Focus Strategies in African Languages: The Interaction of Focus and Grammar in Niger-Congo and Afro-Asiatic

Focus Strategies in African Languages

≥

Trends in Linguistics Studies and Monographs 191

Editors

Walter Bisang (main editor for this volume)

Hans Henrich Hock Werner Winter

Mouton de Gruyter Berlin · New York

Focus Strategies in African Languages The Interaction of Focus and Grammar in Niger-Congo and Afro-Asiatic edited by

Enoch Olade´ Aboh Katharina Hartmann Malte Zimmermann

Mouton de Gruyter Berlin · New York

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter GmbH & Co. KG, Berlin.

앝 Printed on acid-free paper which falls within the guidelines 앪 of the ANSI to ensure permanence and durability.

Library of Congress Cataloging-in-Publication Data Focus strategies in African languages : the interaction of focus and grammar in Niger-Congo and Afro-Asiatic / edited by Enoch Olade´ Aboh, Katharina Hartmann, Malte Zimmermann. p. cm. ⫺ (Trends in linguistics studies and monographs ; 191) Includes bibliographical references and index. ISBN 978-3-11-019593-4 (alk. paper) 1. Niger-Congo languages ⫺ Grammar. 2. Afroasiatic languages ⫺ Grammar. 3. Focus (Linguistics) I. Aboh, Enoch Olade´. II. Hartmann, Katharina. III. Zimmermann, Malte, 1970⫺ PL8026.N44F63 2007 4961.36⫺dc22 2007042927

ISBN 978-3-11-019593-4 ISSN 1861-4302 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. ” Copyright 2007 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording or any information storage and retrieval system, without permission in writing from the publisher. Cover design: Christopher Schneider, Berlin. Printed in Germany.

Contents Focus and grammar: The contribution of African languages Enoch Oladé Aboh, Katharina Hartmann and Malte Zimmermann ........... 1 Part I.

Focus and prosody

Nuclear stress in Eastern Benue-Kwa (Niger-Congo) Victor Manfredi ......................................................................................... 15 Investigating prosodic focus marking in Northern Sotho Sabine Zerbian .......................................................................................... 55 Part II.

Information structure and word order

Preverbal objects and information structure in Benue-Congo Tom Güldemann ........................................................................................ 83 Focus strategies and the incremental development of semantic representations: Evidence from Bantu Lutz Marten .............................................................................................. 113 Part III.

Ex-situ and in-situ strategies of focus marking

Ex-situ focus in Kikuyu Florian Schwarz ....................................................................................... 139 Focus in the Force-Fin system: Information structure in Cushitic languages Mara Frascarelli and Annarita Puglielli ................................................. 161 Coptic relative tenses: The Profile of a morpho-syntactic flagging device Chris H. Reintges ..................................................................................... 185

vi Contents Part IV.

The inventory of focus marking devices

Identificational operation as a focus strategy in Byali Brigitte Reineke ........................................................................................ 223 Exhaustivity marking in Hausa: A reanalysis of the particle nee/cee Katharina Hartmann and Malte Zimmermann ........................................ 241 Part V.

Focus and related constructions

Narrative focus strategies in Gur and Kwa Anne Schwarz and Ines Fiedler ................................................................ 267 Focused versus non-focused wh-phrases Enoch Oladé Aboh ................................................................................... 287

List of contributors ................................................................................... 315 Subject index ............................................................................................ 317 Language index ........................................................................................ 323

Focus and grammar: The contribution of African languages Enoch Oladé Aboh, Katharina Hartmann, and Malte Zimmermann Starting with Chomsky (1971) and Jackendoff (1972), the formal study of information structure and grammar has received an ever increasing interest, and led to the formulation of various proposals centering on the expression of focus in natural languages and its function in the architecture of grammar (e.g. Dik 1981, Vallduvi 1990, Cinque 1993, Lambrecht 1994, Reinhart 1997, Drubig 2007). This book aims at contributing to the ongoing discussion of focus by investigating a range of African languages hitherto under-represented in the literature. The volume consists of a selection of articles that developed from presentations at a Workshop on “Topic and Focus: Information Structure and Grammar in African Languages” held at the University of Amsterdam in December 2004. Notwithstanding the wide range of approaches to focus found in the literature, the contributions to the present volume all agree on the following very general notion of focus as a pragmatic category that interacts with grammar (Jackendoff 1972, Dik 1981, Lambrecht 1994): (1)

Focus refers to that part of the clause that provides the most relevant or most salient information in a given discourse situation.

Typically, an expression will be most relevant or most salient if it is either new or contrasted with another element in the preceding or subsequent discourse. The non-focused part of a clause is often referred to as the background: (2)

Background refers to that part of the clause that contains the presupposed and/ or given information, where givenness implies having been mentioned in the preceding discourse.

In this volume, the background is also referred to as the out-of-focus part by some authors (e.g. Reineke, A. Schwarz & Fiedler).

2 Enoch Aboh, Katharina Hartmann and Malte Zimmermann The information-structural category of focus must be kept apart from the notion of focus marking, which refers to the overt realization of focus by special grammatical means, which is subject to cross-linguistic variation: Languages can mark focus syntactically, or prosodically, or morphologically, or they can use combinations of these grammatical means. At the same time, non-contrastive focus is not always marked, as shown by some of the languages discussed in the present volume. The articles in this volume look at focus strategies in a variety of African languages (Niger-Congo, Afro-Asiatic) from several theoretical and methodological perspectives. Their common aim is to deepen our understanding of how the information-structural category of focus is represented and marked in the languages of the world. This inquiry into the focus systems of African languages will have repercussions on existing theories of focus because it reveals new focus strategies as well as fine-tuned focus distinctions that are not discussed in the literature, which is almost exclusively based on intonation languages. Departing from the purely descriptive studies that we are familiar with when it comes to African languages, this book is – to our knowledge – unique in its effort to combine careful empirical study and theoretical analysis of the prosody, morphosyntax, and semantics of focus. Though every single paper in this volume addresses more than one issue in connection with the interaction of focus and grammar, the whole collection revolves about the following theoretical and empirical topics: (i) Focus and prosody; (ii) information structure and word order; (iii) ex-situ and in-situ strategies of focus marking; (iv) the inventory of focus marking devices; (v) focus and related constructions.

1.

Focus and prosody

Studies on intonation languages (e.g. Germanic, Romance) show that these use stress assignment (sometimes combined with syntactic transformations) for marking focus (e.g. Selkirk 1984, 1995). The underlying stress is typically expressed in form of a nuclear pitch accent, as shown for the following English examples (the focused constituent is underlined, the accented syllable is printed in capitals): (3)

a. Where did Peter buy the cassava? He bought them at the MARket.

Focus and grammar: The contribution of African languages 3

b. What did Peter buy at the market? He bought casSAva at the market. c. Who bought the cassava at the market? PEter bought them. In contrast, some African tone languages (e.g. from the Kwa, Gur, Bantu, and Chadic families) make use of syntactic transformations (e.g. fronting rules) for the same purpose. A plausible assumption is that the presence of tone in these languages somehow diminishes the prominence of intonation (or stress assignment). The following examples from Gungbe would support this view a priori: (4)

à n2a sà ? a. Ét( w(a 2sg HAB sell what FOC ‘What do you habitually sell?’ b. Hwèví w(a ùn n2a FOC 1sg HAB fish ‘I sell FISH.’

sà. sell

Things are not so clear-cut however. Kanerva (1990), for example, shows that focus marking in Nkhotakota Chichewa (Bantu) has an effect on the prosodic phrasing. In Chichewa, the right-edge of a prosodic phrase is indicated by penultimate lengthening and tone lowering on the phrase-final vowel, e.g. nyumbá > nyúumba in (5bc). The examples in (5a–c) show that the expression of focus affects the prosodic phrasing of the Chichewa clause in that a prosodic phrase boundary must be inserted after the focused constituent, i.e. after the VP in (5a), after the object NP nyúumba in (5b), and after the verb anaméenya in (5c), respectively: (5)

a. What did he do? (VP focus) (anaményá nyumbá ndí mwáála) he.hit house with rock ‘He hit the house with a rock.’ b. What did he hit with the rock? (object NP focus) (anaményá nyuúmba) (ndí mwáála) c. What did he do to the house with the rock? (V focus) (anaméenya) (nyuúmba) (ndí mwáála)

4 Enoch Aboh, Katharina Hartmann and Malte Zimmermann The examples in (5) show that certain tone languages can resort to prosodic devices for marking focus (see also Yip 2002), giving rise to the following questions: How do prosodic focus marking and lexical tone interact? To what extent is prosody inherent to the expression of focus? And do speakers always employ prosodic cues that are associated with the focused expression? The articles by Manfredi and Zerbian in this volume address these issues for a number of African tone languages and provide a closer inspection of the role of prosody in information packaging in these languages. The two articles take different starting points and come to radically different conclusions. Based on a careful experimental study of the prosodic properties of Northern Sotho, Zerbian shows that prosody plays no role in the marking of postverbal focus constituents in this language. Manfredi, in contrast, takes a universalist perspective as his starting point. According to his analysis, African tone languages do not differ significantly from intonation languages when it comes to focus marking. Universally, focus is marked by means of abstract stress (prominence), which is spelt out as a pitch accent in intonation languages and as a high tone in Kimatuumbi and Luhaya. It remains to be seen whether Manfredi’s analysis can be extended to Northern Sotho, and if so, how abstract stress is prosodically realized in this language.

2.

Information structure and word order

In many languages, information structure affects word order. Such interaction has been described for so-called discourse-configurational languages (É. Kiss 1995). In German, for instance, constituents expressing old information tend to precede the focus constituent in the middle field, sometimes leading to reordering of the basic word order (scrambling). For instance, the unmarked order of indirect object (IO) and direct object (DO) in the neutral clause in (6a) is IO > DO. However, when the indirect object is focused, it must follow the backgrounded direct object, as shown in (6b). Notice that this word order variation correlates with the definiteness/ indefiniteness distinction: (6)

a. Peter hat einem Mann ein Peter has a man a ‘Peter gave a book to a man.’

Buch book

gegeben. given

Focus and grammar: The contribution of African languages 5

b. Q: A1:

To whom did Peter give the book? Peter hat das Buch einem Peter has the book a A2:# Peter hat einem Mann das Peter has a man the

Mann man Buch book

gegeben. given gegeben given

Similar effects of information structure on word order are found, for instance, in Hungarian and in the Slavic languages. Even though some Bantu languages have been observed to exhibit discourse-configurationality as well (e.g. Bresnan & Mchombo 1987), the correlation between information structure and word order variation in African languages has not been subject to in-depth study. The articles by Marten and Güldemann in this volume are the first systematic studies in this domain: Marten gives an account of the various word orders in Bantu as the result of an interaction between syntax and pragmatics. Güldemann shows that the shift from SVO to SOV in a number of Bantu and Kwa languages is triggered by the information-structural status of the preposed object, comparable to scrambling processes in Germanic.

3.

Ex-situ and in-situ strategies

Certain African language families, such as Kwa, Bantu, and Chadic, mark focus syntactically by means of focus fronting (cf. e.g. Ameka 1992, Newman 2000). This ex-situ strategy was illustrated in (4) for Gungbe, but it also exists in Hausa, as illustrated by the question-answer pair in (7): (7)

a. Mè sukà kaamàà ? what 3pl.rel.perf catch ‘What did they catch?’ nèè sukà b. Dawaakii FOC 3pl.rel.perf horses ‘They caught HORses.’

kaamàà. catch

Depending on the context, though, the same question may also be answered by using an in-situ strategy in many of these languages: In the Hausa example (8), the focus constituent appears in-situ in its base position.

6 Enoch Aboh, Katharina Hartmann and Malte Zimmermann (8)

Sùn kaamà dawaakii. 3pl.perf catch horses ‘They caught HORses.’

In other words, these languages allow for two focus strategies (at least), one of which involves overt displacement and optional morphological marking. The existence of two structural focus strategies gives rise to two additional questions: Firstly, are there semantic/pragmatic differences between the two focus strategies? And if so, what are they? Secondly, do the ex-situ versus in-situ strategies comply with the new information versus contrastive focus dichotomy as proposed for certain intonation languages (e.g. É. Kiss 1998, Drubig 2001)? While the literature on certain African languages often deals with the focus strategy involving displacement and/or morphological marking, it is still unclear whether or how the in-situ strategy is marked, and what semantic effects arise with in-situ foci. Several articles in the volume address these questions to some extent: Concerning the formal marking of in-situ focus, Reintges shows that in-situ focus in Coptic is marked by means of special relative morphology, whereas Zerbian’s production and perception studies suggest that some languages (e.g. Northern Sotho) do not mark focus at all, neither in-situ nor ex-situ. With respect to interpretation, Aboh proposes that there is no semantic difference between the two focus positions. In the same vein, Hartmann & Zimmermann argue that the semantic notion of exhaustivity should not be linked to a particular syntactic position in Hausa, but results from the presence of an exhaustivity marker.

4.

Inventory of focus marking devices

The African languages under discussion exhibit a rich variety of focus strategies, sometimes even within one and the same language, and consequently the present book provides an extensive overview of possible focus marking devices. Apart from syntactic dislocation, which is attested in most languages investigated here, focus can also be marked morphologically by means of special focus markers or by a change in the aspectual system: The morphological strategy is found, for instance, in Byali (Reineke), Egyptian Coptic (Reintges), Somali (Frascarelli & Puglielli), Ewe, Akan, Lelemi, Buli, Dagbani (A. Schwarz & Fiedler), and Kikuyu (F. Schwarz).

Focus and grammar: The contribution of African languages 7

The observed variety of focus marking devices should not obscure the fact that many of the African languages discussed here exhibit a curious absence of grammatical focus marking, especially when compared to intonation languages, where sentences must have a pitch accent. In most of these languages, what has been referred to as new information focus (e.g. in answers to wh-questions) tends to be unmarked. The absence of such grammatical focus marking may constitute a serious challenge to the view that focus must be marked somehow (see e.g. Gundel 1999). At the same time, it might also turn out that tone and intonation interact in an intricate way to mask the marking of the in-situ focused constituent. To our mind, the articles in this volume contribute to a better understanding of such focus related phenomena and will pave the way to some refinements of existing focus theories.

5.

Focus and related constructions

Another issue addressed in this book is the way in which clauses containing an overtly marked focus expression, often informally referred to as focus constructions in the volume, are related to other discourse-related clause types, such as wh-questions, narratives, relative clauses, and clefts. A phenomenon that is often mentioned, but still not fully understood, is the formal identity between focus constructions and these other clause types. A case in point is the identical shape of the aspectual marker in focus constructions, wh-questions, and relative clauses (e.g., Hartmann & Zimmermann for Hausa, Reintges for Coptic). In addition, A. Schwarz & Fiedler observe a much less studied structural correlation between ex-situ focus and narrative structures in Kwa and Gur. Aboh discusses questionanswer pairs in Gungbe, showing that focused wh-questions and nonfocused wh-questions require different formal licensing as well as different information structuring of the answer. Finally, Frascarelli & Puglielli (Somali), and Reineke (Byali) point to well-known but still controversial structural identities between focus constructions and clefts, wh-questions, and equative sentences, respectively. As mentioned above, many of the articles in this volume address more than one of the theoretical and empirical issues outlined above. Nonetheless, it is possible to subsume each article under one of the five issues discussed above according to its main focus of interest. The volume thus consists of five independent but interrelated parts. Part I, Focus and prosody,

8 Enoch Aboh, Katharina Hartmann and Malte Zimmermann takes up the issue of whether and how focus is marked prosodically in African languages and contains the articles by Manfredi and Zerbian. Part II, Information structure and word order, deals with the influence of information structure on word order and contains the articles by Güldemann and Marten. Part III, Ex-situ and in-situ strategies of focus marking, examines syntactic focus marking strategies in more detail and contains the articles by F. Schwarz, Frascarelli & Puglielli, and Reintges. Part IV, The inventory of focus marking devices, deals with the question of which functional elements should be considered bona fide focus markers, and which ones should not. It contains the articles by Reineke and Hartmann & Zimmermann. Part V, Focus and related constructions, investigates the relationship of sentences containing a syntactically marked focus with other discourse-related clause types, such as wh-questions, relative clauses and clefts. It contains the articles by A. Schwarz & Fiedler and Aboh. We hope that this ordering will allow prospective readers an easier access to those articles reflecting their particular research interests. For ease of exposition, the maps at the end of the introduction provide an overview of the languages discussed in this volume, and where in Africa they are spoken. Map 1 gives a general overview. Map 2 gives a more detailed overview of the languages spoken in Nigeria and neighboring countries. Acknowledgments Work on the present volume would not have been possible without the generous financial support from the following institutions: The Amsterdam Center for Language and Communication (ACLC) and the Nederlandse Organisatie voor Wetenschappelijk Onderzeok (NWO), Vidi-grant 276–75– 003 (Enoch Oladé Aboh); and the Deutsche Forschungsgemeinschaft (DFG), and in particular the DFG-sponsored collaborative research centre SFB 632 “Information Structure”, located at Potsdam University and Humboldt University of Berlin (Katharina Hartmann, Malte Zimmermann). We would further like to thank Felix Ameka, Jeff Good, Nancy Chongo Kula, Roland Pfau, Harold Torrence, and an anonymous reviewer for helping us in the reviewing process. Special thanks are due to Lars Hermann and Anne Schwarz for their editing work in preparing the final manuscript, to Maria Höger for helping with the indexes, and to Lars Marstaller for his proof-reading.

Focus and grammar: The contribution of African languages 9

Map 1. Africa complete

10 Enoch Aboh, Katharina Hartmann and Malte Zimmermann

Map 2. Nigeria and neighboring countries

Focus and grammar: The contribution of African languages 11

References Ameka, Felix 1992 Focus Constructions in Ewe and Akan: A Comparative Perspective. MIT Working Papers in Linguistics 17, 1–25. Bresnan, Joan, and Sam Mchombo 1987 Topic, Pronoun, and Agreement in Chichewa. Language 4, 741–782. Chomsky, Noam 1971 Deep structure, surface structure and semantic interpretation. In Semantics: an interdisciplinary reader in linguistics, philosophy and psychology, Danny D. Steinberg and Leon A. Jacobovits (eds.), 183– 216. Cambridge: Cambridge University Press. Cinque, Guglielmo 1993 A null theory of phrase and compound stress. Linguistic Inquiry 24: 239–298. Dik, Simon et al. 1981 On the typology of focus phenomena. In Perspectives on functional grammar, Teun Hoekstra (ed.), 41–74. Dordrecht: Foris. Drubig, Bernhard 2001 Focus Constructions. In Language Typology and Language Universals. An International Handbook, Martin Haspelmath, Ekkehard König, Wulf Oesterreicher, and Wolfgang Raible (eds.), 1079–1104. Berlin: de Gruyter. 2007 Phases and the typology of focus constructions. In On Information Structure, Meaning and Form, Kerstin Schwabe and Susanne Winkler (eds.), 33–66. Amsterdam: John Benjamins. É. Kiss, Katalin 1995 Discourse Configurational Languages. Oxford: Oxford University Press. 1998 Identificational Focus Versus Information Focus. Language 74: 245– 273. Gundel, Jeanette K. 1999 On Different Kinds of Focus. In Focus: Linguistic, Cognitive, and Computational Perspectives, Peter Bosch and Rob van der Sandt (eds.), 293–305. Cambridge: Cambridge University Press. Jackendoff, Ray 1972 Semantic Interpretation in Generative Grammar. Cambridge, MA: MIT Press. Kanerva, Jonni 1990 Focusing on Phonological Phrases in Chichewa. In The PhonologySyntax Connection, Sharon Inkelas and Draga Zec (eds.), 145–161. Chicago: CSLI.

12 Enoch Aboh, Katharina Hartmann and Malte Zimmermann Lambrecht, Knud 1994 Information Structure and Sentence form. Cambridge: Cambridge University Press. Newman, Paul 2000 The Hausa Language. An Encyclopedic Reference Grammar. New Haven & London: Yale University Press. Reinhart, Tanya 1997 Interface Economy: Focus and Markedness. Studia Grammatica 40: 146–169. Selkirk, Elisabeth 1984 Phonology and Syntax. The Relation between Sound and Structure. Cambridge, MA: MIT Press. 1995 Sentence Prosody: Intonation, Stress, and Phrasing. In Handbook of Phonological Theory, John Goldsmith (ed.), 550–569. Oxford: Blackwell. Vallduví, Enric 1990 The Informational Component. Ph.D. dissertation, University of Pennsylvania, Philadelphia. Yip, Moira 2002 Tone. Cambridge: Cambridge University Press.

Part I Focus and prosody

Nuclear stress in eastern Benue-Kwa (Niger-Congo) Victor Manfredi

Abstract In ‘Bantu’ (eastern Benue-Kwa, [BK]), standard analyses of information structure assume autosegmental tonemes and templatic verb morphology (Odden 1984, Hyman 1999). Dispensing with both devices, inflectional tones reduce to phrasal accents (Bamba 1991, Idsardi & Purnell 1997), a VP constituent emerges from aux and clitic shells (Keach 1986, Myers 1998, Kinyalolo 2003a,b), and focus prosody appears Romance-like, with topics dislocated away from nuclear stress (Vallduví 1990, Zubizarreta 1998). The revised view enhances learnability, narrowing the set of possible languages while capturing language-internal generalizations.

1.

Caveat stressor

In generative studies of Germanic and Romance, the term nuclear stress refers to default phrasal accentuation which codes information focus (Chomsky 1971, 199f., Vallduví 1990, Zubizarreta 1998). The concept is not universally applied, thanks to a stack of gratuitous assumptions. (i) Prosodic phonology (Pierrehumbert 1980, Selkirk 1995) views accent as “non-phonemic prominence structure having to do with various pragmatic things people do” and thus as an unlikely source for “symbolic representations of phonological units that determine pitch differences” (D. Odden, p.c.). In English, however, nuclear stress is phonologically conditioned (Chomsky & Halle 1968, 17, Arregi 2002) and also shows “genuine ambiguity” between broad and narrow focus interpretations (Ladd 1996, 201, cf. Gussenhoven 1983b), therefore, accent can’t be excluded from derivations and a null theory of accent is the null hypothesis (Cinque 1993). (ii) Cartographic syntax (Brody 1990, Rizzi 1997) treats ex-situ focus as attraction to a formal/“criterial” feature with no intrinsic PF correlate. In Hungarian, however, subordinating main stress (the “Eradication Rule”) tracks focus movement (Kornai & Kálmán 1988, Szendröi 2003). (iii) Relativistic de-

16 Victor Manfredi scriptions of prosody (§1.1) and morphology (§1.2) underanalyze focus phenomena in Kimatuumbi (§2) and other ‘Bantu’ languages (§3). 1 1.1. Autosegmental luggage rack Tonemes – alphabetic representations of paradigmatic pitch contrasts (Pike 1948) – were ported into generative grammar as unscreened conceptual baggage three decades ago (Williams 1971, Leben 1973, Goldsmith 1976). Less than robustly detected in instrumental and perceptual experiments (Rouget 1965, LaVelle 1974, Hombert 1976, Abramson 1979, Gandour 1983, Connell 2000, Dilley 2005), tonemes have proven inseparable from metrical effects like interpolation, skewed distribution, and juncture (McCawley 1978, Abramson 1978, Akinlabip 1985, Sietsema 1989, Haraguchi 1991, Bamba 1991, Lapnipran 1992, Liberman et al. 1993, Liberman 1995, Idsardi & Purnell 1997). Despite such clues, tonemes are immune to disproof if accents are considered as just another tier in an “autosegmentalmetrical” array (Leben 1976, Halle & Vergnaud 1982, Goldsmith 1982, Beckman & Pierrehumbert 1986, Gussenhoven 2004). The main justification for tonemic autonomy is that phonological contrasts obey phonetic biuniqueness, as in Hyman’s slogan “tone = pitch features present lexically” (1989, 116) or in Odden’s sorting PF properties into “stress” versus “tone” while rejecting “shared formal machinery” despite “certain apparent similarities” (1999, 212). But biuniqueness loses generalizations in taxonomic discovery procedures (Halle 1959, 23), translation wordlists don’t reliably find minimal pairs, and language-internal puzzles result. (i) Emic tone demands ad hoc junctures. In Kimatuumbi, for example, Odden observes “tonal upsweep” – a gradual pitch rise from “L” to “H” across multi-syllables and a phrase boundary – and writes it as a string of H tonemes separated by [¡], the mark of “phonetic upstepping” (Pike & Wistrand 1974, Meir et al. 1975). But only “the final H in such a sequence is at the pitch level typical for an H-tone” (Odden 1996, 6), so to specify each middle syllable as H entails an absurd rule of leftward downdrift. 2 (ii) Many BK listemes divide into ‘tone classes’ – lexically latent pitch contrasts surfacing only in phrasal contexts – e.g. Gikuyu and Umbundu “nouns” (Benson 1964, xxi–xxviii, Clements & Ford 1979, Clements 1984, Schadeberg 1986) and Izgbo “verbs” (Swift et al. 1962, EpmepnanjoB 1981, Nwapchukwu 1983). Such effects neatly diagnose prelinked phrasal accent (Pulleyblank 1982, Déchaine 1993, Purnell 1998) but are more often handled as postlexical “tonal morphemes” (Sharman & Meeussen 1955,

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 17

Welmers 1959) or “floating/boundary/register” tones (Hyman 1974, 1985a, Clements 1981, Snider 1990). (iii) Tonemes obscure evidence of a phrase boundary after the BK “noun prefix” (Welmers 1973b) – an obligatory closed-class item displaying restricted tone and licensing a definite interpretation even in the absence of an overt demonstrative (Stahlke 1971, Manfredi 1993, 2003, 2004a, Déchaine & Manfredi 1998, Ajipbopyez 2005). Tonemes also overstate the prosodic diversity of natural language: their frequency is broadly complementary to suffixation and/or to the option of branching rimes (Manfredi 2003). Therefore if “a major goal for linguistic theory is to define the notion ‘possible language’” (Fromkin 1978, 1), it’s reasonable “to reinterpret autosegmental phonology as a special case of metrical phonology” (Leben 1982, 189). Accordingly, Duanmu eliminates underlying monomoraic tone contours in Chinese (1a), wiping out the initial plausibility argument for autosegmental notation (Goldsmith 1976, 21– 30), and arriving at the universal in (1b). In the same vein, consider West Germanic (2) and ‘Bantu’ (3).

(1)

(2)

(3)

Duanmu (1994, 567; 2005) a. “… stressless syllables either do not carry underlying tones… or will lose their underlying tones” b. “… while languages can differ in word stress, all languages have the same rule for compound and phrasal stress” Wagner (2005, 34, 273, 275) 3 a. “If A and B are sisters and A is the functor and B its argument, … B is [accentually] more prominent than A unless A already contains an argument …” (cf. Cinque 1993, 244). b. “Only associative domains are built in a single cycle. … Each cycle consists of a right-branching structure.” c. [An informationally given constituent] “becomes a functor” (cf. Steedman 2000). Hyman (1999, 153) a. “[–focus] o tonal integration = reduction” b. “[+focus] o tonal finality = end demarcation”

Despite the morphosyntactic distance between these languages, patterns (1) – (3) are empirically nondistinct. This convergence could be due to extrin-

18 Victor Manfredi sic factors (4a), or it could show that the faculté de langage handles stress and tone uniformly (4b). (4)

a. prosodic relativity: focus cues are incommensurable grammaticalized “strategies” (Vallduví & Engdahl 1996), interpreted beyond the semantic interface (Reinhart 1997) under a generic functional label of “prominence” (Truckenbrodt 1995). b. prosodic unity: all natural languages compute semanticallyrelevant prosody as phrasal accent (‘nuclear stress’).

(4a) is unfalsifiable (cf. Kaye 1988), both because it’s evaluated in a global output procedure (economy/OT) and because it admits any imaginable focus cue (pitch, duration, word order, morphological Gestalt…). 4 To limit indeterminacy, it’s sometimes suggested that a focus-sensitive feature can’t also support lexical contrasts – e.g. “… the phonetic cue to sentence accent is duration in Bantu languages” since “Bantu languages cannot use the prevalent component of pitch changes because of their tonal nature” (Zerbian 2005, 15) – but such an inference is false in general. In East Asian languages, lexical tone doesn’t block “parallel encoding” of focus information as F0 pitch (Xu & Xu 2005, cf. Potisuk et al 1996, Xu 2004), and in ‘nontonal’ Chimwiini-Kiswahili vowel duration is lexically contrastive as well as governed by focus-sensitive phrasing (Kisseberth 2002). Appeals to lexical tone are circular anyway: it’s “difficult to draw a dividing line between languages with contrastive tone on (almost) all syllables and languages with tone contrasts in more restricted locations in the word. Standard Chinese and Swedish are … both tone languages by this definition” (Gussenhoven 2004, 47). Eastern BK is equivocal in these terms, because lexical pitch is underspecified in all conceivable ways: paradigmatic (Meeussen 1963, Stevick 1969), syntagmatic (Voorhoeve 1973) and relative to syntactic category (Odden 1988, Kimenyi 2002). The literature does report one plausible instance where duration necessarily substitutes for pitch as a focus cue: in second-occurrence focus – however, this effect depends on configurational anaphora not lexical contrast sets and implicates metrical structure (Rooth 1996, 219) consistent with (2c). (4b), on the other hand, could be wrong but finds support beyond the studies cited under (1) – (3). In an elegant analysis of Spanish and Catalan, for example, verbs and nouns follow different stress rules reflecting the difference between high and low pronunciation of the root in its extended phrasal shell (Arregi & Oltra-Massuet 2005). To be sure, it’s not obvious

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 19

from inspection of BK citation forms that every (OCP) H tone recovers an accent, thus implying the presence of a local weak (non-H) position in keeping with metrics’ “relational” property (Liberman 1975, 51). It was not even obvious in English, to English-speaking linguists, just 60 years ago: Anticipating ToBI (Pierrehumbert 1980), Pike misses the phrasal distribution of stress (Bolinger 1951, Chomsky et al. 1956, 72) and sets up four “relative but significant levels (pitch phonemes)” (1945, 25), stringing them into intonational lines with a redundant word-level stress [°] for each “2” tone. 5

(5)

Pike (1945, 27–30, cf. Trager & Smith 1949) a. He wanted to do it. [°2–4] b. I want to go home. [3–°24] c. The boy in the house is eating peanuts rapidly. [3–°23–3–°23–3–°23–°23–°24]

In sum, the tonemic valise packs a load of warrantless assumptions plus a bundle of pitch features in search of analysis. In eastern BK, this search meets a second type of impediment. 1.2. The diacritical imperative Some Bantuists identify a focus slot right-adjacent to the finite verb (Watters 1979, Ndayiragije 1998, v.d. Wal 2005), but this can’t be the verb’s syntactic complement – as in Hyman’s (1985b) analysis of Aghem – if one recognizes a polysynthetic “verbal word” (Nurse 2003, 90, cf. Meeussen 1967, Goldsmith 1985, Hyman 2003), a templatic string of argument-type clitics, an aux, the lexical root plus derivational “extensions” (minimally, a default vowel), cf. (6a). Canonized in conjunctive orthography (Guthrie 1948), the template is reborn in generative theory (Baker 1988, 1996, Odden 1996, 71, 228f., Carstens 2001) and entails computational explosion: Odden (1981, 17) reportedly estimates “that some 16,000,000,000,000 distinct forms can be built around a single verb radical, not counting the differences induced by distinctive tonal features of various morphemes” (Sietsema 1989, 90). 6 Luckily, a range of evidence shows that the left edge of the verb root is separated by ordinary phrase boundaries from subject “agreement” (CLs ), which is “merged as a DP, an independent syntactic object” (Kinyalolo 2003b, cf. Keach 1986, Bresnan & Mchombo 1987, Myers 1998, Russell 1999, Kinyalolo 2003a), cf. (6b). 7

20 Victor Manfredi (6)

a. b.

[X° CLs–(aux)–(CLo)–verb root–extensions ] [DP+foc __ ] [TP [DP CLs] [T' (aux) [KP [DP (CLo) ] [VP [V° verb root– extensions] [DP+foc __ ] ] ] ] ]

The choice between (6a) and (6b) has a clear consequence for focus: (6b) allows a surface VP constituent, whereas under (6a) a VP must be simulated by stratally-ordered lexical phonology (Pulleyblank 1983, Odden 1996, 228ff.) plus a focus diacritic. This diacritic is actually embraced: “Bantu languages in particular are known for their ‘focus prominence’ … as when a tense is marked differently according to whether the verb is included in the focus or not. … Whether one is a syntactician or semanticist wishing to study focus or one is a phonologist wishing to study tone, one must consider all aspects of the grammatical system of a Bantu language. … [T]o not do so would be to risk drawing the tempting – but wrong – conclusion that there is a direct link between semantic focus and pitch in these languages.” (Hyman 1999, 151, 174, his italics)

Similarly, Odden rescues (6a) with a semantic parameter, making Kimatuumbi a language where “focal-sensitivity is a general property of an entire grammar” banning “two items focused in a clause” and deploying “morphological processes whose sole purpose is marking focus”; the existence of examples like “Who likes only meat? or Tom likes Sally” are said to show that such a ban “is not found in languages like English” (Odden 1984, 277, his italics, cf. 1996, 71). On second thought, however, these arguments for (6a) fail. Hyman’s rhetoric above is misdirected and overblown. Misdirected: It’s not the “syntactician or semanticist” but rather the prosodic phonologist (Pierrehumbert 1980, Gussenhoven 1983a) or interface economist (Reinhart 1997) who’s “tempted” to draw “a direct link between semantic focus and pitch”. Overblown: No reason is given why the road of indirectness must implicate “all aspects of the grammatical system of a Bantu language” (my italics) since just one grammatical feature (syntactic phrasing) may suffice. Odden’s English examples involve operator absorption not multiple independent foci (Krifka 1991), and an argument type expression modified by the Kimatuumbi item translating only, “whose sole purpose is marking focus” (cf. Rooth 1985), requires a “noun focal tense” (Odden 1984, 292). In any case, semantic parameters are doubtfully learnable (Gavarró et al 2005), so they are slim support for exotic morphosyntax. 8

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 21

The remaining task is to reconsider focus in eastern BK without tonemic and templatic distractions. Although I lack firsthand experience with these languages, the standard literature contains numerous helpful hints.

2.

Kimatuumbi

Odden holds that “there are no data in Kimatuumbi which help choose among competing definitions of focus” (1984, 278), but empirical differences do distinguish the diacritic/tonal (§2.1) and syntactic/accentual (§2.2) analyses. 2.1. H deletion, diminishing returns Odden generates Kimatuumbi verbal prosody by lexical H-tone assignment, followed by morphosyntactic tone mapping: “Every verb is assigned a floating H-tone, which is mapped … to the third vowel after the subject prefix in the subjunctive and participial; it is mapped to the second mora in dependent clauses; otherwise it is mapped to the rootinitial mora. … [T]o the extent that H is assigned to different moras in the stem, we have support for the independence of these processes” (Odden 1996, 191f.)

By this premise, every Kimatuumbi verb should end up with one H, but actual occurrences range from zero to two, and that’s where focus comes in. Sticking to main clause indicatives, verb forms are subclassified along a “three-way contrast in focal properties”: (7a/b) are “neutral”, (7c/d) “nounfocal”, and (8a–d) “verb-focal” (Odden 1984, 289, 295 fn), although all the forms in (8) are systematically ambiguous between focus on VP as a whole and narrow V. In accordance with the Bantuist template (6a), even the auxiliated, double-H forms in (8) count as “simple verbs” (Odden 1996, 71). 9 (7)

a. b. c. d.

CLs CLs CLs CLs

kalang-iBtee ñapma. a kaplang-iBtee ñapma. kalang-aa ñapma. kalang-a-ee ñapma.

‘… recently fried meat.’ ‘…fried meat.’ ‘…is frying meat.’ ‘…was frying meat.’

(8)

a. b. c. d.

CLs CLs CLs CLs

tiBp kalaang-apa ñapma. a-tipB kalaang-apa ñapma. endap kalaang-apa ñapma. ende-ep kalaang-apa ñapma.

‘…recently fried meat/fried meat.’ ‘…fried meat/fried meat.’ ‘…is frying meat/frying meat.’ ‘…was frying meat/frying meat.’

22 Victor Manfredi In retrospect, these data are less friendly to morphological focus than advertised. The “focus-neutral” status of (7a/b) – illustrated in (9) – (11) below – belies the notion that “focal-sensitivity is a general property of an entire grammar” (Odden 1984, 277). A three-valued focus feature overgenerates, providing no clue as to why Kimatuumbi lacks a “noun focal” past tense; why a “neutral” progressive specially requires periphrastic ‘be’; or why the “nine forms of the future [are] all focally neutral” (Odden 1996, 62f., my italics). 10 Barring accident, these gaps show that focus is not a morphological feature; rather, focus interpretations emerge as a compositional product of freestanding elements. Even taking the morphology at face value, if natural language has at least one generative engine – a syntax – and conceding that morphology can emulate at least some syntactic effects (Keenan & Stabler 2003), then for any given interpretation – including focus – a Kimatuumbi child needs to decide whether syntax or morphology is driving it. Odden admits that the motor is non-morphological in certain “neutral” forms which yield narrowly verb-focal readings even though unassisted by dedicated “verb focal” inflection (1996, 62; 1984, 281): 11 (9)

a. A tel-iBke tuBp. 3S cook-PERF only ‘S/he only cooked.’ b. ?Mamboondop a a M.H

kaplaang-iBte, a a

3S AUX H.fry-PERF

yapn-iBte

3S AUX H.forge-PERF

liBpiB-liBp. NEG-NEG.H

‘M. [something], he didn’t [something].’ From (9b), Odden reasonably concludes that “the focal properties of the verb are determined independently and are not a direct result of the selection of the verbal morphology” (1984, 281f., fn. 4). Logically, the next steps would be (i) to name these independent factors and (ii) to check whether the same factors also explain “noun-focal” and “verb-focal” effects. Attaining step (ii) would mean that morphological focus diacritics perform no indispensable work in Kimatuumbi. As to step (i), notice that both examples in (9) display object pro-drop. Assuming that this ellipsis is necessary in order to obtain narrow verb focus in “neutral” forms, it follows that focus is only accidentally narrow in (9) – just as expected for nuclear stress in a phonetically nonbranching VP. 12 In the same vein, consider a wider sample of focal interpretations which can contextually elicit a “neutral” form (Odden 1984, 280f.):

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 23

(10) a. A a kaplang-iBtee ñapma. ‘S/he .’ (How did s/he feed the children?) b. A a tepl-iBke kindooplo. ‘S/he cooked <sweet potato(es)>.’ (What did s/he cook?) c. KiBwiBiByop a a wiBpiB-le. ‘’ (Why are you crying?) d. A a wipB-le KiBwiBpiByo. ‘ died.’ (Who died?) e. A a kaplang-iBte Mambopondo. ‘<M.> fried [something].’ (Who fried [something]?) f. A a twep-tiB kiBndolop chaapngu. ‘S/he took <my> sweet potato.’ (Whose sweet potato did s/he take?) With direct object in-situ, the neutral form returns either broad VP focus (10a) or narrow focus on the object (10b) – a situation familiar from English. Narrow subject focus (10d/e) requires VS order, as in Italian (Antinucci & Cinque 1977, 124). 13 Focus on the possessor of the complement (10f) is also consistent with nuclear stress, assuming that the postnominal possessor of Kimatuumbi is c-commanded by its possessum. Not only does the “neutral” form have a semantics consistent with nuclear stress, it also has the phonetic earmarks. Apparently without exception, a main verb bears H in two circumstances: after an aux (7b, 8, 9b, 10) or phrase finally – even with an aux absent (11e). But with no aux, any clausemate phonetic material to the right of the verb, as in (11a–d), is sufficient to block H on the verb itself (Odden 1996, 62, 233, 287) – conforming to the first clause of (2a) above. 14 (11) a. A kalang-iBte yopopapta eepla. ‘S/he recently fried [something] to get money.’ b. A kat-iBte kaapmba. ‘S/he recently cut rope.’ c. A tel-iBke ñama tupB. ‘S/he recently cooked only meat.’ d. A tel-iBke Mambopondo. ‘M. recently cooked [something].’ e. Mamboondop a tepl-iBiBke. ‘M. recently cooked [something].’

24 Victor Manfredi As mentioned, Odden pre-syntactically assigns H to all verb forms, then maps this H to the left edge of the root in finite main clauses and lastly deletes the same H in case the verb is not clause-final (7a, 9a, 11a–d) by “Perfective Tone Loss” (1996, 233). Curiously, though, a deletion sensitive to the righthand context happens to fail just in case the root is introduced by an overt aux – a lefthand context exception which accommodates the tonal contrast between the “recent” (7a) and “remote” (7b) perfectives. The (7b) form shows that the absence of root H depends on a larger domain than [VO]. (7b) aside, the more general problem with the sequence of lexical H insertion followed by phrasal H deletion is that it is theory-internal: There can be no independent evidence for the deletion, because Kimatuumbi happens to be a language “lacking lexical tones in verbs, … similar to Safwa, Kinga, Makua, and Kikuria” (Odden 1996, 165, cf. Cheng & Kisseberth 1979, 31). The opposite conclusion – that the H in (7b, 11e) is epenthetic and phrasal, and therefore accentual in the normal meaning of the term – is compelled by an additional fact: H also shows up in unauxiliated progressives (forms which are not “perfectives”) when these are preceded by an object clitic (Odden 1996, 192), thus, the H-less forms in (7c/d) contrast minimally with (12a–b): 15 (12) a. CLs n teplek-y-a kiBndooplo. ‘… is cooking sweet potato(es) for her/him.’ [‘her/him’ = n] b. CLs niB teplek-y-a-e kiBndooplo. ‘… was cooking sweet potato(es) for me.’ [‘me’ = niB] To handle (7c/d) as well as (12), Odden suggests to “postulate a special rule deleting H from a mora that is both stem initial and superstem initial … in noun focal tenses” (1996, 195). 16 This deletion won’t occur in (12), because the H in (12) is not superstem-initial – Odden’s “superstem” being a sub-word constituent which includes the object clitic. Of course, this “special rule” is a mere restatement of the problem, but worse, it is inadequate: Due to the presence of the object clitic between the subject clitic and the root in (13), H deletion should fail to apply, but no verb form H is reported either in applicative (13a), in which direct object scrambling puts default focus on the applied object, or in causative (13b), where the “superstem”initial clitic receives a causee interpretation (Odden 1984, 74, 296): 17 (13) a. Ñamap niBiB-n kalang-y-a Mambopondo. ‘I’m frying meat for M.’ [‘M.’ = n]

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 25

b. TuBuB-n kalang-iBy-a LiBbuBlupBlee ñapma. ‘We’re making L. fry meat.’ [‘L.’ = n] The dim prospect of fiddling with “special” H-deletion so as to affect (13) without wrongly also applying in (12) shows that tonemic investment yields diminishing returns: Morphological, focus-sensitive tone rules lack generality because tone deletion lacks a coherent structural description, needing a new patch-up rule for every new context. 2.2. Clitics all the way up If tonemes are the lexical hypostasis of accent (§1.1 above), a natural alternative to H-deletion as an account of focus prosody is to turn the toneme ‘on its head’ and express the complement set of H-deletion contexts as accent-driven H-epenthesis. Doing this for Kimatuumbi, given the absence of lexical pitch contrasts on verb roots, leaves syntactic phrasing as the sole basis for verb accentuation in the language. As already discussed (§1.2), this is blocked by the morphological template (6a), as assumed by Odden (1996, 71, 228f), but it’s possible with the phrasal syntax of (6b) which admits a surface VP and so brings focus effects in reach of nuclear stress, as in (2). Tentative results can be glimpsed in diverse constructions. Unauxiliated progressives, being “noun-focal tenses”, necessarily “have no prepausal form in main clauses” (Odden 1996, 195), thus, the nuclear stress rule (2a) can’t explain H epenthesis in these forms – nor should it if extrinsic factors are responsible. 18 One such factor, seen in (12) above, is where stacking an object clitic between the subject clitic and the root evokes a lexically spurious H. Assume with Seidl (2001) that the object clitic is licensed in a KP shell whose null head doubles thematic material in VP: the applicative formative [ú ] ~ [y]. 19 Then the cross-root dependency between KP shell and applicative extension, flagged in (12' ) by coindexing, diagnoses a spellout domain, as in (2b): The lexical predicate can’t be construed with two structurally external arguments symmetrically, therefore, KP is accented. This accent is however realized neither on the object clitic nor on the null head of KP but on the nearest accentable syllable, which happens to be on the left edge of VP. Auxiliated, nonapplicative (7b) is parallel: The verb root is linearized between the overt aux [a] and the aspectual suffix [úte]; regardless of whether this suffix is an argument-type expression (Manfredi 2005b) or a predicate operator, its functional composition within TP is nonassociative, therefore by (2b), Asp inhabits a differ-

26 Victor Manfredi ent cycle from the subject/tense, so AspP is accented. Completing the parallel to the KP shell in (12' ), the AspP shell in (7' b) is phonetically subminimal – in this case null – and again, accent is realized on the nearest accentable material, namely the left edge of VP. 20

(12' )

TP

3 3KP g T 3 CL g SPEC 3VP ‡i g K 6 CL g

SPEC

‡j

(7' b )

teplek-yj-a(-ei) kindooplo

TP

3 SPEC 3AspP g T 3 VP g Asp 6 CL a g ‡i

kaplang-iBteei napma

The presence of VP-initial accent in (12) contrasts with its absence in (13), though the object clitic occurs throughout. This difference is beyond the expressive power of tonemic-templatic description, as pointed out above: No revision of the H-deletion rule can cover both (12) and (13) at once. Considered in phrasal terms, however, a differentia specifica does appear: For construction-particular reasons affecting the position of the direct object, both (13a) and (13b) display a discontinuous VP constituent, and the relevance of this fact to accentuation is consistent with (2a). 21 Another lexically spurious, phrasally-assigned H appears at the right edge of any XP before a conjoint/adjoint phrase or clause. The rule’s name, “Phrasal Tone Insertion” (Odden 1996, 234–38), tells the whole story: 22 (14) a. mpuBuBngap [ConjP na kiBndooplo]. ‘rice and sweet potato(es)’

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 27

b. mpuBuBnga ntepepenga uBp [ConjP naa ñapma ] ‘wet rice and meat’ c. Ñamap [TP a a liBpiB-le]. ‘S/he ate meat.’ (cf. ñama ‘meat’) d. MuBnduBuB ntokopBma uBp [TP a wipBiB-le]. ‘A sluggish person died.’ (cf. muBnduBuB ntokopBmauB ‘sluggish person’) A third type is the “verb-focal” paradigm, cf. (8) above. The phonetic generalization is that the inflected domain (aux plus verb) displays two pitch peaks (call them H1 and H2) whose distributions are patently phrasal. H1 occurs at the right edge of the (minimally CV) aux or aux stack. 23 H2 is found at the right edge of the lexical V° (extension included). The tonemictemplatic approach treats H2 as derivationally identical to the H that shows up immediately after the aux in “focally neutral” (7b); by Odden’s rule of Focal Flop, the first H gets “delinked” and “set afloat – later … [d]ocking … to the final syllable” of the verb (1996, 193). Viewed accentually, however, the root-initial H in (7b) is only superficially linearized on the verb but is generated on a later cycle. As to H2 in (8), its own appearance depends on the appearance of H1, moreover its position apparently at the right edge of VP draws semantic support from the fact in (8) that the object cannot bear narrow scope: What is in focus in (8) is either the verb by itself or the entire V+O sequence if any. It’s incorrect to describe the interpretation of (8) as “translated into English with contrastive stress on the verb” (Odden 1984, 279). That may be true in (15a), necessarily read with narrow scope on the verb, and it may indeed be “preferable to topicalize an object noun phrase when a verb-focal tense is selected” (Odden 1984, 296), but (15b/c) with the object not topicalized are both cited as felicitous answers to questions with broad VP scope (1984, 280, 290). A narrowing of focus to in examples like (15a) could be imposed contextually, easier than a widening from to . 24 (15) a. Ne endap ly-apa ñapma, ‘I’m <eating> meat.’ b. A a tiBp telek-ap kindooplo. ‘S/he ’ c. E endap kalaang-apa ñapma. ‘S/he’s .’

(… not it) (What did s/he <do>?) (What is s/he <doing>?)

28 Victor Manfredi The hard question for accent is how the position of H at the right edge of V° tracks VP focus. It must be the case that the object is evacuated from VP without removing it from the focus domain, but this surprising inference is supported by several considerations. Phonetically, Odden wonders why “the final H-tone which derives via Focal Flop cannot undergo Retraction” onto a preceding double vowel (1996, 199), as expected in the tonemic framework, yielding something like the ungrammatical [*…kalaapng-aa …] in (8) or (15b). This failure of retraction is banal under the accentual assumption that the verb-final position of the H in (8) and (15) is already retracted from the complement, as expected for nuclear stress. Indeed, Kimatuumbi presents independent evidence for retraction in (8) and (15): the concomitant failure to shorten the vowel of the verb root, which normally “applies when some word follows within the VP (Odden 1996, 226). For example, the root ‘fry’ appears as [kalang] throughout (7) but as [kalaang] in (8) and (15) where the complement is overt, as well as in (16) and (17) where the complement is null. Note that (16) is not morphologically “verb focal” but rather “neutral”. 25 (16) a. CLs kaplaang-iBte. ‘… recently fried [something].’ b. CLs a kaplaang-iBte. ‘… fried [something].’ (17) a. CLs tiBp kalaang-ap. ‘… recently fried [something].’ b. CLs a tipB kalaang-ap. ‘… fried [something].’ “It is not clear how this exceptionality is to be handled” says Odden (1996, 227) when shortening fails before an overt object in (8) and (15). But the syllabic and tonal facts follow if the object is phrased external to VP in those cases, as required by the accentual hypothesis. As to interpretation, more Kimatuumbi data are needed, but consolation comes from a remarkably similar phenomenon in Italian. According to Cardinaletti, the logical object in (18) is deaccented in-situ, no less than in (19) where it is patently dislocated when doubled by a clitic. In both contexts, there is a “low pitch intonation contour … separated from the clause by an intonation break (signalled by a comma)” (2002, 30 fn 1). Despite the similar contour, Cardinaletti argues for a structural difference between (18) and (19). In (19b),

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 29

for example, the participle is XP-extractable by itself, but this extraction fails in (18b). Conversely, (18c) but not (19c) allows a quantified direct object (Cardinaletti 2003, 36ff.). 26 (18) a. Ho

già comprato, il giornale. already bought the newspaper ‘I’ve already bought, the newspaper.’ b. *Comprato, non ho, il giornale. NEG AUX.1S the newspaper bought c. Non ha invitato Gianni nessuno. NEG AUX.3S invited G. nobody ‘G. hasn’t invited anybody.’ AUX.1S

(19) a. L’ho già comprato, il giornale. already bought the newspaper 3S-AUX.1S ‘I’ve already bought it, the newspaper.’ b. Comprato, non l’ho, il giornale. NEG 3S-AUX.1S the newspaper bought ‘I haven’t bought the newspaper.’ c. *Non l’ha invitato Gianni, nessuno. NEG 3S-AUX.3S invited G. nobody In sum, both Kimatuumbi “verb focus” (8) and Italian emarginazione (18) display a right-peripheral direct object which is outside of nuclear stress but still within sentential nuclear scope – though it may be excluded from that scope by further operations including dislocation. §3 concludes this paper with some parallels elsewhere in eastern BK. 27

3.

“Disjoint” forms

Unlike Kimatuumbi (§2), most of eastern BK shows accentual prelinking (‘lexical tone’) in verb roots to some degree, but this difference doesn’t undermine the accentual treatment of focus as presented above. As illustrated in (20) from Byarushengo et al. (1976, 196, 199) and in (21) from Hyman (1999, 153, 155), word order in Luhaya correlates with the prosody of penultimate syllables. [HL] is found in the penults of certain argument type phrases just in case they are sentence-final (20, 21a), narrowly focused (20b), or clitic-doubled after the verb (20c/d, 21c), otherwise

30 Victor Manfredi the penult is simple [H] (20a, 21a/b). As for certain finite verbs, the penult is L (20a/b, 21a) unless an object clitic precedes, in which case the penult is [H] if the verb is in narrow focus (20c/d) or is followed by a nondoubled argument (21a), otherwise the penult is HL (21b/c). To highlight these alternations, I use a simplified transcription exploiting tone/length redundancy, writing [xx] for HL, [x] for H and no brackets for no H. 28 (20) a. Abak[a]zi ba bon’ omw[aa]na. women CL see child ‘The women see the child.’ [broad focus] b. Ba bon’ omw[aa]n’ abak[aa]zi. CL see child women ‘They see the chíld, the women.’ [narrow focus on direct object] c. Ba mu b[o]n’ abak[aa]zy’ omw[aa]na. CL CL see women child ‘They sée him, the women, the child.’ [narrow focus on the verb] d. Ba mu b[o]n’ omw[aa]n’ abak[aa]zi. CL CL see child women ‘They sée him, the child, the women.’ [narrow focus on the verb] (21) a. Abak[a]zi ni ba bal[i]la omw[a]na emb[uu]zi. ASP CLs counted.for women child goats ‘The women are counting the goats for the child.’ [broad focus] b. Abak[a]zi omw[a]na emb[u]zi ni ba zi mu bal[ii]la. women child goats ASP CLs CLo CLo counted.for ‘The women, the child, the goats, they are counting them for him.’ c. Ni ba zi mu bal[ii]la abak[aa]zi omw[aa]na emb[uu]zi. ASP CLs CLo CLo counted.for women child goats ‘They are counting them for him, the women, the child, the goats.’ Hyman maps H tonemes to surface distributions by autosegmental spreading and deletion, filling in L tonemes. A rule changing penult [H] to [HL] is triggered before %, a diacritic that “marks assertive focus, i.e. the end of an assertion” (1999, 154). This entails that examples (21a/b), each with a single [HL] contour, count as simplex assertions, but that (21c) with a total of four such contours contains four % diacritics and therefore “four completed assertions” in one sentence (1999, 156). While saving the diacritic approach to focus (§1.2) and the tonemic hypothesis (§1.1), Hyman also tacitly rejects a treatment of Luhaya % as a clause boundary, i.e. a formal ob-

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 31

ject with only indirect, syntactically-mediated impact on focus interpretation. This nondiacritic view assimilates a Luhaya ‘noun’ after a % to clitic “right dislocation” in Romance languages, i.e. as a strategy that “allows speakers to remove an argument, or other bit of information from within the scope of assertion” (Byarushengo et al. 1976, 197, crediting Francesco Antinucci p.c.). 29 The same view, according to which “the effect of the % boundary is to ‘defocus’ non-asserted information” (Byarushengo et al. 1976, 198), is close to the syntactic treatment of Kimatuumbi in §2.2 above, modulo the extra option of prelinked H in certain verb roots as well as the richer, triple contrast of syllable prosodies (HL, H, ‡). Both considerations favor the accentual treatment of H: A phrasal regime of nuclear stress expects the nonparsing of H in an unperturbed, broad focus VP lacking object clitics (20a/b), as well as the positional “enhancement” (Halle & Vergnaud 1987, 37) of H to [HL]. Neither phenomenon fits in a phonemic framework: The former requires an arbitrary deletion rule, and the latter entails mysterious action-at-a-distance from the % trigger. 30 Based on available published data, it’s hard to argue directly against Hyman’s (1999) theory-internal claim of “four completed assertions” in (21c). A sharper problem for the claim is posed by narrow focus data like (20b–d): The most natural interpretation of the English glosses allows one assertion per example despite the presence of multiple [HL] contours, so either the glosses or the multiple focus analysis must be wrong. 31 Conversely, the freedom of post-verb word order in (20c/d) can be taken as evidence for the nondiacritic 1976 treatment of % – as a sentence boundary lacking intrinsic semantic content – in view of the parallel word order freedom in Spanish examples like (23), which has been argued to diagnose right dislocation (Zubizarreta 1998, 156). 32 (22) a. Le envió un regalo, María, a CL sent a gift M. to ‘She sent her a present, Maria, to Mom.’ Mamá, b. Le envió un regalo, a CL sent a gift to mom ‘She sent her a present, to Mom, Maria.’

Mamá. mom María. M.

Another reason not to build semantic assertion into the % boundary is that real semantic assertion is transparently implicated in a different Luhaya contrast, discussed by Hyman (1999, 160–62) under the rubric of “conjoint/disjoint” phrasing of post-verb material (cf. Schadeberg 2004). A

32 Victor Manfredi relevant pattern in both languages is that lexically prelinked H is “suppressed” in environments of what Hyman & Watters (1984) call “auxiliary focus”. Because this distribution differs from the suppression of verb root H in (20) and (21), conditioned by % and tied by hypothesis to a focus diacritic, Hyman is obliged to introduce “a secondary focus that has become morphologized and which, therefore, only imperfectly corresponds to the semantics that motivates it” (1999, 162). Assuming with Hyman that both phenomena are indeed focus-related, his dilemma can be blamed on use of a focus diacritic which has at most two values, minus (3a) and plus (3b). An indirect analysis which reads focus from syntax naturally admits more than one kind of focus related phrasing. In Kirundi, as first observed by Meeussen (1959, 119–28), the “disjoint” aux ra is necessary for the appearance of H in the lexically accented root ‘pick’ in (23a). Absence of ra suppresses the H (23b). The prosodic contrast is missing with a lexically unaccented root, as in (24) from Ndayiragije (1998), leaving auxiliation as the only cue. The correlation of auxiliation and lexical H in (23) follows from nuclear stress if the aux and VP form separate cycles, as is guaranteed by the pleonastic character of ra, which would otherwise be blocked by economy. As to interpretation, the presence of ra allows either the verb root alone or the whole VP to constitute new information (24a), whereas a narrow information focus on the object, as in a content question (24b), entails ra’s absence. The correlation is 100%, because the ra auxiliary is pleonastic apart from focus considerations, much like affirmative (i.e stressed, declarative) English do, as helpfully hinted by Ndayiragije’s translation of (24a). (23) a. N-ra áam-uur-a intore. 1S-ra pick-EXT-V plum ‘I’m picking plums.’ [disjoint] b. N aam-uur-a intore. plum 1S pick-EXT-V ‘I’m picking the plum.’ [conjoint] (24) a. Yuvinari a-á-ra somye ibitabo. Y. a-a-ra read books ‘Y. read/did read books.’ [disjoint] b. Yuvinari a-á (*-ra) somye iki? Y. a-a-ra read what ‘What did Y. read?’ [conjoint]

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 33

According to Goldsmith, Kirundi’s “-ra- Focus marker … is itself a recent innovation shared with Kinyarwanda but no other languages (a reanalysis of what was formerly the present tense marker, presumably)” (1985, 127). Even so, the focus effect of auxiliation is not limited to this pleonastic form, as shown by Setswana. Literal pause is not necessary after a disjoint verb (Chebanne et al. 1997, 56), but disjoint phonetic traits remain prosodic in the broad sense: a language-particular mix of pitch/timing and affixation/auxiliation, both consistent with an analysis by phrasing. Creissels (1996, 110f.) reports systematic tonal minimal pairs in the Setswana present perfect (25) and negative nonpast (26). In the superimposed paradigms in (27), the tonal difference appears in tandem with auxilation, whether the aux in question is pleonastic (affirmative nonpast) or a contentful sentence operator (future). If the aux is substantive, the audible conjoint/disjoint distinction reduces to tone alone. In contexts with a nonroot character, including those labeled pluperfect or consecutive, the distinction is phonetically neutralized altogether, yielding radical ambiguity, (28). (25) a. Bap jep-lez lep boznep. 3P eat-PERF with 3PL ‘Theyi have eaten, even theyi.’ [disjoint] b. Bap jez-lep lep boznep. 3P eat-PERF with 3PL ‘Theyi have eaten with themj.’ [conjoint] (26) a. Gaz

bap bipn-ez lep boznep. 3P dance-V with 3PL ‘Theyi don’t dance/aren’t dancing, even theyi.’ [disjoint] b. Gaz bap bipn-ep lep boznep. NEG 3P dance-V with 3PL ‘Theyi don’t dance/aren’t dancing with themj.’ [conjoint] NEG

(27) a. Bap {az/tlapaz} bipn-ap lep boznep. AUX dance-V with 3PL 3P ‘Theyi {‡/will} dance, even theyi.’ [disjoint] bipn-az lep boznep. b. Bap {‡/tlapaz} AUX dance-V with 3PL 3P ‘Theyi {‡/will} dance with themj.’ [conjoint]

34 Victor Manfredi (28)

Bap nez bap bipn-nep lep PERF 3P dance-PERF with 3P a. ‘Theyi had danced, even theyi.’ [disjoint] b. ‘Theyi had danced with themj.’ [conjoint]

boznep. 3PL

Setswana differs from Kirundi semantically in that the Setswana disjoint form puts narrow focus on the verb excluding the direct object (Creissels 1996, 113f.). A similar difference even divides Kirundi from Kinyarwanda, where the ra form excludes an adverbial (29a) or a direct object (30a) from the domain of new information (Givón 1975b, 194; tone outside the aux not marked in the source); and where the ra form is also impossible in a negative or relative predicate (reported but not illustrated in the source). The non-ra form can’t be followed by a discourse-old (pronominalized or scrambled) object (31b), which the ra form allows (31a). (29) a. *Yohani y-à-rá koze vuuba/ mumusozi. Y. y-a-ra work fast/ in the village [disjoint] b. Yohani y-à koze vuuba/ mumusozi. Y. y-a work fast/ in the village ‘Y. worked fast/in the village.’ [conjoint] (30) a. *Yohani y-à-rá riiye Y. y-a-ra eat b. Yohani y-à riiye Y. y-a eat ‘Y. ate (a) fish.’ [conjoint]

iffi. fish [disjoint] iffi. fish

(31) a. Yohani y-à-rá yi-riiye (iffi). Y. y-a-ra CL-eat fish ‘Y. ate it (the fish).’ [disjoint] b. *Yohani y-à yi-riiye. CL-eat [conjoint] Y. y-a In Setswana, auxiliation is neither necessary nor sufficient for a disjoint form which is narrow: (25a) is disjoint with no aux, whereas the nonfuture variant of (27a) is indeed auxiliated but could also be conjoint depending on the tone (27b). In Kirundi by contrast, all disjoint forms described by Meeussen are auxiliated, with the further prosodic correlate noted above, and they are all broad. Crucially, in the Kirundi conjoint, the verb root is

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 35

deaccented – a robust correlate of old information (Williams 1997) – so the fact that it’s narrow is no surprise. Now, why is the Setswana conjoint broad? Prosody distinguishes two subcases. Assuming that accent appears as a [HL] pitch contour, an accented verb is broad conjoint after a ‘middle field’ aux – one which follows the subject clitic, as in (27b) – but narrow disjoint otherwise – if nothing separates the subject clitic and the verb root, as in (25), (26), and (28). In other words, focus projects from an accented verb iff an aux directly precedes. This generalization also holds in Kirundi, in fact trivially so because verb root accent (in this language, a simple H) is limited to auxiliated forms. Turning to Setswana sentences which lack a middle field aux, an accented verb is narrow/does not project, but this is not a comparative problem because Kirundi ra has no Setswana counterpart, for the independent reason already stated. The remaining case is an unstressed (and therefore nonauxiliated) verb in Kirundi, which is narrow conjoint; the corresponding form in Setswana is broad. Both languages share this principle of stress-to focus mapping: focus projects from a stressed verb to VP (is broad) only after a middle field aux. In the absence of either VP-internal stress or an aux, focus includes the object (is conjoint) in both languages. Thus far unexplained is the difference that neutral phrasing is broad in Setswana but narrow in Kirundi; tentatively, this could be related to the different accentuation of argument type phrases (‘nouns’), presumably an independent fact.

4.

Perspective

The foregoing shows that the choice to forego tonemes and Bantuist templates brings many BK prosodic phenomena under a general theory of phrase structure, a hypothesis of prosodic unity (4b), as well as a modular architecture of information structure (2c). Numerous language-particular consequences across BK and more widely remain to be drawn, e.g. the appearance of left-peripheral focus and “compulas”, i.e. copular focus markers (Manfredi 1987, 110, cf. Bergvall 1987).

Notes

This paper revises Manfredi (2004b, 2005a). Thanks to E. O. Aboh, ’SB. AdepsBoBlap, A. Akinlabip, A. Bachrach, M. Bamba, L. Brunetti, A. Cardi-

36 Victor Manfredi

1.

2.

3.

4.

5.

naletti, G. Cinque, R.-M. Déchaine, L. Domínguez, S. Duanmu, I. Fiedler, J. Frampton, G. Giusti, F. del Gobbo, M. Halle, K. Hartmann, A. Kimenyi, K. Kinyalolo, G. Longobardi, J. Lowenstamm, C. Mayr, A. Nevins, D. Odden, J. Rooryck, V. Samek-Lodovici, B. Reineke, T. Schadeberg, R. Schuh, A. Schwarz, A. Seidl, K. Szendröi, C. Upchezchupkwu, M. Wagner, J. v.d. Wal, M. Zimmermann and M.-L. Zubizarreta. The term Bantu is sharp ideology but fuzzy linguistics. In apartheid Zuid Afrika it was the state synonym for “black” – as in this parliamentary speech by Minister P.W. Botha in 1964 (quoted by McGreel 2006): “I am one of those who believe that there is no permanent home for even a section of the Bantu in the white area of South Africa and the destiny of South Africa depends on this essential point. If the principle of permanent residence for the black man in the area of the white is accepted then it is the beginning of the end of civilisation as we know it in this country.” Turning the tables, Sowetan students replied to Botha’s Bantustans with black consciousness (Biko 1972). The Bantu mystique persists in U.S. Afrocentrism (Holloway & Vass 1993) and also in linguistics, even though it is “impossible to draw a clear line between Bantu, however defined, and non-Bantu Niger-Congo” (Nurse & Philippson 2003, 5). In Neogrammarian terms, the nearest affiliation of any two ‘Bantu’ languages above the local cluster is the Benue-Kwa “dialect continuum” (Williamson & Blench 2000, 17f., 27, cf. Greenberg 1963, 39, Givón 1975a), also called Volta-Congo (Stewart 1976, 1994). Odden (p.c.) notes that “[l]exicality is not the real issue, though that is often a fact that makes people comfortable with the idea of something being phonological.” Indeed, and once ‘tone’ is phrasal, a reanalysis in terms of nuclear stress comes within reach. The “unless” clause of (2a), accentually weighting the predicate over the subject, can be dropped if the default focus operator is the middle field assertion head, Sigma (Piñón 1992, Surányi 2004, Manfredi 2004b, cf. Gleitman 1969, Laka 1990). A “non-associative domain”, which by (2b) must be multi-cyclic, is defined as one in which compositional order affects truth value. Bamba & Liberman state that in Manding (Niger-Congo) “(some of) the functions of English intonational focus are performed by explicit and ordinary morphological marking” (1999, 1) but then go on to note that this marking is homophonous with the non-negative copula – anything but “ordinary”! The tonemic format of (5) may have a sociological explanation: Pike “utilized the study of English intonation as one step in the teaching of [S.I.L.] students how to reduce tone languages to writing” (1947, 131), but susceptibility to commonsense Africanist analysis persisted half a century

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 37

6.

7.

8.

9.

later when, for example, Ladd defended “tonal targets” in English intonation based on “the existence of languages like Yoruzbap in which it is uncontroversial that the system of lexical tones is based on distinctive levels … Once we have such a theory, it is plausible to assume that it will apply to English or Dutch as well” (1996, 61). Guthrie exempts from the template “some of the languages of the extreme north-western part of the Bantu area” (1948, 24 fn. 2). Templatics and tonemics have a close affinity, e.g. in Izgbo (BK), Clark’s (1989) templatic, level ordering analysis – developing Goldsmith (1976) in tune with Clements & Goldsmith (1980) – is obliged to posit a phonemic “downstep” (an unpredictable tonemic juncture) whereas a syntactic, accentual analysis can derive this juncture from phrase boundaries, e.g. between aux and V or within the Genitive DP (Déchaine 1992, 1993, 497–520, Manfredi 1993, cf. Clark 1980). Odden (1996, 228f.) considers the possibility that narrow focus on V results from verb raising, i.e. if the X0 in (6a) = “Infl”, or alternatively from extraposition of [-focus] arguments. Neither kind of rephrasing is independently motivated, however. In (6b), the label V° is intended to cover the categorial conflations of “lexical syntax” à la Hale (1995). I can’t find a Kimatuumbi example with a ‘perfective’/focally ‘neutral’ verb and a noun phrase or PP modified by ‘only’. One may exist, given the situation with wh-expressions: Odden initially says that these “cannot appear with the verbfocal tense, but may appear only with the noun focal tense” (1984, 292) but later observes that “Intrinsically focused elements such as wh-words may appear in clauses containing a focally neutral verb” (1996, 62). The crucial case for focus-sensitivity would combine a neutral verb, a wh expression, and a distinct nominal or prepositional phrase modified by ‘only’. See also discussion of the data in (9) below. Both Hyman’s “focus prominence” and Odden’s “focus sensitivity” recall the functional parameter of subject prominence vs. topic prominence (Li & Thompson 1976). Throughout (7) and (8), the verb root is kala(a)ng ‘fry’ and the logical object ñapma ‘meat’; the remaining morphemes, glossable only with difficulty, are discussed in the text. Italics in the translations mark obligatory focus as described in the source. I’ve cosmetically enhanced Odden’s transcription by changing hyphen to wordspace between the aux and the lexical verb root and by bolding the inflectional H tones. The consistently lengthened final vowel of the verb throughout (7) and (8), spelled as a double letter, has no morphological import (Odden 1996, 253ff.), just reflecting a phonetic constraint on the left edge of bisyllabic nouns (e.g. ñama ‘meat’).

38 Victor Manfredi 10. The nine Kimatuumbi futures include both “periphrastic” forms and not. In the closely related language Makua, an effable focus distinction in the future is reported (Stucky 1979a, 363), but without illustration. 11. The question mark on (9b) denotes “acceptable, if not optimal” grammaticality (Odden 1984, 281). In the translation of (9b) and henceforth, mark actual scope elicited in context versus italics which indicate obligatory focus as described without explicit tests. In the published translation of (9b), I correct a typo : “…forget”. (9b)’s interlinear gloss notes phrasally assigned H tones on the subject, verb, and negation – without prejudging their status as tonemes versus accents. In the glosses, I mark person/number (e.g. 3S) rather than noun class. In examples (9)– (11), the subject clitic is [a], denoting a third person singular human subject which is traditionally called class 1 (Odden 1996, 34). Examples (7b, 8b, 9b, 10, 11) include an aux [a], labeled “remote” (Odden 1996, 57) and linearized immediately after the subject clitic. Thus, (9b, 10, 11) show homophony between the default/epenthetic aux and third singular human pro. Welmers (1973b) expresses scepticism about standard Bantuist tense labels denoting multiple degrees of remoteness. 12. A phrasal origin of focus in (9) is further supported by the appearance of the doubled, specifically phrase-final form of negation (lùú-lú ) in the second clause. A desideratum is defining the general freedom of object prodrop in this language. 13. In contrast to Italian as well as to Kimatuumbi, VS order denotes broad (“all new”) focus in Catalan (Vallduví 1990), and VOS is generally possible for narrow subject focus in Spanish (Zubizarreta 1998). Plausible sources of this variation – case assignment and clitic doubling – remain controversial in these languages. 14. The locution complement is used here rather than direct object in order to denote an argument phrase in immediate post-verb position, thus including the focused, inverted subjects in (10d/e). This accords with Cinque’s view that, for purposes of nuclear stress, a non-topical post-verb argument is more deeply embedded than the verb itself “on the recursive side”. Nevertheless, the configurational definition is distinguished from an edgebased theory by cases of right detachment; faced with these, the edgebased theory needs to proliferate prosodic boundaries, while the syntactic view can refer to phrasal discontinuity. 15. For me, epenthetic describes minimal overt material not drawn from the numeration. For example, prosodic unity (4b) claims that children know how to construct metrical grids ex nihilo although of course they can learn lexicalized accents. 16. On the same page, Odden floats a second, more technical solution without enthusiasm.

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 39 17. Both exceptional forms have two overt phrasal objects; I assume that scrambling of the direct object is obligatory in (13a). Focus on the applied object is implied in the description of (13a); nothing is said about focus in the causative (13b). Although relevant data are lacking, it is implied that the rule designed for (12) cannot be collapsed with “Perfective Tone Loss”, i.e. that even with an object clitic an unauxiliated recent past still lacks stem H, as in hypothetical (i), based on Odden (1996, 74, ex 112): (i) [CLs n telek-iB Mamboondo kiBndooplo.] [‘…recently cooked sweet potato(es) for M. (= n)’] 18. Apart from progressives, the source mentions no other “noun focal tenses”. 19. Cf. Kisseberth & Mmusi (1990) for similar facts in Setswana. The polysynthetic view of applicatives (Baker 1988b, c), saving templatic (6a), derives the “verb extension” as a lexical head raised from an abstract PP in a downward lexical cascade (Baker 1988a). Hyman (2003) reviews problems for applicative incorporation – problems eluded by the inverse architecture in (6b) and (12' ) where the applicative licensor is above, not below, its spellout position, consistent with upwardly exploded treatments of clausal superstructure (Rizzi 1997, Manzini & Savoia 1998, Cinque 1999). 20. In rightward H displacement forms like (7' b), Goldsmith marks the onsetless aux [a] as “a ‘post-High’ morpheme” (1985, 123). 21. Note the scopal difference between (12) and (13a). In lieu of a phrasing for (13), I offer the slogan flat syntax, flat prosody. 22. The phrase structure of conjunction is notoriously murky, but the two environments potentially form a natural class (Williams 1989). The similarity is underlined by the accentuation of the scrambled object in (14c). 23. “In general, for all verb-focused tenses there is an H-tone on the last mora of the tense-aspect prefix” (Odden 1996, 194). 24. For example, if the object in (15a) is anaphoric; Kenesei (2005) argues that narrowing is the only way to get narrow V focus. 25. Odden (1996, 225) gives examples in which descriptive shortening applies twice in one form: once to the root vowel and once to the complex of applicative extension plus tense suffix. However, the likely operation of vocalic epenthesis in the latter context, though little explored in Bantuist literature, places the theoretical relevance of shortening in doubt. Instrumental data may clarify the correlation between length and pitch dynamism along the lines of the Chinese literature referred to above. Odden speaks of shortening “providing a second argument for syntactic constituency in potentially unclear cases” (p. 233), hinting at acoustic unclarity. 26. Examples tweaked. The presence of subject focus in (18c) is not indicated in Cardinaletti’s data but is implied in descriptive remarks (2002, 32). Emarginazione is the closest Italian counterpart to Germanic “metrical in-

40 Victor Manfredi visibility” (Zubizarreta 1998, 49). Although the typical example in both languages is anaphoric/topical, the requirement is less stringent as shown by (18c) and by (i) from Cinque (1993, 255 fn. 19 citing G. Grewendorf p.c., cf. also Wagner 2005, 211 and references cited there): (i) dass der Árzt bereitwillig [FP jeden Patienten [untersúchte]]. that the doctor willingly every patient investigated ‘…that the doctor willingly examined every patient.’ In (18c), nontopical material in XP is deaccented, and in (i), the transitive verb fails to deaccent despite occupying final position. 27. My treatment of Kimatuumbi differs notationally from Pulleyblank’s (1982) use of a Halle & Vergnaud accentual grid plus tonemes. The similarity is that reference to accent lets Pulleyblank avoid rules which first assign tones lexically and then delete them in phrasal contexts. A substantive difference – ignored here – stems from Pulleyblank’s emphasis on nominal prosody. 28. Hyman (1999) abstracts away from automatic vowel elision and glide formation, both of which are marked as apostrophes by Byarushengo et al (1976), as in (20). Hyman’s glosses for (21c/d) contain the apparent typo “have counted” for ‘are counting’. 29. Right dislocated items were accordingly described as either “afterthoughts” or “recapitulation” (1976, 201f.). Similarly, Hyman & Byarushengo state that “the focus (or assertion) of a Haya utterance is placed last in a clause” (1984, 70, their italics). 30. Unitary formal treatment of phonetic [HL] has a precedent in eastern BK: Luganda (Hyman & Katamba 1993, 49). 31. Corresponding forms with narrow focus on a sentence final verb, like (i) from Byarushengo et al. (1976, 201), or a broad-focus sentence-final verb, like (ii) from Hyman & Byarushengo (1984, 94), contain no instances of [HL], but it is not clear how the 1999 analysis can avoid generating [HL] there, or on the finite verbs in (20c/d) for that matter, while still correctly producing [HL] contours on the verbs in broad focus sentences like (21 b/c). (Throughout, the verb root belongs to the prelinked H class.) (i) a. Kak[u]lw’ abak[a]zi ba mu b[o]na. (ii) Ba mu k[o]ma. K. women CL CL see CL CL tie.up ‘Kakulu, the women (they) sée him.’ ‘They tie him/her up.’ b. Abak[a]zi Kak[u]lu ba mu b[o]na. women K. CL CL see ‘The women, Kakulu, they sée him.’ 32. In (22), nuclear stress is transcribed as underlining.

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 41

References Abramson, Arthur S. 1978 Static and dynamic acoustic cues in distinctive tones. Language & Speech 21, 319–25. 1979 The noncategorical perception of tone categories in Thai. In Frontiers of speech communication research, Björn Lindblom and Sven Ohman (eds.), 127–134. London: Academic Press. Ajipbopyez, OB. 2005 Topics on Yoruzbap nominal expressions. Dissertation, University of British Columbia, Vancouver. Akinlabip, Akinbiyi M. 1985 Tonal underspecification and Yoruzbap tone. Dissertation, University of Izbazdazn. Antinucci, Francesco, and Guglielmo Cinque 1977 Sull’ordine delle parole in Italiano; l’emarginazione. Studi di grammatica italiana 6, 121–46. Arregi, Karlos 2002 Focus on Basque movements. Dissertation, M.I.T., Cambridge, MA. Arregi, Karlos, and Isabel Oltra-Massuet 2005 Stress-by-structure in Spanish. Linguistic Inquiry 36, 43–84. Baker, Mark 1988a Incorporation; a theory of grammatical-function changing. University of Chicago Press. 1988b The mirror principle and morphosyntactic explanation. Linguistic Inquiry 16, 373–415. 1989 Elements of a typology of applicatives in Bantu. In Current Approaches to African Linguistics 7, John P. Hutchison and Victor Manfredi (eds.), 111–24. Dordrecht: Foris. 1996 The Polysynthesis Parameter. Oxford University Press. Bamba, Moussa 1991 De l’interaction entre tons et accent. Dissertation, Université du Québec à Montréal. Bamba, Moussa, and Mark Liberman 1999 Focus in Manding. http://www.ldc.upenn.edu/myl/mawu_LSA_handout.pdf Beckman, Mary E., and Janet Pierrehumbert 1986 Intonational structure in Japanese and English. Phonology Yearbook 3, 255–309. Benincà, Paola 1988 L’ordine degli elementi della frase e le construzioni marcate. In Grande grammatica italiana della consultazione 1, Lorenzo Renzi (ed.), 115–94. Bologna: Il Mulino. Benson, Thomas G. (ed.)

42 Victor Manfredi 1964 Kikuyu-English Dictionary. Oxford University Press. Bergvall, Victoria L. 1987 Focus in Kikuyu and Universal Grammar. Dissertation, Harvard University. Biko, Steve 1972/78 Let’s talk about Bantustans. I Write What I Like; a selection of his writings edited with a personal memoir by Aelred Stubbs. 81–86. London: Bowerdean. Bolinger, Dwight 1951 Intonation; levels vs. configurations. Word 7, 199–210. Reprinted in Forms of English; accent, morpheme, order, 3–16. Cambridge, Mass.: Harvard University Press. Bresnan, Joan, and Sam A. Mchombo 1987 Topic, pronoun and agreement in Chichewa. Language 63, 741–82. Brody, Michael 1990 Remarks on the order of elements in the Hungarian focus field. In Approaches to Hungarian 3; Structures & Arguments, István Kenesei (ed.), 95–122. Szeged: JATE. Byarushengo, Ernest Rugwa, Larry M. Hyman, and Sarah Tenenbaum 1976 Tone, accent and assertion in Haya. In Southern California Occational Papers in Linguistics 5: Studies in Bantu Tonology, Larry M. Hyman (ed.), 183–205. Los Angeles: University of Southern California. Cardinaletti, Anna 2002 Against optional and null clitics; right dislocation vs. marginalization. Studia Linguistica 56, 29–57. Carstens, Vicki 2001 Multiple agreement and Case deletion; against M-incompleteness. Syntax 4, 147–63. Chebanne, A., Creissels, Denis, and Nkhwa, H. 1997 Tonal Morphology of the Setswana Verb. München: Lincom Europa. Cheng, Chin-Chuan, and Charles W. Kisseberth 1979 Ikorovere Makua tonology (part 1). Studies in the Linguistic Sciences 9, 31–63. 1981 Ikorovere Makua tonology (part 3). Studies in the Linguistic Sciences 11, 181–202. Chomsky, Noam 1971 Deep structure, surface structure and semantic interpretation. In Semantics; an interdisciplinty reader in linguistics, philosophy & psychology, Danny D. Steinberg and Leon A. Jakobovits (eds.), 183– 216. Cambridge University Press. Chomsky, Noam, and Morris Halle 1968 The Sound Pattern of English. New York: Harper.

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 43 Chomsky, Noam, Morris Halle, and Fred Lukoff 1956 On accent and juncture in English. In For Roman Jakobson, Morris Halle, Horace G. Lunt, Hugh McLean, Cornelis H. van Schooneveld (eds.), 64–80. The Hague: Mouton. Cinque, Guglielmo 1991 Types of A’-Dependencies. Cambridge, Mass.: MIT Press. 1993 A null theory of phrase and compound stress. Linguistic Inquiry 24, 239–98. 1999 Adverbs & Functional Heads, a cross-linguistic perspective. Oxford University Press. Clark, Mary M. 1980 On the treatment of syntactically-distributed downstep. Studies in African Linguistics 11, 101–37. 1989 The Tonal System of Izgbo. Dordrecht: Foris. Clements, George N. 1981/83 The hierarchical representation of tone features. In Harvard Studies in Phonology 2, George N. Clements (ed.), 50–108/ In Current Approaches to African Linguistics 1, Ivan R. Dihoff (ed.), 145–76. Dordrecht: Foris. 1984 Principles of tone assignment in Kikuyu. In Autosegmental Studies in Bantu Tone, George N. Clements and John Goldsmith, 281–339. Dordrecht: Foris. Clements, George N., and Kevin Ford 1979 Kikuyu tone shift and its synchronic consequences. Linguistic Inquiry 10, 179–210. Clements, George N., and John Goldsmith 1980 What is downstep? A reply to Clark [1980]. Studies in African Linguistics 11, 239–54. Connell, Bruce A. 2000 The perception of lexical tone in Mambila. Language & Speech 43, 163–82. Creissels, Denis 1994 La tonalité des finales verbales et la distinction entre formes verbales conjointes et formes disjointes en tswana. Africana linguistica 11 (= Annales du Musée Royale de l’Afrique Centrale, Sciences humaines 142), 27–45. 1996 Conjunctive and disjunctive verb forms in Setswana. South African Journal of African Languages 16, 109–15. Déchaine, Rose-Marie 1992 Inflection in Izgbo and Yoruzbap. MIT Working Papers in Linguistics 17, 95–120. 1993 Predicates across categories; towards a category-neutral syntax. Dissertation, UMass, Amherst.

44 Victor Manfredi 2001 On the left edge of Yoruzbap complements. Lingua 111, 81–130. Déchaine, Rose-Marie, and Victor Manfredi SVO ergativity and abstract ergativity. Recherches Linguistiques de 1998 Vincennes 27, 71–94. Dilley, Laura 2005 The phonetics & phonology of tonal systems. Dissertation, M.I.T., Cambridge, Mass. Duanmu, San 1994 Against contour tone units. Linguistic Inquiry 25, 555–608. 2004 Tone and non-tone languages; an alternative to language typology and parameters. Language & Linguistics 5, 891–923. <wwwpersonal.umich.edu/~duanmu/ToneTyp04.pdf>. 2005 The tone-syntax interface in Chinese; some recent controversies. In Cross-Linguistic Studies of Tonal Phenomena, Historical Development, Tone-Syntax Interface & Descriptive Studies, Shigeki Kaji (ed.), Institute for the Study of Languages & Cultures of Asia & Africa, Tokyo University of Foreign Studies. <wwwpersonal.umich.edu/~duanmu/ToneSyntax05.pdf>. EpmepnanjoB, E. ’N. 1981/85 Auxiliaries in Izgbo Syntax. Dissertation, University of Izbazdazn/ Indiana University Linguistics Club, Bloomington. Fromkin, Victoria 1978 Introduction. In Tone; a linguistic survey, Victoria Fromkin (ed.), 1– 4. New York: Academic Press. Gandour, Jack 1983 Tone perception in far eastern languages. Journal of Phonetics 11, 149–75. Gavarró, Anna, Ana Teresa Pérez-Leroux, and Thomas Roeper 2005/ to appear Definite and bare noun contrasts in child Catalan. In The Romance Turn, Vicenç Torrens and Linda Escobar (eds.), Amsterdam: Benjamins. Givón, Talmy 1975a Serial verbs and syntactic change: Niger-Congo. In Word order and word order change, Charles N. Li (ed.), 49–112. Austin/ London: University of Texas Press. 1975b Focus and the scope of assertion. Studies in African Linguistics 6, 185–205. Gleitman, Lila R. 1969 Coordinating conjunctions in English. In Modern Studies in English: Readings in Transformational Grammar, David A. Reibel and Sanford A. Schane (eds.), 80–112. Englewood Cliffs, New Jersey: Prentice-Hall.

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 45 Goldsmith, John 1975 Tone melodies and the autosegment. OSU Working Papers in Linguistics 20, 135–47. 1976 Autosegmental phonology. Dissertation, M.I.T., Cambridge, Mass. 1978/81 English as a tone language. Communication & Cognition 11, 453– 76. /In Phonology in the 1980’s, Didier L. Goyvaerts (ed.), 287– 308. Ghent: Story-Scientia. 1982 Accent systems. In The Structure of Phonological Representations Part 1, Harry van der Hulst and Norval Smith (eds.), 47–63. Dordrecht: Foris. 1984 Tone and accent in Tonga. In Autosegmental Studies in Bantu Tone, George N. Clements and John Goldsmith (eds.), 19–51. Dordrecht: Foris. 1985 Bantu –a–; the far past in the far past. Studies in African Linguistics Supplement 9, 123–34. 1987 The rise of rhythmic structure in Bantu. In Phonologica 1984. Proceedings of the Fifth International Phonology Meeting, Wolfgang U. Dressler, Hans C. Luschützky, Oscar E. Pfeiffer, and John R. Rennison (eds.), 65–78. Cambridge University Press. Greenberg, Joseph H. 1963 The Languages of Africa. The Hague: Mouton. [= International Journal of American Linguistics 29.1]. Gussenhoven, Carlos 1983a Focus, mode and the nucleus. Journal of Linguistics 19, 377–417. [= Gussenhoven 1984, 11–62]. 1983b Testing the reality of focus domains. Language & Speech 26, 61–80. [= Gussenhoven 1984, 151–75]. 1984 On the Grammar & Semantics of Sentence Accents. Dordrecht: Foris. 2004 The Phonology of Tone & Intonation. Cambridge University Press. Guthrie, Malcolm 1948 Bantu Word Division; a new study of an old problem. [= International African Institute Memorandum 22.] Oxford University Press. Hale, Ken 1995 Universal grammar and the necessity of linguistic diversity. Presidential address, Linguistic Society of America, New Orleans. Published as: Universal grammar and the roots of linguistic diversity. MITWPL 28 (1996),137–61. Halle, Morris 1959 The Sound Pattern of Russian. The Hague: Mouton.

46 Victor Manfredi Halle, Morris, and Jean-Roger Vergnaud 1982 On the framework of autosegmental phonology. In The Structure of Phonological Representations Part 1, Harry van der Hulst and Norval Smith (eds.), 65–82. Dordrecht: Foris. 1987 An Essay on Stress. Cambridge, Mass.: MIT Press. Holloway, Joseph E., and Winifred K. Vass 1993 The African Heritage of American English. Bloomington: Indiana University Press. Hombert, Jean-Marie 1976 Perception of tones in bisyllabic nouns in Yoruzbap. Studies in African Linguistics Supplement 6, 109–21. Hyman, Larry M. 1974 The great Izgbo tone shift. In Third Annual Conference on African Linguistics, Erhard Voeltz (ed.), 111–25. Bloomington: Indiana University Press. 1985a Word domains and downstep in Bamiléké-Dschang. Phonology Yearbook 2, 47–83. 1985b Dependency relations in syntax; the mysterious case of the empty determiner in Aghem. Studies in African Linguistics Supplement 9, 151–56. 1989 Accent in Bantu; an appraisal. Studies in the Linguistic Sciences 19, 115–34. 1999 The interaction between focus and tone in Bantu. In The Grammar of Focus, Georges Rebuschi and Laurice Tuller (eds.), 151–77. Amsterdam/Philadelphia: Benjamins. 2003 Suffix ordering in Bantu; a morphocentric approach. In Yearbook of Morphology 2002, Geert Booij and Jaap van Marle (eds.), 245–81. Kluwer Academic Publishers. 2004 How to become a ‘Kwa’ verb. Journal of West African Languages 30, 69–88. Hyman, Larry M., and Ernest Rugwa Byarushengo 1984 A model of Haya tonology. In Autosegmental Studies in Bantu Tone, George N. Clements and John Goldsmith (eds.), 52–103, Dordrecht: Foris. Hyman, Larry M., and Francis X. Katamba 1993 A new approach to tone in Luganda. Language 69, 34–67. Hyman, Larry M., and John R. Watters 1984 Auxiliary focus. Studies in African Linguistics 15, 233–73. Idsardi, William J., and Thomas C. Purnell 1997 Metrical tone and the Elsewhere Condition. Rivista di Linguistica 9, 129–56.

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 47 Kaye, Jonathan 1988 The phonologist’s dilemma: a game-theoretic approach to phonological debate. GLOW Newsletter 21. [Barrett-]Keach, Camillia 1986 Word-internal evidence from Swahili for Aux/Infl. Linguistic Inquiry 17, 559–64. Keenan, Edward, and Edward Stabler 2003 Bare Grammar. Stanford, Calif.: CSLI. Kenesei, István 2005 Focus as identification. In The Architecture of Focus, Valéria Molnaar and Susanne Winkler (eds.), Berlin: Mouton. . Kisseberth, Charles W. 2002 Prosody and the phonology-syntax interface: Chimwiini revisited. M.I.T. Linguistics Colloquium, 15 February. Kisseberth, Charles W., and Sheila Mmusi 1990 The tonology of the object prefix in Setswana. Studies in the Linguistic Sciences 20, 151–61. Kornai, András, and László Kálmán 1988 Hungarian sentence intonation. In Autosegmental Studies on Pitch Accent, Harry van der Hulst and Norval Smith (eds.), 183–95. Dordrecht: Foris. Krifka, Manfred 1991 A compositional semantics for multiple focus constructions. SALT 1, 127–58. Ladd, D. Robert 1996 Intonational Phonology. Cambridge University Press. Laka, Itziar 1990 Negation in syntax; on the nature of functional categories & projections. Dissertation, M.I.T., Cambridge, Mass.

48 Victor Manfredi Lapnipran, Y. 1992

Intonation in tone languages; the phonetic implementation of tones in Yoruzbap. Dissertation, Cornell University, Ithaca, N.Y.

LaVelle, Carl 1974 An experimental study of Yoruzbap tone. Studies in African Linguistics Supplement 5, 185–204. Leben, William R. 1973 Suprasegmental phonology. Dissertation, M.I.T., Cambridge, Mass. 1976 The tones in English intonation. Linguistic Analysis 2, 69–107. 1982 Metrical or autosegmental. In The Structure of Phonological Representations Part 1, Harry van der Hulst and Norval Smith (eds.), 177– 90. Dordrecht: Foris. Li, Charles N., and Sandra A. Thompson 1976 Subject and topic; a new typology of language. In Subject and Topic, Charles N. Li. (ed.), 457–85. New York: Academic Press. Liberman, Mark Y. 1975 The intonational system of English. Dissertation, M.I.T., Cambridge, Mass. 1995 The sound structure of Mawu words. In Invitation to Cognitive Science, Lila R. Gleitman and Mark Y. Liberman (eds.), 55–86. Cambridge, Mass.: MIT Press. Liberman, Mark Y., J. Michael Schultz, Soonhyun Hong, and Vincent Okeke 1993 The phonetic interpretation of tone in Izgbo. Phonetica 50, 147–60. Manfredi, Victor 1987 Anti-logophoricity as domain extension in Izgbo and Yoruzbap. NigerCongo Syntax & Semantics 1, 97–113. African Studies Center, Boston University. 1993 Spreading and downstep; prosodic government in tone languages. In The Phonology of Tone; the Representation of Tonal Register, Harry van der Hulst and Keith Snider (eds.), 133–84. Berlin: De Gruyter. 1995 Tonally branching s in Yoruzbap is [LH]. Niger-Congo Syntax & Semantics 6, 171–82. African Studies Center, Boston University. 2003 A fonosyntactic parameter within Benue-Kwa and its consequences for EBzdop. In Typologie des langues d'Afrique et universaux de la grammaire, vol. 2: Benue-Kwa, Soninke, Wolof, Patrick Sauzet and Anne Zribi-Hertz (eds.), 127–62. Paris: Presses Universitaires de Vincennes/ Éditions de l’Harmattan. 2004a A toneless theory of 2-and-a-half tonemes in Gbez. UParis-7, 1 December. 2004b The prosodic infrastructure of focus in Benue-Kwa. Universiteit van Amsterdam, 4 December.

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 49 2005a

Conjoint/disjoint in western Benue-Kwa. Humboldt-Universität Berlin, 7 October; Leiden University Centre for Linguistics, 18 October. 2005b Aspect or the serialization parameter. Institute for African Studies, Universität Leipzig. 12 October. 2006 Nuclear stress in western Benue-Kwa. Dipartimento Scienze del Linguaggio, Università di Venezia Ca’ Foscari, 13 June; Workshop on Minority Languages, Università di Roma-3, 17 June; submitted to ZAS Working Papers in Linguistics, Berlin. Manzini, Maria Rita, and Leonardo M. Savoia 1998 Clitics and auxiliary choice in Italian dialects; their relevance for the person-ergativity split. Recherches Linguistiques de Vincennes 27, 115–37. McCawley, James 1970 Some tonal systems that come close to being pitch accent systems but don’t quite make it. Chicago Linguistic Society 6, 526–32. 1978 What is a tone language? In Tone; a linguistic survey, Victoria Fromkin (ed.), 113–31. New York: Academic Press. McGreal, Chris 2006 Brothers in arms; Israel’s secret pact with Pretoria. The Guardian [U.K.], 7 February. . Meeussen, Achille E. 1959 Éssai de grammaire rundi. Musée Royal du Congo Belge, Tervuren. 1963 Morphotonology of the Tonga verb. Journal of African Languages 2, 72–92. 1967 Bantu grammatical reconstructions. Africana linguistica 3, 80–122. Meier, Paul E., Ingeborg Meier, and John T. Bendor-Samuel 1975 A Grammar of Izzipiz, an Izgbo Language. Summer Institute of Linguistics/ Wycliffe Bible Translators, Normal, Oklahoma. Myers, Scott AUX in Bantu morphology and phonology. In Theoretical Aspects of 1998 Bantu Tone, Larry M. Hyman and Charles W. Kisseberth (eds.), 231–64. Stanford, Calif.: CSLI. Ndayiragije, Juvénal 1998 TP-internal focus in Kirundi and “Attract-F”. In Proceedings of Workshop on Focus, Elena Benedicto, Maribel Romero, and Satoshi Tomioka (eds.), 175–90. [= University of Massachusetts Occasional Papers in Linguistics 21.] Amherst, Mass.: G.L.S.A.

50 Victor Manfredi Nurse, Derek 2003 Aspect and tense in Bantu languages. In The Bantu languages, Derek Nurse and Gerard Philippson (eds.), 90–102. London: Routledge. Nurse, Derek, and Gerard Philippson 2003 Introduction. In The Bantu languages, Derek Nurse and Gerard Philippson (eds.), 1–12. London: Routledge. Odden, David 1981 Problems in tone assignment in Shona. Dissertation, University of Illinois, Urbana-Champaign. [Not personally consulted; cited by Sietsema 1989.] 1984 Formal correlates of focusing in Kimatuumbi. Studies in African Linguistics 15, 275–99. 1988 Predictable tone systems in Bantu. In Autosegmental Studies on Pitch Accent Systems, Harry van der Hulst and Norval Smith (eds.), 225–51. Dordrecht: Foris. 1996 The Phonology & Morphology of Kimatuumbi. Oxford University Press. 1999 Typological issues in tone and stress in Bantu. In Cross-linguistic Studies of Tonal Phenomena, Tonogenesis, Typology & Related Topics, Shigeki Kaji (ed.), 187–215. Institute for the Study of Languages & Cultures of Asia & Africa, Tokyo University of Foreign Studies. 2003 Rufiji-Ruvuma (N10, P10–20). In The Bantu languages, Derek Nurse and Gerard Philippson (eds.), 529–45. London: Routledge. Pierrehumbert, Janet 1980 The phonetics & phonology of English intonation. Dissertation, M.I.T., Cambridge, Mass. Pike, Eunice V., and Kent Wistrand 1974 Step-up terrace tone in Acatlan Mixtec. In Advances in Tagmemics, Ruth Brend (ed.), 83–104. Amsterdam: North-Holland. Pike, Kenneth L. 1945/47 The intonation of American English. Washingon, D.C.: Foreign Service Institute/ Ann Arbor: University of Michigan Press. 1948 Tone languages; a technique for determining the number & type of pitch contrasts in a language, with studies in tonemic substitution & fusion. Ann Arbor: University of Michigan Press. 1965 On the grammar of intonation. In Proceedings of the 5th International Congress of Phonetic Sciences (Münster), Eberhard Zwirner and Wolfgang Bethge (eds.), 105–19. Basel: Karger. Piñón, Christopher 1992 Heads in the focus field. In Approaches to Hungarian 4; the Structure of Hungarian, István Kenesei and Csaba Pléh (eds.), 99–121. Szeged: JATE.

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 51 Potisuk, Sirpong, Jack Gandour, and Mary P. Harper 1996 Acoustic correlates of stress in Thai. Phonetica 53, 200–20. Pulleyblank, Douglas 1982 Accent in Kimatuumbi. In Current Approaches to African Linguistics 2, Jonathan Kaye, Hilda Koopman, Dominique Sportiche, and André Dugas (eds.), 195–215. Dordrecht: Foris. 1983 Tone in lexical phonology. Dissertation, M.I.T., Cambridge, Mass. Purnell, Thomas C. Purnell 1998 Principles & parameters of phonological rules; evidence from tone languages. Dissertation, University of Delaware. Reinhart, Tanya 1997 Interface economy; focus and markedness. Studia Grammatica 40, 146–69. Rizzi, Luigi 1997 The fine structure of the left periphery. In Elements of Grammar: Handbook in generative syntax, Liliane Haegeman (ed.), 281–337. Dordrecht: Kluwer. Rizzi, Luigi, and Leonardo M. Savoia 1992 Conditions on /u/ propagation in southern Italian dialects; a locality parameter for fonosyntactic processes. In Syntactic Theory & the Dialects of Italy, Adriana Belletti (ed.), 252–318. Torino: Rosenberg & Sellier. Rooth, Mats 1985 Association with focus. Dissertation, UMass, Amherst. 1996 On the interface principles for intonation. SALT 6, 202–226. Rouget, Gilbert 1965 Analyse des tons du guzn par le «détecteur de mélodie» de l’institut de Phonétique de Grenoble; rapport d’expériences. Langage & comportement 1, 31–48/ 50, 3–29. Russell, Kevin 1999 What’s with all these long words anyway? MIT Occasional Papers in Linguistics 17, 119–130. Schadeberg, Thilo C. 1986 Tone cases in Umbundu. Africana linguistica 10, 423–47. 2004 Conjoint & disjoint; grammaticalized information structure in Bantu. Universiteit van Amsterdam, 3 December. Seidl, Amanda Hallie 2001 An Optimality-Theoretic account of applicative asymmetries and their phonological domains. ApplFest, Dept. of Linguistics, M.I.T., Cambridge, Mass, 28 January.

52 Victor Manfredi Selkirk, Elisabeth 1995 Sentence prosody; intonation, stress and phrasing. In The Handbook of Phonological Theory, John Goldsmith (ed.), 550–69. Oxford: Blackwell. Sharman, John C., and Achille E. Meeussen 1955 The representation of structural tones, with special reference to the tonal behavior of the verb, in Bemba, Northern Rhodesia. Africa 25, 393–404. Sietsema, Brian M. 1989 Metrical dependencies in tone assignment. Dissertation, M.I.T., Cambridge, Mass. Stahlke, Herbert 1971 The noun prefix in Ez8ez. Studies in African Linguistics Supplement 2, 141–59. Steedman, Mark 2000 Information structure and the syntax-phonology interface. Linguistic Inquiry 34, 649–89. Stevick, Earl W. 1969 Tone in Bantu. International Journal of American Linguistics 35, 330–41. Stewart, John M. 1976 Towards Volta-Congo reconstruction. Leiden: University Press. 1994 The comparative phonology of Gbå and its significance for that of Kwa and Volta-Congo. Journal of African Languages & Linguistics 15, 175–93. Stucky, Susan 1979a Focus of contrast aspects in Makua; syntactic and semantic evidence. Berkeley Linguistic Society 5, 362–72. 1979b The interaction of tone and focus in Makua. Journal of African Languages & Linguistics 1, 189–98. Surányi, Balazs 2004 The left periphery and cyclic spellout; the case of Hungarian. In Peripheries; syntactic edges & their effects, David Adger, Cecile de Cat, and George Tsoulash (eds.), 49–73, refs. 407–33. Dordrecht: Kluwer. Swift, Lloyd B., A. AzhapghiBpotuz, and E. UzgopBjiBp 1962 Izgbo Basic Course. Washington, D.C.: Foreign Service Institute. Szendröi, Kriszta 2003 A stress-based approach to the syntax of Hungarian focus. Linguistic Review 20, 37–78. Trager, George, and Henry Lee Smith 1949 An Outline of English Structure. Washington, D.C.: Foreign Service Institute.

Nuclear stress in eastern Benue-Kwa (Niger-Congo) 53 Truckenbrodt, Hubert 1995 Phonological phrases; their relation to syntax, focus & prominence. Dissertation, M.I.T., Cambridge, Mass. Vallduví, Enric 1990 The informational component. Dissertation, University of Pennsylvania, Philadelphia. Vallduví, Enric, and Elisabet Engdahl 1996 The linguistic realization of information packaging. Linguistics 34, 459–519. Voorhoeve, Jan 1973 Safwa as a restricted tone system. Studies in African Linguistics 4, 1–22. Voorhoeve, Jan, Achille E. Meeussen, and K. de Blois. 1969 New proposals for the description of the Izgbo completive phrase. Journal of West African Languages 6, 79–84. Wagner, Michael 2005 Prosody and recursion. Dissertation, M.I.T., Cambridge, Mass. Wal, Jenneke van der 2005 Prosody and focus in Setswana. Ms., Leiden University. Watters, John R. 1979 Focus in Aghem: a study of its formal correlates and typology. In Aghem grammatical structure, Larry M. Hyman (ed.), 137–97. [= Southern California Occasional Publications in Linguistics 7]. Los Angeles: Department of Linguistics, University of Southern California. Welmers, William E. 1959 Tonemics, morphotonemics and tonal morphemes. General Linguistics 4, 1–9. 1973a Functional and vestigial noun class systems. African Language Structures, 184–210. Berkeley: University of California Press. 1973b Verbal constructions in Niger–Congo. African Language Structures, 343–83. Berkeley: University of California Press. Williams, Edwin 1971/76 Underlying tone in Margi and Izgbo. Linguistic Inquiry 7, 462–84. 1989 The clause as a coordinate structure. Ms., University of Massachusetts, Amherst. 1997 Blocking and anaphora. Linguistic Inquiry 28, 577–628. Williamson, Kay 1989 Niger-Congo overview/ Benue-Congo overview. In The NigerCongo Languages, John T. Bendor-Samuel and Rhonda L. Hartell (eds.), 3–45, 247–74. Lanham, Md.: American Universities Press.

54 Victor Manfredi Williamson, Kay, and Roger Blench 2000 Niger-Congo. In African languages: an introduction, Bernd Heine and Derek Nurse (eds.), 11–42. Cambridge: Cambridge University Press. Xu, Liejiong 2004 Manifestation of informational focus. Lingua 114, 277–99. Xu, Yi, and Ching X. Xu 2005 Phonetic realization of focus in English Declarative intonation. Journal of Phonetics 33, 159–97. Zerbian, Sabine 2005 Why a prosodic approach to focus in Northern Sotho does not work. Bantu Network Meeting, S.O.A.S., London, 13 May. Zubizarreta, Maria Luisa 1998 Prosody, Focus and Word Order. Cambridge, Mass.: MIT Press.

Investigating prosodic focus marking in Northern Sotho Sabine Zerbian

Abstract The paper contributes to the ongoing debate on the expression of information structure in the languages of the world by investigating prosodic focus marking in Northern Sotho, a Bantu language spoken in South Africa. It presents the results of a production and perception study that shows the absence of prosodic focus marking in this tone language. It thereby challenges both the universality of the prominent status of focused constituents as well as the understanding of re-phrasing as a universal cue to focus. However, Northern Sotho is not unique in this respect. The finding therefore supports further research on a context-dependent theory of focus interpretation.

1.

Introduction

Over the last two decades, the expression of information structure in grammar has become a prominent topic in linguistic research. This paper contributes to the ongoing debate by investigating prosodic focus marking in Northern Sotho, a Bantu language spoken in the northern part of the Republic of South Africa, thereby deepening the understanding of whether and how the category of focus is represented and marked in the languages of the world. Although the Bantu languages form the largest language family in Africa, consisting of approximately 300 languages (Nurse & Philippson 2003: 2f), comparatively little research has been carried out on the prosodic expression of the information-structural categories topic and focus. Exceptions are Kanerva (1990), Jokweni (1995), Hyman (1990), and Downing et al. (2004). This paper joins these studies by providing data on the prosodic expression of in-situ focus in one further Bantu language. It applies experimental methodology that has been tested on Indo-European languages and thereby makes the data comparable cross-linguistically.

56 Sabine Zerbian The production and perception studies carried out show the absence of prosodic marking in the verbal and postverbal domain in Northern Sotho. The lack of grammatical marking of new information focus therefore challenges existing theories of focus that predict focus to be marked by prosodic prominence. More specifically with respect to focus prosody, the results reported here show that neither culminative stress prominence (Selkirk 2004) nor (re-) phrasing (Ladd 1996, Hayes & Lahiri 1991) are universal cues to focus. At the same time, the results underline the importance of typological research, in this case of African languages, for the development and improvement of linguistic theories. By investigating a tone language, this paper also contributes to the research of the interaction of (lexical) tone and intonation. Whereas non-tonal languages (e.g. Germanic languages like English and German) typically use pitch accent for focus marking, it is reported that tone languages (for Bantu, see Bearth 2003) often use syntactic transformations for the same purpose. The role of prosody in African tone languages remains under-investigated (Bearth 1999, Creissels 1996). Contrary to prior views and in accordance with work on non-African tone languages, research on the Bantu languages Chichewa, Kinande, and Xhosa has given evidence for the fact that tone languages do make use of prosodic means to indicate focus. In showing the absence of prosodic focus marking in Northern Sotho, this paper reveals language-specific variation among Bantu languages and thereby encourages further research into the microvariation of Bantu prosody. The paper is organized as follows. Section 2 discusses phrasal phonology in the Bantu language Chichewa in order to illustrate the kind of suprasegmental parameters that can be influenced by focus in a tone language. It relates to Northern Sotho in presenting the kind of prosodic changes to be expected under focus in this language. In section 3, the production study is presented that has been carried out in order to investigate the prosodic expression of in-situ focus in Northern Sotho. Section 4 deals with the perception study. Finally, section 5 summarizes the results of the studies and discusses their implications.

2.

Prosody in Bantu tone languages

Contrary to prior views (Cruttenden 1986), research on a range of different tone languages has shown that these languages use suprasegmental means in order to indicate focus. Among the suprasegmental means used in tone

Investigating prosodic focus marking in Northern Sotho 57

languages are the widening of the overall pitch range (Xu 1999 for Chinese), the insertion of boundary tones (Hyman 1990 for Kinande), the insertion of phonological phrase boundaries with resulting tonal effects (Kanerva 1990 for Chichewa; Jokweni 1995 for Xhosa), and the manipulation of downdrift (Downing et al. 2004 for Chichewa). Section 2.1 presents the prosodic expression of focus in Chichewa (Kanerva 1990, Downing et al. 2004), thereby illustrating how focus can be indicated in a Bantu tone language. Section 2.2 illustrates further use of suprasegmental means for conveying discourse-pragmatic meaning. Section 2.3 relates to the prosody of Northern Sotho. Section 2.4 summarizes and formulates the research question that will be investigated in section 3. 2.1. Influence of focus on phrasal phonology in Bantu Bantu languages have two-tone systems in which high tones (H) are specified underlyingly and low tones (L) are inserted by default (Kisseberth & Odden 2003). Following common practice, in the examples cited here, high tones are marked by acute and underlying high tones are underlined. Surface realizations of tone patterns are determined by principles governing the realization of underlying tones and the inventory and application of tonal processes as well as by the syntactic position of a word. The setting of these parameters is language-specific. Research on Chichewa, a Bantu language spoken in Malawi, has shown that, besides syntactic structure, it is also focus that influences the surface tone pattern of a word (Kanerva 1990, Downing et al. 2004). Focus is shown to insert phonological phrase boundaries that determine the application of tonal processes and thereby result in surface tone sandhi. Among the cues to phonological phrase boundaries in Chichewa are the lack of high tones on phrase-final vowels and lengthened vowels in the penultimate syllable of phrase-final words (See Kanerva 1990 for a detailed discussion of phonological processes that motivate phonological phrases in Chichewa). Phonological phrase boundaries are indicated by parentheses, focus structure is indicated by square brackets.

(1)

Chichewa (Kanerva 1990: 98) a. Q: ‘What did he do?’ A: ([anaményá nyumbá ndí he hit house with ‘He hit the house with a stone.’

mwáála]F)PP stone

58 Sabine Zerbian b. Q: A:

‘What did he hit the house with?’ (anaményá nyumbá [ndí mwáála]F)PP

c. Q: A:

‘What did he hit with the rock?’ (anaményá [nyuúmba]F)PP (ndí

d. Q: A:

‘What did he do to the house with the rock?’ (nyuúmba)PP (ndí ([anaméenya]F)PP

mwáála)PP

mwáála)PP

Under wide focus, (1a), the entire VP forms a single phonological phrase. The same phrasing is observed for narrow focus on the sentence-final prepositional phrase, as in (1b). Narrowing the focus within the sentence, as to the object in (1c) and to the verb in (1d), results in additional phonological phrase boundaries. The focused constituent is followed by a phonological phrase boundary. For further details of the analysis concerning the phrasing of post-focal constituents, see Kanerva (1990) and Truckenbrodt (1995, 1999). For comparable expression of focus by suprasegmental means in Xhosa, see Jokweni (1995) and a reanalysis in Zerbian (2004). 2.2. Intonation in Bantu Apart from focus, suprasegmental means also convey further discoursepragmatic meaning in Bantu languages, such as sentence types. Again, there is microvariation concerning the phonetic implementation (see, e.g., Downing 1996 for the Bantu language Jita). In Northern Sotho and Southern Bantu languages more generally, raised overall pitch and lack of length on the penultimate syllable distinguish declaratives from yes/no-questions (Poulos & Louwrens 1994: 374, Jones et al. 2001 in a detailed phonetic study on Xhosa). Independent of the lexical tone patterns of the words used in an utterance, the yes/no-question displays a raised overall pitch, which is indicated by the arrow in (2). Furthermore, the penultimate syllable of the sentence-final word is not lengthened in yes/no-questions. (2)

a.ÇO dumediUhC dikgarebe ? ladies 2nd greet ‘Are you greeting the ladies?’

Investigating prosodic focus marking in Northern Sotho 59

b. Ee, ke dumedKUhC dikgare:be. greet ladies yes 1st ‘Yes, I am greeting the ladies.’ Also, a non-lexical high tone on the ultimate syllable of a clause indicates continuation, as shown in (3). (3)

Kgarebe é

búla lefasetere H% mo-kgalabje ó

tswálela lematí.

CL9.lady SM9

open window

close

CL1-old.man SM1

door

‘The lady opens the window, the old man closes the door.’ The examples in (2) and (3) show that Northern Sotho and Bantu languages in general make use of intonation for discourse-pragmatic meaning in accordance with cross-linguistic tendencies. 2.3. Phrasal phonology in Northern Sotho Just as in Chichewa, Northern Sotho shows suprasegmental and tonal processes that are sensitive to prosodic domains. The two most prevalent phrasal processes are penultimate lengthening and the application of the finality restriction at prosodic boundaries. Vowel length in Northern Sotho is not contrastive. It can therefore be exploited to convey grammatical meaning. Like in Chichewa, in Northern Sotho, the penultimate syllable of a word in phrase-final position is regularly lengthened (Doke 1954: 125). Length is indicated by a colon in the examples to follow. The rule of high tone spread (HTS), which is common in Bantu languages, accounts for the spreading of an underlying high tone to the immediately right-adjacent syllable (see also Zerbian 2006b). This rule applies within a word, as shown in the verbs in (4a). If, however, the word appears in phrase-final position, spreading to a phrase-final syllable is blocked. Such a restriction that exempts a final syllable from prosodic processes is called finality restriction and is common across Bantu (Kisseberth & Odden 2003: 64). The tonal effect of the finality restriction is exemplified in (4b).

(4)

Northern Sotho a. Ke thúshá mosá:di. woman 1st help

b.

Ke 1st

a A

thú:UhC help

60 Sabine Zerbian Ke 1st Ke 1st

Uhómá work bóná see

polase:-ng. farm-at gagó:lo. well

Ke 1st Ke 1st

a A

a A

Uhó:ma. work bó:na. see

The blocking of HTS is therefore an indicator of prosodic domain edges. The prosodic domain for lengthening and the application of the finality restriction in Northern Sotho is smaller than the entire clause. In (5), length and tone is manipulated on the verb which is in non-final position. (5)

Ke a mo thú:UhC st A CL1 help 1 ‘I am helping her, the woman.’

mo-sá:di. CL1-woman

Prosodic structure therefore depends on syntactic structure in Northern Sotho. Zerbian (2006a) argues that prosodic boundaries occur at the right edges of syntactic IPs in this language. In the study reported in the next section, it is tested if the prosodic domain which penultimate lengthening and the finality restriction give evidence for is sensitive to focus. 2.4. Summary Focus-induced variation in phrasing in some Bantu languages and the use of intonational changes for conveying discourse-pragmatic meaning in Northern Sotho motivate the hypothesis that suprasegmental means also play a role in postverbal in-situ focus in Northern Sotho. At the same time, language-specific variation among Bantu languages with respect to the parameters that determine and influence surface tone realization justifies research on the focus prosody of further Bantu languages. The next section presents a production and perception study on the prosodic marking of in-situ focus in Northern Sotho.

3.

Production study

This section presents the production study that was carried out in order to investigate the means of prosodic marking of in-situ focus in Northern Sotho. 1 The expectation based on comparative studies is that prosodic marking of focus in Northern Sotho results in perceptible suprasegmental

Investigating prosodic focus marking in Northern Sotho 61

changes, either concerning the realization of underlying tone patterns, realization of length on the penultimate syllable (as discussed in section 2.1 and 2.3), or overall register height (see section 2.2). For more details on the studies, the reader is referred to Zerbian (2006a: 149ff). 3.1. Production set up In order to establish if there are prosodic differences between sentences displaying narrow focus on the verb phrase, the verb, the object, or the adverb, a production study was carried out with native speakers of Northern Sotho. Sentence pairs with different focus structures were recorded. The individual pairs consist of a trigger sentence and a target sentence. The trigger sentence is a question that determines the focus structure of the corresponding answer, the target sentence. Thereby, this study follows the underlying assumption, first used in Paul (1880) and applied in experimental set ups by Gussenhoven (1983), Oppenrieder (1989), Birch & Clifton (1995), and Xu (1999) among others, that the focus of an utterance is controlled by a preceding question. An example for the question/answer-pairs is given in (6). (6)

a. What do you do? Ke [mémá mo-hú:mi]F. invite CL1-rich 1st ‘I [invite the rich man]F.’ b. Who do you invite? Ke mémá [mohú:mi]F. ‘I invite [the rich man]F.’ c. What are you doing with the rich man? Ke [mémá]F mohú:mi. ‘I [invite]F the rich man.’

The current study tried to prevent reading intonation (Gussenhoven 1983, referring back to Brazil et al. 1980: 83) by having two native speakers interact in posing and answering questions which have been provided on cards. The subjects were instructed to answer in a full sentence repeating all constituents. Although this is not the most natural answer (discourse-old constituents are often reduced or deleted), this methodology is justified by the need for a set-up in which focus is controlled for.

62 Sabine Zerbian The production study consisted of 69 sentences altogether, having five different syntactic target structures: intransitive verbs followed by an adverb, and transitive and ditransitive structures both with and without adverb. Examples are given in (7). (7)

a. Intransitive sentences with adverb Ke bíná monyányé-:ng. dance party-LOC 1st ‘I am dancing at the party.’ b. Transitive sentences Ke mémá mo-hú:mi. 1st invite CL1-rich ‘I invite the rich man.’ c. Transitive sentences with adverb moséga:ré. Ke rémá morúlá chop marula tree midday 1st ‘I chop the marula tree at midday.’ d. Ditransitive sentences Ke fá mmá hém:pe. give mother shirt 1st ‘I give mother a shirt.’ e. Ditransitive sentences with adverb malómé lengwáló Ke néá st 1 give uncle letter ‘I give uncle a letter at the dwelling place.’

bo-dúló:-ng. CL14-live-LOC

Narrow focus on each constituent of these sentences plus wide focus yielded 69 target sentences, displaying different focus structures. Eight native speakers of Sotho languages participated in the study. The five speakers who provided the target sentences were all speakers of the Sepedi dialect of Northern Sotho.

Investigating prosodic focus marking in Northern Sotho 63

3.2. The target sentences The target sentences were chosen in such a way that a maximal suprasegmental contrast, both in length and tone, could have been observed if focus induced boundaries in Northern Sotho had indicated differences in information structure (see section 2.3). This hypothesis is illustrated in (8). (8)

a. VP and object focus Ke mémá mohú:mi)PP. ‘I invite [the rich man]F.’ b. Verb focus Ke mé:ma)IP mohú:mi)PP. ‘I [invite]F the rich man.’

In (8a), the verb stem initial high tone spreads onto the immediately rightadjacent vowel under wide focus and object focus. If focus induced a phrase boundary, the application of HTS would be blocked because the finality restriction would apply, as shown in (8b). Also, the syllables that undergo penultimate lengthening are expected to differ under different focus structures, as is also shown in (8). 2 3.3. Quantitative analysis A sample of the recorded data, viz. the basic SVO/SVAdv structure, was chosen for the analysis. The reason for the decision in favor of basic SVO/SVAdv structures, as in (7a,b), as compared to structures in which two or more constituents follow the verb, as in (7c-e), is that if there is prosodic expression of focus, it will most clearly emerge within short sentences. In short sentences, the register size at disposal for pitch manipulations towards the end of the sentence is wider than in long sentences, due to downdrift. The sentences are given in (9). (9)

Sentences of prosodic analysis a. i. Ke mémá mo-hú:mi. invite CL1-rich 1st ‘I invite the rich man.’ ii. Ke rwálá mo-rwá:lo. carry CL3-load 1st ‘I am carrying a load.’

64 Sabine Zerbian b. i.

ii.

Ke bíná mo-nyányé:-ng. dance CL3-party-LOC 1st ‘I am dancing at the party.’ Ke bóá mo-lálé:-ng. return CL3-field-LOC 1st ‘I am returning from the field.’

Although the sets are different with respect to the grammatical function of the postverbal constituent, they are identical in their phrasal and tonal make-up. It has been shown in Zerbian (2006a: 122ff) that objects and adverbs are phrased parallel in Northern Sotho. This insight is exploited in the analysis. Although the sentences in (9a) differ from those in (9b) in the absolute number of syllables (the locative suffix -ng in (9b) is syllabic), this difference is not crucial as the syllables of interest are the two syllables of the verb and the following nominal prefix. Moreover, all sentences in (9) contain a high-toned disyllabic verb which is followed by a constituent that has a low-toned nominal prefix but bears a high tone on the noun stem initial syllable. This tonal environment allows HTS to occur on the verb and has been chosen because it has been shown that high tones are indeed manipulated at IP-boundaries in Northern Sotho (see section 2.3). The sounds of these utterances were segmented according to common phonetic practice (see e.g. Ladefoged 2001, 2003). Two phonetic parameters were measured: pitch height and duration. 3 These two parameters are phonetic correlates of stress in intonation languages, and also the tonal and durational changes expected under phrasing differences in Bantu tone languages are captured by these parameters. Changes in the tonal structure due to the application of the finality restriction and the resulting blocking of HTS will result in differences in F0. Changes in penultimate lengthening due to the insertion of prosodic boundaries after focused constituents will result in longer duration of vowels. 3.4. Results – Fundamental frequency In the following diagrams, the F0-average values for each of the five speakers are represented in a plot. The three horizontal lines correspond to the three different focus conditions: VP focus, object focus, and verb focus. The values indicated in the diagrams are average values of the respective segment over the four utterances indicated in (9). 4

Investigating prosodic focus marking in Northern Sotho 65

On the x-axis, the vowels of the different constituents in the utterance are displayed. The second and third value correspond to the verb, the fourth till final value correspond to the object or locative adverbial. The first value refers to the vowel of the personal pronoun (ke). On the y-axis, the average F0-value is given in Hertz. Four speakers are female. The fifth speaker is male wherefore the lines occur in a lower pitch range. The dashed line shows VP-focus, the continuous line represents focus on the verb, and the dotted line indicates focus on the object or local adverbial. Diagrams 1–5: Average F0

300 250 200 150 100 50 0

Obj/Adv-focus V-focus

/A dv

O

O

O

bj

/A dv

v

bj

Ad

/A dv

V

bj

V

VP-focus

SM

F0 (Hz)

Diagram 1: Speaker DE

Diagram 2: Speaker LY 250

F0 (Hz)

200 150 100 50 0 SM

V

V

Obj/Adv

Adv

Obj/Adv

Obj/Adv

66 Sabine Zerbian Diagram 3: Speaker MA 350 300 F0 (Hz)

250 200 150 100 50 0 SM

V

V

Obj/Adv

Adv

Obj/Adv Obj/Adv

F0 (Hz)

Diagram 4: Speaker Mak 350 300 250 200 150 100 50 0 SM

V

V

Obj/Adv

Adv

Obj/Adv Obj/Adv

Diagram 5: Speaker MO 180 160

F0 (Hz)

140 120 100 80 60 40 20 0

SM

V

V

Obj/Adv

Adv

Obj/Adv

Obj/Adv

Investigating prosodic focus marking in Northern Sotho 67

The general tonal information, discussed in section 2.3, emerges from the diagrams. The verb bears a high tone on the stem initial syllable underlyingly. The peak of F0 is only reached in the second syllable of the verb, which is captured by the process of high tone spread (HTS). After the F0 peak in the second syllable of the verb, F0 drops again. The noun stem initial syllable, i.e. the second syllable of the noun, is transcribed in the examples as bearing a high tone. This is motivated by the tonal realization the noun receives if it is pronounced in isolation. Within the sentence, no F0 peak is visible. This is due to declination, which narrows the pitch range available to speakers towards the end of the utterance (see Gussenhoven 2004 for an overview of downtrends in intonation). In the speech of LY, MA, Mak, and MO, we find a nearly perfect overlap in what concerns the F0 for all three focus conditions. F0 raises in the verb due to HTS and falls smoothly towards the end of the sentence. Interestingly, in the speech of DE a slightly higher F0 can be found for the initial syllable of the object when it is in focus. Also, a slightly raised pitch and a lower initial syllable of the object can be found when the verb is in focus, which results in a sharper fall in pitch. Although the differences are only small, it is nevertheless important to follow up on these differences as the second phenomenon, viz. the sharper fall after a focused verb, has been reported as a focus strategy in Chichewa (Downing et al. 2004). By the term ‘anti-accent’, Downing et al. (2004: 177) describe that in Chichewa the focused element is “made prominent by raising the pitch enough to make the following elements relatively much lower in pitch”. However, what we do not find in Northern Sotho is a pitch accent on the focused constituent. The diagram shows that F0 falls towards the end of the sentence (due to declination), which is also the case if the object is the focused constituent. Judging from the pitch accent distribution in intonationonly languages like English and German, if focus were expressed by pitch accent in Northern Sotho as well, one would expect to find some reversal of the trend in the condition of object focus. In addition to the lack of evidence for the use of a pitch accent to indicate focus, it is interesting to note that tone sandhi do not emerge as a consequence of different focus conditions either. Against expectation, no speaker shows a fall of F0 on the second syllable of the verb, as predicted if focus inserted a phrase boundary that blocks the application of high tone spread (HTS). Also, a third option reported for tone languages, namely the raising of the overall pitch register, is not mirrored in the diagrams 1–5.

68 Sabine Zerbian The preliminary conclusion is that focus is indicated neither by pitch accent nor by tone sandhi in Northern Sotho. The perception study presented in section 4 checks this preliminary conclusion, and also checks if the slightly sharper fall after a focused verb in the speech of DE is interpreted as a focus strategy, parallel to the ‘anti-accent’ described in Downing et al. (2004) for Chichewa. 3.5. Results – duration In diagrams 6–10, the duration of each vowel over the three different focus conditions is given for all five subjects separately. The duration values represent relative duration of the segments. Because of the uneven number of syllables in (9a) and (9b), the absolute duration of the segment has been divided by the absolute time of the whole utterance in order to account for differences which are not only induced by utterance length but also by speech tempo etc. Again, on the x-axis the vowels of the different constituents of the utterance are displayed. Each line represents the average values within one focus condition. On the y-axis, the relative duration of the segments is given in seconds. The diagrams are discussed below. Diagrams 6–10: Average duration

0,25 0,2 0,15 0,1 0,05 0

Obj/ Adv-focus V-focus

v

v bj

/A d

/A d O

v

v /A d

Ad

bj O

O

bj

V

VP-focus

V

SM

relative duration (sec)

Diagram 6: Speaker DE

Investigating prosodic focus marking in Northern Sotho 69

relative duration (sec)

Diagram 7: Speaker LY 0,25 0,2 0,15 0,1 0,05 0 SM

V

V

Obj/Adv

Adv

Obj/Adv Obj/Adv

Diagram 8: Speaker MA

relative duration (sec)

0,25 0,2 0,15 0,1 0,05 0 SM

V

V

Obj/Adv

Adv

Obj/Adv Obj/Adv

relative duration (sec)

Diagram 9: Speaker Mak 0,25 0,2 0,15 0,1 0,05 0 SM

V

V

Obj/Adv

Adv

Obj/Adv Obj/Adv

70 Sabine Zerbian

relative duration (sec)

Diagram 10: Speaker MO 0,25 0,2 0,15 0,1 0,05 0 SM

V

V

Obj/Adv

Adv

Obj/Adv Obj/Adv

A quite homogeneous picture emerges for all five speakers in all focus conditions. The speech of DE, Mak, and MO again shows a nearly perfect overlap with respect to duration in different focus conditions. The lengthening of the vowel in the penultimate syllable is clearly discernible in the diagrams 6–10. In the speech of LY and MA, we find a slightly lengthened verb final syllable. However, the implication of this observation with respect to the encoding of information structure is doubtful. First, the increase in duration is not as expected in the penultimate syllable of the focused constituent and also is not as high as the lengthened penultimate syllable of the sentence. Second, the lengthening of the final syllable is not restricted to one focus condition. It occurs with verb focus in the speech of MA but with object focus in the speech of LY. Again, the preliminary conclusion from a phonetic investigation of the production data suggests no prosodic expression of focus through lengthening. However, the lengthening of the verb final syllable, although slight, might be significant enough for native speakers to serve as an indicator for the focused constituent in the utterance.

4.

Perception study

The prosodic analysis revealed that across all speakers there are no systematic, significant differences between different focus structures if tested for the common stress-related parameters pitch and duration. The objective of the perception study is therefore to investigate if there are nevertheless prosodic cues in the production data that enable native speakers of Northern

Investigating prosodic focus marking in Northern Sotho 71

Sotho to identify instances of in-situ focus. These prosodic cues might be very subtle so that they have escaped an acoustic analysis of the data, or they might differ from the stress-related phonetic correlates that were investigated in section 3. Furthermore, it needs to be established if the existing suprasegemental differences seen in the speech of LY, MA, and DE are indeed related to differences in focus structure of the utterances. Perception tests for establishing distinct categories in focus structures have been carried out by Gussenhoven (1983), Oppenrieder (1989), Jannedy (2002) a.o. for focus in sentences and by Swerts et al. (2002) for focus in NPs. 4.1. Perception set up The target sentences obtained by the recordings are used in the perception study in order to test if prosodic differences in the non-identical pairs can be related to information structure (study 1) and if there are perceivable prosodic differences in the pairs that sound identical (study 2). The perception study asked for appropriateness judgments: Subjects were asked to listen to an identical question followed by two differing answers in direct comparison. The task is illustrated in (10). (10)

Question: ‘Which of these question-answer pairs matches best?’ a. O mémá má:ng? ‘Who do you invite?’ Ke mémá [mohú:mi]F. ‘I invite [the rich man]F.’ b. O mémá má:ng? ‘Who do you invite?’ Ke [mémá]F mohú:mi. ‘I [invite]F the rich man.’

As indicated in (10), one of the answers is the original answer to the question. The competing answer is taken from a different focus-context. Subjects had to decide between the first and the second question/answer-pair. Study 1 contained utterances with perceptible suprasegmental differences. Four lexically different contexts were used, which are given in (11). Study 2 contained utterances with no perceptible difference. Three lexically different contexts were used, which are given in (12).

72 Sabine Zerbian (11)

(12)

Study 1 a. Ke néá 1st give b. Ke néá 1st give c. Ke nyálwá 1st marry d. Ke rwálá 1st carry Study 2 a. Ke já 1st eat b. Ke rémá 1st chop c. Ke fá 1st give

náré lé:e. buffalo egg molámó lemá:o. uncle needle labo:ne. four times morwá:lo. load

‘I give the buffalo an egg.’ ‘I give the uncle a needle.’ ‘I marry for the fourth time.’ ‘I carry a load.’

namú:ne. ‘I eat an orange.’ orange morúlá moséga:ré. ‘I chop the marula at midday.’ marula tree midday m´má hém:pe. ‘I give mother a shirt.’ mother shirt

In the perception study, the focus structures of a target sentence were checked against all possible constituent questions. The expectation for this study is that the speaker chooses one of the two alternative answers as more appropriate if there are subtle prosodic differences in the realization of various in-situ foci. The set up in terms of minimal answer pairs was chosen in order to focus the listener’s attention on possible subtle differences in the realization of different focus structures. The appropriateness judgments made by the four subjects participating in this study were recorded by the experiment software (Praat, Boersma & Weenink 2005) and evaluated. 4.2. Results The responses in experiment 1, in which the question/answer-pairs showed a slight difference in prosody, were distributed according to chance both within and across speakers. The same result was obtained for experiment 2, in which the question/answer-pairs did not show prevalent perceptible differences (for details see Zerbian 2006a). Therefore, one can conclude that the existing prosodic differences are not related to the focus structure of an utterance (study 1) and furthermore, that the utterances with no perceptible

Investigating prosodic focus marking in Northern Sotho 73

differences do not contain subtle prosodic cues that could be related to focus structures (study 2). A closer inspection of the answers to the perception study reveals that suprasegmental differences influence the judgments independent of focus contexts. To give an example: in one test item (DE_15), downstep occurs after the focused verb. However, it is not considered a significant cue to focus structure. Although all subjects perform exceptionally well when choosing the downstepped answer as the inappropriate answer, they do not choose the downstepped answer as the appropriate one in the verb focus context. Also, ordering effects become visible in the responses (Ladd & Morton 1997, Schiefer & Batliner 1991). Whereas one subject nearly always chooses the second answer, another nearly always chooses the first answer.

5.

Conclusion and discussion

The experimental part had as its objective to test if there is a prosodic expression of focus in Northern Sotho when the focused constituent is in-situ and is followed by discourse-old material. Studies on Chichewa and Xhosa report changes in the realization of the tonal pattern or lengthening of penultimate syllables. Northern Sotho tonology shows characteristics which are comparable to Chichewa and Xhosa and which lead one to expect prosodic changes under focus in Northern Sotho as well. The production study tested if Northern Sotho native speakers produce prosodic differences which are correlated to diverging focus structures. Target sentences were prepared in such a way as to allow tonal as well as durational changes. Question/answer-pairs that were controlled for different focus conditions were recorded from native speakers. A phonetic analysis of the data with respect to fundamental frequency and duration showed no systematic prosodic expression of focus. A follow-up perception study showed that the existing prosodic differences were not interpreted with respect to focus. Neither did any other prosodic cue emerge as relevant for the encoding of focus in Northern Sotho. The conclusion drawn from this production/perception study is that in syntactically identical sentences there is no suprasegmental marking of focus in Northern Sotho. The study therefore showed that in Northern Sotho, in contrast to Chichewa and Xhosa, focus does not affect phrasing. Differences in production of diverging underlying information structures

74 Sabine Zerbian can only be attributed to idiolects or differences in speech tempo, but they cannot be regarded as language-inherent functional means of marking focus. The production and perception study has shown that there are no prosodic means available in the grammar of Northern Sotho for marking insitu focus in the verbal and postverbal domain. The absence of prosodic marking has also been shown for other languages (e.g. Wolof in Rialland & Robert 2001). However, Northern Sotho shows the simultaneous lack of any other grammatical focus marking, such as morphological (as in Wolof) or syntactic (see example (8) and also Zerbian (2006a)). Hartmann & Zimmermann (2004, 2007) report similar findings for Tangale and Hausa, two Chadic languages. In general, the findings in these languages question the universal need for focus marking in grammar. However, the conclusion does not imply that Northern Sotho lacks grammatical means of indicating the focused constituent in the verbal and postverbal domain. For example, Northern Sotho uses deletion, pronominalization, and dislocation of discourse-old constituents; processes which conspire to the tendency that the focused constituent is often final in a clause. Neither does the lack of prosodic marking of postverbal in-situ focus mean that Northern Sotho does not show any prosodic reflexes of information structuring. For example, a right-dislocated discourse-old constituent is set off from the sentence by an optional pause, preboundary lengthening, and tone sandhi, see (5). However, the necessity of an object marker on the verb when being dislocated shows that the prosodic changes are dependent on the syntactic structure and not a means on their own. As pointed out above, the absence of prosodic means with the simultaneous absence of syntactic marking in languages like Northern Sotho poses a problem for standard theories of focus (from Jackendoff 1972 to Schwarzschild 1999 and beyond) that generally assume that focus must be marked somehow in the grammar, e.g. by pitch accent in languages like English and German. Following Hartmann & Zimmermann (2007), who discuss the theoretical implications of the lack of grammatical focus marking in Hausa, there are two ways to account for this lack. First, one could argue that focus does not exist as a grammatical category in the grammar of Northern Sotho, which is reflected by the lack of formal F-features. However, syntactic processes of deletion, pronominalization, and dislocation (also called indirect focus marking strategies by Bearth 1999) make indirect reference to formal F-features as only non-F-

Investigating prosodic focus marking in Northern Sotho 75

marked constituents are targeted by these processes. Furthermore, the subject cannot be F-marked in its canonical preverbal position (Zerbian 2006a, to appear). The alternative view is that Northern Sotho, just like Hausa, is among the languages that do not require obligatory F-marking in grammar. This characterization is reminiscent of the distinction between obligatory and non-obligatory focus-marking systems in Heine & Reh (1983). This distinction was first proposed with respect to morphological F-marking but seems to be applicable more widely also with respect to prosodic and syntactic focus marking. From the study of African languages such as Hausa (Chadic) and Northern Sotho (Bantu), evidence against the obligatoriness of marking focus in grammar emerges. Therefore, existing theories of focus marking cannot lay claim to universality, and future research has not only to bring further evidence for such systems but also to accommodate the strategies of those languages in appropriate focus theories (but see Manfredi, this volume, for an alternative). A first proposal has been advanced in Hartmann & Zimmermann (2007), drawing attention to topic marking and intersentential marking as two alternative strategies of information structuring.

Notes

This work is part of the project ‘The expression of information structure in languages of Southern Africa’ funded by the DFG. Further financial assistance also came from a fellowship from the DAAD. I wish to express my gratitude to Makopi Mogodi, Mogoke Mogodi, Simon Mashanya, Mampaka Lydia Mojapelo, Matseleng Mokhwesana, and Deborah Mampuru who made this study possible in the first place. Furthermore, I wish to thank Bernd Pompino-Marschall for practical advice, and Hubert Truckenbrodt, Laura J. Downing, and two anonymous reviewers for helpful comments. 1. The investigation is restricted to in-situ focus in the verbal and postverbal domain as focus on the subject cannot be expressed with having the subject in its canonical preverbal position in Northern Sotho (see Zerbian 2006a, to appear). 2. Furthermore, there are independent phonetic restrictions on the use of segmental material in the target utterances to assure that changes in fundamental frequency are indeed related to extra-phonetic facts like discourse-pragmatics. On the one hand, obstruents interrupt and disturb the F0-contour by their inherent phonetic properties. On the other hand, with

76 Sabine Zerbian short vowels, the pitch reaches its peak only at the end of a following consonant (Ladd et al. 2000 for Dutch). Because of the inherent phonetic properties of obstruents as well as the lack of underlyingly long vowels in Northern Sotho, the target sentences of the current study are constructed in such a way that they contain sonorants only, which allows a theoretically undisturbed F0-contour. 3. Intensity is left out for two reasons: first, it is the parameter that indicates stress in intonation-only languages least reliably. Second, it is the parameter that is the most sensitive to changes in the recording surrounding, such as distance to the microphone, background noises that might lead the subject to speak louder. As these parameters were not controlled for (flexible distance to the microphone and no sound-proof box), it is expected that deviation in intensity values might not alone be attributed to focus. 4. Fundamental frequency is influenced by the preceding sound. Sometimes, no F0-value could be calculated for the vowel following the plosive of the subject marker, as is the case in the speech of DE. Along the same lines, it was sometimes impossible to calculate a F0-value for the final vowel of an utterance. Because of the unequal number of syllables in (9a) and (9b), the fifth value in the diagrams is only calculated for the examples in (9b) and therefore represents the average of two utterances only. The penultimate and last syllable of all examples have been calculated together. This is especially important for the duration measurement, as the duration measurement, as the penultima is the lengthened syllable.

References Bearth, Thomas 1999 The contribution of African linguistics towards a general theory of focus. Update and critical review. Journal of African Languages and Linguistics 20: 121–156. 2003 Syntax. In The Bantu languages, Derek Nurse and Gerard Philippson (eds.), 121–142. London: Routledge. Birch, Stacy, and Charles Clifton 1995 Focus, Accent, and Argument Structure: Effects on Language Comprehension. Language and Speech 38: 365–391. Boersma, Paul, and David Weenink 2005 Praat: doing phonetics by computer (Version 4.3.01) [Computer program]. Retrieved Apr 23, 2004 from http://www.praat.org/. Brazil, David, Malcolm Coulthard, and Catherine Johns 1980 Discourse intonation and language teaching. Essex: Longman.

Investigating prosodic focus marking in Northern Sotho 77 Creissels, Denis 1996 Conjunctive and disjunctive verb forms in Setswana. South African Journal of African Languages 16: 109–115. Cruttenden, Alan 1986 Intonation. Cambridge: Cambridge University Press. Doke, Clement Martyn 1954 The Southern Bantu Languages. London, New York, Cape Town: Oxford University Press. Downing, Laura J. 1996 The tonal phonology of Jita. München: Lincom Europa. Downing, Laura J., Al Mtenje, and Bernd Pompino-Marschall 2004 Prosody and Information Structure in Chichewa. In Papers in Phonetics and Phonology, Susanne Fuchs and Silke Hamann (eds.), 167– 187. Berlin: ZAS. Gussenhoven, Carlos 1983 Testing the Reality of Focus Domains. Language and Speech 26: 61–80. 2004 The Phonology of Tone and Intonation. Cambridge: Cambridge University Press. Hartmann, Katharina, and Malte Zimmermann 2004 Focus Strategies in Chadic - the Case of Tangale Revisited. In Interdisciplinary Studies on Information Structure 1. Working Papers of the SFB 632, Shinichiro Ishihara, Michaela Schmitz, and Anne Schwarz (eds), 207–243. Potsdam: Universitätsverlag Potsdam. 2007 In Place - Out of Place: Focus in Hausa. In On Information Structure, Meaning and Form, Kerstin Schwabe and Susanne Winkler (eds.), 365–403. Amsterdam: John Benjamins. Hayes, Bruce, and Aditi Lahiri 1991 Bengali Intonational Phonology. Natural Language and Linguistic Theory 9: 47–96. Heine, Bernd, and Mechthild Reh 1983 Grammaticalization and Reanalysis in African Languages. Hamburg: Helmut Buske Verlag. Hyman, Larry M. 1990 Boundary Tonology and the Prosodic Hierarchy. In The PhonologySyntax Connection, Sharon Inkelas and Draga Zec (eds.), 109–126. Chicago: The University of Chicago Press. Jackendoff, Ray 1972 Semantic Interpretation in Generative Grammar. Cambridge (Mass.): Massachusetts Institute of Technology.

78 Sabine Zerbian Jannedy, Stefanie 2002 Hat Patterns and Double Peaks: The Phonetics and Psycholinguistics of Broad versus late narrow versus double focus intonations. PhDthesis, Ohio State University. Jokweni, Mbulelo Wilson 1995 Aspects of Isixhosa Phrasal Phonology. PhD-thesis, University of Illinois. Jones, Jackie, Jacobus Abraham Louw, and Justus Christian Roux 2001 Queclaratives in Xhosa: An acoustic analysis. SAJAL Supplement 36: 3–18. Kanerva, Jonni M. 1990 Focus and Phrasing in Chichewa Phonology. New York, London: Garland Publishing. Kisseberth, Charles W., and David Odden 2003 Tone. In The Bantu languages, Derek Nurse and Gerard Philippson (eds.), 59–70: London & New York: Routledge. Ladd, D. Robert 1996 Intonational Phonology. Cambridge: Cambridge University Press. Ladd, D. Robert, and Rachel Morton 1997 The perception of intonational emphasis: continous or categorical. Journal of Phonetics 25: 313–342. Ladefoged, Peter 2001 Vowels and Consonants - An Introduction to the Sounds of Language. Malden, MA: Blackwell. 2003 Phonetic Data Analysis - An Introduction to Fieldwork and Instrumental Techniques. Malden: Blackwell Publishing. Nurse, Derek, and Gerard Philippson 2003 Introduction. In The Bantu languages, Derek Nurse and Gerard Philippson (eds.), 1–12. London & New York: Routledge. Oppenrieder, Wilhelm 1989 Fokus, Fokusprojektion und ihre intonatorische Kennzeichnung. In Zur Intonation von Modus und Fokus im Deutschen, Hans Altmann, Anton Batliner, and Wilhelm Oppenrieder (eds.), 267–280. Tübingen: Niemeyer. Paul, Hermann 1880 Prinzipien der Sprachgeschichte. Leipzig: Niemeyer. Poulos, George, and Louis J. Louwrens 1994 A Linguistic Analysis of Northern Sotho. Hatfield: Via Afrika. Rialland, Annie, and Stephane Robert 2001 The intonational system of Wolof. Linguistics 39: 893–939.

Investigating prosodic focus marking in Northern Sotho 79 Schiefer, Liselotte, and Anton Batliner 1991 A ramble round the order effect. Forschungsberichte des Instituts für Phonetik und Sprachliche Kommunikation der Universität München (FIPKM) 29: 125–180. Schwarzschild, Roger 1999 GIVENness, AvoidF and other Constraints on the Placement of Accent. Natural Language Semantics 7: 141–177. Selkirk, Elisabeth 2004 Bengali intonation revisited. In Topic and Focus: A Cross-Linguistic Perspective, Chungmin Lee, Matthew Gordon, and Daniel Büring (eds.), 217–246: Kluwer Academic Publishers. Swerts, Marc, Emiel Krahmer, and Cinzia Avesani 2002 Prosodic marking of information status in Dutch and Italian: a comparative analysis. Journal of Phonetics 30: 629–654. Truckenbrodt, Hubert 1995 Phonological phrases: Their relation to syntax, focus, and prominence. PhD-thesis, Massachussetts Institute of Technology. 1999 On the Relation between Syntactic Phrases and Phonological Phrases. Linguistic Inquiry 30: 219–255. Xu, Yi 1999 Effects of tone and focus on the formation and alignment of f0 contours. Journal of Phonetics 27: 55–105. Zerbian, Sabine 2004 Phonological Phrases in Xhosa. In Papers in Phonetics and Phonology, Susanne Fuchs and Silke Hamann (eds.), 71–100. Berlin: ZAS. 2006a Expression of Information Structure in Northern Sotho. PhD thesis, Humboldt-University Berlin. 2006b High tone spread in the Sotho verb. In Selected Proceedings of the 35th Annual Conference on African Linguistics, John Mugane, John P. Hutchinson, and Dee A. Worman (eds.), 147–157. Somerville, MA: Cascadilla Proceedings Project. to appear The subject/object-asymmetry in Northern Sotho. In Information Structure and the Architecture of Grammar: A Typological Perspective, Kerstin Schwabe and Susanne Winkler (eds.), Amsterdam: John Benjamins.

Part II Information structure and word order

Preverbal objects and information structure in Benue-Congo Tom Güldemann

Abstract It is well known that Africa hosts a considerable number of languages with an inconsistent word order profile, including a typologically rare clause word order S(Auxiliary)-O-V-Other. The most recent non-generative accounts of this pattern (Claudi 1993, Heine and Claudi 2001) try to explain it exclusively in terms of grammaticalization scenarios. I will address its possible motivation in BenueCongo where important preconditions for the proposed grammaticalizations do not hold. By observing that preverbal objects are here regularly restricted to subclasses of objects (e.g. extrafocal participants, pronouns) or to special clause types (e.g. negation and certain TAM categories), I will try to show that this language group gives evidence for another factor triggering O-V: this order is arguably associated with the status of the object as less focal or even extrafocal, non-asserted information.

1.

Introduction

A number of African languages of different genealogical and areal affiliation show a variable object position, namely before or after the verb, as exemplified in (1) from Idoma (Idomoid). (1)

vs. ó lç$cí ó mç$cí 3S see:tree 3S ?:tree ‘She saw the tree.’ (Abraham 1951: 18)

má see

As argued by Gensler and Güldemann (2003) and Güldemann (forth.), this feature has a biased distribution on the continent: geographically it is particularly frequent in a large East-to-West belt South of the Sahara; in genealogical terms, it is a property of many sub-groups of Niger-Congo. While early non-generative 1 assessments of such “inconsistent” word or-

84 Tom Güldemann ders in the area had assumed an older O-V stage, in particular for NigerCongo (e.g. Givón 1975), the standard explanation today is in terms of grammaticalization and reanalysis from a basic V-O clause to innovative O-V patterns (e.g. Claudi 1993, Heine and Claudi 2001). The two basic developments, starting (a) in an auxiliary periphrasis involving a nominalized complement and (b) in a serial verb construction, can be schematized as follows: 2 (a) > (b) >

Verbx Auxiliaryx [Verbx Auxiliaryx -

[Genitivey Objecty Object] Object -

Verbal nounz] Verbz Verb Verb -

Other Other Other Other

The grammaticalization hypothesis can account for a number of languagespecific cases, as argued by Heine and Claudi (2001) for languages of the Kwa (Niger-Congo), Moru-Mangbetu (Central Sudanic), and Ju (a.k.a. “Northern Khoisan”) families. However, looking at the phenomenon only from this perspective runs into problems in many other cases; see Gensler (1997) for a detailed assessment of Claudi (1993). From a methodological point of view, it often requires one to invoke the ad-hoc idea that processes of structural generalization and/or analogical leveling that target the new O-V pattern have erased all or most traces of earlier grammaticalization. 3 Moreover, the validity of a grammaticalization scenario does not exclude other explanations, even within one and the same language. That is, the word order alternation in Africa might well be the result of “multiple causation”, and there is no a priori reason why it should be an entirely unitary phenomenon with a single evolutionary explanation for all its attested cases. The factual problems of the two grammaticalization scenarios have mostly to do with the assumed input features that make a change possible in the first place. As these will be illustrated in section 2 by examples from Benue-Congo languages, I will just list them at this point. First, the word order alternation may not be coupled, in a given language, with genitivenoun order and/or verb serialization. There are even cases where genitivenoun order actually holds but the object shift still occurs, according to the language-specific source, without the involvement of a genitive. Rather, the marked O-V structure represents a true preverbal object complement controlled by a nominalized verb (cf. in this respect German non-finite structures like Schuhe putzen ‘polishing, to polish shoes’, Fahrkarten kaufen

Preverbal objects and information structure in Benue-Congo 85

‘buying, to buy tickets’, etc.). These languages suggest the existence of at least one additional grammaticalization scenario, in which, however, the preverbal object position would remain an unexplained phenomenon: (c) >

[Object Verbx Auxiliaryx - Object -

Verbal nouny] Verby -

Other Other

A further empirical problem with the grammaticalization hypothesis is that the relevant word order alternation sometimes applies without the involvement of an auxiliary or can fail to apply despite the presence of an auxiliary; here, auxiliation cannot, without additional assumptions, be the triggering factor. Finally, the available grammaticalization scenarios have not properly addressed the fact that the word order alternations can concern constituents other than direct objects, including adverbial adjuncts, and/or multiple participants. This paper is an attempt to remedy some of the shortcomings of the currently dominant explanation for the alternating word order in African languages. It will look more closely at the phenomenon in a single language group, Benue-Congo 4 , where (a) the above problems associated with the exclusive grammaticalization approach are particularly salient and (b) the relevant word order changes are recurrently associated with a specific functional effect, namely that the marked preverbal position for objects (and other participants) is typical for information units that are less salient in the pragmatic context.

2.

Object order alternation in Benue-Congo

2.1. Family overview Benue-Congo languages have a fairly consistent head-initial word order (cf., e.g. Williamson and Blench (2000: 30–36) and the relevant chapters in Bendor-Samuel (ed., 1989) for overviews of the basic word order in these languages): (a) they show a basic V-O clause while O-V order is a marked alternative in specific grammatical contexts, and (b) the noun phrase displays a head-modifier pattern, which, except for Nupoid, also includes genitive constructions. This situation is in line with the currently prevailing reconstruction of Benue-Congo and Niger-Congo as S-V-O (Heine 1980,

86 Tom Güldemann Claudi 1993, Manfredi 1997). Nevertheless, as will be shown below, a possible alternation between post- and preverbal objects is widespread in the group. While a generally head-initial noun phrase would exclude the grammaticalization scenario under (a) in section 1 for almost all of BenueCongo, the development under (b) is also not viable for many languages, especially in the eastern part of the group, because they lack verb serialization of the type found in Kwa and western Benue-Congo. Another salient feature of several Benue-Congo languages is the partial irrelevance of auxiliaries for the occurrence of certain O-V structures. This property in particular has led me to pursue a slightly different approach to the general problem. That is, instead of looking only for the pattern S-Auxiliary-O-VOther, I have undertaken a general survey of O-V and related structures, irrespective of whether an auxiliary is involved. The result of this still preliminary investigation is given in Table 1. This list is not exhaustive because it omits several languages which are genealogically close to those given in the table and show similar behavior. Table 1: Attested cases of preverbal objects in Benue-Congo __________________________________________________________________ Language Group Sub-branch(es) Source ______________________________________________________________________________________________________________________________________________________________________

Nupe

Nupoid

Idoma Yoruba Igbo Ibibio Leggbo Kana Kaje Mambila Vute Tikar Naki Aghem Bafut

Idomoid Defoid Igboid Cross River Cross River Cross River Platoid Bantoid Bantoid Bantoid Bantoid Bantoid Bantoid

Nen Bantoid Ewondo Bantoid Proto-Bantu* Bantoid

Lower Cross Upper Cross Kegboid Central Plateau Northern, Mambiloid Northern, Mambiloid Southern Southern, Beboid Southern, Grassfields Southern, Grassfields Southern, Mbam Southern, Narrow Bantu Southern, Narrow Bantu

George 1971, Kandybowicz and Baker 2003 Abraham 1951 Manfredi 1997 Manfredi 1997 Urua 1997 Good 2004 Ikoro 1996 Wolff and Gerhardt 1977 Perrin 1994 Thwing and Watters 1987 Stanley 1991 Good 2005 Watters 1979 Chumbow and Tamanji 1994, Tamanji 2002 Mous 1997, 2005 Redden 1979 Meeussen 1967

______________________________________________________________________________________________________________________________________________________________________

Note: * except Northwest

Preverbal objects and information structure in Benue-Congo 87

In the following, I will present data from individual languages grouped according to certain recurrent factors that trigger the shift of an object, and partly of other participants, to the marked preverbal clause position. 2.2. Pronominal objects A first group of cases is characterized by the shift of PRONOMINAL objects. For example, this is reported in Ibibio (Cross River) for an apparently isolated context, namely reflexive experiencer pronouns. Compare (2), with a nominal object canonically following the verb, and (3), where the 1stperson singular object m@ precedes the experiencer predicate biák ‘hurt, be painful’; note that no auxiliary is involved in the construction. (2)

Òkôn á dêp ébót PROP 3S buy goat ‘Okon is buying a goat.’ (Urua 1997: 201)

(3)

úbç@k á m@ biák hand 3S 1S be.painful ‘My hand hurts.’ [lit.: ‘The hand hurts me.’] (Urua 1997: 204)

In Vute (Bantoid), the marked preverbal position of pronoun objects is more frequent in that it applies to human referents in all clauses with imperfective aspect. This is shown by the difference between (4) with a nominal object and (5) with a pronominal object, both of which involve an auxiliary. (4)

ƾgée júk yáyá taá sè-nà ni tá PROP PST:IPFV wash-IO ? father 3S.POSS clothes ‘Yaya was washing clothes for his father.’ (Thwing and Watters 1987: 100)

(5)

ni ƾgáb á ƾgé jìm 3P IPFV 3S beat ? ‘They are beating him.’ (Thwing and Watters 1987: 100)

Preverbal pronoun objects are even more salient in Kana (Cross River) since here the word order change is possible or obligatory for a wider range of preverbal auxiliaries and predicate inflections (Ikoro 1996: 206-13). One

88 Tom Güldemann marker which generally triggers the change encodes past tense; compare (6) where the 1st-person singular pronoun mE# precedes a serial verb structure. (6)

we#è mE#-te#e#ra# p"#" # 3S:PST 1S-run meet ‘He ran to me.’ (Ikoro 1996: 212)

However, example (7), which is semantically identical to (6), illustrates an important fact about the general phenomenon in Kana: as soon as the pronominal object is pragmatically salient, even a past tense clause no longer has the object in the marked preverbal position. (7)

te#e#ra# píí n#da# we#è 3S:PST run meet 1S.EMPH ‘He ran to ME.’ (Ikoro 1996: 212)

An even stronger correlation between the presence of an auxiliary and the preverbal position of a pronoun object is found in certain Plateau languages such as Kaje and Afusare (Wolff and Gerhardt 1977: 1527–8). Another example is Ewondo (Bantoid, Bantu A70), as described by Redden (1979: 126, 166–7). Compare (8), with a postverbal object pronoun in a clause without an auxiliary, and (9), with a preverbal object pronoun in a clause with an auxiliary (habitual kad); the latter example also shows that Ewondo is a language where preverbal position can apply to multiple objects. (8)

ma-v‹" Ջ Ջ" 1S-give 3S.H 3S ‘I am giving it to him.’ (Redden 1979: 166)

(9)

a-kad m‹ dzn v‹" 3S-HAB 1S 3S give ‘He usually gives it to me.’ (Redden 1979: 167)

2.3. Negation Another recurrent context for the preverbal occurrence of, often multiple, objects is negation. In Vute (Bantoid), this applies to perfective clauses, as shown in (10).

Preverbal objects and information structure in Benue-Congo 89

ƾgé cene (10) a. mvèìn yi ˜wáb-na tí chief REM.PST buy-IO ? 3S chicken ‘The chief bought him a chicken.’ ƾgé cene ˜wáb-na wá b. mvèìn yi ní chief REM.PST NEG 3S chicken buy-IO PFV.NEG ‘The chief did not buy him a chicken.’ (Thwing and Watters 1987: 101) In other languages, negative clauses cause the object to precede the verb in a more generalized fashion. Compare in this respect (11a) with (11b) from Leggbo (Cross River, Upper Cross), or (12a) with (12b) from Bafut (Bantoid, Grassfields). Similar cases are reported in other closely related languages, for example Lokaa (Iwara 1982) from Upper Cross, and Nweh (Nkemnji 1995: 112–47) and Mbili (Ayuninjam 1998: 338–53) from Grassfields Bantu. (11) a. wàdum s'" e-nii wà'" s'" ntààmi man DEF 3S-give child DEF gift ‘The man gave the child a gift.’ b. wàdum s'" wà'" s'" ntààmi eè-nii man DEF child DEF gift 3S:NEG-give ‘The man didn’t give the child a gift.’ (Good 2004) (12) a. Sùù kì kó mbà PROP TA catch animal ‘Suh killed an animal.’ kì wa#'à mbà ko# b. ka#a# Sùù NEG PROP TA NEG animal catch ‘Suh did not kill an animal.’ (Chumbow and Tamanji 1994: 224) 2.4. Non-finite verb phrases A third typical context for preverbal objects in Benue-Congo is non-finite verb phrases which may or may not be controlled by an auxiliary. This occurs recurrently inter alia in Western groups of Benue-Congo; it is reported for Nupoid (cf. Hyman and Magaji (1971: 28–33) for Gwari, and Kandybowicz and Baker (2003: 131–145) for Nupe), Idomoid (cf. Abraham (1951: 71–7) for Idoma), Defoid, and Igboid. A representative example is given here from Standard Igbo (Igboid).

90 Tom Güldemann (13)

óB mára-na bùbá a-kwá 3S know-PFV blouse NOM-sew ‘S/he knows how to sew blouses.’ (Manfredi 1997: 98)

Several facts should be noted in connection with the construction in (13). First, this particular combination of verbal noun and object cannot be used outside the present context after the controlling verb mára ‘know’ – for example, it cannot appear as a sentence subject. Second, this is the only pattern possible as long as the object is semantically specific. As soon as the object is generic and – similar to a cognate object – notionally contained in the verb, an interesting variability can be observed: one can either use the structure given in (13), as shown in (14a), or have recourse to a nominalization type with a different nominalizer, where the object FOLLOWS the verb, as in (14b). 5 (14) a. óB mára-na akwà a-kwá 3S know-PFV cloth NOM-sew b. óB mára-na iB-kwá akwà 3S know-PFV INF-sew cloth ‘S/he knows how to sew.’ (Manfredi 1997: 98) The following example of a non-finite verb phrase with a preverbal object is from Standard Yoruba (Defoid); according to the author, this particular nominalization type need not be preceded by a complement-taking verb like ‘learn’ in (15). (15)

ó kóB ìwé kí-kà 3S learn book NOM-read ‘S/he learned the art of reading.’ (Manfredi 1997: 96)

While the O-V order in this Yoruba construction is fixed with verbs implying a lower object that is generic, like ‘know’ and ‘learn’, the quite similar structure after the verb ‘want’ allows for both marked O-V, as in (16a), and normal V-O, as in (16b) (note again the different nominalizers). (16) a. mo 1S

féB want

asBoB cloth

óB-hun NOM-weave

Preverbal objects and information structure in Benue-Congo 91

b. mo féB é-hun asBoB 1S want NOM-weave cloth ‘I want to weave (some) cloth.’ (Manfredi 1997: 96) The Upper Cross languages Leggbo and Lokaa, already mentioned in Section 2.3, are also cases where preverbal objects are typical for nominalizations in general. As the following examples from Leggbo show, the feature is not only characteristic of verbal nouns controlled by another verb, like phasal ‘begin’ in (17); it also occurs when the nominalized verb phrase functions as the subject, as in (18). This last example also demonstrates that the alternation in object position not only can be independent of an auxiliary but also that it can apply to more than one participant. (17)

(18)

b a-ttnngn lídzil gèdziE$ 3P 3P-begin food eat:NOM ‘They began eating food.’ (Good 2004) n$tààmi bE$'" gèniE$ nEE$-dàa m gift children give:NOM 3S:HAB-please 1S ‘I like giving gifts to children.’ (Good 2004)

2.5. Preverbal inflections and auxiliaries As is well known from the literature on the topic, the shift between a postand a preverbal object is often tied to the absence or presence of an auxiliary or some other preverbal inflection. While in the cases discussed in the previous sections certain additional conditions pertain (pronominal object, negation, nominalization, etc.), the phenomenon can be more general as, for example, in Tikar (Stanley 1991: 102–28, 132–58). This Bantoid language has S-V-O order in perfective and modal clause types without a preverbal auxiliary; this pattern is shown in (19). (19)

wù sh-ê mùn 2S say-IRR 1S ‘Si tu me l’avais dit, ...’ (Stanley 1991: 71)

With preverbal inflections and auxiliaries, however, the object occurs before the verb, as with the imperfective marker tă in (20) and the phasal auxiliary Èunmi ‘begin’ in (21).

92 Tom Güldemann (20)

à tă nshe sh' 3S IPFV luggage carry ‘Il porte le bagage.’ (Stanley 1991: 103)

(21)

à Èunmi nun ji f' 3S begin:PFV 3S food give ‘Il a commencé à lui donner de la nourriture.’ (Stanley 1991: 133)

However, such a situation does not seem to be typical for Benue-Congo as a whole; more often O-V patterns are restricted to a smaller set of inflections/auxiliaries (see Manfredi’s (1997) discussion of “aspectual licensing and object shift”). Important for the following discussion is the observation that several of these cases seem to be sensitive to the pragmatic aspect of information structure. One such structure, fairly wide-spread in the Westernmost subgroups of Benue-Congo, is the likely result of the grammaticalization scenario mentioned in section 1 under (b) involving an earlier verb ‘take, have, hold, get’ (cf. Lord 1982). Nupe (Nupoid) is a case in point. Past clauses display a distinction between a plain V-O phrase, as in (22a), and an O-V structure, as in (22b); the latter involves an element á which derives etymologically from a verb ‘take’. tsùku$) (22) a. Musa zu@) PROP break stick ‘Musa broke the stick.’ b. Musa á tsùku$) zu@) PROP PERF stick break ‘Musa got the stick broken/ has broken the stick.’ (George 1971: 90, 93) While both sentences convey a perfective event, they differ in another respect: the first pattern with unmarked V-O would be a response to a question like “What did Musa do?” or “What did Musa break?” while the second structure with marked O-V answers “What did Musa do with (or to) the stick?”, where the object noun is given (also cf. Lord (1982) for other cases where the preverbal object correlates with definiteness). Since the áconstruction in Nupe is characterized by a focus on the predicate, including its aspectual operator of perfectivity, it has often been interpreted in terms of aspect, namely as a resultative~perfect.

Preverbal objects and information structure in Benue-Congo 93

The example pair in (23) shows the á-construction with a verb and a cognate-like object. Here, the object can either precede the verb or follow it. This suggests that the structure involves more than aspect and that the object position itself is more directly related to information structure. That is, it can be argued that the preverbal object itself characterizes sentences where the focus is on the predicate as the carrier of the assertion. (23)

Musa á eci bi or á bi (e)ci PROP PERF race run PERF run race ‘Musa (really) ran (well) / ran off.’ (Manfredi 1997: 100)

Verb serialization with ‘take’ is also at the base of the constructions with preverbal objects reported by Adive (1989: 80–1, 130–1) for Ebira, by Hyman and Magaji (1971: 56–8, 63) for Gwari (both Nupoid), by Manfredi (1997: 97) for OBwoBroB Yoruba (Defoid), and by Abraham (1951: 18) for Idoma (Idomoid, cf. (1)). While not all authors mention a functional difference vis-à-vis the unmarked clause with a postverbal object, it is likely that the situation is – at least from a diachronic perspective – similar to that in Nupe, where the marked O-V pattern can be characterized as a combination of perfectivity and focus on the predication instead of the object. Another case where a preverbal object seems to be associated with a tense-aspect form that involves a specific information structure can be identified in southern Igbo varieties (Igboid). For example, in Avu` Igbo, the future auxiliary gà can be followed by two phrase types involving different verbal nouns; in one, the object follows the nominalization, as in (24a), and in the other, the object precedes it, as in (24b). gà e-rí rin ahù` (24) a. ó` 3S FUT NOM-eat food that ‘S/he is going to eat that food.’ b. ó` gà rín ahù` n@-ri 3S FUT food that NOM-eat ‘S/he must (certainly) eat that food.’ [?< ... WILL eat ...] (Manfredi 1997: 99) The two structures are associated with a subtle meaning distinction which can be interpreted in terms of a difference in information structure: while the form with a postverbal object is a plain future, the one with a preverbal object renders an “(epistemic or deontic) obligative future.” Like the previ-

94 Tom Güldemann ous case of auxiliary perfects, this can arguably be analyzed as a combination of tense-aspect and information structure, namely as a future with focus on the truth value of the proposition; from this configuration the modal reading of obligation can be derived. 2.6. Information structure A final set of structures displaying variable object position in Benue-Congo languages is overtly and explicitly associated with an alternating information structure. To date, these structures have been reported exclusively from Bantoid languages which are spoken in a fairly restricted area along the Southern stretch of the border between Nigeria and Cameroon, that is, Mambila from Mambiloid (Perrin 1994), Naki from Beboid (Good 2005), Aghem from Grassfields (Watters 1979), and Nen from Mbam (Mous 1997, 2005). I will start with Aghem because it is the case first reported in the literature and has been researched fairly extensively. While the overall focus system is morphologically and syntactically far more complex than can be discussed here, the basic relevant facts, according to Watters (1979), can be summarized as follows. The linear position immediately after the verb is the site for focused constituents while the position before the verb can be used for pragmatically less salient participants; this mechanism is largely insensitive to different semantic relations. Thus, the sentence in (25a) is used when the exclusive focus is on the direct object, and the time adverb is given information. This pragmatic relation between the two non-subject participants is reversed in (25b); here, the time adverb is the new information, and the object is defocused. If both participants are extrafocal, both occur before the verb, which then ends up being clause-final, as in (25c). zí kí-bE@ (25) a. fíl á mç$ á'zç@ç friends 3P PST yesterday eat CL-fufu ‘The friends ate FUFU (not yams) yesterday.’ b. fíl á mç$ bE@-'kí zí á'zç@ç friends 3P PST fufu-CL eat yesterday ‘The friends ate fufu YESTERDAY (not two days ago).’ c. fíl á máà bE@-'kç@ á'zç@ç zí friends 3P PST.PF fufu-CL yesterday eat ‘The friends DID TOO eat fufu yesterday.’ (Watters 1979: 149, 148, 150)

Preverbal objects and information structure in Benue-Congo 95

A largely comparable situation has been described by Perrin (1994) for Mambila. A postverbal object is associated with focus, as shown in (26a). Extrafocal constituents are placed before the verb, as happens with the time adverb in (26b). It is also possible that both these constituents are focused, as in (26c); in such cases, the last one would be more salient. An overt defocalization of the object is illustrated in (26d) with exclusive focus on the time adverb and in (26e) with focus on the verb. (26) a. mè ƾgeé naâ cngn 1S buy PST cloth ‘I bought cloth.’ or ‘It was cloth that I bought.’ b. mè léílé ƾgeé naâ cngn 1S yesterday buy PST cloth ‘It was cloth that I bought yesterday.’ c. mè ƾgeé naâ cngn léílé 1S buy PST cloth yesterday ‘I bought cloth yesterday.’ d. mè cngn ƾgeé naâ léílé 1S cloth buy PST yesterday ‘It was yesterday that I bought cloth.’ e. mè naâ cngn ƾge 1S PST cloth buy ‘I bought cloth.’ [assertive verb focus] (Perrin 1994: 233) The situation in Nen appears to be quite different at first glance but it turns out to be comparable to the two previous cases. According to Mous (1997, 2005), the preverbal position for an object is unmarked and conveys assertive focus on this participant, as in (27a). The postverbal object, which must be preceded by a particle such as á, as in (27b), is associated with contrastive focus. índì (27) a. àná mònE@ 3S:PST money give ‘S/he gave money.’ b. àná índì á mònE@ 3S:PST give ? money ‘S/he gave MONEY.’ (Mous 1997: 126)

96 Tom Güldemann As opposed to Aghem and Mambila, the preverbal object in Nen is not extrafocal. However, on a more abstract level, there is still a common denominator across the three languages in that the preverbal object is pragmatically less salient vis-à-vis its postverbal counterpart. 2.7. Summary Table 2 gives a summary of the grammatical profile of O-V structures in the Benue-Congo languages treated above, as far as the information is available in the source. If a language has multiple O-V constructions which differ according to the features surveyed here, they are treated in separate lines. Two major observations can be made. First, many cases are characterized by the irrelevance of preverbal auxiliaries/inflections and by the possibility of subjecting more than one object or a non-object participant to the word order shift. These cannot, without further assumptions, be claimed to be instances of the grammaticalization scenarios mentioned in section 1 and thus call for another explanation. Second, O-V order in Benue-Congo is in general conditioned by RECURRENT grammatical factors. Some relate to a property of the affected object, in that it must be a pronoun, or refer to a generic concept, or have an extrafocal status. Others relate to the type of predicate in a wider sense: it is negated, has a certain tense-aspect feature, or is characterized by an auxiliary in general. Table 2: Contexts of preverbal objects in Benue-Congo

___________________________________________________________ Language (Section number)

Specific object feature

Mul- Non- Nega- Spe- Non- Pretiple object tion cific finite verbal objects TAM verb aux./ phrase infl. __________________________________________________________________ Nupe (2.4) YES YES ? YES (2.5) Non-focal NO ? PERF YES Idoma (2.4) ? ? NO (YES) (2.5) NO YES PST YES Stand. Yoruba (2.4) (Generic) ? NO (YES) OBwoBroB Yoruba (2.5) NO ? PERF YES Stand. Igbo (2.4) NO NO (YES) YES Avu` Igbo (2.5) ?Non-focal NO NO (FUT) YES YES Ibibio (2.2) (Pronoun) NO NO

Preverbal objects and information structure in Benue-Congo 97 Leggbo

(2.4) YES YES YES (2.3) YES YES YES Kana (2.2) Pronoun ? NO YES Kaje (2.2) Pronoun ? ? YES Mambila (2.6) Non-focal YES YES Vute (2.2) (Pronoun) NO NO (IPFV)YES (2.3) YES ? (YES) (PFV) YES Tikar (2.5) ? NO IPFV YES (2.5) YES NO YES Naki (2.6) Non-focal YES YES (2.3) YES ? (YES) YES Aghem (2.6) Non-focal YES YES Bafut (2.3) YES YES YES YES Nen (2.6) Less focal YES NO Ewondo (2.2) Pronoun YES YES YES Proto-Bantu (3) Pronoun (YES) (YES) YES __________________________________________________________________ Note: Blank = feature not relevant, (...) = restricted relevance, ? = no information

A look at other languages with object order alternations reveals that such factors are not restricted to Benue-Congo but can be identified elsewhere – in particular, including languages which are genealogically and/or areally unrelated. One context of preverbal objects noted above for BenueCongo but also found outside this group is progressive and imperfective clauses, for example, in some languages of Moru-Mangbetu (Central Sudanic, Andersen 1984, Watson 1997) and Gbe (Kwa, Lefebvre and Brousseau 2002); it is also found in Kaansa and Cefo (Gur, Elders p.c.). Negation triggers a different object position in some Gurunsi languages (Gur, Elders p.c.) as well as in parts of Surmic (Unseth 1986) – like Moru-Mangbetu, a family which is unrelated to Benue-Congo. To have a different position for pronominal vis-à-vis nominal objects is cross-linguistically quite common (cf. object clitics in Romance); it is also attested in other West African languages, like some of the Oti-Volta group (Gur, Elders p.c.) and most of the Atlantic family (Childs forth.). In some Potou-Tano (Kwa) languages and in Mandarin Chinese, the preverbal object is associated with the status of definiteness and discourse-givenness (Stewart 1963, Lord 1982). This has not been described explicitly for Benue-Congo; it is, however, noteworthy that the verbal object prefixes in Savannah Bantu, which arose from an earlier S-(Auxiliary)-O-V pattern, correlate with a definite postverbal object (cf., e.g. Creissels 2000: 235). Finally, information structure plays a

98 Tom Güldemann role in shifting object position in Doyayo (Adamawa, Elders 2004) and in the unrelated Southern Cushitic group (Mous 1993, Kießling 1994).

3.

Information structure as a potential common denominator of O-V

At first glance, these multiple factors seem fairly heterogeneous. I will, however, propose in the following that a common denominator does exist. I will start out from cases where O-V has the most straightforward function, namely where it correlates with some specific information structure. In languages like Aghem, Mambila, and Naki, the preverbal object is explicitly extrafocal, and the focus is on another constituent or on a clause feature. As mentioned, the situation in Nen is similar on a more abstract level: the preverbal object is not extrafocal but it is still less focal than its postverbal counterpart. I will now argue that a decrease in pragmatic salience of the object can also be identified for other, more grammaticalized contexts of O-V structures. A good case for this assumption can be made with regard to preverbal object pronouns. The use of a pronoun is largely correlated with discourse anaphora and implies the status of its referent as given/presupposed. This means that a pronoun object is not part of the new asserted information (recall in this respect the case of Kana in Section 2.2: as soon as the object pronoun is emphatic, O-V no longer applies). A similar argument can be made for preverbal definite objects; definiteness also implies a discoursegiven status, hence, such an object is extrafocal or less focal. The situation for preverbal generic objects is slightly different. Here, the object is not necessarily given information; it is, however, also not asserted independently but instead demoted so as to become a semantic subpart of a complex information unit. This analysis also applies in part to objects that occur within non-finite verb phrases. The non-asserted status in a kind of “incorporation” structure can correlate with a lower discourse saliency (see, e.g., Mithun 1984). At least since Hyman and Watters’ (1984) work on so-called “auxiliary focus”, it has been known that certain features of the predication, such as polarity, aspect, modality etc., can interact in a systematic way with the information structure of a clause. The authors propose that these specific predication operators have inherent focus. I follow this position and assume that some of the factors that trigger or favor O-V are related to this phenomenon.

Preverbal objects and information structure in Benue-Congo 99

That negation, in particular of the “metalinguistic” type (cf. Horn 1989), is an operator with inherent focus has been argued by a number of authors such as Givón (1978), Heine and Reh (1983), Marchese (1983), and Güldemann (1996, 1999). Thus, the fact that preverbal objects recurrently co-occur with negation can be taken as a reflex of their less focal status in this context. Other inherently focused features, according to Hyman and Watters (1984), are the progressive and the perfect. For the progressive, this has been discussed in more detail by Güldemann (2003) for Bantu languages and some other geographically and genealogically unrelated cases. Preverbal objects tied specifically to the progressive have not yet been found in Benue-Congo. However, certain phenomena which come close to this do exist (see also below regarding the historical relation between progressives and imperfectives). Recall from Section 2.5 that clauses in Tikar with imperfective tă generally have O-V order. However, the situation is slightly more complex with “semi-intransitive verbs”, which are characterized by optional locative complements. The position of such complements is variable and this is associated with a semantic distinction that implies a direct correlation between O-V and the progressive. If the complement follows the verb, the clause has a habitual reading, as in (28a); if it occurs in the marked preverbal position, the clause renders a progressive state of affairs, as in (28b). fumban (28) a. à tă kE$n 3S IPFV leave PROP ‘Il a l’habitude de partir à Foumban.’ b. à tă fumban kE$nni 3S IPFV PROP leave ‘Il est en train de partir à Foumban.’ (Stanley 1991: 114) Within the present framework, the cases of the perfect in Nupe, OBwoBroB Yoruba, and possibly other languages of the area, as well as the “obligative” future in Southern Igbo treated in section 2.5 are more straightforward. They can be analyzed as configurations of information structure where a predication operator is in focus; this causes the object to be pragmatically less salient and to occur outside its canonical postverbal position. I will conclude the discussion with those cases where preverbal object position is relevant for auxiliary periphrasis more generally and an exclusive explanation in terms of information structure is problematic. One

100 Tom Güldemann could, of course, first investigate the question of whether periphrasis has a general tendency to focus on the auxiliary category and thereby defocus participants. There are, however, other possible scenarios for how a language can come to apply an O-V order to many or even all periphrastic clause types, irrespective of whether the object has a decreased information value or not. For one thing, it is conceivable that the grammatical pattern O-V with preverbal predication operators becomes the target of analogical generalization and gradually expands in the language. A second possibility is that an inherently focused verb category develops into another category which is less sensitive to a particular information structure. In view of the results of grammaticalization research, this hypothesis is particularly attractive for an imperfective with general O-V, because this category frequently originates, according to e.g. Bybee, Perkins and Pagliuca (1994: 140–9), from a progressive, which I argue is inherently focused. In both scenarios, analogy and grammaticalization, the marked word order would be retained but the original explanation in terms of less focal objects would no longer need to apply. The general argument for the V-O > O-V alternation in the relevant languages is tied to an important point which should be stated in a more explicit fashion. I have repeatedly pointed out that preverbal position of the object or, more generally, of a non-subject participant is the marked pattern and that this is associated with a less salient information status of this element. Conversely, then, the position after the verb is the unmarked position for focus on a term like an object, adjunct, or adverb in the sense of Dik (1997); the focus is mostly assertive but in a few cases (also) contrastive, as in Mambila and Nen. Consequently, in the cases treated in section 2.6, where information structure is synchronically the factor transparently controlling the word order shift, the alternation not only applies to a single direct object but to multiple objects as well as to non-objects. This syntactically expressed pragmatic opposition lies at the heart of my hypothesis for the object order alternation: in a V-O clause, the object is (part of) the assertive focus while in an O-V clause the object is defocused or at least less salient vis-à-vis a postverbal object. The present account of the relevant word order alternations in terms of information structure does not contradict in principle the explanation of OV structure in terms of grammaticalization given in Section 1, that is, an origin of some language-specific structures in a concrete syntactic configuration of auxiliary + nominalized complement or verb serialization. Rather, it is compatible with and/or entails these historical scenarios insofar as they

Preverbal objects and information structure in Benue-Congo 101

ultimately cause the object to occur early in the clause and hence to iconically precede other, informationally more salient elements. The present proposal is also not meant to exclude the existence of other factors that may contribute to the existence of O-V structures. To mention only one other promising line of research for the future: the relatively frequent occurrence of pronominal items before the verb is likely to be related to their overall phonetic weight: as light elements they may tend to precede heavier ones like lexical verbs. While the order alternation regularly affects a wide range of constituents in some languages (see, e.g. section 2.6), the predominant situation in Benue-Congo is that it has a clear bias towards the object. Why should it be like that, given that an object is generally viewed as a grammatical relation that refers primarily to a semantic role implied by the verb? What is the special connection between the syntactic concept of object and the pragmatic notion of assertive clause focus that would cause pragmatically sensitive word order manipulations to target the object in particular? My basic idea is that all clause types are inherently associated with some type of information structure. This also applies to so-called “unmarked” clauses in fixed-word-order languages, which are often neglected in the discussion of focus phenomena. My more specific claim is that these clauses are unmarked not just in terms of basic syntactic configuration but also with respect to information structure. That is, they are “in-situ” focus constructions with a default configuration of assertive focus (= Dik’s (1997) “completive” focus). In V-O languages with a grammaticalized subject relation that conflates the semantic agent-role complex and the pragmatic function of topic, this makes it possible to assume an even more specific situation for basic clauses. All other things being equal, the default in an intransitive S-V sentence is assertive focus on the predicate, and in a transitive S-V-O sentence assertive focus on the object, possibly including the predicate. In the latter case, even if the verb is also new, the object can still be granted a higher information value in that the iconic linear progression is from less to more salient information. Moreover, and, given that in true O-V languages the preverbal object still seems to be more prominent than the final verb, possibly more importantly, nominal referents may be more central to the encoding of information structure than predicates. To the extent that the basic transitive clause of a language is a salient construction type, it can be argued that the grammatical relation between object and verb is inherently tied not only to certain semantic roles but also to a pragmatic function. The default position of an object is closely associ-

102 Tom Güldemann ated with its status as newly asserted information. This correlation can be exploited by a language: it can place the object in an unusual position in contexts where it has a non-canonical information status, i.e. where it is pragmatically given or less salient vis-à-vis a more marked focus on another constituent or a predication operator. This option can be seen as a direct function of at least two language-specific factors: (a) the degree to which a language allows for word order flexibility, and (b) the degree to which it highlights pragmatic function in addition to semantic role in the grammatical object relation. In proposing this fairly abstract common denominator for the relevant word-order alternations in general and for preverbal objects in particular, I do not want to claim that, synchronically, the explanation accounts for all cases in a straightforward way. Clearly, the recurrent occurrence of an object in preverbal position does not directly reflect its extrafocal information status but results initially from a particular grammatical construction. Moreover, these conditioning constructions are quite heterogeneous, even in one and the same language (cf., e.g. Nupe and similar languages where the ‘take’-serializations and phrases involving a verbal noun generate two quite different types of preverbal objects). Thus, while O-V under negation in the relevant languages comes fairly close to overt defocalization, auxiliary periphrases are overall different in that they often trigger O-V for purely grammatical reasons. Thus, control auxiliaries such as ‘know’, ‘want’, ‘start’, ‘be at’, etc. may simply require a non-finite verb phrase, which in the language may happen to require a preceding object complement. Also, in a core serialization, the patient object of ‘take, get, hold’ automatically ends up before the second verb and any additional participant. In both cases, the information status of the object vis-à-vis other elements is only indirectly reflected and may be an effect rather than the immediate cause. However, information status sometimes seems to be relevant even in such problematic cases. The Standard Yoruba examples (15) and (16) showed that the semantics of control verbs can bear on whether an object appears before or after the controlled content verb. The phrase ‘learn to VERB X’ often implies an activity that, once learned, is carried out routinely and then is associated preferably with a non-individuated, nonspecific, and thus pragmatically less salient object. The phrase ‘want to VERB X’ does not seem to have this habitual connotation and thus is more open with respect to the pragmatic status of the lower object. While the

Preverbal objects and information structure in Benue-Congo 103

former requires O-V order in the non-finite verb phrase, the latter allows for both O-V and V-O. A different situation can hold in the case of objects which are semantically entailed by the content verb: these are least likely to attract the clause focus and hence are the maximal opposite of a specific and/or referential object. In the case of ‘know to VERB X’ in Standard Igbo, this would seem to make a defocusing manipulation “superfluous” – as in (14), where V-O is possible besides normal O-V, in contrast to (13), where the semantically more specific object must precede the verb. The non-focus status of cognate-like objects might also be at the basis of the alternation in (23) from Nupe: here, the V-O pattern simply asserts a complex predicate while the O-V pattern arguably focuses on the truth of the predicate, which demotes the object even further in terms of salience. Since O-V patterns of some sort can pervade entire language groups, the present proposal has potential historical implications for earlier language states, and the question emerges as to what situation merits reconstruction for what chronolect. Gensler (1994, 1997) has gone some way in this direction by proposing to reconstruct the syntagm S-Auxiliary-O-V-Other, besides S-V-O-Other, back to Proto-Niger-Congo. The fuller range of data and their discussion in this paper throw new light on this issue. I would agree with Gensler in general that O-V is, alongside V-O, a historically older pattern in Niger-Congo and not just the ever-recurring outcome of a set of grammaticalization paths that just start out from V-O and its concomitant structures. At the same time, a reconstruction which entails such a morphosyntactic alternation should also address and answer questions regarding the distribution of and the ultimate motivation for this variability, and this on the basis of the particular syntactic profile of each modern group considered. Drawing from the above discussion, I would like to propose the following refinements to Gensler’s hypothesis. A reconstruction of a generalized S-(Auxiliary)-O-V-Other syntagm is likely to overstate the case for most sub-groups of Niger-Congo. A case in point is Proto-Bantu. The Northwest Bantu languages with preverbal pronouns like Ewondo and the modern morphotactic pattern in Savannah Bantu languages, where object affixes in a complex agglutinative verb usually precede the verb stem (cf. Meeussen 1967: 96–9), do not warrant such a far-reaching hypothesis. Instead, it would be more faithful to the available evidence to assume for the relevant proto-language a situation similar to that in Kana, Kaje, and Ewondo where

104 Tom Güldemann all or at least many auxiliaries triggered the positional shift of pronominal objects only. Such a situation with a clear grammatical restriction on O-V would plausibly have evolved from an earlier ancestor language with a similar, yet different profile. This can be modeled in a preliminary fashion on the empirical data presented here from Benue-Congo (irrespective of whether this group turns out after more research to be a valid genealogical sub-group within Niger-Congo). In order to account for the variation across BenueCongo, I posit that an earlier language state displayed a basic V-O syntax which, however, was different from that of “canonical” S-V-O languages in a few important respects. Sentence positions in this early language state, including the position after the verb, had besides their association with a typical semantic profile a greater pragmatic load than is usually recognized by standard syntactic analyses. This feature would have assigned to the postverbal object a stronger association with the role of assertive focus of the clause. This could well have gone hand in hand with a weaker syntactic bond between the verb and its object. Apart from possible external factors (e.g. contact with languages of a different word-order type), these properties would have facilitated the development of an optional O-V pattern which was associated with defocusing the object. Other groups in Greenberg’s Niger-Congo like Adamawa-Ubangi, Kwa, Gur, and Kru display an overall similar situation regarding a flexible object position (cf. Gensler and Güldemann 2003). These, together with BenueCongo, form what could be viewed as the more secure core of a genealogical unit (cf. Stewart 1976 and Bennett and Sterk 1977 regarding such concepts as “Volta-Congo” and “Central Niger-Congo”, respectively). Thus, the situation briefly outlined above may possibly have already taken place in a far older chronolect. Clearly, more research on this domain is necessary. I would venture, however, that a realistic and tangible morphosyntactic reconstruction of a proto-language is only possible if the whole range of synchronic variation in the daughter languages is taken into account and an attempt is made to relate the different patterns by means of observable functional correlations.

Notes

This paper was presented previously at the following occasions: at the conference “Topic and Focus: Information Structure and Grammar in Af-

Preverbal objects and information structure in Benue-Congo 105

1.

2.

3.

4.

5.

rican Languages”, University of Amsterdam (03/12/2004); at the “Work in Progress” series, MPI-EVAN Leipzig (14/12/2005); and at the “Jour fixe - Afrikalinguistisches Kolloquium” series, Universität Leipzig (15/12/2004). Thanks are due to the respective audiences for interesting discussions and to Orin Gensler and two reviewers for comments on a previous draft. The following abbreviations are used in the paper: CL noun class, DEF definite, EMPH emphatic, FUT future, H human, HAB habitual, INF infinitive, IO indirect object, IPFV imperfective, IRR irrealis, NEG negative, NOM nominalization, O object, P plural, PERF perfect, PF predication focus, PFV perfective, POSS possessive, PROP proper name, PST past, REM remote, S singular or subject, TA(M) tenseaspect-(modality), V verb. Elements of examples which are focused on in the discussion are bold-faced; question marks in the glossing indicate elements whose function is unclear to me. My comments within quoted material are given in square brackets. Accounts within the generative framework generally assume some form of constituent movement which I don't follow here (see Koopman (1984) on the phenomenon in Kru languages as one representative example). The elements which differ in their category status but are related to each other historically are co-indexed by sub-script letters. The term “Other” refers to postverbal participants other than direct objects. Cf., for example Heine and Claudi’s (2001: 64–8) discussion of the Mande case. Following Kastenholz (2003) and pace Claudi (1994), this family shows very little internal evidence for a transition from an earlier V-O to a modern O-V stage; rather the S-Auxiliary-O-V-Other pattern can be reconstructed to the proto-language. A major reason for entertaining an original V-O order and its later change to O-V by means of grammaticalization seems to be Mande’s assumed genealogical affiliation with NigerCongo - a hypothesis that still awaits the presentation of convincing evidence. Benue-Congo is used in the modern sense of “New Benue-Congo” (see, e.g. Bender-Samuel (ed.) 1989) without, however, any commitment to the hypothesis that this is a genealogical entity. Green and Igwe (1963: 167–71) distinguish these two non-finite verb forms by means of the terms “participle” (with prefixal a-/e-) and “infinitive” (with prefixal i-/iB-).

References Abraham, Roy C. 1951 The Idoma language. London: University of London Press.

106 Tom Güldemann Adive, John R. 1989 The verbal piece of Ebira. (Publications in Linguistics 85). Dallas/ Arlington: Summer Institute of Linguistics and University of Texas. Andersen, Torben 1984 Aspect and word order in Moru. Journal of African Languages and Linguistics 6: 19–34. Ayuninjam, Funwi F. 1998 A reference grammar of Mbili. Lanham/ New York/ London: University Press of America. Bendor-Samuel, John T. (ed.) 1989 The Niger-Congo languages. Lanham: University Press of America. Bennett, Patrick R., and Jan P. Sterk 1977 South Central Niger-Congo: a reclassification. Studies in African Linguistics 8: 241–273. Bybee, Joan L., Revere Perkins, and William Pagliuca 1994 The evolution of grammar: tense, aspect, and modality in the languages of the world. Chicago: The University of Chicago Press. Childs, G. Tucker forth. It’s everywhere! The S-AUX-O-V-OTHER syntagm in Atlantic. Studies in African Linguistics. Chumbow, Beban, and Pius N. Tamanji 1994 Bafut. In Typological studies in negation, Peter Kahrel and René van den Berg (eds.), 211–236. (Typological Studies in Language 29). Amsterdam: John Benjamins. Claudi, Ulrike 1993 Die Stellung von Verb und Objekt in Niger-Kongo-Sprachen: Ein Beitrag zur Rekonstruktion historischer Syntax. (Afrikanistische Monographien 1.) Köln: Institut für Afrikanistik, Universität zu Köln. 1994 Word order change as category change: the Mande case. In Perspectives on grammaticalization, William Pagliuca (ed.), 191–231. (Current Issues in Linguistic Theory 109). Amsterdam: John Benjamins. Creissels, Denis 2000 Typology. In African languages: an introduction, Bernd Heine and Derek Nurse (eds.), 231–258. Cambridge: Cambridge University Press. Dik, Simon C. [ed. by Kees Hengeveld] 1997 The theory of functional grammar, part 1: the structure of the clause. Berlin/ New York: Mouton de Gruyter. Déchaine, Rose-Marie, and Victor Manfredi (eds.) 1997 Object positions in Benue-Kwa: papers from a workshop at Leiden University, June 1994. (Holland Institute of Generative Linguistics Publications 4). The Hague: Holland Academic Graphics.

Preverbal objects and information structure in Benue-Congo 107 Elders, Stefan 2004 Distributed predicative syntax in Doyayo: constituent order alternations and cliticization. In Proceedings of the 4th World Congress of African Linguistics, New Brunswick 2003, Akinbiyi Akinlabi, and Oluseye Adesola (eds.), 189–197. Köln: Rüdiger Köppe. Gensler, Orin D. 1994 On reconstructing the syntagm S-Aux-O-V-Other to Proto-NigerCongo. Berkeley Linguistics Society (special session) 20: 1–20. 1997 Grammaticalization, typology, and Niger-Congo word order: progress on a still-unsolved problem. Journal of African Languages and Linguistics 18 (1): 57–93. Gensler, Orin D., and Tom Güldemann 2003 S-Aux-O-V-Other in Africa: typological and areal perspective. Paper presented at the 4th World Congress of African Linguistics, Workshop “Distributed predicative syntax (S P O V X)”, Rutgers University, June 21, 2003. George, Isaac 1971 The á-construction in Nupe: perfective, stative, causative, or instrumental? In Papers in African Linguistics, Chin-Wu Kim and Herbert Stahlke (eds.), 81–100. (Current Inquiry in Language and Linguistics 1). Edmonton/ Champaign: Linguistic Research, Inc. Givón, Talmy 1975 Serial verbs and syntactic change. In Word order and word order change, Charles N. Li (ed.), 149–188. Austin/ London: University of Texas Press. 1978 Negation in language: pragmatics, function, ontology. In Pragmatics, Peter Cole (ed.), 69–112. (Syntax and Semantics 9). New York/ San Francisco/ London: Academic Press. Good, Jeffrey 2004 When arguments become adjuncts: split configurationality in Leggbò. Unpublished manuscript: University of California Berkeley. 2005 A fixed-position focus construction in Naki. Paper presented at the International Conference on Focus in African Languages, ZAS Berlin, October 6–8, 2005. Green, Margaret Mackeson, and Georgewill Egemba Igwe 1963 A descriptive grammar of Igbo. (Veröffentlichungen des Instituts für Orientforschung 53). Berlin: Deutsche Akademie der Wissenschaften. Güldemann, Tom 1996 Verbalmorphologie und Nebenprädikationen im Bantu: Eine Studie zur funktional motivierten Genese eines konjugationalen Subsystems. (Bochum-Essener Beiträge zur Sprachwandelforschung 27). Bochum: Universitätsverlag Dr. N. Brockmeyer.

108 Tom Güldemann 1999

2003

forth.

The genesis of verbal negation in Bantu and its dependency on functional features of clause types. In Bantu historical linguistics: theoretical and empirical perspectives, Jean-Marie Hombert and Larry M. Hyman (eds.), 545–587. (CSLI Lecture Notes 99). Stanford: Center for the Study of Language and Information (CSLI). Present progressive vis-à-vis predication focus in Bantu: a verbal category between semantics and pragmatics. Studies in Language 27 (2): 323–361. The Macro-Sudan belt: towards identifying a linguistic area in northern Sub-Saharan Africa. In Africa as a linguistic area, Bernd Heine and Derek Nurse (eds.).

Heine, Bernd 1980 Language typology and linguistic reconstruction: the Niger-Congo case. Journal of African Languages and Linguistics 2: 95–112. Heine, Bernd, and Mechthild Reh 1983 Diachronic observations on completive focus marking in some African languages. Sprache und Geschichte in Afrika 5: 7–44. Heine, Bernd, and Ulrike Claudi 2001 On split word order: explaining syntactic variation. General Linguistics 38 (1): 41–74. Heine, Bernd, and Derek Nurse (eds.) 2000 African languages: an introduction. Cambridge: Cambridge University Press. Horn, Laurence R. 1989 A natural history of negation. Chicago: Chicago University Press. Hyman, Larry M., and Daniel J. Magaji 1971 Essentials of Gwari grammar. (Occasional Publications 27). Ibadan: Institute of African Studies, University of Ibadan. Hyman, Larry M., and John R. Watters 1984 Auxiliary focus. Studies in African Linguistics 15 (3): 233–273. Ikoro, Suanu M. 1996 The Kana language. (CNWS Publications 40). Leiden: Research School CNWS, Leiden University. Iwara, Alexander 1982 Phonology and grammar of Lokaa. M.A. thesis: University of London. Kandybowicz, Jason, and Mark C. Baker 2003 On directionality and the structure of the verb phrase: evidence from Nupe. Syntax 6 (2): 115–155. Kastenholz, Raimund 2003 Auxiliaries, grammaticalization, and word order in Mande. Journal of African Languages and Linguistuics 24 (1): 31–53.

Preverbal objects and information structure in Benue-Congo 109 Kießling, Roland 1994 Eine Grammatik des Burunge. (Afrikanistische Forschungen 13). Hamburg: Research and Progress. Koopman, Hilda J. 1984 The syntax of verbs: from verb movement rules in the Kru languages to universal grammar. Dordrecht: Foris. Lefebvre, Claire, and Anne-Marie Brousseau 2002 A grammar of Fongbe. (Mouton Grammar Library 25) Berlin: Mouton de Gruyter. Lord, Carol 1982 The development of object markers in serial verb languages. In Studies in transitivity, Paul J. Hopper and Sandra A. Thompson (eds.), 277–299. (Syntax and Semantics 15). New York/ San Francisco/ London: Academic Press. Manfredi, Victor 1997 Aspectual licensing and object shift. In Object positions in BenueKwa: papers from a workshop at Leiden University, June 1994, Rose-Marie Déchaine and Victor Manfredi (eds.), 87–122. (Holland Institute of Generative Linguistics Publications 4). The Hague: Holland Academic Graphics. Marchese, Lynell 1983 On assertive focus and the inherent focus nature of negatives and imperatives: evidence from Kru. Journal of African Languages and Linguistics 5 (2): 115–129. Meeussen, Achille E. 1967 Bantu grammatical reconstructions. In Africana Linguistica 3, 79– 121. (Annalen Wetenschappen van de Mens 61). Tervuren: Koninklijk Museum voor Midden-Afrika. Mithun, Marianne 1984 The evolution of noun incorporation. Language 60 (4): 847–894. Mous, Marten 1993 A grammar of Iraqw. (Kuschitische Sprachstudien 9). Hamburg: Helmut Buske. 1997 The position of the object in Tunen. In Object positions in BenueKwa: papers from a workshop at Leiden University, June 1994, Rose-Marie Déchaine and Victor Manfredi (eds.), 123–137. (Holland Institute of Generative Linguistics Publications 4). The Hague: Holland Academic Graphics. 2005 The innovative character of object-verb word order in Nen (Bantu A44, Cameroon). In Studies in African comparative linguistics with special focus on Bantu and Mande, Koen Bostoen and Jacky Maniacky (eds.), 411–424. (Collectie Menswetenschappen 169). Tervuren: Royal Museum for Central Africa.

110 Tom Güldemann Nkemnji, Michael 1995 Heavy pied-piping in Nweh. Ph.D. thesis: Department of Linguistics, University of California Los Angeles. Perrin, Mona J. 1994 Rheme and focus in Mambila. In Discourse features of ten languages of West-Central Africa, Stephen H. Levinsohn (ed.), 231– 241. (SIL Publications 109). Arlington: University of Texas. Redden, James E. 1979 A descriptive grammar of Ewondo. (Occasional Papers on Linguistics 4). Carbondale: Department of Linguistics, Southern Illinois University. Stanley, Carol 1991 Description morpho-syntaxique da la langue tikar (parlée au Cameroun). Société International de Linguistique. Stewart, John M. 1963 Some restrictions on objects in Twi. Journal of African Languages 2: 145–149. 1976 Towards Volta-Congo reconstruction (inaugural lecture). Leiden: Leiden University Press. Tamanji, Pius N. 2002 Negation, verb movement and word order in Bafut. Journal of West African Languages 29 (1): 45–64. Thwing, Rhonda, and John R. Watters 1987 Focus in Vute. Journal of African Languages and Linguistics 9 (2): 95–121. Unseth, Pete 1986 Word order shift in negative sentences of Surma languages. Afrikanistische Arbeitspapiere 5: 135–143. Urua, Eno E. 1997 Object movement in Eastern Lower-Cross. In Object positions in Benue-Kwa: papers from a workshop at Leiden University, June 1994, Rose-Marie Déchaine and Victor Manfredi (eds.), 189–206. (Holland Institute of Generative Linguistics Publications 4). The Hague: Holland Academic Graphics. Watson, Richard 1997 OV order in Ma’di? Occasional Papers in Sudanic Linguistics 7, 103–113. Nairobi: Summer Institute of Linguistics. Watters, John R. 1979 Focus in Aghem. In Aghem grammatical structure, Larry M. Hyman (ed.), 137–197. (Southern California Occasional Papers in Linguistics 7). Los Angeles: Department of Linguistics, University of Southern California.

Preverbal objects and information structure in Benue-Congo 111 Williamson, Kay, and Roger Blench 2000 Niger-Congo. In African languages: an introduction, Bernd Heine and Derek Nurse (eds.), 11–42. Cambridge: Cambridge University Press. Wolff, Ekkehard, and Ludwig Gerhardt 1977 Interferenzen zwischen Benue-Kongo- und Tschad-Sprachen. In Zeitschrift der Deutschen Morgenländischen Gesellschaft, Supplement 3: 1518–1543.

Focus strategies and the incremental development of semantic representations: Evidence from Bantu Lutz Marten

Abstract Bantu languages, such as Nsenga, Swahili, and Tumbuka, exhibit word-order variation associated with specific discourse-pragmatic contexts, such as topicalizing or focusing, both at the left and at the right periphery while expressing the same semantic or truth-conditional content. By employing the formal architecture of Dynamic Syntax (Cann et al. 2005), the paper proposes an analysis of topic and focus constructions that reflects the distinction between discourse-pragmatic function and truth-conditional meaning. Topic and focus interpretations are purely pragmatic notions, which arise from the combination of contextual information and independently available syntactic processes such as *Adjunction and LINK structures. The analysis thus dispenses with syntactically primitive notions of topic and focus.

1.

Introduction

In this paper, I discuss different word orders found in different Bantu languages which are associated with specific discourse-pragmatic contexts, such as topicalizing or focusing a particular constituent, both at the left and at the right periphery, but which express the same semantic or truthconditional content. I will argue that the distinction between discoursepragmatic function and truth-conditional meaning is important and should be reflected in the syntactic analysis of topic and focus constructions. In particular, I will show that Dynamic Syntax (Kempson et al. 2001, Cann et al. 2005), which models how hearers build semantic representations from the time-linear string of words encountered in context, provides the tools to express this distinction formally. One consequence of the Dynamic Syntax analysis is that notions like ‘topic’ and ‘focus’ are indeed pragmatic notions but are not part of the syntactic vocabulary of natural language (Kempson et al. 2004). Rather, syntactic configurations can be exploited for the ex-

114 Lutz Marten pression of topic and focus without the need to postulate dedicated syntactic topic and focus projections, as is often done in Principles and Parameters approaches to syntax (e.g. Rizzi 1997), or as primitive predicates in a feature structure matrix, as in Lexical Functional Grammar (e.g. Bresnan and Mchombo 1987). In this, I hope that the analysis presented in this paper raises an alternative to current analyses of topic and focus which will contribute to a better understanding of these notions. The paper is organized as follows: in Section 2, a short background discussion of the relevant notions and approaches to word order, topic and focus is provided, and a brief introduction to the tools of Dynamic Syntax is given. Section 3 is dedicated to the discussion of presentational focus, identificational focus, and background topics at the right periphery, while Section 4 looks at focus and topic at the left periphery. Section 5 presents conclusions from the analysis presented and indicates problems and directions for further research.

2.

Background

Before starting the discussion of word order and information structure in Bantu, a short background discussion is provided in this section about the position of the analysis I am going to develop with respect to wider work on topic and focus, as well as of the model of Dynamic Syntax and the tools which are relevant in the present context. 2.1. Syntax and information structure Information structure has become a central topic in syntax in the last two decades or so. That discourse-pragmatic functions like topic and focus play an important role for word order is especially clear when looking at Bantu languages. For example, Bresnan and Mchombo (1987) observe for Chichewa, that, in the presence of an object clitic functioning as an incorporated pronoun, all permutations of S, O, and V of a transitive clause are possible: 1 (1)

a. Njûchi zi-ná-wá-lum-a a-lenje 10.bees SC10-PAST-OC2-bite-FV 2-hunters ‘The bees bit them, the hunters.’ b. Zináwáluma alenje njûchi (VOS)

[Chichewa]

Focus strategies and semantic representations: Evidence from Bantu 115

c. d. e. f.

Alenje zináwáluma njûchi Zináwáluma njûchi alenje Njûchi alenje zináwáluma Alenje njûchi zináwáluma

(OVS) (VSO) (SOV) (OSV)

Bresnan and Mchombo’s (1987) main interest was to show that subject and object clitics in Chichewa can be analyzed as incorporated pronouns. 2 And while they discuss the relation of subject and object clitics with topicalized full NPs quite extensively, the discussion of focus is restricted to a few remarks. However, from their discussion it is clear that they assume that NPs can be co-indexed with focus and topic predicates at f-structure and that topics may stand in an anaphoric agreement relation with the subject and object clitics. Similar observations about word order freedom in Xhosa are found in du Plessis and Visser (1992: 13), who also draw attention to the relation between different word orders and prosodic information, a point which is further discussed in Downing et al. (2004, 2005), who, again using Chichewa, show that different word orders are related to specific patterns of phonological phrasing. However, no fully worked-out analysis of patterns like the one illustrated in (1) has yet been proposed. In the following sections, I will outline an analysis of different word orders using the Dynamic Syntax tools of LINK and *Adjunction and show how these relate different word orders to the context in which they are felicitous and to their possible discourse-pragmatic functions. Before doing this, however, I give a brief introduction to Dynamic Syntax in the next section. 2.2. Dynamic Syntax Dynamic Syntax (Kempson et al. 2001, Cann et al. 2005) models the process by which hearers construct semantic representations of content from words in context as a model of linguistic competence. Semantic representations, or ‘logical forms’, corresponding to the hearer’s representation of what she thinks is the intended message of the speaker are formally given in the model as annotated trees which transparently show the predicate argument structure of the proposition: 3

116 Lutz Marten (2)

Daudi likes Muna. Tn(0), Ty(t), Fo(like’(muna’)(daudi’)), ¸

Fo(daudi’), Ty(e)

Ty(e o t), Fo(like’(muna’))

Fo(muna’), Ty(e)

Fo(like’), Ty(e o (e o t))

The tree shows that verb phrase interpretation is a function of the interpretation of the main predicate and the interpretation of the object and that the interpretation of the sentence, the main proposition at the top node Tn(0), is a function of applying the interpretation of the subject to the interpretation of the VP. Note, however, that the trees do not show natural language syntax but purely semantic composition: the tree-nodes are decorated with semantic information from words but not by the words directly, and the structure of the tree only reflects semantic composition, e.g. predicateargument structure, without projecting word order as part of the tree. The fact that the predicates are on the right-hand side branches and arguments on the left-hand side branches is a matter of convention and not related to natural language syntax and word order. In fact, the whole point of the Dynamic Syntax enterprise is that natural language syntax reflects the way humans are able to build complex semantic structures like the one in (2) from a linear string of sounds. 4 Generalizations about syntax, like word order, grammaticality, and well-formedness, are expressed through the process of tree growth from a minimal tree as a starting point through a succession of partial trees to a fully annotated logical form: the dynamics of the system lie in the incremental mapping from linearly-ordered words to structured semantic representations, during which trees ‘grow’ as a result of syntactic transition rules or lexical actions. 5 For illustration, here is a sample derivation for the string in (3): (3)

Daudi likes Muna.

At the outset of the derivation, a minimal tree is assumed with just one node and no branches. The node is annotated with Tn(0), indicating that it is the root node, and ?Ty(t), indicating that at this node a requirement holds for an expression of type Ty(t), i.e. a proposition. This expresses hearers’

Focus strategies and semantic representations: Evidence from Bantu 117

justified expectation for information of propositional type which may interact with currently held assumptions (cf. Sperber and Wilson 1995). Finally, the diamond indicates that the node is the current node (somewhat trivially, since there is only one node so far): (4)

Tn(0), ?Ty(t), ¸

However, since in most cases not the whole proposition is communicated at once, syntactic rules license the introduction of subtasks in this situation: (5)

Tn(0), ?Ty(t) ?Ty(e o t)

?Ty(e), ¸

The rationale behind this move is that the satisfaction of both subtasks leads automatically to the satisfaction of the overall goal and that the subtasks may be easier to accomplish than the task at the root node. The fact that the argument node (on the left-hand side) becomes the current node, rather than the predicate node on the right, is a parametric value of SVO languages. 6 At this stage, the first lexical information is scanned, namely information from Daudi. Lexical information in Dynamic Syntax is modeled as procedural and as directly interacting with the tree annotations: (6)

‘Daudi’

IF THEN ELSE

?Ty(e) put(Fo(daudi’), Ty(e)) abort

The IF statement in the lexical information from Daudi states that the word can be introduced into the derivation if there is a current node with a requirement ?Ty(e). If this is so, then at that node ‘Fo(daudi’)’ and ‘Ty(e)’ can be added. On the other hand, if Daudi is parsed when the current node does not have a requirement for Ty(e), the parse ends. In the case at hand, the condition of the IF clause is met, as the current node in (5) has a requirement for Ty(e), and the tree can be further developed:

118 Lutz Marten (7)

Tn(0), ?Ty(t) ?Ty(e o t), ¸

Fo(daudi’), Ty(e)

The next step is the expectation of the development of the predicate node and the parsing of likes (ignoring tense and agreement for the moment): (8)

‘like’

IF THEN ELSE

?Ty(e o t) make(), put(Fo(like’), Ty(e o (e o t))), go(n), make(), go(), put(?Ty(e)) abort

The information from like shows that lexical information does not only decorate existing nodes but may also build new nodes. The actions of the THEN clause result in the building of a new predicate node and a corresponding argument node with a requirement ?Ty(e), which becomes the current node: (9)

Tn(0), ?Ty(t)

Fo(daudi’), Ty(e)

?Ty(e o t)

?Ty(e), ¸

Fo(like’), Ty(e o (e o t))

The next word is Muna, which comes with lexical information similar to that of Daudi, and can fulfill the requirement at the current node: (10)

Tn(0), ?Ty(t)

Fo(daudi’), Ty(e)

?Ty(e o t)

Fo(muna’), Ty(e), ¸

Fo(like’), Ty(e o (e o t))

Focus strategies and semantic representations: Evidence from Bantu 119

In the final tree of the derivation, all information established during the parse is accumulated, and all requirements are fulfilled: (11)

Daudi likes Muna. Tn(0), Ty(t), Fo(like’(muna’)(daudi’)), ¸

Fo(daudi’), Ty(e)

Ty(e o t), Fo(like’(muna’))

Fo(muna’), Ty(e)

Fo(like’), Ty(e o (e o t))

The tree is identical to the one in (2), corresponding to the logical form associated with the string Daudi likes Muna. 2.3. *Adjunction and LINK structures So far, the tree development, i.e. the order of transitions, matched the SVO word order of our example (recall that the arrangement of the nodes in the tree does not reflect word order). However, this is not always the case, as there is word order variation both between and within languages. For example, information may be presented early or late, or in other words, at the left or the right periphery. One way of modeling this is by employing structurally underspecified tree relations established by introducing unfixed nodes and dominated only by the root node by *Adjunction (named after the Kleene* operation, the reflexive-transitive closure over tree nodes): (12)

Muna, … ?Ty(t), ¸

Ty(t), Ty(e), Fo(muna’) In (12), the information from Muna is projected onto an unfixed node, which means that the node will be part of the eventual tree but that, at the time of the introduction of the word, it is not yet clear at which position it

120 Lutz Marten will be exactly. *Adjunction can be employed for the analysis of clauseinitial wh-words or for fronted NPs such as Muna in (13): (13)

Muna, Daudi likes.

After the introduction of the information from Muna at an unfixed node, the tree will be developed as usual. The unfixed node will remain part of the tree until a suitable stage in the derivation can be found at which the information at the unfixed node can be incorporated into the tree (which has to be found within the current tree). For example, a requirement might be introduced at the object node from the verb where the unfixed node can merge: (14)

Muna, Daudi likes. ?Ty(t)

Ty(t), Ty(e), Fo(muna’)

Ty(e), Fo(daudi’)

?Ty(e o t)

?Ty(e), ¸

Fo(like’), Ty(e o (e o t))

The eventual tree for (13) will be identical to the trees in (2) and (11), expressing the fact that, semantically, (3) and (13) are identical. On the other hand, the two analyses differ, not in the result but in the steps of transitions which were involved in deriving the final tree. It is this difference in derivation which expresses the pragmatic and information structure differences between the two examples, such that in (14), for example, *Adjunction can be used to express focus on the preposed NP. I will discuss this point in more detail below in relation to *Adjunction and the corresponding Late *Adjunction, which can be used to introduce information late. A second mechanism for introducing information outside of canonical position is called LINK transition. It allows for the building of two parallel trees, linked by a shared term, in which one tree can be exploited to provide a particular context for the other. LINK structures are used in Dynamic

Focus strategies and semantic representations: Evidence from Bantu 121

Syntax for example for the analysis of relative clauses, conjunction and topic constructions: (15)

Muna, Daudi likes her.

In examples like this one, the initial term is projected onto a linked tree which provides the background for the main tree to be developed. Formally, the tree to be developed carries a requirement that the formula value of the linked term be part of the new tree (?Fo(muna’)), thus ensuring that the new tree is built within the context set up by the LINK structure and that a term is shared across the LINK structure. Note that the LINK transition does not provide a copy of the formula value but merely introduces a requirement that such a copy should be part of the tree to be built, and so the presence of this information in the tree following the LINKed node has to be ensured by some other means, for example by the presence of a ‘resumptive’ pronoun: (16)

Muna, … Fo(muna’), Ty(e) LINK Tn(0), ?Ty(t), ?Fo(muna’), ¸

The main tree is then built as previously, but this time the object position is filled by a pronoun: (17)

Muna, Daudi likes her. Fo(muna’), Ty(e) LINK Tn(0), ?Ty(t), ?Fo(muna’)

Ty(e), Fo(daudi’)

?Ty(e o t)

Ty(e), Fo(U), ?x(Fo(x), Female(x)), ¸

Fo(like’), Ty(e o (e o t))

122 Lutz Marten The lexical specification of the pronoun is the same for both its ordinary use and the ‘resumptive’ use as in this example: it specifies its type and introduces an underspecified formula value with the metavariable U and the requirement that the metavariable be enriched to a full formula value (restricted by Female(x) to be female, following the gender specification in her). 7 This enrichment is a pragmatic process and may involve formula values from the context. The reason for the ‘resumptive’ interpretation is that any choice other than Fo(muna’) as formula value in object position will not satisfy the requirement at the root node that Fo(muna’) be part of the tree. However, once Fo(muna’) is chosen as the formula value, the tree will end up as being identical to (2) and (11) (except for the linked ‘topic’ node) – again, indicating that all of the examples (3), (13), and (15) are semantically identical but different in pragmatic meaning, which is expressed as a function of the interaction between the different transitions involved in the establishment of the final trees and the context. It is a general feature of Dynamic Syntax that grammaticality (and felicity) are determined by the set of transitions, and not by the final tree alone. With these tools at hand, I will now return to word order and information structure in Bantu, starting with focus (and topic) at the right periphery.

3.

The expression of focus and topic at the right periphery

The tendency to place new or focused information late in the utterance has often been observed, and examples of this tendency can also be found in Bantu languages, for example in presentational and identificational focus constructions. In presentational focus constructions in Nsenga, for example, the unmarked SV order is reversed, and the subject is introduced after the verb: 8 (18) a. À-léndò à-fwík-à 2-visitors SC2.PAST-arrive-FV ‘(The) guests have arrived.’ b. À-fwík-à à-léndò SC2.PAST-arrive-FV 2-visitors ‘There have arrived (the) guests.’

[Nsenga]

[Nsenga]

VS order expressing presentational focus is freely available in Nsenga and is not restricted to a subset of verbs (such as unaccusative verbs); it is also

Focus strategies and semantic representations: Evidence from Bantu 123

found with transitive verbs, with the subject following the verb and the object: (19) a. Kàtíshà ó-wéléng-à má-bùkù [Nsenga] Katisha SC1.PAST-read-FV 6-books ‘Katisha was reading books.’ b. Ó-wéléng-à má-bùkù Kàtíshà [Nsenga] SC1.PAST-read-FV 6-booksKatisha ‘Katisha was reading books.’ (‘There is reading books Katisha.’) Inversion structures like (18b) and (19b) are part of a number of constructions expressing new information or emphatic focus (also including locative inversion and subject-object reversal) and are used to emphasize the newness of (the referent of) the subject in the discourse. An example is provided in (20) from Swahili (from Bearth 1995: 198), where the two new participants of the story are introduced through the use of VS structures: (20)

... watu wa-ka-pand-a fiwi. Ka-j-a ... people SC2-CONS-plant-FV beans SC1.PERF-come-FV Talafa... A-ka-toke-a Mzee Mgomba ... [Swahili] Tafala… SC1-CONS-appear-FV Mzee Mgomba … ‘... the people were planting beans.’ ... ‘There came Talafa.’ ... ‘Then arrived Mzee Mgomba.’ ...

The analysis of these structures involves the interplay between the pronominal nature of Bantu subject clitics and structural underspecification, in particular Late *Adjunction. Following Bresnan and Mchombo (1987), I assume that subject clitics may function like pronouns, encoding a metavariable as formula value which can be updated from context (as well as restrictions as to the permissible kinds of substitutions, e.g. for class 2, that the substituend be human (Human(x)) and a group (PL(x)), cf. Cann et al. 2005, Marten and Kempson 2002 for further discussion of the analysis of Bantu agreement in DS). Thus, if it is clear that we are talking about guests, no overt subject is necessary: (21)

À-fwík-à SC2.past-arrive-FV ‘They (i.e. the guests) have arrived.’

[Nsenga]

124 Lutz Marten VS inversion structures can be used for focussing the post-verbal subject as well as in ‘afterthought’ topic constructions, discussed further below. VS structures involving focus arise in situations where it is not clear from the context who the intended referent of the subject clitic is. There may be various reasons for this. For example, the referent may be discourse new, as in (20), or may be provided as a response to a clarification question, giving rise to identificational focus (example (26) below). These different uses are a function of the particular context and receive the same structural, syntactic analysis in Dynamic Syntax, namely that the meta-variable projected from the subject clitic cannot be assigned a value, and the task to provide a suitable formula value remains open even after the predicate node is built: (22)

À-fwík-à .... (‘they came’) Tn(0), Tns(Past), ?Ty(t) ‘??????’ Ty(e), Fo(U) ?x(Fo(x), PL(x), Human(x)), ¸

Ty(e o t), Fo(fwík’)

The lack of an appropriate referent from the context (indicated by the question marks in the bubble) means that the parse cannot be completed after the introduction of the verb and that further information is necessary. This information is provided by the post-verbal subject, which is introduced into the parse by Late *Adjunction, i.e., the application of *Adjunction not at the outset but at the final stage of the parse when all nodes are typecomplete, but with an outstanding requirement for a formula value only. (23)

À-fwík-à (‘they came ...’)

Focus strategies and semantic representations: Evidence from Bantu 125

Tn(0), Tns(Past), ?Ty(t)

Tn(00), Ty(e), Fo(U), ?x(Fo(x), PL(x), Human(x)), ¸

Ty(e o t), Fo(fwík’)

Tn(00), ?Ty(e), ¸ Like *Adjunction, Late *Adjunction allows the building of an unfixed node, but this time the unfixed node is dominated by the argument node with the outstanding requirement (‘Tn(00)’). It is decorated with a requirement for a Ty(e) expression, a requirement which the information from àléndò can fulfill (for reasons of space, only the unfixed node is shown): (24)

Tn(00), Fo(àléndò’), Ty(e), ¸ As a final step, the unfixed node is merged at subject position, and the final tree has all requirements completed: (25)

À-fwík-à à-léndò (‘They came the visitors.’) Tn(0), Tns(Past), Ty(t), Fo(fwík’(àléndò’)), ¸

Ty(e), Fo(àléndò’)

Ty(e o t), Fo(fwík’)

The final tree of both the SV and the VS orders is identical, reflecting their identical predicate-argument structure. However, the intermediate trees leading to the derivation of the final tree differ, and the difference between the two versions lies only in the steps which have been taken to reach it. The claim is that, semantically, the two utterances in (18) (as well as those in (19)) are identical, and this is reflected in the identical final trees, but

126 Lutz Marten that they differ in pragmatic felicity. In particular, the VS order in the focus examples here works only in a context where the subject clitic cannot be fully interpreted from the context, and the postverbal subject is focused. The use of *Adjunction for VS structures means that the left and the right periphery are analyzed by the same structural means and that asymmetries between left and right periphery are a function of the incremental nature of structure building and context (cf. Cann et al. 2004). The advantage of this analysis is that it distinguishes between structural aspects of the left and right periphery, which are modelled as uniform, and the asymmetry between the two peripheries, which is explained in the present analysis as resulting from the difference in contextual information available at the outset and at the end of the parse. As mentioned above, the same structural analysis can be given to postverbal identificational focus, as in the following question-answer pair from Chichewa. Downing et al. (2005), from which the example is taken, comment on this that there is identificational focus on àlèéndó in the answer, as it gives a choice from a known list of possibilities: 9 (26) Q: À-ná-m-dyéts-á SC2-PAST-OC1-eat.CAUS-FV ‘Who fed him fish?’ A: À-ná-m-dyèts-à SC2-PAST-OC1-eat.CAUS-FV ‘The visitors fed him fish.’

nsóòmbà ndàáni ? [Chichewa] fish who nsóòmbà fish

àlèéndó 2-guest

The structural analysis in Dynamic Syntax for examples like these is identical to the analysis of presentational focus: the subject at the right periphery is introduced by Late *Adjunction. What is different in (26) from (20) is that a set of potential referents is available in the context, giving rise to an identificational reading. Notice that the analysis provides a formal, syntactic reflex of an observation often made in the literature on focus, namely that the focused element provides a value to an open proposition or gives rise to ‘alternative’ propositions (e.g. Lambrecht 1994, Rooth 1996, Bearth 1999). In the *Adjunction analysis, there is transparently a stage in the derivation at which the open proposition Ȝx[fwík’(x)] is entertained and a stage at which Ȝx[fwík’(x)](àléndò’) is entertained. Yet, semantically, in terms of truth conditions, both SV and VS end up as Fo(fwík’(àléndò’)). This analysis differs from analyses with designated focus (and topic) projections (as in

Focus strategies and semantic representations: Evidence from Bantu 127

P&P) or with primitive TOP and FOC attributes (as in LFG). The Dynamic Syntax claim is that structural configurations (such as late unfixed nodes) can be exploited for specific pragmatic effects, arising in the process of tree construction in a given context, but that information theoretic notions are not part of the eventual (semantic) representation (LF) or any other representational level (such as LF’, proposed by Vallduví 1990). Information structure and propositional structure can thus be seen as intertwined but distinct aspects of structure building in natural language. It is worth pointing out that there is another instance of VS order, and that is so-called afterthought constructions: (27)

À-fwík-à, à-léndò SC2.PAST-arrive-FV 2-guests ‘They have arrived, the guests.’

[Nsenga]

(28) Q: À-ná-m-tàání nyàání à-lééndó ? [Chichewa] SC2-PAST-OC1-do.Q baboon 2-visitors ‘What did the visitors do to the baboon?’ A: À-ná-m-dyèts-à nsóòmbà ! à-lèéndó SC2-PAST-OC1-eat.CAUS-FV fish 2-guests ‘The visitors fed him fish.’ (Downing et al. 2005) Afterthought constructions differ from focus constructions in their prosody, as they involve an intonational break between verb and subject (27) or are marked by tonal downstep (indicated by ‘!’) as in (28), and in their pragmatic felicity, as here the referent of the subject is not introduced as new into the context but is given as a reminder or clarification of the context; it is thus more (background) topical than focused. As the referent is not discourse new, the difference to the focus constructions discussed above is that here the subject clitic is interpreted from the context and that the postposed subject provides information to ensure that it is interpreted correctly. The emerging tree is thus different, in that, in afterthought constructions, the subject clitic is completed and the post-verbal subject needs to match the interpretation assigned to it; these constructions are thus in a sense the inverse of hanging topic constructions, discussed above, where a LINK relation ensures the identity of the left-dislocated topic and the ‘resumptive’ pronoun. Taking this parallelism as a starting point, the Dynamic Syntax analysis of afterthought topic constructions does not involve Late *Adjunction but the building of a LINK structure after the proposition has

128 Lutz Marten been built, with the requirement that the formula value of the linked node, i.e. Fo(àléndò’) in (27), be part of the proposition, which is fulfilled if the hearer has picked the right referent from context when interpreting the subject clitic. Like the use of *Adjunction, LINK structures provide a single structural means for introducing information at the left and right periphery, either as providing a context for an assertion yet to be developed or as clarifying an intended context as an afterthought. But these different functions are solely the result of contextual information available at the different stages in structure building reached when the linked node is built, and do not need to be stipulated as part of the structural specification of the LINK transition, as will also be seen when looking at the left periphery in the next section.

4.

The expression of focus and topic at the left periphery

In addition to the focus (and topic) effects on the right periphery, information structure can be expressed at the outset of the parse, at the left periphery, involving *Adjunction and initial LINK structures. However, there is a further point to be raised with respect to the left periphery, and that is the role of subjects and subject clitics in Bantu. The pronominal nature of Bantu subject clitics seems to indicate an analysis where subject clitics are projected directly onto the subject node of the emergent tree structure, or as locally unfixed nodes (Marten and Kempson 2002, Marten 2005). In either case, this would allow for the option that overt subject NPs are in general introduced at linked or unfixed nodes, an assumption supported by the word order flexibility of Bantu languages illustrated earlier. But this implies that the initial unfixed node is not available for other constituents when it is filled by the subject (as there can only be one unfixed node at any one time), and, furthermore, that pragmatic effects are reduced since the introduction of unfixed nodes is a standard strategy for introducing subjects. Still, initial focus examples can be found (Downing 2005: 6): 10 (29)

Ngô:ma ti-zamu-limilir-a 9.maize SC1PL-FUT-weed-FV ‘Maize we will weed tomorrow.’

namach!ê:ro tomorrow

[Tumbuka]

Focus strategies and semantic representations: Evidence from Bantu 129

(30)

Ma-bû:ku wa-ka-p!ás-a !wâ:na 6-books SC2-PAST-give-FV children ‘They gave the children books.’

[Tumbuka]

On the example in (29), Downing comments that maize is being contrasted with some other possible crop and on (30) that it is an answer to ‘What did they give to the children?’, indicative of contrastive and new information focus. Furthermore, prosodic evidence (the initial NP is in a separate phonological phrase) as well as the absence of an object clitic (otherwise strongly preferred with left-dislocated objects) support the status of the construction as focus related. In the Dynamic Syntax analysis, the initial constituent is projected onto an unfixed node by *Adjunction, as shown in the introduction. It is interesting that in both examples no overt subject is expressed, leaving the unfixed node to be used by the focused object. Another example of initial focus is the answer in (31) from Chichewa: (31) Q: Ndàání á-ná-m-dyéts-á nsóòmbà [Chichewa] Who SC1-PAST-OC1-eat.CAUS-FV fish ‘Who fed him fish?’ A: À-lèéndó à-ná-m-dyèts-à nsóòmbà 2-guests SC2-PAST-OC1-eat.CAUS-FV fish ‘The visitors fed him fish.’ (Downing et al. 2005) The subject àlèéndó is in its own phonological phrase, indicating new information focus, as is also indicated by the context, and contrasts with the subject àlèndó in (32) through the absence of the high tone on the penultimate syllable and penultimate lengthening, the two criteria Downing et al. (2005) identify as indicative of phonological phrase boundaries: 11 (32) Q: Kù-nà-chítík-á chìyáànì ? SC17-PAST-happen-FV what ‘What happened?’ A: À-lèndó à-ná-m-dyèts-à 2-guests SC2-PAST-OC1-eat.CAUS-FV ‘The visitors fed him fish.’

[Chichewa]

nsóòmbà fish

That the use of initial *Adjunction is possible for the expression of focus is further supported by examples with multiple focus, where both *Adjunction and Late *Adjunction are employed in the same structure:

130 Lutz Marten (33)

Pa-mu-pâ:nda zi-ka-d!úk-a mb!û:zi [Tumbuka] 16-3-wall SC10-PAST-jump-FV 10.goats ‘Over the wall jumped goats.’ (‘The goats jumped over the wall.’) (And something else jumped over something else.) (Downing 2005: 7)

In (33), both the locative NP pa-mu-pâ:nda and the subject mb!û:zi are contrasted with something else, are prosodically marked as constituting separate phonological phrases, and are hence, according to Downing (2005), focused. Both NPs are dislocated, at least under the reasonable assumption that the locative phrase is part of the VP, and so, in Dynamic Syntax terms, (33) can be analyzed as involving *Adjunction for the locative NP, which merges with a fixed node supplied by the predicate, and Late *Adjunction for the subject, which is introduced as unfixed with respect to the subject node which lacks a full interpretation from the context. The examples discussed here show that *Adjunction can be used for the expression of focus in Bantu, even though, for the reasons outlined at the outset of this section, other strategies are more common. One of these other strategies is, in fact, to employ LINK structures, more usually associated with topichood, to introduce new information: (34) Q: Ba-ntfwana, ba-ba-nik-e-ni ? 2-children SC2-OC2-give-PAST-what ‘What did they give to the children?’ A: Tin-cwadzi, ba-ti-nike ba-ntfwana 10-books SC2-OC10-give-PAST 2-children ‘Books, they gave (them) to the children.’

[Swati]

Both the question and the answer in (34) have a left dislocated NP which is co-referential with an object clitic. Bantfwana in the question appears to be a discourse topic and may as such be analyzed as linked to the main tree, where the information from the linked node is introduced through the object clitic -ba-. However, tincwadzi in the answer provides pragmatically new information, but it is co-referenced with an object clitic in the verb, which is commonly analyzed as being cross-referenced to topics, not to focused elements. From the Dynamic Syntax perspective, the structure may be analyzed as involving either a LINK structure, implying that there is no one-to-one correspondence between structural operations and pragmatic function, or as an unfixed node under the assumption that object clitics in

Focus strategies and semantic representations: Evidence from Bantu 131

Swati can be merged with information from full NPs. This latter option appears to be more in line with the argument presented here, but a full discussion will have to await a more in-depth Dynamic Syntax study of the function of object clitics in Swati, and in Bantu more widely.

5.

Conclusion

In this paper, I have shown how different structural possibilities at the right and left periphery are exploited for the expression of focus (and topic) in Bantu. On a theoretical level, my main concern was to show that the model does not assign any syntactic meaning to pragmatic notions like topic and focus, in contrast to most alternative analyses. From the Dynamic Syntax perspective, pragmatic effects arise from the particular building steps involved in constructing the semantic representation associated with the utterance but are not reflected in the final representation itself, thus providing a formal reflex of the distinction between propositional, semantic structure and pragmatic information structure. Furthermore, an important part of the Dynamic Syntax analysis is the way in which utterances are tied to the context, as identical structural analyses can represent different pragmatic readings (e.g. the difference between identificational and presentational focus) as a function of different contexts. From a functional perspective, this might be seen as an instance of the versatility of natural language, which expresses an infinite range of meanings by limited structural means. On the Bantu side, there are a number of questions outstanding, which have to be addressed on another occasion, partly due to reasons of space, and partly due to the fact that the relevant Dynamic Syntax analyses are still in progress. Amongst those, two in particular deserve a brief mention. First, all structures I have discussed show the expression of focus at the clausal periphery. However, it is well known that the immediate postverbal position in Bantu is associated with focus (e.g. Bearth 1999). The extension of the Dynamic Syntax analysis to such examples presupposes a Dynamic Syntax analysis of the Bantu VP, which has yet to be fully developed. 12 Second, with respect to presentational focus structures, another set of data needs to be mentioned, and that is presentational focus constructions with locative subject clitics:

132 Lutz Marten (35) a. Kwà-fwík-á à-léndò SC17.PAST-arrive-FVconj 2-visitors ‘There have arrived guests.’ b. *A-lendo kwa-fwik-a 2-visitors SC17.PAST-arrive-FV Intd.: ‘There have arrived guests.’

[Nsenga]

[Nsenga]

By assumption, the postverbal subject is introduced, like in the cases discussed in section 3, by Late *Adjunction. But the function of the subject clitic is more difficult to analyze, as it may be taken as a full locative clitic, as an expletive element, or as something in between the two. In addition, the verb form in (35) is marked as ‘conjoint’ by the high tone on the final vowel, in contrast to verbs in VS order with subject agreement as discussed here, presumably indicating that the following NP is part of the core clause. I will leave a more detailed discussion of examples like (35) to a future occasion, as it would require more space than is available here. To conclude, I hope to have shown the interest and relevance to questions of information structure of the new framework of Dynamic Syntax, and of data from Bantu, which, with about 400 different languages, remains a valuable and largely untapped resource for linguistic theory formation.

Notes

I would like to thank Laura Downing, Ruth Kempson, Nancy Kula, Al Mtenje, Laura Mutti, Clara Simanga, Nhlanhla Thwala and two reviewers for comments and discussion of the points made in the article. Swati examples were provided by Nhlanhla Thwala. Nsenga examples were provided by Clara Simanga during a research visit at the Centre of Language Studies of the University of Malawi in April 2004, whose hospitality is hereby gratefully acknowledged. Parts of this research were supported by AHRB award B/RG/AN8675/APN16312. 1. The following abbreviations are used in the glosses: SC = subject concord; OC = object concord; FV = final vowel; CONS = consecutive tense; PERF = perfect; CAUS = causative; FUT = future; Q = question particle. Numbers refer to noun classes. 2. More precisely, Bresnan and Mchombo argue that the object clitic is always an incorporated pronoun while the subject clitic can also function as a marker of grammatical agreement. Discussion of this point would lead too far afield, but see e.g. Demuth and Johnson (1989), Marten and Kempson (2002).

Focus strategies and semantic representations: Evidence from Bantu 133 3. Tree annotations in this example are: Tn – treenode identifier, Ty – logical type, Fo – formula value, ¸ (diamond) – indicating the current node. The type value ‘e’ stands for ‘entity’ and corresponds roughly to NPs, ‘t’ stands for ‘truth-evaluable’, that is, a proposition, corresponding to a sentence, and combinations of the two indicate functions, e.g. ‘e o t’, a function from an entity to a proposition, corresponding to VP interpretation, or to intransitive verbs. More annotation will be introduced in the course of the discussion. The type value is similar to the types used in type-logical grammar, and, correspondingly, formula values can be thought of as lambda terms. Like in Montague or Categorial Grammar (e.g. Steedman 2000), logical form is the only level of (syntactic and semantic) representation. 4. Or, more correctly, Dynamic Syntax is concerned with how information from the suitably phonologically parsed input-string is built up into larger structures. The parsing perspective adopted here is shared with models like Government Phonology (Kaye 1989) in phonology and with several pragmatic theories, in particular Relevance Theory (Sperber and Wilson 1995). 5. Note that, although the perspective adopted is related to parsing, the Dynamic Syntax model is not a parsing model in the traditional sense, as crucially, Dynamic Syntax does not presuppose an independently defined competence model. Rather, the claim is that the dynamics of building semantic representations is all there is to syntax. 6. For more discussion of pointer movement see Cann et al. (2005). 7. In addition, locality restrictions on interpretation take care of binding effects, but I ignore those here; see Cann et al. (2005). 8. The Nsenga past tense marker à fuses with the preceding subject concord as follows: SC2 à + à = à; SC1 ú + à = ó; SC17 kù + à = kwà. 9. Chichewa examples are tone marked as in the source. In the following examples (as well as in examples (28) and (31)), there appear to be tonal differences between the questions and answers. Although Downing et al. (2005) do not comment on this, the difference may be related to the expression of the different clause types. 10. It is sometimes claimed that initial focused constituents are impossible in Bantu since Bantu languages express focus on the right periphery. However, the Dynamic Syntax analysis seems to be preferable in that it allows initially focussed constituents, but also has something to say about why they are rare. 11. I am grateful to Al Mtenje for making this point clear to me. 12. But see Marten (2002) for some Dynamic Syntax discussion of verb phrase structure and Bantu applicatives.

134 Lutz Marten References Bearth, Thomas 1995 Wortstellung, Topik und Fokus. In Swahili - Handbuch, Gudrun Miehe and Wilhelm J.G. Möhlig (eds.), 173–205. Köln: Köppe. 1999 The contribution of African linguistics towards a general theory of focus. Update and critical review. Journal of African Languages and Linguistics 20: 121–156. Bresnan, Joan, and Sam Mchombo 1987 Topic, pronoun, and agreement in Chichewa. Language 63: 741– 782. Cann, Ronnie, Ruth Kempson, Lutz Marten, Masayuki Otsuka, and David Swinburne 2004 On the left and on the right. In Peripheries: Syntactic Edges and their Effects, David Adger, Cécile de Cat, and George Tsoulos (eds.), 19–47. Dordrecht: Kluwer. Cann, Ronnie, Ruth Kempson, and Lutz Marten 2005 The Dynamics of Language: An Introduction. Oxford: Elsevier. Demuth, Katherine, and Mark Johnson 1989 Interaction between discourse functions and agreement in Setawana. Journal of African Languages and Linguistics 11: 21–35. Downing, Laura 2005 The prosody of some focus-related enclitics in some Southern Bantu languages. Paper presented at the 5th Bantu Grammar: Description and Theory Meeting, SOAS. Handout available on-line at http://mercury.soas.ac.uk/users/lm5/bantu_project.htm. Downing, Laura, Al Mtenje, and Bernd Pompino-Marschall 2004 Prosody and information structure in Chichewa. ZAS Papers in Linguistics 37: 167–186. Available on-line at http://mercury.soas.ac.uk/ users/lm5/bantu_project.htm. 2005 Non-accentual prosodic cues to focus in a tone language: the case of Ntcheu Chichewa. Paper presented at Between Tone and Stress, University of Leiden, 16–18 June 2005. Du Plessis, J.A., and M. Visser 1992 Xhosa Syntax. Pretoria: Via Afrika. Kaye, Jonathan 1989 Phonology: A Cognitive View. Hillsdale, NJ: Lawrence Erlbaum. Kempson, Ruth, Ronnie Cann, and Jieun Kiaer 2004 Topic, focus and the structural dynamics of language. Ms. Available on-line at http://semantics.phil.kcl.ac.uk/ldsnl/papers/. Kempson, Ruth, Dov Gabbay, and Wilfried Meyer-Viol 2001 Dynamic Syntax: The Flow of Natural Language Understanding. Oxford: Blackwell.

Focus strategies and semantic representations: Evidence from Bantu 135 Lambrecht, Knut 1994 Information Structure and Sentence Form. Cambridge: Cambridge University Press. Marten, Lutz 2002 At the Syntax-Pragmatics Interface: Verbal Underspecification and Concept Formation in Dynamic Syntax. Oxford: OUP. 2005 Passive, locative inversion and subject-object reversal. Ms. SOAS. Marten, Lutz, and Ruth Kempson 2002 Pronouns, agreement, and dynamic construction of verb phrase interpretation: A Dynamic Syntax approach to Bantu clause structure. Linguistic Analysis 32: 471–504. Rizzi, Luigi 1997 The fine structure of the left periphery. In Elements of Grammar, Liliane Haegeman (ed.), 281–337. Dordrecht: Kluwer. Rooth, Mats 1996 Focus. In The Handbook of Contemporary Semantic Theory, Shalom Lappin (ed.), 271–297. Oxford: Blackwell. Sperber, Dan, and Deirdre Wilson 1995 Relevance: Communication and Cognition. 2nd ed. Oxford: Blackwell. Steedman, Mark 2000 The Syntactic Process, Cambridge, Mass.: MIT Press. Vallduví, Enric 1990 The Information Component. Unpublished PhD dissertation, University of Pennsylvania.

Part III Ex-situ and in-situ strategies of focus marking

Ex-situ focus in Kikuyu Florian Schwarz

Abstract This paper discusses ex-situ focus constructions in Kikuyu, focusing on the particle ne which thereby plays a crucial role. The two analyses of ne proposed in the literature, the focus phrase analysis (Clements 1984; Schwarz 2003) and the cleft analysis (Bergvall 1987), are compared in detail. I argue that the focus phrase analysis is more successful in accounting for a number of central properties of focus constructions with ne. Among other things, it accounts for focus projection and the relation between in-situ and ex-situ focus. At the same time, these points constitute serious problems for the cleft analysis.

1.

Introduction 1

Kikuyu 2 has in-situ and ex-situ focus constructions and wh-constructions. Both the focus and the wh-constructions exhibit the same syntactic pattern. This chapter is primarily concerned with the syntactic analysis of ex-situ focus constructions. A crucial player in the ex-situ focus constructions is the particle ne. It appears in a number of places, namely before the focused phrase (or the fronted wh-phrase), in simple copula constructions, and in the immediately preverbal position in certain declarative sentences. Accounting for its distribution and function is a central task for any analysis of the ex-situ focus and wh-constructions. Much of this chapter is therefore concerned with the syntactic analysis of the ne-constructions. The two analyses that have been proposed to account for the distribution of ne are the focus phrase analysis (Clements 1984; Schwarz 2003) and the cleft analysis (Bergvall 1987). By comparing the two analyses, I argue that the focus phrase analysis, though not without problems, is by far more promising than the cleft analysis, which faces a number of serious problems. The chapter is organized as follows: Section 2 provides a brief overview of the crucial properties of ne. Section 3 presents the focus phrase analysis and further data supporting it, concerning focus projection and the relation

140

Florian Schwarz

between in-situ and ex-situ focus. The cleft analysis and the problems it faces are discussed in section 4. Section 5 gives a conclusion.

2.

Properties of ne

The particle ne appears in three environments: in ex-situ focus and whconstructions, in simple copula constructions, and preverbally in regular declarative sentences. Examples of these are presented in section 2.1. Ne is also subject to some crucial distributional restrictions, which are discussed in section 2.2. 2.1. Environments in which ne appears An example of the first environment in which ne appears, namely ex-situ focus and wh-constructions, is given in (1): 3 (1)

a. PG³ MGG #DFWN C³ TC³ P[W³ KT³ '? FM- what A. SM- T- drink- ASP- FV ‘What did Abdul drink?’ b. PG OCG #DFWN C³ TC³ P[W³ KT³ ' FM 6.water A. SM- T- drink- ASP- FV ‘Abdul drank WATER.’

In the ex-situ question in (1a), the focus marker ne combines with the question word kee in the sentence initial position. The sentence in (1b) is a possible answer to this question, and here the object mae ‘water’, which is focused due to the preceding question 4 , appears in the same position as the question word in (1a), adjacent to ne. All questions, except for subject questions, also have an in-situ version, which does not contain ne, as shown in (2): (2)

a. #DFWN C³ TC³ P[W³ A. SM- Tdrink‘What did Abdul drink?’ b. #DFWN C³ TC³ P[W³ A. SM- Tdrink‘Abdul drank WATER.’

KT³ ASP-

' FV

MGG? what

KT³ ASP-

' FV

OCG 6.water

Ex-situ focus in Kikuyu 141

The discussion in this paper will mostly concern the ex-situ cases, but the relationship between in-situ and ex-situ constructions will be relevant in the discussion of focus projection in section 3.3. One important point about the ex-situ constructions is that ne can be preceded by other material (both in focus and wh-constructions), as was first noted by Schwarz (2003). Examples of this are given in (3) where a topicalized subject (3a) and a topicalized adverbial clause (3b) appear before the fronted object with ne: 5 (3)



a. #DFWN PG OCG C³ TC³ P[W³ KT³ ' A. FM 6.water SM- Tdrink- ASP- FV ‘Abdul drank WATER.’ b. OD'T' [C P[QODCPG OCG #DFWN C³TC³ P[W³KT³ ' in-front 9.A 9.house FM 6.water A. SM-T- drink-ASP- FV ‘In front of the house, Abdul drank WATER.’

These examples will play a crucial role in the argument for the focus phrase analysis developed in the following section. The second construction involving ne is that of a simple copula clause. An example is given in (4a): (4)  

a. CDFWN *(PG) Ø OQ³ A. FM COP 1‘Abdul is a teacher.’ b. CDFWN (PG) C³ C³ TG A. FM SM- T- be ‘Abdul was a teacher.’

TWVCPK teacher OQ³TWVCPK 1- teacher

The obligatory presence of ne in (4a) might suggest that ne itself can function as the copula. However, once we consider cases that are not in the third person present tense form, ne is no longer obligatorily present (although it still can precede the copula verb), as can be seen in (4b). Instead, the copula verb stem re appears with the usual inflectional morphology. 6 The analysis commonly adopted for this pattern is that the underlying form of (4a) contains a phonologically null form of the copula verb, as indicated in (4a) (cf. Bergvall 1987, Clements 1984, Schwarz 2003). A central question in this respect is why ne is obligatory when the verb is phonologically null. I propose a tentative answer to this at the end of section 3.3.

142

Florian Schwarz

According to this analysis, ne appears in the immediately preverbal position in copula constructions. These are then a special case of the last environment in which ne appears, namely preverbally in regular declarative and interrogative sentences (which only differ in intonation), as is illustrated in (5): (5)

a.#DFWN (PG)³ C³ TC³ P[W³ A. FM- SM- T- drink‘Abdul drank water.’ b.#DFWN (PG)³ C³ TC³ P[W³ A. FM- SM- T- drink‘Did Abdul drink water?’

KT³ ' ASP- FV

OCG 6.water

KT³ ' ASP- FV

OCG? 6.water

While ne is not obligatory in these sentences as far as the syntax is concerned, leaving it out changes the interpretation with respect to what is in focus. I will discuss this in more detail in section 3.3. 2.2. Distributional restrictions of ne There are several restrictions on the distribution of ne: it can only appear once per clause, its distribution in embedded clauses is limited, and it cannot co-occur with the regular verbal negation marker ti. The first point is illustrated in (6): (6)

a. *PG OCG #DFWN FM 6.water A. ‘Abdul drank water.’

PG³ C³ TC³ P[W³ KT³ ' FM- SM- T- drink- ASP- FV

This sentence is fine when ne only appears in one of the two positions, but ungrammatical as soon as it appears in both. The second restriction is that ne cannot appear in certain embedded clauses. For example, the sentence in (7a) becomes ungrammatical when ne is added in the relative clause (see (7b)): (7) 

a.OQ³TWVCPK Q³ &QO³KT³ ' K³ $WMW 1- teacher SM- read- ASP- FV 5- book PG³ C³ TC³ P[W³ KT³ ' OCG FM- SM- T- drink- ASP- FV 6.water ‘The teacher who read a book drank water.’

Ex-situ focus in Kikuyu 143



b.* OQ³ 1PG³ FM-

TWVCPK PG³ Q³ &QO³ KT³ ' K³ $WMW teacher FM- SM- readASP- FV 5- book C³ TC³ P[W³ KT³ ' OCG SM- T- drink- ASP- FV 6.water

Clauses that are embedded by a bridge verb (i.e. think, know, say, etc.), on the other hand, do allow ne. This is not surprising since such clauses behave in many ways like matrix clauses. The last restriction concerns co-occurrence with verbal negation. The regular negation marker ti appears in the verbal complex between the subject marker and the tense marker. When ti is present in this position, inserting ne into the sentence leads to ungrammaticality: (8)



a. #DFWN C³ VK³ TC³ P[W³ KT³ ' A. SM- NEG- Tdrink- ASP- FV ‘Abdul didn’t drink water.’ b.* #DFWN PG³ C³ VK³ TC³ P[W³ KT³ ' A. FM- SM- NEG- T- drink- ASP- FV Intended meaning: ‘Abdul didn’t drink water.’ c.*PG OCG #DFWN C³ VK³ TC³ P[W³ FM 6.water A. SM- NEG- TdrinkIntended meaning: ‘Abdul didn’t drink water.’

OCG 6.water OCG 6.water KT³ ' ASP- FV

Note, however, that there is an alternative negation marker ta which can occur on the main verb with ex-situ focus constructions. I will come back to this when discussing the problems of the focus phrase analysis.

3.

The focus phrase analysis

The main challenge posed by the data presented in the preceding section is to account for the different types of occurrences of ne in a unified manner while also making the correct predictions about its distributional restrictions. The focus phrase analysis deals with this challenge by assuming that ne appears in a syntactic focus phrase within an extended CP-projection (Brody 1990; É. Kiss 1998; Rizzi 1997). The position that ne appears in is always the same then, and the different constructions involving ne are derived by having different elements move to the focus phrase. The general structure that this account is based on is the following:

144 (9) 

Florian Schwarz

FP ei SpecFP F' ei XPF F YP [+F] 6 … XPF …

A strong feature in the head of the focus phrase triggers the movement of an XP bearing a focus feature to the specifier of the focus phrase. There are two slightly different possible theoretical implementations with respect to ne. First, ne might be the head of the focus phrase (a common assumption for focus markers). We then have to say that it cliticizes onto the material in its specifier to get the right word order, namely ne XP (cf. Muriungi 2004 for a proposal along these lines for the closely related language Kitharaka). Alternatively, we could say that the focus feature on the XP gets spelled out as ne when it appears in the specifier of the focus phrase. Most of the following is compatible with either of these accounts, and I will simplify the representations by putting ne XP in the specifier of the focus phrase. 3.1. Accounting for the different occurrences of ne Based on the idea that ne appears in a syntactic focus phrase, how can we account for the different constructions involving ne in detail? Let us first turn to the ex-situ focus construction. Take the example of (3b) above, repeated here as (10a). (10) a. #DFWN PG OCG C³ TC³ P[W³ KT³ ' A. FM 6.water SM- T- drink- ASP- FV ‘Abdul drank WATER.’ The fronting of the object mae ‘water’ can be captured by moving it to the specifier of the focus phrase. Furthermore, the subject abdul is topicalized so that it occurs in the sentence initial position. We can then represent the derivation (with many simplifications) as follows: 7

Ex-situ focus in Kikuyu 145

(10) b. 

CP ei SpecCP FP #DFWN ei SpecFP IP PGOCG ei SpecIP VP ei #DFWN V DP CTCP[WKT' OCG

When the subject remains in its base position, we get the ne-initial order found in (1b). The case of preverbal ne is derived in a similar fashion. In this case, the entire IP moves to the focus phrase, and the subject moves on to the same topic position as in (10b). Moving the entire IP into the focus phrase is motivated by the fact that immediately preverbal ne expresses focus on the entire sentence (as will be discussed in more detail below). The sentence in (5a), repeated below, can then be analyzed as in (11b) 8 : (11) a.#DFWN PG³ C³ TC³ P[W³ KT³ ' A. FM- SM- Tdrink- ASP- FV ‘Abdul drank water.’ b.

OCG 6.water

CP ei SpecCP FP  #DFWN q SpecFP PG+ IP IP ei 6 SpecIP VP #DFWNCTCP[WKT'OCG #DFWN ei V DP  CTCP[WKT' OCG One difference between this case and the ex-situ focus case in (10) is that the topicalization of the subject is obligatory here. This does not fall out of the theory at this point. The only explanatory speculation that I can offer in this respect at the moment is that if the subject was not topicalized, the structure would be string identical to the ex-situ focus construction in (1b),

146

Florian Schwarz

and hence the formal marking of different foci would be less perspicuous. Assuming that copula constructions are a special case of preverbal ne, their analysis will be as in (11b) with the only difference that the head of the verb phrase is phonologically null in the third person singular case. 3.2. Accounting for the distributional restrictions The preceding section has shown how the focus phrase analysis can account for the different occurrences of ne. Now we need to make sure that we can also account for the distributional restrictions. First, why is it that ne only appears once per clause? According to the focus phrase analysis, this is simply because ne is tied to a particular syntactic phrase which only appears once per clause. The second question is why ne cannot appear in relative clauses. The focus phrase assumed in the above analysis is part of the extended Csystem. There seems to be good evidence indicating that the fully extended C-system is not present in relative clauses. For example, topicalization is not possible in relative clauses either, which is just what we expect if the topic position is also part of the extended C-system. The absence of ne in relative clauses is then simply a reflection of the absence of the extended Csystem. Finally, we have to explain why ne cannot co-occur with the regular negation marker. The answer to this question is not so obvious. One tempting possibility might be to say that ne and the negation ti appear in the same syntactic position. However, this is hard to reconcile with their surface distribution (see (8a)). Another possibility is to say that negation is somehow inherently linked to focus (cf. Hyman 1999), but it is unclear how to spell this out in detail in the present framework. Perhaps one option would be to say that the negative head is capable of checking the focus feature on the fronted element and thus make the focus phrase unnecessary. Unfortunately, I cannot explore this option in more detail here. So, for present purposes, it must suffice to say that while a more detailed answer has to be developed by further research, there is no reason to believe that this issue poses a problem that is particular to the focus phrase analysis.

Ex-situ focus in Kikuyu 147

3.3. Focus projection: In-situ vs. ex-situ focus Up to this point, I have only discussed simple focus constructions where the focused object appears ex-situ. However, looking at a larger variety of focus constructions, including in-situ focus and cases of focus projection, lends further support to the analysis developed here. Furthermore, these cases distinguish the focus phrase analysis from the cleft analysis, which cannot account for the facts presented here, as will be discussed in the next section. According to the focus phrase analysis, ne marks focus, and we expect to find complex patterns with respect to what exactly is in focus semantically given a particular formal marking of focus, just as we find such cases of so-called focus projection in pitch accent languages. Let us start with cases where the entire verb phrase is focused. Assuming, as above, that we can force a particular focus structure on a declarative sentence by putting it in the context of a question, the following question answer pair illustrates a case of VP-focus: (12) a. PG³ CVGC CDFWN GM³ KT³ '? FM- what/how A. (SM)-do- ASP- FV ‘What did Abdul do?’ b. PG OCG CDFWN C³ P[W³ KT³ ' SM - drink- ASP- FV FM 6.water A. ‘Abdul [drank WATER]F.’ Formally marking the object for focus is apparently sufficient for focusing the entire verb phrase semantically. This is exactly the type of focus projection we find for pitch-accent languages like English. The same is true for in-situ focus; i.e. the sentence in (2b), where the object is focused in-situ, could also express focus on the verb phrase, e.g. as an answer to an in-situ version of the question in (12a). 10 Next, let us turn to sentence focus, which is what we find in so called out of the blue-contexts or as answers to questions like What happened? (13) [Context: Abdul drank non-purified water and got sick. A just got back and wants to know from B what happened.] A: PG³ MGG MG³ QTW? FM- what CLbad ‘What’s wrong?’ or ‘What happened?’ [literally: ‘What is bad?’]

148

Florian Schwarz

B:#DFWN PG³ C³ TC³ A. FM- SM- T‘Abdul drank water.’

P[W³ KT³ drink- ASP-

' FV

OCG 6.water

In this context, we find ne in the immediately preverbal position marking sentence focus. The fact that the entire IP is in focus here motivated the analysis in (11b) where the IP moves to the focus phrase. Note that neither in-situ nor ex-situ focus marking of the object (as in (1b) and (2b)) can express sentence focus. This is different from English where focus projects from the stressed object all the way up to the sentence level. Apart from marking sentence focus, preverbal ne can also express a particular emphasis on the truth of a statement (i.e. verum focus), as is frequently discussed in traditional grammars of Kikuyu. The following example (adopted with a few changes from Armstrong (1940:297)) illustrates this use of preverbal ne: (14) A: ‘Where did you put it?’ B: ‘I put it in the granary.’ A: ‘I didn’t see it there – are you sure?’ B: PG³ P³ FC³ K¢³ C K³  FM- SM- putASP- FV CL‘I DID put it in the granary.’

MQQODG granary

Sentence focus and verum focus are the only two functions that preverbal ne has, contrary to claims made in the literature that it could express focus on the predicate, i.e. verb or verb phrase focus (Güldemann 1996). 11 In order to express narrow focus on the verb, the verb has to appear in its infinitival form in the ex-situ position with ne as well as in its base position in the inflected form: (15) A: CDFWN PG³ C³ &GM³ KT³ A. FM- SM- laugh- ASP‘Did Abdul laugh?’ B: C5C CDFWN PG³ MQ³ TGT³ no A. FM- SM(INF)- cry‘No. Abdul CRIED.’

'? FV C C³ TGT³ KT³ ' FV SM- cry- ASP- FV

Although a full analysis of this has to await another occasion, one might be able to account for the two occurrences of the verb within the copy-theory

Ex-situ focus in Kikuyu 149

of movement. 12 The sentence in (15) would then be an example of a situation where both copies are pronounced, presumably because after morphological reanalysis the higher copy of the verb becomes invisible to Kayne’s LCA and, therefore, to deletion (Nunes 2004). 13 In summary, focus marking on the object (either in sentences without ne and with the object in-situ or in sentences with ex-situ focus and ne) can express focus on the object or on the verb phrase, and ne in preverbal position can express sentence focus or verum focus. While the technical details of focus projection have to be worked out in future work, the fact that there is focus projection in constructions with ne fits in very naturally with the focus phrase analysis. If we make just one additional assumption, namely that every sentence has to have a focus, several further facts that would otherwise be surprising (in particular from the viewpoint of the cleft analysis, as will be discussed in the next section) fall out of the theory. First, a seemingly odd fact about Kikuyu is that “when a sentence consists of an affirmative finite verb only (e.g. nƭokire, he came), nƭ [= ne; FS] is indispensable” (Barlow (1951: 34)). 14 Under the current analysis, this follows because there is no object that could introduce any in-situ focus marking, and hence, the only possibility for introducing a focus into the sentence (apart from having ex-situ subject focus with ne) is to have preverbal ne. Second, there is an interaction between the availability of in-situ focus and ne. The in-situ focus on the object (or the verb phrase) in (2b) is no longer available if preverbal ne is introduced. This is, of course, exactly what we expect if we analyze ne as a focus marker which triggers movement of the focused constituent (either the object or the entire IP) to the focus phrase. Finally, we are able to explain the obligatory presence of ne in third person present tense copula constructions if we make the additional assumption that the focus feature on the focused XP in in-situ focus constructions is in some way licensed by the lexical verbal head. It is commonly assumed that phonologically null heads have limited licensing capacities. Since the third person present tense form of the copula is phonologically null, it cannot license in-situ focus on the object and hence, the only way to introduce a focus in such copula sentences is to let ne do the job.

150

Florian Schwarz

3.4. Remaining problems Although the focus phrase analysis makes promising predictions with respect to both of the two different environments in which ne appears, its distributional restrictions and the facts connected to focus projection discussed in the preceding subsection, there are two remaining problems. The first concerns the details of the technical implementation of ne mentioned at the beginning of this section. I have nothing more to say about this here. The second problem concerns a number of morphological changes that appear on the verb phrase when it is preceded by an ex-situ focus or whconstruction: negation changes from ti to ta, the third person subject marker changes from a to o, and the post-verbal downstep is deleted (for details, see Clements 1984). These changes appear to be identical to the ones we observe in relative clauses. This has been taken as support for the cleft analysis since on that analysis we are in fact dealing with relative clauses (see Bergvall 1987, and the discussion in the next section). However, this argument is not as straightforward as it may seem at first sight. As Bergvall herself points out (Bergvall 1987: 114), once we consider more complex constructions involving multiple clauses, with the focused element originating in the lowest clause, these changes affect different domains: the subject marker only changes in the lowest clause, negation only changes in the highest embedded clause, and the tonal changes affect all embedded clauses. These phenomena presumably are general effects of A'-movement, which need an independent account. Therefore, they do not pose a problem that is particular to the focus phrase analysis, and the competing cleft analysis has to account for them independently as well.

4.

The cleft analysis

I now sketch out the cleft analysis (Bergvall 1987) and discuss some of the problems it faces. Its starting point is the occurrence of ne in copula constructions although it does not assume that ne is the copula. As above, copula constructions are taken to be a special case of the preverbal occurrence of ne. The cleft analysis differs from the focus phrase analysis in the case of ex-situ focus and wh-constructions. The latter are taken to be yet another variant of the preverbal occurrence, where the verb is the phonologically null form of the copula which is part of a cleft.

Ex-situ focus in Kikuyu 151

Another difference between the two analyses lies in the role that is assigned to ne. Since cleft constructions have a well known impact on focus structure, it is unnecessary to assign ne the role of a focus marker. What role does ne play then? According to Bergvall’s cleft analysis, it is an assertion marker that appears in the head of the IP. I will come back to this point below after introducing the analysis in some more detail. 4.1. Accounting for the different occurrences of ne With ne generated in the head of the IP (see (16b)), the immediately preverbal cases can be straightforwardly accounted for. Assuming that the subject appears in the specifier of the IP (or, alternatively, in the specifier of CP), the word order of the preverbal case can be derived without any difficulties: (16) a. #DFWN PG³ C³ TC³ A. FM- SM- T‘Abdul drank water.’ b. 

P[W³ KT³ drink- ASP-

' FV

OCG 6.water

IP ei SpecIP I' #DFWN ei I VP PG ei V DP CTCP[WKT' OCG

As in the focus phrase analysis, copula constructions are accounted for in exactly the same way as the preverbal case by assuming that a phonologically null form of the copula appears as the head of the verb phrase. Finally, ex-situ focus and wh-constructions are analyzed as bi-clausal, with a phonologically null expletive subject and a phonologically null copula in the first clause. In a slightly simplified form, the analysis as proposed by Bergvall then is as in (17b) (Bergvall 1987: 123): (17) a. PG OCG #DFWN C³ TC³ P[W³ KT³ ' FM 6.water A. SM- T- drink- ASP- FV ‘Abdul drank WATER.’

152

Florian Schwarz

b. 

IP ei SpecIP I' G ei I VP PG tgp V DP CP 6 Ø OCGK Opi CDFWNCTCP[WKT' ti

The index and the operator in the lower CP link mae to its base position. According to this analysis, all occurrences of ne are reduced to the preverbal case, and the focusing effect of the ex-situ construction is attributed to the cleft construction. Crucially, this dissociates ne per se from focus. 4.2. Accounting for the distributional restrictions Given the analysis sketched above, how does the cleft approach account for the distributional restrictions? With respect to the limitation to one ne per clause, it says that each independent clause can only make one assertion (and hence can only contain one ne, which is taken to be a marker of assertion). The absence of ne in embedded clauses, in particular in relative clauses, is explained by the fact that relative clauses (at least restrictive ones) are presupposed, and their content is therefore not part of what is asserted. Finally, the complementary distribution of ne and the negation marker ti is explained by saying that ti is a marker of assertion as well (with opposite polarity) so that it would not make sense to have both, a positive and a negative assertion marker, in one clause. 4.3. Problems Note that all of the points concerning distributional restrictions rest on the assumption that ne is a marker of assertion. This characterization of ne is problematic, given that ne routinely occurs in questions and other types of speech acts that are not assertions. In addition to this issue concerning the role of ne, the cleft analysis faces a number of further problems. First, it cannot account for topics preceding ex-situ focus constructions. Second, it is hard to reconcile with the fact that

Ex-situ focus in Kikuyu 153

ex-situ wh-constructions can be part of multiple wh-questions. Third, it cannot account for certain cases of multiple ne’s in complex clauses. Let us turn to the first point in more detail. Topicalized elements can precede the ex-situ focus constructions with ne, as shown in (3) above. The cleft analysis, however, which assumes that ex-situ constructions do involve a relative clause, falsely predicts this to be impossible, given that topicalization out of relative clauses is impossible, as the data in (18) illustrate for the PP topic in front of the house: (18) a. P[KPC QQTGC Y³ nP³ KT³' K³$WMWOD'T' [C P[QODC mother DEM

SM- see- ASP-FV 5-book infront 9.A 9.house

‘the mother who saw the book in front of the house’ b. * Q\LQD RRUHD OD'T' [CP[QODCY³nP³KT³' mother DEM

K³$WMW

infront 9.A 9.house SM-see-ASP-FV 5-book

c. * OD'T' [CP[QODCQ\LQD RRUHD Y³ nP³ KT³ ' K³$WMW infront 9.A 9.house mother DEM

SM- see- ASP-FV 5-book

Furthermore, topicalization beyond clause boundaries appears to be impossible, as shown in (19), which speaks against any bi-clausal treatment of exsitu focus constructions. (19) a. CDFWNPG- W¢- KT- ' CVG P[KPC A. FM- sayASP- FV that mother PG- nnn- KT³ ' K³$WMWOD'T' [C P[QODC FM- seeASP- FV 5-book in-front 9.A 9.house` ‘Abdul said his mother saw the book in front of the house.’ KT- ' DWH OD'T' [C b. DEGXO PG- W¢- A. FM- sayASP- FV that infront 9.A  P[QODC Q\LQD PG³ nnP³ KT³ ' K³$WMW  9.house motherFM- seeASP- FV 5-book



c.# OD'T' [C P[QODC CDFWN PG³ W¢³ KT³ infront 9.A 9.house A. FM- say- ASPCVG P[KPC PG³ nnP³ KT³ ' K³$WMW that mother FM- seeASP- FV 5-book

' FV

 In (19b), the prepositional phrase is fronted within the embedded clause, and the meaning is the same as in (19a). The sentence in (19c), on the other hand, can only be understood in such a way that ‘in front of the house’ is

154

Florian Schwarz

the location where the saying took place and not where the mother saw the book. Given this restriction on topicalization, the cleft analysis makes false predictions about topicalization in ex-situ focus constructions (cf. (3)). The second problem concerns the fact that ex-situ wh-constructions with ne can form part of a multiple wh-question, as illustrated in (20): (20)

PQQ Y³ 'PF³ KTK³ FM.who SM- sellASP‘Who sold what?’

' FV

MGG? what

This is problematic because, cross-linguistically, cleft based questions can normally not form part of multiple wh-questions. 15 Unless we can find good reason to believe that we are dealing with an exception here, this point speaks against the cleft analysis. Finally, there are constructions involving multiple clauses which can contain multiple ne’s. In particular, Kikuyu allows for long-distance whextraction out of certain clauses. When the wh-word is extracted from the lowest clause, ne can appear both with the fronted wh-word and preverbally in the lowest clause: (21) a. PG³ MQ PIQ¢G C³ W¢³ KT' CVG MCOCW FM- where N. SM- sayASP that K.  PG V n³ nP³ KT' MCPCM' FMSM- seeASP kanake ‘Where did Ngoge say that Kamau saw Kanake ?’ [‘Where, according to Ngoge, did Kamau see Kanake?’] b.* PIQ¢G C³ W¢³ KT' CVG PG³ MQ MCOCW N. SM- say- ASP that FM- where K.  PG V n³ nP³ KT' MCPCM' FMSM- seeASP kanake However, this is only possible if the wh-word moves all the way to the highest clause, and not if it remains in the lower clause, as indicated in (21b). This is problematic for the cleft analysis because it explains the restriction that ne can only occur once per clause in semantic terms by saying that each clause can be marked for assertion only once (however this is to be understood for the question cases; the same point would apply to the corresponding case of focus fronting). But semantically, the wh-word belongs to the lower clause so that in this respect both ne’s should be counted

Ex-situ focus in Kikuyu 155

as belonging to the same clause. Therefore, the cleft analysis falsely predicts (21a) to be ungrammatical. Note that this construction is not at all problematic for the focus phrase analysis. Assuming successive cyclic movement, we expect the wh-word to move through the specifier of the focus phrase of the lower clause. Apparently, moving the wh-word through this position is compatible with having an overt ne in the lower focus phrase. 16 This analysis gains further support from the fact that in the closely related language Kitharaka the morpheme equivalent to ne is obligatorily present in the lower phrase, a fact which has been taken as an argument for an analysis in terms of successive cyclic movement by Muriungi (Muriungi 2004). In addition to these major problems, the cleft analysis also cannot account for the facts about focus projection and the related issues in section 3.3. According to the cleft analysis, the focusing effect of ex-situ focus constructions is due to the syntactic configuration of the cleft, and ne only plays its general role as an assertion marker in these cases. Beyond such special constructions that directly affect focus structure, we thus have no reason to expect interactions between ne and focus on this account. But as we saw above, the possibility of in-situ focus depends on the absence of ne. I do not see how this can be accounted for if we assume that ne is an assertion marker. Concerning the fact that we find focus projection with the ex-situ focus construction (namely focus on the verb phrase when the object is fronted with ne), there also is a problem for the cleft analysis, as clefts typically do not allow focus projection. Finally, there does not seem to be a way to account for the seemingly odd facts discussed at the end of section 3.3. Why is ne obligatory in sentences that only consist of an intransitive verb? Surely not because these always have to be emphatically marked for assertion, but that is all that the cleft analysis could say about this. And why is ne obligatory with third person present tense copula constructions? Again, the role that the cleft analysis assigns to ne, namely that of a marker of assertion, does not provide any help in explaining this. Taken together, these problems seem to provide a good case against the cleft analysis. 17 Furthermore, as already mentioned above, it is unclear what the status of ne on this analysis could reasonably be, given that the assertion marker analysis is incompatible with its presence in questions and other types of speech acts.

156 5.

Florian Schwarz

Conclusion

I have discussed Kikuyu focus constructions involving ne and compared two analyses, the focus phrase analysis and the cleft analysis. I have argued that the former makes out better, both with respect to the syntax and the semantics of the construction. Although there are plenty of questions for further research, I hope to have convinced the reader that it is most promising to approach these from the viewpoint of the focus phrase analysis.

Notes 1. Parts of this paper have been presented at SOAS (London), ZAS (Berlin), and at the workshop ‘Topic and Focus: Information Structure and Grammar in African Languages’ (Amsterdam). I would like to thank all the participants for helpful comments and discussion, and in particular Enoch Aboh, Rajesh Bhatt, Lisa Cheng, Laura Downing, Katharina Hartmann, Angelika Kratzer, Victor Manfredi, Yukiko Morimoto, Brigitte Reineke, Anna Szabolcsi, Sabine Zerbian, and Malte Zimmermann. Special thanks are due to Manfred Krifka for crucial guidance while I was working on my Master’s thesis on which this paper is based. Special thanks are also due to my Kikuyu consultant, Sam Kinuthia. I gratefully acknowledge support for this research from the ZAS. 2. Kikuyu is an SVO Bantu language spoken in Kenya. Its label in Guthrie’s (1967) classification system is E50. 3. All data has been elicited from my consultant, unless otherwise indicated. The following abbreviations are used in the glosses: FM: focus marker, SM: subject marker, T: tense, ASP: aspect, FV: final vowel, A: associative, COP: copula, NEG: negation, DEM: demonstrative. Numerals preceding nouns indicate the noun-class. Tones are not marked. Although the more detailed study of tonal effects in relation to focus is an important topic for further research, I believe that the syntactic points made in this chapter hold independently of such possible effects. Details concerning tense and aspect are omitted. See Johnson (1980) for a comprehensive discussion of Kikuyu tense and aspect. 4. I assume throughout that focus can be reliably manipulated by different question contexts. This is independent of the issue of whether a theory of focus ultimately needs to make reference to question-answer correspondence. 5. I use topicalization in a syntactic sense here, without making any direct claims about its discourse properties. In syntactic terms, there is evidence for an additional Topic Phrase between the CP and the FP, since the sen-

Ex-situ focus in Kikuyu 157 tence in (11) could be embedded by a bridge verb and would then be preceded by the complementizer ate which presumably occurs in C0. Therefore, the topicalized element cannot appear in the specifier of CP. This is, of course, perfectly consistent with the standard analysis of the extended left periphery (Rizzi 1997). 6. The stem re is actually ambiguous: Apart from the copula meaning, it also has a possessive meaning (i.e. (4b) can also mean ‘Abdul has a teacher’). The one place where the possessive and the copula paradigms diverge is in the third person present tense form, where the null form unambiguously has the copula meaning, whereas re only has the possessive meaning. 7. Movement is indicated by crossing out elements of a syntactic chain that aren’t pronounced. The IP-level is ignored to keep things simple. The topicalized subject is represented in the specifier of CP to keep things simple. See footnote 5 on the need for a distinct topic phrase inside of the CP. 8. One potentially problematic aspect of this analysis, pointed out to me by Rajesh Bhatt, is that the apparent possibility of movement out of the moved IP is somewhat unexpected. A possible alternative analysis would leave the IP in its base position and have ne assign focus to it from the head of FP. 9. Interestingly, however, the order FM-S-V-O can express sentence focus in the closely related language Kitharaka (Muriungi 2004). 10. An interesting question that was pointed out to me by Katharina Hartmann is why focus on the verb phrase cannot be expressed by moving the VP to the focus phrase. Perhaps this is blocked by the alternative option of just moving the object, which is more economical. 11. Again, there is an interesting contrast with Kitharaka, where the order SFM-V-O can express focus on the verb as well as sentence focus (Muriungi 2004). 12. Thanks to Lisa Cheng for bringing this to my attention. 13. As one of the reviewers points out, more needs to be said about this. For example, it is unclear, given this brief description, why the object in exsitu focus constructions is not pronounced in both positions. 14. I assume that this holds both for intransitive verbs and transitive verbs that only have an object marker and no overt object noun phrase, but my data on this are incomplete. 15. Thanks to Anna Szabolcsi for pointing this out to me. 16. This fact might speak in favor of the second analysis of ne above, which assumes it to be generated in the head of the focus phrase and then cliticizes it onto the XP in its specifier, since otherwise we would have the feature on the focused XP spelled out twice. 17. Yet another problem that the cleft analysis probably has to face is the absence of tense in the cleft copula. I do not have the relevant data to make

158

Florian Schwarz this point, but Bergvall’s discussion of the interpretation of tense in clefts (Bergvall 1987: 130–132) suggests that there are no clefts with past tense copulas, which is unexpected on her account (thanks to Rajesh Bhatt for pointing out this issue to me).

References Armstrong, Lilias E. 1940 The Phonetic and Tonal Structure of Kikuyu. London: Oxford University Press. Barlow, Arthur Ruffell 1951 Studies in Kikuyu Grammar and Idiom. Edinburgh: Blackwood & Sons. Brody, Michael 1990 Some Remarks on the Focus Field in Hungarian. UCL Working Papers 2: 201–225. Bergvall, Victoria Lee 1987 Focus in Kikuyu and universal grammar. Ph.D. diss., Harvard University. Clements, George N. 1984 Binding Domains in Kikuyu. Studies in the Linguistic Sciences 14: 37–56. É. Kiss, Katalin 1998 Identificational Focus versus Information Focus. Language 74: 245– 273. Güldemann, Tom 1996 Verbalmorphologie und Nebenprädikation im Bantu. Eine Studie zur funktional motivierten Genese eines konjungationalen Subsystems. Bochum: Brockmeyer. Guthrie, Malcolm 1967 The Classification of Bantu Languages. London: Dawsons. Hyman, Larry M. 1999 The Interaction between Focus and Tone in Bantu. In The Grammar of Focus, Georges Rebuschi and Laurice Tuller (eds.), 151–177. Amsterdam, Philadelphia: John Benjamins. Johnson, Marion R. 1980 A semantic description of temporal reference in the Kikuyu verb. Studies in African Linguistics 11: 269–320.

Ex-situ focus in Kikuyu 159 Muriungi, Peter 2004 Wh-movement in Kitharaka as focus movement. Paper presented at the Workshop: Syntax, Semantics and Pragmatics of Questions at ESSLLI 11. Nunes, Jairo 2004 Linearization of Chains and Sideward Movement. Cambridge: MIT Press. Rizzi, Luigi 1997 The Fine Structure of the Left Periphery. In Elements of Grammar, Liliane M. Haegeman (ed.), 281–337. Dordrecht: Kluwer Publications. Schwarz, Florian 2003 Focus Marking in Kikuyu. In Questions and Focus, Regine Eckardt (ed.), 41–118. ZASPIL 30. Berlin: ZAS.

Focus in the Force-Fin system: Information structure in Cushitic languages Mara Frascarelli and Annarita Puglielli

Abstract This paper investigates the Force-Fin system in two Cushitic languages – Somali and Afar – in which the interpretation of discourse grammar categories depends on specific morphosyntactic conditions. In particular, the analysis deals with the activation of the “Focus field” and constitutes an argument for the assumption of a [+foc] feature in the C-domain, playing a crucial role in the interpretation of different focus-related categories. The relevant discussion also provides substantial support for a cartographic approach to Information Structure and shows the existence of AGREE relations between (some) functional features in the left periphery of the sentence. Finally, a crucial connection is shown between Focus and the illocutionary Force of the sentence and a structural distinction between matrix and embedded C-domains is therefore proposed.

1.

Introduction

This paper investigates information structure and the Force-Fin system (cf. Rizzi 1997) in two Cushitic languages, Somali and Afar. In particular, the activation of the focus field will be analyzed in different clausal types, showing the existence of a crucial connection between focus-related categories and the illocutionary force of a sentence. 1.1. The fine architecture of the C-domain In a cartographic approach (cf. Rizzi ed. 2004), the original CP-node (a recursive phrase, targeted by different functional categories) has been reanalyzed as an array of functional projections each dedicated to a specific function related to information structure. The C-domain thus provides an interface between the propositional content (the IP-node) and specific dis-

162

Mara Frascarelli and Annarita Puglielli

course roles. 1 In particular, Rizzi (1997) proposes that the left periphery of the sentence is included between force and finiteness: “Complementizers express the fact that a sentence is a question, a declarative, an exclamative, a relative, a comparative, an adverbial of a certain kind, etc., and can be selected as such by a higher selector. This information is called the specification of Force […] the C-system expresses a specification of finiteness, which in turn selects an IP system with the characteristics of finiteness: mood distinctions, subject agreement licensing nominative case, overt tense distinctions.” (Rizzi 1997: 283, 284)

Force and finiteness can be expressed on a single head “in simple cases” but are forced to split in marked constructions, namely with “the activation of the Topic-Focus field” (Rizzi 1997: 314). The Force-Fin system thus includes (at least) the FocP projection, that is the syntactic locus for [+foc] interpretation (cf., among others, É. Kiss ed. 1995, Rizzi 1997, Frascarelli 2000), and different TopPs in which topic constituents are located according to their specific discourse properties 2 (Frascarelli and Hinterhölzl 2007). The C-domain also includes a functional projection connected with the interrogative force of the selected clause, called IntP, located in a position that is higher than FocP (cf. Rizzi 2001). The Force-Fin system can thus be represented as follows (the asterisk indicates recursion): (1)

[ForceP

[AboutTopP

[IntP

[FocP

[FamTopP*

[FinP

[IP

Languages like Somali and Afar, in which information structure is morphosyntactically marked, support a cartographic approach and provide evidence for some refinement. In particular, data show that the activation of [+foc] plays a crucial role in the interpretation of other discourse categories, so that an interaction must be posited between functional heads in the C-domain. 1.2. Somali and Afar: Basic properties 3 Somali is a polysynthetic language (in the sense of Baker 1996). This condition entails that T-roles are only assigned through incorporation into the verbal head (the so-called “Morphological Visibility Condition”, MVC). Hence, argument structure is realized by means of clitic pronouns that are disposed in the Verbal Complex (VC) in a rigid SOV order (Puglielli 1981, Svolacchia and Puglielli 1999) while full DPs are merged in non-argument position and connected to the sentence by means of resumptive pronouns,

Focus in the Force-Fin system: Information structure in Cushitic languages 163

which bind constituents to their thematic roles. Somali is also a focusprominent language, so that one constituent must be overtly identified as the focus of the sentence. In particular, focused DPs occur in preverbal position followed by the focus marker (FM) baa: 4 (2)

Axmed gurigii SHALAY buu (baa+uu) pro 5 Axmed house.ANAPH yesterday FM.SCL3SGM OCL3 nooga (= na+u+ka) qaaday. OCL1PL-for-from take.PAST.3SGM ‘YESTERDAY Axmed took it from the house for us.’ [lit.: ‘Axmed, the house, YESTERDAY he took it from it for us.’]

The pronominal argument analysis in Somali is supported by a number of standard diagnostics, e.g. obligatoriness of clitic pronouns in the VC ((2) is ungrammatical if the SCL uu or the OCL na is omitted), absence of nonfinite clauses and of DP anaphors (cf. Svolacchia and Puglielli 1999). In particular, the non-argument position of full DPs (independent of focus) is shown below by means of ‘disjoint reference effects’: (3)

SHALAY bayk jabisay Maryank yesterday FM.SCL3SGF break.PAST.3SGF Maryan ‘*Shek broke Maryank’s knife YESTERDAY.’

mindideedk knife-her

As we can see, a coreferential reading is possible in (3) between the subject clitic ay and the R-expression Maryan (while this interpretation is excluded in English). This shows that full DPs in Somali do not sit in A-position and must therefore be considered as adjuncts. Afar is an SOV language as well. However, its morphosyntactic properties are quite different from Somali, so that a comparison of the two languages is extremely effective for the issues of the present study. Afar is, in fact, an inflectional pro-drop language in which DPs carry argument role and pronouns are “strong” elements realized as objects of verbs and postpositions (cf. (4b)) 6 : (4)

a. Ali mootar-at bilu-k yemeete Bilu.ABS-from come.PAST.3SGM Ali car.ABS-by ‘Ali came by car from Bilu.’

164

Mara Frascarelli and Annarita Puglielli

b. anu kaa ko-h ruube I him you-to send.PAST.1SG ‘I sent him to you.’ c. xaa’b-e ‘koo ka’l-e stop.PAST.1SG leave.COMP you ‘I prevented you from leaving.’ As is shown, the grammar of Afar allows non-finite subordinate clauses (4c), and pronominal DPs are not obligatory for morphological visibility (4a). In section 6, we will see that clitic resumption is not present in topic constructions as well – an additional, major difference with respect to Somali.

2.

Focus as predicate

In recent works, Frascarelli and Puglielli (henceforth, F&P 2005a, 2005b, to appear) have argued for a Small Clause (SC) analysis of focus constructions in focus marking languages. Drawing from diachronic and comparative studies (cf. Hetzron 1980, Lamberti 1983) and on the basis of synchronic evidence, F&P show that focus markers (FM) were originally copular forms, which may have included a 3rd person (clitic) pronoun. Focus constructions therefore imply the presence of a SC in which the focused DP is the (specificational) predicate while presupposed information is carried by a relative clause (sitting in subject position). To provide an illustration, consider the following focus construction in Somali: 7 (5)

CALI baa [Op Cali FM (=COP) ‘CALI is (that is) Somali.’

soomaali ah ]REL. Somali be.RED

A sentence like (5) takes as a presupposition the existence of “(some) person who is Somali”, which is expressed through a free relative 8 (defining a set). Since it is a headless relative, an “external” specification is needed for the relative operator: this is provided by the DP that moves from predicate position (in the SC) to Spec,FocP (to check the [+foc] feature). The focus thus gains scope over the SC and specifies a value for the operator-variable chain in the relative clause:

Focus in the Force-Fin system: Information structure in Cushitic languages 165

(6)

a. [FocP [Foc’ baa [SC [DP [CP OP [IP tOP soomaali ah ]]] [DP Cali ] ]]]Æ b. [FocP CALI [Foc’ baa [SC [DP [CP OP [IP tOP soomaali ah ]]] tCali ]]]

Hence, focus is not merged as a part of the vP-phase containing the verb in the relative clause. It is simply “reinterpreted” as a part of it after identification of the operator contained therein. In other words, focus provides a value for a variable that is not its own copy. 9 This analysis accounts for the connection between focus and relativization that was pointed out by many scholars (since Schachter 1973), derives focus strategies and discourse-semantic properties (Frascarelli 2005), and explains a number of focus-related phenomena such as the so-called Antiagreement Effect (cf. Ouhalla 1993, Frascarelli 1999, F&P 2005a, 2005b). This is illustrated below (from Somali): (7)

(8)

NIMANKÁAS baa Hilibka meat.ART men-those.ABS FM ‘THOSE MEN are eating the meat.’

a. * Hilibka meat.ART b. * Hilibka meat.ART c. * Hilibka meat.ART

NIMANKÀASU men-those.NOM NIMANKÁAS men-those.ABS NIMANKÁAS men-those.ABS

cunayá. eat.PROG.RED

baa FM

baa FM

bay FM.3PL

cunayá. eat.PROG.RED cunayaan. eat.PROG.3PL cunayá. eat.PROG.RED

When the subject is focused, it cannot show NOM case but the “default” ABS case (as used for citation and predicative DPs). Furthermore, it triggers the presence of the so-called “reduced paradigm” (RED) 10 and, finally, it makes it impossible for the (focused) subject to be resumed by a pronoun. In a theory in which the Focus DP is not an argument, but a predicate, antiagreement effects can be accounted for without stipulations: what we understand as the “subject” is only reinterpreted as such after the identification of the relative operator. Antiagreement can thus be attributed to the presence of an empty subject (i.e. the variable) in the relative clause. Since Somali is not a pro-drop language, this obtains a reduced (i.e. kind of participial) form of agreement: 11 (7’)

[TopP hilibka [FocP NIMANKÁAS [Foc’ baa [SC [DP [CP OP [IP tOP cunayá ] tnimankáas ]]]]]]

166

Mara Frascarelli and Annarita Puglielli

The morphosyntactic properties of focus constructions in Afar (a non focus-prominent language) provide important support for the present analysis. Indeed, the focus shows ABS case independent of its role, and the verb shows a “nominalizer suffix” whose nature is not clear in the literature (cf. Bliese 1981). According to the present analysis, these properties obtain a comprehensive explanation: the verb is embedded in a relative clause, and the “verb suffix” –m is therefore a COMP (in final position, consistent with the head-final quality of the language). 12 Consider the following: (9)

[FocP ÀWKA [Foc’ (y) [SC [DP [CP OP [IP tOP huurí-h adda-l boat-of inside-in boy.ABS 3SG kullumta habte-m]]] tàwka ]]] fish leave.PAST.3SGF-COMP.NOM ‘THE BOY left the fish inside the boat.’ [lit.: ‘THE BOY (he is) that left the fish in the boat.’]

The relative DP is the subject of the SC that we posit (hence, marked for case) while focus is the predicative DP (showing ABS case) that moves to FocP. The COMP head originates from the feminine noun im (‘part/thing’) and is connected with a null operator that triggers a default 3SGF agreement. Finally, the optional y (following the focus) is the remnant of an old copula (cf. Parker and Hayward 1985). Hence, like a FM, it is a lexicalization of Foc°, attracting the predicative DP in its Spec (for checking requirements). Let us now investigate the interaction of focus with other discourse categories in the Force-Fin system, starting with the functional projection that “looks downstairs” into the propositional content, namely FinP. NOM

3.

Focus and the specification of finiteness

3.1. Force in “verb focus” constructions When the focus field is not activated in Somali, the VC is interpreted as new information and sentences are characterized by the presence of a specific marker, waa, traditionally considered as a “verb FM” (Puglielli 1981). The syntactic properties of waa sentences, however, show that waa cannot be considered as a Foc head. 13 Consider the subject resumption in the following sentences:

Focus in the Force-Fin system: Information structure in Cushitic languages 167

(10)

buuk / *baa Calik MARYAM FM.3SGM FM Cali Maryam ‘(As for Cali), he saw MARYAM.’

arkay. see.PAST.3SGM

(11)

Calik Maryam wuuk / waa arkay. WAA.3SGM WAA see.PAST.3SGM Cali Maryam ‘(As for Cali), he saw her, Maryam.’

In both (10) and (11), Cali – that is understood as the subject of the following sentence – is realized as a left-hand topic. Consistent with the MVC (cf. Section 1.2), clitic resumption (uu) is obligatory in the focus construction (10) while it is only optional in the presence of waa (cf. (11)). This asymmetry is explained if we assume that waa is connected with subject agreement features which are not available to a FM like baa. In other words, baa and waa do not sit in the same functional projection. Indeed, waa is strictly connected with overt tense distinctions and NOM case marking (cf. (12)(13)) while baa excludes the latter and requires a reduced form of verb inflection (cf. (8a-b) above). Moreover, no TopP projection is available between waa and the VC (cf. (14)) while topics are allowed between baa and the rest of the sentence (cf. (17) below): (12)

Waa sheegaa/

sheegay/

sheegayaa/

sheegayay

say.PRES.3SGM/say.PAST.3SG/say.PRES.PROG.3SGM/say.PAST.PROG.3SGM

‘He says /

said /.

is saying /

was saying.’

(13)

Wiilku / *wiilka waa boy.ART.ABS boy.ART.NOM ‘The boy is Somali.’

soomaali. Somali

(14)

( soomaali) Cali (soomaali) waa (*soomaali) ahaa (soomaali). Somali Cali be.PAST.3SGM ‘Cali was Somali.’

Finally, waa is in complementary distribution with other sentential markers such as ma in questions (cf. Section 4.1) and ha in imperative clauses (e.g. ha i labin, ‘don’t’ bother me’). We thus conclude that waa is not a FM, but a modality marker (cf. also Saeed 1984). It is therefore a lexicalization of Fin° that activates [+decl] sentential force 14 in the selected IP. As a Fin head, it is connected with subject agreement and tense specifications

168

Mara Frascarelli and Annarita Puglielli

through AGREE (Chomsky 2001, 2002) with AgrS°. Subject omission is thus permitted (cf. (11), (12)). The “focus interpretation” on the VC must therefore be considered as a “by-product” of [+decl] force in the absence of [+foc] constituents. In other words, when the focus field is not activated, sentential and illocutionary force coincide: the relevant declarative sentence has no other “intention” than expressing a statement. 3.2. Force in narrow focus constructions When the focus field is activated, the illocutionary force of the sentence is crucially modified. The relevant sentence is merged as a SC construction, in which new information is only carried by the predicative DP and the verb is embedded in a relative clause (as presupposed information). This means that in focus constructions Fin does not select an IP-node, and a DECL marker like waa is therefore excluded. Since AGREE is a “local relation between two adjacent heads” (cf. Chomsky 2001), Fin° cannot be related with AgrS° and subject omission is therefore not permitted. 15 This explains the asymmetry shown in (10)–(11), the structures of which are given, respectively, in (15) and (16) below: (15)

[TopP Calik [FocP MARYAM [Foc’ baa+uuk [FinP [Fin’ ‡ [SC [DP[CP OP [AgrSP tuu tOP arkay] tMaryam ]]]]]]

(16)

[TopP Calik [TopP M.z [FocP[FinP [Fin’waa[+decl] [AgrSP (uu)k[AgrS’ ‡z arkay ]]]]] [+FIN, +AGR]

Sentence like (15), however, show that Fin° (though empty) is not completely “inert” in narrow focus constructions because it serves as a licensing head for the subject trace in Spec,AgrSP. This is in line with Rizzi’s (1997) explanation of “anti-adjacency effects”: in Somali (like in French), Fin° can be endowed with AgrS features in virtue of its selecting properties. This allows subject extraction. In this respect, also consider the following sentence: (17)

Shalay CALI baan

anigu ku arkay,

yesterday Cali FM-SCL1SG I.NOM 16 at

jaamacadda.

see.PAST.1SG university.ART

‘Yesterday, I saw CALI, at the university.’

Focus in the Force-Fin system: Information structure in Cushitic languages 169

In (17), subject extraction occurs across a topic (anigu), which evidently does not create minimality effects. 17 In this respect, the Somali pattern is the same as in French, but different from English (examples from Rizzi 1997): (18) a. Je ne sais pas qui [ ton livre [ t pourrait l’acheter ]] b.* A man who [ liberty [ t should never grant t to us]] We suggest that the relevant asymmetry is connected with clitic resumption and argue for a structural distinction between topic constructions, which depends on the availability of clitic pronouns in a given language. In Frascarelli (2004), evidence is provided that clitic-resumed topics are merged in the C-domain (where they form a binding chain with the clitic) while non clitic-resumed topics are moved from an IP-internal position. This distinction accounts for topic properties from a cross-linguistic perspective: in languages like French and Somali, topics are inserted in TopP and, as such, they do not interfere in movement chains: (16’)

[FocP CALIk [Foc’ baa+aanj [TopP anigu [FinP t’j FIN° [SC [DP[CP OP [AgrSP tj tOP ku arkay] tCali ]]]]]]

In a language like English (in which clitics are not available), on the other hand, topics are moved and induce minimality effects on subject extraction. Functional projections in the left periphery thus show different properties. Some of them are connected with operator movement while others contain non-quantificational constituents that are merged in the C-domain. This is a crucial distinction for the internal composition of the Force-Fin system.

4.

Focus and questions

The realization of interrogative force in Cushitic languages shows the existence of a connection between focus, finiteness, and a specific position in which the [+int] feature is encoded for interface interpretation.

170

Mara Frascarelli and Annarita Puglielli

4.1. Yes-no questions A yes-no question in Somali has different realizations according to what is the “centre” of the interrogative force. Specifically, a question must be either centered on the VC or on the focused DP, yielding different orders and marking combinations. Consider the following sentences: (19) a. Cali hadiyad ma / muu Cali present QM QM.3SGM ‘Did Cali bring a present?’ b.* Cali ma / muu hadiyad keenáy?

keenáy ? bring.PAST.3SGM

keenáy ? (20) a. Cali ma HADIYAD *baa / buu Cali QM present FM FM.3SGM bring.PAST.3SGM ‘Is it a PRESENT that Cali brought?’ b. ma HADIYAD *baa / buu Cali keenáy? c.* Cali HADIYAD buu ma keenáy? As we can see, yes-no questions require the presence of a question marker (QM), namely ma, which must gain scope over the constituent that is questioned. Hence, it precedes the VC in (19) and is located in a position higher than the FocP in (20). Note in particular that ma in (19) shows the same syntactic properties discussed for waa in [+decl] sentences (cf. Section 3.1): the subject pronoun can be dropped (19a), and no topic is allowed between ma and the VC (19b). When ma is used in combination with baa, on the other hand, the properties of a narrow focus construction are found (compare (20) with (10)). It is thus possible to conclude that ma is a [+int] marker that is merged in different functional projections according to the specific illocutionary force of the sentence. In particular, when the focus field is activated, the interrogative force must gain scope over it. This forces the activation of a higher projection in the C-domain that is connected with questioning, namely IntP (Rizzi 2001). After operator movement of the predicate-focus, ma is merged in Int°: (21)

[TopP Calik [IntP [Int’ ma [FocP HADIYAD [Foc’ baa+uu [FinP ‡ [SC [DP[CP OP [AgrSP tuu tOP keenáy] thadiyad ]]]]]]]]]

Focus in the Force-Fin system: Information structure in Cushitic languages 171

When the focus field is not activated, on the other hand, ma is merged in Fin° and activates the [+int] feature over the embedded IP, through the (more economical) relation AGREE with the Int° head: (22)

[TopP Calik[TopP hadiyadz [IntP [Int’ [FocP [FinP [Fin’ ma [AgrSP (uu)k [Agr’ ‡z keenay ]]]]]]]? [+INT]

This analysis leads to the important generalization that [+int] is activated on the functional head that takes scope over the informative part of the sentence. 18 Yes-no questions in Afar present similar properties, thus supporting the present analysis. Since Afar is not a focus prominent language, questions can be realized without specific markers through prosodic means (a falling intonation, cf. (23a)). However, an alternative option is available in which the QM ma is used (23b): (23) a. selteè ? finish.PAST.2SG b. ma selte ? QM finish.PAST.2SG ‘Did you finish?’ Like in Somali, Afar ma activates [+int] force on the predicate it selects. Hence, it obtains a yes-no question interpretation when an IP follows (cf. (24)), and a wh-question when it precedes a focused DP (as in (25)): (24)

awkí ma biyaakitta ? boy.NOM QM be sick.3SGM ‘(As for the boy), is he sick?’

(25)

ma

ÀWKA QM boy.ABS ‘Which BOY is sick?’

(y) 3SG

biyaakittam ? be sick.3SGF-COMP.NOM

The latter interpretation, triggered by the combination of [+foc] and [+int], leads us to explore the correlation between FocP and IntP in more detail.

172

Mara Frascarelli and Annarita Puglielli

4.2. Wh-questions Given the information-structural parallelism between focus and whconstituents (they are both syntactic operators connected with new information), FocP has been very often considered to be some sort of “multifunctional” projection (cf. Horvath 1986, Rizzi 1997, Frascarelli 2000) targeted by different focus-related categories (new and contrastive foci, whphrases, negative polarity items, etc.). Indeed, wh-constituents and foci do not co-occur in the same clause, and this can be explained by assuming that they compete for the same functional position (Rizzi 1997). This assumption, however, is only partially correct. Frascarelli (2005) shows that the realization of focus and wh-constituents presents several asymmetries cross-linguistically and concludes that different projections must be posited for their interpretation in the C-domain. In particular, the interpretation of (genuine) wh-questions depends on the interaction between Foc° and Int°. For this, the internal structure of wh-phrases in Cushitic languages provides immediate support. Consider the following (Somali): (26)

Cali muxuu (= ma+wax+baa 19 +uu) cunay ? QM+thing+FM+3SGM eat.PAST.1SG Cali ‘As for Cali, what did she eat?’

As is clear, the wh-phrase maxaa (‘what’) is obtained through the combination of two functional heads ([+int] ma and [+foc] baa) with the generic NP wax. 20 This is also the case in Afar, in which the QM ma combines with the relative pronoun iyya (cf. (27)) and with the generic noun waxa ‘thing’ (cf. (28)): (27)

lee MIYYA-y water who.ABS-3SG ‘Who drank the water?’

torobem ? drink.PAST.3SGF-COMP.NOM

(28)

MAXA-t baahtam ? awka lee girl water what.ABS-with bring.PAST.3SGF-COMP.NOM ‘With what brought the girl water?’

We argue that wh-constituents are not inherently specified for a discourse feature connected with question, even though they are potential candidates

Focus in the Force-Fin system: Information structure in Cushitic languages 173

for expressing such a discourse property. Wh-constituents take on a genuine (i.e. information-seeking) interrogative force only if they combine with [+foc]. For this reason, a chain is created between Int° and Foc° through AGREE. This endows Int° with the [+foc] feature, inducing interrogative force on the constituent in its scope. 21 The derivation of (26) thus is the following: (29)

[TopP Calik [IntP [Int’ ma[+int] [FocP WAX [Foc’ baa[+foc]+uuk [FinP ‡ [SC [DP[CP OP [AgrSP tuu tOP cunáy] twax ]]]]]]]]] [lit.: ‘As for Cali, was there ANYTHING that he ate?’]

The [+int] feature also triggers operator-movement to Spec,IntP. This distinguishes a question focused on a DP from a (quantified) wh-question. Consider the following contrast: baa yimid ? man FM come.PAST.RED ‘Is there a MAN who came?’ b. NINMAA (nin+ma+baa) yimid ? come.PAST.RED man.QM.FM ‘WHICH MAN came?’

(30) a. ma

NIN

QM

(31) a. Haa (NIN baa yimid). ‘Yes (A MAN came).’ b. CALI baa yimid. ‘CALI came.’

(ok for (30a); * for (30b)) (* for (30a); ok for (30b))

In both (30a–b), the activation of the focus field defines yimid (‘who came’) as presupposed information. However, (30a) is a yes-no question while (30b) is not, as is shown by the possible answers in (31a–b). This shows that the interpretation of wh-questions is obtained through a Spechead relation in IntP when Foc° is activated. This analysis is further supported by wh-phrases like goormaa ‘when’ in Somali: (32)

GOORMUU

(goor+ma+baa+uu)

time.QM.FM.3SGM

(33)

yimid ? come.PAST.3SGM

‘When [lit: ‘in which time’] did he arrive?’ [IntP GOOR [Int’ ma[+int] [FocP [Foc’ baa[+foc]+uu [FinP ‡ [SC [DP[CP OP [AgrSP tuu tOP yimid ] tgoor ]]]]]]]]]

174

Mara Frascarelli and Annarita Puglielli

In conclusion, the interpretation of genuine wh-questions requires the SC derivation discussed for focus in Section 2. A wh-question implies the focalization of a generic NP (i.e. PERSON, THING, TIME, etc.) that is in the scope of [+int]. Since a generic NP cannot specify a value for the operator in the relative clause, the resulting identification is “vacuous” and the information-seeking value is thus obtained. A final consideration concerning the syntactic nature of the interactions shown between functional positions in the C-domain is in order. The question might arise whether a base-generation analysis could not account for them in a more economic way. Movement constraints and scopal properties, however, show that the operation Move (in a cartographic approach) is needed. Evidence in this direction is provided by WCO effects (in Afar 22 ) and scope ambiguities in focus constructions, as is shown in the following Somali sentence: (34)

DHAMMAN BUUGAAGTA bay

saddexda arday gateen

all

three.ART

books.ART

FM.3PL

student buy.PAST.3PL

‘The three students bought EVERY BOOK.’ Sentence (34) allows for a double reading: either each of the three students bought all the books (of a given list) or the books were all bought by the three students (together). The possibility of the first reading shows that the quantified phrase dhamman bugaagta can be interpreted in the scope of the DP saddexda, that is to say, in the position of reconstruction at LF.

5.

Focus and illocutionary force

The data discussed so far have shown that FocP is not a “multifunctional” position: FocP is only dedicated to the syntax and interpretation of informational focus. The activation of [+foc], on the other hand, is crucial for the interpretation of focus-related discourse categories. In this respect, it is important to point out that (genuine) focus information is only available in a matrix C-domain. Consider the following data from Somali (for cross-linguistic evidence, cf. Frascarelli 2005):

Focus in the Force-Fin system: Information structure in Cushitic languages 175

(35)

* hilibka meat.ART

waan

sheegay [ in say.PAST.1SGM that NIMANKÁAS baa cunayá] men-those.ABS FM eat.PROG.RED ‘I said that THOSE MEN are eating the meat.’ DECL.SCL1SG

(36)

[ Isaga [ oo isbitaalka (*baa) ku CONJ hospital.ART FM in he buu dhintay FM.SCL3SGM die.PAST.3SGM ‘He died while he was in the hospital.’

jira]] stay.PAST.RED

(37)

Wiilka [ Maryam (*baa) la hadlayá] waa walaal-kay. boy-ART Maryam FM with speak.PROG.RED DECL brother-my ‘The boy that is talking to Maryam is my brother.’

As we can see, no [+foc] feature is available in complement (35), adverbial (36), or relative (37) clauses. Indeed, wh-phrases in embedded domains are realized as generic NPs heading a relative clause: (38)

waxaan (wax+baa+aan) weyddiyay [goorta /*goormaa thing.FM.1SG ask.PAST.1SG time-ART / time-QM.FM Cali imaanayó] Cali come.PRES.3SGM.DEP ‘I was wondering when Cali is going to come.’ [lit.: ‘The thing that I was asking (is) the time (that) Cali is coming.’]

Sentences like (38) show that genuine (i.e. focused) wh-phrases cannot be realized in embedded clauses. This shows that “IntP is inherently endowed with a wh [i.e., quantificational] feature” (Rizzi 2001: 293); not with a [+foc] feature, however. This combination is obtained through AGREE with Foc°, which can only occur in a matrix C-domain. In line with recent proposals 23 , we thus conclude that new information is strictly connected with the root illocutionary force.

176 6.

Mara Frascarelli and Annarita Puglielli

Topic constituents and information structure in the C-domain

All along this presentation, we have provided evidence for a number of interactions occurring between functional projections when the focus field is activated. The presence of topic constituents, on the other hand, has proved a marginal role in this respect: [-foc] extraposed DPs do not interfere with operator-movement and the specification of force. Nevertheless, a closer look at the syntax-discourse properties of topics can help provide a deeper understanding of information structure in the C-domain. First of all, it is important to note that, even though dislocation of [-foc] constituents is a major feature of Cushitic languages, two positions are totally excluded for topics, that is to say below FinP (cf. (39)) and between IntP and FocP (cf. (40)): (39)

Cali Maryam ma (*Maryam) QM Cali Maryam ‘(As for) Cali, did he see her, Maryam?’

arkay ? see.PAST.3SGM

(40)

Cali ma (*Cali) HADIYAD buu keenáy ? QM present FM.3SGM bring.PAST.3SGM Cali ‘(As for) Cali, is it a PRESENT that he brought?’

In a cartographic approach of functional projections, we propose that this is not a language-specific property but the effect of a structural condition for interface interpretation. In Sections 3 and 4, we have shown the crucial role of AGREE relations between Fin° and AgrS° and between Int° and Foc°. We therefore argue that TopP projections are excluded in these positions to avoid topic interference with UG requirements. Data from Cushitic languages also show that TopP positions are associated with different discourse properties (cf. Frascarelli and Hinterhölzl 2007). Consider the following asymmetry in the realization of topics in Somali yes-no questions: (41) a. hadiyadda (*hadiyad) ma CALI baa keenáy ? QM Cali FM bring.PAST.RED present.ART ‘(As for the present) did CALI bring it?’ b. Ma CALI baa hadiyad (*hadiyadda) keenáy? ‘Did CALI bring a present?’

Focus in the Force-Fin system: Information structure in Cushitic languages 177

In (41), the interrogative force is centered on the focused constituent (marked by baa) while the rest of the sentence is presupposed information. It is thus interesting to note that on the topic (hadiyad) an opposite [definiteness] requirement depending on its position is imposed, which is connected with its role in the discourse (as is made clear by translations). Specifically, the high topic in (41a) must be definite and obtains an “aboutness” role while the low topic in (41b) is indefinite and is simply part of background information (as a “familiar” element). Somali data thus support an analysis in terms of “different types of topics” (cf. note 2): even though [-foc] DPs are merged in extra-sentential position, it is not the case that just any kind of topic can occupy any position in the C-domain. Instead, there is a systematic correlation between discourse roles and grammatical properties of topics, which is encoded in a strict hierarchy of functional heads. In particular, the topic position that is higher than the focus field defines “what the sentence is about” and cannot be used to express “familiarity” (which is encoded in the structural domain between Foc and Fin). For further support, consider the “broad focus” question in (42) and its possible answers in (43): (42)

Cali muxuu

sameeyay ?

Cali

do.PAST.3SGM

QM.thing.3SGM

‘What did Cali do?’ (43) a. Cali MARYAM buu dilay ‘Cali beat MARYAM.’ b. MARYAM buu dilay, Cali c. *MARYAM baa Cali dilay As we can see, when a DP (Cali, in the relevant case) is presented as the aboutness topic in the question, it cannot be realized as a low topic in the answer. It can either be maintained in that position or realized as a righthand topic (as an “afterthought”). To conclude this section, let us briefly consider topicalization in Afar. As is shown in (44)–(45), what is “given” information in the question must be realized in a fixed order in the answer, which is exactly the order of arguments in unmarked sentences (namely S>OI>OD):

178

Mara Frascarelli and Annarita Puglielli

(44) Q: ‘Who gave water to Maryam ?’ Maryam-a lee a. OMAR tee’em Omar give.PAST.3SGF-COMP.NOM Maryam-to water ‘OMAR gave water to Maryam.’ b. Maryam-a lee OMAR tee’em. c.* OMAR tee’em lee Maryam-a. d.* lee Maryam-a OMAR tee’em. (45) Q: ‘What did Omar give to Maryam ?’ tee’em Omar a. LEE water give.PAST.3SGF-COMP.NOM Omar ‘Omar gave WATER to Maryam.’ b. Omar Maryam-a LEE tee’em. c.* LEE tee’em Maryam-a Omar. b.* Maryam-a Omar LEE tee’em.

Maryam-a. Maryam-to

To provide an explanation, it is important to point out that in Afar cliticresumption is not available. According to our proposal (cf. Section 3.2), this core-grammar property entails that topics are merged in IP-internal position. It is thus possible to argue that the rigid ordering imposed on topics is due to movement constraints (i.e. minimality). Hence, given information is either in its basic position (so that right-hand topicalization ((44a)– (45a)) is in fact PF marginalization; cf. Cardinaletti 2002) or, after verb raising, the remnant VP is moved to the C-domain (possibly to Spec,GP; cf. Poletto and Pollock 2004) to obtain left-hand topicalization ((44b)–(45b)). Further investigation is needed to gain deeper understanding of the structure and interpretation of topic constructions. The connection between discourse roles and formal properties of topics is the subject of ongoing research. 7.

Conclusions

This paper has provided substantial evidence that discourse roles are connected with specific positions in the C-domain and that AGREE relations must be posited between functional heads for interpretative requirements. We have shown the crucial role of the [+foc] feature for the activation of different kinds of new-related information and its importance for the definition of the illocutionary force of the sentence. Focus and force are thus strictly connected and, in this respect, a structural distinction must be pos-

Focus in the Force-Fin system: Information structure in Cushitic languages 179

ited between matrix and embedded clauses that is centered on the presence of a “focus field”. Topic constituents, on the other hand, are independent of the illocutionary force of the sentence and are present in both root and embedded clauses. Their different position in the C-domain is only dependent on their specific discourse role.

Notes

1.

2.

3.

4.

5.

We thank Istvan Kenesei, Victor Manfredi, Chris Reintges, and the audience of the Workshop on Focus in African Languages in Amsterdam (December, 3–4 2004) for helpful comments and discussion. Investigation of the left periphery of the sentence has gained significant impulse in the Minimalist framework, given the centrality of Interfaces (the only “conceptually necessary” levels, cf. Chomsky 1995) and the specification that Internal Merge is connected with scopal features and discourse requirements (cf. Chomsky 2002, 2004). Contra a “free recursion” analysis of topicalization (cf. Rizzi 1997), Frascarelli and Hinterhölzl (2007) provide substantial evidence that different TopP projections must be posited to realize different “types of topics”. In particular, the [aboutness] feature is realized in the highest TopP node, while low topics are connected with [familiarity] and used for continuity in the discourse. On this point, see also Section 6. When not otherwise indicated, examples are taken from original data. In this respect, we thank Axmed Cabdullaahi Axmed, Cabdalla Omar Mansur (Somali) and Mohammed Ali Mahmoud (Afar) for their help and patience in testing and discussing our data. The list of the abbreviations used in the glosses is the following: ABS = absolutive case ANAPH = anaphoric article ART = definite article DECL = declarative (marker) F = feminine FM = Focus Marker M = masculine NOM = nominative case OCL = object clitic PAST = past tense PL = plural PROG = present progressive QM = question marker RED = reduced paradigm SCL = subject clitic SG = singular The 3rd person object clitic (both genders and numbers) has no phonetic realization and must be interpreted as an object pro. Evidence for this analysis is provided by the interpretation of sentences like (i) below. As is shown in the English translation, this empty category always obtains a referential reading, so that a pseudo–intransitive use is excluded in this language (for details, cf. Puglielli 1981, Svolacchia and Puglielli 1999):

180

Mara Frascarelli and Annarita Puglielli (i)

Cali baa pro arkay. Cali FM see.past.red ‘CALI saw it/him/her/them.’ 6. The ABS label is used to indicate non-nominative case, as in use among scholars of African languages. 7. The traditional arguments for positing a relative clause analysis are discussed in F&P (2005a). A comparative analysis of the syntactic properties of relative clauses in Somali and Afar is provided in F&P (2005b). 8. The literature on free relatives is very rich. The present approach is in line with an analysis of the free relative as a DP that takes a complement CP (cf. among many others, Groos and van Riemsdijk 1981, Grosu 1994). In a sentence like (5) the DP-head is the generic (null) noun ‘PERSON’. 9. In Frascarelli (2005), this analysis is discussed in detail and proposed for a unified theory of focus structure and interpretation, in which identification and spell-out conditions depend on core grammar rules (the Phase Impenetrability Condition) and interface requirements. 10. The reduced paradigm has three forms in Somali: one for 3SGF, one for 1PL and one for all other persons (cf. Puglielli 1981). 11. It is important to underline that reduced agreement is not a specific property of focus constructions, but a morphosyntactic effect arising whenever the subject is the head of an operator-chain. Consider, for instance, the following sentence in which nimankaas is the head of a relative clause (for the analysis of waa, cf. Section 3.1 below): (i) nimankaas hilibka cunayá waa saaxiibboday men-those meat.ART eat.PROG.RED friends-POSS1SG ‘Those men that are eating meat are friends of mine.’ The present analysis thus provides a unified account for anti-agreement and related effects. 12. Although –m plausibly derives from an original lexical form (roughly meaning ‘thing’), it cannot be considered the NP-head of the relative clause. Evidence for its COMP status is provided by the focalization (and relativization) of PPs. In this case, the preposition cannot be postponed to –m, forcing a pronominal resumption strategy within the relative clause. This is illustrated below: (i) MOHAMMADA Ilyas maruw ka-h ye’em him-to Mohammed.ABS Ilyas ram give.PAST.3SG.COMP.NOM ‘Ilyas gave a ram to MOHAMED.’ 13. For space limitations, the syntax of waa sentences cannot be addressed in detail. For the sake of discussion, we simply point out that waa does not trigger a contrastive reading of the verb, nor does it lead to anti-agreement effects. Indeed, this property is fully expected in the present analysis. As is discussed in F&P (2005a), waa constructions do not involve a func-

Focus in the Force-Fin system: Information structure in Cushitic languages 181 tional relation between an operator and a variable in a relative clause. Quantifying implies, in fact, partitioning of the universe and this operation only pertains to nominal elements. Anti-agreement effects with focused verbs are therefore not expected cross-linguistically; possible counterexamples would be ascribed to a nominalization strategy for verb focusing (which is attested in a number of languages). 14. As is discussed in Chierchia and McConnel-Ginet (1990), the sentential force refers to the conventional pragmatic force associated with a sentence type. The illocutionary force, on the other hand, is connected with speaker’s intentions. 15. Indeed, the FM baa does not show any type of number/gender agreement with the focus phrase in Spec,FocP. 16. Case is transmitted to topic DPs through binding chains (cf. Baker 1996). 17. Note that subjects can be extracted independent of the syntactic role (argument vs. adjunct) of the constituent in topic position, as is shown by the following sentence: (i) anigu CALI baan shalay ku arkay, jaamacadda. ‘I saw CALI, yesterday, at the university.’ 18. This is consistent with the Extension Condition (Chomsky 1995, 2001) and with the fact that, cross-linguistically, it is not possible to focus a constituent and question another part of the sentence. When a question contains a focused DP, this is the predicate in the scope of the [+int] force. The rest of the sentence is presupposed information: (i) Is JOHN going out? [ = ‘Someone is going out (presupposition). Is it JOHN (Focus)?’] 19. The phonological process that obtains the FM –aa in clusters is a debated question that cannot be addressed in detail. Lamberti (1983) suggests that baa is derived from an original form *awaa and, indeed, the alternation [w]|[b] is widely attested in the derivation of Somali verbal nouns (e.g., duq ‘old’ Æ duqow ‘to get old’ Æ duqoobid ‘getting old’; for further details, cf. F&P 2005a). It is therefore possible that, in clusters with the FM, an underlying [w] is present and simplified (a common morphophonological operation, as is shown by ma+wax in (26). 20. This analysis is very much in the spirit of Cheng’s (1991) proposal of whitems as free-choice indefinites (with the interrogative interpretation being brought about by dedicated question particles). The non-referential interpretation of the noun wax in Somali is supported by independent evidence, as is shown by examples like the following: (i) Wax waan arkay. thing decl.1sg see.past.1sg (‘I saw something.’) 21. Despite the “Freezing Condition”, Rizzi (2004: note 8) does not exclude the possibility of a “combination of features”. In particular, the author

182

Mara Frascarelli and Annarita Puglielli suggests the possibility of a complex Q+Foc head (after movement of Q to Foc). 22. WCO effects are excluded in polysynthetic Somali, since full DPs never bind a trace in A-position (cf. Section 1.2). An illustration from Afar whquestions is given below: MIYYAY k torobem? (i) * kayk lee his water who.abs-3sg drink.past.3sgf-comp.nom 23. See Haegeman (2002) on adverbial clauses and Bayer (2001) on emphatic topicalization in Bavarian.

References Appleyard, David L. 1989 The relative verb in focus constructions: an Ethiopian areal feature. Journal of Semitic Studies 34.2: 291–305. Baker, Mark 1996 The Polysynthesis Parameter. Oxford: Oxford University Press. Bayer, Josef 2001 Asymmetry in emphatic topicalization. In Studia Grammatica 52: Adiatur Vox Sapientiae, Caroline Féry and Wolfgang Sternefeld (eds.), 15–47. Berlin: Akademie Verlag. Bliese, Loren F. 1981 A Generative Grammar of Afar. Arlington: The Summer Institute of Linguistics and University of Texas. Cardinaletti, Anna 2002 Against optional and zero clitics. Right Dislocation vs. Marginalization. Studia Linguistica 56.1: 29–57. Cheng, Lisa Lai-Shen 1991 On the Typology of Wh-questions. MIT, Ph.D. dissertation, Chierchia, Gennaro, and Sally McConnell-Ginet 1990 Meaning and Grammar. Cambridge Mass., MIT Press. Chomsky, Noam 1995 The Minimalist Program. Cambridge, Mass.: MIT Press. 2001 Derivation by phase. In Ken Hale: A life in language, Michael Kenstowicz (ed.), 1–52. Cambridge, Mass.: MIT Press. 2002 An interview on Minimalism. In On Nature and Language, Adriana Belletti and Luigi Rizzi (eds.), 92–161. Cambridge: Cambridge University Press. 2004 Beyond explanatory adequacy. In Structures and Beyond. The Cartography of Syntactic Structures. Vol. 3, Adriana Belletti (ed.), 104– 131. Oxford: Oxford University Press.

Focus in the Force-Fin system: Information structure in Cushitic languages 183 É. Kiss, Katalin (ed.) 1995 Discourse Configurational Languages. Oxford: Oxford University Press. Frascarelli, Mara 1999 Subject, nominative case, agreement and Focus. In Boundaries of Morphology and Syntax, Lunella Mereu (ed.), 195–215. Amsterdam: John Benjamins. 2000 The Syntax-Phonology Interface in Focus and Topic Constructions in Italian. Dordrecht: Kluwer Academic Publishers. 2004 Dislocation, clitic resumption and minimality: A comparative analysis of left and right Topic constructions in Italian. In Romance languages and linguistic theory 2002, Reineke Bok-Bennema, Bart Hollebrandse, Brigitte Kampers-Manhe, and Petra Sleeman (eds.), 99–118. Amsterdam: John Benjamins. 2005 Focus as predicate and Spell-out conditions. Deriving crosslinguistic strategies from a unified account. Ms. Univ. Roma Tre. Frascarelli, Mara and Annarita Puglielli 2005a The Focus System in Cushitic Languages. In Proceedings of the 10th Hamito-Semitic Linguistics Meeting (Quaderni di Semitistica, 25), Pelio Fronzaroli and Paolo Marrassini (eds.), 333–358. Firenze: Università degli Studi di Firenze. 2005b A comparative analysis of restrictive and appositive relative clauses in Cushitic languages. In Contribution to the IGG XXX, Laura Brugè, Giuliana Giusti, Nicola Munaro, Walter Schweikert, and Giuseppina Turano (eds.), 279–303. Venezia: Cafoscarina. to appear Focus Markers and Universal Grammar. In Cushitic and Omotic languages: Proceedings of the fourth International Conference on Cushitic and Omotic Languages (Leiden, 10–12 april 2003), Azeb Amha, Maarten Mous, and Graziano Savà (eds.), Cologne: Rüdiger Köppe Verlag. Frascarelli, Mara, and Roland Hinterhölzl 2007 Types of Topics in German and Italian. In On Information Structure, Meaning and Form, Susanne Winkler and Kerstin Schwabe (eds.), 87–116. Amsterdam: John Benjamins. Groos, Anneke, and Henk van Riemsdijk 1981 Matching effects in free relatives: a parameter of core grammar. In Theory of Markedness in Generative Grammar, Adriana Belletti, Luciana Brandi, and Luigi Rizzi (eds.), 171–216. Pisa: Scuola Normale Superiore. Grosu, Alexander 1994 Three Studies in Locality and Case. London: Routledge.

184

Mara Frascarelli and Annarita Puglielli

Haegeman, Liliane 2002 Anchoring to speaker, adverbial clauses and the structure of CP. Georgetown University Working Papers in Theoretical Linguistics 2, 117–180. Hetzron, Robert 1980 The Limits of Cushitic. Sprache und Geschichte in Africa 2, 7–126. Horvath, Julia 1986 Focus in the Theory of Grammar and the Syntax of Hungarian. Dordrecht: Foris. Lamberti, Marcello 1983 The origin of the Focus particles in Somali. In Sprache, Geschichte und Kultur in Afrika, Vossen von Rainer and Ulrike Claudi (eds.), 59–112. Hamburg: Helmut Buske Verlag. Ouhalla, Jamal 1993 Subject-extraction, negation and the anti-agreement effect. Natural Language & Linguistic Theory 11: 477–518. Parker, E. M., and R .J. Hayward 1985 An Afar-English-French Dictionary (with Grammatical Notes in English). London: SOAS. Puglielli, Annarita (ed.) 1981 Studi Somali 2. Sintassi della Lingua Somala. Roma: Ministero AA EE. Rizzi, Luigi 1997 The Fine Structure of the Left Periphery. In Elements of Grammar. Handbook in Generative Syntax, Liliane Haegeman (ed.), 281–337. Dortrecht: Kluwer Academic Publishers. 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. 2004 (ed.) The Structure of CP and IP: The Cartography of Syntactic Structures. Vol. 2. Oxford: Oxford University Press. Saeed, John Ibrahim 1984 The Syntax of Focus and Topic in Somali. Hamburg: Helmut Buske. Schachter, Paul 1973 Focus and relativization. Language 49:19–46. Svolacchia, Marco, and Annarita Puglielli 1999 Somali as a polysynthetic language. In Boundaries of Morphology and Syntax, Lunella Mereu (ed.), 97–120. Amsterdam: John Benjamins.

Coptic relative tenses: The profile of a morphosyntactic flagging device Chris H. Reintges

Abstract A number of languages morphologically distinguish relative clauses and questions as well as focus constructions from simple declarative clauses by means of special inflectional morphology. This morphological pattern thus provides prima facie evidence for these constructions as a natural class. The present paper comprises a case-study of the morpho-syntax of so-called relative tenses in Coptic Egyptian (Afro-Asiatic, 3rd-13th c. CE). Different from well-studied cases of special inflection, Coptic relative tenses are selected when the questioned or focused constituent appears in-situ in a clause-internal position, yet displays the scope and interpretational properties of a displaced constituent. To accommodate these facts, an analysis in terms of covert wh-movement will be proposed.

1.

Introduction

Since Chomsky’s (1977) ground-breaking study on wh-movement, considerable attention has been paid to the syntactic similarities between such apparently diverse construction types as relative clauses, constituent questions, and contrastive focus structures. What these sentence types have in common, is that they can be most insightfully analyzed as involving an open position or “variable” that is assigned an interpretation by a scopetaking operator. In the generative-transformational framework, the operator-variable relation has been analyzed as the outcome of a movement operation that links two or more positions in the syntactic structure. The displaced operator (a question word or focus phrase) is relocated to a position in which its scope and interpretational properties are determined while its case and thematic properties are interpreted in the original position (or intermediate ones along the movement path). The evidence for the unity of operator-variable constructions is largely indirect: since operator-variable dependencies typically involve syntactic

186

Chris H. Reintges

reordering, the point of origin of the movement process is generally phonetically empty. English constituent questions represent a clear example where the operator is visible in the form of the fronted wh-phrase while the associated position in which the wh-phrase is interpreted is not, e.g. Whom did you meet __ yesterday evening? (where __ indicates the position from which movement has occurred). Indeed, in the standard analysis of thatrelative clauses (Browning 1987), neither the operator nor the variable is phonetically realized, e.g. the man [Op that you met __ yesterday evening], where Op stands for the null counterpart of the relative pronoun whom. There is, however, direct morphological evidence for the naturalness of operator-variable constructions as a class. A number of genetically related and unrelated languages distinguish relative clauses, questions, and a range of declarative focus constructions from pragmatically neutral clauses by means of a special type of inflectional morphology. A particularly well studied example of this morphological pattern comes from relative aspects in Hausa (Chadic, Nigeria). In this language, the perfective and imperfective forms divide into two paradigms: the “general” and the “relative”. The latter is selected in relative clauses, wh-fronting questions, and focusfronting constructions. See Tuller (1986), Green (1997, in press), Newman (2000), Jaggar (2001), Hartmann & Zimmermann (this volume) for a detailed description and analysis. 1 (Relative aspect forms appear in boldface; small caps indicate focus.) (1)

a. Relative clause kƗ ga bƗ̖Ýî-n 2SG.M.PERF

[dà

see guest-DD.PL COMPREL

sukà

(Hausa) isǀ yànzu]?

REL.PERF.3PL

arrive

now

‘Did you see the guests who just arrived?’ (Jaggar 2001, 527) b. Wh-question sukà zǀ ? wƗ̖ dà wƗ̖ REL.PERF.3PL come who.pl ‘Who came?’ (Newman 2000, 488) c. Declarative focus sentence Amir˾kƗwƗ nƝ̖ sukà fƗrà zuwƗ̖. Americans COP.PL REL.PERF.3PL be.first go.VN ‘(Who got to the moon first?) It was the AMERICANS who got (there) first.’ (Jaggar 2001, 494) The Coptic conjugation system shows a similar subdivision of tense-aspectmood (TAM) markers which fall into a “general” and a “relative” class. 2

Coptic relative tenses: The profile of a morpho-syntactic flagging device 187

Thus, consider the following sentence pair in which both the declarative clause and the corresponding wh-questions employ the Perfect marker a, which appears in front of the subject. (The basic word order pattern is subject-verb-object SVO.) But while pragmatically neutral declaratives like (2a) are well-formed with the tense-aspect word alone, wh-questions like (2b) require the presence of the relative marker ‹ntí besides the TAM marker in order to be grammatical. (2)

a. Neutral declarative clause (general TAMs) awn a ruhe šnpe ero-f J‹O p-ma and

PERF night

happen to-3SG.M in

(Coptic) ‹n-Taȕenİse

DEF.SG.M-place of-Tabenêse

‘And night-fall surprised him in the area of Tabenêse.’ (V. Pachôm. 136, 24–25) b. Wh-question (relative TAMs) šnpe ‹mmo-k pa-þoeis p-‹rro ‹nt-a u REL-PERF

what happen to-2SG.M

DEF.SG.M.1PL-lord DEF.SG.M-king

‘What happened to you, our lord and king?’ (Eudoxia 36, 24) On the face of it, Coptic relative tenses look like a clause-typing device that distinguishes questions from declarative sentences. A clause-typing analysis would, however, be at odds with the broad distribution of relative TAMs across different sentence types. Just like in Hausa, Coptic relative TAMs appear in relative clauses, wh-questions, and focusing sentences. 3 (3)

a. Relative clause nhİt-f ] e-p-ma [nt-a-k-kynt-f REL-PERF-2SG.M-find-3SG.M inside-3SG.M to- -place ‘the place where you found it’ (Ac. A&P 204,145-146) b. Wh-question ȕnk e-pe.k-hİt ? ‹nt-a u REL-PERF what come to-DEF.SG.M-2SG.M-heart ‘What has come into your heart?’ (AP Chaîne, no.139, 31:7) c. Declarative focus sentence ‹mmnn alla ‹nt-a-u-s‹ȕtot-f REL-PERF-3PL-prepare-3SG.M no but ‹m-p-diaȕolos m‹n ne.f-aggelos for-DEF.SG.M-devil with DEF.PL-3SG.M 3sm-angels ‘(Is the Purgatory prepared for us?) Not at all! It is rather prepared FOR THE DEVIL AND HIS ANGELS.’ (B. Hom. 14, 19-21)

188

Chris H. Reintges

The presence of operator-variable dependencies in relative clauses, questions, and contrastive focus constructions provides a necessary, but not sufficient condition for the introduction of relative tenses into the structure: the relevant operator must also be in the appropriate configuration. As shown by the contrast between (4a) and (4b) below, relative TAMs occur in wh-in-situ questions, so called because the question word is located in the same position as its non-interrogative counterpart in declarative sentences. By contrast, no such special inflection is needed when the question word moves to the left of the clausal subject. (4)

a. Wh-in-situ question (relative TAMs) awo nt-a-u-ei eȕol tnn ? and REL-PERF-3PL-come PCL where ‘From where did they come?’ (Apoc. 7, 13; ed. Budge) b. Wh-fronting question (general TAMs) eȕol tnn a-tet‹n-ei e-pei-ma ? PCL where PERF-2PL-come to-DEM.SG.M-place ‘From where did you come here?’ (B. Martyr. 220, 8)

These distributional patterns highlight the two important aspects of Coptic relative tenses: on the one hand, this special inflection unifies a class of constructions while, on the other hand, it is subject to parametric variation in terms of the syntactic conditions that govern its presence. The paper is organized as follows. Section 2 sets the stage by discussing the typological profile of special inflection. Section 3 provides some background information about Coptic clause structure with particular attention paid to the topic/focus field. The focus of section 4 is on the categorical properties and morphological alternations of relative TAM markers, which will be analyzed as alternating relative complementizers. Section 5 examines the distribution of relative TAMs across different sentence types. Elaborating on Reintges, LeSourd & Chung’s (2006) comparative study, section 6 presents a syntactic analysis of Coptic wh-in-situ questions as “hidden” movement configurations. Section 7 examines the relation between morpho-syntactic flagging and the overt versus covert movement distinction. Section 8 summarizes the main results of the paper.

Coptic relative tenses: The profile of a morpho-syntactic flagging device 189

2.

Parameters of cross-linguistic variation

The use of special inflectional morphology is attested cross-linguistically as one of the options for the structural encoding of features associated with the standard inventory of operator-variable constructions (e.g. Hyman & Watters 1984, Haïk 1990, Watanabe 1996, Chung 1998). Reintges, LeSourd & Chung (2006, 166ff.) identify a set of parameters of crosslinguistic variation, which include (A) the location of special inflection, (B) the structural information encoded, (C) the recursiveness property in longdistance movement construction, and (D) the inventory of operator-variable constructions thus flagged. A.

LOCATION. The morphology of special inflection is associated with a restricted set of syntactic heads. According to Zaenen (1983), this type of morphology surfaces on verbs, inflectional heads (TAM markers), or left-peripheral heads like complementizers.

In Hausa, this morphological pattern is visible in the so-called personaspect complex (PAC), which occurs in preverbal person and is morphologically independent of it. The PAC inflection consists of a weak pronoun and a TAM marker, which are sometimes fused together (Newman 2000, 564ff.). Depending on how the PAC is decomposed in such cases, one might infer that inflection is either fused with a morpheme that encodes person, number, and gender features (i.e. phi-features) or else encodes phi-features itself. On either analysis, Hausa represents the typologically marked case of a language in which this morphological pattern is sensitive to phi-features. Furthermore, moved focus/wh-phrases may be modified by a pronominal copula that recurs in predicational sentences. Green (1997, in press) analyzes the copula as a focus marker that is merged in the head position of a designated focus phrase. As illustrated in (5), the copula cƝ̖ (SG.F) agrees in number and gender with the question word wƗ̖cƝ ‘who’ (SG.F). (5)

Wh-question with relative aspect and focus copula cƝ̖ ta mutù ? wƗ̖cƝ who.SG.F COP.SG.F 3SG.F.REL-PERF die ‘Who died?’ (Jaggar 2001, 514)

(Hausa)

The semantic contribution of the optional focus copula is subtle, but it seems to reinforce an exhaustive listing interpretation. In this respect, it

190

Chris H. Reintges

differs from fully grammaticalized focus markers (Hartmann & Zimmermann, this volume). Green & Reintges (2005) view the focus copula as a different manifestation of special inflectional morphology. However, in having semantic import, it differs from relative aspects that merely signal an application of wh-movement. B.

STRUCTURAL INFORMATION. In some languages, special inflection only indicates operator-variable dependencies. In other languages, it also indicates the categorial features as well as the case and thetarole of the moved constituent (cf. the term “wh-agreement”).

Chamorro, an Austronesian VSO language spoken on the Mariana Islands, has a standard inventory of wh-constructions all of which exhibit the hallmarks of wh-movement (displacement of the wh-phrase to the left periphery, obligatory gap at the extraction site, sensitivity to islands). According to Chung (1994, 1998), wh-movement feeds two distinct morphological operations. One operation triggers the introduction of “wh-agreement” marking on finite verbs and adjectives. The other operation involves alternations in the form of the lexical complementizer, which may lack phonetic content (indicated as [C Ø]). Wh-agreement registers the case of the variable left behind by whmovement while Operator-C agreement marks both the categorial status and the thematic role of the moved constituent. Thus, compare the declarative clause in (6a), where the verb assumes neutral inflection, with the corresponding wh-questions in (6b–d) below. In (6b), the nominative case of the moved wh-subject is flagged by the wh-agreement morpheme -um- on the finite verb. In (6c), the objective case of the wh-object is registered by the infix -in- plus possessor agreement. In (6d), the complementizer ni marks both the nominal category and the locative theta-role of the fronted wh-adverb manu ‘where’. (6)

a. Neutral declarative VSO clause (Chamorro) Ha-fa!gasi si Juan i kareta. AGR-wash Juan the car ‘Juan washed the car.’ b. Wh-agreement; wh-subject with nominative case Hayi [C Ø] fuma!gasi i kareta ? WH[NOM].wash the car who COMPQ ‘Who washed the car?’

Coptic relative tenses: The profile of a morpho-syntactic flagging device 191

c. Wh-agreement; wh-object with oblique case si Henry pära hagu? Hafa [C Ø] fina!gasése-nña Henry for you what COMPQ WH[OBJ].wash.PROG-AGR ‘What is Henry washing for you?’ (Chung 1998, 236, (52)–(53a–b)) d. Operator-C agreement, wh-adverb Manu ni mañ-ásaga siha ? where COMPQ AGR.live.PROG they ‘Where are you living?’ (Chung 1998, 58, (80a)) The morphological effects of wh-agreement proper are induced by the trace left behind by the moved wh-phrase, which thus behaves just like any overt nominal in that it must be case-marked by the inflected verb. C.

RECURSIVENESS. Languages vary according to the domain in which the designated syntactic heads are affected. In cases of longdistance movement across a clause-boundary, special inflection surfaces either on every designated head along the extraction path or on the highest one (Haïk 1990, 354–62).

A salient feature of Modern Irish wh-movement constructions is the presence of the alternating complementizer particles aL (the ‘direct relative particle’) and aN (the ‘indirect relative particle’). The COMP-particle aL is selected in contexts in which wh-dependencies terminate in a gap while the particle aN is chosen when such dependencies terminate in a resumptive pronoun (McCloskey 1990, 2001). In cases of long-distance wh-movement across a clause boundary, every complementizer that intervenes between the moved wh-phrase and the variable in the base position assumes its characteristic form aL. 4 (7)

Multiple occurrence of alternating complementizer (Irish) dúradh léithi [a cheannódh_i é]? Céi a who COMP.aL was-said with-her COMP.aL would-buy it ‘Who was she told would buy it?’(McCloskey 2001, 94 fn. 23)

A different situation obtains in Hausa long-distance wh-movement structures where only the highest TAM inflection assumes the relative aspect form. 5

192 (8)

D.

Chris H. Reintges

Single occurrence of relative aspect (Hausa) yârƗi HàlƯmà ta tàmbayƗ̖ wàÎànnè children Halima 3SG.F.REL-PERF ask which.PL [kǀ___ sun sƗ̖ci kuÎîn] ? COMPQ 3PL.PERF steal money ‘Which children did Halima ask whether they stole the money?’ (Newman 2000, 502) DISTRIBUTION OVER SENTENCE TYPES.

Special inflection always flags canonical wh-constructions, yet in some languages it also applies to yes-no questions, conditionals, temporal adverb clauses, predicative adjuncts, as well as narrative use (Hyman & Watters 1984, 256ff.; Haïk 1990, 350f).

In Hausa, the relative perfective may appear in conditional clauses as well as in series of past events, which constitute the foregrounded portions of the narrative (Tuller 1986, 112ff.; Jaggar 2001, 162ff, in press). (The discourse fragment in (9a) is from a personal “brush-with-death” narrative.) (9)

a. Conditional clause (Hausa) in kin/kikà kintsƗ̖, zƗ mù tàfi. if/when 2SG.F.PERF/2SG.F.REL-PERF be.ready FUT 1PL go ‘If/when you are ready, we’ll go.’ (literally: ‘If/when you have got ready…’) (Jaggar 2001, 609) b Narrative relative perfective (…) mukà bnjÎè Ýǀfàr˾˾

yi

saurƯ,

1PL.REL-PERF open door.of car.DD 1PL.REL-PERF do

speed,

sai

wani mùtûm

then

DEM

mukà

men

mǀtàr˾, mukà

ya

bnjÎƝ̖

mîn

3SG.M. REL-PERF

open

IO.1SG

yi saurƯ, mukà

jƝ, mukà

1PL.REL-PERF do speed 1PL.REL-PERF go

(…)

bnjÎƝ̖ (…)

1PL.REL-PERF open

‘(... ) we opened the car door and moved quickly, then some man opened (it) for me (…), and we moved quickly and went and opened (it) (…)’ (Jaggar 2001, 163) One might wonder whether the narrative use of relative aspects registers the presence of an operator-variable dependency. Since there is no overt focus constituent, the spell-out of special inflection in foregrounded chains of

Coptic relative tenses: The profile of a morpho-syntactic flagging device 193

events may be related to the presence of a null focus operator (Tuller 1986, 117). 3.

The cartography of the Coptic left periphery

In this section, I will lay out the background assumptions about Coptic syntax that the discussion of relative tense formation in the reminder of this study is based on. Coptic Egyptian may be classified as a discourseconfigurational language where topic and focus prominence involve a departure from the canonical SVO surface order. See Reintges (2004, chap. 10) for a more detailed discussion of the correlation between word order and discourse structure. The markers of tense-aspect-mood are phonological clitics, which display characteristic properties of auxiliary verbs, such as impoverished agreement and the compatibility with several clausal positions. Most TAMs occur in clause-initial position, preceding both the subject and the main verb. Root modals, on the other hand, appear in a lower syntactic position, following the subject and preceding the main verb. For the purposes of this paper, I will not further explore the complex interaction between the two inflectional positions but assume, following Rizzi (1997), that clause-initial TAMs are directly merged into the ‘Finiteness Phrase’ (FINP) while root modals are either merged or move together with the main verb to the head of a clause-internal ‘Mood Phrase’ (MODP) (see Reintges 2001 for a more detailed discussion). Just as in Hungarian, Hausa, and the Kwa languages, the discourseconfigurational syntax of Coptic Egyptian involves an articulated topicfocus field (see, among various others, É. Kiss 1998, Rizzi 1997, Green 1997, in press, Green & Reintges 2004a, 2005, Aboh 2004). To begin with, fronted focus/wh-phrases appear following the subordinating complementizer þe ‘that’, as in (10a), or the dedicated interrogative particle eye, as in (10b), either of which is merged into the highest functional head of the clause, viz. the C0/Force0-node. From this it follows that the moved focus/wh-phrase is located in the specifier position of a functional projection below C0 and above the FINP.

194

Chris H. Reintges

(10) a. ti-þ‹nu ‹mmn-t‹n [þe h‹n u ‹n-šaþe (PRES)1SG-ask PREP-2PL COMP with what of-word a-tent‹n-mute ero-i (…)] PERF-2PL-say about-1SG ‘I ask you with which reason do you say about me (...).’ (Acts 10, 29) tet‹n-ynr‹m ‹nsn-n hns

b. eye etȕe u Q

for

what (PRES)2PL-gaze at-1PL

as.if

‹nt-a-‹r REL-PERF-1SG-make

pai

h‹n

te.n-kyom

İ

ten-m‹nt-euseȕİs ?

DEM.SG.M

through

DEF.SG.F.1PL-power

or

DEF.SG.M.1PL-piety

‘For what reason do you gaze at us as if we have done this (thing) out of our (own) power or piety?’ (Acts 3, 12) Topicalized constituents may either precede or follow fronted focus/whphrases. (11) a. anon de etȕe ne.n-noȕe we

PCL

for

mar-‹n-opt-‹n

DEF.PL.1PL-sin OPT-1PL-lock.up-1PL

mawaa-n SELF-1PL

‘(As for) us, BECAUSE OF OUR SINS, let’s lock ourselves up!’ (AP Chaîne no.41, 8:28–29) b. awo n-aš and

‹n-he

in-what of-manner

anon

t‹n-snt‹m

p-wa

we

(PRES)1PL-hear

DEF.SG.M-one

p-wa

hrai

h‹n

te.f-ape

DEF.SG.M-one

PCL

in

DEF.SG.F.3SG.language

‘And how do we hear each one (of us) in his native tongue?’ (Acts 2, 8) The availability of two topic positions can be accommodated straightforwardly under Rizzi’s (1997) left-periphery analysis, according to which a non-recursive ‘Focus Phrase’ (FOCP) is embedded into a topic field. Since preposed time and location adverbs as well as free-standing pronouns typically appear to the right of fronted focus/wh-phrases, Green & Reintges (2004a, 2005) propose that such ‘lower’ topics are base-generated in the specifier position of the FINP. The relative ordering of complementizers, topic constituents, and fronted focus/wh-phrases motivates the following hierarchical structure of the Coptic clause.

Coptic relative tenses: The profile of a morpho-syntactic flagging device 195

(12)

The fine structure of the Coptic left periphery CP 3 TOPP C0 3 TOPIC TOPP 3 FOCP Top0 3 FOCUS/WH-XP FOCP 3 FOC0 FINP 3 FINP TOPIC 3 FIN0 MODP TAM 3 SU MODP ru VP MOD0 TAM

If we assume that moved wh-phrases or focus constituents compete for the same position Spec, FOCP, a number of gaps in the Coptic data receive a principled explanation. Thus, neither multiple focus fronting nor a combination of wh-movement and focus-fronting is attested.

4.

The morpho-syntax of relative TAMs

The focus of this section is the morphological derivation and the syntactic positioning of Coptic TAMs. This special morphological pattern is derived by adding a relative complementizer to the TAM marker, which is independent of the verb. Despite their categorical status, relative markers appear in Foc0 rather than in C0. 6

196

Chris H. Reintges

4.1. A paradigmatic split in the Coptic tense-aspect-mood system Coptic Egyptian has a complex inflectional system, which comprises more than twenty different verbal tenses. At the foundation of this richness of morpho-semantic distinctions is the division of the four absolute tenses (the Present, the Habitual, the Future, and the Perfect) into a general and a relative class, cf. table 1. (The verb nau ‘to see’ has been chosen to illustrate a typical paradigm: a-f-nau ‘he has heard’.) Table 1. The inflectional paradigms of Coptic affirmative tenses PRESENT

general 1SG 2SG.M 2SG.F 3SG.M 3SG.F 1PL 2PL 3PL

ti-nau k‹-nau te-nau f‹-nau s‹-nau t‹n-nau tet‹n-nau se-nau

PERFECT

relative

general

e-i-nau e-k-nau e-te-nau e-f-nau e-s-nau e-t‹n-nau e-tet‹n-nau e-u-nau

1SG 2SG.M 2SG.F 3SG.M 3SG.F 1PL 2PL 3PL

relative

general

e-i-na-nau e-k-na-nau e-te-na-nau e-f-na-nau e-s-na-nau e-t‹n-na-nau e-tet‹n-na-nau e-u-na-nau

1SG 2SG.M 2SG.F 3SG.M 3SG.F 1PL 2PL 3PL

FUTURE

general 1SG 2SG.M 2SG.F 3SG.M 3SG.F 1PL 2PL 3PL

ti-na-nau k‹-na-nau te-na-nau f‹-na-nau s‹-na-nau t‹n-na-nau tet‹n-na-nau se-na-nau

relative a-i-nau a-k-nau are-nau a-f-nau a-s-nau a-n-nau a-t‹n-nau a-u-nau

‹nt-a-i-nau ‹nt-a-k-nau ‹nt-are-nau ‹nt-a-f-nau ‹nt-a-s-nau ‹nt-a-n-nau ‹nt-a-t‹n-nau ‹nt-a-u-nau

HABITUAL

relative ša-i-nau ša-k-nau šare-nau ša-f-nau ša-s-nau ša-n-nau ša-t‹n-nau ša-u-nau

e-ša-i-nau e-ša-k-nau e-šare-nau e-ša-f-nau e-ša-s-nau e-ša-n-nau e-ša-t‹n-nau e-ša-u-nau

There is a clear morphological relationship between the two sets of absolute tenses: relative tenses are derived from general tenses by adding a relative marker e– or ‹nt– in front of the TAM expression (a null morpheme in the present tense). The system of relative TAM formation is organized around a marked value represented by the Relative Perfect marker ‹nt– and the ‘elsewhere’ form e–. Besides the affirmative relative forms listed in table 1, Coptic has two modal forms, the Conditional conjugation e-f-šannau ‘if/when he sees’ and the Third Future e-f-e-nau ‘he shall see’, which both show the relative marker e– but lack the corresponding non-relative form.

Coptic relative tenses: The profile of a morpho-syntactic flagging device 197

(13)

awn on ne-snİu [e-u-šan-amelei] PCL DEF.PL-brothers REL-3PL-COND-neglect and e-k-e-nehse ‹mmn-u h‹m p‹-nomos REL(-PRES)2SG.M-PREP-wake PREP-3PL in DEF.SG.M-law e-p‹-nute to-DEF.SG.M-god ‘And also (concerning) the brothers, if they become negligent, you shall raise them up in the law of God.’ (V. Pachôm. 94, 14–16)

Since Coptic makes productive use of relative TAMs in the modal system, these forms can even be described as the predominant system. The alternations in the shape of the relative TAM marker provide prima facie evidence for the correlation between special inflection and tense/finiteness. 4.2. The syntactic position of relative TAMs Coptic relative clauses are externally headed and contain no relative pronoun or other subordinator besides the relative TAM marker. In expressing the core functions of relativization (attribution, subordination), such relative markers can be classified as [+finite] relative complementizers (see de Vries 2002, chap. 5). The relative marker itself does not encode any nominal-functional features of the relative head; all these features are present on a resumptive pronoun in the embedded clause. (14)

[e-ša-u-ȕi n-toot-ui nei-esowi DEM.PL-sheep REL-HAB-3PL-take from-hand-3PL m-pe.ui-šos] PREP-DEF.SG.M.3PL-shepherd ‘(like) sheep from which their shepherd is taken away’ (V. Pachôm 92, 30–93, 1)

It is possible to combine the relative marker e– with the relative complementizer et–, as seen in (15). When preceding a nominal subject, the marker e– assumes a longer form ere– (Reintges 2004, 414ff. §11.1.2). (15)

p-İii

[et-ere

pei-šİre

DEF.SG.M-house COMPREL-REL DEM.SG.M-boy

šİm mowt nhİt-fi] little die inside-3SG.M

‘the house where this little boy died’ (Ac. A&P 206, 163–164)

198

Chris H. Reintges

The multiple occurrence of such relative elements shows that Coptic relative clauses have an articulated left-periphery. While the initial complementizer et– is merged into the C0-node, the default relative marker e– / ere– occupies a lower functional position which can be identified with the head of the focus projection. Further empirical support for a lower leftperipheral position of relative TAM markers comes from clitic leftdislocation structures like (17) where the topicalized pronoun anok ‘I’ precedes the relative marker e– (cf. Cinque 1990). (16)

p-mai

[ anok e-ti-na-ȕnk I REL-1SG-AUX-go ‘the place where I am going to’ (John 8, 21) DEF.SG.M-place

ero-fi ] to-3SG.M

Coptic relative tenses display a considerable number of context-dependent alternations in the form of the relative marker. Reintges (2003, 400f.) views these complementizer alternations as a morphological reflex of the movement and the incorporation of the TAM auxiliary into the relative marker in Foc0. This is shown in tree diagram (17). (OP stands for the abstract relative operator.) (17)

The left periphery of Coptic relative clauses CP 2 OP CP 2 FOCP C0 et– 2 FOCP 2 FOC0 FINP ere– 2 MODP FIN0 TAM 2 SU MODP 2 T0-FOC0-movement MOD0 VP

Coptic relative tenses: The profile of a morpho-syntactic flagging device 199

T(ense)0-to-FOC(us)0 movement seems to be a corollary of T0-to-C0 movement (Rizzi 1996, Pesetsky & Torrego 2001). Since the morphological effects of incorporation surface on the FOC0 head, it acts as the attracting category or probe, presumably because it has unvalued tense/finiteness features that need to be checked. 7

5.

Distribution across sentence types

This section surveys the non-relative environments in Coptic where relative TAMs are allowed and/or grammatically required. The presence of relative TAMs is obligatory in wh-in-situ questions as well as a range of declarative focus constructions. In yes-no questions, on the other hand, the presence of relative TAMs seems to be optional. 5.1. Wh-in-situ questions In Cheng’s (1991) typological framework, Coptic can be classified as an optional fronting language, in which wh-fronting and wh-clefting are available as marked alternatives to the canonical wh-in-situ pattern. Wh-in-situ has a broad syntactic distribution, appearing in main and embedded clauses. Moreover, neither wh-arguments nor wh-adjuncts show any resistance to wh-in-situ. (18) a. Wh-in-situ object question na-k ? e-i-na-þe u REL(-FUT)-1SG-AUX-say what to-2SG.M ‘What shall I say to you?’ (AP Chaîne no. 28, 5:25) b. Wh-in-situ prepositional object question e-k-nİu tnn, pa-son ? REL(-PRES)-2SG.M-come where DEF.SG.M.1SG-brother ‘Where did you come from, my brother?’ (Ac. A&P 198:64–65) c. Wh-in-situ adverb question ‹n-he ? ‹nt-a-k-ei e-pei-ma ‹n-aš REL-PERF-2SG.M-go to-DEM.SG.M-place in-what of-manner ‘How did you get to this place?’ (B. Martyr. 206, 29) Wh-in-situ is not restricted to matrix questions but can also occur in embedded questions. When the relative marker surfaces to the right of the

200

Chris H. Reintges

subordinating complementizer þe ‘that’, the wh-in-situ phrase takes the embedded scope and the resulting interpretation is that of an indirect question. (19)

n-ti-sow‹n

an [ þe

nt-a

NEG-(PRES-)1SG-know NOT COMP REL-PERF

u

šnpe

mmo-s ]

what happen to-3SG.F

‘I don’t know what happened to her.’ (Hilaria 7, 30–31) As we can see from (20), wh-in-situ questions with relative TAMs can be further modified by an interrogative particle like eye. (20)

nuþe eȕol h‹n nim ? eye ere ne.t‹n-šİre Q REL (PRES-)DEF.P.2P-son cast PCL in who ‘In whom are your sons casting out (demons)?’ (Luke 11, 19)

Coptic question particles are compatible with wh-in-situ, wh-fronting, as well as wh-cleft constructions. In view of their tolerance towards the surface position of the wh-phrase, they do not seem to be involved in any kind of licensing relationships. In conveying positive or negative presuppositions as well as corroborative focus readings, such interrogative particles are closely connected to the modal domain (Reintges 2003, 378ff.). 5.2. Declarative focus constructions The syntactic derivation of declarative focus structures runs entirely parallel to that of wh-questions, with the clause-internal placement of the focus phrase conditioning relative TAMs. In the longstanding research tradition on focus, question-answer sequences have been used as diagnostic tools for identifying the new information focus within a clause. The focus corresponds to the sentence element in the response that provides a value for the variable introduced by the wh-constituent in the question while the presupposition underlying the question and the corresponding answer remains constant (Chomsky 1971, Jackendoff 1972, Erteshik-Shir 1986). (21) Q: e-r-ȕİk e-tnn ? REL-(PRES)-2SG.F-goSTAT to-where ‘Where are you (woman) going to?’

Coptic relative tenses: The profile of a morpho-syntactic flagging device 201

A: e-i-ȕİk e-p‹-topos REL(-PRES)-1SG-goSTAT to-DEF.SG.M.-shrine of ‹nta-šlİl. ‹n Apa Mİna Apa Mêna CONJ.1SG-pray ‘I am on my way TO THE SHRINE OF APA MÊNA to pray.’ (Mêna, Mirc. 27b:22–25) As Rooth (1992, 84) puts it, focus in an answer evokes a set of alternatives that qualify as potential answers in the context of the question. In doing so, it marks a contrast between the asserted answer and other potential answers (in our example, other holy places to go to). The same situation applies to exhaustive listing focus, which specifies an exhaustive set of which the proposition holds true and excludes other possibilities (Kuno 1972, É. Kiss 1998). Its prototypical syntactic frame in Coptic is the ‘not X but Y’ construction which provides a straightforward way of rejecting a previous utterance and offering an alternative specification of the variable (Horn 1989, Erteschik-Shir 1997, Herburger 2000). (22)

‹mpnr

pa-šİre,

no

DEF.SG.M.1SG-son NEG.PERF-3PL-appoint-2SG.M

e-ti-oikonomia,

‹mp-u-toš-‹k alla ‹nt-a

to-DEM.SG.F-service but

e-u-sols‹l

gar

p‹-þois

REL-PERF DEF.SG.M-lord

‹n-ne-snİu

to-INDEF.SG-comfort for-DEF.PL-brothers

PCL

toš-‹k appoint-2SG.M

et-waaȕ (…) COMPREL-be.holySTAT

‘No, my son, for you have not been destined (lit. they have not destined you) for this career (as a hermit), but the Lord has appointed you AS A COMFORT FOR THE HOLY BROTHERS (…)’ (B. Martyr. 216, 33–217, 1) Relative TAMs are found in presentational focus contexts, in which a new discourse referent is introduced “out of the blue”. In including only information that is not construable from the previous context, presentational focus differs systematically from the new information focus in the response to a preceding question (Green & Reintges 2005). 8 (23)

e-w‹na-f-s

h‹n

n-et-ka

ma

REL-HAVE-3SG.M-3SG.F

among

DEF.PL-COMPREL-put

place for-3SG.M

na-f

‘He has it (a place) among THOSE WHO MAKE A PLACE FOR HIM.’ (Shenoute III 85:14)

202

Chris H. Reintges

Relative TAMs may also be used as narrative tenses, indicating that a particular event is of special relevance for the subsequent discourse (Hopper 1979, Reinhart 1984). Such foregrounded event chains typically respond to cause/reason questions that require a discourse as a felicitous answer. In the following discourse fragment, the relative TAM marker ere appears in a present tense sentence with a stative verb form (kyal2w ‘to be entrusted’), which is clearly atelic (Reintges 2004, 216f. §6.2.3, 259-60 §7.3.2.2). The selection of a relative TAM in this context seems to be motivated by the desire to single out the situation that is crucial for the development of the narrative. (24) Q: ahro-s why-3SG.F

tei-h‹ll2

[ e-s-rime

n-tei-he ] ?

DEM.SG.F-old.woman REL(-PRES)-3SF-weep in-DEM.SG.F-way

‘What about this old woman who is weeping in such manner?’ A: eti ere pe.s-hai onah ere ADV

REL

(PRES-)DEF.SG.M-3SG.F-husband

kyal2w

h‹n-‹nka

ero-f

be.alive

‹nte

(PRES-) INDEF.PL-thing enthrustSTAT to-3SG.M of

REL

u-r2me INDEF.SG-man

a-f-mu

de

h‹n-u-š‹p-en-š2p

‹nweš ‹n-šaþe (…)

PERF-3SG.M-die

PCL

suddenly

without of-word

‘What about her, this old woman, who is weeping in such a manner? When her husband was still alive, someone’s things were entrusted to him. But he died all of a sudden without a word (…)’ (AP Chaîne no. 225, 65:16–18) The generalization that emerges from the previous discussion is that in-situ focus in Coptic Egyptian is compatible with argument, predicate, and sentence focalization, with the obtained new information or contrastive focus interpretation being governed by the discourse context (see Green & Jaggar 2003 for parallel facts in Hausa). 5.3. Interrogative focus constructions Coptic yes-no questions fall into two groups: those which are introduced by interrogative particles and those that lack any morphological marking of interrogative force. Both types of yes/no questions may condition relative tenses, as seen in (25a–b).

Coptic relative tenses: The profile of a morpho-syntactic flagging device 203

nsn-i e-i-na-ȕnk (25) a. e-k-na-kynšt REL(-FUT)-2SG.M-AUX-look after-1SG REL(-FUT)-1SG-AUX-go e-p-tako ? mpnr, pa-þoeis! to-DEF.SG.M-perdition NO DEF.SG.M.1SG-lord ‘Would you watch me rushing to perdition? Oh no, my Lord!’ (Ac. A&P 200, 90–91) b. mİ e-k-na-kaa-t nsn-k REL(-FUT)-2SG.M-AUX-put-1SG behind-2SG.M Q etȕe u-hnȕ mmate ? for INDEF.SG-thing only ‘Will you abandon me (lit. put me behind you) BECAUSE OF ONE THING ONLY?’ (Ac. A&P 200, 81–82) If relative tenses were connected to interrogative force, their broad syntactic distribution across non-interrogative sentence patterns (relative clauses, declarative focus constructions) would be left unexplained. For this reason, it seems more promising to relate their presence in yes-no questions to focus rather than to clause-typing. In (25b), for instance, the exhaustive listing interpretation of the cause adverbial etȕe u-hnȕ ‘because of one thing’ is brought about by the focus-sensitive particle mmate ‘only’ (Herburger 2000, 105ff.). Alternative questions entail a choice between two members of a disjunctive set that contradict each other. The selection of one member will therefore automatically eliminate the alternative option (Erteschik-Shir 1997). As we can see from (26), the contrastive focus reading of alternative yes-no questions may condition the presence of relative TAMs. (26)

p‹-þoeis

e-k-tšn

DEF.SG.M-lord REL(-PRES)-2SG.M-say

na-n

‹n-tei-paraȕolİ

to-1PL

PREP-DEM.SG.F-parable

þen

e-k-tšn

‹mmo-s

e-won nim ?

or

REL(-PRES)-2SG.M-say

PREP-3SG.F

to-one every

‘Oh Lord, do you tell this parable TO US, or do you tell it TO EVERYONE?’ (Luke 12, 41) What we seem to be dealing with is a case of apparent optionality: the relevant trigger for relative TAMs in yes-no questions seems to be the presence of focus.

204 6.

Chris H. Reintges

The “hidden” movement configurationality of Coptic wh-in-situ

In contrast to Hausa relative aspects and Chamorro wh-agreement, Coptic relative tenses are restricted to focus/wh-in-situ contexts. Elaborating on Reintges, LeSourd & Chung’s parametric analysis of special inflection (2006), I will argue in what follows that Coptic wh-in-situ instantiate a “hidden” movement configuration where syntactically invisible whmovement leaves a footprint in the relative TAM marking. The main evidence for a movement approach to Coptic wh-in-situ comes from the parallelism in scope and interpretation between the wh-in-situ and the whfronting pattern. 6.1. Wh-movement and interrogative interpretation The construal of a genuine question interpretation of wh-items supports a movement analysis of Coptic wh-in-situ. Consider the following triplets of examples which involve the wh-words nim ‘who’ and u ‘what’. In the aexamples, the wh-word remains in-situ and relative marking surfaces in the left periphery. As we can see from comparison with the b-examples, the wh-item is interpreted as a question word in exactly the same way as its fronted counterpart. This contrasts with the c-examples in which the TAM auxiliary assumes the non-relative form and the wh-in-situ phrase is interpreted as a specific indefinite. Such an indefinite interpretation is not available for fronted wh-phrases (Reintges 2003, 380ff., 2004, 144ff.§ 4.2.1.). na-f n-tei-hypomenİ ? (27) a. nt-a nim þpo REL-PERF who achieve for-3SG.M PREP-DEM.SG.F-endurance ‘Who has achieved for himself such endurance?’ (Hilaria 12, 29) a-f-ent-k e-pei-ma ? b. nim who PERF-3SG.M-bring-2SG.M to-DEM.SG.M-place ‘Who brought you here?’ (KHML I 3:7–8) c. ša-u-eime [ þe a nim p-šİre HAB-3PL-know COMP PERF who DEF.SG.M-child þrn h‹m p-agnn ] ‹n-nim DEF.SG.M-contest of-who win in ‘They know that so-and-so, the son of so-and-so, won the contest.’ (B. Apoc. 134, 14–15)

Coptic relative tenses: The profile of a morpho-syntactic flagging device 205

na-k ? (28) a. e-i-na-ti u REL(-FUT)-1SG-AUX-give what to-2SG.M ‘What shall I give you?’ (Genesis 30, 31) b. u se-þi wa ero-i h‹m what (PRES-)3PL-speak malice against-1SG in p-e-ti-š‹p hmot anok haro-f ? DEF.SG.M-COMPREL(-PRES-)1SG-take grace I for-3SG.M ‘What could they say maliciously against me because of what I give thanks (lit. take grace) for?’ (1 Cor. 10, 30) c. a-i-ti u mn u ehun e-pei-ma. PERF-1SG-give what and what PCL to-DEM.SG.M-place ‘I gave such and such thing to this place.’(Shenoute IV 105, 16) Since both wh-fronting structures and relative-marked wh-in-situ constructions are associated with an interrogative interpretation, there is compelling evidence for equating the co-occurrence of wh-in-situ and relative tenses with a movement configuration. By contrast, the specific indefinite interpretation of in-situ wh-words instantiates the non-movement option. 6.2. Wh-movement and scope The scope of an overtly moved wh-phrase is contingent on its landing site. When a wh-phrase moves overtly to the embedded focus position to the right of the subordinator þe ‘that’, as in (29a), it takes embedded scope and the entire construction is interpreted as an indirect question. If, on the other hand, the wh-phrase undergoes long-distance wh-movement across a clause boundary, as in (29b), it takes matrix scope and the resulting interpretation is that of a direct question. 9 (29) a. ti-þ‹nu ‹mmn-t‹n [þe h‹n u ‹n-šaþe (PRES).1SG-ask PREP-2PL COMP with what of-word a-tent‹n-mute ero-i …] PERF-2PL-say about-1SG ‘I ask you with what reason do you say about me...’(Acts 10, 29) ‹n-he ‹ntok k‹-tšn ‹mmo-s b. ‹n-aš in-which of-manner you.SG.M (PRES-)2SG.M-say PREP-3SG.F [þe tet(n)-na-‹r r‹mhe] ? COMP (FUT-)2PL-AUX-make free.man ‘How do you say that you will become free?’ (John 8, 33)

206

Chris H. Reintges

Parallel facts can be observed for embedded wh-in-situ questions. In finite complement clauses, the wh-in-situ constituent generally takes embedded scope, which yields an indirect question interpretation. The relative marker appears after the complementizer þe ‘that’. (30)

‹mpe-f-þoo-s

[ þe

‹nt-a-f-kaa-f

tnn ]

NEG.PERF-3SG.M-say-3SG.F COMP REL-PERF-3SG.M-put-3SG.M

where

‘He did not tell where he had put it.’ (AP Chaîne no 235, 65:18) However, examples are also attested where the in-situ wh-phrase scopes out of the embedded finite clause and the entire construction is interpreted as a direct question. In this context, the relative TAM marker surfaces at the left edge of the higher clause. (31)

eye ntntn e-tetn-þn mmo-s

ero-i

Q

about-1SG

you.PL

REL-2PL-sayPREP-3SG.F

[ þe COMP

ang nim]? I

who

‘Who are you saying of me that I am?’ (Mark 8, 29) The matrix scope of embedded wh-in-situ and the presence of relative TAMs in the matrix clause indicate that successive-cyclic movement through the specifiers of CP has taken place. The Coptic facts therefore provide counterevidence to Rackowski & Richards’ (2005) claim that morphosyntactic flagging reflects an AGREE relation within the verbal (vP) domain from which the wh-phrase has been extracted. A different situation obtains in infinitival wh-in-situ questions, in which only the direct question interpretation seems to be available. Once again, relative TAMs flag the highest clause over which the wh-in-situ takes scope. Since they lack TAM markers, Coptic infinitival clauses do not project an articulated left periphery, including the FINP and the topic-focus field. Since there is no designated scope position inside the infinitival clause, the embedded wh-in-situ phrase must move to the specifier position of the matrix FOCP, where its scope is determined (McDaniel 1989, Dayal 1991). (32) a. e-tet‹n-wnš [ e-tra-ka nim nİ-t‹n REL(-PRES)-2PL-want to-CAUS.INF.1SG-place who for-2PL eȕol h‹m pe-snau ] ? PCL from DEF.SG.M-two ‘Who do you want me to release to you out of the two?’ (Matthew 27, 21)

Coptic relative tenses: The profile of a morpho-syntactic flagging device 207

b. e-k-wnš [ e-tre-n-snȕte na-k REL(-PRES)-3SG.M-want to-CAUS.INF-1PL-prepare for-2SG.M tnn ‹m-p‹-paskha e-wom-f ] ? where PREP-DEF.SG.M-Paskha.meal to-eat.INF-3SG.M ‘Where do you want us to prepare the Pasha meal for you to eat?’ (Matthew 26, 17) To conclude, the absence of clear semantic contrasts between wh-in-situ and wh-fronting questions suggests that both patterns are derived by the same movement operation. In wh-in-situ questions, this movement process is syntactically invisible since the wh-phrase is apparently frozen in place. Yet, covert wh-movement feeds overt morphology, triggering the spell-out of relative TAMs. 6.3. The syntactic derivation of wh-in-situ and wh-fronting structures The previous discussion on scope and interrogativity raises a non-trivial question concerning the representational level at which covert whmovement is operative: does it take place at narrow syntax or at LF? Assuming with Chomsky (2005, 16ff.) that LF operations have no access to the morpho-phonological cycle, the presence of relative TAMs in Coptic wh-in-situ questions clearly shows that covert movement operation takes place in the narrow syntax before the derivation is shipped to PF. Adopting the copy theory of movement, Reintges, LeSourd & Chung (2006) derive the word order contrast between wh-in-situ and wh-fronting constructions from different pronunciation sites of the scope-taking operator: the phonological realization of the overtly moved wh-phrase in the target position (Spec,FOCP) produces wh-fronting structures while the phonological realization of the relevant operator in the original position gives rise to the hidden movement configurationality of wh-in-situ. Moreover, the semantic non-distinctness of the wh-in-situ and the wh-ex-situ patterns receives an immediate explanation since the spell-out of the moved whphrase is a matter of the morpho-phonological rather than the semantic component. 10 In current Minimalist views on movement, in order for one category (the goal) to move to the vicinity of another (the probe), the two must stand in an agreement relation through which features on maximal projections and features on heads are matched. Both the probe and the goal have active (that is, uninterpretable or unvalued) features whose values can be set by

208

Chris H. Reintges

the corresponding features of the other. Subsequently, movement is implemented to bring the goal into a local configuration with the probe, an operation that leaves the original copy of the goal untouched. Once the probe and the goal are in a local configuration within a single projection, the uninterpretable features on both categories are erased (Chomsky 2000, 122ff., 2001, 4f.). If this theory is applied to the case at hand, the probe for wh-movement is the left-peripheral focus head and the goal is the wh-phrase. With respect to their feature specification, I assume, following Chomsky (2000, 128), that wh-phrases have an uninterpretable feature [uwh] and an interpretable interrogative feature [Q] that matches the uninterpretable interrogative feature [uQ] on the focus head. 11 It may very well be the case that the focus head inherits its uninterpretable [uQ] feature from the force-indicating C0 head (Chomsky 2005, 9). If the uninterpretable wh-feature on the in-situ wh-phrase cannot be eliminated via movement to the designated scope position, it can no longer receive an interrogative interpretation. As a result, such wh-in-situ items in Coptic can only receive a specific indefinite interpretation. Movement of the wh-phrase to the specifier position of the focus phrase yields the proper checking configuration. If the lowest copy of the whphrase is pronounced, then the relative marker is spelled out overtly, as schematically represented in diagram (33). (Strikethrough indicates material that is left unpronounced. The relative marker on the TAM word is represented as REL. Arrows indicate the feature-matching relationship between the focus head and tense.) (33)

Covert wh-movement with relative tense marking [ C0 [FOCP WH [ FOC REL ][TP TPERF [MODP SU [MOD [VP SU] [VP V WH]]]]

feature inheritance

T0-to-Foc0 movement

If, on the other hand, the highest copy of the wh-phrase is pronounced, then the relative marking, which would occur in the same projection, is left unpronounced, as schematically represented in diagram (34).

Coptic relative tenses: The profile of a morpho-syntactic flagging device 209

(34)

Overt wh-movement without relative tense marking [ C0 [FOCP WH [FOC REL] [TP TPERF [MODP SU [MOD [VP SU] [VP V WH ]]]]]

feature inheritance The analysis proposed here departs from Chomsky’s (2001) “Derivation by Phase” model, according to which wh-movement proceeds successivecyclically through different derivational domains (or phases), including the verb phrase with full argument structure and the CP. Phases are domains that are eligible for phonological realization, which means that a constituent has to target the left edge of each phase in order to undergo subsequent movement. There is, however, no independent evidence in Coptic Egyptian for the phasal status of the vP, as argued in Rackowski & Richards (2005). In lacking tense and finiteness, Coptic verbs seem to be defective categories. All these features are morphologically manifest on the TAM word and thus define opaque domains for core syntactic processes like case and extraction (cf. Reintges 2001). Complementizer alternations provide strong evidence for a feature matching agreement relation between the left-peripheral FOC0 head, which acts as a probe for wh-movement, and finite tense. It is clear that the probe enters into some agreement relationship not only with the goal (the wh-phrase) but also with the head of FINP which is productively involved in the formation of wh-dependencies (cf. Chomsky 2004, 116).

7.

Spellout conditions on special inflection

In the canonical wh-in-situ pattern, the relative TAM overtly marks a local specifier-head relationship between the designated functional head and the topmost copy of the displaced wh-phrase which is not phonologically realized. The question that arises is why relative TAMs are systematically absent in wh-fronting structures where the topmost copy of the moved whphrase is phonologically realized. In this section, I propose an economy explanation for the complementary distribution between wh-fronting and relative TAMs, which registers the presence of operator-variable dependen-

210

Chris H. Reintges

cies. I will then briefly discuss Tuller’s (1986) specificity filter for Hausa relative aspect inflection. 12 7.1. An economy condition on morpho-syntactic flagging Adopting the late insertion view of Distributed Morphology, I assume that the syntactic component is essentially phonology-free (Halle & Marantz 1993). The core task of the syntax is to create structural dependencies through syntactic operations while spell-out operations are confined to the morpho-phonology (Bobaljik 2002). Spell-out operations fall into two classes, namely (i) the phonetic realization of copies in the course of lexical insertion, and (ii) the conversion of a hierarchical structure into linear order. The decision which copy to pronounce may be dictated by languagespecific constraints but may also fall out from the language’s parameter setting. In short, nothing is deleted, but not everything is pronounced. On this view, two issues concerning phonological realization of whmovement structures are involved: firstly, whether the higher or the lower copy of a moved wh-phrase is pronounced, and secondly, which of the copies triggers the morphological effects of special inflection. In Reintges, LeSourd & Chung’s (2006, 183f.) analysis, the relative marker is inserted into the structure roughly at the point when the moved wh-phrase is pronounced. In other words, in Coptic, the spell-out of wh-movement structures is governed by two language-specific morphological requirements. On the one hand, the creation of operator-variable dependencies in the narrow syntax must be marked overtly by the pronunciation of the topmost copy or by morphological flagging. On the other hand, there is an economy condition preventing the over-coding or double marking of such dependencies by the simultaneous spell-out of both the topmost wh-copy and relative TAM marking. This situation is reminiscent of the Doubly-filled COMP Filter in English since either the head or the specifier of the target projection of whmovement can be realized, but not both (Chomsky & Lasnik 1977). Alternatively, one might think of an economy condition of wh-scope marking where the scope of a moved wh-phrase is either marked overtly by the pronunciation of the attracted category in the target position or by inserting a morphological scope marker into the attracting Foc0 head.

Coptic relative tenses: The profile of a morpho-syntactic flagging device 211

7.2. A specificity filter for Hausa relative aspects? The economy condition of morpho-syntactic flagging does not carry over to Hausa, where wh-fronting requires relative aspect marking, while this is prohibited in the corresponding Coptic structures. Conversely, relative aspect marking is blocked in Hausa in non-echo wh-in-situ questions but obligatory in Coptic. Thus, Hausa is the “mirror image” of Coptic with respect to the syntactic conditions on special inflection (Green & Reintges 2004a, 2005, Reintges & Green 2004). (35) a. Wh-fronting with relative aspects sukà jƝ ? ìnƗ where 3PL-REL-PERF go ‘Where have they gone?’ (Jaggar 2001, 517) b. Wh-in-situ with general aspects sunƗ̖ fitôwa dàgà ìnƗ ? 3PL-IMPERF come.out.VN from where ‘Where are they coming out from?’ (Newman 2002, 496) In Hausa, overt wh-movement feeds overt morphology while covert movement leaves no morphological footprint. The prohibition against relative aspects in wh-in-situ questions can be accounted for if wh-in-situ constituents move to their target position at LF, which has no direct access to the morphological operations. To exclude the combination of moved wh-phrases and non-relative aspects, Tuller (1986, 73 and 109ff.) postulates an interpretative filter for Hausa which prohibits [–definite] heads from being specified by the feature [focus]. In her system, operators in the specifier position of CP are assigned focus. The relative aspect qualifies as a [+definite] head, witnessed by the narrative use of the relative perfective to describe specific events in the past. As a [+definite] head, the relative aspect in T0 is compatible with the focus features associated with the displaced wh-phrase in Spec,CP. The Coptic facts are problematic for Tuller’s analysis. As we have seen in section 5.2, relative TAMs may be used as narrative tenses in much the same way as Hausa relative aspects and would therefore be eligible as a [+definite] head. Nevertheless, it is impossible to use this special inflection in Coptic focus/wh-fronting contexts. This suggests that special inflection in both languages is not controlled by semantically driven filters but rather functions as a morpho-syntactic flagging device for operator-variable de-

212

Chris H. Reintges

pendencies. As such, relative TAMs are subject to language-particular spellout conditions.

8.

Conclusions

Coptic provides yet another example of a language in which classical whconstructions (relative clauses, wh-questions, declarative focusing constructions) are flagged by relative tense-aspect-mood categories. The relative marker could be identified with a finite relative complementizer, which is inserted into the head of a designated focus phrase in the clausal leftperiphery. The presence of operator-variable dependencies provides a necessary, but not sufficient condition for the presence of morpho-syntactic flagging. The presence of relative TAMs is required in wh-in-situ questions while they are absent in wh-fronting structures. Coptic hosts either the fronted wh-phrase in the specifier position or the relative marker in the head position of the left-peripheral focus phrase, but these must not cooccur within the same projection. Coptic wh-in-situ questions display the typical characteristics of whmovement as far as scope and interrogative readings are concerned. The hidden movement configuration of wh-in-situ has been accounted for in terms of covert syntactic movement and the phonological realization of the moved wh-phrase in the base position. Syntactically invisible whmovement feeds overt morphology in the form of relative TAM marking. From this perspective, the difference between ex-situ and in-situ constructions reduces to the question of where the moved wh-phrase is phonologically realized, viz. in the base position or in the target position of movement. Since Coptic wh-in-situ is derived by wh-movement followed by the pronunciation of the lowest wh-copy, the presence of relative tenses as the morphology of extraction falls into place.

Notes

Many thanks to Victor Manfredi and Malte Zimmermann for detailed written comments on a previous version of the present paper. For discussion on the Coptic facts, I am indebted to Lisa Cheng, Sandy Chung, Hamida Demirdache, Melanie Green, Phil Jaggar, and Lutz Marten. I am also grateful to Jeroen van de Weijer for the correction of my English. The usual disclaimers apply. The research reported here is funded by the

Coptic relative tenses: The profile of a morpho-syntactic flagging device 213

1.

2.

3.

4.

Dutch Organization for Scientific Research N.W.O. (Vidi-grant 276-70008). The following abbreviations are used in the glosses: 1 ‘first person’; 2 ‘second person’; 3 ‘third person’; AGR ‘agreement’; AUX ‘auxiliary verb’; CAUS.INF ‘causative infinitive’; COMP ‘subordinating complementizer’; COMPQ ‘interrogative complementizer’; COMPREL‘relative complementizer’; COND ‘conditional’; CONJ ‘conjunctive’; FUT ‘future’; IMPERF ‘imperfective’; INF ‘infinitive’; INTERJ ‘interjection’; IOM ‘indirect object marker’; NOM ‘nominative case’; OBJ ‘objective case’; OBL ‘oblique case’; PCL ‘particle’; PERF ‘perfect’; PL ‘plural’; PREP ‘preposition’; PRES ‘present tense’; Q ‘yes/no question particle’; REL ‘relative marker’; SG ‘singular’; SUBJ ‘subjunctive’; VN ‘verbal noun’; WH ‘wh-agreement’. Glosses are given in parentheses for morphemes that have no surface-segmental shape. Coptic Egyptian is the indigenous language of late-antique and early medieval Christian Egypt (from about the 3rd to the 11th century CE) and represents the most recent stage of the Ancient Egyptian language. Coptic is actually a dialect cluster with at least six regional varieties, two of which gained supra-regional importance: Sahidic Coptic, the vernacular of Upper Egypt, and Bohairic Coptic, the vernacular of Lower Egypt. The data in this study are exclusively taken from Sahidic Coptic, which, due to its early records and rich literature represents the main reference dialect. Jaggar (2001, 162 fn. 4,5) suggests that the term “relative” be replaced with the cover term “focus” to provide a semantic characterization of the environments in which relative aspects are grammatically allowed and/or required. Layton (2000, 352–5) re-analyzes Coptic declarative and interrogative main clauses that contain relative TAMs (traditionally called “second tenses”) as focusing constructions, which does, however, not include relative constructions, temporal adverbial clauses, and secondary predicates. Although Jaggar’s reductionist approach is appealing from a conceptual point of view, it is not entirely clear or obvious whether a semantic ‘focus’ analysis carries over to relative clause constructions, where relative aspect marking is obligatory, or conditional and temporal adjunct clauses, where it is optional. It is worth pointing out that the information structure of this construction type has received alternative analyses in terms of topicalization (see Green & Reintges 2004b, 86f. for further discussion). According to McCloskey (1990), the phonological and morphological distinction between the alternating complementizers is sometimes not reflected in the phonological shape of the complementizer itself but rather in the initial phonological mutation induced on the following sentence element (typically a verb). The abbreviations aL and aN reflect the fact that the former complementizer induces lenition on the adjacent verb whereas

214

Chris H. Reintges

5.

6.

7.

8.

9.

the latter induces nasalization on the initial segment of a following verb. Complementizer alternations are subject to dialect variation (McCloskey 1990, 242–3, fn.7, 2001, 68, fn.2). Although this generalization holds for most Hausa speakers, Tuller (1986, 120) notes that some speakers permit relative aspect marking to surface on every inflectional head in the path of movement. This appears to represent a case of dialect variation since most speakers of Standard (Kano) Hausa do not produce such constructions (Green & Reintges 2005). Victor Manfredi proposes to treat the relative marker e– “not as a morpheme of any kind, but rather as an epenthetic mora motivated by phrasing (i.e. assuming prosodic conditioning, analogous to the destressing pattern of anaphora in English)”. In fact, the e– formative derives historically from the aspectual auxiliary jw ‘to come’, which performs various clausechaining and subordinating functions in Pre-Coptic Egyptian. Both the complementizer allomorphs ‹nt– and et– come from the relative complementizer ntj ‘that’, which is itself derived from the genitival linker nj ‘of’. See Polotsky (1944) and Loprieno (1995) for a more detailed discussion of the precursors of Coptic relative TAMs. A question remains with respect to the auxiliary verb na– ‘go’, which indicates future tense. I assume that the so-called First Future f-na-sot‹m ‘he is going to hear’ is actually a present progressive. On this view, the null morpheme of the present tense is incorporated into the relative marker e– in the Foc0-node. Indefinite subjects that correspond to presentational focus are generally not construed with relative TAMs. The reason for this is that indefinite subjects trigger the insertion of the copular verbs w‹n ‘be’ and m‹n ‘not be’ into the structure in the present and the future tense, which thus assume the form of existential sentences (Reintges 2004, 259 §7.3.2.1). (i) a. wen u-noky ‹n-r2me š2ne h‹m p‹-palation INDEF.SG-great of-man be.sick in DEF.SG.M-palace be ‘A nobleman in the palace who became sick.’ (Hilaria 10, 27) na-taho-u b. w‹n u-m‹nt.eȕeiİn be (FUT-)INDEF.SG-misery AUX-come.upon-3p ‘A misery will come upon them.’ (V. Pachôm. 90, 28–91, 1) The complementary distribution between copula support and special inflection in presentational subject focus contexts needs to be clarified in future research. Since Coptic is a dead language, it is impossible to establish that wh-in situ questions are sensitive to islands. However, there are no attested examples of wh-in-situ questions in which the wh-phrase originates inside an island and takes scope outside the island. In wh-fronting structures, the long-distance movement of a topical wh-phrase out of an adjunct clause is saved by resumptive pronominalization, as seen in (i).

Coptic relative tenses: The profile of a morpho-syntactic flagging device 215 nimi gar ‹mmn-t‹n [e-fi-weš-ket u-pyros] who PCL among-2PL REL(-PRES).3SG.M-want-build INDEF.SG-tower an ‹n-šor‹p mİ n-fi-na-hmoos Q neg(-fut)-3sg.m-aux-sit not at-first p-np ‹n-te.f-danapİ? n‹-fi-fi conj-3sg.m-carry def.sg.m-number of-def.sg.f.3sg.f-cost ‘Who among you, who intends to build a tower, would he not first sit and estimate the total amount of its costs?’ (Luke 1, 28) 10. The possibility of pronouncing the lowest, as opposed to the highest copy of a moved wh-phrase has been explored by Groat & O’Neil (1996), Pesetsky (1997, 1998, 2000), Bobaljik (2002), Nunes (2004), and various others. The phonetic realization of the lowest wh-copy is widely believed to be a last resort option, available only when the pronunciation of the highest chain link does not result in a convergent derivation at PF (Boškoviü 2002, Nunes 2004). The Coptic facts clearly show that lower copy pronunciation may instantiate a parameterized option of a language’s whinterrogative paradigm (Reintges, LeSourd & Chung 2006). 11. This feature composition implements the idea that wh-movement is a means of clause typing, which specifies a clause as a question. The feature specification of the attracting probe, however, is a matter of debate in the current literature. For instance, Cheng & Rooryck (2000) argue that the C0 node has an interpretable [Q] feature and an unvalued [wh] feature. 12. I am indebted to Malte Zimmermann for clarifying the issues concerning the spell-out conditions on special inflectional morphology addressed in this section. (i)

References Aboh, Enoch Oladé 2004 The morphosyntax of complement-head sequences. Clause structure and word order patterns in Kwa: Oxford studies in comparative syntax. Oxford, New York: Oxford University Press. Bobaljik, Jonathan D. 2002 A-chains at the PF-interface: copies and ‘covert’ movement. Natural Language & Linguistic Theory 20: 197–267. Boškoviü, Željko 2002 On multiple wh-fronting. Linguistic Inquiry 33: 351–383. Browning, Maggie 1987 Null operator constructions. Ph.D. dissertation. MIT, Cambridge, Mass.

216

Chris H. Reintges

Cheng, Lisa Lai-Shen 1991 On the typology of wh-questions. Ph.D. dissertation. MIT, Cambridge, Mass. Cheng, Lisa Lai-Shen, and Johan Rooryck 2000 Licensing wh-in-situ. Syntax 3: 1–19. Chomsky, Noam 1971 Deep structure, surface structure, and semantic interpretation. In Semantics: an interdisciplinary reader in philosophy, linguistics and psychology, Danny D. Steinberg and Leon Jakobovits (eds.), 183– 216. Cambridge: Cambridge University Press. 1977 On wh-movement. In Formal syntax, Peter W. Culicover, Thomas Wasow, and Adrian Akmajian (eds.), 71–132. New York: Academic Press. 2000 Minimalist inquiries: the framework. In Step by step: Essays on minimalist syntax in honor of Howard Lasnik, Roger Martin, David Michaels, and Juan Uriagereka (eds.), 89–155. Cambridge, Mass.: MIT Press. 2001 Derivation by Phase. In Ken Hale: A Life in Linguistics, Michael Kenstowicz (ed.), 1–52. Cambridge, Mass.: MIT Press. 2004 Beyond explanatory adequacy. In Structures and beyond, Adriana Belletti (ed.), 104–131. The carthography of syntactic structures. Vol. III. Oxford: Oxford University Press. 2005 Three factors in language design. Linguistic Inquiry 36, 1–22. Chomsky, Noam, and Howard Lasnik 1977 Filters and control. Linguistic Inquiry 8: 425–504. Chung, Sandra 1994 Wh-agreement and “referentiality” in Chamorro. Linguistic Inquiry 25: 1–44. 1998 The design of agreement: evidence from Chamorro. Chicago/London: The University of Chicago Press. Cinque, Guglielmo 1990 Types of A'-dependencies. Linguistic Inquiry monograph 17. Cambridge, Mass.: MIT Press. Dayal, Veneeta 1991 Subjacency effects at LF: The case of Hindi wh. Linguistic Inquiry 22: 762–770. Erteschik-Shir, Nomi 1986 Wh-questions and focus. Linguistics and Philosophy 9: 117–149. 1997 The dynamics of focus structure. Cambridge: Cambridge University Press.

Coptic relative tenses: The profile of a morpho-syntactic flagging device 217 Green, Melanie 1997 Focus and copular constructions in Hausa. Ph.D. dissertation. School of Oriental and African Studies, London. To appear in Publications of the Philological Society. Oxford: Blackwell. Green, Melanie, and Philip J. Jaggar 2003 Ex-situ and In-situ Focus in Hausa: syntax, semantics and discourse. In Research in Afroasiatic grammar II: Selected Papers from the Fifth Conference on Afroasiatic Languages, Paris 2000, Jacqueline Lecarme (ed.), 187–214. Amsterdam/Philadelphia: Benjamins. Green, Melanie, and Chris H. Reintges 2004a Syntactic anchoring in Hausa and Coptic wh-constructions. In Proceedings of the 27th Annual Meeting of the Berkeley Linguistics Society, Andrew Simpson (ed.), 61–72. University of California, Berkeley. 2004b Review article of Jaggar (2001). Lingua 114: 44–91. 2005 Syntactic conditions on special inflection: evidence from Hausa and Coptic Egyptian interrogative and focus constructions. University of Sussex Working Papers in Linguistics and English Language 14 at http://www.sussex.ac.uk/linguistics/1-4-1.html. Groat, Erich, and John O’Neil 1996 Spell-out at the LF interface: achieving a unified syntactic computational system in the Minimalist framework. In Minimal ideas: syntactic studies in the Minimalist framework, Werner Abraham, Samuel D. Epstein, Höskuldur Thráinsson, and Jan-Wouter Zwart (eds.), 113–139. Amsterdam/Philadelphia: Benjamins. Haïk, Isabelle 1990 Anaphoric, pronominal and referential INFL. Natural Language and Linguistic Theory 8: 347–374. Halle, Morris, and Alec Marantz 1993 Distributed Morphology. In The view from building 20: essays in honour of Sylvain Bromberger, Kenneth Hale and Samuel J. Keyser (eds.), 111–176. Cambridge, Mass.: MIT Press. Hartmann, Katharina, and Malte Zimmermann this vol. Exhausitivity marking in Hausa: A reanalysis of the particle nee/cee. Herburger, Elena 2000 What counts: focus and quantification. Cambridge, Massachusetts: MIT Press. Hopper, Paul J. 1979 Some observations on the typology of focus and aspect in narrative language. Studies in Language 3: 37–64. Horn, Laurence R. 1985 Metalinguistic negation and pragmatic ambiguity. Language 61: 121–174.

218

Chris H. Reintges

Hyman, Larry M., and John R. Watters 1984 Auxiliary focus. Studies in African Linguistics 15: 233–273. Jackendoff, Ray 1972 Semantic interpretation in generative grammar. Current studies in linguistics 2. Cambridge, Mass.: MIT Press. Jaggar, Philip J. 2001 Hausa. London Oriental and African language library. Amsterdam/Philadelphia: John Benjamins. in press The Hausa perfective tense-aspect used in wh-/focus constructions and historical narratives: a unified account. To appear in West African linguistics: descriptive, comparative and historical studies in honor of Russell G. Schuh, Larry M. Hyman and Paul Newman (eds.), Studies in African Linguistics, special issue. Kuno, Susumo 1972 Functional sentence perspective: A case study from Japanese and English. Linguistic Inquiry 3: 269–320. Layton, Bentley 2000 A Coptic grammar with chrestomathy and glossary. Porta Linguarum Orientalium N.S. 20. Wiesbaden: Harrassowitz. Loprieno, Antonio 1995 Ancient Egyptian: a linguistic introduction. Cambridge: Cambridge University Press. McCloskey, James 1990 Resumptive pronouns, A'-Binding, and levels of representation in Irish. In The syntax of Modern Celtic languages, Randall Hendrick (ed.), 199–248. Syntax and Semantics 23. New York: Academic Press. 2001 The morphosyntax of WH-extraction in Irish. Journal of Linguistics 37: 67–100. McDaniel, Dana 1989 Partial and multiple WH-movement. Natural Language and Linguistic Theory 7: 565–604. Nunes, Jairo 2004 Linearization of chains and sideward movement. Cambridge, Mass.: MIT Press. Newman, Paul 2000 The Hausa language: an encyclopedic reference grammar. New Haven and London: Yale University Press. Pesetsky, David 1987 Wh-in-situ: movement and unselective binding. In The representation of (in)definiteness, Eric J. Reuland and Alice G.B. ter Meulen (eds.), 98–129. Current studies in linguistics 14. Cambridge, Mass.: MIT Press.

Coptic relative tenses: The profile of a morpho-syntactic flagging device 219 1997

Optimality theory and syntax: movement and pronunciation. In Optimality Theory: an overview, Diana Archangeli and D. Terence Langendoen (eds.), 134–170. Oxford: Blackwell. 1998 Some optimality principles of sentence pronunciation. In Is the best good enough?, Pilar Barbosa, Danny Fox, Paul Hagstrom, Martha McGinnis, and David Pesetsky (eds.), 337–383. Cambridge, Mass.: MIT Press. 2000 Phrasal movement and its kin. Linguistic Inquiry monographs 37. Cambridge, Massachusetts: MIT Press. Pesetsky, David, and Esther Torrego 2000 T-to-C movement: causes and consequences. In Ken Hale: A life in linguistics, Michael Kenstowicz (ed.), 355–426. Cambridge, Mass.: MIT Press. Polotsky, Hans Jakob 1944 Études de syntaxe copte. Publications de la société d’archeologie copte. Cairo. Rackowski, Andrea and Norvin Richards 2005 Phase edge and extraction: a Tagalog case study. Ms., MIT, Cambridge, Mass.. Reinhart, Tanya 1984 Principle of Gestalt perception in the temporal organization of narrative texts. Linguistics 22: 779–809. Reintges, Chris H. 2001 Aspects of the morphosyntax of subjects and objects in Coptic Egyptian. In Linguistics in the Netherlands 2001, Ton van der Wouden and Hans Broekhuis (eds.), 177–188. Amsterdam/ Philadelphia: John Benjamins. 2002 A configurational approach to Coptic Second Tenses. Lingua Aegyptia 10: 343–388. 2003 Syntactic conditions on special inflection in Coptic interrogatives. In Research in Afroasiatic grammar II. Selected papers from the fifth conference on Afroasiatic languages, Paris 2000, Jacqueline Lecarme (ed.), 363–408. Current Issues in Linguistic Theory 241. Amsterdam/Philadelphia: Benjamins. 2004 Coptic Egyptian (Sahidic dialect): A learner’s grammar. Afrikawissenschaftliche Lehrbücher 15. Cologne: Rüdiger Köppe Verlag . Reintges, Chris H., and Melanie Green 2004 Coptic second tenses and Hausa relative aspects: a comparative view. Lingua Aegyptia 12: 157–177.

220

Chris H. Reintges

Reintges, Chris H., Phil LeSourd, and Sandra Chung 2006 Movement, wh-agreement, and ‘apparent’ wh-in-situ. In press. To appear in Wh-movement moving on, Lisa Lai-Shen Cheng and Norbert Corver (eds.), Cambridge, Mass.: MIT Press. Rizzi, Luigi 1996 Residual Verb Second and the wh-criterion. In Parameters and functional heads: essays in comparative syntax, Adriana Belletti and Luigi Rizzi (eds.), 63–90. Oxford/New York: Oxford University Press. 1997 The fine structure of the left periphery. In Elements of Grammar: Handbook in generative syntax, Liliane Haegeman (ed.), 281–337. Dordrecht: Kluwer. Rochemont, Michael S. 1986 Focus in generative grammar. Amsterdam/Philadelphia: Benjamins. Rooth, Mats 1992 A theory of focus interpretation. Natural Language Semantics 1: 75– 116. Tuller, Laurie 1986 Bijective relations in Universal Grammar and the syntax of Hausa. Ph.D. dissertation. University of California at Los Angeles. Watanabe, Akira 1996 Case absorption and wh-agreement. Dordrecht: Kluwer. Vries, Mark de 2002 The syntax of relativization. Ph.D. Dissertation. University of Amsterdam. LOT Dissertation Series 53. Utrecht: LOT. Zaenen, Annie 1983 On syntactic binding. Linguistic Inquiry 14: 469–504.

Part IV The inventory of focus marking devices

Identificational operation as a focus strategy in Byali Brigitte Reineke

Abstract This article investigates the focus system of Byali, a Gur language spoken in the north-western part of Benin. It is argued that the essential feature of focus marking in this language is an identificational operation which takes the form of an ex-situ construction. The latter is analyzed as containing two interrelated predications, namely a constituent that obligatorily precedes an identifying 'to be'-verb – as such representing a nominal predication – and a verbal process or state established in previous discourse. The grammatical operation of interrelating the two predications results in a biclausal focus construction. The identifying 'to be'-verb, which has been grammaticalized as a focus marker, but palpably still bears verbal features, establishes a relation between the focused constituent and the predicative statement representing the background. Whereas a combination of two propositions within a single utterance is shown for the ex-situ constructions, no such overlap can be established for the in-situ focus. As a result, I will argue that in-situ focus constructions exhibit a higher degree of grammaticalization.

1.

Introduction

This paper sheds light on the focus system of Byali and tries to show that the essential feature of the focus strategy in this language is an identificational operation clearly retrievable from the ex-situ constructions. The latter are envisioned to contain two interrelated predications, a constituent which obligatorily precedes an identifying 'to be'-verb – as such representing a nominal predication – and a pre-established verbal process or state. The operation of interrelating the two predications results in a focus construction. As Caron (2000: 28) puts it: ”l’imbrication, la présentation simultanée, dans un seul énoncé, d’un élément identifié (avec ou sans contraste) avec une place non-instanciée d’une relation prédicative.” The identifying 'to be'-verb that has been grammaticalized as a focus marker but palpably still

224 Brigitte Reineke bears verbal features establishes the relation between the focused constituent and the predicative statement. In-situ focus constructions are considered to be more grammaticalized as thought of before. The Byali language, along with Ditammari, Nateni, Waama, and MD'NKO', belongs to the eastern subgroup of the Oti-Volta-languages of the Gur family. It is spoken in the Atakora mountains in the north-western part of Benin. All the languages of this sub-group, except for Ditammari, display a full-fledged noun class system characterized essentially by suffixation of the class marker to the noun stem. Ditammari has prefixes in addition to suffixes, which distinguishes it from most Gur languages which are mainly characterized by suffixes. The languages have a fully intact agreement system, as shown by the existence of anaphoric pronouns and of nominal class markers linked to the modifiers of the noun which indicate their syntactic dependencies. In addition, the verbal system exhibits a grammaticalized aspectual system characterized by the binary opposition of perfective and imperfective aspect. Also, these languages display a verbal derivation having semantic as well as syntactic functions. On the prosodic level, tone plays an important part in the grammar and the lexicon. The basic word order in a canonical sentence is SVO; only in case of pronominal objects, the order is SOV. This paper is structured as follows: part 2. illustrates the essential characteristics of the Byali focus system. As concerns the concrete focus phenomena, the distinction between non-verbal focus (3.) and verbal focus (4.) is made, and the asymmetry between non-subject focus (3.1) and subject focus (3.2) inherent in non-verbal focus is paid special attention to. In (5.), an interpretation of the presented facts is given. The paper ends with a summary in (6.).

2.

Overview on the essential characteristics of the focus system

To express focus grammatically, the language employs morphological, syntactic, as well as phonological means. At the level of morphology, grammatically marked focus is realized by a focus marker immediately following the constituent in focus. This holds true for the labelling of term focus in the same way as for verbal focus. Concerning the polarity, a single specific focus marker exists for affirmative ((l)G) 1 and negative focus (PYC), respectively. Different syntactic focus constructions manifest themselves in the ex-situ and in the in-situ structures. In the ex-situ construc-

Identificational operation as a focus strategy in Byali 225

tions, the focused constituent is located in sentence initial position. Although it shares such a position with the topic constituent, the focused term is integrated into the structure of the predicate and fulfills a grammatical function whereas the topic constituent always stands outside the predication and, from a semantic and structural perspective, can be seen to be optional. 2 In ex-situ constructions, the focus constituent represents the in-focus part of the sentence that is followed by the out-of-focus part after a break. In addition to this multiple marking of the focused term – i.e. syntactically by the sentence-initial position, morphologically by the focus marker, and (in case of the ex-situ constructions) phonologically by phrasing – the non-focal part of the focus construction exhibits relative structure, an isomorphism frequently encountered across languages. The out-of-focus part is marked as pragmatically dependent in two respects: Besides the relative structure, there is a pragmatically conditioned verbal morphology that is restricted to the perfective aspect. The examples show that neither in ex-situ nor in insitu focus constructions is the perfective verb joined by the suffix ³U‹YJKEJJCUan assertive meaning, cf. (1) and (2) vs. (3). 3 The latter example (3) is unmarked with respect to focus. Thus, the perfective verb takes the suffix ³U‹ Pragmatically marked focus constructions, the suffix ³U‹ is absent: (1)

D¸"I‹P7PF‹" DCPCPC G child buy.PFV banana FM ‘The child bought BANANAS.’

as an answer to a constituent question: (2)

D¸"I‹P7PF‹" DCCT‹? child buy.PFV what ‘What did the child buy?’

No marking of focus, the suffix ³U‹ appears: (3)

D¸"I‹ P7PF‹"³ U‹ DCPCPC child buy.PFV ASS banana ‘The child bought bananas.’

226 Brigitte Reineke Concerning the functional differentiation between the two syntactic focus constructions that has been observed in the literature on many languages of the world, 4 as far as the current research goes it does not seem to apply to Byali. It is often claimed that the ex-situ constructions represent the socalled identificational focus where an element is chosen out of the possible alternatives in the context. This constituent would be marked as the one which the predication would hold for to the exclusion of all other alternatives. At the same time, the focused element would stand in contrast to the other options so that the features of exhaustivity and contrast would be highlighted as typical for the ex-situ constructions. The in-situ type as information structure refers to the part of predication which represents the new information, i.e. the focused constituents fill a gap within the pragmatic information of the listener. While in Byali new information focus is realized primarily in-situ and identificational focus is very frequently associated with exhaustivity and contrast, neither of the two types of structures seem to be bound unequivocally to one of the semantic-pragmatic meanings. According to the judgment of the informants, both syntactic focus constructions allow for either of the interpretations.

3.

Non-verbal focus

The differentiation between non-subject and subject focus applies to nonverbal focus, whether the focused term is the argument of a verb or an adverbial phrase not governed by the valency of the verb. Such differentiation appears to be necessary insofar as this asymmetry known from many other languages is typical for Byali as well even though from a structural point of view there is no difference regarding non-subject focus in sentence initial position. While non-subject focus can be realized either ex-situ or in-situ, subject focus obligatorily is expressed by an ex-situ construction. For the better understanding of the following data illustrating both exsitu as well as in-situ constructions, some preliminary remarks shall be given before an interpretation is carried out in section 5. 1. The relative structure that marks the non-focused part of the ex-situ construction and which represents the background is characterized by a high toned nasal placed in front of the verb, as well as by the class marker of the reference noun following the verb. In (5), the relative nasal precedes the verb [CtOD‹"‘see.PFV’ that is followed by the class marker of the noun in focus, D¸³¸ I‹‘child’, here in the allomorphic

Identificational operation as a focus strategy in Byali 227

variant M‹ From this example, it is evident that the syntactic postverbal base-position of the argument marked for focus by the focus marker, here the object position, is empty, i.e. the position in the presupposed predication is not filled with an anaphoric pronoun. 2. The interrogative clauses show that in Byali, in contrast to many other languages, the question words are never followed by the focus marker. Contrary to answers, where the focus marker obligatorily marks the focused constituents in in-situ as well as in ex-situ constructions. This fact indicates that in Byali question words are inherently focus-marked. 3. In sentences where a focus marker follows the object in in-situ constructions (cf., for example, (7) and (11)), narrow focus on the object as well as focus on the verb together with its succeeding object can occur, so that in such a case the scope of the focus stretches over the whole of the verbal phrase. Sentences where the object is focus-marked in-situ are therefore ambiguous because they may as well represent a reply to the question “What did she do?”. Their meaning can only be determined by the context. 4. The combination of negated verb forms with marked focus that is acceptable in many languages does not apply to Byali. 3.1. Non-subject focus For the presentation of the relevant data, question/answer-pairs are used as a reliable method for elicitation. Questions point towards the interest of the interrogator for information. The answers focus on the entities representing the desired information. In examples (4) to (11) where the object is focused, the WH-question together with the corresponding answer are shown in both construction types, namely ex-situ and in-situ, respectively. 3.1.1. Object ex-situ (4) Q: YG"GV‹" Coffi P" [CtOD who Coffi REL see.PFV ‘WHO is it that Coffi has seen?’ (5)

W"? CL.REL

A: D¸¸³ I³ G W P" child CL FM CL.SUBJ. REL ‘It is the CHILD that (s)he has seen.’

[CtOD‹" see.PFV

M‹ CL.REL

228 Brigitte Reineke in-situ (6) Q: Coffi [CtOD‹" YG"GV‹"? Coffi see.PFV who ‘WHOM has Coffi seen?’ (7) 

A: W [CtOD‹" D¸¸³ I³ G  CL.SUBJ see.PFV child CL FM ‘He has seen the CHILD.’ (DK"K"³I‹‘child’)

ex-situ (8) Q: DCCT‹" W P" P7PF‹" ¹? what CL. SUBJ REL buy.PFV CL.REL ‘WHAT is it that (s)he has bought?’ (9)

A: DCPCPC G W P" P7PF‹" JCp banana FM CL.SUBJ REL buy.PFV CL.REL ‘It is BANANAS that (s)he has bought.’

in-situ (10) Q: W P7PF‹" DCCT‹? CL.SUBJ buy.PFV what ‘WHAT has (s)he bought?’ (11) A: W P7PF‹" DCPCPC G CL.SUBJ buy.PFV banana FM ‘(S)he has bought BANANAS.’ Negation of an object focused in-situ is realized by a combination of the preverbal negation morpheme R‹ and a focus marker following the object. 5 (12)  

W R‹ P7PF‹" DCPCPC PYC CL.SUBJ NEG buy.PFV banana FM.NEG (S)he has not bought BANANAS.’

3.1.2. Adjuncts When adjuncts are focused, the same regularities apply as in the case of focused objects, i.e. they can occur in ex-situ as well as in in-situ constructions and they are always followed by the focus marker. In ex-situ construc-

Identificational operation as a focus strategy in Byali 229

tions, the out-of-focus part of the utterance also exhibits the features of a relative clause. The class marker that is referring to adjuncts and which is postponed to the verb as a discontinuous part of the relative construction is always -m in relation to the m-noun class. In (13) to (16), examples with a local adjunct are given: ex-situ (13) Q: OC"PV‹"" W P" [CtOD‹" O" UC"D³ where CL.SUBJ REL see.PFV CL.REL book ‘WHERE is it that (s)he has seen the book?’

¹? CL

(14) A: MCT³ º [CtJ‹ G W P" [CtOD‹""  market CL inside FM CL.SUBJ REL see.PFV O" UC"D³¹  CL.REL book CL ‘It is ON THE MARKET that (s)he has seen the book.’ in-situ (15) Q: W [CtOD‹"" UC"D³ ¹ CL.SUBJ see.PFV book CL ‘WHERE has (s)he seen the book?’

OC"PV‹? where

(16) A: W [CtOD‹"" UC"D³ ¹ MCT³ º[CtJ‹G CL.SUBJ see.PFV book CL market CL inside FM ‘(S)he has seen the book ON THE MARKET.’ 3.2. Subject focus As already mentioned, the focusing of subjects is restricted to the ex-situ strategy. In Byali, the asymmetry between non-subject focus and subject focus is expressed solely by this restriction of subject focus to ex-situ constructions. A hint at the ex-situ realization of subject focus is constituted by the fact that the predication is categorically taking a relative-like structure, as is exactly the case with ex-situ focused non-subjects, cf. the sentences given in (17) to (20). Therefore, the subject constituent marked for focus is directly followed by the relativizing nasal. Also in this context, attention has to be paid to the fact that because of focus being inherent to it the question pronoun asking for the subject is not followed by any focus marker.

230 Brigitte Reineke (17) Q: YG"GV‹"P" U+¹P [CtOD W" who REL yesterday see.PFV CL.REL ‘WHO is it that saw the child yesterday?’

D¸"³ child

(18) A: Coff" G P" U+¹P M‹ [CtOD Coffi FM REL yesterday CL.OBJ. see.PFV ‘It is COFFI that saw it yesterday.’ (19) Q: DCCT‹" P" FGG what REL fall.PFV ‘WHAT has fallen?’

I‹? CL W" CL.REL

¹? CL.REL

(20) A: UC"D³  G P" FGG  book CL FM REL fall.PFV ‘It is a BOOK that has fallen.’

¹ CL.REL

Structurally, subject focus constructions can also serve to express sentence focus and are therefore ambiguous. This isomorphism holds valid for many Gur languages. Apparently, the non-existence of a topic in case of sentence focus, which constitutes a thetic construction, determines subject focus structure. 6 The following sentence containing a focus-marked subject is a reply to the question “What happened?”. (21)

4.

P UCP³JWt  NG P" UYCP‹" JWp 1POSS car CL FM REL break.PFV CL.REL ‘MY CAR BROKE DOWN.’

Verbal focus

To mark focus in the verbal domain grammatically, in Byali, the same focus markers are used as with term focus 7 , i.e. (l)G for affirmative and PYC for negative focus. The focus marker immediately follows the verb. (22)

W UQWO PG CL.SUBJ smoke.IPFV FM ‘(S)he SMOKES.’

Identificational operation as a focus strategy in Byali 231

(23) 

W UºPF‹" G OCPU¸P¹ CL.SUBJ turn on FM Computer ‘(S)he HAS TURNED ON the computer.’

If the verb in focus is negated, the combination of the preverbal negation morpheme R‹and the negative post-verbal focus marker turns out to be obligatory: (24)

O R‹ UºPF‹" PYC 1SG NEG turn on.PFV FM.NEG ‘I haven’t TURNED ON the computer.’ 8

OC"PUKP¹ Computer

With the focused sentence-initial position being restricted to nominal constituents, the essential feature of verb focus is that it exclusively occurs in in-situ constructions. In some cases, the Byali language makes use of a nominal periphrasis strategy for marking focus on the verbal phrase by nominalizing the verb that forms an associative construction together with its original object (Note that verbs are nominalized by integrating them into the m-class). Thus, the nominalization of the verb-object phrase is a precondition for it to be marked for focus and at the same time for it to occur in ex-situ constructions. However, the nominalizing strategy for marking predicate focus only occurs when the verbal phrase is represented by a progressive construction consisting of the dummy verb D‹"³RW"¢¸‘to hold’ + a nominalized verb. ex-situ (25) 

EQ³

W

UCJCp³

P"PG W

O" RW"¢‹"

O"

porridge CL preparation CL FM CL.SUBJ REL hold.IPFV CLREL

‘It is the PREPARING OF PORRIDGE that she does.’ in-situ (26)

W RW"¢¸ EQ³ W UCJCp³ P" CL.SUBJ hold.IPFV porridge CL preparation CL ‘She IS PREPARING PORRIDGE.’

PG FM

It would lead to ungrammaticality if the nominalized verb is copied and used as a finite verb instead of RW"¢¸, cf. (27): (27)

*EQWUCJCp³PPGWO"UCJCpO"

232 Brigitte Reineke 5.

Interpretation

In order to understand how focus phenomena in Byali can be interpreted, attention shall be centered on the ex-situ constructions. The following data will show why the identificational strategy is considered decisive for marking focus in the Byali language. The ex-situ focus constructions contain two interrelated predications, a constituent followed obligatorily by a 'to be'-verb with identifying meaning and placed at the left periphery, and a verbal expression referring to a presupposed process or state. The sentence-initial term is identified as a referent and at the same time as the focused term by the 'to be'- verb functioning as focus marker; it is brought into an equational relation with the verbal predication in which the syntactic base position of an argument is empty. The operation of interrelating these two predications results in a focus construction with a biclausal character. The in-focus part of the overall structure has to be evaluated as nominal predication. This is especially indicated by the possible connection of the 'to be'-verb with tense markers, cf. section 5.1. In verbal predication, it is characteristic of the out-of-focus part of an utterance to be realized by means of relative-like structures, an operation observed very frequently across languages. 5.1. Encoding of identity statements The encoding of identity statements in Byali makes it quite obvious that the focus marker lè, in specific cases reduced to è, and its negative form PYC have to be traced back to a verb 'to be', or 'not to be' respectively, which has an identifying meaning and that is used in both types of identifying structures, first, the presentational type, and second, the identifying nominal predication. 5.1.1. The presentational type The presentational construction makes the identity of a referent known to the hearer, i.e. “presentation statements provide an object or a class of objects with a 'name', which may from now on be used by the hearer in referring to it” (Stassen 1997). In following Lambrecht’s (2001) approach, Tröbs (2002) emphasizes that one-argument identificational clauses exhibit a marked information structure where the scope of assertion stretches over

Identificational operation as a focus strategy in Byali 233

the entire proposition (in the sense of Sasse’s thetic utterances), i.e. the construction is to a large extent discourse-pragmatically motivated. In Byali, this one-argument identificational statement consists of a nominal or pronominal constituent + the 'to be'-verb; the form lè occurring after pronouns is very often reduced to è in connection with a noun. In case of negation, the affirmative (l)è is substituted by PYC The presupposition of such identificational statements is not explicitly expressed, only situationally implied. In this way, the following statements are understood. (28)     

CNG WNG V‹NG D‹NG UC"D³C"G CNG  D¸¸³IG(< D¸¸³I‹G) M‹NG

Negation: (29) PPYC  WPYC DG"IC"PYC  D¸¸³U¸PYC

‘It is you.’ ‘It is him/her.’ ‘It is us.’ ‘It is them.’ ‘It is the books.’ ‘It is them.’ (= the books < UC"D³C") ‘It is the child.’ ‘It is it.’ (= the child < D¸¸³I‹)

‘It is not me.’ ‘It is not him/her.’ ‘It is not the chief.’ ‘It is not the children.’

5.1.2. The identifying nominal predication Identifying nominal predications can be classified as equational; two entities refer to one and the same object. According to Stassen, both noun phrases have the same semantic status and both can function as subject or predicate, but their position evidently depends on discourse conditions. So, the following sentences differ pragmatically: (30)

CQHHK G U[GN³ ¹ O[CV³ ¹ Coffi be.ID village CL teacher CL ‘Coffi is the teacher of the village.’

(31) 

U[GN³ ¹ O[CV³  G CQHHK village CL teacher CL be.ID Coffi ‘The teacher of the village is Coffi.’

234 Brigitte Reineke The fact that in both types of identity statements the affirmative (l)è and the negative PYC can be linked to the subject personal pronoun and can be preceded by tense markers hints at the verbal features of these morphemes. The emphatic pronominal forms cannot occur as subjects neither in oneargument identificational clauses nor in identifying nominal predications. with past marker ([¹): (32) a. C [¹ NG 2SG PAST be.ID ‘It was you.’ *CYG[¹NG(CYG= 2.P.SG. emphat.) (cf.CYGC[¹NG‘You, it was you.’, where CYGrepresents the topic of the sentence.) b. W[¹PYC  ‘It was not him.’ c. CQHHK[¹NGU[GN¹O[CV¹  ‘Coffi was the teacher of the village.’ d. CQHHK[¹PYCU[GN¹O[CV¹  ‘Coffi was not the teacher of the village.’ Following subject pronouns or tense markers, the phonetically non-reduced form of the affirmative identifying 'to be'-verb is obligatorily used, as (28), (32a,b), and (33a) show. with future marker ([¸): With verbs of stative meaning, the future tense marker [¸is always followed by a nasal. (33) a. W[¸P"FG(< NG) b. W[¸P"FGU[G"N¹O[CV¹ c. CQHHK["P"PYCU[G"N¹O[CV¹

‘It will be him.’ ‘He will be the teacher of the village.’ ‘Coffi won’t be the teacher of the village.’

In this connection, notice that in predicational relations characterized by a subject and predicate structure, i.e. in predications in which the subject is not identical to the entity specified in the predicate nominal, the classifying copula PYCO occurs. It is also linked to personal and anaphoric subject pronouns, not to emphatic ones, and can co-occur with tense markers. In these sentences, the predicate nominal is marked for focus.

Identificational operation as a focus strategy in Byali 235

(34) a. b. c. 

WPYCOO[CVG ‘He is a TEACHER.’ W[¹PYCOO[CVG ‘He was a TEACHER.’ ¹PYCOb[C"N‹DG(< b[C"N‹D‹G)‘You are BYALEBE, ¹R‹PYCOPCV[GOD‹PYC you are not NATEMBA.’

5.2. Focus constructions The same phenomenon, i.e. the possible combination of (l)è with tense markers, occurs when (l)è functions as focus marker, as the following example (35) containing the past morpheme [¹shows: (35) 

D¸¸³ I‹ [¹ NG W P" [¹ [CtOD‹" M‹ child CL PAST FM CL.SUBJ REL PAST see.PFV CL.REL ‘It was the CHILD that (s)he had seen.’

However, the variant without the tense marker in the in-focus part is the preferred one; the temporal reference of the whole predication is designated by the obligatory occurrence of the tense marker before the verb: (36) 

D¸³¸  I(‹)³G W P" [¹ EJKNFCL FM CL.SUBJ REL PAST ‘It was the CHILD that (s)he had seen.’

[CtOD‹" M‹ see.PFV CL.REL

It is interesting to note that the combination of the focus marker with tense morphemes is not restricted to the fronted constituent; its occurrence is also possible, even if infrequent and unusual, with constituents focused in-situ: (37)

W [¹ [CtOD‹" D¸¸³ CL.SUBJ PAST see.PFV child ‘(S)he had seen the CHILD.’ or: W[¹[CtOD‹"D¸¸³I(‹)³G

I‹ [¹ NG CL PAST FM

This applies also to instances of verbal focus: (38)

W [¹ F¹ [¹ NG EQ³ W CL.SUBJ PAST eat.PFV PAST FM porridge CL ‘(S)he had EATEN the porridge.’ or: W[¹F¹¹GEQW

236 Brigitte Reineke The possible occurrence of the focus marker with tense morphemes in exsitu as well as in in-situ position of the focused constituents indicates very clearly the original verbal feature of NGand supports the interpretation of the sequence N - (tense marker) - focus marker as an original nominal predication. Another observation supporting the hypothesis of the verbal origin of the focus marker is the use of personal pronouns and anaphoric pronominal forms in subject function when they are focused. The use of their free or emphatic variants respectively is ungrammatical: (39)

C NG P" F W EQ³ W 2SG FM REL eat.PFV CL.REL porridge CL ‘It was YOU that has eaten the porridge.’ *CYG(l)GP"F¹¹WEQW(CYG= 2. SG. emphat.)

It should be remembered that emphatic pronouns are also illicit in oneargument identificational clauses and in identifying nominal predications. The examples in (37), (38), and (39) show that the focus marker always appears in its non-reduced form NG when immediately following tense markers and subject pronouns. 5.3. In-situ constructions: More grammaticalized? What about in-situ constructions? How could they be interpreted? Whereas the ex-situ constructions are characterized by an imbrication of two propositions within a single utterance, such an imbrication cannot be established for in-situ focus. One might be able to consider sentences in which objects are focus marked in-situ (40), or in which adjuncts take a sentence-final position (41), to consist of two subsequent predications as well: (40) 

W P7PF‹" DCPCPC G CL.SUBJ buy.PFV banana FM ‘(S)he has bought BANANAS.’ = ?? (S)he has bought // it is bananas

(41) 

Coffi [CtOD‹" D¸¸³ I‹ MCT³ º [CtJ‹ G Coffi see.PFV child CL market CL inside FM ‘Coffi has seen the child ON THE MARKET.’ = ?? Coffi has seen the child // it is on the market.

Identificational operation as a focus strategy in Byali 237

Such an interpretation would be somewhat more difficult for sentences with focused verbs taking one or two arguments; there, the verb is separated from its argument(s) by (l)è, cf. the following sentence: (42) 

Coffi F¹¹ G EQ¸³ W Coffi eat.PFV FM porridge CL ‘Coffi has EATEN the porridge.’ = ?? Coffi has eaten // it is // the porridge.

A similar problem arises with adjuncts that can also take the position between the verb and its argument in addition to sentence-final position, so that apart from SVOA the sequence SVAO is possible as well. Cf. (43) vs. (41), where the adjunct MCT¹[CtJ‹G ‘(it is) on the market’ occurred in sentence-final position: 9 (43)

Coffi [CtOD‹" MCT³ ¹ [CtJ‹ G D¸¸³ Coffi see.PFV market CL inside FM child ‘Coffi has seen the child ON THE MARKET.’ = ?? Coffi has seen // it is on the market // the child.

I‹ CL

In none of the cases can in-situ constructions be evaluated as an imbrication of two propositions within a single utterance. In general, one might interpret the current status of these constructions as more grammaticalized.

6.

Summary

First, in tracing back the (affirmative) focus marker (l)è to a verb 'to be' with identifying meaning, the identificational strategy is considered decisive for explaining ex-situ focus constructions in Byali. The identifying 'to be'-verb exhibits by itself discourse-pragmatical features; the constructions in which it occurs are to a large extent pragmatically motivated. Second, the ex-situ focus construction contains two interrelated predications, a presupposed verbal process (or state) representing the background and a constituent followed by the 'to be'-verb with identifying meaning and placed at the left periphery for which the verbal predication holds true. The operation of interrelating these two predications results in a focus construction with biclausal character. The identifying 'to be'-verb functions in this

238 Brigitte Reineke construction as focus marker while still possessing verbal features, as evidenced by its co-occurrence with subject pronouns and tense markers. The out-of-focus part of ex-situ constructions is marked as pragmatically dependent in two ways: First, the non-focal part exhibits relative structure; second, there is a clearly pragmatically conditioned verbal morphology, restricted to the perfective aspect. Third, whereas an imbrication of two propositions within a single utterance could be shown for the ex-situ constructions, no such overlap can be established for the in-situ focus. The in-situ constructions are considered as more grammaticalized. Abbreviations ASS CL DEP EMPH FM IPFV NEG OBJ PFV POSS REL SG SUBJ

assertive suffix class marker dependent marker emphatic focus marker imperfective negation particle object perfective possessive pronoun relative marker singular subject

Notes 1. Usually the phonologically reduced variant Gis employed. If the focused entity ends with a nasal, the focus marker (l)G) is realized with its allomorph variant nè. 2. If both categories, topic and focus, appear in a single sentence, the sequence topic – focus corresponds to the universal fact that, contrary to focus, topic is external to the sentence. 3. In the same way, the non-occurrence of the assertive verbal suffix in relative clauses, temporal and conditional clauses, sequential clauses as well as in negative sentences is due to their pragmatics, cf. Frajzyngier (2004)

Identificational operation as a focus strategy in Byali 239

4. 5. 6. 7.

8.

9.

who emphasizes the pragmatical dependency of these clauses in languages of the Niger-Congo as well as the Chadic families. Cf. e.g. E. Kiss (1998). In ex-situ focus constructions, the negative focus marker follows the negated entity placed in sentence initial position as well. Cf. Sasse 1987, 1995. This fact conflicts with the situation in the Ditammari language belonging to the same subgroup. In Ditammari, nominal constituents and verbs are focused by different markers. If it is not the negated verb that stands in the scope of focus but the entire utterance meaning something like “It’s true that they didn’t steal it.”, then the negation morpheme R‹ and the affirmative focus marker (l)Gcan cooccur in a sentence. A combination of [R‹+ Verb] [(l)G] in a sentence represents the statement of the truth value of a negated proposition whereas the combination [R‹+ Verb] [PYC] represents the negation of a focused verb or term, e.g. OR‹N‹UºPF‹"G‘(It is true), I HAVEN’T TURNED IT ON.’ cf. Reineke (2006). A pragmatic difference between the two positional variants was negated by the informant. Still, further research will have to resolve whether or not the immediate post-verbal position is related to stronger emphasis.

References Bearth, Thomas 1993 Satztyp und Situation in einigen Sprachen Westafrikas. In Beiträge zur afrikanischen Sprach- und Literaturwissenschaft, Wilhelm J.G. Möhlig (ed.), 91–104. Köln: Rüdiger Köppe. Caron, Bernard 2000 Assertion et préconstruit: topicalisation et focalisation dans les langues africaines. In Topilisation et focalisation dans les langues africaines, Bernard Caron (ed.), 7–42. Louvain-Paris: Peeters. É. Kiss, Katalin 1998 Identificational focus versus information focus. Language 74 (2): 245–273. Frajzyngier, Zygmunt 2004 Tense and aspect as coding means for information structure: a potential areal feature. Journal of West African Languages 30: 53–67. Hopper, Paul 1979 Some observations on the typology of focus and aspect in narrative language. Studies in Language 3: 37–64.

240 Brigitte Reineke Reineke, Brigitte 2006 Verb- und Prädikationsfokus im Ditammari und Byali (to appear). In “Mama Miehe”. Festschrift für Gudrun Miehe zum 65. Geburtstag, Kerstin Winkelmann and Dymitr Ibriszimow (eds.), 163–180. Köln: Rüdiger Köppe. Sasse, Hans-Jürgen 1987 The thetic/categorical distinction revisited. Linguistics 25 (3): 511– 580. 1995 ‘Theticity’ and VS order: A case study. In Verb-subject order and theticity in European languages (Sprachtypologie und Universalienforschung, 48: 1/2), Yaron Matras and Hans-Jürgen Sasse (eds.), 3– 31. Berlin: Akademie-Verlag. Stassen, Leon 1997 Intransitive predication: Oxford studies in typology and linguistic theory. Oxford: Clarendon Press. Tröbs, Holger 2002 Some notes on one-argument-identificational clauses in Manding (Western Mande, Niger-Congo). Afrika und Übersee, 85: 129–144.

Exhaustivity marking in Hausa: A reanalysis of the particle nee/cee Katharina Hartmann and Malte Zimmermann

Abstract The article presents a reanalysis of the Hausa particle nee/cee, which is frequently analyzed as a focus marker (Tuller 1986, Green 1997). The syntactic distribution of nee/cee is shown to be not typical of genuine focus markers: nee/cee is optional. It is not restricted to a particular position as it can occur with ex-situ and in-situ focus (with different frequencies). And, nee/cee can associate with focus at a distance. Taking up observations by Green (1997), it is shown that nee/cee is not optional from a semantic point of view, as its presence triggers an exhaustive interpretation of focus. Nee/cee is therefore analyzed as a focus-sensitive marker introducing a conventional implicature, which is responsible for the exhaustive interpretation. This reanalysis of nee/cee also accounts for its infelicity in non-exhaustive contexts.

1.

Introduction

Hausa focus constituents are often accompanied by the particle nee or its feminine counterpart cee. The literature usually describes this particle as a focus marker. Green (1997:29) mentions that the particle leads to an exaustive interpretation of the focus but does not further develop this idea. In this article, we take up Green’s observation and show that nee/cee always indicates exhaustivity. Our reanalysis of the particle is based on several observations showing that nee/cee does not share most of the typical properties of grammatical focus markers. It is similar to focus markers in being focusrelated. It differs from them in at least three respects: First, it is optional, even if focus is not marked by other strategies. Second, it can associate with the focus at a distance, an untypical property of focus markers. Third, if present, the particle nee/cee has a semantic impact in form of a conventional implicature: it causes an exhaustive interpretation of the focus. It is therefore excluded in non-exhaustive environments such as mention-some

242 Katharina Hartmann and Malte Zimmermann contexts, or in contexts where a property is known to hold of more than the focused entity.

2.

Focus in Hausa

Hausa 1 is a tone language with three lexical tones: a high tone, which is not marked in the examples, a low tone (`), and a falling tone (^). Its basic word order is SVO. The uninflected verb is preceded by a separate morpheme that encodes temporal, aspectual, and agreement specifications (the auxiliary). In the perfective and continuous aspects, the auxiliary has two different morphological forms, depending on whether some constituent of the sentence is fronted (Tuller 1986). We follow the traditional terminology and call the auxiliary in clauses without fronting the absolute auxiliary. In clauses with fronting, we refer to it as the relative auxiliary. Hausa has two strategies to express focus. A focus constituent can either be fronted (ex-situ focus) or it can remain in its base-position (in-situ focus). Ex-situ focus (cf. Tuller 1986, Green 1997, Newman 2000, Jaggar 2001, Hartmann and Zimmermann 2007) is repeatedly marked: It is syntactically marked through fronting. In addition, it is morphologically marked since syntactic fronting triggers the relative auxiliary. The fronted focus constituent can be followed by the particle nee (or its feminine form cee). Finally, focus fronting is prosodically marked by an intonational phrase boundary between the ex-situ constituent and the rest of the clause (cf. Leben, Inkelas and Cobler 1989). An example of ex-situ focus is given in (1), with the focus printed in bold face. 2 Here, as in most other examples, focus is pragmatically controlled for by means of question-answer pairs. (1)

Q: Mèe sukà kaamàa ? what 3PL.REL.PERF catch ‘What did they catch?’ A: Kiifii (nèe) sukà kaamàa. PRT 3PL.REL.PERF catch fish ‘They caught FISH.’

In-situ focus, on the other hand, is not marked morpho-syntactically (Jaggar 2001, 2004, Green and Jaggar 2003, Hartmann and Zimmermann 2007). Due to the absence of focus movement, the auxiliary appears in its absolute form. In-situ focus is also not marked prosodically. Furthermore, it is only

Exhaustivity marking in Hausa: A reanalysis of the particle nee/cee 243

rarely accompanied by the particle nee, which is often shortened to ne in sentence-final position. (2) gives a typical question-answer pair. Since whphrases are almost always fronted, the relative auxiliary occurs in the question. In the answer, however, the focus (kiifii, ‘fish’) is not moved, hence the auxiliary is absolute. There is no formal indication of the focus in the answer at all, which can therefore only be determined pragmatically. (2)

Q: Mèe sukà kaamàa ? what 3PL.REL.PERF catch ‘What did they catch?’ A: Sun kaamà kiifii. fish 3PL.ABS.PERF catch ‘They caught FISH.’

While ex-situ focus is restricted to maximal projections, in-situ focus is possible for any constituent (heads and maximal projections alike), with the exception of subjects. If a subject is focused, the auxiliary must be relative, indicating ex-situ focus. This is illustrated in (3). (3)

Q: Wàa yakèe kirà-ntà ? who 3SG.REL.CONT call-her ‘Who is calling her?’ A: [NP Dauda] (nèe) yakèe / *yanàa kirà-ntà. PRT 3SG.REL.CONT/ 3SG.ABS.CONT call-her D. ‘DAUDA is calling her.’

Our analysis of the distribution and meaning of the particle nee/cee diverges from the opinion held in the recent literature on focus in Hausa where it is proposed that nee/cee is a focus marker (Green 1997, 2004, Newman 2000). Since nee/cee always appears together with focus, this analysis seems to be plausible at first glance. There are three arguments against this position, however. In a nutshell: Assuming that focus markers are required to mark focus, the optionality of nee/cee is unexpected (section 3). Second, the frequency of the particle considerably differs between exsitu and in-situ focus: nee/cee freely associates with ex-situ focus, but is quite restricted with in-situ focus. Given that, at least with question-answerfocus, the in-situ strategy is the prominent strategy (for a quantitative study of in-situ and ex-situ focus, cf. Hartmann and Zimmermann 2007), the rare occurrence of nee/cee in these cases would be unaccounted for if it was

244 Katharina Hartmann and Malte Zimmermann indeed a focus marker. Finally, the semantic behavior of nee/cee is untypical of focus markers: it always appears in exhaustive environments (section 4). From these considerations, we conclude that nee/cee is not a focus marker, but a focus-sensitive exhaustivity marker.

3.

Syntactic properties of nee/cee

In this section, we further investigate the syntactic distribution of nee/cee. We first discuss the particle after ex-situ focus and then turn to its more restricted occurrence with in-situ focus. 3.1. Ex-situ focus It is well-known from the literature on Hausa that the particle nee/cee is focus-related, i.e. that it only appears if a constituent is focused (Parsons 1963, Schachter 1966, Tuller 1986, Green 1997, Newman 2000, Jaggar 2001, Green and Jaggar 2003, Hartmann and Zimmermann 2007). The particle can occur after a fronted focus. The following examples illustrate subject focus (4ab), object focus (4c), PP focus (4d), and (nominalized) VP focus (4e) (examples (4ade) are from Newman 2000: 187ff.). (4)

a. [NP Dèelu] cèe takèe sôn àgoogo. D. PRT 3SG.F.REL.CONT want watch ‘DEELU wants a watch.’ zoo. b. [NP Kànde da Hàwwa] nee sukà K. and H. PRT 3PL.F.REL.PERF come ‘KANDE AND HAWWA came.’ Dèelu takèe sô. c. [NP Àgoogo] nèe watch PRT D. 3SG.F.REL.CONT want ‘Deelu wants A WATCH.’ sòokee shì. d. [PP Dà wuÝaa] nèe ya with knife PRT 3SG.REL.PERF stab him ‘He stabbed him WITH A KNIFE.’ hàrÄaajì-n] nee Tankò ya yi. e. [VP Biyà-n paying-GEN taxes-DEF PRT T. 3SG.REL.PERF do ‘It was PAYING THE TAXES that Tanko did.’

Exhaustivity marking in Hausa: A reanalysis of the particle nee/cee 245

The particle has a tonal peculiarity in that it always carries polar tone, i.e. a tone opposite to the preceding tone (cf. Parsons 1963:166). We further assume that the particle nee/cee is formally unspecified: it neither carries a tense specification nor is it specified for agreement features, with the exception of gender. It is only with feminine singular noun phrases that gender is specified and cee is used instead of nee (4a). In all other cases, e.g. with masculine NPs (see (1), (3), and (4c)), plurals (including coordinated feminine NPs (4b)), PPs (4d), and VPs (4e), nee must occur (cf. Parsons 1963). Hausa scholars usually analyze nee/cee after ex-situ focus as an emphatic marker (Schachter 1966) or as a focus marker (cf. Tuller 1986, Green 1997, Green 2004, Newman 2000, Jaggar 2001, Green and Jaggar 2003). Given the existence of focus markers in a large variety of other African languages (see Bearth 1999 for an overview), this assumption is not far-fetched. Green (1997) represents the most elaborate analysis of focus in Hausa. In her account, nee/cee is the head of a focus phrase (FP). Provided with focus features, the particle attracts the focus phrase to its specifier: (5)

biyà teelà]]]] [FP [NP BintàF] [F’ [F cee [S tsubj takèe PRT 3SG.F.REL.CONT pay tailor B. ‘BINTA paid the tailor.’

All analyses acknowledge that the particle is optional. The examples in (4) are equally grammatical in the absence of nee/cee. In other words, the presence of nee/cee is not obligatory for focus marking. It is still optional if there is no word order variation, e.g. with focused subjects, which appear in the same linear position as unfocused subjects. The only indication of subject focus in (6) is the relative auxiliary. (6)

Tankò (nee) ya biyà hàrÄaajì-n. T. PRT 3SG.REL.PERF pay taxes-DEF ‘TANKO paid the taxes.’

Recall from section 2 that the absolute-relative distinction within the auxiliary paradigm is only attested in the perfective and continuous aspect. In the future and habitual aspect, the auxiliary has the same form independent of focus fronting. If nee/cee was a focus marker, one might expect it to be obligatory when focus is not marked by other morpho-syntactic means (word order, relative auxiliary), such as subject focus in the subjunctive,

246 Katharina Hartmann and Malte Zimmermann future, and habitual aspect. However, nee/cee may be absent even then. (7) illustrates subject focus in the future aspect. 3 (7)

Q: Wàanee nèe 4 zâi tàfi PRT FUT.3SG go Who ‘Who will go to Germany?’ A: Audù zâi tàfi Jamùs. go Germany Audu FUT.3SG ‘AUDU will go to Germany.’

Jamùs ? Germany

In light of these data, the hypothesis that the particle nee/cee is a focus marker appears to be unwarranted. Its possible absence in sentences with no other morpho-syntactic signs of focus marking suggests that the primary function of the particle is not that of a focus marker. In section 4, we present an alternative analysis showing that the presence of nee/cee adds a conventional implicature which leads to an exhaustive interpretation of the focus. We will argue that the particle is a focus-sensitive exhaustivity marker, rather than a syntactic focus marker. 3.2. In-situ focus In section 2, we pointed out that focus constituents do not have to be fronted but may remain in their base position. In-situ focus is quite frequent, it even represents the predominant focus-strategy for newinformation focus (cf. Hartmann and Zimmermann 2007). In addition, we argued that in-situ focus need not be marked at all. As concerns the particle nee/cee, it is optional with in-situ foci as well although it seems to occur much less frequently with these. If nee/cee appears, it generally follows the in-situ focus (Jaggar 2001:497). There seem to be two positions for nee/cee with in-situ focus. It can either appear in the sentence-final position (8-A1), or “at the end of the core sentence but before adverbial adjuncts or complements”, cf. (8-A2) (Newman 2000:546). (8)

Q: Mèenee nèe Tankò ya sàyaa PRT T. 3SG.REL.PERF buy what ‘What did Tanko buy at the market?’ A1: Tankò yaa sàyi [NP kàazaa][PP à chicken at T. 3SG.ABS.PERF buy

à at

kàasuwaa ? market

kàasuwaa] nè. market PRT

Exhaustivity marking in Hausa: A reanalysis of the particle nee/cee 247

A2: Tankò yaa sàyi [NP kàazaa] nèe [PP à kàasuwaa]. 5 ‘Tanko bought CHICKEN at the market.’ Notice that the particle does not have to be adjacent to the in-situ focus, see (8-A1) where the particle follows the locative adverbial, which belongs to the informational background. In the following, some further examples are presented that provide more evidence for the two particle positions with in-situ focus. First, when the right edge of the focus extends to the right periphery of the clause, nee/cee has to appear in clause-final position. This is shown for in-situ object focus (9), locative focus (10), predicate focus (11), and sentence focus (12). (9)

Q: Mèenee nèe Audù PRT A. What ‘What did Audu buy?’ A: Audù yaa sàyi A. 3SG.ABS.PERF buy ‘Audu bought a RING.’

ya sàyaa ? 3SG.REL.PERF buy [NP zoobèe] ne. ring PRT

(10) Q: (A) ìnaa nèe Tankò ya sàyi kiifíi ? PRT T. 3SG.REL.PERF buy fish where ‘Where did Tanko buy fish? A: Tankò yaa sàyi kiifii [PP à kàasuwaa] nè. at market PRT T. 3SG.ABS.PERF buy fish ‘Tanko bought fish AT THE MARKET.’ (11) Q: Mèe Hàwwa ta yi ? do what H. 3SG.F.REL.PERF ‘What did Hawwa do? A: Hàwwa taa [VP yankà naamàa] ne. H. 3SG.F.ABS.PERF cut meat PRT ‘Hawwa CUT THE MEAT.’ (12) Q: Mèe ya fàaru ? what 3SG.REL.PERF happen ‘What happened?’ A: [IP Muusaa yaa yi M. 3SG.ABS.PERF do ‘MUSA TALKED TO ME.’

minì màganàa] ne. me speech PRT

248 Katharina Hartmann and Malte Zimmermann The particle can also follow an in-situ focus in non-final position, cf. (8A2), as well as (13): (13) Q: Mèenee nèe màkaanikèe ya gyaaràa PRT mechanic 3SG.REL.PERF repair what ‘What did the mechanic repair at the garage?’ A: Màkaanikè yaa gyaarà [NP mootàr] car mechanic 3SG.ABS.PERF repair ‘The mechanic repaired the CAR at the garage.’

à gaareÄ ejì ? at garage nee PRT

à g. at g.

In addition, nee/cee can associate with the focus at a distance, as already observed in connection with (8-A1). A further example is given in (14). (14) Q: Wàacee cèe ka ganii à makarÄantar ? who.fem PRT 2SG.REL.PERF see at school ‘Whom did you see at school?’ A: Naa gaa [NP Dèelu] à makarÄantar nè. D. at school PRT 1SG.ABS.PERF see ‘I saw DELU at school.’ The fact that the particle does not have to follow the focus immediately corroborates our conclusion from section 3.1 that nee/cee is not a typical focus marker. Grammatical markers are usually adjacent to the constituent they mark. We propose that the position of the particle is not primarily determined syntactically (as in Green 1997), but follows from prosodic requirements instead: Nee/cee always occurs before a prosodic phrase boundary. 6 In Hausa, there are obligatory phrase boundaries between an ex-situ focus constituent and the rest of the clause, and between the direct object and subsequent embedded clauses and/or adverbials (cf. Leben, Inkelas and Cobler 1989). 7 As it happens, these are exactly the positions where nee/cee appears. It goes without saying that the end of a sentence also demarcates a prosodic boundary, hence, the occurrence of clause-final nee/cee is predicted here, too. That the particle is sensitive to its prosodic environment receives further support from the fact that it is sensitive to another phonological property of the preceding material, i.e. its tone. Recall that nee/cee has polar tone, a tone with opposite pitch to the preceding one. Note that there is no prosodic phrase boundary between the verb and the object NP in transitive sentences. It is therefore not surprising that nee/cee

Exhaustivity marking in Hausa: A reanalysis of the particle nee/cee 249

is banned from this position. This restriction holds even if the verb is narrowly focused. Such cases are illustrated in (15) and (16). If the particle is present, it must appear after the direct object. 8 (15) Q: Mèenee nèe màkaanikèe ya yi PRT mechanic 3SG.REL.PERF do what wà mootàr à gaarÄeejì ? with car at garage ‘What did the mechanic do with the car at the garage?’ A: Màkaanikèe ya [V gyaarà] (*nee) [NP mootàr] (nee) à gaarÄeejì. ‘The mechanic REPAIRED the car at the garage.’ (16) Q: Mèe Tanko ya yi wà hàraajì-n ? taxes-DEF what T. 3SG.REL.PERF do to ‘What did Tanko do with the taxes?’ A: Tanko yaa [V biya] (*nèe) hàraajìn (ne). Tanko PAID the taxes.’ The examples in (15) and (16) suggest a close structural relationship between the verb and the object. It might seem unexpected that verb focus does not lead to a restructuring of the prosodic structure, as it happens for instance in some Bantu languages (Kanerva 1990). But recall from section 2 that in-situ focus is absolutely unmarked, even prosodically. Hence, insitu focus has no repercussion on the prosodic structure in Hausa, and the tight connection between verb and object remains even under verb focus. Nee/cee may also occur between the indirect and the direct object in double object constructions, cf. (17). On a prosodic account, this is expected given that “there is typically a phrase boundary between the two objects of double object constructions” (Inkelas & Leben 1990:19). (17) Q: Wàacee cèe Ìbrahìm ya bai wà kud’ii ? whom.fem PRT I. 3SG.REL.PERF give to money ‘To whom did Ibrahim give the money?’ A: Ìbrahìm yaa bai wà tsoohuwarsà nee kud’ii. PRT money I. 3SG.ABS.PERF give to mother ‘Ibrahim gave the money to his MOTHER.’

250 Katharina Hartmann and Malte Zimmermann We conclude this section with a further observation. A clause-final particle is incompatible with ex-situ focus. Such examples are consistently judged ungrammatical. This is shown for ex-situ subject and object focus: (18) Q: Wàanee nèe ya zoo ? PRT 3SG.REL.PERF come who ‘Who came?’ A:*Audù ya zoo nè. A. 3SG.REL.PERF come PRT ‘AUDU came.’ (19) Q: Mèenee nèe Harúuna ya kaawoo dàgà Jamùs ? PRT H. 3sg.rel.perf bring from Germany what ‘What did Haruna bring from Germany? A:*ěeediyòo Hàruunà ya kaawoo (dàgà J.) ne. radio H. 3SG.REL.PERF bring from G. PRT ‘Haruna brought a RADIO from Germany.’ Our language consultants unanimously agreed that (18A) and (19A) are only grammatical as yes-no questions where the final particle functions as a question tag. 9 A declarative reading of these sentences is not available. At present, the source of this additional restriction is mysterious to us. The data in (18) and (19) appear to fall neatly under the syntactic account proposed by Green (1997) in (5): As nee/cee heads the FP, it must be rightadjacent to the fronted focus constituent in Spec,FP. On the other hand, Green’s analysis does not easily account for the sentence-internal occurrences of nee with in-situ focus in (13) and (15). We will leave this matter open for further research. To summarize, Hausa has a particle nee/cee which optionally appears after the focus constituent, whether in-situ or ex-situ. The properties of the particle described in the present section lead us to assume that it does not behave like a typical focus marker. Typical focus markers, as employed in many other African languages, are obligatory. They consistently mark the focus in a sentence. The particle nee/cee, on the other hand, is optional even if focus is not marked by any other means. Moreover, it can associate with the focus at a distance. This property is typical of focus-sensitive particles but not of grammatical focus markers. We conclude that focus in Hausa does not imply the presence of nee/cee. Rather, the reverse holds: if

Exhaustivity marking in Hausa: A reanalysis of the particle nee/cee 251

nee/cee occurs, a focus must occur to its left. Since such a dependency on focus is typical of focus-sensitive particles, we conclude that nee/cee is a focus-sensitive particle, rather than a focus marker.

4.

Nee/cee as a focus-sensitive exhaustivity marker

As we concluded in the last section, the distribution of nee/cee is not primarily determined by structural factors. Instead, we will argue that its occurrence is motivated by semantic considerations only. More precisely, we show that the presence of nee/cee introduces a conventional implicature triggering an exhaustive focus interpretation. 4.1. Green’s (1997) observation In her dissertation, Green (1997) observes a semantic distinction between cases of focus fronting where nee/cee is present and cases where it is absent. Usually, a sentence containing a focus may be followed by an afterclause that introduces an alternative to the focus constituent. This is shown for Hausa in (20) (from Green 1997:29). The fronted focus à kân teebùr ‘on the table’ is extended in the afterclause by another PP. Such an extension is illicit if the fronted focus is followed by nee. If nee is present, the focus receives an exhaustive interpretation: A focus constituent is interpreted exhaustively if the property denoted by the backgrounded part of the clause holds of the entity denoted by the focus constituent, and only of this entity. With respect to (20), this means that the books are put on the table and nowhere else. (20)

À kân teebùr (*nee) sukà sâ lìttàttàfai, upon table PRT 3PL.REL.PERF put books dà kuma cikin àkwàatì. and also inside box ‘They put the books on the table, and also inside the box.’

Green (1997) accounts for this observation by treating nee/cee as an exhaustive focus marker. In the following sections, we provide new data that corroborate Green’s claim that nee/cee adds exhaustivity to the semantic interpretation. As pointed out above, though, we analyze nee/cee as a focus

252 Katharina Hartmann and Malte Zimmermann sensitive exhaustivity marker, rather than as a proper focus marker. In what follows, we therefore gloss the particle as EXH for exhaustivity marker. 4.2. *Nee/cee in non-exhaustive contexts The data discussed in this section have in common that the focused entity is not the only one satisfying the property denoted by the background, to the effect that an exhaustive interpretation of the focus becomes impossible. This is achieved by adding an afterclause in which the same backgrounded property is predicated of an alternative value. In all such contexts, the particle nee/cee is illicit. Notice first that we were able to reproduce the facts discussed in (20). If the focus in the main clause is followed by nee/cee, extension by an alsophrase is excluded. The presence of nee/cee excludes all focus alternatives except for the focused entity itself. In (21A), nee forces the interpretation that nobody else apart from Musa returned from Kano. Similarly in (22), no additional individuals may be added to the denotation of the predicate satisfying the focused object, if this is followed by cee. (21) Q: Wàa ya daawoo dàgà Kano ? who 3SG.REL.PERF return from Kano ‘Who returned from Kano?’ A:#Musa nèe ya daawoo dàgà Kano EXH 3SG.REL.PERF return from Kano M. dà kuma Hàliimà cee ya daawoo dàgà Kano. EXH 3SG.REL.PERF return from Kano and also H. ‘MUSA returned from K. and HALIMA, too, returned from K.’ (22)

Hàwwa (#cèe) mukà ganii. EXH 1PL.REL.PERF see H. Kuma mun ga Hàliimà dà Dèelu. and D. also 1PL.PERF see H. ‘We saw HAWWA, also we saw Halima and Deelu.’

The examples in (20) to (22) show that the meaning component introduced by nee/cee cannot be cancelled. This suggests that nee/cee introduces a conventional implicature in the sense of Karttunen & Peters (1979). Second, nee/cee is illicit when the focus denotes an entity in a domain that is explicitly introduced as containing more than the focused entity, as

Exhaustivity marking in Hausa: A reanalysis of the particle nee/cee 253

illustrated in the following examples. If, as in (23a), a pluralic group is introduced (mutàanee dà yawàa ‘many people’), a focus with nee cannot pick a unique individual from this group (23b). In the absence of nee, the focused entity can be one among others in the denotation of the predicate. This is emphasized by the possibility of the additive particle maa ‘also’. (23) a. Naa san mutàane dà yawàa know people many 1SG.PERF dà sukà sayar dà àyàbà à kàasuwaa. that 3PL.REL.PERF sell banana at market ‘I know many people that sold bananas at the market.’ b. Maalàm Shehù #née / (maa) ya sayar dà àyàbà. EXH also 3SG.REL.PERF sell bananas Mr. S. ‘MR. SHEHU nee / (also) sold bananas.’ Assuming that nee/cee is an exhaustivity marker, the infelicity of (23b) with nee follows directly: The presence of nee in this sentence indicates that the property under discussion, i.e. the selling of bananas, only holds of a unique individual. This is in contradiction with the plural group introduced in (23a). Third, nee/cee is also illicit in mention-some environments. Consider the following context and the subsequent question-answer pair. (24)

Context: Musa knows that many students have passed last year’s exam. In order to prepare for this year’s exam, Musa wants to talk to one of them beforehand. (He has no time to talk to all of them). Unfortunately, Musa does not know who passed the exam, but he does know that his friend Amadu knows everybody who passed. Therefore Musa addresses Amadu in the following way: M: Kaa san wad’àndà sukà ci jar r àbâwaa ? 3PL.REL.PERF eat exam you know who.PL ‘Do you know who passed the exam?’ A: Î, dàgà ciki Ùmarù #nee/ maa ya ci j. EXH also 3SG.REL.PERF eat exam yes from among U. ‘Yes, among them UMARU passed the exam.’

254 Katharina Hartmann and Malte Zimmermann Amadu mentions to Musa one of the students that passed the exam last year. In the answer, he cannot use the exhaustivity particle nee after the focused subject since this would entail that only Umaru and nobody else passed. This would contradict the contextual condition that both, Musa and Amadu, know that many students were successful in the exam. The infelicity of nee/cee in mention-some contexts can be mended by means of accommodation: the property under discussion is specified in such a way that it applies to a unique individual, in congruence with the exhaustivity requirement. Reconsidering Amadu’s answer, the per se infelicitous presence of nee can trigger an accommodation such that the property under discussion is not only that of passing the exam but that of passing it in a special way, e.g. with the highest or lowest score. This property can now apply to the unique individual Umaru, as shown in (24’): (24’) M: Kà fàÍaa minì: Wàa ya ci jar r àbâwaa ? 2SG.SUBJ tell me who 3SG.REL.PERF eat exam ‘Tell me: Who passed the exam?’ A: Ùmarù nee ya ci jar r àbâwaa Umar EXH 3SG.REL.PERF eat exam ‘UMAR passed the exam (with the highest/lowest score etc.).’ A similar observation holds with respect to example (23). Below is a slightly extended context, which is followed by a question-answer pair. (25)

Context: Maalam Haruna wants to buy bananas at the market. He knows that there are many people selling bananas but not who exactly. He does not have much time and only wants to get the name of one of them. Therefore he asks his friend Maalam Shehu: sayar dà àyàbà ? H: Kaa san waÍàndà sukèe 2sg.masc know who.PL 3PL.REL.CONT sell bananas ‘Do you know who sells bananas?’ S: Îi, dàgà ciki Hamiidù nee yakèe sayar-waa. yes from among Hamidu EXH 3SG.REL.CONT sell-NMLZ ‘Yes, among them HAMIDU always/certainly sells bananas.’

Again, nee may follow the focused subject in Mr. Shehu’s answer, even though an exhaustive focus interpretation contradicts the mutual knowledge of Haruna and Shehu that many people sell bananas at the market. And

Exhaustivity marking in Hausa: A reanalysis of the particle nee/cee 255

again, the presence of nee can be licensed by accommodation, leading to an inherent quantification over times. It is understood that, among all the banana-sellers at the market, Hamidu always sells bananas. Thus, we are faced with a methodological problem to be reckoned with: due to the possibility of accommodation, native speakers will often judge nee/cee in mention-some contexts as acceptable. 4.3. Inferences based on (strong) exhaustivity The following example is a variation of (24). Recall that the context given required a non-exhaustive interpretation of the focus. Accordingly, nee/cee was illicit (without accommodation). (26)

Context: A student D’ (as in Íaalìbii ‘student’) who is anxious that he might have failed a test approaches teacher M (as in maalàamii ‘teacher’) and asks: ‘Can you tell me whether I have passed or not?’ Unfortunately, teachers are by law forbidden to tell a student directly about his or her result. However, there is no law forbidding them to talk about other students’ performances. D’: (Koo) naa ci jar r àbâwaa ? Q 1SG.PERF eat exam ’Have I passed the exam?’ M: Bà zâ-n gayà makà ba NEG FUT-1SG tell you NEG àmmaa Musà (nee) bà-i ci jar r àbâwaa ba. but M. PRT NEG-3SG eat exam NEG ’I will not tell you, but MUSA (nee) has not passed the test.’ Æ with nee: D’can assume that he has passed. Æ without nee: D’ cannot find out anything about himself.

The context in (26) allows for nee in the answer in principle. However, the amount of information differs depending on whether or not the teacher decides to use the particle. If the particle is absent after the focus constituent Musa, the student learns about Musa’s result, but he cannot draw any conclusions concerning his own score. If the particle is present, the student can deduce that he passed the exam in the following way: Since the particle marks the focus as exhaustive, Musa must be the only student who did not pass. The student D’ can therefore infer that he must have passed the test,

256 Katharina Hartmann and Malte Zimmermann although this is not explicitly asserted. If nee/cee were an optional focus marker, there should be no asymmetry in interpretation between the two variations. More precisely, the presence of nee/cee should not allow for any inference which is based on exhaustivity. 4.4. Nee/cee and other exhaustivity markers With adverbial exhaustivity markers, such as kawài ‘just, only, merely, simply’, kaÍai ‘only, alone’, or sai ‘only, just, except’, nee/cee is typically or often (Newmann 2000:190, Jaggar 2001:511) omitted. ganii. (27) a. Sai Gar ba mukà only Garba 1PL.REL.PERF see ‘It’s only Garba we saw.’ b. Hakà kawài zaa-kà yi. FUT-2SG do this only ‘That is just what you have to do.’ The analysis of nee/cee as an exhaustivity marker predicts the typical omission of nee/cee with adverbial exhaustivity markers on grounds of redundancy. When present, nee/cee can serve to reinforce kawái or kaÍai (Jaggar 2001:511). In contrast, an analysis of nee/cee as a plain focus marker leaves the highly restricted occurrence of nee/cee with other exhaustivity markers unexplained. Even though the adverbial exhaustivity markers kawái or kaÍai ‘only’ and nee/cee have similar semantic effects, the two kinds of expressions are not identical in meaning. It shows that nee/cee is semantically weaker than the adverbial exhaustivity markers. Compare (26-M) above, with nee present, to (28-M), with nee replaced by kawài. According to our consultant’s judgments, the difference between the two answers is the following: (28-M) asserts that only Musa has not passed the exam, so that the student knows for sure that he has passed, while (26-M) (with nee) makes the student only assume that he must have passed. (28) D’: (Koo) naa ci jar r àbâwaa ? Q 1SG.PERF eat exam ’Have I passed the exam?’

Exhaustivity marking in Hausa: A reanalysis of the particle nee/cee 257

M: Bà

zâ-n gayà makà ba àmmaa FUT-1SG tell you.m NEG but Musà kawài bà-i ci jar r àbâwaa ba. M. only NEG-3SG eat exam NEG ’I will not tell you, but only MUSA has not passed the test.’ Æ Student knows for sure that he has passed. NEG

The difference in interpretation between the mininal pair (26-M) and (28M) shows that the adverbials kawài and kaÍai introduce exhaustivity into the assertion as part of their truth conditions. The exhaustivity marker nee/cee, on the other hand, is weaker in that it does not add exhaustivity to the assertion. Nee/cee only adds a conventional implicature to this effect. It therefore does not translate as ‘only’. (Often, it does not translate at all, which might also have led to the erroneous impression that it is a grammatical focus marker.) The presence or absence of nee/cee does not change the truth-conditions of clauses. However, if nee/cee is dropped, the exhaustivity effect disappears. This shows that the semantic effect is detachable. That the semantic import of a lexical item is not cancelable, but detachable is a typical property of conventional implicatures. We therefore conclude that nee/cee triggers a conventional implicature. Putting the results of this section together, we assume the following meaning of nee/cee (where S stands for the clause containing nee/cee): (29)

[[nee/cee S]] = [[S]] = p defined iff f 0 (Æ focus-sensitivity) i. [[S]] z { [[S]] } (Æ exhaustivity) ii. p’  [[S]]f : Âp’ Æ p’ = [[S]]0

Nee/cee is a propositional operator that denotes a partially defined identity function: When applied to an arbitrary clause denoting the proposition p, it gives back the value p iff (i.) S has a non-trivial focus value (i.e. it contains a focus) and (ii.) the only focus alternative that is true is p. The first clause accounts for the focus-sensitivity of nee/cee, the second for the exhaustivity effect. Finally, by comparing the paradigms of focus-sensitive particles in Hausa and English (or German) we observe that the Hausa paradigm is more complete. While English only has a truth-conditional focus particle with universal force (only), Hausa has both truth-conditional particles (kawài, kaÍai) as well as a non-truth-conditional particle (nee/cee) with universal force.

258 Katharina Hartmann and Malte Zimmermann 4.5. Summary In this section, we have presented ample evidence in support of the claim, originally hinted at in Green (1997), that nee/cee is an exhaustivity marker. The presence or absence of nee/cee in a clause has semantic effects beyond the introduction of those presuppositions that are usually associated with focus: Nee/cee exhibits typical exhaustivity effects. First, it is infelicitous or highly marked when the context suggests non-exhaustivity of the focus domain. Second, it is typically left out in the presence of other exhaustivity markers, such as kawài or kaÍai ‘only’. Nee/cee triggers an exhaustivity effect by means of a conventional implicature, and, unlike only, not as part of its truth-conditions. Finally, like exhaustivity markers in other languages, nee/cee is focus-sensitive, which accounts for its dependence on focus. Being focus-sensitive, nee/cee can associate with focus constituents at a distance, accounting for the non-adjacency with in-situ foci (see also section 3.2). Altogether, these properties make an analysis of nee/cee as a purely grammatical focus marker highly implausible. 10 Finally, the optionality of the exhaustivity marker nee/cee with ex-situ foci and in wh-questions may have repercussions for the theory of focus and questions in general. First, the dependency of an exhaustive interpretation of ex-situ foci on the presence of nee/cee shows that exhaustivity is not structurally encoded in Hausa. In this respect, Hausa differs from Hungarian, where, following É. Kiss (1998), exhaustive focus is always fronted. Second, the optionality of the exhaustivity marker nee/cee in wh-questions (see n.4 in section 3.1) suggests that wh-questions are not inherently exhaustive in Hausa. This is in contradiction to what has been claimed for questions in other languages by Groenendijk & Stokhof (1984) but in accordance with claims in Heim (1994) and Beck & Rullmann (1999).

5.

Conclusion

In this paper, we have argued that – despite first appearances – the Hausa particle nee/cee is not a grammatical focus marker, but a focus-sensitive exhaustivity marker. The particle nee/cee does not exhibit the typical properties of grammatical focus markers. Rather, its presence or absence is governed solely by semantic factors while its syntactic distribution seems to depend on prosodic factors. However, since nee/cee is focus-sensitive, its presence can serve as an indirect indicator for focus because focus-sensitive

Exhaustivity marking in Hausa: A reanalysis of the particle nee/cee 259

elements need a focus in order to be licensed. The lesson to be learned is that not every grammatical formative that frequently co-occurs with focus constituents is best analyzed as a grammatical focus marker. From a crosslinguistic perspective, Hausa seems to differ from many other West African languages in that it does not have a grammatical focus marker. At the same time, our findings for Hausa should be tested against other instances of apparently optional focus markers in other African languages in order to find out whether these elements are genuine grammatical focus markers, or not just focus-sensitive exhaustivity markers. Notes

1.

2.

3.

4.

We would like to thank our Hausa consultants Dan Asabe, Balarabe Zulyadaini, Rabiu Shehu, Umar Ibrahim, Sa’adatu Garba, Aisha Mahmud, and Mu’awiya Jibir for their help and cooperation. Many thanks to Daniel Büring for his helpful comments and suggestions. This article was written within project B2 “Focusing in Chadic Languages” as part of the SFB 632 “Information Structure”, funded by the German Science Association (DFG). We thank the participants of the international workshop “Topic and Focus: Information Structure and Grammar in African Languages”, held in Amsterdam in December 2004, for comments on a preliminary version of this paper, as well as Lutz Marten and Florian Schwarz for comments on the present version. Hausa is a Chadic language spoken primarily in northern Nigeria. The Chadic languages belong to the Afro-Asiatic family. With more than 35 million speakers, Hausa is the biggest representative of the Chadic group. We use the following abbreviations: 1,2,3 = person number markers, sg = singular, pl = plural, perf = perfective, cont = continuous, rel = relative, abs = absolute, fut = future, subj, = subjunctive, fem = feminine, masc = masculine, NEG = negation, NMLZ = nominalizer, PRT = particle, EXH = exhaustivity marker, DEF = definite. Notice that subject focus in the aspects under discussion is marked prosodically by local High-tone raising, as is the case with all other instances of ex-situ focus (cf. Leben, Inkelas and Cobler 1989). Given this, it could be argued that nee/cee is absent in (7A) because focus is marked prosodically after all. The question-pronouns for ‘who’ and ‘what’ can be either morphologically simple (wàa, mèe), or they can be complex (wàanee nèe, mèenee nèe). In the latter case, they contain the particle nee/cee. If the particle occurs in the question, its presence in the answer seems to be obligatory.

260 Katharina Hartmann and Malte Zimmermann 5. The judgements of our language consultants, all L1 speakers living mostly outside of the Hausa heartland, varied considerably concerning the gender of the particle with in-situ focus. Since kàazaa ‘chicken’ is feminine, some speaker preferred cee here. While cee is always obligatory with feminine singular ex-situ focus, the picture is not so clear with in-situ focus. We cannot offer an account for the gender variation other than contributing it to dialectal differences, cf., e.g., Abubakar (2001). 6. The prosodic unit that Leben, Inkelas and Cobler (1989) call an intonational phrase is referred to as phonological phrase in other approaches (cf., e.g., Nespor and Vogel 1986). The term accent phrase is also used sometimes, primarily for accent languages (e.g. Uhmann 1991). 7. The location of prosodic phrase boundaries can be tested by a number of prosodic processes, the application or blocking of which is sensitive to their presence, cf. Leben, Inkelas and Cobler (1989:47–49). 8. A similar observation has been made with respect to the Hausa discourse particle fa. As Zec and Inkelas (1990:369ff) show, fa can only appear at intonational phrase boundaries and is also excluded after verbs. 9. Note that double occurrences of nee/cee in declarative sentences are equally ruled out. This is surprising given the possibility of multiple whquestions in Hausa (i-Q), see also Green (1997:116), as well as the possibility to combine an ex-situ with an in-situ focus in the corresponding answer (i-A). (i) Q: Suwàa sukà ganii à ìnaa ? who.pl 3pl.rel.perf see at where ‘Whom did they see where?’ A: Muusaa (nèe) na ganii à kàasuwaa. PRT 1sg.rel.perf see at market M. ‘I saw MUSA at the MARKET.’ Double occurrences of nee/cee are expected to be grammatical as long as one particle follows the in-situ focus. However, the only possible reading of such sentences is that of a yes/no question where the “in-situ particle” is interpreted as a question tag, indicating a certain degree of uncertainty or suspicion: (ii) Muusaa nèe ya sha ruwaa nè? water Q M. PRT 3sg.rel.perf drink ‘Musa drank water, (didn’t he)?’ not: ‘MUSA drank WATER.’ (as an answer to ‘Who drank what?’) A tentative solution to the impossibility of double occurrences of nee/cee in declaratives would go as follows: Sentences with two instances of focus are marked and require strong contextual licensing, e.g. in form of multiple wh-questions (cf. i-Q). According to our consultants, the corresponding wh-question for (ii) Who drank what? has the strong presupposition that there are various people drinking various beverages. In section 4,

Exhaustivity marking in Hausa: A reanalysis of the particle nee/cee 261 though, we will argue that nee/cee triggers an exhaustivity implicature. As a result, the implicature of (ii), when interpreted as a declarative, would be that only Musa drank only water (and nobody else drank anything else). This implicature is incompatible with the presupposition of the licensing question, ruling out (ii) as an answer. 10. Note that nee/cee is always obligatory in predicative constructions, where the particle is usually described as a copula verb (cf. McConvell 1973, Green 1997, 2004, Newman 2000, Jaggar 2001). (i) and (ii) illustrate this for adjectival and nominal predicates: (i) Teebùr ÝanÝanèe *(nee). (ii) Nii Bàtuurìyaa *(cèe). table small COP I European.fem COP ‘The table is small.’ ‘I am a European.’ The obligatory occurrence of nee/cee in predicative constructions can be derived from the fact that predicates necessarily involve focus: in the standard case, a (new) property is predicated of a (given) entity (see Green 2004). The proposed analysis of nee/cee as an exhaustivity marker predicts that the property denoted by the predicate is the only property (under discussion) that holds of the subject. Further research has to show whether this prediction for predicative constructions holds in general.

References Abubakar, Abdulhamid 2001 Isoglosses in Hausa Dialectology. In Maiduguri Journal of Linguistic and Literary Studies (MAJOLLS) 3, Mairo Kidda Awak (ed.), 60– 83. Bearth, Thomas 1999 The contribution of African linguistics towards a general theory of focus: Update and critical review. Journal of African Languages and Linguistics 20. 121–156. Beck, Sigrid, and Hotze Rullmann 1999 A Flexible Approach to Exhaustivity in Questions. Natural Language Semantics 7. 249–297. É. Kiss, Katalin 1998 Identificational Focus Versus Information Focus. Language 74. 245– 273. Green, Melanie 1997 Focus and Copular Constructions in Hausa. Ph.D. dissertation, SOAS, London.

262 Katharina Hartmann and Malte Zimmermann Green, Melanie 2004 Predication, Equation and Information Structure: Evidence from Hausa Copular Sentences. Ms. University of Sussex. Green, Melanie, and Philip J. Jaggar 2003 Ex-situ and in-situ focus in Hausa: syntax, semantics and discourse. In Research in Afroasiatic Grammar 2 (Current Issues in Linguistic Theory), Jacqueline Lecarme (ed.), Amsterdam: Benjamins. 187– 213. Groenendijk, Jeroen, and Martin Stokhof 1984 Studies on the Semantics of Questions and the Pragmatics of Answers. Ph.D. dissertation, Universiteit van Amsterdam. Hartmann, Katharina, and Malte Zimmermann 2007 In Place - Out of Place? Focus in Hausa. In On Information Structure, Meaning and Form, Kerstin Schwabe and Susanne Winkler (eds.), 365–403. Amsterdam: John Benjamins. Heim, Irene 1994 ‘Interrogative Semantics and Karttunen’s Semantics for Know’. In Proceedings of the Ninth Conference of the Israeli Association for Theoretical Linguistics (IATL 1), Rhonna Buchalla and Anita Mittwoch (eds.), 128–144. Jerusalem: Akademon. Inkelas, Sharon, and William R. Leben 1990 Where phonology and phonetics intersect: the case of Hausa intonation. In Papers in Laboratory Phonology I. Between the grammar and physics of speech, John Kingston and Mary E. Beckman (eds.), 17–34. Cambridge: Cambridge University Press. Jaggar, Philip J. 2001 Hausa. Amsterdam: John Benjamins. 2004 More on in situ Wh- and Focus Constructions in Hausa. Ms., SOAS, London. To appear in: Chadic Linguistics/Linguistique Tchadique/ Tschadistik, vol. 3., Dymitr Ibriszimow, Henry Tourneux, and Ekkehard Wolff (eds.), Cologne: Rüdiger Köppe. Kanerva, Jonni M. 1990 Focus and Phrasing in Chichewa Phonology. New York: Garland. Karttunen, Lauri, and Stanley Peters 1979 Conventional Implicature. In Syntax and Semantics 11, Choon-Kyu Oh and David A. Dinneen (eds.), 1–56. New York: Academic Press. Leben, William R., Sharon Inkelas, and Mark Cobler 1989 Phrases and Phrase Tones in Hausa. In Current Approaches to African Linguistics, Paul Newman and Robert D. Botne (eds.), 45–61. Dordrecht: Foris. McConvell, Patrick 1973 Cleft Sentences in Hausa? A Syntactic Study of Focus. Ph.D. dissertation. SOAS, London.

Exhaustivity marking in Hausa: A reanalysis of the particle nee/cee 263 Nespor, Marina, and Irene Vogel 1986 Prosodic Phonology. Dordrecht: Foris. Newman, Paul 2000 The Hausa Language. New Haven & London: Yale University Press. Parsons, Frederick William 1963 The Operation of Gender in Hausa: Stabilizer, Dependent Nominals and Qualifiers. African Language Studies IV. SOAS, University of London: 166–207. Schachter, Paul 1966 A Generative Account of Hausa Ne/Ce. Journal of African Languages 5. 34–53. Tuller, Laurice 1986 Bijective Relations in Universal Grammar and the Syntax of Hausa. Ph.D. dissertation. UCLA. Uhmann, Susanne 1991 Fokusphonologie. Tübingen : Niemeyer. Zec, Draga, and Sharon Inkelas 1990 Prosodically Constrained Syntax. In The Phonology-Syntax Connection, Sharon Inkelas and Draga Zec (eds.), 365–378. Chicago: CSLI.

Part V Focus and related constructions

Narrative focus strategies in Gur and Kwa Anne Schwarz and Ines Fiedler

Abstract The paper considers certain morphosyntactically marked focus constructions in five Ghanaian languages which display special structural features in the out-offocus part of the sentence compared to the canonical clause. Such a phenomenon is widespread in West African languages, and these non-canonical structures are often labeled “relative” because they seem to be typical for relative clauses. However, while the parallels with relative clauses are never complete in the languages studied, there is a constant structural isomorphism with clauses encoding sequential events in narration. Consequently, it is suggested that these focus constructions should be analyzed as extra-clausal structures in which the non-focal information is presented as a narrative clause.

1.

Introduction

In this paper, we examine a phenomenon that is well known from several West African languages like Hausa, Fulfulde, and others (cf. Bearth 1993, Frajzyngier 2004), namely the parallels between morphosyntactically marked focus constructions on the one hand and relative and narrative clause morphology (and sometimes the morphology of other clause types) on the other. In Hausa for example, one can differentiate between two morphosyntactic aspectual codings: a canonical paradigm A and a paradigm B which is found in focus constructions, relative clauses, and – with respect to the perfective aspect – in narrative contexts. This second paradigm is often called the “relative” tense/aspect. Our investigation shows that parallels similar to these can also be found in languages of the Gur and Kwa group. It also shows that part of the focus constructions can be analyzed as morphosyntactically marked constructions containing a narrative clause. Our language sample consists of five Ghanaian languages which we investigated during our field research in summer 2004. We deal with two Gur languages: Buli and Dagbani which belong to different subgroups of the Oti-Volta branch and are spoken in the Northern area of Ghana, as well as

268 Anne Schwarz and Ines Fiedler with three Kwa languages of different subgroups: inland dialect of Ewe (Gbe), Asante dialect of Akan (Potou Tano), and Lelemi, which is treated as a Togo remnant or Ghana-Togo-Mountain language (na-Togo) which according to Rongier 1997 (cited in Blench 2001) are still under discussion as to whether they belong to Gur or to Kwa. The following grammatical features shared by all of the five languages will be relevant for our discussion: the strict SVO structure of canonical sentences and the lack of morphological agreement on the verb phrase in the presence of nominal arguments in simple sentences, with the exception of Lelemi. In this language, lexical subjects co-occur with verb initial agreement markers. We restrict the topic of this article to morphosyntactically marked constituent focus structures (ex-situ) in affirmative utterances where a focused nominal constituent takes sentence initial position. 1 The following findings concerning the parallels between these constructions and relative as well as narrative clauses emerge in our language sample: (1) there is no complete correspondence with (restrictive) relative clauses while (2) the correspondence with narrative clauses is surprisingly obvious. Our concern here is to give evidence for the latter by presenting first results of ongoing work. The similarity of ex-situ focus constructions with narrative clauses shows up primarily in non-subject focus constructions. We will discuss the implications of these observations for the analysis of ex-situ focus constructions and provide a diachronic approach to account for these patterns and the language-specific differences concerning the degree of grammaticalization of the narrative clause. The paper is structured as follows: First, we present evidence for the asymmetry between ex-situ subject und non-subject focus constructions (henceforth SF resp. NSF) (2.1.). Then we demonstrate the parallelism between ex-situ non-subject focus constructions (NSF) and narrative clauses (NAR) (2.2.). Following this, we discuss the narrative pattern found in ex-situ focus constructions and the various degrees of grammaticalization of narrative sequential clauses into parts of focus constructions for each of the sample languages (3.). Finally, we summarize our findings comparatively in the conclusion (4.).

2.

Recurring features of ex-situ focus constructions

In the following, we will demonstrate two observations concerning the structure of morphosyntactically marked focus constructions. 2

Narrative focus strategies in Gur and Kwa 269

2.1. SF vs. NSF asymmetry There is a constant structural asymmetry between SF and NSF constructions which shows up in several ways and to different degrees in the selected languages. We will start with weaker developed cases and then move on to languages showing stronger asymmetries. 2.1.1. Ewe The characteristics of ex-situ focus constructions in Ewe are as follows: First, the focused element can be marked by the focus marker (henceforth FM) ([)G"(cf. Ameka 1992) which is obligatory for subject focus and optional for non-subject focus with the exact constraints for the optionality not being clear yet. Second, there are special subject pronouns for 2nd and 3rd person singular which are used only in NSF constructions (2) while in SF no pronominals are preposed to the verb (1). SF (1)

NSF (2)



P"VUW"³C"³G" VUn"³G man-DEF-YE take-O:3sg ‘The MAN took it.’(not the woman) GF\(³G") YQ- ÇW DEP:3sg eat top(-YE) ‘He WON.’ (i.e., He was on TOP.) cf. the canonical sentence G" ÇW F\¸ 3sg eat top ‘He WON.’ (i.e., He was on TOP. ~ He WAS ON TOP.)

2.1.2. Akan In Akan, the constructions for both SF and NSF are characterized by two features. They obligatorily make use of the FM nà (cf. inter alia Boadi 1974, Bearth 2002) and in the perfective aspect they display tonal changes at the verb in the out-of-focus part: compare the falling tone at the verb in the focus construction in (3) with the low verb tone in the following ca-

270 Anne Schwarz and Ines Fiedler nonical sentence (cf. also the so-called “link tone” in Bearth 2002; cf. Schachter 1973 as well). In SF (3), an invariable expletive subject pronoun ('³), an anaphoric subject pronoun, is used alternatively. 3 In NSF (4), on the other hand, there is no general syntactic filler in the postverbal position. Instead, an anaphoric object pronoun has to be selected due to semantic criteria. While animates require a pronoun (cf. 4a), especially when human, inanimates do not allow one. In the absence of a pronominal object, the perfective verb in sentence-final position carries the suffix -[' (cf. 4b). SF (3)

'³[' CDTGYC" PQ" PC '³F¸º CFWC" PQ EXPL-COP old.woman DEF NA EXPL-eat beans DEF ‘It is the OLD WOMAN who ate the beans.’ cf. the canonical sentence CDTGYC" PQ" Fºº CFWC" PQ eat beans DEF old.woman DEF ‘The old woman ate the BEANS.’ ~ ‘... woman ATE THE BEANS.’

NSF (4) a. PG MTC"OC"P PC POSS:3sg dog NA ‘The man carried his DOG.’ b. PG DC"IG PC POSS:3sg bag NA ‘He carried his BAG.’

RCRC" PQ man DEF

UW"C carry

PQ O:3sg

n³UW"C³['" 3sg-carry-YE

2.1.3. Lelemi In Lelemi, the difference between subject and non-subject focus constructions lies above all in the verbal morphology. Lelemi has two sets of TMAmarkers: one used in simple tenses, the other in so-called “relative” tenses (Allan 1973). Not every simple tense has a counterpart in the relative tenses. The “relative” tenses (relative past, relative present, relative future, and relative present for verbs of state) appear in SF constructions. They display only one invariant form prefixed to the verb in each tense for all persons.

Narrative focus strategies in Gur and Kwa 271

The simple tenses, on the other hand, are used in NSF. They always indicate tense and person (incl. gender agreement) in form of verbal prefixes. The morphological coding device for NSF consists of the morpheme nà following the focused constituent (cf. the homophone FM in Akan). Yet, while some informants treated it as obligatory, others claimed that the syntactic marking, i.e. the sentence-initial position, already makes clear that this constituent is in focus. SF (5)

NSF (6)

a. nPCCDº WOYº R'" On-F¸ boy one only REL.PRS(dyn)-eat ‘Only ONE boy is eating an orange.’ b. WNGMW" n"On PC"³Fº woman DEF REL.PAST-eat ‘THE WOMAN ate them.’

not simple tense form: MWVW" *nn-Fº orange *n"³F

CMC"D" C"YnF" (PC) WNGMW n"On n"³F beans raw (NA) woman DEF 3sg.PAST-eat ‘The woman ate RAW BEANS.’ cf. the canonical sentence WNGMW" n"On n"³F CMC"D" woman DEF 3sg.PAST-eat beans ‘The woman ate BEANS.’ ~ ‘The woman ATE BEANS.’

2.1.4. Buli Buli has a preposed affirmative FM ká (negative suppletive FCC) in SF as well as in NSF. The affirmative morpheme seems to be optional in both constructions. In SF, the focused constituent is always followed by the conjunction NG while in NSF we typically find the conjunction VG. 4 In NSF, the occurrence of NGis less common but not totally excluded. Verb tone deviates in both focus constructions from that in simple sentences and depends on the respective conjunction. Buli has three paradigms distinguished by grammatical tone on dynamic perfective verbs in clausefinal position: a canonical paradigm A in simple sentences, a paradigm B after conjunction NG(SF), and a paradigm C that shows up in clauses with the conjunction VG (NSF). 5 In the out-of-focus part of both marked con-

272 Anne Schwarz and Ines Fiedler structions, there are no pronouns coreferent or agreeing with the focused constituents. SF (7)

(MC") O+¸ NG EJG0 (FM) 1sg:DISJ CNJ go ‘I went. ~ It is ME who went.’ (not you)

Paradigm B, not: *VG

cf. the canonical sentence O¸ EJG"0³MC" UC"PFGO Paradigm A 1sg:DISJ go-FM 6 Sandema ‘I went to SANDEMA.’ ~ ‘I WENT TO SANDEMA.’ NSF (8) a. (MC") UC"PFGO VG O¹ EJG0 Paradigm C, rare: NG (FM) Sandema CNJ 1sg:DISJ go ‘It is SANDEMA where I went.’ (not Navrongo) b. LQJP VG O+¸ HnD CNJ 1sg:DISJ slap John ‘It is JOHN whom I slapped.’ 2.1.5. Dagbani Dagbani provides a strong structural asymmetry between SF and NSF. The common SF construction contains a postponed syllabic nasal called “emphatic” by Olawsky (1999: 59f.) 7 . In NSF, the so-called FM MC (Olawsky 1999: 63) appears at the beginning of the out-of-focus part. Similar to Buli, both focus construction types are distinguished from each other by grammatical verb tone and deviate from the verb paradigm in simple sentences. In general, neither focused verb arguments nor other focused constituents are additionally expressed in the out-of-focus part. SF (9)

RC"¢C" OC"xC"O Dn"N³n Paradigm B woman DEF N call-O:3sg ‘The WOMAN called him.’ cf. the canonical sentence Q DnN³NC" IGQTIG Paradigm A 8 George sg call-FM ‘She called GEORGE.’ ~ ‘She CALLED GEORGE.’

Narrative focus strategies in Gur and Kwa 273

NSF (10)

IGQTIG MQ"³xQ" DnN+º Paradigm C George KA-3sg call ‘She called GEORGE.’ ~ ‘It is GEORGE whom she called.’

2.2. NSF-NAR parallelism Our second observation concerns the fact that there is a recurring formal parallelism between the out-of-focus part of ex-situ non-subject-focus constructions (NSF) and narrative clauses (NAR). In some of our sample languages, the narrative structure extends to SF as well. We use the term NAR for clauses that encode the succession of events in realis mood and that serve to continuatively develop the main story line. Labov regards this function as follows: “Each clause … describes an event that is understood to shift reference time, i.e. it follows the event immediately preceding it, and precedes the event immediately following it.” (1972, cited in Schiffrin 1994: 284). Syntactically, this succession of events is encoded iconically by coordinated clauses which have special, language-specific structural features: clausal conjunctions “and (then)”, particles, special pronouns, special verb forms, etc. In the following, we illustrate that the formal parallels show up in several ways in the selected languages. 2.2.1. Ewe In Ewe, one indication of the structural similarities between NSF and NAR consists in the use of special subject pronouns (restricted to 2nd and 3rd person singular) 9 in both constructions, as can be seen in ex. (11) (former (2)) and (12). Westermann (1930: 61) mentions that these pronouns are used “in the continuation of a sentence, or closely to connect one sentence with a preceding one”. Additionally, the FM ([)G"resembles a clause coordinating conjunction (G")[G" 10 ‘and (then)’ which is found in narrative contexts and which is often reduced to [G". NSF (11)

GF\(³G") YQ³ ÇW top(-YE) DEP:3sg eat ‘He WON.’ (i.e., He was on TOP.)

(= ex. 2)

274 Anne Schwarz and Ines Fiedler NAR (12)

ÇGX"C" NnP G"[G" YQ³ F\G child.DEF agree CNJ DEP:3sg arrive ‘The child agreed and started the journey.’

On" path

2.2.2. Akan Akan has a clausal sequential conjunction (é)nà with the meaning ‘and (then)’ (Bearth 2002) which is identical with the FM – with the restriction that the occurrence of the generally rare initial vowel has not been attested so far within focus constructions. In the perfective aspect, the verbal morphology including the verb tone changes in both clauses (cf. -F¸º³ in (13)– (14) vs. -Fºº³ in the canonical simple sentence), although this remains a matter of further research. 11 NSF (13)

CFWC" PC n³F¸º³['" beans NA 3sg-eat-YE ‘He ate BEANS.’

NAR (14) a. OC"COG" PQC"C CFWC" PC P³CFC"OHWQ F¸º³['" Maame cook beans CNJ POSS:3sg-friend eat-YE ‘Maame cooked beans and her friend ate them.’ b. RCRC" PQ" HCC PG DC" PC n³UW"C PQ" man DEF take POSS:3sg child CNJ 3sg-carry O:3sg ‘The man took his child and carried it.’ 2.2.3. Lelemi In Lelemi, NSF and NAR clauses show identical features: In both, the simple tense is used. Furthermore, the FM is homophone with the narrative conjunction ‘and (then)’ which coordinates two clauses, and we suppose that it is the same morpheme. It is segmentally identical with the “relative past” tense morpheme as well (cf. ex. 5c). NSF (15)

CMC"D" C"YnF" (PC) WNGMW n"On n"³F beans raw (NA) woman DEF 3sg.PAST-eat ‘The woman ate RAW BEANS.’

(= ex. 6)

Narrative focus strategies in Gur and Kwa 275

NAR (16) 

‘The youngest child went …’ PCW"³V¹ WNW I'y'y CNJ 3sg.PAST-take road right ‘and he took the right road.’

n"On DEF

2.2.4. Buli Buli, too, displays a striking parallel between NSF and NAR. First, in both structures, the same clausal conjunction VG ‘and’ is used. Second, the identity pattern stretches onto the verb in both: it bears the same grammatical tone (paradigm C) after VG and differs thus from (i) the canonical paradigm A and (ii) the paradigm B that is found in combination with the marker NG (i.e. mainly with SF, cf. 2.1.4.). It should be noted, however, that beyond the perfective aspect some structural peculiarities can be observed (cf. 3.4.). NSF (17)

NAR (18)

(MC") UC"PFGO VG YC (FM) Sandema CNJ 3sg ‘It is SANDEMA where he went.’

EJG0 go

‘... and his mother was happy with him’ VG DC FºI LGPVC0C... CNJ 3pl cook soup.DEF ‘and they cooked the soup ...’

2.2.5. Dagbani Finally, Dagbani also affirms the parallel pattern between NSF and NAR convincingly. The morpheme kà which follows the focused constituent in NSF constructions has a clausal conjunction counterpart kà ‘and’ in narrative contexts. Furthermore, the grammatical tone of dynamic verbs in such clauses differs in the same way from the subject congruent verb tone in simple clauses, irrespective whether we deal with a real narrative context or a focus construction. The coding of the second clause in ex. (20b) demonstrates Olawsky’s (1999: 44) observation that if the subject of the clause introduced by kà is coreferent with the subject of the preceding clause, it has to be elided.

276 Anne Schwarz and Ines Fiedler NSF (19)

[¸N‹" OC"xC" Pº MQ"³xQ" Fº in KA-3sg eat house DEF ‘In the HOUSE she ate.’

NAR (20) a. ‘... and the mother sent the youngest child’ MC D¸¸ OC"xC" EJC0 CNJ child DEF go ‘and the child went ...’ b. RC"¢C" OC"xC" FC"C"³xNC" RGVGT MC" x0OG"³Q push-FM Peter CNJ hit-O:3sg woman DEF ‘The woman pushed and hit Peter.’ not: MC*Q0OG³Q 2.3. Overview Summarizing our observations concerning affirmative ex-situ focus constructions and narrative clauses, the ex-situ focus construction minus the focus constituent itself (F) resembles more or less a narrative clause. All of the five Ghanaian Gur and Kwa languages considered here show this phenomenon and most of them display the parallelism especially between NSF and NAR. Table 1 Ewe

Akan

Lelemi

Buli

Dagbani

SF

F (y)é

('[') F nà expl. ' tone change

F relative tense

(ká) F (à)lG tone B

FN tone B

NSF

F (y)é dep. pron.

('[') F nà tone change

F nà simple tense

(ká) F (à)tè tone C

F kà tone C

NAR

... (é)yé dep. pron.

... (é)nà tone change

... nà simple tense

... (à)tè tone C

... kà tone C

Thus, we conclude that the parallelism between the out-of-focus part of morphosyntactically marked (N)SF and narrative clauses (NAR) is no coincidence but is due to a systematic narrative basis of the respective focus constructions.

Narrative focus strategies in Gur and Kwa 277

3.

Narrative hypothesis

From the structural analyses above, it is evident that the parallelism between (N)SF and NAR is a systematic pattern. We propose that in fact a narrative clause constitutes the non-focal part of such ex-situ focus constructions and suggest the following structure, here exemplified for the nonsubject focus construction: (21)

Extra-clausal structure of NSF focal part: DP / clause [Non-Subject+/-focus marker]

non-focal part: narrative clause [CNJ 12 Subject Predicate]

In a simple transitive sentence with only one complement, the predicate in the non-focal part contains the verb and only in certain cases also pronominal objects. The conditions of the pronouns’ occurrence are the same as in real sequential narrative contexts and they are language-specifically determined by the animacy hierarchy. We therefore do not follow the movement hypothesis as suggested, for example, by Aboh (2004) and Green (1997) for Gbe and Hausa respectively. Like in the cleft hypothesis assumed for various languages (Schachter 1973, Heine/Reh 1984, Givón 1990, Lambrecht 2001, Drubig 2003), our proposal considers the constructions as involving a clausal boundary, but unlike in the cleft hypothesis the non-focal part does not represent a relative clause. Even if there are close structural correspondences to restrictive relative clauses in most of these languages, some essential elements like the obligatory relative pronoun resp. particle and the common clause-final determiner are missing. For a more detailed discussion of the incomplete relation between relative clauses and the non-focal clause of ex-situ focus constructions compare Fiedler & Schwarz (2005: 120–126). While the non-focal part of the ex-situ construction undoubtedly represents a clause of its own, the focal part has in most occurrences no clausal status. For some languages, it is analyzed as a reduced clause with a zero copula in affirmation (cf. Drubig 2003: 32f).

278 Anne Schwarz and Ines Fiedler 3.1. Ewe The synchronic FM yé differs from the conjunction (G")[é ‘and (then)’ only in the prefixed vowel. Thus, we assume – also based on the evidence in the other languages of the sample – that the FM has developed out of the conjunction leading to a functional split between conjunction and focus marker. In normal speech, the latter is usually further eroded to vowel é and suffixed to the preceding DP, with phonological reduction being a common feature of grammaticalization processes. Additionally, Ewe displays a homophone morpheme yé occurring in nominal predications like ‘It is a pen.’ – RGP[G". Here, its function is comparable to a copula verb but not identical with the identifying/classifying copula in Ewe 13 . This structure can therefore be seen as an elliptical focus construction (cf. Ameka 1992) which the background information is dispensed with. The narrative analysis is supported by the requirement for special subject pronouns (in some persons) in NSF. Note that the correspondence becomes less close in SF where no coreferent subject pronoun is allowed in the non-focal part even though it is required in narrative contexts. 3.2. Akan An argument for the narrative hypothesis in Akan is the identity of FM and narrative conjunction, at least if we disregard the conjunction’s rare initial vowel. Our informants perceived nà in focus constructions still as conjunction so that if there is a certain degree of grammaticalization at all, as suggested by its description as FM by some authors, this could only be due to a functional split in the very inceptive stage. The first part of the extra-clausal focus construction is often only represented by a DPAlternatively, the initial part starts with '[', i.e. an expletive pronoun plus copula verb ‘to be’ (cf. ex. 3). ̓In the non-focal clause, the subject position must always be filled although variation between the anaphoric form and an expletive pronominal form is possible in SF as well as in narratives. 3.3. Lelemi In Lelemi NSF, the non-focal clause is formally identical with the narrative clause, i.e. any grammaticalization of the narrative clause is restricted to the

Narrative focus strategies in Gur and Kwa 279

functional level and has no effects on the structural level. Accordingly, we consider the morpheme nà in both functions a conjunction. In SF, the conjunction is missing and “relative” tense forms are used instead of the simple tense forms. Nevertheless, we can assume it to be present in one of the “relative” tenses/aspects. The TMA morpheme for the “relative past” tense is high toned ná. We analyze it as a conglomeration of the conjunction nà (with inherent Low tone) plus a High tone which is borne by the subject pronoun in the simple past. Like in Akan, the slot for the subject has to be filled even though this is only achieved by using an invariant form. 3.4. Buli In Buli, the narrative hypothesis is valid for the prototypical NSF construction which is formed with conjunction VG and tone paradigm C. Since these features are shared by sequential clauses in narrative contexts as well, the non-focal part in NSF can be regarded to be represented by a narrative clause. Some structural peculiarities support the proposed narrative clause type. While non-perfective verb forms are naturally excluded from sequential events in real narrative contexts, they are subject to certain accommodation in NSF constructions: in the imperfective aspect, dynamic verbs prefer a subjunctive encoding while stative verbs appear in the same tonal form as in canonical simple sentences. The SF construction, on the other hand, requires the conjunction NG which cannot be related to the narrative conjunction as such but is segmentally identical with the NP coordinating conjunction NG‘and, with’. This structural similarity among the two le-junctors might be an indication for a semantico-syntactically closer conceptualization of the SF construction as one single information structural unit compared to the evidently extraclausal NSF organization with VG. 3.5. Dagbani Like in Buli, the extra-clausal NAR pattern is restricted to NSF in Dagbani. There are no constitutional structural differences between the non-focal part in NSF construction and the basic NAR clause. The so-called “focus marker” MC (Olawsky 1999: 63) which requires verb tone paradigm C (for

280 Anne Schwarz and Ines Fiedler dynamic verbs) is in both contexts analyzed as conjunction by us though it has some potential for grammaticalization into a FM. Interesting is the parallel to Buli as the “emphatic” (Olawsky 1999: 59f.) marker N in SF resembles the DP coordinating conjunction (mì)nì~P ‘with, and’. It seems that Dagbani has a similar tendency towards a more intraclausal organization of SF compared to NSF and hence does not make use of the coordinative narrative pattern in SF.

4.

Comparative summary

As we have shown, there are striking similarities on the morphosyntactic level between focus constructions and narrative clauses. In all the languages we are concerned with here, this parallelism can be accounted for by acknowledging that the conjunction of the narrative clause is an appropriate device to link together focus constituent and non-focal part in a nonhierarchical way. In some of the languages, the inceptive stages of grammaticalization processes of the clause-initial conjunction into FM can be perceived, a grammaticalization chain that may even stretch further into a copula-like predicative morpheme, as noticed by Stassen (1997: 85). The actual stage of such grammaticalization chains in our sample languages is shown in table 2: Table 2 CNJ Ewe (é)yé Akan (é)nà Lelemi nà Buli (à)tè Dagbani kà

o

FM o yé, -é (nà) (nà) -(kà)

COP~PRD yé -----

Three of the languages, namely Akan, Lelemi 14 , and Dagbani, display the same pattern insofar as they have a conjunction which has been interpreted by some authors as a right-adjacent FM (Boadi 1974, Ameka 1992, Olawsky 1999). According to our research, the respective morphemes do have the potential for such a functional split but apparently that stage has not yet been reached since we could not notice any relevant categorial or structural changes of the conjunction towards a FM.

Narrative focus strategies in Gur and Kwa 281

As for the clausal Buli conjunction, there are no indications at all that it might take the grammaticalization path into a focus and then also into a predicative marker in the near future. Responsible for that is firstly its restriction to NSF, a fact that the Buli conjunction shares with the respective Dagbani conjunction. Secondly, the Buli clause conjunction is in affirmative focal contexts relatively often accompanied by the FM ká left to the focus constituent while such a frequently used affirmative counterpart is missing in Dagbani. If the focused constituent is negated, all five languages need, however, to use negative copula forms. We conclude that the rarer the copula forms in affirmation are the higher are the chances for reanalysis of the clausal conjunction as FM. Contrary to the rather inceptive stage of grammaticalization – if existent at all in most of the languages – the development in Ewe seems to have been longer. The original conjunction already shows signs of erosion when functioning as FM and often being suffixed to the constituent in focus. As noted in 3.3., in Lelemi, the conjunction nà has taken a special direction in grammaticalization. Together with the High tone borne by the subject prefix in other syntagmata, it has become a “relative past” tense marker in SF. Such development from a conjunction denoting the accomplishment of actions to a past marker was also shown by Hopper (1979) for Malay, an Austronesian language. Table 3 Lelemi

CNJ nà

o

“Relative Past” ná (m nà + ´ )

An interesting picture in our small language sample is displayed by the distribution of the narrative structures in ex-situ focus constructions, as shown in table 4. Table 4

Akan, Ewe

Lelemi

Buli, Dagbani

SF

SF

SF

NSF

NSF

NSF

282 Anne Schwarz and Ines Fiedler In the Kwa languages studied, including Lelemi, the narrative pattern is more or less overtly extending into SF constructions. On the other hand, in the two Gur languages studied here, SF constructions do not participate in the narrative pattern. Schwarz (in preparation) shows that in languages of this group SF rather tends to be represented by a syntactically more hierarchical intraclausal construction and that the distribution of the two ex-situ focus constructions can be accounted for on discourse organizational grounds taking into account the notion of topic. Our aim here was to propagate the existence of a constant narrative pattern in ex-situ focus constructions as an alternative to cleft and movement approaches. We showed that not only the in-focus part but also of the outof-focus part is important for an adequate analysis of ex-situ focus constructions. In this respect the functional load verb morphology, including tone, should not be underestimated in African languages. Abbreviations CNJ COP DEF DEM DEP DISJ DYN EXPL F FM NAR NSF O PAST POSS PRD PRS PROG REL SF STAT TMA

conjunction copula definite marker demonstrative (pronoun) dependent (pronoun) disjunctive (pronoun) dynamic (verb) expletive pronoun focus constituent focus marker narrative (clause) non-subject focus (construction) object past tense possessive (pronoun) predicator present (tense) progressive marker relative (tense) subject focus (construction) stative (verb) tense-mood-aspect

Narrative focus strategies in Gur and Kwa 283

Notes

This article was written within project B1 “Focus in Gur and Kwa Languages” as part of the SFB 632 “Information Structure” funded by the German Science Association (DFG). We thank our colleagues who commented the preliminary version of this paper which was presented at the international workshop “Topic and Focus: Information Structure and Grammar in African Languages”, December 2004, Amsterdam, and an anonymous reviewer for critical comments. 1. For practical reasons, we will use the term focus constituent here also in those cases in which only part of the clause constituent is focal. 2. Please note that the data is transcribed with surface tones and that versalia in the English translation indicate the respective focal part of the utterance. The following diacritics are used for tone marking: high tone ( ´ ), mid tone (  ), low tone ( ` ), and downstepped high tone ( !´ ). 3. Bearth et al. (2002) describe the change of the subject pronouns as restricted to human referents. This distribution is not supported by our data. 4. Both conjunctions are sometimes provided with an initial vowel (CNGCVG). This vowel occurs with other clause-inital conjunctions as well as with clause-initial serialized verbs and is always correlated with a prosodic break before the clause. 5. For clause-final verbs in the indicative perfective, paradigm B is characterized by an “instable rising tone” (Schwarz 2004: 38) and paradigm C by an invariable low tone. 6. Please note that in the canonical indicative sentence in which either the complement or the whole VP is focal FM MC"is encliticized to the verb. In ex-situ focus constructions, any postverbal occurrence of this morpheme is completely excluded. 7. There is another SF construction formed with post-subjectival NG"G" which is almost restricted to questions and not considered here. 8. In case of complement or VP focus, suffixed FM ³NC"occurs in Dagbani indicative sentences. Like Buli FM MC" this suffix is excluded from the postverbal position in ex-situ focus constructions. 9. This has already been noted by Duthie (1996: 53) and Ameka (2004: 17). 10. Concerning the tonal behavior of the conjunction, we follow Clements (1977: 172) who analyzes it as having overall high tones. The first high is raised to extra-high in the beginning of a clause, therefore giving it a extra-high – high contour which is often interpreted as high – mid contour (cf. Westermann 1954). 11. Bearth (2002) postulates the existence of a “link tone” on the verb as well as the existence of a so-called “dependent” morpheme [' in ex-situ focus

284 Anne Schwarz and Ines Fiedler constructions while our own data exhibit these two features in other contexts, too, including sequential events with clause-initial conjunction nà 12. In some languages (for example, Knnni, a Gur language closely related to Buli), the most common sequential narrative clause type as well as the out-of-focus clause occur without clause-initial conjunction (cf. Schwarz, in preparation). 13. Cf. Ndayiragije 1992 who gives evidence for the non-verbal status of the cognate FM w'in the closely related Gbe language Fon. 14. As can be seen in the table, Akan and Lelemi exhibit homophone morphemes which could be a result of borrowing from Akan to Lelemi since loans from Akan are common in the Togo mountain area.

References Aboh, Enoch Oladé 2004 The morphosyntax of complement-head sequences. Clause structure and word order patterns in Kwa: Oxford studies in comparative syntax. Oxford, New York: Oxford University Press. Allan, Edward Jay 1973 A grammar of Buem: The Lelemi language, School of Oriental and African Studies, University of London: PhD Thesis. Ameka, Felix 1992 Focus constructions in Ewe und Akan: A comparative perspective. In Proceedings of the Kwa comparative syntax workshop M.I.T. 1992, Chris Collins and Victor Manfredi (eds.), 1–15. Cambridge: MIT. 2004 Grammar and cultural practices: The grammaticalization of triadic communication in West African languages. Journal of West African Languages 30: 5–28. Bearth, Thomas, et al. 2002 Ali Akan. Version 56. CD-ROM. Bearth, Thomas 1993 Satztyp und Situation in einigen Sprachen Westafrikas. In Beiträge zur afrikanischen Sprach- und Literaturwissenschaft, Wilhelm J. G. Möhlig (ed.), 91–104. Köln: Köppe. 2002 Fokus im Akan - Outline. Handout of a presentation at the Linguistic Colloquium of the Seminar of African Studies at the HumboldtUniversity, 28.05.2002. Blench, Roger 2001 Comparative Central Togo: what have we learnt since Heine? Paper presented at 32nd Annual Conference on African Linguistics, Cambridge.

Narrative focus strategies in Gur and Kwa 285 Boadi, Lawrence A. 1974 Focus-marking in Akan. Linguistics 140: 5–57. Clements, George N. 1977 Four tones from three: The extra-high tone in Anlo Ewe. In Language and linguistic problems in Africa. Proceedings of the VII. conference of African linguistics, Paul F. A. Kotey and Haig DerHoussikian (eds.), 168–181. Columbia: Hornbeam Press. Drubig, Hans Bernhard 2003 Toward a typology of focus and focus constructions. Linguistics 41: 1–50. Duthie, Alan S. 1996 Introducing Ewe linguistic patterns. Accra: Ghana University Press. Fiedler, Ines, and Anne Schwarz 2005 Out-of-focus encoding in Gur and Kwa. Interdisciplinary Studies on Information Structure 3, Shinichiro Ishihara, Michaela Schmitz, and Anne Schwarz (eds.), 111–142. Potsdam: Potsdam University. Frajzyngier, Zygmunt 2004 Tense and aspect as coding means for information structure: a potential areal feature. Journal of West African Languages 30: 53–67. Givón, Talmy 1990 Syntax. A functional-typological introduction. vol. 2. Amsterdam: John Benjamins. Green, Melanie 1997 Focus and copula constructions in Hausa, School of Oriental and African Studies, University of London: Ph.D thesis. Heine, Bernd, and Mechthild Reh 1984 Grammaticalization and reanalysis in African languages. Hamburg: Helmut Buske. Hopper, Paul J. 1979 Some observations on the typology of focus and aspect in narrative language. Studies in Language 3: 37–64. Hopper, Paul J., and Elizabeth Closs Traugott 1993 Grammaticalization. Cambridge: Cambridge University Press. Labov, Wilhelm 1972 The transformation of experience in narrative syntax. In Language in the Inner City, 354–96. Philadelphia: University of Philadelphia Press. Lambrecht, Knud 2001 A framework for the analysis of cleft constructions. Linguistics 39: 463–516. Ndayiragije, Juvénal 1992 Structure syntaxique des clivées en Fon. Journal of West African Languages 22: 3–56.

286 Anne Schwarz and Ines Fiedler Olawsky, Knut J. 1999 Aspects of Dagbani grammar. With special emphasis on phonology and morphology: LINCOM studies in African linguistics; 41. München, Newcastle: Lincom. Rongier, Jacques 1997 Langues autonomes du Togo, entre Gur et Kwa?, Ms., n.p. (cited in Blench 2001). Schachter, Paul 1973 Focus and relativization. Language 49: 19–46. Schiffrin, Deborah 1994 Approaches to discourse. Oxford, Cambridge: Blackwell. Schwarz, Anne 2004 Aspekte der Morphosyntax und Tonologie im Buli. Mit Schwerpunkt auf dem Buli von Wiaga. Dissertation, Humboldt-Universität zu Berlin. in prep. Topic/Focus-Intersection in discourse: Distribution of copulative and narrative patterns in Gur. Stassen, Leon 1997 Intransitive predication: Oxford studies in typology and linguistic theory. Oxford: Clarendon Press. Westermann, Diedrich 1930 A study of the Ewe language. London: Oxford University Press.

Focused versus non-focused wh-phrases Enoch Oladé Aboh

Abstract This paper discusses the relation between focused versus non-focused wh-phrases and seeks to identify how they frame the information structure in wh-questionanswer pairs. It is argued that the well-known complementary distribution between focused constituents and wh-phrases masks another reality: Wh-phrases need not be focused. Cross-linguistic evidence indicates that, in languages where whphrases lack focus morphology, these do not interact with focused constituents, unlike focused wh-phrases which appear sensitive to focused constituents. Accordingly, it is suggested that focused wh-phrases and non-focused wh-phrases require different formal licensing: They target different structural positions in syntax and consequently require different information structuring.

1.

Introduction

This paper discusses the relation between focused versus non-focused whphrases and seeks to identify how they frame the information structure in wh-question-answer pairs. I informally define focused wh-phrases as those wh-phrases that are displaced to a designated focus position (where they may co-occur with a focus marker). Non-focused wh-phrases, on the other hand, represent wh-phrases that occur in other positions than the focus position. It is commonly assumed that focused constituents and wh-phrases are closely related because they interact in question-answer pairs and appear to exclude each other in many languages. In terms of Hovarth (1986) and much related work, such interaction is regarded as piece of evidence that wh-expressions are inherently focused. In a framework that assumes the existence of a focus feature [F] in syntax, this amounts to saying that whphrases and focus phrases bear the feature [F] and therefore fall under the same formal licensing constraints (e.g. Haegeman 1995, Rizzi 1996, 1997, 2004, Aboh 2004a among others).

288 Enoch Oladé Aboh This paper shows that such characterization might be misleading. I suggest instead that the apparent complementary distribution of wh-phrases and focused expressions masks the fact that wh-phrases need not be focused. Cross-linguistic evidence indicates that in languages where whphrases do not embed any focus features these do not interact with focused constituents. On the other hand, only focused wh-phrases appear sensitive to focused constituents. With regard to syntax, these facts suggest that focused wh-phrases and non-focused wh-phrases require different formal licensing and therefore target different structural positions. I further propose that wh-phrases of different types require different information structures. In question-answer pairs, for instance, focused wh-phrases require an answer containing a focus marked constituent, unlike non-focused whphrases which don’t seem to impose such a restriction on the answer. Instead, non-focused wh-phrases only convey non-presupposed information for which the comment holds (É. Kiss 1998), but the constituent expressing such information cannot be formally focus marked. In languages with a focus marker, for instance, it appears that such responses do not typically involve a constituent associated with a focus marker. The paper is structured as follows. Section 2 discusses the interaction between focus phrases and wh-phrases and concludes that there are two focus projections within the clause – at the clause periphery and at the VPperiphery – where these elements can occur. Section 3 distinguishes between focused and non-focused wh-phrases and shows that focused whphrases, on a par with focused constituents, may occur within the clause left periphery or at the VP-periphery where they realize the specifiers of focus projections. Non-focused wh-phrases, on the other hand, occur in various IP-internal positions. With regard to information structure of question-answer pairs, it is proposed in section 4 that the focus or non-focus status of the relevant constituent in the answer depends on the focus or nonfocus status of the wh-phrase in the question. This view implies a direct mapping of information structure onto clause structure, as already proposed in the literature (Drubig 2003). Section 5 concludes the paper.

2.

The interaction between focused phrases and wh-phrases

The relation between focused items and wh-phrases has always been an intricate one. The following sections discuss their interaction with regard to clause structure and information structure.

Focused versus non-focused wh-phrases 289

2.1. Clause structure, focused phrases, and wh-phrases In the following Gungbe examples (1a–b) and (1a’–b’), the focused constituents and the wh-phrases are parallel because they occur in a designated position to the left of the focus marker w(a. 1 These Gungbe examples could be taken as representative of languages where focused expressions and whphrases appear to follow the same syntactic path. (1)

w(a dà Àsíàbá a. S(sínú Sessinou Foc marry Asiaba ‘SESSINOU married Asiaba.’ w(a dà Àsíàbá ? a’. M(nù who Foc marry Asiaba ‘WHO married Asiaba?’ S(sínú dà b. Àsíàbá w(a Asiaba Foc Sesinou marry ‘Sessinou married ASIABA.’ w(a S(sínú dà ? b’. M(nù who Foc Sessinou marry ‘WHO did Sessinou marry?’

[Gungbe]

The ungrammatical sentences in (2) further show that this position is unique in Gungbe (and most Kwa languages) because focused constituents and wh-phrases are mutually exclusive in it. 2 (2)

a.* Àsíàbá Asiaba b.* M(nù w(a who Foc

w(a m(nù w(a Foc who Foc Àsíàbá w(a Asiaba Foc

dà ? marry dà ? marry

[Gungbe]

Similar facts are reported in typologically different languages. Rizzi (1997, 2001), for instance, indicates that focused expressions and wh-phrases are in complementary distribution in Italian main clauses (3). As already illustrated by the Gungbe examples in (2), this incompatibility persists in Italian too, irrespective of whether the focused constituent precedes or follows the wh-phrase. (3)

a.* A chi questo hanno detto (non qualcos’altro)? [Italian] To whom THIS they said (not something else)?

290 Enoch Oladé Aboh b.* Questo a chi hanno detto (non qualcos’altro)? THIS, to whom they said (not something else)? In many languages where there seems to be no focus movement, the counterparts of the Gungbe sentences under (1) quite often translate as clefts. Such a systematic correspondence between focus movement constructions and cleft constructions is commonly taken as evidence for a close relation between clefts, focus sentences, and questions. In certain French varieties, for instance, questions involve clefting, as shown in (4a–b). Observe, however, from the ungrammatical examples (4c–d) that such questions cannot include other clefted constituents. (4)

a. C’est qui qui It.is who who ‘WHO came?’ b. Qu’est ce What.is it ‘WHAT did you do?’ c. * C’est quoi, It.is what d. * Qu’est-ce que What.is it that

est be

venu ? come

[French]

que that

tu you

as have

fait ? done

c’est it.is c’est it.is

Jean John Jean John

qui who qui who

a has a has

vu ? seen vu ? seen

These data indicate that clefted constituents and clefted questions exclude each other in a way similar to wh-phrases and focused constituents in Gungbe and Italian. If clefting is related to focusing to some extent, we may further conclude that focused constituents and wh-phrases are mutually exclusive. Similar comparisons abound in the literature and give credibility to the idea that focused expressions and wh-phrases are in competition for the same position in syntax. In a framework that assumes a single specifier position, the ungrammatical sentences in (2) and (3) result from the impossibility of merging two distinct elements (or phrases) in the same position. The same reasoning extends to the cleft constructions in (4), where ungrammatical sentences (4c-d) can be accounted for by saying that cleft constituents and wh-phrases target the same position (e.g. in French). In terms of Rizzi (1997), focused constituents and wh-expressions occur in the specifier of a focus phrase whose head may be overtly realized in some languages. This analysis clearly accounts for data from Gungbe-type

Focused versus non-focused wh-phrases 291

languages where there is a focus particle that is always adjacent to the focused phrase or the wh-phrase, as shown in (5a–b) for the Gungbe sentences (1a–a’) respectively, see Aboh (2004a) and references cited there for discussion. (5)

a. [CP [FocP S(sínú [Foc w(a [TP tS(sínú dà Àsíàbá]]]] b. [CP [FocP M(nù [Foc w(a [TP tm(nù dà Àsíàbá]]]]

The idea that wh-phrases and focused phrases belong to the same class of (quantificational) elements led to formalisms such as Hovarth’s (1986: 118) focus constraint on wh-operators which suggests that “A non-echo question interpretation can be derived only if the Wh-Q operator bears the feature focus”. It appears from Hovarth’s definition that wh-phrases triggering questions must be focused (and quantificational). However, close scrutiny reveals that this generalization does not hold for all the relevant cases because wh-phrases are not always focused cross-linguistically. 2.2. Information structure, focused phrases, and wh-phrases In terms of information structure, the interaction of focused constituents and wh-phrases seems to correlate with the fact that wh-questions trigger focused answers in question-answer pairs, as represented in (6). Here, the Gungbe sentences in (6b) and (6c) represent felicitous replies to the question in (6a). In example (6b), the in-situ focused constituent l(sì acts as new information focus. In the sentence (6c), however, the same constituent encodes identificational focus and has been fronted. It is worth mentioning that, provided the appropriate context, the choice of answer (6c) over (6b) implies a contrast or requires a continuation of the sentence (e.g. he ate rice and became sick). (6)

Kòfí ˜ù ? a. Ét( w(a what Foc Kofi eat ‘WHAT did Kofi eat?’ ˜ù l(sì b. É 3sg eat rice ‘He ate rice.’ é ˜ù (bò c. L(sì w(a rice Foc 3sg eat and ‘He ate RICE (and became sick).’

[Gungbe]

b( start

àwútù) sickness

292 Enoch Oladé Aboh Within the generative framework, studies on the interaction between information structure (or discourse semantics) and clause structure (e.g. É. Kiss 1998) regard the data in (6) as partial evidence that in-situ focused constituents (6b) and ex-situ focused constituents (6c) target two focus positions in the clause. The post-verbal position is specialized in encoding new information focus, but the peripheral position expresses exhaustive/identificational focus. I will return to this asymmetry in (6) in section 4 (see also Zubizarreta 1998, Drubig 2003, Belletti 2002, and references cited there for discussion). At this stage of the discussion, however, it is worth mentioning that under such a view of clause peripheral and clause internal focus positions the traditional ex-situ versus in-situ distinction appears to be misleading. Indeed, the idea that clause structure involves two focus positions (i.e. clause periphery vs. VP-periphery) where focused elements occur implies that exsitu strategies include focused elements that appear in the clause periphery (i.e. within CP) or within the VP-periphery. This suggests that focused elements necessarily move to a designated position. On the other hand, genuine in-situ strategies would involve elements that are internally focused (e.g. a DP-internal focus, see Aboh 2004b) and which occur in positions (including first merge positions) other than designated focus positions. This in turn implies that the term in-situ should not be understood to hold for first merge positions only. In this paper, I argue that syntactically focused elements appear in the VP-peripheral or clause peripheral focus positions, but non-focused constituents are excluded from such positions and must realize some other positions within the clause. Going back to (6) and the assumed informational difference between the VP-peripheral focus position and the clause peripheral one, a look at Gungbe subject wh-phrases suggests that things might not be so clear-cut. Indeed, Gungbe subject wh-phrases do not allow for an answer where the target subject remains IP-internal. Given the question in (7a), for instance, the ungrammatical sentence (7b) indicates that a reply without the focus marker is infelicitous, unlike (7c), which includes the focus marker. (7)

a M(nù w(a ˜ù what Foc eat ‘WHO ate rice?’ b.* Kòfí ˜ù l(sì Kofi eat rice ‘KOFI ate rice.’

l(sì ? rice

[Gungbe]

Focused versus non-focused wh-phrases 293

c. Kòfí w(a ˜ù Kofi Foc eat ‘KOFI ate rice.’

l(sì rice

Put together, these Gungbe facts lead us to conclude that there seem to be two focus positions involved in object wh-question-answer pairs (clause peripheral vs. VP-peripheral) with possible differences in interpretation (e.g. contrast, D-linking). On the other hand, subject wh-question-answer pairs appear to exploit just one (i.e. clause peripheral) position. Some facts from Italian wh-questions provide us with complementary data on this issue. Based on the distribution of new information focused constituents in Italian question-answer pairs, Belletti (2002) argues that Italian post-verbal focused subjects occur in a low focus position immediately dominating the VP, where they express new information focus only. On the other hand, sentence-initial focus is used to realize contrastive focus. Accordingly, sentence (8b) is a felicitous answer to the subject whquestion in (8a), unlike example (8c). (8)

a. Chi ha parlato? Who has spoken ‘Who spoke?’ b. Ha parlato GIANNI has spoken Gianni ‘Gianni spoke.’ c.* Gianni ha parlato

[Italian]

In the same context, however, sentence (9), pronounced with primary stress (or emphasis) on Gianni, can only be interpreted as contrastive focus. (9)

GIANNI ha parlato (non tutta la classe). ‘Gianni has spoken (not the whole class).’ ‘GIANNI spoke (not the whole class).’

[Italian]

Belletti (2002) and much related work therefore concludes that the clause structure involves two focus positions that correlate with informationally different focus expressions: new information focus versus exhaustive focus (É. Kiss 1998). Brunetti’s (2003) discussion of object wh-questions suggests a different picture. According to the author, sentence (10b), with the object in-situ, is

294 Enoch Oladé Aboh an appropriate answer to the question (10a) and expresses new information. On the other hand, sentence (10c), even though marginally felicitous, does not imply a contrast (Brunetti 2003). (10) a. Che cosa ti ha regalato Gianni? ‘What did Gianni give you as a present?’ b. Gianni mi ha regalato [un libro] ‘Gianni me has given a book.’ c. ??[Il libro]F mi ha regalato, non il cd. the book me (he) has given not the cd ‘He gave me the book, not the cd.’

[Italian]

Brunetti (2003) further shows that the contrast in (10b–c) disappears in non-interrogative contexts where both post-verbal and clause peripheral focus positions are used without pragmatic differences. Starting with the statement in (11a), the reactions in (11b–c) indicate that both positions can encode contrastive focus. (11) a. La felpa, l’ha vinta Gianni. ‘As for the sweatshirt, Gianni won it.’ b. No, Gianni ha vinto LA MAGLIETTA. ‘No, Gianni won the T-shirt.’ c. No, LA MAGLIETTA ha vinto Gianni. ‘No, it is the T-shirt that Gianni won.’

[Italian]

Brunetti (2003) concludes from this that focus, whether realized at the clausal periphery or at the VP-periphery, has the same informative value at any of these interfaces. Put together, the Italian and Gungbe facts therefore support the view that there are two linear focus positions within the clause, one at the clausal left periphery and one at the VP-periphery (i.e. post-verbal position). However, these data are inconclusive as to whether these two positions map onto informative differences such that the low focus position is restricted to information focus while the higher position is devoted to contrastive focus. In recent years, the expression of clause peripheral focus has been shown to be a property of a focus projection (FocP) within the complementizer system (Rizzi 1997). The specifier of this projection hosts the focused constituent while its head may be realized by a focus marker, as is the case in certain African languages (e.g. Kwa; Aboh 2004a).

Focused versus non-focused wh-phrases 295

Not much is known, however, about the structural properties of the VPperipheral focus position described in previous paragraphs, and there is still some doubt as to whether a focus phrase could project in the lower portion of the clause where the post-verbal focus is licensed. The following section provides additional empirical evidence supporting the existence of such a focus projection. The discussion shows that an asymmetry between certain Kwa (e.g. Gungbe), Romance (e.g. Italian), and Bantu languages (e.g. Aghem) with regard to expression of focus could reduce to the question whether they solely use the clause peripheral or VP-peripheral focus projection, or both. 2.3. Focus at the VP-periphery This section discusses data from Aghem (Watters 1979, Hyman 1979, 2005), but the reader is referred to Biloa (1997), Sabel & Zeller (2002), Downing, Mtenje and Pompino-Marschall (2004), Good (2005), and references cited there for discussions of post-verbal focus in other Bantu languages (e.g. Chitumbuka, Zulu). Aghem is an SVO Grassfield Bantu language spoken in Cameroon that displays the basic structure exemplified in (12a) and illustrated in (12b). 3 (12) a. T,-bvÊ tËa-bËaghà m2¶ zËa kË-b( n( dogs two Past eat fufu today ‘The two dogs ate fufu today.’ [Hyman 2005: 1] b. S... Aux… Verb… (Focus)… Object… Adjunct

[Aghem]

Starting from representation (12b), the examples in (13) and (14) indicate that a contrastively focused constituent as well as a wh-phrase must occur in the position immediately after the verb. The sentence in (13a) is an instance of object focusing. Here, the focused object is right adjacent to the verb while the subject occurs in the canonical position (i.e. pre-verbally). In addition, the locative adjunct phrase án ‘sóm ‘in the farm’ is displaced to a pre-verbal position. This example contrasts, however, with the subject focus construction in (13b) where an expletive realizes the canonical subject position and the associate DP-subject occurs post-verbally. Finally, the adjunct focus in (13c) indicates that the focused adjunct must occur immediately to the right of the verb.

296 Enoch Oladé Aboh (13) a. Éná2 m2a án ‘sóm zi [b(-k2] [Aghem] Inah Past in farm eat fufu ‘Inah ate FUFU at the farm.’ b. Á m2a zi [éná2] b(-k2 án ‘sóm Expl Past eat Inah fufu in farm ‘INAH ate fufu at the farm.’ c. T,-bvÊ tËa-bËaghà m2¶ zËa n( b(-k2 dogs two Past eat today fufu ‘The two dogs ate fufu TODAY.’ [Hyman 2005: 1] Similarly, Aghem wh-phrases occur in the same position as focused constituents, that is, immediately after the verb, as shown in (14) (Hyman 2005). The sentence in (14a) illustrates adjunct wh-phrases, while the example in (14b) represents subject wh-phrases. Note from sentence (14c) that in multiple wh-questions one wh-phrase immediately follows the verb (presumably in the focus position) while all the others follow successively (see Biloa 1997, Hyman 2005). (14) a. Fil a-m2¶ zËa zËn friends SM.Past eat when ‘WHEN did the friends eat fufu?’ b. À m2a zËa ndúgh2 Expl Past eat who ‘WHO ate fufu today?’ c. À m2a zËa ndúgh2 Expl Past eat who ‘WHO ate what when?’

b(-k2 ? fufu

[Aghem]

b(-k2 ? fufu kw2ak2a zËn ? what when

The distribution of focused constituents and wh-phrases in Aghem as well as the correlation between a focused subject and the presence of an expletive in the sentence (e.g. 12a vs. 13b) suggest that this language does have a fixed focus position immediately after the verb. This description is further supported by the fact that focusing on the truth value of the proposition requires the presence of the Aghem focus marker nó in the position immediately after the verb, as illustrated in (15a). The example in (15b) indicates that a contrastively focused object may left adjoin to the focus particle nó. This example shows that the verb and the focus particle do not form a constituent. Instead, it seems as if the focus particle nó scopes over the constituent immediately left adjacent to it.

Focused versus non-focused wh-phrases 297

m2¶ zËa nó (15) a. T,-bvÊ tËa-bËaghà dogs two Past eat Foc ‘The two dogs ATE fufu.’ [Hyman 2005: 1] b. ZËa b(-k2 nó eat fufu Foc ‘eat FUFU’

b(-k2 [Aghem] fufu

If we take the Gungbe facts discussed previously, where – in order to be somewhat comparable to the Aghem data – the focus marker w(a occupies a focus head position within the clausal left periphery and attracts a focused element to its left (5), we are lead to conclude that the Aghem functional category expressed by the particle nó plays a function similar to the Gungbe focus category realized by w(a. The two languages differ, however, because the Aghem focus category occurs within the VP-periphery only. More precisely, I suggest that the Aghem particle nó occupies a focus head position immediately to the right of the surface position of the verb. Even though Gungbe, Italian, and Aghem differ in many respects as to how they realize the category focus, I regard the facts discussed thus far as representative of the distribution of structural focus positions crosslinguistically. These typologically different languages therefore suggest the following description, where languages only differ with respect to whether they realize the clause peripheral or VP-peripheral focus positions or both. 4 Gungbe [Focus[subject/object, adjunct]- w(a ...V…Focus[object/ adjunct]…..] Italian [Focus[subject/object, adjunct]…… V…Focus[subject]…..] Aghem […………………………….V…Focus-[subject/object/ adjunct]-nó] 5 Clause structure

[Focus….[subject…verb…Focus…]]

This description is compatible with Belletti’s (2002) representation of the clause structure in (16), where there is a topic/focus articulation both at the clausal left periphery and at the VP-periphery. In languages of the Gungbeor Aghem-type, focus markers head these projections (Aboh 2004a). (16)

[ForceP [TopP [FocP [FinP …[ TP ....[TopP [FocP [VP…]]]]]]]]

Under this approach, the Italian post-verbal focus occurs in the VPperipheral focus position. In a sentence like (8b), the subject follows the verb because the latter raises to some higher position in the clause to be

298 Enoch Oladé Aboh licensed for tense and aspect as schematized in (17a). The same reasoning applies to the Aghem example (15b). Following Aboh’s (2004a) proposal of V-to-Asp movement as determining VO versus OV structures in Kwa as well as Sabel and Zeller’s (2002) analysis of Nguni (Zulu) as involving Vto-T movement, I propose (by analogy) that the verb in Aghem must move to an aspect position in overt syntax in order to be licensed for aspect. 6 As a result of this movement, the verb necessarily precedes constituents or whphrases that move to the VP-peripheral focus position, as illustrated in (17b). This analysis extends to wh-phrases in Aghem. (17) a.

AspP

2 spec Asp'

b.

AspP

2 spec Asp'

2 2 Asp° FocP Asp° FocP parlatoi 2 zËa 2 spec Foc' spec Foc' 2 Gianni 2 b(-k2 Foc° VP Foc° VP 6 (nó) 6 tGianni tparlato tzi tb(-k2 With regard to information structure, we have already seen from the discussion on focus constituents in Gungbe and Italian (section 2.2) that there seems to be no systematic mapping of É. Kiss’ (1998) identificational focus onto clause peripheral focus and new information focus onto VP-peripheral focus. Nevertheless, given the interaction between focused constituents and wh-phrases, one could still ask whether such a dichotomy may arise with wh-questions instead. More precisely, one wonders whether wh-phrases are sensitive to the contrastive versus new information distinction such that, in question-answer pairs, the focus value of the answer would match that of the wh-phrase in the question. Under such a view, a contrastive focus whquestion would require a contrastive focus constituent in the clausal left periphery of the answer while a new information focus wh-question would require a constituent carrying new information to occur within the VPperiphery of the answer (É. Kiss 1998, Belletti 2002). The next section deals with these issues.

Focused versus non-focused wh-phrases 299

3.

Focused wh-phrases versus non-focused wh-phrases

The discussion in section 2 suggests that wh-phrases and focus phrases exclude each other because they target the same position. This leads to the conclusion that wh-phrases have something focal about them and must be focused cross-linguistically (Hovarth 1986). However, close scrutiny indicates that wh-phrases do not always occur in a focus position crosslinguistically. Instead, it appears that some languages distinguish between focused and non-focused wh-phrases. In Italian matrix clauses (cf. 3), for example, the incompatibility between focused constituents and wh-phrases that serves as empirical evidence for regarding wh-phrases as inherently focused elements appears less conclusive when it comes to embedded clauses. Consider the following sentences. (18) a. */?

Mi domando [a chi] QUESTO abbiano detto (non qual cos’altro). ‘I wonder to whom THIS they have said (not something else).’ b. */? Mi domando QUESTO [a chi] abbiano detto (non qualcos’ altro). ‘I wonder THIS to whom they have said (not something else).’ c. Mi domando A GIANNI [che cosa] abbiano detto (non a Piero). ‘I wonder TO GIANNI what they have said (not to Piero).’

As already discussed in Rizzi (2001) and much related work, these sentences show that when the focused element is the direct object and the whphrase is a prepositional phrase both orders are degraded as in (18a–b). On the other hand, when the wh-phrase is the direct object and the focused phrase involves a prepositional phrase, as in (18c), the order focus preceding wh-phrase becomes acceptable (or marginal). Rizzi (2001) concludes that when wh-phrases are not forced to move to the focus position (as in the Italian matrix clauses), they will not move there. In embedded clauses, therefore, wh-phrases move to a lower position than the focus position, as indicated by sentence (18c), where the focus element necessarily precedes the wh-phrase. These data suggest that wh-phrases must be focused in matrix clauses but need not be in embedded clauses. Put another way, whphrases do not always occur in a focus position because they are not always

300 Enoch Oladé Aboh focused, contra Hovarth (1986), see Frascarelli & Puglielli (this volume for similar conclusion). 7 Yiman (1988: 370) reports similar facts in Oromo (a Cushitic SOV language) that displays both non-focused and focused wh-phrases. According to the author, the question in (19a), based on the presupposition that someone came, requires the answer in (19b). (19) a. EeĖĖu GCuf-e ? who come-3sg-Past ‘Who came?’ b. Túlluu (GCuf-e) Tulluu come-3sg-Past ‘Tulluu (came).’

[Oromo]

In traditional approaches to question-answer pairs, the wh-phrase in (19a) and the DP-subject in the response (19b) are regarded as focused elements. Yet, it is important for our discussion to observe that these elements are not formally marked for focus in Oromo. Instead, they seem to occur in some other derived position and therefore fall in the class of non-focused categories, as described here. A piece of evidence supporting this description is that the wh-phrase in (19a) and the DP-subject in the response are bare and therefore lack nominative case morphology (Yiman 1988: 371). In contrast to the examples (19a–b), the sentences in (20) involve a case-marked DP-subject in (20a) while its correlated reactions involve a focus marked wh-phrase in the question and a focused DP-subject in the answer (20b-c). These sentences imply a context of controversy where Túlluu’s arrival is denied, as in (20a). This statement triggers the response in (20b), which in turn leads to the reaction in (20c). According to Yiman (1988: 371), such a context requires the expression of contrastive focus, which is why the wh-phrase and the target constituent in the response are focus marked with the particle tu. In addition, the absence of case morphology on these focused elements could be analyzed as evidence that the focused constituent does not transit through the canonical subject position on its way to the peripheral focus position (see section 4 for discussion). (20) a. Túlluu-n hin-GCuf-n-e Tullu-Nom Neg-come-neg-Past ‘Tulluu did not come.’

[Oromo]

Focused versus non-focused wh-phrases 301

GCuf-e b. EeĖĖu-tu who-Foc come-3sg-Past ‘Who is it that came?’ GCuf-e c. Fayyisaa-tu Fayyisaa -Foc come-3sg-Past ‘It is Fayyisaa who came?’ The examples in (19) and (20) and the distinct replies they are associated with therefore indicate that Oromo non-focused subject wh-questions occur in a derived position different from the position in which focused subject wh-questions occur. Put another way, Oromo subject wh-questions involve non-focused and focused wh-phrases with each strategy being associated with a distinct information structure. 8 In wh-questions with non-focused wh-phrases, the target constituent in the response is not focused (19). In wh-questions containing a focused wh-phrase (20b), however, the latter cooccurs with the focus marker tu, which necessitates an answer including a focused constituent marked by the same marker (20c). I conclude from this that the two strategies in (19) and (20) relate to two distinct discourse properties where, in a question-answer pair, a focused wh-phrase requires a focused constituent in the answer while a non-focused wh-phrase requires a non-focused constituent. I further propose that whether such a focused element occurs in the clausal left periphery or within the VP-periphery is subject to parametric variation. Recall from section 2 that Aghem resorts to the VP-peripheral focus position for all types of focus (e.g. new information vs. identificational). Gungbe, on the other hand, mainly uses the clause left peripheral area for focus purposes while Italian seems to use both the clause peripheral and the VP-peripheral focus positions. Following Brunetti (2003), I therefore suggest that left peripheral focus and VP-peripheral focus are identical at the discourse-syntax interface. 9 The following section discusses data on subject-object asymmetry in question-answer pairs that further support the proposed characterization.

4.

Question-answer pairs and information structure

This section analyzes subject-object asymmetry in question-answer pairs which further supports the idea that the information-structural status of the

302 Enoch Oladé Aboh wh-phrase in the question (focused vs. non-focused) determines the status of the target constituent in the response. I start with data from Lele. 4.1. On subject versus object asymmetry in wh-questions Lele forms wh-questions by means of a wh-phrase and a sentence-final question marker gà. In subject questions, the wh-phrase must be fronted to the left of the focus marker ba, as in (21a). The ungrammatical example (21b) indicates that the wh-phrase cannot remain IP-internal (Frajzyngier 2001: 282). (21) a. Wéy ba é who FM go ‘Who went away?’ b.*Wéy é gà ? who go Inter

gà ? Inter

[Lele]

Object wh-phrases, however, display both IP-internal and clause peripheral strategies. The sentence in (22a) illustrates the IP-internal context where the wh-phrase follows the verb, and the sentence involves the sentence-final question marker. In the clause peripheral context, however, the object whphrase is fronted to the clause periphery, where it occurs to the left of a focus marker (22b), see (Frajzyngier 2001: 284/86). 10 (22) a. Mè ày wéy 2sg[F] marry who ‘Who did you marry?’ b. Me ba gol What FM see ‘What did he see?’

gà ? Inter dí 3sg[M]

[Lele]

gà ? Inter

At this stage of the discussion, a brief digression is in order. Indeed, it is worth mentioning that the presence of the question particle in Lele suggests that wh-phrases are not inherently interrogative or focused by definition (see Aboh & Pfau 2006 for discussion). This would mean that in languages where there is no question particle of the Lele-type and where wh-phrases must be fronted, such fronting might as well derive from the necessity of the wh-phrase to be under the scope of some functional element. In the case of Lele, the obligatory occurrence of the focus particle in example (22b) is

Focused versus non-focused wh-phrases 303

strong evidence that fronting in this case is contingent to a clause peripheral focus requirement. I therefore conclude that wh-phrases move only if they are forced to, that is, when there is some informational and syntactic reason for them to front. Put differently, wh-phrases derive their semantics from their surface syntactic positions, not (exclusively) from their internal structure. This conclusion is compatible with Lipták’s (2001) proposal that whphrases better qualify as variables (i.e. with no inherent quantificational meaning) and derive their various quantificational meanings from their surface syntactic positions as well as the type of quantificational elements that bind them. In terms of this approach, the description provided here implies that focused wh-phrases acquire their quantificational meaning by virtue of being focused syntactically (i.e. moved to an operator position). On the other hand, non-focused wh-phrases derive their (quantificational) interpretation from their binder (e.g. an interrogative operator). With this in mind, I conjecture that there is no question-focus correlation such that in question-answer pairs a wh-phrase will necessarily trigger an answer containing a focus marked constituent. I therefore propose the general description in (23) and (24). (23)

The discourse properties of the wh-phrase determine the discourse properties of the answer.

(24)

A focused wh-phrase (in the higher or lower focus position) requires a focused constituent in the response, but a non-focused wh-phrase doesn’t.

Empirical evidence from Amharic supports this view (Drubig & Schaffar 2001, Demeke & Meyer 2003). According to Demeke (2003), Amharic, an SOV language, displays multiple wh-phrase constructions where the whphrases can occur in-situ or ex-situ (in the classical sense). The examples in (25) are instances of in-situ wh-questions involving arguments as well as adjuncts. (25) a. Kassa lä-man mïn sät’t’ä ? Kassa to-whom what gave ‘What did Kassa give to whom?’ b. Kassa mäce lä-man mïn Kassa when to-whom what ‘What did Kassa give to whom and when?’

[Amharic]

sät’t’ä ? gave

304 Enoch Oladé Aboh With regard to ex-situ wh-questions, Demeke (2003) claims that Amharic does not exhibit any syntactic constraints (e.g. superiority) on the ordering of fronted wh-phrases. Instead, this order is mapped onto the informative structure that the speaker expects from the hearer. The following questionanswer pairs where the (b) patterns are responses to the (a) patterns illustrate this correlation. (26) a. Mäce man mïn When1 who2 what3 b. Tinant Kassa ‘Yesterday1 Kassa2

gäza ? bought mäs’hafgäza a book3

[Amharic]

bought’

(27) a. Mäce mïn man When1 what3 who2 b. Tinant mäs’haf Yesterday1 book3

gäza ? bought Kassa Kassa2

gäza bought

(28) a. Mïn mäce man gäza ? who2 bought What3 when1 b. Mäs’haf tinant Kassa gäza yesterday1 Kassa2 bought Book3 According to Demeke, the leftmost wh-phrase is interpreted as the most prominent (i.e. the focused wh-phrase in our terms). It is still unclear what the exact syntax of multiple fronted wh-phrases is in Amharic, but it is worth noting that the sequencing of the wh-phrases in the question determines rigid ordering of the constituents in the answer. This makes Amharic a strict discourse-configurational language where information structure and clause structure work in tandem. Similarly, the described facts represent strong empirical evidence that wh-phrases are not inherently specified for discourse properties such as focus or interrogation. Instead, I suggest that just like ordinary DPs they constitute potential candidates for establishing focus-presupposition or question-answer relations at the interface level. The idea that wh-phrases mainly derive their semantics from surface structure is compatible with the systematic asymmetry found between subject and object wh-questions, which I turn to now. Recall from the Lele examples in (21) and (22) that the subject wh-phrases must be fronted to

Focused versus non-focused wh-phrases 305

the left of the focus marker while object wh-phrases need not. The latter may occur in-situ (22a) or ex-situ when focused (22b). Similar facts are found in Gungbe. As mentioned previously in the examples in (6), repeated here as (29), an object wh-question (29a) may trigger two possible answers, as in (29b–c). Kòfí ˜ù ? (29) a. Ét( w(a what Foc Kofi eat ‘WHAT did Kofi eat?’ b. É ˜ù l(sì 3sg eat rice ‘He ate rice.’ é ˜ù (bò c. L(sì w(a rice Foc 3sg eat and ‘He ate RICE (and became sick).’

[Gungbe]

b( start

àwútù) sickness

In addition, certain speakers allow for ‘optional’ realization of the focus marker in object wh-questions, as illustrated in (30). (30)

M(nù %(w(a) Kòfí who Foc Kofi ‘WHO did Kofi marry?’

dà ? marry

[Gungbe]

Sentences like (30) led many authors working on the Gbe languages to assume that the focus marker is optional in object wh-questions. I turn to this discussion shortly and show that example (30) actually masks two different derivations (see Aboh 2004a for discussion). What matters for the present discussion is that no such variation arises in subject wh-questions, and speakers only accept sequences where the whphrase immediately precedes the focus marker, as shown in (31). (31)

dà M(nù *(w(a) who Foc marry ‘WHO married Kofi?’

Kòfí ? Kofi

[Gungbe]

Similarly, in subject wh-question-answer pairs (7), repeated here as (32), a reply to (32a) which does not involve subject focusing is infelicitous (32b). Instead, the response requires a structure like (32c) where the subject oc-

306 Enoch Oladé Aboh curs in the left peripheral focus position and immediately precedes the focus marker. (32) a. M(nù w(a ˜ù who Foc eat ‘Who ate rice?’ l(sì b.* Kòfí ˜ù Kofi eat rice ‘KOFI ate rice.’ c. Kòfí w(a ˜ù Kofi Foc eat ‘KOFI ate rice.’

l(sì ? rice

[Gungbe]

l(sì rice

These data indicate that there is a structural constraint in Gungbe that prevents subject wh-phrases from exploiting the VP-peripheral focus position in question-answer pairs. Similarly, the Gungbe subject wh-phrase cannot occur in the canonical subject position (i.e. [spec TP]). As a consequence, subject wh-question in Gungbe necessarily involve a focused wh-phrase, which in turn requires a focused subject DP in the response, hence the infelicitous example (32b). Sabel & Zeller (2002) discuss data on subject wh-questions in Zulu (Bantu) that mirror the Gungbe facts. Zulu has both in-situ and ex-situ whquestions, with ex-situ wh-questions being realized as clefts. These correspond to focused wh-questions in our terms. The sentences in (33) illustrate these options for object wh-questions. (33) a. U-bona-ni ? [Zulu, Sabel & Zeller 2002:1, 2] 2sg-see-what9 ‘What did you see?’ b. Y-ini o-yi-bona-yo ? COP-what9 RC2sg-OC9-see-RS ‘What is it that you saw?’ Interestingly, subject wh-questions do not allow the subject DP to appear in its canonical position (i.e. [spec TP]), hence the ungrammatical example (34a). Instead, two strategies are available: one that involves a cleft in parallel to the object wh-question, as in (34b), and, most crucially, one that involves a post-verbal position with the canonical subject position being filled by an expletive (34c).

Focused versus non-focused wh-phrases 307

(34) a.* Ubani u-fike ? who1a SP1a-arrived ‘Who arrived?’ b. Ng-ubani o-fikile COP-who1a RC1a-arrived ‘Who arrived?’ c. Ku-fike bani ? Expl-arrived who1a ‘Who arrived?’

[Zulu]

In terms of the two focus positions proposed in this study (i.e. clause peripheral and VP-peripheral), Zulu appears to exploit the clause peripheral and the VP-peripheral focus positions for both subject and objects, while in Gungbe subjects are restricted to the clause peripheral focus position only while objects can occur in the clause peripheral position and a derived position within IP. Further, observe from examples (34b–c) that Zulu subject wh-phrases cannot occur in the canonical subject position (i.e. [spec TP]) but rather in post-verbal position. The Gungbe subject wh-phrase is also excluded from the canonical subject positions but is restricted to the clausal left peripheral position. Accordingly, Gungbe subject wh-questions involve the focus marker and require a response where the DP-subject is fronted to the left of the focus marker w(a (32a–c). These facts clearly indicate that subject whphrases are excluded from [spec TP] in both Gungbe and Zulu, although the two languages circumvent this constraint differently. 4.2. When the EPP-feature competes with the focus feature Under Sportiche’s (1988) VP-internal subject hypothesis, the ban on subject wh-phrases in [spec TP] reduces to the competition between two syntactic features: the EPP-feature and the focus feature [F] (Chomsky 1995, Rizzi 1997, Rizzi and Shlonsky 2005). Indeed, the exclusion of the Gungbe subject-DP in the VP-peripheral focus position could be explained if we assume that the subject must move to [spec TP] to check the EPP-feature under T in this language (Chomsky 1995). This movement, however, cannot pass through the lower [spec FocP], a freezing position as shown by Rizzi (1996, 1997, 2004), Haegeman (1995), Rizzi and Shlonsky (2005), and much related work. Under the assumption that focus positions are criterial positions (Rizzi 2004) where the satisfaction of the focus criterion ter-

308 Enoch Oladé Aboh minates the focus chain, this would mean that movement of the subject whphrase to the intermediate [spec FocP] would freeze the subject in place. Consequently, such derivation crashes in Gungbe because the nominative subject fails to check the EPP-feature under T. As proposed by Rizzi and Shlonsky (2006), however, [spec TP] is also a freezing position. Accordingly, the subject wh-phrase cannot move to [spec TP] on its way to [spec FocP] within the left periphery. I therefore propose that Gungbe subject wh-phrases obligatorily move to the clause peripheral focus position, where they control (and identify), under c-command, an empty category (e.g. pro) in [spec TP]. This empty category checks the EPP under T. I further suggest that obligatory control of the subject wh-phrase in [spec FocP] over the empty category in [spec TP] in Gungbe correlates with the paucity of agreement morphology as well as the lack of DP-expletives in Gbe. Under Agree (Chomsky 1999), it is conceivable that the presence of agreement morphology and a DPexpletive in Gbe could have allowed for a derivation where the empty category is directly licensed in [spec TP], say under spec-head agreement, while the subject DP moves to the VP-peripheral focus position. This, however, is impossible, and the only option Gungbe is left with is to displace the subject wh-phrase to the peripheral focus position, where it controls the element in [spec TP], as illustrated in (35). Projections irrelevant to this discussion are ignored. (35)

[FocP M(nù [Foc w(a [.....[TP prom(nù l(sì]]]]]]]]

[AspP [FocP [

Foc

[VP tm(nù˜ù

This analysis also holds for the response where the answer of a focused whphrase (i.e. a focused subject DP) cannot remain VP-internal and cannot move to the intermediate [spec FocP] on its way to [spec TP], where it checks the EPP under T. As a consequence, the focused DP-subject has to move to [spec FocP] at the clausal periphery from where it controls the empty category in [spec TP], as sketched in (36). (36)

[FocP Kòfí [Foc w(a [.....[TP proKòfí l(sì]]]]]]]]

[AspP [FocP [

Foc

[VP tKòfí ˜ù

The same argument holds for the Zulu facts where the VP-peripheral focus position is available for both objects and subjects. Under the assumption that the subject must move to [spec TP] due to the EPP, movement of the

Focused versus non-focused wh-phrases 309

subject wh-phrase to the VP-peripheral focus position would force a derivation where the subject will pass through [spec FocP] on its way to [spec TP]. Given that [spec FocP] is a freezing position, however, subjectmovement to this position freezes the subject there, and the derivation would crash unless the EPP-feature under T is checked otherwise. In this regard, it is interesting to see that Zulu, unlike Gungbe, exhibits an expletive that checks the EPP under T, and the derivation converges. Like most Bantu languages, Zulu also displays a rich agreement morphology, contrary to Kwa languages in general. The EPP constraint and the freezing nature of [spec TP], however, do not affect object wh-question-answer pairs, and languages consequently show no asymmetry in this respect (e.g. Lele vs. Zulu, Gungbe vs. Zulu, and Zulu vs. Italian). As suggested in the Gungbe example (29), an object wh-question may lead to two structurally different responses where the object DP either stays in-situ or fronts to the clausal left periphery, where it precedes the focus marker (29b–c). This would mean that object whphrases have access to both the VP-peripheral and the clause-peripheral focus positions. Considering this, I propose that the optionality in (30) is only apparent. I suggest that this example masks two different constructions, as illustrated in (37a) and (37c) with their corresponding responses in (37b) and (37d) respectively. Kòfí dà ? (37) a. %M(nù who Kofi marry ‘Who did Kofi marry?’ b. Kòfí dà Màrí. Kofi marry Mary ‘Kofi married Mary.’ Kòfí dà ? c. M(nù w(a who Foc Kofi marry ‘Who is it that Kofi married?’ dà. d. Màrí *(w(a) Kòfí Mary Foc Kofi marry ‘Kofi married MARY.’

[Gungbe]

Example (37a) corresponds to a situation where the set of possible girls that Kòfí can marry is unknown and the speaker is looking for this new information as part of the topic of discussion, hence the response in (37b). Sentence (37c), on the other hand, corresponds to a situation where there is a pre-

310 Enoch Oladé Aboh existing set of potential girls that Kòfí can marry and the speaker wants to know who Kòfí marries eventually. The response here must contain a focused object as in (37d). This suggests that even though the wh-phrase fronts in (37a) and (37c), it does not target the same positions, and it does not trigger the same information structure in the responses. This is shown by the contrast in (37b) and (37d), which leads me to conclude that there is a projection FP within the clausal left periphery that is lower than FocP but higher than TP and whose specifier [spec FP] may host non-focused whphrases, as illustrated in (38). (38)

[FocP [Foc [FPM(nù.....[TP Kòfí [AspP [FocP [ Foc [VP tKòfí dà tm(nù]]]]]]]]

This analysis is compatible with the conclusion we reached while discussing the Italian data in section 3 that wh-phrases do not always move to the focus position. Instead, it appeared from the discussion that in certain contexts wh-phrases occur in a derived position between the focus position and their first merge position. I further assume that in the response of such Gungbe wh-questions the target constituent also appears in a derived position that is not a designated focus position. With regard to the focused object wh-phrases, I propose that they necessitate focused object DPs in the response, where the focused object is in the clause peripheral focus position and precedes the focus marker. This is illustrated in (39). (39) a. [FocP M(nù [Foc w(a [FP tm(nù [TP Kòfí [AspP [FocP [ Foc [VP tKòfí dà tm(nù]]]]]]]] b. [FocP Màrí [Foc w(a [FP tMàrí [TP Kòfí [AspP [FocP [ Foc [VP tKòfí dà tMàrí]]]]]]]] Extending this discussion to the question of wh-movement versus wh-insitu languages, the proposed analysis here indicates that, in languages which resort to the VP-peripheral position only (e.g. Aghem), the so-called in-situ position does not correspond to a VP-internal argument position. Instead, what superficially appears to be a first merge position is actually the designated VP-peripheral focus position where focused constituents occur in these languages. Granted this is the right characterization, it seems reasonable to propose that the constraints on accessing this lower focus position can only be language specific. Note, for instance, that while the Gungbe focused subject

Focused versus non-focused wh-phrases 311

DPs or wh-phrases cannot target this position, no such problem arises in Aghem, Zulu, or in Italian where the languages involve expletive or pronominal elements that can fill the canonical subject position (i.e. [spec TP]), allowing the focused subject to move to the intermediate VPperipheral focus position.

5.

Conclusion

This paper shows that languages distinguish between focused and nonfocused wh-phrases, which target different positions within the clause. The discussion of the distribution of focused wh-phrases supports recent research within the cartography framework on the existence of two focus projections within the clause structure, that is, within the clause periphery and the VP-periphery. However, cross-linguistic data suggest that there is no semantic (or pragmatic) distinction between these two focus positions. Instead, it is argued, that in question-answer pairs, the position of the whphrase within the question determines that of the target DP in the response. This implies that the surface position of the target constituent in the response depends on that of the wh-phrase in the question. In this regard, it has been shown that a focused wh-question requires a focused constituent in a focus position while a non-focused wh-phrase requires a non-focused constituent. Data from languages with focus markers show that, in focused wh-phrase questions, the wh-phrase occurs with a focus marker which also marks the target constituent in the response. In non-focused wh-questions, however, the wh-phrase occurs in a derived position that also conditions the surface position of the target constituent in the response. I conclude from this that there is no systematic correlation such that in question-answer pairs a wh-question will necessarily require a response including a focused constituent.

Notes 1. Small caps in the translations indicate focused constituents. 2. See Aboh (2004a) and references cited there for the discussion on focus constructions and wh-questions across Kwa. 3. Examples are reproduced as they appear in the cited sources.

312 Enoch Oladé Aboh 4. It is crucial to note that the Gungbe focus marker and the Aghem focus marker are not isomorphic. For instance the focus marker does not seem to co-occur with wh-phrases in Aghem, unlike in Gungbe. 5. Other Bantu languages (e.g. Zulu) are also relevant here (see Sabel & Zeller 2002). 6. See also Manfredi (1997) for further discussions of VO vs. OV in Kwa. 7. See also Lipták (2001) for arguing that wh-phrases are not inherently quantificational. Similarly, Boskovic’s (2002) work on multiple whfronting languages indicates that wh-phrases undergo movement in these languages depending on whether they are focused or not, suggesting that wh-questions in those languages may involve focused versus non-focused instances. 8. See section 4 for a similar conclusion concerning object wh-questions in Gungbe. Put together, these data suggest that subject and object whphrases show parallel distribution cross-linguistically even though they might behave differently language-internally. 9. See Lipták (2001) for similar conclusion for English in-situ focus. 10. Adjunct focusing also displays these two strategies. Whether the Lele data underscore the VP-peripheral and clause peripheral focus positions is not clear for the time being and more study is needed before we reach a precise characterization of these facts.

References Aboh, Enoch Oladé 2004a The morphosyntax of complement-head sequences. Clause structure and word order patterns in Kwa: Oxford studies in comparative syntax. Oxford, New York: Oxford University Press. 2004b Topic and Focus within D. Linguistics in The Netherlands 21: 1–12. Aboh, Enoch Oladé, and Roland Pfau 2006 What’s a wh-word got to do with it? Ms. University of Amsterdam. Belletti, Adriana 2002 Aspects of the low IP area. In The Structure of IP and CP. The Cartography of Syntactic Structures 2, Luigi Rizzi (ed.), New York: Oxford University Press. Biloa, Edmond 1997 Functional Categories and the Syntax of Focus in Tuki. Lincom Studies in African Linguistics 02. Newcastle: Lincom Europa. Boškoviü, Željko 2002 On Multiple Wh-fronting. Linguistic Inquiry 33: 351–383.

Focused versus non-focused wh-questions 313 Brunetti, Lisa 2003 A Unification of Focus. Ph.D. dissertation, Università di Firenze. Cheng, Lisa Lai-Shen 1991 On the Typology of Wh-questions. Ph.D. dissertation, MIT. Chomsky, Noam 1995 The minimalist Program. Cambridge, Mass.: MIT Press. 1999 Derivation by phase. MIT Occasional Papers in Linguistics 18. Department of Linguistics, MIT. Demeke, Girma Awgichew 2003 Wh-questions in Amharic. Ms. University of Tromsoe. Demeke, Girma Awgichew, and Ronny Meyer 2003 Information Structuring in Amharic. Ms. University of Tromsoe. Drubig, Hans Bernhard 2003 Toward a Typology of Focus and Focus Constructions. Linguistics 41: 1–50. Drubig, Hans Bernhard, and Wolfram Schaffar 2001 Focus Constructions. In Language Typology and Language Universals, Martin Haspelmath, Ekkehard König, Wulf Oesterreicher, and Wolfgang Raible (eds.), Berlin: Mouton. Downing, Laura, Al Mtenje, and Bernd Pompino-Marschall 2004 Prosody and information structure in Chichewa. ZAS Papers in Linguistics 37: 167–186. É. Kiss, Katalin 1998 Identificational Focus versus Information Focus. Language 74: 245– 273. Frajzyngier, Zygmunt 2001 A Grammar of Lele. Stanford Monographs in African Languages, Stanford, California: CSLI Publications. Good, Jeff 2005 A fixed-position focus construction in Naki. Paper presented at the Conference on Focus in African languages, ZAS, Berlin, October 2005. Haegeman, Liliane 1995 The Syntax of Negation. Cambridge: Cambridge University Press. Hovarth, Julia 1986 Focus in the Theory of Grammar and the Syntax of Hungarian, Dordrecht: Foris. Hyman, Larry M. (ed.) 1979 Aghem grammatical structure. Southern California Occasional Papers in Linguistics 7. Los Angeles, University Southern California.

314 Enoch Oladé Aboh Hyman, Larry M. 2005 Focus Marking in Aghem: Syntax or Semantics? Paper presented at the Conference on Focus in African languages, ZAS, Berlin, October 2005. Hyman, Larry M., and John R. Watters 1984 Auxiliary Focus. Studies in African Linguistics 15: 233–273. Lipták, Anikó Klára 2001 On the Syntax of Wh-items in Hungarian. LOT dissertation series. 45 Manfredi, Victor 1997 Aspectual Licensing and Object Shift. In Object Position in BenueKwa, Rose-Marie Déchaine and Victor Manfredi (eds.), 87–122. The Hague: The Holland Academic Graphics. Rizzi, Luigi 1996 Residual Verb Second and the wh-criterion. In Parameters and Functional Heads: Essays in Comparative Syntax, Adriana Belletti and Luigi Rizzi (eds.), 63–90. New York: Oxford University Press. 1997 The Fine Structure of the Left Periphery. In Elements of Grammar, Liliane Haegeman (ed.), 281–337. Dordrecht: Kluwer. 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. New York: Elsevier. 2004 Locality and Left Periphery. Ms. University of Siena. Rizzi, Luigi, and Ur Shlonsky 2005 Strategies of Subject Extraction. Ms. University of Siena. Sabel, Joachim, and Jochen Zeller 2002 Wh-question formation in Nguni. To appear in Oxford Working Papers in Linguistics. Oxford. Sportiche, Dominique 1988 A Theory of Floating Quantifiers and its Corollaries for Constituent Structure. Linguistic Inquiry 19: 425–449. Yiman, Baye 1988 Focus in Oromo. Studies in African Linguistics 19: 365–384. Watters, John R. 1979 Focus in Aghem: a study of its formal correlates and typology. In Aghem grammatical structure, Larry Hyman (ed.), Southern California Occasional Papers in Linguistics 7: 137–197. Zubizarreta, Maria Luisa 1998 Topic, Focus and Word Order. Cambridge, Mass.: MIT Press.

List of contributors Enoch Oladé Aboh General Linguistics, Languages and Literature Faculty of Humanities University of Amsterdam Spuistraat 210 NL - 1012 VT Amsterdam [email protected] Ines Fiedler Humboldt University of Berlin SFB 632 Information Structure Unter den Linden 6 / Location: Mohrenstr. 40-41 D - 10099 Berlin [email protected] Mara Frascarelli Università degli Studi Roma 3 Dipartimento di Linguistica Via Ostiense, 236 I - 00146 Roma [email protected] Tom Güldemann Seminar für Allgemeine Sprachwissenschaft Universität Zürich Plattenstrasse 54 CH - 8032 Zürich [email protected]

Katharina Hartmann Humboldt Universität zu Berlin Institut für deutsche Sprache und Linguistik Unter den Linden 6 / Location: Dorotheenstr. 24 D - 10099 Berlin [email protected] Victor Manfredi African Studies Center Boston University 270 Bay State Road Boston MA 02215 U.S.A. [email protected] Lutz Marten Department of African Languages and Cultures School of Oriental and African Studies Thornhaugh Street Russell Square UK - London WC1H 0XG [email protected] Annarita Puglielli Università degli Studi Roma 3 Dipartimento di Linguistica Via Ostiense, 236 I - 00146 Roma [email protected]

316 List of contributors

Brigitte Reineke Eichenring 4a D - 16341 Panketal [email protected] Chris Reintges PO box 9515 NL - 2300 RA Leiden [email protected]. nl Anne Schwarz Humboldt University of Berlin SFB 632 Information Structure Unter den Linden 6 / Location: Mohrenstr. 40-41 D - 10099 Berlin [email protected]

Florian Schwarz Department of Linguistics 226 South College University of Massachusetts Amherst, MA 01003 U.S.A. [email protected] Sabine Zerbian Department of Linguistics University of the Witwatersrand Private Bag 3 Wits 2050 South Africa [email protected] Malte Zimmermann Universität Potsdam Institut für Linguistik Karl-Liebknecht-Str. 24-25 D - 14476 Golm [email protected]

Subject index accent 2; 4; 7; 15–19; 21; 24–32; 35–51; 56; 67–68, 74; 76; 79; 134; 147; 260 pitch 2; 4; 7; 47; 49–50; 56; 67; 74; 147 *Adjunction 113; 115; 119–120; 123–132 Late *Adjunction 120; 123–132 afterthought construction 124; 127 Agree 1; 103; 161; 168; 171; 173; 175–176; 178; 206; 308 agreement 11; 19; 42; 47; 115; 118; 123; 132; 134–135; 162; 165– 167; 180–181; 183–184; 190– 191; 193; 204; 207; 209; 213; 216; 220; 224; 242; 245; 268; 271; 308; 309 anaphoric 115 anti-agreement 180–181; 184 wh-agreement 190–191; 204; 213; 216; 220 assertion marker 151–152; 155 asymmetry 79; 126; 167–169; 176; 182; 224; 226; 229; 256; 268– 269; 272; 292; 295; 301; 302; 304; 309 subject vs. non-subject 224; 229; 268–269; 272 subject vs. object 79; 301–304 auxiliary periphrasis 84; 99; 102 C-domain 161–162; 169–170; 172; 174–179 Clause declarative 185; 187; 190 embedded 142; 150; 152–153; 175; 179; 197; 199; 206; 248; 299 narrative 267–268; 273; 276– 280; 284 neutral 4; 186

relative 7; 8; 121; 142; 146; 150; 152–153; 164–166; 168; 174–175; 180–181; 183; 185– 186; 188; 197–198; 203; 212– 213; 229; 238; 267–268; 277 sequential 238; 268; 279 small clause (SC) 164 clause-typing 187; 203 cleft 7; 8; 139–140; 147; 149–158; 199–200; 262; 277; 282; 285; 290; 306 clitic 15; 19; 24–31; 35; 38–39; 42; 49; 97; 107; 114–115, 123– 132; 144; 157; 162–164; 167; 169; 178–182–183; 193; 198; 283 locative subject clitic 131 subject clitic 24–25; 35; 38; 123–128; 131–132; 163; 179 complementizer 157; 162; 188– 200; 206; 209; 212–214; 294 relative 188; 195; 197; 212–214 conjoint (form) 31–35; 43; 132 conjunction 39; 44; 121; 271–284 construction declarative focus 186; 199–200; 203 extra–clausal 279 focus, see focus intra-clausal 280; 282 interrogative focus 202 narrative 7; 192; 202; 211; 267– 268; 273–286 relative 213; 229 wh- 139–141; 150–154; 190; 192; 217 copula 36; 139–142; 146; 149; 150–151; 155–158; 164; 166; 189–190; 214; 217; 234; 261– 262; 277–278; 280–286 defocusing 103–104

318

Subject index

discourse 1; 4–8; 11; 34; 57–61; 73–76; 97–98; 110; 113–115; 123–124; 127; 130; 134; 156; 161–162; 165–166; 172–174; 176–179; 183; 192–193; 201– 202; 217; 223; 233; 237; 260; 262; 282; 286; 292; 301–304 categories 162; 166; 174 discourse-pragmatic function 113–115 disjoint (form) 29; 31–35 duration 18; 64; 68–70; 73; 76 Dynamic Syntax 113–117; 122– 135 EPP-feature 307–309 exhaustivity 6; 86; 189; 201; 203; 226; 241; 244; 246; 251–261; 292–293 marker 6 ; 244 ; 246 ; 251–253; 256–261 expletive pronoun 278; 282 focus assertive 30; 95; 100–101; 104; 109 ambiguity 15; 21; 227; 230 auxiliary 32; 46; 98; 108; 218; 314 contrastive 6; 95; 100; 129; 172; 185; 188; 202–203; 293– 296; 298; 300 focus copula 189–190 exhaustive listing 201; 189; 201; 203 ex situ 7; 15; 139–140; 143– 155; 229; 232; 237; 239; 242– 245; 248; 250; 259–260; 268– 269; 276–277; 281–283; 292 focus alternatives 252; 257 focus constituent 4; 5; 192; 195; 225; 241–242; 246; 248; 250– 251; 255; 258–259; 276; 280– 283; 298

focus construction 7; 11; 47; 101; 107; 113; 122; 127; 131; 139; 143–157; 164; 166–170; 174; 180–188; 199–203; 217– 218; 223–226; 230; 232; 235; 237; 239; 262; 267–285; 291; 295; 311; 313 focus feature 22; 144; 146; 149; 211; 245; 287–288; 307 focus fronting, see focus movement focus marker 2–8; 33; 35; 55– 56; 74–75; 108; 137; 140; 144; 148–151; 156; 159; 163–164; 179; 183; 189–90; 221–259; 269; 277–279; 282; 285; 287– 288; 292; 294; 296–297; 300– 314 affirmative 237; 239 negative 239 focus movement 5; 15; 154; 159; 195; 242; 245; 251; 290 focus particle 184; 257; 291; 296; 302 focus phrase 139–151; 155– 157; 181; 185; 189; 194; 200; 208; 212; 245; 287–288; 290; 295; 299 focus position 6; 205; 287; 292–312 focus prominence 20; 37; 166; 171; 193 focus sensitivity 18; 25; 37; 203; 241; 244; 246; 250–251; 257–259 focused wh-question 7; 175; 287–288; 299–311 identificational 11; 114; 122; 124; 126; 158; 226; 239; 261; 291; 292; 298; 313 inherent 98–99; 109

Subject index 319 in situ 6; 7; 55–56; 60; 71; 74– 75; 101; 147; 149; 155; 202; 217; 223–224; 236; 238; 241– 250; 260; 262; 291–292; 312 multiple 31; 47; 129; 195 new information 7; 56; 129; 200–201;226; non-focused wh-question 287– 288; 299; 301; 303; 310; 311 non-subject focus 224–229; 268–270; 273; 277; 282 object focus 63; 64; 67; 70; 228; 244; 247; 250; 295 presentational 114; 122; 126; 131; 201; 214 sentence focus 147–149; 157; 230; 247 subject focus 23; 38; 39; 149; 214; 224; 226; 229–230; 244– 246; 259; 269; 295; 305 verbal focus 224; 230; 235 verb phrase focus 148 verum focus 148–149 focus-marking 75; 227; 230; 285 lack/absence of 7; 55–56; 74; 129; 202; 242; 246 obligatory 37–38 non-obligatory 75 optional 256; 259 prosodic 4; 55–56 morphological 22 syntactic 8; 75 force 161–162; 166–179; 193; 202–203; 208; 297 illocutionary 161; 168; 170; 174–175; 178–181 interrogative 162; 169–170; 173; 177; 202–203 Force-Fin system 161–163; 166; 169 fronting, see movement f-structure 115 fundamental frequency 64; 73–76 Government Phonology 133

grammaticalization 77; 83–86; 92; 96; 100; 103–108; 223; 268; 278; 280–281; 284–285 implicature 241; 246; 251–252; 257–262 conventional 241; 246; 251– 252; 257–258; 262 inflection 22; 43; 87; 91–92; 96; 167; 185; 188–192; 197; 204; 209–211; 214; 217; 219 special 185; 188–192; 197; 204; 209–211 ; 214–215; 217; 219 inflectional morphology 141; 185–186; 189–190; 215 information structure 1–8; 11–12; 15; 35; 51–52; 55; 63; 70–83; 92–94; 97–101; 104; 114; 120; 122; 127–135; 156; 161–162; 172; 176; 183; 213; 226; 232; 239; 259; 262; 279; 283; 285; 287–288; 291–292; 298; 301; 304; 310; 313 intonation 2–7; 12; 28; 36–37; 42; 45; 47–61; 64; 67; 76–79; 142; 171; 262 Lexical Functional Grammar (LFG) 114; 127 lexical information 117–118 LINK 113; 115; 119–121; 127– 128; 130 Logical Form (LF) 115–116; 119; 127; 133; 174; 207; 211; 216– 217; 242 morpho-syntactic flagging 188; 210–212 movement 3; 5;15; 41; 47; 105; 109–110; 133; 144–145; 149– 150; 154–155; 157; 159; 169178; 182; 185–186; 188– 191; 195; 198–200; 204–220; 242; 245; 251; 277; 282; 290; 298; 302–303; 307– 312 covert 188; 207; 211; 215 long distance 215

320 Subject index wh-movement 159; 185; 190– 191; 195; 204–212; 215–216; 218; 220; 310 negation 38; 47; 83; 88; 91; 97; 99; 102; 106–108; 110; 142–143; 146; 150; 152; 156; 184; 217; 228; 231; 233; 238–239; 259; 313 object definite 98 object clitic 24–26; 30–31; 39; 97; 114–115; 129–132; 179 object shift 84; 92; 109; 314 preverbal 83–86; 88–93; 96– 102 pronominal 87–88; 91; 104; 224; 270; 277 operator 20; 25; 33; 36; 92; 98– 102; 152; 164–166; 169–176; 180–181; 185– 186; 188–193; 198; 207–212; 216; 257; 291; 303 operator-variable dependency 185; 188; 190; 192; 209–210; 212 out-of-focus part 1; 225; 229; 232; 238; 269–273; 276; 284–285 perception 6; 41–44; 46; 55–56; 60; 67; 71–74; 78; 219 periphery left 51–52; 114; 128; 131; 135; 157; 159; 161–162; 169; 179; 184; 190; 193–195; 198; 204; 206; 220; 232; 237; 288; 294; 297–298; 301; 308–310; 314 right 113–114; 119; 122; 126; 128; 133; 247 VP-periphery 288; 292; 294– 295; 297; 301; 311

Phrasing 3; 12; 18; 20; 25; 31–33; 35; 37; 39; 52; 55–60; 64; 73; 78; 115; 214; 225; 262 Phonological 115 Syntactic 20; 25 postverbal position 99; 131; 270; 283 postverbal subject 126; 132 predicate-argument structure 125 pronoun 11; 42; 65; 83; 87–88; 96– 98; 103; 114–115; 121–123; 127; 132; 134–135; 162–165; 169–170; 172; 186; 189; 191; 194; 197–198; 218; 224; 227; 229; 233–234; 236; 238; 259; 269–273; 277–279; 282–283 incorporated 114–115; 132 proposition 94; 115–117; 126–128; 131; 133; 161; 166; 201; 223; 233; 236–239; 257; 296 prosody 2; 4; 7; 12–16; 18; 21; 25; 29; 35; 39–40; 47; 52–54 prosodic information 115 prosodic phrase boundary 3; 248 prosodic prominence 56 question constituent question 72; 185– 186; 225 question–answer pair 5; 71; 126; 242–243; 253–254; 287– 288; 291; 293; 298; 300–306; 309; 311 question marker 170; 179; 302 wh-question 7–8; 153–154; 171–174; 182; 186–189–190; 200; 212; 216; 227; 258; 260; 287; 291; 293; 296; 298; 301– 314 wh-in situ question 188; 199200; 206–207; 211–212; 214

Subject index 321 yes/no-question 58; 170–171; 173; 176; 192; 199; 202–203; 213; 250; 260 Relevance Theory 133 scope 27; 29; 31; 38; 44; 47; 164; 170–174; 181; 185; 200; 204– 214; 227; 232; 239; 236; 302 semantic composition 116 semantic representation 113–116; 127; 131; 133 stress 2–4; 11–56; 64; 70–71; 76; 134; 293 nuclear 15; 18; 22–23; 28–32; 36; 38; 40; 49 TAM (tense-aspect-mood) 83; 96; 186–214 TAM-marker 186–189; 195; 197–198; 202; 204; 206; 210; 212 Relative TAM 187; 188; 195; 106; 197–214 templatic morphology 15 tense 20–27; 33; 37–39; 50; 88; 93–94; 96; 105–106; 118; 132– 133; 141; 143; 149; 155–158; 162; 167; 179; 185; 188; 193; 196–199; 202–205; 208–214; 218–219; 232–239; 245; 267; 270–271; 274; 276; 279; 281– 282; 285; 298 relative 185; 187–188; 193; 196; 198; 202–209; 212; 267; 270; 276; 279 tone 3–4; 7; 12; 16–21; 24–30; 33– 34; 36–38; 40–67; 74; 77–79; 129; 132–134; 156; 158; 224; 226; 242; 245; 248; 259; 262; 269–276; 279; 281–285 floating 17; 21 polar 245; 248 tonal change 150; 269 tonal morphology 42 verb tone 269–275; 279

topic 1–2; 11; 15; 27; 37; 40–42; 48; 50; 55; 75; 79; 91; 101; 104; 113–115; 121–134; 145– 146; 152–153; 156–157; 162; 164; 167; 169–170; 176–184; 188; 193–195; 206; 225; 230; 234; 238; 259; 268; 282–283; 286–297; 309; 312; 314 background 114; 127 clitic-resumed 169 topicalization 146; 153–154; 156; 177–179; 182; 213 topic-focus field 162; 206 underspecification 41; 123; 135 structural 123 verb non–finite 89; 90; 98; 102; 103; 105 serial 44; 84; 88; 107; 109 verbal morphology 22; 225; 238; 270; 274 wh-extraction, see movement wh-phrase 139; 172–173; 175; 186; 189; 190–195; 200; 204–215; 287–312 wh-word 37; 120; 154–155; 204– 205; 312 word order 2–5; 8; 18; 29; 31; 44; 54; 81–85; 87; 96; 100–110; 113– 116; 119; 122; 128; 144; 151; 187; 193; 207; 215; 224; 242; 245; 284; 312; 314 VS order 23; 38; 122–127; 132; 240 word order alternation 84–85; 100; 102

Language index Afar 161–164; 166; 171–172; 174; 177–182; 184 Aghem 19; 46; 53; 86; 94; 96–98; 110; 295–298; 301; 310–314 Akan 6; 11; 268–271; 274; 276; 278–281; 284–285 Amharic 303; 304–313 Bafut 86; 89; 97; 106; 110 Buli 6; 267; 271–276; 279–284; 286 Byali 6–7; 223–240 Chamorro 190; 204; 216 (Nkhotakota) Chichewa (Nyania) 3; 11; 42; 56–57; 59; 67–68; 73; 77–78; 114–115; 126–129; 133–134; 262; 313 Chitumbuka 295 (Egyptian) Coptic 6–7; 185–188; 193–219 Dagbani 6; 267; 272; 275–276; 279–281; 283; 286 Ditammari 224; 240; 339 Ewe 6; 11; 268–269; 273; 276; 278; 280–281; 284–286 Ewondo 86; 88; 97; 103; 110 Fon 284–285 Fulfulde 267 Gungbe 3; 5; 7; 289–298; 301; 305–310; 312 Hausa 5–7; 12; 74–75; 77; 186– 193; 202; 204; 210–211; 214; 217–220; 241–263; 267; 277; 285 Ibibio 86–87; 96

Idoma 83; 86; 89; 93; 96; 105 Igbo 86; 89; 93; 96; 99; 103; 107 (Avu) Igbo 93; 96 Konni 284 Kaje 86; 88; 97; 103 Kana (Khana) 86–88; 97; 103; 108 Kikuria (Kuria) 24 Kikuyu (Gikuyu) 6; 42–43; 139; 148; 154; 156; 158–159 Kimatuumbi (Matumbi) 4; 16; 20– 25; 28–31; 37–38; 40; 50–51 Kinande (Nande) 56–57 Kinga 24 Kinyarwanda 33–34; 47 Kirundi (Rundi) 32–35; 49 Leggbo (Legbo) 32–35; 49 Lele 302; 304; 309; 312–313 Lelemi 6; 268; 270; 274; 276; 278; 280–282; 284 Luhaya 4; 29–31 Makua 24; 38; 42; 52 Mambila 43; 86; 94–98; 100; 110 Mandinka (Manding) 36; 41; 240 Naki 86; 94; 97–98; 107; 313 Nateni 224 Nen (Njen) 86; 94–98; 100; 109 Northern Sotho 4; 6; 54–64; 67; 73–79 Nsenga 113; 122–123; 127; 132– 133 Nupe 86; 89; 92–93; 96; 99; 102– 103; 107-108 Oromo 300; 301; 314 Proto-Bantu 86; 97; 103

324 Language index Safwa 24; 53 Setswana (Tswana) 33–35; 39; 42– 43; 47; 53; 77 Somali 6; 7; 161–184 Swahili 18; 47; 113; 123; 134 Swati 130–132 Tikar 86; 91; 97; 99; 110 Tumbuka 113; 128–130 Umbundu 16; 51

Vute 86–88; 97; 110 Waama 224 Xhosa 56–58; 73; 78–79; 115; 134 Yoruba 86; 90; 93; 96; 99; 102 Zulu 295; 298; 306–312