock language. He proposes a unified analysis for interrogative constructions of the Wh and yes/no types, as well as for structures with the conditional operator txibo ‘whether’. His proposal, although based on the original proposal by Rizzi (1997) of a structured CP layer in the left periphery, differs from Rizzi’s in that it proposes that in Karaja a strong wh-feature is located in the head of Finiteness Phrase. Maia provides evidence from locality restrictions for a movement analysis of fronted wh-constituents. He also analyzes them as phrases with internal subconstituents mirrored in the complex morphology of these words. Of particular interest is his analysis of yes/no interrogatives. These involve a Q operator (aõbo) in the specifier of Finiteness Phrase and the obligatory topicalization of some constituent. Evidence for the topicalized nature of this fronted constituent is found in the need for an obligatory resumptive pronoun inside IP in this type of questions. The crucial piece of evidence for a dependency relation between wh-operators in spec of FinP and IP-internal material is the presence of the verbal suffix {-keki}, a morphological correlate of the conditional operator. A similar morphological reflex {-my} is found as a suffix added to subordinate verbs in contexts in which a main verb may select a wh-word. Maia’s proposal also differs from previous proposals found in the literature (Cinque 1999, Speas 2004) in the treatment of the evidential particle ràki, for which he proposes a noncartographic analysis along the lines of a previous proposal by Matthewson, Davis and Rullmann (2007). Finally, the article “Agree and the licensing of wh-words and polarity sensitive items in Southern Quechua” by Liliana Sánchez deals with the interaction between the C-layer and the T-layer. Sánchez proposes an analysis of the dual nature of wh-words in Southern Quechua (as wh-words and as polarity sensitive items) as the result of an Agree relationship triggered either by a C-level or a T-level operator. In her proposal wh-words can be the spell out of Q features when the trigger is a C-level operator or they can be the spell out of non-veridicality features when the trigger is an event-related, non-veridicality operator located in T of the type proposed by Chierchia (2006). In interrogative sentences, wh-words agree with a Q operator in C (more specifically in Speech Act, one of the projections proposed by Cinque 1999). In declarative sentences with non-veridicality operators such as negation, they are Negative Polarity Items or in the case of imperatives or free relatives, Free Choice Items that agree with a T-related

14

José Camacho, Rodrigo Gutiérrez-Bravo and Liliana Sánchez

operator. The availability of operators at different levels gives rise to potential intervening effects. In order to avoid them, wh-movement is required for a Q interpretation. In the analysis proposed by Sánchez, whfronting is a last resort movement required to avoid intervention effects.

References Baker, Mark 1988 Incorporation: A Theory of Grammatical Function Changing. Chicago: The University of Chicago Press. Belletti, Adriana 2005 Extended doubling and the VP periphery. Probus 17: 1-35. Bosque , Ignacio 1999 On Focus vs. Wh-Movement: The case of Caribbean Spanish. Sophia Linguistica 44-45: 1-32. Camacho, José 2006 In Situ Focus in Caribbean Spanish: Towards a Unified Account of Focus. In Selected Proceedings of the 9th Hispanic Linguistics Symposium, Nuria Sagarra and Almeida Jacqueline Toribio (eds.), 13-23. Somerville, MA.: Cascadilla Proceedings Project. http://www.lingref.com/cpp/hls/9/paper1362.pdf Chafe, Wallace 1976 Givenness, contrastiveness, definiteness, subjects, topics and point of view. In Subject and Topic, Charles N. Li (ed.), 25-55. New York: Academic Press. Chierchia, Gennaro 2006 Broaden your views: implicatures of domain widening and the “Logicality” of language. Linguistic Inquiry 37 (4): 535-590. Cinque, Guglielmo 1999 Adverbs and Functional Heads: A Cross-Linguistic Perspective Oxford: Oxford University Press. Costa, João 1998 Word order variation: A constraint-based approach. Ph.D. diss., University of Leiden. Cozier, Franz 2006 The co-occurrence of predicate clefting and Wh-questions in Trinidad dialectal English. Natural Language and Linguistic Theory 24: 655-688. Culy, Christopher 1994 Aspects of logophoric marking. Linguistics 32: 1055–1094.

Introduction 15 Erteschik-Shir, Nomi 2007 Information Structure: The Syntax-Discourse Interface. Oxford: Oxford University Press. Gutiérrez Bravo, Rodrigo 2005 Structural Markedness and Syntactic Structure. London: Routledge. Gutiérrez Bravo, Rodrigo 2008 La identificación de los tópicos y los focos [Identifying topics and foci]. Nueva Revista de Filología Hispánica 56: 363-401. Halliday, M. A. K. 1967 Notes on transitivity and theme in English (Part 2). Journal of Linguistics 3: 199-244. Jayaseelan, K. A. 1996 Question-word movement to focus and scrambling in Malayalam. Linguistic Analysis 26: 63-83. Jayaseelan, K. A. 2001 IP-internal Topic and Focus Phrases. Studia Linguistica 55: 39-75. Kiss, É. Katalin 1998. Identificational focus versus information focus. Language 74: 245-273. Lambrecht, Knud 1994 Information Structure and Sentence Form. Cambridge: Cambridge University Press. Lambrecht, Knud 2001 Dislocation. In Language Typology and Language Universals. Martin Haspelmath (ed.), 1050-1078. Berlin: Mouton de Gruyter. Matthewson, Lisa, Henry Davis and Hotze Rullmann 2007 Evidentials as epistemic modals: Evidence from St'át'imcets. Linguistic Variation Yearbook, Jeroen Van Craenenbroeck, (ed.), 201–254. Mel’þuk, Igor 1974 Opyt teorii lingvistiþeskix modelej “Smysl ֞ Tekst” [An outline of a theory of "Meaning ֞ Text" linguistic models]. Moscow: Nauka. Mel’þuk, Igor 1981 Meaning-Text Models: A recent trend in Soviet linguistics. Annual Review of Anthropology 10: 27–62. Muysken, Peter 1977 Syntactic Developments in the Verb Phrase of Ecuadorian Quechua. Lisse: Peter de Ridder Press. Prince, Alan, and Paul Smolensky 2004 Optimality Theory: Constraint Interaction in Generative Grammar. Malden, MA: Blackwell. Rizzi, Luigi 1997 The fine structure of the Left Periphery. In Elements of Grammar, Liliane Haegeman (ed.), 281-337. Kluwer: Dordrecht.

16

José Camacho, Rodrigo Gutiérrez-Bravo and Liliana Sánchez

Speas, Margaret 2004 Evidentiality, logophoricity, and the syntactic representation of pragmatic features. Lingua 114: 255-276. Speas, Margaret 2008 Review of A. Aikhenvald, Evidentiality. Lingua 115: 754-759. Vallduví, Enric 1992 The Informational Component. New York: Garland. Willett, Thomas 1988 A cross-linguistic survey of the grammaticization of evidentiality. Studies in Language 12: 51–97

Language contact and pragmatic notions: Tariana in its multilingual context Alexandra Y. Aikhenvald

1. What this paper is about Languages of different typological profiles are known to employ a wide array of morphosyntactic devices to encode information structure (thus supporting Rizzi 1997). In languages with flexible, or discourse-dependent, order of clausal constituents, linear position in a sentence or clause often correlates with marking topic or focus (cf. Pereltsvaig 2004). Alternative grammatical means encoding information structure may include particles encoding focus or topic, as in Urarina, an isolate from Peru (Olawsky 2006; also see cross-linguistic discussion in Lambrecht 1994), specialized marking on verbs (e.g. focus systems in Western Austronesian languages: see Dixon and Aikhenvald 1997 for a brief survey), or dedicated person marking such as proximal-obviative in Algonquian languages (e.g., Junker 2004; also see Frascarelli 2000). Valency-changing or valency-manipulating devices, such as applicatives, may have an additional pragmatic function, casting highly topical participants as core arguments (see Mithun 2001, for a general perspective). North Arawak languages of adjacent areas of Brazil, Venezuela and Colombia employ different sets of personal cross-referencing prefixes for grammatical marking of information structure (see Aikhenvald 1995, on Baniwa of Içana, and Bare; Aikhenvald 1998 on Warekena of Xié). Tariana, a North Arawak language closely related to Baniwa of Içana,1 is rather unusual for the Arawak family in that it employs cases on noun phrases as a major means for encoding topical and focal constituents. This is in addition to a plethora of other grammatical means including clausal constituent order, an applicative-like verbal derivation and placement of enclitics. In this paper I investigate the grammatical means for encoding pragmatic notions in Tariana, and outline the dynamics of their development in the situation of intensive language contact with genetically unrelated Tucanoan languages, and the increasingly dominant Portuguese, the local lingua franca. Tariana is the only Arawak language spoken in the multilingual area of the Vaupés River Basin. Its position, and the role of contact-induced change

18

Alexandra Y. Aikhenvald

in its history, are addressed in §2. I then turn to a brief analysis of grammatical means encoding pragmatic notions in the Traditional Tariana (§3). This is accompanied by a comparison with similar patterns in Tucanoan languages. Intensive diffusion of patterns between Tariana and other languages of the area has brought about a plethora of changes in the ways of information structure is marked in the language of innovative speakers (§§45). The last section (§6) contains a summary.

2. Tariana in its areal context The Vaupés River basin in north-west Amazonia (spanning adjacent areas of Brazil and Colombia) is a well-established linguistic area. Its major feature is an obligatory societal multilingualism which follows the principle of linguistic exogamy: ‘those who speak the same language with us are our brothers, and we do not marry our sisters’. Marrying someone who belongs to the same language group is considered akin to incest and referred to as ‘this is what dogs do’. Language affiliation is inherited from one’s father, and is a badge of identity for each person. Languages traditionally spoken in the area belong to three unrelated genetic groups: East Tucanoan, Arawak and Makú. Speakers of East Tucanoan languages (Tucano, Wanano, Desano, and a few others), and of the only Arawak language, Tariana, participate in the exogamous marriage network which ensures obligatory multilingualism (see Aikhenvald 2002, 2006 for further details). A striking feature of the Vaupés linguistic area is a strong cultural inhibition against language mixing: borrowed forms are condemned as tokens thereof. Long-term interaction based on institutionalized multilingualism between East Tucanoan languages and Tariana has resulted in the rampant diffusion of grammatical and semantic patterns (though not so much of forms) and calquing of categories. As a result, the Vaupés area provides a unique laboratory for investigating how contact induced changes take place, which categories are more prone to diffusion, and which are likely to remain intact.2 Tariana is an endangered language, currently spoken by about 100 people in two villages along the Vaupés River (the minor dialectal differences are addressed in Aikhenvald 2003; also see Aikhenvald 2006). At present, Tucano is rapidly gaining ground as the major language of the area, at the expense of other languages in the Brazilian Vaupés (see Aikhenvald 2006, for a summary of historical reasons for this shift).

Language contact and pragmatic notions

19

According to traditional patterns of language use, a father would speak his own language to the members of his immediate family; and a mother would teach her children her father’s language. These patterns of language use are still being maintained by a few surviving older speakers (60-80 years of age) who use a somewhat archaic variety which we call Traditional Tariana. Younger speakers effectively breach this pattern, by using the dominant Tucano language in their day-to-day life. Consequently, the Innovative Tariana spoken by these people displays more Tucano-like patterns in their language than do traditional speakers. These newly introduced patterns reflect on-going changes produced as the result of gradual and imminent shift to the dominant language. Most speakers of Tariana are also highly proficient in Portuguese, the national language of Brazil, conceptualized as the language of prestige and status in the ‘white man’-dominated world (see Aikhenvald 2002: 175-86). The ensuing influence of Portuguese on the indigenous languages — Tariana among them — is reflected in the emergence of new devices for encoding pragmatic notions.

3. Encoding pragmatic notions in Traditional Tariana A major grammatical device for encoding information structure in Traditional Tariana is case marking for encoding topical, and focussed, constituents (§3.1). Both phrasal and clausal constituent order are used for expressing pragmatic functions of a constituent — but in other ways, displaying different limits on flexibility (§3.2). These patterns partly stem from intensive areal diffusion from East Tucanoan languages into Tariana: in §3.3 we show which ones are fully and which ones are partly shared with East Tucanoan languages of the Vaupés River Basin. A word on terminology is in order. Following general consensus, topic is understood as what a stretch of text is about (Cruse 2006: 182-3), or, in Pereltsvaig’s (2004: 327) words, ‘the part of the clause that denotes discourse-accessible information that is the matter of common concern for the speaker and the addressee’. Focus is understood as new, salient, or nonpresupposed, part of a clause or a sentence. Contrastive focus implies picking out an element from a set of alternatives, as opposed to other alternatives (also see Zubizarreta 1998). 3.1. Case marking Grammatical relations in Tariana are marked in two ways. A feature Tariana shares with most Arawak languages is marking grammatical relations with

20

Alexandra Y. Aikhenvald

personal prefixes, roughly on an active-stative basis (see Dixon 1994). Every verbal root in Tariana is either prefixed or prefixless. Prefixed verbs can be transitive (e.g. -pitaneta ‘name someone’), ditransitive (-bueta ‘teach’), ambitransitive (type A = Sa, e.g. -wapa ‘wait, attend to something’, or type O = Sa, e.g. -thuka ‘break’) or active intransitive (Sa, e.g. -emhani ‘walk around’).3 Most prefixless verbs are stative intransitive (e.g. kasitana ‘be annoyed’). A few are A = So ambitransitives (e.g. nhesiri ‘enjoy (not food)’) or O = So ambitransitives (hui ‘enjoy (food); be tasty’) (see Aikhenvald 2003: 235-43 for further examples). There is no object marking on the verb. Both O and So acquire zero-realization (whereas in Baniwa of Içana, a closely related language, both are marked with pronominal enclitics). Grammatical relations are also marked by case enclitics, on a subject/non-subject basis. Case marking of subjects, and of non-subjects — which include direct objects (O), a variety of third (‘extended’) arguments of ditransitive verbs, and numerous obliques, such as recipient, beneficiary, locative, manner, instrument and time — is determined by semantic and pragmatic properties of the constituent, in addition to its grammatical function. The principles of case marking in Tariana are summarized in Table 1. Table 1. Grammatical relations and core cases in Tariana DISCOURSE NOUNS PRONOUNS

GRAMMATICAL FUNCTION

STATUS

subject (A/S)

nonfocussed

focussed non-subject (non nonA/S) topical topical

subject form Subject form (pronominal (noun-Ø) prefix + emphatic formative -ha) subject form + clitic -ne/-nhe4 subject form pronominal prefix + (noun-Ø) object suffix -na subject form + clitic -naku/-nuku5

Examples (1) - (2) illustrate a non-focussed subject as opposed to the subject in focus. (1) appears at the beginning of a narrative. The tapir is not contrasted to any other participant (its sentence-initial position correlates with its being the topic of the stretch of discourse). (1)

Hema hinipuku di-hña-pidana. tapir garden 3sgnf-eat-REM.P.REP ‘A tapir (reportedly) ate (fruits of) a garden.’

Language contact and pragmatic notions

21

In (2), ‘tapir’ — a main protagonist of the story, and the major culprit — is focussed. The subject in (2) has strong contrastive overtones: the tapir — and not a village thief nor an evil spirit — ate the fruits of the garden. (2)

Hema-ne hinipuku di-hña-pidana. tapir-FOC.A/S garden 3sgnf-eat-REM.P.REP ‘A/the tapir (not anyone else) (reportedly) ate (the fruits of) a garden.’

The tapir here is not a new participant: it is effectively being reintroduced as an important player in the subsequent narrative, and at the same time contrasted to other potential thieves. The clause-initial position of this constituent reflects its topicality, while case marking indicates that it is contrastive. In (1) and (2), ‘garden’ is not marked as a topic. A topical non-subject is illustrated in (3) — here, the garden is what is being talked about. A topical non-subject may be definite or indefinite. In (3), the subject is accompanied by a demonstrative determiner (the noun phrase is in square brackets): (3)

Hema [h̙ hinipu-naku] di-hña-pidana. tapir this garden-TOP.NON.A/S 3sgnf-eat-REM.P.REP ‘A tapir (reportedly) ate (fruits of) this garden (we are talking about).’

If the subject is focussed, and the non-subject is topical, a sentence can contain both a marker of focussed subject and of a topical non-subject: (4)

hema-ne hƭ hinipu-naku di-hña-pidana. tapir-FOC.A/S this garden-TOP.NON.A/S 3sgnf-eat-REM.P.REP ‘A/the tapir (not anyone else) (reportedly) ate (fruits of) this garden (we are talking about).’

Tariana does not have any dedicated marking of definiteness (such as definite articles). Depending on the context, the subject in (1) - (4) can be definite or indefinite (independently of whether it is focused or not). The pragmatically and semantically determined marking of non-subjects is reminiscent of differential object marking as described by Bossong (1985, 1991) (also see Aissen 2003 and further references there). However, unlike other instances described there, pragmatically and semantically based marking in Tariana extends beyond direct objects. It also applies to obliques (or adjuncts). The presence of the overt subject marking in Tariana depends on whether or not the subject is focussed. This ‘differential’ subject marking

22

Alexandra Y. Aikhenvald

is superficially reminiscent of semantic marking of the subject (A/S) in Australian languages (see Dixon 2002: 132-3) where volition and degree of agency of the subject correlates with the presence of overt marking. In some such languages, case marking is also used for disambiguating who did what to whom, and may thus acquire an additional highlighting discourse function comparable to the focussed subject marker in Tariana (see Dixon 1994: 2833). Nouns distinguish two additional cases marking obliques — locative -se ‘in, into, from’ and instrumental-comitative -ne ‘with, by’. The latter is partially homophonous with the focussed subject marker whose allomorphs are -ne and -nhe (see (2)). The ‘topical non-subject’ and oblique cases can mark a constituent with the same grammatical function, and can occur together. In (5a), the locational oblique is marked with the locative -se: (5)

a. Hema hinipuku-se di-eku di-a-pidana. tapir garden-LOC 3sgnf-run 3sgnf-go-REM.P.REP ‘A tapir (reportedly) ran into (or in) the garden.’

Using the locative case with nouns which typically refer to location (such as ‘garden’, ‘village’, ‘world’) imparts overtones of definiteness to the noun — this is reflected in the translation of (5a). If the noun is indefinite, the locative case marker is likely to be omitted, as in (5b): (5)

b. Hema hinipuku di-eku di-a-pidana. tapir garden 3sgnf-run 3sgnf-go-REM.P.REP ‘A tapir (reportedly) ran into (or in) a garden.’

In (5c), the garden is topical — the example comes from a story about a man’s garden that was constantly under attack from various creatures, including village thieves, tapirs and evil spirits. The oblique ‘garden’ is definite and is the topic. It then takes both the locative -se and the topical non-subject marker -naku: (5)

c. Hema hinipuku-se-naku di-eku tapir garden-LOC-TOP.NON.A/S 3sgnf-run di-a-pidana. 3sgnf-go-REM.P.REP ‘A tapir (reportedly) ran into (or in) the garden (which we are talking about).’

Language contact and pragmatic notions

23

This constitutes a typologically unusual instance of double marking of syntactic function on one noun phrase (or double case: see Aikhenvald 2002). As can be seen from Table 1, nouns and personal pronouns differ in their principles of case marking. Firstly, nouns in a non-subject function can be unmarked for case (as in 5b), while pronouns are always marked. This is shown in (6): (6)

Di-ka-pidana di-na nawiki-nhe. 3sgnf-see-REM.P.REP 3sgnf-OBJ man-FOC.A/S ‘The man (not anyone else) (reportedly) saw him (the tapir).’

Secondly, the subject form of a pronoun can be analyzed as consisting of a pronominal prefix + a formative -ha — di-ha (3sgnf-FORMATIVE) ‘he’, nu-ha (1sg-FORMATIVE) ‘I’. If a pronoun is marked as a topical nonsubject or as a focussed subject, the marking is attached to the subject form, as in (7). (7)

Di-eku-hu-pidana diha-ne.6 3sgnf-run-AWAY-REM.P.REP he-FOC.A/S ‘He (the tapir, not anyone else) (reportedly) ran away.’

If a pronominal non-subject is topical, it can take the non-subject case marker. An example is in (8). Such examples are rare in texts and conversations, and are highly pragmatically marked (Aikhenvald 2003: 147 provides further examples): (8)

Diha-naku7 kiaku di-kwisa-pidana nawiki-nhe. he-TOP.NONA/S strongly 3sgnf-scold-REM.P.REP man-FOC.A/S ‘The man (focussed) (reportedly) strongly scolded him (that is, tapir, the one we were talking about).’

The presence of the non-subject marker correlates with the position of its referent on the nominal hierarchy (see Dixon 1994: 85): personal pronouns have to be case-marked, while other types of noun phrase heads are casemarked depending on their topicality. In contrast, the marker of focussed subject is used in the same way with any type of noun phrase head. The use of pragmatically-motivated case markers correlates with the linear position of topics and foci within a sentence. This takes us to the next section.

24

Alexandra Y. Aikhenvald

3.2. Constituent order Traditional Tariana is a language with flexible order of clausal constituents, whose freedom is constrained by discourse-pragmatic functions. 8 Tariana discourse is elliptical — a feature the language shares with many other highly synthetic languages with pronominal cross-referencing on the verb (cf. Mithun 1987, Hale 1992). As a result, clauses with overtly marked A (transitive subject) and O (transitive object) are not at all frequent. (In fact, such clauses constitute not more than 25% of the corpus). The order of modifiers within noun phrases operates on a different basis, largely depending on the modifier. Numerals and adjectives precede a noun if its referent is definite and specific; and follow it, if it is not. So, ñhamepa emi-peni (two+CL:HUMAN child-PL) means ‘the two children’, and emipeni ñhamepa (child-PL two+CL:HUMAN) means ‘two (indefinite) children’. Such a noun phrase can occur at the beginning of a story, introducing new referents: (9)

Emi-peni ñhamepa alia-pidana. child-PL two+CL:HUMAN be-REM.P.REP ‘There were two children (reportedly).’

Demonstratives always precede the noun head, as in (4) (see Aikhenvald 2003: 476-8). Unlike other languages, such as Western Romance (see Bernstein 2001), a process of right dislocation does not apply to demonstrative modifiers. Clausal constituent order operates on a different principle. The clauseinitial and the pre-predicate position of a noun phrase strongly correlates with topicality of a constituent. We saw in examples (3), (4) and (5c) that non-subject constituents marked with -naku ‘topical non-subject’ occur in the pre-predicate position. So do the non-focussed subjects, as illustrated in (1), (3) and (8). In each of these cases, the subjects are introduced as future topics of the stretch of discourse. A constituent which is more topical than the subject tends to occur at the left periphery of the clause, as in (10) (‘in her garden’). This is the beginning of the story about a tapir who was stealing fruit from the garden: (10) Payapese-pidana uphedo-pidana long.time.ago-REM.P.REP widow-REM.P.REP hinipuku-se-nuku] hema di-hña garden-LOC-TOP.NON.A/S tapir 3sgnf-eat

[duha she di-ya-nhi. 3sgnf-stay-ANT

Language contact and pragmatic notions

25

‘It was (reportedly) a long time ago. There (reportedly) was a widow. In her (topical) garden (reportedly) a tapir was eating.’ In contrast, the post-predicate position is usually reserved for a constituent in contrastive focus. Examples (6)-(8) illustrate the focussed subject on the right periphery of a clause. In (11), an object in the postpredicate position is in contrastive focus: (11) Awakada-se dinu di-emhani-pidana itœiri. jungle-LOC 3sgnf+kill 3sgnf-walk-REM.P.REP game ‘He (reportedly) went around in the jungle hunting game’ (not little birds). An oblique in contrastive focus also occupies a post-predicate position on the right periphery: (12) Wyaka di-a di-emhani-pidana far 3sgnf-go 3sgnf-walk-REM.P.REP [awakada ieriku-se]. jungle in.depth.of-LOC ‘He (reportedly) went far deep inside the jungle’ (not at the periphery of it). No other constituent can appear in this position. A topical constituent can also be contrasted to something else. Then, it would occur at the right periphery of the clause: (13) Ha-kada dinu di-uma-pidana that-CL:DAY 3sgnf+kill 3sgnf-seek/try-REM.P.REP hema-nuku. tapir-TOP.NON.A/S ‘That day he (reportedly) tried to kill the tapir’ (the one we are talking about; not any other animal). In summary, clausal constituent order and differential case marking in Traditional Tariana interact with pragmatic notions of topic and focus. Postpredicate position correlates with contrastive focus (similar to the situation discussed by Xu 2004, for Chinese), while position at the left periphery of a sentence is preferred for topics. This correlation between linear order and the expression of pragmatic notions comes as no surprise — similar correlations

26

Alexandra Y. Aikhenvald

have been described for numerous languages (see, for example, Aissen 1992; Pereltsvaig 2004). A focussed subject acquires a special marker, clitic -ne/-nhe, and a topical non-subject (the category which subsumes core arguments and obliques) is marked with the clitic -nuku/-naku.

3.3. Grammatical relations and pragmatic notions in Tariana: an areal perspective The principles of case marking for topical non-subjects in Tariana have distinct areal roots. The situation with marking the focussed subject, and using constituent order for encoding pragmatic notion is less straightforward.

3.3.1. Case marking for topical non-subjects The differential non-subject case marking based on semantic and pragmatic properties of the referent is shared by Tariana with its East Tucanoan neighbours. An NP in a non-subject function takes a marker — East Tucanoan -re (Barnes 1999: 219-20) and Tariana -naku/-nuku — if its referent is topical and has already been introduced into the discourse. The parallelism in structure between the use of cases in Tucano and Tariana is illustrated with the following example which comes from an almost identical story recorded in the two languages. An evil spirit demands that a man should give him his own heart. The man is trying to cheat by offering the spirit the heart of a monkey rather than his own. Both ‘heart’ (the direct object) and ‘evil spirit’ (the beneficiary) are topical, and therefore appear marked with the topical non-subject case (underlined): Tucano (Ramirez 1997, Vol. III: 193; my own fieldwork) (14) a. [K̙̣ yaá ehêri põra-re] he POSS ‘breath’ CL:LEAF.LIKE-TOP.NON.A/S(=heart) o'ô-p̣'. miî wãti-re take evil.spirit-TOP.NON.A/S give-REM.P.REP ‘He (reportedly) took his (monkey’s) heart (and) gave it to the evil spirit.’ Tariana (my own fieldwork) b. Di-kale-da-nuku 3sgnf-heart/breath-CL:ROUND-TOP.NON.A/S(=heart) dhita ñamu-nuku di-a-pidana.

Language contact and pragmatic notions

27

3sgnf+take evil.spirit-TOP.NON.A/S 3sgnf-give-REM.P.REP ‘He (reportedly) took his (monkey’s) heart (and) gave it to the evil spirit.’ The presence of the topical non-subject case marker in East Tucanoan languages correlates with the position of its referent on the nominal hierarchy (see Dixon 1994: 85), and the degree of its individuation. A pronominal argument or a proper name is always case-marked. A noun with an uncountable inanimate referent is less likely to be case-marked than a noun with an animate or with a human referent (see Ramirez 1997, Vol. I: 224 on Tucano; Stenzel 2004: 219-25 on Wanano). In agreement with the hierarchy, an animate object is likelier to acquire case marking than an inanimate object. In ditransitive constructions, the recipient or the benefactive is typically animate. Consequently, there is a strong tendency throughout the East Tucanoan family to case-mark second objects of ditransitive verbs (e.g. Tucano examples in Ramirez 1997, Vol. I: 226). If an inanimate ‘gift’ is definite, and the recipient is indefinite, the ‘gift’ is casemarked. That is, definiteness ‘overrides’ animacy and individuation in choosing a case marker. The overt case marking also correlates — throughout the family — with the pragmatic properties of a constituent. For instance, in Desano the nonsubject case -re appears on nouns referring to ‘specific individuals already on stage in the discourse’ (Miller 1999: 58-60). The case marker does not occur on nouns that have ‘just been introduced to the discourse’ (also see Morse and Maxwell 1999: 111 on Cubeo, and Kinch 1977 on Yurití). When used on locatives and time words, the marker indicates that the constituent ‘will have further significance in the discourse’ (Barnes 1999: 220; also see Stenzel 2004: 177-8, 241-2). Along similar lines, the use of -re with locative and temporal constituents in Tucano correlates with their topicality, rather than their definiteness or specificity. The patterns of usage of the topical non-subject enclitic -nuku/-naku in Tariana displays striking similarity with the pan-Tucanoan -re. Both occur on any non-subject constituent which is, or is going to be, the topic of a narrative (Aikhenvald 2003: 145-6). The semantically and pragmatically based marking of non-subjects is a feature spread from Tucanoan into Tariana. As shown in Aikhenvald (2006), the non-subject marker developed from a locative morpheme (following a well-attested grammaticalization path from marking location to marking of topicality: see Radetzky 2002). In summary: a combination of semantic properties — definiteness, specificity and animacy — determines case marking of non-subject core

28

Alexandra Y. Aikhenvald

constituents across the Vaupés area. In some languages topicality is an additional factor. But since topics tend to be definite in Tariana this may be a corollary of the definiteness requirement. Overt non-subject case marking of locative and temporal constituents is based entirely on their pragmatic functions — topicality and focality.9

3.3.2. Case marking for focussed subjects No other Arawak language — except Tariana — has any pragmatically based subject marking. In contrast, some East Tucanoan languages do. But, unlike the non-subject marker -re which is uniform throughout the family, markers of contrastive subjects vary. Desano (Miller 1999: 161-2) employs the contrastive suffix -p̣, which ‘most frequently occurs with the subject’ but ‘can be attached to any noun phrase in the sentence’. The suffix -se’e in Wanano (Stenzel 2004: 175-6; Waltz and Waltz 1997: 45) tends to mark focussed subject. Tucano -‘a, p̣ and pe’e (Ramirez 1997, Vol. I: 231-2; 21819) mark contrastive focus, with a tendency to occur on subjects.10 In all likelihood the Tariana pattern of focussed subject marking is a Tucanoan-based innovation. However, unlike the topical non-subject marker, the Tariana focussed subject clitic -ne/-nhe lacks a one-to-one correspondent morpheme in Tucano, the major contact language. We will see, in §4, how this impacts the reinterpretation of the focussed subject marker in Innovative Tariana, and results in the reduction of its use.

3.3.3. Constituent order East Tucanoan languages are predominantly verb-final, with a tendency towards an AOV and SV order (see Ramirez 1997, Vol. I: 367-72 on Tucano, Miller 1999: 2 on Desano; Morse and Maxwell 1999: 141-2 on Cubeo). Topical constituents tend to occur at the left periphery of the sentence, while constituents in contrastive focus occur sentence-finally. In Tucano, sentence-final constituents on the right periphery are separated by a short pause, and are used for disambiguation and clarification (Ramirez 1997, Vol. I: 372). That is, unlike Traditional Tariana, Tucanoan languages have a fixed clausal constituent order which partly correlates with grammatical function of the core arguments. Tucanoan languages do not employ phrasal constituent order as a means of marking topicality or focality of arguments. Sentence-final tendency for linear position of constituents in contrastive

Language contact and pragmatic notions

29

focus is attested in other East Tucanoan languages, e.g. Barasano (Jones and Jones 1991: 169), and Wanano (examples in Stenzel 2004: 175). We now turn to the ways in which patterns of encoding pragmatic notions in Innovative Tariana mirror those found in Tucanoan languages.

4. Encoding pragmatic notions in Innovative Tariana Innovative Tariana maintains the basic patterns of differential case marking of non-subjects, and of subjects, as described for the traditional language. However, it comes closer to the patterns found in East Tucanoan languages, in particular, Tucano, in the following aspects. Firstly, Innovative Tariana has a fixed verb-final constituent order with a strong verb-final tendency. Just as in the traditional language, sentenceinitial position is strongly associated with topical material. The sentenceinitial topicalization is frequently associated with left-dislocation: the leftmost constituent is separated from the rest of the sentence with a pause. Example (15) is a variant of (5b) produced by an innovative speaker: (15) Hema [PAUSE] hinipuku di-eku di-a-pidana. tapir garden 3sgnf-run 3sgnf-go-REM.P.REP ‘A tapir (reportedly) ran into (or in) a garden.’ Along similar lines, a pause accompanies left-dislocation in Tucano (Ramirez 1997, Vol. I: 367). A similar phenomenon was described for Portuguese in Callou et al. (1993). Secondly, the post-verbal position is restricted to disambiguation, clarification and afterthought. The post-verbal subject is always preceded by a pause (cf. Donati and Nespor 2003 on the association between focus and pause). The use of the focussed subject marker -ne/-nhe is most frequently restricted to disambiguating who did what. An example is at (16). We can recall, that in (2), from Traditional Tariana, the focussed subject is not separated with a pause. (16) Di-eku di-a-pidana 3sgnf-run 3sgnf-go-REM.P.REP

[PAUSE] hema-ne tapir-FOC.A/S

di-eku di-pinita-pidana [PAUSE] nawiki-ne. 3sgnf-run 3sgnf-follow-REM.P.REP man-FOC.A/S ‘The tapir (reportedly) ran, the man (reportedly) followed.’ (lit. He ran, the tapir, he followed, the man)

30

Alexandra Y. Aikhenvald

Innovative speakers hardly ever employ subjects marked for contrastive focus in positions other than post-predicate one. This agrees with the Tucano pattern whereby focus tends to be in the position of the right periphery. Younger speakers tend to use the focussed subject marker less frequently than do traditional ones. Some younger speakers freely omit the focussed subject marker in examples like (16). The tendency to establish a one-to-one correspondence between sentencefinal position and right-dislocation, with an accompanying pause, is paralleled by the process of right-dislocation as a means of marking an antitopic and afterthought in Portuguese (see Koch et al. 1996, and Tarallo and Kato 1996: 327). The loss of frequency in the usage of the focussed subject marker -ne/ -nhe is partly due to the more prominent role of constituent order for encoding pragmatic notions in Innovative Tariana than in the traditional language. That the focussed subject marker does not have an exact parallel in the dominant Tucano is also a contributing factor. An additional circumstance contributing to the loss of the focussed subject marker is partial phonological obsolescence in Innovative Tariana. Due to the linguistic pressure from Tucano, innovative speakers lose the distinction between aspirated and non-aspirated nasals. Phonemes not found in any East Tucanoan language, such as aspirated nasals, tend to be lost; therefore, the form -ne becomes a general one for both contrastive subject and the instrumental-comitative — for many innovative speakers -ne/-nhe is almost always pronounced as -ne. This makes the focussed subject look the same as the instrumental -ne. In (17a), a traditional speaker said waru-nhe (parrot-FOC.A/S), and one innovative speaker repeated this on several occasions as waru-ne, as in (17b). He then translated the sentence into Portuguese as ‘He died, with the parrot’, confirming that for him, -ne covers both contrastive subject and comitative. (17) a. Di-yami-pidana hƭ waru-nhe . 3sgnf-die-REM.P.REP this parrot-FOC.A/S ‘The parrot (not anyone else) (reportedly) died.’ b. Di-yami-pidana [PAUSE] hƭ waru-ne. 3sgnf-die-REM.P.REP this parrot-FOC.A/S ‘He (reportedly) died, the parrot’, translated as ‘He (reportedly) died, with parrot.’

Language contact and pragmatic notions

31

Examples like (17b) do not make sense to many speakers of Innovative Tariana. As a result, they avoid using -ne with subjects altogether. That is, linguistic pressure from Tucano results in the obsolescence of the unusual differential subject marker in Tariana.

5. Further means of encoding pragmatic notions in Traditional and in Innovative Tariana Traditional Tariana has a number of additional means employed for encoding topical core arguments and obliques. Two of these illustrate rather striking differences between the traditional, and the innovative, language. When the causative-applicative derivation -i is added to a transitive verb in Traditional Tariana, it serves to foreground a topical oblique constituent. In (18), the verb -wapa ‘wait, attend’ appears as a simple transitive verb: (18) Nu-na di-wapa-pidana. 1sg-OBJ 3sgnf-wait-REM.P.REP ‘He was (reportedly) waiting for me.’ In (19), the verb takes the causative-applicative suffix -i; its effect consists in foregrounding the oblique which traditionally accompanies ‘waiting’ — such as ritual offering. The ritual offering is the topic of the stretch of discourse: (19) Di-walita-nipe-ne-pidana nu-na di-wapa-i.11 3sgnf-offer-NOM-INS-REM.P.REP 1sg-OBJ 3sgnf-wait-APPL ‘He was (reportedly) waiting for me with the ritual offering (topic).’ The applicative derivation is not fully productive. The type of oblique depends on the semantics of the verb. Each verb has a preferential oblique which can be cast as topic through using the causative-applicative derivation. In many cases, the knowledge of which oblique to choose depends on the cultural context — for instance, for the verb ‘bless’ the expected foregrounded oblique involves the instrument of blessing, and for the verb ‘call’ it involves the purpose of calling. As a result of the encroaching obsolescence of Tariana, many younger and innovative speakers do not have enough cultural or linguistic knowledge to confidently use the causativeapplicative derivation as a topicalizing device. The device itself becomes obsolescent in the young speakers’ language.

32

Alexandra Y. Aikhenvald

Tense, aspect and evidentiality in Traditional Tariana are expressed with floating clitics, whose position depends on the information structure. If the sentence contains no established topic, they occur on the verb, as in (1) - (4) — which is their functionally unmarked position. If there is a new topic, clitics attach to it, as in (19). Tucanoan languages express the same categories of tense, aspect and evidentiality with suffixes attached to the verb. The positioning of affixes is never used to encode topics. Innovative speakers of Tariana, who speak mostly Tucano in their day-to-day lives, follow the Tucano mould: they put the clitics exclusively on the verb, and do not employ them as an additional topicalizing device. So far, one may have the impression that Innovative Tariana is impoverished as compared to the traditional language. But this is not exactly so. Innovative Tariana is developing an incipient cleft-like construction, under Portuguese influence. A young speaker spontaneously produced (20a) — which is a loan translation from a cleft construction in the local Portuguese reproduced in (20b): (20) a. Maria-naka kwaka wa-na du-ni du-waketa. Maria-PRES.VIS what 1pl-OBJ 3sgf-do 3sgf-join ‘Maria is the one who helps us (we see it).’ b. Maria é que nós ajuda. Maria is what us helps ‘Maria is the one who helps us.’ Such cleft constructions are used more and more often, as more and more younger speakers of Tariana acquire proficiency in Portuguese — which is the symbol of status in indigenous communities, and especially mission centres where many of the Tariana reside. They are rarer than constructions with left dislocation marked by a pause (also a phenomenon widespread in colloquial Portuguese, including Amazonian Portuguese). Such cleft constructions have not been so far documented for Tucanoan languages.

6. To conclude Both Traditional and Innovative Tariana constitute prime examples of languages which combines portmanteau mechanisms for marking grammatical relations and information structure. Both employ special marking for topical non-subjects —a feature resulting from intensive areal

Language contact and pragmatic notions

33

diffusion from Tucanoan languages. Traditional Tariana has a number of other ways of encoding topics: these include variable position for tense, aspect and evidentiality clitics, and an applicative verbal derivation with a foregrounding effect. In Traditional Tariana, the clitic -ne/-nhe consistently marks a focussed subject. Just like its relatives from the Arawak family, Traditional Tariana has discourse-dependent clausal constituent order which does not correlate with grammatical function. That is, grammatical relations do not correlate with the syntactic position of a constituent within a clause or a sentence. Alternative orders of clausal constituents bear different pragmatic overtones, with left periphery associated with topicality and right periphery with focus. Gradual language obsolescence and intensified spread of contact-induced patterns from the dominant Tucano conspire to make Innovative Tariana appear more Tucano-like in the ways it encodes pragmatic concepts. As a consequence of the influence of the dominant Tucano, with its predominantly verb-final clausal constituent order, Innovative Tariana is becoming predominantly verb-final. Under the influence of the dominant Tucano, the order of clausal constituent order is acquiring a strong verb-final tendency. The position of left-periphery is now strongly associated with topicality and left dislocation of a constituent (marked with an audible pause). The position of right periphery is associated with disambiguation and anti-topics. These patterns are also shared with Portuguese, another dominant language of the area. In addition, Innovative Tariana is developing an incipient cleft construction under Portuguese influence. The focussed subject marker is becoming obsolescent which is partly due to its coalescence with the comitative-instrumental marker as a result of the loss of the phonological contrast between simple n and aspirated nh. We conclude that in Innovative Tariana linear position in a sentence or clause is acquiring new functions to do with marking information structure, partially replicating the Tucanoan pattern, and partially aligning with the patterns in Portuguese. Information structure remains encoded in the syntax and morphology of the innovative language — but the means of encoding are simpler in some ways, and more complex in others. These results corroborate the diffusability of (a) clausal constituent order and (b) encoding pragmatic concepts, in a situation of intensive language contact with a pervasive societal multilingualism.

34

Alexandra Y. Aikhenvald

Notes 1. I am grateful to all my teachers of Tariana, the Britos of Santa Rosa and the Muniz of Periquitos, for teaching me their remarkable language. The data on Tariana and Tucano come from original fieldwork. Thanks are equally due to R. M. W. Dixon for incisive comments and insights. 2. Comparison between Tariana and closely related Arawak languages, such as Baniwa of Içana, helps distinguish between archaic inherited features on the one hand, and innovative and diffused traits on the other. This lies beyond the scope of this paper. 3. Abbreviations: 1, 2, 3 - first, second, third person; A - transitive subject; ANT anterior; APPL - applicative; CL - classifier; EMPH - emphatic; FOC.A/S - focussed subject; INS - instrumental; LOC - locative; NOM - nominalizer; O object of transitive verb; OBJ - object marker; PL - plural; PRES.VIS - present visual; REM.P.REP - remote past reported; S - subject of intransitive verb; Sa subject of intransitive active verb; So - subject of intransitive stative verb; sgf - singular feminine; sgnf - singular nonfeminine; TOP.NON.A/S - topical nonsubject. 4. The allomorph -nhe is used if the noun does not contain an aspirated consonant, e.g. nawiki-nhe (person-FOC.A/S). Otherwise, the allomorph -ne is selected, e.g. hema-ne (tapir-FOC.A/S), ñhamepa-ne (two+CL:HUMAN-FOC.A/S) ‘they two’. 5. A more archaic allomorph -naku used by the representatives of the older generation. The variant -nuku is used by younger people: in Innovative Tariana, enclitics tend to undergo assimilation between the two final vowels. 6. Here are throughout this paper, the formative -ha is not provided with a gloss, since forms like diha ‘he’, nuha ‘I’ etc can be considered fossilized. 7. Despite a formal similarity between -na ‘object marker on pronouns’ and -naku ‘topical non-subject’, there is no evidence for any historical link between these (see Aikhenvald 2006). 8. Alternatively, Tariana can be considered ‘discourse-configurational’ (cf. Kiss 1995). Note that I refrain from using the term ‘word order’ since it does not allow us to distinguish between order of individual words within a phrase (or phrasal constituent order) and order of clausal constituents. 9. Case marking in East Tucanoan languages and in Tariana also correlates with the position of the object argument: an unmarked argument with a generic referent is likely to occur in the preverbal position. (This may result in OV constructions being interpretable as instances of noun incorporation: see Barnes 1999: 220; Morse and Maxwell 1999: 70-1.) Exact correlations between constituent order and information structure in East Tucanoan languages require further investigation. 10. The lack of a common morpheme for focussed or contrastive subject among the East Tucanoan languages may suggest that this is a recent innovation. However,

Language contact and pragmatic notions

35

West Tucanoan languages Koreguaje and Siona spoken outside the Vaupés area have a marker marking subject (Aikhenvald 2006; Cook and Levinsohn 1985: 92-100; Wheeler 1967: 61-3; Wheeler 1987: 124-6), suggesting older origins of this pattern. 11. For ease of reference, the form is given in its archaic shape; the contemporary Traditional Tariana shape is di-wape.

References Aikhenvald, Alexandra Y. 1995 Person-marking and discourse in North-Arawak languages. Studia Linguistica 49: 152-95. Aikhenvald, Alexandra Y. 1998 Warekena. In Handbook of Amazonian Languages, Desmond C. Derbyshire and Geoffrey K. Pullum (eds.), vol. 4, 215-439. Berlin: Mouton de Gruyter. Aikhenvald, Alexandra Y. 2002 Language contact in Amazonia. Oxford: Oxford University Press. Aikhenvald, Alexandra Y. 2003 A Grammar of Tariana, from North-west Amazonia. Cambridge: Cambridge University Press. Aikhenvald, Alexandra Y. 2006 Semantics and pragmatics of grammatical relations in the Vaupés linguistic area. In Grammars in contact: a cross-linguistic typology, Alexandra Y. Aikhenvald and R. M. W. Dixon (eds.), 237-66. Oxford: Oxford University Press. Aissen, Judith L. 1992 Topic and Focus in Mayan. Language 68: 43-80. Aissen, Judith L. 2003 Differential object marking: Iconicity vs economy. Natural Language and Linguistic Theory 21: 435-83. Barnes, Janet. 1999 Tucano. In The Amazonian languages, R. M. W. Dixon and Alexandra Y. Aikhenvald (eds.), 207-26. Cambridge: Cambridge University Press. Bernstein, Judy B. 2001 Focusing the "right" way in Romance determiner phrases. Probus 13: 129. Bossong, Georg 1985 Empirische Universalienforschung. Differentielle Objekstmarkie- rung in den neuiranischen Sprachen. Tübingen: Gunter Narr. Bossong, Georg 1991 Differential object marking in Romance and beyond. In New analyses in

36

Alexandra Y. Aikhenvald

Romance linguistics. Dieter Wanner and Douglas A. Kibbee (eds.), 14370. Amsterdam: John Benjamins. Callou, Dinah, João Moraes, Yonne Leite, Mary A. Kato, Célia T. de Oliveira, Elenice Costa, Mônica Orsini, Violeta Rodrigues 1993 Topicalização e deslocamento à esquerda: sintaxe e prosódia. In Gramática do Português Falado, Vol. III. As abordagens, Ataliba Teixeira de Castilho (ed.), 315-60. Campinas: Editora da Unicamp/FAPESP. Cook, Dorothy M. and Stephen Levinsohn 1985 Coreguaje: Domains of Focus Markers. In From phonology to discourse: studies in six Colombian languages, Ruth M. Brend (ed.), 91-116. Dallas: SIL. Cruse, Alan 2006 A glossary of semantics and pragmatics. Edinburgh: Edinburgh University Press. Dixon, R. M. W. 1994 Ergativity. Cambridge: Cambridge University Press. Dixon, R. M. W. 2002 Australian languages: their nature and development. Cambridge: Cambridge University Press. Dixon, R. M. W. and Alexandra Y. Aikhenvald 1997 A typology of argument-determined constructions. In Essays on Language Function and Language Type, Joan Bybee, John Haiman and Sandra Thompson (eds.), 71-113. Amsterdam: John Benjamins. Donati, Caterina and Marina Nespor 2003 From focus to syntax. Lingua 113: 1119-1142. Frascarelli, Mara 2000 The syntax-phonology interface in focus and topic construction in Italian. Amsterdam: Kluwer. Hale, Kenneth L. 1992 Basic word order in two "free word order" languages. Reprinted in Pragmatics of word order flexibility, Doris L. Payne (ed.), 63-82. Amsterdam: John Benjamins. Jones, Wendell and Paula Jones. 1991. Barasano Syntax. (Studies in the languages of Colombia 2.) Arlington: Summer Institute of Linguistics and the University of Texas at Arlington. Junker, Marie-Odile 2004 Focus, obviation, and word order in East Cree. Lingua 114: 345-365. Kinch, R. A. 1977 El enfoque temático vs el enfoque no temático en yuriti. Estudios Tucanos II: 129-75. Instituto Lingüístico de Verano: Lomalinda, Meta. Kiss, Katalin É. 1995 Introduction. In Discourse configurational languages, Katalin É. Kiss (ed.), 3-27. Oxford: Oxford University Press.

Language contact and pragmatic notions

37

Koch, Ingedore G. Villaça, Clélia Cândida A. Spinardi Jubran, Hudinilson Urbano, Leonor L. Fávero, Luiz Antonio Marcuschi, Maria do Carmo O. T. Santos, Mercedes Sanfelice Risso 1996 Aspectos do processamento do fluxo de informação no discurso oral dialogado. In Gramática do Português Falado, Vol. 1. A Ordem, Ataliba Teixeira de Castilho (ed.), 143-84. Campinas: Editora da Unicamp/FAPESP. Lambrecht, Knud 1994 Information structure and sentence form. Topic, focus and the mental representation of discourse referents. Cambridge: Cambridge University Press. Miller, Marion 1999 Desano grammar. Arlington: Summer Institute of Linguistics and the University of Texas at Arlington. Morse, Nancy L. and Maxwell, Michael B. 1999 Gramática del Cubeo. Santa Fé de Bogotá: Editorial Alberto Lleras Camargo. Mithun, Marianne 1987 Is basic word order universal?. In Coherence and grounding in discourse, R. S. Tomlin (ed.), 281-328. Amsterdam: John Benjamins. (Reprinted in Pragmatics of word order flexibility, Doris L. Payne (ed.), 15-61, Amsterdam: John Benjamins. Mithun, Marianne 2001 Understanding and explaining applicatives. Proceedings from the Parasessions of the Thirty-seventh Meeting of the Chicago Linguistic Society, 37-1: 73-97. Olawsky, Knut 2006 A grammar of Urarina. Berlin: Mouton de Gruyter. Pereltsvaig, Asya 2004 Topic and focus as linear notions: Evidence from Italian and Russian. Lingua 114: 325-344. Radetzky, Paula K. 2002 The functions and evolution of topic and focus markers. PhD Dissertation, University of California Berkeley. Ramirez, Henri 1997 A fala Tukano dos Yepâ-masa. Tomo I. Gramática. Tomo III. Método de aprendizagem. Manaus: Inspetoria Salesiana Misionária da Amazônia, CEDEM. Rizzi, Luiggi 1997 The fine structure of the left periphery. In Elements of grammar. Handbook in generative syntax, Liliane Haegeman (ed.), Dordrecht: Kluwer.

38

Alexandra Y. Aikhenvald

Stenzel, Kristine S. 2004 A reference grammar of Wanano. PhD Dissertation. University of Colorado. Tarallo, Fernando and Mary Kato 1996 Preenchimento em fronteiras de constituintes. In Gramática do Português Falado, Vol. II. Niveis de Análise Lingüística, Rodolfo Ilari (ed.), 315-353. Campinas: Editora da Unicamp/ FAPESP. Waltz, Nathan and Caroline Waltz 1997 El Agua, La Roca y El Humo. Estudios sobre la cultura wanana del Vaupés. Santafé de Bogotá: Instituto Lingüístico del Verano. Wheeler, Alva 1967 Grammatical structures in Siona discourse. Lingua 19: 60-77. Wheeler, Alva 1987 Gantëya Bain. El Pueblo Siona del río Putumayo, Colombia. Tomo 1: Etnología, Gramática, Textos. Bogotá: Instituto Lingüístico de Verano. Xu, Liejiong 2004 Manifestation of informational focus. Lingua 114: 277-299. Zubizarreta, Maria Luisa 1998 Prosody, focus, and word order. Cambridge, Massachusetts: MIT Press.

Communicative structure in Lushootseed syntax: Thematicity and focalization David Beck

1. Introduction The surface form of sentences in the Salishan language Lushootseed, as in other languages in its family, is conditioned to a remarkable degree by Communicative Structure (Mel’ìuk 2001, a.k.a. Information Structure — Lambrecht 1994; Vallduví 1992). While previous attempts to come to terms with this aspect of Salishan syntax have made use of concepts such as Topic and Comment (Davis and Saunders 1978) and Discourse Topic (Beck 2000; Kinkade 1990), this paper will attempt to account for a wider range of the effects of Communicative Structure on Lushootseed syntax by applying the model of Semantic Communicative Structure (Sem-CommS) outlined by Mel’ìuk (2001) in the framework of Meaning-Text Theory. It will be argued that, contrary to traditional approaches to syntax, which give a priori primacy to lexical and syntactic categories in clause structure, Lushootseed gives precedence to Communicative Structure in the organization of the clause. Some preliminary steps will also be taken towards formalizing a system of ranked constraints on the expression of particular elements of the SemCommS in Deep Syntactic Structure, and it will be shown that some of the typological differences separating Salishan from more familiar languages can be accounted for by differential ordering of the same basic set of constraints. 2. Communicative Structure in a Meaning-Text Model Meaning-Text Theory (Mel’ìuk 1974; Mel’ìuk 1981; Mel’ìuk 1988; Žolkovskij and Mel’ìuk 1965) is an approach to language, used to date primarily in the fields of lexicography and text-synthesis, which models linguistic structures in terms of mappings from a Semantic Representation (SemR) through Syntactic and Morphological Representation to a Phonological Representation. Mappings between levels are treated as rule- or constraintgoverned correspondences allowing for one-to-many and many-to-one mappings rather than as transformations or derivations. SemR consists in part of

40

David Beck

Figure 1. SemS ™ DSyntS correspondences for the boy looked for the dog

a Semantic Structure, a network representing the propositional content of an utterance; likewise, Syntactic Representation includes a Syntactic Structure, represented as a dependency tree. Figure 1 shows a simplified mapping between the SemS and the first level of syntax, Deep Syntactic Structure (DSyntS).1 The SemS on the left of Figure 1 shows the two-place predicate ‘seek’ governing two Semantic Actants (SemAs) (numbered 1 and 2 for book-keeping purposes) and itself governed by a predicate indicating timeof-utterance. Each node of the SemS corresponds to a semantic unit or semanteme, which is expressed in the DSyntS by lexical, syntactic, or morphological means. Lexical meanings are instantiated as nodes in the DSyntS via a process called lexicalization; morphologically-expressed inflectional meanings like tense are indicated as subscripts on the syntactic node of the lexeme with which they are associated. Syntactic relations between heads and dependants are indicated by arrows labeled for the type of relation that holds between them. The top node of the syntactic tree in Figure 1 is the lexicalization of the semantic predicate ‘seek’, the verb LOOK (FOR) ‘X seeks Y’, whose diathesis specifies that X be expressed as its first Deep Syntactic Actant (DSyntA I) and Y as its DSyntA II.2 The process of mapping between SemS and DSyntS is referred to as syntacticization. In addition to a SemS representing the propositional content of the utterance, SemR also contains other structures, including the Semantic Communicative Structure (Sem-CommS), which specifies the values of eight communicative categories — Thematicity, Focalization, Givenness, Perspective, Emphasis, Presupposedness, Unitariness, and Locutionality (Mel’ìuk 2001). Of these eight, the two of interest here are Thematicity and Focalization. Thematicity entails the division of the SemR into areas or subnetworks that are specified as either Thematic (what the utterance is about) or Rhematic (what is being communicated about the Theme).3 Focalization refers to the designation of an area of the SemS as being “logically prominent” or the focal point of a speaker’s attention (Mel’ìuk 2001: 181). Focalization cor-

Communicative structure in Lushootseed 41

Figure 2. Sem-CommS of the dog, the boy looked for it

responds roughly to what Chafe (1976) calls “Contrastiveness”, the notion of contrast being related to Focalization in that something that is contrastive is necessarily Focalized, but something that is Focalized is not necessarily contrastive (Mel’ìuk 2001: 183). Thematicity and Focalization are independent categories, and the Focalized area of a SemS may be coterminous with either the Thematic (Th) or the Rhematic (Rh) area of a SemS. Figure 2 above shows the SemS in Figure 1 with a Focalized Semantic Theme (Focalization being indicated by the shaded circle). This SemR corresponds to the English sentence the dog, the boy looked for it. Because the sentence is “about” the dog, ‘dog’ is contained in the Semantic Theme; what is being said about the dog is that the boy sought it, so ‘seek’ and its SemA 1 ‘boy’ are contained in the Semantic Rheme. This sentence would be appropriate in a context that focuses attention on the Thematic event-participant, which appears in the Focalized area of the SemR. The SemR is mapped onto a DSyntS in which the expression of ‘dog’ is iterated, once as the DSyntA II of the verb LOOK(FOR) (pronominalized by the correspondence rules with SSyntR) and once as an Appenditive dependant (the left-dislocated NP). Co-reference between the two nodes is indicated by the double-headed arrow (part of the Anaphoric Structure of the DSyntR). The DSyntR also has a Communicative Structure, including its own Theme and Rheme which are not necessarily in exact alignment with the Semantic Theme and Rheme; this has effects on the surface form of some sentences in some languages, but does not affect the present discussion. The transition between the SemR and DSyntR is mediated by rules and constraints that define the acceptable mappings between configurations of semantemes and their lexical and morphosyntactic expressions. One of the goals of this paper is to outline certain constraints on such mappings that have a profound impact on Lushootseed syntax (Section 4), following a brief sketch of the relevant aspects of Lushootseed clause structure in Section 3.

42 David Beck

3. Lushootseed clause structure Lushootseed, like other Salishan languages, uses a system of pronominal clitics, agreement-markers, and a fairly rigid VSO word-order to encode grammatical relations (Czaykowska-Higgins and Kinkade 1998). The simple transitive clause is illustrated in (1):4 (1)

a. ɎVHɯȓÍ‫ޙ‬ȓEÍȓEUJTRɯȓCBZɎ ɍV‫ޠ‬HɮȒÍަ‫ޠ‬U


ÍȒE



UJ
T

RɮȒCBZɍ
PFV–seek–ICS 1SG.SUB DEF dog ‘I looked for the dog.’




b. ɎVHɯȓÍ‫ޙ‬ȓEÍȓȦUJTRɯȓCBZɎ


ɍV‫ޠ‬HɮȒÍަ‫ޠ‬U


ÍȒȥ



UJ


TRɮȒCBZɍ
PFV–seek–ICS 1PL.SUB DEF dog ‘We looked for the dog.’




c. ɎVHɯȓÍ‫ޙ‬ȓEÍȓYɯUJTRɯȓCBZɎ


ɍV‫ޠ‬HɮȒÍަ‫ޠ‬U


ÍȒYɮ


UJ


TRɮȒCBZɍ
PFV–seek–ICS 2SG.SUB DEF dog ‘You looked for the dog.’




d. ɎVHɯȓÍ‫ޙ‬ȓEÍȓMȓQUJTRɯȓCBZɎ


ɍV‫ޠ‬HɮȒÍަ‫ޠ‬U


ÍȒMȒQ


UJ


TRɮȒCBZɍ
PFV–seek–ICS 2PL.SUB DEF dog ‘You guys looked for the dog.’




e. ɎVHɯȓÍ‫ޙ‬ȓEUJTRɯȓCBZɎ


ɍV‫ޠ‬HɮȒÍަ‫ޠ‬U


˜


UJ



TRɮȒCBZɍ
PFV–seek–ICS 3SUB PROX dog ‘He/she/it/they looked for the dog.’ (Hess 1995: 10)

Pronominal subjects are marked by one of a series of matrix subject clitics, the third person in this series being Ø and not making a distinction for number (Beck 2000). A peculiarity of Lushootseed, shared to a certain extent by some other languages in the family (Gerdts 1988; Kinkade 1990), is that transitive sentences with both an NP subject and an NP object are disallowed (Hess 1973). Sentences with third-person subject and object undergo obligatory pronominalization of the subject, surfacing as sentences like (1e). An interpretation of (1e) where the NP following the verb is interpreted as subject/AGENT (i.e., ‘the dog looked for him/her/it/them’) is disallowed by what

Communicative structure in Lushootseed 43

Gerdts (1988) refers to as the One-Nominal Interpretation Law. Interpretation of transitive clauses in discourse is facilitated by a reference-tracking system built around a strong constraint that subjects be topical (Beck 2000; Kinkade 1990). In contexts where the identity of the subject of a transitive clause is not recoverable from discourse, or where both event-participants must be specified for communicative reasons, the passive voice is used: (2)


ɎVHɯȓÍ‫ޙ‬UȓCɎȓUJÍ‫ޙ‬BÍ‫ޙ‬BTUJTRɯȓCBZɎ
ɍV‫ޠ‬HɮȒÍަ‫ޠ‬U‫ޠ‬C
ɍȒ
UJ
ÍަBÍަBT
UJ
TRɮȒCBZɍ PFV–seek–ICS–PASS PR DEF child DEF dog ‘The dog was looked for by the boy.’ (Hess 1995: 23, ex. 6a)

Like the English passive, the passive in Lushootseed promotes the direct object to subject and demotes the active voice subject to an oblique agentive complement (AgCo) phrase, introduced by a preposition (Ɏȓ). The order of arguments can be either Sub >> AgCo or, as shown here, AgCo >> Sub, the order in (2) being more prevalent. One of the more remarkable characteristics of Lushootseed syntax is the flexibility it displays with respect to which parts of speech are eligible syntactic predicates (Beck 2002). As in most Salishan languages (Kinkade 1983), words corresponding to English verbs, nouns, adjectives, adverbs, and demonstrative pronouns are all potential clausal predicates, as in the nominally-predicated expressions in (3): (3) 

a. ɎBDJȦUBMCJYɯÍȓE
  ɍBDJȥUBMCJYɮ
ÍȒE Indian 1SG.SUB ‘I am an Indian.’ (Hess and Hilbert 1976: vol. I, 36)



b. TɎVMBEYɯUJɎJȦ
  TɍVMBEYɮ
UJɍJȥ salmon DIST ‘That is a salmon.’ (Hess and Hilbert 1976: vol. I, 7)

Non-verbal predicates like ɎBDJȦUBMCJYɯ

‘Indian’ in (3a) take the same subject inflections as do the verbal predicates in (1) above; simple copular constructions with NP or demonstrative subjects like that in (3b) simply jux-

44 David Beck

tapose subject and predicate, the latter appearing in sentence-initial position. Unlike nouns and other nominal elements used as arguments, nominal predicates do not take determiners unless the sentence identifies the subject with a specific individual. Non-verbal predicates are not confined to simple expressions of identity like those in (3); constructions like that in (4), with a nominal predicate and a complex nominal acting as subject, are quite routine: (4)

XJX‫ޙ‬TVUJɎȓɎɎVÍBMBEUJɎȓɎTRɯȓCBZɎ XJXަTV
UJɍȒɍ
ɍV‫ޠ‬ÍBMB‫ޠ‬E
UJɍȒɍ
TRɮȒCBZɍ
children PROX PFV–chase–ICS PROX dog ‘Those who chased the dog are the children.’ (Hess 1995: 99)

The syntactic predicate in (4) is the bare noun XJX‫ޙ‬TV ‘children’, while the subject is a relative clause headed by the determiner UJɎȓɎ (Beck 2002). It is the predicative use of nouns in constructions like those in (3) and (4) which have led some researchers (e.g., Jelinek and Demers 1994; Kinkade 1983; Kuipers 1968) to argue that there is no distinction between nouns and verbs in Salishan languages. While this is probably an over-reaction,5 it is the case that part of speech plays less of a role in the selection of the syntactic predicate of a sentence than it does in most other languages, where there is a strong tendency for sentences to have verbal predicates. Instead, predicate selection in Lushootseed depends crucially on Communicative Structure — specifically, on Thematicity. 4. Effects of Thematicity on syntactic structure The component of Sem-CommS that has the greatest effect on Lushootseed syntax is Thematicity, which is the driving force behind the selection of the syntactic predicate. A similar effect is discussed for another Salishan language, Nuxalk (Bella Coola), in Davis and Saunders (1978). In Davis and Saunders’ terms, the structure of a Nuxalk clause involves a bi-partition between the part of the sentence that is the “Comment” and that which is “Topic”. In our terms, this translates into the Nuxalk sentence being organized in such a way that the Sem Rheme is expressed as the syntactic predicate and the Sem Theme is expressed as its subject. Beck (1997) shows that the same pattern is found in Lushootseed, as illustrated by question-and-answer pairs such as those in (5):

Communicative structure in Lushootseed 45

(5)










a. ɎVɎȓᆪJELɯJLJLȓXJÍ


ɍV‫ޠ‬ɍȒሜJE
LɮJ
LJ‫ޠ‬LȒXJÍ PFV–what.happen REM ATTN–hunchback ‘What happened to Little Hunchback?’ — ɎVʇ ś‫ޙ‬BM‫ޙ‬CVTBYɯȓCEVCVUUJɎJȦLJLȓXJÍ
  ɍV
śަBMަ
CȒɍV‫ޠ‬TBYɮȒC‫ޠ‬EYɮ‫ޠ‬CVU
UJɍJȥ
LJ‫ޠ‬LȒXJÍ INTJ also ADD=PFV–run–DC–REFL DIST ATTN–hunchback ‘Oh, Little Hunchback also managed to escape.’ [DM Basket Ogress, lines 79 – 80] b. TUBCUJɎVÍBMBUȓCɎȓUJɎJȦXJX‫ޙ‬TV


TUBC
UJ
ɍV‫ޠ‬ÍBMB‫ޠ‬U‫ޠ‬C
ɍȒ
what DEF PFV–hit–ICS–PASS PR ‘What was chased by the children?’

UJɍJȥ
XJXަTV
DIST children

— TRɯȓCBZɎUJɎVÍBMBUȓCɎȓUJɎJȦXJX‫ޙ‬TV


TRɮȒCBZɍ
UJ
ɍV‫ޠ‬ÍBMB‫ޠ‬U‫ޠ‬C
ɍȒ
UJɍJȥ
XJXަTV DEF PFV–hit–ICS–PASS PR DIST children dog ‘The one chased by the children was a dog.’ (Hess 1995: 98 – 99)

The question in (5a) is a narratively-focused question asking about an event in which a particular Thematic event-participant is involved, and elicits a narratively-focused response with a Rhematic verbal predicate. The question in (5b), however, asks for the identity of an unknown participant in a Thematic event, this event being expressed as a headless relative clause in subject position of a sentence whose predicate is the interrogative word TUBC ‘what?’. The response mirrors this structure exactly, substituting the requested information for the interrogative word, giving us a sentence with a Rhematic nominal predicate. Subsequent work has found that this pattern also occurs more generally in Lushootseed narrative and other discourse contexts where the event is Thematic (and, generally, Given) and the Rheme is an event-participant or some other non-verbal element of the sentence (Beck 2000; Beck 2002). This is an interesting situation from a theoretical point of view in that traditional approaches to syntax generally approach clause structure as being built around phrasal projections of lexical elements whose part of speech (or the projections of the functional/inflection categories associated therewith) determine whether particular elements are realized in predicate or argument positions. Thematicity in such approaches tends to be treated as a secondary phenome-

46 David Beck

non with less import for clause structure; however, data like that in (5) shows that Thematicity plays a much more fundamental role in the grammars of Salishan languages, and suggests that it might be worthwhile to re-examine some of these traditional assumptions. In the following section, I will illustrate how the effects of Thematicity on syntactic structure can be modeled from the perspective of text generation, using some of the formalisms from Meaning-Text Theory introduced in Section 2 above, beginning with the selection of the syntactic predicate (Section 4.1), and then moving on to constraints on the use of grammatical voice (4.2). 4.1. Constraints governing the selection of syntactic predicates From the point of view of text synthesis, the initial step in mapping between SemR and DSyntR is the selection of what is called the entry node, that semanteme (or configuration of semantemes) in the SemR that will be lexicalized as the top node of the DSyntS (the matrix predicate). In the SemR in Figure 1, for example, the entry node is ‘seek’, which is lexicalized as the verb LOOK(FOR), the matrix predicate in the DSyntR. One of the reasons that ‘seek’ is chosen is that it is the Communicatively Dominant Node in its sub-network of the SemR. In this context, Communicatively Dominant means that the subnetwork of the SemS ‘past’ → ‘seek’ → ‘dog’ can be reduced to ‘seek’ without changing its referent — that is, its referent is a seeking event rather than, say, a dog (if ‘dog’ were the Comm-dominant node in this configuration, the corresponding DSynt tree would be a relative clause, the dog that was sought).6 The second consideration that determines the selection of ‘seek’ as the entry node is the part of speech with which its meaning is lexicalized. In English, ‘seek’ corresponds to the meaning of a verb, LOOK(FOR), which is selected as the top node of the DSyntS. In most languages, part of speech is one of the most important factors governing the selection of entry nodes: by far the majority of the world’s languages select as entry node semantemes that are semantic predicates and whose most natural expression is a verb.7 However, Lushootseed departs from this pattern in preferring elements that are Rhematic, whatever their lexical class, allowing for a wide range of nonverbal predicates in syntactic structure. Procedural rules for the selection of entry nodes in languages like English, Russian, and French are set out in work by Iordanskaja and Polguère (Iordanskaja 1990; Iordanskaja and Polguère 1988) and discussed in the context of Sem-CommS by Mel’ìuk (2001). Without getting bogged down in

Communicative structure in Lushootseed 47

technical details, the relevant rules proposed by these authors can be restated in terms of the ranked constraints in (6): (6)

Semantically-Predicative Entry Node (SPEN) The entry node is a semantic predicate. VEN The entry node is most naturally lexicalized as a verb. Comm-Dominant Entry Node (CDEN) The entry node is the Comm-dominant node. Rhematic Entry Node (REN) The entry node is a part of the Sem Rheme.

In most languages, these constraints are ranked in the order given here, expressing the near-universal preference for verbal syntactic predicates: (7)

SPEN, VEN » CDEN » REN

In Lushootseed, on the other hand, the same four constraints are implemented, but they are ordered differently, as in (8): (8)

REN » CDEN » SPEN, VEN

These rankings express the Lushootseed preference for Rhematic syntactic predicates, irrespective of their lexical class. The differences in these two ranking systems account neatly for the differences in the selection of syntactic predicate seen between English and Lushootseed. The most straightforward case, a narratively-focused sentence with a verbal predicate such as (1e), is shown in Figure 3. This sentence represents a common case in narratives, which are typically structured around one or more Topical participants. Topical participants tend to be expressed at the sentence-level as Themes, whose actions and (to a lesser extent) the other participants with whom they interact tend to be Rhematic at the sentencelevel. Thus, the SemR in Figure 3 on the next page is partitioned into two SemComm areas, the Theme containing the topical event-participant, ‘boy’, and the Rheme containing the semantic predicate ‘chase’ and its SemA 2 ‘dog’ — this sentence being an appropriate response to a question such as “What did the boy do?” or as part of a story relating the boy’s activities.

48

David Beck

Figure 3. ƝugȿΩþ’Ωd (ti þ’aþ’as) ti sqȿΩbayƝ ‘(the boy) looked for the dog’

In both English and Lushootseed, the SemR in Figure 3 is expressed as an ordinary verbally-predicated sentence, given that the node ‘seek’ satisfies all four constraints for being the top node in DSyntR — that is, it is a semantic predicate, it is naturally expressed as a verb, it is a Comm-Dominant node (indicated by underlining), and it belongs to the Sem Rheme. This is not the case for the other nodes in the SemR, as can be seen in Table 1 below, which presents —using an Optimality-Theory (Prince and Smolensky 2004) style tableau — all five candidate entry nodes in the SemR in Figure 3 as evaluated against the four constraints, ordered as per the Lushootseed rankings shown in (8). ‘Seek’ is the superior candidate as it incurs no constraint violations whatsoever, whereas the remainder of the candidates violate at least two of the constraints. Because ‘seek’ satisfies all four constraints, the difference in rankings between (7) and (8) are of no consequence, and English and Lushootseed syntacticize this SemR in essentially the same way (although Lushootseed requires pronominalization of the NP subject in SSyntR, as noted in Section 2). Table 1. Lushootseed constraint rankings for Figure 3 Candidate Node ) ‘seek’ ‘boy’ ‘dog’ ‘def’Rh ‘def’Th

REN

CDEN

*!

*!

*! *! *

SPEN

VEN

* *

* * * *

For sentences with non-verbal predicates such as that in (3b), shown in Figure 4 on the next page, we see the same principles at work, although here the difference in constraint rankings results in very different structures. This sentence, a felicitous response to the question “What is that?”, is straightfor-

Communicative structure in Lushootseed 49

Figure 4. TɎVMBEYɯUJɎJȦ ‘that is a salmon’

ward in propositional terms, its Sem S consisting of a semantic predicate ‘be’ with two Semantic actants (SemAs), indicating the identification of a particular entity pointed to by the speaker (SemA 1)— as a member of a designated class of entities (SemA 2). In languages with a copula, the semanteme ‘be’ would be realized as the language-equivalent of the English lexeme BE. Even though ‘be’ is not a Communicatively-dominant node in the SemR (i.e., the utterance is an assertion of “salmonhood” rather than of “being”), the constraint-rankings given in (7) guarantee that in English it will be realized as the syntactic predicate, giving us the English sentence this is a salmon. Thus, as shown in Table 2, although ‘be’ violates the second-ranked constraint CDEN, it satisfies SPEN, whereas the nodes ‘salmon’ and ‘thing’ do not. The next-best candidate, ‘dist’, is not part of the Sem Rheme and does not have a natural verbal expression, and therefore is also excluded. Table 2. English constraint rankings for Figure 4 Candidate Node ) ‘be’ ‘salmon’ ‘thing’ ‘dist’

SPEN

VEN

*! *!

*! *! *!

CDEN *

*

REN

* *

The different rankings Lushootseed accords to the constraints in (6) result in the selection of a different entry node, shown in Table 3 on the next page. As in English, the fact that ‘be’ is not the Communicatively-Dominant node of the Sem Rheme triggers a violation of CDEN; however, this constraint is ranked more highly that SPEN, and so ‘be’ is rejected as a candidate. The non-Rhematic nodes, ‘thing’ and ‘dist’, are eliminated by the highest-ranked constraint, REN. It should also be pointed out that in Lushootseed there is no equivalent of BE and, hence, no lexical element to appear in the DSyntS. Instead, SemA 2 of ‘be’, ‘salmon’, becomes the DSynt predicate and takes SemA 1 as its first Deep-Syntactic actant (DSyntA I), the seman-

50

David Beck

teme ‘be’ finding its expression in the construction NPRED → N rather than in a lexical item, as it does in most languages. Table 3. Lushootseed constraint rankings for Figure 4 Candidate Node ) ‘salmon’ ‘be’ ‘thing’ ‘dist’

REN

CDEN

SPEN *

*! *! *!

* *

VEN * * * *

In terms of its Sem-CommS, the sentence in Figure 4 can be compared with the sentence in (9), whose syntactic predicate is the proximal demonstrative UJɎȓɎ and which would be the answer to a question such as “Which stone is it?”: (9) UJɎȓɎUȓÍ‫ޙ‬ś‫ޙ‬BɎ



UJɍȒɍ

UȒ




ÍަśަBɍ PROX NDEM stone ‘The stone is this one.’ (Hess 1995: 81, ex. 5) The SemR for this sentence is shown in Figure 5:

Figure 5. UJɎȓɎUȓÍ‫ޙ‬ś‫ޙ‬BɎ ‘the stone is this one

Unlike Figure 4, however, the DSyntR in Figure 5 takes the Rhematic SemA 1 of ‘be’ as its syntactic predicate rather than SemA 2 — in other words, the entity whose identity is established by the predication becomes the DSynt-predicate, rather than the identity attributed to it, as in Figure 4. The source of this difference lies in the Sem-CommS. In Figure 4, SemA1 (the entity being identified) belongs to the Sem Theme and SemA 2 (the identity attributed to it) lies within the Sem-Rheme. The fact that Lushootseed re-

Communicative structure in Lushootseed 51

quires that the syntactic predicate belong to the Rheme results in SemA 2 surfacing as the DSynt predicate in Figure 4 and Sem A 1 becoming predicate in Figure 5 — the CommS of the sentence, rather than its propositional content, determining the DSyntS. The contrast in constraint-rankings quite easily accounts for these differences. The most natural English sentence expressing this SemR is something like this is the stone, where the entry node continues to be the semantic predicate ‘be’. This outcome is predicted by Table 4. Table 4. English constraint rankings for Figure 5 Candidate Node ) ‘be’ ‘stone’ ‘thing’ ‘prox’ ‘non-dem’

SPEN

VEN

*! *!

*! *! *! *!

CDEN *

REN *

* *

*

Once again, although ‘be’ is not the Communicatively Dominant node, it is selected as entry node because it is a semantic predicate and because its most natural expression is verbal. ‘Stone’ and ‘thing’ are not semantic predicates, whereas the semantic predicates ‘prox’ and ‘non-dem’ are predicates, but have no natural verbal expression in English.For Lushootseed, the relatively low-ranking of the constraints SPEN and VEN select ‘thing’ as the entry node as in Table 5: Table 5. Lushootseed constraint rankings for Figure 5 Candidate Node ) ‘thing’ ‘stone’ ‘be’ ‘prox’ ‘non-dem’

REN

CDEN

*!

*!

*! *! *!

SPEN * *

VEN * * * * *

As a result, the syntactic predicate becomes UJɎȓɎ
‘this one’, lexicalizing the configuration of semantemes ‘prox’ → ‘thing’ as a demonstrative. The same principles come into play in expressions like (4), shown in Figure 6, which answers the question “who chased the dog?”.

52

David Beck

Figure 6. XJX‫ޙ‬TVUJɎȓɎɎVÍBMBEUJɎȓɎTRɯȓCBZɎ ‘those who chased the dog are the children’

The English constraint-rankings for Figure 6 are shown in Table 6. In this case, the high ranking of SPEN and VEN ensure that ‘chase’, which has a lexical equivalent in the verb CHASE, is selected as the entry node, giving us the children chased the dog, the marked prosody indicating that, exceptionally, the Thematic element is not subject. Table 6. English constraint-rankings for Figure 6 Candidate Node ) ‘chase’ ‘children’ ‘dog’ ‘prox’

SPEN

VEN

CDEN

*! *!

*! *! *!

* *

REN * * * *

As shown in Table 7, the Lushootseed treatment of the same SemR gives different results: Table 7. Lushootseed constraint-rankings for Figure 6 Candidate Node

) ‘children’ ‘chase’ ‘dog’ ‘prox’

REN

CDEN

SPEN *

VEN *

*! *! *!

* *

*

* *

Communicative structure in Lushootseed 53

Here, as in Figure 3, there is an eligible semantic predicate in the SemS, ‘chase’ which is most naturally lexicalized as a verb, Ì"-"% ‘chase’; however, ‘chase’ is Thematic rather than Rhematic, and so the constraint rankings select the node ‘children’ rather than the node ‘chase’, reflecting the relatively low-ranking of the part-of-speech related constraints SPEN and VEN. The high precedence Lushootseed gives to the Sem-Comm status of elements thus leads to the creation of the nominally-predicated sentence XJX‫ޙ‬TVUJɎȓɎɎVÍBMBEUJɎȓɎTRɯȓCBZɎ

‘those who chase the dog [are] the children’. The form of the headless relative clause that appears as DSyntA I (that is, as the argument that will become subject in SSyntR) is the result of the interaction of further sets of constraints governing syntacticization (specifically, constraints requiring the realization of DSyntAs of verbs as nouns or nominal expressions).8 4.2. Thematicity and voice Lushootseed makes use of passive voice in a way that parallels its function in many of the world’s languages, that of ensuring the alignment of sentencelevel Theme (and, by extension, Discourse Topic) and the syntactic relation subject (Keenan 1976; Li and Thompson 1976). We can begin to examine this issue by comparing the SemR ৑ DSyntR transition shown in Figure 3 above with that in Figure 7, which represents the sentence in (2). The SemR in Figure 3 represents a sentence where the SemA 1 of ‘seek’, ‘boy’, is Thematic and is realized as the DSyntA I cum subject of the active voice form HɯȓÍ‫ޙ‬ȓE ‘seek’. However, in the SemR in Figure 7 (a felicitous answer to the question “What happened to the dog?” or part of a discourse episode centred on the dog), the Theme is SemA 2, the PATIENT ‘dog’. In such a context, a DSyntR analogous to that shown in Figure 3 is ruled out by the constraint given in (10): (10) Thematic DSyntA I (ThDAI) DSyntA I of the main predicate must be Thematic. This constraint is a formalization of the strong tendency in natural languages for subjects to be Thematic and Topical. In Lushootseed, it requires the use of the passive ÍBMBUȓC ‘be sought’, whose government pattern assigns the expression of the SemA 2 of ‘seek’ to the role of DSyntA I rather than assigning this role to SemA 1 as in the active voice.

54

David Beck

Figure 7. ɎVHɯȓÍ‫ޙ‬UȓCɎȓUJÍ‫ޙ‬BÍ‫ޙ‬BTUJTRɯȓCBZɎ ‘the dog was looked for by the boy’

Because of the central role that Thematicity plays in Salishan referencetracking, the constraint in (10) is virtually inviolable for Lushootseed, the only higher-ranked constraint being an essentially morphological restriction on passives which blocks the realization of first- and second-person agentive complements (AgCos) (Hukari 1976; Jelinek and Demers 1983): (11)

3rd Person Passive AGENT (3PassAgt) The DSynt II of a passive verb must be third-person.

This makes the passive in Lushootseed a much more straightforward proposition than it is in, say, English or many other familiar languages where it is more closely tied to textual, stylistic, and discourse-level considerations which are related to, but potentially independent of, Sem-CommS proper (see, for example, Givón 1994; Shibatani 1988). The Sem-CommS in Figure 7 can also account for another feature of passive sentences in Lushootseed — the potential variation in constituent order alluded to earlier that would allow for either the AgCo » Sub order given in (2) or the alternative Sub » AgCo order (i.e., ɎVHɯȓÍ‫ޙ‬UȓC UJ TRɯȓCBZɎ Ɏȓ UJ Í‫ޙ‬BÍ‫ޙ‬BT). Because the linear ordering of constituents in a dependency-based syntax is considered a means of expressing the syntactic relations represented by dependency trees (on a par with affixation and other morphological means), the linearization of SyntR is accomplished by rules and constraints governing the correspondences between SSyntR and Morphological Representation, which in Lushootseed include the two in (12): (12)

Contiguous Syntactic Rheme (CSR) The linear contiguity of the Syntactic Rheme must not be interrupted by a non-Rhematic NP.9

Communicative structure in Lushootseed 55

SyntPred >> Sub The subject must immediately follow the syntactic predicate. These constraints are ranked in the order they are given in (12), which ensures that in cases like that in Figure 7, where the Sub is Thematic and the AgCo is Rhematic, the AgCo appears adjacent to the verb (preceding the Sub) in order to maintain the contiguity of the Rheme at the expense of the SyntPred » Sub constraint. The alternate order, Sub » AgCo, occurs in those cases where neither constraint is violated — that is, in those cases where the Sub and the AgCo belong to the same part of the Sem-CommS. This situation occurs in one of two situations, either where the sentence is entirely Rhematic (as at the beginning of a discourse episode),10 or where both the Sub and the AgCo are Thematic, as in Figure 8 on the next page. Sentences like this are found in contexts where both participants are Topical or Given, but the speaker feels it necessary to express both as full NPs. During syntacticization, the Rhematic element is selected as entry node, while choice of DSyntA I (which determines the voice of the matrix predicate) is governed by the constraint against two non-oblique nominal constituents in transitive clauses: in order to specify both participants without triggering pronominalization of the subject, it is necessary to passivize the verb and express the AGENT as an oblique AgCo. This DSyntS can be linearized so that it satisfies both of the constraints in (12), resulting in the order V » Sub » AgCo. These constraints are two of many constraints on linearization based on Thematicity, and in this respect Lutosheed is no different from many other languages where Communicative Structure plays a central role in determining the linear ordering of sentence constituents.

Figure 8. ɎVHɯȓÍ‫ޙ‬UȓCUJTRɯȓCBZɎɎȓUJÍ‫ޙ‬BÍ‫ޙ‬BT ‘the dog was looked for by the boy’

56

David Beck

5. Effects of Focalization on syntactic structure Like Thematicity, Focalization has certain (albeit less spectacular) effects on Lushootseed syntax, being marked by the particle EJȦ, which indicates a Focalized Theme (ū Lambrecht’s 1994 “focalized topic”) and appears in DSyntS as a modifier of the matrix predicate: (13) EJȦɎVRBEBEJD



EJȦ  ˜   ɎV‫ޓ‬RBEB‫ޓ‬EJ‫ޓ‬U‫ޓ‬T FOC 3SUB PFV–steal–SS–ICS–1SG.OBJ ‘He’s the one who robbed me of it.’ (Hess 1998: 67, line 67) This sentence comes from a story at a point where Mink catches sight of another character, %VLɯJCȓȦ ‘Changer’, who stole Mink’s roasting salmon in the previous discourse episode. The SemR of this sentence, based on the three-place predicate ‘rob’ (that is, ‘X takes Y from Z against Z’s wishes’), is shown in Figure 9. The Focalized area of the SemR is coterminous with the Semantic Theme, which is realized as DSyntA I; the entry node for syntacticization, as determined by the constraint rankings in (8) is ‘rob’, which corresponds to the trivalent verb RBEBEJE ‘rob somebody of something’. The government pattern of this verb expresses the person robbed (ɎȓDB ‘I’) as DSyntA II (the eventual direct object, realized as a pronominal suffix in MorphR) and the THEME (TɎVMBEYɯ ‘salmon’) as DSyntA III — the latter being elided between DSyntR and SSyntR. Likewise, the DSyntA I is pronominalized in SSyntR. The focalizing particle itself appears as a modificative dependant of the verb, behaving syntactically as an adverbial particle (Hess 1995) which takes semantic scope over the Thematic DSyntA I/subject of the clause. Precisely the same principles come into play in sentences with nominal predicates and Focalized Themes like that in (14) on the next page.

Figure 9. EJȦɎVRBEBEJD ‘he’s the one who robbed me of it’

Communicative structure in Lushootseed 57

(14) EJȦUVCBETLɯJɎVɎBUȓCȓE



EJȥ

UVCBE‫ޠ‬T





LɮJ

ɍV‫ޠ‬ɍBUȒCȒE FOC PAST=father–3PO REM PFV–die ‘The one who died was his father.’ (based on Hess 2006: 31, line 227)11 The sentence occurs at a place in a text where a character, Coyote, has faked his own death and is now impersonating his son (whom he has tricked into becoming lost in the Sky World so that he (Coyote) can appropriate his son’s wives). Coyote, dressed as his son, puts on a show of mourning and repeatedly declaims in a loud voice that the deceased is his father (that is, Coyote himself). The deceased in such a context is Thematic, but it is also Focalized in that the deceased is the focal point of the speaker’s attention; the Rhematic content of the utterance is the identity (‘his father’) of that focal point. This gives us the SemR in Figure 10 below.12 Once again, the selection of the entry node is made by using the constraint-rankings shown in (8). In this case, the configuration of semantemes ‘father’ –1–> ‘person’ is lexicalized as the relational noun BAD ‘father’ in the DSyntR and becomes the DSynt predicate. The Thematic area of the SemR contains the predicate ‘die’, but also overlaps with the Rheme to the extent that it includes the SemA 1 of ‘die’, ‘person’, as its Comm-Dominant node (hence, the expression of the Theme as a headless relative clause in DSyntR).13 The entire Semantic Theme is Focalized and, in DSyntR, is realized as DSyntA I.

. Figure 10. EJȦUVCBETLɯJɎVɎBUȓCȓE ‘the one who died was his father’

One thing that is not entirely clear from the examples so far is whether or not the scope of EJȦ is determined by syntactic or communicative considerations — in other words, whether the Focalizing particle takes scope over the

58

David Beck

DSyntA I or over the Thematic NP in DSyntRGiven the near-absolute correspondence between subjecthood and Thematicity, this distinction is difficult to tease apart; however, the sentence in (15) seems to show that it is Sem-CommS rather than syntax that determines the scope of the Focalizing particle: (15) EJȦDȓȦHɯȓLɯȓEBEUJɎJȦ




EJȥ

DȒȥ



HɮȒLɮȒEB‫ޠ‬E

UJɍJȥ FOC 1PL.SUB SBJ=taken–ICS DIST ‘That guy’s the one we should get.’ (Hess 2006: 18, line 167) This sentence is taken from a context where two brothers, looking for someone to help rescue their sister who is trapped in a tree, have been discussing the talents of a particular person (ś‫ޙ‬ȓś‫ޙ‬JR‫ޙ‬ġȓE ‘Sapsucker’). The SemR of (15) is given in Figure 11 below. Much as in Figure 10, we see that the Focalized area of the SemR corresponds to the Semantic Theme, and the entry node for the DSyntS is selected according to the constraint-rankings given in (8). Note, however, that contrary to the constraint given in (10), which requires that the Theme be realized as the DSyntA I (which in a case like this would normally require the passive form of the verb), here the Theme is realized as DSyntA II, passivization being blocked (as per the constraint in (11) above) by fact that the agentive SemA 1 is first-person plural. Nevertheless, the Focalization clearly pertains to the Thematic DSyntA II rather than DSyntA I (ÍȓȦ ‘we’), showing that the scope of the Focalizing particle is determined by Thematicity rather than by syntactic relations.14

Figure 11. EJȦÍȓȦHɯȓLɯȓEBEUJɎJȦ ‘that guy’s the one we should get’

Communicative structure in Lushootseed 59

6. Conclusion As the preceding discussion illustrates, the Communicative Structure of Lushootseed sentences is one of the primary determinants of their syntactic structure. This is seen most dramatically in the constraints governing the selection of the syntactic predicate, which in Lushootseed and many other Salishan languages depends on Thematicity rather than on part of speech. The priority placed on Thematicity in this context offers a striking typological contrast to the majority of the world’s languages, which place a much higher priority on considerations of semantics and part of speech in the selection of syntactic predicates. Approaching the issue of Lushootseed clause structure from the point of view of Communicative Structure also has the advantage of resolving some long-standing debates in Salishan studies concerning the (non-)existence of the distinction between nouns and verbs, allowing us to keep this useful (and probably universal) distinction while at the same time accounting for the unique properties of Salishan syntax. Notes 1. In Meaning-Text Theory there are two levels of Syntactic Representation, Deep and Surface, the former based on a limited number of Deep Syntactic Relations (DSyntRels) such as Actantial (= argument) and Modificative, and the latter based on Surface Syntactic relations such as subject, direct object, etc., which have more language-specific properties (Mel’ìuk 1988). 2. DSyntA I maps onto Subject in Surface Syntactic Representation (SSyntR) by a general correspondence rule, while DSyntA II maps onto an oblique object governed by the preposition FOR (as specified by the government pattern of LOOK(FOR)). Definiteness in the SemR and the determiners that express it have not been included to avoid complicating the discussion. 3. The terms Rheme and Theme are also used by Halliday (1970); essentially the same contrast is commonly referred to as Topic vs. Comment (Sapir 2004) or Topic vs. Focus (Lambrecht 1994; Sgall, Hajiìová, and Benešová 1973). Frequently, the notion of Theme is equated with the notion of Topic in the sense of discourse topic. Although the two concepts are clearly related, they are also potentially independent of one another, Theme being a sentence-level concept and Topic belonging to the level of discourse. 4. The abbreviations used in glosses are: 1,2,3 = first-, second-, third-person; ADD = additive; ATTN = attenuative; DEF = definite; DIST = distal; ICS = internal causative; INTJ = interjection; NDEM = non-demonstrative; OBJ = object; PASS = passive; PFV = perfective; PL = plural; PR = preposition; PROX = proximal; REM = remote; SBJ = subjunctive; SG = singular; SS = secondary suffix; SUB = subject.

60 David Beck 5. Cf. van Eijk and Hess (1986), Kroeber (1999), and Beck (2002). 6. A more detailed discussion of Comm-Dominance and methods for determining the Comm-Dominant node can be found in Iordanskaja and Polguère (1988) and Mel’ìuk (2001: 29ff.). 7. This, of course, may depend on language-specific characteristics of the lexicon such as the (non-)existence of verbal expressions for certain semantemes (the case in point being the absence of a verbal expression of ‘be’ in Lushootseed — see below). The term “most naturally” is deliberately vague, allowing for lexical and other types of idiosyncrasies to override more rigid considerations such as the existence in the lexicon of a direct expression of a particular semantic predicate (viz. the case of the English expression BE HUNGRY which is the most common expression of the stative predicate ‘hunger’, in spite of the existence of the verb HUNGER). 8. The elision of the co-referential NPs is part of pronominalization in the DSyntR ৑ SSyntR component of the rules. 9. It can, however, be interrupted by subject-markers and other types of clitics. 10. It should also be noted that an all-Rhematic “out-of-the-blue” transitive sentence in Lushootseed will always be passive, given the constraint against the realization of both an NP subject and NP direct object in the clause, which would normally trigger obligatory pronominalization (see the discussion of (1e) above). Given the absence of an antecedent for a pronominal in an out-of-the-blue context, passivization is the only viable option. 11. The sentence as it appears in the text is EJȦUVCBET UVTCJBX LɯJɎVɎBUȓCȓE
‘the one who died, it was his father, Coyote’. The appositive phrase UVTCJBX ‘Coyote’ has been removed to avoid cluttering the SemR given below. 12. This diagram has been simplified to avoid getting bogged down in some technical issues. Most notably, temporal and spatial deixis has been omitted from the SemR. The demonstrative pronoun that serves as the head of the relative clause in the DSyntR owes its value (‘proximal’) to considerations of discourse which have yet to be carefully unraveled, and are in any case somewhat orthogonal to our discussion. I have also treated CȓEBɎɎȓTCJBX ‘son of Coyote’ as a proper name rather than as a periphrastic expression which should be decomposed in SemR and represented by a tree in DSyntR. 13. On overlaps in the Sem-CommS, see Melìuk (2001: 275 – 276). 14. Unlike Focalization of Themes, the Focalization of Rhemes (the CommS corresponding to clefts in English and related languages) in Lushootseed is not overtly marked for non-verbal elements. There is a special construction of Focalized verbal predicates that uses the nominalization of a verb phrase as the head of a non-verbal predication.

Communicative structure in Lushootseed 61

References Beck, David 1997 Theme, Rheme, and Communicative Structure in Lushootseed and Bella Coola. In Recent trends in Meaning-Text Theory, Leo Wanner (ed.), 93– 135. Amsterdam: John Benjamins. Beck, David 2000 Semantic agents, syntactic subjects, and discourse topics: How to locate Lushootseed sentences in space and time. Studies in Language 24: 277– 317. Beck, David 2002 The Typology of Parts of Speech Systems: The Markedness of Adjectives. New York: Routledge. Chafe, Wallace 1976 Givenness, contrastiveness, definiteness, subjects, topics, and point of view. In Subject and Topic, Charles N. Li (ed.), 25–55. New York: Academic Press. Czaykowska-Higgins, Ewa, and M. Dale Kinkade 1998 Salish languages and linguistics. In Salish Languages and Linguistics: Theoretical and Descriptive Perspectives, Ewa Czaykowska-Higgins and M. Dale Kinkade (eds.), 1–68. Berlin: Mouton de Gruyter. Davis, Philip W., and Ross Saunders 1978 Bella Coola syntax. In Linguistic Studies of Native Canada, Eung-Do Cook and Johnathan Kaye (eds.), 37–65. Vancouver: University of British Columbia Press. Gerdts, Donna B. 1988 Object and Absolutive in Halkomelem Salish. New York: Garland. Givón, Talmy 1994 Voice and Inversion. Amsterdam: John Benjamins. Halliday, M.A.K. 1970 Language structure and language function. In New Horizons in Linguistics, John Lyons (ed.), 140–165. Harmondsworth: Penguin. Hess, Thomas M. 1973 Agent in a Coast Salish language. International Journal of American Linguistics 39: 89–94. Hess, Thomas M. 1995 Lushootseed Reader with Introductory Grammar, Volume I. Missoula: University of Montana Occasional Papers in Linguistics.

62 David Beck Hess, Thomas M. 1998 Lushootseed Reader, Volume II. Missoula: University of Montana Occasional Papers in Linguistics. Hess, Thomas M. 2006 Lushootseed Reader, Volume III. Missoula: University of Montana Occasional Papers in Linguistics. Hess, Thomas M., and Vi Hilbert 1976 Lushootseed: An Introduction, Books 1 and 2. Seattle: American Indian Studies, University of Washington. Hukari, Thomas E. 1976 Person in a Coast Salish Language. International Journal of American Linguistics 42: 305–318. Iordanskaja, Lidija N. 1990 Ot semantiìeskoj seti k glubinno-sintaksiìeskomu derevu: pravila naxoždenija veršiny dereva [From semantic sets to deep syntactic trees: Rules for the selection of the top nodes of trees]. In Metody formalne w opisie j‫ؚ‬zyków słowiaēskich, Festschrift Jurij Apresjan, Z. Saloni (ed.), 33–46. Białystok: Universytet Warszawski. Iordanskaja, Lidija N., and Alain Polguère 1988 Semantic processing for text generation. Proceedings of the International Computer Science Conference '88: 310–318. Jelinek, Eloise, and Richard A. Demers 1983 The agent hierarchy and voice in some Coast Salish languages. International Journal of American Linguistics 49: 167–85. Jelinek, Eloise, and Richard A. Demers 1994 Predicates and pronominal arguments in Straits Salish. Language 70: 697–736. Keenan, Edward 1976 Toward a universal definition of “subject”. In Subject and Topic, Charles N. Li (ed.), 303–333. New York: Academic Press. Kinkade, M. Dale 1983 Salishan evidence against the universality of “noun” and “verb”. Lingua 60: 25–40. Kinkade, M. Dale 1990 Sorting out third persons in Salish discourse. International Journal of American Linguistics 56: 341–360. Kroeber, Paul D. 1999 The Salishan Language Family: Reconstructing Syntax. Lincoln, NB: University of Nebraska Press.

Communicative structure in Lushootseed 63 Kuipers, Aert H. 1968 The categories verb-noun and transitive-intransitive in English and Squamish. Lingua 21: 620–626. Lambrecht, Knud 1994 Information Structure and Sentence Form: Topic, Focus, and the Mental Representation of Discourse Referents. Cambridge: Cambridge University Press. Li, Charles N., and Sandra A. Thompson 1976 Subject and topic: A new typology of language. In Subject and Topic, Charles N. Li (ed.), 457–489. New York: Academic Press. Mel’ìuk, Igor A. 1974 Opyt teorii lingvistiìeskix modelej “Smysl ඼ Tekst” [An outline of a theory of "Meaning ඼ Text" linguistic models]. Moscow: Nauka. Mel’ìuk, Igor A. 1981 Meaning-Text Models: A recent trend in Soviet linguistics. Annual Review of Anthropology 10: 27–62. Mel’ìuk, Igor A. 1988 Dependency Syntax: Theory and Practice. Albany, NY: SUNY Press. Mel’ìuk, Igor A. 2001 Communicative Organization in Natural Language: The SemanticCommunicative Structure of Sentences. Amsterdam: John Benjamins. Prince, Alan, and Paul Smolensky 2004 Optimality Theory: Constraint Interaction in Generative Grammar. Malden, MA: Blackwell. Sapir, Edward 2004 Language: An Introduction to the Study of Speech. New York: Dover. Sgall, Petr, Eva Hajiìová, and Eva Benešová 1973 Topic, Focus, and Generative Semantics. Kronberg/Taunus: Kromber. Shibatani, Masayoshi 1988 Passive and Voice. Amsterdam: John Benjamins. Vallduví, Enric 1992 The Informational Component. New York: Garland. van Eijk, Jan, and Thomas M. Hess 1986 Noun and verb in Salishan. Lingua 69: 319–31. Žolkovskij, A.K., and Igor A. Mel’ìuk 1965 O vozmožnom metode i instrumentax semantiìeskogo sinteza [On a possible method and instruments for semantic synthesis]. Nauìnotexniìeskaja informacija 5: 23–28.

Null subjects in Shipibo switch-reference systems José Camacho and José Elías-Ulloa

1. Introduction The availability of null subjects varies from language to language. On one extreme of the continuum, we have English-like languages where sentences must have phonologically overt subjects, as shown in (1). (1)

a. I bought a car. b. *pro bought a car

At the other end, we find Spanish-like languages, where subjects can be either explicit or null, as seen in (2). When they are null, the general assumption is that there is a subject pronoun in the structure of the sentence, pro, but that the pronoun has no phonological content. (2)

a. Yo compré un carro. I bought a car ‘I bought a car.’ b. pro compré un carro.

Between these two extremes, a number of languages show mixed properties in regards to the availability of null subjects; that is, in some languages subjects in some grammatical persons must be obligatorily overt while others can optionally be null. This is the case of Shipibo, a Panoan language, in which 1st and 2nd person subjects are obligatory but 3rd person ones are optional. In order to understand how unusual the Shipibo null subject sytem is, we will start by examing the best-known cases of languages with mixed nullsubject system; namely, Hebrew and Finnish, which are the mirror image of Shipibo. As Vainika and Levy (1999) and Levy & Vainikka (2000) show, 1st and nd 2 person subjects can be null but 3rd person subjects must be explicit in both Hebrew and Finnish. The data in (3) and (4) show this for Hebrew (Levy & Vainikka 2000:365).

66 José Camacho and José Elías-Ulloa

(3)

(Ani) aliti al ha-rakevet. (I) step-PAST/1SG on the-train

‘I boarded on the train.’ (4)

*Ala al ha-rakevet. stepped/3sgm on the-train ‘*(He) boarded on the train.’

A complete paradigm of the mixed null subjects of Hebrew (Artstein 1998:3) is shown in (5). The table presents the verb ‘to eat’ conjugated in past and future tenses (third and fourth columns respectively) for all the grammatical persons. Pronouns in parentheses can be omitted, the rest are obligatory. Observe that in (5), all the verb forms have morphological person agreement with the exception of the past tense 3rd person singular masculine. Thus, it is not possible to claim that, in Hebrew, subjects become obligatory when there is no person agreement morphology. Moreover, Vainikka & Levy (1999:640) note that Hebrew only allows for null subjects when “a discernable phonological similarity” exists between the subject pronouns and the agreement morphology of the verb. This “discernable phonological similarity”, as Vainikka & Levy call it, seems to be relatively abstract and not of a synchronic nature. (5)

Hebrew Mixed-Null Subjects in Past and Future Tenses

Person 1.sg 2.m.sg 2.f.sg 3.m.sg 3.f.sg 1.pl 2.m.pl 2.f.pl 3.m.pl 3.f.pl

Pronoun (ani) (ata) (at) Hu Hi (anaxnu) (atem) (aten) Hem Hen

ate+infl. axal-ti axal-ta axal-t axal axl-a axal-un axal-tem axal-ten axl-u axl-u

infl.+will.eat+infl. ٧-oxal t-oxal t-oxl-i y-oxal t-oxal n-oxal t-oxl-u t-oxal-na y-oxl-u t-oxal-na

Table (6) shows the Hebrew the agreement morphemes that attach to verbs in past and future tenses (from Vainikka & Levy 1999:644)1 and the corresponding subject pronoun. The grey columns indicate those cases for

Null Subjects in Shipibo switch-reference systems

67

which Vainikka & Levy (1999) assert there is no phonological similarity between subjects and the agreement morphology. (6)

Hebrew: Subjects and Agreeement for Past and Future Tenses

Past Future Pronouns Past Future Pronouns

1.sg -ti eani 1.pl -nu neanaxnu

2.sg.m -ta teata 2.pl.m -tem te- -u Atem

2.sg.f -t te- -i at 2.pl.f -ten te- -na aten

3sg.m yehu 3.pl.m -u ye- -u hem

3sg.f -a teHi 3.pl.f -u te- -na Hen

Thus, overt subject pronouns are obligatory in Hebrew when there is no phonological similarity between the subject pronoun and the corresponding person agreement on the verb. This occurs in the 3rd person in the past and future (see Table (6)) and in the entire present tense, which does not mark person distinctions, although it does mark gender and number (see data in (7) from Vainikka & Levy 1999:646). In those two cases, subjects become obligatory, they cannot be null. (7)

*Zoxer et ha-tshuva. Remember-PRESENT/SGM ACC the-answer ‘(I/you/he) remember(s) the answer.’

Putting aside the phonological relation between subject pronouns and agreement and its effect on the licesing of null subjects, it is worth noting that Hebrew does allow the occurrence of null subjects in the 3rd person but in very specific contexts: certain embedded clauses, expletive constructions and the generic impersonal construction (Vainikka & Levy 1999:647). Given that two unrelated languages like Hebrew and Finnish present the same mixed null-subject system, Vainikka & Levy (1999) and Levy and Vainika (2000) and Artstein (1998) independently coincide in asserting that there cannot exist a language that is the mirror image of Hebrew and Finnish; namely, that subjects of the 3rd person can be null whereas subjects of the 1st and 2nd persons are obligatory. Vainikka & Levy (1999) and Levy & Vainikka (2000) support this conclusion by claiming that languages with mixed null-subject systems are derived from the different positions in which the features of person agreement can be based-generated together with a licesing principle. Thus, in Spanish-

68 José Camacho and José Elías-Ulloa

like languages, these features are based-generated in the subject position, whereas in English-like languages, they are based-generated in AGR. In the case of languages with a mixed null-subject system, like Hebrew and Finnish, 3rd person features of subjects are generated in AGR so subjects cannot be null. In contrast, the features of 1st and 2nd person subjects are basedgenerated in the subject position, therefore they can be null. Levy & Vainikka add that UG does not allow languages that are the mirror image of Hebrew and Finnish because of pragmatically based notions of discourse participants and referentiality. Thus, for 1st and 2nd person, the set of potential referents are restricted in a conversational exchange (namely, speaker and hearer); whereas for 3rd person, the set of referents is open. Levy & Vainikka claim that this pragmatic distinction is encoded in the syntax of mixed null-subject systems so that the 1 st and 2nd persons can only be basedgenerated in the subject position and the 3rd person in the AGR position. Shlonsky (2009) proposes a slightly different explanation of the Hebrew person partition, assuming, as we will, a Speech Act Participant functional projection, and that 1st and 2nd person pronouns cliticize onto this projection. Within Optimality Theory (Prince & Smolensky 2004), Artstein (1998) aligns the person and animacy scale as proposed in Aissen (1999) (see 8) with a reduction scale (see 9). The result is two universal constraint rankings that predict the attested occurrence of non-null subject English-like languages and null subject Spanish-like languages. They also predict the attested existence of mixed null subject Hebrew/Finnish-like languages. Furthermore, they ban the fourth logical possibility: a mixed null subject language where the 3rd person can be omitted but 1st and 2nd persons cannot. (8)

Person/Animacy Scale (Aissen 1997, 1999) 1st/2nd p.>Proper Noun 3rd>Human 3rd>Animate 3rd> Inanimate 3rd

(9)

Reduction Scale Null > Overt

In this analysis, elements appearing on the high end of a hierarchy and the low end of another give rise to marked configurations that the language avoids. After applying Alignment, Artstein (1998:2-5 and 1999:4-6) obtains the universal rankings in (10).2 (10)

Universal Rankings Cnull: *NULL/3.INANIMATE >> … >> *NULL/3.HUMAN>> *NULL/1,2

Null Subjects in Shipibo switch-reference systems

69

Covert: *OVERT/1,2 >> *OVERT/3.HUMAN >> … >>*OVERT/3.INANIMATE Artstein (1998, 1999) ties the availability of null subjects in languages to the markedness relations expressed by the rankings in (10) and obtains two important implications from those rankings. (11)

Implication I: If a language allows null arguments in the 3rd person, it will also allow them in the 1st and 2nd persons.

(12)

Implication II: If a language requires 1st and 2nd person arguments to be overt, it will also require 3rd person arguments to be overt.

Thus, Artstein (1998, 1999) concludes that for mixed null subject systems: “if a language has a null subject system conditioned by person, it must work in the same direction as in Hebrew”. As we will see shortly, Shipibo presents a clear counterexample to these analyses. Shipibo is a language in which subjects of the 1st and 2nd person cannot be null in declarative clauses, but 3rd person subjects can. In addition, unlike Hebrew and Finnish, Shipibo does not have person agreement morphology (except for the verbal suffix –kan, which indicates 3rd person plural), which means that the presence of a phonological relation between subject pronouns and person agreement like that found in Hebrew and Finnish is not necessary for a mixed null-subject system to emerge.

2. Shipibo null-subjects As first noted by Elías-Ulloa (2001, 2008) for Capanahua, Pano languages have a mixed null-subject system. Like Hebrew and Finnish, Shipibo also groups 1st and 2nd person vs. 3rd person, but the pattern is the opposite: 1st and 2nd person subjects must be overt, 3rd person can be null. Thus, example (13)a can only be interpreted as having a null 3rd person subject. If an overt 1st (or 2nd) person pronoun is inserted, as in (13)b-c, then the subject is interpreted as 1st (or 2nd) person. Notice that there are no changes in verb morphology, in particular, there is no agreement morphology on the verb, (but see below). (13)

a. Lima-n-ra noko-ke. Lima-DIR-EVID arrive-PERF ‘He/she went to Lima.’

70 José Camacho and José Elías-Ulloa

b. Ea-ra Lima-n noko-ke. 1P- EVID Lima-DIR arrive-PERF ‘I arrived in Lima.’ c. Mia-ra Lima-n noko-ke. 2P- EVID Lima-DIR arrive-PERF ‘You arrived in Lima.’ The pattern shown in (13) is fairly robust, in particular, speakers presented with null clauses such as (13)a do not interpret them as null versions of (13)b or c, and when prompted to do so, they either reject them, or return an example with an overt subject. Agreement between a 3rd person plural pronoun and a verb with an overt agreement marker –kan is obligatorily, as seen in (14)-(15). The suffix –kan is mutually exclusive with subject pronouns of 1st and 2nd person, as shown for 2nd person in (cf. (14)a vs. b). 3rd persons plural pronouns must appear with –kan (cf. (15)a vs. (15)b), but a 3rd person singular pronoun cannot (cf.(15)c). Thus, for overt agreement only happens in 3rd p. pl. (14)

a. *Maton-ra atsa pi-kan-ke. 2P.PL-EVID yucca eat-PL-PERF b. Maton-ra atsa pi-ke. 2P.PL-EVID yucca eat-PL-PERF ‘You (pl.) ate yucca.’

(15)

a. *Westiora ipo FKachiԓon-ra A carachama catch-PRIOR.SAME.SUB.TR-EVID jabaon Quique kena-ke. 3P.PL Quique call-PERF ‘They caught a carachama (type of fish) and called Quique.’ b. Jabaon-ra pelota manoma-kan-ke. loose-PL-PERF 3P.PL-EVID ball ‘They lost the ball.’ FKachiԓon-ra ja c. Westiora ipo A carachama catch-PRIOR.SAME.SUB.TR-EVID 3P.SG Quique kena-ke. Quique call-PERF ‘He caught a carachama and called Quique.’

To summarize, main clauses show a split null-subject pattern, where 1st/2nd persons are obligatory, but 3rd person is optional, and only 3rd person plural subjects are overtly marked on the verb.

Null Subjects in Shipibo switch-reference systems

71

2.1 Null-subjects in main and embedded clauses Shipibo has a class of embedded clauses, switch-reference (SR) clauses, headed by a SR morpheme that indicates, among other things, whether the subject is the same or different as the subject of the reference clause3 (Valenzuela 2003, Camacho 2008a). The pattern of null subjects in these clauses complements the picture we have just presented in interesting ways. When a SR clause appears with a main clause, the null subject paradigm partially resembles the main clause situation described above. In particular, 1st/2nd persons cannot be null in both clauses, as illustrated in (16), which can only be interpreted as a 3rd person, not as a 1st/2nd person subject (‘I/you/we grew yucca and cut the leaves.’). When the subject is 1st person, the pronoun can appear in either clause, as shown in (17), with no clearly discerenible preference for either. When speakers were asked to repeat a sentence like (17), they tended to include an overt subject in the second (main) clause, although they accept it as it stands, without an overt subject in the main clause. (16)

(17)

Jawen atsabo oroԓon-ra pei-bo ate-ke. POSS yucca grow-PRIOR.SAME.SUB.TR-EVID leave-PL cut-PERF ‘He grew yucca and cut the leaves.’ a. En westiora ipo chachi-ԓon-ra carachama catch-PRIOR.SAME.SUB.TR-EVID 1P a Quique kena-ke. Quique call-PERF ‘I caught a carachama (type of fish) and called Quique.’ b. En westiora ipo chachi-ԓon-ra carachama catch-PRIOR.SAME.SUB.TR-EVID 1P a Quique kena-ke. Quique call-PERF ‘I caught a carachama (type of fish) and called Quique.’ en c. Westiora ipo chachi-ԓon-ra a carachama catch-PRIOR.SAME.SUB.TR-EVID 1P Quique kena-ke. Quique call-PERF ‘I caught a carachama (type of fish) and called Quique.’

en 1P

2nd person subjects can also appear either in both clauses, in the main clause, or in the SR-clause (cf. (18)a-c respectively), and if the subject is null

72 José Camacho and José Elías-Ulloa

in both, it can only be interpreted as 3rd person subject, as already mentioned for example (16). (18)

a. Mia Lima-n nokó-ԓon-ra min jawen 2P Lima-DIR arrive-PRIOR.SAME.SUB.TR-EVID 2P POSS artesania-bo toe-ke. handcraft-PL break-PERF ‘You have arrived in Lima and have broken his/her handcrafts.’ min jawen b. Lima-n nokó-ԓon-ra Lima-DIR arrive-PRIOR.SAME.SUB.TR-EVID 2P POSS artesania-bo toe-ke. handcraft-PL break-PERF ‘You have arrived in Lima and have broken his/her handcrafts.’ jawen c. Mia Lima-n nokó-ԓon-ra 2P Lima-DIR arrive-PRIOR.SAME.SUB.TR-EVID POSS artesania-bo toe-ke. handcraft-PL break-PERF ‘You have arrived in Lima and have broken his/her handcrafts.’

Finally, 3rd person subjects can be null in the SR-clause, in the main clause or in both, as seen in (19).4 In (19)a, the subject pronoun jabaon appears in the main clause (after the evidential morpheme –ra), in (19)b, it appears in the SR-clause, in (19)c, the subject is null in both clauses, and in (19)d, the subject is overt in both. Although speakers accepted this last sentence, when they were asked to repeat it back, they dropped the second pronoun, producing the equivalent of (19)b. (19)

a. Nonon oin-ԓon-ra jabaon nonti canoe duck see-PRIOR.SAME.SUB.TR-EVID 3P.PL rambe-kan-ke. turn.around-PL-PERF ‘Seeing the duck, they turned around the canoe.’ b. Jabaon nonon oin-ԓon-ra nonti duck see-PRIOR.SAME.SUB.TR-EVID canoe 3P.PL rambe-kan-ke. turn.around-PL-PERF ‘Seeing the duck, they turned around the canoe.’ c. Nonon oin-ԓon-ra nonti duck see-PRIOR.SAME.SUB.TR-EVID canoe rambe-kan-ke. turn.around-PL-PERF ‘Seeing the duck, they turned around the canoe.’

Null Subjects in Shipibo switch-reference systems

d.

73

Jabaon nonon oin-ԓon-ra jabaon nonti duck see-PRIOR.SAME.SUB.TR-EVID 3P.PL canoe 3P.PL rambe-kan-ke. turn.around-PL-PERF ‘Seeing the duck, they turned around the canoe.’

As these examples show, the distribution of null subjects in SR clauses shows the same overall person asymmetry as in main clauses: 1st and 2nd persons must have an overt subject, but 3rd persons need not. However, the overt subject can appear either in the main or in the SR clause. This distribution is schematized in (20)-(21). (20) 1st/2nd person: a. [SR-clause subj1/2p V] ∅SUB Vmain b. [SR-clause ∅SUB V] subj1/2p Vmain c. [SR-clause subj1/2p V] subj1/2p Vmain d. *[ SR-clause ∅SUB V-SR] ∅SUB Vmain (21) 3rd person: a. [SR-clause subj3p V] ∅SUB Vmain b. [SR-clause ∅SUB V-SR] subj3p Vmain c. [SR-clause subj3p V] subj3p Vmain d. [SR-clause ∅SUB V] ∅SUB Vmain

3. The Typology of null subjects 3.1. The Null-subject parameter As noted earlier, Shipibo raises an important issue for the typology of null subjects. While the best known split systems group 1st/2nd person vs. 3rd person, Hebrew, Finnish and Nez Perce allow null subjects in 1st/2nd person but not in 3rd, whereas Shipibo allows null subjects in 3rd person but not in 1st/2nd. This mirror image does not seem like a coincidence. (22) Null 1st/2nd Null 3rd

Shipibo No Yes

Hebrew, Finnish, Nez Perce Yes No

74 José Camacho and José Elías-Ulloa

As mentioned, Vainika and Levy (1999) account for the person split by arguing that 1st and 2nd persons are necessary participants in the discourse, and as such they are constantly indentified. 3rd person, on the other hand, is neither fixed nor constantly identified in discourse. From this, they derive the Hebrew/Finnish pattern. Shipibo, however, has the mirror image, suggesting that the analysis of split paradigms needs to be revised. To account for the observed typology of null subject languages, we first assume that all clauses have subjects (the so-called EPP requirement), and that null subjects must be locally licensed (cf. Rizzi 1982). The second requirement will provide a mechanism for the null subject to be associated to a discourse referent. Consider, for example, (23) in Spanish. The subject of the last clause is usually identified as the same subject in the first one (Marta). This is possible because the null pro has 3rd person features identified through the agreement on the verb. (23) Marta salió a la calle. Después,[pro subió al bus]. Marta went.out to the street. After pro got.on to-the bus ‘Marta went out to the street. After, she got on the bus.’ Our main intuition is that the availability of null subjects can be accounted for within the theory of binding: null subjects are anaphors that require local binding. In order to develop this intuition, we will adopt a movement version of Binding Theory (cf. Lebaux 1983, Hornstein 2007). Hornstein (2007) proposes an account of Binding Principles A and B that attempts to dispense with binding-specific constraints. In this proposal, anaphors are traces of moved categories, whereas bound pronominals are last resort items inserted when no other option is available. His analysis assumes that reflexives are anaphors that are a result of movement, and bound pronouns are last resort lexical insertion. A numeration containing a null category can end up with an overt anaphor if the anaphor can locally move, or with a last-resort pronoun, if movement is not local. For our present purposes, we adapt Hornstein’s approach to Binding as follows: we assume that some null subjects are anaphors, and as such, they are traces of movement, whereas others are pronominals, as in (24). As anaphors, they are locally bound, as in (25). (24)

A null pro is an anaphor or a pronoun.

(25)

a) Anaphors are traces of movement within a strong phase (a CP). b) Anaphors must have an identifiable copy.

Null Subjects in Shipibo switch-reference systems

75

In addition, we assume that the lexical entry for referential categories (pronouns, DPs) minimally include a set of ϕ-features, possibly a referential index, whether they are thematic or not, and an anaphoric feature that signals the locality of their grammatical antecedent, as illustrated in (26). Arg stands for whether the pronoun must be obligatorily argumental or not, ref for a referential index, and anaph for the fact that the pronoun must have a local antecedent. Thus, they has a referential index and an argumental feature that and there lacks. Pro has unspecified person, number and gender, it may or may not be anaphoric and it has an unspecified thematic role.5 (26)

a. b. c. d. e.

They There Herself Pro Pro

[3p.pl, ref, arg] [3p.sg] [3p.sg. fem, arg, anaph] [_p._num._gen, _arg, anaph, _ș] [_p._num._gen, _arg, _ș]

In addition to these possible pronominal items, we assume that inflectional morphology may have pronominal features, i.e. bear a referential index, in addition to having ϕ-features (cf. Borer 1986, 1989, Ordóñez and Treviño 1999, Camacho 2008b, Shlonsky 2009).

3.1.1. Pro as an anaphor With these assumptions in mind, consider the possible typological configurations of null subjects when pro is an anaphor, as illustrated in (27): pro must have an antecedent within its strong phase (its binding domain), and this antecedent can be either an overt DP, agr or another pro. (27)

Typology of possible copies a) agr-ϕ… proanaph, ϕ b) DP-ϕ … pro-ϕanaph, ϕ c) pro-ϕ… pro-ϕanaph, ϕ

The setting in (27)a yields languages such as Spanish, Italian, Portuguese, etc. In Spanish, for example, a clause like (28)a will have the structure in

76 José Camacho and José Elías-Ulloa

(28)b. The lower copy of the verb and pro are both deleted and the higher copy of the verb’s ϕ-features acts as an overt antecedent for pro. (28) a. Salieron a la calle. went.out to the street ‘They went out to the street.’ b. [IP pro[3p.pl. anaph] [I' salieron[3p.pl., ref, arg] [VP pro[3p.pl. anaph] salieron[3p.pl.] a la calle ]]] The representation in (28)b should be seen as a chain of ϕ-features, one of whose copies surfaces overtly (the one on the inflection of the verb), and allows for the referential index to be anchored in discourse, as we will see below.6 We assume that what is copied in this particular case is a fairly abstract representation of ϕ-features, although in languages such as Capanahua, Elías-Ulloa (2008) has proposed that the subject doubling construction illustrated in (29), where the subject pronominal is obligatorily doubled, could be an instance where the ϕ-features surface partially on the verb, and fully in the two identical upper copies, as illustrated in (29).7 (29)

a. Min ta٧ min yu٧a ȕana-ni-ԙ-ki. You EVID you manioc plant –REM.REM.PAST-1/2P.-DECL ‘You planted manioc (a long time ago)’ b. Min2p ta٧ min2p yu٧a ȕana-ni-ԙ-ki. 2p

The second setting in (27)b yields English-type languages, where the ϕfeatures always surface on the subject, never on the verb, with the exception of 3rd person, singular, present -s. The structure is presented in (30).8 (30)

a. They left yesterday. b. [IP they[3p.pl., ref, arg] [I'left [VP they[3p.pl., ref, arg] left yesterday]]]

The third setting in (27)c directly yields a subset of null subjects in Chinesetype languages. As Huang (1984, 1989) points out, some instances of subject pro in Chinese are obligatorily coreferential with the matrix subject, as in (31)a, his (18). Huang’s analysis treats these instances of pro as cases of control, where the embedded subject lacks an accessible SUBJECT (because there is no AUX), hence the controlling category is the matrix clause. In our

Null Subjects in Shipibo switch-reference systems

77

approach, which follows Hornstein’s (2007), control is in this case an instance of raising pro to the matrix clause, yielding (31)b. (31)

a. Zhangsan qi ma qi de [pro hen lei]. Zhangsan ride horse ride till very tired ‘Zhangsan rode a horse until he got very tired..’ b. DP[3p.sg.] … [CP pro[3p.sg. anaph] [IP pro[3p.sg.anaph] [I' V [VP ] ] ] ]

There is another set of cases in Chinese that need not have an overt antecedent in the immediate clause, which we will turn to below.

3.1.2. Pro as a pronominal A number of languages have null subjects that do not require an antecedent in the same clause, but have it in discourse. In these cases, we would argue that pro is a pronominal, it does not require local licensing, and its interpretation has to be done through a discourse operator. The Chinese examples in (32) (Huang’s (42)a) illustrated this option. I (32)

[pro yi hui dao jia], Zhangsan jiu ku. once return to home Zhangsan then cry ‘As soon as hei/j arrived home, Zhangsani began to cry.’

n Huang’s analysis, the preverbal adjunct clause does not have a controlling domain, because it is not c-commanded by the higher SUBJECT (Zhangsan).9 In one of the interpretations of (32), the coreferential one, the null subject is bound by a topic operator linked to Zhangsan. In the other (noncoreferential interpretation), its antecedent is not present in this piece of discourse.

3.2. Split paradigms We now turn to split paradigms such as Shipibo, Hebrew and Finnish, which raise two independent issues: how to account for why some persons can have null subjects but not others, and why the split paradigms are divided in 1st/2nd person vs. 3rd person. Regarding the first question, we will propose that split

78 José Camacho and José Elías-Ulloa

paradigms instantiate two of the options in (27) above, within the same grammar. In the case of Shipibo, 1st/2nd person are like English. (33)

1st/2nd p. obligatory subjects in Shipibo a. Ea-ra Lima-n noko-ke. 1P- EVID Lima-DIR arrive-PERF ‘I arrived in Lima.’ b. CP Ea[1p, ref, arg]

C' -ra

IP ea[1p, ref, arg]

I'

VP ea[1p, ref, arg] Liman

nokoke V' nokoke

3rd person, on the other hand, is like Chinese, where the antecedent of pro is the higher clause, as shown in (34)-(35). (34)

3rd p. optional subjects in Shipibo Lima-n-ra noko-ke. Lima-DIR-EVID arrive-PERF ‘He/she went to Lima.’

(35)

DP[3p.sg] [CP pro[3p.sg] [CP Limanra [IP pro[3p.sg] [VP pro[3p.sg,] [I' nokoke]]]

In the case of Hebrew and Finnish and Nez Perce, 1st/2nd person are like Spanish/Italian, whereas 3rd person is like English.10 We now turn to the second question, i.e. why are 1st/2nd person grouped vs. 3rd person. As suggested above, languages like Hebrew, Finnish, Nez Perce (cf. Deal 2005) and Shipibo all have split paradigms that allow null subjects in either 1st/2nd person or in 3rd person.11 Let us assume that 1st/2nd person project a Speech Act Participant, as in (36) (cf. Poletto 2000:139, who proposes a HearerP and a SpeakerP below a

Null Subjects in Shipibo switch-reference systems

79

number of other possible left-peripheral positions, and Shlonsky 2009). SAP is restricted to 1st/2nd person subjects. With this additional projection in mind, let us see how subjects with different persons would be derived. In order to do so, consider first the partial list of lexical entries for pronominals in Shipibo in (37)12. (36)

[SAPP SAP [IP I …]]

(37)

Shipibo pronominal lexical settings a) Pro [ _p, _num, _gen, _arg, anaph] b) Ea ‘I’ [1p., sg., arg, ref] c) Ja ‘he/she’ [3p., sg., arg, ref]

Suppose that a pro with the lexical specification in (36)a is merged into a derivation, resulting in (38). In order for it to be interpreted as 1st/2nd person, one option is to have an overt 1st/2nd person pronominal in SAP, as in (39). This corresponds to the setting in (30)b above. (38)

SAPP SAP

IP

pro[_p._gen, _anaph, _arg] I VP (39)

I

SAPP ea[1p. arg, ref]

SAP' SAP

IP

pro[1p. arg, anaph, ref] VP

I' I

If SAPP is left null, there is no local antecedent in the general case. However, in SR clauses it is possible to have a null 1st/2nd person subject, a possibility that we turn to below.

80 José Camacho and José Elías-Ulloa

In Hebrew, Finnish and Nez Perce, since 1st/2nd person are pronominal, that means that they do not require an overt antecedent, and therefore SAPP can remain null. By restricting the reference of 1st/2nd persons to a designated projection (SAPP), we are able to predict split systems along the lines of discourse participants vs. 3rd person. In this sense, the analysis is not very different from that of Levy & Vainikka, however, by allowing null subjects to be either anaphors or pronominals, we are able to account both for the Hebrew-type of split paradigm, and the Shipibo type. 4. Non-local subject 1st/2nd person antecedents in Shipibo: SR clauses Let us consider the possibility that an anaphoric pro in Shipibo is not licensed by SAPP in its clause. In the general case, this derivataion will not be possible, since there is no antecedent for the anaphor. However, in certain cases, if the anaphor can move outside of its local clause, we predict that it could have an overt antecedent in another clause. This is precisely what we observed in SR clauses in section 2.2: if the SR clause has a null subject, the main clause must have an overt (1st/2nd person) one. The schematic configuration is presented in (40) below. (40)

[SAPP ea[1p., arg, anaph] [IP pro[_p., arg, anaph] V [SR-clause [pro[_p., arg, anaph] V]]]]

Following Camacho (2008a), we assume that SR-clauses extend the domain for movement of the anaphor because they have a mechanism of case concord and subject coreference that acts as a bridge to the higher clause. As a result, the anaphor can move from the lower clause to the higher one. The proposed analysis views the availability of an overt subject either in the SR-clause or the reference clause as a case of deleting the higher or lower copy of the anaphor. However, we do not know what drives the preference for deleting one copy or the other. When there are overt, coreferential subjects in both clauses, we assume a derivation similar to the one we have proposed, where neither copy is deleted, as in (41). (41)

[SAPP ea[1p., arg, anaph] [IP pro[_p., arg, anaph] V [SR-clause [ea[_p., arg, anaph] V]]]]

An alternative possibility would be to have distinct instances of the pronominal in each clause, not copies. Thus, there would be no movement from

Null Subjects in Shipibo switch-reference systems

81

one clause to the other, as in . Presumably, one of these pronominals would be focused. In the first structure, the focused pronoun is in the SR clause, in the second structure, it is in the reference clause. (42)

a. [SAPP ea[1p., arg ] [IP pro[_p., arg, anaph] V [SR-clause [ea[_p., arg, foc] V]]]] b. [SAPP ea[1p., arg. foc] [IP pro[_p., arg, anaph] V [SR-clause [ea[_p., arg] V]]]]

The reason to chose (41) over (42) relates to the contrast between 1 /2 persons and 3rd persons discussed in examples (16)-(19). As noted in footnote 4, speakers find two overt 3rd person pronouns much less acceptable than the repetition of 1st/2nd person. This may suggest that 1st/2nd persons have the structure in (41), i.e., they are simple copies of a single pronoun, whereas 3rd person overt pronouns have a different structure, perhaps one along the lines of (42). st

nd

5. Conclusion We have argued that the mixed null subject system of Shipibo can be accounted for by assuming that pro is anaphoric and must have a local antecedent in the case of 1st and 2nd persons. The locality can be extended from its minimal domain to the adjacent clause whenever the minimal domain is a SR-clause. 3rd person subjects, on the other hand, are considered pronominals that do not require an antecedent. We have argued that other split systems like Finnish and Hebrew are similar in having an SAP projection, but differ with respect to whether the null subject is anaphoric or pronominal. The proposed analysis captures the insights of previous accounts of Hebrew and Finnish, namely that the partition of 1st/2nd person vs. 3rd is not a coincidence, but also explains why Shipibo is exactly the mirror image of those languages. Notes 1. Levy & Vainikka 1999:644 do not provide the agreement suffixes for the subject pronouns hen 'they feminine' and aten 'you plural feminine'. We have added them in Table (8) taking as a reference Artstein (1999). 2. In (10), *NULL/X means realize overtly the feature X and *OVERT/X, do not realize overtly the feature X.

82 José Camacho and José Elías-Ulloa 3. Some SR suffixes in Shipibo are sensitive to the transitivity of the verb in the the clause they attach to: suffixes -aԓ, -i, and -noԓ indicate that the adjacent verb is intransitive, and -ԓRQ, -kin, and -noԓRQ. SR morphemes also indicate whether the event happened before, at the same time or after the event of the reference clause. See Camacho (2008a). 4. A speaker suggested that the preferred pattern would have the overt pronoun initially, as in (i), whereas having both subjects overtly (ii) soulds offensive to the addressee, as if the speaker were complaining. This seems to suggest that an overt 3rd person pronoun has a focused, emphatic meaning.

5.

6.

7. 8. 9. 10. 11.

(i.) Jan nonon bi-ԓon-ra Sara kena-ke 3P. duck buy-PREC.SAME.SUB.TR-EVID Sara call- PERF ‘He bought duck and called Sara..’ jan Sara kena-ke (ii.) Jan nonon bi-ԓon-ra 3P. duck buy-PREC.SAME.SUB.TR-EVID 3P. Sara call- PERF ‘He bought duck and called Sara..’ This approach makes some of the parametric variation in null subjects a function of lexical entries: a language may have anaphoric pro or may have pronominal pro as a lexical entry, cf. Deal (2005), and Camacho (2008b). Whether pro obligatorily appears in the higher position depends on a number of theoretical and empirical considerations, in particular, whether the EPP must be satisfied in the Spec, IP. The higher copy is interpreted as a topic, cf. Elías-Ulloa (2008). We are assuming that the pronounced copy of the verb is the lower one in English, although this does not affect our analysis. In that framework, a SUBJECT is either a grammatical subject, or AUX. Or perhaps like Chinese, given that 3rd p. subjects can be null if controlled (cf. Borer 1989, Deal 2005). The data from Sprouse and Vance (1993) on the Swiss Rhaeto-Romance dialect of Surselvan seems to contradict this tendency. This language only allows null subjects in out-of-the-blue matrix clauses with 2nd person subjects, and noninitial verb, as in 0-(iii). Given the right context (for example in embedded clauses), 3rd person can also be null (and 1st person, presumably, as Vance, p.c. has suggested). (i.)

(ii.)

Ier veis cumprau ina tastga. yesterday have-2p.pl bought a bag ‘Yesterday you bought a bag.’ *Ier han cumprau ina tastga. yesterday have-3pl bought a bag

Null Subjects in Shipibo switch-reference systems

83

(iii.) *Veis cumprau ina tastga ier. have-2p.pl bought a bag yesterday We will tentatively leave this possibility aside, in the absence of more evidence, and we assume that the person grouping displayed by the other four languages above (and others reported in the literature: Germanic dialects, Paduan and Venetian, cf. references in Deal 2005) deserves an account. 12. See also Fleck 2006 for similar issues in Matses.

References Aissen, Judith 1997 Voice and Person Hierarchy. Talk given at the Hopkins Optimality Theory Workshop. University of Maryland Mayfest, Baltimore. Aissen, Judith 1999 Markedness and Subject Choice in Optimality Theory. Natural Language and Linguistic Theory 17: 673-711. Artstein, Ron 1998 Hierarchies. Manuscript. New Brunswick, Rutgers University. Artstein, Ron 1999 Person, Animacy and Null Subjects. In Tina Cambier-Langeveld, Aniko Liptak, Michael Redford and Erik Jan van der Torre (eds.), Proceedings of Console VII, pp. 1-15. Leiden: SOLE. Black, Andrew 1992 South american verb second phenomena: Evidence from shipibo. Syntax at Santa Cruz, 1, 35-63. Borer, Hagit 1986 Borer, H. 1986. "I-Subjects", Linguistic Inquiry, 17.3. Borer, Hagit 1989 Anaphoric AGR. In Osvaldo Jaeggli and Ken Safir (eds.). The Null Subject Parameter, . Dordrecht: Kluwer. Camacho, José 2008a On Case Concord: the Syntax of Switch-reference Clauses. To appear in Natural Language and Linguistic Theory. Camacho, José 2008b Syntactic Variation: The case of Spanish and Portuguese Subjects. Studies in Hispanic and Lusophone Linguistics, 1.2. Deal, Amy Rose 2005 Pro-drop, topic-drop, and the functional lexicon. A constructional account of null arguments. Honors Thesis. Brandeis University.

84 José Camacho and José Elías-Ulloa Elías-Ulloa, José 2001 Subject Doubling in Capanahua. Manuscript. New Brunswick, Rutgers University. Elías-Ulloa, José 2008 Subject Doubling and the Mixed Null Subject System of Capanahua. In Proceedings of the Conference on Indigenous Languages of Latin America-III (CILLA-III), Universty of Texas at Austin. Fleck, David 2006 Antipasive in Matses. In Studies in Language 30 (3): 541-73. Hornstein, Norbert 2007 Pronouns in a Minimalist Setting. In The Copy Theory of Movement, Norbert Corver and Jairo Nunes (eds.), 351-385. Amsterdam: John Benjamins. Huang, James 1984 On the Distribution and Reference of Empty Pronouns. Linguistic Inquiry 15(4): 531-574. Huang, James 1989 Pro-Drop in Chinese: A Generalized Control Theory. In The Null Subject Parameter, Osvaldo Jaeggli and Ken Safir (eds.), 185-214. Dordrecht: Kluwer. Lebeaux, David 1983 A distributional difference between reciprocals and reflexives. Linguistic Inquiry 14.723-730. Levy, Yonata and Anne Vainika 2000 The Development of a Mixed Null Subject System: A Cross-Linguistic Perspective with Data on the Acquisition of Hebrew. Language Acquisition, 8(4), 363–384. Loriot, James, Ewin Lauriault and Dwight Day 1993 Diccionario Shipibo-Castellano. Lima: Ministerio de Educación-Instituto Lingüístico de Verano. Ordóñez, Francisco and Esthela Treviño 1999 Left Dislocated Subjects and the Pro-Drop Parameter: A Case Study of Spanish. Lingua, 107, 39-68. Poletto, Cecilia 2000 The figher functional field: evidence from Northern Italian dialects. Oxford and New York: Oxford University Press. Prince, Alan and Paul Smolensky 2004 Optimality Theory: Constraint Interaction in Generative Grammar. Wiley-Blackwell. Rizzi, Luigi 1982 Issues in Italian Syntax. Dordrecht: Foris.

Null Subjects in Shipibo switch-reference systems

85

Shlonsky, Ur 2009 Hebrew as a Partial Null-subject Language. Studia Linguistica. 63: 133157. Sigurdsson, Halldór 1993 Icelandic Finite Verb Agreement. Lingua. 89: 247-280. Sprouse, Rex A. and Barbara Vance. 1999. An explanation for the loss of null subjects in certain Romance and Germanic languages. In Michel DeGraff (ed.) Language Creation and Language Change: Creolization, Diachrony, and Development. Cambridge: MIT Press. 257-284. Vainika, Anne and Yonata Levy 1999 Empty Subjects in Finnish and Hebrew. Natural Language and Linguistic Theory 17: 613-671. Valenzuela, Pilar 2003 Transitivity in Shipibo-Konibo Grammar. PhD Dissertation, University of Oregon. Valenzuela, Pilar 2006 Syntactic distributions and co-referentiality in shipibo-konibo. In L'ergativité en Amazonie. Paris, CNRS. http://celia.cnrs.fr/FichExt/ Documents de travail/Ergativite/3iValenzuela.pdf

Cartography of expanded CP in Kuikuro (Southern Carib, Brazil) Bruna Franchetto and Mara Santos

1. Introduction The aim of this article1 is to propose a cartography of the CP (Complementizer System) in the Kuikuro language, spoken at the southern edge of Southern Amazonia. This agglutinative and head final language, endowed with an extremely rich morphology, provides evidence corroborating the internal structure of what has been called, since the seminal work of Rizzi (1997), ‘the left periphery,’ where the syntax of the sentence interfaces with the pragmatics of the utterance. Since this turning point, the former CP, a Phrase, a building block, has become a domain to be explored. After an introduction where we present general information on the Kuikuro people and the basic grammatical features of their language, we offer a careful, albeit summarized, description of the constructions that present an active and explicitly complex left periphery, using interrogative sentences/utterances as our starting point. The description of wh and yes/no questions (section 3) leads us to Focus and Topic constructions (sections 3.3 and 5), and to the almost omnipresent epistemic particles or markers, linked to illocutionary force and the speaker’s ‘point of view’. In section 4 we present data on conditional sentences that suggest the existence of the Fin(itude) Phrase as the lowest projection of the left periphery’s Force-Fin System. Finally in Section 6 we offer a tentative proposal for the internal structure of Kuikuro’s left periphery, an overview of its cartography, trying to ground the existence of projections in CP according to each relevant element of the Force-Fin system.

2. The language and its speakers Since the second half of the 19th century, Kuikuro has been the name found in the ethnographic literature for one of the Carib speaking groups2 leaving in the Upper Xingu river region, on the southern edge of Amazonia (Brazil).

88

Bruna Franchetto and Mara Santos

In fact, the Upper Xingu Carib language, a southern branch of the Carib family, comprises two main variants, one spoken by the Kuikuro people, and the other spoken by the Kalapalo, Nahukwá and Matipú (Meira & Franchetto 2005). Kuikuro is highly agglutinative and head final, a typical member of its family (Derbyshire 1999). Kuikuro is an ergative language, from the point of view of morphosyntactic typology.3 It has a unique set of pronominal prefixes, with nouns, verbs and postpositions. The verbal word has person markers and detransitivizers as prefixes and transitivizers, verbalizers, aspects, moods, plural, collective, tense (future) and negation as suffixes: (1)

ut-uã-te-tagü4 1DTR-cover-VBLZ-CONT ‘I am covering myself.’

(2)

hite heke u-keünti-nhuN-ne-nügü wind ERG 1-cold-VBLZ-TR-PNTC ‘the wind made me feel cold.’

(3)

apa akaN-toho-te-lü-ingo u-heke father seat-INSTNR-VBLZ-PNCT-FUT 1-ERG ‘I will give my father a seat.’

The nominal word also has prefixes (person markers) and suffixes (nominalizers, relational, plural and collective, among others): (4)

ku-gi-tü-gü-ko 12-head-NMLZ-REL-PL ‘Our (all of us) head(s).’

Morphological classes determine the rich allomorphy of most of the inflectional and derivational suffixes (Franchetto 2006; Santos 2002 and 2007): (5)

ahu-nügü ahu-lü ahu-jü close.door-PNCT pound-PNCT fill-PNCT he-toho hihi-tsoho api-goho kill-INSTNR scrape-INSTNR beat-INSTNR

Cartography of expanded CP in Kuikuro

89

Fixed order and strict adjacency characterize the essential relation between a head (V, N, P) and its argument. No other element can come between them. This relation results in a single phonological word, whose main stress is on the last syllable of the argument, the point of its merging with the head. Observe that the main stress of the isolated word is on its penultimate syllable. In the following examples, the point of the argument-head merging is underlined: (6)

u-ingü hutu-ti-lü 1-cloth wet-VBLZ-PNCT ‘My cloth.’

(7)

kangamuke atsaku-lü child run-PNCT ‘The child runs/ran.’

(8)

kangamuke ahetinho-mba-lü itaõ heke child helper-VBLZ-PNCT woman ERG ‘The woman helps/helped the child.’

(9)

kangamuke kanga-gü child fish-REL ‘The child´s fish.’

(10) kangamuke kaenga child near ‘Near the child’ From the examples (6), (7) and (8) above, as well as from the nominalizations in (11) and (12) below, we see that intransitive argument behaves exactly like the Object, both being internal arguments: (11) t-ogopi-si-nhü5 AN-return-PTP-AINR ‘That one who returns/returned.’ (12) t-enge-nhü AN-eat.PTP-AINR ‘That one who is eaten /can be eaten (edible).’

90

Bruna Franchetto and Mara Santos

Moreover, as also shown in (6) and (7), there is no distinction between inergative and unaccusative intransitive verbs (they are all unaccusative). Absolutive is the structural, non-marked and default case, for internal arguments. The external argument is ergatively marked by the postposition heke and can be separated from the OV unit by adverbs and particles; when fully nominal, it has its own autonomy and mobility: (13) a. kangamuke ahetinho-mba-lü leha itaõ heke child helper-VBLZ-PNCT CMPL woman ERG ‘The woman helped the child.’ b. itaõ heke leha kangamuke ahetinho-mba-lü woman ERG CMPL child helper-VBLZ-PNCT ‘The woman helped the child.’ Pronominal external arguments, also ergatively marked, have no such autonomy and they occur always after the OV unit: (14) kangamuke ahetinho-mba-lü leha i-heke child helper-VBLZ-PNCT CMPL 3-ERG ‘she/he helped the child.’ *iheke leha kangamuke ahetinho-mba-lü Another remarkable characteristic is the minimal distinction between nominal and verbal inflections. Compare the morphology of two of the three verbal aspects and of the nominal relational suffixes: (15) a. u-ügü-lü 1-fish.hook-REL ‘My fish hook.’ b. u-te-lü 1-go-PNCT ‘I go/went.’ (lit. my going) c. u-api-lü (hege-i) i-heke 1-beat-PNCT DDIST-COP 3-ERG ‘He beats/beat me.’ (my being beaten by him) (16) a. u-te-ta-gü 1-go-CONT-REL ‘I am/was going.’

Cartography of expanded CP in Kuikuro

91

b. u-kanga-gü 1-fish-REL ‘My fish.’ Summarizing the basic morphosyntactic features in Kuikuro: (i) verbal inflection is minimally distinguishable from the morphology that expresses the nominal genitive relation; (ii) fixed order and a kind of phonological incorporation realize the relation between any lexical head (verbal, nominal, postpositional) and its argument; (iii) the merging between any head and its internal argument is needed for the attribution of the unmarked structural case (absolutive); (iv) the ergative case of the external argument (licensed by transitive v) is lexical and realized by a postposition.

3. Interrogative constructions and interrogative words In this section we present a description of Kuikuro wh and yes/no interrogative clauses. Wh-questions are introduced by pure interrogative markers, situated in CP, to which questioned constituents are moved in syntax. Yes/no interrogatives clauses show a set of particles, also in CP, carrying interrogative propositional content and semantic features of epistemic modality. Interrogative words always occur at the beginning of interrogative sentences; there is one interrogative word for yes/no questions and two for wh-questions. They are frequently associated with epistemic particles or markers (EM); moreover the yes/no interrogative particle itself has an epistemic value that cannot be dissociated from the ‘pure’ interrogative value.

3.1. Wh-questions Wh-questions are introduced by one of two wh-words - tü and uN – and are characterized by the syntactic movement of the questioned constituent triggered by the feature [+int] of the wh-word. 3.1.1. tü wh-questions - tü is the [+int] element for questioned arguments of the type: S (argument of mono-argumental verbs), O (internal argument of transitive verbs), A

92

Bruna Franchetto and Mara Santos

(external argument of transitive verbs), argument of a noun in a genitive construction, and argument of a postposition. tü is frequently followed by an EM compatible with the interrogative illocutionary force. Moreover, different patterns characterize tü wh-questions: nominalizations plus copula are characteristic of S and A interrogative clauses, but not of O and other arguments with interrogative clauses. We describe each of these patterns below. Despite this ergative alignment, a nominative-like alignment characterize interrogative constructions for internal and external arguments. If nominalizations characterize S and A questions, in O questions we find a deergativized and non-nominalized form of the verb: S argument: stative nominalization of the verb + copula A argument: agentive nominalization of the verb (+ copula) O argument: de-ergativized verb + copula Usually the answer to these questions (A in the examples) repeats the same nominalized or de-ergativized patterns, also characteristic of Focus and relative constructions, as we shall demonstrate in Section 5.1. As exemplified in (11) and (12), stative nominalization is characteristic of any internal S/O argument nominalization. Stative nominalization only occurs in S questions. The wh-word tü occurs obligatory at the very beginning of the clause and is often followed by the dubitative EM -ma or, less frequently, by other EM like niküle. The copula -i closes the whole structure: (17) tü-ma t-ügü-nuN-ta-ti-nhü-i WH-DUB AN-sick-VBLZ-CONT-PTP-AINR-COP ‘Who is sick?’ (lit. who is the sick (person)?) A: u-muku-gu t-ügü-nuN-ta-ti-nhü-i 1-son-REL AN-sick-VBLZ-CONT-PTP-AINR-COP ‘My son is sick’ (lit. my son is the sick (person)) Usually, A interrogative clauses, where the questioned argument is the external argument of a transitive verb, have a similar structure to the S interrogative clauses, but with an agentive nominalization rather than a stative one: wh tü agentive nominalization6 of the verb. Copula -i is suffixed to the nominalized transitive verb only when its internal argument is a person marker (compare (18) with (19)). The WH-word tü introduces the clause, often followed by one of the following EM: nile, niküle, the dubitative -ma:

Cartography of expanded CP in Kuikuro

93

(18) tü-ma uãke e-hihi-nhi-i WH-DUB EM.past 2-scrape-AENR-COP ‘Who scraped you back then?’ (lit. who was your scraper)? A. ku-pihi-nhi-ha apa-i 12-scrape-AENR-AF father-COP ‘My father was the one who scraped us.’ (lit. my father was our scraper) (19) tü-ni(le) kanga enge-ni WH-EM fish eat-AENR ‘Who eats/ate (the) fish?’ (lit. who is/was the eater of the fish?) A: uge t-enge-ni, ahütü ekise-hüngü 1D AN-eat-AENR NEG 3DDIST-NEG ‘I am/was the one who eats/ate it, not him.’ (lit.: I am/was its eater, not him) Although less productive, there is an alternative construction for A that parallels the one used to question the argument of a postposition, as we can see in the short dialogue below (X is the speaker, Y his interlocutor): (20) X: tü-heke a-ane-nügü WH-ERG 2-burn-PNCT ‘Who burnt you?’ Y: kangamuke heke child ERG ‘(The/A) child.’ X: tü-heke-ma WH-ERG-DUB ‘Who?’ O internal argument-type questions for transitive verbs present a distinct structure, which Franchetto has called ‘de-ergativized’ (Franchetto 1999, 2002). The loss of the O absolutive argument, moved to CP, triggers a pseudo-detransitivization of the verb: (21) tük-nile kuguagi ngk-enge-tagü-i WH-EM vulture OM-eat-CONT-COP ‘What is the vulture eating?’

94

Bruna Franchetto and Mara Santos

(22) tük-ma e-ngk-iku-tse-gagü-i WH-DUB 2-OM-sap-VBLZ-CONT-COP ‘Whom are you painting?’ Following Marantz (1984), we can say that the prefix ng- (OM), co-indexed to the moved argument, occupies the internal argument slot and receives the Theme Theta role, but is incapable of receiving the structural case (Absolutive), which is then attributed to the ‘promoted’ A. Thus, the verb only superficially appears to be intransitive, lacking the ergatively marked A. In ‘genitive’ wh-questions, the wh-element tü once more introduces the interrogative clause; the whole genitive NP (NP-REL) is attracted to CP and, therefore, associated to EM particles: (23) tü kanga-gü-ma ige-i WH fish-REL-DUB DPROX-COP ‘Whose fish is this?’ A: u-muku-gu kanga-gü ige-i 1-son-REL fish-REL DPROX-COP ‘This is my son’s fish’ (24) tü tolo-gu nile ese-i tahitse-i WH pet-REL EM 3DPROX-COP macaw-COP ‘whose pet is this, the macaw?’ In interrogatives for posposition arguments (such as adverbial adjuncts), the whole PP moves to CP, resulting in a surface structure in which the whelement tü is always followed by suffixes or postpositions and, not always but often, by EM, like the dubitative -ma: (25) tü-ki akaN-toho ha-nügü e-heke WH-INST seat-INSTNR make-PNCT 2-ERG ‘What did you use to make the bench?’

(26) tü-ake e-te-lü tunga-ka WH-COM 2-go-PNCT water-AL ‘With whom did you go to the lagoon?’ (27) tü-inha-ma ege-i Malu heke kamisa humi-lü WH-BEN-EM DDIST-COP Malu ERG shirt send-PNCT ‘To whom has Malu sent the shirt?’

Cartography of expanded CP in Kuikuro

95

(28) tü-tomi e-ini-luN-tagü WH-FIN 2-cry-VBLZ-CONT ‘Why are you crying?’

3.1.2.

uN wh-questions

The wh-element uN also introduces non-nuclear questions, where the questioned constituent is an adverbial adjunct, but exclusively of the ‘how’ type (manner). In what follows, the possible internal structure of the wh uN word is given with some examples: (29) uN-ngu-a-ma pano ita-gü WH-NMLZ-as-DUB cloth width-REL ‘How wide is the cloth?’ (30) uN-ngu-a nile e-ine-po-lü i-heke WH-NMLZ-as EM 2-poison-HIP-PNCT 3-ERG ‘How would she poison you?’ (31) uN-ngu-hungu-ma i-kapohoN-tu WH-NMLZ-like-DUB 3-size-NMLZ ‘How/what is his size?’ (32) uN-te-ma inde imbe t-indi-nhü WH-LOC-DUB D pequi AN-fall-AIRN ‘Where (can I find) fallen pequi here?’ (33) u-haüN, uN-na e-te-tagü 1-cousin WH-AL 2-go-CONT ‘My cousin, where are you going?’ 3.2. Yes/no questions Yes/no questions are characterized by the [+int-EM] particle angí7 alone or very often followed by another particle of epistemic modality: niküle, with its reduced forms nika and ka. As we have already mentioned, yes/no interrogative words can be analyzed as carrying both [+int] and EM features. The particle angí introduces yes/no questions and, as explained by the Kuikuro speakers themselves, expresses

96

Bruna Franchetto and Mara Santos

the speaker’s desire to know what the addressee thinks about the described event. Angí has scope over the whole sentence and is marked by a special intonational contour with pitch and rising intonation on the final syllable. In contrast, the questioned sentence has the same intonational contour as an affirmative utterance (final falling intonation): (34) angí uitu heke e-ihe-tagü INT sleep ERG 2-catch-CONT ‘Are you sleepy?’ (lit.: is sleep catching you?)  Angí is very often followed by specific EM particles: niküle, nika, ka(ha), which we have already seen occurring with the wh-words in wh-questions: (35) angí niküle og-opi-jü-ingo INT EM 2DTR-return-PNCT-FUT ‘Would you really come back?’ (36) angí nika pape humi-pügü u-inha INT EM paper send-PERF 1-BEN ‘Was this the letter sent to me?’

3.3. The interrogative interpretation of the EM particles in Focus constructions Epistemic markers (EM) occur in the majority of Kuikuro utterances and we have seen some of them in interrogative clauses. In wh (tü and uN) questions, the most frequent EM is the ‘dubitative’ suffix -ma, but the particles nile and taka also appear, after the wh-constituent. In yes/no questions, the interrogative word angí has per se EM value and is often followed by the EM particle niküle or by its reduced forms, ka(ha) and nika. These are all second position elements. Moreover, the presence of EM niküle/ka(ha)/nika alone leads to an interrogative interpretation of a Focus construction, which will be the issue of the next section. The simple elimination of the EM particle results in the corresponding declarative Focus sentence: (37) a. kogetsi ka-ha ito ugi-jü-ingo e-heke-ni tomorrow EM-AF fire blow-PNCT-FUT 2-ERG-PL ‘Will you make fire tomorrow?’

Cartography of expanded CP in Kuikuro

97

b. kogetsi-ha ege-i ito ugi-jü-ingo e-heke-ni tomorrow-AF DDIST-COP fire blow-PNCT-FUT 2-ERG-PL ‘Tomorrow you will make fire’ The interrogative utterances above can be interpreted as kinds of rhetorical questions. The meaning of the set of EM present in interrogative clauses is connected to an act of wondering, seeking an answer or confirmation, along a continuum spanning from the speaker’s uncertainty to certainty. At one end, the more frequent dubitative EM -ma gives particular strength to the interrogative speech act conveying the lack of knowledge of an uncertain speaker and, at the same time, the urgency of an answer. At the other end, the particles niküle/nika/ka(ha) convey a feeling of strong probability, the almost certainty of the speaker who seeks a confirmative answer from the interlocutor; the speaker communicates that she/he knows that the partner knows the answer.8

4. EM and conditional constructions In 3.2 we saw the word angí in yes/no questions and determined it to be characterized by the features [+int] and by a value of epistemic modality. The same angí followed by naha, another EM, introduces a conditional sentence. The verbs in both parts of the sentence are in Hypothetical Mode: (38) angí naha telo tapü-gü i-po-lü i-heke if EM other foot-REL take-HYP-PNCT 3-ERG inhalü leha et-uho-po-lü-i NEG CMPL DTR-clear-HYP-PNCT-COP ‘If he takes the foot of another person. It isn’t clear.’ (Context: the speaker explains why it is not acceptable to count beyond the number 20, using the toes on another person’s feet) Using naha the speaker communicates the wish that the hearer will accept and understand the explanation and that she/he is confident that the hearer will agree with the statement. The EM naha is also present as enaha in the same conditional constructions: (39) enaha EM(if)

t-iküte-la e-i-ho-lü AN-NEG 2-be-HYP-PNCT

98

Bruna Franchetto and Mara Santos

kuge kotuN-ho-lü person angry-VBLZ-HYP-PNCT ‘If you don’t divide and share (food), people get angry.’ It is important to observe that EM naha (after angí or as enaha) determines the hypothetical mood in both components of the conditional construction. It introduces a peripheral adverbial clause, as Haegeman says, which “express a premise which is entertained by the speaker and which serves as privileged context for the processing of the main clause ... the relevant adverbial clause provides discourse background that serves as restrictors for processing, against such backgrounds the matrix clause yields particular contextual implicatures” (Haegeman 2004:161).

5. Focus and Topic 5.1. Focus As we have already mentioned in 3.1, nominalizations and the de-ergative form of the verb characterize interrogatives. They also appear in relative clauses9 as well as Focus constructions for S, A and O arguments: Focused S: (40) u-ingaNtsu-ha ekise-i t-iniluN-ta-ti-nhü-i 1-sister-AF 3DDIST-COP(FM) AN-cry-CONT-PTP-AINR-COP ‘It was my sister who was crying.’ Focused A: (41) u-ingaNtsu-ha ekise-i hikutaha enge-ni-mbüngü 1-sister-AF 3DIST-COP(FM) turtle eat-AENR-SUBS ‘It was my sister who ate the turtle.’ Focused O: (42) hikutahai-ha ege-i u-ingaNtsu ngi-enge-tagü turtle-AF DDIST-COP(FM) 1-sister OM-eat-CONT ‘It was a turtle that my sister was eating.’ The focused constituent in initial position is followed by a copular expression formed by a deictic plus copula, which we consider as the Focus marker (FM). Here it is important to provide a brief description of Kuikuro

Cartography of expanded CP in Kuikuro

99

deictics. These are specified by the semantic features of animacity, visibility, distance/proximity (in space and time, to the speaker), person and number in the case of personal deictics.10 In Focus constructions, the selection of one or other deictic is motivated by the interaction between its semantic features, the semantic features of the verb and the context. Below we provide examples of sentences containing the FM and nominalization of the verb of a Small Clause; the morpheme -mbüngü, suffixed to the nominalization, is also an essential element in the Focus construction since it functions as an operator determining one among the various potential members of a set: (43) a. u-ingaNtsu-ha eekise-i hikutaha enge-ni-mbüngü 1-sister-AF 3DDIST-COP(FM) turtle eat-AENR-SUBST ‘That one is my sister, the one who ate the turtle’ (context: the speaker indicates the agent [+animate, +visible, +distant] of ‘eat’) b. u-ingaNtsu-ha ese-i hikutaha enge-ni-mbüngü 1-sister-AF 3DPROX-COP(FM) turtle eat-AENR-SUBST ‘This one is my sister, the one who ate the turtle’ (context: the speaker indicates the agent [+animate, +visible, -distant] of ‘eat’) c. *u-ingaNtsu-ha ege-i hikutaha enge-ni-mbüngü 1-sister-AF DDIST-COP(FM) turtle eat-AENR-SUBST ‘That one is my sister, the one who ate the turtle.’ (context: the speaker indicates the agent [-animate, +distant] of ‘eat’) The ungrammaticality of (43c) arises from the conflict between the feature [animated] of the deictic ege and the selection feature [agent +animated] of the verb enge, ‘to eat.’ On the other hand, in Focus constructions where the verb is not nominalized and where the movement of the focused constituent to the left periphery is evident, only ege/ige deictics [-animated, +distant/-distant] such as the FM can occur. In this case, the interpretation depends on a single context: the account of an event, near or distant in time in relation to the speaker:

(44) a. [u-ingaNtsu heke-ha ege-i [hikutaha enge-lü____] 1-sister ERG-AF DDIST-COP(FM) turtle eat-PNCT ‘It was my sister who ate the turtle’ or ‘my sister ate the turtle.’ (context: the speaker is telling a story) b. *[u-ingaNtsu heke-ha ekisei-i [hikutaha enge-lü__] 1-sister ERG-AF 3DDIST-COP(FM) turtle eat-PNCT

100

Bruna Franchetto and Mara Santos

‘It was the my sister who ate the turtle’ or ‘my sister ate the turtle.’ (context: the speaker indicates the agent [+animate, +visible, +distant] of ‘eat’) c. *u-ingaNtsu-ha ekise-i hikutaha enge-ni-mbüngü 1-sister-AF 3DDIST-COP(FM) turtle eat-AENR-SUBS ‘My sister was that one who ate the turtle.’ (context: the speaker is telling a story) The same interplay between semantic features of the FM and the verb can be observed in the interrogative sentences in which the interrogated element is focused on and which for this reason requires a response that identifies it precisely: (45) tü-ma kanga enge-ni WH-DUB fish eat-AENR ‘Who ate the fish?’ A: kuge / ngene people / animal

(generic answer)

(46) a. tü-ma ese-i kanga enge-ni WH-DUB 3DPROX-COPfish eat-AENR ‘who was this one who ate the fish?’ A: uingaNtzu / tago (specific answer) ‘My sister’ / ‘the otter’ b. *tü-ma egei kanga engeni WH-DUB DDIST-COP fish eat-AENR ‘Who was that one who ate the fish?’ c. tüma egei kanga engeni ‘what/who was it who startled the fish?’ A: ehu itsu (specific answer) canoe noise ‘The noise of the boat’ The reasons for the ungrammaticality of (46b) are identical to those we identified for (43c), namely the conflict between features of the deictic and the verb; in (46c), on the other hand, the grammaticality of what appears to be the same sentence is due to the features of the verb enge when interpreted as ‘startle’ (emotional consumption).

Cartography of expanded CP in Kuikuro

101

Focus may also occur in YES/NO questions and in wh-questions that do not contain nominalization: (47) ungua nile ege-i a-a-tühügü agugu-i WH EM DDIST-COP(FM) 2-be-PERF thin-COP ‘Why exactly have you become so thin?’ (48) tü-inha-ma ege-i Malu heke kamisa humi-lü WH-BEN-EM DDIST-COP(FM) Malu ERG shirt send-PNCT ‘To whom exactly has Malu sent the shirt?’

5.2. Topic The canonical position of wh-words, as seen up to now, is at the very beginning of the interrogative clause. Nevertheless, they may also occur in medial and final positions: (49) kuge hu-toho-ma uN-te person represent-INSTNR-DUB WH-LOC ‘Where are the ‘representations of persons’?’ (50) uitu heke e-ihe-tagü angí sleep ERG 2-catch-CONT INT ‘You are sleepy, aren’t you?’ It is clear from the semantic and pragmatic interpretation of the above utterances that the apparent non-canonical position of the wh-word is the result of the topicalization of the NP [kanga ingugu], through movement, to the left periphery of the clause in (49), and of the whole VP or vP in (50). It is also clear that the landing site of the topicalized constituents is above the interrogative marker. Evidence for topicalization deriving from movement to the CP is seen in (51), a standard topic construction: (51) ekise=tsü naha kangak, u-ingaNtsu Nk-enge-pügü 3DDIST+EV EM fish 1-sister OM-eat-PERF ‘That fish, my sister ate.’

102

Bruna Franchetto and Mara Santos

Recall that the nasal prefix in N-enge-pügü, glossed as O(bject)M(arker), can be interpreted as a kind of resumptive pronoun left behind by the internal argument of ‘eat’ moved to the Topic position. Also observe the clitic tsü, a reportative evidential, followed by the EM naha, whose meaning, as we saw earlier, is something like ‘the speaker is seeking confirmation from her/his interlocutor’. In Kuikuro, differences in the order of constituents, in declarative utterances, can be taken to be outputs of processes of topicalization. (52) below is an example of a very common repetition of the same utterance, with a slight difference in terms of word order. The two utterances therefore stand in a parallelistic relation, a means of providing complementary ‘point of views’ on the same event, a fact characteristic not only of formal discourse and poetry, but also of ordinary speech (Franchetto 2003): (52) a. u-etiko-gu-pe leha tü-kahisi-nhü-pe t-atanhe-ti 1-belt-REL-ex CMPL AN-red-AINR-ex AN-lose-PTP ‘My former belt, that which was red, is lost.’ b. tatanheti leha uetikogupe tükahisinhü ‘It is already lost, my former red belt.’ In the above example, the particle leha occurs at the border between the topicalized constituent and the rest of the clause. In general, leha simultaneously possesses an aspectual value of accomplishment and an organizing function at the level of the overall discourse, sequencing the described events in a temporal/logical chain.

6. A cartography of the expanded CP in Kuikuro The Kuikuro data described up to now show a complex left periphery of the clause, the side of the utterance linking what is said to the speaker’s communicative intentions and feelings. The rich morphology of this highly agglutinative language offers a remarkable diversity of means for encoding different projections inside the left periphery, in other words, inside CP. In order to explore the Kuikuro left periphery, we begin by considering in succession the main proposals regarding the expansion of CP. The starting point is, inevitably, Rizzi (1997) with his analysis of CP as an internally complex system, with a maximal number of functional projections, whose heads are filled by phonologically realized or null wh-elements or phrases, illocutionary force particles, focus and topic markers and other functionally

Cartography of expanded CP in Kuikuro

103

equivalent markers. The specifiers of these projections contain the constituents (questioned, focalized, topicalized, etc.) attracted by the activated projections. Force phrase (ForceP) and Finitude Phrase (FinP) are the highest and the lowest borders of the CP system, called by many authors the Force-Fin system. ForceP is the locus of the speaker’s communicative intentions and, therefore, of the distinction between the different types of utterances according to distinct speech acts (questions, commands, constative, etc.). FinP, the lowest projection, is the interface between CP and IP, the two layers of the sentence/utterance structure, determining or interacting with “the finite or non finite nature of the clause, mood distinctions, overt tense distinctions, subject agreement licensing nominative case” (Rizzi 1997: 283, 284). Between ForceP and FinP, other projections include Focus Phrase (FocP) and Topic Phrase (TopP). Rizzi 2001 provides an even more detailed description of the structure of expanded CP by postulating at least two non-contiguous projections for Topic. The proposed structure of CP would therefore be the following: [ForceP [AboutTopP [ IntP [ FocP [ FamTopP* [FinP [IP Particularly interesting in terms of an understanding of the Kuikuro data is the article by Puglielli & Frascarello (n.p.) on the Force-Fin system in Somali and Avar, two Cushitic languages. The authors show that the activation of [+foc] plays a crucial role in the interpretation of other discourse categories and emphasize that an interaction must be posited between functional heads in the C-domain. The evidence provides by the Kuikuro data leads us to postulate the following structure of CP: [ForceP [TopP [ IntP [ FocP [FinP

6.1. Force and Finitude Force-Fin is an active system where interrogative operators and the Focus Marker are lodged in functional projections between Force Phrase and Finitude Phrase. Regarding epistemic particles we have already seen their relevance in Kuikuro,11 where EM markers are present in most of the sentences that become utterances with communicative efficacy. Kuikuro seems to make

104

Bruna Franchetto and Mara Santos

explicit, through a rich repertoire of forms, the nature of Force, convincingly synthesized by Haegeman (2004:164): “…(Force) guarantees anchoring to the speaker and is implicated in the licensing of, among other things, illocutionary force and epistemic modality...(Epistemic modality) expresses the speaker’s stance concerning the likelihood of the state of affairs/event, which is anchored to speech time.” The interaction between Force and the other components of the CP system (Topic, Focus and WH) is one of the main points of Puglielli & Frascarelli and is underlined by Haegeman (epistemic modality depends on Force, topicalization depends on Force). Despite the recognition of the obvious relation between Force and EM, we have not yet a satisfying formal analysis for EM. Nevertheless we don’t think that they are head of some specific Force or EM Phrase in CP, but that they behave more like operators parasitic to any functional projection, following the suggestions of Blain & Déchaine (2007) and Matthewson (to appear). Looking now to the lowest projection of the expanded CP, evidence for the existence of FinP comes from the conditional constructions briefly described in section 4. In Kuikuro, FinP determines or selects moods, realized by suffixes bound to the verbal stem. The unmarked default Mood is Declarative; Imperative, Hortative, Intentional (immediate future) and Conditional moods are all realized by explicit suffixal forms. ForceP interacts actively with FinP. Consequently, we postulate that the Kuikuro iflike forms (angí naha, enaha), as particles with epistemic value, lodged in ForceP or having it in their scope, are crucial for the activation of FinP and the selection of the Hypothetical Mood in both parts of the conditional complex sentence. We shall now examine the data given so far on Kuikuro step-by-step in order to ground the existence of the other CP components.

6.2. Questions First of all, we need to explain the interrogative structures for the arguments S and A containing nominalizations. Here we adopt Puglielli & Frascarelli’s analysis of Somali and Avar Focus constructions, not only for Kuikuro S and A questions, but also, as we shall show in a moment, for the structurally homologous S and A Focus utterances. The examples (17) and (18) are

Cartography of expanded CP in Kuikuro

105

reproduced below as (53) and (54) and are followed by the proposed structure of their left periphery: S argument question: (53) a. tü-ma [ ____ t-ügü-nuN-ta-ti-nhü]-i WH-DUB AN-sick-VBLZ-CONT-PTP-AINR-COP ‘Who is sick?’ (lit. who is the one who is ill?) b. [ForceP [IntP tük ma [FinP [SC tk tügünuNtatinhü-COP]]]] A argument question: (54) a. tü-ma [ _____ a-ane-tinhi]-i WH-DUB 2-burn-AENR-COP ‘Who burns/burnt you (who is your burner)?’ b. [ForceP [IntP tük -ma [FinP [SC tk aanetinhi-COP]]]] The scope of the WH-EM sub-complex is the questioned argument; in other words, the agentive or stative nominalizations of the verb, a Small Clause (SC). Both the S and the A questions result from the movement of the questioned constituent from the subject position of the small clause to the specifier of IntP. The movement is motivated by the active feature [+int] of the wh-element, head of IntP. In A questions alternative to nominalized ones, it is clear that the questioned argument, the whole ergatively marked NP, moves from its position as specifier of v, to the specifier of IntP. We reproduce (20) as (55):

(55)

a. X: tü heke [a-ane-nügü ___ ] WH ERG 2-burn-PNCT ‘who burns/burnt you?’ b. [ForceP [IntP tü-hekek -ma [FinP[IP aanenügü tk]]]]

The same structure can be posited for wh-questions of adjuncts, where the whole Postpositional phrase is moved ((56) reproduces (26)):

(56) a. tü-ake [e-te-lü tunga-ka __] WH-COM 2-go-PNCT water-AL ‘With whom did you go to the lagoon?’

106

Bruna Franchetto and Mara Santos

b. [ForceP [IntP tü-akek [FinP [IP etelü tungaka tk]]]] Internal argument questions for transitive verbs do not contain a nominalization but a de-ergativized clause, where, as we have already seen, the underspecified nasal prefix of the verb, a general Object Marker, can be analysed as a kind of a resumptive pronoun co-indexed with the moved object. Hence, the example (21), reproduced here as (57), can be analyzed in the structure below: (57) a. tük-nile kuguagi Nk-enge-tagü-i WH-EM vulture OM-eat-CONT-COP ‘What is the vulture eating?’ b. [ForceP [IntP tük -nile [FinP[IP kuguagi ngkengetagü-COP]]]] The wh-word uN and the interrogative particle angí of yes/no questions are lodged, alternatively, in the same functional projection, IntP with its [+int] feature. These too are very often associated with epistemic particles. The examples (36) and (44) are reproduced in (74) and (75), respectively and are followed by their postulated structure: (58) [unguak-ma [pano ita-gü tk]] WH-DUB cloth width-REL ‘How wide is the cloth?’ (59) a. angí uitu heke e-ihe-tagü INT sleep ERG 2-catch-CONT ‘Are you sleepy?’ (lit.: is sleep catching you?) b. [ForceP [IntP angí/uN [FinP [IP/NP]]]]

6.3. Focus The Kuikuro FM is a copular form, as in Somali and Avar, the languages analyzed by Puglielli & Frascarelli. In fact, Puglielli & Frascarelli argue that Somali and Avar Focus Markers (FM) were originally copular forms and that they imply the presence of a Small Clause in the Focus constructions. The structure of (43a) is similar to the one proposed for the corresponding Somali sentence, with a functional relation between an operator and a variable in a relative nominalized clause (enge-ni-mbüngü):

Cartography of expanded CP in Kuikuro

107

(60) [ForceP [FocP[u-ingaNtsu k [Foc’ ekisei [FinP[REL hikutaha engenimbüngü tk]]]]]] With FM ege, the focused A argument, or the whole predicate (S/O V), or an adjunct, moves to the specifier of FocP, attracted by its active [+focus] feature. We should recall that the movement always involves the whole constituent. The structure of (44a) can be represented as: (61) [ForceP [FocP[hüati hekehak [Foc’ ege-i[FinP[IP ekise hekutelü tk]]]]]] We saw that in Kuikuro wh-questions can be constructed with a Focus structure, as in (48), reproduced as (62): (62) tü-inha-ma ege-i Malu heke kamisa humi-lü WH-BEN-EM DDIST-COP(FM) Malu ERG shirt send-PNCT ‘To whom has Malu sent the shirt?’ Looking to the internal structure of the Force/Fin system, we observe that the interrogative phrase is higher than FocP. We can therefore represent the above interrogative sentence, parallel to a Focus sentence, by means of the structure below, where both IntP and FocP are activated and linked: (63) [ForceP [IntP tüinhak -ma [FocP[ Foc’ egei [FinP [IP Malu heke kamisa humilü tk]]]]]] Moreover, we must postulate an interaction between Focus and Fin, recalling that the default [-human] deictics ige and ege also carry a temporal value anchored to the speech time and, therefore, to the speaker: proximity or distance in space and time, respectively.

6.4. Topic Given the present state of our research, only one Topic position can be posited in our cartography of the Kuikuro expanded CP: a TopP higher than IntP (Rizzi 2001). We saw in section 5.2 that in interrogative sentences with a topicalized constituent, this topicalization occurs to the left of the interrogative word, which stays above the Focus Marker. Hence, the sentence below (reproducing (49)) could be represented by the structure (64b):

108

Bruna Franchetto and Mara Santos

(64) a. kuge hu-toho-ma uN-te person represent-INSTNR-DUB WH-LOC ‘The ‘representations of persons,’ where (are they)?’ b. [ForceP [TopP kuge hutoho-ma [Top’[IntP uNte]]]]] In addition it seems to us that the topicalized element moves from its base position to TopP, as it was seen in (51) and can be seen in the following examples: (65)

ekise=tsü naha kangamukek isk-ügünuN-tagü tsuei 3DDIST=EV EM child 3-sick-VBLZ-CONT much ‘That child, he is very sick.’

(66)

ekise=tsü naha kangamukek tsue sahuNtu 3DDIST=EV EM child much tucunaré ik-heke 3-ERG ‘That child, he killed a lot of tucunaré fish.’

ik-heke kill-PNCT

Phenomena related to Topic are still under scrutiny and we need a more careful investigation of their occurrence in different genres of discourse. 7. Conclusions Kuikuro speakers exhibit a sophisticated meta-linguistic thought and discourse, with many explicit meta-linguistic categories, often rendered as metaphors. Interrogative words, epistemic and focus markers, aspectual and discourse floating particles – in other words, all those elements we have examined in this article – are called tisakisü enkgu-toho (arrive.to.portINSTNR), translatable as ‘to make our words arrive at the port (at the place of rest, where a trip comes to the end; to beach safely)’. The Kuikuro also say that a sentence without these elements is like a dead body, a skeleton without flesh and blood. Franchetto described tisakisü enkgutoho in her thesis (Franchetto 1986:346) as tiny words that weave the discourse’s texture and outline the constituents at their borders, connecting them, transforming propositions/sentences into utterances, thereby bridging words and the world, frequently grouped together in compositions of deixis, aspectuality and focalization. Tisakisü enkgutoho are thus members of a natural class whose linguistic and cognitive identity is located within the domain of the speaker’s

Cartography of expanded CP in Kuikuro

109

awareness. This article is a first, and still incomplete, approach to understanding tisakisü enkgutoho beyond the descriptive level.. Starting out from different and alternative constructions for wh and yes/no questions, we have tracked the emergence of Focus, Topic, the illocutionary force of the epistemic particles and the active presence of the Finitude phrase, leading us to postulate an internal structure of the CP layer and to look inside the socalled Force-Fin system. Kuikuro is an agglutinative, head final language, and, we would add, still ‘unpolluted’ by a long history of alphabetical writing and writing habitus. As such, it is a promising laboratory for testing and parametrizing universal hypotheses about the packaging of the information in utterances and the complexity of the left periphery.

Notes 1. We are grateful to all those that contributed to this paper: the editors and the reviewers, our colleagues Márcia Dámaso Vieira, Marcus Maia and Rosana Costa, and especially, among our Kuikuro consultants, Mutua Mehinaku. 2. According to the last census, the Kuikuro population numbers about 600 people, living in four villages; the Kalapalo, Nahukwá and Matipú number some 750 people, living in seven villages. 3. The Upper Xingu Carib language represents the outcome of a history of reanalyzing ergatively marked nominalized subordinate clauses in the Carib languages classified by Gildea as System I, as root clauses in the few ergative languages of System II (Gildea 1998). 4. The Kuikuro data are transcribed using the current phonemic orthography established by the Kuikuro teachers and by ourselves, the linguists. The correspondences between written and IPA symbols (where these differ) are as follows: ü (̣ , j (j), g (uvular flap), ng (І), nh (Ј), nkg (Іg); N represents a subspecified nasal feature. The abbreviations for glosses in the interlinear morpheme-to-morpheme translations are: 1 first person, 12 first inclusive dual, 13 first exclusive plural, 2 second person, 3 third person, 1D first person deictic, 2D second person deictic, 3DPROX third person deictic proximate to the speaker, 3DDIST third person deictic distant from the speaker AENR nominalizer for external argument, AF constative, AINR nominalizer for internal argument, AN anaphoric, CMPL completive, COM comitative, CONT continuative aspect, COP copula, DDIST deíctic.distance, DPROX deíctic.proximity, DES desiderative, DTR detransitivizer, EM epistemic, ERG ergative, EV evidential, ex nominal past, FUT future, GNR generic nominalizer, HYP hypothetical. INT interrogative particle, LOC locative, NEG negation, NMLZ nominalizer, OM object marker, PERF perfect/perfective, PL plural, PNCT punctual aspect, PRSP perspective, PTP participial, PURP purpose, REL

110

Bruna Franchetto and Mara Santos

relational, Temp temporal subordination, TR transitivizer, VBLZ verbalizer, WH wh interrogative particle. 5. Stative or non-agentive nominalization is formed by adding the non-agentive nominalizer (AINR) -nhü to the participial form of the verb, whose structure is tV-ti (t- being the generic anaphoric filler of the argument position, and -ti being one of the participial phonological exponents, conditioned by the morphological inflectional classes: see Santos 2007). t-V-ti-nhü is the Kuikuro reflex of the proto-Carib nominalized participle *t-V-se-mï reconstructed by Gildea (1998, Ch. 8). 6. The external argument or agentive nominalization is formed by adding the suffixes -tinhi, -ni or -nhi (glossed as AENR) to transitive and transitivized verbal stems. 7. These interrogative particles may be related to angi, an indefinite deictic, but they are distinguished by the pitch accent: on the last syllable in the former, on the penultimate syllable (general and predictable pattern) in the latter. Angi is used in answer to questions of the kind ‘where is X?’ and is interpreted as indicating X in the space near and visible to both speaker and listener: (i) pape ekuatsangeangikapehe book EM D high ‘The book is high over there.’ Angi with the nominalizer -ni means ‘someone/some’: (ii) angi-ni-ko kangagamuke-ko tsügütse D-NMLZ-PL child-PL only t-ügü-nuN-ta-ti-nhü-i AN-sick-VBLZ-CONT-PTP-AINR-COP ‘Only some children are sick.’ 8. We acknowledge the inspiring work of anthropologist Ellen Basso, author of many profound observations on the meaning of epistemic markers in Kalapalo, the other ‘sister’ variant of the Upper Xingu Carib Language (Basso 1987, 1995). See also Franchetto (2007) for a preliminary analysis of epistemic modality and evidentiality in Kuikuro. 9. Examples of relative clauses: (i) itão ingi-lü u-heke [t-atsaku-nhü] woman see-PNCT 1-ERG AN-run-AINR ‘I saw the woman who ran (is a runner).’ (ii) Mutua heke itaoN ingi-lü kajü ilaN-te-ni-mbüngü M. ERG woman see-PNCT monkey food-VBLZ-AENR-SUBS ‘Mutuá saw the woman who cooked the monkey’

Cartography of expanded CP in Kuikuro (iii)

111

pape ingilü u-heke [Ahukaka N-humi-pügü] paper see 1-ERG A. OM-send-PERF ‘I saw the letter that Ahukaka sent.’

10. The Kuikuro third person deictics that occur in FM are as follows: ige [-animate, +proximate (to the speaker)]; ege [-animate, -proximate (to the speaker)]; ese [+animate, +proximate]; ekise [+animate, -proximate]; ago [+animate, +proximate, +plural]; akago [+animate, -proximate, +plural]. First and second person deictics can also occur in FM: uge [1]; tisuge [13]; kukuge [12]; ege [2]; amago [2, +plural]. 11. The reader is almost certainly wondering about the meaning and the nature of an almost omnipresent particle, the clitic ha, glossed as AF(firmative), a provisory and less than coherent label. Derbyshire (1985) calls the same particle in Hixkaryana, a northern Carib language, ‘intensifier’. Ha, compared by the young literate Kuikuro to ‘a comma’, will receive special attention in future work; it seems to be a purely discursive phenomenon, but it is undoubtedly implied in some processes of information packaging that, at present, we are unable to analyze satisfactorily. However, we should like to offer at least an example of its occurrence with thisexcerpt from a personal narrative (Fem_eginhoto, 240-42): (i) lepene lepene-ha gehale-ha u-i-nhügü Patsi tolo itajope-gü-i then then-AF also-AF 1-be-PNCT Patsi tolo asker-REL-COP gehale also ‘Then I also became the ‘asker’ for Patsi’s tolo ritual.’ (ii) Patsi kagahuku-gu tüi-lü-ha ti-heke gehale Patsi fence-REL make-PNCT-AF 13-ERG also ‘We also made Patsi’s fence.’ (iii) uge-ha i-hotu-gü-i t-a-ti-nhü-i 1D-AF 3-beak-REL-COP AN-stay-PTP-AINR-COP ‘I was the first one to stay as ‘asker’.’

References Basso, Ellen 1987 In Favour of Deceit. A Study of Tricksters in an Amazonian Society, Tucson: The University of Arizona Press. Basso, Ellen 1995 The Last Cannibal. Austin: The University of Texas Press. Blain, Eleanor M. and Déchaine, Rose-Marie 2007 Evidential types: evidence from Cree dialects. International Journal of American Linguistics 73, no 3: 257-91.

112

Bruna Franchetto and Mara Santos

Derbyshire, Desmond 1985 Hixkaryana and Linguistic Typology. The Summer Institute of Linguistics and The University of Texas at Arlington. Derbyshire, Desmond 1999 Carib. The Amazonian Languages, R.M.W. Dixon and A. Y. Aikhenvald (eds.), 23-64. Cambridge: Cambridge University Press. Franchetto, Bruna 1986 Falar Kuikuro. Estudo etnolingüístico de um grupo caribe do Alto Xingu [Speaking Kuikuro. An ethnolinguistic study of an Upper Xingu Carib speaking group]. PhD Thesis, PPGAS/MN, Universidade Federal do Rio de Janeiro. Franchetto, Bruna 1990 Ergativity and Nominativity in Kuikúro and Other Carib Languages. In Amazonian Linguistics. Studies in Lowland South American Languages, Doris L. Payne (ed.), 407-428. Austin: University of Texas Press. Franchetto, Bruna 2002 Kuikuro. Uma língua ergativa no ramo meridional da família Karib (Alto Xingu) [Kuikuro. An ergative language of the Carib family Southern branch (Upper Xingu)]. Ergatividade na Amazônia I, 15-44. Paris, Brasília: Centre d'études des langues indigènes d'Amérique (CNRS, IRD); Laboratório de Línguas Indígenas (UnB). Franchetto, Bruna 2003 L’autre du même: parallélisme et grammaire dans l’art verbal des récits Kuikuro (caribe du Haut Xingu, Brésil) [The other same: parallelisms and grammarin the verbal art of Kuikuro narratives (Karib, Upper Xingu)]. Amerindia 28, numéro Langues caribes, 213-248. Paris: AEA. Franchetto, Bruna 2006 Are Kuikuro Roots Lexical Categories? Lexical Categories and Root Classes in Amerindian Languages, Ximena Lois and Valentina Vapnarski (eds), 33-68. Bern: Peter Lang. Franchetto, Bruna 2007 Les marques de la parole vraie en Kuikuro, langue caribe du HautXingu (Brésil)” )[Markers of the true speech in Kuikuro, a Carib language of the Upper Xingu (Brazil)]. L’Énonciation médiatisée II. Le traitement éistémologique de l’information : illustrations amerindiennes et caucasiennes, Z. Guentchéva & I. Landaburu (eds), 173-204. Paris: Éditions Peeters. Gildea, Spike O. 1998 On Reconstructing Grammar: Comparative Cariban Morphosyntax. Oxford: Oxford University Press. Haegeman, Liliane 2004 Topicalization, CLLD and the Left Periphery. Proceedings of the Dislocated Elements Workshop, ZAS Berlin, November 2003 (2

Cartography of expanded CP in Kuikuro

113

Volumes). ZAS Papers in Linguistics 35. B. Shaer , F. Werner and C. Maienborn (eds.), 157-192. Marantz, Alec 1984 On the Nature of Grammatical Relations. Cambridge MA: MIT Press. Matthewson, Lisa; Davis, Henry and Rullman, Hotze 2007 Evidentials as epistemic modals: evidence from St’át’imcets. The Linguistic Variation Yearbook. Jeroen Van Craenenbroeck & Johan Rooryck (eds.), 201-254 John Benjamin Publishing Co. Meira, Sérgio and Franchetto, Bruna 2005 The Southern Cariban Languages and the Cariban Family. International Journal of American Linguistics, vol 71, n. 2: 127-190. Puglielli, Anna Rita and Frascarelli Mara 2007 Focus in the Force-Fin System: Information Structure in Cushitic Languages. in A. Enoch, K. Hartmann and M. Zimmermann (eds.), 161–184. Focus Strategies: Evidence from African languages, Berlin, Mouton de Gruyter. Rizzi, Luigi 1997 The Fine Structure of the Left Periphery. In Elements of Grammar. Handbook of Generative Grammar. Liliane Haegeman (ed.), 281-337, Dordrech: Kluwer Academic Publishers. Rizzi, Luigi 2001 On the Position Int(errogative) in the Left Periphery of the Clause. In Current Studies in Italian Syntax. Essays offered to Lorenzo Renzi, Guglielmo Cinque and Giampaolo Salvi (eds.), 287-296, Amsterdam, New York: Elsevier. Santos, Gélsama M. Ferreira. 2002 Morfologia Kuikuro: as categorias ‘nome’ e ‘verbo’ e os processos de transitivização e intransitivização [Kuikuro Morphology: ‘noun’ and ‘verb’ categories and the processes of transitivization and intransitivization]. Universidade Federal do Rio de Janeiro, M. A. Dissertation. Santos, Gélsama M. Ferreira. 2007 Morfologia Kuikuro: gerando nomes e verbos [Kuikuro Morphology: Generating Nouns and Verbs]. Universidade Federal do Rio de Janeiro, Ph.D. thesis.

On word order and information structure in Yaqui Lilián Guerrero and Valeria A. Belloro

1. Introduction This paper explores word order patterns in Yaqui (Uto-Aztecan) as they manifest in a corpus of oral narratives, and advances a preliminary account of the relation of word order and information structure in the language. Word order patterns can be conceived of as the arena where syntactic and pragmatic forces meet. The way in which languages resolve the potential tensions arising from this encounter serves as a typological criterion under which different language-types can be distinguished. According to this criterion, languages can be characterized in terms of whether the order of constituents in main clauses is primarily dependent on syntactic principles (e.g. grammatical relations) or on pragmatic ones (e.g. the (assumed) cognitive status of the referents involved) (Thompson 1978; Payne 1990, 1992). In those languages where word order is constrained by syntactic principles, the encoding of information structure is frequently carried out exclusively by prosodic means, leaving word order intact. These languages can be said to be syntactically rigid (Van Valin 1999). English is one example of this type. On the other hand, word order may be constrained by pragmatic principles. For instance, a language may ban the assignment of focus to preverbal subjects, as Italian (Bentley 2008), or reserve a specific syntactic position for particularly newsworthy information, as Cayuga (Mithun 1992). In Van Valin’s typology, languages in which the potential domain for focus assignment excludes certain syntactic positions in the clause are considered pragmatically rigid. Rigidity and flexibility are conceived as tendencies along a continuum, and should not be understood as absolute opposites. Syntactically rigid languages may be either flexible or rigid pragmatically. English and French represent examples of each type, respectively. In both languages word order is mainly determined by grammatical relations. However, in French, the syntactic rigidity of the SVO order meets a further pragmatic constrain which bans preverbal focus (with some exceptions that do not concern us here). Hence, to the question ‘What happened?’ (which triggers a response where the full sentence is in focus) an English speaker may answer with a canonical SVO structure; for instance ‘My car broke down’. A speaker of French, on the contrary, will have to produce an alternative in

116

L. Guerrero and V. Belloro

which both the syntactic constraint (SVO order) and the pragmatic constraint (no preverbal focus) are satisfied, as in ‘J’ai ma voiture qui est en panne’ (lit. ‘I have my car that is in breakdown’, Lambrecht 1994:15). Pragmatically flexible languages may be syntactically rigid (like English), or syntactically flexible (like Russian). Russian is like English in that there is no constraint in the clause with respect to the domain for focus assignment. It is different from English, however, in that its word order is not determined by grammatical roles, so that it can be altered depending on the particular pragmatic status of the participants. Finally, pragmatically rigid languages may be syntactically rigid as well (as French) or syntactically flexible (like Italian). In Italian, unlike in French, the order of grammatical roles can be altered to accommodate the existing pragmatic constraints. Consequently, the Italian equivalent to the sentences above would be ‘Mi si é rotta la macchina’, a structure which avoids a preverbal focal subject with less dramatic effects on the syntax than the French alternative (Lambrecht 1994; Bentley 2008; Van Valin 1999, 2005). English, French, Italian and Russian are examples of the four crosslinguistically attested results of the struggle between syntactic and pragmatic constraints over word order. Roughly, in the English type syntax overcomes pragmatics, and in the Italian type it is pragmatics which overcomes syntax. In languages like Russian there are no strong constraints of either type, and order depends on information structure (among other factors). Finally, in the French type both the syntax and pragmatics must be satisfied, potentially resulting in a complex structure. The analysis of the principles governing word order in different languages, thus, offers a privileged view of the interplay between syntax and pragmatics and, more generally, opens a venue for studying motivation in grammar. With this as general background, the aim of this paper is to examine the interplay between word order and information structure in Yaqui.1 So far, most of the studies on Yaqui have focused on the syntactic and/or semantic aspects of its grammar, and little has been said about word order and its possible correlation with the expression of information structure. In this regard, the goal of this paper is twofold: (i) to determine whether the basic word order assumed for the language is corroborated by the analysis of spontaneous corpora or whether it should be revised, and (ii) to examine to what extent word order variations are determined by discourse-pragmatic constraints. Establishing if a language is syntactically flexible or not is, in our opinion, best determined by the analysis of texts. The more rigid the syntax of the language, the easier it should be to determine a “basic” word order based

On word order and information structure in Yaqui

117

on frequency counts. On the other hand, one of the most straightforward ways to determine whether a language is pragmatically flexible or not is via direct elicitation, since this methodology makes it possible to create unambiguous contexts which trigger clearly distinct focus structures. Thus one may test, for instance, whether the same syntactic structure used to respond to ‘What happened?’ may also appear in relation to questions which assign different pragmatic roles to each the participants involved (suggesting pragmatic flexibility) or, instead, if a different constituent order or a different syntactic construction has to be employed (suggesting pragmatic rigidity). In this paper we have taken both methodological approaches. First, we will present a brief sketch of Yaqui clause structure and its commonly assumed basic word order (Section 2), which will be evaluated with respect to the patterns found in a corpus of oral narratives (Section 3). The corpus data challenge the expected syntactic flexibility of the language with respect to the ordering of nominal arguments, and indicate that word order in Yaqui is strongly determined by grammatical relations (Section 3.1). It is found that, in fact, even so-called “second position” subject clitics frequently appear in the sentence-initial position expected for nominal subjects (Section 3.2). The facts discussed in this section show that the language should be categorized as syntactically rigid (in terms of the typology mentioned above) and, further, as strongly SOV. The most frequent marked construction involves the use of postnuclear subjects or objects. Finally, in Section 4 we draw on direct elicitation to provide a first description of the preferred means for encoding different focus structures in the language, based on Lambrecht’s (1994) characterization of focus-types. The evidence presented here indicates that Yaqui is pragmatically flexible and information structure is mainly encoded prosodically. However, marked syntax can be used to encode some discourse-pragmatic categories; we found evidence of postnuclear subjects and objects occurring in a right-dislocated position and functioning pragmatically as “antitopics” (Lambrecht 2001).

2. Basic clause structure Yaqui is a synthetic/agglutinative language, with a nominative-accusative case system. Lexical nominatives are morphologically unmarked, and accusatives are marked by the suffix -ta (1a). There is no dative case; rather, “indirect objects” are marked by directional postpositions (1b). Multiple object constructions (1c) are also very common (Guerrero and Van Valin 2004).2

118

L. Guerrero and V. Belloro

(1) a. Amani wa’a rancheo-∅ wakas-ta o kaba’i-ta ta’aru-k. there DEM farmer-NOM cow-ACC or horse-ACC loose-PFV ‘This farmer lost a cow or a horse over there.’ (Jiak nokpo’s 5: 10) b. U o’ou-Ø jamut-ta-u nooka-k. DET man-NOM woman-ACC-DIR talk-PFV ‘The man talked to the woman.’ c. Aurelia-Ø Ivan-ta mo’obei-ta jinu-ria-k. Aurelia-NOM Ivan-ACC hat-ACC buy-APPL-PFV ‘Aurelia bought Ivan a hat.’ The examples above also illustrate the canonical arrangement of constituents in the Yaqui sentence, which justifies its categorization as a verb-final language. As most of the languages of the same type, Yaqui employs postpositions (2a) and verbal suffixes (2b), and within a genitive phrase the genitive precedes the head noun (2c). (2) a. Bw’awa-ta betchi’ibo. soup-ACC for ‘For the soup.’ b. Ili uusi-Ø bwaan-taite-k. little child-NOM cry-begin-PFV ‘The little child started crying.’ c. Joan-ta juubi = ne bicha-k. Joan-ACC wife =1SG.NOM see-PFV ‘I saw John’s wife.’ The examples in (1) also show an arrangement where subjects precede objects, although it is commonly argued that the ordering of subjects and objects with respect to each other is relatively free (Escalante 1990; Rude 1996; Dedrick and Casad 1999: 39; Félix 2000; Guerrero 2006). Both SOV and OSV patterns are judged equally grammatical, the preference of one over the other presumably dependent on pragmatic factors. This word order flexibility is illustrated in (3), from Escalante (1985: 36). (3) a. Juan-∅ Maria-ta María-ACC Juan-NOM ‘John saw Mary.’ b. Maria-ta Juan-∅ María-ACC Juan-NOM

‘John saw Mary.’

bicha-k. see-PFV bicha-k. see-PFV

On word order and information structure in Yaqui

119

With respect to the position of pronouns, it depends on their type. Traditionally, Yaqui pronouns have been divided into independent (“full”) and dependent (“reduced”) forms (Langacker 1977: 124-6 for Uto-Aztecan). “Full” pronouns are expected to behave as lexical elements in terms of their distribution. On the other hand, “reduced” nominative pronouns have been traditionally analyzed as occupying a fixed position in the clause (namely, as “second position” clitics). “Reduced” accusatives (available only for third person) cliticize to the verb. A third set of “reduced” forms occur as objects of postpositions. The relevant paradigms are presented in Table 1. Table 1. Yaqui pronominal system Nominative Accusative 1 Sg 2 Sg 3 Sg 1 Pl 2 Pl 3 Pl

inepo empo aapo itepo eme’e bempo

=ne =’e =Ø =te =’em =mme

Nee enchi apo’ik a= Itom enchim apo’im am=

Object of postposition neeaitoemoame-

The possible relation between alternative word orders and the expression of pragmatic categories has received, up to now, limited attention. Dedrick and Casad (1999: 43-45) comment on a structure where the first position in the sentence is occupied by a topicalized element. Characteristically, this constituent is separated from the rest of the clause by a so-called “introducer element” serving as discourse connector, such as into ~ intok3 ‘and (thus)’ (4a) and bea ‘then’ (4b). (4) a. Kauwis-∅ intok pocho’o-kun-bicha-u bwite-k. fox-NOM and woods-to-site-to run-PFV ‘And the fox ran toward the woods.’ (Dedrick and Casad 1999: 43) b. Name’e-∅ bea nee weeya-n reso-po. These-NOM then 1SG.ACC guide-PASTC pray-LOC ‘And these (women) guided me in the prayers.’ (Maejto: 14) Dedrick and Casad also mention a structure where the sentence-initial position is occupied by a focal element, like the pronominal aapo ‘he’ in (5). It is unclear, however, whether the authors believe that the positions occupied by the pronoun in (5) and the topicalized constituents in (4) should be structurally distinguished. 4

120

(5)

L. Guerrero and V. Belloro

Aapo intok ejea-ta-betuk taawa-k. 3SG.NOM and ironwood:tree-ACC-under stay-PFV ‘And he stayed under the ironwood tree.’ (Dedrick and Casad 1999: 44)

To summarize, it is generally agreed upon that Yaqui is strongly verb final and that, typically, subjects precede objects, although SOV and OSV patterns are equally possible. With respect to pronominal arguments, it is commonly held that full forms may exhibit the same ordering as their corresponding nominal counterparts, whereas the order of reduced forms is fixed: accusatives attach to the verb and nominatives occur in second position, cliticizing to whatever element appears first. So far, two constructions where word order correlates with information structure have been described; one where the sentence-initial constituent is interpreted as topical, and another where it is interpreted as focal. For the most part, however, these claims are based exclusively on direct elicitation data. In this regard, the analysis of naturally-occurring texts appears a necessary step in order to examine whether the posited patterns are found in natural, connected discourse and, ideally, what are the functions served by ordering alternatives. With these goals in mind, the following section examines word order patterns in a corpus of oral narratives. The two sub-sections concern nominal arguments and full pronouns (3.1.) and the position of subject clitics (3.2.).

3. Word order in Yaqui texts 3.1. Nominal arguments and strong pronouns The following discussion on word order patterns is based on a corpus formed by the oral narratives listed in Table 2. The sample includes six folktale stories from Jiak nokpo’s collection (Silva et al. 1998), and three life stories (Silva 2004, n.d.; Félix n.d.), providing a total of nearly 1400 active clauses.5

On word order and information structure in Yaqui Table 2. Yaqui oral texts Text Jiak nokpo’s (Silva et al.) Hilario (Silva) Maejto (Silva) CB (Félix)

121

Clauses 337 478 154 432 1401

As mentioned above, it is traditionally assumed that the order of subject and object with respect to each other is relatively free when these arguments are realized by nouns. In order to examine whether this claim is consistent with natural discourse data, all transitive clauses with nominal Ss and Os were selected from the corpus; full pronouns and demonstrative pronouns (cf. (7)) were also included since they show a similar behavior within the clause. The relative frequency of each attested order is presented in Table 3. Table 3. Word order in transitive clauses SOV OSV Jiak nokpo’s 49 Hilario 21 Maejto 13 CB 18 1 101 (78%) 1 (1%) 102 (79 %)

OVS SVO 4 9 5 3 2 1 4 12 (9%) 16 (12%) 28 (21 %)

As expected, both arguments usually precede the verb (79%). Note, however, that there is only one instance of OSV. This suggests that this order, if not fully ungrammatical, is highly marked. The examples in (6) illustrate the most frequent SOV pattern, where S is a lexical noun (6a) or a full pronoun (6b); (7) presents the only attested OSV example. kosineo-∅ itom tea-ka. cook-NOM 1PL.ACC find-PFV ‘The cook found us.’ (CB: 118) b. Itepo into i-me’e yoeme-m aman 3PL.NOM and DEM-PL man-PL there ‘We sent those men there.’ (Hilario: 69)

(6) a. U

DET

(7)

bittua-kan. send-PASTC

Bea ju-ka in jaboi-∅ ta-ta’a-k. then DET-ACC 1SG.GEN grandfather-NOM RED-know-PFV ‘Then, my grandfather met that one.’ (CB: 17)

122

L. Guerrero and V. Belloro

The data in Table 3 also show that, after sentence initial, the next most frequent position for subjects is in fact postverbal (although no full pronouns nor demonstratives where found in this position). All postverbal subjects in the corpus are marked as definite by a determiner that is optional with preverbal subjects.6 Some instances of OVS are presented in (8). (8) a. Kia itom aman jika-u go’ota-∅ juna-me’e peron-im. only 1PL.ACC there up-DIR throw-PRES DET-PL pelón-PL ‘They only throw us up (onto the train), those pelones.’ (Hilario: 39) b. Junaman itom to’o-bwite-k juna’a Mayor-∅ there 1PL.ACC leave-run-PFV DEM Mayor-NOM ‘He abandoned us there, the Mayor.’ (Hilario: 185) The third most common word order of nominal arguments found in the corpus is SVO. These postverbal objects exhibit the same morphosyntactic constraints as postverbal subjects: they exclude full pronouns and demonstratives, and the nominal form is obligatorily marked by a definite determiner. Further, postverbal objects must co-occur with a pronominal element occupying the canonical object position. The SVO pattern is illustrated with the examples in (9). (9) a. junum bea inepo ai ansu-k u-ka eskuela-tai. like.this and 1SG.NOM 3SG.ACC finish-PFV DET-ACC school-ACC ‘I finished the school like this.’ (CB: 346) b. chukula itepo mura-m-met ami pu’akta-k later 1PL.NOM mule-PL-LOC 3PL.ACC load-PFV ju-me’e koko-ka-me ito-betanai. DET-PL die.PL-PFV-REL 1PL-side ‘Later on, we loaded our deaths in the mules.’ (Hilario: 155) With respect to intransitive clauses, subjects also occur most commonly in preverbal position (Table 4). Table 4. Word order in intransitive clauses SV Jiak nokpo’s 60 Hilario 28 Maejto 9 CB 59 156 (85%)

VS 17 7 3 27 (15%)

On word order and information structure in Yaqui

123

This preference is independent of the verb type, occurring for instance with both unergative (10a) and unaccusative (10b) verbs. (10) a. Ju-me’e Caantora-m bwi-bwika-o. DET-PL singer-PL RED-sing-when ‘When the singers sang.’ (Maejto: 26). b. Junama in yoyo’owa-m jo’ak. there 1SG.GEN RED-parent-PL live.PFV ‘My parents lived there.’ (CB: 8) Postverbal intransitive subjects obey the same constraints that apply to transitive clauses with respect to definiteness. An example of VS is presented in (11). (11) Taawa-k ju’u boobok bo’o-jo-reo. stay-PFV DET toad road-make-AGT ‘The wandering toad stayed.’ (Jiak nokpo’s 3: 29) In sum, the frequencies from the narrative data question the previously assumed flexibility of preverbal nominal arguments and suggest that, instead, the order of nominal arguments is strongly determined by grammatical relations. Alternatives to the most frequent SOV order are relatively infrequent, and most commonly involve postverbal nominative and accusative phrases, which must be marked as definite (and in the case of accusatives, co-occur with a pronominal counterpart). The next section analyzes the ordering restriction of “reduced” forms, in particular so-called “second position subject clitics”.

3.2. Subject clitics We mentioned in Section 2 that it is commonly argued that Yaqui exhibits a widespread trait among Uto-Aztecan languages, namely the existence of socalled “second position clitics”. The occurrence of these forms was first attested by an 18th century grammarian, who in describing the language “states as one of his most infallible rules that the pronominal subject [read clitic pronoun-Steele] must be the second word or element in the sentence...” (Steele 1977:553, citing from Mason 1923, emphasis in the original). This constraint is clearly observed in Tarahumara, a language closely related to Yaqui, belonging also to the Taracahita branch. In Tarahumara, second posi-

124

L. Guerrero and V. Belloro

tion is defined in strictly linear terms, as illustrated in (12), where the subject clitic occurs after the first word of the initial constituent. (12) Semati ne napaha rarimea aré. Nice 1SG shirt gonna:buy probably ‘I am probably going to buy a nice shirt’ (Llaguno 1970: 29) Although the pervasiveness of “second position clitics” in Yaqui is widely accepted, the literature is not clear on the criteria under which this second position should be determined. It has been claimed that a subject clitic occurs “following the first word” in the sentence (Dedrick and Casad 1999: 242), following the first “element” (Steele 1977:541), or “following the first constituent, whatever that may be” (Escalante 1990: 48). It is therefore ambiguous whether in Yaqui second position should be understood, as in Tarahumara, in strictly linear terms or in syntactic ones (i.e., after the first phrasal constituent). The examples provided in the grammars suggest that the second interpretation allows for a more consistent analysis (13). (13) a. siika= ne go = 1SG.NOM ‘I am going now.’ (Dedrick and Casad 1999: 242) b. Tutuka-biako-o = ne koche=su woo’o Yesterday time ago-when = 1SG.NOM sleep=EMP mosquito nee ke’e-ka. 1SG.ACC bite-PFV ‘Last night while I was sleeping, a mosquito bite me.’ (Dedrick and Casad 1999: 48) Be it as it may, the analysis of our corpus shows that either criterion (first word or first phrase) if far from uncovering an absolute pattern. In fact, subject clitics seem to exhibit a much greater flexibility in their ordering possibilities that was previously held. Table 5 illustrates the distribution of subject clitics in Yaqui transitive clauses (X stands for any non-subject phrasal constituent, either an object or an adjunct). Table 5. Transitive clauses with “subject clitics” XSclV XXSclV Jiak nokpo’s 8 3 Hilario 22 12 maejto 21 5 CB 15 14 66 (41%) 34 (21%)

SclXV 7 19 13 22 61 (38%)

On word order and information structure in Yaqui

125

In terms of ordering possibilities, we did not find any instance where the reduced form occupied a “second position” cliticized to a verb (cf. (13a)). There were also no instances in the corpus where “second position” can be unambiguously defined, as in Tarahumara, in strictly linear terms. Whereas the first constituent in (14a) and (14b) can be analyzed either as a word or as a phrase, the last two examples are clear instances of phrasal constituents. (14) a. Ka=ne ame beas kop-kopte-∅. NEG = 1SG.NOM 3PL.ACC about RED-forget-PRES ‘I don’t forget them.’ (CB: 368) b. Aman= te kari-ta am ya’a-ria-k. there = 1PL.NOM house-ACC 3PL.ACC do-APPL-PFV ‘There, we built them a house.’ (Hilario: 13) c. Bea sejtul ta’a-po ket=ne a-u waate Then one day-LOC too=1SG.NOM 3SG-DIR remember ‘I also remember one day.’ (CB: 108) d. Tajkai huevona-m ke=ne bwa'e into jaiba-m. and crab-PL tortilla huevona-PL no.yet=1SG.NOM eat ‘I haven’t eaten yet ‘tortillas huevonas’ and crab.’ (Hilario: 260) Even if “second position” is defined in terms of syntactic constituents, note that the assumed tendency is also belied by instances where reduced subject pronominals also appear following more than one phrasal element, as in (15). (15) a. Amak goki-m-mea amak bisikleeta-t bea then sometimes foot-PL-INSTR sometimes bike-LOC = ne wee-ne. 1SG.NOM go-POT ‘Sometimes I go by walking, sometimes by bike.’ (CB: 192) b. Aman beseo-m=nee jajase-o. 7 there calf-PL=1SG:ACC track-when ‘When I track the calf over there.’ (Hilario: 2) c. Kaa bwe’ura Lio-nok-ta reso-ta jiba=ne NEG big God-word-ACC pray-ACC only=1SG.NOM ta’a-n. know-PASTC ‘But I didn’t know all the “God words”, only the prayers.’ (Maejto: 5)

126

L. Guerrero and V. Belloro

d. Junak bea batte cien metro-po bat=te kaate. Later then almost 100 metro-LOC ahead=1PL.NOM go.PFV ‘And later, we were ahead by almost one hundred meters.’ (Hilario: 138). An anonymous reviewer suggested that some of these examples could be analyzed as small clauses functioning as matrix objects, e.g., the calf over there in (15b), in which case they could be consistent with the “second position” rule, if the small clause is considered to occupy the first position. Still, the relevance of “second position” with respect to this phenomenon is questioned by the fact that it can be defined with respect to a word, a phrase, or a small clause (under such analysis), and also by the instances where the pronominal follows more than one phrase. Further, note that Table 5 shows that the “reduced” subjects appear sentence-initially almost as frequently as they occur in this putative “second position”. Some examples are presented in (16). (16) a. Ne kaa into nappat a tekipanoa-∅ ian tajti. 1SG.NOM NEG and either 3SG.ACC work-PRES now until ‘Up to now, I cannot work either.’(Maejto: 98) b. Ne in maala jariu-bae. 1SG.NOM 1SG.GEN mother look-DESID ‘I want to look for my mother.’ (Hilario: 221) c. Ne soda-ta je'e-ka merkao-wi. 1SG.NOM soda-ACC drink-PFV market-DIR ‘I drank a soda in the market.’ (Hilario: 165) In sum, the analysis of the corpus data presents some interesting puzzles for the common view of reduced nominative pronominals in Yaqui as “second position clitics”. That this view should be refined is particularly clear in light of examples such as those in (16), where there is no preceding material to which the pronoun may cliticize, and cases both as those in (15) and (16) where the pronoun appears out of its expected second position. These data suggest that more detailed studies of these forms should be pursued, which could further uncover the motivations behind their alternative orderings. The findings about subject clitics presented in this section challenge current assumptions and justify the need for further studies related to the phenomena of cliticization in Yaqui, including the detailed analysis of the factors behind the occurrence of subject clitics in one position or the other and

On word order and information structure in Yaqui

127

the presence of multiple pronouns. Leaving these issues for future endeavors, in the next section we go back to Van Valin’s (1999) typology on the interaction between syntax and pragmatics in the determination of word order patterns. The analysis of corpus frequencies suggested that in Yaqui word order of major constituents is mainly determined by grammatical relations, justifying the categorization of the language as syntactically rigid. Now we examine how main information structure types are encoded in the clause, and how the language should be characterized pragmatically. 4. Information structure in Yaqui We saw that the corpus data strongly indicates that clausal constituents follow a basic SOV order. It is not surprising that this order is also the one used to encode predicate focus, since this structure also constitutes the unmarked focus type (Lambrecht 1994). A predicate focus structure in Yaqui is illustrated in (17). In this structure, the subject (typically a pronoun) is the topic, whereas the object and the verb constitute the focus (marked with small caps).8 (17) a. Jita-sa u jamut-∅ ya’a-∅? What-Q DET woman-NOM do-PRES ‘What is the woman doing?’ /aapo KAFE-TA b. U jamut-∅ DET woman-NOM 3SG.NOM coffee-ACC ‘The woman / she is toasting the coffee.’

BWASA-∅.

toast-PRES

In Yaqui there is no constraint over focal preverbal subjects. Just as in English and other pragmatically flexible languages, the same basic word order is used in sentence-focus constructions, as illustrated in (18): (18) a. Jita-sa weye-k? What-Q go.SG-PFV ‘What happened?’ JAMUT -∅ TOTO’I-M JINU-K. b. U DET woman-NOM hen-PL buy-PFV ‘The woman bought the hens.’ The canonical SOV order is also used for narrow focus constructions, regardless of whether the focus targets the object (19) or the subject (20).

128

(19)

(20)

L. Guerrero and V. Belloro

Q: -‘What did the woman buy?’ TOTO’I-M A: - U jamut-∅ DET woman-NOM hen-PL ‘The woman bought hens.’

jinu-k. buy-PFV

Q: -‘Who bought the hens?’ JAMUT-∅ toto’i-m jinu-k. A: -U DET woman-NOM hen-PL buy-PFV ‘The woman bought the hens.’

On the other hand, our informants consistently considered OSV order ungrammatical, regardless of the focus structure assigned to it. This is illustrated in (21). Recall that we only found one such instance in the corpus (cf. Table 4). Their low frequency and the judgment of native speakers about their acceptability indicate that, if not fully ungrammatical for all speakers, OSV order is highly dispreferred. (21)

Q: -‘Who bought the hens?’ / ‘What did the woman buy?’/ ‘What happened?’ jinu-k. A: * Toto’i-m u jamut-∅ hen-PL DET woman-NOM buy-PFV ‘The woman bought the hens.’

The fact that the canonical SOV appears consistently in response to different focus structures indicates that, like English, Yaqui is pragmatically flexible, in as much as there does not seem to be any constraints in terms of focus assignment to any syntactic position within the clause. Speakers offer clauses with the canonical SOV order regardless of what constituent is in focus, and different interpretations are achieved by differences in the prosodic prominence associated with the different constituents. We saw, however, that in connected discourse we found a marked pattern where a lexical argument, either S or O, occurred postverbally. As mentioned above, all postverbal arguments must be marked as definite and, in the case of accusatives, the canonical object position must occupied by a pronominal form. These postverbal constituents occupy a right-dislocated position and, consequently, cannot be focal. The examples in (22) illustrate the infelicity of

On word order and information structure in Yaqui

129

VS, OVS and SVO orders with a focal interpretation of the postverbal constituent (pragmatically unacceptable sentences are marked #). (22)

Q: ‘Who danced?’/ ‘What happened?’ a. #Ye’e-k, U QUETA-∅ DET Queta-NOM dance-PFV ‘Queta danced.’ Q: Who bought the hens? b. #U-me toto’i-m jinu-k, U JAMUT-∅ . DET-PL hen-PL buy-PFV DET woman-NOM ‘She bought the hens, the woman.’ Q: What did the woman buy? c. #U jamut-∅ ami = jinu-k U-ME TOTO’I-Mi. DET woman-NOM 3PL.ACC = buy-PFV DET-PL hen-PL ‘The woman bought them, the hens.’

The right-dislocated position in Yaqui is associated with the pragmatic function of “antitopic” (Lambrecht 2001). Note that, as is the case for most languages (cf. Lambrecht 2001 and references cited therein) the data suggest that Yaqui antitopics not only must be definite, and hence identifiable, but further, the source of the identifiability of the referent must be in the previous context. In all cases in the corpus the postverbal element refers to a “discourse-old” element (Prince 1992). We could not find any instances where the postposed element referred to an entity not previously introduced in the discourse context, even in the cases where these constituents denoted elements or characters which could be considered identifiable because of their widely known status in the community. The antecedent for the postverbal phrase is typically mentioned a few clauses prior in the context, suggesting that it is not maximally “active” (Chafe 1994) when the antitopic construction is used. Below we illustrate some of the contexts where the examples of postnuclear arguments mentioned in Section 3 (repeated below) occur. The relevant sentences illustrate OVS (23), SVO (24) and VS (25) orders, respectively. (23) Context: Abion ito-t siika, tettebote-sime. Bweta int-uchi notte-ka bea, ta’abwi=si itom tebo-tua-k. Metrayadora-e itom mumuki-su-k... Yokoria-po beja te aman bwij-wa-k. Nau konta-wa-k. Ju-me’e

130

L. Guerrero and V. Belloro

Aguilenyo-m tea-me ama rejte into ju-me’e peron-im.9 Junak bea te Pitaya-u te kom saja-k. Pitaya-u lula bea te weiya-wa. Junaman riel bo’o-u te toji-wa-ka, ama ja’abwa-wa-k, junaman bea te desarmaroa-wa-k wiko’e-ka-me. Bweta juna-me’e into kaa parke-k. ¿Jita-e-sa nassua-ne kia bea wiko’e-ka? Ama rejte. Nakbea te vagone-u kima’awa-k. ‘And an airplane flew above us, greeting us. But then it returned and greeted us very differently. It shot us with a machine gun… The next day, we were captured. We were surrounded. The so-called Aguileños and the Pelones were there. Then, we went down to Las Pitayas; we were taken directly to Las Pitayas. And there, after we were taken to the railways, we were lined up, and those having rifles were disarmed. But those [rifles] didn’t have bullets. How could we fight, just with the rifles? There we were. And later, we were put into a wagon.’ (Hilario: 28-38) Kia itom aman jika-u go’ota-∅ juna-me’e peron-im. only 1PL.ACC there up-DIR throw-PRESDET-PL pelón-PL ‘They only throw us up (onto the train), those pelones.’ (Hilario: 39) (24) Context: Junak bea u jamut beeme inen au jiia-k:“Nolia in o’ola ji’ibwa e; empo jumaa si tebaure kia ne kaa a’abo weye-n”. U yoeme into au nattemae-k:“¿Jaisaakai?”.“Bwe’ituk ne si obisaane-n”. U yoeme kia “mmm-po” taawa-k, kaachini into au jiia-k. Junuen luula yesjte-se-kan, aapo into bachi-ta chiu-taite-k. ‘Then, the young woman said: “C’mon, old man, eat; perhaps you are hungry and I wasn’t coming [to bring you food]”. And then the man asked: “Why?”. “Because I was very busy”. And the man just said “mmmhm”, and said no more to her. At noon, he began to clean the corn.’ (Jiak nokpo’s 2:18-25) Jiba ai bitchu-k u-ka’a jamut-tai. always 3SG.ACC watch-PFV DET-ACC woman-ACC ‘He observed her all the time, the woman.’ (Jiak nokpo’s 2: 26) (25) Context: Ju’u boobok juchi notte-k. Ju’u boobok bo’o-jo-reo-ta-u en jia-k: tu’i achai o’ola mabet-wa-ka, ta inim koobi’iku jiba yumjo’e-ne. Bwe tu’i chuba-la ne yumjoe-bae. Yooko ne ket jakko sim-bae, junuen jia-ka.

On word order and information structure in Yaqui

131

‘The toad came back. He told this to the wandering toad: “it is ok old man, you were accepted, but you must stay in this corner only”. Ok then, I will rest for a little while. I will leave tomorrow, he said.’ (Jiak nokpo’s 3: 24-28) Taawa-k ju’u boobok bo’o-jo-reo. stay-PFV DET toad road-make-AGT ‘The wandering toad stayed.’

(Jiak nokpo’s 3: 29)

With respect to the coding of arguments, the expected tendency was confirmed, where “reduced” pronominals encode continuous topics, full pronouns typically carry a contrastive meaning and nominals are the preferred means for introducing new participants. This is illustrated with the following passage from one of the folktales, where two turtles are discussing how to beat a coyote in a racing contest. Note the use of the full pronominals empo and inepo in (26b), (26c) and (26d) for contrasting the tasks assigned to speaker and addressee. In (27) the sequences of reduced forms after a full pronoun are used to encode maximally continuous topics. (26) a. Ika’a = te ya’a-ne=ti bea a-u jiia: that.ACC = 1PL.NOM do-POT CLM then 3SG-DIR say ‘We will do this, (the turtle) says to her:’ b. Empo bea bo’o jinko’ola naate’e-po yej-ne road competition begin-LOC stand-POT 2SG.NOM then ‘You will be where the racetrack begins’ c. Inepo into gojo’oria-po bo’o jinko’ola-ta 1SG.NOM and hole-LOC road competition-ACC lu’ute’e-po yej-ne. end-LOC stand-POT ‘And I will stand in the hole where the racetrack ends.’ d. Into bea empo bwia-ta betuk emo and then 2SG.NOM land-ACC under REFL bwiti-bae = ti jiu-ne. run.SG-DESID = CLM say-POT ‘And then you will say (you) want to run underground.’ e. Apo’ik bwiiti-taite-k-o 3SG.ACC run-start-PFV-when ‘When he starts running,’ f. Empo junama’a ji’ibweji-taiti-ne, there scratch-start-POT 2SG.NOM ‘You will start scratching.’

132

L. Guerrero and V. Belloro

(27) a. Inepo 1982-po = ne mujti-tua-wa-k 1SG.NOM 1982-LOC = 1SG.NOM consecrate-CAUSE-PASS-PFV ‘I was consecrated in 1982.’ b. junu-ka’a cargo-ta = ne mak-wa-k. DET.ACC cargo-ACC= 1SG.NOM give-PASS-PFV ‘I was given that position.’ c. Ne kia tukaria-t naabujti ino majta-k. 1SG.NOM also night-LOC in.addition 1SG.REFL learn-PFV ‘In addition, I also prepared myself at nights.’ (Maejto: 103-105) The evidence presented in this section indicates that Yaqui is pragmatically flexible and focus can be assigned to any position in the clause. However, this preliminary look at corpus data suggests that right-dislocated structures are used to encode relatively topical subjects and objects if they are not assumed to be maximally active in the mind of the interlocutor. Maximally active referents (i.e. continuous topics), on the other hand, are typically encoded with reduced pronominal forms.

5. Final comments The aim of this paper has been to provide a first pass in the study of word order and information structure in the Yaqui language. It was shown that a basic SOV order can be determined with respect to frequencies in a corpus of oral narratives. The data called into question, however, the assumed flexibility of the language with respect to the ordering of nominal arguments, showing a strong tendency for placing subjects sentence-initially. We presented data which shows that Yaqui speakers make use of a right-dislocated position for placing antitopics. We did not find, however, clear arguments for distinguishing a left-dislocated structure used for topicalized participants. In terms of the typology mentioned in the introduction, we presented evidence indicating that the language is syntactically rigid but pragmatically flexible. There are however, many new questions waiting for investigation. We hope that this preliminary exploration can serve as a basis for future studies, and that through this we can begin to incorporate a discourse-pragmatic perspective into the study of Yaqui and other Uto-Aztecan languages.

On word order and information structure in Yaqui

133

Notes 1.

2.

3.

4.

5.

6.

7.

8.

The Yaqui language is spoken mainly in Mexico, by more than 15,000 people living along the Yaqui River in the Central West part of the State of Sonora. Across the U.S.-Mexican border, in Pascua, Arizona, there are an estimated 1,000 speakers, who migrated there at the beginning of the 20th century. There are several grammatical studies on Yaqui; among the most significant are Crumrine (1961), Johnson (1962), Lindenfeld (1973), Escalante (1990), Jelinek and Escalante (2000), and Demers, Escalante and Jelinek’s (1999) article on the principles that account for prosodic prominence in isolated words, all based on the Arizona dialect. Dedrick and Casad (1999), Félix (2000), Hernández (2002), Guerrero and Van Valin (2004), Silva (2004), Guerrero (2004a,b, 2006), Martinez (2007), as well as several articles in Estrada et al. (2008) and Estrada et al. (2007) are all on the Sonoran dialect. Abbreviations; ACC: accusative, AGT: agentive, APPL: applicative, CLM: clause linkage marker, COM: commitative, DESID : desiderative, DET: determiner, DIR: directional, FUT: future; GEN: genitive, LOC: locative, NEG: negation, NOM: nominative, NMLZ: nominalizer, PASTC: past continuative, POT: potential, PPIO: participle, PFV: perfective, PASS: passive, PL: plural, PRES: present, PURP: purpose, Q: question particle, RED: reduplication, REFL: reflexive, SG: singular, TNS: unmarked tense. The distribution and functions of into(k) within coordinated units have been explored by Martinez (2007). The author claims this particle cannot be considered a second position clitic (2007: 114) but nothing is said about its pragmatic functions. The extent to which into(k) and the other discourse connectors are used to set off the topicalized or focused material is still an open question. In his study of Uto-Aztecan languages, Langacker (1977:25ff) recognizes as a common trait of this language family the displacement to sentence-initial position as a way to mark emphasis, although again this interpretation is linked to both topical and focal readings. We did not find clear evidence for these structures in our data and therefore they will not be discussed in this paper. Non-predicative clauses and matrix clauses with speech act predicates were excluded from this analysis. We owe our gratitude to Cresencio Buitimea, Carlos Silva and Rolando Félix for letting us use their oral narratives. In Yaqui, definiteness is associated with the presence of determiners. Determiners are marked by –ta when modifying an accusative NP, but with –e if the NP is marked by a postposition; if the N is plural, then the determiner is likewise plural. In adverbial and complement clauses, pronominal subjects are accusative and nominals are marked by -ta (Guerrero 2006). Except for this morphological coding, the distribution of major constituents in main and dependent clauses is the same. The analysis in this section follows Lambrecht’s (1994) characterization of focus types as adopted within the framework of Role and Reference Grammar.

134

9.

L. Guerrero and V. Belloro For details, we refer the reader to Van Valin and LaPolla (1997:199-241) and Van Valin (2005:68-88). Some studies dealing with information structure within this model include Bentley (2008) for Italian and Sicilian; Shimojo (1995, in press) for Japanese; Belloro (2007, 2009) for Spanish. There is a prior mention of this participant, which was not included in the selected context for simplicity.

References Belloro, Valeria 2007 The syntax-pragmatic interface. A study of Spanish clitic doubling. Ph.D. diss., University at Buffalo, The State University of New York. Belloro, Valeria 2009 Spanish datives: remarks on the information-structure side of the story. In Studies in Role and Reference Grammar. Lilián Guerrero, Sergio Ibáñez, and Valeria A. Belloro (eds.). México: UNAM. Bentley, Delia 2008 The interplay of focus structure and syntax: evidence from two sister languages. In Investigations of the Syntax-Semantics-Pragmatics Interface. Robert D. Van Valin, Jr. (ed.). Amsterdam: John Benjamins. Barreras, Aguilar 1998 Orden de palabras básico en el guarijío de Sonora. Memorias del IV Encuentro de Lingüística en el Noroeste. A. Acosta, Z. Estrada Fernández, M. Figueroa, and G. López Cruz (eds). Universidad de Sonora, Hermosillo, 37-68. Buitimea, Cresencio 2007 Preesio betana nottiwame. Hermosillo: Universidad de Sonora. Chafe, Wallace 1994 Discourse, Consciousness and Time. Chicago: University of Chicago Press. Cumrine, Lynne S. 1961 The phonology of Arizona Yaqui, with texts. Tucson: The University of Arizona Press. Dedrick, John and Eugene H. Casad 1999 Sonora Yaqui Language Structure. Tucson: The University of Arizona Press. Demers, Richard, Fernando Escalante and Eloise Jelinek. 1999 Prominence in Yaqui words. International Journal of American Linguistics 65 (1):40-55. Escalante, Fernando 1990 Voice and Argument Structure in Yaqui. Ph.D. diss., The University of Arizona.

On word order and information structure in Yaqui

135

Escalante, Fernando 1985 A preliminary view of the structure of Yaqui. M.A thesis., The University of Arizona. Estrada, Zarina 1985 El orden de palabras en pima bajo. Paper presented at the Friends of Uto-Aztecan Working Conference, Tucson, Az. Estrada, Zarina and Lilián Guerrero 2007 Grammatical borrowing in Yaqui. In Grammatical Borrowing in CrossLinguistic Perspective. Matras Yaron and Jeanette Sakel (eds.). Berlin: Mouton de Gruyter. Estrada, Zarina, Albert Álvarez, Lilián Guerrero and María Belem Carpio 2007 Mecanismos de voz y formación de palabra [Mechanisms of voice and word formation]. México: Editorial Plaza y Valdes. Estrada, Zarina, Søren Wichmann, Claudine Chamoreau, and Albert Álvarez 2008 Studies in voice and transitivity (Estudios de voz y transitividad). Munich: Lincom Europa. Félix, Rolando 2000 Las relaciones gramaticales en yaqui: un análisis en el marco de la Gramática del Rol y Referencia [Grammatical relations in Yaqui: a Role and Reference Grammar analysis]. M.A. thesis, Universidad de Sonora. Félix, Rolando n.d. Life story of CB (oral text). Manuscript. Guerrero, Lilián 2004a The syntax-semantic interface in Yaqui complex sentences, a Role and Reference Grammar analysis. Ph. D. diss., University at Buffalo, The State University of New York. Guerrero, Lilián 2004b Verbos de movimiento y posición en yaqui [Motion and position verbs in Yaqui]. In Estudios en lenguas amerindias. Homenaje a Ken Hale. Zarina Estrada, Ana Fernández, and Albert Álvarez (eds.), 199-228. Hermosillo: Universidad de Sonora. Guerrero, Lilián 2006 The Structure and Function on Yaqui Complementation. (Studies in Native American Linguistics 54). Munich: Lincom. Guerrero, Lilián and Robert D. Van Valin, Jr. 2004 Yaqui and the analysis of primary object languages. International Journal of American Linguistics 70: 290-319. Hernández Doode, Gabriela 2002 Marcadores discursivos: una exploración en la narrativa yaqui [Discourse markers: A Yaqui narrative exploration]. M.A. Thesis., Universidad de Sonora.

136

L. Guerrero and V. Belloro

Jelinek, Eloise and Fernando Escalante 2000 Unaccusative and unergative verbs inYaqui.Uto-Aztecan: structural, temporal and geographic perspectives. Eugene H. Casad and Thomas L. Willet (eds). México: Universidad de Sonora. Johnson, Jean B. 1962 El idioma yaqui [The Yaqui language]. México: Instituto Nacional de Antropología e Historia. Lambrecht, Knud 1994 Information Structure and Sentence Form. A Theory of Topic, Focus, and the Mental Representations of Discourse Referents. Cambridge: Cambridge University Press. Lambrecht, Knud 2001 Dislocation. In Language Typology and Language Universals. Martin Haspelmath (ed.), 1050-1078. Berlin: Mouton de Gruyter. Langacker, Ronald W. 1977 Studies in Uto-Aztecan Grammar. Vol. 1 An overview of Uto-Aztecan Grammar. Texas: Summer Institute of Linguistics and University of Texas at Arlington. Lindenfeld, Jelinek 1973 Yaqui Syntax. University of California Press. Llaguno, Jose 1970 El tarahumara sin esfuerzo (in colaboration with D. Brambila and E. Uranga). Manuscript. Mason, J. Alden 1923 A preliminary sketch of the Yaqui language. UCPAAE 20: 195-212. Martínez, Constantito 2007 Yaqui coordination. (Studies in Native America Linguistics 59). Munich: Lincom. Mithun, Marianne 1992 Is basic word order universal? In Pragmatics of Word Oder Flexibility. Doris Payne (ed.), 15-61. Amsterdam and Philadelphia: John Benjamins. Payne, Doris 1987 Information structuring in Papago Narrative Discourse. Language 63: 783-804. Payne, Doris 1990 The Pragmatics of Word Order: Typological Dimensions of Verb Initial Languages.Berlin: Mouton de Gruyter. Payne, Doris 1992 Introduction. In Pragmatics of Word Oder Flexibility. Doris Payne (ed.), 1-13. Amsterdam and Philadelphia: John Benjamins.

On word order and information structure in Yaqui

137

Prince, Ellen 1992 The ZPG letter: Subjects, Definiteness, and Information Status. In Discourse Description: Diverse Analysis of a Fund Raising Text. Amsterdam and Philadelphia: John Benjamins. Rude, Noel 1996 Objetos dobles y relaciones gramaticales: el caso del yaqui [Double objects and grammatical relations: the case of Yaqui]. In Memorias del Tercer Encuentro de Lingüística en el Noroeste. Vol. 2. Zarina Estrada Fernández, Max Figueroa Esteva, Gerardo López Cruz (eds.), 491-522. Hermosillo: Editorial Unison. Shimojo, Mitsuaki 1995 Focus Structure and Morphosyntax in Japanese: wa and ga, and Word Order Flexibility. Ph. D. diss., University at Buffalo, The State University of New York. Shimojo, Mitsuaki in press Focus structure and beyond Discourse-pragmatics in Role and Reference Grammar. In Studies in Role and Reference Grammar. Lilián Guerrero, Sergio Ibáñez, and V. Belloro (eds.). México: UNAM. Silva Encinas, Carlos 2004 Life Story of Hilario (oral text). In La secuencia temporal en el discurso narrativo Yaqui [The temporal sequence of narrative discourse in Yaqui] M.A. Thesis. Hermosillo: Universidad de Sonora. Silva Encinas, Carlos n.d. Life story of Maejto (oral text). Manuscript. Silva Encinas, Carlos, Pablo Álvarez Romero, and Cresencio Buitimea 1998 Jiák nokpo étejoim, Pláticas en Lengua Yaqui [Conversations in Yaqui language]. Hermosillo: Universidad de Sonora. Steele, Susan 1977 Clisis and diachrony. In Mechanisms of syntactic change. Charles N. Li (ed.), 539-579. Austin: The University of Texas Press. Thompson, Sandra 1978 Modern English from a Typological Point of View: Some Implications of the Function of Word Order. Linguistische Berichte 54:19-35. Van Valin, Robert D. 1999 A typology of the interaction of focus structure and syntax. In Typology and Linguistic Theory: From Description to Explanation. E. Raxilina and J. Testelec (eds.), 511-24. Moscow: Languages of Russian Culture. Van Valin, Robert D. 2005 Exploring the Syntax-Semantics Interface. Cambridge: Cambridge University Press. Van Valin, Robert D. and Randy J. LaPolla 1997 Syntax. Structure, Meaning and Function. Cambridge: Cambridge: Cambridge University Press.

On the nature of word order in Yucatec Maya* Rodrigo Gutiérrez-Bravo and Jorge Monforte y Madera

1. Introduction It is a standard assumption that in languages that have SVO as their unmarked word order, this order is the result of a “strong” EPP requirement. This requirement is understood in Minimalist analyses as the presence of an EPP feature in I0 or C0, and in Optimality Theoretic analyses as resulting from the ranking EPP » STAY. In this paper we argue that Yucatec Maya (a Mayan language from the Yucatán Peninsula, Mexico: henceforth, Yucatec) instantiates a different kind of SVO language. Yucatec displays a discrepancy between clauses where there is a phonetically-overt object, where the unmarked word order is SVO, and clauses with no object, which are by and large VS. We propose that this phenomenon be referred to as split word order and show that standard analyses of the SVO order cannot account for it. We develop an alternative Optimality Theoretic analysis, where we argue that Yucatec has the ranking STAY » EPP characteristic of verb-initial languages. However, we propose that a higher ranked constraint, which we name DISTINCT, prevents the subject and the object from appearing simultaneously inside the VP. This constraint forces the subject DP to move outside the VP, thus producing the SVO order. Since this constraint is sensitive to whether both the subject and the object have phonetic content, when there is no overt object the resulting order is VS. The paper as a whole is structured as follows. In Section 2 we address the problem of the unmarked word order of transitive clauses in Yucatec and provide evidence that SVO is the unmarked word order, contrary to what is usually claimed in the literature. In Section 3 we address the issue of the word order of clauses without a direct object and conclude that it is mostly VS. We introduce the term split word order to refer to this word order discrepancy. In Section 4 we develop an Optimality Theoretic analysis that accounts for this split, and in Section 5 we present our conclusions.

140

R. Gutiérrez-Bravo and J. Monforte y Madera

2. Yucatec Maya as an SVO language 2.1. Background Mayan Languages (like the majority of Mesoamerican languages) are mostly verb-initial. As such, Yucatec is generally classified as a VOS language, where the SVO order is taken to be the result of topicalization of the subject. Yucatec indeed displays VOS clauses in texts and spontaneous speech. 1 (1) Je’ bin

k-u

la’ach-ik-ø

u

jo’ol x-nuk

ASV CIT HAB-ERG.3 scratch-IND-ABS.3sg ERG.3 head FEM-great

reyna-o’. queen-CL ‘And the great queen truly scratched her head...’

(Gigante-97)

However, none of the works that have dealt in detail with word order in Yucatec (Durbin and Ojeda (1978), Hofling (1984), Briceño Chel (2002)) has actually concluded that the language’s unmarked word order is VOS. Instead, what these works observe is the notorious frequency of the SVO order. This is illustrated in (2), taken from Góngora (1990). (2) Le

ko’olel-o’ t-u

ts’-aj-ø

u

ma’alob nook’.

DM woman-CL CP-ERG.3 put-PRF-ABS.3sg ERG.3 good clothes ‘...and the woman put on her good clothes...’ (Si’ipil-51)

For instance, Durbin and Ojeda (1978) and Hofling (1984) both conclude, on the basis of qualitative criteria, that Yucatec is best understood as a language with two basic word orders: SVO y VOS. Hofling (1984) further acknowledges that if frequency alone is considered, SVO should be taken to be the basic word order. Similarly, based on frequency Briceño (2002) concludes that SVO is the unmarked word order of the language. A similar conclusion is arrived at in Gutiérrez-Bravo and Monforte (2008). As a first step in solving this puzzle, in 2.4 we briefly repeat the evidence which lead us to the conclusion in Gutiérrez-Bravo and Monforte (2008) that Yucatec is not a VOS language. In 2.4.4, we further present evidence from our own corpus of Yucatec oral narratives to support this conclusion.2 Before presenting the relevant evidence, we lay out our assumptions about clause structure in Yucatec. We also briefly discuss the relation between topicalization, focus fronting and the SVO order in this language.

On the nature of word order in Yucatec Maya

141

2.2. Transitive clauses in Yucatec Transitive clauses in Yucatec consist minimally of the verb and a proclitic (glossed ERG) cross-referencing the subject of the verb.3 Most of the time, the proclitic is preceded by an auxiliary particle or verb. The main verb in turn displays a series of suffixes (glossed ABS) that agree with the object.4 This minimal structure is shown in (3). (3)

K-in

w-il-ik-ech.

HAB-ERG.1s EP-see-IND-ABS.2s

‘I see you.’ The subject and object DPs then appear to the left or right of this minimal structure, as in (1) and (2). Following Aissen (1996) and Aissen (1999a), we assume that the VP in Yucatec projects its specifier to the right. Hence we analyze the VOS order of (1) as a structure where both the subject and the object remain in their VP-internal positions, whereas for SVO we assume that the subject has been fronted from [Spec, V] to [Spec, I]. (4)

IP (XP)

I'

(Subject)

I

VP

k-in V' V w-il-ik-ech

DP DP (Subject) Object

2.3. Preverbal subjects, topics and foci As mentioned, in this paper we conclude that the unmarked word order of transitive clauses in Yucatec is SVO. Our definition of unmarked word order corresponds to the one developed in Costa (1998) and Gutiérrez-Bravo (2005). Specifically, following these works we assume that there is a set of purely syntactic constraints that determine the relative order of constituents in a language. These purely syntactic constrains in turn interact constantly with pragmatic, semantic, intonational, and discourse constraints that have their own word order requirements. Unmarked word order is then defined as

142

R. Gutiérrez-Bravo and J. Monforte y Madera

the order that results from the purely syntactic constraints in the language, whereas perturbations of this order resulting from the non-syntactic constraints are understood as derived orders. Because of this interaction between syntactic and non-syntactic word order requirements, in order to determine the unmarked word order of Yucatec it is necessary to simultaneously take into account at least the word order perturbations that result from Information Structure considerations (topicalization and focusing). In terms of the problem we are addressing here, what is necessary is to shown that there are instances of the SVO order where the preverbal subject is not the result of focusing or topicalization. As in any other Mayan language (England (1991), Aissen (1992)), transitive subjects in Yucatec can appear in the preverbal position as the result of topicalization or focusing. Topicalized preverbal constituents (i.e. sentence topics), including the subject, typically take the topic clitic =e’, as in (5). (5)

Juan=e’ t-u ts'on-aj-ø kéej. Juan=TOP CP-ERG.3s shoot-PRF-ABS.3s deer ‘(As to) Juan, he shot a deer.’

Focused transitive subjects also appear to the left of the verb, and they further trigger the Agent Focus form of the verb (Bricker 1978), as happens in many other Mayan languages. In Yucatec, this form differs from the canonical structure of transitive clauses in (3) in that it lacks both the ergative proclitic and the auxiliary that appears to its left (Bricker (1978), Bohnemeyer (2002), Tonhauser (2003), Stiebels (2006)).5 This is shown in (6). (6) Juan il-ik-ech. Juan see-IND-ABS.2sg ‘JUAN sees you.’

(Tonhauser 2003: 211)

Although these two processes have been widely reported in the literature on Yucatec, it has mostly gone unobserved that preverbal transitive subjects in this language can also appear without the topic clitic or the agent focus form. This is illustrated with the preverbal subject leti’ ‘he’ in (7) and máak ‘person’ (i.e. ‘someone’) in (8). (7) Leti’ t-u 3SG

ordenar-t-aj-ø

ka’a túul

saapo

CP-ERG.3 order-TRNS-PRF-ABS.3sg two NUMC toad

On the nature of word order in Yucatec Maya

143

bi-s-ej-ø. ERG.3 go-CAUS-IRR-ABS.3sg u

‘He commanded two toads to take it away (a box).’

(Sapo-1)

(8) Pero wa máak u y-ojel u ts’ak-ø=e’, but if person ERG.3 EP-know ERG.3 cure-ABS.3sg=TOP pues séeb u ts’a’ak-al. then fast ERG.3 cure+PASS-IND ‘But if someone knows how to cure it, then it is cured fast.’ (Sonámbulo-26)

As such, Yucatec displays three different kinds of transitive preverbal subjects. This is sketched in Table 1, where the numbers refer to the example sentences that illustrate each of the preverbal subject types. Table 1. Preverbal transitive subjects Morphosyntactic cue Topicalized subject (5) Focused subject (6) “Neutral” subject (7) and (8)

=e’ Agent Focus Form None

In what follows we present evidence that the subjects labeled as “neutral” above are pragmatically and syntactically unmarked subjects. This points to the conclusion that the unmarked word order of Yucatec is SVO.

2.4. Evidence for SVO as unmarked 2.4.1. Out-of-the-blue questions (“what happened?”) It has been observed that answers to this kind of interrogative are only felicitous when they have the language's unmarked word order (Contreras (1976), Vallduví (1992), Lambrecht (1994); for Spanish see also Gutiérrez-Bravo (2005), and Gutiérrez-Bravo (2008). In order to test for this felicity contrast, different pairs of sentences (one SVO, the other one VOS) were read out loud to five native speakers (in random order with respect to one another) so that they would identify which one was the most natural answer to the question in (9).

144

R. Gutiérrez-Bravo and J. Monforte y Madera

(9)

Ba’ax k-u y-úuch-ul? what HAB-ERG.3 EP-happens-IND. ‘What’s happening?’

(10) Le

koolnáal-o’ t-u jats’-aj-ø le máak-o’. DM peasant-CL CP-ERG.3 beat-PRF-ABS.3sg DM person-CL ‘The peasant beat the man.’

(11) T-u

jats’-aj-ø le máak ASP-ERG.3 beat-PRF-ABS.3sg DM person Idem.

koolnáal-o’.6 DM peasant-CL le

We present the results of this test in Table 2, where it can be seen that speakers consulted expressed their preference for SVO as the answer to (9) 80% of the time. We take this to be a first piece of evidence that the unmarked word order of transitive clauses in Yucatec is SVO. Table 2. Preference to ‘Out of the blue’ questions order

percentage

SVO

8/10

80%

VOS

2/10

20%

2.4.2. Definiteness effects Many languages display definiteness effects in their derived (i.e. marked) word orders, but not in their unmarked word order (Gutiérrez-Bravo 2008). In other words, in constructions displaying different kinds of perturbations from the unmarked word order (topicalization, object shift, existential constructions, etc.) it is not rare to find that the relevant perturbation is only tolerated if the argument displaced from its canonical position is either obligatorily definite or obligatorily indefinite. In contrast, to the best of our knowledge no such effects are observed in a language’s unmarked word order. Hence, the presence of definiteness effects can be taken to be an indication that the order in which they are observed is not the language’s unmarked word order. For Mayan languages in particular, this can be

On the nature of word order in Yucatec Maya

145

illustrated with Tz’utujil, a Mayan language from Guatemala. Tz’utujil, like any other Mayan language, allows SVO constructions like (12). (12) TZ’UTUJIL Ja Taa’ ma t-uu-ya’ r-paq r-xaayii.l SVO the señor NEG ASP-A3SG-give A3-money A3SG-wife ‘The Señor doesn’t give Money to his wife.’ (Dayley 1985: 314) However, as noted in Aissen (1999a), it is fairly clear that SVO is a derived order in Tz’utujil and not its unmarked word order. This is because the preverbal subject needs to be definite in these cases, as shown by the contrast between (13) and (14). This definiteness restriction on preverbal subjects is an indication that the preverbal subject must be a sentence topic, and consequently that the SVO order of (12) is derived by subject topicalization. (13) X-in-ruu-ti’ CMPL-B1SG-A3SG- eat

jun a

kaab’. wasp

VS

‘A wasp bit me.’ (14) *Jun kaab’ x-in-ruu-ti’. a wasp CMPL-B1SG-A3SG-eat

SV (Aissen 1999a: 172)

In contrast, this definiteness effect is not observed in Yucatec SVO clauses. Indefinite DPs and bare NPs in the preverbal position are accepted by speakers without hesitation even in the absence of any context, as shown by elicitation of clauses like (15). SVO clauses with indefinite DPs and bare subject NPs are also readily found in texts, as in (8) and (16). In (16), from a monolingual speaker, the preverbal NP is actually the first instantiation made in the narrative of the referent in question. (15) Jun túul máak t-u kíin-s-aj-ø le one NUMC person CP-ERG.3 die-CAUS-PRF-ABS.3sg DM koolnáal-o’. peasant-CL ‘A person killed the peasant.’ (16) Jun túul máak one NUMC person

te’

Peto ku

y-a’al-ik-ø

LOC Peto HAB-ERG.3 EP-say-IND-ABS.3sg

146

R. Gutiérrez-Bravo and J. Monforte y Madera

teen=e’… 1.SG=TOP

‘A man from Peto used to tell me…’

(Don Pablo-170)

To corroborate this observation further, we applied an elicitation-bytranslation test, the results of which indeed show that the contrast indefiniteverb-definite of (15) does not affect the SVO interpretation of this kind of clause. The elicitation-by-translation test was carried out as follows: two different clauses like (15) were read to five Yucatec speakers (ten trials in total). The speakers were asked to provide a translation in Spanish for each sentence (see also Skopeteas and Verhoeven 2005). One of the trails was discarded because the subject provided an answer unrelated to the current experimental conditions (hence Table 3 below shows the results of only nine trials). From the translation it is possible to determine if these clauses are being interpreted as SVO (a person killed the peasant) or OVS (the peasant killed a person). Crucially, in this test an OVS interpretation indicates that definiteness, and not word order, is the factor that determines which of the two DPs is interpreted as the subject/agent. Following our assumptions about the relation between definiteness effects and word order, this would point to the conclusion that the order argument-verb-argument is derived (i.e. a marked word order). As shown in Table 3, this was not what was observed. The preferred interpretation of this kind of clause was still SVO. This is further evidence that SVO is not a derived order. Table 3. Preferred interpretation of indefinite-verb-definite clauses Interpretation Percentage SVO

8/9

88.88%

OVS7

1/9

11.11%

We further applied this same test with clauses like (17), in which both argument DPs appear after the verb (we henceforth refer to this order as verb-argument-argument order). In these clauses the nominal expression to the immediate right of the verb (the canonical object position) was definite, whereas the second DP was indefinite.8 Again, each speaker was presented with two different clauses, giving ten trials in total.

On the nature of word order in Yucatec Maya

147

(17) Tun

jats’-ik-ø le máak jun túul DUR+ERG.3 beat-IND-ABS.3sg DM person one NUMC koolnáal-o’. peasant-CL

In principle, clauses like (17) have two possible interpretations, ‘A peasant beat the man’ (i.e. VOS) or ‘The man beat a peasant’ (i.e. VSO). The first interpretation corresponds to what has been traditionally claimed to be the canonical word order of Yucatec. The second interpretation instead indicates that a definite post-verbal DP is preferably interpreted as an agent even if it does not occupy the canonical position of the subject/agent. Now, in sharp contrast with what is observed in Table 3, in verb-argumentargument clauses definiteness has a robust effect, to the extent that it thoroughly overrides a potential VOS interpretation. This is shown in Table 4.9 Table 4. Interpretation of Verb-Definite-Indefinite clauses order

percentage

VOS

0/10

0%

VSO

10/10

100%

Under our assumption that definiteness only affects derived orders, this result points to the conclusion that verb-initial orders in Yucatec transitive clauses are derived and not unmarked.

2.4.3. Neutralization of the preverbal position We also carried out a test that can be described as “neutralization of the preverbal position”. This is also an elicitation by translation test in which we “neutralized” the preverbal position by having an interrogative or relative operator occupy it, as in (18). (18) [interrogative or relative pronoun] verb-argument-argument We constructed a number of clauses with this structure (an example is presented in (19)), which were then read to native speakers. The native speakers were asked to provide a translation in Spanish of each example.

148

(19)

R. Gutiérrez-Bravo and J. Monforte y Madera

Tu’ux t-u jats’-aj-ø le máak le koolnáal-o’? where CP-ERG.3 beat-PRF-ABS.3sg DM person DM peasant-CL

In principle, each clause like (19) can have two possible interpretations; ‘Where did the peasant beat the man?’ (i.e. a VOS interpretation where the second DP is taken to be the agent) or ‘Where did the man beat the peasant?’ (i.e. a VSO interpretation). This test was applied to corroborate one of two possible hypotheses. (20)

Hypothesis A. Yucatec is an SVO language. Under this hypothesis, the interpretation of which of the two postverbal DPs is the subject/agent should be affected by the presence of a third element in the preverbal position. This is because the preverbal position is the canonical subject position, but it has now been occupied by a different constituent. Yucatec is a strict head-marking language, so now there is no cue left to determine which of the two postverbal DPs in (19) is the subject. As a result, we expect the interpretation of these constructions not to be systematically VOS.

(21)

Hypothesis B. Yucatec is a VOS language. In this case, the presence of a third element in the preverbal position should not affect the interpretation of clauses like (19). This is because neither the canonical subject position nor the canonical object position (both postverbal) are being affected by the presence of this third element. Interpretation is expected to be systematically VOS.

This test was applied to six speakers of Yucatec. Five different clauses like (19) were presented to each speaker, adding up to thirty trials in total. Out of these, four trails were discarded because the subjects provided answers unrelated to the current experimental conditions. Our results are presented in Table 5, where it can be seen that VOS, while being the preferred interpretation, was still clearly not the systematic interpretation of these clauses. We take this to be support for Hypothesis A and consequently as further evidence that the unmarked word order of Yucatec transitive clauses is SVO and not VOS.

On the nature of word order in Yucatec Maya Table 5. XP-Verb-Argument-Argument clauses Interpretation VOS VSO

16/26 10/26

149

Percentage 61.53% 38.46%

2.4.4. Text count and comparison with Tzeltal Finally, we present evidence from a frequency count of the occurrences of SVO and VOS in four oral texts from Yucatec speakers.10 Like all text counts of this kind, ours is also subject to the limitation that the frequency of occurrence in texts of transitive clauses with two full DP arguments is fairly low. Still, we believe that our results are significant, especially when they are subject to cross-linguistic comparison, as we do below. The texts in our sample were segmented by clause (not sentence). After segmentation, we counted the total number of occurrences of SVO and VOS declarative clauses.11 We present our results in Table 6, where it can be seen that the frequency of occurrence of SVO is three times higher than the frequency of VOS. Table 6. Text Count in Yucatec oral texts 1 Order No. of occurrences

% (N=372)

SVO

9

2.41

VOS

3

0.80

However, our result of the total number of SVO clauses in Table 6 includes cases where the subject has been clearly topicalized (see Table 1). In order to distinguish these cases from cases where SVO cannot be attributed to topicalization, the number of occurrences of each was calculated separately. For this purpose, our criteria for identifying sentence topics were: (a) the fronted subject DP showed the topic clitic =e' or, (b) the subject DP was fronted outside the clause it originated in (i.e. either long topicalization had taken place, or the topic appeared to the left of an interrogative pronoun or a complementizer). When clauses with topicalized subjects are considered as a separate category, the results are clearly more complex. In fact, as shown in Table 7, they actually reflect the original observations of Durbin and Ojeda (1978) and Hofling (1984) in that Yucatec appears to have two unmarked word orders.

150

R. Gutiérrez-Bravo and J. Monforte y Madera

Table 7. Text Count in Yucatec oral texts 2 Order No. of occurrences

% (N=372)

S=TOPVO

6

1.61

SVO

3

0.80

VOS

3

0.80

In order to shed light into this situation, we compared our results with two word order text counts from Tzeltal Mayan, a Mayan language for which there is very strong evidence that its unmarked word order is VOS (Keenan (1978), Robinson (2002)). The results of these text counts are presented below, where it can be observed that the comparative frequencies of occurrence of SVO and VOS in Yucatec and Tzeltal are considerably different. Specifically, even allowing for subject-initial constructions as the result of subject topicalization (or focusing), VOS in Tzeltal has a higher frequency than SVO, contrary to what is observed in Yucatec. We conclude from this comparison that Yucatec is not showing the behavior characteristic of true VOS languages. Table 8. Tzeltal text count 1: Keenan (1978: 280) Order No. of occurrences

% (N=288)12

SVO

2

0.7

VOS

16

5.55

Table 9. Tzeltal text count 2: after Robinson (2002: 60-61) Order No. of occurrences

% (N=495)

SVO

32

6.46

VOS

67

13.53

Summing up, by themselves the results of our text count are not enough to conclude that Yucatec is an SVO language. But when we compare them with the behavior observed in a VOS language like Tzeltal, then our results point to the conclusion that the unmarked word order of Yucatec is not VOS. This, added to the evidence presented in earlier subsections, leads us to conclude that the unmarked word order of Yucatec transitive clauses is SVO.

On the nature of word order in Yucatec Maya

151

3. Split word order However, in sharp contrast with the results in Tables 6 and 7, Skopeteas and Verhoeven (2005) had already noted that VS is much more frequent than SV in intransitive clauses. We in fact observed from the analysis of our own corpus that this was true not only of intransitive clauses, but also of transitive clauses where the object is pro-dropped. In order to bring intransitive and object-drop clauses together for comparative purposes we henceforth refer to these clauses as “subject-only” clauses. As mentioned, we took clauses where there is no overt object to belong to this category. However, we do not include in this definition clauses that display a sentential complement (CP, IP) instead of a DP object. Neither did we consider clauses where the subject DP occupies the postverbal position when an interrogative or relative pronoun had been fronted, since Yucatec displays obligatory inversion in these cases. Examples of “subject-only” clauses are presented below. (22) T-u

tuukl-aj-ø

bey-o’

CP-ERG.3 think-PRF-ABS.3s thus-CL

le

‘The youngster reasoned it thus’. (23) Chan k DIM

janal

ERG.1.pl eat

xi’ipal-e’.

DM boy-CL (Gigante-14)

to’on. we

‘We are going to eat.’

(Don Pablo-675)

(24) Lekan lúub-uk-ø le ka’anal ja’… when fall-IRR-ABS.3s DM tall water ‘When the rain falls...’

(Cultivo-21)

The verb in Yucatec shows different aspectual and mood morphology depending on whether the verb is transitive or intransitive, so there is no doubt that (22) is a transitive object-drop clause, whereas (22) and (24) are intransitive clauses. The results of our text count of “subject-only” clauses are presented below: Table 10. Text count in Yucatec oral texts 3: “subject-only” clauses Order No. of occurrences % (N=372) SV

17

4.57

VS

58

15.59

152

R. Gutiérrez-Bravo and J. Monforte y Madera

Hence, on the one hand we have considerable evidence that the unmarked word order of transitive clauses in Yucatec is SVO. However, in the absence of an (overt) object, it is quite clear that the most frequent order is VS. We propose that this discrepancy between transitive declaratives and “subjectonly” declaratives be referred as split word order. Split word order should be kept distinct from the phenomenon observed in Spanish, German, Japanese, etc. whereby different kinds of clauses display different word orders depending on the semantic role of the arguments of the verb (Gutiérrez-Bravo 2007). In split word order the split/discrepancy in word order is instead (mostly) dependent on whether one or two arguments of the verb are overtly expressed. The split word order observed in Yucatec represents a considerable analytical challenge. If the SVO order is analyzed as the result of a strong EPP requirement in a functional projection above VP, as is typically the case in SVO languages, then the analysis incorrectly predicts that the strong EPP requirement will produce an SV order in “subject-only” clauses. On the other hand, an analysis that claims that Yucatec is a verb-initial language where SVO is the result of subject topicalization (as in the analysis of Greek and Spanish in Alexiadou and Anagnostopoulou 1998), is incompatible with the evidence presented so far that SVO is the unmarked word order of Yucatec transitive clauses. In the following section we argue that Optimality Theory, where constraints are ranked and violable, is a framework that is ideally suited to provide an analysis of this word order split.

4. An Optimality-theoretic analysis 4.1. Deriving SVO and VS The classic analysis of SVO languages in the OT literature is developed in Samek-Lodovici (1996), Grimshaw (1997), Costa (1998) and Costa (2001). In this analysis, the SVO order results from a constraint that requires [Spec, I] to be filled (henceforth EPP, following the definition in Gutiérrez-Bravo 2007) outranking the STAY constraint (Grimshaw 1997), which penalizes the presence of traces in the syntactic representation. (25) EPP The specifier of the highest inflectional projection must be filled.

On the nature of word order in Yucatec Maya

153

(26) STAY No traces This analysis assumes that the subject DP is generated in [Spec, V], and this is the position from where it moves directly to [Spec, I]. As stated before, following standard analyses of Mayan languages like Aissen (1996) and Aissen (1999a), we assume that [Spec, V] is projected to the right in Yucatec. These assumptions are schematized in (27). Lastly, recall that I0 in Yucatec transitive clauses is always filled by an ergative proclitic and (in most cases), an auxiliary. (27)

IP I' I

VP V' V

DP DP

(Subject)

(Object)

Here it is important to note that Yucatec also shows VSO clauses, although much less frequently than either SVO or VOS clauses. Our data shows that in VSO clauses the object is discourse-old. For instance, in (28) the object maaya ‘Mayan’ had been previously introduced in the discourse two clauses earlier. (28) K'abéet a kan-ik-ø necessary ERG.2 learn-IND-ABS.3sg ‘You need to learn Mayan.’

teech maaya. 2SG Mayan (Aurelia-3)

As such, we take VSO clauses be to cases where the object corresponds to the tail in the sense of Vallduví (1992). Hence, we assume the object is rightdislocated in these cases. With respect to the structure in (27), this means that the object is right-adjoined to IP (or perhaps VP). The result is a different structure altogether from that of VOS clauses, and so it is not targeted by the constraints introduced in what follows. Now, returning to the issue of split word order, if we analyze the SVO order of Yucatec transitive clauses as being indeed the result of the ranking

154

R. Gutiérrez-Bravo and J. Monforte y Madera

EPP » STAY, we get an incorrect result for “subject-only” clauses, as illustrated in (29). This is because under the ranking EPP » STAY, the winning candidate for subject-only clauses is the one displaying an SV order. (29) Subject only clauses = (22) INPUT: < think (x,y), x=the boy, y=pro > ³ a. [IP [DP le xi'ipale' ]i tu [VP tuklaj [DP pro ] beyo' hi ]] SV: the boy thought.it thus ) b. [IP tu [VP tuklaj [DP pro ] beyo' [DP le xi'ipale' ]]] VS: thought.it thus the boy

EPP

STAY *

*!

Now observe that the opposite ranking does not give us the correct result either. Following Costa (1998) and Costa (2001), we can take the unmarked word order of verb initial languages to be the result of the ranking STAY » EPP. As the reader can verify for himself, this ranking will give the correct result for subject-only clauses, but now we get an incorrect result for transitive clauses. This is because the STAY » EPP ranking incorrectly selects the verb-initial VOS candidate as the winner. This is shown in (30). (30)

Transitive clauses = (2) INPUT: < put (x,y), x=the woman, y=her good clothes>

³ a. [IP tu [VP ts'aj [DP u ma'alob nook'] [DP le ko'olelo' ]] VOS: put.on her good clothes the woman ) b. [IP [DP le ko'olelo']i tu [VP ts'aj [DP u ma'alob nook'] hi] SVO: the woman put.on her good clothes

STAY

EPP *

*!

Hence, by itself the interaction between STAY and EPP is insufficient to account for the word order split observed in Yucatec. We now propose that Yucatec indeed has the ranking STAY » EPP characteristic of verb-initial languages, but that STAY is in turn outranked by a well-formedness constraint that forces the subject to move outside the VP. Working on word order phenomena in Greek and Italian, Roussou and Tsimpli (2006) propose that two DPs cannot appear in the same syntactic domain if they share the same features (see also Mohanan (1994), Belletti (2004)). In their proposal, the syntactic domains in consideration are the lexical, inflectional and complementizer layers of the clause. They use this proposal to explain the absence of the VSO order in Italian. As illustrated in (31), from Roussou and Tsimpli (2006: 318), in this order both the subject

On the nature of word order in Yucatec Maya

155

and object DPs are found inside the lexical domain (which we label as VP below). Since in Italian these DP are identical in all of their features with overt morphological realization, this configuration is ruled out. In contrast, VSO is freely allowed in Greek because overt case morphology in this language makes the subject and object DPs formally distinct. (31) ITALIAN *Ha riparatoi [VP [DP Gianni ] hi [DP il mio computer]]. has repaired Gianni the mine computer We adopt Roussou and Tsimpli’s proposal in its essentials, with the following modifications. First, it seems to us that the definition of domain in their proposal can be simplified to specific maximal projections (VP, IP, CP) if unexploded versions of these projections are adopted (as we do in the OT analysis that follows). More importantly, we further propose that the crucial features that trigger the effect in (31) are category features, ([D], [N], [V], etc.), not general feature identity. 13 Finally, following recent work where it has been argued that constituents must have phonetic content in order to meet certain well-formedness conditions (Holmberg (2000), Landau (2007)), we propose that the effect in (31) requires that the XPs that share the same category feature have both phonetic content (see also Belletti 2004). Taking these considerations into account, we arrive at the following constraint, which we name DISTINCT. (32) DISTINCT Two phonetically overt phrases XP and YP with the same category feature [Z] should not be dominated by the same set of maximal projections. Crucially, any structure where the subject is located in [Spec, I] satisfies DISTINCT. This is because both argument DPs are dominated by IP, but only the object DP is dominated by VP. In contrast, when the subject DP remains in its VP-internal position (thus deriving a verb-initial order, as in the VOS schema in (27)), DISTINCT is violated. These distinctness effects were in fact reported in an investigation on word order in Yucatec carried out independently of this one in Skopeteas & Verhoeven (2008).14

156

R. Gutiérrez-Bravo and J. Monforte y Madera

(33)

IP I'

DPi (Subject)

I

VP V' V

hi DPi

(Object)

Now, having introduced the DISTINCT constraint it is possible to account for the word order split observed in Yucatec with the ranking DISTINCT » STAY » EPP. The analysis of “subject-only” clauses under this ranking is presented in (34). (34) “Subject-only” clauses = (22) INPUT: < think (x,y), x=the boy, y=pro > ) a. [IP tu [VP tuklaj pro beyo' le xi'ipale']] VS: thought.it thus the boy b. [IP le xi'ipale'i tu [VP tuklaj pro beyo' hi ]] SV: the boy thought.it thus

DIST

STAY

EPP *

*!

First observe that, even though the clause under consideration is transitive, both candidates satisfy DISTINCT. This is because the direct object is realized as a pro, which is not phonetically overt. Since both candidates satisfy the highest-ranked constraint, movement of the subject to [Spec, I] does not improve the structure in any respect. Hence it gratuitously violates STAY, and this makes the SV candidate (34) suboptimal when compared to the VS candidate (34). In this way, the VS order of “subjectonly” clauses is derived. However, a different situation results when both argument DPs are overt. In this case there is no default compliance of DISTINCT, and now this constraint plays a crucial role in selecting the winning candidate, as shown in tableau (35).

On the nature of word order in Yucatec Maya

(35)

157

Clauses with overt subject and object = (2)

INPUT: < put (x,y), x=the woman, y=her good clothes>

DIST

a. [IP tu [VP ts’aj [ u ma’alob nook’] [ le ko’olelo’]] VOS: put.on her good clothes the woman ) b. [IP le ko’olelo’i tu [VP ts’aj u ma’alob nook’ hi ] SVO: the woman put.on her good clothes

*!

STAY

EPP *

*

In this case, the verb initial candidate (35) fatally violates DISTINCT because both DPs are dominated by the same maximal projections (VP and IP). In contrast, the candidate where the subject surfaces in [Spec, I] satisfies DISTINCT, and so the SVO structure emerges as the winner. In this way, our analysis accounts for the word order split observed in Yucatec. Observe that our analysis further makes the prediction that when the complement of the verb is not a DP (but instead a PP or a CP), the unmarked word order should be verb-initial.15 Testing this prediction fully requires a corpus much larger than the one used for this paper. However, the data we do have does confirm this prediction. All else being equal, the subject is postverbal when the complement of the verb is not a DP, as shown in (36) and (37). (36) Ken k’uch-uk-ø [DP jun p’éel lu’umkab ] bey [PP te’ when arrive-IRR-ABS.3sg one NUMC mortal thus LOC t-u kaajtal ] bey-o’… DUR-ERG.3 dwell thus-CL ‘When a mortal arrives to where he (the giant) dwells...’ (Gigante-32) (37) K-u

taal

HAB-ERG.3 come

tun

teech [DP k’ank’ubul ][PP ti’ 2SG

jóok’-ol

DUR+ERG.3 exit-IND

u

plague

PREP

a

nal]

ERG.2 corn

y-i’ij-o’ob-o’.

ERG.3 EP-corn.bud-PLUR-CL

‘Plague comes (on you) to your corn as its tips are coming out.’ (Cultivo-75)

More importantly, the two instances in the corpus where the complement is not a DP and the subject is instead preverbal are clearly cases of topicalization of the subject DP. In (38), the preverbal subject bears the topic clitic =e’, and in (39) it appears to the left of the complementizer ka’, which indicates that it has been displaced to a topic position (Aissen 1992).

158

R. Gutiérrez-Bravo and J. Monforte y Madera

(38) [DP Lu’umkab-e’] chéen mix u y-ojel [CP bix only neither ERG.3 EP-know how mortal-TOP ko’ol-ik-ø-i’ ]. clear+PASS-FOC-ABS.3sg-LOC ‘The mortal didn’t even know how (the jungle) had been cleared.’ (Gigante-83)

(39) [DP Le chan lu’umkab Isaac ]-o’, ka’ j-k’uch-ø-i’ DM small mortal Isaac-CL COMP CP-arrive-ABS.3sg-LOC ka’ t-u y-e’es u baj [PP ti’ le rey-o’]. COMP CP-ERG.3 EP-show ERG.3 RFLX PREP DM king-CL ‘The little mortal Isaac, when he arrived there, he introduced himself to the king.’ (Gigante-35) It is now worth addressing two theoretical issues that make our optimality theoretic analysis preferable to possible alternatives that do not make use of OT. First, observe that DISTINCT only states that two XPs with the same category feature should not be dominated by the same set of maximal projections. As such, DISTINCT does not specifically derive movement of the subject DP to [Spec, I]. DISTINCT is equally satisfied if instead the object moves to [Spec, I], or if the subject moves to any position outside the VP. Our OT analysis accounts for the fact that neither option is observed. With respect to the possibility of the object being the DP displaced to [Spec, I], we adopt the OT analysis of unmarked word order developed in Gutiérrez-Bravo (2007). This analysis claims that the highest inflectional specifier of the clause is sensitive to the semantic role of the XP that occupies it. Specifically, structures where this specifier position is occupied by an agent are less marked (all else being equal) than clauses where a patient or theme occupies this position. As such, a structure where the object is displaced to [Spec, I] to comply with DISTINCT will always be more marked than a structure where the subject-agent is displaced to this position. As we leave for the reader to verify, an OVS candidate that satisfies DISTINCT will consequently always be eliminated because of its more severe violation of the relevant markedness constraints proposed in Gutiérrez-Bravo (2007). Secondly, we mentioned that in principle, the subject could move to any position outside the VP and still satisfy DISTINCT. However, this is clearly not what is observed. For instance, although particles, adverbs and some adverbial XPs can appear between the main verb in V and the auxiliary in I0

159

On the nature of word order in Yucatec Maya

in Yucatec, insertion of the subject DP in this intermediate position is robustly ungrammatical. (40) *[IP tu

[le

ko’olelo’]i [VP ts’-aj-ø u ma’alob put-PRF-ABS.3sg ERG.3 good

CP-ERG.3 DM woman-CL

nook’ hi ]]. clothes Our OT analysis provides a straightforward account of this fact. Even though the EPP is at the bottom of the ranking in Yucatec, it is still an active constraint (as in any OT analysis). The structure in (40) satisfies DISTINCT, but it violates EPP because it leaves the highest inflectional specifier empty. Because of this, it is bound to lose against the S Aux VO candidate that satisfies both DISTINCT and EPP. This is shown in the tableau in (41), where the losing candidate (41a) corresponds to ungrammatical (40). Observe that there is no straightforward way to achieve this result in an alternative analysis that makes use of DISTINCT, but where the EPP is not taken to be an active requirement in the language. (41) Active EPP INPUT: < put (x,y), x=the woman, y=her good clothes> a. [IP tu [VP [le ko’olelo’]i [VP ts’aj [u ma’alob AUX the woman put.on her good nook’] hi ]]] clothes ) b. [IP le ko’olelo’i tu [VP ts’aj [u ma’alob nook’] the woman AUX put.on her good clothes hi]]

DIST

STAY

EPP

*

*!

*

Finally, DISTINCT is itself a violable constraint, and consequently we expect its effects not to be observed when it is dominated by another constraint. We suggest that this is the case of true VOS languages like Tzeltal. Specifically, VOS as the unmarked word order is derived with the ranking STAY » DISTINCT, EPP. This is shown for the example in (42) in the tableau in (43).

160

R. Gutiérrez-Bravo and J. Monforte y Madera

(42) TZELTAL (Stross 1978:27; taken from Robinson 2002:61) La laj [VP s-ta-ø [NP alchaxiltik] [NP te winik-e]]. CP HS 3ERG-find-3ABS orange.orchard ART man-CL ‘… the man found an orange orchard.’ (43) TZELTAL INPUT:

STAY

a. [IP te winikei la laj [VP sta alchaxiltik hi ]] SVO: the man found orange.orchard ) b. [IP la laj [VP sta alchaxiltik te winike]] VOS: found orange.orchard the man

*!

DIST

EPP

*

*

In this way, our proposal is compatible with the analysis of verb-initial languages in Costa (1998) and Costa (2001), where it is proposed that what is characteristic of these languages is the very high ranking of STAY relative to the EPP.

4.2. Back to VOS clauses It now remains to be explained what derives VOS clauses like (1). In this final section, we briefly summarize the analysis of Yucatec VOS clauses developed in Gutiérrez-Bravo and Monforte (2008). Following the preceding sections of this paper, we conclude that VOS is a derived order, and specifically, we now propose that VOS is the order that corresponds to thetic judgments. In this we follow Kuroda (1972), Aissen (1999a), and Ladusaw (1994) in assuming a fundamental semantic difference between categorical and thetic judgments. Categorical judgments establish a predication relation between an argument (typically the subject) and the rest of the proposition. Categorical judgments are composed of two parts: a referent of which something is predicated (the logical subject), and the property that is ascribed to this referent. Following Aissen (1999a), we consider subject-initial constructions to prototypically instantiate this kind of judgment. Specifically, in an SVO clause like (2), repeated here as (44), the preverbal DP is not only the grammatical subject of the clause. It is also the logical subject, the entity of which the rest of the clause is predicated. Hence subject-initial clauses can be taken as prototypical instances of categorical judgements.

On the nature of word order in Yucatec Maya

161

(44) Le

ko’olel-o’ t-u ts’-aj-ø u ma’alob nook’. DM woman-CL CP-ERG.3 put-PRF-ABS.3sg ERG.3 good clothes ‘...and the woman put on her good clothes...’ (Si’ipil-51)

In contrast with categorical judgments, thetic judgments present an event in a unitary way. The event is just presented as the simple perception of the situation, without highlighting any of the participants and without establishing the logical subject-predicate configuration of categorical judgments. Again, following Aissen (1999a), we take verb-initial constructions to be a typical instantiation of thetic judgments. Specifically, in a VOS clause like (1), repeated here as (45), there is no logical subject. In other words, in contrast with (44), there is no unique constituent of which the rest of the clause is predicated. What we mean by this is that there is no single constituent that the clause can be claimed to be about. Instead it appears that the event is merely being presented as the simple perception of the situation. (45) Je’

bin

k-u

la’ach-ik-ø u jo’ol ASV CIT HAB-ERG.3 scratch-IND-ABS.3sg ERG.3 head x-nuk reyna-o’. FEM-great queen-CL ‘And the great queen truly scratched her head...’ (Gigante-97)

It does not seem to us that this particular state of affairs is specific to Yucatec. As discussed in Gutiérrez-Bravo and Monforte (2008), we believe that this same phenomenon is what is observed in a well-studied kind of word order alternation, namely, Transitive Expletive Constructions (TECs). TECs show a word order alternation that is very similar to the one observed in Yucatec (see Bobaljik and Jonas 1996, Alexiadou and Anagnostopoulou 1998). This is illustrated with examples from Icelandic (an SVO language) in (46), from Alexiadou and Anagnostopoulou 1998: 492). (46) ICELANDIC a. Einhverjir stúdentar lasu bókina. some students read the book b. Þaÿ lasu [ einhverjir stúdentar ] [ bókina ]. EXPL read some students the book ‘Some students read the book.’

SVO Exp-VSO

162

R. Gutiérrez-Bravo and J. Monforte y Madera

Example (43b) is not strictly speaking a verb-initial construction. However, as argued by Alexiadou and Anagnostopoulou, this is merely a parametric difference, relative to the need to satisfy the EPP with a full XP. Greek, for instance, shows essentially the same SVO/VSO alternation, but no expletives. Following this same reasoning, it is feasible to equate Icelandic TECs with Yucatec VOS clauses, given that the expletive adds no semantic content to the proposition. Observe that, just as in the case of the Yucatec SVO/VOS alternation, the word order alternation in Icelandic TECs does not bring with it any truth-conditional difference. 16 Now, as proposed in Gutiérrez-Bravo and Monforte (2008), TECs and VOS clauses in Yucatec appear to share a fundamental property. Specifically, in both constructions the subject appears in the post-verbal field and there is no target of predication whatsoever. That is, both Yucatec VOS clauses and TECs have no constituent that the clause can be understood to be about. Instead they simply present the content of the proposition in a unitary way. In this way, our claim that VOS in Yucatec is a marked order coexisting with unmarked SVO is really no different from the situation observed in Icelandic in (46). In order to integrate the distinction between categorical and thetic judgments into the OT analysis developed so far, we propose that the thetic nature of a proposition is established in the semantic information of the input (Gutiérrez-Bravo 2005). Hence, transitive inputs like (35) can further be specified with a [thetic] feature ([TH]): (47) INPUT: < [TH] scratch (x,y), x=the great queen, y=her head > The presence of the [thetic] feature in the input implies that the syntactic output should be interpreted as a thetic judgment and not as a logical subjectpredicate configuration. Since subject-initial constructions are typically interpreted as logical subject-predicate configurations, we propose that the manifestation of the [TH] feature in output syntax is regulated by the following constraint.17 (48) THETIC-CONDITION (THET-CON) If a proposition bears the [thetic] feature in the input, then the highest inflectional head asymmetrically c-commands all the arguments of the lexical head of the clause. Observe that this restriction is satisfied by a verb-initial clause like (1), since I0 asymmetrically c-commands both of the arguments of the verb. In

163

On the nature of word order in Yucatec Maya

contrast, it is violated when the subject occupies the [Spec, I] position (as in SVO clauses) since in this case I0 does not c-command the subject. Because of this THET-CON, necessarily conflicts with the requirements of DISTINCT, which forces movement of the subject to [Spec, I] when an object is also present.18 We now account for the word order alternation observed in Yucatec transitive clauses through the interaction of these constraints. When THET-CON outranks DISTINCT, if the input bears the [thetic] feature, the requirement that there should be no argument that is not c-commanded by I0 has priority over the requirement that there shouldn’t be two DPs dominated by the same set of maximal projections. In this case, in contrast with what is observed in (35), it is the verb initial candidate that emerges as the winner. (49) Yucatec: thetic judgments/VOS clauses INPUT: a. [IP x-nuk reynao’ k-u [VP la’achik the great queen scratched u jo’ol ]] her head SVO ) b. [IP __ k-u [VP la’achik u jo’ol scratched her head x-nuk reynao’]] the great queen VOS

THETCON

DIST

STAY

EPP

*

*!

*

*

In this way we account for the presence of VOS clauses in Yucatec even when the unmarked word order of its transitive clauses is SVO.19

5. Conclusions In this paper, we have provided evidence that the unmarked word order of transitive clauses in Yucatec Maya is SVO, contrary to what is often assumed in the literature. We applied four tests to determine whether the unmarked word order of Yucatec is SVO or VOS (answers to out-of-the blue questions, definiteness effects in argument-verb-argument clauses, definiteness effects in verb-argument-argument clauses, and interpretation of XP-verb-definite.NP-definite.NP clauses). The results of none of these tests points to the conclusion that the unmarked word order of Yucatec transitive

164

R. Gutiérrez-Bravo and J. Monforte y Madera

clauses is VOS. In contrast, the first two provide strong evidence that the unmarked word order of Yucatec is SVO. We also carried out a text count to compare the relative frequency of occurrence of SVO and VOS clauses. Although our frequency results in text counts shed no light on whether the basic word order of Yucatec is SVO or VOS, a comparison of these results with Tzeltal indicates that Yucatec does not behave like a VOS language. A frequency count of “subject-only” clauses, however, indicates that these clauses are by and large verb-initial. We have proposed the descriptive term split word order to refer to this discrepancy between the two different kinds of clauses considered. We then developed an Optimality Theoretic analysis to account for this word order discrepancy. We first showed that Yucatec is different from other SVO languages in that its unmarked word order cannot be accounted for as the result of the ranking EPP » STAY. Instead we proposed that the verb-initial nature of “subject-only” clauses in Yucatec results from the ranking STAY » EPP. The SVO order, we argued, is the result of an altogether different constraint outranking STAY in this ranking. We proposed that this constraint is the DISTINCT constraint, which states that two phonetically overt XPs with the same category feature [Z] should not be dominated by the same set of maximal projections. The resulting analysis, which crucially relies on the violable nature of the syntactic constraints involved, provides a straightforward account of the split word order observed in Yucatec.

Notes *

We would like to thank the Academia de la Lengua Maya de Yucatán for the help and support provided throughout this investigation. Many thanks also to Santiago Abam, Lázaro Dzul, Mario Manrique, Pascual Mukul, Luciano Tah, and Bernardino Yaj for their judgments on the Yucatec data presented here. Also, we would like to thank an anonymous reviewer and the audiences at the 2008 Annual Meeting of SSILA and the Cuarto Encuentro de Teoría de Optimidad for their feedback; all errors that remain are our own. This project was supported in part by the National Council for Science and Technology of Mexico (CONACYT), grant SEP-2004-CO1-47613. 1. All examples are presented according to the orthographic conventions of the Academia de la Lengua Maya de Yucatán and so they do not necessarily reflect their phonetic form accurately. The abbreviations we use in the examples in this paper are the following:

On the nature of word order in Yucatec Maya ABS ART ASV CAUS CIT CL CP DM

absolutive definite article assurative causative reportative clitic completive demonstrative

DUR EP ERG FEM FOC HAB HS IND

durative epenthesis ergative feminine (biological) focus habitual hearsay indicative

165

IRR irrealis LOC locative NUMC classifier PASS passive PRF perfect TOP topic TRNS transitive

2. In Gutiérrez-Bravo and Monforte (2008) we presented a comparison between the corpus counts made for Yucatec in Skopeteas and Verhoeven (2005) and the text count made for Tzeltal in Keenan (1978). There we argued that the results of this comparison were problematic for the analysis of VOS as unmarked in Yucatec. In section 2.4.4 of this paper we make use of the same argument, but we now resort to our own corpus count. We include an additional comparison with a second word order frequency count in Tzeltal, from Robinson (2002). 3. In terms or traditional Mayan linguistics, these proclitics correspond to the Series A pronouns. Although glossed as ERG here, it should be noted that ergativity in Yucatec is split on the basis of aspect. Hence not every instantiation of these pronouns corresponds to an ergative function. 4. These correspond to the Series B pronouns of traditional Mayan linguistics. Again, because of split ergativity, these suffixes do not always display an absolutive function. 5. Specifically, the third person proclitic u and the completive aspect particle t- in (3), respectively. Agent focus phenomena in Mayan languages is most often classified depending on whether the agent focus form of the verb agrees with the subject or the object. In Yucatec, however, the verb agrees systematically with the object, and subject agreement is only observed in intransitive constructions in completive aspect or irrealis mood (which follow an ergative-absolutive pattern). Hence, perhaps the most accurate description of the agent focus construction in (6) is that it is a one-place predicate displaying transitive morphology, which further always agrees with the object, and in which the proclitic that cross-references the subject is always absent. See Gutiérrez-Bravo and Monforte (2009). 6. The clitic –o’ attached to the right of the arguments of the verb in these examples is a deictic clitic that typically expresses the relative distance between the speaker and the referent of each DP; this particular clitic can be approximately translated into English as the demonstrative that. Yucatec does not allow more than one of these clitics to appear simultaneously in the postverbal field. Hence in (11) only the second argument DP bears this clitic. 7. It is important to note that, under the correct pragmatic conditions, OVS is a perfectly acceptable (albeit marked) order in Yucatec. These pragmatic

166

R. Gutiérrez-Bravo and J. Monforte y Madera

conditions, however, are not relevant for our study, which concentrates solely on the language’s unmarked word order. 8. Observe that in VOS analyses of Yucatec this second DP thus occupies the canonical subject position and hence is expected not to be subject to definiteness effects, contrary to what our results indicate. 9. In this paper we do not attempt to provide an analysis of the definiteness effects shown in Table 4. This is because VSO is a marked order in Yucatec (see section 4.1) and the scope of this paper is restricted to the analysis of unmarked word order. However, it seems to us that a possible analysis of these definiteness effects can be developed in which definiteness is aligned with the subject grammatical relation, along the lines of the OT analysis in Aissen (1999b). 10. All these texts correspond to the “eastern” variety of Yucatec, spoken in the state of Quintana Roo and the southern tip of the state of Yucatan. 11. We restricted ourselves to declarative clauses since wh-interrogatives and relatives trigger obligatory subject inversion in Yucatec, hence forcing a VS order. We do not know if this variable was controlled in earlier texts counts of Yucatec or the in the Tzeltal text counts we discuss later. 12. In Keenan's (1978) and Robinson's (2002) text counts, N corresponds to the number of sentences, not of clauses. In this respect their text counts are different from ours. 13. Observe that we are not claiming that identical category features are the only features that trigger the effect in (31). What we propose is that identical category features are a necessary condition for this effect to occur in the first place. We will see that in Yucatec it is also a sufficient condition for this effect to occur. Our sense, however, is that in other languages (including other Mayan languages) this might not be by itself a sufficient condition. It is also fairly clear from our data that full NPs and pronouns in Yucatec do not belong to the same category for the purposes of the word order effects reported here. Further research is necessary to clarify this point. 14. The conclusions that Skopeteas & Verhoeven (2008) arrive at with respect to the unmarked word order of transitive clauses in Yucatec are quite different from ours, though. See also Skopeteas & Verhoeven (2009). 15. We are thankful to an anonymous reviewer for bringing this point to our attention. 16. TECs are subject to the requirement that the object be indefinite, as it is well known. In this respect it seems to us to be very meaningful that various kinds of different word orders in Mayan languages are subject to definiteness restrictions (see England 1991 for a survey). We acknowledge the importance of studying this particular property in VOS clauses in Yucatec, but addressing this specific issue is beyond the scope of this paper. 17. See Gutiérrez-Bravo (2005) for an earlier formulation.

On the nature of word order in Yucatec Maya

167

18. THET-CON also necessarily conflicts with the EPP constraint (Gutiérrez-Bravo 2005, Gutiérrez-Bravo and Monforte 2008), but this is not relevant in the case of Yucatec because of the low ranking of EPP in this language. 19. A reviewer asks whether it would be possible instead to account for the SVO/VOS alternation with a stochastic OT analysis. At this point we cannot claim for certain whether such an analysis would be desirable or not. However, there is a reason why we have not attempted to develop this kind of analysis. In a stochastic analysis, the same kinds of clauses are predicted to freely alternate between the SVO and VOS orders. However, we have reason to believe that this is not what is observed in Yucatec. Our data indicate that there is a marked tendency for either the subject or the object to be a bare NP in VOS clauses, but not in SVO clauses. In principle, this could be the result of a semantic or syntactic restriction on VOS clauses that does not apply in SVO clauses, which would make the stochastic analysis undesirable. A full confirmation of this observation, however, requires a much more detailed corpus study than the one we have developed in this paper.

References Aissen, Judith 1992 Topic and Focus in Mayan. Language 68: 43-80. Aissen, Judith 1996 Pied-piping, abstract agreement and functional projections in Tzotztil. Natural Language and Linguistic Theory 14: 447-491. Aissen, Judith 1999a External possessor and logical subject in Tz'utujil. In External Possession, Doris L. Payne and Immanuel Barshi (eds.), 167-193. Amsterdam/Philadelphia: John Benjamins. Aissen, Judith 1999b Markedness and subject choice in Optimality Theory. Natural Language and Linguistic Theory 17: 673-711. Alexiadou, Artemis, and Elena Anagnostopoulou 1998 Parametrizing AGR: word order, V-movement and EPP-checking. Natural Language and Linguistic Theory 16: 491-539. Belletti, Adriana 2004 Aspects of the Low IP Area. In The Structure of CP and IP, Luigi Rizzi (ed.), 16-51. New York: Oxford University Press. Bobaljik, Jonathan, and Dianne Jonas 1996 Subject positions and the roles of TP. Linguistic Inquiry 27: 195-236. Bohnemeyer, Jürgen 2002 The Grammar of Time Reference in Yukatek Maya. Munich: Lincom Europa.

168

R. Gutiérrez-Bravo and J. Monforte y Madera

Briceño, Fidencio 2002 Topicalización, enfoque, énfasis y adelantamiento en el maya yukateco. In La organización social entre los mayas prehispánicos, coloniales y modernos, V. Tiesler Blos, R. Cobos and M. Greene Robertson (eds.), 374-387. Mexico City/Mérida: INAH/UADY. Bricker, Victoria 1978 Wh-questions, relativization and clefting in Yucatec Maya. In Papers in Mayan Linguistics, Laura Martin (ed.), 109-139. Columbia, Missouri: Lucas Brothers. Contreras, Heles 1976 A Theory of Word Order with Special Reference to Spanish. Amsterdam: North Holland. Costa, João 1998 Word order variation: A constraint-based approach, University of Leiden: Ph.D. Dissertation. Costa, João 2001 The emergence of unmarked word order. In Optimality-theoretic Syntax, G. Legendre, J. Grimshaw and S. Vikner (eds.), 171-204. Cambridge, Mass.: MIT Press. Dayley, Jon P. 1985 Tzutujil Grammar . Berkeley, CA: University of California Press. Durbin, Marshall, and Fernando Ojeda 1978 Basic word order in Yucatec Maya. In Papers in Mayan Linguistics, Nora England (ed.), 69-77. Columbia: University of Missouri. England, Nora 1991 Changes in basic word order in Mayan languages. International Journal of American Linguistics 57: 446-486. Góngora, María Luisa 1990 U Tzikbalilo’ob Oxkutzcaab yéetel Maní: Mérida, Maldonado Editores. Grimshaw, Jane 1997 Projection, heads and optimality. Linguistic Inquiry 28: 373-422. Gutiérrez-Bravo, Rodrigo 2005 Structural Markedness and Syntactic Structure. New York: Routledge/Taylor and Francis. Gutiérrez-Bravo, Rodrigo 2007 Prominence scales and unmarked word order in Spanish. Natural Language and Linguistic Theory 25: 235-271. Gutiérrez-Bravo, Rodrigo 2008 La identificación de los tópicos y los focos. Nueva Revista de Filología Hispánica 56: 363-401.

On the nature of word order in Yucatec Maya

169

Gutiérrez-Bravo, Rodrigo, and Jorge Monforte 2008 La alternancia sujeto-inicial/verbo-inicial y la Teoría de Optimidad. In Teoría de Optimidad: Estudios de Sintaxis y Fonología, R. GutiérrezBravo and E. Herrera (eds.), 61-99. Mexico City: El Colegio de México. Gutiérrez-Bravo, Rodrigo, and Jorge Monforte 2009 Focus, agent focus and relative clauses in Yucatec Maya. In New Perspectives in Mayan Linguistics, H. Avelino, J. Coon and E. Norcliffe (eds.), 83-96. Cambridge, Mass.: MIT Working Papers in Linguistics. Hofling, Charles 1984 On proto-Yucatecan word order. Journal of Mayan Linguistics 2: 35-64. Holmberg, Anders 2000 Scandinavian stylistic fronting: How any category can become an expletive. Linguistic Inquiry 31: 445-483. Keenan, Edward 1978 The syntax of subject-final languages. In Syntactic Typology, Winfred Lehmann (ed.), 267-327. Austin: University of Texas Press. Kuroda, S. Yuki 1972 The categorical and the thetic judgment. Foundations of Language 9: 153-185. Ladusaw, William 1994 Thetic and categorical, state and individual, weak and strong. In Proceedings of the Conference on Semantics and Linguistic Theory 4, 220-229. Ithaca, New York: Cornell University. Lambrecht, Knud 1994 Information Structure and Sentence Form. Cambridge: Cambridge University Press. Landau, Idan 2007 EPP Extensions. Linguistic Inquiry 38: 485-524. Mohanan, Tara 1994 Case OCP: A constraint on word order in Hindi In Theoretical Perspectives on Word Order in South Asian languages, M. Butt, T.H. King and G. Ramchand (eds.), 185-216. Stanford: CSLI. Robinson, Stuart 2002 Constituent order in Tenejapa Tzeltal. International Journal of American Linguistics 68: 51-80. Roussou, Anna, and Ianthi-Maria Tsimpli 2006 On Greek VSO again! Journal of Linguistics 42: 317-354. Samek-Lodovici, Vieri 1996 Constraints on Subjects: An Optimality Theoretic Analysis: Ph.D. Dissertation, Rutgers University.

170

R. Gutiérrez-Bravo and J. Monforte y Madera

Skopeteas, Stavros, and Elisabeth Verhoeven 2005 Postverbal argument order in Yucatec Maya. Sprachtypologie und Universalienforschung 58: 347-373. Skopeteas, Stavros, and Elisabeth Verhoeven 2008 Topic constructions in Yucatec Maya. Paper presented at the 2008 annual meeting of SSILA, Chicago. Skopeteas, Stavros, and Elisabeth Verhoeven 2009 Distinctness effects on VOS order: Evidence from Yucatec Maya. In New Perspectives in Mayan Linguistics, H. Avelino, J. Coon and E. Norcliffe (eds.), 157-174. Cambridge, Mass.: MIT Working Papers in Linguistics. Stiebels, Barbara 2006 Agent focus in Mayan languages. Natural Language and Linguistic Theory 24: 501-570. Stross, Brian 1978 Tzeltal Tales of Demons and Monsters. Museum Brief 24. Columbia: Museum of Anthropology of Missouri. Tonhauser, Judith 2003 F-constructions in Yucatec Maya. In The Proceedings of SULA 2, Jan Anderssen, Paula Menéndez-Benito and Adam Werle (eds.), 203-223. Amherst: GLSA, University of Massachusetts. Vallduví, Enric 1992 The Informational Component. New York/London: Garland.

The structure of CP in Karaja Marcus Maia

1. Introduction This paper intends to describe and analyze the left periphery of the clause in Karaja, a Macro-Je stock language spoken by about 3,000 people on and around the Bananal Island in Central Brazil. We will analyze interrogative constructions of the Wh and yes/no types, topic and focus constructions, as well as constructions with the txibo ‘whether’ conditional operator and with the subordinator morpheme –my in order to propose an integrated account for those structures. We also make the proposal that the evidential particle ràki described in Maia (2007) cannot be adequately cartographed in a Mood or Force projection position in the functional hierarchy, but should be considered as a “parasitic” category, following Matthewson, Davis & Rullman (2007). This study elaborates on previous analyses of the Karaja language presented in Maia (1998), Maia, Salanova and Lanes (2000), Maia (2001) and Maia (2007). We follow the Minimalist framework of Chomsky & Lasnik (1993), Chomsky (1995), and particularly, Rizzi (1997) in which an expanded CP system is proposed. We review our previous work on the structure of the complementizer system in order to incorporate the analysis of the conditional operator txibo ‘whether’ which , as we will show, occupies the same position as the interrogative words in the lowest projection of the CP system, capturing a dependence relation between this system and the inflectional specifications of the verbal system. In this sense, we also analyze embedding constructions which display basically the same dependence properties between the CP and the clause inflectional system. The paper is organized as follows. First, a brief introduction to the relevant theoretical topics is presented. Then, we describe interrogative wh constructions, yes/no interrogative structures, as well as their interaction with topic, focus, condition and embedding structures in Karaja. The hearsay particle ràki is then examined. Finally, we present our proposal to account for these facts, considering the internal structure of wh-phrases in the language. The data base gathered for the study includes not only the transcription of sentences in spontaneous narratives and questionnaires, but also the elicitation of grammaticality judgments.

172

Marcus Maia

2. The theoretical framework According to Rizzi (1997), the representation of clauses in Universal Grammar consists of three layers, each of these associated with a specific type of information, as illustrated in (1): (1)

CP IP VP

The VP system is the lexical layer, in which the verbal head assigns theta role to its arguments; the IP system is the inflectional layer, made up by the different functional heads which license morphological features such as case and agreement; the CP system is the layer whose head is typically a free functional morpheme, the complementizer. The CP system also hosts topics and operators such as question and focus markers, evidentials, deictic particles and other elements usually related to the information structure of the sentence. According to Rizzi, the CP system is made up by the articulate arrangement of X-bar projections, as illustrated in (2). (2)

Force P Force' Force

TopP Top' Top˚

Focus P Focus' Focus˚

TopP Top' Top˚

Fin P Fin' Fin˚

IP

The structure of CP in Karaja

173

Rizzi (1997) offers evidence in favor of the expansion of the CP system in the spirit of Pollock’s 1989 analysis of verb movement in which the IP system is expanded in a series of functional projections. Rizzi’s proposal is that the role of the CP system is to provide an interface between a propositional content which is expressed by the IP and the superordinate structure, a higher clause or the articulation of discourse. Therefore, in the two ends of the configuration in (2), Rizzi represents the Force Phrase and the Finiteness Phrase, which express information on the clause type and the dependences between the CP and the IP, respectively. The Force Phrase encodes information such as the interrogative, declarative or exclamative nature of a clause that marks it to be selected by a higher selector. The Finiteness Phrase captures the properties of IP which are replicated in the CP system, such as the relation between the complementizer and the finite or non-finite nature of a predicate. Between these two phrases, Rizzi represents the topic and focus systems, reviewing a series of differences between the two types of constructions which typically involve the left periphery of the clause. A difference which is directly relevant to the purposes of this paper is the distinction between focus and topic which is established on the basis of the correlation with a resumptive clitic. According to Rizzi, only the topic construction allows the resumptive clitic, which is inconsistent with the focalized constituent. In line with the cartographic project, Cinque (1999) and Speas (2004) claim that elements such as evidentials can also occupy a fixed functional position in the universal architecture of the sentence. Matthewson, Davis & Rullman (2007), on the other hand, offer evidence that this single head approach may not be suitable for evidentials, which are analyzed as a parasitic category. The main focus of the present paper is the internal make up of Karaja whquestion words and their position in the CP system. In the Principles and Parameters framework, wh-questions are interrogative structures which involve a wh-phrase which can be non-referential or referential. Wh-phrases move from their base position to a position in the left periphery of the clause, the Spec, CP position. From this site, they can bind their trace in the extraction site. According to minimalist requirements (cf. Chomsky, 1995), syntactic movement is only possible as a last resort in order to satisfy morphological requirements. Morphological features can thus be parameterized as strong or weak. Strong features must be checked via overt syntactic movement, that is, they must raise to the appropriate pre-terminal node, where a complex of abstract features match them, granting the so called convergence of the derivation. If strong features are not checked in

174

Marcus Maia

overt syntax, the derivation will crash since the strong features cannot be interpreted at the Phonetic Form (PF) level. On the other hand, weak features can procrastinate their checking to the level of Logical Form (LF). Chomsky (1995) analyzes the phrase ‘whose book’ as represented in (3): (3)

DP who

D’ ‘s

book

Interrogative words are analyzed as possessing an abstract wh-feature and an abstract element underlying indefinite pronouns. If the interrogative Q feature instantiated in the CP is strong, such as in English, the wh-feature must raise to check the Q feature in the CP. If the wh-feature raises alone, leaving behind the referential expression that accompanies it, the result would be the crash of the derivation at PF. Therefore the whole phrase must raise in overt syntax. That is why in (3) not only the interrogative word is raised but also the residue ’s book. It is important to note, though, that only the wh-feature needs to be raised. The rest of the phrase is automatically dragged along in the operation that became known as ‘pied-piping’. As we intend to show below, the Karaja language offers interesting empirical material to be checked against the fragment of Universal Grammar reviewed above. The non-fusional nature of Karaja morphology allows a clear-cut segmentation of the components of the wh-word. Unlike English, in which the morpheme segmentation in (4) discussed in Tsai (1994), though intuitively interesting, has a fusional nature, the Karaja data allow us to entertain the theoretically interesting possibility that the wh-feature is the head of its phrase. (4) wh-words Wh+at Wh+ere Wh+en

pronominals th+at th+ere th+en

3. Interrogative words in Karaja Karaja interrogative words are invariably formed by the composition of one or more indefinite roots with the wh-feature –bo. The data in (5) describe the basic constitution of wh or bo-words in Karaja.

The structure of CP in Karaja

175

(5) a. Aõ + bo Thing wh ‘What’ b. Mo + bo Person wh ‘Who’ c. Ti + wàse + na + bo Equal nominalizer wh ‘Which’ d. Ti + ki + bo in wh ‘Where’ e. Ti + u + bo time wh ‘When’ f. Ti + my + bo to wh ‘How’ g. Ti + wàse + bo equal wh ‘How many’ h. Aõ + he + re + ki + bo thing emphatic in wh ‘Why’ Let it be established from the outset that wh-interrogatives in Karaja are derived via overt syntactic movement. Note that (6) is a declarative sentence displaying an SOV word order, whereas (7) is an interrogative construction with the wh-word aõbo left dislocated. (8) demonstrates that a grammaticality judgment contrast obtains if the interrogative word remains in situ. (6)

Wataju iheto riwinyreri. Wataju 3-house is making ‘Wataju is making his house.’

(7)

Aõbo Wataju riwinyreri? What Wataju is making ‘What is Wataju making?’

176

Marcus Maia

(8)

?Wataju aõbo riwinyreri? Wataju what is making ‘Wataju what is making?’

Sentences (9), (10) and (11) demonstrate that adjuncts, as well as arguments, must also move to the left periphery of the clause in Karaja: (9)

Kaiboho mawaxinybenykre biurasòmy. you (PL) will fish tomorrow ‘You will fish tomorrow.’

(10) Tiubo kaiboho mawaxinybenykre? when you (pl.) will fish ‘When will you fish?’ (11) ?Kaiboho mawaxinybenykre tiubo? You (PL) will fish when ‘When will you fish?’ As expected for syntactic movement operations, restrictions concerning the locality of movement are operative in Karaja, as demonstrated by the ungrammaticality of the sentence in (12): (12) *Mo-my-bo Arirama a-ko relyyre Person-ACC-wh Arirama you-to said tiubo tii itxirearemy Brasilia-ki? when he met Brasilia-in? ‘Who Arirama said to you when he met in Brasilia?’ Sentence (12) demonstrates that the overt extraction of the wh-phrase momybo from the adjunct clause is ungrammatical in Karaja, indicating that this language is sensitive to the Subjacency Condition (Chomsky, 1977). Note that sentence (12) also offers an example of an interesting grammatical pattern: the interrogative word mobo ‘who’ displays inside it the accusative marker -my. We will turn immediately to the description of this fact. Argumental interrogative words receive postpositional particles in their indefinite component, as exemplified below: (13) Kai waha-my tabita. You my father-ACC saw ‘You saw my father.’

The structure of CP in Karaja

177

(14) Mo-my-bo kai tabita? Person-ACC-wh you saw? ‘Who did you see?’ (15) Aõ-my-bo kai tabita? thing-ACC-wh you saw ‘What did you see?’ Note that some verbs in Karaja mark their direct objects with the accusative suffix -my, as exemplified in (13). Examples (14) and (15) demonstrate that the accusative morpheme is also postposed to the indefinite roots mo ‘person’ and aõ ‘thing’. Besides the particle -my, other postpositions can also occupy this internal position either inside mobo ‘who’, or inside aõbo ‘what’, in Karaja, as exemplified in (16), (17), (18) and (19): (16) Mo-wyna-bo kai tohonyte kau? Person-and-wh you left yesterday ‘With whom did you leave yesterday?’ (17) Mo-dee-bo tii kua wyhy riwahinyra? Person-BEN-wh he that arrow gave ‘To whom did he give that arrow?’ (18) Mo-ràbi-bo kai kaa may temyta? Person-from-wh you this knife grabbed ‘From whom did you grab this knife?’ (19) Aõ-di-bo juwata temyta? Thing-INST-wh piranha caught ‘With what did you catch the piranha?’ Note also that in the referential wh-phrases, that is, those phrases in which a quantified nominal element occurs, this nominal also occurs in the same infixed position, as exemplified in (20) and (21): (20) Aõ-utura-bo kai temyta? Thing-fish-wh you caught ‘Which fish did you catch?’

178

Marcus Maia

(21) Mõ-utura-bo kaa rare? Person-fish-wh this is ‘Whose fish is this?’ If the quantified NP is the internal argument of a verb that requires the accusative marker or any other postpositional particle, both the nominal and the particle will be infixed inside the interrogative word, as illustrated by examples (22) and (23): (22) Aõ-ijyy-my-bo kai telyyta kau? Thing-story-ACC-wh you told yesterday ‘Which story did you tell yesterday?’ (23) Mõ-hawyy-dee-bo kai may tewahinyta? Person-woman-BEN-wh you knife gave? ‘To which woman did you give the knife?’

4. Yes/no questions, topic and focus constructions in Karaja Before we present our analysis of the interrogative words described above, we will discuss some constructions which are also related to the left periphery of the clause, namely, questions of the yes/no type, as well as topic, focus and conditional structures. Yes/no questions: Note that the Karaja wh-word aõbo is also used as an interrogative operator in order to form interrogative constructions of the yes/no type. In this case, aõbo occurs consistently in the second constituent position in the clause, as exemplified by the contrast between the declarative sentence in (24) and the yes/no interrogative in (25): (24) A-biòwa orera-my robira ahu-ki. Your friend alligator-ACC saw lake-in ‘Your friend saw the alligator in the lake.’ (25) A-biòwa aõbo orera-my robira ahu-ki? Your friend Q alligator-ACC saw lake-in ‘Did your friend see the alligator in the lake?’

The structure of CP in Karaja

179

Topic and Focus constructions: Topic constructions in Karaja are formed by the fronting of the topicalized NP to a position in the left periphery of the clause, as exemplified by the contrast between (26) and (27): (26) Isè kua ijadoma-my robira hawa-ki. Her mother that girl-ACC saw village-in ‘Her mother saw that girl in the village.’ (27) Kua ijadoma-my, isè tuu robira hawa-ki. that girl-ACC her mother her saw village-in ‘That girl, her mother saw her in the village.’ Note that the construction in (27) must obligatorily include the third person resumptive tuu, in Karaja. Without the clitic the sentence becomes ungrammatical, as shown in (28): (28) *Kua ijadoma-my, isè robira hawa-ki. That girl-ACC her mother saw village-in ‘That girl, her mother saw (her) in the village.’ In complementary distribution with the interrogative particle aõbo, the free functional morpheme dori occurs consistently to the right of the NP to which it refers. We analyze dori as a focus particle, maybe a cleft construction, which unlike the topic construction introduces new information. The interpretation of a sentence such as (29) differs from the interpretation of a sentence such as (27), because in (27), the NP Kua ijadoma-my expresses given information, whereas in (29) the NP has a focus interpretation, that is, it constitutes the new information in itself. (29) Kua ijadoma-my dori isè robira hawa-ki. That girl-ACC FOC her mother saw village-in ‘It was that girl that her mother saw in the village.’ Observe that, now, as expected, it is not possible to coindex the focused NP with the resumptive clitic. As discussed above, one of the diagnostics to distinguish topic and focus is exactly the impossibility to include the resumptive clitic in focus constructions. This is exemplified in (30): (30) * Kua ijadoma-my dori isè tuu robira hawa-ki. That girl-ACC FOC her mother her saw village-in ‘It was that girl that her mother saw her in the village.’

180

Marcus Maia

Consider now data as (31). This is a yes/no interrogative topic construction, in which the interrogative operator aõbo is in the second constituent position. Sentence (31) includes the third person clitic tuu. If the clitic is omitted, the structure will be ill formed, as in (32): (31) Kua ijadoma-my aõbo, isè tuu robira hawa-ki? that girl –ACC Q her mother her saw village-in ‘That girl, did her mother see her in the village?’ (32) *Kua ijadoma-my aõbo isè robira hawa-ki? That girl-ACC Q her mother saw village-in ‘That girl, did her mother see in the village?’

5. The conditional operator txibo Rizzi (1997) proposes that the finiteness system is expressed by the lowest projection of the CP system, selecting an inflectional system with distinctions whose morphological realization can vary from language to language, but which seem to be related to free functional morphemes in the Finiteness phrase. In Karaja, the conditional operator implies a specific morphological inflection in the verb form, which displays the suffix –keki, as exemplified in (33) and (34): (33) Txibo kua habu ixy r-i-rubuny-keki, i-riorè if that man boar 3-THEME-kill-SUBJ, 3-child ràma r-i-sa-õ-ke. hunger 3-THEME-feel-NEG-COND ‘If that man had killed the boar, his child would not be hungry.’ (34) Txibo kai b-i-heteny-keki, weryry r-a-hiny-kre. if you 2-THEME-hit-SUBJ, boy 3-THEME-cry-FUT ‘If you hit him, the boy will cry.’ Notice that the dependence between the operator txibo and the verbal suffix – keki is exemplified in (35) and (36). In (35), the idea of condition is replaced by the idea of time and accordingly the suffix –kre which indicates future tense must be used rather than –keki, as demonstrated in (36):

The structure of CP in Karaja

181

(35) Kai b-i-heteny-kre-u, weryry r-a-hiny-kre. you 2-THEME-hit-FUT-when, boy 3-THEME-cry-FUT ‘When you hit him, the boy will cry.’ (36) *Kai b-i-heteny-keki-u, weryry you 2-THEME-hit-SUBJ-when boy ‘When you hit him, the boy will cry.’

r-a-hiny-kre. 3-THEME-cry-FUT

In embedded clauses, txibo can also function as a complementizer, but in this case, the suffix –keki cannot be used. Rather, in those constructions, if the verb marks its complement NP with the accusative suffix –my, as discussed above, the embedded clause will be marked with the same suffix on the subordinate verb, as illustrated in (37): (37) Kua habu r-i-ery-õ-reri txibo i-hawyy that man 3-THEME-know-NEG-PRES whether 3-woman r-u-ru-ra-my. 3-THEME-die-PAST-ACC ‘That man does not know whether his wife died.’

6. The subordinative suffix –my Embedded clauses which are arguments of verbs such as –bi- ‘see’, -ery‘know’, -hõtiny- ‘think’ and -lyy- ‘tell’ are generally marked in the final position of the verb by the subordinative –my. In such constructions, the CP system may host a bo-word or it may be empty, as illustrated below: (38) Kua habu r-i-ery-õ-reri tiubo i-hawyy that man 3-THEME-know-NEG-PRES when 3-woman r-u-ru-ra-my. 3-THEME-die-PAST-SUB ‘That man does not know when his wife died.’ (39) Kua habu r-i-ery-õ-reri aõherekibo i-hawyy that man 3-THEME-know-NEG-PRES why 3-woman r-u-ru-ra-my. 3-THEME-die-PAST-SUB ‘That man does not know why his wife died.’

182

Marcus Maia

(40) Kua habu r-i-ery-õ-reri i-hawyy that man 3-THEME-know-NEG-PRES 3-woman r-u-ru-ra-my. 3-THEME-die-PAST-SUB ‘That man does not know his wife died.’

7. The hearsay particle ràki As we showed in Maia (2007), the most typical evidential marker in Karaja is the hearsay particle ràki. Speakers often resort to this particle in order to indicate that they have not witnessed the facts which will be reported. In this sense it functions as a quotative device. (41) Wa-se-riore ràki benora 1POSS-mother-child QUOT tucunare(fish) r-i-waxi-ny-re. 3A-THEME-hook-VERBALIZER-PAST ‘It is said that my brother caught a a tucunaré.’ (42) Benora inatximy ràki habu sohoji sohoji Tucunaré two QUOT man one one r-i-my-ra 3A-THEME-catch-FUT ‘It is said that each man caught two tucunarés’ Even though ràki is frequently used in everyday speech whenever the speaker judges relevant to indicate that a predication was not established out of direct observation or experience, its occurrence in mythical narratives is even more productive. Note that the particle is constantly reiterated along the story: (43) Bydolyke ràki ijoi r-a-hu-re ijyy-my Pirarucu QUOT people 3A-theme-finish-PAST story-DAT ar-e-lyy-kre. 1A-THEME-tell-FUT Aõma ràki, ijoi ràki waximy rare. Tai ta thus QUOT people QUOT to fish went. So then ràki r-e-hemyny-re ràki. QUOT 3A-THEME-arrive-PAST QUOT Tai ta ràki bydoleke ràki r-i-my-my

The structure of CP in Karaja

183

so then QUOT pirarucu QUOT 3A-theme-catch-SUB r-y-i-myhy. 3A-theme-be-CONT Ràki r-i-my-my ràki r-y-i-myhy. QUOT 3A-THEME-catch-SUB QUOT 3A-THEME-be-CONT ‘I will tell the story about the pirarucu which is said to have killed the people. So it is said that the people (it is said) went fishing. So then it is said that they arrived (it is said). Then it is said that they were said to catch (the pirarucu).It is said that catching, it is said, they were.’ Unlike several other Karaja particles, which typically occupy the second constituent position in the sentence, ràki displays extremely free ordering possibilities. As exemplified in (41) and (42), its canonical position seems to be the second constituent position, but, as illustrated in (43), it may also occur in the sentence initial position, establishing a frame for the whole story. It may also appear between the subject and the verb, between discourse particles and the verb, between an auxiliary and a main verb and it may be a sentence final particle. It may even be marked with the emphatic suffix -he, as shown in (44): (44) Tai ta ràki-he hãlòe r-e-hemyny-ra. so then QUOT-EMPHATIC jaguar 3A-THEME-arrive-PAST ‘So then it is said that the jaguar arrived.’ Having such a free distribution in the sentence, ràki poses a challenge to the cartographic project, since it cannot be associated with a specific functional projection. Therefore, we propose that our data on the Karaja hearsay particle ràki adduces evidence in favor of Matthewson, Davis and Rullmann (2007) analysis of evidentials as a parasitic category, which cannot be confined to a single position in the functional hierarchy. These authors review evidence of the north American indigenous language St’át’imcets (Salish) to conclude, against Cinque (1995) and Speas (2004), that evidentials are structurally heterogeneous and do not conform to a fixed universal syntactic ordering. As the Karaja data reviewed above indicate, this proposal seems to be on the right track. 8. The internal configuration of wh-phrases and the structure of CP in Karaja In this section we present our proposal of analysis for the internal configuration of the wh-words, both in the wh-questions and in the yes/no interrogatives, and for the topic, focus and conditional structures described

184

Marcus Maia

above. Following Abney (1987), we explore the possibility that an NP has two projections: a lexical projection whose head is N and a functional projection whose head is D. This structure will allow us to analyze a sentence such as (20) as represented in (45): (45)

Fin P DP Aõ

Fin' D’

NP N’ | N | utura

Fin˚ [+WH] D Bo

IP Kai

I' I

VP V’ V temyta

Note that this representation proposes to instantiate the referential wh-phrase aõ-utura-bo in the specifier positon of the Finiteness Phrase, the lowest projection of the CP complex, in contrast with Rizzi’s analysis for the whphrases in Italian, which occupy the Spec of the Focus Phrase. Basically, our proposal intends to capture the dependency relation between inflectional properties of the IP/VP system and the CP system, as we demonstrate in the next section. We also intend to make explicit Tsai’s (1994) and Chomsky’s (1995) intuition that wh-words are formed by an indefinite element + a whfeature. According to our analysis, the Karaja language would have a strong interrogative feature in CP, requiring the overt checking of the wh-phrase (or bo-phrase in Karaja), which must rise to the position of the specifier of the relevant functional projection in the CP complex, namely, the Finiteness Phrase. As reviewed above, according to Chomsky (1995), the Move Feature operation must take along the necessary material to guarantee the convergence of the derivation. This way, in (45) it is not only the bo-feature which rises, but the whole phrase whose head is the functional element bo. That is why both the indefinite [-human] aõ, and the object NP utura, which is in the complement position of bo, must also be raised. We propose now an analysis for wh-phrases such as (22), in which the postpositional marker required by the verb must also be present in the structure. Our proposal is that the position of the complement of the bo head

The structure of CP in Karaja

185

inside the DP is occupied not by an NP, but by a PP, as represented in (46). We propose the representation in (47) to analyze Yes/No interrogatives as (31), with a topic NP. (46)

Fin P DP Aõ

D’ PP

NP | N | ijyy

Fin'

P | my

Fin˚

IP

D Kai I' | [+WH] bo I VP V’ V | telyyta

(47)

TopP NP

Top'

Kua ijadoma-myi Top

Fin P DP

aõ

Fin'

D' Fin˚ IP | D Isè I' | bo I VP V’ V’

PP

tuu-i V hawa-ki robira Observe that in (47) we capture the syntactic and morphological symmetry between the two types of interrogatives in Karaja, the wh-interrogatives and

186

Marcus Maia

the Yes/No interrogatives. The interrogative operator aõbo is represented in the same position as the wh-phrases. The difference is that the bo head of the Yes/No operator is an intransitive form, as it does not select any complement. The NP Kua ijadoma-my ‘that girl’, occupies the functional projection immediately above, a Topic Phrase, implying that the yes/no interrogatives necessarily topicalize the constituent followed by the interrogative operator aõbo. Note that the position that the NP Kua ijadomamy occupies is Spec, Topic Phrase and not Spec, Focus Phrase since only the topic construction allows the resumptive clitic, as noted above. We, now, analyze in (48) the focus construction in (29): (48)

Focus P NP Kua ijadoma-my

Focus' dori

TopP Top' Top˚

FinP Fin' Fin˚

IP Isè

I I

VP V' | V

PP hawa-ki robira

Note that the focus functional morpheme dori is analyzed as the head of the focus functional projection whose specifier position is occupied by the focalized NP Kua ijadoma-my. As we observed above, the clitic is not allowed in this construction, as expected for focus structures as (43). Finally we represent in (49) the conditional construction exemplified in (34), in which there is a dependence between the bo-word txibo, which occupies the

The structure of CP in Karaja

187

position of Spec, FinP, and the inflectional element –keki. In (50) and (51), we represent the embedded clauses (37) and (40), in which the whole embedded clauses are marked by the suffix –my which is attached to the verb form. (50) analyzes the specifier of the Fin Phrase filled by the txibo operator, whereas (51) demonstrates that that position can remain empty, even if related to the suffix –my in the embedded clause.

(49) (…)

Fin P txibo

Fin' Fin˚ IP | [+WH] Kai

I'

I

VP V’ V biheteny-keki

(50)

(…) txibo

Fin P Fin' Fin˚ IP | [+WH] ihawyy I' I

VP V’ V r-u-ru-ra-my

188

(51)

Marcus Maia

(…)

Fin P Fin' Fin˚

IP ihawyy I

I' VP V’ V r-u-ru-ra-my

9. Conclusions We investigated in this paper the structure of the interrogative wh-words in Karaja, as well as topic, focus, condition and embedding constructions, adopting an analysis which intended to provide an integrated account for these structures. The Karaja language displays wh-features that, as we sought to demonstrate, allow a clear-cut segmentation. Our analysis of the sublexical makeup of the Karaja bo-words proposes that the -bo element be the head of its functional category, allowing us to capture syntactically the morphological parallelism that exists between those constructions. We also explored topic and focus constructions, relating them to the interrogative constructions inside the CP system. We, then, discuss Karaja data on the evidential hearsay particle ràki which poses a challenge to the cartographic project, since it does not seem to conform to a single functional position in the sentence. Finally we analyzed the conditional operator txibo, providing a motivation to represent the bo-words in the Finiteness Phrase, the lowest projection in the CP complex, which selects an inflectional element related to the free functional morpheme txibo in Spec, FinP. We also propose an analysis for the embedded clauses which are marked by the suffix –my, capturing another dependency relation between the Fin P of the CP system and the structure of the IP/VP complex.

The structure of CP in Karaja

189

References Abney, Steve 1987 The English Noun Phrase in its Sentential Aspects. Ph.Diss., MIT. Benincà, Paola 2001 The position of Topic and Focus in the left periphery. In Current Studies in Italian Syntax, Guiglielmo Cinque and Giampaolo Salvi. (eds.) pp. 39–64. Amsterdam: Elsevier. Cheng, Lisa 1993 On the typology of Wh-questions. Ph.Diss. MIT. Chomsky, Noam 1977 On Wh-movement. In Formal Syntax, Peter Culicover, Thomas Wasow, and Adrian Akmajian, (eds.) pp. 71–3132. New York: Academic Press. Chomsky, Noam 1993 A Minimalist Program for Linguistic Theory. MIT Occasional Papers in Linguistics, 1. Chomsky, Noam 1995 The Minimalist Program. Cambridge: MIT Press. Chomsky, Noam 2005 Three factors in language design. Linguistic Inquiry, 36(1): 1–322. Chomsky, Noam 2008 On Phases. In Foundational Issues in Linguistic Theory: Essays in Honor of Jean-Roger Vergnaud, Robert Freidin, Carlos Otero and Maria-Luisa Zubizarreta. (eds). pp. 133–166. Cambridge, Mass: MIT Press. Chomsky, Noam, and Howard Lasnik 1993 The theory of principles and parameters. In Syntax: An International Handbook of Contemporary Research, J. Jacobs et al. (eds.) Vol. 1. pp. 506–569. Berlin: Walter de Gruyter. Cinque, Guiglielmo 1999 Adverbs and Functional Heads: A Cross-Linguistic Perspective. Oxford: Oxford University Press. Maia, Marcus 1998 Palavras Interrogativas em Karajá. Paper presented at the Meeting of the Indigenous languages work group during the XIII National Meeting of the National Association of Research and Graduate Studies in Letters and Linguistics - ANPOLL , Campinas, SP. Maia, Marcus 2000 Syntaxis Comparada de las interrogativas en Karajá, Kayapó y Manchineri. In Essays on Indigenous Languages of Lowland South America, Hein van der Voort and Simon van de Kerke (eds.) .pp. 297– 308. Leiden: CNWS Publications.

190

Marcus Maia

Maia, Marcus 2001 WH-type constructions in Karaja. Kansas Working Papers in Linguistics, 25 ( 2): 21–36. Maia, Marcus, Andres, Salanova and Elder Lanes 2007 Evidentiality Processes in Karajá. In: L' enonciation mediatiseé II - Le traitement épistémologique de l'information: illustrations amérindiennes. Jon Landaburu and Zlatka Guentcheva. (eds). pp. 293– 308. Louvain - Paris: Peeters, II. Matthewson, Lisa, Henry Davis and Hotze Rullmann: 2007 Evidentials as epistemic modals: Evidence from St’át’imcets’, Ms. University. of British Columbia. Rizzi, Luigi 1991 Residual verb second and the wh-criterion. Technical report 2. Université de Genève. Rizzi, Luigi 1997 The Fine Structure of the Left Periphery. Elements of Grammar. Liliane Haegeman (ed). 281-337. Dordrecht: Kluwer Academic Press. Speas, Margaret  (YLGHQWLDOLW\ORJRSKRULFLW\DQGWKHV\QWDFWLFUHSUHVHQWDWLRQRI SUDJPDWLFIHDWXUHV/LQJXD² Tsai, Wei.-Tien. Dylan 1994 On economizing the theory of A-bar dependencies. Ph. Diss. , MIT.

Agree and the licensing of wh-words and polarity sensitive items in Southern Quechua1 Liliana Sánchez 1. Introduction In Southern Quechua varieties, all languages with canonical SOV word order and flexible word order (Cusihuamán [1976] 2001:61) determined by the informational structure of the sentence (Cerrón-Palomino 1989:290), whfronting and wh in situ are possible, as shown in examples (1)-(2) from a variety of Southern Quechua, Cuzco Quechua.2 For many speakers the preferred formulation of a wh- question is the one in (1). In (1) and in (2), the wh-word is marked for accusative case and it is also marked with the suffix m that may be interpreted as focus and evidentiality in declarative sentences but only as focus in interrogative sentences.3 (1)

Ima-ta-m Mariya yacha-n?4 What-ACC-FOC Mariya know-3S ‘What does Mariya know?’

(2)

Mariya ima-ta-m yacha-n? Mariya what-ACC-FOC know-3S ‘What does Mariya know?’

In some contexts such as free relatives and negative sentences, wh-words in situ behave as a polarity sensitive items (PSI).5 In the free relative in (3), the wh-word ima appears as imaymana and is a Free Choice Item, whereas in (4), the wh-word is marked with the suffix –pas and is a Negative Polarity Item. In polarity sensitive contexts, such as negative sentences, the whinterpretation is not possible, as shown in (5) and indicated by the symbol #: (3)

Huwan Mariya-q imaymana miku-sqa-n-ta muna-rqa-n.6 Huwan Mariya-GEN whatever eat-NOM-3S-ACC want-PAST-3S ‘Huwan wanted whatever Maria ate.’

(4)

Mana-n ima-ta-pas ranti-rqa-ni-chu. NEG-FOC what-ACC-ADD buy-PAST-1S-NEG ‘I did not buy anything.’

192

Liliana Sánchez

(5)

#Mana ima-ta-m ranti-rqa-nki-chu. NEG what-ACC-FOC buy-PAST-1S-NEG ‘What did you not buy?’

In this paper, I attempt to answer two questions: 1) Why can items such as ima in Quechua be interpreted as wh-words and as polarity sensitive items in different contexts?, and 2) Why do these items receive different morphological markings in each context? In order to answer these questions, I will propose that the interpretation of these items is dependent on the relations they establish with different operators in the clause and that those relations are spelled out as different forms of agreement according to the operators involved. I follow Sánchez’s (2008) proposal that in Southern Quechua wh-words can be underspecified as wh-words and as polarity sensitive items. The fact that wh-words are frequently polarity items in many languages has been noted from a typological perspective by Haspelmath (1997). In this paper, I will take the view that both wh-words and polarity items in Quechua are indefinite elements that are underspecified for a set of relevant features that may be valued by a non-veridicality operator in T or by a Q operator in C. In the interrogative interpretation of wh-words in Quechua, a Q:wh feature is valued by a C-level operator when there are no other potentially intervening operators, irrespectively of whether the wh-word is fronted or not. In the case of polarity items, the polarity feature is a non-veridicality value [-v] that is valued via agree (Chomsky 1999) with a TP-internal non-veridicality operator (Giannakidou 2001, Chierchia 2006). In the case of free relatives of the type discussed by Dayal (1996), the non-veridicality feature is valued by an operator in the relative clause. The paper is divided as follows. In section 2, I present the distribution of wh-words in main clauses and in embedded sentences in Southern Quechua. This distribution is important to understand how the T and C level operators interact with indefinite wh-words. In section 3, I present the evidence in favor of a [-v] operator inside TP that licenses polarity sensitive items, and in section 4 I discuss intervening effects between C-level and T-level operators.

2. Wh-fronting in main and subordinate clauses In Southern Quechua main interrogative clauses fronting of the wh- is strongly preferred over wh-in situ.7 Sentence (6) illustrates the fronting of the object whereas (7) shows the in situ option. However, the interpretations of

Wh-questions in Southern Quechua

193

(6) and (7) differ. While (6) asks information about Maria’s knowledge, (7) is interpreted as a questioning of Maria’s knowledge: (6)

Ima-ta-m Mariya yacha-n? what-ACC-FOC Mariya know-3S ‘What does Mariya know?’

(7)

Mariya ima-ta-m yacha-n. Mariya what-ACC-FOC know-3S ‘What does Mariya know?’

Fronting is required with intransitive verbs, as shown by the ungrammaticality of (9) and (11): (8)

Pi-n hamu-rqa-n? who-FOC come-ATT PAST-3S ‘Who came?’

(9)

*Hamu-rqa-n pi-n? come-ATT PAST-3S who-FOC? ‘Who came?’

Fronting is also required with adjuncts: (10) May-pi-n Mariya tiya-n? Where-LOC-FOC Mariya live-3S ‘Where does Mariya live?’ (11) *Mariya tiya-n may-pi-n? Mariya live-3S where-LOC-FOC ‘Where does Mariya live?’ In the fronting cases, extraction out of a relative clause results in ungrammaticality. This indicates a movement type of dependency between the wh-element and the gap:8 (12) *Ima-ta-m Mariya [runae ruwa-sqa-n-ta] what-ACC-FOCi Mariya [man e make-NOM-3S] riqsi-rqa-n? know-ATT PAST-3S ‘Mariya knew the man who built what?’

194

Liliana Sánchez

In subordinate clauses fronting out of complement clauses is required. Complement clauses in Southern Quechua are nominalizations, as shown in (13). Sentence (14) shows that an in situ ima in a nominalization cannot be interpreted as a wh-word. (13) Ima-ta-n Huwan muna-rqa-n what-ACC-FOC Huwan want-ATTPAST-3S miku-na-n-paq]? eat-NOM-3 S-FIN] ‘What did Huwan want Mariya to eat?’

[Mariya-q [Mariya-GEN

(14) *Huwan [Mariya ima-ta miku-sqa-n-ta] muna-rqa-n? Huwan [Maria what-ACC eat-NOM-3-ACC] want-ATT PAST-3S ‘What did Huwan want Mariya to eat?’ In some varieties of Southern Quechua such as Bolivia Quechua, wh- in situ is allowed in adjunct nominalizations, although this is not accepted in some varieties of Cuzco Quechua: (15) Huwan-cha ri-n [pi riku-q]? Huwan-DIM go-3S who see-AG ‘Who does Huwan go to see?’ (Lefebvre and Musyken 1988: 160) In Cuzco Quechua, embedded questions also show fronting: (16) Tapuku-ni-n may-pi tiya-nki. ask-1S-FOC where-LOC live-2S ‘I wonder where you live’ To summarize, wh-fronting in Southern Quechua is the preferred option in main clauses and is required out of complement nominalizations but not from adjunct nominalizations in Bolivian Quechua. 3. Agree and polarity interpretations As mentioned before, in addition to their interpretations as wh-words, words such as ima are Polarity Sensitive Items, a term originally proposed by Baker (1970) and subsequently used by many researchers (Progovac 1994, Déprez 1999, Haegeman 1995, among others). Polarity sensitive items such as any in

Wh-questions in Southern Quechua

195

English can only appear in specific contexts involving negation or some form of modality that licenses them and are ungrammatical in other contexts as shown in the contrast between (17) and (18): (17) I did not find any book. (18) *I found any book. Wh-words in Quechua, like English any, are interpreted as Negative Polarity Items (NPIs) in the scope of negation: (19) Mana-n ima-ta-pas ranti-rqa-ni-chu. NEG-FOC what-ACC-ADD buy-PAST-1S-NEG ‘I did not buy anything.’ This interpretation is not available in regular episodic contexts, namely non-generic contexts as noted for other languages by Dayal (1998), Giannakidou (2001), Giannakidou and Cheng (2006) and Chierchia (2006):9 (20) *Mariya ima-ta-pas yacha-rqa-n. Mariya what-ACC-add know-past-3S ‘Maria knew anything.’ (21) *Pi-pas hamu-rqa-n. who-ADD come-ATT PAST-3S ‘Anybody came.’ (22) *Pi-pas ima-ta-pas ranti-rqa-n. who-ADD what-ACC-ADD buy-ATT PAST-3S ‘Anybody bought anything.’ (23) *Ima-ta-pas pi-pas ranti-rqa-n. what-ACC-ADD who-ADD buy-ATT PAST-3S ‘Anybody bought anything.’ Notice the contrast with non-episodic contexts such as the generic use of the present tense: (24) Pi-pas ima-ta-pis ranti-nku-n.10 who-ADD what-ACC-ADD buy-3SPL PRES-FOC

196

Liliana Sánchez

‘Anybody buys anything.’ In other contexts, such as in imperative sentences, any can be interpreted as a Free Choice Item (FCI) (Dayal 1998, Tovena 1999, Giannakidou 2001): (25) Bring any book! This is also true of ima. Notice that in this case it is also modified by the additive suffix -pas: (26) Ima-ta-pas apamu-y! Anything-ACC-ADD bring-IMP ‘Bring anything!’ Ima is also an FCI in contexts involving some form of modality, a fact noted by Dayal (1998) for Hindi. One of such contexts is that of a sentence with a verb modified by a suffix that has been labeled as expressing conditional mood (Cusihuamán 2001):11 (27) Ima-ta-pas ruwa-n-man-mi uywa-n-kuna-rayku-qa. any-ACC-ADD do-3S-COND-FOC/EVID animal-3S-PL-CAUS-TOP ‘(S)he would do anything for her/his animals.’ This is also the case when ima is in the scope of a constituent marked with the yes/no question marker –chu: (28) Uyari-rqa-nki-chu ima noticia-ta-pis naha tutamantan? Hear-PAST-2S-INT what news-ACC-ADD this morning ‘Did you hear any news this morning?’ (Cusihuamán 2001: 107) It is also the case in the context of a free relative clause, as noted by Dayal (1998) for Hindi. In this case the FCI is the word imaymana formed by the root ima and the ending –ymana that may be analyzed as involving the negative word mana. This was shown in (3) repeated here as (29): (29) Huwan [Mariya-q imaymana miku-sqa-n-ta] muna-rqa-n.12 Huwan Mariya-GEN whatever eat-NOM-3S-ACC want-PAST-3S ‘Huwan wanted whatever Maria ate.’ To summarize, wh-words are interpreted as FCIs when they appear in

Wh-questions in Southern Quechua

197

imperative sentences, modality contexts, yes/no questions and free relative clauses. As noted by Giannakidou and Cheng (2006) for other languages, FCIs tend to be modified by an additive particle. In the case of Southern Quechua, they are modified by the additive suffix –pas (~-pis). The only case in which the additive –pas is not present is the free relative clause that shows a different ending, -ymana. These properties indicate that wh-words are licensed as FCIs in non-veridicality contexts (Giannakidou 2001), namely in contexts in which some modality or speech act operator is involved. In the case of the free relatives, wh-words are licensed as FCIs by an operator that, according to Rullman (1995), selects the unique maximal sum of a set when it is applied to a set of individuals ordered by the partwhole relationship. This operator is the one that licenses the FCI in (3)/(29). Given this distribution, wh-words appear to meet Giannakidou’s (1999, 2001) definition of Affected Polarity Items (APIs). They are grammatical in affective contexts such as questions and negative sentences and they also conform to Chierchia’s definition of Polarity Sensitive Items (PSIs); their different interpretations follow from the interaction of their meaning and the syntactic contexts in which they are inserted. I propose an analysis of wh-words in Southern Quechua according to which the differences in their interpretation as wh-words and as polarity items are related to their interaction with sentence-level operators and are determined by their syntactic position. Following Chierchia’s (2006) proposal, I argue that when wh-words are in the scope of a TP-internal eventrelated operator that has non-veridicality values they are interpreted as polarity items. I also argue that in the absence of that type of operator, they must be valued by a Q operator at the CP level and are interpreted as whelements. In the absence of either type of operator, as is the case in episodic contexts in declarative sentences, they are ungrammatical. In Chierchia’s unified theory of polarity sensitivity items, negative polarity items such as English any occur in contexts in which they allow the widening of the set of individuals salient in the context of use. This set is referred to as domain D. To exemplify how polarity items widen a domain let us look at a sentence such as (30): (30) There is a student D (in that building). In sentence (30), D represents the set of students that is relevant to the utterance. Under the scope of negation, a polarity item such as any widens the domain D by allowing the inclusion of more individuals in the set, that is, all students are now part of the relevant set:

198

Liliana Sánchez

(31) There isn’t any student D+ (in that building). Chierchia argues that polarity sensitive items are not felicitous in affirmative contexts because the widening of the domain in such contexts is non-informative since it does not alter the speaker’s knowledge that there is a student in that building as shown in: (32) *There is any student D (in that building). On the other hand, widening the domain of individuals in (31) is informative because it is clear that, even when considering a larger domain, there was no student in that building. In order to have an operational way of introducing the notion of domain widening (a scalar implicature), Chierchia proposes that certain scalar implicatures are introduced by an operator (he uses the symbol [+ı] to represent the operator) and are associated with morphological items such as any. He assumes that “NPIs carry an (uninterpretable) feature (specifically a piece of possibly abstract negative morphology) that needs to be checked by an appropriate (interpretable) operator (namely, ı). NPIs must be checked by ı (i.e., one might say enter an agreement relation with ı). (Chierchia 2006: 557).”

This is precisely the line of analysis that I will adopt here. I will argue that Southern Quechua wh-words exemplify cases of underspecified morphemes that enter agree relationships with negative and modal operators inside TP and with wh- operators in the C-domain. There is one additional characteristic that supports the view that the syntactic position of the wh-word and the relevant operator is crucial for its interpretation: a wh-interpretation is possible despite a negation head in the numeration, when the wh-word ima is outside the scope of negation. This is shown in: (33) Ima-ta-m mana muna-nki-chu? what-ACC-FOC NEG want-2S-NEG ‘What don’t you want?’ This indicates that the relationship between the operators and the morphological items is construed locally. In order for ima to be interpreted as a wh-element, there must be at least one overt copy of the wh-word in a local relationship to the Q:wh operator. By local I mean one in which there are no potentially intervening operators. The main difference between an NPI and a wh-interpretation of wh-words is related to the position of the highest copy in the sentence. When the highest copy is inside the TP and

Wh-questions in Southern Quechua

199

there is a relevant operator such as negation, an agree relationship is established between the TP-internal operator (ı) and the underspecified whword. When the highest copy of the wh-word is outside the scope of a TPinternal operator, it agrees in Q features with a Q:wh operator in C. These different types of agree relationships are possible because wh-words are underspecified for non-veridicality and Q features. Evidence that wh-words enter different agree relationships with different operators is the fact that they are marked with different suffixes when they are in the scope of TP-internal operators and when they are in the scope of CP-internal operators. As FCIs and NPIs inside TP, they are marked with the suffix –pas, also used as a conjunction (see below). When they are interpreted as wh-words, they are marked with the evidentiality/focus marker –mi, as shown in the contrast between (34)-(35) and (36): (34) Ima-ta-pas apa-mu-y! what-ACC-ADD bring-CIS-IMP ‘Bring anything!’ (35) Mana ima-ta-pas muna-ni-chu. Neg what-ACC-ADD want-1S-NEG ‘I do not want anything.’ (36) Ima-ta-m(*-pas) what-ACC-FOC (*-ADD) ‘What do you want?’

muna-nki. want-2S

I take this morphological difference to indicate different agree relationships between the word ima and the C-domain operators: Q:wh and Focus (spelled out as –m in the case of imata-m) and between ima and a TPinternal operator (spelled out as –pas). Evidence in favor of the T-related nature of –pas comes from the fact that it is the conjunction used to coordinate events inside a clause. As noted by Camacho and Sánchez (1996), Southern Quechua has three suffixes that are used as conjunctions: –nti, -wan and -pas. While –nti and –wan are grammatical when used to conjoin arguments (subjects or objects) that are related to a single event referred to by a verb, -pas is the only suffix that can be used to coordinate two different events expressed by finite verb forms. This is shown by the contrast in grammaticality between (37) and (38)-(39):

200

Liliana Sánchez

(37) Phista-pi [tusu-ni-pas, upya-ni-pas].13 party-LOC [dance-1S-ADD, drink-1S-ADD] ‘At the party I dance and (I) drink.’ (38) *Phista-pi [tusu-ni-wan, upya-ni-wan]. party-LOC [dance-1S-CONJ, drink-1S-CONJ] ‘At the party I dance and (I) drink.’ (39) *Phista-pi [tusu-ni-nti, upya-ni-nti]. party-LOC [dance-1S-CONJ, drink-1S-CONJ] ‘At the party I dance and drink.’ In (37) the coordination of the two TPs (‘I dance’ and ‘I drink’) is possible and is marked with the suffix –pas on both conjuncts. Sentences (38) and (39) show that coordination of events with –wan and –nti is not grammatical and contrasts with the grammaticality of these suffixes as conjunctions of arguments, as shown in (40) and (41): (40) Warmi-kuna-wan qhari-kuna-wan llaqta-man woman-PL-CONJ man-PL-CONJ town-DAT puri-rqa-nku. walk-ATT PAST-3 PL ‘Women and men marched (in separate groups) towards the town.’ (41) Warmi-kuna-(ntin) qhari-kuna-ntin llaqta-man woman-PL-(CONJ) man-PL-CONJ town-DAT puri-rqa-nku. walk-ATT PAST-3 PL ‘Women and men marched (collectively) towards the town.’ (Camacho and Sánchez 1996: 33) I take this to indicate that –pas conjoins TPs that refer to independent events. Having established that –pas is a conjunction of TPs denoting events, we can now go back to sentence (19) repeated here as (42):14 (42) Mana-n ima-ta-pas ranti-rqa-ni-chu. NEG-FOC what-ACC-ADD buy-PAST-1S-NEG ‘I did not buy anything.’

Wh-questions in Southern Quechua

201

This sentence raises the question of the location of the operator that licenses imatapas and of its relationship to negation. One possibility would be for negation as a head to be the operator that licenses imatapas. I would like to propose that while negation is involved in the agree relationship, negation is not the probe that agrees directly with ima as an NPI. In order to account for the fact that modality operators can also license wh-words as FCIs without involving a negative interpretation, it would be better to propose that both negation and modality operators license an abstract eventrelated operator that has the relevant properties to license an FCI or an NPI. In this account, the suffix –pas is the spell out of the agree relation between the ı operator located in T and imatapas. Positing the T head as the locus for the operator seems reasonable given its event-related interpretation in conjunctions. In order to be licensed, ima must agree with the ı operator. This ı operator is licensed by other operators responsible for the veridicality of the proposition (Giannakidou 2001). According to Giannakidou, polarity sensitive items are subject to licensing conditions directly related to the veridicality of the operators involved. She proposes a licensing condition and an anti-licensing condition for polarity items: (43) Licensing by non-veridicality “A polarity item Į will be grammatical in a sentence S iff Į is in the scope of a non-veridical operator ȕ in S.”

(Giannakidou 2001: 670) (44) “Anti-licensing by veridicality A polarity item Į will not be grammatical in a sentence S iff Į is in the scope of a veridical operator ȕ in S.”

(Giannakidou 2001: 670) She defines veridicality as: “a property of propositional operators in terms of truth entailment. A propositional operator [Op p] is veridical iff the truth of Op p in c [context that contains a set M of Models relative to an individual x] requires that p be true in some individual x’s epistemic model ME (x) in c. If the truth of Op p in c does not require that p be true in some such model in c Op is nonveridical. A non`-veridical operator is furthermore antiveridical iff the truth of Op p in c requires that p be false in some epistemic model ME (x) in c.” (Giannakidou 2001: 671).

202

Liliana Sánchez

It follows from such definition that negation generates an antiveridical operator. In the scope of the overt negation operator mana, the NPI ima must be licensed. I propose that the syntactic means by which ima is licensed is by checking of the features of a propositional operator in T that is by default veridical (this would be the operator found in episodic sentences). This operator obtains its non-veridical [-v] value from Neg. Ima is underspecified for a [-v] feature or a Q:wh feature. The operator is the probe and ima is the goal and they agree in [-v] features. The spell out of the agree relation between the [-v] operator and the wh-word is the suffix –pas. The representation in (45) shows the relevant agree relationships signaled with an arrow: (45)

NegP Neg’ Neg Mana

TP T’ T Op [-v]

vP pro

v’ v

VP

imata-pas rantirqani-chu [-v] NEG The proposed analysis accounts for the fact that NPI licensing takes place in the scope of negation but is not the same as negative concord. The negative operator mana has an agreeing counterpart in the suffix –chu that marks the constituent being negated.15 In episodic contexts that are not negated, the veridical nature of the operator meets the anti-licensing condition proposed by Giannakidou. There is a veridical operator that does not inherit an anti-veridicality value from negation or from a modal. Therefore, ima as an NPI or a FCI is not licensed and there is no agreement

Wh-questions in Southern Quechua

203

with a non-veridical operator. The lack of such agreement is evidenced in the ungrammaticality of the suffix –pas on ima in an episodic context: (46) *Ima-ta-pas riku-rqa-ni. any-ACC-ADD see-PAST-1S ‘I saw anything/whatever.’ Agreement with a non-veridicality operator can be extended in Southern Quechua to contexts in which ima restricts an NP as in structures of the type ima NP-ta-pis as in sentence (28). In this case, the licensing of the NP as a polarity item is possible in the context of a yes/no question with a focalized fronted constituent. The derivation for sentence (28) is shown in (47). The verb has moved to the head of Focus Phrase and –chu is the spell out of the yes/no (Q:y/n) question features. I assume an articulated view of the Cdomain (Rizzi 1997) and a Speech Act projection (Cinque 1999): (47)

SpActP SpAct’ SpAct Q: y/n

FocP Foc’

Foc Uyarirqanki-chu

TP T’ T Op [-v]

vP pro

v’ v

VP

ima noticiapis nahatutamanta

204

Liliana Sánchez

In this context, the full NP ima noticiapis ‘any news’ agrees with the nonveridicality operator that inherits its [-v] value from the Q:y/n features in Speech Act.16 In the case of conditional mood sentences and imperatives, the nonveridical nature of the operator is not inherited from negation or the Q:y/n features but from a modality operator and a Speech Act operator respectively. Licensing of the wh-word as FCIs in subordinate clauses seems more difficult to explain for several reasons. First in a sentence such as (3) (repeated here with the addition of –pas) as (48) imaymana receives a universal interpretation that is consistent with universal interpretations of polarity items such as any (Dayal 1998). However, unlike in the cases previously discussed (all of them cases of root clauses), the non-veridicality operator is not located in the main clause but in the subordinate clause. Furthermore, -pas is not grammatical as the spell out of the agree relationship between the operator and ima. (48) Huwan [Mariya-q imaymana- (*-pas) miku-sqa-n-ta] Huwan [Mariya-GEN whatever-(*-ADD) eat-NOM-3S-ACC] muna-rqa-n. want-PAST-3S ‘Huwan wanted whatever Mariya ate.’ I would like to argue that imaymana receives a universal FCI interpretation by virtue of being licensed by the operator internal to the free relative proposed by Rullman (1995). This operator is the one that values the [-v] features of ima. The fact that there is no –pas marking on ima can be accounted for by the lack of a proper T-related non-veridicality operator inside the nominalization. Once the checking of the nonveridicality features takes place the interpretation of imaymana as a universal FCI is possible. The agree relation involved is illustrated in the following representation: 17

Wh-questions in Southern Quechua

(49)

205

AgrP Agr’ Agr

NomP Nom’ Nom

vP

Modal, Op [-v] Mariya

v’ v

VP

imaymanaj

V’

V mikhu-sqa-n-ta

tj

In (49) the modal operator is located in Nom, the head of the nominalization and is the probe that agrees with ima in [-v] features.

4. Consequences of the analysis for wh-licensing and wh-fronting Having proposed an account for the licensing of ima as an NPI and an FCI, we can now turn back to the contexts in which ima is licensed as a wh-word. Ima is licensed by establishing an agree relationship with a Q:wh operator in Speech Act. This agree relation is possible because ima is underspecified for a Q:wh feature. When ima is in situ, agree between the Q:wh operator and ima takes place provided there are no intervening operators. When it is fronted to the specifier of Speech Act, the relevant Q:wh feature is checked in a specifier-head relationship, as shown in (50) (other C-related projections such as Focus Phrase are omitted):

206

(50)

Liliana Sánchez

SpActP Imata-mi

SpAct’

SpAct [Q:wh]

TP T’ T

vP v’ v

VP

ti mikunki The main issue is why can ima not be licensed as a wh- inside TP in certain contexts. Inside TP it could agree with the probes Q:wh and Focus in the C-domain. I would like to argue that the licensing of ima (as well as pi and other wh-words) is dependent on structural conditions, namely on its position relative to nonveridicality and Q:wh operators. Nonveridicality operators act as interveners that block the checking of Q:wh features. This is the case in sentences such as (13) where fronting allows the licensing of the Q:wh feature. In (14) on the other hand, licensing is blocked by a [-v] operator. This blocking effect is avoided when the ima word is in the specifier of Speech Act as in (13). Thus, ima checks off Q:wh features when it is in fronted position, in the spec of Speech Act. Given the apparent optionality of the fronting of objects in main clauses, I propose that wh-fronting in Southern Quechua is necessary to avoid intervening effects by [-v] operators. In the fronted position, ima is the spell out of Q:wh features that are checked via a spec-head relation between the Speech Act head and the XP in its specifier position. On the other hand, when it is in a TP-internal DP position in the scope of negation (as in (35)), it is the spell out of [-v] features. In sentences such as (33) (repeated here as (51) for convenience), ima does not check [-v] features, when its highest copy (the one with phonological content) is outside the scope of negation, as shown in:

Wh-questions in Southern Quechua

207

(51) ¿Ima-ta-m mana muna-nki-chu? what-ACC-FOC NEG want-2S-NEG ‘What don’t you want?’ For this account to work, it is crucial that the spell out of the agree relation surfaces only on the highest copy (the one with phonological content). Once agree in Q:wh features takes place, then it is possible for the negative suffix –chu to be the spell out of negative features: (52)

SpAct P

Imata-m

SpAct’ SpAct [Q:wh]

FocP Foc’ Foc

NegP Neg’ Neg mana

TP T’ T Op [-v]

vP v’

v

VP

imata-m muna-nki-chu Support for the fact that in the scope of negation ima is an NPI can be found in negative yes/no questions: (53) Mana-chu ima-ta-pas ni-y-ta muna-nki? NEG-Q any-ACC-ADD say-INF-ACC want-2S ‘Don’t you want to say anything?’

208

Liliana Sánchez

In this sentence, -chu as the spell out of Q: y/n features surfaces on the negative head mana and imatapas as an NPI is licensed inside the subordinate clause.

5. Conclusions In this paper, I have proposed an account for the distribution of wh-words in Southern Quechua as wh-words and as polarity sensitive items that is based on the idea that in Southern Quechua wh-words are underspecified for +Q and [-v] (non-veridicality) features. Their licensing and their interpretation depend on whether they establish an agree relation with a C-level or a Tlevel operator. In interrogative sentences, wh-words such as ima agree with a Q operator in Speech Act when there are no potentially intervening [-v] operators in T. This results in a wh-interpretation and is compatible with focus marking. In declarative sentences involving non-veridicality contexts (negative sentences, imperative sentences and free relatives) they agree in [v] features with an operator in T. This agree relationship allows their interpretation as NPIs or FCIs. The spell out of this agree relation is the additive suffix -pas which is also the suffix that expresses the conjunction of TPs. In interrogative sentences with a [-v] operator in T and a Q operator in C, in order for a wh-interpretation to obtain and to avoid intervening effects by the [-v] operator in T, the highest copy of the wh-word must be outside TP where it checks its Q features with the operator in C. In the absence of a Q and a [-v] operator, as in the case of declarative episodic sentences, whwords are not interpretable and yield ungrammatical sentences.

Notes 1. Thanks to Vidal Carbajal and Hipólito Peralta Ccama for their help as language consultants. The data in this article come from my own fieldwork (recordings of narratives) and from grammaticality judgments provided by Cuzco speakers who participated in a study in 2005 and from the two language consultants mentioned above. The majority of examples correspond to Cuzco Quechua. When indicated examples come from Apurimac Quechua. Data from other sources is cited. Thanks also to two anonymous reviewers and to José Camacho for their comments. All errors are mine. 2. Southern Quechua varieties belong to the Quechua II or Quechua B family of languages. For a detailed presentation of the classification of Quechua dialects see Cerrón-Palomino (1989).

Wh-questions in Southern Quechua

209

3. I use the term suffix to refer to –mi and its allomorph –n following CerrónPalomino (1987) who classifies them as independent suffixes. 4. The following abbreviations are used: 1, 2, 3 – first, second and third person; ACC – accusative; ADD – additive; AG – agentive; ATT PAST –attested past; CAUS - causal; CIS - cislocative; CONJ –conjunction; DIM- diminutive; EUPH – euphonic; FIN- final; FOC/EVID - focus/evidential; GEN-genitive; IMP – imperative; INF-infinitive; LOC – locative; NEG - negation; NOM – nominalization; PAST – past; PL – plural; S – singular; TOP – topic. 5. In this paper, I will focus only on the argument wh-words ima ‘what’ and pi ‘who’ and on their distribution as wh-words and as polarity sensitive items. 6. This sentence is grammatical in the Apurimac variety with only the accusative marker on the wh-word. The form is: ima-ta (what-ACC). 7. Lefebvre and Musyken (1989) note that in Bolivian Quechua, a language that belongs to the Southern Quechua family, wh in situ in main clauses is grammatical. It is also the case for Central varieties of Quechua such as Ancash Quechua as described by Cole and Hermon (1994). 8. In this respect Southern Quechua differs from Ancash Quechua (Cole and Hermon 1994). 9. Episodic contexts are those that refer to only one event (Giannakidou 2001, Giannakidou and Cheng 2006). 10. The suffix –pis is an allomorph of –pas. (Cusihuamán 2001: 237). 11. See Dayal (1998) for an account of how modality modifies episodic contexts allowing for free choice items. 12. The focus marker –m is not allowed inside nominalizations. 13. The original example is from Cerrón-Palomino (1987). 14. The fact that –pas can be used to conjoin direct objects and other arguments does not constitute a counterexample to the analysis according to which –pas coordinates event denoting expressions. A sentence such as: (i) Papa-ta-pas uqa-ta-pas mikhu-rqa-n Potato-ACC-CONJ uqa-ACC-CONJ eat-ATT PAST- 3PL ‘(They) ate potatoes and uqas.’ can be interpreted as two separate events: one in which potatoes were eaten and another one in which uqas were eaten. 15. I assume that the suffix –chu is also underspecified for [Q: y/n], focus and negation features. 16. I follow Cheng and Rooryck (2000) proposal based on Scobbie’s (1991) Attribute Value Phonology that Q is an attribute that “can take either of two (unary) values: [Q:wh] or [Q:y/n]” (Cheng and Rooryck 2000:6). I will assume that the difference between Yes/No questions and wh-questions is a difference in the type of Q feature selected in the numeration ([Q:wh] versus [Q:y/n]) and that only the latter feature involves a relationship with negation. This difference in features is mirrored in the Quechua morphology by the difference between the

210

Liliana Sánchez

suffix –chu (specified for [Q:y/n], focus and negation) and words such as ima and pi (specified for [Q:wh] and PSI features). 17. I assume a uniform view of left headedness. SOV word orders are derived by movement to the spec of VP. I also assume that the head of the nominalization is nominal agreement that projects into an Agreement Phrase (AgrP).AgrP selects a projection headed by a functional category Nom (for Nominalizer) that is specified for N features but also for some V-related features.

References Baker, C. L. 1970 Double negatives. Linguistic Inquiry 1: 169-186. Camacho, José, and Liliana Sánchez 1996 Three types of conjunctions. Proceedings of the 26 Northeastern Linguistic Society Conference. Kiyomi Kusomoto (ed.), 31–42. Amherst, Massachusetts: GSLA Publications. Cerrón-Palomino, Rodolfo 1989 Lingüística Quechua. Cuzco: Bartolomé de las Casas. Chierchia, Gennaro 2006 Broaden your views: implicatures of domain widening and the “Logicality” of language. Linguistic Inquiry 37 (4): 535–590. Cheng, Lisa, and Johan Rooryck 2000 Licensing wh in situ. Syntax 3 (1): 1–19. Chomsky, Noam 1999 Derivation by phase. MIT Working Papers in Linguistics 18. Cambridge, Massachusetts: MIT Press. Cinque, Guiglielmo 1999 Adverbs and Functional Heads: A Cross-linguistic Perspective. Oxford: Oxford University Press. Cole, Peter 1987 The structure of internally-headed relatives. Natural Language and Linguistic Theory 5 (2): 277–302. Cole, Peter, and Gabrielle Hermon 1994 Is there LF WH-Movement? Linguistic Inquiry 25 (2): 239-262. Cole, Peter, and Gabrielle Hermon 1998 The typology of wh-movement. Syntax 1 (3): 221-258. Cusihuamán, Antonio 1976[2001] Reprint. Gramática Quechua. Cuzco-Collao. Lima: Instituto de Estudios Peruanos. Original edition, Lima: Instituto de Estudios Peruanos. Dayal, Veneeta 1996 Locality in wh-quantification. Dordrecht: Kluwer Academic.

Wh-questions in Southern Quechua

211

Dayal, Veneeta 1998 Any as inherently modal. Linguistics and Philosophy 21: 433-476. Déprez, Viviane 1999 The roots of negative concord in French and French based Creoles’, in Michel DeGraff, (ed.), 375-428. Language Creation and Language Change: Creole, Diachrony and Development, MIT Press. Giannakidou, Anastasia 1999 Affective dependencies. Linguistics and Philosophy 23: 367-421. 2001 The meaning of free-choice. Linguistics and Philosophy 24: 659-731. Giannakidou, Anastasia, and Lisa Cheng 2006 (In)definiteness, polarity and the role of wh-morphology in free choice. Journal of Semantics 23: 135-183. Haegeman, Liliane 1995 The Syntax of Negation. Cambridge: Cambridge University Press. Haspelmath, Martin 1997 Indefinite Pronouns. Oxford: Oxford University Press. Lefebvre, Claire, and Pieter Muysken 1988 Mixed Categories: Nominalizations in Quechua. Dordrecht: Kluwer Academic Publishers. Progovac, Ljiljana 1994 Negative and Positive polarity. Cambridge: Cambridge University Press. Rizzi, Luigi 1996 The fine structure of the left periphery. In Elements of Grammar. Liliane Haegeman (ed.), 281-337. Dordrecht: Kluwer Academic Press. Rullman, Hotze 1995 Maximality in the semantics of wh-Constructions Ph.D. Diss. University of Massachusetts, Amherst, distributed by Graduate Linguistic Student Association. Sánchez, Liliana 2008 Agree and the left periphery in Southern Quechua. Ms. Rutgers University. Scobbie, James 1991 Attribute value phonology. Ph.D. Diss. University of Edinburgh. Tovena, Lucia 1999 The Fine Structure of Polarity Items, Garland, New York.

Index absolutive (case), 11, 90, 91, 93, 94 accusative, 176, 177, 178, 181 active referents, 132 active-stative languages, 20 adjunct, 94, 95, 105, 107 affected polarity items, 197 afterthought, 29–30 agent focus (Mayan), 142, 165 fn. 5 agglutinative, 87, 88, 102, 109 AGR (position), 68 Agree, 13 animacy scale, 68 antitopic, 12, 30, 117, 129, 131 applicative, 17, 31–32 areal diffusion, 19, 33 Binding theory, 74 Hornstein’s Binding theory, 74–5 borrowing, 18 bo-words, 174, 181, 186, 188 Brazilian Portuguese, 10 Capanahua, 69, 76 Carib, 87, 88, 109 fn. 3, 110 fn. 8, 111 fn. 11 cartography, 6, 7, 11, 87, 102, 107, 171, 173, 183, 188 case, 9, 17, 20–25, 26–30 case assignment, 6, 11 categorical judgments, 160 causative, 31 Chinese, 76–8 cleft(s), 6, 11, 32–33, 179 clitics, 26, 28, 32–33 subject clitics, 43, 117, 120, 123–126 resumptive, 173, 179, 186 second position, 119, 119–121, 123–126 comment, 39, 44, 59 fn. 3 communicative dominance, 46–51,

56, 59 fn. 6 complementizer, 87, 149, 155, 172, 173, 181 phrase (CP), 172–174, 180, 181, 183, 184, 188 CP-layer, 5 C-level operator, 13, 192 conditional, 87, 97, 98, 104 conditional operator, 13, 171, 180, 188 constituent order 24–26, 28–29, 33 (see also word order) constraint, 6, 10, 39, 41, 43, 46–47, 48, 49, 51–60, 61 fn. 10, 141, 143–144, 154, 156, 158 constraint-ranking, 3, 10, 48, 49, 50, 51, 52, 53, 56, 58, 68, 154, 159, 160 contrastive, 4, 9, 132 feature, 3 topics, 3 focus, 3, 4, 19, 25, 28, 30 cross-referencing, 9, 17, 24, 141 de-ergative, 98 definite, 12, 122–3, 128–129 definiteness, 21–22, 27, 28, 123, 144– 147, 166 fn. 9, 166 fn. 16 dependence, 171, 173, 179, 186 discourse, 1–8, 11, 12 domain, 5, 8 discourse-old, 12, 129 dislocation left-dislocation, 132 right-dislocation, 10, 12, 30, 117, 128–129, 132, 153 displacement, 2, 5 distinctness effects , 139, 155 Eastern Tucano, 9

214

Index

Ecuadorian Quechua, see Quechua, Ecuadorian elicitation-by-translation, 146–148 embedding, 172, 188 English, 2, 6, 8, 10 entry node, 46–58 epistemic, epistemic marker, 87, 96, 98, 108, 110 fn 8 epistemic modality, 91, 95, 97, 104, 110 fn 8 epistemic particle, 87, 91, 103, 106 epistemic value, 91, 93, 104 epistemological, 8 ergativity, 11, 88, 90, 91, 142, 165 fn 3, 165 fn. 4 de-ergative, 11, 92, 98, 106 ergative, 88, 90–92, 109 fn 3 ergative case, 91 evaluative mood, 8 evidentiality, 4, 5, 8, 10, 32, 191, 199 markers, 4, 8, 172, 182, 184, 199 mood, 8 exhaustive identification, 3 exogamy, 18 Extended Projection Principle (EPP), 74, 139, 152, 154 Finiteness Phrase, 5, 103, 109, 173, 180, 187–188 Finnish, 10, 65, 67, 68 focalization, 40–41, 56, 58, 60 fn. 14 focus, 1–6, 8, 9, 11, 17, 19, 21–25, 87, 92, 96, 98, 99–104, 106, 108–111, 115–117, 127–128, 132, 140–143, 171– 173, 178, 179, 183, 184, 186, 188, 191, 199, 203, 205, 206, 208 agent focus (Mayan), 142, 165 fn. 5 focussed subject, 20–21, 25, 26, 28–31 narrative focus, 45, 47

narrow focus, 127 new information, 1, 10, 179 potential domain of, 115 predicate focus, 127 sentence focus, 127 Focus Phrase, 5 Force Phrase, 5, 103, 173 Force-Fin system, 87, 103, 107, 109 free choice items (FCIs), 13, 191, 196, 209 fn. 11 functional projections, 7, 8, 11 ij-features, 75–77 German, 152 grammaticality judgment, 171, 175 grammaticalization, 27 Greek, 152, 154, 155, 162 head-final language, 11, 87, 88, 109 head-marking, 148 hearsay information, 4, 165 hearsay particle, 171, 182, 183, 188 Hebrew, 10, 65–69 Icelandic, 162 identifiable, 129 illocutionary force, 87, 92, 102, 104, 109 implicature, 4 indefinite (pronoun, root), 174, 176, 177, 184, 192 Information Structure, 1–5, 17–19, 39, 115–117, 127–132 instrumental marker, 30, 33 interface, 1, 87, 103, 173 interrogative, 171, 173–176, 178, 179, 183, 184, 185, 188 interrogative clauses, 91–92, 101, 147–148, 173, 176, 178, 180, 183, 185–186, 188 intonation , 2, 6, 96 intonational pattern, 2 inversion (subject inversion), 151, 166 fn. 11

Index 215 Italian, 155 Japanese, 152 Karaja, (Macro-Je) 13, 171–192 Korean, 8 Kuikuro , 11, 87–111 language contact, 17, 18, 26–29, 34 language endangerment, 18 last resort movement, 14, 74 Left-periphery , 5–6, 33, 87, 101–102, 105, 171, 173, 176, 178, 179 linear precedence, 6 linguistic area, 18 local licensing, 77 logical subject , 160, 162 logophoricity, 8, 9 Lushootseed (Salishan), 10, 39–63 Malayalam, 6 Maricopa, 9 Meaning-Text Theory, 9, 10, 39, 46, 59 fn. 1 Minimalism, 9 Mirror Principle, 7 mixed null subject languages, 10 multilingualism, 17–19 negative polarity items, 13, 191, 195, 197, 199, 199, 201, 202, 203, 205, 207, 209 nominal hierarchy, 23, 27 nominalization, 89, 92, 93, 98, 99, 101, 104, 105, 106, 110 non-veridicality operator, 13, 197, 201–204, 206, 208 nuclear pitch accent, 3 null-subjects mixed null-subject systems, 10, 65– 69, 81 mixed null-subject system of Hebrew, 66–68 mixed null-subject system of Shipibo, 69–70

null-subject parameter, 73 typology of null subjects, 75–7 Nuxalk (Bella Coola), 44 object direct, 43, 56, 59 fn. 1, 61 fn. 2 old information, 2 Optimality Theory, 6, 10, 12, 48, 68, 139, 152–164, 167 fn. 9, 167 fn. 19 Operator speech act, 197, 204 Q:wh, 13, 194, 199, 210 out-of-the-blue contexts, 143–144 Pano family, 10 parasitic category, 171, 173, 183 part of speech, 44, 45, 46, 53, 58 passive (voice), 43, 53, 54, 58, 60 fn. 10 person scale, 68 polarity sensitive items , 191, 192, 194, 197, 198, 201, 203, 208 Portuguese influence, 17, 19, 30, 32–33 postverbal position, 122–123, 128– 129, 148, 151 pragmatics pragmatic constraints, 116 pragmatic flexibility, 116–118 pragmatic principles, 115 pragmatic rigidity, 117–118 predicate, nominal, 43, 44, 56 non-verbal 45, 48 semantic, 40, 46–54, 60 fn. 7 syntactic, 43–44, 46–51, 55, 59 presupposed , 1, 2 preverbal position, 122, 142 pro, 3, 12 Projection Speech Act, 205– 207 Speech Act Participant Projection (SAPP), 68, 78–79

216

Index

pronouns, independent, 119 reduced, 121–122, 123–127, 131–132 resumptive, 13, 102, 106 weak, 3 proposition, 2, 3, 4, 5, 108, 160–162, 173, 201 Q: wh operator, see operator Quechua Bolivia, 194 Cuzco, 191, 194, 196, 208 fn. 1 Ecuadorian, 4 Southern, 191–208 questions, 44, 45, 47, 48, 50, 51, 53 wh (see wh-questions) yes/no (see yes/no questions) quotative device, 182 relative, clause, 5, 44, 45, 46, 98, 192, 193 free relatives, 191, 192, 196, 197, 204, 208 headless, 53, 57 operator, 147 rheme (rhematic), 1, 10, 41, 43, 46, 47, 48, 49, 50, 51, 52, 54, 56, 58, 61fn. 3, 61fn. 10, 62 fn. 14 Salishan, 10, 41–55, 57 Sherpa, 9 Shipibo, 10, 65–81 small clauses, 99, 105, 106 Spanish, 143, 152 Caribbean Spanish, 6 speech act, 8, 10, 13 speech act operator, see, operator speech act projection, see projection subjacency (condition), 176

subject, 1–6, 10, 11, 12, 42–45, 48, 52, 53, 55, 56, 58, 59 fn. 1, 59 fn. 2, 59 fn. 4, 66, 67 subject doubling (Capanahua), 76 subject-only clauses, 151–154, 156 switch-reference, 10, 71 syntax, 1, 4, 5, 7, 9, 10, 11 syntactic constraint, 116 syntactic movement, 173, 175, 176 syntactic principles, 115 syntactically rigid, 115–117, 127, 132 T-level operator, 13, 192, 208 tail, 153 Tariana, 9, 27–34 innovative Tariana 31–34 traditional Tariana 17–29, 30–4 tense-agreement domain, 5 text-synthesis, 39, 46 theme (thematic), 2, 10, 41, 44, 47, 50, 53, 56–58, 59 fn. 3, 60 fn. 14 thematic domain, 5 thetic judgments, 160–162 Tibetan, 9 topic, 1–7, 9, 10, 12, 17, 19, 21–33, 39, 43, 44, 47, 53–56, 59 fn. 3, 87, 98, 101–104, 107, 109, 119, 120, 127, 131–132, 141–145, 149–150, 157, 171, 172, 173, 178, 179, 180, 183–185, 186, 188 discourse topic, 3 sentence topic, 3 topical non-subject, 21, 22, 26– 28, 32 topicalization, 12, 101–104, 140, 142, 144–145, 149–150, 152, 157 Topic Phrase, 5, 8 transitive expletive constructions (TECs), 161, 166 fn. 16

Index 217 Trinidad Dialectal English, 6 Tucano, 17–19 Tzeltal (Mayan), 12, 149–150, 159 Tz’utujil (Mayan), 145–146 unaccusative, 90 unergative, 90 Universal Grammar, 172, 174 voice, 43, 46, 53–55 well-formedness conditions, 6, 7 Wh phenomena wh-feature, 174, 188 wh-fronting, 14, 192, 194, 205, 206 wh-in situ, 192, 194, 196, 209 fn. 7 wh-operator, 2, 13 wh-phrase, 171, 173, 176, 183, 184, 186 wh-questions, 2, 11, 91–96, 101, 105, 107, 173, 182, 191, 209 fn. 16 wh-word, 6, 13, 91, 92, 96, 101, 106, 174, 175, 178, 183, 184, 188, 191–208

word order, 2, 9–12, 17, 42, 102, 115–117, 118–127, 139–164, 175 OVS, 12, 121, 122, 129, 146 SOV, 12, 117, 118, 121, 123, 127, 128, 175, 191, 210 fn. 17 SVO, 12, 123–124, 129, 139– 164 VOS, 140, 141, 147–150, 153, 155–157, 160–163 split, 12, 139, 152–153, 264 unmarked, 139–145, 147–152, 154, 157, 158 Yaqui (Yuto-Aztecan), 11–12, 115– 138 yes/no interrogatives 183, 185, 186 yes/no questions, 11, 13, 87, 91, 95, 96, 97, 101, 106, 109, 178, 197, 203 (see also yes/no interrogatives) Yucatec Maya (Mayan), 12, 139–169