Relative Clauses in Cameroonian Languages: Structure, Function and Semantics 9783110469547, 9783110467611

Citation preview

Gratien Gualbert Atindogbé and Rebecca Grollemund (Eds.) Relative Clauses in Cameroonian Languages

Empirical Approaches to Language Typology

Editors Georg Bossong Bernard Comrie Kristine Hildebrandt Jean-Christophe Verstraete

Volume 58

Relative Clauses in Cameroonian Languages Structure, Function and Semantics Edited by Gratien Gualbert Atindogbé and Rebecca Grollemund

ISBN 978-3-11-046761-1 e-ISBN (PDF) 978-3-11-046954-7 e-ISBN (EPUB) 978-3-11-046767-3 ISSN 0933-761X Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at http://dnb.dnb.de. © 2017 Walter de Gruyter GmbH, Berlin/Boston Typesetting: Compuscript Ltd., Shannon, Ireland Printing and binding: CPI books GmbH, Leck ♾ Printed on acid-free paper Printed in Germany www.degruyter.com

Preface This volume, Relative clauses in Cameroonian Languages: Structure, function and semantics, offers analytical descriptions of aspects of relative pronouns and relative clauses in some Cameroonian languages. In nine chapters, the contributors have expatiated on eleven languages by providing descriptive information backed by sound illustrations to enable the reader to understand the structure of relative clauses in the languages studied. With the support of primary source data, a panoramic introduction and a new perspective-opening conclusion, these reflections on ten Bantu and non-Bantu Bantoid languages constitute an original contribution to the morpho-syntax of relativization in Cameroon languages. The goal of this volume therefore, is not only to enrich existing databases on facts of relative clauses, but also, with the analyses provided, to fuel ongoing debates on the nature, structure and function of relative clauses. So, the contributors, each from his/her language of expertise, offer to the scientific community as well as to any interested audience insights into the ways and strategies relative clauses are formed in ten under-described or fully undescribed languages. The papers proposed are not only descriptive, but analytical and thought-provoking. By so doing, they contribute to a volume that establishes a significant relationship between linguistic theory and typology on the one hand, and in-depth language description and areal studies on the other. In grammatical descriptions, a relative clause (RC) is often defined as a clause that describes the referent of a head noun or pronoun. However, the reality is much more complex, as illustrated by the diversity of relative clause strategies found within the languages studied. This confirms De Vries’ (2001) contention that, with 231 relative strategies in 176 languages spoken around the world, there are potentially 4608 types of relative clauses. For the purposes of this volume, we have selected languages from C ­ ameroon for its high linguistic diversity. The term “Africa in miniature”, commonly used to qualify Cameroon due to its concentration of all the characteristics of the African continent in terms of culture, landscape and climate, is also true linguistically. Indeed, Cameroon, situated in west Central Africa, is an incredibly diverse country with 285 languages (Lewis, Simons and Fennig 2016) belonging to three of the four large language families listed for Africa: Afro-Asiatic, NiloSaharan and Niger-Congo. The latter group constitutes the most important one in terms of speakers and geographical spread and it can be subdivided into three subgroups: Atlantic languages, Adamawa-Ubangi languages and Benue-Congo languages (that includes Bantoid and Bantu languages). The selected languages for this volume are spoken in a particularly interesting zone where Bantu and Bantoid languages are in contact, a zone from which the Bantu have split from the Bantoid subgroup. DOI 10.1515/9783110469547-202

vi 

 Preface

The papers contained in this volume are the results of a close collaboration between the editors and the contributors right from the onset. Indeed, the interest in this topic grew rapidly as we, the editors, contacted our colleagues who, not only have scientific interests in the languages of Cameroon, but also have a panoptic view of the complex nature of the central concern of this volume, the relative pronouns and clauses of the African languages. It is a fact that African languages are grossly under-described in general, and consequently still have a great deal to offer to the people speaking them and to the scientific arena as well. Their enthusiasm to the project was evidence that they each had a significant contribution to make. Thus, this project, which started sometime in August 2009 in Cologne at the 6th World Congress of African Linguistics with the then-PhD student, Rebecca Grollemund, finally came to maturity and took shape in 2015. Rebecca and I met in Cologne, and from conversation to conversation, we decided to address some of these fields of African linguistics that are still under-studied. We discussed the concept and contacted the potential contributors. The initial ambition was to launch the volume at the next WOCAL, i.e. the one held in Buea in 2012. Unfortunately, this first bet did not hold due to our various commitments. However, with perseverance, the first draft finally came out in 2015. We approached Mouton’s EALT and the project proceeded further from that point. After the various reviews by the editors of Mouton in 2015 and 2016, all the authors accepted to address the queries raised. We want to take this opportunity to thank all of our contributors for their patience, understanding, hard work, in short their utmost collaboration in making this volume a reality. We are particularly grateful to Rhonda Thwing who, despite her poor health condition by the time we wanted her to look at the queries of the reviewers, succeeded in finding the necessary physical and intellectual strengths to correct her manuscript. Special thanks are equally due to Peter, Larry and Emmanuel for accepting the suggestions made by the reviewers, and for taking the necessary time to effect the corrections. We really appreciate that sign of humility that honours them and teaches us lessons on scientific collaboration. Thanks also go to Jeff Good for stepping in to write the introduction to the volume despite his tight schedule. Our thanks also go to our PhD students, Nsom Christiane Alexandra and Dissake Koumassol Midinette Endurence, who engaged in the tedious task of cross-checking the references of the volume. Finally, we are immensely grateful to Kristine Hildebrandt for her tireless encouragements and her spontaneous offer to proofread this introduction. She enriched the text with valuable suggestions. Right from the beginning, our aim was to have original pieces, i.e. papers that would provide the reader with the real state-of-affairs of the theme, mainly by 1) proposing sound analytical descriptions of relative clauses in the selected languages, and 2) revealing the diverse natures and structures of those subordi-

Preface 



 vii

nate clauses that modify a noun or a noun phrase and give essential information to define or identify the person or thing the speaker is talking about. Thanks to the constructive queries of the external reviewers from Mouton, the authors have brilliantly delivered on our agenda by writing smart, informative pieces that bring readers into the heart of aspects of what relative clauses are in some Cameroon languages. Just like Jeff Good rightly wrote in his introduction to the volume, At first glance, one might not expect to see much grammatical diversity in a volume focusing on languages that are drawn from a subgroup of a subgroup of Niger-Congo, which is how Bantoid languages are currently classified. However, as one goes through the various chapters, it is hard not to be struck by the diversity of relative clause encoding found within them.

That alone is a big reward for us, especially as it comes from a contributor with whom we had not had prior exchanges in that direction. We wish that these bold, visionary and thought provoking papers will be the starting point of more research into the phrasal linguistics of Cameroon languages. We would be particularly excited for example to see subsequent volumes on the subject matter as we are almost convinced that the many other languages from the 284 listed for Cameroon will have their own stories to tell about the ways relative pronouns and relative clauses behave. Gratien G. Atindogbé & Rebecca Grollemund Buea & Columbia, 27 March 2017

Contents Preface v Abbreviations

xi

Jeff Good 1 Kaleidoscopic variations on grammatical themes: Relative clauses in Bantoid languages of Cameroon 01 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman 2 Accessibility and demonstrative operators in Basaá relative clauses Mark Van de Velde 3 The augment as a construct form marker in Eton relative clause constructions 47 4

Théophile Ambadiang Relative clauses and relativization processes in Nugunu

5

Florence A. E. Tabe and Gratien G. Atindogbé Kenyang relative clauses 89

John R. Watters 6 Relative clause in (Western) Ejagham

67

119

Pius N. Tamanji and Blasius Achiri-Taboh 7 A prolegomenon to the syntax of the relative clause in the Eastern Grassfields Bantu borderland 145 Jesse Lovegren and Rebecca Voll 8 Relative clause constructions in two Yemne-Kimbi languages Rhonda Thwing 9 Relative clauses in Vute grammar and discourse Marieke Martin 10 Relative clauses in Wawa

233

209

169

17

x 

 Contents

Denis Creissels 11 Conclusion 261 References Index

277

267

Abbreviations NB: where there is variation or conflict in usage, we provide the abbreviation with both meanings followed by the name(s) of the authors. (a), (b), (c) verbal conjugation class (H) floating high tone (L) floating low tone | light phonetic break || heavy phonetic break 1, 2, 3, … 19(a) noun classes 1PL, 2PL, 3PL plural pronouns 1SG, 2SG, 3SG singular pronouns ADJZ adjectivizer
 ADP additive verb extension: adds participant to event ADSEQ additive verb extension: adds event in sequence with previous event ADSIM additive verb extension: adds event simultaneous with previous event ADV adverb ADVZ adverbializer AGR agreement ALL allative verb extension, direction away from referential center AM associative marker ANA anaphoric demonstrative anaphoric relative marker ANAPH.REL anterior (aspect) ANT APPL applicative ASP aspect ASS associative [Lovegren & Voll] associative (≠genitive) [Martin] ASS AUG augment
 AUX-COMP auxiliary complementizer CAUS causative CC concord consonant CF contrafactual focus CLFT sentence enclitic: cleft sentence consecutive CNS comitative COM DOI 10.1515/9783110469547-204

xii 

 Abbreviations

COMP complementizer CON connective (genitive) CONC concessive/concessive auxiliary or verb extension COND conditional the floating L which marks a consecutive event CONS CONSEC consecutive CONSTF constituent focus CONT continuous/progressive copula COP CP complementizer phrase defocused/dependent form of aspect marker D DAT dative DEFZ definitizer DEF definite DEM demonstrative (definite) determiner DET DF defocused form DIM diminutive directional DIR DIST distal DP determiner phrase DS dummy subject DSF disfluentive Emph emphatic EQ equative auxiliary existential auxiliary EX near future F1 F2 far future FIN final future/polite negative FNEG focus FOC FOC.Q focus question FPPR final form of the personal pronoun FRUST frustal
 FSL far-form-speaker/listener FUT, FUT1...2 future FUT1, 2, 3 near, middle, remote Future final vowel FV G “G-FORM” a multifunctional verbal morpheme



Abbreviations 

 xiii

GEN NP enclitic: association of two nouns or the end of N + RC unit [Thwing] GER gerundive HAB habitual HORT hortative IDEO ideophone IMP imperative IMPERS impersonal (pronoun) IMPFV imperfective INCH inchoative INCL inclusive INDF indefinite (demonstrative) INF infinitive INT intransitive INTERJ interjection INTS intensifier IO verb extension, marks addition of indirect object/benefactee/ addressee [Thwing] IO indirect object [Martin] IPA international phonetic alphabet IPFV imperfective IPRF imperfect IRR irrealis (mood) IT iterative LOC locative (form of noun) general locative LOC.GEN specific locative LOC.SPEC logophoric pronoun LOG LT link tone M mark(ed) verb form N nominal N1 form of a noun that is modified in some way, the head noun of a NP narrative chain NAR noun clause NC NEG negative marker for aspect or mood [Thwing] negative [Martin] NEG NEG.EXIST negative of existence   NF non-final NMLZ nominalizer


xiv 

 Abbreviations

NOM nominalized NP noun phrase NPAH noun phrase accessability hierarchy NPPR non-final form of the personal pronoun relative noun phrase NPREL NR non-restrictive O pronoun used to encode object at the end of a relative clause OBJ object object pronoun ObjPro OBL oblique past tense P0...3 P1 near past tense P2 far past tense P3 remote past tense PART participle PFV perfective (aspect) PL plural PN person’s name sentence enclitic: polite command POL POSS possessive POT potential mood PP past participle PPR personal pronoun PRED predicate PREP preposition PREP preposition PRF perfect PRN pronoun PRO proform PROG progressive proximal (demonstrative) PROX PRS present near, middle, remote past [Ambadiang] PST1, 2, 3 past tense [Lovegren & Voll] PST today past tense [Jenks, Makasso & Hyman] PST1 general past tense [Jenks, Makasso & Hyman] PST2 Q question [Tamanji & Achiri-Taboh] sentence enclitic: question [Thwing] Q QP quantifier phrase QT sentence enclitic: quotation



Abbreviations 

 xv

RC relative clause RE relativizing element RED reduplicant REF reflexive possessive adjective: always refers to the topical participant REF.GEN REL relative operator [Jenks, Makasso & Hyman] REL relative marker [Tabe & Atindogbe] REL relativizer [Watters; Lovegren & Voll] relative [Martin] REL REL.DEF relative definiteness marker relative determiner REL.DET relative marker REL.M repetitive dependent REP.DEP resumptive pronoun RPR RR relative relator inflection with reference to result or consequence RRC RS resultative SBJ subject (agreement) SBJV subjunctive SEQ sequential chain SF subject focus SG singular SIM simultaneous chain subject marker SM SP southern dialect present SUBD subordinator
 particle used to create suspense before a climactic action SUSP tense aspect mood TAM TMN terminative TNS tense topic field form TOP TPZ topicalizer tense phrase TP verb V venitive verb extension, direction toward referential center VEN VENT ventive VFOC verum focus

Jeff Good

1 Kaleidoscopic variations on grammatical themes: Relative clauses in Bantoid languages of Cameroon Abstract: The diversity of relative clause structures found in Bantoid languages of Cameroon is quite surprising given how closely related these languages are to each other and the fact that they are in relatively close contact. Important parameters of variation are: (i) whether or not the head of the relative clause is coded as being modified, (ii) which arguments can be relativized in a given relative clause construction, (iii) the way that the “gap” in the relative clause is encoded, and (iv)  similarities and dissimilarities holding between restrictive relative clauses and functionally related constructions (e.g., non-restrictive relative clauses, adverbial subordinate clauses, and clefts). Areas where further research would seem likely to yield interesting results include detailed consideration of the historical source of relative clause markers, the distribution and nature of resumptive pronouns, and restrictions on the verb forms that head relative clauses.

1 Relative clauses in Cameroon Current reference sources, such as the Ethnologue (Lewis, Simons, and Fennig 2016) and Glottolog (Hammarström et al. 2016), suggest that there are around 300 languages spoken in Cameroon. Representatives of three of Africa’s four major phyla are found in the country, though it is dominated by Niger-Congo languages, in particular Bantoid languages of the Benue-Congo subgroup. This volume reflects that fact. All of the contributions focus on relative clause constructions in Bantoid languages, with the genealogical diversity of this large group, as found in Cameroon, well represented. The Bantoid group includes the Bantu languages, which dominate SubSaharan Africa. The vast majority of these are spoken outside of Cameroon, but they are also found in the south of the country and relativization strategies in Bantu languages are described here for Basaá (Jenks, Makasso, and Hyman), Eton (Van de Velde), and Nugunu (Ambadiang). Bantu’s closest relatives are the non-Bantu Bantoid languages, and, here, there are contributions representing Jeff Good: University at Buffalo, [email protected] DOI 10.1515/9783110469547-001

2 

 Jeff Good

Bantoid languages from a number of its subgroups: Mamfe (Tabe and Atindogbé), Ekoid (Watters), Grassfields (Tamanji and Achiri-Taboh), Yemne-Kimbi (Lovegren and Voll), and Mambiloid (Thwing, Martin). At first glance, one might not expect to see much grammatical diversity in a volume focusing on languages that are drawn from a subgroup of a subgroup of Niger-Congo, which is how Bantoid languages are currently classified. However, as one goes through the various chapters, it is hard not to be struck by the diversity of relative clause encoding found within them. Some of this is, no doubt, connected to wider patterns of morphosyntactic diversity of the Bantoid subgroup, where languages at its northwestern edge show a mix of relatively reduced segmental morphology and increased tonal complexity when set against the wellknown agglutinating patterns of Bantu languages (see, e.g., Hyman 2004 and Good 2012). Consider, for instance, the examples of relative clauses given in (1)–(3). Example (1) is drawn from Chichewa, a Bantu language of Malawi that can stand in for Bantu languages of more or less the “canonical” type, with its relatively transparent agglutinative morphology. The other two examples are drawn from chapters of this volume. The example in (2) is taken from Ambadiang’s description of Nugunu, and the example in (3) is from Watters’ description of Ejagham. (1)

mkángó u-méné alenje á-ku-sáka 3.lion 3-rel 2.hunters 2.rel-prs-hunt.fv “The lion that the hunters are hunting” (Mchombo 2004: 17)

(2) mɔ-ádɔmbɔ́ mɔ́-ɔnyɔ a-bá-ná-bɔla ínoni 1-child 1-dem 3sbj-pst1-m-arrive today “The boy who has arrived today.” (Ambadiang, this volume) (3) ǹ-tèm ò-mé ǹ-nyò à-nàm-éʹ ˊ bì-yù 1-friend 1-my rel-1 1.pfv-buy-ConstF 8-yam “My friend who bought yams” (Watters, this volume) The Chichewa example in (1) reveals a relative clause structure where a modified noun comes before a relative clause introduced by a relativizer. This relativizer is different from a relative pronoun, of the kind that are well-known from European languages – but otherwise are quite rare (Comrie and Kuteva 2013) – where there is a special class of forms signaling the role of the relativized argument in the relative clause, e.g., whom as used to indicate that the relativized noun is functioning as an object in a phrase like the man whom I saw. The Chichewa relativizer does, however, show agreement with the head noun, in this case, by being inflected for Class 3, following standard Bantuist numbering conventions



1 Kaleidoscopic variations in Bantoid relative clauses 

 3

(see Katamba 2003). The verb in (1) also shows different tone marking on its subject prefix than it would if it were in a main clause, adding further encoding of the clause’s relative status (Mchombo 2004: 18). Nugunu, along with Basaá and Eton, is spoken in the southern end of the north-south range of the languages covered in this volume and shows a structure that is roughly analogous to what is found in Chichewa. Northwest Bantu languages like Nugunu are known, in general, for their typological distinctiveness when set against Bantu languages spoken to the south and east of the Bantu area (Nurse and Philippson 2003: 5). However, a comparison of (1) and (2) shows clear similarities: a head noun followed by a relativizing element with prefixal agreement and a verb showing some morphology distinct from what would be found in a main clause, in this case most clearly seen in the form of the -ná- prefix (coding a “marked” verb form in Ambadiang’s terminology). The relativizer in Nugunu is homophonous with a demonstrative, but, otherwise, it is morphologically and functionally quite similar to the Chichewa relativizer exemplified in (1). The relative clause in (3), from Ejagham, spoken to the north and west of Nugunu, overlaps with the other examples in showing a noun followed by a relativizer introducing a relative clause. Here, though, even a casual inspection reveals differences between basic word structure in Ejagham (where roots tend to be shorter than in a narrow Bantu language) and the form of the relativizer, which still shows agreement, but suffixing agreement rather than prefixing agreement. Throughout this volume, one sees patterns like this again and again. Each language partly overlaps in its relativization strategies with some of the others while presenting interesting new “twists” on the possibilities for encoding of relative clauses. In section 2 of this introduction, I will summarize each of the papers in their order of presentation in the volume, which roughly follows a “south to north” trajectory, starting with the languages showing structures closer to canonical Bantu languages, which are likely to be more familiar to most readers, and moving northward through the Cameroonian Grassfields and into the region of the Adamawa plateau. Because there is, by design, significant overlap in many of the descriptive topics covered in each paper, the summaries will attempt to give a sense of the overall coverage of the volume in a way that highlights what makes the relative clause strategies of each language distinctive. In section 3, I will then briefly consider some of the research questions that emerge when looking at all of the chapters together as a way of offering additional context to the volume that complements the remarks made in the concluding chapter by Creissels, where he highlights points of special typological and comparative interest.

4 

 Jeff Good

2 Overview of the chapters The first three papers of the volume discuss relative clauses in languages classified as part of the Narrow Bantu group, all within its A zone, following the classificatory scheme developed by Guthrie (see Maho 2003). Jenks, Makasso and Hyman consider relative clauses in Basaá. Like all of the papers in the volume, the discussion includes a basic descriptive overview of relative clauses in the language. It also considers the relationship between relative clauses and demonstratives, since both are in complementary distribution in some constructions, and, more intriguingly, they are the only elements in the language which permit the appearance of a nominal prefix that they associate with the augment prefix found variously throughout Bantu languages (de Blois 1970). The authors further develop a formal syntactic analysis to account for this pattern, as well as other features of Basaá relative clauses. The typical structure of a Basaá relative clause is not dissimilar to the examples given in (1)–(3). A head noun is followed by an optional relativizer which is homophonous with the “near hearer” demonstrative found in the language (including showing the same range of concordial forms). An example of a relative clause in Basaá is given in (4). ↓jɛ́ (4) í-mut1 (nú) [ ___1 a bí bíjɛ́k ] aug-1.person 1.rel 1.sbj pst2 eat 8.food “The person that ate the food” (Jenks, Makasso and Hyman, this volume)

In (4), the primary marker of the relative clause is the element nú, showing Class 1 agreement. One also finds the vocalic prefix í- on the head noun, which is one of the forms of the augment element mentioned above. Augment morphemes have various functions in Bantu languages, and these can generally be said to be related to notions like definiteness, specificity, or givenness (see, e.g., Maho 1999: 62 and also the discussion of the paper by Van de Velde below). It is not entirely surprising to see an augment appearing on the head noun of a relative clause construction given that the modifying function of relative clauses means that they will be typically be associated with entities that are already given in the discourse. However, this does not explain why the augment is only seen with relative clauses or demonstratives in Basaá. This gives the Basaá augment the interesting morphosyntactic classification of being a head-marking strategy for encoding that a noun is modified by a subclass of possible modifiers in a language. Van de Velde looks at relative clauses in Eton, a language which is relatively closely related to Basaá. Eton shares with Basaá the fact that an augment element is part of the morphosyntactic encoding of relativization. Like Basaá, this element has the shape of a segmental prefix í- or a high tone on a noun’s class prefix, a striking formal overlap. Moreover, the augment in Eton is obligatory in the same



1 Kaleidoscopic variations in Bantoid relative clauses 

 5

contexts where it is found in Basaá, namely, when a noun is modified by a relative clause or a demonstrative. Unlike Basaá, the augment is clearly the primary coding mechanism for relative clauses in Eton since it is the only consistent formal device found in relative clause constructions. This leads Van de Velde to classify it as a construct form marker, following Creissels’ (2009) adaptation of this traditional term from Semitic linguistics to a variety of phenomena in African languages where the presence of nominal modifier is encoded on the noun itself. Eton also employs a relativizer of invariant form yá, though this does not appear to be especially commonly used, as well as special relative verb forms in some cases. Van de Velde argues that the function of the augment as an encoding device should not be understood in terms of a definiteness/indefiniteness distinction, as suggested by Jenks, Makasso and Hyman for Basaá but, rather, in terms of localizing modification -that is, as a kind of modification that helps identify a referent noun phrase in space or discourse. This analysis is then used to sketch a historical scenario of the development of the head-marking strategy of relative clause encoding found in Eton (and, by extension, Basaá), which is significant as a contribution to our understanding of the development of augment markers within Bantu, as well as of grammaticalization pathways for the rise of construct noun forms more generally. Like the other papers in the volume, Jenks, Makasso and Hyman and Van de Velde consider how relativization of different kinds of arguments (e.g., subject, object, possessor, etc.) is encoded, following a long line of work on the so-called Accessibility Hierarchy in relativization (Keenan and Comrie 1977). The basic pattern that emerges from their descriptions is that higher arguments in the hierarchy (e.g., subject and object) are associated with gaps in the relative clause while lower arguments (e.g., obliques and possessives) are associated with resumptive pronouns. This can be seen in the examples given in (5) and (6). (5)

íbá màyêm í-bá mə̀-à-H-jɛ̀m aug-ii.dem 1sg-sp-rel-know “The ones (siblings) that I know” (Van de Velde, this volume)

↓pád (6) íkɔ̂b mə̀tə́ ópùm éèy yɔ̀ ìnə̀ və̀? L H í-kɔ́b mə̀-tə́ -pád òpùmá èːy yɔ̀ ì-nə̀ və̀ aug-9.hook 1sg-prs inf-pick lt orange with ix-ppr ix-be where “Where is the hook with which I pick oranges?” (Van de Velde, this volume)

In example (5) the relativized noun has the role of an object, and there is simply a gap in the relative clause in the postverbal position where the object would otherwise be expected. In example (6) the relativized noun is the object of a

6 

 Jeff Good

preposition, and, in this case, the preposition in the relative clause is followed by a pronominal form showing the same noun class as the head noun of the clause. There are complications, however. For example, in Eton, relativization of possessors of subjects requires a resumptive possessive pronoun while relativization of possessors of non-subjects does not, producing an asymmetry between subject and non-subject arguments, a pattern which appears in different guises in other languages described in this volume, and, as will be discussed here below, in Bafut and Medumba. The paper by Ambadiang on Nugunu treats another Bantu language from Guthrie’s A zone. Like Basaá, relative clauses in Nugunu make use of a relativizer homophonous with a demonstrative (in both cases one involving proximal rather than distal semantics). Another salient coding mechanism involves the use of special verb forms. The opposition between the regular form of the verb and the form used in relative clauses can be seen in the pair of sentences presented in (7). (Example (7b) is repeated from (2).) (7)

a. mɔ-ádɔmbɔ́ mɔ́-ɔnyɔ a-báa-bɔla ínoni 1-child 1-dem 3sbj-pst1-arrive today “This boy has arrived today” b. mɔ-ádɔmbɔ́ mɔ́-ɔnyɔ a-bá-ná-bɔla ínoni 1-child 1-dem 3sbj-pst1-m-arrive today “The boy who has arrived today” (Ambadiang, this volume)

In (7a) the verb is coded as a main clause verb by virtue of lacking any special affix other than a subject marker and a tense marker. In (7b), the verb is coded with an additional -ná- prefix, which codes the relative function of verb form in one of the language’s past tenses. Among other things, this coding results in the element glossed as a demonstrative as not being associated with demonstrative semantics, since it now functions as a relativizer, unlike what is seen in (7a), where demonstrative semantics are associated with the same form. Special verb forms in relative clauses are seen in other languages described in this volume (as well as Africa more generally (Hyman and Watters 1984: 234)), though there is variation in the extent to which there is syncretism between relative and nonrelative forms. For instance, Ambadiang’s description of Nugunu suggests that the distinction between non-relative and relative verb forms is quite pervasive in the language, while, in Eton, Van de Velde describes overt coding of the distinction as being quite restricted, limited to a few tense-mood-aspect configurations and a special form of the verb ‘be’. Verb forms of this kind are not limited to relative clause contexts in Nugunu but can also be found in, for instance, subordinate clauses, clefts, and content questions (also known as Wh-questions). In the case of subordinate clauses, there



1 Kaleidoscopic variations in Bantoid relative clauses 

 7

can be a clear connection between certain kinds of subordination and relative clauses in cases where the subordinate clause is expressing a notion that could be seen as modifying referential elements with meanings like ‘time’, ‘place’, ‘cause’, etc., making the subordinate clause literally translatable along the lines of the time when... or the reason why.... For clefts and content questions, the connection with relative clauses seems to center around the complementarity between focus and backgrounding (a point noted as early as Schachter 1973), where the encoding of one element of a clause as being in focus can be partly achieved by encoding other material in the clause as backgrounded. Relative clauses are similarly backgrounded insofar as one of their core functions is to restrict the interpretation of a noun rather than to provide new information. Tabe and Atindogbé’s study of relative clauses in Kenyang, a non-Bantu Bantoid language of the Mamfe group, reveals a remarkable degree of variation in their encoding. For instance, while Noun-Relative Clause order predominates in the languages described in this volume, Kenyang additionally attests structures where the relativized noun appears to remain within the relative clause by virtue of following the relativizer, as seen in the opposition in the Kenyang examples given in (8). (8) a. m̀-mɔ̀ á-nɛ̀ á wáy m̀-mɛ́n à ʧwè 1-child aug-rel 1.agr.prf kill 9-goat 1.agr.prf enter kɛ̀-nɔ̀ŋ 14-prison “The child who killed the goat was imprisoned” m̀-mɔ̀ á wáy m̀-mɛ́n à b. à-nɛ̀ aug-rel 1-child 1.agr.prf kill 9-goat 1.agr.prf ʧwè kɛ̀-nɔ̀ŋ enter 14-prison “The child who killed the goat was imprisoned” (Tabe and Atindogbé, this volume) Structures like the one found in (8a) are seen throughout the papers in the volume, where a noun is followed by a relativizing element that, in turn, introduces the relative clause. (As with some of the other relativizers discussed to this point, this one shows a similar shape to a proximal demonstrative.) More unusual is the structure in (8b), where the relativizer is still present but the relativized noun follows the relativizer. This pattern for relative clauses is, perhaps, connected to a general freedom of Noun-Modifier order Kenyang, as described by Tabe and Atindogbé. In addition to making use of a relativization strategy involving the use of a relativizer, Kenyang also employs a strategy where there is no relativizer but special relative verb forms are used instead, further adding to the “mixed” nature of its relative

8 

 Jeff Good

clause constructions. A remarkable aspect of these relative verb forms is that the prosodic aspects of their marking involve a fairly elaborated tone reversal: In comparison to non-relative forms, high tones and low tones interchange, as well as rising and falling ones. Tabe and Atindogbé also describe an interesting opposition between the relativizer exemplified in (8) and another relativizer with a stem form of -rɛ́, characterizing the former as being associated with new information and the latter with old information. While the scope of the chapters in this volume does not give them the opportunity to explore this semantic opposition in detail, further investigation into the functions of these two complementizers seems to be a worthwhile area for future study into information structure relations within noun phrases containing relative clauses. This opposition is, to some extent, paralleled by a noteworthy contrast between a relativizer introducing restrictive relative clauses and another introducing non-restrictive ones described for the Mambiloid language Wawa as discussed below. The paper by Watters considers relative clauses in Ejagham, a language classified in the Ekoid group of Bantoid and spoken adjacent to Kenyang. Ejagham relative clauses generally have a structure comparable to what is found in a number of the languages already discussed where a noun is followed by a relativizer, which is in turn followed by the relative clause. The relativizer further agrees with the head noun of the noun phrase, though the agreement is atypical for the languages seen here insofar as it is suffixing rather than prefixing. Furthermore, unique among the languages described here, Ejagham has three series of relativizers that differ with respect to the role of the noun and its modifying phrase. One serves as a general relativizer, another is used for locative modification, and the final one for property modification (i.e., adjectival function). The three locative series show extensive formal overlap but are distinguished tonally in some forms. An example from the locative series is given in (9). (9) à-tə̂m fə́nè à-rúì ǹ-gyìm á-kí-jǐ 1.pfv-bend.knees consequently 1.pfv-get.up 9-standing 2-cont-go nà ǹ-gòn ǹ-nyǐ è-bhǐn with 9-hornbill rel.loc-9 5-farm “She consequently got up and was going with Hornbill to the farm.” (Implicitly: “[on the road which goes] to the farm”) (Watters, this volume) The locative element ǹ-nyǐ in (9) introduces the nominal modifier è-bhǐn ‘farm’ as a modifier of location. It differs from the general relativizer by virtue of its rising tone. (The general relativizer has the form ǹ-nyì.) As can be seen in (9), the modifying phrase in this kind of construction is a not a true clause insofar as it lacks



1 Kaleidoscopic variations in Bantoid relative clauses 

 9

a verb, perhaps indicating that the function of ǹ-nyǐ lies somewhere in between that of a relativizer and a more general linking element (see, e.g., ­Creissels et al. 2008: 121–122 for further discussion). The same observation holds for forms from the adjectival series that also introduce phrases lacking a verb. At the same time, the parallel between these different series of relativizers and relative pronouns in European languages is noteworthy, especially given the typologically unusual status of European-type relative pronouns: In both cases, the element that serves to link a noun to its modifying phrase codes the semantic relation holding between the noun and that phrase. As with other languages in this volume, relative clauses are not only coded through the use of a relativizer in Ejagham but also via special verb forms for verbs coded for the perfective or imperfective aspect. In particular, forms generally associated with constituent focus are employed. These forms are also found in content questions and their associated answers and are discussed in detail in Watters (2010: 357–362). The paper by Tamanji and Achiri Taboh considers relative clauses in two Narrow Grassfields languages, Bafut and Medumba, spoken in the Northwest and West regions of Cameroon respectively. From the perspective of Bantoid, both languages are quite closely related, though they are not particularly close within their subgroup and are not in direct contact with each other. Relative clause formation is similar in the two languages, and Tamanji and Achiri Taboh are especially interested in the use of an element found in both of them, which they label a relative definitizer. This element, with form láá in Bafut and lá in Medumba, appears at the end of the relative clause. An example of its use in Bafut use is given in (10). (10) màŋgjè jìí à ká lò mə́ á lí̵gá sí̵gí̵nì̵ láa woman rel.ded rel.m rp cultivate well rel.def sm tns leave “Any woman who cultivates (farm) well will leave.” (Tamanji and Achiri Taboh, this volume) There are three morphemes encoding the relative clause status of the clause modifying the word màŋgjè ‘woman’ in (10). In addition to the definitizer láá, there is also a relativizer of form jìî which agrees in class with the head noun of the clause, as well as non-agreeing complementizer mə́, which is also used as a more general subordinator (e.g., to introduce reported speech). Medumba codes relative clauses in a similar way, though a general subordinator is not found in relative clauses in the language. None of these coding elements is obligatory, meaning that there are a variety of ways for relative clauses to be formally expressed, though with potential differences in their pragmatic interpretation depending on which of these markers is present.

10 

 Jeff Good

Tamanji and Achiri Taboh give a thorough overview of relative clause formation in these two languages, pointing out the similarities and differences between them and also consider what the best analysis is for the definitizer, concluding that it should be interpreted as a marker of definite noun phrases that have already been introduced in the discourse. This is opposed to the transformationalist analysis of Kouankem (2012) which treats the definitizer along the lines of a primary relativizer whose final positioning is the result of leftward movement of the relative clause from a position following it to a position before it. Bafut and Medumba both, at first glance, appear as though they might disobey the predictions of Keenan and Comrie’s (1977) Accessibility Hierarchy, introduced above. Specifically, their formulation of the hierarchy includes a constraint that predicts that any relative clause formation strategy “must apply to a continuous segment of” the Accessibility Hierarchy. However, in Bafut and Medumba relativized subjects and obliques are associated with a marker in the relative clause (e.g., a resumptive pronoun), while relativized objects can leave a gap. It therefore appears that subjects and obliques are patterning together to the exclusion of objects, in violation of the Accessibility Hierarchy, which places objects between subjects and obliques. In this case, however, the apparent violation seems likely due to grammatical constraints forcing the presence of some element in subject position, not just in relative clauses, thus not making it a clear counterexample (see also Keenan and Comrie 1977: 85–86). Lovegren and Voll present descriptions of relative clauses in two YemneKimbi languages, spoken in the same general area as Bafut, but somewhat further to the north. Yemne-Kimbi is a referential label for a cluster of Bantoid languages spoken at the northern edge of the Grassfields region, with the details of their genealogical classification being otherwise unclear. Regardless of their genealogical affiliation, the two languages they discuss, Mungbam and Mundabli, are in close contact. Representative examples of relative clauses from the Missong dialect of Mungbam and from Mundabli respectively are given in (11) and (12). (11) í-ɕɛ̋ jɛ̄ nə́ í gbè í-bwɩńsə́ jɛ̄ 10-fowl 10.det rel 10.pro fall 10-red 10.det “The fowls which fell are the red ones.” (Lovegren and Voll, this volume) (12) dʒǔ nō̤ bə̄ kə̀ lə̌ kpī̵ y-ɔ̄ (yì) tō 9.goat comp impers pst make die 9-rel 9.pro 7/8.day b-ɔ́ ŋgɔ᷆ kə̀ bān áná būbūbūbū 8-det upon pst be.white like.that ideo.white “The goat which was killed on that day was completely white.” (Lovegren and Voll, this volume)



1 Kaleidoscopic variations in Bantoid relative clauses 

 11

Both examples in (11) and (12) employ an optional non-agreeing relativizing element that introduces the relative clause. In Mundabli, this element can introduce other kinds of subordinate clauses as well, which is why it is glossed as a general complementizer in (12), and, in this regard, it is comparable to Bafut, discussed above. Four of the five dialects of Mungbam also make use of a non-­ agreeing relativizer, as exemplified in (11), but the Biya variety does show agreement on its relativizer (see (13)), which is otherwise of similar phonological shape. Overall, verbs in relative clauses in these languages show comparable coding for categories like tense and aspect as is found in main clauses. Mundabli relative clauses are further coded by an element that is homophonous with the distal demonstrative and appears in immediate postverbal position. This element agrees with the head noun of the relative clause, thus being a Class 9 form yɔ̄ in (12) since the word dʒǔ ‘goat’ is Class 9. In his concluding remarks, Creissels points out that a coding strategy involving an immediate postverbal marker, also found in the nearby Naki language, is surprising. It is perhaps connected to the special grammatical status of the immediate postverbal position in languages of this area, which is typically associated with the expression of focus (see, e.g., Good 2010). As also pointed out in the concluding remarks by Creissels, Mungbam is remarkable in an African context in that its relative clause construction appears to be better understood as a more general noun-modifying construction insofar as it does not require that the head noun have a role in the modifying clause. This is exemplified in (13), drawn from the Biya variety, where the noun ìʤī ‘sound’ is modified using the relative clause construction by a clause that describes the kind of sound being made but where there is no gap or resumptive pronoun corresponding to where the word for ‘sound’ would appear. (13) ì-ʤī ì-nɩ̄ bű gbà ɲà kə̄-tɔ̄ kə̄ 5-sound 5-rel 2.pro cut.ipfv stay.ipfv 12-tree 12.det mə̄ nɔ̀ bâŋ ɲà make.ipfv block.ipfv stay.ipfv 1sg “The sound of them cutting the tree disturbs me.” (Lovegren and Voll, this volume) The final two papers of the volume, by Thwing and Martin, discuss two languages of the Mambiloid group, Vute and Wawa. While Mambiloid languages are classified as Bantoid, they are considered to be more distantly related to Bantu languages than the other Bantoid languages discussed to this point and have the most eastern distribution of the languages covered. Within Mambiloid, Vute and Wawa are currently classified very closely together. The languages differ, however, in their use of a relativizer. Vute lacks any overt relativizer, and a primary means of

12 

 Jeff Good

encoding relativization is simply the presence of a gap or resumptive pronoun in a relative clause, the latter of which is seen in (14). By contrast, Wawa makes use of two relativizers, one for restrictive relative clauses and one for non-restrictive ones, as seen in the examples in (15). (14) njɔdì ŋgə́ yi cuwo njɔdì ŋgə́ yi (H)cù-wò-Ø wealth 3sg p2 pfv.return-ven-pfv.d “The wealth that he returned with [it]”

ɓe ɓe with

kíí kí=í ana.prn =gen (Thwing, this volume)

nòr kí níī Wàwà sɛ́ì (15) a. mə dōníī nòr kí níī Wàwà sɛ́=ì mə dō-níī 1sg like-neg person rel neg Wawa neg.cop=ass “I do not like people who are not Wawa.” b. nòr kə́ mə dagā éérə́ nòr kə́ mə dag-ā ér-ə́ person rel 1sg see-pfv.part leave-pfv “The person, who I saw, left.” (Martin, this volume) Even though it lacks a dedicated relativizer, Vute still codes relative clauses fairly clearly using a “bricolage” of other strategies, including the presence of verb forms associated with out-of-focus contexts, the use of an anaphoric pronoun as a resumptive form, and the presence of a genitive (or associative) marker at the end of a relative clause. All of these are seen in example (14) from Thwing. Only the first of these features is found in all relative clauses. Wawa relative clauses employ a similar range of encoding devices to Vute. For instance, in affirmative clauses, a different verb form is used from main clauses which Martin categorizes as a participle. Relative clauses can also appear with one of two final markers, the first is homophonous with a distal demonstrative, and the second has associative function and can be seen in (15a). These two features of relative clause encoding in Wawa bear clear parallels to what is found in Vute. As already discussed, unlike Vute, Wawa also encodes relative clauses with a relativizer. Many of the languages described in this volume make use of relativizers, but only Wawa is described as having a paradigmatic distinction between a relativizer encoding restrictive relative clauses (15a) and one encoding nonrestrictive relative clauses (15b). Most of the discussion of relative clauses in this volume is focused on restrictive relative clauses, though some authors do discuss how non-restrictive clauses are coded in the languages described. Tamanji, for instance, describes prosodic encoding of the restrictive/non-restrictive distinction in Bafut, and Thwing remarks that non-restrictive relative clauses have the same morphosyntactic encoding as restrictive ones with the additional constraint



1 Kaleidoscopic variations in Bantoid relative clauses 

 13

that only subjects and objects can be relativized with non-restrictive meaning. Wawa is distinctive by virtue of having a salient and obligatory strategy for encoding the restrictive/non-restrictive distinction. Overall, the impression one gets from reading this volume, which only represents a small fraction of Cameroon’s languages, is that there exists a surprising amount of diversity in the encoding of relative clauses given how typologically similar these languages are to each other on a global scale. All, for instance, have basic SVO word order, employ broadly similar noun phrase structure, and code for a similar range of verbal tense and aspect categories. Nevertheless, each language carves out its own way of encoding relative clauses, largely drawing from a common pool of strategies. Remarkably, closely related languages have significant differences in their relative clause marking, showing that there is more to the differences than diachronic drift. In the final section of this introduction, I will briefly discuss some questions raised by the papers here with respect to future areas to consider in the study of relative clauses in the languages of Cameroon.

3 Considerations for future research on relative clauses in the languages of Cameroon In a volume such as this one, where each of the chapters raises interesting questions in their own right, it is admittedly a subjective enterprise to chart out the implications of the collection as a whole for future research. Nevertheless, there are a number of recurrent topics that seem worth highlighting here. First, even though the volume is focused on restrictive relative clauses, a number of the chapters point out ways that the languages described make use of other kinds of noun modification strategies involving verbs and verb phrases. This suggests a need for further studies of these other kinds of nominal modification and the ways that they are similar to, and different from, restrictive relative clauses. The presence of a dedicated strategy to encode non-restrictive relative clauses in Wawa was discussed above. To pick a few other examples, Ambadiang describes a strategy for nominal modification in Nugunu involving an associative element (or “connective” in his terms) followed by a non-finite verb, which has very similar function to a relative clause; Tabe and Atindogbé illustrate a similar kind of modification via an infinitive marker in Kenyang; and Lovegren and Voll discuss how Mundabli nouns can be modified by clauses introduced by a general subordinator in cases where the head noun plays no clear role within the modifying clause.

14 

 Jeff Good

Another question that arises when the chapters are considered together is how to account for variation in the grammatical sources of relative clause markers. Many of the chapters discuss this explicitly, such as Van de Velde’s consideration of the development of the Eton augment, which, as discussed, is the primary coding mechanism for relative clauses in the language. Similarly, Lovegren and Voll lay out a possible route through which one dialect of Mungbam acquired an agreeing relativizer, an important question given that each of the other four dialects makes use of an invariant relativizer. Moreover, while demonstratives showing agreement with the head noun of the relative clause appear to be a typical source of relativizers, we see cases here where the source demonstrative has more proximal semantics (e.g., Basaá and Nugunu) as well as cases where the source demonstrative has more distal semantics (e.g., Mundabli and Wawa). In other languages including most varieties of Mungbam as just discussed, one finds invariant relativizers, suggesting further grammaticalization pathways in need of study. Some languages (e.g., Bafut, Mundabli, Wawa) even allow agreeing and invariant relativizers to be combined in the marking of relative clauses. What factors govern these different pathways of grammaticalization? The distribution of resumptives in relative clauses also appears deserving of deeper consideration. On the one hand, even though the languages described here generally fall in line with the predications of the Accessibility Hierarchy, some positions on the hierarchy show interesting “transitional” behavior. For instance, in Basaá obliques can be relativized either via a gap or a resumptive construction, and they are the only class of arguments allowing this variation – higher arguments on the hierarchy (e.g., subjects and objects) require a gap, and lower arguments (e.g., possessor and objects of comparison) require resumptive pronouns. Watters describes something similar for Ejagham, though, in that case, variability is found in a higher position of the Accessibility Hierarchy, specifically for indirect objects (to use Keenan and Comrie’s (1977) term). In addition, there is also the question of the form that resumptives take. Vute, for example, has a special anaphoric pronoun which can be used for resumptive function, and the language also allows for a nominalized verb to be relativized in a construction where a finite copy of the verb appears within the relative clause in “resumptive” position (e.g., “the love that I love”). One also finds various non-pronominal elements used for resumptive function in certain semantic contexts, such as an invariant word in Eton that is translatable as ‘inside’, which Van de Velde exemplifies as being used in an instrumental context in place of a prepositional phrase. All told, the sense one gets from reading the papers in this volume is that the typology of the use and function of resumptive elements in the languages of Cameroon would be an interesting candidate for further targeted investigation.



1 Kaleidoscopic variations in Bantoid relative clauses 

 15

A final topic raised by these chapters is what types of verb forms are allowed in relative clauses. As made clear above, some languages allow more or less the same range of verb forms in relative clauses as they do in main clauses, while other languages place significant restrictions on verbs in relative clauses. An interesting case in this volume is Wawa, where affirmative relative clauses show strong restrictions on allowable verb forms while negative clauses do not show an asymmetry between relative clauses and main clauses. More detailed investigation of this area across more languages may very well yield interesting implicational patterns. Of course, other readers will be struck by different features of the languages described here. The most remarkable aspect of this volume is clearly not any one of the grammatical patterns described within but, rather, the sheer diversity one finds among a group of languages that are so closely related to each other and spoken in such a small geographic area—at least when we set them against the languages of Africa more generally. One can hope that this collection will inspire similar work in other parts of the continent.

References Blois (de), K.F. 1970. The augment in the Bantu languages. Africana Linguistica 4. 85–165. Comrie, Bernard & Tania Kuteva. 2013. Relativization on subjects. Leipzig: Max Planck Institute for Evolutionary Anthropology. http://wals.info/chapter/122 (accessed 13 November 2016). Creissels, Denis. 2009. The construct form of nouns in African languages. In Peter K. Austin, Oliver Bond, Monik Charette, David Nathan & Peter Sells (eds.), Proceedings of Conference on Language Documentation and Linguistic Theory 2, 73–82. London: School of Oriental and African Studies. Creissels, Denis, Gerrit J. Dimmendaal, Zygmunt Frajzyngier & Christa König. 2008. Africa as a morphosyntactic area. In Bernd Heine & Derek Nurse (eds.), A linguistic geography of Africa, 86–150. Cambridge: Cambridge University Press. Good, Jeff. 2010. Topic and focus fields in Naki. In Ines Fiedler & Anne Schwarz (eds.), The expression of information structure: A documentation of its diversity across Africa, 35–67. Amsterdam: John Benjamins. Good, Jeff. 2012. How to become a “Kwa” noun. Morphology 22. 293–335. Hammarström, Harald, Robert Forkel, Martin Haspelmath & Sebastian Bank. 2016. Glottolog 2.7. Jena: Max Planck Institute for the Science of Human History. http://glottolog.org (accessed 16 November 2016). Hyman, Larry M. 2004. How to become a “Kwa” verb. Journal of West African Languages 30. 69–88. Hyman, Larry M. & John R. Watters. 1984. Auxiliary focus. Studies in African Linguistics 15. 233–273. Katamba, Francis. 2003. Bantu nominal morphology. In Derek Nurse & Gérard Philippson (eds.), The Bantu languages, 103–120. London: Routledge. Keenan, Edward L. & Bernard Comrie. 1977. Noun phrase accessibility and universal grammar. Linguistic Inquiry 8. 63–99.

16 

 Jeff Good

Kouankem, Constantine. 2012. The syntax of the Medumba determiner phrase. Yaounde: University of Yaounde I dissertation. Lewis, M. Paul, Gary F. Simons & Charles D. Fennig (eds.). 2016. Ethnologue: Languages of the world, 19th edn. Dallas, Texas: SIL International. Online version: http://www.ethnologue. com (accessed 16 November 2016). Maho, Jouni. 1999. A comparative study of Bantu noun classes. Göteburg: Acta Universitatis Gothoburgensis. Maho, Jouni. 2003. A classification of the Bantu languages: An update of Guthrie’s referential system. In Derek Nurse & Gérard Philippson (eds.), The Bantu languages, 639–651. London: Routledge. Mchombo, Sam A. 2004. The syntax of Chichewa. Cambridge: Cambridge University Press. Nurse, Derek & Gérard Philippson. 2003. Introduction. In Derek Nurse & Gérard Philippson (eds.), The Bantu languages, 1–12. London: Routledge. Schachter, Paul. 1973. Focus and relativization. Language 49. 19–46. Watters, John R. 2010. Focus and the Ejagham verb system. In Ines Fiedler & Anne Schwarz (eds.), The expression of information structure: A documentation of its diversity across Africa, 349–376. Amsterdam: John Benjamins.

Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman

2 Accessibility and demonstrative operators in Basaá relative clauses Abstract: This paper shows that Basaá (A.43) possesses two relative clause strategies, one involving a gap and another a relative pronoun. The distribution of these two strategies follows the Accessibility Hierarchy (Keenan and Comrie 1977). Relative clauses in Basaá form a natural class with demonstrative modifiers, as they cannot co-occur in relative clauses involving a gap. This restriction does not obtain for relative clauses with the resumption strategy. We show that by adopting the head-raising analysis of relative clauses just for those relative clauses that include a gap; the complementarity of demonstratives and relative operators in Basaá follows. Thus, the Basaá facts provide a novel empirical argument for this analysis of relative clauses. We offer a diachronic perspective on these facts, suggesting the Basaá relative operator is at an intermediate stage of grammaticalization.

1 Introduction Basaá [ɓasaá], (A.43), is a Narrow Bantu language spoken by approximately 300,000 people in southern Cameroon (Lewis, Simons, and Fennig 2016). Narrow Bantu languages are in the Bantoid branch of Benue-Congo, part of the expansive Niger-Congo language family. This paper focuses on the Mbene dialect group of Basaá, primarily spoken in the Centre region, specifically as spoken in the Nyonget-Kellé division. This paper reviews both typical and less-typical properties of relative clauses in Basaá. Two basic relativization strategies are outlined, one involving a gap and another involving a resumptive pronoun. The distribution of these two strategies

Peter Jenks: UC Berkeley, [email protected] Emmanuel-Moselly Makasso: Zentrum für Allgemeine Sprachwissenschaft (ZAS), Berlin, [email protected] Larry M. Hyman: UC Berkeley, [email protected] DOI 10.1515/9783110469547-002

18 

 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman

is shown to follow from the predictions of the Accessibility Hierarchy (Keenan and Comrie 1977). It is further demonstrated that relative clauses form a natural class with demonstratives: both license a definite/specific prefix /í-/ on the head noun, and relative operators are in complementary distribution with demonstratives in relative clauses involving a gap. Additionally we show that this restriction does not obtain for relative clauses with the resumption strategy. We account for these facts by adopting the relatively standard assumption that relative clauses involving a gap involve movement while those with resumptive pronouns do not (McCloskey 2006). Specifically, we show that by adopting Kayne’s (1994) head-raising analysis of relative clauses with a gap, the complementarity of demonstratives and relative operators in Basaá can be derived. We offer several supporting arguments for the head-raising analysis of gapped relatives. Thus, the Basaá facts provide a novel empirical argument for the headraising analysis of relative clauses. Additionally, we discuss these facts from the perspective of historical change, suggesting that the relative operator in Basaá is only partially grammaticalized from a demonstrative. The structure of this paper is as follows. Section 2 introduces relative clauses in Basaá, including their morphosyntactic marking, the range of arguments that can be relativized, and the position of relative clauses within the noun phrase. Section 3 discusses the relationship between definiteness, demonstratives, and relative operators. Section 4 presents a formal syntactic analysis of both relative clauses that involve a gap and relative clauses involving resumption. Section 5 discusses the grammaticalization of the definite/specific prefix and the relative operator in Basaá.

2 Restrictive relative clauses in Basaá In contrast to many other languages in Cameroon, Basaá has been the subject of extensive previous work (e.g. Bot Ba Njock 1970; Lemb and Degastines 1973; Dimmendaal 1988; Bitjaa Kody 1990; Hyman 2003). The grammatical properties of Basaá are typical for Northwest Bantu languages: it is head-initial, exhibiting SVO word order in declarative sentences, head-modifier order, and prepositions. Basaá exhibits the rich noun class system typical of Bantu, as well as subject agreement on verbs and concord on nominal modifiers. Phonologically, Basaá exhibits a binary distinction in H vs. L tone; these tones play an important role in marking morphosyntactic distinctions as well as encoding lexical contrasts between words.



2 Accessibility and demonstrative operators in Basaá relative clauses 

 19

Relative clauses in Basaá have been described by Makasso (2010), which this paper builds on, and more briefly by Hyman (2003) and Dimmendaal (1988).1 Section 2.1 reviews the external properties of relative clauses, including their position relative to other modifiers, the morphology of the relative operator, and the ability of relative clauses to license a nominal prefix that marks definiteness. Section 2.2 demonstrates that Basaá exhibits two basic strategies for relativization, a gap strategy for relativization from argument positions, and a resumption strategy for relativizing possessives.

2.1 Basic characteristics of relative clauses Relative clauses in Basaá follow the head noun. In most cases, the internal syntax of relative clauses is identical to the syntax of declarative sentences, modulo a gap in an argument position created by relativization. The location of this gap will be marked overtly throughout this paper and subscripted with an index matching the head noun for clarity. The main morphological marker of a relative clause is an optional relative pronoun, identical to the ‘near speaker’ demonstrative (see Section 3) that precedes the relative clause and agrees with the head noun in noun class. This is illustrated for a subject (1b) and object (2b) relative clause: (1) Subject relative clause a. mut a bí ↓jɛ́ bíjɛ́ k 1.person 1.sbj pst2 eat 8.food ‘The person ate the food.’ b. í-mut1 (nú) [ ___1 a bí ↓jɛ́ bíjɛ́ k ] aug-1.person 1.rel 1.sbj pst2 eat 8.food ‘The person that ate the food’ (2) Object relative clause a. liwándá lí bí ↓tɛ́ hɛ́ hínuní 5.friend 5.sbj pst2 see 19.bird ‘The friend saw the bird.’

1 Another relevant reference on Basaá relative clauses is Bassong (2010). However, Bassong focuses on the Bakoko dialect group (A43b) spoken in Littoral province, while we focus on the Mbene dialect group (A43a), as stated in the introduction. The judgments and data reported in this paper have been checked with several native speakers of the Mbene dialect.

20 

 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman b. hínuní1 (hí) [ liwándá lí bí ↓tɛ́ hɛ̂ ___1 ] aug.19.bird 19.rel 5.friend 5.sbj pst2 see ‘The bird that the friend saw’

Relativization is also marked by the insertion of a /í-/ prefix on the head noun, realized either as í-, as in (1b), or simply as a H tone on the class prefix of the noun, as in (2b).2 We concur with Dimmendaal (1988: 58) that the high tone on the prefix in (2b) is a reduced form of the í-prefix in (1b). We will return to this prefix in more detail in Section 3.1.3 Relative clauses occur at the right edge of the noun phrase in Basaá. This means that relative clauses must follow adjectives (3), numerals (4), possessive pronouns (5), and demonstratives (see Section 3), which also follow the noun: (3) Adjective > relative clause a. hínuní hi-kéŋí (hí) [ liwándá lí bí ↓tɛ́ hɛ̂ ___1 ] aug.19.bird 19-big 19.rel 5.friend 5.sbj pst2 see ‘The big bird that the friend saw’ b. *hínuní (hí) [ liwándá lí bí ↓tɛ́ hɛ̂ ___1 ] hi-kéŋí (4) Numeral > relative clause a. dínuní dítán (dí) [ liwándá lí bí ↓tɛ́ hɛ̂ ___1 ] aug.13.bird 13.five 13.rel 5.friend 5.sbj pst2 see ‘The five birds that the friend saw’ b. *dínuní (dí) [ liwándá lí bí ↓tɛ́ hɛ̂ 1 ] dítân (5) Possessive pronoun > relative clause a. hínuní hyɛ̂ m (hí) [ liwándá lí bí ↓tɛ́ hɛ̂ ___1 ] aug.19.bird 19.my 19.rel 5.friend 5.sbj pst2 see ‘My bird that the friend saw’ b. *hínuní (hí) [ liwándá lí bí ↓tɛ́ hɛ̂ ___1 ] hyɛ̂ m The right-edge position of relative clauses puts them in a class with demonstratives, which must also occur following other modifiers (Hyman 2003: 270). Unlike other

2 The noun class prefix also has H tone in object position in (2a), but this is due to metatony, whereby in most tenses a H tone occurs on the verb and what follows (Makasso 2012). The citation form for ‘bird’ is hi-nuní, with L on the class prefix. 3 Bassong (2010), chapter 4, reports a distinct type of relative clause headed by the complementizer lɛ́ rather than the demonstrative operator discussed in this paper. For Mbene Basaá speakers, these relative clauses sound somewhat unnatural, and feel like calques from French relative clauses introduced by que, though this may be a dialectal difference (see fn. 1).



2 Accessibility and demonstrative operators in Basaá relative clauses 

 21

nominal modifiers, demonstratives and possessive pronouns can also precede the noun when focused, as Section 3.1 illustrates for demonstratives. We will see below that the connection between demonstratives and relative clauses is deeper than their shared syntactic distribution.

2.2 Accessibility to relativization and the resumptive strategy While clausal positions in Basaá can be relativized, there are two strategies for forming relative clauses depending on the site of relativization. The first strategy, seen earlier in examples (1) and (2), involves a dependency between the head noun and a gap located inside of the relative clause. This strategy extends to relative clauses formed on goals of ditransitive verbs (6) as well as oblique noun phrases, such as the object of a prepositional phrase (7): (6) Relativization from goals of ditransitive verbs a. liwándá lí bí ↓tí njɔ̂ k litám 5.friend 5.sbj pst2 give 9.elephant 5.fruit ‘The friend gave the elephant fruit.’ b. í-njɔk1 (i) [ liwándá lí bí ↓tí ___1 litám ] ‘The elephant that the friend gave fruit (to)’ (7) Relativization from object of preposition a. mut a m ɓíí káar í-ŋɡií têble 1.person 1.sbj pst1 put 9.book loc-top 9.table ‘The person put the book on the table.’ b. í-têble1 (i) [ mut a m ɓíí káar í-ŋɡií ___1 ] ‘The table that the person put the book on top of’ The second strategy for forming relative clauses involves the use of a resumptive pronoun. This strategy is used with objects of prepositions, which are realized as possessive pronouns (8) (Makasso 2010: 152). Resumptive pronouns are obligatory when relative clauses are formed on possessive noun phrases (9): (8) Resumptive pronoun with relativization from object of preposition a. í-têble1 (i) [ mut a m ɓíí káar aug-9.table 9.rel 1.person 1.agr pst1 put 9.book í-ŋɡii yé↓é1 ] loc-top 9.poss ‘The table that the person put the book on top of it’

22 

 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman

(9) Resumptive pronoun with relativization from possessive a. ŋgwɔ́ i maaŋgɛ́ i bí kɔgɔ́ l mɛ̂ 9.dog 9.con 1.child 9.sbj pst2 bite 1sg ‘The child’s dog bit me.’ b. í-maaŋgɛ́ 1 (nú) [ ŋgwɔ́ yé↓é1 i bí kɔgɔ́ l mɛ ] aug-1.child 1.rel 9.dog 9.poss 9.sbj pst2 bite 1sg (lit. ‘the child that his dog bit me’) ‘The child whose dog bit me’ In (9b), the head noun, ‘child’, is coindexed with the class 1 possessive pronoun yeé (the H tone and downstep on the pronoun arise due to high tone spread from ŋgwɔ́ ). If the possessive target of relativization is located in object position, another possibility for relativization becomes available: the noun phrase containing the resumptive possessive pronoun is displaced to the front of the relative clause, and a pronoun is left in its place (10b). Fronting is optional, however. The noun phrase containing the resumptive pronoun can also occur in object position (10c): (10) Relativization from possessor in object position ↓jɛ́ a. mɛ bí bíjɛ́ k bí mááŋgɛ́ 1sg pst2 eat 8.food 8.con 1.child ‘I ate the child’s food’ ↓jɛ́ b. í-maaŋgɛ́ 1 (nú) [ bijɛ́ k *(gwéé1) mɛ bí *(gwɔ́ )1 ] aug-1.child 1.rel 8.food 8.poss 1sg pst2 eat 8.PRN ‘The child whose food I ate’ (lit. ‘the child that his food I ate it’) ↓jɛ́ c. í-maaŋgɛ́ 1 (nú) [ mɛ bí bijɛ́ k *(gwéé1) ] aug-1.child 1.rel 1sg pst2 eat 8.food 8.poss ‘The child whose food I ate’ (lit. ‘the child that I ate his food’) Resumptive pronouns are obligatory in these examples. Thus, relativizing a possessive noun phrase necessarily relies on the resumption strategy. Resumption is also necessary with standards or objects of comparison: (11) Relativization from object of comparative a. ŋgwɔ́ i ye i-kɛ́ ŋí ilɛ́ l maaŋgɛ́ 9.dog 9.sbj prs.be 9-big exceed 1.child ‘The dog is bigger than the child’ b. í-maaŋgɛ́ 1 (nú) [ ŋgwɔ́ i ye i-kɛ́ ŋí ilɛ́ l *(↓nyɛ́ 1) ] aug-1.child 1.rel 9.dog 9.sbj prs.be 9-big exceed 1.him ‘The child that the dog is bigger than’ (lit. ‘the child that the dog is big exceeding him’) We can identify a common structural thread that connects comparative objects to the objects of prepositions and genitive, all of which require resumption. In all



2 Accessibility and demonstrative operators in Basaá relative clauses 

 23

three cases, the relativized argument is not a core argument of the main predicate. Thus, we can generalize over the distribution of relativization strategies in Basaá that the gap strategy is only allowed with arguments of the main predicate. In addition, this distribution follows from the predictions of the Accessibility Hierarchy (Keenan and Comrie 1977: 66), where ‘>’ means ‘is more accessible than’: (12) Accessibility Hierarchy (AH) SU > DO > IO > OBL > GEN > OCOMP As predicted by Keenan and Comrie, the two relativization strategies in Basaá correspond to contiguous segments of the AH, and that the primary strategy in Basaá – marking the relativized site with a gap – forms a contiguous stretch of the AH including subjects. The secondary strategy, marking the relativized site with a resumptive pronoun, includes oblique arguments, optionally, and genitive noun phrases: (13) Accessibility in Basaá a. Strategy 1: Gap b. Strategy 2: Resumptive pronoun

SU > DO > IO > OBL OBL > GEN > OCOMP

The generalization seems to be that while arguments of the main verb can be directly relativized with the gap strategy in Basaá, arguments of these arguments must be relativized with the resumptive strategy. The oblique arguments/prepositional objects are transitional because in many cases the preposition is selected by the verb, though the noun phrase is structurally dependent on the preposition. It would not be surprising to find variation even between different prepositions. In summary, we have seen that relative clauses occur at the right edge of the noun phrase. The primary morphological reflex of relativization is an optional relative operator which is homophonous with a demonstrative modifier. The following section turns to the syntactic status of this operator as well as the status of the augment prefix on the noun.

3 Relative operators, demonstratives, and specificity This section examines the distribution of demonstratives and the augment prefix with relative clauses. After reviewing the distribution of demonstratives in section 3.1, section 3.2 shows that the augment prefix, obligatory with postnominal demonstratives, only occurs with relative clauses in definite noun phrases. Section 3.3 presents a novel finding: demonstratives in Basaá are in complementary distribution with relative operators in the gapped relative clause strategy.

24 

 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman

3.1 Properties of demonstratives Basaá makes a three-way distinction in its demonstrative system. These distinctions identify people or objects near the speaker, near the hearer, and beyond the speaker and hearer. Demonstratives show concord with the noun in noun class, i.e. gender and number. (14) Demonstratives in Basaá (from Hyman 2003: 267) Class



1 2 3 4 5 6 7 8 9 10 19 13

‘this’ (n.s.)

‘that’ (n.h.)

‘that’ (far)

núnú ɓáná únú míní líní máná íní bíní iní íní híní tíní

nû ɓá û mî lî mâ î bî i î hî dî

núú ɓáa úú míí líí máá íí bíí ií íí híí díí

These demonstratives all begin with a H tone except for class 9, whose tonal agreement is L, as in Proto-Bantu. Class 1 is mixed in this regard. While connectives and demonstratives do begin with a H tone, other instances of class 1 agreement have a L tone (Hyman 2003: 266). Demonstratives in Basaá can occur either before or after the noun they modify, as in many Bantu languages (Van de Velde 2005). The prenominal position of demonstratives is associated with contrastive or identificational focus on the demonstrative (cf. Makasso 2010: 149). The postnominal position, while information structurally unmarked, requires that the augment prefix be present on the noun:4 (15) Prenominal and postnominal demonstratives a. líní liwándá b. lí ↓wándá líní 5.this 5.friend aug.5.friend 5.this ‘THIS friend’ (near speaker) ‘this friend’ (near speaker)

4 As discussed earlier, this prefix is realized as H on the noun class marker, or as í- with nouns that lack an overt CV prefix, as in examples (1) and (6–10).



2 Accessibility and demonstrative operators in Basaá relative clauses 

 25

c. lí liwándá d. lí ↓wándá lî 5.that 5.friend aug.5.friend 5.that ‘THAT friend’ (near hearer) ‘that friend’ (near hearer) e. líí liwándá f. lí ↓wándá líí 5.that 5.friend aug.5.friend 5.that ‘THAT friend’ (far) ‘that friend’ (far) Hyman (2003) identifies the í-prefix as a trace of the Proto-Bantu augment due to its identical position before the noun, its H tone, and its connection to definiteness. A homophonous prefix also functions as locative marker in expres­sions such as í-↓ndáp ‘at home’ (Makasso 2010: 148). While the formal connection between the locative and the nominal prefix cannot be ignored, we will refer to í- as the ‘augment prefix’ in the following discussion below. The augment prefix cannot occur on bare (unmodified) nouns. Thus, there is no means of overtly marking definiteness in Basaá in unmodified noun phrases: (16) Definite bare nouns ↓tɛ́ hɛ́ mɛ bí a. yaaní mût ni mudaá 1.person with 1.woman yesterday 1sg pst2 see ‘Yesterday I saw a man and a woman.’ b. (*í-)mut a ɓée ntendéé, (*í-)mudaá kírîk 1.person 1.sbj pst2.be tall 1.woman short ‘The man was tall, but the woman (was) short.’ With the exception of relative clauses and demonstratives, postnominal modifiers do not occur with the augment prefix, including numerals (ma-wándá mátân ‘five friends vs. *má-↓wándá mátân) possessive pronouns (li-wándá jɛ̂m ‘my friendʼ vs. *lí-↓wándá jɛ̂m) and adjectives (li-wándá likéŋí ‘big/important friend’ vs. *lí-↓wándá li-kéŋí). This distinction between the demonstrative and the other modifiers also obtains in their phonological behavior: while H spreads from nouns to numerals and possessive pronouns, it does not spread to demonstratives. To see this, compare ŋgwɔ́ yɛ̂m ‘my dog’ vs. njok yɛm ‘my elephant,ʼ where H spreads to the possessive pronoun from the H toned ŋgwɔ́ , with ŋgwɔ́ iní ‘this dog’ vs. njok iní ‘this elephantʼ, where H does not spread from ŋgwɔ́ onto the following demonstrative.

3.2 The augment prefix and definiteness The previous section demonstrated that the augment prefix appears in Basaá in noun phrases with postnominal demonstratives. In Section 2, we saw that the augment prefix also occurs on nouns modified by relative clauses. However,

26 

 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman

r­ elative clauses and demonstratives differ in one respect: while the augment is obligatory with demonstratives, it can be omitted with relative clauses, resulting in an indefinite interpretation (cf. Makasso 2010: 153–154): ↓gwɛ́ s mût1 (nú) [ ___1 a yé (17) a. mɛ ŋ́ 1sg prs like 1.person 1.rel 1.sbj cop ‘I like a person that is big/important.’ b. mɛ ŋ́ ↓gwɛ́ s í-mut1 (nú) [ ___1 a yé mbóm ] ‘I like the person that is big/important.’

mbóm ] 9.big

↓yéŋ ↓yí (18) a. mɛ ń mááŋɡɛ́ 1 (nú) [ mɛ ń ___1 ] 1sg prs seek 1.child 1.rel 1sg prs know ‘I’m looking for a child that I know.’ b. mɛ ń ↓yéŋ í-maaŋɡɛ́ 1 (nú) [ mɛ ń ↓yí ___1 ] ‘I’m looking for the child that I know.’

The augment in Basaá thus marks definiteness in the presence of relative clauses. We can see this by examining contexts that would allow only one of the sentences in (18). Example (18a) would be felicitous in the following context: there are twenty kids around and five of them that I know. What I am looking for is one of the five. In contrast, example (18b) would be used in a context where there are twenty kids around and I know just one. In other words, the augment in (18b) seems to carry a uniqueness presupposition, a canonical characteristic of definite noun phases. We conclude that the augment is an overt marker of definiteness in noun phrases containing relative clauses.5 In a sense, this makes its exclusive occurrence with demonstratives outside of relative clauses less surprising, as demonstratives are the only nominal modifiers in Basaá that are inherently definite.

3.3 Relative operators and demonstratives We have seen that the distribution of the relative operator (e.g. nú) is independent from the definiteness of the noun phrase, marked by the augment. This is expected if the sole syntactic function of the relative operator is to mark a relative clause.

5 Van de Velde (this volume) notes that the augment in Eton (A. 71) does not express a semantic contrast, and expresses skepticism that it could express any such contrast in neighboring languages. Yet Barlew and Clem (2014) show that Bulu (A. 74), a close relative of Eton, uses augments to mark a semantic contrast very reminiscent of definiteness in the context of relative clauses just like Basaá. Specifically, Barlew and Clem show that the Bulu augment marks an existence commitment or presupposition. In either case, variation among A-group Bantu languages in the distribution and interpretation of an augment strike us as an important topic for future research.



2 Accessibility and demonstrative operators in Basaá relative clauses 

 27

However, the examples below demonstrate that relative operators are in complementary distribution with demonstratives, whether they are postnominal or prenominal. This is shown for subject relative clauses in example (19) and object relative clauses in example (20), both of which use the gap strategy: (19) Complementary distribution of rel. operator and demonstrative (subject r. c.) ↓jɛ́ a. lí↓wándá1 líní/lí↓/líí (*lí↓) [ ___1 lí bí bíjɛ́ k ] aug.5.friend 5-dem 5.rel 5.sbj pst2 eat food ‘This/that friend that ate the food’ ↓jɛ́ b. líní/lí/líí liwandá1 (*lí↓) [ ___1 lí bí bíjɛ́ k ] 5.dem 5.friend 5.rel 5.sbj pst2 eat food ‘THIS/THAT friend that ate the food’ (20) Complementary distribution of rel. operator and demonstrative (object r. c.) ↓yí a. í-maaŋɡɛ́ 1 núnú/nú↓/núú (*nú) [ mɛ ń ___1 ] aug-1.child 1.dem 1.rel 1sg prs know ‘This/that child that I know.’ ↓yí b. núnú/nú↓/núú maaŋɡɛ́ 1 (*nú) [ mɛ ń ___1 ] 1.dem 1.child 1.rel 1sg prs know ‘THIS/THAT child that I know.’ The complementary distribution of demonstratives and relative operators is unexpected. This is because the demonstratives and the relative operator have been shown to have distinct syntactic behaviors – only demonstratives require the augment prefix – and semantic effects – augmentless noun phrases with relative clauses are indefinite. The complementarity between demonstratives and the relative operator is more limited with relative clauses involving the resumption strategy. While the ‘nearhearer’ demonstrative cannot occur with the relative operator, the ‘near-speaker’ and distal demonstrative can occur with the relative operator. This is illustrated for resumptive relatives formed on standards of comparison (21) and possessive noun phrases (22): (21)  Distribution of relative operator and demonstrative (object of comp r.c.) a. í-maaŋgɛ́ 1 núnú/*nú↓/núú (nú) [  ↓ŋgwɔ́ i ye aug-1.child 1.dem 1.rel 9.dog 9.sbj prs.be ↓nyɛ́ ] ilɛ́ l 1 exceed 1.PRN ‘This/that child that the dog is bigger than’ b. núnú/*nú↓/núú maaŋgɛ́ 1 (nú) [ ↓ŋgwɔ́ i ye ikɛ́ ŋí ilɛ́ l ‘THIS/THAT child that the dog is bigger than’

ikɛ́ ŋí 9-big

↓nyɛ́

1 ]

28 

 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman

(22)  Distribution of relative operator and demonstrative (object ­possessor r.c.) ↓jɛ́ bijɛ́ k a. í-maaŋgɛ́ 1 núnú/*nú↓/núú nú [ mɛ bí gwéé1 ] aug-1.child 1.dem 1.rel 1sg pst2 eat 8.food 8.poss ‘This/that child whose food I ate’ ↓jɛ́ b. núnú/*nú↓/núú maaŋgɛ́ 1 nú [  mɛ bí bijɛ́ k gwéé1 ] 1.dem 1.child 1.rel 1sg pst2 eat 8.food 8.poss ‘THIS/THAT child whose food I ate’ With fronted genitives (cf. 10b), the pattern is the same: (23) Distribution of relative operator and demonstrative (fronted object possessor r.c.) a. í-maaŋgɛ́ 1 núnú/*nú↓/núú nú [ bijɛ́ k gwéé1 mɛ aug-1.child 1.dem 1.rel 8.food 1.poss 1sg ↓jɛ́ bí gwɔ́ 1 ] pst2 eat 8.PRN ‘This/that child whose food I ate’ b. núnú/*nú↓/núú maaŋgɛ́ 1 nú [ bijɛ́ k gwéé1 mɛ bí 1.dem 1.child 1.rel 8.food 1.poss 1sg pst2 ↓jɛ́ gwɔ́ 1 ] eat 8.PRN ‘THIS/THAT child whose food I ate’ To summarize, relative clauses formed by resumption pattern together, allowing demonstratives to co-occur with the relative operator, with the exception of the homophonous near-hearer demonstrative. In contrast, in relative clauses formed with the gap strategy, demonstratives cannot co-occur with the relative operator at all. We have seen that relative clauses in Basaá share with demonstratives the ability to license an overt marker of definiteness, the augment prefix. This prefix cannot mark definiteness on bare (unmodified) nouns, however. Furthermore, while the relative operator is segmentally identical to the near-hearer demonstrative, it can occur in indefinite, augment-less noun phrases, leading to the expectation that the relative operator is syntactically distinct from demonstratives. Yet the relative operator and demonstratives are in complementary distribution, though a subset of the demonstratives can occur with the relative operator in relative clauses formed by resumption.



2 Accessibility and demonstrative operators in Basaá relative clauses 

 29

4 An analysis of Basaá relative clauses In this section we present one possible theoretical analysis of relative clauses in Basaá which accounts for the distribution of demonstratives and relative operators. For those who have little interest in formal syntactic theory, this section may not be particularly compelling, and the data may not add much to the overall picture of Basaá relative clauses. In brief, we suggest a head-raising analysis of gapped relative clauses along the lines of Kayne (1994), from which the complementarity between the relative operator and demonstrative in (19–20) follows, given additional assumptions about the syntactic category of relative operators. This analysis cannot be extended to the resumption strategy, for which an analysis involving base generation of relative operator is proposed. This analysis accounts for the non-complementarity of certain demonstratives with the relative operators in these cases (21–23), and offers suggestions about the syntactic position of the augment prefix and the fact that it only occurs with certain modifiers in Basaá.

4.1 The syntactic status of the relative operator This section provides evidence from stacked relatives and non-restrictive relatives that suggest that the relative operator is indeed a relative operator, and hence is part of the relative clause. Once we adopt this conclusion, we can exclude analyses of the complementarity between demonstratives and relative operators above which posit that relative operators are not relative operators at all but simple demonstrative determiners, rendering the observation moot. The first argument that relative operators form a constituent with relative clauses comes from the recurrence of relative operators in stacked restrictive relative clauses in Basaá. (24) and (25) provide definite and indefinite stacked relatives, respectively: (24) Definite stacked relatives ↓gwés liwándá jɛ̂ m lí ŋ́ ńdíkí híŋgɔnda1 (hí) 5.friend 5.my 5.sbj prs like only aug.19.girl 19.rel [ ___1 hí yé diláám ] 19.sbj prs.be 13.beautiful ↓yí *(hí) [ nyaŋ a ń ___1 ] 19.rel mother 1.sbj prs know ‘My friend only likes the girl that is beautiful that his mother knows.’

30 

 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman

(25) indefinite stacked relatives ↓gwés liwándá jɛ̂ m lí ŋ́ ńdíkí diŋgɔnda1 (tí) 5.friend 5.my 5.sbj prs like only 13.girls 13.rel [ ___1 dí yé diláám ] 13.sbj prs.be 13.beautiful ↓yí *(tí) [ nyaŋ a ń ___1 ] 13.rel mother 1.sbj prs know ‘My friend only likes girls that are beautiful that his mother knows.’ If a demonstrative occurs with the stacked relatives in (26), only the first relative operator is prohibited: (26) Stacked relatives with prenominal demonstrative ↓gwés liwándá jɛ̂ m lí ŋ́ tíní diŋgɔnda1 (*tí) 5.friend 5.my 5.sbj prs like these 13.girls 13.rel [ ___1 dí yé diláám ] 13.sbj prs.be 13.beautiful ↓yí *(tí) [ nyaŋ a ń ___1 ] 13.rel mother 1.sbj prs know ‘My friend only likes THESE girls that are beautiful that his mother knows.’ These examples show that the demonstrative that occurs before relative clauses is a true relative operator by virtue of the fact that it must recur in stacked relatives. Example (26) demonstrates that the complementarity between the relative operator and demonstratives only holds of the relative operator closest to the head noun.6 The second argument that the relative operator is syntactically associated with the relative clause comes from non-restrictive relative clauses (Makasso 2010: 152): (27) Non-restrictive relative clauses ↓gwés a. Paul1, nú *(↓nyɛ́1) a ŋ́ P. 19.rel 1.PRN 1.sbj prs like ɓok í-têble first loc-table ‘Paul, who likes to eat, was the first to the table.’

jɛ́, eat

a 1.sbj

bí pst2.be

6 A reviewer has suggested that the requirement that the second demonstrative operator be obligatory might be due to pragmatic or processing considerations, as more complex embedded structures often require more explicit marking. While it such a processing explanation is indeed possible, the very fact that Basaá noun phrases permit, much less allow, two demonstratives just in case relative clauses are stacked would be surprising if these demonstratives were associated with the head noun rather than the relative clause.



2 Accessibility and demonstrative operators in Basaá relative clauses 

 31

↓sál ɓɛ̂ m, ɓá *(↓ɓɔ́ 1) ɓá bí lɔ́ ŋgɛ̂, b. ɓaúrú1 2.student 2.my 2.rel 2.PRN 2.sbj pst2 work well ɓá bí nɛt ɓɔ́ ɓasô agr pst2 succeed all ‘My students, who have been working well, have all succeeded.’

Non-restrictive relative, like restrictive relatives are introduced by the relative operator, but they necessarily use the resumption strategy, as indicated by the free subject pronouns in the examples above, usually used for non-subjects (Hyman 2003: 269). As the commas indicate, the relative operator forms a prosodic constituent with the non-restrictive relative clause, suggesting that the relative operator is not an adnominal demonstrative. Additionally, we see in (27a) that the relative operator occurs with the proper name Paul. As proper names are otherwise prohibited with demonstratives, the demonstrative in example (27a) is licensed by the relative clause.

4.2 Demonstrative-operator complementarity and head-raising relatives The Basaá facts detailed above present two problems: 1) the complementarity of relative operators and demonstratives in ‘strategy 1’ relative clauses containing a gap and 2) distinguishing these cases from the non-complementarity of certain demonstratives and relative operators in ‘strategy 2’ relative clauses containing a resumptive pronoun. The intuition we pursue in this section is that the complementarity between demonstratives and relative operators in Basaá is due to the fact that even true demonstratives can function as relative operators, and relative operators and demonstratives occupy in the same syntactic position in the noun phrase. However, competition does not extend to relative clauses involving resumption, which must receive a different analysis, the topic of section 4.3. Complementarity arises from competition: two syntactic elements compete for the same syntactic position. To see how this solves our puzzle, we begin with prenominal demonstratives in Basaá, which we take to occur in a position in the specifier of DP: (28) [DP núnú/nú/núú [D’ Ø [NP mut ]]]

‘this/that person’

As relative operators compete with demonstratives in all positions, including prenominally, we take prenominal demonstrative operators to occur in the specifier

32 

 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman

of DP as well. Evidence for this claim comes from constituent questions, where wh-operators must occur prenominally: (29) njɛ́ máaŋɡɛ́ a bí ɓok 1.which 1.child 1.sbj pst2.be first ‘Which child was the first to the table?’

í-têble loc-table

(*í-máaŋɡɛ́ njɛ́ )

Given that relative clauses and constituent questions are subtypes of long-distance dependencies with closely related semantics, we expect wh-operators to occupy the same position as prenominal relative operators and focused demonstratives. In addition, demonstrative relative operators have been analyzed as a D-element associated with the head noun in several other Bantu languages (Ngonyani 2001, Zeller 2004, Carstens 2005, Cheng 2006), which is particularly attractive, given the common morphological and diachronic relationship between demonstratives and relative operators (see section 5).7 Of course, relative operators, wh-operators, and prenominal demonstratives cannot co-occur with a postnominal demonstrative. This observation follows if we derive the prenominal position of demonstratives and operators from the postnominal position, which is a lower position within NP or some intermediate functional projection. Following Leu (2015) and many others, 8 we take morphologically complex prenominal determiners, such as demonstratives, to be specifiers which originate from a lower, adjectival position. In Basaá, we can say that this movement is always associated with focus semantically: (30) a. postnominal demonstrative b. Prenominal demonstrative [DP í [NP [NP mut ] nú ]] → [DP nú [DP í [NP [NP mut ] nú ]] The fact that the augment prefix occurs overtly in (30a) but not in (30b) follows from a generalized version of the “Doubly-filled Comp Filter” (e.g. Koopman and

7 The alternative would be to analyze the relative operator as a relative complementizer (cf. Demuth and Harford 1999; Schneider-Zioga 2007; Henderson 2009). However, such an analysis offers no way of accounting for the complementarity of the demonstrative operator and demonstratives, and in fact predicts that they should never be in complementary distribution as they occupy distinct structural positions. Also, as discussed in fn. 3, Bassong (2010) reports that an overt complementizer lɛ́ can occur in relative clauses in other dialects of Basaá. The fact that a complementizer can occur which is distinct from the demonstrative suggests they should not receive the same analysis. 8 Demonstratives show similar alternations in their position in Romance languages, particularly in Romanian and Spanish, and they are similarly seen as alternating between a lower postnominal position to a higher position before the noun. In Spanish, the prenominal position also blocks the definite article. See, e.g., Bernstein (1997); Brugè (2002); Giusti (1997, 2002).



2 Accessibility and demonstrative operators in Basaá relative clauses 

 33

Szabolsci 2000: 4), which states that a head cannot be pronounced if its specifier is filled. That is, D cannot be pronounced in (30b) because the demonstrative or relative operator is occupying its specifier position. Leu (2015) relies on this principle to account for the inability of prenominal demonstratives and definite articles to co-occur in Germanic. With this proposal as background, we turn to the derivation of relative clauses themselves. Kayne (1994: 87) proposes that the nominal heads of relative clauses move to the specifier of the relative CP, which can be selected as the complement of D. This head-raising analysis has been argued to hold in other Bantu languages (Ngonyani 2001; Carstens 2005; Cheng 2006; Henderson 2009), and to a different dialect of Basaá by Bassong (2010) albeit based on different considerations. The head-raising analysis is one of only three available analyses of relative clauses in the literature, the others being the operator-movement analysis and the matching analysis (Bhatt 2002). Because both of these other two analysis assume two distinct noun phrases – one inside the relative clause and another heading the main DP – they do not lend themselves to an explanation of the complementarity of demonstratives and relative operators observed in Basaá, precisely because there are multiple NPs in the structure and hence multiple potential sources of demonstratives. In contrast, the head-raising analysis provides a natural account of this complementarity as well as explaining why only gapped relative clauses manifest this restriction. We now turn to an explicit model of head-raising relative clauses in Basaá, showing how they account for the complementarity of demonstratives and relative operators. First, in this analysis, because the augment is the head of DP we propose that it takes the relative CP as its complement. The head noun and relative operator occur in the specifier of this relative CP: (31) [DP1 í [CP [DP2 nú mut ]i [C’ … ti … ]]] Here, DP1 represents the entire DP, headed by the augment. DP2 is the moved relative head, whose base position is t1. The C head is silent here due to the filled specifier,9 again by the Doubly-filled Comp Filter According to the proponents of the head-raising analysis, the structure in (31) never surfaces, however, because one of two further operations must take place for the derivation to converge. Either 1) the N must move past Dem to the specifier of DP2 (cf. Kayne 1994: 90), or 2) DP2 moves to Spec, DP1. Both steps occur because D1 requires a noun in its “minimal domain” (Bianchi 2000: 128).

9 Again, see fn. 3 and fn. 6 on overt complementizers.

34 

 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman

The movement of the relative operator to the spec of the larger DP is motivated as an instance of focus-associated movement, parallel to the noun phrase-internal operation in (31). NP movement to [Spec, DP] is less clearly motivated; several other options exist in the literature, including of the N moving to a higher CP shell (Zwart 2000), a higher n head (de Vries 2002) and N moving and reprojecting above the CP (Bhatt 2002). However, the noun in the Aug-N-Dem order must be the NP originating within the relative clause, or our account of the complementarity between demonstratives and relatives would be lost. In the first case, the Aug-N-Op/Dem-RC word order results, as is illustrated below: (32) a. í-mut1 nú [ ___1 a aug-1.person 1.rel 1.sbj ‘The person that ate the food’ DP1 b.

D1

bí ↓jɛ́ bíjɛ́ k ] pst2 eat 8.food

CP

í DP2 NP mut

DP2 Op nú



C

D2

Ø

TP DP2 a bí ↓jέ bíjέk

NP

Ø

The movement of the demonstrative to the operator position inside the specifier of DP2 is not represented for simplicity. This derivation above is the only option when the relative operator is a true operator rather than a demonstrative. This is because in order to be focused in the specifier of DP1, the operator must have additional deictic lexical content that can be focused there. Furthermore, we can now see that the configuration in (32b) is the one in which the relative operator is optional, a fact that that we likewise relate to the fact that the operator itself is not in focus in this position. Additionally, indefinite relative clauses that lack an augment can be analyzed as having the structure in (32b), but with an unpronounced indefinite article in D1 instead of the augment.

2 Accessibility and demonstrative operators in Basaá relative clauses 



 35

In the second derivation, the entire relative DP head moves out of CP, resulting in the Dem-N-RC word order with emphasis on the demonstrative: [ ___1 a (33) a. nú mut1 1.that 1.person 1.sbj ‘THAT person that ate the food’ DP1 b. DP2 Dem nú

D1

D2

NP

Ø

mut

Ø

bí pst2



eat

bíjɛ́ k ] 8.food

CP DP2

C Ø



↓jɛ́

TP DP2 a bí ↓jέ bíjέk

The derivation in (33b) is restricted to cases where the operator is a ‘true’ demonstrative (e.g. 20–21). Demonstratives can also remain in [Spec, CP], as in (32b), but they do not receive a focused interpretation there. The derivations in (32–33) for relative clauses thus mirror the two options for demonstratives themselves illustrated in example (30). It follows that either a demonstrative or a relative operator can occur in a restrictive relative clause, but not both, as both occur in the same position internal to DP2. This proposal thus derives the complementarity between demonstratives and relative operators (Section 3.3). In contrast, alternatives to the head-raising analysis of relative clauses cannot account for this fact. This is because these analyses, which include the matching analysis and the operator-movement analysis,10 both posit a second noun phrase which the relative clause adjoins to. If there were a distinct NP from the one in the relative clause, we would expect that it would be able to host a distinct demonstrative under the assumption that this demonstrative modifies NP or one of its projections. At the same time, this analysis raises the deeper question of why demonstratives and relative operators are in competition at all, or, stated differently, why demonstratives can serve as relative operators. This seems to be a language-­ particular property of Basaá: if relative operators are distinguished by some feature Op (cf. McCloskey 2002, Adger, and Ramchand 2005), then Op can also be

10 See Bhatt (2002) for discussion.

36 

 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman

realized by demonstratives in addition to their more contentful deictic features in Basaá. Thus, Basaá is simply a language with a rich set of relative operators, as we will discuss further in Section 5. Standard diagnostics for wh-movement,11 including weak and strong crossover effects (34–35) (Postal 1971) and island constraints (36) (Ross 1967) confirm that the gap strategy for relative clauses involve overt movement, as expected under the analysis above: (34) Weak crossover a. í-maaŋɡɛ́ 1 nú [ ___1 a ŋ́↓ aug-1.child 1.rel 1.sbj prs ‘The child1 that likes his1/2 friend’ b. í-maaŋɡɛ́ 1 nú [ lí-wándá jéé*1/2 aug-1.child 1.rel 5-friend 5.poss ‘The child1 that his*1/2 friend likes’ (35) Strong crossover a. í-maaŋɡɛ́ 1 nû [ ∅*1/2 aug-1.child 1.rel null.PRN ‘The child1 that he*1/2 likes’

a 1.sbj

ɡwɛ́ s like

líwándá 5-friend

lí 5.sbj

ŋ́↓ prs

ŋ́↓ prs

ɡwɛ́ s like

jéé1/2 ] 5.poss

ɡwɛ́ s like

___1 ]

___1 ]

(36) Complex NP island violation in subject (a) and object (b) relative w/gap ↓gwés í-kaat ↓yí a. *liwándá jɛ̂ m lí ŋ́ i [ mɛ́ ń 1 aug-9.book 9.rel 1sg prs know 5.friend 5.my 5.sbj prs like [  í-maaŋɡɛ́ 2 nú [  ___2 a ŋáŋ ___1 ]]] aug-1.child 1.rel 1.sbj read ‘*My friend likes the book that I know the child that read.’ ↓yí b. *liwándá jɛ̂ m lí ń í-maaŋɡɛ́ 1 nú [ mɛ́ ŋ́ -↓gwés 5.friend 5.my 5.sbj prs know aug-1.child 1.rel 1sg prs like [ í-kaat2 i [ ___1 a ŋáŋ ___2 ]]] aug-9.book 9.rel 1.sbj read ‘*My friend knows the child that I like the book that read.’

11 These restrictions on extraction have been shown to constrain all types of wh-movement including relativization in many languages, and are thus typologically robust tests for analyses involving overt movement. See Chomsky (1977) for English and McCloskey (2006) for discussion of the alleviation of island constraints with resumption in Irish and other languages.



2 Accessibility and demonstrative operators in Basaá relative clauses 

 37

Reconstruction facts that have been argued to specifically favor the head-raising analysis of relatives by Bhatt (2002) can also be reproduced in Basaá. For example, relative heads can receive bound readings under quantificational subjects inside the relative clause (37), and adjectives can receive interpretations internal to the relative clause (38): (37) Bound variable reconstruction í-↓fóto1 i [ hígií ŋ̀úrú a bí yɔŋ aug-9.photo 9.rel every 1.student 1.sbj pst2 take ndáp yé ↓é ] 9.house 9.his i ɓée iláám 9.sbj pst2.be 9.nice ‘The picture that every student took of his house was nice.’

___1

(38) Adjective reconstruction pɔ̌ m kaat1 i [  Pierre a ŋ́-kal lɛ́ Victor a ŋáŋ ___1 ] only book 9.rel P. 1.sbj pst1-say that V. 1.sbj read ‘The only book that Pierre said that Victor read’ (only > say, say > only) Example (37) is admittedly not a true case of bound variable reconstruction: the head DP does not contain a bound pronoun. However, the most salient interpretation of (37) is one where each student took a different picture, which would involve reconstruction of the head noun into the relative VP where it could be interpreted under the scope of hígií ‘every.’ Together, then, the facts in (34–38) support the head-raising analysis of restrictive relative clauses involving a gap in Basaá. A remaining challenge for this proposal is the recursion of relative operators in stacked relative clauses, only the first of which is in complementary distribution with demonstratives (27). The proposal above can countenance these facts if relative operators are base-generated in the specifier of a DP (DP2 below) which can take a CP complement (CP2 below). This latter CP must move to a position above the operator, as was shown to be necessary with nouns in simple (nonstacked) relative clauses in example (32b), and parallel to the movement of NP below (cf. Bianchi 2000: 132): (39) a. híŋgɔnda1 (hí) [ ___1 hí yé hiláám ] aug.19.girl 19.rel 19.sbj prs.be 19.beautiful ↓yí [ nyaŋ a ń ___1 ] mother 1.sbj prs know ‘The girl that is beautiful that his mother knows.’

*(hí) 19.rel

38 



 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman DP1

b. 

D1

CP1

H DP2

CP2

Op1

DP3 NP

DP3

hiŋgonda Op2 hí

DP2

C2 Ø

D3 NP

TP2

hí

C1

TP1

Ø

nya ŋ a ń ↓yí DP2

D2 CP2 Ø

DP3 hí yé hiláám

Ø

This structure accounts for the observed properties of stacked relative clauses. First, the configuration in DP3 is similar to the configuration in (34b) where it is adjacent to the upper D head, and the relative operator is optional in both cases. Second, the relative operator in DP2 is not adjacent to a upper D head, and this operator is obligatory. Third, the relative clause on the right, CP1, is predicted to have scope over CP2 on the left, due to the fact that it is structurally higher. This prediction is correct; (41a) has the interpretation of restricting a larger set of beautiful girls to a particular girl that his mother knows. Fourth, demonstratives are only predicted to alternate with Op2, because only Op2 occurs in a DP which takes a NP complement, which is the source of demonstratives (cf. 32). Finally, demonstratives are only predicted to be in complementary distribution with Op2 for the same reason.

4.3 Resumption and non-complementarity Now that the analysis of ‘strategy 1’ restrictive relative clauses involving a gap has been established, ‘strategy 2’ restrictive relative clauses involving a resumptive



2 Accessibility and demonstrative operators in Basaá relative clauses 

 39

pronoun will be shown to have a different structure. This proposal can account for the ability of relative operators to co-occur with demonstratives in relative clauses with resumption (22–24). There is an earlier literature analysis identifying the presence of resumptive pronouns in relative clauses and wh-question as evidence that they were not derived by movement. This has most famously been shown in Irish (McCloskey 1990, 2002, 2006), where there is a morphological distinction on the relative complementizer correlating with the presence of resumptive pronouns, and relative clauses involving resumption are not subject to island constraints. The following examples, which are identical to (36a-b) except for the existence of resumption, demonstrate that resumption alleviates island constraints in Basaá as well. (40) No complex NP island violation in relative clauses w/ resumption ↓yí a. liwándá jɛ̂ m lí ń í-maaŋɡɛ́ 1 nú [ mɛ́ 5.friend 5.my 5.sbj prs know aug-1.child 1.rel 1sg ↓gwés ŋ́ [ í-kaat2 i [ nyɛ́ 1 a ŋáŋ ___2 ]]] prs like   aug-9.book 9.rel 1.PRN 1.sbj read ‘??My friend knows the child that I like the book that he read.’ ↓gwés b. liwándá jɛ̂ m lí ŋ́ í-kaat1 i [ mɛ́ 5.friend 5.my 5.sbj prs like aug-9.book 9.rel 1sg ↓yí ń [  í-maaŋɡɛ́ 2 nú [ ___2 a ŋáŋ yɔ1 ]]] prs know aug-1.child 1.rel 1.sbj read ?? ‘ My friend likes the book that I know the child that read it.’ Unlike their English counterparts, these sentences are fully grammatical in Basaá. Thus, we conclude that relative clauses involving a resumptive pronoun do not involve movement in Basaá. This entails in turn that these relative clauses cannot be derived by the head-raising analysis detailed in section 4.2, but must simply occur as nominal adjuncts with a base-generated relative operator. The analysis of relative clauses involving resumption is illustrated with a possessive relative, which can occur left peripherally in the relative clause (10b): (41) a. í-maaŋgɛ́ 1 (nú) bijɛ́ k aug-1.child 1.rel 8.food ‘The child whose food I ate’

gwéé1 8.poss

mɛ 1sg

bí pst2

↓jɛ́

eat

gwɔ́ 1 8.PRN

40 

 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman b.

DP1 D1

NP

í NP

CP

N DP2

maaŋgéi Op nú

C Ø

D2

NP

Ø

bíjέk gɯééi

TP ˊ i mɛ b ↓jέ gɯɔ

The structure is the same in relative clauses that lack an overt NP complement with the operator. It follows from the structure in (41b) that demonstratives are not in complementary distribution with relative operators because the relative clause is an adjunct. Therefore, the presence or absence of a relative operator would not be expected to affect the internal syntax of the NP to which it attaches. Thus, when an additional demonstrative is present (22–24), it presumably attaches a position between the NP and the CP, and optionally fronts to [Spec, DP], as outlined in the previous section. Two issues remain. The first is the fact that the ‘near hearer’ demonstrative is blocked even in cases involving resumption. The second is how these proposals shed light on the licensing of the augment. One possible explanation for the persistent complementarity between the ‘near hearer’ demonstrative and the relative operator is that it represents a case of haplology, or Menn and MacWhinney’s (1984) Repeated Morph Constraint: when two identical elements are adjacent, one must be deleted: e.g. í-maaŋɡɛ́ (*nú) nú bijɛ́ k gwéé1 mɛ bí ↓jɛ́ gwɔ́ 1 ‘(lit.) the child that (*that) his food I ate it.’ Of course, the problem with this view is that even prenominal near-hearer demonstratives are impossible with the relative operator: *nú maaŋɡɛ́ nú bijɛ́ k gwéé1 mɛ bí ↓jɛ́ gwɔ́ 1 ‘(lit.) that child that his food I ate it.’ To block this latter case, a more powerful constraint would be required, which either blocked adjacent identical elements at any point in the derivation or at a certain distance. Yet the claim that the Repeated Morph Constraint would apply at an earlier point in the derivation which does not surface would be surprising, as this is generally a surface-level phenomenon. The alternative we adopt is that the ‘near hearer’ demonstrative is a default demonstrative, which is underspecified for deixis. Thus, it receives its ‘near-hearer’



2 Accessibility and demonstrative operators in Basaá relative clauses 

 41

interpretation by virtue of neither being distal or proximal. Because it lacks a lexical deictic specification, we take this demonstrative to be semantically vacuous in the presence of a relative clause and hence ruled out by a semantic economy constraint. The second issue is how the augment prefix is licensed in just the presence of a demonstrative and relative clause. The structures for head-raising relative clauses pursued in the previous section offered a potential explanation for the licensing condition, as in these structures the D took a CP complement rather than an NP, and thus was participating in a distinct structure. However, the ability for relative clauses involving resumption (e.g. 43) and demonstratives to license the augment cannot be made to follow from this proposal, as both of these modifiers elements are adjuncts. One promising line of explanation would be to say that demonstratives and relative clauses both attach to a higher position in the noun phrase than other modifiers, and are able to license the augment from this position. Evidence for this analysis comes from the H spreading facts noted at the end of section 3.1, where demonstratives were seen to be distinct from other modifiers in not allowing spreading from the N, a possible effect of their structural height. In general, however, we note with Creissels (this volume) that cases of definiteness being only marked in the presence of modification are not unattested cross linguistically, and with him observe that similar patterns occur in languages of the Balkans.

5 Grammaticalization While the previous section has examined relative clauses in Basaá from a formal perspective, this section puts into a historical context the distribution of the augment prefix in Basaá (section 3.1–3.2) as well as the complementarity between demonstratives and relative operators (section 3.3). These facts receive ready explanations from the idea that Basaá relative operators are not fully grammaticalized from demonstratives, and that, simultaneously, the augment came to be associated with deixis as well as definiteness historically. We will model this process in this section by using informal semantic features that approximate what was and is a more complex semantic representation. To begin, if the augment prefix in Basaá is a remnant of the Proto-Bantu augment, as suggested in Hyman (2003: 267), we suggest, based on its very irregular modern distribution both in Basaá and across the A group, that it once was a productive definite marker that has eroded and semantically bleached in different ways in different languages. This conclusion would not be surprising from a

42 

 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman

broader Bantu perspective, as the augment has different behaviors in different Bantu languages (e.g. de Blois 1970, Hyman and Katamba 1993). Thus, we suggest an initial stage where the augment had the feature [def], and the demonstratives had a feature [deix], relating to the semantic process of identification with a discourse index, as well as the feature [def], corresponding to their definite specification: (42) Stage 1 í- [def] nú [def, deix] At this stage, other modifiers such as relative clauses, adjectives, and possessives lacked the [deix] feature because these modifiers are fundamentally predicative rather identificational. At a second stage of grammaticalization, the augment only occurred with demonstratives. Because we assume that the augment was still definite at this stage, and hence would be expected to freely combine with other modifiers, we posit that this restriction in the distribution of the augment corresponded to the augment gaining a deix feature, meaning that it was only compatible with other modifiers with this feature: (43) Stage 2a í- [def, deix] nú [def, deix] This view is compatible with the suggestion by Makasso (2010) that the augment in Basaá is inherently deictic, and the claim by Van de Velde (this volume) that it serves as a localizing modifier in closely related Eton. Other semantic factors that may have differentiated these two markers, including the fact that the demonstratives contained other features marking proximity, as well as syntactic differences in their category and position. An analogy can be made here with negation and the Jespersen cycle: negation can be adverbial or inflectional, and there are often stages in the development of negation where both types of markers can co-occur, as in French (ne)-V pas. This leads us to the complementarity between relative operators and demonstratives in Basaá. Recorded instances of grammaticalization from demonstratives to relative markers are widespread (Diessel 1999; Heine and Kuteva 2002: 113). In some languages where this process has occurred, demonstrative relative markers and demonstratives themselves freely co-occur. Germanic provides examples of this late stage of grammaticalization. English is one case: that man that I knew; German is another: die Frau die meinen Bruder liebt. From this perspective, Basaá represents a language where the demonstrative has not fully grammaticalized



2 Accessibility and demonstrative operators in Basaá relative clauses 

 43

as a relative marker, but still retains its demonstrative status at some level. Of course, it is also relevant that the unmarked position of demonstratives in Basaá is between the head noun and the relative clause. We can model this stage in the development in the demonstrative using a similar representation to the one used for the augment prefix. At Stage 1, the demonstrative was only a deictic determiner. However, at Stage 2, demonstratives became polysemous due to the innovation of an op feature that is characteristic of relative operators: (44) Stage 2b í- [def, deix] nú1 [def, deix] nú2 [op, def, deix] At this intermediate stage, we expect that demonstrative operators may only have been used in definite noun phrases, and that neither the augment nor the demonstrative operator would be present in indefinite complex noun phrases. The final stage in the development of the augment and operator is the modern stage for Basaá: at some point the demonstrative operator lost its [def] feature, and as a result could be used in either indefinite or definite noun phrases, with only the augment remaining as a marker of definiteness: (45) Stage 3 í- [def, deix] nú1 [def, deix] nú2 [op, deix] Because the relative operator was no longer connected to definiteness, Stage 3 represented a kind of return to form for the augment prefix, as relative clauses provided a limited context where it marked definiteness. At the same time, Basaá relative operators are still at a relatively early stage of grammaticalization by virtue of the fact that they retain their deictic semantics, represented by the [deix] feature. Relative markers in many Bantu languages have fully grammaticalized from demonstratives into verbal prefixes, as in the southern Bantu languages discussed by Zeller (2004), and also Demuth and Harford (1999), where they lack demonstrative semantics. Such markers can be seen as retaining only the [Op] feature. A similar case of bleaching can be seen in languages like German, where relative operators identical to definite determiners can be used in indefinite noun phrases. A different grammaticalization pathway is represented by Bantu A-group ­languages like Eton, discussed by Van de Velde (this volume). In Eton, the augment prefix lacks any semantic contribution, and seems to be fully optional. In the context of the pathway represented above, we can take the augment prefix

44 

 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman

in Eton to have lost its [def] feature, simply serving as a redundant marker of general deictic features. In summary, then, we can identify a contiguous grammaticalization pathway from a Proto-Bantu stage where the augment prefix was purely a marker of definiteness to a later stage of development where the augment prefix only marks definiteness in certain contexts. Similarly, demonstrative operators in Basaá developed due to semantic enrichment of a demonstrative determiner followed by bleaching of the definite semantics typical of demonstratives.

6 Conclusion While certain aspects of relative clauses in Basaá are unsurprising from a crosslinguistic perspective, including the conformity of relativization to the Accessibility Hierarchy (Section 2), other aspects of relative clause formation are more particular to Basaá, such as the ability of relative clauses to license an augment prefix which overtly marks definiteness (Section 3.2) and the complementarity of demonstratives with relative operators (Section 3.3.). This complementarity, and the different distributions of demonstratives with gapped and resumptive relatives, was in turn shown in Section 4 to follow from two standard analyses of relative clauses, one which invokes head-raising, accounting for the complementary pattern, and another that relied on base-generation of the relative operator and adjunction of the relative clause to NP, which allowed multiple demonstratives. Looking forward, two of the main issues discussed in this paper – the contextually restricted marking of definiteness by the augment prefix and the complementarity of demonstratives and relative operators, pose promising topics for future typological research. Both are of historical and theoretical interest, and they are likely to exhibit additional shades of variation when more languages are considered, especially the typologically and genetically related languages spoken in Cameroon.

References Adger, David & Gillian Ramchand. 2005. Merge and move: Wh-dependencies revisited. Linguistic Inquiry 36. 161–193. Barlew, Jefferson & Emily Clem. 2014. The augment morpheme in Bulu. Presentation at the 88th Linguistic Society of America Annual Meeting, January 2–5, 2014. Minneapolis, MN. Bassong, Paul R. 2010. The structure of the left periphery in Basaá. Yaounde: University of Yaounde 1 MA Thesis.



2 Accessibility and demonstrative operators in Basaá relative clauses 

 45

Bernstein, Judy B. 1997. Demonstratives and reinforcers in Romance and Germanic languages. Lingua 102. 87–113. Bhatt, Rajesh. 2002. The raising analysis of relative clauses: Evidence from adjectival modification. Natural Language Semantics 10. 43–90. Bianchi, Valentina. 2000. The raising analysis of relative clauses: A reply to Borsley. Linguistic Inquiry 31. 123–140. Bitjaa Kody, Zachée D. 1990. Le système verbal du basaa. Yaoundé : Université de Yaoundé Thèse de 3ème Cycle. Blois (de), Kornelis F. 1970. The augment in the Bantu languages. Africana Linguistica 4. 85–165. Bot Ba Njock, Henri-Marcel. 1970. Nexus et nominaux en basaa. Paris: Université de Paris-Sorbonne dissertation. Brugè, Laura. 2002. The positions of demonstratives in the extended nominal projection. In Guglielmo Cinque (ed.), Functional structure in DP and IP: The cartography of syntactic structures, Volume I, 15–53. Oxford: Oxford University Press. Carstens, Vicki. 2005. Agree and EPP in Bantu. Natural Language & Linguistic Theory 23. 219–279. Cheng, Lisa L.-S. 2006. Decomposing Bantu relatives. In Christopher Davis, Amy R. Deal & Youri Zabbal (eds.), Proceedings of the 36th Annual Meeting of the North East Linguistic Society, 197–215. Amherst, MA: Graduate Linguistic Student Association. Chomsky, Noam. 1977. On wh-movement. In Peter Cullicover, Adrian Akmajian & Thomas Wasow (eds.), Formal syntax, 71–133. New York: Academic Press. Demuth, Katherine & Carolyn Harford. 1999. Verb raising and subject inversion in Bantu relatives. Journal of African Languages and Linguistics 20. 41–61. Diessel, Holger, 1999. The morphosyntax of demonstratives in synchrony and diachrony. Linguistic Typology 3. 1–49. Dimmendaal, Gerrit J. 1988. Aspects du basaá: Bantou zone A, Cameroun. Paris: Peeters/Selaf. Giusti, Giuliana. 1997. The categorial status of determiners. In Liliane Haegeman (ed.), The new comparative syntax, 95–123. New York: Longman. Giusti, Giuliana. 2002. The functional structure of noun phrases: A bare phrase structure approach. In Guglielmo Cinque (ed.), Functional structure in DP and IP: The cartography of syntactic structures, Volume I, 54–91. Oxford: Oxford University Press. Heine, Bernd & Tania Kuteva. 2002. World lexicon of grammaticalization. Cambridge: Cambridge University Press. Henderson, Brent. 2009. Anti-agreement: Locality of movement or agreement? Proceedings from the Annual Meeting of the Chicago Linguistics Society 45 (1). 89–102. Hyman, Larry M. 2003. Basaá (A43). In Derek Nurse & Gérard Philippson (eds.), The Bantu languages. 257–282. New York: Routledge. Kayne, Richard. 1994. The Antisymmetry of syntax. Cambridge, MA: Massachusetts Institute of Technology Press. Keenan, Edward L & Bernard Comrie. 1977. Noun phrase accessibility and universal grammar. Linguistic Inquiry 8. 63–99. Koopman, Hilda & Anna Szabolszi. 2000. Verbal complexes. Cambridge, MA: Massachusetts Institute of Technology Press. Lemb, Pierre & François de Gastines. 1973. Dictionnaire basaá-français. Douala: Collège Libermann. Leu, Thomas. 2015. The architecture of determiners. Oxford: Oxford University Press. Lewis, M. Paul, Gary F. Simons & Charles D. Fennig (eds.). 2016. Ethnologue: Languages of the world, 19th edn. Dallas, Texas: SIL International. Online version: http://www.ethnologue. com (accessed 16 November 2016).

46 

 Peter Jenks, Emmanuel-Moselly Makasso and Larry M. Hyman

Makasso, Emmanuel-Moselly. 2010. Processus de relativisation en bàsàa : De la syntaxe à la prosodie. In Laura Downing, Annie Rialland, Jean-Marc Beltzung, Sophie Manus, Cédric Patin & Kristina Riedel (eds.), Papers from the Workshop on Bantu Relative Clauses (ZAS Papers in Linguistics 53), 145–158. Berlin: Zentrum für Allgemeine ­Sprachwissenschaft. Makasso, Emmanuel-Moselly. 2012. Metatony in Basaa. In Michael R. Marlo, Nikki B. Adams, Christopher R. Green, Michelle Morrison & Tristan M. Purvis (eds.), Selected Proceedings of the 42nd Annual Conference on African Linguistics: African Languages in Context, 15–22. Somerville, MA: Cascadilla Proceedings Project. McCloskey, James. 1990. Resumptive pronouns, A’-binding and levels of representation in Irish. In Randall Hendrick (ed.), Syntax and semantics 23: The syntax and semantics of modern Celtic languages, 199–248. New York / San Diego: Academic Press. McCloskey, James. 2002. Resumption, successive cyclicity, and the locality of operations. In Samuel D. Epstein & T. Daniel Seely (eds.), Derivation and explanation in the minimalist program, 184–226. Malden, MA: Blackwell. McCloskey, James. 2006. Resumption. In Martin Everaert & Henk van Riemsdijk (eds.), The Blackwell companion to syntax, 94–117. Malden, MA: Blackwell. Menn, Lise & Brian MacWhinney. 1984. The Repeated morph constraint: Toward an explanation. Language 60. 519–541. Ngonyani, Deo S. 2001. Evidence for head raising in Kiswahili relative clauses. Studies in African Linguistics 30 (1). 59–73. Schneider-Zioga, Patricia. 2007. Anti-agreement, anti-locality and minimality: The syntax of dislocated subjects. Natural Language and Linguistic Theory 25. 403–446. Van de Velde, Mark. 2005. The order of noun and demonstrative in Bantu. In Koen Bostoen & Jacky Maniacky (eds.), Studies in African comparative linguistics with special focus on Bantu and Mande, 425–441. Tervuren: Royal Museum for Central Africa. Vries (de), Mark. 2002. The syntax of relativization. Utrecht: Landelijke Onderzoekschool Taalwetenschap dissertation. Zeller, Jochen. 2004. Relative clause formation in the Bantu languages of South Africa. Southern African Linguistics and Applied Language Studies 22 (1–2). 75–93. Zwart, Jan-Wouter. 2000. A head raising analysis of relative clauses in Dutch. In Artemis Alexiadou, Paul Law, André Meinunger & Chris Wilder (eds.), The syntax of relative clauses, 349–85. Amsterdam: John Benjamins.

Mark Van de Velde

3 The augment as a construct form marker in Eton relative clause constructions Abstract: The Narrow Bantu language Eton has two strategies of relativisation, a gap strategy and a resumptive pronoun strategy. In both strategies, relative clause constructions are head marked by means of a prefix ı́- on the relativised noun. This prefix is called the augment in Eton, because it is similar to the augment in other Bantu languages in its form and position and in its demonstrative origin. However, contrary to the augment in most other Bantu languages, it cannot be situated in Greenberg’s (1978) three stage cycle of the development of definite articles to noun markers. Instead, I argue that it should be analysed as a construct form marker that codes the syntactic relation of localising modification, a relation marked in other north-western Bantu languages as well. The chapter closes with a discussion of a number of plausible diachronic scenarios that could have led to this typologically unusual construction type.

1 Introduction1 The only formal characteristic that all restrictive relative clause constructions have in common in Eton is that the relativised noun is marked by means of a prefix ı́- ~ H- traditionally called the augment. Eton does not normally use a relative relator and only a subset of TAM-forms have a special relative verb form, so that for subject relatives the augment on the relativised noun is often the only indication for the presence of a relative clause. Hence, relativisation is basically head marking, and the augment is best analysed as a construct form marker in Eton (see Creissels (2009) for an earlier use of this notion in the analysis of subSaharan languages). I use the term augment for the Eton construct form marker

1 The research for this chapter has been partially funded by a public grant overseen by the French National Research Agency (ANR) as part of the progam “Investissements d’Avenir” (reference: ANR-10-LABX-0083). All Eton data are primary data gathered by the author. Elicited examples have been double-checked with at least four different speakers in different contexts and tend to be clearer examples of constructions also recorded in spontaneous speech. Mark Van de Velde: LLACAN, CNRS, Sorbonne Paris Cité, INALCO, Villejuif and Research Centre for Nigerian Languages, KWASU, Malete) [email protected] DOI 10.1515/9783110469547-003

48 

 Mark Van de Velde

and its cognates in closely related languages such as Ewondo, Fang and Basaá, without implying that this morpheme can be identified as a reflex of a prenominal Proto-Bantu augment as reconstructed by Meeussen (1967: 99). Rather, the term is here used for any morpheme that is synchronically preposed, procliticised or prefixed to a nominal element and that can be argued to be cognate with an augment, pronominal prefix or weak demonstrative in other Bantu languages. Section 2 summarises the grammatical description of relative constructions provided in Van de Velde (2008), which was thoroughly double-checked for this paper. We will see that Eton uses two strategies, a gap-strategy and a resumptive pronoun strategy. I will subsequently describe the dedicated relative verb forms. Section 3 shows that the augment is an inflectional morpheme in Eton. Section 4, finally, proposes a scenario for its origin and development. Eton (ìtón) is the northernmost variety of the Beti-Bulu-Fang language cluster, a group of Narrow Bantu languages known as A70 in Guthrie’s referential classification. It is spoken by an unknown number of speakers, possibly about 400 000 (Van de Velde 2008: 2, fn2), in the Lékié department of the Centre province of Cameroon.

2 Relative clause constructions in Eton Relative clause constructions are here defined as constructions in which a nominal constituent (henceforth the head NP) is modified by a subordinate finite clause (the relative clause) by specifying the role of its referent in the situation described by the relative clause (adapted from Andrews 2007: 206). This definition excludes participial constructions, in which the modifying clause is non-finite (1). (1) (elicited) àjám ḿpùgl ótíìgı̀ à-H-ʤám-H ɴ̀-pùglà ó-tìg-Lgı̀ i-pst-cook-nf 3-sauce iii-be.thick-g2 ‘She made a thick sauce.’

2 I follow the Leipzig Glossing Conventions. Additional abbreviations are found in the list provided. Non-final markers (nf) appear on verbs and pronouns that are not in final position of their syntactic constituent. Link tone (lt) is a submorpheme that appears whenever an infinitive is followed by a nominal form, whatever the syntactic relation between the infinitive and the following nominal (if any). The morpheme glossed g (“g-form”, a multifunctional verbal morpheme) is hard to gloss in a non-arbitrary way. It is a reflex of the so-called Prefinal morpheme in ProtoBantu and appears in a number of TAMN forms. The morphology of Bantu verb forms tends to be



3 The augment as a construct form marker in Eton relative clause constructions 

 49

The definition equally excludes focus constructions that have formal similarities with relative constructions, but that do not imply adnominal modification, or embedding in a matrix clause. In (2) the focused argument dwé ‘name’ is in pre-verbal focus position followed by a focus pronoun and a clause that has the formal characteristics of a relative clause (a relative verb form and a resumptive pronoun after the preposition èèy). Focus pronouns are segmentally identical to personal pronouns, but they always have a falling tone, as opposed to the dissimilating high tone of personal pronouns (low in agreement pattern IX). Therefore focus pronouns may historically derive from personal pronouns preceded by the augment. (2) tɔ̀ dwé dɔ̂ wɔ̀ yı̂ kè èèy dɔ́ L-kə̀ tɔ̀ d-óé dɔ̂ ù-à-H-jı̀ conc 5-name v.fppr 2sg-sp-rel-want inf-go ‘Even if it is the name you want to take with you’

èːj with

dɔ̋ v.ppr

Eton relative clauses are always externally headed. They occur after the head noun and can only be followed by a demonstrative in a nominal constituent (see examples (10) and (16) below). Normally, relative clauses are not introduced by a relative relator in Eton, but my corpus contains a small number of examples with an optional non-agreeing relator yá. Example (3) illustrates this with a relative clause construction headed by a cognate object of the relative verb. Many especially older speakers reject the use of the relator yá, claiming that it is due to Ewondo influence. (3) íkúkwâlɔ̀ [á béŋgákwálɔ̂ èèy ɲé] H-ı̀-kú∼kɔ́ là yá bə́-ŋgá-kɔ́ là èːj aug-7-talk rr ii-rp-talk with ‘The talk that they talked about her...’

ɲɛ́ i.ppr

A pronominal head of a modifying relative clause is expressed by means of an augmented demonstrative (4). The augment is required here for two reasons, nominalising a demonstrative (see section 3) and marking the head of a relative clause.

non-compositional. Arabic numbers in glosses are used to refer to nominal class prefixes (except before sg and pl where they stand for first or second person), roman numbers to agreement prefixes. Nouns that do not have a noun class specification in the glosses belong to the so-called class 1a, which I have analysed elsewhere as being genderless. Superscript H and L stand for floating high and low tones, respectively.

50 

 Mark Van de Velde

(4) íbá [màyêm] ı́-bá mə̀-à-H-jɛ̀ m aug-ii.dem 1sg-sp-rel-know ‘The ones (siblings) that I know’ Relative clauses have a rather heavy functional load in Eton, and are therefore frequent. Relative constructions frequently occur as adverbial clauses. Whenclauses, for instance, are relative clauses with the noun ìjɔ̀ ŋ ‘time, occasion’ as their head. An interesting type of after-clauses are relative clauses headed by a class 5 noun derived from the verb of the adverbial clause and one type of reason clause is a relative clause with the noun m̀mól ‘way, manner’ as its head. A relative strategy is also often employed in circumstances where European languages would use adnominal adjectives, compounding or deverbal nominal derivation. See Van de Velde (2008) for more details.

2.1 Relativisation strategies Eton has two strategies for relativisation: a resumptive pronoun strategy and a gap strategy. The resumptive pronouns strategy is used for the complement of prepositions other than locative á, as well as for possessors. The gap strategy is used elsewhere. SUBJ > OBJ > LOC > OBL > POSS Depending on the construction, the resumptive element used in the resumptive pronoun strategy can be a personal pronoun, a possessive pronoun or the invariable word ı́↓tə́, translatable as ‘inside’ in some contexts. The relativised complement of the preposition ábɔ̂ ‘at [+human]’ is represented within the relative clause by a personal pronoun that follows the preposition in (5b). Example (5b) is ambiguous and can also be read as an example of the gap strategy, meaning ‘the doctor who lives at his place’. A pronominal complement of the preposition ású ‘for’ has the form of a possessive pronoun, also when used resumptively, due to the genitive (aka connective) origin of this prepositional construction ( [CF=N Mod] It is useful to be reminded here that the reflex of Dem in (30), viz. the augment, can still fulfil the role it had at the stage schematised in (30), namely to turn an adnominal modifier into an independent nominal constituent. In fact, according to Redden (1979: 67), the construction in (30) can be observed when a noun is modified by a demonstrative in contemporary Ewondo: “Sometimes the definite article is repeated before the demonstrative, e.g. /é mod é ɲɔ́ / this man; but most of the time only the /é/ before the noun is used.” Starting from the situation in (31), another possible path involves reinforcing of the demonstrative relator by means of a prenominal light demonstrative. The likeliness of such a step is illustrated by Bantu languages with a circumposed demonstrative. This scenario may be more compatible with the current situation in Basaa, where relative clauses are still optionally introduced by a demonstrative relator, and where the augment does not appear when a demonstrative modifier is reinforced by being fronted. All we would need to assume for Eton, then, is that relative clauses used to be introduced by a relator of demonstrative origin. Finally, a more direct scenario, starting from the situation in (30), cannot be ruled out. It involves the use of a weak, prenominal cataphoric that points ahead



3 The augment as a construct form marker in Eton relative clause constructions 

 65

to the information that will allow the identification of the noun’s intended referent (‘this man, the one that I met yesterday, …).

5 Conclusion The most remarkable characteristic of relative clauses in Eton is that they are head marked by means of the augment, which therefore should be analysed as a construct form marker. Given that the gap strategy is the most common one for relative clause formation in Eton and that several tenses do not have a special relative verb form, this construct form marker can be the only formal means for distinguishing between subject relative clauses and non-relative clauses. The scenario for the origin of this construct form marker that I elaborate in the second part of this chapter may shed a new light on the origin and evolution of the augment in the Bantu languages, as well as on that of similar phenomena in other Niger-Congo languages. The proposed initial appositive structure could explain phenomena as diverse as the typologically unusual word order phenomena in the noun phrase of many Benue-Congo languages and the prosodic break that often occurs between nouns and augmented modifiers. More research should point out whether this scenario can be integrated in a general scenario for the origin and evolution of the augment in the Bantu languages, or whether it is one of several independent evolutions.

References Andrews, Avery D. 2007. Relative clauses. In Timothy Shopen (ed.), Language typology and syntactic description. Volume II: Complex constructions, 206–236. Cambridge: Cambridge University Press. Creissels, Denis. 2009. The construct form of nouns in African languages. In Peter K. Austin, Oliver Bond, Monik Charette, David Nathan & Peter Sells (eds.), Proceedings of Conference on Language Documentation and Linguistic Theory 2, 73–82. London: School of Oriental and African Studies. Elderkin, Edward D. 2003. Herero. In Derek Nurse & Gérard Philippson (eds.), The Bantu languages, 581–608. London: Routledge. Leitch, Myles F. (2003). Babole. In Derek Nurse & Gérard Philippson (eds.), The Bantu languages, 392–421. London: Routledge. Meeussen, Achille E. 1967. Bantu grammatical reconstructions. Africana Linguistica III. 79–121. Nsuka Nkutsi, François. 1982. Les structures fondamentales du relatif dans les langues bantoues. Tervuren: Royal Museum for Central Africa. Ondo Mebiame, Pierre. 2001. L’augment en fang ntumu. Revue Gabonaise des Sciences du Langage 2. 60–77.

66 

 Mark Van de Velde

Puèch, Gilbert. 1988. Augment et préfixe nominal en ngubi. Pholia 3. 247–256. Redden, James E. 1979. A descriptive grammar of Ewondo. (Occasional Papers on Linguistics 4). Carbondale IL: Department of Linguistics, Southern Illinois University. Rijkhoff, Jan. 2002. The noun phrase. Oxford: Oxford University Press. Rijkhoff, Jan. 2008. Descriptive and discourse-referential modifiers in a layered model of the noun phrase. Linguistics 46 (4). 789–829. Van de Velde, Mark. 2005. The order of noun and demonstrative in Bantu. In Koen Bostoen & Jacky Maniacky (eds.), Studies in African comparative linguistics with special focus on Bantu and Mande, 425–441. Tervuren: Royal Museum for Central Africa. Van de Velde, Mark. 2008. A grammar of Eton. (Mouton Grammar Library 46). Berlin: Mouton de Gruyter. Van de Velde, Mark. 2010. The syntax of verb complements and the loss of the applicative in Eton (A71). In Karsten Légère & Christina Thornell (eds.), Bantu languages: Analyses, description and theory, 281–294. Cologne: Rüdiger Köppe Verlag. Van Hille, Marion. 1989. Eléments de description du syntagme nominal en puku, langue bantoue de zone A. Bruxelles: Université Libre de Bruxelles MA thesis. Van Houdt, Bettie. 1987. Eléments de description du leke, langue bantoue de zone C. Tervuren: Royal Museum for Central Africa.

Théophile Ambadiang

4 Relative clauses and relativization processes in Nugunu Abstract: This chapter aims at describing and discussing the relative clauses and relativization processes typical of Nugunu, a Bantu language (A62) spoken in the Centre Region of Cameroon. After a brief presentation of Nugunu, the structures and functions involved in the formation of the different kinds of relative clause found in the language are examined. Our analysis suggests that whereas none of the elements and processes involved in relativization is either exclusive to relative clauses or sufficient for relativization to occur in the language, some of them are involved conjointly in the formation of a given relative clause. Lastly, it is noted that restrictive and non-restrictive relative clauses pattern differently regarding intonation, though a more detailed study of prosody may improve our understanding not only of relative clauses, but also of other constructions which subsume processes, such as focus, which seem to be closely related to relativization.

1 Introduction The aim of this paper is to describe the relative clauses and give an account of the relativization processes in Nugunu, a Bantu language spoken in the centre of Cameroon, Mbam and Inoubou Division. This language, whose grammar shows properties typical of Northwest Bantu, is also known as Gunu and even as Yambassa, a name which is more commonly used for the ethnic groups whose languages are subsumed under the Yambassa linguistic complex (A.62). Whereas Guthrie (1953) includes the latter in the Sanaga group (Bantu A.60), Nugunu appears among Bantoid languages under the number 541 in the Atlas linguistique du Cameroun (Dieu and Renaud 1983), and is classified by the 16th edition of Ethnologue (Lewis, Simons, and Fennig 2016) as “Niger-Congo, Atlantic-Congo, Volta-Congo, BenueCongo, Bantoid, Southern, Narrow Bantu, Northwest, A, Sanaga (A. 622). Besides Nugunu, the Yambassa linguistic complex is generally considered to consist of Mmala, Libie, Baca and Kalong, sometimes considered as forming a dialect continuum. The total number of speakers oscillates between 43.000 and 45.000. Nugunu is, of all these languages, the one which has most speakers, 35.000, according to Lewis, Simons and Fennig (2016). It is also the most widely Théophile Ambadiang: Universidad Autónoma de Madrid, [email protected] DOI 10.1515/9783110469547-004

68 

 Théophile Ambadiang

spoken, as evidenced by the fact that most of the Yambassa territory is inhabited by native speakers of Nugunu dialects (Nɔcala, Nukefige, Nɔkɔ́ a and Nɔɔmbɛsa). Traditionally, these dialects have been grouped under two headings, Gunu Nord and Gunu Sud (northern and southern Gunu) which, however, do not match the linguistic situation of the region. The asymmetry between the former, administration-based geography, and the linguistic map of the region shows up clearly in the descriptions proposed in Robinson (1983: 12) and Paulian (1986). Whereas Robinson’s description is consistent with the political and administrative entities (so-called cantons) inherited from the colonial period (cf. Canton Gunu Nord (Guientsing, Bouraka, Balɛ́ ama and Ɔmbɛsa); Canton Gunu Sud (Assala I, Assala II, Guéfigué, and Bakɔ́ a)), it loses sight of the fact that Nɔkɔ́ a, the dialect of Bakɔ́ a, has much more in common with Gunu Nord than the other dialects included in Gunu Sud. Paulian’s description seems to privilege linguistic similarity, since she groups Nɔkɔ́ a along with Gunu Nord, even though she maintains an administrative interpretation of the latter. Nɔkɔ́ a and the dialects included in the dialectal complex subsumed under Gunu Nord will be the focus of our study. In contrast to the differences that occur in the phonology of these dialects, more specifically in vowel harmony (see Ambadiang 1991), syntactic structures do not show much dialectal variation, not even in interaction with tonal processes. The paper is organised in the following way. First, in Section 2, we describe the structures which are typical of relative clauses in Nugunu, as well as constructions which are most commonly associated with the different kinds of relative clauses found in this language. In Section 3 we discuss the relativization processes which occur in Nugunu, and try to account for the role and the effects of the elements involved in the formation of bound relative clauses. On the basis of a brief comparison of relative clauses and other constructions which result from processes such as focus and fronting, for instance, we suggest that whereas none of these elements is exclusive to relative clauses, it is the fact that some of them are involved conjointly in clause formation that is really specific to relativization. A summary is given in Section 4 followed by some concluding remarks.

2 Relative clauses in Nugunu 2.1 Structure of relative clauses 2.1.1 Finite relative clauses Gunu relative clauses follow the head noun or the antecedent, which is part of a main clause. They appear in two forms, depending on whether the verb is inflected



4 Relative clauses and relativization processes in Nugunu 

 69

or not. In both cases, however, a head noun phrase is co-referential with one of the arguments of a clause which modifies it. The most common form, in which inflected verbs are involved, has a structure like the one shown in (1). (1) Relative clauses with finite verbs (Head Noun) – Relativizing element – Subordinate clause From (2) and (3) it appears that the subordinating or relativizing element is similar to one of the demonstrative forms and, more specifically to “the demonstrative with the first, proximity position” (Du Plessis 2010: 8). Due to this fact, the structure in (1) does not, by itself, determine whether a clause is relative or not, as shown in (2), where class 1 mɔ́-ɔnyɔ functions as a demonstrative. (2) mɔ-ádɔmbɔ́ mɔ́-ɔnyɔ a-báa-bɔla 1-dem 1-child 3sbj-pst1-arrive “This boy has arrived today”

ínoni today

In contrast to what is observed in (2), structures such as (3) show some peculiarities which, together with the subordinating element, account for their nature as relative clauses. Such features concern mostly the morphotonology of the verb. (3) mɔ-ádɔmbɔ́ mɔ́-ɔnyɔ a-bá-ná-bɔlá 1-child 1-dem 3sbj-pst1-m-arrive “The boy who has arrived today”

ínoni today

The differences between (2) and (3) are related to morpho-tonological processes which affect the default or unmarked verb forms, turning them into marked verb forms (Patman 1991, 1995: 5)). Such processes consist mainly in the combination of an unmarked verb form with either -(n)á- or -mɔ́ (or their [+ATR] alternants: -(n)é-, -mú)) as shown in (4), with –bana “to read”, a low-tone verb, as an example. (4) Tense Present: “(s)he reads”

Default forms a-bana 3sg-read Past 1: “(s)he has read” a-báa-bana Past 2: “(s)he red” a-á-baná Past 3: “(s)he red” a-mba-baná Future 1: “(s)he is going to read” a-gaá-baná Future 2: “(s)he will read” a-ná-bana Future 3: “(s)he will read” a-nga-baná

Marked forms a-bana-mɔ́ 3sg-read-m a-bá-ná-baná a-ná-baná a-mba-á-baná a-ga-ná-baná a-ná-bana-mɔ́ a-nga-á-baná

As (4) shows, all marked verb forms end in a high tone, independently of their segmental make up. Furthermore, the affixes mentioned above compete in the formation of the marked verb forms. On the one hand, -mɔ́ is used with PRESENT

70 

 Théophile Ambadiang

and FUTURE 2 forms, i.e. in cases in which the default or simple verb form and the marked one do not differ either in their segmental shape or their tonal pattern, except precisely for the suffix, as shown in pairs such as a-bana/a-bana-mɔ́ “(s) he reads” and a-ná-bana/a-ná-bana-mɔ́ “(s)he will read”. The other two markers are in complementary distribution: -ná- is used with PAST1, PAST2 and FUT1 verb forms, i.e. with default forms which have two consecutive vowels in their inflection, whereas -á- is used in the remaining cases, in which no such vowel sequences show in the default forms. 2.1.2 Non-finite relative clauses The second strategy by means of which relative clauses are formed in Nugunu typically involves, in a structure like (5), a connective form and a predicate which, though commonly described as a non-finite verb form, may also be an adjective and even a noun. (5) Non-finite relative clauses (Head Noun) – Connective – ‘Predicate’ In spite of the fact that the constructions subsumed under (5) share the function of attribution (de Vries 2002: 58), only structures with a non-finite verb form as their head are generally considered to be functionally equivalent to finite relative clauses in Bantu languages (Ngonyani 2006). In Nugunu, however, a non-finite verb form may be replaced by an adjective or even a noun, although in the latter case the resulting structure may also be analysed as a noun followed by its complement. More generally, these constructions may be characterized as phrases whose heads are non-finite verb forms, adjectives or nouns which do not refer to definite, specific entities (Adamou and Castaouec 2010), i.e. (non-finite) predicates. Although the structural similarities between relative clauses with finite and non-finite verb forms are more striking than those observed with the other types of structure, the fact that all of them may coordinate with finite relative clauses, as in (6), suggests that, at the least, such constructions are functionally equivalent to finite relative clauses (see de Vries (2002: 58-61) for discussion). (6) a. ba-ábayɔ́ bá-aba bí-íyimene-mú gɔ-bana na bá 2-child 2-re 2sbj-know-m 15-read with 2.con bá-lɛ́ dégédégé gɔ-hɔɔ́ na táŋɔ 15-learn 9.calculus 2sbj-be many “The children who know to read and those who learn calculus are numerous”



4 Relative clauses and relativization processes in Nugunu 

 71

b. ba-ábayɔ́ bá ba-dɔmbɔ́ na bá-aba bí-íyimene-mú 2-child 2.con 2-young prep 2-re 2sbj-know-m gɔ-bana bá-lɛ́ bá-táanɔ 15-read 2sbj-be 2-five “The children who are young and those who know to read are five”. A peculiarity of the structure in (5) is that the subordinating element is a so-called connective or associative form which is followed by a phrase whose head behaves functionally as a modifier (much in the way finite relatives do). As the data in (7) show, there is a strong correlation between the use of infinitive and participial verb forms and the function of the head noun modified by the relative clause. Thus, infinitives are used with head nouns which function as the subject (and less typically as the object) of transitive verbs (7a, b), whereas past participles tend to appear with head nouns syntactically interpreted as subjects of passive verbs (7c) and intransitive verbs (7d). (7) a. mɔ-ádɔmbɔ́ ba gɔ-lafa kálata 1-child 1.con 15-tear 9/10.book “The child who tears books/has torn a book” b. kálata ya gɔ-bana 9/10.book 9.con 15-read “The/A book (which is) to read/A reading book” c. kálata ya lafá-alú 9/10.book 9.con tear-PP “The/A torn book” d. mɛ-ɛ́ tɛ má gɔmá-alú 6a-tree 6a.con grow-PP “(The) trees that have grown” This paper focuses on finite relative clauses, and only some cursory observations are given here with respect to non-finite relatives, whose non-canonical nature has been stressed, among others, by Dixon (2010: 360). Whereas finite relative clauses may be either restrictive or non-restrictive, structures such as those in (7) tend to have only a restrictive interpretation. This is also the case with structures in which an adjective replaces the non-finite verb, as evidenced in the contrast between mɔádɔmbɔ́ ba ɔdɔmbɔ́ “the/a boy who is young” (boy CON young) and mɔádɔmbɔ́ ɔdɔmbɔ́ “a young boy” (boy young), whereas constructions in which the head is a noun, such as mɔɔtɔ ba mmɔ́ nda “the/a man who is bald” (‘man CON baldness’), may also be described as noun phrases (cf. “the/a bald man”). Moreover, the presence of a demonstrative at the right edge of structures such as those in (7) has the effect of adding deictic information to the whole construction,

72 

 Théophile Ambadiang

without modifying the restrictive interpretation of the relative clause. Thus, in the case of (7a), for instance, mɔádɔmbɔ́ ba gɔlafa kálata mɔ́ ɔnyɔ/múunú is glossed as “this/that boy who tears books”.

2.2 The syntax of finite relative clauses The use of relative clauses is generally associated with noun modification (Riedel 2010) and with focus or clefting processes (Zeller 2004; Simango 2006; Du Plessis 2010). Besides the restrictive and non-restrictive interpretations alluded to above and to which we will return briefly below, finite relative clauses may differ according to diverse factors, such as the presence of a head noun phrase (headless or free vs. bound relative clauses), the function and nature of their head (subject, object, locative relative clauses), the passive or active nature of the verb, etc. These differences, only a small portion of which will be considered here, have in many cases interesting implications for the syntactic structure of the relative clauses (Dixon 2010; Keenan 1985; Kuteva and Comrie 2006). We will begin with differences related to the function and nature of the head noun phrase. 2.2.1 Bound relative clauses The most common relative clauses in Bantu languages have a subject or an object noun phrase as their head. The differences between these two types of clause are illustrated in (8) and (9)–(11). In (8), a subject noun phrase is the head of the relative clause. (8) Relative clauses with a subject noun phrase head (8) a. ba-ábayɔ́ bá-aba bí-íyimene-mú gɔ-nɛ́ ŋa bá-yɔnɔ 2-child 2-re 2sbj-know-M 15-swim 2sbj-play naá má-ŋá prep 6a-sea “The children who know to swim play in the sea” b. kálata sɛ́-ɛsɛ sí-é-né-húmé iyó 9/10.book 10-re 10sbj-pst2-m-come out yesterday sɛ́-lɛ dɛ́ a 10-be expensive “The books which came out yesterday are expensive” In (9)–(11) co-reference between the head noun phrase and an argument of the relative clause does not imply that they are equivalent from a functional point of view. Thus, the head noun functions as a subject in the main clause, whereas the



4 Relative clauses and relativization processes in Nugunu 

 73

argument which shares reference with it may function as a complement or even as an adjunct in the relative clause. (9) Relative clauses with an object noun phrase head [direct object in the RC] (9) a. bɛ-la bɛ́-ɛbɛ m-bá-ná-sɔgá ínoni bɛ́-lɛ́ bɛ-nɔ́ gɔ́ 8-cloth 8-re 1sgsbj-pst1-m-wash today 8-be 8-nice “The clothes I have washed today are nice” b. u-gúle ba-mɛ mɔ́-ɔnyɔ u-bé-né-mu-bédígíe 1-friend 1-my 1-re 2sbj-pst1-m-3sgrpr-greet naá bu-elí e-yégel-en-i-e táŋɔ prep 14-work 3sgsbj-learn-it-caus-fv 9.calculus “My friend whom you have greeted at work teaches maths” In contrast to what is observed in (8) and the pairs of sentences in (10)-(11), the presence of an object marker in the verb form in (9b) and its absence in (9a) is linked to the fact that the head is a [+human] noun in (9b) but not in (9a) in which the resumptive pronoun is optional at best. This contrast is not observed with indirect objects and with possessor heads. With the former an object marker is required in the verb form in all cases, as shown in (10), whereas a possessive adjective is necessary with the latter, as in (11). (10) Relative clauses with an object noun phrase head [indirect object in RC]) (10) a. ba-ábayɔ́ bá-aba dɛ-mba-á-bɔ́ -fá kálata 2-child 2-re 1plsbj-past3-m-2rpr-give 9/10.book bá-gaá-bɔl-á nembeli 2sbj-fut1-arrive-fv tomorrow “The children to whom we gave books (will) arrive tomorrow” b. bi-dule bɛ́-ɛbɛ di-bí-gídene-mú mbasa bɛ́-lɛ́ bɛ́-andɛ́ 8-heap 8-re 1plsbj-8rpr-add-m 9/10.corn 8-be 8-two “The heaps to which we add corn are two” (11) Relative clauses with a possessor head (11) a. ba-ábayɔ́ bá-aba ɔ-bá-ná-bɔgɔ́ kálata sá-bɔ́ 2-child 2-re 2sgsbj-pst1-m-seize 9/10.book 10-their bá-a-bɔl-á 2-past1-arrive-fv “The children whose books you seized have arrived” b. mɛ-ɛ́ tɛ má-ama dɛ-á-ná-gásá bɛ-samɔ́ bá-má iyó 1pl-pst2-m-pick 8-fruit 8-their yesterday 6a-tree 6a-re má-gɔma gɔana naá mbɔ́ lɔ ya má-ŋá 6a-grow only prep 9/10.shore 9.con 6a-sea “The trees whose fruits we picked yesterday only grow by the seaside”

74 

 Théophile Ambadiang

(12)–(14) illustrate structures in which the head noun phrase shares reference with an argument which functions as an adjunct of some kind (locative, temporal, instrumental, etc.) in the relative clause. (12) Locative a. ndeme e-túfíe hɔ-ɔ́ ma hɔ́-ɔhɔ baaba Ndeme 3sgsbj-1sgobj-show 19-place 19-re 1a.father a-mba-ɛ́ da-guúle hɔ́ ɔ 3sg-past3-go-come there “Ndeme shows me the place from which my father came” b. ndeme e-ngúlíe naá ma-hana má-ama Ndeme 3sgsbj-1sgobjget out prep 6a-misery 6a-re n-dɔ́-mɔ́ naá máa (má-ama) 1sgsbj-be-m prep 6a-rpr (6a-dem) “Ndeme takes me out of/spares me the misery in which I find myself” (13) Temporal ndeme a-bá-ága bi-binyó bu-dúgú bɔ́ ɔbɔ/ naá Ndeme 3sgsbj-ref-put 8-glove 14-day 14-re/ prep gɛ-camɛna gɛ́-ɛgɛ a-bádáfa-mɔ́ ɔ-ndɔ 7-time 7-re 3sgsbj-touch-m 3-drug “Ndeme puts on gloves (the day/at the time) when he handles chemicals” (14) Other structures with adjuncts a. mɔ-ɔtɔ a-háanana gɔ-yɔgɔnɔ naá hɔ-ɔ́ ma hu-ímée 1-person 3sgsbj-must 15-work prep 19-place 19-all hɔ́-ɔhɔ ba-gɔ́ nɔ́ be-mu-lukemen-i-e-mú hɔ́-ɔ 19-re 2-big/old 3plsbj-3sgobj-sit-caus-fv-m there “Any human being must work in the place where her elders place her” b. alámá e-fúmbe mbámba yɛ́-ɛyɛ gɔ-ɔya-mɔ́ gɔamɔ́ Aláma 3sg-look for 9/10.gun 9-re 2sgsbj-say-m that (2sgsbj-say) ɔ-mbɔ-ɔná ncúbé na yɛ 2sgsbj-past3-kill 9/10.hippo prep 9.rpr “Aláma is looking for the gun with which you say that you killed a hippo” c. mɔ-ɔtɔ mɔ́-ɔnyɔ pege-mú cɔá ndeme 1-person 1-re 1sgsbjthink-m that (1sgsbj- say) ndeme a-mbe-bíné nɛ-hɛ́ e-bé-m-bédigie 3sgsbj-past3-dance prep-3sgrpr 3sgsbj-past1-1sgobj-greet “The person I think Ndeme danced with has greeted me”



4 Relative clauses and relativization processes in Nugunu 

 75

d. ndeme a-nɔ-anan-a na mɔ-ɔtɔ unk-imé mɔ́-ɔnyɔ ndeme 3sgsbj-fight-it-fv prep 1-person 1-all 1-re e-dígie-mú nɛ hɛ́ 3subj-quarrel-m prep 3sg-rpr “Ndeme fights with any person with which (s)he quarrels” The relativizing element (RE) is similar with the singular and plural forms of proximity demonstrative and, moreover, it does not vary, independently of its function within the relative clause. With respect to the resumptive pronouns, intended as pronominal elements which appear in relativized positions (Ngonyani 2006), they are morphologically identical to object pronouns only when they appear in a verb form (cf. (9b, d) and (10)). In the remaining cases, resumptive information is encoded by means of either possessive adjectives (with a possessor head), personal pronoun forms used only with prepositions, or an adverbial such as hɔ́ ɔ which functions as a ‘locative’ resumptive. The data above show, furthermore, that whereas resumptive pronouns in Nugunu do not occur with subject heads, they are obligatory in any relative clause whose head is a [+human] noun functioning as a direct object (9c) and with any noun functioning as an indirect object (10). In the case of oblique heads, temporal head nouns do not require a resumptive pronoun (cf. (13)), whereas in the remaining cases the presence or absence of such a pronoun depends on the syntactic behaviour of the verb of the relative clause as well as on the nature of the head noun (cf. (14b, c, d)). With locative heads, the factors which are more relevant have to do with the semantics of the verbs and their complements. Verbs such as guúlie “remove”, guúle “come”, gulukumenie “make sit”, which express movement, require the presence of resumptive hɔ́ ɔ when the complement is hɔɔ́ ma “place”, in contrast to other complements with which the preposition naá and ‘non-locative’ resumptive pronouns are preferred (cf. (12b)). With other verbs such as gɔgɛ́ ga “stay”, gubébiéne “be born”, etc., which are less movement-oriented, no resumptive pronoun is required when the complement is hɔɔ́ ma “place”, whereas with other complements resumptive information is expressed exclusively by means of hɔ́ ɔ. As can be seen, although relativization is more common with subject and object head nouns, it is also possible with oblique and possessor head nouns. The differences between bound relative clauses have thus to do with the presence of a resumptive pronoun, which depends on the function of the head noun and, to a lesser extent, on the nature of the head noun and the semantics of the verb, as well as with the form of the resumptive pronoun, which may be prepositional or not, in accordance with the subcategorization of the verb. Resumption may also

76 

 Théophile Ambadiang

be effected by means of a possessive determiner (with possessor head nouns) and even an adverbial element (with locatives). In a sense, the use of resumptive pronouns correlates to some extent with the accessibility hierarchy on relative clause formation (Keenan and Comrie 1977). Whereas the use of resumptive pronouns is required with indirect objects as well as adjuncts and possessor phrases, in the case of more accessible noun phrases such as direct object noun phrases, either gaps or resumptive pronouns may be used depending on factors such as the nature of the arguments involved. This is discussed in section 3.

2.2.2 Headless or free relatives The main characteristic of headless relatives has to do with the fact that they lack an antecedent and, by way of consequence, a pronoun initiates the clause in the case of subject clauses (15a), or a verb initiates the clause and the utterance in the remaining cases (15b, c, d, e). (15) a. Subject mɔ́-ɔnyɔ a-lɔ́-mɔ́ na e-dúu ɔ-ɔ́ bɔ 1-re 3sgsbj-be-m prep 6-ear 3sgsbj-hear “Those who have ears must listen” (Lit. The one who has ears hears) b. Direct object ŋg-íyimene bɔ́-ɔbɔ a-lɛ́ galɛga-mɔ́ 1sgsbj-know 14-re 3sgsbj-want-m “I know what (s)he wants” c. Indirect object ŋg-íyimen-e mɔ́-ɔnyɔ ɔ-bá-ná-mɔ-fá mu-inyí 1sgsbj-know-fv 1-re 2sgsbj-past3-m-3sgrpr-give 6a-money “I know (the one) to whom you have given money” d. Prepositional ŋg-íyimen-e mɔ́-ɔnyɔ u-mbe-é -díéné nɛ hɛ́ 1sgsbj-know-fv 1-re 2sgsbj-past3-m-quarrel prep 3sgrpr “I know (the one) with whom you had an argument” e. Possessive ŋg-íyimene mɔ́-ɔnyɔ u-mbe-é -díéné na 1sgsbj-know 1-re 2sgsbj-past3-m-quarrel prep isé bɛ-hɛ́ 1a.father 3sgrpr-his/her “I know the one with whose father you had an argument” As is the case with bound relatives, resumptive pronouns are compulsory only with indirect object, oblique and possessor head nouns.



4 Relative clauses and relativization processes in Nugunu 

 77

2.2.3 The morphosyntax of bound relatives Even though the data adduced in this Section are consistent with the structure describing relative clauses in Nugunu in (1), they suggest that the description is in need of some more precision, concerning particularly the nature of the subordinating element and the configuration of the relative clause. As already noted, the similarity relationship in which the subordinating element is involved only holds for the proximity demonstrative. Moreover, it does not imply functional coincidence or equivalence. The invariability of the subordinating element regarding deixis suggests that this item, originally a demonstrative (Dixon (2010: 364), Keenan (1985: 149)), has undergone a process of grammaticalization in which its deictic function has been lost. On the other hand, the subordinating element may appear in a clause together with a genuine demonstrative which varies not only along the noun class dimension but also along the deictic dimension. The two or three demonstrative forms which correspond to each noun class (e.g. mɔ́ ɔnyɔ, múunú for noun class 1 and báaba, bɔ́ ɔbɔ, béení for noun class 2) may appear at the beginning or (preferably) at the end of the noun phrase and even, though less commonly, in both positions, as in (16a). Whereas the relativizing element follows the head noun phrase and lacks the deictic function, the leftmost position in a noun phrase may only have a deictic function, which is typical of demonstrative pronouns (15a) and adjectives (16a, b). (16) a. bá-aba ba-ábayɔ́ bá-dadɔ́ bá-aba bé-di-íyíméné táŋɔ 2-dem 2-child 2-three 2-dem 3plsubj-neg-know 9/10.calculus “These three children (here) do not know calculus” b. bá-aba ba-ábayɔ́ bá-dadɔ́ bá-aba bí-íyímene-mú táŋɔ 2-three 2-re 3plsubj-know-m 9/10.calculus 2-dem 2-child “These three children who know calculus” One consequence of the idea that the subordinating element is functionally different from demonstratives is that it may not be considered a determiner, i.e. as belonging to the noun phrase headed by the antecedent of the relative clause, in spite of the fact that its position coincides with the one which is typical of demonstratives. This allows not only to characterize it as a pronoun, but also to include it in the relative clause in which, moreover, it has a specific function. Its role as a pivot between the main clause and the relative clause accounts for its position at the left periphery of the latter (de Vries 2002). Furthermore, its morphosyntactic properties, i.e. the fact that it conveys only noun-class related information, account for the way it combines with resumptive pronouns, as the function of the latter inside the verb form or verb phrase is to indicate the case and (redundantly) the noun class of the head nouns with which they are co-referential, as will be shown in the next Section.

78 

 Théophile Ambadiang

Lastly, with respect to the verb, we have already noted that relative clauses require marked verb forms. This, however, does not mean that only such forms may appear in relative clauses nor that they are unique to them. More specifically, as shown in (14b, c), the verb immediately following the relativizing element is the only one which is marked.

3 Relativization processes in Nugunu As observed in Section 2, even though many elements are involved in the formation of relative clauses in Nugunu, none of them is exclusive to such constructions. From this point of view, what may be considered as specific to relative clauses has more to do with the combination of these elements than with (the effects of) each of them taken separately, for any of them may appear in diverse structures which share many of the functional features observed with relative clauses (Nikolaeva 2006; Biloa 1990; Kuteva and Comrie 2006; Aboh, Hartmann, and Zimmermann 2007). In 3.1. we will consider the formal elements whose combination makes a given configuration a relative clause, in contrast to other constructions related to processes such as focus and topicalization, whereas in the remainder of the Section we will look at strategies related to the recoverability of grammatical functions (3.2.), and the prosodic patterns observed in relative clauses (3.3.).

3.1 The specificity of the relativization processes As described above, bound relative clauses include a subordinating element which shares its reference with an antecedent noun phrase, and a verb form which presents certain morpho-tonological features typically different from those observed in ordinary descriptive sentences or assertions (Patman 1995; Keenan 1985; Dixon 2010). Thus, the left periphery of a relative clause is occupied by a relativizing element whereas a genuine demonstrative may appear at its right edge. That the function of this element may not dissociate from the marked verb forms alluded to above is suggested by the fact that it has a demonstrative function and meaning in case it is used with an ordinary, non-marked verb, as evidenced in the contrast between (17a) and (17b), only the latter of which may show a demonstrative at its right edge. (17) a. mɔ-ɔtɔ mɔ́-ɔnyɔ a-dɔ́ mbana kanɛ gu-(cú)-bédigie 1-person 1-this 3sgsbj-pass by prep 15-(1plobj)-greet “This individual passes by without greeting (us)”



4 Relative clauses and relativization processes in Nugunu 

 79

b. mɔ-ɔtɔ mɔ́-ɔnyɔ a-dɔ́ mbana-mɔ́ kanɛ gu-(cú)-bédigie 1-person 1-this 3sgsbj-pass by-m prep 15-(1plobj)-greet “The individual who passes by without greeting (us)” We have also noted that whereas the relativizing element varies in accordance with the noun class of the head noun, it may appear only in the form of the proximity demonstrative, in contrast to genuine demonstratives which have at least two forms in each noun class (for instance, má-ama, mɔ́-ɔmɔ, mé-ení for nouns of class 6a). Crucially, the form of the verb, as described in (4), determines the interpretation and, correlatively, the configuration of the sentence, in the sense that the parsing of the demonstrative together with the head noun (demonstrative interpretation), or separately from it (relative interpretation) depends on it. From a functional point of view, although these items serve to identify or describe the referent of the head noun phrase, in (17a) an individual is identified by means of the demonstrative determiner, and the fact that (s)he does not greet is mere additional information, whereas in a structure with -mɔ́ such as (17b) the relative clause serves to specify the individual by underlining that (s)he characteristically passes by without greeting. In this sense, it gives prominence to a given individual or to a feature that is typical of such a person in a fashion rather similar to what occurs with cleft constructions and, more generally with focus, as briefly discussed below. Focus is typically, though not exclusively, associated with cleft constructions, which consist in Nugunu in an extraposed noun phrase followed by a clause with a marked verb form, besides an element (generally a personal pronoun form) which encodes focus (cf. (20)). The data in (18), using a verb such as gɔ-ɔba “to fall”, show that different arguments of the verb, and even adjuncts, may be extraposed. The moved constituent appears on the left and the pronominal element which encodes focus (e.g. mɔ́ ɔ and bɔ́ ɔ in (18a, b)) is optional except when focus concerns the verb, in which case it is absent, as in egediegediemú ngáláfá “(s)he makes chairs” in contrast to ngáláfá (sɛ́ ɛ) egediegediemú “Chairs are what (s)he makes”. (18) Cleft constructions a. Subject mɔ-ádɔmbɔ́ (mɔ́ ɔ) ɔ-ɔba-mɔ́ 1sg-boy (1tpz) 3SG-fall-fv-m “It is a child who has fallen” b. Direct Object bɔ-ɛ́ tɛ (bɔ́ ɔ) di-ub-i-e-mú 1sg-boy (14tpz) 1subjpl-fall-caus-fv-m “It is a tree (that) we have felled”

80 



 Théophile Ambadiang c. Indirect Object mɔ-ádɔmbɔ́ (mɔ́ ɔ) di-mu-ub- in-i-e-mú i-bíle 1sg-boy (1tpz) 1pl-3objsg-fall-appl-caus-fv-m 5-palm-tree “It is a child we have felled a palm-tree for” d. Adjuncts mbámba (yɛ́ ɛ) di-mu-ub-i-e-mú na yɛ 9/10.gun (9tpz) 1pl-3objsg-fall-caus-m prep 9.rpr “It is a gun we have made him/her fall with” naá gɛ-bala (hɔ́ ɔ) a-bá-ná-ɔ́ bá prep 7-road (there) 3sg-past1-m-fall “It is on the road that (s)he has fallen” na bu-dúgú (gɔ́ ɔ) a-bá-ná-ɔbá (then) 3sg-past1-m-fall prep 7-night “It is in the night that (s)he has fallen”

As (19) shows, the marked verb forms also appear in other structures (adverbial clauses, WH-questions, and copulative clauses) in which they may co-occur with adverbs, as in (19a). (19) Other uses of marked verb forms a. Adverbial clauses anáa ndeme e-bé-né-yímé hɔ́ ɔ dɛ-ga-ná-ɛ́ dá adv ndeme 3sgsbj-m-get lost there 1plsj-fut2-m-go “(The place) Where Ndeme got lost, there we will go” aŋa ndeme a- bá-ná-bɔlá gɔ́ ɔ dɛ- bá-ná-ɛ́ dá adv ndeme 3sgsbj-pst1-m-arrive then 1plsbj-pst1-m-go naá i-sekúle prep 19-school “As/when Ndeme has arrived, (then) we have gone to school” b. WH-questions nkádɛ́ (gɛ́ ɛ) a-námba-mɔ́? What (7.foc) 3sgsbj-cook.prs-m “What does (s)he cook?” áhanɛ́ /nɛɛ́ nɛ/niikí a-bá-ná-bɔlá? Where/how/when 3sgsbj-pst1-m-arrive “Where/how/when has (s)he arrived?” c. Copulative clauses a- lɔ́-mɔ́ u-gúle ba-mɛ 3sgsbj-be-m 1-friend 1.con-my “(S)he is my friend”



4 Relative clauses and relativization processes in Nugunu 

 81

Relative clauses and the other types of clause considered above show many structural similarities. Whereas marked verb forms are common to all of them, the presence of a subordinating element (a proximity demonstrative), a possessive adjective or an adverb and, depending on the syntax and semantics of the verb, of a resumptive pronoun is typical of relative clauses. The remaining structures require an interrogative pronoun, as well as a pronominal form, typically the pronoun corresponding to the third singular or third plural person (optional in cleft constructions and with direct objects in WH-questions as in (19b)), which encodes focus. According to Patman (1995), the similarities observed between these syntactic structures result from the fact that, from a functional point of view, they are based on a “common cognitive strategy for accessing, or activating, concepts and relating them to the context shared by speaker and hearer”. According to our discussion, focus phenomena appear to be relevant in such a strategy (cf. also Simango 2006, as well as Aboh, Hartmann, and Zimmermann 2007). Summing up, since relative clauses and other constructions mainly associated with extra-position processes require marked verb forms, the differences concerning all other aspects of their structures suggest that the specificity of the relativization processes may not dissociate from the presence in a clause of a subordinating element identical to one of the demonstrative pronouns and co-referential with a noun phrase head, as well as, in the case of indirect object, possessive, adjunct or [+human] direct object head nouns, a resumptive pronoun.

3.2 Grammatical function recoverability Even though relative clauses show rather consistent functional similarities in different languages, this is not necessarily the case with the strategies the latter have recourse to in order to indicate grammatical functions in the relative clause. Nikolaeva (2006) identifies four syntactic processes which are characteristic of such strategies: use of a relative pronoun, non-reduction, pronoun retention, and gapping (cf. also Kuteva and Comrie 2006). The first strategy, which implies the presence of a genuine relativizer, is of no use in Nugunu as the data above show. Of the remaining three strategies, pronoun retention involves the use of a personal pronoun which is co-referential with the head noun and encodes morphosyntactic features of the corresponding argument in the relative clause, i.e. a resumptive pronoun. Non-reduction implies the presence of a full noun phrase in the relative clause, whereas in case of gapping there is no overt encoding of morphosyntactic features within the relative clause. Inasmuch as in Nugunu, as in Bantu languages more generally, the presence of some ­relativizing element(s)

82 

 Théophile Ambadiang

is obligatory in any finite relative clause, the focus of our concern here will be on the relationship between such element(s) and the strategies alluded to above, and more specifically, the latter two, since Bantu languages do not have recourse to full noun phrases in relative clause formation, besides lacking genuine relative pronouns. According to the observations adduced in the previous Section, Nugunu relative clauses have recourse both to gapping and resumptive pronouns. The latter are used with indirect object, [+human] direct object, oblique and possession head nouns, and the former in the remaining cases ([-human] direct object noun heads), and come in different forms depending on whether they appear within a verb form or in a preposition phrase. Note, however, that the differences between all these strategies are not easy to draw because of the complex morphology of the Bantu verb paradigms, as well as the morphological indistinctness of direct and indirect objects shown in (20). We will begin with a brief look at the personal pronoun system typical of noun class 1/2 which, because of its complexity -due to formal variation related to features such as person, number and function-, is in sharp contrast with those of the other noun classes, which have at most three different forms (one of them for relativized positions associated with the functions of subject and object). (20) Gunu personal pronouns (noun class 1/2) Full forms Subject Object Prepositional 1sg yɔmɔ́ ɔ n-nmɛ 2sg mamɔ́ ɔ ɔ-gɔhɔ 3sg mɔ́ ɔ a-mɔhɛ 1pl bicóo dɛ-cɔ́có 2pl binyóo nɔ-nyɔ́nyó 3pl bɔ́ ɔ bá-bɔ́bɔ́

Focus mɔ́ ɔ mɔ́ ɔ mɔ́ ɔ báa báa báa

The use of full forms tends to be limited to distributive and emphatic contexts, where marked verb forms are also required, and their presence disallows that of either a relativizing element or a resumptive pronoun. Direct and indirect objects share the same markers which, moreover, are also used as resumptive pronouns in relative constructions. Lastly, independently of their function, prepositional pronouns are obligatory and may not be separated from their head. Processes as those involved in relativization in Bantu languages can be accounted for in different ways. For instance, the position of an object head to the left of the relative clause has received explanations of either a semantic (discursive) or a syntactic nature. Thus, Mchombo (2004: 44) considers the Chichewa noun phrase “in anaphoric agreement with the object marker as a discourselicensed-non-argument phrase [whose] structural position is not determined by



4 Relative clauses and relativization processes in Nugunu 

 83

rules of grammar but by discourse structure”. By way of consequence, relative clause formation in Chichewa does not involve movement operations, and “the object marker functions like a pronominal variable bound by the relative operator”. Proposals of this type are questioned in Ngonyani (2006) on the basis of several structural phenomena, such as dislocation, that we will not discuss here. According to Ngonyani, the pronoun and the head noun phrase are generated inside the relative clause, and resumptive constructions involve movement of the head noun phrase (raising to the specifier of CP) as well as, eventually, the stranding of the pronoun in structures he interprets as unbounded (cf. also Biloa 1990, 1997 and, for further discussion, Aboh, Hartmann, and Zimmermann 2007). Although Mchombo and Ngonyani focus on different structures related, respectively, to objects and adjuncts, for the generalization proposed by Mchombo to be maintained, it seems necessary to draw a distinction between relative clauses and other structures resulting from processes such as focus or cleft construction, whose functional and structural similarities with the former, underlined by Patman (1995), Ngonyani (2006), as well as in Aboh, Hartmann, and Zimmermann (2007), have precisely the effect of questioning such a differentiation. Syntactic asymmetries such as those observed in (21) lend support to the idea that some kind of movement is required with adjuncts. (21) a. gɛ-dɔŋɔ/ n-nyaŋá gɛ́-ɛgɛ/yɛ́-ɛyɛ e-mbeé-bé-biéné hɔ́ ɔ 7-country/9-house 7-re/9-re 3sgsbj-pst3-ref-bear there “The country/house in which (s)he was born” b. naá gɛ-dɔŋɔ//nnyaŋá gɛ́-ɛgɛ/yɛ́-ɛyɛ e-mbe-é-bé-biéné prep 7-country//9/10-house 7-re/9-re 3sgsbj-pst3-m-ref-bear “In the country/the house in which (s)he was born” c. mbámba yɛ́-ɛyɛ bá-bá-ná-ɔ́ ná ncúbé na yɛ 9/10.rifle 9-re 3plsbj-past1-m-kill 9/10.hippo prep rpr “The gun with which they have killed the hippo” d. na mbámba yɛ́-ɛyɛ bá-bá-ná-ɔ́ ná ncúbé prep 9/10.rifle 9-re 3plsbj-past1-m-kill 9/10.hippo “With the gun (with which) they have killed the hippo” The presence of the prepositional phrase na yɛ, obligatory in (21c), would be a source of ungrammaticality in (21d). Insofar as there is a correlation between this fact and the structural differences that exist between these sentences, syntax is involved in the presence of a resumptive pronoun more than suggested by discourse-based analyses. Thus, within the phrase headed by the verb bábánáɔ́ ná the preposition and the adjunct form a constituent whose components move together in (21d) but not in (21c), in which a stranded preposition is followed by a pronoun which typically has resumptive functions, as is also the case with more

84 

 Théophile Ambadiang

adverbial elements, such as hɔ́ ɔ “there” or gɔ́ ɔ “then”, whose use may have a semantic motivation (cf. (18d)). More generally, adjunct as well as indirect object head nouns and [+human] direct object head nouns also require the presence of a resumptive pronoun in the relative clause, although only the markers of the latter may be part of a verb form (cf. (9), (10)). The last observation suggests, however, that syntax alone does not account for all the cases in which a resumptive pronoun appears in a relative clause. As already noted, semantic features such as [+human] are necessary in order to account for the differences observed among direct object head nouns. Likewise, variation that obtains with relative clauses headed by space expressions and noun phrases correlates with the semantics of the verbs involved. As may be seen in (22), with head nouns such as hɔɔ́ ma “place” a resumptive element, such as hɔ́ ɔ “there”, is required in the relative clause with verbs which express movement, as is the case in (22a) in contrast to (22b). Head nouns with more specific spatial reference such as gɛdɔŋɔ “country” and nnyaŋá “house” require resumptive elements more in accordance with their respective noun classes than hɔ́ ɔ “there”, although the latter may also be used in this context due to its semantic content (cf. (12c)). (22) a. hɔ-ɔ́ ma hɔ́-ɔhɔ a-mbɛ-ɛ́ -ɛ́ dá-guúle/bɔla hɔ́ ɔ 19-place 19-re 3sgsbj-past3-m-go-come/arrive there “The place where (s)he came from/(s)he arrived at” b. hɔ-ɔ́ ma hɔ́-ɔhɔ e-mbe-é-bé-biéné / a-gɛ́ ga-mɔ́ 19-place 19-re 3sgsbj-past3-m-refl-give birth / 3sgsbj-stay-m “The place where (s)he was born/(s)he lives” Lastly, relative clauses headed by time expressions and noun phrases characteristically do not require resumptive information, as shown in (13).

3.3 Some observations on the prosody of Gunu relative clauses The discussion above has shown that a particularity of relative clauses in Nugunu, whether restrictive on not, has to do with the morpho-tonology of the finite verb. However, these two types of finite relatives show differences, very briefly discussed here, which have to do with the kind of information they carry and the way they do it. One essential function of restrictive clauses is to identify or to give prominence to a participant. The most common structural effects show in a symmetrical patterning at the edges of the relative clause: there is no pause either between the head noun and the relativizing element or between the last constituent of the relative clause and the elements further to its right, though in case of asymmetry only the right edge of the relative clause may require a pause. By way



4 Relative clauses and relativization processes in Nugunu 

 85

of contrast, non-restrictive relative clauses, which do not specify the participant focused on, tend to require a pause on both edges. Pending a more systematic study of the different types of relative clause, the observations above suggest that the restrictive relative clause and its antecedent are part of the same intonation unit, whereas non-restrictive clauses form such a unit on their own, as long as it is acknowledged that from a phonetic point of view edge effects differ from those which are typically related to complete utterances (see Dixon (2010: 353) and Makasso (2010), among others).

4 Final remarks The aim of this paper has been to describe constructions characterized as relative clauses in Nugunu. After a brief presentation of the Nugunu language, we have discussed the characteristics of these structures as well as the elements and processes involved in them. We have seen that, even though these elements (marked verb forms, relativizing pronouns, resumptive reference) are not necessarily exclusive to relative clauses, their combination is what makes a sentence a relative clause in Nugunu. The examples adduced in Section 2 aim at illustrating the structural patterns of such constructions, focusing on finite relative clauses, whereas the study of the relativization processes in Section 3 is concerned with the way finite relative clauses are formed in Nugunu, the strategies necessary to indicate grammatical functions within relative clauses, as well as the intonational patterns typical of such structures. It has been shown that relativization characteristically involves some kind of movement of the object or adjunct head noun phrase from a position within the relative clause, as well as the presence of a marked verb form and a resumptive pronoun. With respect to the form of resumptive pronouns, we have observed that they do not differ from object pronouns, in contrast to prepositional resumptives. Their distribution has been accounted for on the basis of syntactic factors such as the function of the head noun (only with adjuncts and indirect objects is a resumptive pronoun required), semantic factors related to nouns ([+human] direct object nouns require a resumptive pronoun, as is also the case with nouns with a specific reference to ‘place’, in contrast to nouns with temporal reference) or to verbs (verbs which do not express movement do not require a resumptive pronoun in presence of nouns with general spatial reference). Also, in the last Section, the prosodic properties that characterize relative clauses have been briefly considered. In this respect, it has been suggested that a more detailed study of the prosody and intonation of relative clauses in Nugunu

86 

 Théophile Ambadiang

may improve our understanding of these constructions, as well as the description of more complex structures, such as reciprocal, multiply embedded relative clauses and even WH-questions and cleft constructions, which have been dealt with only in passing in this study.

References Aboh, Enoch O, Katharina Hartmann & Malte Zimmermann. 2007. Focus and grammar: The contribution of African languages. In Enoch O. Aboh, Katharina Hartmann & Malte Zimmermann (eds.), Focus strategies in African languages: The interaction of focus and grammar in Niger-Congo and Afro-Asiatic, 1–12. Berlin: Mouton de Gruyter. Adamou, Evangelina & Denis Castaouec. 2010. Connective constructions in the world’s languages: A functionalist approach. La Linguistique 46(1). 43–80. Ambadiang, Théophile. 1991. La oposición flexión-derivación en la morfología del nombre bantú, con especial referencia al Nugunu (bantú A62) [The opposition inflection-derivation in the morphology of Bantu nouns, with special reference to Nugunu (Bantu A62)]. Madrid: Universidad Complutense de Madrid dissertation. Biloa, Edmond. 1990. Resumptive pronouns in Tuki. Studies in African Linguistics 21(2). 211–236. Biloa, Edmond. 1997. Functional categories and the syntax of focus in Tuki. Munich: Lincom Europa. Dieu, Michel & Patrick Renaud (eds.). 1983. Atlas linguistique du Cameroun. Paris: Agence de Coopération Culturelle et Technique (ACCT), Yaoundé: Centre Régional de Recherche et de Documentation sur les Traditions Orales et pour le Développement des Langues Africaines (CERDOTOLA). Dixon, Robert M.W. 2010. Basic linguistic theory. Volume II: Grammatical topics. Oxford: Oxford University Press. Du Plessis, Jacobus A. 2010. The relative clause in the African languages of South African (Bantu) languages. http://scholar.sun.ac.za/handle/10019.1/4884 (accessed 07 February 2012). Guthrie, Malcolm. 1953. The Bantu languages of Western Equatorial Africa. Oxford: Oxford University Press. Keenan, Edward L. 1985. Relative clauses. In Timothy Shopen (ed.), Language typology and syntactic description. Volume II: Complex constructions, 141–170. Cambridge: Cambridge University Press. Keenan, Edward L & Bernard Comrie. 1977. Noun phrase accessibility and universal grammar. Linguistic Inquiry 8. 63–99. Kuteva, Tania & Bernard Comrie. 2006. The typology of relative clause formation in African languages. In F. K. Erhard Voeltz (ed.), Studies in African linguistic typology, 209–228. Amsterdam/Philadelphia: John Benjamins. Lewis, M. Paul, Gary F. Simons & Charles D. Fennig (eds.). 2016. Ethnologue: Languages of the world, 19th edn. Dallas, Texas: SIL International. Online version: http://www.ethnologue. com (accessed 17 December 2016).



4 Relative clauses and relativization processes in Nugunu 

 87

Makasso, Emmanuel-Moselly. 2010. Processus de relativisation en bàsàa : De la syntaxe à la prosodie. In Laura Downing, Annie Rialland, Jean-Marc Beltzung, Sophie Manus, Cédric Patin & Kristina Riedel (eds.), Papers from the Workshop on Bantu Relative Clauses (ZAS Papers in Linguistics 53), 145–158. Berlin: Zentrum für Allgemeine Sprachwissenschaft. Mchombo, Sam A. 2004. The syntax of Chichewa. Cambridge: Cambridge University Press. Ngonyani, Deo S. 2006. Resumptive pronominal clitics in Bantu languages. In Olaoba F. Arasanyin & Michael A. Pemberton (eds.), Selected Proceedings of the 36th Annual Conference on African Linguistics, 51–59. Somerville, CA: Cascadilla Proceedings Project. Nikolaeva, Irina. 2006. Relative clauses in spoken discourse. In Keith Brown (ed.), Encyclopedia of language and linguistics, Volume VII, 501–508. North Holland: Elsevier. Patman, Frankie. 1991. Tone in the Nugunu verb phrase. In Stephen C. Anderson (ed.), Tone in five languages of Cameroon, 73–90. Arlington: Summer Institute of Linguistics & The University of Texas. Patman, Frankie. 1995. A functional account of syntactic similarities in Nugunu. Yaounde: University of Yaounde MA thesis. Paulian, Christiane. 1986. Les parlers yambassa du Cameroun (bantou A62). In Gladys Guarisma & Wilhelm J. Möhlig (eds.), La méthode dialectométrique appliquée aux langues africaines, 243–279. Berlin: Reimer Verlag. Riedel, Kristina. 2010. Relative clauses in Haya. In Laura Downing, Annie Rialland, Jean-Marc Beltzung, Sophie Manus, Cédric Patin & Kristina Riedel (eds.), Papers from the Workshop on Bantu Relative Clauses (ZAS Papers in Linguistics 53). 211–225. Berlin: Zentrum für Allgemeine Sprachwissenschaft. Robinson, Clinton. 1983. Phonologie du gunu, parler yambassa. Paris: Société d’Études Linguistiques et Anthropologiques de France. Simango, S. Ron. 2006. Verb agreement and the syntax of ciNsenga relative clauses. Southern African Linguistics and Applied Language Studies 24 (262). 277–290. Vries (de), Mark. 2002. The syntax of relativization. Utrecht: Landelijke Onderzoekschool Taalwetenschap dissertation. Zeller, Jochen. 2004. Relative clause formation in the Bantu languages of South Africa. Southern African Linguistics and Applied Language Studies 22 (1-2). 75–93.

Florence A. E. Tabe and Gratien G. Atindogbé

5 Kenyang relative clauses

Abstract: In Kenyang (a Niger-Congo language spoken in Cameroon), relative clauses can be realized structurally or phonologically. Structurally, the unmarked relative clause position is DP REL. The reversed order REL DP is derived by ­focalization. The relative morphemes nɛ̀ and bɛ́ modify DPs. The former is used to relativize DPs in finite clauses while the latter relativizes DPs in ­infinitival clauses. In finite clauses, the relative morpheme has the structure V-CV or V-CV-CV depending on whether the relativized DP is singular or plural. V is the augment which ­co-varies with the phi-features of DPs. Agreement between the DP and the ­relativizer is local in a strictly left - right configuration. In the absence of these morphemes, relativization is marked through aspect and this involves modifying the aspectual tones on the subject markers. This paper i­ dentifies and describes the different relativization strategies in Kenyang as well as the ­relativized ­positions and their local dependencies.

1 Introduction A relative clause is a subordinate clause which is dependent on a noun or pronoun and usually introduced by a relative pronoun or relative adverb to modify various elements in the clause. The relative clause may have an antecedent (that is, a head noun) or lack one as it is the case with headless or free relative clauses. There are cross-linguistic variations in relative clauses. One major typological distinction is in the order of the relative clause and the head noun. In many languages the relative clause appears externally to the head noun. In this strategy, the relative clause follows or precedes the head noun, as in (1-4) for postnominal relativization and (5–6) for prenominal relativization: (1) The man who robbed the bank (2) Wa mwîê rra [nrâ suveharru nrâ toni] nrâ truu numea The woman there 3sg like sm Toni 3sg stay Noumea “The woman [Toni likes ------]lives in Noumea”  Tinrin (Tallerman 1998: 84) Florence A. E. Tabe: University of Yaoundé I, [email protected] Gratien G. Atindogbé: University of Buea, [email protected] DOI 10.1515/9783110469547-005

90 

 Florence A. E. Tabe and Gratien G. Atindogbé

(3) Mo ri ọkunrin [ti John lu l’ ẹsẹ] I see man that John strike on leg “I saw the man [that John struck -----] on the leg”  Yoruba (Tallerman 1998: 84) (4) John mle aye l-ọ vbere John saw woman that-she sleep “John saw the woman who is sleeping” 

Urhobo (Keenan 1985: 152)

In some languages, the relative clause precedes rather than follows the head noun as illustrated in (5) and (6): (5) orixe dâ [txakurra ekarri dabe-n] mutilla that is dog bring aux-comp boy “That is the boy [who ------ brought the dog]”  Basque (Tallerman 1998: 84) (6) [buu-m maa-ča] etiken reindeer-obj kill-prf.participle old man “The old man [who ------- killed the reindeer]” 

Even (Tallerman 1998: 84)

Another possibility that has been identified and appears to be “much rarer crosslinguistically” following Tallerman (1998: 85) is for the head noun to appear inside the relative clause itself. Tallerman does not provide examples of such languages. This is however the case in Basaá, a Cameroonian language of Bantu A. The word order of Basaá, is SVO as shown in (7): (7) mɛ̀ bí kàl wɛ̂ mùt I pst tell you man “I told you of one man” 

wàdá one (Bitjaa Kody, p.c.)

The relative clause in Basaá, is introduced by a discontinuous morpheme í-nû. For example, when the subject, mùt ‘man’, of the preceding clause is relativized, the head noun appears inside the relative morpheme as indicated in the square brackets in (8): (8) [í-mùt-nû] mɛ̀ bí kàl rel-man-rel I pst tell “The man whom I told you of” 

wɛ̂ you (Bitjaa Kody, p.c.)



5 Kenyang relative clauses 

 91

There are languages with internally-headed relative clauses. Creissels (2000: 256) observes that languages of this type are quite difficult to come across. The characteristic property of internally-headed relative clauses is that the head noun appears inside the relative clause. (9) [n nǝ puhi saan-so [1sg comp greeted stranger-rel “The stranger who I greeted has gone” 

la] dem]

ʧaŋya has.gone Dagbani (Peterson 1974: 77)

(10) [‘ehatt gaat kw-akwii]=ve=ch nye-chunuw dog cat rel-sbj-chase] ]=defz=sbj I.obj-bite “The dog that chased the cat bit me”  Mesa (Couro and Langdon 1975: 186-187) A relativized strategy, which is similar and possibly considered a subtype of internally-headed relative clauses, is the correlative relative clause. Like the internally-headed relative clause, the head noun in correlative relative clauses appears inside the clause. They differ however from their internally-headed relative counterparts in that “the relative clause is outside the main clause and is connected anaphorically to a noun phrase in the main clause”. (11) a. [muso mìn tarara], o ye fìn san [woman rel leave], 3sg pst cloth buy “The woman who left bought the cloth”  (Bambara, Bird and Kante 1976: 9) jana y’ o faa b. cɛ ye mìsí min ye man TAM cow rel see lion TAM PRN kill “The lion killed the cow that the man saw”  Bambara (Creissels (2000) cited by Kuteva and Comrie 2006: 214) The other major cross-linguistic distinction in the typology of relative clauses has to do with whether or not there is a gap in the relativized position. In English it is standard to find a gap in a relativized position. It allows prepositions to be stranded (that is with no prepositional objects) at the end of the clause when the objects of the prepositions are relativized. The construction in (12a) and its relativized counterpart in (12b) serve to exemplify the fact: (12) a. Jerry travelled to the city in a black car b. The black car [that Jerry travelled to the city in ----]

92 

 Florence A. E. Tabe and Gratien G. Atindogbé

Contrary to what we find in English, many other languages require the relativized position to contain a pronoun or a full noun phrase. Consider the following examples: (13) wúqád [dà yá káshé tá dà knife rel he killed her with “The knife that he killed her with -----” 

ìtá] it Hausa (Tallerman 1998: 85)

í ɲūī (14) a. wū-ū bàà kì-tʃí he-pst cut 7-tree with cutlass “He cut the tree with a cutlass”  Lamnso’ (Atindogbé 2014, field work) b. ɲūī jē wū-ū bá-án kí-tʃí ʃô cutlass rel he-pst cut-rel 7-tree with.it “The cutlass that he cut the tree with”  Lamnso’ (Atindogbé 2014, field work) (15) a. wù-ù jùìrî wûn í ʃwàà he-pst kill him with knife “He killed him with a knife”  Lamnso’ (Atindogbé 2014, field work) b. ʃwàà wɔ́ wū-ū jùìrî wûn ʃô knife rel he-pst kill him with.it “The knife that he killed him with”  Lamnso’ (Atindogbé 2014, field work) In Hausa (cf. 13), the sentence cannot end with the instrumental ‘with’ like in English; the instrumental preposition must be followed by the pronoun [ítá] ‘it’. In Lamnso’, the morpheme [ʃô] stands for the instrumental preposition and the pronoun ‘it’ as illustrated in (14b) and (15b). When the sentence does not demand the pronoun ‘it’, the instrumental is the morpheme [í ] as shown in (14a) and (15a). In the generative framework, it has been assumed that relative constructions are derived from clauses embedded in matrix sentences having an equivalent noun phrase. For example: The man [S who won the best prize] is very tall. The processes involved in the derivation of relative constructions require that the equivalent NP in the embedded clause be deleted and replaced by a relative pronoun through a transformational rule EQUI (Chomsky 1981). If however, the embedded clause is attached to the object of the matrix clause, a relative pronoun is not mandatory as in John bought the overcoat he liked. Relative clauses share some properties with wh-clauses. Like wh-clauses, they have a constituent



5 Kenyang relative clauses 

 93

missing inside the clause and that is represented by a wh-constituent. A difference between them is that whereas wh-clauses appear in functions in which NPs could also appear (e.g. subject or direct object), relative clauses actually occur only within NPs (Collins and Hollo 2000). Depending on the semantic/pragmatic function, a distinction is usually made between defining/restrictive and amplifying/non-restrictive relative clauses. Restrictive relative clauses define and describe the noun modified by limiting the set of possible objects that the noun specified by the clause can refer to. For example, Jane married the man whom she was in love with (and none other than that one). Non-restrictive relative clauses, in contrast, amplify the character of their referent more exactly by modifying the entire clause with words or phrases to mirror its purely attributive character (e.g. Bob, who is a renowned artist, produced a thrilling performance). A distinction is made between the two sentences in intonation and also in punctuation. Thus, the restrictive relative clause is spoken only with one intonation contour, while the non-restrictive relative clause contains three intonation contours and is set off by commas (Batko 2004). The properties that mirror this distinction hold for English, not universally. In many languages, relative constructions, adjectives and genitives occupy the same syntactic position in relation to the noun modified. This shared relationship with the noun supports the view that attributive adjectives are derived from relative clauses containing predicate adjectives and genitives from attributive prepositional phrases. Thus sentence (16) is as possible as sentence (17). Sentences (18) and (19) show that attributive prepositional phrases can be derived in the same way. (16) The men who were most prominent in the club were elected (17) The men most prominent in the club were elected (18) The men in the club were elected (19) The leaders of the club were elected 

(Lehmann 1976: 195)

Each of the relative clauses presented in the different languages above has an overt antecedent. Many languages also have headless or free relative clauses. These relative clauses have no antecedents. (20) Whoever got involved in human trafficking was punished (21) What I love so much has been taken away (22) Who wins the race will become the next president

94 

 Florence A. E. Tabe and Gratien G. Atindogbé

Sometimes the relative morpheme is implied but not overtly expressed in the sentence. Such clauses are referred to as zero relatives or reduced relatives. They are denoted by ø. (23) a. Sam bought the car that I proposed to him b. Sam bought the car ø I proposed to him bá-nà bénà á tá-nɔ á wàsà (24) a. Elame à sɔ̀ -í Elame 3sg find-pst 2-child 2.rel 3sg pst-rpr 3sg look “Elame found the book he was looking for”  (Duala, Clédor Nseme, p.c.) bá-nà ø á tá-nɔ á wàsà b. Elame à sɔ̀ -í Elame 3sg find-pst 2-child ø 3sg pst-rpr 3sg look “Elame found the children he was looking for”  (Duala, Clédor Nseme, p.c.) The study of Kenyang relative clauses reveals that different relativization strategies are used in the language. These strategies are derived from a range of operations: from purely morphological, morphophonological to morphosyntactic. This paper examines each of these strategies. Focus will be on the description of restrictive relative clauses. The rest of the paper is structured into four sections. Section 2 presents an overview of aspects of Kenyang morphosyntax. Here the word order of Kenyang is defined and the unmarked structure of typical Kenyang clauses identified. The structure of nominal modifiers is also presented. The objective in this respect is to define the linear ordering of elements that co-occur with the noun phrase as well as their relative functions in the language. Section 3 examines the morphosyntax of relative clauses in the language. It looks at the relative syntactic positions of the relative marker with respect to the modified noun phrase. The agreement pattern in Kenyang relativized DPs is discussed in section 4. A conclusion to the paper is given in section 5.

2 An overview of Kenyang syntax and morphology Kenyang is a Niger-Congo language of the Nyang group spoken in the Southwest Region of Cameroon. The word order is SVO, as illustrated in the following sentences: (25) m̀-mɔ̀ à kwù 1-child 1.agr.prf buy “The child bought a book”

ɛ̀ -kátì 7-book

5 Kenyang relative clauses 



 95

(26) àʧùò à-á kó ɛ̀ -nɔ̀ k PN 3sg-iprf climb 7-tree “Achuo is climbing a tree” A simple Kenyang noun (i.e. the form used as the citation) comprises a root and a prefix: ŋ̀-gɔ̀ rɛ́ ‘wife’, bà-ɣɔ̀ rɛ́ ‘wives’, ɛ̀ -kɛ́ t ‘house’, bɛ̀ -kɛ́ t ‘houses’, etc. The (nominal) prefixes determine the singular/plural alternation of the nouns and the agreement system in the language as illustrated in the examples below: (27) a. ɛ̀ -kɛ́ t ɛ̀ 1 7-house 7.agr.cop “The house is big” b. bɛ̀ -kɛ́ t bɛ́

8-house

8.agr.cop

“The houses are big”

ʧèk big ʧèk

big

(28) a. ŋ̀-gɔ̀ rɛ́ ǹ-fɔ̀ 1-wife 1-chief “The chief’s wife” b. bà-ɣɔ̀ rɛ́ bà-fɔ̀ 2-wives 2-chiefs “The chiefs’ wives” Furthermore, in order to understand the structure of relative clauses in Kenyang, it is necessary to examine the different modifiers that co-occur with the noun. Simple nouns are used as the citation form, as predicate nominal and as DPs with varied interpretation including generic, indefinite, or definite depending on the environment in the clause. We are referring here to the various forms of adjectives comprising quantifiers, determiners, qualifiers, possessives among others which function in modifying the nominal element. The different forms of adjectives can either pre-modify or post-modify nominal expressions in the language. It is customary to find the same adjective functioning as a prenominal and post-nominal modifier in Kenyang. The characteristic features accounting for such derivational processes will not be addressed in this paper.

1 In Kenyang, the copula is generally the morpheme ʧí as in ɛ̀ kátì ɛ́ ʧí màndû ‘the book is small’. However, this copula can be dropped when adjectives like ʧèk ‘big’, sàp ‘long’, etc. are involved. Then, the agreement or the personal pronoun plays that role.

96 

 Florence A. E. Tabe and Gratien G. Atindogbé

Modification by quantifiers (29) a. bɛ̀ yà β-ó many 2-people “Many people” b. m̀bɔ̀ g bà-tɔ̀ ŋ some 2-teachers “Some teachers” c. m̀bɔ̀ g mɛ̀ -nyìŋ few 8-things “A few things” Modification by determiners à-nɛ̀ (30) a. ǹ-tà 9-cap 9.agr-this “This cap” b. ɛ̀ -kátì ɛ̀ -mɔ̀ t 7-book 7.agr-one “One book” c. m̀-mɔ̀ m̀bì̵ 1-child 1.agr.first “The first child” Modification by qualifiers (31) a. ǹ-dɛ̀ n ʧú 9-dress red “A red dress” b. ɛ̀ -kìßɛ̀ ɛ̀ ʧèk 7-suitcase 7.agr.cop big “The suitcase is big” à sàp c. ǹ-ní̵k 3-rope 3.agr.cop long “The rope is long” Modification by possessives/genitives (32) a. ɛ̀ -kɛ̀ t ɛ̀ -yâ2 7-house 7.agr-1sg.poss “My house”

2 The root for the possessive is -à, but glide /y/ or /w/ is inserted depending on the preceding vowel.

5 Kenyang relative clauses 



 97

b. ɛ̀ -kɛ̀ t ́ ákò 7-house am pn “Ako’s house” ákò c. ɛ̀ -kɛ̀ t ɛ̀ -nɛ̀ 7-house 7.agr-of pn “The house of Ako/Ako’s house” It is worth commenting on the genitive as a form of nominal modification in Kenyang. Genitive constructions can be derived phonologically as in (b) involving a floating tone or morphologically as in (c) using the particle nɛ̀ which is apparently equivalent to the Saxon genitive ‘of’. The morpheme nɛ̀ is found in noun-noun modification constructions in which there is no operator. The Kenyang particle nɛ̀ however has multifaceted functions in the language. These comprise coordination, preposition and a relativizer, the focus of this paper to be presented in section 3. Kenyang verbs are not inflected for temporal features. The time frame is expressed through aspect and by tone. The latter is hosted by the subject marker: (33) a. ákò à rɔ̀ ŋ ɛ̀ -kátì pn 3sg.prf go 7-school “Ako went to school” rɔ̄ ŋ ɛ̀ -kátì b. ákò à-á pn 3sg-iprf go 7-school “Ako is going to school” (34) a. bà-fɔ̀ bá rɔ̀ ŋ ɛ́ -syɛ́ 2-chiefs 2.agr.prf go 7-market “The chiefs went to the market” b. bà-fɔ̀ bá-á rɔ̄ ŋ ɛ́ -syɛ́ 2-chiefs 2.agr-iprf go 7-market “The chiefs are going to the market” bǎ ʧwè m̀-bòg c. békâ bà-ɣɔ̀ rɛ́ 2pl 2-women 2.agr.prf enter 9-hole “You women entered a hole” d. békâ bà-ɣɔ̀ rɛ́ bà-á ʧwè m̀-bòg 2pl 2-women 2.agr-iprf enter 9-hole “You women are going to enter a hole” Apart from being encoded phonologically (by tone) aspect is also lexicalized in Kenyang. The perfect/imperfect alternation is determined by tones: There are two level tones (high and low) two contour tones (rising and falling). The level tones mark clauses in the perfect, while the contour tones mark clauses in the imperfect.

98 

 Florence A. E. Tabe and Gratien G. Atindogbé

The use of tone serves to mark the instant perfect and imperfect only. The tones do not indicate the different time frames, for instance, past perfect, past imperfect, etc. The properties of the latter are captured and elucidated by adverbs and adverbial modifiers: (35) ákò à mày rɔ́ ŋ nɔ́ kɔ́ pn 3sg.prf asp.used go asp.often “Ako used to go often to the farm”

ɛ̀ -β ì̵ 7-farm

(36) àjók à kàísí nyāká bɛ́ pn 3sg.prf think asp.ant to “Ayuk had thought of working with us”

bɛ̀ -tèk 8-work

kî do

nɛ̀ with

bɛ̀ sɛ́ us

The relationship between the subject marker and the verb is not always local given the presence of intervening particles such as negation, temporal adverbs and modals: (37) ɛ̀ bɔ̀ b à pù kwày nyáká pn 3sg.prf neg ought asp.ant “Ebob could not have worked with us”

bɛ́ to

kì do

bɛ́ -tèk 8.work

nɛ̀ with

bɛ̀ sɛ́ us

Given the clause structures in (35, 36, 37), and by the logic of Baker’s (1985) Mirror Principle (a generalization that expresses that the order of morphemes mirrors the order of functional projections), a proposed verbal accessibility hierarchy of Kenyang should correspond to the structure in (38): (38) 1 sm

2 neg

3 mod

4 asp

5 v

6 asp

Thus, (38) shows the supposed linear ordering of verbal morphemes in Kenyang. As an SVO language, the subject is followed by the verb and the verb by an object or complement. Between the subject and the verb is a subject marker which hosts features of the preceding subject (number, noun class and person). There are a series of other morphemes associated to the verb. Some of these may precede the verb, while some others may follow. Before the verb, the order is always a subject marker (sm) >negation (NEG)>mood (MOD)>aspect (ASP). There is only one morpheme that follows the verb: aspect. Aspectual morphemes can occupy two slots (preverbal and postverbal). Having presented aspects of Kenyang syntax and morphology above, we shall proceed with the examination of the structure of relative clauses in the language.



5 Kenyang relative clauses 

 99

3 Kenyang relative clauses The presentation in section 2 shows that a simple noun is, in fact, a complex structure with several overt morphemes. Relative clauses, as earlier indicated, are clauses which can be used to modify a head noun. There are two types: the restrictive and the non-restrictive relative clauses. Focus in this paper is on the restrictive relative clause. The discussion that follows presents the different relativization strategies in the language. It also examines the complex empirical generalizations underling the surface syntactic patterns of the placement of the relative marker (that is to determine the morphosyntactic information that regulates the different positioning of the relative morpheme), definiteness marking and phi-feature agreement in Kenyang relativized noun phrases. Two types of relative clauses exist in Kenyang: an aspectual relative clause that makes use of the aspectual morphemes of perfect and imperfect on the relativized clause and the nɛ́ -relative morpheme. The former solely requires the use of an aspectual tone. The latter, in contrast, involves the use of the aspectual morpheme alongside the nɛ́ -relative morpheme in the relative clause. Both types of relative clauses appear in the different relativization strategies used in Kenyang. Before we go into describing the structure of each relative clause type, we would like to indicate that the unmarked structure of a relative clause is a head noun followed by a relative marker as shown in (39). This is what we call the determiner phrase plus relative marker order, i.e. DP REL order. (39) m̀-mɔ̀ á-nɛ̀ á wáy m̀-mɛ́ n à 1-child aug-rel 1.agr.prf kill 9-goat 1.agr.prf kɛ̀ -nɔ̀ ŋ 14-prison “The child who killed the goat was imprisoned”

ʧwè enter

However, the head noun can also be preceded by the relative marker (REL DP order) to show focusing. In (40a) we have free focusing while in (40b) we have emphatic focusing, which is expressed not only through the REL DP order, but also the use of the focus marker kɛ̀ . á wáy m̀-mɛ́ n à (40) a. à-nɛ̀ m̀-mɔ̀ aug-rel 1-child 1.agr.prf kill 9-goat 1.agr.prf kɛ̀ -nɔ̀ ŋ 14-prison “The child who killed the goat was imprisoned”

ʧwè enter

100 

 Florence A. E. Tabe and Gratien G. Atindogbé

b. à-nɛ̀ m̀-mɔ̌ kɛ̀ á aug-rel 1-child foc 1.agr.prf “It is the child who killed Eta’s goat”

wáy kill

m̀-mɛ́ n 9-goat

́ am

ɛ̀ tá pn

The frequency of use of each structure is very high. However, the DP REL order is the preferred pattern. The REL DP is used to mark emphasis as indicated in (40b) above. Furthermore, the head noun in Kenyang can be doubly headed by a relative marker: ɛ̀ -nɛ́ ʧù ɛ̀ -nɛ́ ɔ́ kwú nɔ̂ (41) ɛ̀ -kátì 7-book aug-rel red aug-rel 2sg.prf buy anaph ɛ́ nɛ́ rɛ́ 7.agr.cop wet “The book which is red that you bought is wet / the red book that you bought is wet” The head noun can appear between the two relative clauses rather than otherwise in (41) above. (42) ɛ̀ -nɛ́ ɛ̀ -kátì ʧù ɛ̀ -nɛ̀ ɛ́ nɛ́ rɛ́ nɔ̂ aug-rel 7-book red aug-rel 7.agr.cop wet anaph ɛ̀ -yâ ʧí cop 7.agr-1sg.poss “The book which is red that is wet is mine / the red book that is wet is mine” The relative clauses presented above make use of the relative morpheme nɛ̀ whose tone varies depending on the context of occurrence. The aspectual relative clause is discussed in 3.1.

3.1 The aspectual relative clause The aspectual relative strategy lacks the relative morpheme nɛ̀ , rendering the structure of such clauses morphologically bare of relativization. The concept of relativization however, is expressed phonologically by tone and lengthening of the final vowel of the verb. The tone which is hosted by the agreement or subject marker to identify clauses in the perfect and imperfect aspect is modified when the clause is relativized. Indeed, the low tone of the agreement (43a) or subject marker (44a and 45a) becomes high in the relative clauses as shown in (43b, 44b and 45b). In addition, in the relativized sentences, the low tone(s) on the verb become(s) high and its final vowel lengthened.

5 Kenyang relative clauses 



(43) a. m̀-mɔ̀ à kwù ɛ̀ -kátì ǹtáh ákò 1-child 1.AGR.prf buy 7-book for PN “The child bought a book for Ako” á kwú-ú ɛ̀ -kátì ǹtáh b. m̀-mɔ̀ 1-child 1.AGR.prf.rel buy-fv 7-book for ʧwɔ̀ come “The child who bought a book for Ako has come”

ákò PN

(44) a. ɛ̀ tá à-á sò ǹ-dɛ̀ n ǹtáh ájōk PN 3sg-IPRF wash 10-dress for PN “Eta is washing dresses for Ayuk b. ɛ̀ tá á-á sò-ò ǹtáh ájōk ǹ-dɛ̀ n PN 3sg-IPRF.rel wash-fv 10-dress for PN “Eta who washes dresses for Ayuk has died”

 101

à 1.AGR.prf

à 3sg.prf

gwù die

(45) a. ɲɛ̌ ntí à pɔ́ kɔ́ ǹ-tɔ̀ ŋ ɛ̀ -kɛ̀ t bà-βé PN 3sg.prf carry 1-teacher 7-house 2-medicine “Nyenty carried the teacher to the hospital” b. ɲɛ̌ ntí á pɔ́ kɔ́ -ɔ́ ǹ-tɔ̀ ŋ ɛ̀ -kɛ̀ t bà-βé PN 3sg.prf.rel carry-fv 1-teacher 7-house 2-medicine à pɛ̀ ntsɛ̀ m 3sg.prf return “Nyenty who carried the teacher to the hospital has returned” The relative clauses in (43b), (44b) and (45b) concern the relativization of subjects. They all follow the DP REL order. Objects can also be relativized by this strategy. Even though Kenyang is SVO, an OSV word order can also be derived when objects are relativized in the language. The resulting sentences apparently look like topicalized constructions save that the latter contrary to the former are marked by an intonation break and a resumptive pronoun fills the basegenerated position of the topicalized constituent. Furthermore, topicalized constructions do not require modification of the tones on the subject marker and the verbs as it is the case for relative clauses. The relative clauses in (46b) and (47b) have their objects relativized as opposed to the relative clauses in (46c) and (47c), their topicalized counterparts: kwù (46) a. ákò à pn 3sg.prf buy “Ako bought a book”

ɛ̀ -kátì 7-book

102 

 Florence A. E. Tabe and Gratien G. Atindogbé

b. ɛ̀ -kátì ákò á kwú-ú ǹtáh òbì 7-book pn 3sg.prf.rel buy-fv for pn “The book that Ako bought for Obi” c. ɛ̀ -kátì, ákò à kwù yɔ́ ǹtáh òbì 7-book, pn 3sg.prf buy rpr for pn “The book, Ako bought it for Obi” (47) a. ɛ́ nɔ́ à wày sɛ̀ -nɛ̄ n pn 3sg.prf kill 13-birds “Enow killed a bird” b. sɛ̀ -nɛ̄ n ɛ́ nɔ́ á wáy-á ɛ̀ yú 13-birds pn 3sg.prf.rel kill-fv yesterday “The birds that Enoh killed yesterday” c. sɛ̀ -nɛ̄ n, ɛ́ nɔ́ à wày sɔ́ ɛ̀ yú 13-bird, pn 3sg.prf kill rpr yesterday “The bird, Enoh killed it yesterday” Gaps are found when subjects and direct objects are relativized. Locative objects are relativized in the same manner as exemplified below: (48) a. bàkó à rɔ̀ ŋ ɛ̀ -syɛ̀ pn 3sg.prf go 7-market “Bako went to the market” b. ɛ̀ -syɛ̀ bàkó á rɔ́ ŋ-ɔ́ 7-market pn 3sg.prf.rel go-fv “The market which Bako went to” c. ɛ̀ -syɛ̀ , bàkó à rɔ̀ ŋ à-rɛ́ 7-market, pn 3sg.prf go loc-there “As for the market, Bako went there” The relativization of indirect objects through the aspectual strategy is ­considered grammatical and hence acceptable in the language; but Kenyang speakers prefer to have the overt relative morpheme nɛ̀ in these constructions. The use of the latter is quite recurrent in their speech. The structure in (50), contrary to the one in (49), shows the relativization of indirect objects by the aspectual strategy in the language: (49) ɛ̀ bɔ̀ b à tò bísɔ̀ ŋ ɛ̀ sáírí ń-dɛ̀ n pn 3sg.prf send pn nice 9-dress “Ebob sent Besong a nice dress” tó-ó (50) bísɔ̀ ŋ ɛ̀ bɔ̀ b á ɛ̀ sáírí ń-dɛ̀ n pn pn 3sg.prf.rel send-fv nice 9-dress “Besong to whom Ebob sent a nice dress”

5 Kenyang relative clauses 



 103

The objects of prepositions can also be relativized as in (51b): (51) a. áʃù à ʧě ń-súkúrú nɛ̀ ŋ̀-gàk pn 3sg.prf peel 10-orange with 9-knife “Ashu peeled the oranges with a knife” ʧé-é ń-súkúrú b. ŋ̀-gàk áʃù á 9-knife pn 3sg.prf.rel peel-fv 10-orange “The knife with which Ashu peeled the oranges” Kenyang possessive NPs can also be relativized: (52) a. m̀-bɔ́ ŋ ́ ɛ́ -kɛ̀ t ́ à-wà 1-owner am 7-house am 1.agr-1sg.poss bòk mɛ̀ send away prn “My landlord sent me away” b. m̀-bɔ́ ŋ ́ ɛ́ -kɛ̀ t ́ à-wà 1-owner am 7-house am 1.agr-1sg.poss bòk mɛ̀ send away prn “My landlord who sent me away”

à 1.agr.prf

á 1.agr.prf.rel

What we gather from the preceding data using the aspectual form to encode relativization in Kenyang is that the aspectual relative form directly follows the head noun when the subjects are involved. In contrast, the relativization of direct objects, indirect objects and objects of prepositions, locatives, etc. follow this order (DP REL), where the head noun is not local to the relative marker.

3.2 The nɛ̀ relative marker Morphologically, the relative morpheme nɛ̀ introduces relative clauses in Kenyang finite constructions. The morpheme is preceded by a relative conjunction which apparently does not vary as such with respect to person, number and noun class. The relative conjunctions are: à and ɛ̀. They are both used as conjunctions in the relativization of singular and plural head nouns. However, the plural head nouns have their plurality indicated by an agreeing morpheme which follows the conjunct and precedes the relative nɛ̀. (53) a. ɛ̀ -kɛ̀ t ɛ̀ -nɛ́ mɛ̀ ŋ́ 7-house aug-rel 1sg 1.agr.prf “The house that I bought is big”

kwú-ú buy-fv

ɛ́ 7.agr.cop

ʧèk big

104 

 Florence A. E. Tabe and Gratien G. Atindogbé

b. bɛ̀ -kɛ̀ t ɛ̀ -βɛ́ -nɛ́ mɛ̀ ŋ́ 8-house aug-8-rel 1sg 1.agr.prf “The houses that I bought are big”

kwú-ú buy-fv

bɛ́ 8.agr.cop

ʧèk big

(54) a. ǹ-tɔ̀ ŋ á-nɛ̀ á dɛ́ n-ɛ́ β-ɔ́ ǎ mè 1-teacher AUG-rel 1.AGR.prf beat-fv 2-child 1.AGR.COP sick “The teacher who beat the children is sick” b. bà-tɔ̀ ŋ á-βɛ̄ -nɛ́ bá dɛ́ n-ɛ́ β-ɔ́ bá mè 2-teacher AUG-2-rel 2.AGR.prf beat-fv 2-child 2.AGR.COP sick “The teachers who beat the children are sick” (55) a. ǹ-dɛ̀ n kwú-ú á-nɛ̀ ɛ̀ tá á 9-dress AUG-rel PN 1.AGR.prf buy-fv “The dress which Eta bought is torn” b. ǹ-dɛ̀ n ɛ́ -nɛ̄ ɛ̀ tá á kwú-ú 10-dress aug-rel pn 1.agr.prf buy-fv “The dresses which Eta bought are torn”

à 9.AGR.COP

dàk tear

ɛ́ 10.agr.cop

dàk tear

The data presented in (53–55) show that the morpheme that precedes the relative marker is invariant to person, number and noun class. It may not in this connection be appropriate to refer to it as a noun class prefix. In relevant works, it is referred to as conjoin or augment. We will adopt the latter in this work. It is also important to indicate that the intervening agreement marker between the augment and the relative is not overtly marked in some relative clauses. It would appear that the position is not marked when the agreeing nominal morpheme is a vowel (V-syllable), but not because the head noun is singular. Since the concords (usually a and ɛ) have the same phonological shape as the augment, they are assimilated by the augment. This means that (56a) is acceptable while (56b) is ungrammatical. (56) a. ɛ̀ -kɛ̀ t b. *ɛ̀ -kɛ̀ t

ɛ̀ -nɛ́ ɛ̀ -ɛ̀ -nɛ́

ɛ̀ tá ɛ̀ tá

á á

kwú-ú . . . kwú-ú . . .

The morphemes for head nouns that begin with a consonant (CV-syllable) must be marked overtly, as shown below: (57) a. sɛ̀ -nɔ̀ ŋ sɛ́ yɛ̀ m m̀-mɔ̀ 13-iron 13.agr.prf pierce 1-child “The piece of iron pierced the child” gyɛ̀ p sɛ̀ -nɔ̀ ŋ ɛ̀ -sɛ́ -nɛ́ sɛ́ b. ákò à pn 3sg.prf throw 13-iron aug-13-rel 13.agr.prf yɛ́ m-ɛ́ ḿ-mɔ́ pierce-fv 1-child “Ako threw the piece of iron that pierced the child”

5 Kenyang relative clauses 



 105

c. *ákò à gyɛ̀ p sɛ̀ -nɔ̀ ŋ ɛ̀ -nɛ́ sɛ́ yɛ́ m-ɛ́ ḿ-mɔ́ pn 3sg.prf throw 13-iron aug-rel 13.agr.prf pierce-fv 1-child The relative morpheme nɛ̀ is a multifunctional morpheme in Kenyang. It is used to mark coordination (58), applicative (59) and as a preposition (60): (58) ɛ̀ tá nɛ̀ ákò pn and pn “Eta and Ako” (59) ɛ̀ tá à ʧé ǹ-súkúrú nɛ̀ pn 3sg.prf peel 9-orange appl.with “Eta peeled the oranges with a knife”

ŋ-gàk 9-knife

(60) ɛ̀ tá à nɛ̀ β-ɔ̀ rɔ̀ ŋ ɛ̀ -syɛ̀ pn 3sg.prf go 7-market with 2-child “Eta went to the market with the children” It is also used as a proximal demonstrative pronoun. In this function, the morpheme is always preceded by the vowels /a/ or /ɛ/. The structural patterning of the morphemes is similar to what we saw in relative clauses with nɛ. The vowels are treated in this context not as augments but rather as locatives. (61) ɛ̀ -kátì ɛ̀ -nɛ̀ 7-book loc-this “This book” The distinction between the interpretation as a demonstrative and as a relative clause rests also on the tone of nɛ̀ , the aspectual tones on the concords or subject markers and the following verbs. (62) a. ɛ̀ -βà ɛ̀ -nɛ̀ ɛ́ rì̵ ɛ́ -ʧá 7-bag loc.this 7.agr.cop nice 7.agr-more than “This bag is more beautiful than mine” b. bɛ̀ -βà ɛ̀ -βɛ́ -nɛ́ bɛ́ té 8-bag loc-8-rel 8.agr.cop burst “These bags are burst”

ɛ̀ -yâ 7.agr-1sg.poss

(63) a. ɛ̀ -βà ɛ̀ -nɛ́ ɛ́ rí̵ ɛ́ -ʧá ɛ̀ -yâ 7-bag aug-rel 7.agr.cop nice 7.agr-more than 7.agr-1sg.poss “The bag which is more beautiful than mine” ákò à sɔ̀ ŋ b. bɛ̀ -βà ɛ̀ -βɛ́ -nɛ́ bɛ́ té βɔ́ 8-bag loc-8-rel 8.agr.cop burst pn 3sg.prf burn rpr “Ako burnt the bags that are burst”

106 

 Florence A. E. Tabe and Gratien G. Atindogbé

For the distinction between (62) and (63), we note that nɛ̀ carries a low tone as a demonstrative contrary to its relative counterpart nɛ́ with a high tone. Furthermore, the tones on the subject markers vary: demonstratives use the unmarked tone perfect / imperfect aspect, while relative clauses modify the perfect / imperfect tones. In addition, the tones of the main verbs may vary in relative clauses. What this indicates is that the relative nɛ́ is derived from the demonstrative pronoun nɛ̀ through grammaticalization. It is customary, however, to hear Kenyang speakers use the morpheme rɛ́ interchangeably with nɛ́ as a form of relativization. The use of both forms is ­constrained to some extent by the same morphosyntactic rules. They are relative morphemes. They modify the head noun by either preceding or following it. When they precede the head noun, they encode focus on the entire DP. Notwithstanding, there is a major distinction between the two forms. The form nɛ́ relativizes NPs with new information (focus), while rɛ́ is used for old information (topic). (64) ɛ̀ -βà ɛ̀ -nɛ́ ákò á 7-bag aug-rel.foc pn 3sg.prf “The bag which Ako bought is burst”

kwú-ú buy-fv

ɛ́ 7.agr.cop

tē burst

ákò á (65) ɛ̀ -βà ɛ́ -rɛ́ 7-bag aug-rel.top pn 3sg.prf “The bag which Ako bought is burst”

kwú-ú buy-fv

ɛ̀ 3.agr.cop

tē burst

The information about the bag in (64) is new, while in (65) it is old information that both speaker and hearer share about the bag. The morpheme rɛ́ also has multiple functions in Kenyang. It is the morpheme that can be used extensively with question formation in the language including yes/no questions, echo questions, constituent questions, etc. When rɛ́ is used in constituent questions, it functions as a relative marker and a question marker with a focus interpretation. kwù yì (66) a. tábí à pn 3sg.prf buy what “What did Tabi buy? b. yì ɛ̀ -rɛ́ tábí what aug-rel.foc.q pn “What did Tabi buy?”

ɛ̀ -rɛ́ ? aug-rel.foc.q á 3sg.prf

(67) a. ɔ̀ ɣɔ̀ àhá á-rɛ́ 2sg.prf see who aug-rel.foc.q “Who did you see yesterday?”

kwú-ú? buy-fv ɛ̀ yú? yesterday

5 Kenyang relative clauses 



b. àhá á-rɛ́ ɔ́ who aug-rel.foc.q 2sg.prf “Who did you see yesterday?

ɣɔ́ see

 107

ɛ̀ yú? yesterday

Constituent questions as earlier indicated cannot co-occur with the form nɛ́ . This means that (66) will be ungrammatical if the question word occurs with nɛ́, as in (68). (68) *tábí pn

à 3sg.prf

kwù buy

yì what

ɛ̀ -nɛ́ ? aug-rel.foc.q

Yes / no questions also co-occur with the form rɛ́ rather than nɛ́. (69) a. ákò à rɔ̀ ŋ ɛ̀ -βɨ ɛ̀ -rɛ́ ? pn 3sg.prf go 7-farm aug-rel.topic.q “Did Ako go to the farm?” b. *ákò à rɔ̀ ŋ ɛ̀-βɨ ɛ̀-nɛ́? pn 3sg.prf go 7-farm aug-rel.topic.q “Did Ako go to the farm?” It should be indicated that they denote different semantic interpretations depending on the type of question involved. In constituent questions, information is asked about some unknown object. The rɛ́ form in such questions encodes relativization, focus and question. In yes / no questions, in contrast, the presence of rɛ́ is used to signal the questioning of some previous information that both speaker and hearer share. The form rɛ́ in such contexts functions to mark relativization, topic and question. In addition, the morpheme rɛ́ is used to indicate topicalization. kwù ɛ̀ -βà (70) a. ákò à pn 3sg.prf buy 7-bag “Ako bought a bag” ákò à b. ɛ̀ -βà ɛ̀ -rɛ́ 7-bag aug-top pn 3sg.prf “As for the bag, Ako bought it”

kwù buy

yɔ́ rpr

3.3 The anaphoric relator nɔ́ The morpheme nɔ́ , anaphoric relator, can appear, albeit optionally, in all relative clauses in Kenyang. It is used to denote some background information that is shared by those involved in the discourse. Some authors have referred to morphemes with such functions as a definitizer (for example, Kouankem 2010). In Kenyang the

108 

 Florence A. E. Tabe and Gratien G. Atindogbé

relator has two syntactic positions whose semantic interpretation remains invariant. It can directly follow the relative morpheme or directly follow the verb. (71) a. ɛ̀ -kátì ɛ̀ -nɛ́ nɔ̂ ɛ́ nɛ́ rɛ́ . . . 7-book aug-rel anaph.rel 7.agr.cop wet “The book that is wet . . .” b. ɛ̀ -kátì ɛ̀ -nɛ́ ɛ́ nɛ́ rɛ́ nɔ̂ … 7-book aug-rel 7.agr.cop wet anaph.rel “The book that is wet…” ɛ́ -rɛ́ (72) a. ɛ̀ -kátì 7-book aug-rel.top “The book that is wet” b. ɛ̀ -kátì ɛ́ -rɛ́ 7-book aug-rel.top The book that is wet”

nɔ̂ anaph.rel ɛ́ 7.agr.cop

ɛ́ 7.agr.cop nɛ́ rɛ́ wet

nɛ́ rɛ́ wet

nɔ̂ … anaph.rel

The discussion of Kenyang relatives presented above mirrors Keenan and Comrie’s (1977) pattern of accessibility hierarchy in relative clauses: Subject > direct object > indirect object > genitive> object of comparison Consider the following Kenyang constructions showing the relativization of ­subjects using the relative morphemes: (73) a. ɛ̀ -kɛ̀ t ɛ̀ -nɛ́ ɛ́ 7-house aug-rel 7.agr.prf ɛ́ -yâ 7.agr-1sg.poss “The house that burnt is mine” b. ɛ̀ -kɛ̀ t ɛ́ -rɛ̄ ɛ́ 7-house aug-rel 7.agr.prf ɛ́ -yâ 7.agr-1sg.poss “The house that burnt is mine”

sɔ́ ŋ-ɔ́ burn-fv

nɔ̂ anaph.rel

ʧí cop

sɔ́ ŋ-ɔ́ burn-fv

nɔ̂ anaph.rel

ʧí cop

In the rest of the discussion that follows, the form nɛ́ will be used as the relative morpheme. Direct objects can be relativized by nɛ́ : (74) ǹ-tù á-nɛ̄ β-ɔ̀ bá pɔ́ kɔ́ -ɔ́ 3-load aug-rel 2-child 1.agr.prf carry-fv ǎ ɲwɔ̀ p 3.agr.cop heavy “The load that the children carried is heavy”

(nɔ̂ ) anaph.rel



5 Kenyang relative clauses 

 109

Sentence (75) shows the relativization of an indirect object using the relative morpheme. When indirect objects are relativized, there is a locally accessible resumptive pronoun in-situ. This resumptive pronoun obligatorily fills the gap left by the relativized object. The resumptive pronoun in question behaves like a real pronoun rather than a variable, given that it is not bound by island configurations. We shall refer to these construction types as resumptive relatives. Resumptive relatives are a reflex of weak topicalization, given the absence of intonation break following the topicalized constituent (or preceding the relativized morpheme). This ensues when indirect objects are relativized. The resumptive pronoun represents the element relativized in these cases probably because Kenyang does not strand its prepositions, as indicated in sentences (76) and (77). óbɛ̂ n á kwú-ú (75) àjók á-nɛ̀ pn aug-rel pn 3sg.prf buy-fv “Ayuk whom Oben bought a book for”

(nɔ̂ ) anaph.rel

yí rpr.him

ɛ̀ -kátì 7-book

Sentences (76) and (77) which are concerned with the relativization of objects of preposition show that Kenyang does not allow preposition stranding: á-nɛ̄ ḿbí á tɛ́ m-ɛ́ (76) ŋ̀-gó 9-gun aug-rel PN 3sg.prf shoot-fv nɛ̀ wú with rpr.it “The gun that Mbi shot the dog with” (77) *ŋ̀-gó á-nɛ̄ ḿbí á tɛ́ m-ɛ́ 9-gun aug-rel pn 3sg.prf shoot-fv “The gun that Mbi shot the dog with”

(nɔ̂ ) anaph.rel

(nɔ̂ ) anaph.rel

m̀-mú 9-dog

m̀-mú 9-dog

nɛ̀ with

Locative relatives in Kenyang may or may not require a visible correlative element. Where there is no correlative element to encode aboutness of the relativized noun phrase, the locative PP (non-human) pronominal object á-rɛ́ (‘there’) appears in-situ as a resumptive pronoun. Sentence (78) shows the relativization of locative subjects, while sentences (80-84) show the relativization of locative objects: (78) a. ŋ̀-kì̵ à sɔ́ ŋ 9-farm 9.agr.prf burn “The farm burnt” á-nɛ̄ á b. ŋ̀-kì̵ 9-farm aug-rel 9.agr.prf “The farm that burnt”

sɔ́ ŋ-ɔ́ burn-fv

(nɔ̂ ) anaph.rel

110 

 Florence A. E. Tabe and Gratien G. Atindogbé

(79) a. bà-ɣɔ̀ rɛ́ bá rɔ́ ŋ ɛ̀ -β ì̵ 2-women 2.agr.prf go 7-bush “The women went to the farm” b. ɛ̀ -βì̵ ɛ́ -nɛ̄ bà-ɣɔ̀ rɛ́ bá rɔ́ ŋ-ɔ́ (nɔ̂ ) á-rɛ̄ 7-bush aug-rel 2-women 2.agr.prf go-fv anaph.rel loc-there “The farm that the women went to” (80) a. β-ɔ̀ bá tì bɛ́ -rè ɛ̀ -sjɛ̀ 2-child 1.agr.prf sell 8-vegetable 7-market “The children sold vegetable in the market” b. ɛ̀ -sjɛ́ ɛ́ -nɛ̄ β-ɔ̀ bá tí-í 7-market aug-rel 2-child 1.agr.prf sell-fv bɛ́ -rɨ̂ á-rɛ̄ 8-vegetable loc-there “The market where the children sold vegetable”

(nɔ̂ ) anaph.rel

In contrast, when the locative preposition is present, it functions in determining the aboutness of the relativized object. The preposition is often preceded by the morpheme /a/ that is cliticized to its left. The morpheme which is identified in Ewondo as “l’augment” (Onguene Essono 2004) functions as a locative definitizer in Kenyang. (81) a. sɛ́ wɛ̄ rɛ́ bɛ̀ -kyɛ̀ βɛ̀ à-mɛ̀ m 1pl.prf keep 8-box loc-inside “We kept the suitcases in the room” b. mɔ̀ -ɛ̀ -kɛ̀ t à-nɛ̀ sɛ́ wɛ́ rɛ́ dim-7-room aug-rel 1pl.prf keep à-mɛ̀ m loc-inside “The room where we kept the suitcases”

mɔ̀ -ɛ̀ -kɛ̀ t dim-7-room (nɔ̂ ) anaph.rel

â-ǹfàì (82) a. ǹ-tɛ́ m-kɛ̀ ntɛ̀ mɛ́ à βɛ̀ sɛ́ ŋ̄-gó 1-person-hunting 1.agr.prf hide 3-gun loc.top “The hunter hid the gun on top of the hill” ǹ-tɛ́ m-kɛntɛ́ mɛ́ á βɛ́ sɛ́ b. n̄ -jɛ̀ à-nɛ̀ 9-hill aug-rel 1-person-hunting 1.agr.prf hide ŋ̄-gó à-ǹfáì 9-gun LOC-there “The hill where the hunter hid the gun”

bɛ̀ -kyɛ̀ βɛ̀ 8-box

n̄ -jɛ̀ 9-hill (nɔ̂ ) anaph.rel

The preposition ‘on’ is identified as à similar to the locative definitizer aforementioned. When the object of the said preposition gets relativized, the locative



5 Kenyang relative clauses 

 111

­resumptive adverb á-rɛ́ ‘there’ replaces the original preposition -a- ‘on’. This goes to confirm that pure prepositions cannot be stranded in Kenyang. Consider the following examples: (83) a. ɛ́ nɔ́ à tò bà-wɛ́ d à ŋ̀-kòk pn 3sg.prf throw 6-oil on 3-wall “Enow spilled oil on the wall” tó-ó b. ŋ̀-kòk à-nɛ̀ ɛ́ nɔ́ á (nɔ̂ ) bà-wɛ́ d á-rɛ̄ 3-wall aug-rel pn 3sg.prf throw-fv anaph.rel 6-oil loc-there “The wall that Enow spilled oil on/The wall on which Enow spilled oil” c. *ŋ̀-kòk à-nɛ̀ ɛ́ nɔ́ á tó-ó (nɔ̂ ) bà-wɛ́ d á 3-wall aug-rel pn 3sg.prf throw-fv anaph.rel 6-oil on “The wall on which Enow spilled oil” Relativization can be extended to objects of comparison. Here the anaphoric relator is obligatory, as illustrated in (84): (84) a. m̀bɛ̀ ŋ à ʧèk à-ʧá ǹ-fɔ̀ pn 3sg.cop big 1.agr-more than 1-chief “Mbeng is bigger than the chief” b. ǹ-fɔ̀ à-nɛ̀ m̀bɛ̀ ŋ á ʧék á-ʧá 1-chief aug-rel pn 3sg.cop big 1.agr-more than á ʧwɔ́ ɛ́ jū nɔ̂ jí anaph.rel rpr.him 1.agr.prf come yesterday “The chief who Mbeng is bigger than arrived yesterday” Next, we consider the relativization of genitive noun phrases. Recall that the Kenyang equivalent of the Saxon genitive ‘of’ is nɛ́. The latter is always preceded by a morpheme (a or ɛ) having the same shape as the augment in relative clauses. We will treat the morpheme in the same capacity as an augment in genitive relative clauses. (85) a. ɛ̀ -kɛ̀ t ɛ̀ -nɛ́ ɛ́ bàì ɛ́ mókó 7-house aug.poss pn 7.agr.prf collapse “The house of Ebai collapsed/Ebai’s house collapsed” b. ɛ̀ -kɛ̀ t ɛ̀ -nɛ́ ɛ́ bàì ɛ̀ -nɛ́ ɛ́ mókó-ó (nɔ̂ ) 7-house aug.poss pn aug-rel 7.agr.prf collapse-fv anaph.rel “The house of Ebai that collapsed/Ebai’s house that collapsed” Reasoning along these lines, one is tempted to say that Kenyang possessive and relative clause DPs have a uniform structural pattern. That is, possessive and relative clauses are preceded by the morpheme nɛ́. Although the complex noun phrases of both are introduced by the marker nɛ́, each clause type is constrained by specific morphosyntactic features. These will not be examined in this paper

112 

 Florence A. E. Tabe and Gratien G. Atindogbé

(cf. Dikken 2007 for details of these differences in Amharic). It is important to note that in a construction like the one in (85b) above, repeated in (86a), Kenyang speakers will replace the possessive determiner ɛ̀ -nɛ́ with either a floating associative tone marker (86b) or the oblique particle ǹtáh ‘for / to’ (86c): (86) a. ɛ̀ -kɛ̀ t ɛ̀ -nɛ́ ɛ̀ bàì ɛ̀ -nɛ́ ɛ́ mókó-ó (nɔ̂ ) 7-house aug.poss pn aug-rel 7.agr.prf collapse-fv anaph.rel “The house of Ebai that collapsed/Ebai’s house that collapsed” b. ɛ̀ -kɛ̀ t ́ ɛ́ bàì ɛ́ mókó-ó (nɔ̂ ) 7-house am pn 7.agr.prf collapsed-fv anaph.rel “The house of Ebai that collapsed/Ebai’s house that collapsed” c. ɛ̀ -kɛ̀ t ǹtáh ɛ̀ bàì ɛ́ mókó-ó (nɔ̂ ) 7-house for pn 7.agr.prf collapsed-fv anaph.rel “The house of Ebai that collapsed/Ebai’s house that collapsed” A possible reason to this discursive behavior may be to avoid repetition of ɛ̀ -nɛ́ in the same clause. Relative clauses can also be adjoined to pronouns in Kenyang. When this happens, the complementizer agrees with the pronoun in number and gender and partially in person. The presence of the anaphoric relator is optional in these cases: (87) mɛ̀ ǹ-nɛ̀ ń só-ó 1sg aug-rel 1.agr.prf wash-fv “I who washed dresses for Ayuk”

(nɔ̂ ) anaph.rel

(88) wɔ̀ ɔ̀ -nɛ̀ ɔ́ ɣɛ́ p-ɛ́ (nɔ̂ ) 2sg aug-rel 1.agr.prf steal-fv anaph.rel kɛ̀ ɔ́ -ɔ́ ʧwè m̀-bóg foc 1.agr-iprf enter 9-hole “You who stole the chief’s goat will enter the hole” á-nɛ̀ á βɔ́ ŋ-ɔ́ (nɔ̂ ) (89) jí 3sg aug-rel 1.agr.prf have anaph.rel à gwù 1.agr.prf die “S/he who had plenty of money died” (90) bɛ̀ sɛ́ ɛ́ -βɛ́ -nɛ́ sɛ́ síŋ-í 1pl.prn aug-2-rel 2.agr.prf write-fv sɛ́ fwɛ̀ t 2.agr.prf pass “We who wrote the exams well passed”

ǹ-dɛ́ n 10-dresses

ǹtáh for

m̀-mɛ̀ n 9-goat

ǹ-fɔ̀ 1-chief

bɛ̀ yà plenty

(nɔ̂ ) anaph.rel

ŋ́-kāp 3-money

nɛ̄ mɔ̀ exam

sàírí well

àjók pn



5 Kenyang relative clauses 

(91) bèká à-βɛ́ -nɛ́ bá nyísí-í (nɔ̂ ) 2pl aug-2-rel 2.agr.prf refuse-fv anaph.rel bà-á rɔ̀ p ɛ̀ -βì̵ 2.agr-iprf remain 7-bush “You who refused to work will stay in the bush”

 113

bɛ̀ -tì̵k 8-work

(92) bɔ́ á-βɛ̄ -nɛ́ bá kwúrí-í (nɔ̂ ) ʧí β-ɔ̀ 3pl aug-2-rel 1.agr.prf stoop-fv anaph.rel cop 2-child “They who stooped are Ako’s children”

ákò pn

(93) yɔ́ ɛ́ -nɛ́ ɛ́ dyɛ́ rɛ́ (nɔ̂ ) kɛ́ mɛ̀ 3sg aug-rel 3.agr.cop clean anaph.rel foc 1sg ŋ́ kɔ́ ŋ-ɔ́ 1.agr.iprf like-fv “That which is clean is what I like/I like that which is clean”

3.4 The infinitival relative marker bɛ̀ The infinitival morpheme bɛ̀ introduces another type of relative clause in Kenyang. Infinitival relatives are similar to control clauses given the presence of the infinitival marker in both constructions, an anaphor, PRO in the binding domain of a relevant antecedent. They are different from pure control clauses because they lack control verbs: (94) ɛ̀ -fɔ́ nɛ́ bɛ́ dí̵ŋ ɛ̀ -βáy ɛ́ 7-mortar to pound 7-fufu 7.agr.cop “The mortar to pound fufu is broken”

dáká break

(95) ɣɔ́ m̀-mù bɛ́ yá á-ʧwí ǹtáh see 1-person to drive 3-car for “This is the person to drive the chief’s car”

ǹ-fɔ̀ 1-chief

nɛ̀ this

3.5 Headless or free relative clauses Kenyang is a pro-drop language. Constructions with null subjects allow the phifeatures of the missing subjects to be retrieved from the preverbal subject markers. kɔ̀ k ǹ-súkúrú (96) a. ákò à pn 3sg.prf pluck 10-orange “Ako plucked some oranges”

114 

 Florence A. E. Tabe and Gratien G. Atindogbé

b. Ø à kɔ̀ k ǹ-súkúrú pro 3sg.prf pluck 10-orange “S/he plucked some oranges” (97) a. ákò á-nɛ̀ á kɔ́ k ǹ-súkúrú à ʧwɔ̀ pn aug-rel 3sg.prf pluck 10-orange 3sg.prf come “Ako who plucked some oranges has come” b. Ø á-nɛ̀ á kɔ́ k ǹ-súkúrú à ʧwɔ̀ pro aug-rel 1.agr.prf pluck 10-orange 1.agr.prf come “The one who plucked oranges has come” The missing subjects of headless relatives have indefinite reference. They do not relate to a specific NP (impersonal reference). Kenyang uses headless relatives in narratives and proverbs. (98) Ø á-nɛ̀ á-á βɔ́ ŋ pro aug-rel 1.agr-iprf have “One who has ears should hear”

bà-tú 4-ear

á 1.agr.hort

ŋ́gók hear

4 Agreement in relativized DPs This section presents an array of agreement facts exhibited by relativized noun phrases in Kenyang. The discussion of φ-feature agreement in Kenyang complex noun phrases is limited to noun class. The relativizer nɛ̀ observed above can appear in different positions in relativized noun phrases in Kenyang. In particular, it can precede or follow the constituent being relativized in the language. Consider the structures in (99–100): (99) a. m̀-mɔ̀ á-nɛ̀ á 1-child aug-rel 1.agr.prf “The child who bought a book” b. à-nɛ̀ m̀-mɔ̀ á aug-rel 1-child 1.agr.prf “The child who bought a book”

kwú-ú buy-fv

(nɔ̂ ) anaph.rel

ɛ̀ -kátì 7-book

kwú-ú buy-fv

(nɔ̂ ) anaph.rel

ɛ̀ -kátì 7-book

(100) a. ɛ̀ -kày ɛ̀ -nɛ́ pɛ́ pɛ́ p ǹtáh ǹ-fɔ̀ 7-basket aug-rel white for 1-chief The chief’s basket that is white/The white basket for the chief” b. ɛ̀ -nɛ́ ɛ̀ -kày pɛ́ pɛ́ p ǹtáh ǹ-fɔ̀ aug-rel 7-basket white for 1-chief “The white basket for the chief”



5 Kenyang relative clauses 

 115

As a noun class language, Kenyang exhibits different forms of agreement morphemes. Each of these mirrors the nominal prefix of a particular class. The structure of the morphemes conforms to one of these syllable configurations: V or CV. When the relative morpheme co-occurs with the CV pattern, the latter must be immediately preceded by the vowel /a/ or /ɛ/ as shown in the following: (101) a. bɛ̀ -káy bɛ́ sɔ́ ŋ-ɔ́ mɛ̀ ŋkɛ̀ m 8-basket 8.agr.prf burn-fv all “The baskets all burnt” b. bɛ̀ -káy ɛ̀ -βɛ́ -nɛ́ bɛ́ sɔ́ ŋ-ɔ́ (nɔ̂ ) mɛ̀ ŋkɛ̀ m 8-baskets aug-8-rel 8.agr.prf burn-fv anaph.rel all “The baskets that burnt all” c. ɛ̀ -βɛ́ -nɛ́ bɛ̀ -káy bɛ́ sɔ́ ŋ-ɔ́ (nɔ̂ ) mɛ̀ ŋkɛ̀ m aug-8-rel 8-basket 8.agr.prf burn-fv anaph.rel all “The baskets that burnt all” (102) a. sɛ̀ -kwòp sɛ́ kwɛ̀ n ámɛ̀ m ǹ-nòk 13-spoon 13.agr.prf fall into 9-soup “The spoon fell into the soup” b. sɛ̀ -kwòp ɛ̀ -sɛ́ -nɛ́ sɛ́ kwɛ́ n-ɛ́ (nɔ̂ ) 13-spoon aug-13-rel 13.agr.prf fall-fv anaph.rel ǹ-nòk 9-soup “The spoon that fell into the soup” c. ɛ̀ -sɛ́ -nɛ́ sɛ̀ -kwòp sɛ́ kwɛ́ n-ɛ́ (nɔ̂ ) aug-13-rel 13-spoon 13.agr.prf fall-fv anaph.rel ǹ-nòk 9-soup “The spoon that fell into the soup”

àmɛ̀ m into

àmɛ̀ m into

We had earlier indicated above that the relativizer in Kenyang finite clauses is a complementizer-like element composed of two morphemes. The vowel preceding nɛ́ is the augment. The latter may carry phi-features for the class and number of the constituent being relativized, while nɛ́ is the relative morpheme itself. Reasoning along these lines poses a problem on how to analyse each of the morphemes that make up the relativizer in (101) and (102). The issue here is whether each of the morphemes preceding nɛ́ signals number as well as class. If it does so, what accounts for the fact that a nominal element can co-occur with two different agreement features (singular and plural) in the same clause without resulting to interpretability? If the morphemes are interpreted otherwise, what therefore is the function of each in such clauses? One way of going through this is to assume that the relativizer has the structure V-CV in Kenyang.

116 

 Florence A. E. Tabe and Gratien G. Atindogbé

When the structure co-occurs with a nominal having an agreement feature homophonous with its initial morpheme V, the latter deletes under feature identity. This explains why, as earlier mentioned, we have a V-CV pattern rather than a V-V-CV pattern in such constructions. It further explains why we have the contrary pattern V-CV-CV in sentences (101) and (102) above.

5 Conclusion The paper set out to describe the structure of relative clauses in Kenyang, a Bantoid language of Cameroon. The findings reveal that the language employs different relativization strategies which are phonologically and morphosyntactically constrained. The language appears to have two unmarked relative strategies: DP REL and REL DP. However, the order DP REL is dominant in the language. Kenyang uses the REL DP order to denote focus. The analysis of Kenyang data fits into the Accessibility Hierarchy of relative clauses projected by Keenan and Comrie (1977).

References Baker, Mark. 1985. The mirror principle and morphosyntactic explanation. Linguistic Inquiry 16. 373–416. Batko, Ann. 2004. When bad grammar happens to good people: How to avoid common errors in English. New Jersey: The Career Press. Bird, Charles S. & Mamadou Kante. 1976. An ka bamanankan kalan: Intermediate Bambara. Bloomington, Indiana: University Linguistics Club. Chomsky, Noam. 1981. Lectures on government and binding. Dordrecht: Foris. Collins, Peter & Carmella Hollo. 2000. English grammar: An introduction. London: Macmillan Press. Couro, Ted & Margaret Langdon. 1975. Let’s talk ‘Lipay Aa: An introduction to the Mesa Grande Diegueño language. Banning & Ramona, California: Malki Museum Press & Ballena. Creissels, Denis. 2000. Typology. In Bernd Heine & Derek Nurse (eds.), African linguistics: An introduction, 231–258. Cambridge: Cambridge University Press. Den Dikken, Marcel. 2007. Amharic relatives and possessives: Definiteness, agreement, and the linker. Linguistic Inquiry 38. 302–320. Keenan, Edward L. & Bernard Comrie. 1977. Noun phrase accessibility and universal grammar. Linguistic Inquiry 8. 63–99. Keenan, Edward L. 1985. Relative clauses. In Timothy Shopen (ed.), Language typology and syntactic description. Volume II: Complex constructions, 141–170. Cambridge: Cambridge University Press.



5 Kenyang relative clauses 

 117

Kouankem, Constantine. 2012. The syntax of the Medumba determiner phrase. Yaounde: University of Yaounde I dissertation. Kuteva, Tania & Bernard Comrie. 2006. The typology of relative clause formation in African languages. In F. K. Erhard Voeltz (ed.), Studies in African linguistic typology, 209–228. Amsterdam/Philadelphia: John Benjamins. Lehmann, Winfred P. 1976. Descriptive linguistics: An introduction. New York: Random House. Onguene Essono, Louis-Marie. 2004. Syntaxe et fonctionnement de la relative et de l’interrogative en éwondo : Essai d’analyse de la subordination en bantou. Revue Internationale des Arts, Lettres et Sciences Sociales 1 (1). 113–139. Peterson, Thomas H. 1974. On definite restrictive relatives in Moore. Journal of West African Languages 4 (2). 71–78. Tallerman, Maggie. 1998. Understanding syntax. London: Arnold.

John R. Watters

6 Relative clause in (Western) Ejagham Abstract: This paper provides an understanding of the structure and constraints of relative clauses in (Western) Ejagham. This understanding is gained through a review of certain relevant typological features of Ejagham, an analysis of the basic structure of the relative clause and the variations on this structure, the strategies used in forming relative clauses, and finally how these strategies relate to the grammatical relations of noun phrases. Ejagham has a standard RC structure for a S-V-O language. However, it has a number of permissible syntactic variations on this structure, including three sets of relativizers. Ejagham uses two strategies to indicate the NP within the RC that is co-referential with the head NP of the RC: the gap strategy and the resumptive pronoun strategy. Given these two strategies, Ejagham puts no limits on which grammatical relations can be relativized.

1 Introduction Relative clauses provide an effective strategy to narrow the referent of a noun phrase to a more specific referent or set of referents. This more specific referent or set of referents would be a subset of all the possible referents to which the given noun phrase may refer. The problem addressed in this study is to better understand the structure and constraints of relative clauses in (Western) Ejagham in all their varied forms. This understanding will be gained through a review of certain relevant typological features of Ejagham (2.0), an analysis of the structure of the relative clause and its variations (3.0), the strategies used in forming relative clauses (4.0), how these strategies relate to the grammatical relations of noun phrases (5.0), and some final comments. This outline follows the major points made by Payne (1997) and Andrews (2007) in their summaries concerning universal properties of relative clauses. Ejagham is classified as a Niger-Congo, Atlantic-Congo, Volta-Congo, (East) Benue-Congo, South, Bantoid, Ekoid language (Watters 1989: 404–413; Williamson and Blench 2000: 31–35; Lewis, Simons, Fennig 2016). The “Ekoid” category is also referred to as “Ekoid-Mbe” byWilliamson and Blench. It is a cross-border language spoken in the South West Region of Cameroon and the Cross River State in southeastern Nigeria.

John R. Watters: SIL International, [email protected] DOI 10.1515/9783110469547-006

120 

 John R. Watters

It is the only Ekoid language found in Cameroon. It is spoken in the Eyumojok Subdivision of the Manyu Division. The total number of Ejagham speakers is approximately 116,700 (Lewis, Simons, and Fennig 2016). Something around 49,400 speakers of Ejagham reside in Cameroon and 67,300 in Nigeria. They live in the rainforest of the Manyu and Cross River Basin within the triangle formed by Calabar and Ikom in Nigeria and Mamfe in Cameroon. The language consists of three major dialects – Western, Eastern, and Southern Ejagham. There are about a dozen sub-dialects, including a transitional dialect between Western and Eastern Ejagham (see Watters (1981: 8–16) for details). The language has a written form that has gained some use in schools and churches in Cameroon but only limited in Nigeria.

2 Typological considerations Word order: The unmarked active word order in the Ejagham language is S-V-O. The subject prefix of the verb agrees with the noun class or the person and number of the head noun phrase. In the case of (1a) the subject prefix a- on the verb root yám ‘cook’ agrees with the noun class of the subject noun phrase ǹkáé òmé ‘my wife’. The same subject prefix is used with the independent subject pronoun yê ‘3s’ in (1b) where person and number is indicated. If neither a subject noun phrase nor a subject independent pronoun is present, the subject prefix is still required in (1c). The perfective ‘PFV’ in the examples in (1) is indicated by the tone – see Watters (2012) for details. (1) The S-V-O word order S________ V_______ O_______ a. Ǹkáé òmé àyâm édî ájì.1 1.wife 1.my 1.pfv.cook 5.food that.5 ‘My wife cooked that food.’ S________ V_______ O_______ b. Yê àyâm édî ájì. 3s 3s.pfv.cook 5.food that.5 ‘She/he cooked that food.’

1 Examples in Ejagham provide the consonants and vowels in the orthographic form used in ­Cameroon. In terms of tone, all tones are marked even though orthographically only high tones and falling tones are marked. Unless otherwise commented on in terms of the name of the transcribed text and the author of the oral text, examples are constructed from phrases and clauses found in hundreds of pages of transcribed oral texts and translated texts, as well as checking modified phrases and clauses with speakers of Ejagham.



6 Relative clause in (Western) Ejagham 

S________ V_______ O_______ c. [no explicit NP] Àyâm édî 1/3s.pfv.cook 5.food ‘She/he cooked that food.’

 121

ájì. that.5

With verbal forms that are not inflected for noun class or person and number, such as the infinitive, the object may optionally precede the infinitive as with èyámám ‘cooking’ in (2). Example (2a) demonstrates the S-V-[V-O] order, while example (2b) S-V-[O-V] shows the optional order in which the object precedes the infinitive. (2) The Object preceding verbal form not inflected for person and number S________ [V_______ˉ [V_______ˉ O______ˉ ] a. Ǹkáé òmé àróghè èyámám édî ájì 1.wife 1.my 1.pfv.begin 5.cooking 5.food that.52 S________ V_______ [O________ V_____] b. Ǹkáé òmé àróghè édî ájì èyámám 1.wife 1.my 1.pfv.begin 5.food that.5 5.cooking ‘My wife began cooking that food.’ The primary object (e.g. recipient) immediately follows the verb as in (3), while the secondary object (the theme) follows the primary object, producing an S-VPO-SO word order3.

2 For all noun classes the subject prefix on the verb will carry the noun class number of the subject. Besides prefixes on verbs, concord with the given noun class is marked with a prefix on possessive pronouns/adjectives and on numerals; it is marked with a suffix on demonstratives and relativizers; and as in infix on object pronouns and locative adjectival interrogatives. Using noun class 5 as an example, the following forms are used as affixes: Prefixes: Verb é-, Possessive èjˊ-, Numeral jə́ Suffixes: Demonstrative and Relativizer –ji (tone determined by the type of demonstrative or relativizer) Infixes: Object and Locative Interrogative –jə́ The noun classes in Ejagham, using the Bantu numbering system, are 1, 2, 3, 5, 6, 8, 9, 14 and 19. At least one noun is a residual form of class 6a and the plural formations for class 9 suggest class 10 is used for nouns of class 9 when they are counted but class 14 when they are a generalized plural. The abbreviations for verbal moods and aspects as well as other morphological categories are found in the main list. 3 In this study, direct objects are referred to as ‘secondary objects’ and indirect objects as ‘primary objects.’ The use of ‘primary’ and ‘secondary’ does not refer to levels of importance but simply their relative order following the verb when there is a double object construction. So ‘primary objects’ would be recipients and benefactives while ‘secondary objects’ would be themes.

122 

 John R. Watters

(3) The Secondary Object immediately following the verb S________ V___ PO__ SO__ Ǹtèm òmé àkárè ábhǒ édî 1.friend 1.my 1.pfv.give them 5.food ‘My friend gave them food.’ Otherwise the recipient or benefactive can be expressed either as an oblique as in (4) following the preposition m̀bâ or as an object of the second verb in a serial clause construction involving the verb karê ‘to give’ as in (5). Obliques generally follow the verb and objects as in (4). (4) Benefactive as adjunct following the preposition m̀bâ S_________ V_________ O________ Oblique____ Ǹtèm òmé àkárè édî ájì m̀bâ ábhǒ 1.friend 1.my 1.pfv.gave 5.food that.5 to them ‘My friend gave that food to them.’ (5) Benefactive as the object of the second verb in serial construction S________ V1________ O_______ V2________ O___ Ǹtèm òmé àkô édî ájì àkárè ábhǒ 1.friend 1.my 1.pfv.took 5.food that.5 1.pfv.gave them ‘My friend took that food and gave it to them.’ Unlike the subject that is prefixed to the verb, Ejagham does not affix object pronominal references to the verb. Identifying grammatical relations: As should be evident from examples (1) through (5) there is no morphological marking on nouns of grammatical relations within the sentence. Instead, grammatical relations are identified with fixed syntactic positions. In addition, Ejagham does not have a passive or circumstantial voice, meaning that direct objects, indirect objects and obliques cannot be made subjects and the subject cannot be made into an oblique. Ejagham uses the third person plural verbal prefix but with non-specific reference as the equivalent of the passive construction. The verb, in all cases but the imperative, is marked with an initial subject prefix. No other grammatical relation is marked by affixes in the verb. Noun class system: The pronominal system is an integrated part of the noun class system (Watters 1980, 1981). Depending on the preferred analysis and dialect, Ejagham has nine to eleven noun classes, and residual forms from another possible two. The following pronouns participate in the noun class system: subject, object, possessive, demonstrative, and interrogative. Similar to these pronouns the markers used to indicate relative clauses participate in the noun class system as

6 Relative clause in (Western) Ejagham 



 123

well. Like pronouns they agree in noun class with the noun phrase to which they refer. However, these relative markers are not pronominal in nature. They do not refer to any grammatical relation or semantic category that the head noun phrase has within the relative clause. We may refer to them as ‘relativizers’ with noun class features that agree with head noun phrase (cf. Payne 1997: 326, 333–335). More information on relevant forms will be given in section 3. Adjectives: It is not uncommon in the analysis of relative clauses to find analogies with adjectives and adjectival phrases and their relationship with nouns. Relative clauses are often positioned like adjectives in relation to nouns. In Ejagham, numerals as well as possessive and demonstrative adjectives follow the head noun as expected for an S-V-O language as in (6). (6) Adjectives follow the head noun à-nè àbh-ə́ ré á-bháʹé á-bhà

2-people

2-1p.poss

2-two

dem.distal-2

‘Those two people of ours’ (lit: ‘people our two those’) In (6) the possessive adjective àbhə́ ré ‘our’, the numeral ábháʹé ‘two’, and the demonstrative adjective ábhà ‘those’ all follow the head noun ànè ‘people’ in a set order. However, beyond the set of adjectives just mentioned, most common adjectival notions are expressed by verbal forms and, in some instances, by nominal forms.

3 Relative clause structure In this study a relative clause is an adjectival subordinate clause that modifies a noun phrase, whether explicit or implicit, by narrowing the particular referent of the noun phrase to a more specific instance or set of instances of all the referents to which the noun phrase could refer (see Keenan (1985: 141–143), Keenan and Comrie (1977) and (1979: 653), Payne (1997: 325), and Andrews (2007: 206) for other similar definitions). In this section the focus will be on the structure of RCs (3.1) and the possible variations in their structure (3.2).

3.1 The structure of RCs Relative to the structure of RCs in Ejagham the components that make up the RC will be noted. The distinction between restrictive and non-restrictive RCs will also

124 

 John R. Watters

be commented on. First, consider the examples in (7). Square brackets will be used to mark the boundaries of the RCs. (7) Structure of RCs a. … á-chə́ bhè ǹ-jə̀ m á-kʉ̀ r-á nà é-dî ǹ-jǐ á-nám-ʹé. … á-chə́ bhè ǹ-jə̀ m á-kʉ̀ r-á nà é-dî …2.pfv-turn 9-back 2.hab-return-impfv with 5-food [ǹ-jǐ á-nám-ʹé ] rel-5 2.pfv-buy-ConstF ‘…they turned back returning with the food which they bought’ (Ojong Tanyi, Njombui na Mgbe) b. … à-chî (ˊ ) ǹ-ték à-ní-g-ìsá m̀-mǎ á-kúmʹí. … à-chî (ˊ ) ǹ-ték à-ní-g-ìsá [m̀-mǎ á-kúm-ʹí] … 6-head of 9-village 2-four-plus-three rel-6 6.pfv-sit-ConstF ‘… the seven village heads that met.’ (Ayamba Nkom Moses, Otui Nkom Ekaŋ) The head NP that the RC modifies in (7a) is édî ‘food’ and in (7b) it is àchî ǹték ànígìsá ‘the seven village heads’. From these facts we note that the RC follows the NP it modifies. It is also embedded in the larger NP that includes the head NP and the RC. So the RC is generally embedded and post-nominal. The next thing of note is that Ejagham uses a relativizer, ǹjǐ in (7a) and m̀mǎ in (7b). These are not relative pronouns since they do not indicate any grammatical relation within the RC. Instead, they serve as a subordinator between the head NP and the RC, indicating the beginning of the RC. The relativizer agrees with the noun class of head NP. Thus, in (7a) the head NP is édî and belongs to noun class 5. This is indicated by the prefix é-. The relativizer agrees, consisting of the relativizer root ń- and the class 5 agreement affix -jǐ, producing ńjǐ. In (7b) the NP is àchî ǹték ànígìsá … with àchî as the head of the NP. It belongs to noun class 6 as indicated by the prefix à-. In this case the relativizer root m̀- takes the class 6 agreement affix –mǎ, producing m̀mǎ. More abstractly, the relativizer consists of a low tone nasal followed by a floating low tone. We can represent this as Ǹ ˋ-. This nasal relativizer morpheme assimilates to point of articulation of the consonant of the agreement affix. The variation can be seen in (8). In (8) each relativizer has an agreement affix. The affix is marked by low tone in the case of noun classes 1 and 9 and a rising tone in all other classes. The low tone of 1 and 9 derives from the low tone on the affix which combines with the floating low tone of the relativizer. All other affixes in (8) take a high tone. This high tone then combines with the floating low tone of the relativizer to form a rising tone on the affix.



6 Relative clause in (Western) Ejagham 

 125

(8) Relativizers for the nine Ejagham noun classes4 1 ǹ-nyò 8 m̀-bǐ 2 m̀-bǎ 9 ǹ-nyì 3 m̀-mǐ 14 m̀-bǐ 5 ǹ-jǐ 19 m̀-fǐ 6 m̀-mǎ Finally, relative to the constituents within the RC, the normal word order holds. However, there is one modification. This involves the form of the verb when inflected for the perfective or imperfective aspects in the RC. See Watters (2010) for extensive discussion of this alternation. In (9) the alternation with the perfective is exemplified and in (10) that with the imperfective. (9) Alternation in perfective verb forms a. ǹ-tèm ò-mé à-nâm bì-yù 1-friend 1-my 1.pfv-buy 8-yam ‘My friend bought yams’ b. ǹ-tèḿ ò-me ǹ-nyò à-nàm-éʹˊ 1-friend 1-my rel-1 1.pfv-buy-ConstF ‘My friend who bought yams’ (10) Alternation in imperfective verb forms a. ǹ-tèm ò-mé à-nàm-á bì-yù 1-friend 1-my 1.hab-buy-impfv 8-yam ‘my friend buys yams’ b. ǹ-tèm ò-mé ǹ-nyò á-nàm-á 1-friend 1-my rel-1 1.ConstF-buy-impfv ‘my friend who buys yams’

bì-yù 8-yam

bì-yù 8-yam

The perfective verb form in (9a) à-nâm ‘she bought’ differs from the form in (9b) à-nàm-éʹˊ ‘she bought’. In (9b) the verb in the RC is suffixed whereas in (9a) where the verb is in the main clause no suffix is present. Similarly in (10a) à-nàm-á ‘she buys’ differs from á-nàm-á ‘she buys’ in the RC in (10b). The difference is in the prefix tone, with the habitual in (10a) taking low tone on the prefix and the imperfective in the RC in (10b) taking high tone on the prefix. The forms used in the RCs (9b) and (10b) are morphologically marked for ‘constituent focus’

4 The numbers used for noun classes in this article are those used for Bantu languages and serve comparative purposes in understanding how the Ejagham noun classes relate to noun classes found in hundreds of other Bantoid languages.

126 

 John R. Watters

or ‘ConstF’. These forms are used in RCs, with interrogative words and their answers, and so on. Given the structural realities of RCs one final comment should be made concerning the distinction between restrictive and nonrestrictive RCs. Ejagham does not make a formal difference between restrictive and nonrestrictive RCs. Generally RCs are restrictive, and nonrestrictive information is provided in main clauses. However, RCs may be used with nonrestrictive readings. The difference between the restrictive and nonrestrictive reading usually depends on semantic and pragmatic properties of the RC in context5. For example: (11) Restrictive vs. Nonrestrictive readings a. Ǹ-tèm ò-mé ǹ-nyò à-gùr-íʹˊ 1-friend 1-1s.poss rel-1 1.pfv-sell-ConstF à-kí-gyèbhé ḿ-bə́ ŋé 1-cont-be.sweet 9-heart ‘My friend who sold two goats is happy.’

m̀-búì 9-goat

é-bháʹé 9-two

i. RESTRICTIVE reading: ‘My friend who sold two goats is happy.’ ii. NONRESTRICTIVE reading: ‘My friend, who sold two goats, is happy. b. Ǹ-tèm ò-mé à-gúrì m̀-búì 1-friend 1-1s.poss 1.pfv-sell 9-goat ḿ-bə́ ŋé 9-heart ‘My friend sold two goats and is happy.’

é-bháʹé 9-two

à-kí-gyèbhé 1-cont-be.sweet

The restrictive reading of (11a.i) is the common situation in which the friend under discussion is identified from among the entire set of friends as the one ‘who sold two goats’. The set of friends is restricted to the one who sold two goats. However, if the speaker is in the market and gestures toward his friend nearby as he tells the hearer ‘my friend, who sold two goats, is happy’ (11a.ii), there is not a set of friends involved. The friend is uniquely identified in the context, pragmatically, through the gesture in the marketplace. The phrase, ‘who sold two goats’,

5 Given the requirements on the size of articles written for volumes such as this, other aspects of RCs have to remain unreported. Further research and reports in the future would include 1) discourse use of RCs in various genres, 2) the use of pragmatic and semantic dimensions in the interpretation of RCs as either restrictive or non-restrictive, 3) the semantics and syntax of relativization in close knit serial clauses (auxiliary verb plus content verb) versus general serial clause constructions, and 4) the nature of adjectival and stative expressions and their relation to RCs.



6 Relative clause in (Western) Ejagham 

 127

provides information about the friend but does not uniquely identify him. This kind of additional information, however, would more likely be provided in a main clause as in (11b). Or consider the example in (12) (12) Restrictive vs. Nonrestrictive readings a. Ǹ-nyò àmé ǹ-nyì à-nʉ̀ r-íʹˊ 9-mouth 9.my rel-9 9.pfv-swell-ConstF ə̂ m ò-sèŋè me 14-trouble

ò-nʉ̌ t 14-swell

à-kí-kàré 9-cont-give

i. RESTRICTIVE reading: *‘My mouth which is swollen is giving me trouble.’ ii. NONRESTRICTIVE reading: ‘My mouth, which is swollen, is giving me trouble.’ b. Ǹ-nyò àmé à-nʉ̂ t ò-nʉ̌ t à-kí-kàré 9-mouth 9.my 9.pfv-swell 14-swell 9-cont-give ‘My mouth is swollen and is giving me trouble.’

ə̂ m me

ò-sèŋè 14-trouble

The restrictive reading in (12a.i) is not possible since the set of mouths is already semantically restricted by the possessive pronoun ‘my’, and each individual has only one mouth. So the nonrestrictive reading (12a.ii) is the only possible reading, with the relative clause providing symptomatic information as to why the speaker is having trouble with his mouth. Again, this additional information about the mouth being swollen would most likely be provided in a main clause as in (12b). To summarize, the general facts about the Ejagham RC can be presented with the abstract schema in (13). The shorter arrow indicates that the ‘Head NP’ controls the noun class affix on the relativizer (REL). The longer arrow indicates that the ‘Head NP’ is co-referential with an NP within the relative clause, this NP indicated by the variable x. VerbConstF represents the requirement that perfective and imperfective aspects use constituent focus verb forms in relative clauses. (13) [NP [Head NP [RC = REL Clause: (VerbConstF´) (x) ]]] The relative clause functions as a modifier of the head NP. In most cases it restricts the set of referents that the nominal refers to, but it can at times also serve in an appositional relation to the head NP as a nonrestrictive RC as permitted by the semantics and pragmatics of the context.

128 

 John R. Watters

3.2 Variations in the structure of relative clauses In this section variants on this general structure will be presented. Some are cases in which the head NP is absent (3.2.1). Others involve the absence of the relativizer (3.2.2). Others involve a different set of relativizers that connect the head NP to a locative expression or adjectival expression (3.2.3). Others involve the possibility of coordinating RCs (3.2.4). Still others show the relative clause displaced, no longer embedded but separated from what would be the head NP yet still agreeing with it in its relativizer (3.2.5).

3.2.1 Headless relative clauses In some cases an RC does not have an explicit head NP. It remains implicit, to be inferred from the context. These constructions require further study that would take us beyond the scope of this more general study of RCs, but there are at least two grammatical dynamics associated with these headless RCs that should be mentioned. First, in some cases the relativizer in these RCs functions as an introductory marker of adverbial or nominal clauses, indicating either time or the identity of a sentential argument. Second, the relativizer of headless RCs may function as an anaphoric pronoun with possible interpretations such as ‘thatwhich,’ ‘those-who,’ ‘time-when,’ and so on. In the first case, the RCs involve nouns of broad categories such as èbhú ‘time’ and èjûm ‘thing’ – time and thing (i.e. content or identity of an argument). Both of these belong to noun class 5 and would use the relativizer ǹjǐ to introduce RCs. These account for the majority of headless RCs. The most frequent case in recorded texts is that which refers to ‘time’ èbhú. The next is that referring to ‘thing,’ èjûm. The use of the headless RC that refers to ‘time’ commonly serves the role of introducing a temporal adverbial clause. This role is demonstrated in (14a) and (14b). In (14a) the temporal adverbial clause, introduced by the relativizer, is to be interpreted as ‘when they began traveling,’ and in (14b) it is ‘when the mother saw that the child had matured.’ (14) Implicit head NP referring to time a. Mân [ǹjǐ á-rógh-ʹé è-jí-ím], ǹ-kʉ́ í à-sə́ rè Thus [rel.5 2 ˜3p.pfv-begin-ConstF 5-go-inf], 9-tortoise 9.pfv-say:to ‘Thus (at the time) [when they began traveling], Tortoise said to É-tághárá Ǹkʉ̀ nkʉ̀ sè

5-spider

Spider that …’

comp.2s/3s

(Agbor Flavius Ntui, Etaghara Nkunku na Nkʉ́ í)

6 Relative clause in (Western) Ejagham 



 129

b. [Ǹjǐ ǹ-nyěn à-yén-ʹé sè m̀-mǒn á-nyò [rel.5 1-mother 1.pfv-see-ConstF comp.2s/3s 1-child dem.distal-1 à-kúì ],

1.pfv-mature

‘[When the mother saw that that child had matured], á-fî yê á-kîm ǹ-kîm. 2.pfv-seize 3s 2.pfv-circumcise 9-circumcision they seized her and circumcised her.’ (Ojong Tanyi, Ayikobhi na Ntikpoŋ) Another common inference with headless RCs involves the noun èjûm ‘thing.’ In an RC whose head is a word with very general reference like èjûm ‘thing,’ the real semantic content identifying the referent is found in the clause following the relativizer. Thus, in the case of headless RCs involving èjûm ‘thing’ as the head of the NP, it is as though semantically the relativizer plus following clause substitute for the semantically general noun èjûm ‘thing,’ acting as a complex pronoun. In (15) this ‘complex pronoun’ (i.e. headless RC) serves as the object of a verb of perception. In context, (15) refers to what the small bird saw the woman do to the child of her co-wife down by the river. (15) Implicit head NP referring to thing (here ‘content’ of perception) M̀-mǒn ì ì-nón è-tǎt à-rì á-f-ò

1-child

of

19-bird

5-other

1-be

dist1-loc.spec-dist

‘Some small bird was there, à-yén-á [ǹjǐ yê á-yìm-á] 1.hab-see-impfv [rel.5 3s 1.hab-do-ConstF.impfv] watching [what she (the woman) was doing].’ (Anonymous, Egan Afughá-osí (The Story of Catching Fish) Turning to the second grammatical development of headless RCs in which they may function as anaphoric pronouns with possible interpretations such as ‘that-which’ or ‘those-who,’ consider the examples in (16). In (16a) the headless RC serves as the object of the verb ’to say to.’ The relativizer belongs to noun class 2, the plural of persons. Thus, the meaning could be interpreted as ‘those-who’ are involved in the situation. It refers back to a likely reference to those at the market, in this case ‘those-who are selling,’ those who are “sellers” or “habitually sell.” When there is not a specific object it is common to use a cognate object like ògùrí ‘to sell’ as used in (16a). However, (16b) shows that a specific object like bìyù ‘yams’ can also be used to specify “habitual sellers of yams.”

130 

 John R. Watters

(16) a. Á-sə́ rè [m̀-bǎ á-gúr-ʹá ò-gùrí …]6 1.pfv-said.to [rel-2 2.hab-sell-impfv 14-sell …] ‘He said to [those-who were selling …] b. Á-sə́ rè [m̀-bǎ á-gúr-ʹá bì-yù …] 1.pfv-said.to [rel-2 2.hab-sell-impfv 8-yams …] ‘He said to [those-who were selling yams …] c. [M̀-bǎ ǹ-g-á ] á-chə̀ bhè nǒŋ à-bhǒn [rel-2 prox-loc.gen-prox] 2.prf-become like 2-children à-sʉ́ ŋ àbh-ê

2-slave

2-poss.3

‘[These-ones here] became like her slaves.’ (Ojong Tanyi, Asiraniran, Ékpímí é nôp akûk, Itəme í nôp mbâk na Okpere ó nôp osêŋ) d. Sè yén ên, [ǹ-nyì è-jî-kúmì, êt à-túófàm, comp.2s/3s see.imp 2p, [rel-9 1p.pfv-go-sit, 1p 2-chiefs ‘Look all of you, [the-matters-which we went to meet about, we chiefs kǎ à-chî (í) ǹ-ték à-nígìsá kǎ Eyumojok].

at

6-head

of

9-village

6-four.and.three

at

Eyumojok].

from seven villages at Eyumojok].’ (Moses Ayamba Nkom, Otui Nkom Ekaŋ) Example (16c) shows the same relativizer from class 2 in the subject position, in this case referring anaphorically to ‘these-who’ are in this “proximate” location, i.e. ‘here.’ Finally, in (16d) the headless RC uses a relativizer from noun class 9. There is no preceding nominal for this relativizer to refer to. Instead, it refers to the preceding situation in which the chiefs from seven villages came together to decide how the next paramount chief would be identified. The noun class 9 relativizer does not refer back to the “meeting” but to the “matter” discussed. It refers to the general noun ǹchǒt ‘matter, discussion, disagreement.’ So the interpretation used in (16d) is to ‘the-matters-which’. Thus, the headless RCs can have an anaphoric function in which

6 Agentive nominals are derived from such verbal phrases. The difference is that in the formation of the nominal forms the –a suffix is a NOM(inalizer) rather than the exponent of the HAB(itual) aspect. The tone of the NOM is determined by the noun class. Noun classes 1 and 9 require a low tone –à (a), while other classes require a high tone -á (b). (See Watters 1981. 496–500 for more details). a. m̀-kpàŋ-à è-bhǐn ‘farmer’ 1-hoe-nom 5-farm b. à-jə̀ m-á ḿ-fô ‘tailors’ 2-sew-nom 9-cloth

6 Relative clause in (Western) Ejagham 



 131

what would be the head noun with general reference is absent. All that is needed is the relativizer which signals which general noun in context is being referenced anaphorically, and the remainder of the RC provides the relevant information.

3.2.2 Relative clauses with absent relativizer Another variation on the standard RC structure involves RCs without the relativizer present. One important grammatical fact that makes the RC retrievable in such cases is that the verb form of the perfective and imperfective in the RC takes a different form than it has in main clauses. This redundancy allows the RC to be identifiable as an RC in most cases. It is not as easily retrievable if it involves a different aspect or mood. It is not common practice to construct RCs without a relativizer. However, there are some specific circumstances in which the relativizer is absent. Most of these are stylistic choices in that they are not obligatory. But some are strong ­stylistic preferences. First the construction types in which relativizers are more likely to be absent will be presented. Then some relevant nouns involved in the optional absence of relativizers will be discussed. One of the most common constructions where relativizers are absent is in the formulaic summary or closing statements in narratives as in (17a) where the focus marker (FOC) nâ is used, or in (17b) where the formula is introduced with the imperative yén of the verb ‘to see.’ In fact, the focus marker is nearly in a complementary relationship with the relativizer.The underlined position marks the position where the relativizer would usually appear. (17) Absent relativizer a. Nâ è-jûm [___ é-yím-ʹí ánò ] à-kə̂ t nà é-bháré foc 5-thing [___ 5.pfv-do-ConstF thus] 1.pfv-remain with 5-white ‘That is how [it came about in this way] it (hornbill) remained with white é-ngbáŋ kǎ ǹ-gàŋ.

5-plate

at

9-chest.

plates on his chest.’

(Agbor Flavius Ntui, Etaghara Nkunku na Nkʉ́ í ) b. Yén è-jûm [___ á-tóʹé ánò ] sê kpè See.imp 5-thing [___ 2.pfv-say-ConstF thus] comp.pl even ò-kí-jǐ è-bhǐn 2s-cont-go 5-farm ‘See the reason [they say thus] that even if you are going to the farm

132 

 John R. Watters nà m̀-mǒn òw-ǎ, á-tʹó sè ǹ-kí-yʹáré with 1-child 1-poss.2s, 1.cond-say comp.2s/3s 1s-cont-turn.off e-kú, 5-latrine with your child, if he/she says that I am turning off to the latrine, é-nóbhʹé sê ò-gímì áfò 5.pfv-good.ConstF comp.pl 2s.pfv-stand there ò-chìghá yê. 2s.hab-wait.impfv 3s. it is good that you stand there waiting for her/him.’ (Ms. Ojong Tanyi, Ayikobhi na Ntikpoŋ)

Another construction where the relativizer is absent is in the dependent clause that uses the repetitive dependent verbal prefix kpò-. In this case it is obligatorily absent. An example is given in (18). The underlining marks where we might expect to find the relativizer if it were to appear. ___ (18) È-bhú á-ji 5-time dem.Distal-5 ___ ‘When that time arrives, …’

é-kpó-s'í aŋ, …

5-rep.dep-reach ang, … (Moses Ayamba Nkom, Otúi Nkom Ekaŋ)

The remaining occurrences involve the general nouns already mentioned above: namely, ebhú ‘time.’ èjûm ’thing,’ èjǐ ‘place,’ and èkpàk ‘type, kind, manner.’ Examples of all of these without relativizers are present, but the one of note is èbhú ‘time.’ It occurs frequently without the relativizer. This fact sets up a complementary relationship in the case of èbhú ‘time,’ between headless RCs where the implicit head NP is èbhú and RCs where èbhú is explicit but the relativizer is absent. There are many occurrences where both the explicit head NP èbhú and the relativizer are present, but also many where one or the other is absent.

3.2.3 Relative clauses with different relativizers The standard relativizer in Ejagham has a concording form for each noun class as listed in (8). This consists of the relativizer root Ǹ ˋ -suffixed by an exponent for the relevant noun class. These standard forms can be seen below in the first column of Table 1 as Type I Relativizer. However, Tab. 1 specifies other relativizers that are used apart from the standard one.

6 Relative clause in (Western) Ejagham 



 133

Tab. 1: Three Types of Ejagham Relativizers Noun class

1 2 3 5 6 8 9 14 19

Type I: Standard ǹ- ˋ -CV̀ (1 & 9) ˜ -CV́

Type II: Locative ǹ-ˋ -CV́

Type III: Adjectival ǹ-ˋ -CV̀

ǹ-nyò m̀-bǎ m̀-mǐ ǹ-jǐ m̀-mǎ m̀-bǐ ǹ-nyì m̀-bǐ m̀-fǐ

ǹ-nyǒ m̀-bǎ m̀-mǐ ǹ-jǐ m̀-mǎ m̀-bǐ ǹ-nyǐ m̀-bǐ m̀-fǐ

ǹ-nyò m̀-bà m̀-mì ǹ-jì m̀-mà m̀-bì ǹ-nyì m̀-bì m̀-f ì

In the second column is the Type II Relativizer. This relativizer is followed by a locative expression rather than a clause with an inflected verb. The locative expression (see Watters 1981: 266ff) involves a locative pronoun of specific location, namely, m̀fá ‘(right) here’, áfò ‘(right) there’, or ḿfô ‘way over there’; or a locative pronoun of general location, namely, ǹgá ‘nearby’, ágò ‘there (in that direction)’ or ńgô ‘way over there (in that direction)’. The Locative Relativizer differs from the standard form paradigmatically. Unlike the standard relativizer, the Type II relativizer neutralizes the tone distinction between noun classes 1 and 9 on one hand and all other classes. The agreement affix has a high tone in every class as seen in the rising tone throughout the paradigm. In the third column is the Type III Relativizer. This one is followed by an adjectival expression rather than a full clause with an inflected verb. These forms differ from the other types by tone. Like the Type II Relativizer, the distinction between noun classes 1 and 9 and all other noun classes found in Type I is again neutralized. In this case all agreement affixes take a low tone that merges with the floating low tone of the relativizer root to produce a final low tone on the class exponent. Adjective roots used with this Type III include adjectival notions such as nò ‘good’, bí ‘bad, wild’, tǎt ‘other, different’, and fê ‘new’ as well as those expressed by gerunds from class 14. They also include adjective stems derived from infinitives of active verbs with the class 5 prefix è- of the infinitive deleted as in (19). (19) Deriving adjectives from infinitives a. è-gùr-úm ‘to sell’ –> gùr-úm ‘for sale’ b. è-yám-ám ‘to cook’ –> yám-ám ‘for cooking’

134 

 John R. Watters

Turning to some examples of these second and third types of relativizers, (20) demonstrates the use of the Locative Relativizer. (20) Locative Relativizer à-bhə̂ p à-bhǒn [m̀-bǎ ǹ-g-á] … 1.pfv-ask 2-child [rel.loc-2 here-general.location-near] … ‘She asked those children who were nearby …’ (Ojong Tanyi, Asiraniran, Ekpimi Enop Akuk, Itəme Inop Mbak, and Opkere Onop Oseŋ) It is common for RCs using the Locative Relativizer to be headless. This is likely due to the fact that the head noun phrase usually refers to a physical location such as a house or a road, or to a person or persons. Thus, many involve agreement for classes 1 and 2 (humans) and classes 5 and 9 (common locations). In (21a) the implicit head is known from the context to be the word m̀bá ‘road’ of noun class 9 as indicated by the Locative Relativizer ǹnyǐ, with class 9 suffix. Without the context, ǹnyǐ could refer the location of ǹgòn ‘hornbill’ which is also a member of class 9. However, in this story it is the mother who is in the farm, not ‘hornbill.’ In (21b) the Locative Relativizer m̀bǎ has the class 2 suffix related to plural of persons. In the story from which this RC is taken, the implicit head NP is likely the three sisters that were caring for the favored fourth sister. The locative in this case has an anaphoric sense, referring to the previously mentioned sisters. (21) Locative Relativizer a. à-tə̂ m fə́ nè à-rúì ǹ-gyìm 1.pfv-bend.knees consequently 1.pfv-get.up 9-standing á-kí-jǐ nà 2-cont-go with ǹ-gòn [ǹ-nyǐ è-bhǐn] 9-hornbill [rel.loc-9 5-farm] ‘She consequently got up and was going with Hornbill to the farm. (Implicitly: (on the road which goes) to the farm)’ (Ojong Tanyi, Asiraniran, Ekpimi Enop Akuk, Itəme Inop Mbak, and Opkere Onop Oseŋ) b. [m̀-bǎ ǹ-g-á] á-chə̀ bhè nǒŋ [rel.loc-2 here-general:location-near] 2.prf-become like à-bhǒn (ˊ ) à-sʉ́ ŋ àbh-ê 2-child of 2-slave 2-3s.poss ‘Those (sisters) who were mentioned (literally: who were nearby) became like her slaves.’



6 Relative clause in (Western) Ejagham 

 135

Turning to examples of the Adjectival Relativizer consider the examples in (22). (22) Adjectival Relativizer a. Adjective à-nè [m̀-bà tǎt] á-bhà 2-person [rel.Adjectival-2 other] 2.prf-come ‘Some other people have come (lit: people who (are) other have come). b. Gerund ḿ-bà è-gàn [ǹ-jì ò-túp] 1ps.prf-come 5-story [rel.Adjectival-5 14-tell (lit: ‘throw’)] ‘I have come to tell a story.’ (Literally: ‘I have come story which to throw.’) c. Adjective m̀-mǒn [ǹ-nyò fê] ‘a newborn (lit: ‘a child that (is) new’)’ d. Infinitive stem è-yù [ǹ-jì gúr-úm] ‘a yam for sale (lit: ‘a yam which (for) selling)’ Both the Type II Locative Relativizer and the Type III Adjectival Relativizer forms could be considered minimal stative clauses by analogy to the different forms of the copula ‘to be’ in Ejagham. The copula of location is basically à-rǐ (or à-rě) with a low tone prefix and a rising tone on the root as in (23a). The Locative Relativizer uses a similar low tone followed by a rising tone on the relativizer in (23b). (23) Locative Relativizer and Copula of Location a. ǹ-nè à-rě m̀-fá 1-person 1-be loc-prox:specific ‘A person is here’ b. ǹ-nè ǹ-nyǒ m̀-fá 1-person rel-1:loc loc-prox:specific ‘A person who is here’ The copula of identity in (24a) uses all low tones as does the Adjectival Relativizer in (24b). (24) Adjectival Relativizer and Copula of Identity a. ǹ-tûfâm à-rè ǹ-nyěn ì ǹ-nè 1-chief 1-be 1-mother of 1-person ‘The chief is a good person’ b. ǹ-tûfâm ǹ-nyò nò 1-chief rel-1:adj good ‘The chief who is good’

136 

 John R. Watters

The locative and adjectival relativizers introduce clauses without inflected verbs but function as minimal clauses analogous to the copula. In addition, similar to the standard relativizer these forms serve to further specify the referent of the head NP from among all its possible referents.

3.2.4 Coordinate relative clauses Relative clauses may stand in a coordinate relationship. Example (24) involves an adjectival RC coordinated with a standard clausal RC. (25) Coordinated RCs Ǹjòk ǹnyì yúp ǹnyì àmə̂ ŋ á kí f ʹáp ábháʹé … Ǹ-jòk [ǹ-nyì yúp ] [ǹ-nyì à-mə̂ ŋ 9-elephant rel.Adjectival-9 large rel-9 6-teeth á-bháʹé ] … 6-two ‘A large elephant with two tusks …’ (Moses Ayamba Nkom of Eyumojok, Otui Nkom Ekam)

à-kí-f ʹáp 6-cont-come.out

3.2.5 Non-embedded relative clauses - displaced Another variation on the standard relative clause occurs when the RC is no longer embedded within the NP with the head NP but is displaced to another position in the main clause. In (26) the RC occurs after the main verb ósì ‘had arrived’ while it modifies the subject NP òfú ‘day’. (26) Displaced relative clause Kə̀ n ò-fú ó-sì

before

14-day

14:prf-reach

[m̀-bǐ

rel-14

á-kí-jǐ

2-cont-go

è-jèn], …

5-journey,

‘Before the day had arrived on which they were going to travel, …’

Given the noun class exponents, it is clear that the relativizer m̀bǐ ‘REL’ agrees with òfú ‘day’ even though the main verb intervenes. A similar situation holds in (27). (27) Displaced relative clause … sè è-ték è-tǎt

é-rî

[ǹ-jǐ

é-dî

cháŋ]

… comp.2s/3s 5-village 5-other 5-be rel-5 -food 5 neg:Existence ‘… that another village is present where there is no food.’

The relativizer ǹjǐ from noun class 5 agrees with the subject èték ‘village’ even though the main verb érî ‘is’ intervenes. Such displacements of relative clauses

6 Relative clause in (Western) Ejagham 



 137

often occur in relation to the subject of a sentence with a stative or intransitive verb. The principle seems to be the one of placing “heavier” information later in the sentence if possible. The goal of further specifying the referent of the head NP is still achieved even if the RC is no longer embedded in the NP.

4 Strategies in relativization In this section we consider the strategies used in Ejagham to form RCs. These strategies are entirely noun-coding ones in that the verb is not marked in any way for grammatical relations except the usual agreement of the verb prefix with the subject. Ejagham uses two different strategies for relativization. The first strategy deletes the NP that is co-referential with the head NP of the RC. Consider (28). (28) Example sentence to demonstrate the deletion/gap strategy of relativization ǹ-tèm ò-mé à-kárè à-jèné é-dî

1-friend

1-1s.poss

1.pfv-give

2-stranger

‘My friend gave (some) strangers food.’

5-food

In (28) the sentence has three NPs: ǹtèm òmé ‘my friend’, àjèné ‘strangers’, and édî ‘food’. Each of these NPs can be relativized in Ejagham. This is demonstrated in (29 a, b, c). (29) a. Subject relativization ǹ-tèm ò-mé [ǹ-nyò

1-friend



rel-1

à-kàréʹˊ

1.pfv-give

‘my friend who gave some strangers food’ b. Secondary object relativization é-dî [ǹ-jǐ ǹ-tèm ò-mé à-kàréʹ

5-food



1-1s.poss

__

rel-5

1-friend

1-1s.poss

à-jèné

2-stranger

1.pfv-give

‘food that my friend gave to some strangers’ c. Primary object relativization à-jèné [m̀-bǎ ǹ-tèm ò-mé à-kàréʹˊ

2-strangers

rel-2

1-friend

1-1s.poss

‘strangers to whom my friend gave food’

é-dî]

5-food

à-jèné __ ]

2-strangers

1.pfv-gave

__

é-dî ] 5-food

In each case in (29) the NP that becomes the head NP leaves a gap inside the relative clause. This is indicated in each case (29 a, b, c) with a marked space. This strategy can be called the “gap” or “deletion” strategy. The head NP is co-­referential with the NP that would appear in the indicated space as subject in  (29a), secondary object in (29b) and primary object in (29c). This strategy is

138 

 John R. Watters

required when relativizing subjects and secondary objects (themes). It is optional with primary objects (recipients and benefactives). The second strategy that Ejagham uses does not delete the relativized NP but instead leaves a trace in the form of a resumptive pronoun. This is demonstrated in (30). In (30a) the final NP is an adjunct and is the object of the preposition nà ‘with’. This NP then becomes the head of relative clause in (30b). (30) a. Example sentence of the trace/resumption pronoun strategy of relativization Obi à-yîm è-tûm nà ǹ-tèm òw-ě

Obi



1.pfv-do

5-work

with

‘Obi worked with her friend’ b. Comitative ǹ-tèm [ǹ-nyò Obi à-yìm-íʹˊ

1-friend

rel-1

Obi

1-friend

1.pfv-do-ConstF

‘the friend that Obi did work with her’

1-3s.poss

è-tûm

5-work

nà

with

yê ]

1:ObjPro

In (30b) there is no gap. Instead, where the noun phrase ǹtèm ‘friend’ was found in (30a) there is now a third person object pronoun yê. This is not a relative pronoun but the normal object pronoun. The strategy used involves what is often called “a resumptive pronoun” or “a trace”.

5 Ejagham relativization strategies and the Accessibility Hierarchy Given these two strategies, the next question is how these two strategies relate to the various grammatical relations within the RC. The related question is how many of the grammatical relations in the RC can actually be relativized. Keenan and Comrie (1977 and 1979) presented what they called the Accessibility Hierarchy (AH). The AH is relevant to the question of what NPs are accessible to being relativized in terms of their grammatical relation within the RC. (31) Accessibility Hierarchy (AH) Subject > Direct Object > Indirect Object > Oblique > Possessor > OCOMP (“Secondary Object”>”Primary Object”) Relevant to (31) is the AH Constraint that basically says two things. First, languages can relativize the NPs from the subject at the top of the hierarchy and then move down the hierarchy, but secondly, languages can limit how far down the hierarchy relativization can be performed. Some languages only allow subject

6 Relative clause in (Western) Ejagham 



 139

NPs to be relativized, while others allow NPs in all grammatical relations to be relativized, and still others limit the access to relativization to grammatical relations from subject to some middle point in the hierarchy. In addition to this AH Constraint, they proposed the Subject Relative Universal (SRU) that claims that all languages can relativize subjects. In Ejagham, then, we want first to establish how far down the hierarchy NPs are accessible to being relativized, and second to identify which strategy is used for relativizing each grammatical relation that can be relativized. We will see that that all grammatical relations are accessible to relativization and that each strategy is limited to a certain portion of the AH as indicated in (32). (32) AH: Subject > Direct Object > Indirect Object > Oblique > Possessor > Object of Comparison Gap Strategy Resumptive Pronoun Strategy



The following data supports the conclusion in (32). In (29a, b, c) we saw the use of the gap strategy with relativization of the Subject, Secondary Object and Primary Object. It was mentioned that the gap strategy was optional with the Primary Object. In (30b) we saw the use of the resumptive pronoun strategy with the comitative adjunct that uses the preposition nà, a grammatical relation that is not included in the AH but which shows the second strategy available. So the first additional data we will consider concerns the issue of both strategies serving the Primary Object. Second, we will consider the lower end of the AH by confirming the strategy used with the Object of Comparison and Possessors. Third, we will confirm that both strategies are used with Obliques. Turning first to Primary Objects, they may be relativized not only by the deletion strategy as in (29c) but also by the resumptive pronoun strategy as in (33b) where yê ‘1.ObjPro’ occurs. (33) Indirect Object-NP Relativization – Resumptive Pronoun a. ǹ-tèm à-kárè Obi é-dî

1-friend

1:pfv-give

Obi

5-food

‘A friend gave Obi food.’ b. Obi [ǹ-nyò ǹ-tèm à-kàr-éʹˊ Obi rel-1 1-friend 1:pfv-give-ConstF ‘Obi who a friend gave him food’

yê

1:ObjPro

é-dî ]

5-food

In the case of (33b) the use of strategies is optional since there is no danger of ambiguity as to whom the food was given and who gave it. However, if the

140 

 John R. Watters

ambiguity increases, the resumptive pronoun becomes preferred. When relativizing the Primary Object, the least ambiguous case is that in (33b) where the subject is present in the RC. In this case it is clear that ǹtèm ‘friend’ is the agent and that the head NP of the RC is distinct from ǹtèm. Furthermore, it is c1ear that édî ‘food’ must be the Secondary Object. The ambiguity increases when the subject of the relative clause is not present on the surface but only marked by the verbal prefix à-. The tendency in this case is to prefer the use of the resumptive pronoun. The greatest ambiguity arises, however, when the number and noun class of the head NP is identical to that of the underlying subject of the relative clause. In this case, there is no way of knowing whether the head NP is the subject or the Primary Object of the subordinate clause. In fact, if no resumptive pronoun is present it will be interpreted as the subject of the relative clause unless the context strongly indicates otherwise. So in this case the resumptive pronoun would disambiguate the grammatical relations within the subordinate clause. Thus, the determination of which strategy to use is based on semantic or pragmatic reasons rather than syntactic ones. The second set of data to be considered concerns the lower end of the AH by confirming which strategy is used with the object of comparison and possessors. In the case of these grammatical relations, the second strategy of using a resumptive pronoun is found. This strategy is exemplified below with various Possessors being relativized as well as the object of comparison. (34) Subject Possessor-NP Relativization a. ǹ-tèm ì ǹ-tûfâm à-kô

1-friend

of

1-chief

1.pfv-take

ǹ-tí

á-mì

3-money

that-3

‘The friend of the chief took that money’ b. ǹ-tûfâm ǹ-nyò ǹ-tèm òw-ě à-kó-ʹ é

ǹ-tí

á-mì

1-chief rel-1 1-friend 1-3s.poss 1.pfv-take-ConstF 3-money that-3 ‘The chief whose friend took that money’

(35) Secondary Object Possessor-NP Relativization a. ǹ-tèm à–kô ǹ-tí (�) ǹ-tûfâm

1-friend

1.pfv-take

1-chief

REL-1

3-money

(of)

‘The friend took the money of the chief’ b. ǹ-tûfâm ǹ-nyò ǹ-tèm à-kó- �é

1-friend

1-chief

1.PFV-take-ConstF

‘The chief from whom the friend took his money’ (36) Primary Object Possessor-NP Relativization a. ǹ-káé à-kárè ǹ-tèm ì ǹ-tûfâm

1-wife

1.PFV-give

1-friend

1.of

1-chief

‘The wife gave the friend of the chief that money’

ǹ-tí

3-money

ǹ-tí

3-money

ìm-ê

3-3s.POSS

á-mì

that-3

6 Relative clause in (Western) Ejagham 



 141

b. ǹ-tûfâm ǹ-nyò ǹ-káé à-kàr-é �ˊ ǹ-tèm òw-ě ǹ-tí 1-chief REL-1 1-wife 1.PFV-give-ConstF 1-friend 1-3s.POSS 3-money ‘The chief whose wife gave the friend of his that money’ (37) Locative Possessor-NP Relativization a. ǹ-káé à-kárè ǹ-tí kǎ ǹ-jû 1-wife 1.PFV-give 3-money at 9-house (of) ‘The wife gave the money in the house of the chief ’ b. ǹ-tûfâm ǹ-nyò ǹ-káé à-kàr-é �ˊ ǹ-tí 1-chief REL-1 1-wife 1.PFV-give-ConstF 9-money ‘The chief whose wife gave the money at his house’

ǹ-tûfâm

1-chief

kǎ ǹ-jû èy-ě at 9-house 9-3s.POSS

(38) Object-NP Relativization in Comparative Construction a. ǹ-tûfâm à-fâŋ à-chíŋì ǹ-tèm òw-ě

1-chief

1.PFV-big

1.PFV-pass

1-friend

REL-1 1-chief

1-friend

1-3s.POSS

‘The chief if bigger/more important than his friend’ b. ǹ-tèm ǹ-nyò ǹ-tûfâm à-fâŋ à-chíŋì

1.PFV-big

1.PFV-pass

yê

3s.ObjPro

‘The friend who the chief is bigger/more important than’

In (34)–(38) the resumptive pronoun is bolded. No constraint is found as to how far down the AH the resumptive pronoun strategy may be used in relativization. The above examples show that all grammatical relations are accessible to relativization since the very bottom of the AH is accessible. Not surprisingly possessive pronouns are used in the case of possessors (34–37). However, the object of comparison (38) is actually a serial clause construction. It is problematic as to whether the object is really an object of comparison or a secondary object or primary object. We are now left with our third set of data, namely that of the relativization of oblique grammatical relations. We will see that both strategies are possible with obliques in (39). The deletion strategy (39b) apparently is preferred but the resumptive pronoun strategy (39c) is acceptable. (39) Locative-NP (Gap Strategy) a. Obi à-kâk é-dî

Obi 1.PFV-place

5-food

kǎ

in

ǹ-tâk

9-bag

‘Obi put the food in the bag’ b. ǹ-tâk ǹ-nyì Obi à-kàgh-é �ˊ 9-bag REL-9 Obi 1.PFV-eat-ConstF ‘The bag in which Obi put the food’ c. ǹ-tâk ǹ-nyì Obi à-kàgh-é �ˊ

é-dî 5-food

___

é-dî

kǎ ń-nyə́-nè

9-house REL-9 Obi 1.PFV-eat-ConstF 5-food in ObjPro-9-ObjPro ‘The bag in which Obi put the food (in it)’

142 

 John R. Watters

The gap strategy is preferred as long as there is no serious ambiguity in identifying the grammatical relations of the NPs within the RC. Thus, the summary in (32) of how the strategies relate to the grammatical relations of the AH is justified.

6 Topics, universal negative quantifiers, and RCs In terms of discourse, the goal of additional research would be to provide a statistical analysis of the use of RCs in various genres. Impressions of various texts indicate that the use of RCs will vary according to the speaker and genre. Beyond the semantic notion of reducing the possible referents of a given NP, RCs are also used to introduce participants and topics. Another common use of RCs is in the expression of the universal negative quantifier. This expression relies on a formula that uses an optional kpe ‘even’ plus cháŋ ‘negative of existence’. An example is: (40) Kpè

even

ǹ-jû

9-house

ǹ-nyì

REL-9

á-fî

2.PFV-seize

êt

1p

cháŋ

NEG.EXIST

‘There is no household (whatsoever) in which they welcomed us’ = ‘(literally: EVEN household which welcomed us DOES NOT EXIST)’ The structure of the construction varies between these two: (41) a. (kpe)

x REL …

cháŋ

‘(even)

x that …

‘(even)

x does not exist that …’

b. (kpe)

does not exist’

x cháŋ REL …

When kpe ‘even’ is used it provides a sense of emphasis or of completeness in identifying that which does not exist.

7 Conclusion Ejagham has a standard RC structure for a S-V-O language. However, it has a number of permissible syntactic variations on this structure, including three sets of ­relativizers. Ejagham uses two strategies to indicate the NP within the RC that is co-­ referential with the head NP of the RC: the gap strategy and the resumptive pronoun strategy. Given these two strategies, Ejagham puts no limits on which grammatical relations can be relativized. The gap strategy is used for the higher portion of the AH and the resumptive pronoun strategy for the lower portion. The gap strategy is preferred with primary objects and obliques, but the resumptive pronoun is acceptable.



6 Relative clause in (Western) Ejagham 

 143

References to Ejagham story tellers Agbor Flavius Ntui of Ayoke, Cameroon. 1978. Etaghara Nkunku na Nkʉ́ í [Spider and Tortoise]. Transcribed by Patrick Etta Etta of Eyumojok, Cameroon. Anonymous, 1978? Egan Afughá-osí [The Story of Catching Fish]. Ayamba Nkom Moses from Eyumojok, Cameroon. 1977. Otui Nkom Ekaŋ [The Chieftancy of Nkom Ekang]. Transcribed by Patrick Etta Etta of Eyumojok, Cameroon. Ojong Tanyi (Ms) from Eyumojok. Asiraniran, Ekpimi Enop Akuk, Itəme Inop Mbak, and Opkere Onop Oseŋ [Arsiraniran, Bottle with Beautiful Edges, Cutlass with Beautiful Scabbard, Calabash with Beautiful Carving]. Ojong Tanyi (Ms) from Eyumojok, Cameroon. Approx 1977. Ayikobhi na Ntikpoŋ [Ayikobhi and Ntikpoŋ] Transcribed by Patrick Etta Etta of Eyumojok, Cameroon. Ojong Tanyi (Ms) (female) from Ekoneman, Cameroon. Approx 1977. Njombui na Mgbe [The Goat and the Leopard]. Transcribed by Patrick Etta Etta of Eyumojok, Cameroon.

References to academic works Andrews, Avery D. 2007. Relative clauses. In Timothy Shopen (ed.), Language typology and syntactic description. Volume II: Complex constructions, 206–236. Cambridge: Cambridge University Press. Keenan, Edward L. 1985. Relative clauses. In Timothy Shopen (ed.), Language typology and syntactic description. Volume II: Complex constructions, 141–170. Cambridge: Cambridge University Press. Keenan, Edward L. & Bernard Comrie. 1977. Noun phrase accessibility and universal grammar. Linguistic Inquiry 8. 63–99. Keenan, Edward L. & Bernard Comrie. 1979. Noun phrase accessibility revisited. Language 55. 649–664. Lewis, M. Paul, Gary F. Simons & Charles D. Fennig (eds.). 2016. Ethnologue: Languages of the world, 17th edn. Dallas, Texas: SIL International. Online version: http://www.ethnologue .com. Payne, Thomas E. 1997. Describing morphosyntax: A guide for field linguists. Cambridge: Cambridge University Press. Watters, John R. 1980. The Ejagham noun class system: Ekoid Bantu revisited. In Larry M. Hyman (ed.), Noun classes in the Grassfields Bantu borderland. Southern California Occasional Papers in Linguistics 8: 99–137. Los Angeles: University of Southern California. Watters, John R. 1981. A phonology and morphology of Ejagham. Los Angeles: University of California, Los Angeles dissertation. Watters, John R. 1989. Bantoid. In John T. Bendor-Samuel (ed.), Niger-Congo, 400–420. New York: University Press of America. Watters, John R. 2010. Focus and the Ejagham verb system. In Ines Fiedler & Anne Schwarz (eds.), The expression of information structure: A documentation of its diversity across Africa, 349–376. Amsterdam: John Benjamins. Watters, John R. 2012. Tone in Western Ejagham (Etung): Lexical tone on the “minimal” verb forms. In Matthias Brenzinger & Anne-Maria Fehn (eds.), Proceedings of the 6th World Congress of African Languages, 309–319. Cologne: Rüdiger Köppe Verlag. Williamson, Kay & Roger Blench. 2000. Niger-Congo. In Bernd Heine and Derek Nurse (eds.), African Languages, 11–42. Cambridge: Cambridge University Press.

Pius N. Tamanji and Blasius Achiri-Taboh

7 A prolegomenon to the syntax of the relative clause in the Eastern Grassfields Bantu borderland1 Abstract: In this chapter, we examine the similarities in the morpho-syntactic and prosodic properties of relative clauses in two Eastern Grassfields Bantu languages, Bafut and Medumba, showing that every argument position in them is relativizeable. We discuss strategies of relative clause formation, showing that both Bafut (in the Ngemba sub-group) and Medumba (in the Nun sub-group), employ a relative determiner in agreement with the relativized antecedent, although Bafut, in addition, also makes use of the regular (relative) complementizer either in alternation or collocation with the relative determiner. It is shown that the two languages exhibit other striking similarities with respect to the syntax and prosody of the relative clause. Specifically, we discuss the occurrence, in both languages, of subject markers after (relativized) lexical subjects, phonetic breaks between overt relative markers/complementizers and overt subjects, resumptive pronouns in every relativized position, and a relative definiteness marker at the end of the matrix relative clause. Finally, we look at the possibility of A-bar movement within the relative clause of both languages, showing that empirical generalisations from the distribution of idiom chunks, binding, and reconstruction effects support a movement analysis for relative clause constructions in Nun and Ngemba languages.

1 The original version of this chapter is the result of work done exclusively by Pius N. Tamanji. I only came in after the first review of the work, on the heels of Pius’s passing away. After addressing the reviewers’ comments to the best of my ability, I made my own modest contribution to the chapter by means of necessary additional data, empirical and conceptual descriptions and analyses. On a personal note, I would like to posthumously express sincere appreciation to Pius for the pioneering work on the chapter. On behalf of both of us, I would also like to express gratitude to our consultants for the additional data they kindly provided. Finally, we have benefitted from research allowances both from our respective universities and from the Cameroonian Ministry of Higher Education for which we are grateful. Pius N Tamanji: University of Yaoundé I, [email protected] Blasius Achiri-Taboh: University of Buea, [email protected] DOI 10.1515/9783110469547-007

146 

 Pius N. Tamanji and Blasius Achiri-Taboh

1 Introduction Eastern Grassfields Bantu languages are spoken in the grassland (Western and North-western) Regions of Cameroon. The group is made up of the Nun, North-East, Bamileke, and Ngemba sub-groups with zone 4, 5, 10, and 11 languages, respectively. An examination of two languages2 from two of these sub-groups – Bafut, in the Ngemba subgroup and Medumba, in the Nun subgroup – shows that they exhibit interesting characteristic similarities in terms of strategies of relative clause formation and prosody. For example, both Bafut and Medumba exhibit a particle láá/lá which marks the right edge of the relative clause, and has been subject to various interpretations (Mfonyam 1989; Tamanji 2009; Kouankem 2012). Note that one general role of relative clauses is to provide definiteness to the DPs they modify. Thus, it is possible to conjecture that the particle is an overt relative definiteness marker, although, adopting views in formal syntax such as those of Rizzi (1997) and subsequent works (cf. De Vries 2002; and Aboh 2004, 2005 and 2006), Kouankem has rather viewed it as the relative complementizer stranded at the base of the relative clause by the raising of the embedded tense phrase (TP). In this chapter, therefore, we refer to it as a (relative) definiteness marker and analyse it as such. However, while both languages employ more or less the same relative clause formation strategies, there are differences in the number, type and the morphology of relative elements. The present chapter seeks to illustrate the phonological and morphosyntactic properties of relative clauses in the two Eastern Grassfields Bantu languages with keen interest in the use of the relative particles. Thus, in Section 2 that follows, we discuss the different strategies of relative clause formation, highlighting the relative morphemes and showing how they distinguish Bafut in the Ngemba sub-group from Medumba in the Nun sub-group. In Section 3, we examine the different relativizable positions and show that with the exception of direct object positions, all relativized positions in both languages are marked with a pronominal element known in the literature as resumptive pronoun (RP), the idea of its existence being resumption. In Section 4, we examine and account for the presence of phonetic breaks within the relative clause in the two languages. Finally, in Section 5, we then look at possibilities of A-bar movements.

2 Pius Tamanji is a native speaker of Bafut, and provided the relevant Bafut data. For parts of the data on which doubts were expressed, we solicited the help of an outside consultant, a 40-year old Bafut Noble ‘Nka’ala’ Ndonwi Joel with logistical support from Ivo Akoso and Raymond Neba Tanjong. As for the Medumba data, we counted mainly on Kwankem’s (2012) study of that language. Like for Bafut, we sought assistance from another Medumba native speaker for additional data, and the 25-year old Segolene Tchaptcheu was very instrumental with logistical support from Simon Charly Tinto. All our data in the chapter are represented using the IPA symbols, and interlineally glossed following the Leipzig glossing rules.



7 A prolegomenon to the syntax of the relative clause 

 147

2 Strategies of relative clause formation Following standard assumptions, the relative clause is used to complement an antecedent DP by describing or delimiting its potential domain of reference.3 In Bafut and Medumba, it occurs to the right of the relativized DP, and the form of the verb and tense-aspect markers within the subordinate clause is not different from the form we find in ordinary declarative sentences. Relative clauses in Grassfields Bantu languages are generally signaled by a variety of elements including a relative determiner (rel.det) (which can be seen as a relative pronoun), a relative marker (rel.m) (which usually co-functions as a regular declarative complementizer), tone, and a relative definiteness marker (rel.def). While some of these elements are found in all the languages, some are restricted to language sub-groups. Thus, as we see in the following subsections, Bafut makes use of the rel.det, the rel.m and the rel.def, while Medumba is limited to the rel.det and the rel.def.

2.1 The relative determiner (rel.det) The use of the rel.det to introduce the relative clause is a strategy that is employed by both Bafut and Medumba. Its use in Medumba is illustrated as follows. Medumba (1) a. mɛ́ nzwì náʔ sɔ̀ g nzwə̀ woman tns wash cloth “The woman washed the cloth.” b. mɛ́ nzwì zə̀ à náʔ sɔ̀ g woman rel.det she tns wash “the woman who washed the dress” (2) a. mə̀ nù ntsə̀ sbj.1sg drink water “I drank water.” b. ntsə̀ mì mə̀ water rel.det sbj.1sg “the water that I drank”

nù drink

nzwə̀ cloth

lá rel.def

lá rel.def

3 Since Abney (1987), all definite expressions including nominals like the chair in the garden and proper names like Paul have been taken to have the status of the determiner phrase (DP), these having been analyzed before the mid 1980s to have the status of the noun phrase (NP).

148 

 Pius N. Tamanji and Blasius Achiri-Taboh

(3) a. à fá ŋkáb núm bún [tʃ]ə̂ n lì sbj.3sg give money to children these here “He gave money to these children.” b. bún [tʃ]ə̂ n lì tsə̀ à fá ŋkáb children these here rel.det sbj.2sg give money bú lá obj.3pl rel.def “these children to whom he gave money”

núm to

As the facts show, the rel.det is concordial to the relative head. Thus, in (1b), the rel.det agrees with the Class 1a noun mɛ́ nzwì ‘woman’ as zə̀ ; in (2b), it agrees with the class 1b noun ntsə̀ ‘water’ as mì; and in (3b), with bún ‘children’ as tsə̀ . As indicated already, Bafut equally makes use of the rel.det as shown in the following paradigms – observe that, within the TP, Bafut exhibits a subject marker (SM) (or a clitic) after a lexical subject (as opposed to a functional/pronominal one); as glossed in (4b), we shall later describe this SM as RP (i.e., resumptive pronoun) when the subject position is relativized. Bafut (4) a. màŋgjè wá à kì̵ sìʔí tsə̀ ʔə̀ woman the sm tns wash dresses “The woman washed dresses.” b. màŋgjè wá jìí à kì̵ sìʔí woman the rel.det rp tns wash “the woman who washed dresses” (5) a. mə̀ kì̵ nó mì̵lùʔù sbj.1sg tns drink wine “I drank water.” b. mì̵lùʔù mìí mə̀ kì̵ wine rel.det sbj.1sg tns “the wine which I drank”

nó drink

tsə̀ ʔə̀ dresses

láà rel.def

láà rel.def

(6) a. mə̀ kì̵ fá ŋkábə́ mbó bɔ́ ɔ̀ sbj.1sg tns give money to children “I gave money to children.” ŋkábə́ mbó bó láà b. bɔ́ ɔ́ bìí mə̀ kì̵ fá children rel.det sbj.1sg tns give money to rp rel.def “the children to whom I gave money” The rel.det in Bafut is derived from the far-from-speaker/listener (FSL) demonstrative determiner, the difference being the tone and the type of concord it bears.

7 A prolegomenon to the syntax of the relative clause 



 149

Following is Tab. 1 which shows the rel.det and FSL demonstrative determiners and their respective concord consonants (CC) – note that the initial consonants of these determiners are identical to those of the nominal prefixes except in the case of Classes 1, 3, 7 and 9 where the nominal prefix begins with either a vowel or a nasal consonant. Tab. 1: Concord consonants of Bafut relative pronouns and FSL demonstratives. Noun class

CC

rel.det

CC

FSL DEM

 1 ø-, N 2 bì̵ 3 ì̵-, N 5 nì̵ 6 mì̵ 7 à 8 ì̵ 9 N10 N19 ì̵-

jbjnmjdʒjdʒf-

jìí jìí jìí nìí mìí jìí dʒìí jìí dʒìí f ìí

wbwnmjdʒjdʒf-

wíì bíì wíì níì míì jíì dʒíì jíì dʒíì fíì

2.2 The relative marker/complementizer Bafut (like other languages of the Ngemba sub-group) employs the rel.m which, in this case, is mə́ – as indicated earlier, Medumba does not exhibit the rel.m. In Bafut, this is used as an alternative strategy to mark the relative clause as shown in the following examples.4 (7) a. mú wá mə́ à kì̵ bòó ŋkwú láa à kwò-mə̂ child the rel.m rp tns meet ghost rel.def sm die-pfv “The child who met a ghost has died.” b. mì̵lùʔù mjá mə́ bìʔì kì̵ nó láa mí tí̵ wine the rel.m we tns drink rel.def sm strong “The wine which we drank is too alcoholic.”

4 Note, however, that the same relative clause marker is used as a regular declarative complementizer to introduce a variety of subordinate clauses as seen below: mà ʒìi mə́ nì̵bàʔ á ʒì sbj.1sg know that Neba sm come “I know that Neba is coming.” It can therefore also be called ‘relative complementizer’, although, in this chapter, we stick to the term relative marker.

150 

 Pius N. Tamanji and Blasius Achiri-Taboh

As noted in the previous subsection, the use of the rel.det is in agreement with the relativized antecedent. The rel.m mə́ is therefore a neutral form to mark the relative clause. However, both the rel.det and the rel.m are often used in collocation to mark the same relative clause, the rel.det preceding the rel.m, to place emphasis on the antecedent in a contrastive focus fashion, arguably so because the rel.det is marked for concord with the relativized antecedent. Consider the following examples. (8) a. mú wá jìí mə́ à kì̵ bòó ŋkwú child the rel.det rel.m sm tns meet ghost láa à kwò-mə̂ rel.def sm die-pfv “The child who met a ghost has died.” mə́ bì?ì kì̵ nó láa mí tí̵ b. mì̵lùʔù mjá mìí wine the rel.det rel.m we tns drink rel.def sm strong “The wine which we drank is too alcoholic.” In other words, in the context of (8a), for example, the reading that we literally get from the perspective of the native speaker is roughly the child, the one who met the ghost has died implying no other child but the one who met a ghost has died. Note, however, that, although cleft-constructions are also used in Bafut to mark contrastive focus (as in it is the children who did the work (and not us)), the use of both the rel.det and the rel.m in collocation in cleft-constructions to mark contrastive focus is rather optional than compulsory as shown in the following examples – specifically, any/both of them can be used. (9) à ní̵ ŋú (jìí) mə́ mfɔ̀ kì̵ fá it cop person rel.det rel.m chief tns give mbó láà wá ɣú the him to rel.def “It is the person to whom the chief gave the picture.”

fútò picture

2.3 The Relative definiteness marker (rel.def) As pointed out already, the relative definiteness marker (rel.def) (lá in Medumba and láá in Bafut) occurs at the right edge of the relative clause. In a non-sentence-final position, in Bafut, as seen in the further examples that follow, it bears a clause boundary tone which rises sharply and then falls midway, while in a sentence-final position, it takes a falling tone.



7 A prolegomenon to the syntax of the relative clause 

 151

Bafut (10) a. ŋgàŋàfàʔà dʒá dʒìí mə́ dʒí bɔ́ rí̵kə́ láa bó workers the rel.det rel.m rp lazy rel.def sm kwɛ́ ɛ̀ ká tns go.home “The workers who are lazy will return home.” b. à ní̵ mì̵lùʔù mìí mə́ kì̵ nó láà it cop wine rel.det sbj.1sg tns drink rel.def “It is the wine which I drank.” Medumba (11) a. mɛ́ n zə̀ mə̀ jə̀ n lá ù áʔ nɛ́ n kàamɛ̀ ɲàm child rel.det sbj.1sg see rel.def sm tns go heaven “The child that I saw will go to heaven.” b. bún tʃə̂ n lì tsə̀ à fá ŋkáb nùm children these here rel.m sbj.3sg give money to bú lá bù nɛ́ n á ntànə̀ them rel.def sm go to market “These children to whom he gave money are going to the market.” As indicated earlier, we conjecture that this particle is an overt relative definiteness marker, although in her 2012 study, Kouankem has analyzed it to be a relative complementizer stranded at the base of the relative clause by the raising of the embedded TP.5 One thing that shows that this particle is actually used as a relative definiteness marker is the fact that it sticks to (the right edge of) the relative clause as seen in the following unacceptable Bafut example: (12) *bə̀ bjá bìí mə́ mfɔ̀ à tsɔ́ ʔɔ̀ wáá people the rel.det rel.m chief sm choose rp bó jɔ́ ʔɔ̀ láà sm marry rel.def “The people whom the chief has selected can get married.”

mbə́ cond

5 Kouankem’s analysis of the status and grammatical function of the particle lá depends crucially on her assumption that the particle is the head of the CP. That is, following Essono’s (2004) analysis of the Ewondo relative marker é, she concludes that lá in Medumba is equivalent to regular complementizer like if in English and si in French. This conclusion poses a problem in Eastern Grassfields languages including those of the Ngemba sub-group like Bafut in which, as we have seen, a regular complementizer is used to serve as a relative marker in relative clauses and co-occur with the relative definiteness marker. Besides, recall that there is also the relative determiner which, in Bafut, is optionally used in collocation with the relative marker or used alone as the relative marker; as we have also seen, a similar relative determiner is compulsorily used in Medumba as a relative marker.

152 

 Pius N. Tamanji and Blasius Achiri-Taboh

That is, it cannot occur in a non-relative-clause-final position. Besides, the fact that it occurs on free relatives in Bafut as in the following examples makes the point clear. mf ɔ̀ (13) a. bɔ́ ɔ́ bjá ʒì-mə́ láa wá twú wáà children the come-pfv rel.def chief the pay them “When the children came (as I have already told you), the chief paid them.” b. màŋgjè wá nò-mə́ mí̵lùʔù mjá láa mə̀ woman the drink-perf wine the rel.def sbj.1sg bí̵ín ŋkwɛ́ ɛ́ ̵ ì̵ get.up go.home “When the woman drank the wine (as I told you), I got up and went home.” Its definiteness role can clearly be seen in the following Bafut construction in which it serves to mark emphasis in a context where some event or object had been mentioned earlier and is already familiar to the participants in the discourse context. (14) à ká fá láa á sbj.3sg tns give rel.def to “Who is it that he will give it to?”

mbó hands

wò ? who

In this example, the object of give is omitted, but is however understood in the discourse context. That is, the rel.def láá which immediately follows the verb renders the speaker’s request for information about who will receive the object more emphatic. With the facts in (13) therefore, we can then assume that both the rel.det and the rel.m are suppressed in free relatives (as the relativized head is not spelled-out overtly) such that the rel.def is what actually marks the relative clause.

3 Relativizable positions and resumption Conditions at the various relativized positions in the relative clause (and in the dependent TP) generally differ depending on whether the relativized position is subject, object, oblique or pre/post possessor. These are examined in the subsections that follow.

3.1 Subject relatives In both Bafut and Medumba, when the subject is relativised, it leaves behind a subject marker (SM) – a corresponding subject pronoun. Consider examples



7 A prolegomenon to the syntax of the relative clause 

 153

from both languages – as we see later in this subsection, such subject markers are known in the literature as resumptive pronouns (RPs) as indicated earlier in Section 2.1, and we mark them thus. Bafut (15) a. màŋgjè jìí mə́ á lí̵gə́ sí̵gí̵nì̵ láa woman rel.ded rel.m rp cultivate well rel.def à ká lò sm tns leave “Any woman who cultivates (farm) well will leave.” b. wò jìí mə́ ó jə́ rí̵ mbí láa ò obl.2sg rel.ded rel.m rp steal goats rel.def sm ká ɣɛ̀ ɛ́ á báŋʃe tns go to hell “You who steals goats will go to hell.” c. jú jìí mə́ á kótí̵ ŋgwú láa obl.3sg rel.ded rel.m rp catch chickens rel.def à ká ɣɛ̀ ɛ́ á búrí̵ sm tns go to heaven “He who catches chickens will go to heaven.” Medumba (16) a. mɛ́ nzwì zə́ à náʔ sɔ̀ g nzwè lá woman rel.ded rp tns wash cloth rel.def “the woman who washed the dress” b. mè yèn mɛ̀ n zè à làb ù lá sbj.1sg see person rel.ded rp beat you rel.def “I have seen the person who beat you.” c. wù zè ù bú ndzwì mbúè lá obl.2sg rel.d rp hab steal goat rel.def áʔ nɛ́ n kàamɛ̀ ɲàm ù sm tns go heaven “You who steals goats will go to heaven.” In (15a), a lexical DP is relativized with a corresponding subject pronoun left in the relativized subject position. The examples in (15b) and (15c) illustrate relativization of pronouns. Here, the antecedent pronoun takes the oblique case form, while the subject pronoun (i.e., RP) left in the relativized subject position takes the regular nominative (or subject) case form. As the examples in (16) show, a similar scenario can be observed in Medumba. Personal

154 

 Pius N. Tamanji and Blasius Achiri-Taboh

pronoun forms in both Bafut and Medumba are presented in Tabs. 2 and 3 for the reader’s appreciation.6 Tab. 2: Bafut personal pronouns. Subject

Object

Oblique

mə́ (I) ò (you) à (he/she) bìʔì (we) bù/nì̵ (you) bó (they)

ɣá (me) ɣó (you) jí (him/her) jíʔí (us) ɣúú (you) wáá (them)

mə́ (me) wò (you) jú (him/her) bìʔì (us) bù (you) bó (them)

Tab. 3: Medumba personal pronouns. Subject

Object

Oblique

mə́ (I) ù (you) à (he/she) bàg (we) bìn (you) bù (they)

mə́ / àm (me) wù (you) jí (him/her) bàg / yàg (us) bín / jín (you) bú / júb (them)

mə́ (me) wù (you) jí (him/her) bàg (us) bín (you) bú (them)

Data on subject relativization in these languages raise a number of interesting questions. One is about the use of the oblique object form when a subject pronoun is relativised. A probable reason for that is that, in these languages, oblique object pronouns are also used as emphatic forms of subject pronouns. This is what happens, at least, in Bafut as illustrated in the following paradigm of simple declarative sentences. (17) a. ò ká ɣɛ̀ ɛ́ sbj.2sg tns go “You will go to hell.” b. wò ká sbj.2sg.Emph tns “YOU will go to hell.”

á to ɣɛ̀ ɛ́ go

báŋʃê hell á to

báŋʃê hell

6 Note that 3rd person pronouns may refer to non-human entities, in which case, (as in many Bantu languages) they take a variety of forms depending on noun class concord. For example, the subject non-human forms in Bafut are as follows: à/ní/á/á/ fí (it) and bí/mí/jí/jí (they).



7 A prolegomenon to the syntax of the relative clause 

 155

If we assume, therefore, that a relativized constituent is subject to focal stress, then the reason is clear. The second concern is on the fact that a relativized subject position remains filled by a nominative form pronoun. In the literature (as pointed out earlier), pronouns in the relativized position are generally known as resumptive pronouns (RP), the general phenomenon of their occurrence called resumption. Since the 1970s, a wide range of accounts has been presented to explain the phenomenon. For an in-depth perspective on the use of RPs, see, for example, accounts of McCloskey (1979, 1990, 2000, 2002), Borer (1984), Chomsky (1986), Aoun and Li (1990), Shlonsky (1992), Boeckx (2003) and Achiri-Taboh (2004, 2006). The third and final worry with subject relativization in Bafut and Medumba is that, if the antecedent (complex) DP that embeds the relative clause is the subject of a main clause, a subject pronoun recurs after the relative clause. It might seem though that this pronoun is a RP in Spec,VP, but this cannot be the case as it precedes the main clause tense. It might also be thought of as the same subject marker observed earlier in Section 2.1 occurring after a simple lexical DP subject in Bafut. In this respect, it would then be considered that it actually occurs null in Medumba after simple DP subjects. However, since it only occurs overtly in Medumba after the complex DP subject, we are obliged here to further motivate its occurrence. Thus, we can follow ideas from Citko (2002) to consider that it functions as an appositive pronoun used (in these languages) to remark the relativized head of the complex DP subject for the purpose of proper parsing, given the presence of the intervening relative clause. In other words, the subject is repeated after the relative clause as a strategy to enhance the parsing of the sentence. In the present study, we shall not attempt to further explore the syntax of (this) appositive modification.

3.2 Direct object relatives Unlike relativized subjects, relativized direct objects in Bafut and Medumba generally do not leave behind RPs. However, in Bafut (though not in Medumba), if the object is human, the RP is optional.7

7 Like in subject relativization, when the relativized object in Bafut is a pronoun, the latter takes an oblique form, while the RP it leaves takes the regular object form. Consider the following examples. mə́ nó ì̵ kì̵ lòó (ɣó) láa ò ká ɣɛ̀ ɛ (i) wò sbj.2sg.Emph rel.m snake sm tns bite rp rel.def you tns go “You, who was beaten by a snake will go.” wá mə́ bì à kì̵ twòŋ (jí) láa à ká tsí (ii) jú sjb.3sg the rel.m Bih sm tn invite rp rel.def sm tns stay “She/he, whom Bih invited will stay.”

156 

 Pius N. Tamanji and Blasius Achiri-Taboh

Bafut (18) a. nsòò wá jìí mə́ bì lɛ́ nlì̵gí̵ láa bɔ́ ŋə́ farm the rel.det rel.m Bih tns cultivate rel.def good “The farm which Bih cultivated is nice.” ŋkwú à b. mú wá mə́ kì̵ fɛ̀ ɛ́ (jí) láà child the rel.m ghost sm tns slap rp rel.def “the child whom a ghost slapped” Medumba (19) a. mə̀ jen mbú zə̀ ù sbj.1sg see child rel.ded obj.2sg “I saw the child that you beat.” b. tsàŋ zə̀ mə̀ zhù lá food rel.ded sbj.1sg eat rel.def “the food that I have eaten”

làb beat

lá rel.def

3.3 Indirect objects We just pointed out that, unlike relativized subjects, relativized direct objects in both Bafut and Medumba generally do not leave behind RPs. However, like relativized subjects in these languages, indirect object relativization results in compulsory resumption. This is seen in the following examples. Bafut (20) mə̀ ʒí ŋú jìí mə́ mfɔ̀ kì̵ sbj.1sg know person rel.ded rel.m chief tns fútò wá *(ɣú) mbó láà to rel.def picture the rp “I know the person to whom the chief gave the picture.”

fá give

Medumba (21) bún tʃə̂ n lì tsə̀ à fá ŋkáb nùm *(bú) lá rp rel.def children these here rel.det sm give money to “these children to whom he gave money” Interestingly, indirect objects in these languages function like oblique objects: they must follow a preposition and, as personal pronouns, they take the oblique form – we further explore the (strange) precedence of the RP over the preposition in (20) in the next subsection.



7 A prolegomenon to the syntax of the relative clause 

 157

3.4 Oblique object relative In Bafut, the relativization of an oblique object (or object of a preposition) leaves behind the resumptive element ɣú which corresponds to the locative adverbial there. In the relative construction, the distribution of ɣú depends on what it resumes. Thus, if the relativized antecedent is the name of a place (i.e., a locative), the preposition is dropped. Consider the paradigm in (22) below. Bafut (22) a. àdzɔ̀ ŋ já á lɛ́ ŋɣɛ̀ ɛ́ á àlá?à giant the sm tns go to village “The giant went to the village.” b. àlá?à já mə́ àdzɔ̀ ŋ já á lɛ́ the sm tns village the rel.m giant “the village that the giant went to”

ŋɣɛ̀ ɛ́ go

ɣú there

láà rel.def

On the other hand, if the antecedent is of any other kind, be it human or nonhuman, ɣú is used preceding the preposition as in the following paradigms. Bafut (23) a. àtì̵rì̵ já á kì̵ tì̵ŋnə́ mbíì já á mbɛ̀ ɛ́ fool the SM tns tether goat the at side “The fool tethered the goat near a tiger.” b. tákwúrə́ wá mə́ àtì̵rì̵ já á kì̵ tì̵ŋnə́ tiger the rel.m fool the sm tns tether mbíì já ɣú mbɛ̀ ɛ̀ láà goat the there side rel.def “the tiger beside which the fool tethered the goat” (24) a. mə̀ lí̵n ntǝ á mbɛ̀ ɛ̀ màngjə̀ sbj.1sg tns stand at near woman “I stood near a woman.” mə̀ lí̵n ntǝ b. màngjè wá mə́ woman the rel.m sbj.1sg tns stand “the woman by whom I stood”

ɣú there

tákwúrì̵ tiger

mbɛ̀ ɛ̀ near

láà rel.def

The situation in Medumba is slightly different. If the relativized oblique object DP is human, it leaves behind a RP which corresponds to the oblique form as in the following paradigm.

158 

 Pius N. Tamanji and Blasius Achiri-Taboh

Medumba (25) a. nùmí kù swə̀ n kùlò bù bún tʃə̀ nlì Numi prog sell plantains with children these “Numi sells plantains with these children.” b. bún tʃə̂ nlì tsə̀ nùmí kù swə̀ n kùlò children these rel.det Numi prog sell plantains bù *(bú) lá with rp rel.def “these children with whom Numi sells plantains” However, in the case of a non-human antecedent, the place adverbial corresponding to there is used as a resumptive element as shown in the following paradigms. (26) a. mɔ́ ndʒìm nɛ̀ n laʔ man go village “The man has gone to the village.” b. láʔ dʒə̀ ʔ zə̀ mɔ́ ndʒìm nɛ̀ n village the rel.det man go “the village where the man has gone to”

dʒìʔ there

lá rel.def

(27) a. à ndɔ̀ ʔɔ mɛ́ ntʃìŋ zà bɔ́ lɔ̀ zà fít nǝ it only someone rel.det foolish that can to kfí̵t mbvǝ máŋə́ nzwì tether goat near panther “It is only a fool that can tether a goat near a panther.” b. nzwì zà mɔ́ ndʒìm à kfí̵t mbvǝ máŋ dʒìʔ lá panther rel.det man sm tether goat near there rel.def “the panther by which the man tether the goat”

3.5 Pre/post-possessor relatives Both languages exhibit post-possessor DPs with the (covert) of-genitive structure “DP-of-DP”, but with a slight difference in that, in Medumba, the structure takes an additional pre-possessor determiner (or marker). (28) a. Bafut tá bì̵nì̵bàʔ à ʒì-mə́ father of.Neba sm come-pfv “Neba’s father has come.”



7 A prolegomenon to the syntax of the relative clause 

b. Medumba zə̀ bɔ̀ bɔ́ ɔ̀ nɔ́ nɔ̀ ɔ poss father Nana “Nana’s father has come.”

 159

sə̀ ʔǝ come

As the facts show, therefore, post-possessor DPs in these languages are like oblique objects which result in compulsory resumption when relativized. This is exactly what happens, although with a nuance in Medumba. (29) a. Bafut mú wá mə́ tàá *(jì) kì̵ kwó láa à ʒì-mə́ child the rel.m father rp tns die rel.def sm come-pfv “The child whose father died has come.” b. Medumba mɛ́ n zə̀ zə̀ bɔ̀ bɔ́ ɔ̀ kfú lá child rel.det poss father die rel.def “The child whose father died has come.”

sə̀ ʔǝ come

Recall from (25) above that, in Medumba (like in Bafut), oblique object relativization generates compulsory resumption. The situation in (29b) may therefore be attributed to the presence of the pre-possessor determiner. More specifically, since the RP is basically a pronoun (McCloskey 2002:192) and personal pronouns belong to the category of D-pronouns (i.e., they are determiners) (Postal 1966; Abney 1987; Longobardi 1994; and Lyon 1999), the presence of the prepossessor determiner means that the need for a post-possessor RP is obviated and unexpected.

4 Prosody: phonetic breaks in restrictive and non-restrictive relative clauses Studies on Bantu languages (Kisseberth and Abasheikh 1974; Kisseberth 2005, 2010a, 2010b; Beltzung, Rialland & Aborobongui 2010; Cheng and Downing 2010; etc.) have revealed four elements that are prominent in their prosody. These are accent, tone, vowel length, and phonetic breaks. In Bafut and Medumba, the prominent prosodic elements are tone and a perceptible phonetic break. In this section, we limit our scope to the occurrence of phonetic breaks which are clearly manifested in their relative clauses.

160 

 Pius N. Tamanji and Blasius Achiri-Taboh

4.1 Restrictive relative clauses In restrictive relative clauses in Bafut and Medumba, two phonetic breaks can be observed. One occurs immediately after the rel.det/rel.m and the other after the relative definiteness marker (rel.def). In regular speech in both languages, the break after the rel.def is clearly perceptible (or “heavy”), while the one after the rel.det/rel.m is rather light. Consider the following examples of restrictive relative clauses in which we use the IPA diacritic [|] to mark a light break and [||] to mark a heavy one. Bafut (30) a. mì̵lùʔù mjá mə́   | mf ɔ̀ wá kì̵ nó láa || mí lìì wine the rel.m chief the tns drink rel.def sm sweet “The wine which the chief drank is sweet.” bí̵ kɔ́ ŋ ndánwì b. mə̀ kì̵ bìtí̵ bɔ́ ɔ́ bjá mə́  | sbj.1sg tns ask children the rel.m rp like church já mə́     | ì̵ tsí á tú ntáʔá láa || mə́ nwì the rel.m rp stay on top hill rel.def that God tsí á búrí̵ lɛ́ stay in heaven q “I asked the children who like the church (which is) on the hill top whether God is in heaven.” Medumba (31) ŋgùŋgùn tsə́     | ù jùn lá   || bù nɛ́ n girls rel.det you see rel.def sm go “The girls whom you saw are going to the market.”

á to

ntànə̀ market

Note that, in (30b) which contains multiple embedding, the rel.def occurs only once – at the end of the relative clause – followed by a heavy break, since all the embedded relative clauses form part of the main relative clause that embeds them. An important question that arises here is with respect to the phonetic breaks, that is, what accounts for them? With respect to the light break after the rel.det/m, one hypothesis could come from simple observation. Note that, in these languages, like in many other Bantu languages, the relative marker or the complementizer (depending on which one is used to relativize) is compulsory (i.e., it must be filled overtly) whatever is relativized, and when the subject position is relativized, a RP is left in the latter position as we have seen. On the basis of these observations, we suggest that there is a post-COMP-subject-spellout effect that creates a break between the relative marker (or overt C(omplementizer)) and



7 A prolegomenon to the syntax of the relative clause 

 161

an overt subject of the embedded TP. If this is correct, then it leads us to the PostCOMP-Pre-Subject-Spellout Break Hypothesis that can be stated as follows: (32) The Post-COMP-Pre-Subject-Spellout Break Hypothesis A phonetic break occurs between an overt relative marker/complementizer and an overt subject. In other words, after an overt relative marker and/or overt complementizer, there is a phonetic break before the subject if the latter is also overt. Since both COMP (i.e., either the relative marker or the complementizer) and the subject in these languages are always compulsory (as the subject position is always resumed when relativized), evidence in support of this hypothesis can only come cross-linguistically from a language in which both of these elements are not always compulsorily overt. In other words, where either the COMP or the subject is covert but not both, we do not expect the break to occur. Indeed, this is what happens in English, for example. Consider the following paradigms.8 (33) a. I know that John is powerful. b. I know John is powerful.

[with a break] [without a break]

(34) a. The house which/that John built is great. b. The house John built is great.

[with a break] [without a break]

(35) The house which/that collapsed was new.

[without a break]

Let us now turn to the post-relative-clause heavy break. A similar analysis can be said to be in play here. That is, the presence of the rel.def at the end of the embedded relative clause (linked to the relative marker) followed by a subject pronoun (as in (30a) and (31)) or a compulsory Comp (as in (30b)) means that the Post-COMP-Pre-Subject-Spellout Break Hypothesis also applies, giving rise to the post-COMP-subject-spellout effect. If this is what happens, then an interesting problem that further arises is the fact that, as the facts show, this break is more prominent than the post-relative marker break discussed earlier. Although we do not suggest any concrete account for this disparity, we think that this must be connected to the concordial nature of the rel.det exhibited in both languages.

8 In (33a)/(34a), after an overt spellout of the complementizer that, a little pause (i.e., the phonetic break) follows before the pronunciation of the subject DP John, unlike in (33b)/(34b) without (an overt spellout of) that or (35) without (an overt spellout of) the subject.

162 

 Pius N. Tamanji and Blasius Achiri-Taboh

4.2 Non-restrictive relative clauses In non-restrictive relative clauses, the location of the phonetic break in both languages is significantly different from what we find in restrictive relatives. As the facts in (36) and (37) below show, there is a heavy break rather ­preceding the rel.m/det, and in this case, the expected break after the latter disappears. Bafut (36) ndá jà || mə́ ò ʒí láa || house my rel.m you know rel.def “My house, which you know, has collapsed.”

ì sm

wò-mə́ fall-pfv

Medumba (37) nʃùn àm || zə̀ náná jùn má ndá lá   || friend my rel.det Nana see in house rel.def bù á ŋkùmdìáŋ sm foc carver “My friend, whom Nana saw in the house, is a carver.” Note that relative clauses are so-called because they relate to the complex DP within which they are embedded. So an easy account for the pre-relative clause break in (36) and (37) is that, as in Traditional Grammar (Quirk and Greenbaum 1973: 383) and in modern Generative Grammar (as of Chomsky and Lasnik 1977; Sag 1997; and Citko 2002), non-restrictive relative clauses (unlike restrictive ones) do not modify/describe the antecedent, but rather only provide additional optional information (or an “after-thought”) about the latter. As a result, the non-restrictive relative clause does not form part of the DP per se, but only adjoins to it. To now account for the disappearance of the expected (light) break after the relative marker (as with restrictive relative clauses), let us then assume that the preceding heavy break triggers a closure between the rel.m/rel.det and the subject of the embedded TP. If this reasoning is correct, then we expect that other post-relative marker light breaks in restrictive relative clauses embedded within a non-restrictive relative clause would not be affected. Indeed, this is what we find in the following Bafut example. mə́  | ì̵ (38) mə̀ kì̵ bìtí̵ lùm || mə́ bí̵ kɔ́ ŋ ndánwì já sbj.1sg tns ask Lum rel.m rp like church the rel.m rp á tú ntáʔá láa || mə́ nwì tsí á búrí̵ lɛ́ tsí stay on top hill rel.def that God stay in heaven q “I asked Lum, who likes the church (which is) on the hill top whether God is in heaven.”



7 A prolegomenon to the syntax of the relative clause 

 163

To sum-up the discussion so far, it is clear from Tab. 4 below that Medumba and Bafut, as two Eastern Grassfields Bantu languages, exhibit a high degree of resemblance with respect to relative clause construction, with only minor differences, that is, specifically in the use of the relative complementizer and in the combined use of both the relative determiner and the relative complementizer. Tab. 4: Defining features of relative clauses in Medumba and Bafut. Defining Feature

Medumba (Nun) Bafut (Ngemba)

Use of a relative marker/complementizer Use of a relative determiner (pronoun) Combined use of relative marker and relative determiner Use of a subject marker after a (relativized) lexical subject Use of relative definiteness particle Presence of phonetic breaks or post-COMP-subjectspellout effect Subject, direct object, indirect object, oblique object and post-possessor relativization Subject resumption Direct object resumption Indirect object resumption Oblique object resumption Post-possessor resumption

No Yes No Yes Yes Yes

Yes Yes Yes Yes Yes Yes

Yes

Yes

Yes Yes Yes Yes Yes

Yes Yes Yes Yes Yes

5 Possible A-bar movement in the Bafut/ Medumba relative clause In this section, we examine Bafut for the possibility that relativization in it and Medumba involves A-bar movement. Indeed, there is ample evidence in Bafut that this is the case (see Cheng & Downing 2010 and Cheng 1997 for literature on A-bar movement). In addition to the fact that a RP surfaces in some of the positions relativized, empirical generalisations from the distribution of idiom chunks, binding, and reconstruction effects support a movement analysis in Nun and Ngemba languages. We illustrate with the case of reconstruction (scope properties) and bound pronoun interpretation in Bafut. Reconstruction effects in Bafut are illustrated by the following examples which are similar to “portrait of himself” constructions.9

9 “Portrait-of-himself constructions” are a set of specific constructions which have been used in generative syntax to argue for various kinds of reconstruction effects.

164 

 Pius N. Tamanji and Blasius Achiri-Taboh

(39) a. ŋùŋgàŋ ntsì̵m ká jə́ ŋgáŋmí̵ɣɔ́ ɔ́ jíbáà doctor every tns see patients two “Every doctor will see two patients.” b. mə̀ kì̵ twòŋ ŋgáŋmí̵ɣɔ́ ɔ́ já jíbáá mə́ sbj.1sg tns call patients the two rel.m ŋùŋgàŋ ntsí̵m ká jə́ láà doctor every tns see rel.def “I called the two patients that every doctor will see.” In these examples, the nominal head can be interpreted as having narrow scope with respect to another quantifier within the relative clause. (39a) contains the object QP Ngáŋmí̵ghɔ́ ɔ́ jíbáà “two patients” which can be interpreted as having narrow scope with respect to the subject QP “every doctor”, that is, there can be twice as many patients as doctors. In (39b), the relativized nominal “(the) two patients” can also be interpreted as in (39a), that is, as having narrow scope with respect to the subject QP in the relative clause. This fact indicates that the Head can be interpreted in the direct object position, and hence that it must have been raised from that position. The distribution of bound pronouns also exhibits reconstruction effects. jì mə́ mú (40) a. fútò tàá ntsì̵m kɔ̀ ŋ láa photo father his rel.m child every like rel.def à ní̵ jìŋwì sm is big “A statue of his father which every child likes is a big one.” b. bìʔì kì̵ jə́ fútò tàá jì mə́ mú ntsì̵m kɔ̀ ŋ láà we tns see photo father his rel.m child every like rel.def “We saw the photo of his father that every child likes.” In the examples in (40), the relativized nominal binds a pronoun contained within the relative clause. This is evidence that there has been movement from the position occupied by the bound pronoun (i.e., the RP) inside the relative clause.

6 Conclusion In this chapter, we have described the morpho-syntactic properties of relative clause constructions in two Eastern Grassfields Bantu languages. Thus, to start with, we have discussed the various strategies used in relative clause constructions in both languages, showing that while Bafut in the Ngemba sub-group



7 A prolegomenon to the syntax of the relative clause 

 165

of the Eastern Grassfields Bantu languages employs a relative determiner, in agreement with the relativized antecedent, that may either alternate or collocate with a regular (relative) complementizer, Medumba in the Nun sub-group employs a relative determiner alone, equally in agreement with the relativized antecedent. We have also shown that the two languages however exhibit striking similarities with respect to prosody, the exhibition of resumptive pronouns and in the use of the particle láá /lá whose principal role has been shown to mark definiteness in the relative clause. Specifically, we have discussed the presence of post-COMP-subject-spellout effects (i.e., phonetic breaks) within the relative clause in both languages, usually at both the left and right peripheries, showing that these mainly occur between an overt relative marker/ complementizer and an overt subject. We have further discussed the presence of resumptive pronouns in relativized positions, showing that all relativized positions in both languages exhibit resumption of some sort. Finally, we have looked at the possibility of A-bar movement within the relative clause of the languages and shown that empirical generalizations from the distribution of binding and reconstruction effects support a movement analysis for relative clause constructions in them.

References Abney, Steven P. 1987. The English noun phrase in its sentential aspect. Massachussetts: Massachussetts Institute of Technology dissertation. Aboh, Enoch O. 2004. The Morphosyntax of complement-head sequences: Clause structure and word order patterns in Kwa. New York/Oxford: Oxford University Press. Aboh, Enoch O. 2005. Deriving relative and factive clauses. In Laura Bruge, Giuliana Giusti, Nicola Munaro, Walter Schweikert & Guiseppina Turano (eds.), Contributions to the thirtieth incontro di grammatical, 265–285. Venezia: Libreria Editrice Cafoscarina. Aboh, Enoch O. 2006. Complementation in Saramaccan and Gungbe: The case of c-type modal particles. Natural Language and Linguistic Theory, 24 (1). 1–55. Achiri-Taboh, Blasius. 2004. Wh-extraction and resumption in relative clause constructions. In Evangelia Daskaliki, Napoleon Katsos, Marios Mavrogiorgo & Matthew Reeve (eds.), Proceedings of the University of Cambridge Second Postgraduate Conference in Language Research, 309–316. Cambridge: Cambridge Institute of Language Research. Achiri-Taboh, Blasius. 2006. A morpho-syntactic analysis of Moghamo-Ngamambo: The syntax of resumption. Frankfurt: University of Frankfurt-Main dissertation. Aoun, Joseph & Audrey Li Yen-hui. 1990. Minimal disjointness. Linguistics 28. 189–203. Boeckx, Cedric. 2003. Islands and chains: Resumption as derivational residue. Amsterdam/ Philadelphia: John Benjamins. Beltzung, Jean-Marc, Annie Rialland & Martial Embanga Aborobongui. 2010. Les relatives possessives en mbochi (C25). In Laura Downing, Annie Rialland, Jean-Marc Beltzung,

166 

 Pius N. Tamanji and Blasius Achiri-Taboh

Sophie Manus, Cédric Patin & Kristina Riedel (eds.), Papers from the Workshop on Bantu Relative Clauses 53, 7–31. Berlin: Zentrum für Allgemeine Sprachwissenschaft. Borer, Hagit. 1984. Resumptive relatives in Modern Hebrew. Natural Language and Linguistic Theory 2. 219–260. Cheng, Lisa L.-S. & Laura Downing. 2010. Locative relative clauses in Durban Zulu. In Laura Downing, Annie Rialland, Jean-Marc Beltzung, Sophie Manus, Cédric Patin & Kristina Riedel (eds.), Papers from the Workshop on Bantu Relative Clauses 53, 33–51. Berlin: Zentrum für Allgemeine Sprachwissenschaft. Chomsky, Noam. 1986. Knowledge of language. New York: Praeger. Chomsky, Noam. 1995. The minimalist program. Cambridge, MA: Massachusetts Institute of Technology Press. Chomsky, Noam & Howard Lasnik 1977. Filters and control. Linguistic Inquiry 8. 425–504. Citko, Barbara. 2002. (Anti)reconstruction effects in free relatives: A new argument against the Comp account. Linguistic Inquiry 33. 507–11. Downing, Laura & Al Mtenje, 2010. The prosody of relative clauses in Chewa. In Laura Downing, Annie Rialland, Jean-Marc Beltzung, Sophie Manus, Cédric Patin & Kristina Riedel (eds.), Papers from the Workshop on Bantu Relative Clauses 53, 53–67. Berlin: Zentrum für Allgemeine Sprachwissenschaft. Kisseberth, Charles. 2005. Accent and phrasing in Chimwiini. In Shigeki Kaji (ed.), Proceedings of the Symposium: Cross-linguistic Studies of Tonal Phenomena, 129–145. Tokyo: Research Institute for Languages and Cultures of Asia and Africa (ILCAA) Tokyo University of Foreign Studies. Kisseberth, Charles. 2010a. Optimality theory and the theory of phonological phrasing: The Chimwiini evidence. In Nomi Erteschik-Shir & Lisa Rochman (eds.), The sound patterns of syntax, 217–246. Oxford/New York: Oxford University Press. Kisseberth, Charles. 2010b. Phrasing and relative clauses in Chimwiini. In Laura Downing, Annie Rialland, Jean-Marc Beltzung, Sophie Manus, Cédric Patin & Kristina Riedel (eds.), Papers from the Workshop on Bantu Relative Clauses 53, 109–144. Berlin: Zentrum für Allgemeine Sprachwissenschaft. Kisseberth, Charles & Abasheikh Mohammad Imam. 1974. Vowel length in ChiMwini – a case study of the role of grammar in phonology. In Anthony Bruck, Robert A. Fox & Michael W. La Galy (eds.), Papers from the Parasession on Natural Phonology, 193–209. Chicago: Chicago Linguistic Society. Kouankem, Constantine. 2012. The syntax of the Medumba determiner phrase. Yaounde: University of Yaounde I dissertation. Longobardi, Giuseppe. 1994. Reference and proper names. Linguistic Inquiry 25. 609–66. Lyon, Christopher. 1999. Definiteness. Cambridge: Cambridge University Press. McCloskey, James. 1979. Transformational syntax and model theoretic semantics: A case study of modern Irish. Dordrecht: Reidel. McCloskey, James. 1990. Resumptive pronouns, A’-binding and levels of representation in Irish. In Randall Hendrick (ed.), Syntax and semantics 23: The syntax and semantics of modern Celtic languages, 199–248. New York / San Diego: Academic Press. McCloskey, James. 2000. Quantifier float and wh-movement in an Irish English. Linguistic Inquiry 31. 57–84. McCloskey, James. 2002. Resumption, successive cyclicity, and the locality of operations. In Samuel D. Epstein & T. Daniel Seely (eds.), Derivation and explanation in the minimalist program, 184–226. Malden, MA: Blackwell.



7 A prolegomenon to the syntax of the relative clause 

 167

Mfonyam, Joseph. 1989. Tone in orthography: The case of Bafut and related languages. Yaounde: University of Yaounde dissertation. Postal, Paul M. 1966. On so-called pronouns in English. In Francis Dinneen (ed.), Nineteenth monograph on language and linguistics. Washington DC: Georgetown University Press, [reprinted in David A. Reibel, and Sandford Schane (eds.). 1969. Modern studies in English. Englewood Cliffs, NJ: Prentice-Hall, 201–24]. Quirk, Randolph & Sidney Greenbaum. 1973. A grammar of contemporary English. Essex: Longman, Harlow. Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman (ed.), Elements of grammar, 281–237. Dordrecht: Kluwer. Sag, Ivan. 1997. English relative clause constructions. Journal of Linguistics 33. 431–83. Shlonsky, Ur. 1992. Resumptive pronouns as a last resort. Linguistic Inquiry 23. 443–468. Tamanji, Pius. 2009. A descriptive grammar of Bafut. Köln: Rüdiger Köppe Verlag. Vries (de), Mark. 2002. The syntax of relativization. Utrecht: Landelijke Onderzoekschool Taalwetenschap dissertation.

Jesse Lovegren and Rebecca Voll

8 Relative clause constructions in two Yemne-Kimbi languages Abstract: Relative clause constructions are profiled for two Yemne-Kimbi (Bantoid) languages of Lower Fungom (North West Region), Mungbam and Mundabli. Both languages have complex phonological systems, with large vowel inventories, four tone levels, and non-affixal morphological processes. Noun class concord systems are relatively well-preserved in these languages (though moreso in Mungbam than in Mundabli) compared to Grassfields languages further to the South. Properties surveyed include ordering of constituents within the noun phrase, the accessibility of different arguments to relativization, possible historical sources of the relativizer, and differential realization of Tense, Aspect, Mood, Polarity and Focus categories in relative clauses. Parallel coverage of topics offers insight into fine grammatical differences between these two related languages. Data are drawn from texts and elicitations collected in the field between 2008–2012 by the authors.

1 Introduction The purpose of the present paper is to describe basic facts about relative clause constructions in two related languages spoken at the northern edge of the ­Cameroonian Grassfields, Mungbam and Mundabli. Though the two languages are not mutually intelligible, there is considerable contact between their speakers, Acknowledgements: This work was made possible by support from NSF grant BSC-0853981, University at Buffalo College of Arts and Sciences Humanities Institute, Leiden University Fund and Leiden University Centre for Linguistics at Leiden University. Thanks to Jeff Good for helpful comments on earlier drafts of this paper. We owe a special debt to a large number of language consultants who provided data, foremost of which are Kang Protus (Biya), Nchang Adeline (Abar), Ngong Belta (Munken) and Yung Donatus Kungmba (Mundabli). Mungbam data were collected by Lovegren during two field trips, in 2010 and 2012, totaling seven months. Access to Ngun consultants was limited in the first trip and non-existent in the second trip, so Ngun data is rather sparse compared to data from the other dialects. Mundabli data were collected by Voll during three field trips, in 2008, 2009 and 2012, totaling eight months. Jesse Lovegren: Texas Commission on Environmental Quality, [email protected] Rebecca Voll: Leiden University College for Linguistics (LUCL), Leiden University, [email protected] DOI 10.1515/9783110469547-008

170 

 Jesse Lovegren and Rebecca Voll

and basic as well as non-trivial formal properties relevant to relative clause ­constructions are to a large extent isomorphic in the two languages. We therefore find it profitable to present relative clause constructions in the two languages as variations on a single basic system.

1.1 Organization of the paper In the remainder of § 1, we give basic geographic information about the two languages and where they are spoken (§ 1.2.1), and outline their major typological properties, especially those relevant to the analysis of relative clause constructions (§ 1.2.2). In sections 2–3, we proceed to a more detailed description of the properties of relative clause constructions for Mungbam, and then for Mundabli. These sections focus on parallel themes: the linear order of the relative clause with respect to the head noun, and with respect to other nominal modifiers; properties of the relativizer and comments on possible grammaticalization sources; the status of what are typically referred to as resumptive pronouns, or “representative nominals,” as we call them here, and the accessibility of different types of formally distinct grammatical relations to relativization.1 Section 4 concludes the paper with a discussion summarizing the key points of similarity and difference between the two languages, and draws attention to the most typologically interesting points.

1.2 Language background 1.2.1 Geographic and sociolinguistic background Mungbam and Mundabli are both spoken within a small part of Cameroon’s Northwest Region near the Nigerian border known as Lower Fungom. As can be seen from the map in Fig. 1 (Di Carlo, 2011: 57), no more than 10 km separates Mundabli from the most distant of the Mungbam villages. The acronym “Mungbam”, associated with ISO 693–3 code [mij], is used to refer to the speech varieties used predominantly in the villages of Munken, Ngun, Biya, Abar and Missong (see Fig. 1). There is no locally recognized name to refer to these five more or less mutually intelligible speech varieties, and no

1 Here we refer to the concept developed in the works of Keenan & Comrie (1977, 1979a, b).

8 Relative clause constructions in two Yemne-Kimbi languages 



 171

Mashi Overside

e m

n

Munken Missong Mashi

Biya

Y Yemgeh Mekaf

Kung

Zhoa Fungom

Buu Ajumbu

Fang

Subum

M To

Nyos

10 km

LEGEND

Footpath Motorable road Ring Road River Uninhabited

Yemne-Kimbi Mungbam [mij] Ji group [boe] Koshin [kid] Fang [fak] Ajumbu [muc]

je

isa

Weh

m

u oW

T

Koshin

Abar

Ngun

b u m

Small Mekaf

M

Isu

b i K i m

e

Mundabli

Mufu

Beboid Naki [mff]

Nigeria Chad

Central Ring Kung [kfl] Mmen [bfm] Bum [bmv] West Ring Isu Weh Zhoa

Mixed community

N

Ring Road Bamenda Cameroon Yaoundé Equatorial Guinea

Fig. 1: Lower Fungom and surrounding area (Di Carlo, 2011).

Central African Republic

Gabon Republic of Congo

172 

 Jesse Lovegren and Rebecca Voll

realistic chance that Mungbam speakers would agree to a name which groups the five villages together (see Di Carlo & Good (2014) for discussion on this point). Mundabli is the name of the speech variety used in the village of the same name in Lower Fungom. Though the Ethnologue entry for “Mundabli” (associated with ISO 693–3 code [boe]) also includes the dialects spoken in the villages of Mufu and Buu (see Fig. 1), this chapter only concerns the variety spoken in Mundabli proper (Lewis et al., 2015). Mundabli, Mufu and Buu are referred to as the Ji cluster in Good et al. (2011). The genetic classification of the two languages is uncertain at this point. Although it is uncontroversial to consider both as Bantoid languages under the Benue-Congo branch of Niger-Congo, the nature of their relationship to neighboring languages and to each other will remain uncertain until further comparative work is undertaken. For the meantime, the referential classification term “YemneKimbi”, comprising the Lower Fungom languages Mungbam, Ji, Koshin, Fang and Ajumbu, has been proposed by Good et al. (2011) to replace the presently-­ unsupported genetic label “Western Beboid.” This point notwithstanding, however, our ­experience studying the two languages has made it clear that the languages overlap considerably in their overall grammatical structure. What is uncertain is the extent to which this overlap is due to contact, rather than common inheritance. As is shown in Fig. 1, there are non-Yemne-Kimbi languages spoken in close vicinity to both Mungbam and Mundabli. The most relevant of these is the Beboid language Naki [mff], which is in especially tight contact with both languages. The Mungbam villages Biya and Ngun, for example, are as close to the Naki-speaking village Small Mekaf as they are to any other Mungbam-speaking village. Likewise for Mundabli, the closest village is not a Ji-speaking village, but the Naki-speaking village Mashi. Various West and Central Ring languages are spoken in and around Lower Fungom. North of Lower Fungom on the Cameroonian side of the Nigeria-Cameroon border are found Jukunoid languages, including Yukuben, Akum, Beezen and Baazem (not shown in Fig. 1, but see Breton (1993)). 1.2.2 Typological background Basic information about the grammatical systems of Mungbam and Mundabli, to the extent they were understood at the time of publication, is presented in Good et al. (2011). Since our understanding of the two languages has evolved since the time of publication of that article, any conflicting information presented in the present work should be considered as superseding the earlier article. Lovegren (2013), a grammar of Mungbam, was completed during the time that this volume was under editorial review.



8 Relative clause constructions in two Yemne-Kimbi languages 

 173

Both languages are rather complex phonologically. Vowel inventories are relatively large, and vowels in Mundabli contrast for pharyngealization. Both languages contrast four level tones in addition to contour tones, and make extensive use of tone in their verbal morphologies. Words tend to be disyllabic in Mungbam, and mostly monosyllabic in Mundabli. Both languages have well-articulated systems of noun class agreement: All nouns are lexically associated with a noun class, and modifiers of the noun exhibit concord which is controlled by the noun class of the head noun. The two languages differ in that while noun class is for the most part overtly marked on nouns in Mungbam via prefixation, noun class in Mundabli is instead to a considerable extent “covert”, as it is for the most part only observable via its ability to control concord agreement.2 Table 1 illustrates this point, showing some cognate forms from different noun classes in Mundabli and the Mungbam dialect of Munken. Tab. 1: Forms illustrating the loss of segmental noun class prefixes in Mundabli vs. their retention in Munken. Mundabli

Munken

gloss

gbɔ̄ dzɔ̄ yɩ ̋ yɩ ̋ kʊ̄ kʊ̄ ʤǔ ʤű

ú-kpe̋ í-kpe̋ í-dzɛ̋hɛ á-dzɛ̋hɛ à-kə᷆ fə bī-kə᷆ fə ì-bé í-be̋

‘cl3.house’ ‘cl4.houses’ ‘cl3.eye’ ‘cl7a/cl6.eyes’ ‘cl7/cl12.bone’ ‘cl8.bones’ ‘cl9.goat’ ‘cl10.goat’

In both languages, verbs may be divided on the basis of tonal alternations into three different conjugation classes, labeled ‘(a)’, ‘(b)’, and ‘(c)’. Cognate verbs in the two languages tend to fall into the same conjugation class. Verbs inflect via tone changes and stem vowel mutations, the latter of which are conditioned by a perfective/imperfective aspectual distinction. Tense marking is produced by a combination of verb stem changes and overt preverbal tense markers. Both Mundabli and Mungbam distinguish four degrees of remoteness in the past (glossed p0, p1, p2, and p3 here). Mungbam has one future tense, while Mundabli has two (glossed fut1 and fut2). In Mungbam, tone changes in the verb encode a four-way formal distinction: between perfective and imperfective aspect on one dimension,

2 See Good (2012: §4) for a discussion, with specific reference to Mundabli and Mungbam.

174 

 Jesse Lovegren and Rebecca Voll

and between realis and irrealis mood on the other (see Tab. 2). The latter distinction is mostly a formal one, as the category “Irrealis” has an idiosyncratic membership in Mungbam: subjunctive, jussive, remote past, and one type of negation construction, but not future and not a second type of negation construction. The tonal morphology of Mundabli verbs is a bit more complicated, and cannot be succinctly summarized here (but see Voll (2012) for details). Tab. 2: Verb conjugation in Mungbam. Tone diacritics are interpreted as follows: x̀ = L, x̄ = M, x᷆ = ML, x́ = H, x̂ = HL, x̋ = S(uperhigh). abl = ‘ablaut’. On disyllabic verbs, the patterns are identical, with the pitch modulation spread across the two syllables.

a b c

realis irr realis irr realis irr

pfv

ipfv

x̀ x̄ x̋ x᷆ x̂ x́

x̀ + abl x̄ + abl x̂ + abl x́ + abl x̂ + abl x́ + abl

Basic constituent order in both languages is SV/SVO, though arguments are frequently dislocated to different positions within the clause for focus-marking. More specifically, the position immediately after the verb (IAV) is considered a structural focus position in both languages. This means that VS constituent order is observed in subject-focus constructions, and OV constituent order is observed in object-defocalization constructions. In Mungbam, a productive process of verbal reduplication is available for the encoding of verum focus (contrastive or assertive focus on the truth value of a clause). Reduplication is however under partial grammatical control (cf. Hyman & Watters (1984: 243)) when certain marked constituent orders are observed, as well as in relative clauses (see § 2.6). Both languages make extensive use of serial verb constructions in which all of the verbs form a continuous block in the middle of the clause, without any intervening arguments. In this paper the term “verbal complex,” taken over from Kießling (2011), refers to this continuous block of one or more verbs. With some minor differences, the two languages have broadly similar systems for formally marking different types of grammatical relations, with the following formally distinct possibilities available: –– Subject (ø-marked, normally appears before the verb). Some preverbal subject pronouns differ from postverbal object pronouns. –– Object (ø-marked, normally appears immediately after the verb) –– Comitative (preceded by particle glossed com)



8 Relative clause constructions in two Yemne-Kimbi languages 

 175

–– Dative (optionally preceded by a general-purpose preposition, followed by a particle glossed dat)3 –– Locative (optionally preceded by a general-purpose preposition, followed by one of several possible postpositions) –– Genitive (ø-marked, follows the head noun) We will occasionally refer to these grammatical relations in our discussion of accessibility.

1.3 Notes on terminology and transcription Transcription symbols used in the examples are mostly consistent with the corresponding IPA symbols, with a few minor exceptions. The use of the [x̤ ] diacritic, normally reserved to indicate breathy voice, is used to indicate what has been described as pharyngealisation of vowels in Mundabli. Also, the symbol [y] indicates a palatal glide in Mundabli. In Mungbam a palatal glide is indicated by the IPA symbol [j]. Additionally, we prefer to use the now outdated symbol [ɩ] rather than the currently-recommended [ɪ] to transcribe a high, front, unrounded rootretracted vowel, simply for the reason that when tone diacritics are included, the latter symbol is easily confused with dotless [ı]. For terminological convenience, we use the term ‘head nominal’ to refer to the head of a noun phrase which contains a relative clause, and the term ‘representative of the head nominal’ (or simply ‘representative nominal’), to refer to a noun or pronoun coreferent with the head nominal which occurs inside of the relative clause itself. Both terms are due to Lehmann (1986: 664, 673). We also use the term ‘matrix NP’, due to Andrews (2007: 206), to refer to the noun phrase headed by the head nominal. Likewise, we use ‘matrix clause’ to designate the clause containing the matrix NP. Since a discussion of noun

3 The comitative is so called because it is prototypically used to encode accompaniment. It is also used with instrumental function. The dative is so called because it is prototypically used to encode recipients. It is also used with benefactive function. We employ the terms “comitative phrase” and “dative phrase” as necessary to refer to a noun phrase together with its associated comitative- or dative-marking morphology. Dative-marked arguments in Mungbam and Mundabli overlap in function somewhat with arguments licensed by reflexes of the verbal suffix *-ɩl- in Narrow Bantu languages, commonly called “applicative,” though called “dative” by Schadeberg (2003: 74). In referring to the grammatical function in Mungbam and Mundabli, we prefer the latter term, since “applicative” is typically associated with head-marking strategies for encoding grammatical roles, while “dative” is typically associated with dependent-marking strategies.

176 

 Jesse Lovegren and Rebecca Voll

class concord figures prominently in several places in this paper, we use the term ‘concordant’ as a short way of referring to a grammatical particle, viz. a relativizer, which shows noun class concord with the head nominal. Likewise, ‘non-concordant’ will designate a grammatical particle which does not show concord. In examples, we enclose relative clauses in square brackets, and indicate head nominals and their representatives within the relative clause with underlining. In § 2, a name in parentheses following the free translation refers to the dialect from which the example has been taken. Examples with no such annotation are Mundabli examples. One final comment concerns the source of the example sentences. Example sentences drawn from naturally-occurring speech are generally to be preferred. However, an elicited sentence is often the best way to clearly exemplify a particular type of structure to the audience of linguists. At a more practical level, elicitation is the most convenient way for a visiting field linguist to test hypotheses that they develop through the examination of texts, and is often the only sensible way to do so, given time constraints. Examples used in this paper, then, come from both elicited and naturally-occurring speech. Elicited sentences are indicated by a ◊ symbol in the free translation. Ungrammatical examples, all of which are elicited, are indicated by an asterisk (with no ◊ symbol).

2 Mungbam 2.1 Introduction Data from Mungbam will be drawn freely from all five varieties, with the understanding that the same facts apply to each variety, except where otherwise mentioned. We first consider superficial properties of the relative clause construction, including the order of the relative clause with respect to other nominal modifiers (§ 2.2). We then consider properties of the relativizer, including its likely historical source (§ 2.3). Section 2.4 concerns the representative nominal, and § 2.5 discusses the accessibility of different grammatical relations to relativization. Finally, § 2.6 compares the possibilities for marking various clause-level properties in main and relative clauses. Special attention is given to focus-marking possibilities.



8 Relative clause constructions in two Yemne-Kimbi languages 

 177

2.2 Basic order of constituents in the NP The unmarked constituent order within NP’s in Mungbam is such that the head noun is initial. The head noun may be modified by possessive pronouns, demonstratives, adjectives, numerals, the definite determiner, and relative clauses, all of which, save the relative clause, show some form of concord with the noun class of the head noun in all Mungbam varieties.4 The order of certain other modifiers of the head noun is as in (1), exemplified in (2). Adjectives and numerals, when modifying the head noun in a noun phase,5 have a more variable ordering with respect to other nominal modifiers, so they cannot be easily incorporated into the schema given in (1). Since the relative clause appears after the head nominal, but before the definite determiner, as in (2), it is rather straightforward to argue that the relative clause is internal to the matrix NP. (1) N –Poss –Dem–RC– Det (2) í-sɛ᷇ hɛ mə́ jɛ̂ n nə́ [ ɲ̋-ɲɛ́ mē cl4a-place isg.poss cl4a.dem.prox rel cl6a-water cl6a.det ɲɛ̀ ] jɛ̄ (a)stay.pfv cl4.det ◊‘This my place where the water is…’ (Munken)6 Departures from the determiner-final constituent order given in (1) are found under different scopal interpretations of definiteness with respect to other constituents of the NP. While the normal state of affairs is for either all or none of the information in the NP to be given a definite interpretation, corresponding to either the appearance of the determiner NP-finally or its complete absence, some cases are possible where the head nominal is given a definite interpretation, but one of its modifiers is given an indefinite interpretation. Example (3), for example, where the relative clause follows the definite determiner, is felicitously uttered in

4 Concord on possessive pronouns is, however, significantly eroded. Only a two-way tonal distinction is realized on pronouns, depending on whether the possessum noun class is one of {1, 5, 9}, or some other class. 5 Adjectives and numerals may also head noun phrases. 6 Glossing abbreviations follow the Leipzig Glossing Rules (Bickel et al., 2008). Africanist-specific and uncommon abbreviations are defined in the introductory chapter of this volume. Verbs not glossed as ‘ipfv’ are in their perfective forms. Mungbam verbs not glossed as ‘irr’ are in their realis forms.

178 

 Jesse Lovegren and Rebecca Voll

a context where the listener is familiar with the group of fowls being discussed, and familiar with the set of red fowls within that group, and also familiar with the fact that some fowls within the group fell, but did not know beforehand that the fowls having fallen were the red ones. (3) í-ɕɛ̋ jɛ̄ nə́ [ í gbè] í-bwɩ́nsə́ cl10-fowl cl10.det rel cl10 (a)fall cl10-red ◊‘The fowls which fell are the red ones.’ (Munken)

jɛ̄ cl10.det

2.3 The relativizer In the Munken examples (2)–(3), the relativizer, which immediately precedes and introduces the relative clause, does not show concord with the head nominal. In all of the Mungbam varieties, except for Biya (to be discussed below), the relativizer is non-concordant. These (non-)agreement facts are noteworthy because they not only separate Mungbam from other Grassfields languages further to the South, which have concordant relativizers,7 but they also can offer a clue about possible grammaticalization sources for the relativizer. If the relativizer is nonconcordant, then it is less likely that it could have grammaticalized from an element which regularly shows concord, such as a demonstrative.8 Ruling out the demonstrative as a grammaticalization source is significant since demonstratives are a common grammaticalization source for relativizers in African languages (Heine, 2011: 706), and the “most frequent” source cross-linguistically (Heine & Kuteva, 2002: 115). 2.3.1 Possible sources for grammaticalization of the relativizer In Biya, the relativizer does show concord with the head nominal. In that variety, the relativizer coincides exactly in form with the word -nɩ ‘some, another’, which takes a noun class prefix and whose tone is also controlled by the class of the noun

7 Cf. for Isu (Ring) (Kießling, 2011: 39, passim), Noni (Beboid) (Hyman, 1981: 91), Bafut (Ngemba) (Tamanji, 2009: 96–7). Mankon (Bamiléké) (Leroy, 2007: 415), and Ngwe (Bamiléké) (Nkemnji, 1995: 77). An exception is the Bamiléké language FeˀFeˀ (Chumbow, 1977: 289), which otherwise has lost much of its concord morphology (Hyman, 1972: §VII). 8 A logical possibility is of course that the relativizer grammaticalized from a concordant form which eventually lost its concord. However, with the exception of possessive marking, the concord system of Mungbam quite well-preserved.

8 Relative clause constructions in two Yemne-Kimbi languages 



 179

it agrees with.9 Examples are given in (4). The only difference between the two is that the prefix of the relativizer is occasionally omitted in casual speech, as in (4c). Since the agreement facts between the two morphemes are identical, it seems likely that the relativizer in Biya has grammaticalized from -nɩ ‘some, another’. (4) a. ī-ɕɛ̄ jī à tù-ā ī-ɕɛ̄ ì-nɩ̄ cl9-fowl cl9.det sbj (a)peck-prf cl9-fowl cl9-other ◊‘The fowl has pecked another fowl.’ (Biya) b. ū-bí ū-nɩ́ [ú-kpə̋ kə̄-ŋwāa sū cl3-year cl3-rel cl3-house cl12-book (a)start.irr kwí á wáhá] wɩ̄ (c)enter.irr prep Zhoa cl3.det ‘The year when the school at Zhoa was opened…’ (Biya) c. wə̌ kə́ -mjɩ̋ nɩ́ [ɲ́-ɕî ń-dé ɕī then cl12-matter rel 1sg-(b)want 1sg-(c)talk.irr (a)discuss.irr lē] kə̄ ja᷆ vent.irr cl12.det thus ‘That’s the problem that I want to talk about…’ (Biya) In the other dialects, where concord between the relativizer and the head nominal is not observed, the forms of the relativizer are quite similar, though the forms of the ‘some, other’ morpheme tend to vary (see Tab. 3), such that this morpheme is not the most obvious grammaticalization source. In Tab. 3 we give forms for two other morphemes which might be candidates for grammaticalization sources. Tab. 3: Forms of relativizer, and of three suspected grammaticalization sources for the relativizer in Mungbam. Lack of tone marking indicates that the tone of the form depends on the prefix it bears. Variety

rel

‘some’

‘reason’

‘own’ (associative)

Munken Ngun Biya Abar Missong

nə́ nɩ ́ -nɩ ɲí ná

-lɛ -ne -nɩ -lehe -lɛ

á-mjű kə-mjɩ ̋ kə-mjɩ ̋ kə-ɲű kí-na̋

-ɲəmə -kjɩ -ɲi -nə̄

We note that in Abar, the word translatable as ‘one’s own’, a type of emphatic associative marker, coincides in segmental form (but not in tone) with the relativizer.

9 It bears a mid tone for agreement with class 1, 5 and 9, nouns, and a high tone for all agreement with nouns of all other classes.

180 

 Jesse Lovegren and Rebecca Voll

The associative marker as a grammaticalization source for the relativizer is not mentioned in Heine & Kuteva (2002: 335), though Ohori (2011: 641–2) considers it to be a source “… with clear conceptual-semantic ground …” Such a grammaticalization scenario is claimed for some Chadic languages (Frajzyngier, 1996: §11), and a relativizing particle in Vute (Mambiloid) is treated by Maxey (1994: §3.1) as being the same as an associative marker. A more relevant case is the neighboring language Naki, where formally identical particles serve as both a pre-RC relativizer and an associative marker (Jeff Good, p.c.).10 The root of the word meaning ‘reason, problem, matter’ in Missong coincides with the relativizer in that language. The root -mjV̋ in Biya, Munken, and Ngun also means ‘word’11 (similar forms with that meaning are attested in Abar and Missong), and it appears to have supplanted the original term in Ngun and Biya. Biya and Munken retain roots kə-nő, á-ɲı ̋ ‘thing’, of the same noun class as Abar kə-ɲı.̋ A possible scenario for grammaticalization of the relativizer from a word meaning ‘matter’, which at this point remains speculative, is as follows: The word translatable as ‘reason, problem, matter’, was originally found in all five varieties in a form close to that observed in Missong or Abar. This form was the original source for the relativizer in all of the dialects, and possibly was the source for the emphatic associative marker translated as ‘own’ in Abar, Missong and Munken. Since the prefix in the original word was fixed, the relativizer never showed concord with the head noun, and this prefix was eventually lost. Some time after the grammaticalization of the relativizer from *kV-nV, the original root meaning ‘matter’, was replaced in Biya, Munken, and Ngun by the root whose original meaning was ‘word’. The relativizer in Biya was eventually reanalyzed as an instantiation of the morpheme -nɩ ‘other’. Heine & Kuteva (2002: 211–2, 295) note that a word with the meaning ‘thing’ or ‘matter’ tends to be the grammaticalization source for a complementizer, though no such source is reported for a relativizer. In summary, we have identified two of the most likely historical sources for the relativizer in Mungbam: an emphatic associative marker, and a lexical item ‘thing, matter’. At present it is not clear which of these is the more likely.

10 See (51), p. 197 for an example of a Naki relative clause. 11  Interestingly, in Munken, a subordinating particle mjú, clearly related to á-mjű, is attested in a type of aversive construction, translatable as ‘lest’. (1) bə́ ʤû bɛ̋hɛ mjú ɲ̋-ɕám mē á tɔ̀ŋ=ā cl2 (c)go.away (b)exit lest cl6a-blood cl6a.det neg (a)touch=prf á bə̄ mə̄ prep cl2.loc.obj loc.at ‘They have moved away lest the blood [of the freshly slaughtered pig] touch them.’ (Munken)



8 Relative clause constructions in two Yemne-Kimbi languages 

 181

2.3.2 Optionality of the relativizer The the relativizer is optional in Mungbam. There are, however, differences between different types of relative clauses with respect to the rate of its omission. While the relativizer is usually omitted in cleft sentences, it is usually not omitted in non-cleft sentences. Example (5) shows two repetitions of the same line in a folk tale told by different speakers, which differ by the presence vs. the absence of the relativizer. In (5a), the original telling of the story, the relativizer is omitted from the cleft construction, and the copula verb is reduced. In (5b), a careful repetition of the same line by a different speaker, the relativizer is present. ǎ fwe̋ [ù ɕô fwôm lànɔ̀ (5) a. ds.cop there cl1 (b)stay.day.ipfv (b)struggle.ipfv (a)walk.ipfv kɔ̄ mɛ̀ bì-dzāŋ bī fwə̀ mhɔ̄ wə̀ í-dāŋ ipfv com cl8-fly cl8.det (a)disturb cl1 loc.cl5-comb ì ŋə̄ n] cl5.det loc.on ‘It’s there [i.e., that’s why] he is constantly struggling with flies disturbing him [landing] on his comb.’ (Abar) b. à lɛ̋ fwe̋ ɲí [ù ɕô fwôm ds (b)cop there rel cl1 (b)stay.day.ipfv (b)struggle.ipfv lànɔ̀ kɔ̄ mɛ̀ bì-dzāŋ bī í-dāŋ (a)walk.ipfv ipfv com cl8-fly cl8.det loc.cl5-comb ì ŋə̄ n] cl5.det loc.on ◊‘It is there that [i.e., that’s why] he is constantly struggling with flieson  his comb.’ (Abar)

2.4 The representative of the head nominal All relative clauses in Mungbam may contain a representative of the head nominal, though its presence is only mandatory when the representative nominal functions as the subject of the relative clause.12 In a subject relative clause such as (6), a pronoun will always follow the relativizer.

12 A similar asymmetry between extracted subject and non-subjects is reported for Vata (Kru), which requires an overt representative nominal for subject relatives, but requires its absence for non-subject relatives (Koopman & Sportiche, 1986: 361).

182 

 Jesse Lovegren and Rebecca Voll

(6) kpòa ná [bű gbə̀ ] bū five rel cl2 (a)fall cl2.det ◊‘The five that fell…’ (Missong) For all other types of relative clauses, inclusion of the representative nominal is optional. Example (7) shows two sentences containing an object relative clause which differ only by the presence or absence of a representative nominal in the relative clause. m̀bʊ̀ŋ ɲí [Na̋ŋ ha̋ lì̵ŋ] (7) a. ɔ̀ ù pí∼pı̋ cl1.cow rel N. p1 (a)look.for cl1.det cl1 vfoc∼(b)die ◊‘The cow that Nang looked for died.’ (Abar) b. m̀bʊ̀ŋ ɲí [Na̋ŋ ha̋ lì̵ŋ wù ] ù13 ù cl1.cow rel N. p1 (a)look.for cl1 cl1.det cl1 pí∼pı vfoc∼(b)die ◊‘The cow that Nang looked for[it] died.’ (Abar) Example (8) contains a relative clause where the representative nominal may or may not appear as the object of the locative phrase á…mī ‘inside of…’. As (8) shows, locative phrases may be “stranded” in relative clauses. Object-less locative phrases, however, are grammatical in main clauses as well (see (12), below). (8) àkə̂ŋ kɛ̀ lɛ̋ á-ka̋ ŋ ɲɩ̋ [bwe̋ ʤù cl12.dem.prox cl12.det cop cl12-pan rel cl2 (a)put.ipfv í-lì̵m̄ ɛ̀ á (kī) mī] kɛ̀ cl4a-nkwi.bark cl4a.det prep cl12 loc.in cl12.det ◊‘This is the dish that nkwi bark is put inside of.’ (Abar) In the case of comitative and dative relatives, if the representative nominal is omitted from the relative clause, then so must the comitative or dative marker itself (9). (9) m̄-fə̋ tsì̵n∼tsɔ᷇ ŋɔ ú-ɕɛ̋ nə́ [Na̋ŋ fə̋ kə̀ m í-ɕɛ̋ 1sg-p1 vfoc∼(b)see cl3-knife rel N. p1 slaughter cl10-fowl jɛ̄ (bə̄ wù)] cl10.det (com cl3) ◊‘I did see the knife that Nang slaughtered the fowls (with it).’ (Munken) When the head nominal refers to a place where the event described by the relative clause took place, the relative clause may optionally contain a locative phrase translatable as ‘there’, as in (10).

13  The form of the class 1 determiner in Abar is sensitive to the presence of a preceding vowel.



8 Relative clause constructions in two Yemne-Kimbi languages 

 183

(10) -fə̋ tsì̵n∼tsɔ᷇ ŋɔ í-sɛ̋hɛ nə́ [Na̋ŋ fə̋ kə̀ m í-ɕɛ̋ 1sg-p1 vfoc∼(b)see cl5-place rel N. p1 (a)slaughter cl10-fowl jɛ̄ (á fē)] cl10.det (there) ◊‘I did see the place that Nang slaughtered the fowls (there).’ (Munken) The conditions on the appearance of a representative nominal within a relative clause exactly parallel the conditions on the presence of overt arguments in main clauses. Just as subject relative clauses are the only type of relative clause which must contain a representative nominal, subjects are the only argument which is not omissible from a main clause. The behavior of comitative and dative arguments is also parallel: in both relative and main clauses, a comitative or dative argument may be omitted, provided that the comitative/dative marker is itself omitted. Examples (11)–(13) show main clauses with an omitted object, locative complement, and dative, respectively. ɲà bí-tə̋m. (11) ù ŋ́kì̵ŋkì̵n jùɔ cl1 now (a)cut.vegetables.ipfv (a)stay.ipfv cl8-vegetables m=ù-ú gjɛ᷅lə then=cl1-fut (a)cook.irr ‘Now she’s cutting vegetables. Then she’ll cook [them].’ (Munken) (12) á-ɲı̋ bə́ tî wân bí-bûŋ á cl12-thing cl2 (c)hang.ipfv (b)keep.ipfv cl8-dress prep kī mə̀ . bə́ tî wânhə á-bûŋ cl12 loc.on cl2 (c)hang.ipfv (c)keep.ipfv cl12-dress ā-lɔ ́ u᷇ 14 mə̀ cl12-some prep-cl1.loc.obj loc.on ‘…a thing they hang dresses on. They’re hanging a dress on [it].’ (Munken) (13) jɛ̀ ù fɛ᷆ kə́ -nő kə́ -dú-lə ɲ̄-ɲɛ́ Comp cl1 (b)give.irr cl12-thing cl12-(b)carry.water-adjz cl6a-water ‘…that he should give [them] the thing for carrying water…’ (Biya)

14 This word is transcribed and glossed ú ‘loc’ in Lovegren (2013: 407), though the present gloss and tonal transcription is more consistent with the analysis and examples in Lovegren (2013: §9.1). In Munken, locative object pronouns always bear mid tone, and those beginning with a vowel or glide may coalesce with the preceding preposition. I.e., u᷇ mə̀ is a shorter variant of á wū mə̀ ‘prep cl1.loc.obj loc.on’. In an earlier version of this paper, ú mə̀ was treated as being a variant of á mə,̀ with the vowel of the preposition harmonizing to [u] following the preceding back vowel. More likely is that the speaker simply lost track of the noun class of the object of the locative phrase (á kī mə̀ ‘prep cl12.loc.obj loc.on’ was intended).

184 

 Jesse Lovegren and Rebecca Voll

2.5 Accessibility to relativization The formation of relative clauses in Mungbam is not restricted according to the grammatical relation of the representative nominal within the relative clause. A relative clause may be formed whose representative functions as a subject, an object, a comitative argument, a dative/benefactive argument, or the object of a locative argument ((14)–(18), respectively). A noun which functions as the possessor in a genitive NP within a relative clause may also be relativized (19). (14) ù-nò né [wə̌ te᷆ mù ɲa᷆ m cl1-person rel cl1.fut (b)come.irr then.cl1 (b)secretly.irr dzɛ́ bɛ́ hɛ̀ kō ì-jā] (c)call.irr (b)exit.irr (a)go.irr cl5-name ‘The person who comes and whispers [to me] the name…’ (Ngun) (15) bə̀ -mbɔ̀ ŋə̄ mə́ nə́ [Na̋ŋ tsa̋m] bē pí-a̋ cl2-cow 1sg.poss rel N. (b)beat cl2.det (b)die-prf ◊‘My cows which Nang beat have died.’ (Munken) (16) ú-ɕɛ̋ mɔ̀ [à ɕɛ̀ lə bə̄ wű ú-kpe̋ cl3-knife top 2sg (a)do.housework com cl3 cl3-house ú-gjɛ̀ lə mē] cl3-(a)cook.adjz loc ‘That knife that you work with in the kitchen’ (Munken) (17) ù-nɛ̀ nə̄ [mə̄ lē fɛ̋ ú-ɕɛ̋ wù=nə̄] wə̀ =ə́ cl1-person rel 1sg p2/3 (b)give cl3-knife cl1=dat cl1=then fə̋ gbè p1 (a)fall ◊‘The man whom I gave a knife to fell.’ (Munken) (18) ŋ̀-gbə̀ -nə̀ ná [Na̋ŋ ká gbə̀ wə̄ mī] cl1.nmlz-(a)fall-nmlz rel N. p1 (a)fall cl1 loc.at wū ká bí̵∼bāhā cl1.det p1 vfoc∼(c)bad ◊‘The way that Nang fell [i.e., the falling that Nang fell on] was bad.’ (Missong)  (19) ù-wə̀ nɔ̀ ù lɛ̋ cl1-dem.prox cl1 (b)cop ī-ɕɛ᷆ jì] cl9-fowl 3sg.poss ◊‘This

ù-nò cl1-person

ɲī rel

[ń=ʤā 1sg=(a)steal.irr

one is the man whose fowl I stole.’ (Abar)

A relative clause may also be formed from a noun heading an adjunct NP, as in (20) (a corresponding main clause is given in (21)).

8 Relative clause constructions in two Yemne-Kimbi languages 



 185

á-mjű nə́ [Na̋ŋ fə̋ ɲʊ̀nə bə̄ (20) -tsì̵tsa᷅ 1sg-vfoc∼know cl12-matter rel N. p1 (a)fight com í-wɔ̋ ŋ jɛ̄ ] kə̄ cl10-pig cl10.det cl12.det ◊‘I know the reason that Nang was fighting with the pigs.’ (Munken) (21) -bɔ̂ fə 1sg-(b)ask.ipfv

á-mjű cl12-matter

ú-kpɔ᷇ fə cl3-money

mə́ 1sg.poss

‘I’m asking because of my money.’ (Munken)

Relative clauses may also be formed where the referent of the head nominal has an obvious logical connection with the meaning expressed by the relative clause, but the grammatical relation of the head nominal within the relative clause itself is unclear, as in (22). (22) ì-jə̄nə ì-ɲı᷆ nə́ [ù-nɛ̀ ɲɛ̀ ù cl5-dem.prox cl5-honey rel cl1-person (a)stay.pfv cl1 tsə᷆ n] (b)be.drunk.irr ‘This is the honey that a person can get drunk [on].’ (Munken) There exist further cases of sentences with an identical structure to relative clauses, in texts and elicited, where the relativized noun does not play a semantic or syntactic role within the relative clause,15 but instead refers to the event itself or some logical consequence of it. These cannot be translated into idiomatic English with a relative clause. Some examples are given in (23)–(25). (23) ì-ʤī ì-nɩ̄ [bű gbà ɲà kə̄-tɔ̄ cl5-sound cl5-rel cl2 (a)cut.ipfv (a)stay.ipfv cl12-tree kə̄] nɔ̀ bâŋ ɲà mə̄ cl12.det (a)make.ipfv (b)block.ipfv (a)stay.ipfv 1sg ◊‘The sound of them cutting the tree disturbs me.’ (Biya) (24) á-mjű cl12-matter

nə́ rel

[bə́ -kjɩ᷇ŋ cl2-children

bē cl2.det

ʤà (a)steal

ì-ɕɛ᷆ cl9-fowl

mə᷆ ] kə̄ fwə̋mfə mə᷆ 1sg.poss cl12.det (b)worry 1sg ◊‘The fact that the children stole my fowl concerns me.’ (Munken)

15 This is taken to be a defining characteristic of relative clauses by most commentators, e.g., Downing (1978: 378), Lehmann (1986: 664), Andrews (2007: 206).

186 

 Jesse Lovegren and Rebecca Voll

(25) bə́ kwɛ̋ ì-tù ú-kpe̋ [bə́ kwɛ̋ ú-kpɔ̋ hɔ] cl2 (b)have cl5-species cl3-house cl2 (b)have cl3-money ‘They have the kind of house [that makes it seem like] they have money.’ (Munken) As Comrie (1998: §3.2) argues on the basis of similar facts for Japanese, there are languages for which the relative clause construction is not a formally distinct construction, but instead is subsumed under a larger noun-modifying clause construction. Comrie suggests that in such languages extraction (and therefore accessibility to relativization) is not a very useful concept for analyzing relative clauses. Mungbam may be such a language, given the presence of relative clauses which would have to be analyzed, in an extraction analysis, as being derived from an ungrammatical sentence in a main clause. Example (26) illustrates how the verb ban ‘climb’, when it takes a complement, must take a locative phrase and not simply a bare NP. Example (27) shows that the same verb may appear in a clause with no complement at all (provided that it is reduplicated when clause-final, (cf. § 2.6.2)). When the same verb appears in a relative clause (28), the relative clause may or may not contain a locative complement with a representative of the head nominal. If the version of (28) lacking the locative complement were to be treated as derived from a main clause *Na̋ŋ bàn ú-kpe̋ wə,̄ with the noun ú-kpe̋ ‘house’ extracted, it would leave us with the unhappy prospect of deriving a relative clause from a main clause which is in fact ungrammatical. When the relative clause is simply modeled as a noun modified by a clause, with no extraction relationship between the two, no such difficulty arises. (26) Na̋ŋ fə̋ bàn ú-kpe̋ wə̄ *(á N. p1 (a)climb cl3-house cl3.det prep ◊‘Nang climbed *(on top of) the house.’ (Munken)

fə̀ mə̄) loc.top

(27) Na̋ŋ fə̋ bì̵m∼ba᷆ n N. p1 vfoc∼(a)climb ◊‘Nang climbed.’ (Munken) ú-kpe̋ nə́ [Na̋ŋ fə̋ bàn (á wū (28) -fə̋ tsɔ̋ ŋ 1sg-p1 (b)see cl3-house rel N. p1 (a)climb prep cl3 fə̀ mə̄)] wə̄ loc.top cl3.det ◊‘I saw the house that Nang (climbed / climbed on top of).’ (Munken)



8 Relative clause constructions in two Yemne-Kimbi languages 

 187

2.6 Asymmetries between main and relative clause properties A well-attested phenomenon in African languages is for relative clauses and main clauses to have different inflectional or focus-marking possibilities. A typical scenario is for fewer inflectional categories to be available in relative clauses, or for in-focus marking to be restricted in relative clauses with respect to main clauses (Hyman & Watters, 1984). Furthermore, in some languages there are “relative tenses,” or differences between the marking of tense in relative clauses vs. in main clauses. In light of observations of this type, we prefer to make comparison of main and relative clauses even in areas where the two show no differences in behavior. Of course, we devote the larger part of the discussion to the part of the grammar where differences between the two clause types are observed, viz., in focus marking (§ 2.6.2). 2.6.1 Tense, aspect, mood, polarity In Mungbam, we find that the inflectional possibilities available to verbs are the same in relative clauses as they are in main clauses, with both the realis/ irrealis and the perfective/imperfective distinction available to verbs in a relative clause. While most examples in this paper are of verbs in their perfective realis forms, imperfective and irrealis forms may be found in examples (8), and (14), respectively. It should also be added that no inflectional categories have been attested which are found in relative clauses, but not in main clauses. Furthermore, no differences between main and relative clauses have been found as concerns the marking of tense and aspect, whether by verb stem changes or the presence of tense markers. This situation contrasts with that seen for Mundabli (§ 3.6). Examples (29)–(31) show relative clauses in each of the past tenses P1–P3. Although the tense markers have not elaborated on due to space limitations, verb tones can be verified against those given in Tab. 2. (29) ì-jə̄ n dà ì-bwɔ́ ì-nɩ̄ [Na̋ŋ fə̋ bwɔ̋  ] nə́ cl5-dem.prox neg cl5-(b)tired.inf cl5-other N. p1 (b)tired frust ◊‘This is not the fatigue that Nang was tired [i.e., that Nang experienced].’ (p1, Biya)

188 

 Jesse Lovegren and Rebecca Voll

(30) ǎ ù-nò ɲí [Mû kə̀ lè tʊ̂ 16 ds.cop cl1-person rel M. p2/3 (a)make (c)show ì-bwé Kúlɔ́ =né] cl9-goat K.=dat ◊‘This is the person by whom Mu made the goat be shown to Kulo.’ (p2, Abar) i i nə́ mɔ̄ wə̄ , ù nə́ [à (31) fɛ᷆ wù b=ú-tɔ̋ fə (b)give.irr cl1 com=cl3-sense top cl3.det cl1 rel 2sg lē fɛ᷆ dā] wə̄ p2/3 (b)give.irr before cl3.det ‘Give him that wisdom, the one which you have given before [to others].’ (p3, Munken) Negation is not restricted or differentially expressed in relative clauses (see (40) for an example). The possibility of non-declarative illocutionary force in relative clauses has not been investigated for Mungbam. 2.6.2 Focus marking As for focus-marking, all of the types of focus constructions which are grammatical in main clauses are also permitted in relative clauses. Formal focus-marking processes in Mungbam include focalization and defocalization of clausal arguments (realized by word order changes) and verum focus (realized by reduplication of the final verb in the verbal complex).17 Term focus The representative nominal may be focalized or defocalized within the relative clause. The examples (32) show relative clauses wherein the subject has been focused by displacement to IAV (immediately after verb) position. In (32a) the representative of the head nominal itself is in focus, while in (32b) the representative of the head nominal is not the term which is in focus.

16 The tense markers for p2 and p3 are identical in all Mungbam varieties except for in Biya. The tenses are distinguished by the fact that in p3 the verb must be in its irrealis form. 17 A more detailed treatment of focus in Mungbam is found in Lovegren (2013: §11). The relevant notion of focus assumed here is that of Hyman & Watters (1984).



8 Relative clause constructions in two Yemne-Kimbi languages 

 189

nə́ [à gbè wù] wə̄ ù pí∼pı̋ (32) a. ù-nɛ̀ cl1-person rel ds (a)fall cl1 cl1.det cl1 vfoc∼(b)die ◊‘The man that he fell died.’ (Munken) i i b. ù-nɛ̀ nə́ [à ɕɛ̂ Na̋ŋ wù] wə̄ ù pí∼pı̋ cl1-person rel ds (c)insult N. cl1 cl1.det cl1 vfoc∼(b)die ◊‘The man that Nang insulted him died.’ (Munken) i i The representative of the head nominal may also be defocalized within the relative clause. Objects in Mungbam are defocalized when they are dislocated away from IAV position. In example (33), the representative nominal is a defocalized object within the relative clause. It displays the areally prevalent SOV word order found in some negated clauses (Güldemann, 2007: §2.3). (33) ù-nɛ̄ nə́ [Na̋ŋ wù á ɕɛ̄ hə̄] cl1-person rel N. cl1 neg (a)insult.irr neg2 pí∼pı̋ vfoc∼(b)die ◊‘The man that Nang did not insult him died.’ (Munken) i i

wə̄ cl1.det

ù cl1

Verum focus The area in the focus-marking system where differences between relative clauses and main clauses have been found is in the expression of verum focus. To illustrate this difference, we will first have to give an overview of the relevant properties of verum focus marking in main clauses. Verum focus marking, which is realized by the reduplication of the final verb in a clause, is in some cases optional (under “pragmatic control,” in the terminology of Hyman & Watters (1984: 243)), but in other cases it is under grammatical control: either it is mandatory and its absence results in ungrammaticality; or it is forbidden and its presence results in ungrammaticality. The restrictions for main clauses may be summarized as follows: (34) a. If a verb is the final element in a clause, it must be reduplicated. b. In a negated clause, the verb must not be reduplicated. c. If a subject argument is dislocated to IAV, the verb must not be reduplicated. d. Otherwise, reduplication is under pragmatic control. Restriction (34a) applies when a verb has no object argument (i.e., is intransitive) (35a) or has a fronted, defocalized object argument (35b). Corresponding sentences with a non-reduplicated verb are ungrammatical (36).

190 

 Jesse Lovegren and Rebecca Voll

(35) a. Na̋ŋ gbù∼gbe᷆ N. vfoc∼(a)fall ◊‘Nang has fallen.’ (Biya) b. Na̋ŋ í-bwe̋ jī kì̵ŋ∼kə᷆m N. cl10-goat cl10.det vfoc∼(a)slaughter ◊‘Nang did slaughter the goats.’ (Biya) (36) a. *Na̋ŋ gbè b. *Na̋ŋ í-bwe̋ jī kə̀ m The ban on reduplication in negated clauses is illustrated in (37). (37) a. Na̋ŋ á bám hɔ̄ N. neg (c)ascend.irr neg ◊‘Nang did not accept.’ (Abar) b. *Na̋ŋ á búm∼bām hɔ̄ Restriction (34c) helps to draw a formal distinction between focused subjects in IAV and objects in IAV, which are unmarked for focus: a focused subject blocks reduplication (cf. the ungrammaticality of (39a)), but an object argument in IAV does not affect the possibility of verum focus marking (cf. the grammaticalness of (39b)). (38) a. à gbè Na̋ŋ ds (a)fall N. ◊‘Nang fell.’ (Biya) b. Na̋ŋ kə̀ m í-bwe̋ jī fə̋-fjɔ̂ ŋ N. (a)slaughter cl10-goat cl10.det cl16-loc.stream ◊‘Nang killed the goats at the stream.’ (Biya) (39) a. *à gbù∼gbe᷆ Na̋ŋ ◊Na̋ŋ kì̵ŋ∼kə᷆m í-bwe̋ jī fə̋ fjɔ̂ ŋ b. Whereas in main clauses, it can be argued that an in situ object is unspecified for focus, even though it is in IAV position, a different situation obtains in relative clauses. Here it can be argued that an object representative nominal is treated as being in focus if it is in IAV position, since it behaves analogously to a focused subject in IAV. Example (40) shows two relative clauses which differ only by the presence or absence of a representative nominal in the relative clause. When the representative nominal is absent, the verb may or may not be reduplicated (40a). However, when the representative nominal is present, the verb may not be reduplicated. (40) a. m̀bʊ̀ŋ ɲí [Na̋ŋ ha̋ {lì̵ŋ / lì̵n∼lı̵ ᷆ŋ}] ɔ̀ cl1.cow rel N. p1 ((a)look / vfoc∼(a)look) cl1.det pí∼pı̋ vfoc∼(b)die ◊‘The cow that Nang {looked for/ did look for}died.’ (Abar)

ù cl1



8 Relative clause constructions in two Yemne-Kimbi languages 

 191

b. m̀bʊ̀ŋ ɲí [Na̋ŋ ha̋ {lì̵ŋ / *lì̵n∼lı̵ ᷆ŋ} wù] ù18 cl1.cow rel N. p1 {(a)look / vfoc∼(a)look} cl1 cl1.det ù pí∼pı̋ cl1 vfoc∼(b)die ◊‘The cow that Nang (looked for / *did look for) it died.’ (Abar) i i Recall that in main clauses, in situ objects are associated with pragmatic control of verum focus, omitted objects force reduplication, and focalized subjects block reduplication. In relative clauses, on the other hand, an object representative nominal is not associated with the same properties as an object in main clauses: when the representative nominal is omitted, reduplication is under pragmatic control, and when the representative nominal is present and in situ, reduplication is blocked. From these facts it can be argued that the absence of an object representative nominal is associated with the relativized noun being focus neutral, while an overt representative object nominal is considered to be in focus. Further support for treating an object representative nominal as in focus comes from an interesting type of construction where the representative nominal is not a pronoun, but instead a modifier of the head nominal, as exemplified by (41). Here the representative nominal is not strictly coreferential with the head noun, but instead refers to a subset of the entities referred to by the head noun, thereby narrowing its reference. This kind of construction has a main clause counterpart wherein part of an object NP is fronted (and defocalized), and one of its modifiers remains in IAV. This type of construction, exemplified in (42), has the effect of putting in focus only the part of the NP which is in IAV. Example (42) could be used, for example, in a situation where the listener was unaware of the number of pigs which were beaten, or mistakenly believed that some number of pigs other than three were beaten. (41) í-jə̂n jī í-gɔ̋ ŋ mə̋ ī-nɩ́ cl10-10.dem cl10.det cl10-pig 1sg.poss cl10-rel ʨa̋m í-tɛ̄  ] jī (b)beat cl10-three cl10.det ◊‘These are my pigs that Nang beat three [of them].’ (Biya) (42) Na̋ŋ í-gɔ̋ ŋ mə̋ à ʨa̋m N. cl10-pig 1sg.poss ds (b)beat ◊‘Nang beat three of my pigs.’ (Biya)

[Na̋ŋ N.

fə̋ p1

í-tɛ̄ cl10-three

The representative nominal in (41) can be recognized as being in focus by considering that the new information is not that the number of beaten pigs is three, but rather the identification of the head nominal ‘my pigs’ with the three beaten pigs.

18 See footnote associated with (7b).

192 

 Jesse Lovegren and Rebecca Voll

Relative clauses also do not show the same aversion to clause-final non-­ reduplicated verbs that main clauses do, admitting simple intransitive clauses with a non-reduplicated verb, as in (43). (43) ù-nò ù-nɩ̄ [ù gbè] wə̄ ù kpú∼kpe̋ cl1-person cl1-rel cl1 (a)fall cl1.det cl1 vfoc∼(b)die ◊‘The man who fell died.’ (Biya) In fact, relative clause-final reduplicated verbs are very rare in texts, and the two Abar consultants did not agree in accepting sentences like the reduplicated version of (40a) as grammatical. Biya consultants do not accept a reduplicated version of (43) either. The dispreference for verum focus marking in relative clauses extends to clefts, where consultants uniformly reject clefts with reduplicated verbs. The marginal status of verum focus marking in relative clauses (which complicates the analysis given above for object representative nominals) is likely explainable by appeal to pragmatic factors: relative clauses in most contexts contain assertions whose truth is presupposed, or readily accomodated.

3 Mundabli 3.1 Introduction This section deals with the structure of the relative clause in Mundabli. We first treat the position of the relative clause with respect to the head nominal and to other noun modifiers (§ 3.2). In § 3.3 we consider how relative clauses are marked. Section 3.4 deals with how the head nominal is represented within the relative clause, and the accessibility of nouns to relativization, depending on their grammatical role within the relative clause, is treated in § 3.5. Finally, we consider differences in how various inflectional categories are marked in relative and main clauses, including tense and aspect, focus marking, illocutionary force and negation (§ 3.6).

3.2 Basic order of constituents in the NP In order to frame the succeeding discussion on Mundabli relative clauses, it is important to take a look at the structure of the noun phrase and the position of the relative clause relative to the head nominal and to other noun modifiers.



8 Relative clause constructions in two Yemne-Kimbi languages 

 193

In the unmarked case, all modifiers within an NP occur to the right of the head noun.19 The head noun may be modified by possessive pronouns, demonstratives, adjectives, numerals, the definite determiner, all of which show concord with the noun class of the head noun and by relative clauses. Also the relativizer shows concord with the noun class of the head noun.20 See Good et al. (2011: 130) for an overview of the Mundabli noun class system. Like all noun modifiers, the relative clause follows the head nominal. In nearly all examples of relative clauses found in natural texts, the relative clause is the only noun modifier and is thus placed directly after the noun. If other modifiers are present, though, the relative clause occurs at the end of the noun phrase, following all other noun modifiers, including the determiner. The schema provided in (44) shows the unmarked order of noun modifiers. Given that no other modifier follows the relative clause, it is difficult to determine whether the relative clause is to be treated as embedded in, or adjoined to, the matrix NP. (44) N – Poss – Adj – Dem – Num – Det – RC (45) ŋwàtᵼ̀ bi̋ bī-fyᵼ̋ŋ b-ɛ́ n bi̋-tɔ᷇ cl7/8.book cl8.3sg.poss cl8-new cl8-dem.prox cl8-three b-ɔ́ nō̤ [wù fə̌ ta̋ ŋ b-ɔ́ Bàmɛ́ ndà] cl8-det subd cl1 p1 (b)buy cl8-rel B. ◊‘these her three new books which she bought in Bamenda’ It is worth noting that the semantically bleached nouns ní̵ŋ ‘thing, matter’ and dɛ̀ ‘place’ are frequently used as head nominal in cases where other languages might use a headless relative clause. Although head-less relative clauses are possible, they are uncommon.

19 Demonstratives can precede the noun they modify, which evokes a more emphatic reading and genitive phrases whose possessor is a first person pronoun exhibit head-final word order (and the use of the free pronoun rather than the possessive pronoun) when headed by the noun wān ‘child’. 20 Both proximal and distal demonstratives and thus also determiners and relativizers, which are identical in shape with the latter, are glossed as prefix-stem sequences. However, the separation of the segmental prefix and the stem by a hyphen is somewhat misleading. The prefix actually consists of a consonant plus a tone which is realized on the stem vowel. This tone is mid for Class 1 and 9 and high for all other noun classes.

194 

 Jesse Lovegren and Rebecca Voll

3.3 Relative clause-marking Having shown how the relative clause relates to its environment, this section discusses relative clause marking, i.e., the strategies used to identify a relative clause as such. Relative clauses in Mundabli are marked at least by a concordant relativizing enclitic attached to the rightmost verb in the verbal complex, here called the “postverbal relativizer,” which is identical in shape with the demonstrative. In addition, relative clauses are optionally introduced by the non-concordant subordinating conjunction nō̤ , which also introduces certain kinds of adverbial clauses, and which we call the “clause-initial subordinator.” In this section we discuss first the postverbal relativizer (§ 3.3.1) and then the clause-initial subordinator (§ 3.3.2). 3.3.1 Postverbal relativizer The postverbal relativizer, exemplified in (46), is identical in shape with the definite determiner and the distal demonstrative. It agrees with the head nominal in noun class and must immediately follow the verb complex of the relative clause, irrespective of the definiteness of the matrix NP or of the syntactic-semantic role of the head noun within the relative clause. (46) wù dzé āyī, n=dᵼ̋ yə́ tʃɩ́n sɛ́ , cl1 (b)say interj 1sg.top=fut1 (c)go.up there cl3/7a.attic n=gān də̄ bān nᵼ́ŋ [kī lǣ 1sg.top= (a)go (a)see clearly cl7.thing cl7 (a)make.ipfv ɲɩ̄m tō̤ k-ɔ́ gū w-ɔ́ ] (c)extinguish.ipfv (b)move.away.ipfv cl7-rel cl3/7a.fire cl3-det ‘She said: Ayi! I will go up to the attic and find out what is putting out the fire.’ The postverbal relativizer is not to be confused with a resumptive pronoun. Firstly, as Tab. 4 shows, the two clearly differ in shape.21 Secondly, although the representative pronoun is often absent, there are numerous cases (e.g., (47)) of relative clauses containing both a postverbal relativizer and a representative nominal in the form of a pronoun.

21 Object pronouns of noun classes other than Class 1, 2 and 9 differ from subject pronouns in their tonal pattern. Object pronouns of these classes carry a super high tone. Apart from this tonal difference, subject and object pronouns are identical.

8 Relative clause constructions in two Yemne-Kimbi languages 



 195

Tab. 4: Subject pronouns and determiners Class 1 2 3 4 5 7 8 9 10 19 18b 6a 14

sbj pro

det

wù bɔ̋ wū yī wū kī bī yì yī fī mū mū bī

wɔ̄ bɔ́ wɔ́ yɔ́ wɔ́ kɔ́ bɔ́ yɔ̄ yɔ́ fɔ́ mɔ́ mɔ́ bɔ́

(47) first nᵼ́ŋ nō̤ [n = káā first cl7.thing subd 1sg.top=fut2 yɛ̄ ... comp … ‘The first thing I will do, is: […]’)

lə́ (a)do

k-ɔ́ cl7-rel

ki̋ ] cl7

dᵼ̌ (b)be

Although it is cognate with the definite determiner and the distal demonstrative and is probably derived from one of these historically, the postverbal relativizer has lost its status as a modifier of the head nominal. This is supported by its position in the middle rather than at the end of the relative clause (see (46)–(47)) and by the fact that the postverbal relativizer is always present, ­irrespective of the definiteness of the matrix NP or of the ability of the head nominal itself to be modified by a determiner. This latter point is made clear by examples such as (48), which contains a postverbal relativizer even though the head nominal is a 2sg pronoun, which cannot be modified by a demonstrative or a determiner22. (48) wān w-ɛ̄ n, dᵼ̌ cl1.child cl1-dem.prox (b)be w-ɔ̄ ná mᵼ̄ wān cl1-rel as 1sg.foc cl1.child

wà 2sg.foc w-ɔ̄ cl1-det

nō̤ [à lə̄ subd 2sg.top (a)make lɛ̄ fán (a)get.lost.ipfv here

22 Relative clauses modifying pronouns as in (48) are possible, though not common. When the head nominal is a first or second person pronoun, the relative marker always shows Class 1 agreement.

196 

 Jesse Lovegren and Rebecca Voll

gbɔ̄ cl3.house

kúŋ] behind

‘This child, you are the one who made my child23 get lost behind this house.’ When a relative clause modifies a non-third person pronoun, as in (48), the postverbal relativizer shows Class 1 agreement. 3.3.2 Clause-initial relativizer Relative clauses can be introduced by the subordinating conjunction nō̤ 24 (see (49)), which also introduces certain adverbial clauses. (49) dʒǔ nō̤ [bə̄ kə̀ lə̌ kpᵼ̄ y-ɔ̄ (yì) tō cl9.goat subd impers p3 (a)make (b)die cl9-rel cl9 cl7/8.day b-ɔ́ ŋgɔ᷆ ] kə̀ bān áná būbūbūbū cl8-det upon p3 (b)be.white like.that ideo.white ◊ ‘The goat which was killed on that day was completely white.’ Every relative clause can be introduced by this subordinator, but its presence is never obligatory. The same subordinator also obligatorily introduces certain adverbial clauses, such as reason clauses and specific kinds of time and manner clauses. In order to better understand Mundabli relative clauses, it is useful to take a brief look at these adverbial clauses which are introduced by the same subordinator. Adverbial clauses introduced by nō̤ contain a particle ná which follows the verb, just like the postverbal relativizer in a relative clause (see 50). (50) then from then, mᵼ̄ m=fə̋ kì-yùŋnɩ̀ bɔ̌ then from then 1sg.foc 1sg.top=(b)give cl7-thanks also gbàm lā nō̤ wù fə̋ ná kpɒ̋ ŋgɔ́ cl7/8.god dat subd cl1 (b)give as cl3/7a.money upon w-ɔ́ ndá lā cl3-det 1sg.dat dat ‘Then, from then, I [would] give thanks to God, as he has given me the money.’

23 The phrase mᵼ̄ wān ‘my child’ is a fixed lexicalized expression. While possessive phrases are usually head-initial, consisting of a head noun followed by a possessive pronoun which agrees with the noun class of the head nominal, in this fixed expression, the noun ‘child’ is simply juxtaposed to the focus form of the 1sg pronoun. 24 The subordinator has a phonetic variant nə̄ which often occurs in fast speech. The two variants occur in free alternation.



8 Relative clause constructions in two Yemne-Kimbi languages 

 197

Unlike relative clauses, these adverbial clauses usually stand at the end of the sentence, following whatever occurs last in the main clause, which is often a verb but may also be a dative phrase, as in (50). Nevertheless, they can also follow the noun, like relative clauses (see (64)). While the similar marking of relative clauses on the one hand and of the described adverbial clauses on the other makes a historical connection between the two very likely, it is unclear whether one function of the particle is historically derived from the other.

3.3.3 Grammaticalization source of the relativizing markers Considering the origin of the postverbal relativizer, it is rather obvious that it must have grammaticalized from the determiner or the distal demonstrative (recall that the three are identical in shape). The grammaticalization of a relative marker from a distal demonstrative, possibly via a determiner, which we propose to have taken place in Mundabli, is likely, given both the language contact situation and universal tendencies of language change. Demonstratives are a common grammaticalization source for relativizers in African languages (Heine, 2011: 706), and the “most frequent” source crosslinguistically (Heine & Kuteva, 2002: 115). While demonstratives seem to be a common grammaticalization source for relativizers, among the Yemne-Kimbi languages and in the wider area, the only language we are aware of that has a postverbal relativizer (shown in (51)) comparable to the one found in Mundabli is Naki (Beboid). The fact that a variety of Naki, namely Mashi, is spoken in a village of the same name which is directly adjacent to Mundabli, suggests that language contact may have played a role and that, if the postverbal relativizer is a recent innovation in Mundabli, it may have adopted this particular relative-clause marking strategy from Mashi. It is unlikely that it was the other way round. First, “oral histories regarding … the Mashi place their origins outside of Lower Fungom” (Di Carlo & Good, 2014: 16), which makes it more likely that Naki was the original source, introducing a new structure to Lower Fungom, and second, Naki is also spoken in several villages more distant from Mundabli, in and outside of Lower Fungom (cf. Di Carlo & Good (2014: 5)). The innovation would have had to be adopted in Mashi and then have spread to the other villages. (51) ŋku᷄ŋ wì [l’ àjī wə̀ cl1.chief cl1.ass 3s (a)eat.ipfv.p1.dsf cl1.det ◊ ‘the chief that was eating fufu’ (Naki) (Jeff Good p.c.)

ūnə̄] cl14.fufu

198 

 Jesse Lovegren and Rebecca Voll

The subordinator which may introduce a relative clause, on the other hand, does not have an equivalent in Naki. The similarity in both position and phonological shape between the Mundabli subordinator nō̤ and the clause-initial relativizers in the Mungbam dialects seems to indicate that they are either cognate or related through borrowing. It should be noted that multiple marking of relative clauses seems to be a regional tendency. It is attested in numerous Grassfields languages spoken to the South of Lower Fungom, namely Bafut (Ngemba) (Tamanji, 2009), Limbum (Nkambe) (Fransen, 1995), Shupamem (a.k.a. Bamun) (Nun) (Nchare, 2012: 188–9, 454) and non-Grassfields languages, such as Ngbaka Ma’bo (Ubangi) (Thomas, 1963: 270). Some of these cases even involve an invariable marker which is similar in shape with the Mundabli subordinator. Nevertheless, in none of these languages relativization involves a postverbal marker, like in Mundabli.

3.4 The representative of the head nominal According to Keenan (1985: 147), the encoding of the role of the head noun in the embedded sentence is, cross-linguistically, one of the most significant parameters from the viewpoint of typological variation. In Mundabli, the head nominal can always be represented within the relative clause. This is generally done by use of a pronoun which takes the same position in the relative clause as in a main clause. The presence of a representative nominal is only obligatory when the representative nominal functions as the subject of the relative clause (see (52)). In all other types of relative clauses, the use of a representative nominal is ­optional. (52) mɔ̀ [wù kə̀ dzé w-ɔ̄ dʒu᷇ gbàm cl1.man cl1 p3 (b)say.ipfv cl1-rel cl3a.word cl7/8.god tō k-ɔ́ ŋgɔ᷆ ] kə̀ dᵼ̄ Pǎ Pᵼ̋tà Kìá cl7/8.day cl7-det upon p3 (b)be Pa P. K. ‘The person who was preaching on that day was Pa Peter Kia.’ The object relative clause in (53) may or may not contain a representative nominal. (53) dʒǔ nō̤ [bə̄ kə̀ lə̌ kpᵼ̄ y-ɔ̄ (yì) tō cl9.goat subd impers p1 (a)make (b)die cl9-rel cl9 cl7/8.day b-ɔ́ ŋgɔ᷆ ] kə̀ bān áná būbūbūbū cl8-det upon p3 (b)be.white like.that ideo.white ◊‘The goat which was killed on that day was completely white.’



8 Relative clause constructions in two Yemne-Kimbi languages 

 199

In a locative relative clause, the object of the locative phrase may be omitted, as shown in (54) so that the locative phrase gets stranded. The use of the locative preposition ɩ ̋(omitted in (54)) is, as always, optional. (54) kpē w-ɔ́ nō̤ [bɔ̋ fə̋ dʒī kūo ʃī w-ɔ́ cl3.pot cl3-det subd cl2 p1 (a)put (c)enter (a)go.down cl3-rel dɔ̄ w-ɔ́ (wū) mᵼ́], wū fə̋ ga̋ a̋ mò̤∼mò̤ cl3a.beans cl3a-det cl3 in cl3 p1 (b)be.big advz one∼advz ◊‘The pot into which they put the beans was very big.’ Unlike the type of locative relative clause exemplified by (54), locatives which do not describe the spatial relation to an object or location, but rather location at a certain place, henceforth referred to as “absolute locative relative clauses”, never contain a representative nominal. Absolute locative relative clauses always take the semantically bleached noun dɛ̀ ‘place’ as head nominal and the postverbal relativizer agrees with a locative proform glossed ‘prox’, likely a remnant of Proto-Bantu locative class 16, which encodes proximity (55). (55) ká à wɔ̄ ŋ mɛ̄ , kə́ m̀=mū, mᵼ̄ when 2sg.top (a)squish (a)finish when 1sg.top=(b)drink 1sg.foc n=tsɔ̀ dɛ̀ nō̤ [wān w-ā kə̀ 1sg.top=(a)show cl9/10.place subd cl1.child cl1-2sg.poss p3 fᵼ̄ f-ɔ́ ] (b)pass prox-rel ‘When you have finished squishing [the small berries], when I will have drunk [the juice], I will show [you] where your child has gone.’ A dative phrase is optionally introduced by the locative preposition  (see (56a)), and requires the dative postposition lā. When the representative nominal in a relative clause is the argument of a dative phrase, it can be omitted so that the dative phrase gets stranded (56b). In this case, the locative marker  usually gets omitted so that the dative postposition lā is left alone (56b). (56) a. mbɛ̄ nō̤ [wù kə̀ dʒìɛ́ b-ɔ́ (ɩ̋  ) bɔ́ cl2.people subd cl1 p3 (b)cook.ipfv cl2-rel (loc) cl2.loc lā], bɔ̋ kə̀ fa᷆ n dat cl2 p3 (a)be.rich ◊‘The people for whom she used to cook were rich.’ b. mbɛ̄ nō̤ [wù kə̀ dʒìɛ́ b-ɔ́ lā], bɔ̋ kə̀ cl2.people subd cl1 p3 (b)cook.ipfv cl2-rel dat cl2 p3 fa᷆ n (a)be.rich ◊‘The people for whom she used to cook were rich.’

200 

 Jesse Lovegren and Rebecca Voll

In a comitative relative clause, the representative nominal can be omitted only if the comitative marker ā is also absent, as shown in (57), see (61) for an example of a comitative relative clause in which the comitative phrase is not omitted. (57) sɨ̀ŋ nō̤ [wù fə̌ ta̋ n y-ɔ̄ sɔ̀ y-ɔ̄ ], yì fə̌ cl9.knife subd cl1 p1 (b)cut cl9-rel cl9.meat cl9-det cl9 p1 dṳ᷆ (a)be.blunt ◊‘The knife that she cut the meat [with] was blunt.’

Comparison with main clauses The situation regarding the representation of the head nominal within the relative clause in Mundabli is almost the same as in Mungbam. Just as only subject relative clauses must obligatorily contain a resumptive pronoun, the subject is also the only obligatory argument in a main clause. Additionally, stranding of a locative phrase is possible in main clauses as well as in relative clauses. Alternatively, in both, the locative phrase can be omitted entirely. Concerning absolute locatives, locative phrases can always be omitted in main clauses. There are no locative pronouns, which may explain the complete absence of representative nominals in absolute locative relative clauses. In both main and relative clauses, the dative phrase can either be completely omitted (as in Mungbam) or get stranded, which is not possible in Mungbam. Just like in Mungbam, in both relative and main clauses a comitative argument may only be omitted if also the comitative marker is omitted. Thus, the differences in Mundabli and Mungbam relative clauses are reflected in the differences they show in main clauses. In both languages, the conditions for omission of arguments are exactly the same in main and relative clause.

3.5 Accessibility to relativization Another typologically relevant factor in relative clause structure concerns the permissible grammatical functions of the head nominal within the relative clause (Andrews, 2007: 207). In Mundabli, there is no restriction on the grammatical function of the head nominal within the relative clause. The representative nominal within a relative clause may be the subject, object, dative argument or the comitative argument of the relative clause, or it may be the argument of a locative phrase (see (58), (59), (60), (61) and (62), respectively). The representative nominal can also be the possessor in a genitive phrase, as in (63).



8 Relative clause constructions in two Yemne-Kimbi languages 

 201

(58) mɔ̀ [wù kə̀ dzé w-ɔ̄ dʒu᷇ gbàm cl1.man cl1 p3 (b)say.ipfv cl1-rel cl3a.word cl7/8.god tō k-ɔ́ ŋgɔ᷆ ] kə̀ dᵼ̄ Pǎ Pi̋tà Kìá cl7/8.day cl7-dist.dem upon p3 (b)be Pa P. K. ‘The person who was preaching on that day was Pa Peter Kia.’ (59) dʒǔ nō̤ [bə̄ kə̀ lə̌ kpᵼ̄ y-ɔ̄ (yì) tō cl9.goat subd impers p1 (a)make (b)die cl9-rel cl9 cl7/8.day b-ɔ́ ŋgɔ᷆ ] kə̀ bān áná būbūbūbū cl8-det upon p3 (b)be.white like.that ideo.white ◊‘The goat which was killed on that day was completely white.’ (60) wàn w-ɔ̄ nō̤ [m=fə̋ fə̋ w-ɔ̄ kpɒ̋ cl1.child cl1-det subd 1sg.top=p1 (b)give cl1-rel cl3/7a.money ɩ̋ wú lā] kɛ́ tʃǔ a̋ kè∼kè loc cl1.loc dat (c)return (b)come advz cl9.hand∼advz ◊‘The child to whom I gave the money came back with empty hands.’ (61) ŋkɔ̀ m nō̤ [ntí fə̋ lɔ̄ w-ɔ̄ ā wù]᷆ cl1/2.hoe subd N. p1 (a)go.bush cl1-rel com cl1 kwa f  ᵼ̌ (a)break (b)pass ◊‘The hoe with which Ntie went to the farm broke.’ (62) kpē w-ɔ́ nō̤ [bɔ̋ fə̋ dʒī kūo ʃ ī w-ɔ́ cl3.pot cl3-det subd cl2 p1 (a)put (c)enter (a)go.down cl3-rel dɔ̄ w-ɔ́ mᵼ́], wū fə̋ ga̋ a̋ mò̤ ∼mò̤ cl3a.beans cl3a-det in cl3 p1 (b)be.big advz one∼advz ◊‘The pot into which they put the beans was very big.’ (63) wān nō̤ [mán mű dᵼ̋ w-ɔ̄ cl1.child subd cl11/12.name cl12.3sg.poss (b)be cl1-rel ngàʃǎ] fə̋ lɔ̀ ɲu᷇ N. p1 (a)go.bush cl3/7a.farm ◊‘The child whose name is Ngasha went to the farm.’ It is important to mention that in Mundabli it is impossible to form a relative clause where the head noun plays no obvious grammatical role within the relative clause. While this is possible in Mungbam (see 2.5), in Mundabli such a situation requires the use of the subordinate construction which was introduced in § 3.3.3. Example (64) is the translation equivalent of the Mungbam relative clause given in (23).

202 

 Jesse Lovegren and Rebecca Voll

(64) dzɛ̋ nō̤ bə̄ gbā ná kpān w-ɔ́ cl3/7a.sound subd impers (a)cut.ipfv as cl3.tree cl3-det fyá mᵼ̄ ā fa̋ m (b)give.ipfv 1sg.foc com cl7/8.problem ◊‘The sound of them cutting the tree disturbs me.’

3.6 Asymmetries between main and relative clause properties As pointed out above (§ 2.6), relative clauses and main clauses in African languages commonly differ regarding their inflectional or focus marking possibilities with typically fewer possibilities available in the relative clause as compared to the main clause. In Mundabli, inflectional possibilities are nearly the same in main and relative clause, but there are slight differences. All temporal and aspectual distinctions exist in both main and relative clauses. A relative clause can be interrogative, in which case basically the same construction is used as in a main clause. Also focus marking is expressed in the same way in main and relative clause. Only the way negation is marked in relative clauses differs considerably from the way it is marked in main clauses. All these aspects are treated in turn. Since the data are not in every case sufficient to support a clear statement regarding tonal effects, some further differences between main and relative clauses not presented here may eventually be discovered. Mundabli has no restrictions on the occurrence of temporal or aspectual distinctions in relative clauses. Tense is marked by preverbal particles and specific tone patterns on the verb. All tenses can occur in both main and relative clauses and also the aspectual perfective/imperfective distinction, marked by ablaut and specific verb tone patterns, exists in both main and relative clauses. We are not aware of a difference between tonal marking in main and in relative clauses. Relative clauses can also be interrogative clauses (see (65) and (66)). Apart from the relativizer, which follows the verb in the relative clause, the same construction is used for interrogative main and relative clauses. (65) mɔ̀ nō̤ [wù yᵼ́ŋ w-ɔ̄ gbɔ̄ ndɛ́ ] kpᵼ̌ cl1.man subd cl1 (c)build cl1-rel cl3.house who (b)die ◊‘The man who built whose house died?’ dᵼ̋ sᵼ̀ ŋ nō̤ [bə̄ fə̋ ta̋ n (66) yɛ́ n cl9-this (b)be cl9.knife subd impers p1 (b)cut mān  ā    yì] what  com  cl9 ◊‘This is the knife that they cut what with?’

y-ɔ̄ cl9-rel



8 Relative clause constructions in two Yemne-Kimbi languages 

 203

The word order is the same in a relative clause as in a main clause, with focusinduced word order changes possible. Focus marking is basically the same in main and relative clauses. In order to be focused, the subject of a main clause can occur in IAV-position. The same happens in relative clauses, as can be seen in (67) and (68) which are opposed to a relative clause with unmarked word order in (69). (67) bɩ̄ tsè mɔ̀ nō̤ [fə̋ gīa 1pl (a)search cl1.man subd p1 (a)steal dʒǔ   y-ɔ̄ ] cl9.goat  cl9-det ◊‘We look for the man that [he] stole the goat.’

w-ɔ̄ wù cl1-rel cl1

(68) n̋ =kɔ̀ ŋ sə̄ k-ɔ́ (nō̤ ) [ta̋ ŋ k-ɔ́ 1sg.top=(a)love cl7/8.clothes cl7-det subd (b)buy cl7-rel ɲùŋfù  (ki̋)] N.     cl7 ◊‘I like the piece of clothes that Nyungfu bought.’ (69) n̋ =kɔ̀ ŋ sə̄ k-ɔ́ (nō̤ ) [ɲùŋfù ta̋ ŋ 1sg.top=(a)love cl7/8.clothes cl7-det subd N. (b)buy k-ɔ́     (ki̋)] cl7-rel  cl7 ◊‘I like the piece of clothes that Nyungfu bought.’ In (67), it is the representative nominal which is in focus, in (68), a nominal other than the representative nominal is in focus. Note that the postverbal relativizer precedes the focused subject. Verum focus can also be expressed in a relative clause, as in (70), although this is not very common. Just like in a main clause, it is marked by the particle tə́ which precedes the verbal complex. (70) fán dᵼ̋ mɔ̀ nō̤ [wù tə́ kə̀ tʃū here (b)be cl1.man subd cl1 vfoc p3 (b)come ◊‘Here is the man who did come.’

w-ɔ̄ ] cl1-rel

While main and relative clauses are identical regarding tense, aspect, etc., they differ with regard to negation. Two different negation strategies are attested in main clauses, the choice between which is semantically determined. In relative clauses on the other hand, no such distinction is made. In order to show how negation in relative clauses differs from negation in main clauses, it is necessary to first explain the two negation strategies which are attested in main clauses. In the regular case, a main clause is negated by adding the discontinuous negative marker ā ... wɔ̄ around the verb complex, as in (71). In cases where the negation

204 

 Jesse Lovegren and Rebecca Voll

can be translated as ‘not yet’, the second negation strategy is applied, see (72). It requires the use of a copula auxiliary and will be referred to as ‘auxiliary negation’. wɔ̄ ɲɔ̄ ni̋ (71) bī ā wṳ́ cl8 neg (b)hear.ipfv neg (b)talk cl1.mother.3sg.poss ‘They are not listening to their mother’s advice25.’ ɲɔ̋ m wɔ̄ ɲa᷆ m (72) n=da̋ 1sg.top=(b)be.neg (b)turn.fufu neg cl14.fufu ◊‘I have not turned fufu yet.’ In (72), the copula d̵ and the negative morpheme ā are contracted to [dȁ]. In relative clauses, only auxiliary negation is attested, see (73). Mundabli differs in this respect from Mungbam, which employs auxiliary negation in main clauses much in the way that it is used in Mundabli,26 but allows the regular negation construction (see (33)) in relative clauses. (73) bɔ̋ fə̋ mù bɔ̀ k-ɔ́ nō̤ [bɩ̄ fə̋ d̵ cl2 p1 (a)take cl7/8.bag cl7-det subd 1pl p1 (b)be k-ɔ́ a̋ dǝ̀ wɔ̄ ], bɔ̋ mí̵ dʒì kǐ ɩ̋ cl7-rel neg (a)see neg cl2 consec (a)put cl7 loc kìa mᵼ̄ . cl9.basket in ◊‘They took the bag we did not see and put it into a basket.’

4 Discussion We conclude this chapter by discussing the key points of similarity and difference between relative clause constructions in Mungbam and Mundabli, and draw attention to some relevant typological issues. A still-unresolved issue concerning the Yemne-Kimbi languages is their exact genetic affiliation. While the languages were on the basis of early survey work grouped together under the low-level genetic unit “Western Beboid” (Hombert, 1980), subsequent work (Good & Lovegren,

25  The class 8 pronoun can be used in an abusive manner to refer to human beings, equating them with inanimate things. 26 A negation construction is found in Mungbam which uses dà or dā (depending on the dialect) as the negator, and which has similar semantic properties to Mundabli auxiliary negation. This morpheme may also function as a negative copula, as its apparent cognate in Mundabli does. However, a positive copula verb with the form dɨ is found only in Missong.



8 Relative clause constructions in two Yemne-Kimbi languages 

 205

2009; Good et al., 2011) has suggested that a more cautious interpretation is in order, given various lexical and morphological dissimilarities between the languages, and a lack of clear shared innovations to motivate the grouping. In the present study we have covered relative clauses in both languages in a parallel fashion, hoping that in doing so we might more easily uncover bits of evidence which will shed further light on the historical relationship between the two languages. Though in this study we have uncovered several interesting parallels between the two languages, and have also noted some curious points of difference, we do not find ourselves in a position to say anything new about the level of genetic relationship between the two languages. Both languages make use of a non-concordant relativizing particle, with similar phonetic shape, which introduces the relative clause. This particle is optional in both languages, though it is omitted more frequently in Mundabli. Mundabli additionally has a concordant relativizer which is cliticized to the end of the verbal complex. Such a feature is not found in Mungbam. Relative clauses in Mundabli can then be doubly marked, once with a pre-RC relativizer nō̤ , and again with the concordant verbal enclitic. Although we mentioned examples of other languages in the area which double-mark their relative clauses, it should be recalled that more than one relative marker is still rare cross-linguistically (Kuteva, 2009: 13, fn. 2). Concerning the concordant clitic in Mundabli, it can be noted that, although it is not uncommon for relative clauses to contain a relativizing marker on the verb, this type of marking usually occurs in verb-final languages (Andrews, 2007: 231). Exceptions to this generalization are, however, attested in several Narrow Bantu languages. In Shingazidja (G44a, Comoros), with basic SVO word order, for example, relative clauses contain a different final vowel from corresponding main clauses, which effectively indicates that the host verb is part of a relative clause: (74) a. e=mw-idzí ha-ib-í e=n-dovu y-á hahe aug1=cl1-thief 1(pst)-steal-fv aug9=cl9-elephant cl9-of his “The thief stole his elephant.” b. e=mw-idz’ yá-ib-a n-dovu aug1=cl1-thief cl1(rel-pst)-steal-fv cl9-elephant ha-ʈáw-a cli(pst)-run.away-fv “The thief who stole an elephant ran away.” (Patin, 2010: 196–7) As we speculate in § 3.3.3, the double-marking of Mundabli relative clauses might be explainable as the result of a contact scenario, since a similar marking feature is witnessed in the neighboring language Naki. The ultimate source of this feature, however, remains unclear in light of the fact that in the other Beboid

206 

 Jesse Lovegren and Rebecca Voll

languages for which data are available, Noni (Hyman, 1981: 91– 94), and Nchane (Boutwell, 2010: 18–9), no such postverbal particle is found. Neither Mungbam nor Mundabli shows restrictions on relative clause formation which might be explained by the accessibility hierarchy. Both languages, however, display a situation contrary to the tendency suggested by Keenan & Comrie (1977: 92), and confirmed on the basis of a larger data set by Comrie & Kuteva (2011a, b), where pronoun retention27 is a required relativization strategy in subject relatives, but not in non-subject relatives. As we point out in sections 2.4 and 3.4, the prevalence of the gap strategy in non-subject relatives, and the pronoun retention strategy in subject relatives, is to a large extent explained by similar restrictions on main clauses: subject arguments may not be omitted from main clauses, but non-subject arguments may, when their reference is inferable from context. Finally, we note that there are two Mundabli constructions which correspond to the Mungbam noun-modifying clause construction. One Mundabli construction corresponds more or less to relative clause constructions in other languages, with a semantic requirement concerning the relationship between the head noun and the relative clause. The second is a type of more general subordinating construction (shown in (64)). Relative clauses in Mungbam, on the other hand, are subsumed under a single noun-modifying clause construction (cf. Comrie (1998)) which has much looser requirements concerning the relationship between the head noun and the modifying clause. The broader significance of this distinction between the two languages is still to be explored more fully.

References Andrews, Avery D. 2007. Relative clauses. In Timothy Shopen (ed.), Language typology and syntactic description. Volume II: Complex constructions 2nd edn., 206–236. Cambridge: Cambridge University Press. Bickel, Balthasar, Bernard Comrie & Martin Haspelmath. 2008. Leipzig glossing rules: Conventions for interlinear morpheme-by-morpheme glosses. http://www.eva.mpg.de/ lingua/pdf/Glossing-Rules.pdf (accessed 26 March 2017). Boutwell, Richard L. 2010. A sketch grammar of the Nchane language. Yaoundé: SIL Cameroon. https://www.sil.org/resources/ archives/43200 (accessed 27 March 2017). Breton, Roland. 1993. Is there a Furu language group? An investigation on the CameroonNigeria border. Journal of West African Languages 23(2). 97–118.

27 That is, the use of a pronoun which is obligatory in relative clauses but only optionally present in main clauses (Comrie & Kuteva, 2011b).



8 Relative clause constructions in two Yemne-Kimbi languages 

 207

Chumbow, Beban. 1977. Relatives as determiners — a case from Ngemba. In Paul F. A. Kotey & Haig Der-Houssikian (eds.), Language and linguistic problems in Africa, 283–302. Columbia, South Carolina: Hornbeam Press, Inc. Comrie, Bernard. 1998. Rethinking the typology of relative clauses. Language design 1. 59–86. Comrie, Bernard & Tania Kuteva. 2011a. Relativization on obliques. In Matthew S. Dryer & Martin Haspelmath (eds.), The world atlas of language structures online, Munich: Max Planck Digital Library. http://wals.info/chapter/123. (accessed 27 March 2017). Comrie, Bernard & Tania Kuteva. 2011b. Relativization on subjects. In Matthew S. Dryer & Martin Haspelmath (eds.), The world atlas of language structures online, Munich: Max Planck Digital Library. http://wals.info/chapter/122. (accessed 27 March 2017). Di Carlo, Pierpaolo. 2011. Lower Fungom linguistic diversity and its historical development: proposals from a multidisciplinary perspective. Africana Linguistica XVII. 53–100. Di Carlo, Pierpaolo & Jeff Good. 2014. What are we trying to preserve? diversity, change, and ideology at the edge of the Cameroonian Grassfields. In Peter K. Austin & Julia Sallabank (eds.), Endangered languages: Beliefs and ideologies in language documentation and revitalisation, 229–262. Oxford University Press. Downing, Bruce T. 1978. Some universals of relative clause structure. In Joseph H Greenberg (ed.), Universals of human language, vol. 4, 357–418. Stanford: Stanford University Press. Frajzyngier, Zygmunt. 1996. Grammaticalization of the complex sentence: a case study in Chadic. Amsterdam: John Benjamins. Fransen, Margo Astrid Eleonora. 1995. A grammar of Limbum, a Grassfields Bantu language spoken in the North-West Province of Cameroon. Amsterdam: Vrije Universiteit te Amsterdam dissertation. Good, Jeff. 2012. How to become a “Kwa” noun. Morphology 22(2). 293–335. Good, Jeff & Jesse Lovegren. 2009. Reassessing Western Beboid. Paper Presented at the Third International Conference on Bantu Languages. Royal Museum for Central Africa, Tervuren, 26 March. http:// www.acsu.buffalo.edu/~jcgood/goodlovegren-Bantu3.pdf (accessed 27 March 2017). Good, Jeff, Jesse Lovegren, Patrick Mvee, Carine Nguanguep, Rebecca Voll & Pierpaolo Di Carlo. 2011. The languages of the Lower Fungom region of Cameroon: Grammatical overview. Africana Linguistica XVII. 101–164. Güldemann, Tom. 2007. Preverbal objects and information structure in Benue-Congo. In Enoch Oladé Aboh, Katharina Hartmann & Malte Zimmermann (eds.), Focus strategies in African Languages: The interaction of focus and grammar in Niger-Congo and Afro-Asiatic. Trends in Linguistics—Studies and Monographs 191, 83–111. Berlin: Mouton de Gruyter. Heine, Bernd. 2011. Grammaticalization in African languages. In Heiko Narrog & Bernd Heine (eds.), The Oxford handbook of grammaticalization, 696–707. Oxford: Oxford University Press. Heine, Bernd & Tania Kuteva. 2002. World lexicon of grammaticalization. Cambridge: Cambridge University Press. Hombert, Jean-Marie. 1980. Noun classes of the Beboid languages. In Larry M Hyman (ed.), Noun classes in the Grassfields Bantu borderland, 83–98. Los Angeles: University of Southern California Department of Linguistics. Hyman, Larry M. 1972. A phonological study of Fe’Fe’. Studies in African Linguistics 3(Supplement 4). Hyman, Larry M. 1981. Noni grammatical structure: With special reference to verb morphology. Los Angeles: University of Southern California: Southern California Occasional Papers in Linguisitcs no. 9.

208 

 Jesse Lovegren and Rebecca Voll

Hyman, Larry M & John R Watters. 1984. Auxiliary focus. Studies in African Linguistics 15(3). 233–273. Keenan, Edward L. 1985. Relative clauses. In Timothy Shopen (ed.), Language typology and syntactic description. Volume II: Complex constructions 1st edn., 138–164. Cambridge: Cambridge University Press. Keenan, Edward L & Bernard Comrie. 1977. Noun phrase accessibility and Universal Grammar. Linguistic Inquiry 8(1). 63–99. Keenan, Edward L & Bernard Comrie. 1979a. Data on the Noun Phrase Accessibility Hierarchy. Language 55(2). 333–351. Keenan, Edward L & Bernard Comrie. 1979b. Noun Phrase Accessibility revisited. Language 55(3). 649–664. Kießling, Roland. 2011. Verbal serialisation in Isu (West-Ring): A Grassfields language of Cameroon. Köln: Rüdiger Köppe Verlag. Koopman, Hilda & Dominique Sportiche. 1986. A note on long extraction in Vata and the ECP. Natural Language and Linguistic Theory 4(3). 357–374. Kuteva, Tania. 2009. Grammatical categories and linguistic theory: Elaborateness in grammar. In Peter K Austin, Oliver Bond, Monik Charette, David Nathan & Peter Sells (eds.), Proceedings of conference on language documentation and linguistic theory 2, 13–28. London: School of Oriental and African Studies. Lehmann, Christian. 1986. On the typology of relative clauses. Linguistics 24(4). 663–680. Leroy, Jacqueline. 2007. Le mankon: Langue bantoue des Grassfields (province Nord-Ouest du Cameroun). Paris: Peeters. Lewis, Paul M, Gary F Simons & Charles D Fennig. 2015. Ethnologue: Languages of the world, eighteenth edition. Dallas, Texas: SIL International. http://www.ethnologue.com/18 (accessed 27 March 2017). Lovegren, Jesse. 2013. Mungbam grammar. Buffalo: University at Buffalo dissertation. Maxey, James Allen. 1994. Relative clauses in Vute. The University of Texas at Arlington MA thesis. Nchare, Abdoulaye Laziz. 2012. The grammar of Shupamem: lingBuzz/001441: New York University dissertation. Nkemnji, Michael Akamin. 1995. Heavy pied-piping in Nweh. Los Angeles: University of California, Los Angeles dissertation. Ohori, Toshio. 2011. The grammaticalization of subordination. In Heiko Narrog & Bernd Heine (eds.), The Oxford handbook of grammaticalization, 636–645. Oxford: Oxford University Press. Patin, Cédric. 2010. The prosody of Shingazidja relatives. In Laura Downing, Annie Rialland, Jean-Marc Beltzung, Sophie Manus, Cédric Patin & Kristina Riedel (eds.), Zentrum Allgemeine Sprachwissenschaft papers in linguistics no. 53: Papers from the workshop on Bantu relative clauses, 187–209. http://www.zas.gwz-berlin.de/zaspil.html. (accessed 27 March 2017). Schadeberg, Thilo C. 2003. Derivation. In Derek Nurse & Gérard Philippson (eds.), The Bantu languages, 71–89. London: Routledge. Tamanji, Pius N. 2009. A descriptive grammar of Bafut. Köln: Rüdiger Köppe Verlag. Thomas, Jacqueline M. C. 1963. Le parler Ngbaka de Bokanga. The Hague: Mouton. Voll, Rebecca. 2012. Tonal variation in the tense system of Mundabli, Western Beboid (Bantoid, Cameroon). In Matthias Brenzinger & Anne Maria Fehn (eds.), Proceedings of the 6th World Congress of African Linguistics, Köln, 17–21 August 2009, 533–544. Köln: Rüdiger Köppe.

Rhonda Thwing

9 Relative clauses in Vute grammar and discourse Abstract: This paper examines the structure of relative clauses in the grammar and discourse of Vute, a mambiloid language spoken by 15,000–25,000 people in Cameroon in the Centre and Adamawa Regions with a few small villages in the East Region. As expected with head-initial constituent order, all relative clauses in Vute are postnominal. The head of a relative clause may be either a noun, pronoun, or nominalized verb. Since the Vute language lacks any overt relative pronoun or relativizer, the language uses three morphosyntactic means to indicate a relative clause: 1) relativized noun phrase (NPREL) marking, 2) d ­ ependent/ defocused verbal aspect marking, 3) genitive marker and some other NP-final elements. Within the NPREL marking there are two strategies, gapping and a particular pronoun that indicate a relative clause. Within verbal aspect marking, the defocused verb form indicates a relative clause. As for genitive marking and other elements, they occur at the final border of the relative clause. Furthermore, Vute limits relative clauses to two or three functions in discourse: as markers of discontinuity, in locally relevant adverbial constituents, and as markers of thematic prominence.

1 Introduction 1.1 Vute population, location and genetic classification of the Vute language Vute (ví̵tè) is a mambiloid language spoken by 15,000–25,000 people in Cameroon according to Maxey (1994: 1) in the Centre and Adamawa Regions with a few small villages in the East Region (Thwing, 1987: 1). It is classified by Ethnologue, (16th edition, 2009) as “Niger-Congo, Atlantic-Congo, Volta-Congo, BenueCongo, Bantoid, Northern, Mambiloid, Suga-Vute, Vute”. ALCAM classifies it as “bantoid, non-bantu, mambiloid, #720” (39). The Vute language represented in

Rhonda Thwing: SIL Cameroon, [email protected] DOI 10.1515/9783110469547-009

210 

 Rhonda Thwing

this paper, nyindí ví̵tèé ‘the speech of the Vute people’, is the standardized Vute language of the Centre Region of Cameroon, Mbam and Kim Division.1

1.2 Vute constituent order patterns and sentence structure Vute has predominantly head-initial (S V O) constituent order with S AUX O V order in imperfective aspect and in negative forms in the independent clause. The subject always precedes the verb and the beneficiary/addressee/indirect object (IO) always precedes the direct object/ patient even in S AUX (IO) O V order. In genitive NPs, the head noun precedes the genitive noun,2 with the enclitic3 genitive marker (GEN) following the genitive noun rather than occurring between the head and possessor. Pre-/postpositional phrases are formed with the preposition ɓe ‘with’ or with the postpositions no ‘in, inside’ or yá ‘at, outside, on top of’. Vute is a clause chaining language, with different events marked as chained together in one sentence or sometimes even one paragraph. Tense-aspect is marked only on the initial verb of the chain, which is typical of head initial languages. Temporal inter-relationships within the chain of clauses are marked as either NAR ‘narrative’: events in contingent succession (Longacre, 1996: 10), SIM ‘simultaneous event’, with the first event more prominent than its simultaneous event, or SEQ ‘sequential event’, with the SEQ event likely or habitually following the previous event(s).

1.3 The structure of Vute nouns A large part of the current vocabulary of Vute shows regular correspondences to Guthrie’s list of Proto-Bantu roots (Thwing, 1987: 73–94). The language shows remnants of a noun class system marked by both prefixes, which are derivational,

1 To my Vute language consultants through the years: Apolinaire Ndjomna and Martin Yaya, both now deceased, Alfred Oumarou and Valentin Yakoura who helped me with the Vute language as teachers and gave the most insightful answers to my questions, I am very thankful to you. I am also grateful for my Cameroon SIL colleagues, especially Stephen Anderson, Robert Hedinger, Mona Perrin, and John Watters for their consultant help. I also wish to thank the government of the United Republic of Cameroon who permitted my husband and I to carry out linguistic research and language development among speakers of the Vute language. 2 The only exception to the pattern Head N+Genitive N is found in locative phrases when the genitive noun precedes the head noun in Vute. 3 The genitive marker is an enclitic, not a suffix, as it attaches to the last word of the noun phrase.



9 Relative clauses in Vute grammar and discourse 

 211

and suffixes, which may be either inflectional or derivational (Thwing, 1987: 26–7) and which combine by fusion. Vute nouns may carry a plural suffix (PL), usually –b, or –y4, or PL may be marked by a change of the final consonant and/or the length of the final vowel, e.g., dúhé ‘house’, dúhí̵b ‘house.PL’; lə̀ r ‘bush’, lààì ‘bush.PL’; kùndų̀ ‘plantain’, kùndùùŋ ‘plantain.PL’. Vute nouns may also carry a locative suffix, which often contains a nasal consonant or nasalized vowel, e.g., lə̀ r ‘bush’, làín ‘bush.LOC’; kúr ‘village’, kwé ‘village.LOC’. Any two or three of these suffixes/enclitics, the plural and locative suffixes, and the GEN enclitic may be combined in the following order and combinations, LOC + GEN, PL + LOC, PL + GEN, PL +LOC + GEN. I will use lə̀ r ‘bush’ as an example: làné ‘bush.LOC.GEN’, làìbìn ‘bush.PL.LOC’; làìbì ‘bush.PL.GEN’, làìbìnè ‘bush.PL.LOC.GEN’ Some more common nouns also have a head noun form labeled N1, which differs from the citation form, e.g., nu ‘pot’ (citation form)’ nɔk ‘pot.N1’ (head noun form); cὶ̧˗ ‘hunger’, càŋ ‘hunger.N1’; dúhé ‘house’, dú ‘house.N1’; mwar ‘field’ mwáí ‘field.N1’ (Thwing, 1987: 55). The N1 form is used when nouns that have an N1 form are followed by a modifier of any type.

2 Different functions of relative clauses in Vute texts5 In addition to the usual functions of restricting the identity of a head noun or adding additional information about the head noun, in Vute the head noun plus relative clause (N + RC) construction has semantic functions that are fulfilled by adverbial clauses or modal auxiliaries in a European language6. A generic head noun plus relative clause is the most frequent way to communicate adverbial information in Vute, and may encode location in time and space, manner, condition, reason, and purpose relationships. In Vute texts, relative clauses are also used to add prominence to topics or participants which are thematic in a given section of discourse. Relative clauses that are functioning in this way appear to be restrictive relative clauses, but do

4 The -y plural is written with i in the Vute orthography. 5 Almost all the examples in this chapter are taken from Vute texts, either a book of folktales Gèìn Ví̵tè or a booklet about raising children, all cited with each example and listed in Appendix A. 6 There are clear syntactic distinctions between relative clauses and adverbial clauses in Vute: adverbial clauses do not end with the genitive marker, nor do they require defocused/dependent verb forms. See § 3.2–3.

212 

 Rhonda Thwing

not function to restrict the identity of their head noun, instead they add thematic prominence to the information contained in them.

3 The structure of relative clauses in Vute As expected with head-initial constituent order, all relative clauses in Vute are postnominal. The head of a relative clause may be either a noun, pronoun, or nominalized verb. There is no relative pronoun or relativizer that introduces a Vute relative clause, but there are three other morphosyntactic markers which signal a relative clause. I will follow the organization and terminology used by Thomas Payne in Chapter 11: Clause Combinations (1997: 325–336).

3.1 Marking the relativized noun phrase (NPREL) Within the NPREL marking, there are two strategies: (a) gapping the position of NPREL and (b) indicating the position of NPREL with a pronoun. 3.1.1 Gapping is used at the upper end of the Noun Phrase Accessibility Hierarchy (NPAH)7, that is, when the subject, recipient/beneficiary/addressee or theme/ patient of the relative clause is the NPREL. The fairly rigid and limited constituent order patterns in Vute seem to eliminate possible confusion. (1) Head Noun + RC with direct object gap: Kì̵cáàb Ndóŋ Mvèìn S 6.4 nacum mɨ mí̵ŋgòŋ moò yà moò tɨ sotɨna mé. (H)nà-cù-m mɨ mí̵ŋgòŋ8 moò yà moò tɨ imp.give-again-imp 1sg corn 1sg.gen mother 1sg.gen p1 (H)sòtì̵-nà-Ø mé (Ø)9 pfv.grill-io-pfv.d 1sg.io.gen do.gap “Give me back the corn my mother roasted for me.”

7 NPAH (Keenan, 147). Subject > Direct object > Indirect object > Object of pre- or postposition > Possessor 8 The head noun is underlined and the relative clause is in bold. 9 The recipient/beneficiary always precedes the theme/patient in Vute. Neither one is marked with a pre-/postposition.



9 Relative clauses in Vute grammar and discourse 

 213

(2a) Head Noun + RC with subject gap: Dùrù ɓe Lì S 3.2 cuní̵b yi kú ye kíì nɨ ní̵m fainndé sé. cuní̵b (Ø) yi (H)kú-Ø ye kí=ì nɨ ní̵m chicken.pl (s.gap) p2 pfv.die-pfv.d day ana=gen neg thing (H)fàìn-ndé sé. pfv.count-pfv.gen eq.neg [When the young women saw them,10] the chickens that died that day were uncountable.” (2b) contains two Head + RC constructions, both restrict the identity of the noun/ pronoun which is the subject of the relative clause, the second RC has the anaphoric pronoun kí̵ as its head, which refers back to ɗúùní̵m, the head noun of the first clause. (2b) Dùrù ɓe Lì S 2.6 Dùrù í̵ pììcùnà ŋgə́ , ŋgə́ í̵ tà, “Tá, ɗûní̵m wu ɗú mé ndóŋtí̵ kí̵ mɨ ɗú wé” dùru í̵ pìì-cù-nà(L) ŋgə́ , ŋgə́ í̵ tà(L), baboon nar answer-again-io. cons 3sg 3sg nar say.cons tá, ɗûní̵m wu (H)ɗú-Ø mé11 (Ø) father, love.nom 2SG pfv.love-pfv.d 1sg.io.gen do.gap (H)ndóŋ-tí̵ kí̵ mɨ (H)ɗú-Ø wé (Ø) pfv.pass-pfv ana.prn 1sg pfv.love-pfv.d 2sg.io.gen do.gap “Baboon replied, ‘Friend, the love you have for me is greater than that which I have for you.” (3) Head Noun + RC with recipient gapped, based on Gɔ̀ Tùkur yi Gɔklé Par. 4 mvom ŋgə́ yi pɨɨmhó̧ fè rè. mvom ŋgə́ yi (H)pì̵ì̵m-hó̧-Ø (Ø) fè rè tortoise 3sg p2 pfv.cheat-out-pfv.d (io.gap) meat prox “This tortoise from whom he swindled the meat”

10 I include the English translation of the first clause in the sentence to make the context clear. 11  This construction (nominalized verb as NPREL modified by a relative clause containing the same verb) is often found in Vute texts. It adds thematic prominence to the preceding event and its result by repetition of the verb as well as a relative clause.

214 

 Rhonda Thwing

3.1.2 Marking the position of NPREL with the human/personal referential pronoun nò/nùb ‘REF/REF.PL’12 or the inanimate anaphoric pronoun kí̵/ kí/ kí̵b ‘ANA.PRN/ANA. PL.PRN’ at the lower end of the NPAH (Noun Phrase Accessibility Hierarchy): instrument, accompaniment13, and the possessor. In (4) the head noun is co-referenced by the genitive form of the topical, referential human pronoun (nò ‘REF’). (4)­­ nì̵ŋgwá kù yi yíkí̵ ndèɗú noò rè14 nì̵ŋgwá kù yi (H)yí-kí̵-Ø ndèɗú noò rè man indf.prn p2 pfv.steal-comp. -pfv.d sheep ref.gen prox “The man whose sheep was/were stolen.”15 The relative clause in (5) is unusually long and contains a complement clause that follows the verb ɗúnɨ ‘to need’ which also has an irrealis consecutive clause following it that is still within the limit of the relative clause, as marked by the genitive marker on kí ‘ANA.PRN’. In addition, the head noun is co-referenced within the relative clause by the anaphoric singular pronoun instrument (kí̵ /kí ‘ANA.PRN) at the end of the embedded irrealis sequential clause inside the relative clause: (5) Gùr Yáyàà S 2.6 Ŋgə́ í̵ túùmnà mɨ pə́ p du ɗúnɨ mɨ gùttɨ, ɓwâ kí̵ mɨ a fàìn ní̵m ɓe kíí. túùm-nà(L) mɨ pə́ p Ŋgə́ í̵ du ɗú-nɨ mɨ 3sg nar show-io.cons 1sg glasses ipfv.d need-inf 1sg gùt-tɨ Ø ɓwâ kí̵ mɨ a fàìn(L) ní̵m buy-sbjv.d O.gap time ana 1sg irr.seq read.cons thing ɓe kí=í. with ana.prn=gen  “He showed me some glasses that I needed to buy (them) then I would read with them.” The extent of this long relative clause is indicated by a) the dependent/defocused form of the IPFV aspect marker (D), b) the gapped object (§ 3.2 below) and

12 The human referential pronoun may only refer to the participant that is topical/thematic at that place in the text. 13 Both instrument and accompaniment are the object of the preposition ɓe ‘with’ in Vute. 14 This example is based on a sentence from the Vute text Ndèɗú ɓe Màŋgù ‘Sheep and Hyena’. 15 English and other European languages often use passive voice to get around ungrammatical constructions at the lower end of the NPAH.



9 Relative clauses in Vute grammar and discourse 

 215

c) the enclitic genitive marker on the inanimate anaphoric pronoun which occurs in sentence final position (§ 3.3 below).

3.2 Dependent/defocused verbal aspect marking All verbs in Vute are marked to indicate whether they fall within or outside of the scope of focus16 of the sentence within which they occur. The verb in a relative clause in Vute is marked as dependent or defocused (D). The perfective and imperfective aspect markers both change form to indicate that a verb is part of dependent or defocused information (Thwing and Watters, 1987: 117). The independent/focused form of the perfective aspect marking carries a prefixed floating H tone and either a H-tone copy of the final vowel of the verb stem as in (6a) below, or the suffix -tí̵ ‘PFV’ is added at the end of the verb as the first verb in (6b). The dependent/defocused form for the perfective aspect carries only the prefixed H tone as in the second verbs in (6b) and in (6c). (6a) Focused/independent aspect marking in both clauses: Dùrù ɓe Lì S 4.4 Mvii yi tamcií lâs, ŋgə́ áyè yó. Mvii yi (H)tàm-cì-í lâs, ŋgə́ á-yè yó. sun p2 PFV.become_strong-already-PFV all 3SG Ipfv-still house.loc “The sun had already become fully strong, and he was still in the house.” (6b) Focused/independent aspect marking in the matrix clause with defocused/ dependent aspect marking in the complement clause for the reason that its subject lì is in focus and the verb falls outside the scope of focus of this ­complement clause. Dùrù ɓe Lì S 4.7 No njɔ̧̀ ɔ̧ ŋgə́ yi ɗəəŋtí̵ lì yi í̵ éí ní ə́ r kí̵ ru. No njɔ̧̀ ɔ̧ ŋgə́ yi (H)ɗə̀ ə̀ ŋ-tí̵ lì yi í̵ ref.gen liver.loc 3 sg P2 PFV.know-PFV dog P2 sf (H)éí-Ø17 ní ə́ r kɨ ru PFV.work-PFV.D log work ana clft “In his heart, he knew that it was dog who had done that to him.”

16 The Scope of Focus of a sentence is here defined as the information in a sentence that is new or contrary to expectation. The verb may or may not fall within the scope of focus of a particular sentence. When the verb falls outside of the scope of focus it is said to be out-of-focus or defocused. The aspect markers on a verb tell whether the speaker deems that verb to be within the scope of focus or out of focus. Verbs in relative clauses are always marked as out of focus or defocused. 17 This occurrence of defocused/dependent aspect marking for the verb éí-nɨ ‘work’ is due to the fact that the verb does not fall within the scope of focus of the sentence, in addition to its occurrence in a relative clause. The subject of the complement clause ‘dog’ is the focused element.

216 

 Rhonda Thwing

(6c) Dependent/defocused perfective aspect marking: Dùrù ɓe Lì. S 1.1 Ókye ní̵m yi ndɨŋ dùrù ɓe lì ducɨ bí̵ ɓwàjììb séé. H)ók-yè ní̵m Ø yi (H)ndì̵ŋ-Ø dùrù ɓe imp.hear-before thing s.gap p2 pfv.make-pfv.d baboon with lì du-cɨ18 bí̵ ɓwàjììb sé=é. dog eq.d-already pf.neg.3pl friend.pl eq.neg =gen “Listen first to the thing that caused Baboon and Dog not to be friends anymore.” The imperfective aspect, which expresses both habitual and progressive meanings in Vute, is marked periphrastically with an auxiliary preceding the main semantic verb in the infinitive form. The independent/in focus form of the imperfective auxiliary is á ‘IPFV’ as in the second clause of (6a) above and the first clause of (7a) below, while the dependent/defocused form of this auxiliary is du ‘IPFV.D’ (7b) below. (7a) Kì̵cáàb Ndóŋ Mvèìn S. 1.2 Ní̵m á lə́ ə̀ mnɨ ɗí̵m kí̵ yi yawo ɓwâ Mèín yi nyaŋhȩ ɗɔ̀ ɔ́ bí á. ní̵m á lə́ ə̀ m-nɨ ɗí̵m kí̵ yi (H)yà-wò-Ø19 1pl ipfv think-inf custom ana p2 pfv.begin-ven-pfv.d ɓwâ Mèín yi (H)nyàŋhȩ̀-Ø20 ɗɔ̀ɔ́b=í á. time God p2 pfv.make-pfv.d world=gen foc “We think that this custom began when God created the world.”21 (7b) Dùrù ɓe Lì S 1.2 Sà̧ à̧ mɨ du wu átì̵nànɨ rè sà̧ à̧ mɨ du wu átì̵-nà-nɨ22 story 1sg ipfv.d 2sg say-io-inf “The story I am telling you”

Ø rè do.gap prox

18 Note that equative predicates as in this example, as well as existential, locative, possessive, and descriptive predicates also use the auxiliaries á and du. 19 The pfv.d suffix here results from the fact that this verb and its subject fall outside of the scope of focus and they are part of the information the storyteller assumes he shares with his listeners/readers. 20 This pfv.d suffix results from its function within a relative clause even though the noun phrase (head + relative clause) of which it is a part is in focus here. Syntactic rules tend to dominate pragmatic rules in Vute. 21 literally: the time that God created the world 22 The verb suffix -na ‘-io’ indicates the presence of a recipient/addressee/beneficiary/or a person harmed by the action of the verb ‘malafactee’. The -na always precedes what is often called the indirect object.



9 Relative clauses in Vute grammar and discourse 

 217

3.3 Genitive marker (GEN) and other NP-final Elements Many relative clauses in Vute carry an enclitic genitive marker (GEN) on the last word in the relative clause as if, in some way, a head noun plus relative clause parallels a head noun plus a possessor noun (noun + (noun = GEN))23. The genitive marker is an enclitic, and not a suffix. It may take the form of a lengthened final vowel or an extra syllable added to the last word in a genitive noun phrase or relative clause. Its function seems to be to signal the end of the NP. Other possible phrase-final elements for a NP which contains a RC are demonstratives and quantifiers. When the final demonstrative (or quantifier) modifies the head noun of the embedding NP, that demonstrative replaces the genitive marker. The demonstratives, quantifiers, and/or GEN indicate the right boundary of the noun phrase and set it off from the matrix clause. NP-final demonstratives which modify the head noun are not marked with the GEN (8a), only those that modify the noun in final position in the relative clause are (8b, 9). However, the inanimate anaphoric pronoun kí̵ (ANA.PRN) is marked with the GEN in relative clause-final position, when it functions as a trace pronoun to mark the position of NPREL (8c). (8a) Dùrù ɓe Lì S 3.7 kúkwí dúrú tɨ leehó̧ lúu kí̵ kúkwí dùrù tɨ (H)lèè-hó̧-Ø bone baboon p1 pfv.throw-out-pfv.d “The bone(s) that Baboon threw outside” (8b) Nyó̧ ó̧ yó ku S 4.7 nò tɨ tómwona ní nyó̧ ó̧ kíì nò Ø tɨ (H)tóm-wò-nà-Ø ref s.gap p1 (H)send-ven-io-pfv.d “The person who sent her that snake”

Ø (do.gap)

ní log

nyó̧ó̧ snake

lúu outside

kí̵ ana

kí=ì ana=gen

(8c) Wàánjí ɓe Nì̵wàtì̵b S 7.1 njɔdì ŋgə́ yi cuwo ɓe kíí njɔdì ŋgə́ yi (H)cù-wò-Ø ɓe kí=i 24 wealth 3sg p2 pfv.return-ven-pfv.d with ana.prn =gen “The wealth that he returned with [it]”

23 Note that the genitive marker follows the possessor noun rather than occurring between the two nouns as in a Bantu language. 24 In comparing the final words of (8b, c) above, it is possible to see that the tone melody of kí̵ the anaphoric pronoun + gen differs from that of kí̵ the anaphoric demonstrative + gen.

218 

 Rhonda Thwing

(9) Fùgamè á Wu Bìnɨ S 3.4 nò yi kwíhɨ fùgamè rèè nò Ø yi (H)kwíhì̵-Ø fùgamè ref s.gap p2 pfv.kill-pfv.d buffalo “The person who killed this buffalo”

rè=è prox=gen

Even verbs that occur finally in a relative clause may be followed by the genitive marker. The shape of the GEN after a clause-final verb is determined by the aspect of the predicate. In the perfective (PFV) aspect, the GEN enclitic takes the form of the suffix -lé (or its morphophonological alternates -ré, -ndé). In the imperfective (IPFV), the GEN is indicated by lengthening the vowel of the infinitive suffix -nɨ with a L tone copy, which creates a ML downglide, -niì ‘INF. GEN’. In (10) there are two relative clauses. In the preposed circumstantial constituent25, the genitive enclitic attaches to the infinitive suffix of the IPFV and creates the -niì ‘INF =GEN’ in clause-final position; the relative clause that modifies the subject of the second clause within the direct quotation ends with a verb in the perfective aspect, with the perfective aspect, -lé/ -ré/ -ndé genitive form attached to a PFV-marked verb occurring at the end of that relative clause. (10) Ɓwàjìrí ɓe Kùŋ S 8.6  Hȩ ŋgə́ í̵ gò ə́ ə̀ ŋniì, kù í̵ tànà ŋgə́ , “Yée, péye, ŋgar kù taré á nò sí̵màŋhȩ̀ wa rè naá, ɓwâ kí̵ ɓwàjìrí woò a ɓâr yáá.” Hȩ ŋgə́ (H)gò-Ø ə́ə̀ŋ-niì, kù í̵ way 3sg pfv.come-pfv.d stand-inf.gen indf.prn nar tà-nà(L) ŋgə́ , “Yée, (H)pé-yè, ŋgar ku say-io.cons 3sg Pardon! imp.see-before medicine indf.prn (H)tà-ré (Ø) á nò sí̵m-àŋhȩ̀ wa pfv.say-pfv.gen do.gap foc ref jump-across.sbjv fire rè naá, ɓwâ kí̵ ɓwàjìrí woò a ɓár(L) yáá.” prox susp time ana friend 2sg.gen irr.seq live.cons qt

25 Even though this clause functions like a circumstantial adverbial clause, from a syntactic point of view it is still a (generic) head noun plus a relative clause which carries all the markers one expects in a relative clause: the aspect of the main verb is marked as dependent/defocused, in contrast with the verb in complement clauses, and the last word in the relative clause carries the GEN.



9 Relative clauses in Vute grammar and discourse 

 219

“Because26 he came to stand there, someone told him, ‘Excuse me, first understand that the medicine that was prescribed (lit: the medicine that someone said) is that someone should jump across this fire, then your friend will live.’ ” In the relative clause in (11), nò ‘REF’, the pronoun subject of the matrix clause, functions as head for the RC and is immediately followed by the main verb of the RC, a complement-taking verb marked PFV.D. The final verb of the RC is an infinitive of purpose that ends with the -niì suffix. (11) Kì̵cáàb Ndóŋ Mvèìn S 5.12,13 “Kə́ í, muti̧ hȩ wu, mvá ti samɓàì, nò ta wu gò mɨ ŋgáár ŋgwéé hèìwúniì á na?” “Kə́ í, mutí hȩ sam-ɓàì, wu, mvá ti nò Ø hey! child way 2sg rat dim nothing-very ref s.gap (H)tà-Ø wu gò mɨ ŋgáár ŋgwéé pfv.say-pfv.d 2sg come.sbjv 1sg hand head.loc hèì-wù-niì á na? touch-up-inf.gen foc who “Hey! a child like you, a worthless little rat, who is the one who said you should come to touch my head?” (Lit: the person that said you should come touch my head is who?).

3.4 Nonrestrictive relative clause structure Nonrestrictive relative clauses in Vute show all the morphosyntactic characteristics of restrictive relative clauses except that only subjects and objects can be relativized in nonrestrictive RCs (12). A nonrestrictive relative clause begins with the pronoun nò ‘REF’ which is used outside of nonrestrictive relative clauses in Vute to refer to a referential, though not necessarily definite, human or anthropomorphized being, as in (8b, 9a, 9b, 11) above. Nonrestrictive relative clauses are exceedingly rare in Vute texts. Their only function is to highlight thematic information or topical participants. Example (12) contains two non-restrictive relative clauses, which state the most thematic quality of each of the two main characters in the tale, Dùrù ɓe Lì. Dog is labeled a glutton because he went outside to gather all the bones that baboon had thrown out, mistakenly also

26 hȩ ‘manner’ is a generic head noun followed by a relative clause which functions adverbially, so I translate hȩ as ‘because’ in this context. (See § 3.3)

220 

 Rhonda Thwing

grabbing the bony thigh of one of the old women. Baboon’s red buttocks were revealed when he was forced to show himself in public without his loincloth because dog had stolen it. (12) Dùrù ɓe Lì S 5.1 Cùr kí̵ no, á tí̵gwìn yi ta, á ɗúnɨ lì, nò du mvècɔɔ̀ ŋ, ɓùbîˉŋ nì̵gwi í̵ ndôŋ dùrù, nò du ɓe wa lɨríì, ŋgə́ tì̵ toŋkí̵ ní̵ kúr yáá. Cùr kí̵ no, á tí̵gwìn yi (H)tà-Ø, á 27 behind ana in foc father_in_law p2 pfv.say-pfv.d ipfv ɗú-nɨ lì nò du mvècɔɔ̀ŋ ɓù-bîˉŋ nì̵gwi need-inf dog ref e q.d glutton.gen marry-whole.sbjv woman í̵ ndóŋ(L) dùrù nò du ɓe wa nar surpass.cons baboon ref ipfv.d with fire lɨrí=ì, ŋgə́ tì̵ (H)tòŋ-kí̵ ní̵ kúr yáá. buttock.loc=gen 3sg irr pfv-burn-comp. ant village qt “After that, the father-in-law said that it was better that dog, who was a glutton, marry his daughter rather than baboon, who had fire on his ­buttocks, lest he burn down the village.” Example (13) contains a nonrestrictive relative clause which refers to the protagonist in the text, a sheep who is introduced in the first sentence of the folktale. The previous episode of the folktale contains the information that hyena’s favorite meat is sheep and that he allowed sheep to leave. We hear hyena’s thoughts here, how he is amazed and disgusted that he let sheep trick him into letting him go instead of eating him. (13) Ndèɗú ɓe Maŋgù S 3.2 Ŋgə́ í̵ tà nò ní mbí̵k gìnɨ rè ndèɗú á, fè níì ɗûní̵mí ánɨ ru. tà(L) Ŋgə́ í̵ nò ní (H)mbí̵k-Ø gì-nɨ rè 3sg nar say.cons ref log pfv.leave-pfv.d go-inf prox ndèɗú á fè ní-ì ɗûní̵m=í á-nɨ ru sheep foc meat.np1 log=gen love.nom=gen eq-cf clft “He said that this one that he allowed to leave is a sheep, and it is sheep that is his favorite meat.”

27 cùr ‘behind’ is a noun in Vute and thus can be qualified by the anaphoric demonstrative.



9 Relative clauses in Vute grammar and discourse 

 221

4 Generic noun plus relative clause constructions In Vute there are nine generic nouns, which when modified by a relative clause have a) adverbial functions, or b) encode the potential mode: àm ‘word, problem’, ɓwâ ‘time’, ye ‘day’, yá /yí ‘place’, hȩ ‘way’, ɓè ‘that place’, ní̵m ‘thing’, sàmàìn ‘possibility’, and nyòme ‘strength’. Reason, manner, location in time and space, etc. do not seem to take part in the NPAH; this is reflected in Vute by the fact that there is no marking of NPREL by a gap or a pronoun in these constructions. I will begin with those generic noun + relative clause elements that have adverbial functions.

4.1 Generic N + RC constructions which function as adverbial constituents Even though the N + RC constructions in § 4.1 function like adverbial subordinate clauses, from a morphosyntactic point of view they are still a head noun plus a relative clause which carries the markers one expects in a relative clause: the aspect of the main verb is marked as dependent/defocused, and the last word in the relative clause either carries the GEN or is a demonstrative adjective or quantifier. Therefore I regard them through “morphosyntactic lenses” as noun phrases rather than as subordinate clauses. 4.1.1 One of the common ways to encode a semantic reason is by means of the generic noun àm ‘word, problem’ modified by a relative clause containing the reason. Reason constituents introduced by àm can never occur in sentence-initial position28. They also must follow the result for which they are the reason, like all reason constituents in Vute. (14) Reason constituent Gɔ̀ Tùkur yi Gɔklé S 3.2 ŋgə́ í̵ tà kí du àm íí níì tɨ pé fèhí á. tà(L) kí du ŋgə́ í̵ àm 3sg nar say.cons ana.prn eq.d word (H)pé-Ø fèhí á (H)see-pfv.d meat.gen foc “He said it was because his eyes saw the meat.”

íí eye

ní=ì log=gen

tɨ p1

28 Adverbial reason clauses introduced by ɓetí ‘because’ may occur in sentence-initial position following their result which appears in the previous sentence. Ɓetí is often used this way to introduce a proverb into a discussion.

222 

 Rhonda Thwing

Sometimes àm is preceded by ɓe ‘with’ so that the reason constituent functions syntactically as if it were a NP object of a preposition. (15) Reason constituent Gùr Yáyàà S 8.1 Kí á mɨ ɗú i̧yáànaá mɨ nì̵mɨr, mɨ sáktì̵yàyè Mèín ɓe àm ŋgə́ ndɨŋ i̧yáànaá mɨ fó̧ ó̧ mì̵yí̵ kí̵, mɨ kúwá. Kí á mɨ (H)ɗú-Ø i̧yáànaá29 mɨ nì̵mɨr, mɨ ana.prn foc 1sg pfv.want-pfv.d comp 1sg owner 1sg sáktì̵-yà-yè Mèín ɓe àm ŋgə́ (H)ndì̵ŋ-Ø thank-begin-first.sbjv God with word 3sg pfv.do-pfv.d i̧yáànaá mɨ (H)fó̧ -Ø mì̵yı̵̌ kí̵ mɨ kú-wá.30 comp 1sg pfv.leave-pfv.d place ana 1sg die-pfv.neg “So I want that I myself, I thank God because he made it that I left that place, I did not die.” 4.1.2 A purpose constituent may also be encoded by means of the generic noun àm plus a relative clause. Like an àm-introduced reason constituent, an àm-introduced purpose constituent may not occur first in the sentence. The difference between a reason constituent and such a purpose constituent is the form taken by the verb of the RC. Reason constituents have finite, realis verb forms in the perfective aspect or one of the past tenses while the verbs contained in àm-introduced purpose constituents have nonfinite, future, or irrealis marking of some type. (16) contains two purpose constituents. The first purpose constituent is introduced by a verb in the infinitive form whose object NP contains a relative clause. It tells the immediate purpose of the action of measuring the feet of the young women. The final clause in the sentence is an àmintroduced purpose constituent that relates the ultimate purpose for the action. (16) Àm Níì á du Ɗúwá kùb Ɓáín Mwin Kwíhì̵r? S. 3.3  Ŋgáb í̵ yà ɗóòtèènànɨ mɨkóómiib lâs mánɨ nò guur noò ɓa èrì̵nɨ ɓe táb kíì àm yèrímà ɓùhá ŋgéé. ɗóò-tèè-nà-nɨ ŋgáb í̵ yà(L) mɨkóómiib 3pl nar begin.cons measure-around-io-inf young_woman.pl lâs [má-nɨ [nò guur noò ɓa èrì̵-nɨ ɓe táb all search-inf ref foot ref.gen ints fit-inf with shoe.np1 kí=ì] [àm yèrímà ɓù-há ŋgéé]] ana=gen word prince marry-only.sbjv 3sg.gen

29 i̧ yáànaá functions like a complementizer here. It is composed of three morphemes: i̧ ‘how?’, yáà ‘polite command (POL)” and naá ‘increasing suspense (SUSP)’. 30 Notice the perfective verb forms in the reason clause.



9 Relative clauses in Vute grammar and discourse 

 223

“They measured all the young women to search for the one whose foot really fit the shoe so that the prince could marry her.” The àm-introduced purpose constituent in (17) relates a negative purpose with the marker tì̵ IRR in the verb phrase. (17) Hààmɗə́ ə̀ ŋhȩ̀ ȩ́ Mutíí Section 2, S 5.11 tûmnam ŋgə́ ɗûní̵m, wu a nà ŋgə́ tɔ̀ ɔ́ ɓe óòní̵m hȩ Kéké ya nakomndé, àm yȩȩ́ ɓe ye-cùrìì, ŋgə́ tì̵ ɓandɨcuré wu no kwee. (H)tûm-nà-m ŋgə́ ɗûní̵m wu a nà(L) ŋgə́ imp.show-io-imp 3sg love 2sg irr.seq give.cons 3sg tɔ̀ ɔ́ ɓe óòní̵m hȩ Kéké ya (H)nà-kòm-ndé31 teach.ger with listen.nom way Lord P3 pfv.give-usually-pfv.gen àm yȩȩ́ ɓe ye cùrì=ì ŋgə́ tì̵ word tomorrow with day behind=gen 3sg irr (H)ɓàndì̵-cù-ré wu no kwee pfv.turn-again-in 2sg loc there.gen “show him love, you should give him teaching with understanding as the Lord has often given, lest tomorrow or the day after tomorrow he turn on you.” 4.1.3 The generic noun ɓè ‘this place’ can introduce a conditional constituent. In (18), taken from a hortatory text about how to discipline children, there are two ɓè -introduced clauses. (18) Hààmɗə́ ə̀ ŋhȩ̀ ȩ́ Mutíí S. 10.4 Wu a tûmnàɓwê ŋgə́ ə́ rí̵b kwéè: ɗȩ̀ ȩ́ , ə́ rí̵b káhȩ̀ né ɓè du mwin nì̵ŋgwáà; ɗȩ̀ ȩ́ , sìì, ɓe ə́ rí̵b yóò ɓè duce mwin nì̵gwiì. tûm-nà-ɓwê(L) kwé=è ŋgə́ ə́ r-í̵b wu a 2sg irr.seq show-io-adp.cons 3sg work-pl village.loc=gen ɗȩ̀ ȩ́ , káhȩ̀ né ə́ r-í̵b ɓè du mwin nì̵ŋgwá=à farm.ger work-pl forest.loc.gen place eq.d child man=gen ɗȩ̀ ȩ́ sìì ɓe ə́ rí̵-b yó-ò ɓè farm.ger cook.nom with work-pl house.loc=gen place du-ce mwin nì̵gwi=ì eq.d-adsim child woman=gen “You should also show him/her the work of the village: farming, work in the forest if he is a boy; farming, cooking, and housework if she is a girl.”

31 For a manner constituent which is also a N + RC, see § 4.1.4.

224 

 Rhonda Thwing

In (19) there are two conditional constituents. The first is a subordinate conditional clause much like conditional clauses in a European language. The second conditional constituent is the complement of the verb ɗə̀ ə̀ ŋnɨ ‘to know’ and is introduced by the noun ɓè ‘this place’. (19) Ndèɗú ɓe Màŋgù S 2.4 Kí á màŋgù ta ní ɗúú ŋgə́ mìsówò nà̧ á̧ ; óo nà̧ á̧ natí̵, ní a ɗə̀ ə̀ ŋgì ɓè ŋgə́ du mèín mì̵ténèé yáá. ní Kí á màŋgù (H)tà-Ø (H)ɗú-ú ŋgə́ ana foc hyena pfv.say-pfv log pfv.want-pfv 3sg mì-só-wò ní nàá; óo nàá (H)nà-tí̵, a mist-down-VEN.SBJV rain if rain pfv.rain-pfv log irr.seq ɗə̀ ə̀ ŋ-gì(L) ɓè ŋgə́ du mèín mì̵ténè=é yáá. know-after.cons place 3sg eq.d god little=gen qt “So then Hyena said that he (hyena) wanted him (sheep) to make it rain; if it rained, then he (hyena) would know if he (sheep) was a little god.”32 4.1.4 The manner constituent in (20) occurs at the end of the sentence in the unmarked position for manner expressions (See also (17 above). It is a noun phrase consisting of the generic head noun hȩ ‘way, manner’ plus a restrictive relative clause. (20) Múkúr ɓe Mvì S 12.3 Kí á múkúr nì̵gwi rè ta ŋgə́ óóye hȩ ŋgə́ ɓa ŋgə̀ mniì yáá. Kí á múkúr nì̵gwi rè (H)tà-Ø ŋgə́ ana.prn foc old_woman woman prox pfv.say-pfv.d 3sg óó-yè hȩ ŋgə́ ɓa ŋgə̀m-niì yáá. hear-before.SBJV way 3sg f1 praise-inf.gen qt “So then the old woman said that he should first listen to the way he would praise (his ancestors).” 4.1.5 The three temporal constituents in (21) all have the generic noun ɓwâ ‘time’ as their head noun.33 In the first and third sentences ɓwâ is modified by relative clauses; in the second it is modified by kí̵ ‘ANA’ the anaphoric demonstrative. All three occur in the unmarked position for temporal constituents at the end of the sentence and have local relevance in the sentences in which they occur rather than functioning as discontinuity markers in the larger parapraph or discourse.

32 I have included the nouns to which the pronouns refer to help with understanding. 33 Temporal constituents composed of noun + RC may also have ye ‘day’ as head noun.



9 Relative clauses in Vute grammar and discourse 

 225

(21) Kì̵cáàb Ndóŋ Mvein S 1.2–4 Ní̵m á lə́ ə̀mnɨ ɗí̵m kí̵ yi yawo ɓwâ Mèín yi nyaŋhȩ ɗɔ̀ ɔ́bí á. Éìye, ɗí̵m kí̵ yi nɨ yàwòwá ɓwâ kí̵. Yi yawonɨ ɓwâ kù yi fein Kì̵cáàb Ndóŋ Mvèìn á. ní̵m á lə́ ə̀ m-nɨ ɗí̵m kí̵ yi (H)yà-wò-Ø ɓwâ 1pl ipfv think-inf custom ana p2 pfv.begin-ven-pfv.d time mèín yi (H)nyàŋhȩ̀-Ø ɗɔ̀ɔ́b=í á éìye, ɗí̵m kí̵ God p2 pfv.make-pfv.d earth=gen foc no custom ana yi nɨ yà-wò-wá ɓwâ kí̵ yi (H)yà-wò-nɨ ɓwâ p2 neg begin-ven-pfv.neg time ana p2 pfv.begin-ven-cf time kù yi (H)fèìn-Ø kì̵cáàb ndóŋ mvèìn á. indf.prn p2 pfv.give_birth-pfv.d cleverness pass chief.gen foc [S 1.1 Now when Vute people give birth to a child, they name him whatever they want.]34 S 1.2–4 “We think that this custom began when God created the earth. No, it did not begin then. It began when Clever-Pass-Chief was born.” 4.1.6 The locative constituent in (22) has the generic noun yá ‘place’ as head noun while (23) has the locative noun yí ‘place’ as head; both are followed by a relative clause. In (22) the relative clause modifies the pronoun kù-b ‘INDFpn-PL’ which is the subject of the matrix clause. (22) Hààmɗə́ ə̀ ŋhȩ̀ ȩ́ Mutíí 2 S 4.7 Cùr kí̵ ya, kùb yá ŋgwé kí̵klé pə́ cê, dùgì̵cé gùúm àmɨrì, ɗàà. cùr kí̵ ya kù-b yá ŋgwé (H)kí̵k-lé behind ana on indFpn-pl place head pfv.put-pfv.gen pə́ -cé(lm) dùgì̵-cé gùúm àmɨr=ì ɗàà see-adseq.fneg conc-adseq marriage true=gen ipfv.neg “After that, her family [Lit:those at the place where (she) puts (her) head] won’t see (her) anymore, not even a true marriage.” In (23) the locative noun yí ‘place’ plus relative clause occurs at the end of the sentence in the unmarked position for locative constituents. (23) Ndèɗu ɓe Màŋgù S1.2 Ndèɗú, kùb ya súŋ ŋgə́ ŋgwàrí yí jírí̵b du tàŋlí̵ŋnɨ cɔɔ̀ ŋ. ndèɗú kù-b ya (H)súŋ-Ø ŋgə́ ŋgwàrí sheep indf-pl P3 pfv.tie-pfv.d 3sg savannah.loc

yí place

34 I include the English translation of the first sentence in (21) so that the context of the passage will be clear.

226 

 Rhonda Thwing

jírí̵b du tàŋ-lí̵ŋ-nɨ cɔɔ̀ŋ goats_and_sheep ipfv.d eat-often-inf food.gen “Sheep, they had tied him in the savannah where goats and sheep usually eat.”

4.2 Generic head N + RC constructions with modal functions The generic nouns sàmàìn ‘permission’ and nyòme ‘strength’ are used to encode modal meanings. Generic noun sàmàìn is used to encode the potential mode idea of ‘be able, can’ (Payne, 244–247). There are two relative clauses in (24) the second of which has sàmàìn as the head noun. I have put both of them in square brackets ([ ]) and used underlining and bolding as in other examples with multiple ­relative clauses. (24) Tùkur ɓe Mì̵nyà S 6.9  Tùkur í̵ pììnà ŋgə́ , “Ɓwàjìrí moò, ní̵m wu ta mɨ ndì̵ŋnɨ kí̵, sàmàìn mɨ ndɨŋndé ɗàà.” pìì-nà(L) tùkur í̵ ŋgə́ , ɓwàjìrí moò, [ní̵m wu hare nar answer-io.cons 3sg friend 1sg.gen thing 2sg (H)tà-Ø mɨ ndì̵ŋ nɨ kí̵] [sàmàìn mɨ pfv.say-pfv.d 1sg do.sbjv cf ana permission 1sg (H)ndì̵ŋ-ndé (Ø)] ɗàà pfv.do-pfv.gen do.gap ex.neg “He answered him, ‘My friend, the thing you said I should do, I can not do (Lit.:the permission that I do (it) does not exist).’” Example (25) contains three different generic noun plus relative clause structures, with the third N + RC embedded within the second. The expression of modality is found in the third relative clause. (25) Mutí ɓe Ŋgàɗə́ r S 2.1 Hȩ ŋgàɗə́ r lè yi óó u̧ nèé, á ŋgə́ ɓéhó̧ wàánjí rè lúu, ŋgə́ í̵ yà ŋgə́ sóòŋnɨ àm ŋgə́ tɨ ta sàmàìn ŋgàɗə́ r kwíhɨ níì ɗàwá kí̵. [hȩ ŋgàɗə́ r lè yi (H)óó-Ø u̧ nèé], á ŋgə́ way shape_shifter prox p2 pfv.hear-pfv.d thus.gen foc 3sg (H)ɓé-hó̧ -Ø wàánjí rè lúu, ŋgə́ í̵ yà(L) pfv.call-out-pfv.d youth prox outside 3sg nar begin.cons ŋgə́ sóòŋ-nɨ [àm ŋgə́ tɨ (H)ta -Ø [sàmàìn 3sg advise-inf word 3sg p1 pfv.say-pfv.d permission ŋgàɗə́r (H)kwíhì̵-Ø ní=ì] ɗàwá kí̵ ] ]. shape_shifter pfv.kill-pfv.d log=gen ex.neg ana



9 Relative clauses in Vute grammar and discourse 

 227

“Since the shape-shifter heard this, he called the youth outside, and began to counsel him because he said that the shape-shifter could not kill him [lit: the permission for the shape-shifter to kill him did not exist].” The modified head noun form is called N1 by Thwing (1987: 55)35. The N1 form of the noun nyòme ‘strength’ is nyòm ‘strength.N1’. It is used in (26) to encode the potential mode idea ‘be able to’. The relative clause with nyòm as its head noun is embedded in another relative clause with cɔŋ ‘food’ as its head noun. Note both of the GEN markers that close first the embedded RC (inner square brackets) and then the embedding RC, which is enclosed within the outer pair of brackets. (26) Wàánjí ɓe Nì̵wàtì̵b S.6.7 Ŋgə́ í̵ nà ŋgə́ cɔŋ ŋgə́ yi du nyòm taŋcáaŋndé ɗàwáá. nà(L) ŋgə́ í̵ ŋgə́ [cɔŋ ŋgə́ yi du [nyòm 3sg nar give.cons 3sg food 3sg p2 ipfv.d strength.np1 (H)tàŋ-cáàŋ-ndé (Ø)] ɗàwá=á ]. pfv.eat-finish-pfv.gen do.gap ipfv.neg=gen “He gave him (so much) food that he was unable to finish eating it.”

5 The discourse functions of NPs containing relative clauses Relative clauses in Vute discourse occur fairly frequently, possibly more frequently than in most Cameroonian languages. However, in spite of their frequency, relative clauses in Vute texts have only two basic functions whether they qualify a head noun that is being introduced for the first time or a generic noun or noun that has already been introduced. They function either as a marker of discontinuity or as a means to highlight topical participants or thematic information.

5.1 As markers of discontinuity As Dooley and Levinsohn (2001: 37) state, discontinuities tend to occur in a text when there is a change of time or place, called a situational discontinuity, a

35 Not all nouns have special NP1 forms in Vute today. Only the most frequently used nouns retain this marking.

228 

 Rhonda Thwing

change of thematic participant or topic, called a referential discontinuity, or a change from event information to non-event information or events out-of-order, e.g., flashbacks, switch from action to conversation, to narrator comments, or irrealis information or vice-versa, called an action discontinuity.36 There are three noun + RC constituents in (27), which comes from a text whose background situation is taking a wife. The first N + RC is a temporal constituent, with generic head noun ɓwâ ‘time’ in sentence-initial position, and functions as a marker of action discontinuity in this text, marking a change from main events to non-event information: the narrator’s humorous comments. The other two RCs highlight information of thematic importance, emphasizing, and even exaggerating the great welcome feast when the young girls saw the handsome bachelors (§ 5.2 below for more explanation). I include the English translation of the previous sentence (S 3.1) for context: (27) Dùrù ɓe Lì S. 3. 2 Ɓwâ mɨkóómɨb yi pé ŋgábè, cuní̵b yi kú ye kíì nɨ ní̵m fainndé sé. [ɓwâ mɨkóómɨb yi (H)pé-Ø ŋgáb=è ] [cuní̵b (Ø) yi time young_girls p2 pfv.see-pfv.d 3pl=gen chicken.pl s.gap p2 (H)kú-Ø ye kí=ì ] nɨ [ní̵m (H)fàìn-ndé ] sé PFV.die-PFV.D day ana=gen neg thing pfv.count-pfv.gen eq.neg [When baboon and dog finished sharing these confidences, they arrived in the village of the young maidens.] “When the young girls saw them, the chickens that died that day were uncountable (Lit.: not something counted).” In (28) a locative constituent consisting of head noun yí ‘place’ plus relative clause occurs at the beginning of the sentence as a marker of situtional discontinuity. I include the English translation of the previous sentence (S 7.1) for context. (28) Tùkur ɓe Mì̵nyà S 7. 2 Ácè yí tùkur gɨce yòknɨ ri, mì̵nyà í̵ yà kí̵ŋnɨ, “Ní̵m á rò, ní̵m á yokkí̵ kaín ndè!”. A-cè [yí tùkur (H)gɨ-ce-Ø yók-nɨ ri] ipfv-adsim place hare pfv.go-adsim-pfv.d hide-inf dist mì̵nyà í̵ yà(L) kí̵ŋ-nɨ ní̵m á rò, ní̵m scorpion nar begin.cons yell-inf 1pl Ipfv prox 1pl á (H)yok-kí̵ kaín ndè ipfv pfv.hide-cmpl here prox [Because the villagers saw this [hare has just hit his father-in-law on the head with his hoe], they aggressively hunted for hare.]

36 For categories, functions and formal correlates of types of information in Vute narrative text see Thwing, 2008, 3–18.

9 Relative clauses in Vute grammar and discourse 



 229

“Also, wherever hare went to hide, scorpion started to yell, “Here we are, we’re hidden in this place!” The manner constituent in (29) consists of a N + RC with head noun hȩ ‘way, manner’. Sentence initial position is a marked position for manner or circumstance constituents which normally fall at the very end of the sentence. In this position, a circumstantial constituent serves as a marker of discontinuity, here a referential as well as an action discontinuity caused by the change of the main participant from tortoise to hare and from quoted speech to action. The second N + RC contains thematic information since it concerns tortoise’ revenge on hare. I include the English translation of the previous context, S 4. 4–6 that tells about the lie tortoise told to trick hare. (29) Gɔ Tùkur yi Gɔklé S 5.1 Hȩ tùkur ya du ní̵ ɓe nò ɓe kùŋ duré ɗàwáá, á ŋgə́ yi lu, .... [hȩ tùkur ya (H)du-Ø ní̵ ɓe [nò

way

hare

p3

pfv.stay-pfv.d

ant

with

ref

ɓe

with

kùŋ

neighbor

(H)du-ré ]37 ɗàwá=á ] á ŋgə́ yi (H)lu-Ø pfv.stay-pfv.gen ex.neg=gen foc 3sg p2 pfv.run-pfv.d [So hare arrived at the village with the package of meat on his head. He asked them what was going on and tortoise replied, “God wants to destroy the earth; all people need to be two-by-two. Whoever is all alone will die.”] “Since hare did not have another person to live with, he ran, ... .”

5.2 As markers of thematic information In Vute, relative clauses introduce only main characters. In (30), taken from the tale Young Man and Shape-shifter, the young man as the main character is introduced into the tale for the first time as ‘the son of the people that the shapeshifter killed’. This young man is also the protagonist of the folktale. He is introduced with both a relative clause and the proximal demonstrative PROX which functions as marker of highlighted thematic information in Vute38.

37 This circumstantial constituent has another relative clause embedded in it whose head is the pronoun nò ‘REF’. Note both GEN markers. 38 The proximal demonstrative rè ‘PROX’ and directional verb suffixes, sometimes called verb extensions in Bantu and Chadic languages, also indicate that a NP refers to a thematic participant or information. (For verbal extensions in Vute see Thwing, 2008: 21.)

230 

 Rhonda Thwing

(30) Mutí ɓe Ŋgàɗə́ r S 1.4 Ye kwè ŋgáb yaá du̧ ú̧ dúú mbə́ rə́ ə̀ ɓe mwin nùb ŋgə́ yi kwíhe rè. Ye kwè ŋgáb yaá du̧ ú̧ dúú mbə́ rə́ =ə̀ day indf 3pl p2.ipfv sit.part house.np1 corn_wine=gen ɓe mwin nùb ŋgə́ yi (H)kwíhe-Ø rè. with child.np1 ref.pl 3sg p2 pfv.kill-pfv.d prox “One day, they were seated in the house (where they serve) corn wine with the son of the people he (the shape-shifter) had killed.” In (31), taken from the folktale Ndèɗú ɓe Màŋgù, the relative clause is nonrestrictive with a pronoun head, whose referent is the protagonist of the tale who has already been clearly identified. In § 3.4 and (12) above we saw that nonrestrictive relative clauses serve to highlight thematic information. Note also the proximal demonstrative at the end of the RC which also functions to signal highlighted thematic information and topical participants, in addition to physical proximity. In (31) we hear hyena’s thoughts when he realizes that he just let his favorite meat escape.39 (31) Ndèɗú ɓe Màŋgù S 3.2 Ŋgə́ í̵ tà nò ní mbí̵k gìnɨ rè ndèɗú á, fè níì ɗûní̵mí ánɨ ru. Ŋgə́ í̵ tà(L), nò ní (H)mbí̵k-Ø gì-nɨ rè 3sg nar say.cons ref log pfv.allow-pfv.d go-inf prox ndèɗú á fè ní=ì ɗûní̵m=í á-nɨ ru. sheep foc meat.np1 log=gen like.nom=gen foc-cf clft “He said that the one he allowed to leave was a sheep; (it) was his favorite meat.

6 Concluding remarks Relative clauses in Vute show some similarities to relative clauses in head-initial European languages, but they also differ from them. While many languages, e.g., English, use the gap strategy to refer to NPREL, not as many use the trace pronoun strategy for positions lower on the NPAH. Instead relativization is limited to the higher positions on NPAH or syntactic means are used to create grammatical relative clauses. Both European languages and Vute use relative clauses to restrict the identity of referents.

39 Note the contrastive focus marker (CF) in the verb phrase of the second clause.



9 Relative clauses in Vute grammar and discourse 

 231

Knowledge of the typological parameters of relativization in a given language is necessary to produce clear, accurate materials for the development of a language and its community of speakers. It is in the discourse functions of relative clauses that we most see the differences between Vute and European languages. Many European languages, e.g., English and French, use relative clauses to introduce new participants or props, whether or not they are thematically prominent, they add further information about participants and topics in non-restrictive relative clauses, and may even introduce new events in a relative clause, i.e., She gave the vase of flowers to her mother, who placed them on the table. However, Vute limits relative clauses to three or four functions in discourse: as markers of discontinuity, in locally relevant adverbial constituents, as ways to express certain moods, and as markers of thematic prominence both in introducing only important characters and in emphasizing important thematic or topical information.

References Breton, Roland & Bikia Fohtung. 1991. Atlas administratif des langues nationales du Cameroun. Paris: Agence de Coopération Culturelle et Technique (ACCT), Yaoundé: Centre Régional de Recherche et de Documentation sur les Traditions Orales et pour le Développement des Langues Africaines (CERDOTOLA), & Centre de Recherches et d’Études Anthropologiques (CREA). Dooley, Robert A & Stephen H. Levinsohn. 2000. Analyzing discourse. Dallas, Texas: SIL International. Doussam, Jean. 1997. Hààmɗəə̀ ŋhȩ̀ ȩ́ Mutíí [raising and disciplining children]. Yoko: Centre de Littérature Ví̵tè. Keenan, Edward L. 1985. Relative clauses. In Timothy Shopen (ed.), Language typology and syntactic description. Volume II: Complex constructions, 141–170. Cambridge: Cambridge University Press. Lewis, M. Paul (ed.). 2009. Ethnologue: Languages of the world, 16th edn. Dallas, Texas: SIL International. Online version: http://www.ethnologue.com (accessed 12 January 2017). Longacre, Robert E. 1996. The grammar of discourse, 2nd edn. New York/London: Plenum Press. Maxey, James A. 1994. Relative clauses in Vute. Texas: University of Texas at Arlington MA thesis. Payne, Thomas E. 1997. Describing morphosyntax: A guide for field linguists. Cambridge: Cambridge University Press. Thwing, Rhonda A. 1987. The Vute noun phrase and the relationship between Vute and Bantu. Texas: University of Texas at Arlington MA thesis. Thwing, Rhonda A. 2008. Verbal Extensions in Vute. Paper presented at SIL Cameroon Academic Forum, May 2008. Thwing, Rhonda A & John R. Watters. 1987. Focus in Vute. Journal of African Languages and Linguistics 9. 95–121. Thwing, Ronald & Rhonda A. Thwing. 1980. Deux contes en langue ví̵tè:La mort par suite de buffle et Qui est le vrai voleur. Unpublished manuscript submitted to ONAREST, Yaounde. Yakoura, Valentin, Alfred Oumarou, Jean Doussam & James Maxey. 1998. Gèìn Ví̵tè. Yoko:Centre de Littérature Ví̵tè.

232 

 Rhonda Thwing

Appendix Vute Folktale Sources (Gèìn Vítè): Àm Níì á du Ɗúwá kùb Ɓáín Mwin Kwíhì̵r? Why is it that orphans should not be harassed? Friend and Neighbor Ɓwàjìrí ɓe Kùŋ Dùrù ɓe Lì Baboon and Dog The Sharing that Hare Shared Gɔ̀ Tùkur yi gɔklé Gùr Yáyàà Yaya’s Trip Kì̵cáàb Ndóŋ Mvèìn More Clever than Chief Máálàm Malam Múkúr ɓe Mvì The Old Woman and the Toad Youth and Shape-Shifter Mutí ɓe Ŋgàɗə́ r Sheep and Hyena Ndèɗú ɓe Màŋgù Hare and Scorpion Tùkur ɓe Mì̵nyà The Young Man and Evil Spirit Wàánjí ɓe Nì̵wàtì̵b Hortatory Source: Hààmɗə́ ə̀ ŋhȩ̀ ȩ́ Mutíí

Raising and Disciplining a Child

Historical Narative: Fùgamè á wu Bìnɨ

Death by Cape Buffalo

Marieke Martin

10 Relative clauses in Wawa Abstract: Wawa is an endangered Mambiloid language of Cameroon, spoken by about 3000–5000 people and comprising four major dialects. In Wawa, relative clauses (RCs) consist of a relative marker, a verb (and an object if the verb is transitive), as well as a relative determiner or the associative clitic; further nominals and complements can be added but are not compulsory. There is a distinction between restrictive and non-restrictive relative clauses, using two different markers; kí for restrictive and kə́ for non-restrictive RCs. Every NP constituent in a sentence can be relativised. Wawa RCs feature common cross-linguistic characteristics, e.g. the relative marker/pronoun strategy. Subject and object RCs are additionally marked by a gap, while possessor and indirect object focused RCs follow the resumptive pronoun strategy. In addition, Wawa RCs show different verbal morphology to that of main clauses.

1 Introduction1 This chapter surveys the structure of relative clauses in Wawa and has been adapted from the author’s PhD thesis “A grammar of Wawa – an endangered language of Cameroon (2012). Wawa is an endangered Mambiloid language (Bantoid) spoken in the Adamawa Region of Cameroon on the border with Nigeria. The number of speakers was estimated at 3000 in 1991 (Lewis, Simons & Fennig 2013), the data of a later census in the early 2000s was destroyed in a fire. The language comprises four main dialects: Oumyari, Gandoua, Ndi and Mbenguedje Wawa. Within Mambiloid, Wawa is most closely related to Vute (Guarisma 1987, Thwing n.d.), however, relative clauses (RCs) are formed differently in these two languages, as can be seen in the chapter on relative clauses in Vute by Rhonda Thwing in this book (also Thwing 2004). In Wawa, an RC is marked by several constituents, though not all are obligatory: a relative marker, a relative determiner and the associative suffix, and the verb must be derived for either imperfective or perfective participle.

1 Many thanks to Peter Jenks, Sean Bowerman, Bonny Sands, Lutz Marten, and an anonymous reviewer for comments and discussion of earlier drafts. Marieke Martin: University of Cape Town, [email protected] DOI 10.1515/9783110469547-010

234 

 Marieke Martin

Section 2 of this paper gives a brief overview over Wawa basic syntax and morphology occurring in RCs, which is needed to understand the structure of the different types of RCs examined in detail in section 3. In section 4, Wawa RCs are discussed in how they fit into the broader typology of RCs.

2 Morphosyntactic overview 2.1 Sentence structure The typical sentence structure in Wawa is SVO. This is shown in example (1). (1)

mə éérə́ no bóóŋgə̀ . mə ér-ə́ no 1sg leave-pfv dir “I went to the river”

bóóŋgə̀ river

In the imperfective aspect, word order is sometimes used to emphasise the verb. Example (2) is a sentence in the imperfective aspect, whereas in example (3) the object is placed between copula and infinitive to shift focus onto the verb. (2) mə də̄ sììnə̀ náábə̀ . mə də̄ sii-nə̀ 1sg cop cook-inf I am cooking fufu

náábə̀ fufu

(3) mə də̄ náábə̀ sììnə̀ . mə də̄ náábə̀ sii-nə̀ 1sg cop fufu cook-inf “I am cooking fufu (not eating it)” Wawa is generally right-branching (except for possession, which is covered in the next paragraph) and all modifiers and relative clauses follow the head noun phrase (NP) in a sentence. Example (4) shows a noun modified by two adjectives: (4) nàg dʒə́ ̄r bubûr. nàg dʒə́ ̄r bubûr cow big white “A big white cow” There are two possibilities to mark possession in Wawa. The first way is by adding the suffix -ɔ̀ to the possessor noun which precedes the possessed NP, as, in sentence (5) where the 3sg pronoun mū and the noun tá ‘father’ are marked with the possessive suffix -ɔ̀ .



(5) mɔ̄ tâ nàgbə̀ bəmbə̄ bubúrbə̀ . mū-ɔ̀ tá-ɔ̀ nàg-bə̀ 3sg-poss father-poss cow-pl “His father’s two white cows”

10 Relative clauses in Wawa 

bəmbə̄ two

 235

bubûr-bə̀ white-pl

The second way is by marking the possessor with the possessive suffix and the associative suffix and reversing the order possessor and possessed NP. This is discussed below in section 2.3.

2.2 Participles Relative clauses, when positive, include a verb in participle form. The participle forms for imperfective and perfective are marked on the verb root. The verb form in negative sentences is discussed below in 2.2.3. In Wawa ­sentences, the main verb must be marked for aspect or mood; they do not co-occur in subjunctive and potential mood but do in other moods, such as irrealis. Tense is not obligatory and the present tense is unmarked, while other tenses are marked by particles before the verb. Aspect, tense and mood cannot be distinguished on a morphological basis, as irrealis mood is also marked by a particle before the verb, while aspect and mood can be marked by verbal extensions and suffixes. Thwing describes very similar structures from Vute in her paper on Vute Verb Extensions (2007), where she calls them ‘adverbial extensions’. However, their function in Wawa is purely aspectual/modal. When a verb carries an aspect or mood marked by verbal extension, it must be marked with an imperfective or perfective aspect in addition, which both are expressed through suffixes. When a mood or aspect is marked by a suffix, the verb cannot carry a second suffix. 2.2.1 Imperfective participle For the imperfective aspect, the suffixes -də̄ /-lə̄ or the toneless suffix -ii are added to the verb root. The particular form of the suffix depends on the last sound of the verb root; e.g. verb roots ending in long vowels, diphthongs and /ɡ/ take the [-ɾə̄ ] allomorph. The [-lə̄ ] allomorph attaches to verb roots ending in [r]; the phoneme /d/ includes three allophones, [nd] after nasals, [ɾ] intervocalically and [r] stemfinally; therefore roots ending in a nasal take [-ndə̄ ], as in sentence (6). Please note that /l/ is not an allophone of /d/, the phonological process creating the [-lə] allomorph is lateralisation. In other Mambiloid languages [l] and [r] or [ɾ] are allophones, e.g. in Kwanja (Weber & Weber 2008) and Mambila (Perrin & Hill 1969). The toneless -ii suffix is taken by verb roots ending in short vowels. In the case

236 

 Marieke Martin

of schwa the suffix takes its place. The suffix takes on the tone of the verb root. Verb roots can carry tone 1 (high), 2 (higher mid), 3 (lower mid) but tone 4 (low) has not been observed. Verb roots with tone 2 are rare and often loan words from Fulfulde. Tab. 1 summarises the different forms of this derivation:

Tab. 1: Imperfective participle derivation. Suffix

Verb stem

Participle

Gloss

-ii: verb root ending in schwa or short vowel

jɛ́ bsə́ mésakə-

jɛ́ b-íí s-íí mé-íí sak-ii

‘try’ ‘hurt’ ‘stink’ ‘scratch’

-də [-ɾə]: verb root ending in vowel cluster, long vowel or /g/

kwáíbɛ́ ɛ́bēghēg-

kwáí-rə̄ bɛ́ ɛ́-rə̄ bēg-rə̄ hēg-rə̄

‘borrow’ ‘call’ ‘broil’ ‘avoid’

-də [ndə]: verb root ending in a nasal stop

tāŋdondoŋ-

tāŋ-ndə̄ don-ndə̄ doŋ-ndə̄

‘eat ‘shut’ ‘seed’

-lə: verb root ending in [r]

nérwúrsēr-

nél-lə̄ wúl-lə̄ sēl-lə̄

‘disunite’ ‘ascend’ ‘be drunk’

The participle is a non-finite verb form and functions as a modifier – not a verb – and is thus part of the NP in examples (6) through (8). Even though the s­ tructure of RCs has not yet been explained in detail, it is worth mentioning here that the structure of these agentive type compounds is very similar to that of RCs. However, the main differences are that in RCs the relative marker kí/kə́ is obligatory and that that in an RC the participle can be either imperfective or perfective, while in these compounds, the participle is always imperfective. (6) nòr tāŋndə̄ náábì nòr tāŋ-ndə̄ man eat-ipfv.part “Fufu-eating man”

náábə̀ =ì fufu=ass

(7) nòr wəmndē jáámbì nòr wəm-ndə̄ person hunt-ipfv.part “Leopard-hunter”

jáámbə̀ =ì leopard=ass

10 Relative clauses in Wawa 



(8) nòr jaŋndə̄ bíínì nòr jaŋ-ndə̄ person dance-ipfv.part “Dancer”

 237

bíínə̄ =ì dance=ass

In examples (6), (7) and (8), the last constituent of the NP is marked for associative. This suffix will be more closely examined in section 2.3; it essentially marks the three words as an NP constituent. The imperfective participle is employed in two very different contexts. The first function is the afore-mentioned occurrence in agentive type NPs. Imperfective participles cannot occur in NPs modified with adjective-type meaning without featuring an object, as above in (6)–(8). This means that it is not possible to have the construction ‘eating man’ (without an object) as in ‘man who is eating’, without a relative marker, compare (9) and (10): (9) *nòr tāŋndīˋ nòr tāŋ-ndə̄ -ì man eat- ipfv.part-ass “Eating man”/ “man who is eating” (10) nòr kə́ tāŋndīˋ … nòr kə́ tāŋ-ndə̄ -ì man rel eat- ipfv.part-ass “The man, who is eating, …” In Wawa, most modifiers are deverbal adjectives, derived from inchoative verbs, though a small class of true adjectives exists, too. The deverbal adjectives must be in a gerundive form, marked by the suffix -ááŋgə̄ : (11) nòr tāŋ-ááŋgə̄

“man for eating” (tāŋááŋgə̄ usually means “edible”)

(12) bíínə̀ jaŋ-ááŋgə̄

“danceable dance”

(13) nòr éé-ŋgə̄

“leaving man”

The second context the participle form occurs in is in sentences with fronted constituents: whenever a constituent of a sentence is fronted with a focus cleft ­construction, the verb occurs in participle form – either imperfective or perfective. Note that the first example, (14), is very similar to an RC construction, while example (15) does not need a relative marker: (14) də̄ mə kí éllə̄ tìí. ér-lə̄ tìí də̄ mə kí cop 1sg rel go-ipfv.part home “It was me who was going home.”

238 

 Marieke Martin

(15) də̄ mūn mə dagáˉrī. də̄ mūn mə dagə́ -ā=rī cop child 1sg see.pfv-pfv.part=det “That is the child that I saw.” In both examples, (14) and (15), the subject is fronted by means of a focus cleft, or in other words, a copula-NP construction. The same sentence pattern can be observed when an adverb or question word is fronted to put it into focus, as in examples (16) and (17): (16) dààlí mə tʃuii. dààlí mə tʃu-ii now 1sg return-ipfv.part “It’s now I was returning.” (17) ŋgí kə́ wu nəmā ? ŋgí kə́ wu nəm-ā what rel 2sg do-pfv.part “What have you done?” It is possible to postpose the subject in a construction like (17) to “ŋgí kə́ nəmā wu?”, which shifts the focus onto the subject, while in (17) the focus is on the verb “to do”.

2.2.2 The perfective participle The perfective participle is derived with the suffix -ā on the verb root. The perfective participle is employed in several contexts: in relative clauses (18) and in narratives, where the participle marks anteriority (19) as well as in sentences with fronted constituents, as above in (14) and (15). (18) nòr kə́ tāŋā ɲàm éérə́. nòr [kə́ tāŋ-ā ɲàm] man [rel eat-pfv.part meat] “The man, who ate meat, left.”

ér-ə́ leave-pfv

(19) nòr tāŋā, éérə́. nòr tāŋ-ā ér-ə́ man eat-pfv.part leave-pfv “Having eaten, the man left.” Constructions as example (19) are very commonly found in narratives where the action of the previous sentence is repeated before moving on to the next action, e.g. example (19) could continue as the following:



10 Relative clauses in Wawa 

 239

(20) nòr ér-ā=rī éér-ə́ kwɛ́ ɛ̄ ér-ā man leave-pfv.part=det leave-pfv town.loc leave-pfv.part kwɛ́ ɛ̄=rī nòr gur-ə́ dáŋāī town.loc=det man buy-pfv horse “After leaving, the man went to town. Having gone to town, the man bought a horse.”

2.2.3 Verbs in negative sentences In negative relative clauses, verb morphology is identical to that of negative indicative clauses with finite verbs. This means that there is no negative form of the participle and participles and the negative marker ‘níī’ cannot co-occur in the same phrase. The verb forms used in the negative sentences below are not participial. Example (21) is a negative perfective main clause and sentence (22) is a negative perfective relative clause. Note that the difference in marking imperfective and perfective clauses is that the negative marker níī occurs before and as a suffix after the verb stem in imperfective clauses, and only as a suffix on the verb stem in perfective clauses. Double negatives thus do not make a positive in Wawa, as also example (35) demonstrates. (21) mū gùnníī ɲàm. mū gur-níī ɲàm 3sg buy-neg meat “He did not buy meat.” (22) mū, kə́ gùnníī ɲàmndī, éérə́ . mū kə́ gur-níī ɲàm=ndī 3sg rel buy-neg meat=det “He, who did not buy meat, left.”

ér-ə́ leave-pfv

(23) nàgbə̀ bə̀ mbə́ kí wu níī kāmjèníīrī də̄ dɔ̀ . nàg-bə̀ bə̀ mbə́ [kí wu níī kām-je-níī=rī ] cow-pl two [rel 2sg neg milk-inch-neg=det] “The two cows that you have not yet milked are here.”

də̄ dɔ̀ cop here

Also the negative copula can be the predicate of a relative clause, as in examples (24) and (25): (24) nòr kí mɔ̄ jú ̄g kɛ́ də̄ gūnnə̀ dɔ̀ . nòr [kí mū-ɔ̀ jú ̄g kɛ́ ] də̄ gún-nə̀ person [rel 3sg-poss house exist.neg] cop sleep-inf “The person who does not own a house is sleeping here.”

dɔ̀ here

240 

 Marieke Martin

(25) saì kí mə dɔ̀ kɛ́ -níī də̄ bíì. saì [kí mə dɔ̀ kɛ́ -níī ] time [rel 1sg here exist.neg-neg] “The time I am not here is bad.”

də̄ cop

bíì bad

2.2.4 Overview of verbal morphology in relative clauses To summarise the previous three sections on the imperfective and perfective participles and negative verbal morphology in RCs, Tab. 2 gives an overview of the differences in verbal morphology in perfective and imperfective, subject and object relative clauses in Wawa. This table only shows non-restrictive relative clauses, as the verbal morphology is the same for both restrictive and non-restrictive RC types. Tab. 2: Morphology in relative clauses. Aspect

Main clause inflection

rel phrase modifies

rel phrase inflection

pfv

mə tāŋ-ə́ ɲàm 1sg eat-pfv meat I ate meat.

Subject

mə, kə́ tāŋ-ā ɲàm,… 1sg rel eat-pfv.part meat I, who had eaten meat,…

pfv neg

mə tāŋ-níī ɲàm 1sg eat-neg meat I did not eat meat.

Subject

mə, kə́ tāŋ-níī ɲàm,… 1sg rel eat-neg meat I, who did not eat meat,…

pfv

mə tāŋ-ə́ ɲàm 1sg eat-pfv meat I ate meat.

Object

ɲàm, kə́ mə tāŋ-ā=rī,… meat rel 1sg eat-pfv.part=det The meat, that I ate, …

pfv neg

mə tāŋ-níī ɲàm 1SG eat-NEG meat I did not eat meat.

Object

ɲàm, kə́ mə tāŋ-níī=rī,… meat rel 1sg eat-neg=det The meat, that I had not eaten, …

pfv

mə tāŋ-ə́ 1sg eat-pfv I ate.

Subject, no object

mə, kə́ tāŋ-ā=rī,… 1sg rel eat-pfv.part=det I, who ate, …

pfv neg

mə tāŋ-níī 1sg eat-neg I did not eat.

Subject, no object

mə, kə́ tāŋ-níī=rī,… 1sg rel eat-neg=det I, who had not eaten, …

ipfv

mə də̄ tāŋ-nə̀ ɲàm 1sg cop eat-inf meat I am eating meat.

Subject

mə, kə́ tāŋ-ndə̄ ɲàm,… 1sg rel eat-ipfv.part meat I, who am eating meat, …

ipfv neg

mə níī tāŋ-níī ɲàm 1sg neg eat-neg meat I am not eating meat.

Subject

mə, kə́ níī tāŋ-níī ɲàm,… 1sg rel neg eat-neg meat I, who am not eating meat, …

10 Relative clauses in Wawa 



 241

Tab. 2 (continued) Aspect

Main clause inflection

rel phrase modifies

rel phrase inflection

ipfv

mə də̄ tāŋ-nə̀ . 1sg cop eat-inf I am eating.

Subject, no object

mə, kə́ tāŋ-ndə̄ =rī,… 1sg rel ipfv.part=det I, who am eating, …

ipfv neg

mə níī tāŋ-níī 1sg neg eat-neg I am not eating.

Subject, no object

mə, kə́ níī tāŋ-níī=rī,… 1sg rel neg eat-neg=det I, who am not eating, …

ipfv

mə də̄ tāŋ-nə̀ ɲàm. 1sg cop eat-inf meat I am eating meat.

Object

ɲàm, kə́ mə tāŋ-ndī ̀… meat rel 1sg eat-ipfv.part-ass The meat, that I am eating, …

ipfv neg

mə níī tāŋ-níī ɲàm 1sg neg eat-neg meat I am not eating meat.

Object

ɲàm, kə́ mə níī tāŋ-níī meat rel 1sg neg eat-neg The meat, that I am not eating, …

Relative clauses involving serial verb constructions and verb chains are constructed similarly to those involving single verbs. A verb chain in Wawa consists of one auxiliary verb and a main verb or serial verb construction. In a verb chain the auxiliary verb is inflected, while the main verb stands in infinitive form. Likewise, in relative clauses, only the auxiliary is derived for participle. Sentence (26) is an example of a verb chain, in which the auxiliary verb dó ‘want’ carries the perfective participle suffix -ā, and the serial verb gur-wur ‘sell-ascend’ following is in the infinitive form: (26) nàg kə́ mɔ̀ tá dó ̄ gùrwùnnə̀ júgə́. nàg [kə́ mə-ɔ̀ tá dó-ā gur-wur-nə̀ ] cow [rel 1sg-poss father want-pfv.part sell-dir-inf] “The cow, which my father wanted to go sell, fled.”

júg-ə́ run-pfv

A serial verb construction is a series of verbs of which only the last verb is marked for aspect; all preceding verbs appear as tone-modified stem-only forms2. Example (27) shows a sentence with the serial verb construction ‘fall-descend’ in which the second verb ‘descend’ is derived for the participle.

2 Verbs roots carry tone 1 (high), 2 or 3. In serial verb constructions, the stem of tone 1 verbs changes to tone 2 and the stem of tone 3 verbs changes to tone 4 (low). Tone 2 verbs do not change their tone; there is evidence that these verbs are a recent invention, especially as the tone 2 verb group mostly contains Fulfulde loan words.

242 

 Marieke Martin

(27) mūn kí gùsēārī bóírə́ . mūn [kí gu-sé-ā=rī ] child [rel fall-descend-pfv.part=det] “The child that fell down cried.”

bóí-rə́ cry-pfv

To summarise, verbal morphology in positive RCs varies from that in positive main clauses. The verb must be derived for either imperfective or perfective participle in an RC. The verbal morphology in negative RCs is the same as in negative main clauses. Verb chains and serial verb constructions also behave in the same way, meaning that in positive RCs a participle derivation is required, while in negative RCs the verbal morphology is not different to that of a main clause.

2.3 Associative The associative suffix plays an important role in compounding in Wawa. Its primary function is to create one constituent out of several words, but it can also appear on single words. Example (28) is similar to examples (6)–(8), where the associative is attached to the last constituent of a number of words to make them one NP. Therefore the author has marked the suffix with ‘=’ as its distributional pattern is not strictly that of a suffix but a clitic in this context. It is possible that the associative’s function here is similar to what Welmers (1973: 42) describes for Kpelle, however, the extent to which the associative nominalises the RC has not been assessed yet in Wawa. (28) nòr tāŋā náábì, (ndén) gúnə́ wə̄ m. nòr taŋ-ā náábə̀ =ì ndén gún-ə́ man eat-pfv.part fufu=ass then sleep-pfv “A man, having eaten fufu (then) slept” literally: “a having-eaten-fufu man (then) slept a sleep”

wə̄ m sleep

The associative is also employed in common compounds, such as animal and plant names, which are qualified for forest (kandaīnə̀ ‘bush, forest’) in contrast to village (tìí ‘compound/home.LOC’), e.g. túŋnī kandaīnì ‘forest cat, wild cat’ vs. túŋnī tìí (the locative stem of ‘compound/home’ doesn’t take the associative). The associative also occurs in possessed NPs, when the order of possessor and possessed NP is switched. This is illustrated in the following examples (29) to (32): (29) tâ tá tá-ɔ̀ tá father-poss father “Father’s father”

10 Relative clauses in Wawa 



 243

(30) tá táwòì tá tá-ɔ̀ =ì father father-poss=ass “Father of the father” (31) nāà mūn nāā-ɔ̀ mūn mother-poss child “Mother’s child” (32) mūn nāāwòì mun nāā-ɔ̀ =ì child mother-poss=ass “Child of the mother” It has not been able to establish that there is any difference between these two ways to express possession; speakers felt that they were the same. The associative is also used in possessive predicates: (33) dʒuraì də̄ mòì. dʒuraì də̄ mə-ɔ̀ =ì present cop 1sg-poss=ass “The present is mine.” (34) júg náwòì də̄ ndáì. júg ná-ɔ̀ =ì house 2dual-poss=ass “Our house is good.”

də̄ cop

ndáì good

In RCs, the last constituent of the clause is sometimes marked with the associative, as shown in (35). As in NPs, the associative only ever occurs at the end of the RC: (35) mə dōníī nòr kí níī Wàwà sɛ́ ì. nòr [kí Wàwà mə dō-níī níī 1sg like-neg person [rel neg Wawa “I do not like people who are not Wawa.”

sɛ́ =ì ] neg.cop=ass]

It has not been possible to establish whether the occurrence of the associative in RCs is linked to a certain structure, as the author has examples of the same structure with and without it. This may be due to language change or the endangered state and morphological loss apparent in the language today. The author’s personal impression was that sentences with the associative marker were regarded as ‘better Wawa’, but people used both constructions on a daily basis.

244 

 Marieke Martin

2.4 The relative determiner /dī/ The third constituent of the RC marking is the relative determiner /-dī/ (SG) or /-rə̄ bə̀ / (PL). /dī/ is added as a clitic to the last constituent of the RC – much like the associative marker, but the two do not co-occur. Its exact function could not be determined from the data, and its use is not obligatory. The author collected the same RC with and without the addition of /dī/ on different occasions. It occurs most commonly when the verb is the last constituent of the RC. The demonstrative dī functions in other contexts as a distal demonstrative and determiner, in contrast to the proximal dè. They appear as dè and dī when functioning as pronouns, e.g. as in dè də̄ bíì “this is bad” or dī də̄ ndáì “that is good”. When they function as determiners, they attach to the noun stem (SG or PL). Both dè and dī have different forms, depending on the last sound of the word they attach to. There is only one plural form of the suffix, as all plural markers end in a vowel. Tab. 3: Relative demonstrative /dı̄/. Form

Noun

-rè/-rī: follow- nàg ing vowels and /g/ tí -ndè/-ndī: following nasal stops -lè/-lì: after /d/

mūn ɲàm kàr ŋgá ̄r

Noun+DEM

Noun PL

Noun PL+DEM

Gloss

nàg-rè nàg-rī tí-rè tí-rī

nàg-bə̀

nàg-bə̀ -rə̀ -bə̀ nàg-bə̀ -rə̄ -bə̀ tí-mə̄ -rə̀ -bə̀ tí-mə̄ -rə̄ -bə̀

“cow”

mūn-ndè mūn-ndī ɲàm-ndè ɲàm-ndī kàr-lè [kàllè ] kàr-lī [kàllī ] ŋgá ̄r-lè [ŋgá ̄llè ] ŋgá ̄r-lī [ŋgá ̄llī ]

tí-mə̄ mūn-bə̄ ɲam-ərə̄ kàr-bə̀ ŋgár-bə̄

mūn-bə̄ -rə̀ bə̀ mūn-bə̄ -rə̄ bə̀ ɲam-ərə̄ -rə̀ bə̀ ɲam-ərə̄ -rə̄ bə̀ kàr-bə̀ -rə̀ -bə̀ kàr-bə̀ -rə̄ -bə̀ ŋgár-bə̄ -rə̀ -bə̀ ŋgár-bə̄ -rə̄ -bə̀

“forehead” “child” “meat” “calabash” “hand”

A summary of the phonological variants is shown in Tab. 3. The -dī may well be a feature of more languages within Mambiloid, as the closely related language Kwanja also takes a demonstrative in the relative clause (Joan Weber, p.c.).

2.5 Restrictive and non-restrictive relative clauses The most distinctive way by which relative clauses are marked in Wawa, is by the use of a relative marker. Wawa formally distinguishes restrictive and non-restrictive



10 Relative clauses in Wawa 

 245

relative clauses by the use of two different markers: kə́ (non-restrictive) and kí (restrictive). These markers are obligatory in a RC. Restrictive relatives can be defined as restricting “the event or situation that they predicate to the class of objects specified in the relative clause, whereas non-restrictive […] relatives do not.” Non-restrictive relatives “give additional information about the head [or antecendent] that is not necessary or essential in order to identify the latter” (Castillo 2003: 26). (36) nòr kí mə dagā éérə́ . nòr [kí mə dag-ā ] ér-ə́ person [rel 1sg see-pfv.part] leave-pfv “That person I saw (and only that specific one, I talked about earlier) left.” In (36), the use of the marker kí marks this sentence as a restrictive relative clause. It is important to the sentence that I saw the man before he left, whereas in example (37) the fact that I saw the man is just additional information and could be left out. (37) nòr kə́ mə dagā éérə́ . nòr [kə́ mə dag-ā ] person [rel 1sg see-pfv.part] “The person, who I saw, left.”

ér-ə́ leave-pfv

Both markers, kə́ and kí, have other functions in syntax, which will be briefly discussed in the following section. 2.5.1 Other functions of relative markers The markers kí and kə́ have other functions in syntax besides introducing relative clauses. Kí functions as an anaphoric demonstrative or determiner.3 The marker kə́ also introduces subordinate purposive clauses and can thus be called a ‘clause introducer’. Anaphoric demonstrative/determiner kí In relative clauses, kí introduces restrictive relative clauses. Its additional function is that of an anaphoric determiner with the meaning of ‘that same one’ or ‘the afore-mentioned’. The difference between the two functions, relative marker and anaphoric determiner, is syntactic and morphological. When kí functions as a relative marker, it cannot be marked for plural; thus the basic form is not changed. When kí functions as a discourse determiner, it must agree in number

3 All demonstratives in Wawa can also function as determiners when they are suffixed to the NP they modify. When they function as demonstratives, they stand alone.

246 

 Marieke Martin

with the NP it refers to. The interpretation of the function of kí in examples (38) to (41) must be made based on context. The main difference between these example sentences is that sentences marked with kí as det are main clauses, e.g. as in examples (39) and (41) while sentences with relative marker kí are NPs and cannot stand alone, as in examples (38) and (40): (38) Nòbə̀ kí éérə́ kwɛ́ ɛ̄ tʃuə́ . nòr-bə̀ [kí ér-ə́ kwɛ́ ɛ̄ ] tʃu-ə́ person-pl [rel go-pfv town.loc] return-pfv “Those people who went into town returned.” (39) nòbə̀ kə́bə́ éérə́ kwɛ́ ɛ̄ (tʃùkə́ ). nòr-bə̀ -kí-bə̀ ér-ə́ kwɛ́ ɛ̄ (tʃu-kə́ ) person-pl-det-pl go-pfv town.loc (return-cns) “The afore-mentioned people went into town (and returned).” If the NP kí refers to is singular, the difference between the two clauses is mostly in pronunciation, morphology and syntax: in sentence (40) ‘nòr kí’ is pronounced as two separate words with a short pause between them, in sentence (41) ‘nòrkí’ is pronounced as one word. The syntactic difference is that sentence (41) is a main clause and the last verb, ‘tʃukə́ ’, is optional, while sentence (40) is a main clause and a subordinate clause and therefore the verb ‘tʃuə́ ’ is obligatory. (40) nòr kí éérə́ kwɛ́ ɛ̄ tʃuə́ . nòr [kí ér-ə́ kwɛ́ ɛ̄ ] tʃu-ə́ person [rel go-pfv town.loc] return-pfv “That man who went into town returned.” (41) nòrkí éérə́ kwɛ́ ɛ̄ (tʃùkə́ ). nòr-kí ér-ə́ kwɛ́ ɛ̄ (tʃu-kə́ ) person-det go-pfv town.loc (return-cns) “The afore-mentioned man went into town (and returned).” In example (41) kí functions as anaphoric determiner and the verb ‘return’ is optional. If kí is added to the sentence, the verb is inflected for consecutive aspect to mark the actions as a sequence. Perfective aspect inflection is possible here grammatically, but speakers agree that consecutive aspect marking ‘sounds better’. The consecutive aspect is usually employed when there is a sequence of actions; in example (41) the consecutive aspect stresses that the actions of leaving and coming back are done in that order, in a sequence. A series of actions marked by consecutive aspect is less immediate than when marked with a serial verb (e.g. he left-returned); instead the focus is on the sequence of events, i.e. he left and then he returned.

10 Relative clauses in Wawa 



 247

Purposive subordinator kə́ In relative clauses, kə́ marks non-restrictive relative clauses. Its other syntactic function is that of a purposive subordinator. Unlike kí, kə́ never serves as a determiner. Its two subordinating functions are distinguished by word order and aspect marking: the purposive subordinator kə́ (FIN) can only be used in subordinate purposive clauses. The aspect marking possible is limited to the imperfective participle derivation plus the associative marker -ì in the imperfective aspect, exemplified in (43). In the perfective aspect, an inflection with reference to result or consequence (RRC) is used, as in sentence (42). (42) mə siə́ mə kə́ tāŋ. mə si-ə́ mə kə́ 1sg cook-pfv 1sg fin “I cooked in order to eat.”

tāŋ-∅ eat-rrc

In the imperfective aspect, only intonation and context can differentiate the meaning of a sentence, such as example (43) from a relative clause. If one pauses after the first part of the sentence, bū də̄ bā n ̀ ‘they are coming’, the second part bū kə́ tāŋndī ̀ could be understood as ‘they who are eating’, as a reiterated and further modified topic. Therefore, interpretation relies on context and intonation. (43) bū də̄ bā ̀n bū kə́ tāŋndī ̀. bū də̄ bā‐nə̀ bū 3pl.indf4 cop come-inf 3pl.indf “They are coming in order to eat.”

kə́ fin

tāŋ-ndə̄ -ì eat-ipfv.part-ass

The syntax of perfective relative and purposive clauses is different from each other. When kə́ functions as a subordinate subordinator, the structure of the subordinate purposive clause is {S (T) kə́ (Obj) V}, while the word order in a subject RC is {kə́ S V Obj}. The structural difference can be seen when comparing sentence (42) to ­sentence (44): (44) mə, kə́ siā, tāŋə́ . mə kə́ si-ā 1sg rel cook-pfv.part “I, who cooked, ate.”

tāŋ-ə́ eat-pfv

4 Wawa distinguishes between two 3pl pronouns, bū and bɔ́ kɔ̀ . Bɔ́ kɔ̀ is a lexicalised form of bū-kwòì (3pl.-dem.indf), which has come to mean the opposite. Bū refers to people not known to the speaker, while bɔ́ kɔ̀ refers to known ones. Therefore bū is glossed as 3pl.indf, in contrast to 3pl.def for bɔ́ kɔ̀ .

248 

 Marieke Martin

To summarise this section, Wawa employs two different relative markers that introduce the RC. There are two markers, kə́ and kí. Kə́ introduces non-restrictive RCs and kí marks restrictive RCs. These markers are obligatory in a RC. Each marker also has different functions in Wawa syntax. Kí functions as an anaphoric demonstrative in main clauses, and kə́ also functions as a clause introducer, marking subordinate purposive clauses. The two markers do not co-occur.

2.6 Position of relative clauses in Wawa syntax Relative Clauses always occur after the head of NP. If the head N is an N that contains several modifiers, the relative clause occurs after the last modifier constituent of this NP. Example (45) is a complex NP with a noun and two modifiers. There is no limit to the number of modifiers a noun can take. ­Sentences (46) and (47) are examples of relative clauses modifying a head NP with several modifiers. (45) júgbə̄ dʒə́ rbə̄ tāābə́ júg-bə̀ dʒə́ ̄r-bə̀ house-pl big-pl “Three big houses”

tāābə́ three

(46) jú ̄g dʒə́ ̄r kí guārī báŋkə́ rə́ . jú ̄g dʒə́ ̄r [kí gu-ā-rī] house big [rel fall-pfv.part-det] “The big house that fell is broken.”

báŋkə́ rə́ break-pfv

Example (47) shows a possessed head NP which is followed by several modifiers and a relative clause. Please note that while in both examples, (46) and (47), the glossing is “is/are broken”, the concept is expressed with a gerundive in example (47) and a perfective form in example (46). Wawa features a very limited set of true adjectives, using stative verbs instead, in this case “to be broken”. The meaning of both constructions is the same. (47) mɔ̀ tâ júgbə̄ dʒə́ rbə̄ tāābə́ kí gúkà ̄rəbə̀ də̄ báŋkáŋbə̄ . mə-ɔ̀ tā-ɔ̀ jú ̄g-bə̀ dʒə́ ̄r-bə̀ tāābə́ 1sg.poss father-poss house-pl big-pl three gú-kə̀ -ā=rə-bə̀ ] də̄ báŋkə-ááŋgə̄ -bə̀ fall-pl-pfv.part=det-pl] cop break-ger-pl “My father’s three big houses that fell are broken.”

[kí [rel



10 Relative clauses in Wawa 

 249

When the head NP is modified by a proximal (48) or distal (49) determiner suffix, the determiner stays on the noun, when the head NP is followed by a relative clause. In example (48) the determiner on the head NP is -rè, the proximal determiner. Example sentence (49) shows that determiners attach to the last constituent of the NP, here the adjective ‘small’, and that the determiner suffix -rī stays in place when the head NP is followed by a relative clause. (48) mə dō ̀n jú ̄grè kí wu dagrə̄ rī. mə dó-nə̀ jú ̄g-rè [kí 1sg want-inf house-det [rel “I want the house that you see.”

wu 2sg

(49) … mūn ŋgásùīrī kí jūgā jú ̄grī jūg-ā … mūn ŋgásùī-rī [kí … child small-det [rel run-pfv.part “… that small child that ran away” lit. “that small child that ran a flight”5

dag-rə̄ =rī ] see-ipfv.part=det]

jú ̄g=rī ] flight=det]

A relative clause can have a conjoint NP as head NP. In this case the relative clause modifies both NPs. (50) nəmbáábə̄ bá gwííbə̄ kí tāŋndə̄ ɲàm dʒéròì dīkū jók. nəmbá ̄r-bə̀ bá gwɛ́ ̄-bə̀ [kí tāŋ-ndə̄ ɲàm and woman-pl [rel eat-ipfv.part meat man-pl dʒér-ɔ̀ -ì ] dīkū6 jók goat-poss-ass] exist many “There are many men and women who eat goats’ meat.” The conjoined NPs must be connected by a conjunction, such as bá ‘and’ or a disjunction e.g. kó ‘or’. The previous sections have described in detail how RCs are marked in Wawa and to what extent the morphology of RCs differs to that of a main clause. The first constituent in an RC is a relative marker, kə́ (non-restrictive) or kí (restrictive). The verb in a positive RC must be derived for a participle form, either perfective or imperfective. In negative RCs, the verbal morphology does not differ from that of a main clause.

5 There are a number of verb noun compounds, often with shared etymologies, e.g. ‘to dance a dance’, ‘to walk a stroll’ etc. usually occurring with unaccuative verbs. 6 dīkū is an existential particle derived from the verb ‘to sit’ də- and the locative syntactic demonstrative/determiner kū.

250 

 Marieke Martin

3 Word order of relative clauses It is possible to modify any NP constituent of a clause with an RC in Wawa. The word order of subject RCs and object RCs is different and determines whether the head NP is understood as subject or object of the main clause. The different types of RCs will be discussed in the following sections, starting with subject RCs.

3.1 Subject-focused relative clauses The word order within a subject relative clause is 1. relative marker – 2. Verb – 3. object – (4. complements), as shown in examples (18)–(20). The relative clause includes a gap in the position of the subject, i.e. after the relative marker. The structure of a sentence including a subject relative clause is thus: {Head-NP [kə́ /kí s v-part Obj (X)] V (Obj) (IO) (X)} (51) nəmbá ̄r kí dagā mə tānnə̀ rī éérə́ . nəmbá ̄r [kí dag-ā mə tān-nə̀ =rī ] man [rel see-pfv.part 1sg market-loc=det] “The man who saw me in the market left.” (52) nàgbə̀ , kə́ érārī, dagə́ mə. nàg-bə̀ [kə́ ér-ā=rī ] cow-pl [rel leave-pfv.part=det] “The cows, which left, saw me.”

dag-ə́ see-pfv

ér-ə́ leave-pfv

mə 1sg

3.2 Object-focused relative clauses In object RCs, the RCs also follows the head NP. The RC includes a gap in the object position, after the verb. Within the RC the word order is 1. relative marker – 2. subject – 3. verb – (4. complements) and the structure of the whole clause is thus: {Head-np [kə́ /kí s v-part (X)] V (Obj) (IO) (X)} (53) nàgbə̀ kə́ mə ākə̀ níīrə̄ bə̀ éérə́ tānnə̀ . nàg-bə̀ kə́ mə akə-níī=rə̄ -bə̀ ér-ə́ tān-nə̀ cow-pl rel 1sg sell-neg=det-pl go-pfv market-loc “The cows, which I did not sell, went to the market.” The difference to a subject RC is the word order inside the RC. The following examples (54) and (55) are a minimal pair illustrating how the position of the participle

10 Relative clauses in Wawa 



 251

dagā and the pronoun ‘mə’ influences whether the head NP is interpreted as the subject or object of the RC. Subject RC: (54) nàgbə̀ , kə́ dagā mə, éérə́ . nàg-bə̀ [kə́ dag-ā mə] cow-pl [rel see-pfv.part 1sg] “The cows, which saw me, left.” Object RC: (55) nàgbə̀ , kə́ mə dagā, éérə́ . nàg-bə̀ [kə́ mə dag-ā ] cow-pl [rel 1sg see-pfv.part] “The cows, which I saw, left.”

ér-ə́ leave-pfv

ér-ə́ leave-pfv

3.3 Indirect object or complement of a preposition-focused RC In indirect object- or complement of a preposition-focused RCs, the structure of the relative clauses is more complex. Example (56) shows a relative clause modifying the complement of the preposition bá ‘with’ as head NP. Note that bá also functions as a conjunction meaning ‘and’ in other contexts. The head NP is followed by 1. relative marker, 2. subject, 3. verb, 4. bá and 5. the anaphoric demonstrative with determiner. The whole sentence structure is: {Head-NP [kə́ /kí NP V-PART PREP kí-DET X] V (Obj) (IO) (X)}. The indirect object in the RC is the demonstrative kí. The preposition bá introduces a barrier which forces resumption; i.e. the preposition cannot be left hanging at the end of the RC with the anaphoric demonstrative kí. (56) sə̂ r kə́ mə kwírə̄ bá kírī báŋkə́ rə́ . sə̂ r [kə́ mə kwí-rə̄ bá hoe [rel 1sg cultivate-ipfv.part with “The hoe, with which I cultivate, broke.”

kí=rī ] dem=det]

báŋkə-rə́ break-pfv

Note that the kí in this relative clause (56) functions as an anaphoric demonstrative, not a relative marker. See section 2.5.1 for other functions of relative markers for further detail. A similar structure to example (58) also applies when the complement is marked for locative. In this case the object position is taken by the locative counterpart of kí - kū, as shown in example (57). (57) ŋgùr kə́ mū daā no kū dàlì də̄ dʒə́ ̄r. ŋgùr [kə́ mū də-ā hole [rel 3sg stay-pfv.part “The hole he is in now, is big.”

no in

kū dem.loc

dàlì ] now]

də̄ cop

dʒə́ ̄r big

252 

 Marieke Martin

It is also possible to relativise an indirect object by deriving the verb for applicative7 (valency-increasing verbal extension). This can be seen in comparing the following two sentences, (58) and (59). In (58) the verb is derived for perfective participle and followed by the preposition bá and the anaphoric demonstrative kí. In example (59) the verb ‘to speak’ is derived for applicative and perfective participle. The preposition bá is not needed in this sentence, and its use would be ungrammatical, as the applicative has increased the valency of the verb ‘to speak’. The same structure is employed again in example (60). If the verb is derived for applicative, the structure is: 1. relative marker, 2. subject, 3. verb-APPL-PART, thus: {Head-NP [kə́ /kí NP V-APPL-PART (X)] V (Obj) (IO) (X)}. (58) nùgwɛ́ ̄ kə́ mə ɲə́ nā bá kírī éérə́ . nùgwɛ́ ̄ [kə́ mə ɲə́ n-ā woman [rel 1sg speak-pfv.part “The woman, who I spoke to, left.”

bá with

kí=rī ] dem=det]

(59) nùgwɛ́ ̄ kí mə ɲə́ nnāì éérə́ . nùgwɛ́ ̄ [kí mə ɲə́ n-nā-ā=ì ] woman [rel 1sg speak-appl-pfv.part=ass] “The woman I spoke to left.”

ér-ə́ leave-pfv

ér-ə́ leave-pfv

(60) sóŋ, kə́ bɔ́ kɔ̀ bānāì, gerə́ . sóŋ [kə́ bɔ́ kɔ̀ bā-nā=ì ] ge-rə́ honey [rel 3pl.def come-appl=ass] be.finished-pfv “The honey, which they came for, is finished.” Example (59) can actually be read in two ways: as a sentence including a relative clause or as one sentence with an [NP V] structure (61). The difference is only marked by intonantion. The glossing would also be different, namely: (61) nùgwɛ́ ̄kí mə ɲə́ nnāì éérə́ . nùgwɛ́ ̄-kí mə ɲə́ n-nā-ā=ì woman-det 1sg speak-appl-pfv.part=ass “The woman I spoke to left.”

ér-ə́ leave-pfv

In example (61) “the woman I spoke to” is one NP constituent, marked by the associative and followed by the verb. The kí functions as anaphoric determiner and refers the addressee back to a person that has been mentioned previously in the conversation. Example (60) could also be expressed with an alternative construction as seen in (58).

7 The applicative is marked with a verbal extension, -nā- (originally the verb ‘to give’), inserted between verb stem and aspect marking. It increases the valency of the verb.



10 Relative clauses in Wawa 

 253

3.4 Possessor relative clauses In possessor RCs, the relative clause introduces the possessor by means of a resumptive possessive pronoun that agrees with the head N in NP in number and gender. Possession is thus not marked on the head NP but on the resumptive pronoun only. The word order in possessor RCs is 1. relative marker – 2. resumptive possessive pronoun – 3. possessed NP – 4. Verb – (5. complements), and the whole sentence has a structure as follows: {Head-NP [kə́ /kí PRON NP V-PART (X)] V (Obj) (IO) (X)} The possessive marker suffix -ɔ̀ needs a constituent to attach to in the RC. Therefore the resumptive pronoun is an obligatory part of this kind of RC. In possessor RCs, the determiner on the verb seems to be obligatory; it was present in all collected examples. Example (62) shows the structure of a possessed NP in a main clause. Examples (63) and (64) are examples of possessor RCs: (62) nəmbá ̄rɔ̀ jú ̄g gurə́ . nəmbá ̄r-ɔ̀ jú ̄g gu-rə́ man-poss house fall-pfv “The man’s house fell.” (63) nəmbá ̄r kí mɔ̄ jú ̄g guārī də̄ dɔ̀. nəmbá ̄r [kí mū-ɔ̀ jú ̄g gu-ā=rī ] də̄ dɔ̀ man [rel 3sg-poss house fall-pfv.part=det] cop here “The man whose house fell is here.” (Lit: the man who his house fell…) (64) nòr kə́ mɔ̄ nàg kúārī éérə́ kwɛ́ ɛ̄ . nòr [kə́ mū-ɔ̀ nàg kú-ā=rī ] ér-ə́ kwɛ́ ɛ̄ person [rel 3sg-poss cow die-pfv.part=det] go-pfv town.loc “The person whose cow died went to town.” Examples (63) and (64) both show that in this case Wawa formally marks the gap in the relative clause by the use of a resumptive pronoun. This pronoun is obligatory and must be marked for possession, while the head noun does not feature the possession suffix /-ɔ̀ /.

4 Complex relative clauses 4.1 Coordinated relatives with a shared head A head N can be modified by more than one relative clause. When two relative clauses refer back to the same head NP, they are connected with a conjunction, e.g.

254 

 Marieke Martin

bá ‘and’, an adverb ndén ‘then’, or a disjunction, e.g. kó ‘or’. The relative clauses do not need to be complete clauses; for example (66) shows that the second relative clause consists of only a verb. The first relative clause must include a relative marker and NP. For the second clause, the relative marker and subject/object are optional. (65) mə kâ mə dō ̀n nòbə̀ [kí gúndə̄ dɔ̀ ] bá [étʃèrí ̄kwá sāŋî ]. mə kâ mə dó-nə̀ nòr-bə̀ [kí gún-ndə̄ 1sg foc 1sg like-inf person-pl [rel sleep-ipfv.part dɔ̀ ] bá [é-tʃè=rī=ì kwá sāŋí=ì ] here] and [leave-inch=det=ass next day=ass] “I like people who are sleeping here and who leave the next day.” (66) mə kâ mə dōníī nòbə̀ [kí lɔrkə̀ tī kúkə́ bə́ ] ndén8 [èrə̀ níīrī ]. mə kâ mə dó-níī [kí nòr-bə̀ 1sg foc 1sg like-neg person-pl [rel lɔr-kə̀ -tə̄ -īī kúkí-bə́ ] ndén [erə-níī=rī ] throw-pl9-caus-ipfv.part bone-pl] then [sweep-neg=det] “I do not like people who are throwing bones and then don’t sweep.”

4.2 Multiply embedded relative clauses It is possible to have multiply embedded RCs in Wawa. The first RC modifies the head NP and the following RC modifies the NP of the RC. Sentence (67) exemplifies how the second RC modifies the NP ‘child’ of the first RC. It is not possible to stack RCs referring back to the same head NP without the addition of the word ‘and’, as shown above in section 4.1. (67) nugwɛ́ ̄ kə́ túgā mūn kə́ bá mūn kɛ́ . nugwɛ́ ̄ [kə́ túg-ā woman [rel possess-pfv.part mūn]] kɛ́

[mūn [child

kə́ rel

túg-ā possess-pfv.part

child]] exist.NEG “The woman who has a child who has a child is not here.”

8 Probably a loan from Fulfulde 9 Pluractional derivation

10 Relative clauses in Wawa 



 255

4.3 Free relative clauses Free relative clauses, or headless relative clauses, are also commonly used in Wawa. They are formed the same way as other types. Free RCs form the first constituent in the main clause, as in example (68). It is also possible to start the main clause with a demonstrative, as shown in (69). Interestingly, free relative clauses can have either relative marker: (68) kí mə dagā bàì də̄ bíìǃ [kí mə dag-ā bàì ] də̄ bíì [dem 1sg see-pfv.part down.there] cop bad “What I saw down there is bad.” (This implies that the speaker has already explained what he saw.) (69) dè, kí mə dagā bàì də̄ bíìǃ dè [kí mə dag-ā dem [rel 1sg see-pfv.part “What I saw down there is bad.”

bàì ] down.there]

(70) kə́ nám tāŋārī níī ɲàm sɛ̄ níī. nám tāŋ-ā=rī ] [kə́ [rel 1pl.incl eat‐pfv.part=det] “What we ate was not meat.”

níī neg

də̄ cop

ɲàm meat

bíì bad

sɛ̄ cop.neg

níī neg

In sentences (68) and (69), the relative markers function as head NPs. In sentence (69) a demonstrative is used as head NP. Wawa speakers agree that the demonstrative is not obligatory and that sentences (68) and (69) are grammatically correct.

5 Discussion 5.1 Typology of Wawa relative clauses This section discusses the typological properties of Wawa RCs. The terminology is based on Givón (2001). As seen in the previous sections, all RCs make use of the relative markers kí and kə́ , thus following the ‘relative pronoun strategy’. These markers distinguish between restrictive and non-restrictive RCs. Kí is derived from an anaphoric demonstrative of the same shape, while kə́ is a clause introducer that also functions as a purposive subordinator in other contexts. Therefore, it could be argued that kí is a relative pronoun, while kə́ is not.

256 

 Marieke Martin

In addition, Wawa employs two more strategies to further distinguish relative clauses. For subject- and object-focused RCs (Figs. 1 and 2, respectively), the gap-strategy is used. This means that the position of the coreferential argument in the RC is empty. For possessor RCs (Fig. 3) and RCs relativizing the indirect object or complement of a preposition (Fig. 4) the anaphoric pronoun strategy is used: in the place of the missing coreferential argument a resumptive (possessor RC) or an anaphoric demonstrative pronoun (IO RC) is used. In the case of possessor RCs, the use of the resumptive pronoun is obligatory to provide a component for the possessive suffix to attach to, as it cannot attach to the REL marker. The anaphoric demonstrative in a RC relativising the complement of a preposition, the position of the coreferential argument must be filled with an anaphoric demonstrative, as the preposition cannot be left hanging without a complement. S VP

Subj [NP] N

V

Rel [S] Subj

VP V

Obj Fig. 1: Subject RC. “The man who saw me left.” (Example 51, simplified)

nǝmbá- r [Ø] [kí dagā mǝ] éérǝ´

S Subj [NP] N

VP

Rel [S] Subj [NP] N

nágbǝ´ [kǝ´ mǝ

V VP

V Obj [NP] dagā

Ø]

éérǝ´

Fig. 2: Object RC. “The cows, that I saw, left.” (Example 55)

10 Relative clauses in Wawa 



 257

S Subj [NP]

VP Rel [S]

N

Subj [NP]

VP

Det

N

V

nǝmbá- r [kí mɔ-

júg

guārī]

Cop

Adv

dǝ-

dɔ´

Fig. 3: Possessor RC. “The man whose house fell is here.” (Example 63)

S Subj [NP] N

VP V

Rel [S] Subj [NP]

VP V

PP Prep

sǝr̂ [kǝ´ mǝ kwírǝ

bá

NP kí] báŋkǝrǝ´

Fig. 4: RC modifying the complement of a preposition. “The hoe, with which I cultivate, is broken.” (Example 56)

5.2 Associative and relative determiner A relative clause can be marked by either a relative determiner or an associative marker, which both function as clitics to create one whole, i.e. one NP constituent, out of the whole sentence. These markers do not co-occur. The data suggests that neither is obligatory, and this may be down to language change: Wawa is in the process of losing verbal extensions in favour of Fulfulde adverbs, suffixes (e.g. the perfective is often marked by only tone), and constructions such as passive, which is considered archaic by Wawa speakers. Therefore it may be possible that sentences featuring neither marker are now considered grammatical, while twenty years ago, they would not have been.

258 

 Marieke Martin

At this point it is not clear what the difference between RCs marked by the relative determiner and those marked by the associative is. Evidentiality is not overtly marked in Wawa and the only grammatical device to mark certainty is the potential mood, which is used to mark an event as ‘possible’ versus ‘known’. The author considers it unlikely that the demonstrative/associative distinction is made on the basis of evidentiality. However, on the basis of her language intuition, she does not dismiss this theory. It is also possible that the relative determiner incorporates the associative suffix and its tone is deleted. It is not possible to determine this phonetically, as the vowel would not be lengthened following Wawa phonemic rules. The tones of suffixes generally do not get deleted. The author is, however, aware of some cases where this happens, one being the 3sg.poss form mɔ̄ , where the low tone suffix of the possessive marker -ɔ̀ is deleted. The author suspects that the determiner -di adds emphasis to the RC. Determiners are reduplicated frequently in the language to add emphasis to an NP, e.g. “that child” can be expressed as mun-ndī (child-det) but also as mun-ndī-rī (childdet-det), or even mun-ndī-rī-rī (child-det-det-det), adding more emphasis to “child” in each case. The associative’s main function is to create one complement out of a series of words (c.f. section 2.3), the determiner’s primary function in the RC may therefore well be to mark the RC with importance to the context. Unfortunately, there is currently not enough data available to assert this analysis.

5.3 Concluding remarks In this chapter the author has presented a diachronic account of the morphology and syntax of Wawa RCs. RCs require special verbal morphology in positive sentences, forcing verbs to be derived for either imperfective or perfective participles, while negative sentences show no difference in morphology. Wawa formally distinguishes restrictive and non-restrictive relative clauses by use of two different relative markers. It was shown that the restrictive marker kí also functions as anaphoric demonstrative and can thus be analysed as a relative pronoun, while the marker kə́ functions as a subordinating clause introducer in relative and purposive clauses. RCs are right-branching and the relative clause follows the head noun. This structure is very common across Niger Congo (Dryer 2008) and the number of relative clauses in one sentence is not restricted. Typologically, Wawa shows very common RC-marking strategies, in addition to the relative markers, subject and object RCs feature a gap, while possessor and indirect object RCs require resumptive pronouns to be grammatical.



10 Relative clauses in Wawa 

 259

References Castillo, Concha. 2003. The difference between restrictive and non-restrictive relative clauses. Atlantis. 25–38. Dryer, Matthew S. 2008. Order of relative clause and noun. In Martin Haspelmath, Matthew S. Dryer, David Gil & Bernard Comrie (eds.), The world atlas of language structures online, chapter 90. Munich: Max Planck Digital Library. http://wals.info/feature/90 (accessed 6 February 2015). Givón, Talmy. 2001. Syntax: An introduction. Volume I. Amsterdam/Philadelphia: John Benjamins. Guarisma, Gladys. 1987. Dialectométrie lexicale de quelques parlers bantoïdes non bantous du Cameroun. In Gladys Guarisma & Wilhelm J. G. Möhlig (eds.), La méthode dialectométrique appliquée aux langues africaines, 281–299. Berlin: Verlag von Dietrich Reimer. Lewis, M. Paul, Gary F. Simons & Charles D. Fennig (eds.). 2013. Ethnologue: Languages of the world, 17th edn. Dallas, Texas: SIL International. Online version: http://www.ethnologue. com (accessed 6 February 2015). Martin, Marieke. 2013. A Grammar of Wawa: An endangered language of Cameroon. Kent: University of Kent dissertation. Perrin, Mona & Hill Margaret. 1969. Mambila (parler d’Atta) : Description phonologique. Dallas, Texas: SIL International. Thwing, Rhonda A. 1987. The Vute noun phrase and the relationship between Vute and Bantu. Texas: University of Texas at Arlington MA thesis. Thwing, Rhonda A. 2004. Word order in Vute. Manuscript. SIL International. Thwing, Rhonda A. 2007. Verb extensions in Vute. Manuscript. SIL International. Thwing, Rhonda A. (n.d.). FLEX database of Vute lexicon. Dallas, Texas: SIL International. http:// www.sil.org/computing/fieldworks/FLEx/index.htm (accessed 15 November 2016). Weber, Joan. (n.d.). Kwanja grammar notes. Manuscript. SIL International. Weber, Joan & Martin Weber. 2008. Précis d’orthographe pour la langue kwanja (dialecte súndànɨ). Yaoundé : SIL Cameroon. Welmers, William E. 1973. African language structures. Los Angeles: University of California Press.

Denis Creissels

11 Conclusion By way of a conclusion to this volume, I would like to highlight its contribution to the general typology of relativization. Generally speaking, collections of papers investigating a particular phenomenon in a set of individual languages and written by language experts are extremely useful to typologists, since they provide reliable and detailed cross-linguistic data on the phenomenon in question in a format easily accessible to general linguists. The present volume does not contradict this general appreciation, quite the contrary. A considerable amount of books and articles dealing with the typology of relativization has been published since the publication of Keenan and Comrie’s seminal paper on Accessibility Hierarchy - Keenan and Comrie (1977). As regards more specifically Sub-Saharan Africa, Kuteva and Comrie (2006) put forward some generalizations about relative clause formation in African languages on the basis of a sample of 54 languages covering all major genetic families and including in particular two languages from the area covered by the present volume (Babungo and Ngemba). They distinguish between six relativization strategies: (i) relative pronoun, (ii) correlatives, (iii) internally-headed relatives, (iv) paratactic, (v) pronoun retention, and (vi) gap. In their sample, they identify only three of these six types, namely gap, pronoun retention, and correlative. The correlative type has been identified in Mande and Senufo languages only, and therefore it comes as no surprise that the only strategies found in the languages represented in this volume are the pronoun retention strategy and the gap strategy. However, Kuteva and Comrie also draw the attention to “the diversity that African languages exhibit with respect to the parsimony/abundance with which they encode relativization strategy”. In this respect, and in several other important aspects of the typology of relativization, the articles gathered in this volume provide very interesting data. Among the questions that must be addressed in a general typology of relativization, the case studies gathered in this volume provide particularly interesting data on the following points, which will be briefly commented below: –– resumptive pronouns and accessibility hierarchy, –– asymmetries between main and relative clause properties, –– possible types of relativizers, –– generalized relativization, or noun-modifying clause construction, –– emergence of head marking in the ‘noun + relative clause’ construction.

Denis Creissels: University Lumière (Lyon 2), denis.creissels@univ-lyon2 DOI 10.1515/9783110469547-011

262 

 Denis Creissels

Resumptive pronouns and accessibility hierarchy It has been observed that, in languages in which the resumptive pronoun strategy and the gap strategy are both available, the gap strategy is typically used for relativizing grammatical relations at the top of Accessibility Hierarchy, whereas the resumptive pronoun strategy tends to be preferred for grammatical relations lower than some middle point in the hierarchy. Among the languages described in this volume, Ejagham is a perfect illustration of this tendency. However, there are also in this volume languages that contradict it. For example, in Mungbam and Mundabli, resumptive pronouns are obligatory in subject relativization, but optional in the relativization of other grammatical relations.

Asymmetries between main and relative clause properties An important parameter in the typology of relativization strategies is the use of finite or non-finite verb forms in relative clauses, and it is generally assumed that the use of non-finite verb forms is common in pre-nominal relative clauses, whereas post-nominal relative clauses tend to make use of finite verb forms. The case studies included in this volume confirm that this generalization is problematic for Sub-Saharan languages, in which post-nominal relative clauses with non-finite verb forms (participles) are not rare, as illustrated in this volume by Wawa. Moreover, it is not uncommon to find among Sub-Saharan languages situations that cannot be characterized straightforwardly in terms of a binary contrast between finite an non-finite relativization strategies, and this observation is confirmed by several of the languages included in this volume. For example, in Wawa, the verb forms found in positive relative clauses are participles, whereas those in negative relative clauses do not differ from the negative indicative forms found in independent negative clauses. In other languages, the ambiguous status of relative clauses with respect to finiteness follows from the fact that the verb forms found in relative clauses can also be found in independent clauses, but some of the verb forms found in independent clauses cannot be used in relative clauses. In other words, a common situation among Sub-Saharan languages is that the verb forms used in relative clauses are not different from those found in independent clauses, but fewer inflectional categories are available in relative clauses. However, here again, the languages examined in this volume show that some generalizations need to be taken with caution. Languages in which focus-marking

11 Conclusion 

 263

constitutes an inflectional category of verbs are not rare in Sub-Saharan Africa, and it has been observed that, in such languages, focus-marking tends to be neutralized in relative clauses. However, this generalization does not hold for Mungban and Mundabli.

Possible types of relativizers In ‘noun + post-nominal relative clause’ constructions, it is particularly common that a relativizer is found between the head noun and the relative clause. However, several subtypes of relativizers must be distinguished. They may be invariable, or express categories such as gender or number. In some languages, they indicate a grammatical relation within the relative clause, which supports an ­extraction/ movement analysis independently of any particular theoretical framework. Moreover, the position between the head noun and the relative clause is not the only possible position for relativizers, and languages with post-nominal relative clauses may also have relativizers at the end of the relative clause, or attached to the verb of the relative clause. The languages described in this volume confirm that, as a rule, Sub-Saharan languages don’t have true relative pronouns (in the sense of relativizers expressing a grammatical relation within the relative clause), but in other respects, they illustrate a wide range of possibilities as regards the nature of relativizers and the positions they occupy: –– Most of the languages represented in this volume (Basaá, one of the Mungbam varieties, Ejagham, Nugunu, Medumba, Bafut, and Kenyang) have relativizers that precede the relative clause and express class agreement with the head noun. Most of the time, they have the same form as a demonstrative. Interestingly, in Nugunu, the relativizer is identical to a demonstrative with finite relative clauses, but with non-finite relative clauses (in which the verb is in the infinitive), it is identical to the linking element that introduces noun phrases in genitival modifier function. –– Mundabli has a relativizer agreeing with the head noun and identical to a demonstrative, which cannot be analyzed as having a function in the relative clause, but which in spite of that is located inside the relative clause, in postverbal position. –– Invariable relativizers preceding the relative clause are also found, either alone (Mungbam), in combination with a relativizer showing class agreement that also precedes the relative clause (Medumba and Bafut), or in combination with a variable relativizer in post-verbal position (Mundabli).

264 

 Denis Creissels

–– Vute and Wawa have a relativizer attached to the last word or the relative clause and identical to an enclitic that mark noun phrases in genitival modifier function. The postverbal relativizer of Mundabli is particularly intriguing, since the grammaticalization of demonstratives that immediately follow the head noun as relativizers is an extremely common phenomenon, but it is difficult to imagine how (or why) a relativizer created in this way could subsequently move to a position inside the relative clause.

Generalized relativization, or noun-modifying clause construction Comrie (1998) argued that, in some languages, the relative clause construction as it is commonly defined is not a formally distinct construction, but rather a particular case of a larger ‘noun-modifying clause construction’ that does not imply identifying the head noun with a given position in the construction of the relativized clause. Rather, the modified noun in the generalized ‘noun-modifying clause construction’ must be identified with an element of a scenario evoked by the modifying clause, without any syntactic constraint on the interpretation of its role in the event in question. For example, in such languages, sentences that could be rendered literally as ‘the sound that trees are falling’ are possible with the interpretation ‘the particular type of sound typically associated with situations that can be described by the sentence trees are falling’. Lovengren and Voll’s paper in this volume is to the best of my knowledge the first mention of an African language in which relative clauses as they are commonly defined are just a particular case of a larger noun-modifying clause construction in the sense of Comrie (1998). Interestingly, this chapter deals with two closely related languages, Mungbam and Mundabli, and it turns out that generalized relativization is productive in Mungbam, but rejected by speakers of Mundabli. Moreover, it is interesting to observe that in all other respects, the typological profile of Mungbam is not different from that of the other languages dealt with in this volume, and is very different from that of Japanese and the other languages in which generalized relativization has been signaled so far. This suggests that the existence of generalized relativization is probably not conditioned by a particular type of syntactic organization that would characterize the languages in which this type of construction is found.

11 Conclusion 

 265

Emergence of head marking in the ‘noun + relative clause’ construction Jenks, Makasso and Hyman’s chapter on Basaá and Van de Velde’s chapter on Eton deal with closely related languages. In both languages, a prefix í‑ called augment is found with nouns modified by a relative clause, with however a very interesting difference: according to Jenks, Makasso and Hyman’s description, this prefix is not obligatory in Basaá, and its presence vs. absence expresses ­definiteness distinctions, whereas according to Van de Velde’s description, in Eton, this prefix is an obligatory element of the ‘noun + relative clause’ ­construction, and consequently fulfills a purely syntactic function. Both situations, which can be analyzed as successive stages in the same grammaticalization process, are typologically interesting: –– According to Jenks, Makasso and Hyman, in Basaá, the use of this prefix to express definiteness distinction is limited to nouns modified by a relative clause. Typologically, definiteness distinctions conditioned by the presence of a given type of noun modifier are not unknown, but this constitutes a somewhat unusual phenomenon, and in Baltic and Slavic languages, which are the best-known cases of languages illustrating this kind of situation, the conditioning factor is the presence of an adjective. –– As regards Van de Velde’s analysis of Eton, head marking (or in other words, the use of a construct form of nouns) in the ‘noun + relative clause’ construction is rarely if ever mentioned in the literature: construct forms of nouns (i.e., noun forms signaling that the noun combines with a given type of modifier) are mainly found with nouns heading genitival ­constructions. To conclude, this volume constitutes a perfect illustration of the general principle according to which progress in our understanding of linguistic diversity is conditioned by the interaction between two complementary types of investigation: on the one hand, large scale typological investigations based on samples of languages as diverse as possible, and on the other hand, detailed investigations of a given phenomenon within the limits of sets of languages delimited according to genetic or areal criteria. Not surprisingly, the case studies that constitute this volume support many of the generalizations already put forward in the literature about relativization strategies in the languages of the world, or about relativization strategies in Sub-Saharan languages. However, they also suggest reconsidering some of these generalization in the light of the novel data they provide.

266 

 Denis Creissels

References Comrie, Bernard. 1998. Rethinking the typology of relative clauses. Language Design 1. 59–86. Keenan, Edward L. & Bernard Comrie. 1977. Noun phrase accessibility and universal grammar. Linguistic Inquiry 8. 63–99. Kuteva, Tania & Bernard Comrie. 2006. The typology of relative clause formation in African languages. In F. K. Erhard Voeltz (ed.), Studies in African linguistic typology, 209–228. Amsterdam/Philadelphia: John Benjamins.

References Abney, Steven P. 1987. The English noun phrase in its sentential aspect. Massachussetts: Massachussetts Institute of Technology dissertation. Aboh, Enoch O. 2004. The Morphosyntax of complement-head sequences: Clause structure and word order patterns in Kwa. New York/Oxford: Oxford University Press. Aboh, Enoch O. 2005. Deriving relative and factive clauses. In Laura Bruge, Giuliana Giusti, Nicola Munaro, Walter Schweikert & Guiseppina Turano (eds.), Contributions to the thirtieth incontro di grammatical, 265–285. Venezia: Libreria Editrice Cafoscarina. Aboh, Enoch O. 2006. Complementation in Saramaccan and Gungbe: The case of c-type modal particles. Natural Language and Linguistic Theory 24(1). 1–55. Aboh, Enoch O, Katharina Hartmann & Malte Zimmermann. 2007. Focus and grammar: The contribution of African languages. In Enoch O. Aboh, Katharina Hartmann & Malte Zimmermann (eds.), Focus strategies in African languages: The interaction of focus and grammar in Niger-Congo and Afro-Asiatic, 1–12. Berlin: Mouton de Gruyter. Achiri-Taboh, Blasius. 2004. Wh-extraction and resumption in relative clause constructions. In Evangelia Daskaliki, Napoleon Katsos, Marios Mavrogiorgo & Matthew Reeve (eds.), Proceedings of the University of Cambridge Second Postgraduate Conference in Language Research, 309–316. Cambridge: Cambridge Institute of Language Research. Achiri-Taboh, Blasius. 2006. A morpho-syntactic analysis of Moghamo-Ngamambo: The syntax of resumption. Frankfurt: University of Frankfurt-Main dissertation. Adamou, Evangelina & Denis Castaouec. 2010. Connective constructions in the world’s languages: A functionalist approach. La Linguistique 46(1). 43–80. Adger, David & Gillian Ramchand. 2005. Merge and move: Wh-dependencies revisited. Linguistic Inquiry 36. 161–193. Agbor Flavius Ntui of Ayoke, Cameroon. 1978. Etaghara Nkunku na Nkʉ́ í [Spider and Tortoise]. Transcribed by Patrick Etta Etta of Eyumojok. Ambadiang, Théophile. 1991. La oposición flexión-derivación en la morfología del nombre bantú, con especial referencia al Nugunu (bantú A62) [The opposition inflection-derivation in the morphology of Bantu nouns, with special reference to Nugunu (Bantu A62)]. Madrid: Universidad Complutense de Madrid dissertation. Andrews, Avery D. 2007. Relative clauses. In Timothy Shopen (ed.), Language typology and syntactic description. Volume II: Complex constructions, 206–236. Cambridge: Cambridge University Press. Anonymous. 1978? Egan Afughá-osí [The Story of Catching Fish]. Aoun, Joseph & Audrey Li Yen-hui. 1990. Minimal disjointness. Linguistics 28. 189–203. Ayamba Nkom Moses from Eyumojok, Cameroon. 1977. Otui Nkom Ekaŋ [The Chieftancy of Nkom Ekang]. Transcribed by Patrick Etta Etta of Eyumojok, Cameroon. Baker, Mark. 1985. The mirror principle and morphosyntactic explanation. Linguistic Inquiry 16. 373–416. Barlew, Jefferson & Emily Clem. 2014. The augment morpheme in Bulu. Paper presented at the 88th Linguistic Society of America Annual Meeting, Minneapolis, MN, 2–5 January. Bassong, Paul R. 2010. The structure of the left periphery in Basaá. Yaounde: University of Yaounde 1 MA Thesis.

DOI 10.1515/9783110469547-012

268 

 References

Batko, Ann. 2004. When bad grammar happens to good people: How to avoid common errors in English. New Jersey: The Career Press. Beltzung, Jean-Marc, Annie Rialland & Martial Embanga Aborobongui. 2010. Les relatives possessives en mbochi (C25). In Laura Downing, Annie Rialland, Jean-Marc Beltzung, Sophie Manus, Cédric Patin & Kristina Riedel (eds.), Papers from the Workshop on Bantu Relative Clauses 53, 7–31. Berlin: Zentrum für Allgemeine Sprachwissenschaft. Bernstein, Judy B. 1997. Demonstratives and reinforcers in Romance and Germanic languages. Lingua 102. 87–113. Bhatt, Rajesh. 2002. The raising analysis of relative clauses: Evidence from adjectival modification. Natural Language Semantics 10. 43–90. Bianchi, Valentina. 2000. The raising analysis of relative clauses: A reply to Borsley. Linguistic Inquiry 31. 123–140. Biloa, Edmond. 1990. Resumptive pronouns in Tuki. Studies in African Linguistics 21(2). 211–236. Biloa, Edmond. 1997. Functional categories and the syntax of focus in Tuki. Munich: Lincom Europa. Bird, Charles S & Mamadou Kante. 1976. An ka bamanankan kalan: Intermediate Bambara. Bloomington, Indiana: University Linguistics Club. Bitjaa Kody, Zachée D. 1990. Le système verbal du basaa. Yaoundé: Université de Yaoundé Thèse de 3ème Cycle. Blois (de), Kornelis F. 1970. The augment in the Bantu languages. Africana Linguistica 4. 85–165. Boeckx, Cedric. 2003. Islands and chains: Resumption as derivational residue. Amsterdam/ Philadelphia: John Benjamins. Borer, Hagit. 1984. Resumptive relatives in Modern Hebrew. Natural Language and Linguistic Theory 2. 219–260. Bot Ba Njock, Henri-Marcel. 1970. Nexus et nominaux en basaa. Paris: Université de ParisSorbonne dissertation. Breton, Roland & Bikia Fohtung. 1991. Atlas administratif des langues nationales du Cameroun. Paris: Agence de Coopération Culturelle et Technique (ACCT), Yaoundé: Centre Régional de Recherche et de Documentation sur les Traditions Orales et pour le Développement des Langues Africaines (CERDOTOLA), & Centre de Recherches et d’Études Anthropologiques (CREA). Brugè, Laura. 2002. The positions of demonstratives in the extended nominal projection. In Guglielmo Cinque (ed.), Functional structure in DP and IP: The cartography of syntactic structures, Volume I, 15–53. Oxford: Oxford University Press. Carstens, Vicki. 2005. Agree and EPP in Bantu. Natural Language & Linguistic Theory 23. 219–279. Castillo, Concha. 2003. The difference between restrictive and non-restrictive relative clauses. Atlantis. 25–38. Cheng, Lisa L.-S & Laura Downing. 2010. Locative relative clauses in Durban Zulu. In Laura Downing, Annie Rialland, Jean-Marc Beltzung, Sophie Manus, Cédric Patin & Kristina Riedel (eds.), Papers from the Workshop on Bantu Relative Clauses 53, 33–51. Berlin: Zentrum für Allgemeine Sprachwissenschaft. Cheng, Lisa L.-S. 2006. Decomposing Bantu relatives. In Christopher Davis, Amy R. Deal & Youri Zabbal (eds.), Proceedings of the 36th Annual Meeting of the North East Linguistic Society, 197–215. Amherst, MA: Graduate Linguistic Student Association. Chomsky, Noam & Howard Lasnik. 1977. Filters and control. Linguistic Inquiry 8. 425–504. Chomsky, Noam. 1977. On wh-movement. In Peter Cullicover, Adrian Akmajian & Thomas Wasow (eds.), Formal syntax, 71–133. New York: Academic Press.



References 

 269

Chomsky, Noam. 1981. Lectures on government and binding. Dordrecht: Foris. Chomsky, Noam. 1986. Knowledge of language. New York: Praeger. Chomsky, Noam. 1995. The minimalist program. Cambridge, MA: Massachusetts Institute of Technology Press. Citko, Barbara. 2002. (Anti)reconstruction effects in free relatives: A new argument against the Comp account. Linguistic Inquiry 33. 507–11. Collins, Peter & Carmella Hollo. 2000. English grammar: An introduction. London: Macmillan Press. Comrie, Bernard & Tania Kuteva. 2013. Relativization on subjects. Leipzig: Max Planck Institute for Evolutionary Anthropology. http://wals.info/chapter/122 (accessed 13 November 2016). Comrie, Bernard. 1998. Rethinking the typology of relative clauses. Language Design 1. 59–86. Couro, Ted & Margaret Langdon. 1975. Let’s talk ‘Lipay Aa: An introduction to the Mesa Grande Diegueño language. Banning & Ramona, California: Malki Museum Press & Ballena. Creissels, Denis, Gerrit J. Dimmendaal, Zygmunt Frajzyngier & Christa König. 2008. Africa as a morphosyntactic area. In Bernd Heine & Derek Nurse (eds.), A linguistic geography of Africa, 86–150. Cambridge: Cambridge University Press. Creissels, Denis. 2000. Typology. In Bernd Heine & Derek Nurse (eds.), African linguistics: An introduction, 231–258. Cambridge: Cambridge University Press. Creissels, Denis. 2009. The construct form of nouns in African languages. In Peter K. Austin, Oliver Bond, Monik Charette, David Nathan & Peter Sells (eds.), Proceedings of Conference on Language Documentation and Linguistic Theory 2, 73–82. London: School of Oriental and African Studies. Demuth, Katherine & Carolyn Harford. 1999. Verb raising and subject inversion in Bantu relatives. Journal of African Languages and Linguistics 20. 41–61. Den Dikken, Marcel. 2007. Amharic relatives and possessives: Definiteness, agreement, and the linker. Linguistic Inquiry 38. 302–320. Diessel, Holger. 1999. The morphosyntax of demonstratives in synchrony and diachrony. Linguistic Typology 3. 1–49. Dieu, Michel & Patrick Renaud (eds.). 1983. Atlas linguistique du Cameroun. Paris: Agence de Coopération Culturelle et Technique (ACCT), Yaoundé: Centre Régional de Recherche et de Documentation sur les Traditions Orales et pour le Développement des Langues Africaines (CERDOTOLA). Dimmendaal, Gerrit J. 1988. Aspects du basaá: Bantou zone A, Cameroun. Paris: Peeters/Selaf. Dixon, Robert M.W. 2010. Basic linguistic theory. Volume II: Grammatical topics. Oxford: Oxford University Press. Dooley, Robert A & Stephen H. Levinsohn. 2000. Analyzing discourse. Dallas, Texas: SIL International. Doussam, Jean. 1997. Hààmɗəə̀ ŋhȩ̀ ȩ́ Mutíí [raising and disciplining children]. Yoko: Centre de Littérature Vᵻ́tè. Downing, Laura & Al Mtenje. 2010. The prosody of relative clauses in Chewa. In Laura Downing, Annie Rialland, Jean-Marc Beltzung, Sophie Manus, Cédric Patin & Kristina Riedel (eds.), Papers from the Workshop on Bantu Relative Clauses 53, 53–67. Berlin: Zentrum für Allgemeine Sprachwissenschaft. Dryer, Matthew S. 2008. Order of relative clause and noun. In Martin Haspelmath, Matthew S. Dryer, David Gil & Bernard Comrie (eds.), The world atlas of language structures online, chapter 90. Munich: Max Planck Digital Library. http://wals.info/feature/90 (accessed 6 February 2015).

270 

 References

Du Plessis, Jacobus A. 2010. The relative clause in the African languages of South African (Bantu) languages. http://scholar.sun.ac.za/handle/10019.1/4884 (accessed 07 February 2012). Elderkin, Edward D. 2003. Herero. In Derek Nurse & Gérard Philippson (eds.), The Bantu languages, 581–608. London: Routledge. Giusti, Giuliana. 1997. The categorial status of determiners. In Liliane Haegeman (ed.), The new comparative syntax, 95–123. New York: Longman. Giusti, Giuliana. 2002. The functional structure of noun phrases: A bare phrase structure approach. In Guglielmo Cinque (ed.), Functional structure in DP and IP: The cartography of syntactic structures, Volume I, 54–91. Oxford: Oxford University Press. Givón, Talmy. 2001. Syntax: An introduction. Volume I. Amsterdam/Philadelphia: John Benjamins. Good, Jeff. 2010. Topic and focus fields in Naki. In Ines Fiedler & Anne Schwarz (eds.), The expression of information structure: A documentation of its diversity across Africa, 35–67. Amsterdam: John Benjamins. Good, Jeff. 2012. How to become a “Kwa” noun. Morphology 22. 293–335. Guarisma, Gladys. 1987. Dialectométrie lexicale de quelques parlers bantoïdes non bantous du Cameroun. In Gladys Guarisma & Wilhelm J. G. Möhlig (eds.), La méthode dialectométrique appliquée aux langues africaines, 281–299. Berlin: Verlag von Dietrich Reimer. Guthrie, Malcolm. 1953. The Bantu languages of Western Equatorial Africa. Oxford: Oxford University Press. Hammarström, Harald, Robert Forkel, Martin Haspelmath & Sebastian Bank. 2016. Glottolog 2.7. Jena: Max Planck Institute for the Science of Human History. http://glottolog.org (accessed 16 November 2016). Heine, Bernd & Tania Kuteva. 2002. World lexicon of grammaticalization. Cambridge: Cambridge University Press. Henderson, Brent. 2009. Anti-agreement: Locality of movement or agreement? Proceedings from the Annual Meeting of the Chicago Linguistics Society 45(1). 89–102. Hyman, Larry M & John R. Watters. 1984. Auxiliary focus. Studies in African Linguistics 15. 233–273. Hyman, Larry M. 2003. Basaá (A43). In Derek Nurse & Gérard Philippson (eds.), The Bantu languages. 257–282. New York: Routledge. Hyman, Larry M. 2004. How to become a “Kwa” verb. Journal of West African Languages 30. 69–88. Katamba, Francis. 2003. Bantu nominal morphology. In Derek Nurse & Gérard Philippson (eds.), The Bantu languages, 103–120. London: Routledge. Kayne, Richard. 1994. The Antisymmetry of syntax. Cambridge, MA: Massachusetts Institute of Technology Press. Keenan, Edward L & Bernard Comrie. 1977. Noun phrase accessibility and universal grammar. Linguistic Inquiry 8. 63–99. Keenan, Edward L & Bernard Comrie. 1979. Noun phrase accessibility revisited. Language 55. 649–664. Keenan, Edward L. 1985. Relative clauses. In Timothy Shopen (ed.), Language typology and syntactic description. Volume II: Complex constructions, 141–170. Cambridge: Cambridge University Press. Kisseberth, Charles & Abasheikh Mohammad Imam. 1974. Vowel length in ChiMwini – a case study of the role of grammar in phonology. In Anthony Bruck, Robert A. Fox & Michael W. La Galy (eds.), Papers from the Parasession on Natural Phonology, 193–209. Chicago: Chicago Linguistic Society.



References 

 271

Kisseberth, Charles. 2005. Accent and phrasing in Chimwiini. In Shigeki Kaji (ed.), Proceedings of the Symposium: Cross-linguistic Studies of Tonal Phenomena, 129–145. Tokyo: Research Institute for Languages and Cultures of Asia and Africa (ILCAA) Tokyo University of Foreign Studies. Kisseberth, Charles. 2010a. Optimality theory and the theory of phonological phrasing: The Chimwiini evidence. In Nomi Erteschik-Shir & Lisa Rochman (eds.), The sound patterns of syntax, 217–246. Oxford/New York: Oxford University Press. Kisseberth, Charles. 2010b. Phrasing and relative clauses in Chimwiini. In Laura Downing, Annie Rialland, Jean-Marc Beltzung, Sophie Manus, Cédric Patin & Kristina Riedel (eds.), Papers from the Workshop on Bantu Relative Clauses 53, 109–144. Berlin: Zentrum für Allgemeine Sprachwissenschaft. Koopman, Hilda & Anna Szabolszi. 2000. Verbal complexes. Cambridge, MA: Massachusetts Institute of Technology Press. Kouankem, Constantine. 2012. The syntax of the Medumba determiner phrase. Yaounde: University of Yaounde I dissertation. Kuteva, Tania & Bernard Comrie. 2006. The typology of relative clause formation in African languages. In F. K. Erhard Voeltz (ed.), Studies in African linguistic typology, 209–228. Amsterdam/Philadelphia: John Benjamins. Lehmann, Winfred P. 1976. Descriptive linguistics: An introduction. New York: Random House. Leitch, Myles F. (2003). Babole. In Derek Nurse & Gérard Philippson (eds.), The Bantu languages, 392–421. London: Routledge. Lemb, Pierre & François de Gastines. 1973. Dictionnaire basaá-français. Douala: Collège Libermann. Leu, Thomas. 2015. The architecture of determiners. Oxford: Oxford University Press. Lewis, M. Paul (ed.). 2009. Ethnologue: Languages of the world, 16th edn. Dallas, Texas: SIL International. Online version: http://www.ethnologue.com (accessed 12 January 2017). Lewis, M. Paul, Gary F. Simons & Charles D. Fennig (eds.). 2013. Ethnologue: Languages of the world, 17th edn. Dallas, Texas: SIL International. Online version: http://www.ethnologue.com (accessed 6 February 2015). Lewis, M. Paul, Gary F. Simons & Charles D. Fennig (eds.). 2016. Ethnologue: Languages of the world, 19th edn. Dallas, Texas: SIL International. Online version: http://www.ethnologue.com (accessed 17 December 2016). Longacre, Robert E. 1996. The grammar of discourse, 2nd edn. New York/London: Plenum Press. Longobardi, Giuseppe. 1994. Reference and proper names. Linguistic Inquiry 25. 609–66. Lyon, Christopher. 1999. Definiteness. Cambridge: Cambridge University Press. Maho, Jouni. 1999. A comparative study of Bantu noun classes. Göteburg: Acta Universitatis Gothoburgensis. Maho, Jouni. 2003. A classification of the Bantu languages: An update of Guthrie’s referential system. In Derek Nurse & Gérard Philippson (eds.), The Bantu languages, 639–651. London: Routledge. Makasso, Emmanuel-Moselly. 2010. Processus de relativisation en bàsàa: De la syntaxe à la prosodie. In Laura Downing, Annie Rialland, Jean-Marc Beltzung, Sophie Manus, Cédric Patin & Kristina Riedel (eds.), Papers from the Workshop on Bantu Relative Clauses (ZAS Papers in Linguistics 53), 145–158. Berlin: Zentrum für Allgemeine Sprachwissenschaft. Makasso, Emmanuel-Moselly. 2012. Metatony in Basaa. In Michael R. Marlo, Nikki B. Adams, Christopher R. Green, Michelle Morrison & Tristan M. Purvis (eds.), Selected Proceedings

272 

 References

of the 42nd Annual Conference on African Linguistics: African Languages in Context, 15–22. Somerville, MA: Cascadilla Proceedings Project. Martin, Marieke. 2013. A Grammar of Wawa: An endangered language of Cameroon. Kent: University of Kent dissertation. Maxey, James A. 1994. Relative clauses in Vute. Texas: University of Texas at Arlington MA thesis. McCloskey, James. 1979. Transformational syntax and model theoretic semantics: A case study of modern Irish. Dordrecht: Reidel. McCloskey, James. 1990. Resumptive pronouns, A’-binding and levels of representation in Irish. In Randall Hendrick (ed.), Syntax and semantics 23: The syntax and semantics of modern Celtic languages, 199–248. New York / San Diego: Academic Press. McCloskey, James. 2000. Quantifier float and wh-movement in an Irish English. Linguistic Inquiry 31. 57–84. McCloskey, James. 2002. Resumption, successive cyclicity, and the locality of operations. In Samuel D. Epstein & T. Daniel Seely (eds.), Derivation and explanation in the minimalist program, 184–226. Malden, MA: Blackwell. McCloskey, James. 2006. Resumption. In Martin Everaert & Henk van Riemsdijk (eds.), The Blackwell companion to syntax, 94–117. Malden, MA: Blackwell. Mchombo, Sam A. 2004. The syntax of Chichewa. Cambridge: Cambridge University Press. Meeussen, Achille E. 1967. Bantu grammatical reconstructions. Africana Linguistica III. 79–121. Menn, Lise & Brian MacWhinney. 1984. The Repeated morph constraint: Toward an explanation. Language 60. 519–541. Mfonyam, Joseph. 1989. Tone in orthography: The case of Bafut and related languages. Yaounde: University of Yaounde dissertation. Ngonyani, Deo S. 2001. Evidence for head raising in Kiswahili relative clauses. Studies in African Linguistics 30(1). 59–73. Ngonyani, Deo S. 2006. Resumptive pronominal clitics in Bantu languages. In Olaoba F. Arasanyin & Michael A. Pemberton (eds.), Selected Proceedings of the 36th Annual Conference on African Linguistics, 51–59. Somerville, CA: Cascadilla Proceedings Project. Nikolaeva, Irina. 2006. Relative clauses in spoken discourse. In Keith Brown (ed.), Encyclopedia of language and linguistics, Volume VII, 501–508. North Holland: Elsevier. Nsuka Nkutsi, François. 1982. Les structures fondamentales du relatif dans les langues bantoues. Tervuren: Royal Museum for Central Africa. Nurse, Derek & Gérard Philippson. 2003. Introduction. In Derek Nurse & Gérard Philippson (eds.), The Bantu languages, 1–12. London: Routledge. Ojong Tanyi (Ms) (female) from Ekoneman, Cameroon. Approx. 1977. Njombui na Mgbe [The Goat and the Leopard]. Transcribed by Patrick Etta Etta. Ojong Tanyi (Ms) from Eyumojok, Cameroon. Approx. 1977. Ayikobhi na Ntikpoŋ [Ayikobhi and Ntikpoŋ] Transcribed by Patrick Etta Etta of Eyumojok, Cameroon. Ojong Tanyi (Ms) from Eyumojpk. Asiraniran, Ekpimi Enop Akuk, Itəme Inop Mbak, and Opkere Onop Oseŋ [Arsiraniran, Bottle with Beautiful Edges, Cutlass with Beautiful Scabbard, Calabash with Beautiful Carving]. Ondo Mebiame, Pierre. 2001. L’augment en fang ntumu. Revue Gabonaise des Sciences du Langage 2. 60–77.



References 

 273

Onguene Essono, Louis-Marie. 2004. Syntaxe et fonctionnement de la relative et de l’interrogative en éwondo: Essai d’analyse de la subordination en bantou. Revue Internationale des Arts, Lettres et Sciences Sociales 1(1). 113–139. Patman, Frankie. 1991. Tone in the Nugunu verb phrase. In Stephen C. Anderson (ed.), Tone in five languages of Cameroon, 73–90. Arlington: Summer Institute of Linguistics & The University of Texas. Patman, Frankie. 1995. A functional account of syntactic similarities in Nugunu. Yaounde: University of Yaounde MA thesis. Paulian, Christiane. 1986. Les parlers yambassa du Cameroun (bantou A62). In Gladys Guarisma & Wilhelm J. Möhlig (eds.), La méthode dialectométrique appliquée aux langues africaines, 243–279. Berlin: Reimer Verlag. Payne, Thomas E. 1997. Describing morphosyntax: A guide for field linguists. Cambridge: Cambridge University Press. Perrin, Mona & Margaret Hill. 1969. Mambila (parler d’Atta) : Description phonologique. Dallas, Texas: SIL International. Peterson, Thomas H. 1974. On definite restrictive relatives in Moore. Journal of West African Languages 4(2). 71–78. Postal, Paul M. 1966. On so-called pronouns in English. In Francis Dinneen (ed.), Nineteenth monograph on language and linguistics. Washington DC: Georgetown University Press, [reprinted in David A. Reibel, and Sandford Schane (eds.). 1969. Modern studies in English. Englewood Cliffs, NJ: Prentice-Hall, 201–24]. Puèch, Gilbert. 1988. Augment et préfixe nominal en ngubi. Pholia 3. 247–256. Quirk, Randolph & Sidney Greenbaum. 1973. A grammar of contemporary English. Essex: Longman, Harlow. Redden, James E. 1979. A descriptive grammar of Ewondo. (Occasional Papers on Linguistics 4). Carbondale IL: Department of Linguistics, Southern Illinois University. Riedel, Kristina. 2010. Relative clauses in Haya. In Laura Downing, Annie Rialland, Jean-Marc Beltzung, Sophie Manus, Cédric Patin & Kristina Riedel (eds.), Papers from the Workshop on Bantu Relative Clauses (ZAS Papers in Linguistics 53). 211–225. Berlin: Zentrum für Allgemeine Sprachwissenschaft. Rijkhoff, Jan. 2002. The noun phrase. Oxford: Oxford University Press. Rijkhoff, Jan. 2008. Descriptive and discourse-referential modifiers in a layered model of the noun phrase. Linguistics 46(4). 789–829. Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman (ed.), Elements of grammar, 281–237. Dordrecht: Kluwer. Robinson, Clinton. 1983. Phonologie du gunu, parler yambassa. Paris: Société d’Études Linguistiques et Anthropologiques de France. Sag, Ivan. 1997. English relative clause constructions. Journal of Linguistics 33. 431–83. Schachter, Paul. 1973. Focus and relativization. Language 49. 19–46. Schneider-Zioga, Patricia. 2007. Anti-agreement, anti-locality and minimality: The syntax of dislocated subjects. Natural Language and Linguistic Theory 25. 403–446. Shlonsky, Ur. 1992. Resumptive pronouns as a last resort. Linguistic Inquiry 23. 443–468. Simango, S. Ron. 2006. Verb agreement and the syntax of ciNsenga relative clauses. Southern African Linguistics and Applied Language Studies 24(262). 277–290. Tallerman, Maggie. 1998. Understanding syntax. London: Arnold. Tamanji, Pius. 2009. A descriptive grammar of Bafut. Köln: Rüdiger Köppe Verlag.

274 

 References

Thwing, Rhonda A & John R. Watters. 1987. Focus in Vute. Journal of African Languages and Linguistics 9. 95–121. Thwing, Rhonda A. (n.d.). FLEX database of Vute lexicon. Dallas, Texas: SIL International. http:// www.sil.org/computing/fieldworks/FLEx/index.htm (accessed 15 November 2016). Thwing, Rhonda A. 1987. The Vute noun phrase and the relationship between Vute and Bantu. Texas: University of Texas at Arlington MA thesis. Thwing, Rhonda A. 2004. Word order in Vute. Manuscript. SIL International. Thwing, Rhonda A. 2007. Verb extensions in Vute. Manuscript. SIL International. Thwing, Rhonda A. 2008. Verbal Extensions in Vute. Paper presented at SIL Cameroon Academic Forum, May 2008. Thwing, Ronald & Rhonda A. Thwing. 1980. Deux contes en langue vɨ́tè : La mort par suite de buffle et Qui est le vrai voleur. Unpublished manuscript submitted to ONAREST, Yaounde. Van de Velde, Mark. 2005. The order of noun and demonstrative in Bantu. In Koen Bostoen & Jacky Maniacky (eds.), Studies in African comparative linguistics with special focus on Bantu and Mande, 425–441. Tervuren: Royal Museum for Central Africa. Van de Velde, Mark. 2008. A grammar of Eton. (Mouton Grammar Library 46). Berlin: Mouton de Gruyter. Van de Velde, Mark. 2010. The syntax of verb complements and the loss of the applicative in Eton (A71). In Karsten Légère & Christina Thornell (eds.), Bantu languages: Analyses, description and theory, 281–294. Cologne: Rüdiger Köppe Verlag. Van Hille, Marion. 1989. Eléments de description du syntagme nominal en puku, langue bantoue de zone A. Bruxelles: Université Libre de Bruxelles MA thesis. Van Houdt, Bettie. 1987. Eléments de description du leke, langue bantoue de zone C. Tervuren: Royal Museum for Central Africa. Vries (de), Mark. 2002. The syntax of relativization. Utrecht: Landelijke Onderzoekschool Taalwetenschap dissertation. Watters, John R. 1980. The Ejagham noun class system: Ekoid Bantu revisited. In Larry M. Hyman (ed.), Noun classes in the Grassfields Bantu borderland. Southern California Occasional Papers in Linguistics 8: 99–137. Los Angeles: University of Southern California. Watters, John R. 1981. A phonology and morphology of Ejagham. Los Angeles: University of California, Los Angeles dissertation. Watters, John R. 1989. Bantoid. In John T. Bendor-Samuel (ed.), Niger-Congo, 400–420. New York: University Press of America. Watters, John R. 2010. Focus and the Ejagham verb system. In Ines Fiedler & Anne Schwarz (eds.), The expression of information structure: A documentation of its diversity across Africa, 349–376. Amsterdam: John Benjamins. Watters, John R. 2012. Tone in Western Ejagham (Etung): Lexical tone on the “minimal” verb forms. In Matthias Brenzinger & Anne-Maria Fehn (eds.), Proceedings of the 6th World Congress of African Languages, 309–319. Cologne: Rüdiger Köppe Verlag. Weber, Joan & Martin Weber. 2008. Précis d’orthographe pour la langue kwanja (dialecte súndànɨ). Yaoundé : SIL Cameroon. Weber, Joan. (n.d.). Kwanja grammar notes. Manuscript. SIL International. Welmers, William E. 1973. African language structures. Los Angeles: University of California Press. Williamson, Kay & Roger Blench. 2000. Niger-Congo. In Bernd Heine and Derek Nurse (eds.), African Languages, 11–42. Cambridge: Cambridge University Press.



References 

 275

Yakoura, Valentin, Alfred Oumarou, Jean Doussam & James Maxey. 1998. Gèìn Vɨ́tè. Yoko : Centre de Littérature Vᵻ́tè. Zeller, Jochen. 2004. Relative clause formation in the Bantu languages of South Africa. Southern African Linguistics and Applied Language Studies 22(1–2). 75–93. Zwart, Jan-Wouter. 2000. A head raising analysis of relative clauses in Dutch. In Artemis Alexiadou, Paul Law, André Meinunger & Chris Wilder (eds.), The syntax of relative clauses, 349–85. Amsterdam: John Benjamins.

Index absence 56, 74, 76–78, 89, 91, 105, 125, 144, 147, 193, 197–198, 205–207, 216, 281 accessibility 31, 61, 68, 92, 102, 114, 124, 132, 159, 185–186, 191–192, 202, 208, 222, 224, 277–278, 282, 286 account 20, 30, 34, 45, 49–51, 54–55, 74, 83–85, 93, 100, 103, 131, 144, 162, 171, 176–178, 182, 274, 285, 289 action 15, 232, 238, 244–245, 254, 262 add 11, 45, 89, 227–229, 247, 274 addition 13, 23, 25, 30, 39, 48, 52, 116, 122–123, 138, 152, 155, 161, 179, 189, 210, 227, 230–231, 246, 249, 251, 260, 270, 272, 274 addressed 111, 135, 277 addressee 13, 226, 228, 232, 268 adjacent 24, 54, 56, 78, 213 adjectival 24–25, 48, 61, 137, 139, 142, 144, 149, 152, 284 adjective 15, 36, 41, 53, 58, 66, 80, 86–87, 89, 91, 93, 97, 109, 111, 137, 139, 149, 193, 209, 237, 250, 253, 264–265, 281 adjunct 55–57, 89–90, 92, 95, 97, 99–101, 138, 154–155, 200 adverb 11, 96–97, 105, 114, 127, 254, 270, 273 adverbial 17, 58, 66, 75, 91–92, 96, 100, 114, 144, 173–174, 210, 212–213, 225, 227, 234, 237, 247, 251 affix 22, 85, 137–138, 140, 143, 149 Africa 5, 17, 22, 31–32, 62, 81–82, 102–103, 159, 182, 223, 277, 279, 285–288, 290–291 African 5–6, 21, 27, 31, 61–62, 81–82, 102–103, 132–133, 159, 194, 203, 213, 218, 222–224, 247, 275, 277, 280, 282–291 agree 24–25, 27, 35, 128, 136, 139–140, 152, 164, 188, 195, 208, 210, 212, 215, 261–262, 269, 271 agreement 11, 15, 18–20, 24, 27, 30, 34, 40, 61–62, 65, 74–75, 77, 79, 98, 103, 110–111, 115–116, 120, 130–132, 140, 149–150, 153, 161, 166, 181, 189, 194–195, 211–212, 279, 285–286, 289 DOI 10.1515/9783110469547-013

aim 6, 83, 101 allophone 251 allow 30–31, 42, 46, 76, 81, 93, 107, 125, 129, 147, 154–155, 220, 246 alternate 54, 181, 234 alternation 48, 111, 113, 141, 161, 189, 212 alternative 48, 51, 56, 80, 165, 268 analogy 58, 139, 151 analyse 5, 20–21, 26, 33–34, 45, 47–49, 51–55, 57, 60–63, 73–74, 76–79, 83, 99, 131–133, 135, 138–139, 158, 161–162, 167, 177, 179, 181, 186, 199, 202, 208, 274, 279, 281, 283–284, 289, 291 anaphoric 11, 28, 30, 75, 80, 98, 123, 127–128, 144–146, 150, 229–231, 233, 236, 240, 261–262, 264, 267–268, 271–272, 274 answer 25, 142, 226, 229, 242 antecedent 84, 92–94, 101, 105, 109, 129, 161, 163, 166, 169, 171, 173–174, 178, 181 apply 26, 56, 71, 177, 192, 205, 267 approach 2, 61, 102, 283, 286 area 19, 24, 26–27, 29, 31, 187, 203, 205, 213, 221, 277, 285 argument 17–18, 21–22, 30, 33–35, 39, 45–46, 65, 85, 88–90, 92, 95, 97–98, 144, 161, 182, 185, 190–191, 199–200, 204–206, 215–216, 222, 272, 285 article 48–50, 63, 73, 75, 80, 141–142, 188, 277 aspect 5, 7, 11–15, 22, 24–25, 27, 29, 31, 60, 97, 105, 110, 113–114, 116, 122, 137, 141–143, 146–147, 163, 181, 189, 203, 208, 218–219, 225–226, 230–232, 234, 237–238, 250–251, 257, 262–263, 268, 277, 283 aspectual 105, 115–116, 118–119, 121, 189, 218, 251 assumption 34, 45, 51, 163, 167 asymmetry 22, 31, 84, 99–100, 197, 277 augment 9, 11, 20–21, 30–31, 39–45, 48–50, 56–61, 63–65, 67, 69, 71–81, 105, 120–121, 126–127, 131, 281, 283–284, 288 auxiliary 11–12, 70, 142, 220, 227, 232, 257

278 

 Index

bag 121–123, 157, 220 bantou 103, 133, 275, 286, 289 bantoue 81–82, 224, 288, 290 Bantu 5, 9, 17–23, 27, 31–34, 40–42, 48–49, 57–64, 73–77, 79–83, 86, 88, 97–98, 102–103, 106, 137, 141, 159, 161–163, 170, 175–176, 179–182, 191, 215, 221, 223–224, 226, 233, 245, 247, 275, 283–291 Basaá, 9, 17, 19–22, 30, 33–37, 39–49, 51–53, 55, 57–61, 106, 279, 281, 283, 286 basket 130–131, 220 bear 28, 99, 164, 166, 195, 199 behavior 30, 41, 43, 58, 128, 199, 203 benefactive 137–138, 154, 191, 200 beneficiary 67–68, 226, 228, 232 beside 95, 98, 173, 261 book 4, 37, 52–53, 55, 69, 87–89, 110–112, 116–118, 121, 124–125, 130, 195, 209, 227, 249, 277 boundary 140, 166, 233 break 11, 81, 117, 125, 129, 161–162, 175–179, 181, 217, 264, 267 Cameroon 5–7, 9, 17, 25, 29–30, 33–34, 60, 64, 80, 83, 103, 105, 110, 132, 135–136, 159, 162, 186, 188, 222–226, 247, 249, 275, 283, 288–290 Cameroonian 1, 3, 5, 19, 106, 161, 185, 188, 223, 243 case 18–19, 21–23, 25–26, 29–31, 35, 39, 42, 45–47, 50–51, 53–54, 56–59, 66, 68, 75, 78, 85–89, 91–95, 97–101, 105–106, 117, 125, 128, 136, 138, 140, 143–150, 153, 155–157, 165, 169–171, 174, 178–179, 181–183, 193, 196, 198, 201, 205, 209–210, 214–215, 218–219, 223, 251, 264–265, 267, 269, 272, 274, 277–278, 280–281, 283, 286, 288 category 27, 29, 45, 58, 102, 135, 137, 139, 144, 175, 185, 190, 203, 208, 224, 244, 278–279, 284 Central 5, 62, 81–82, 187–188, 223, 288, 290 change 34, 75–76, 189, 203–204, 213, 219, 223, 227, 231, 243–245, 257, 259, 273 characteristic 5, 35, 42, 59, 63, 65, 81, 92, 97, 101, 107, 111, 162, 201, 235, 249

circumstance 66, 147, 245 circumstantial 138, 234, 245 claim 48, 56, 58, 72, 155 classification 20, 26, 32, 64, 188, 225, 287 clause 5–7, 9, 13–15, 17–31, 33–39, 41–73, 75–81, 83–103, 105–111, 113–117, 119–125, 127–129, 131–133, 135–139, 141–145, 147–149, 151–157, 159, 161–163, 165–169, 171, 173, 175–183, 185–187, 189–247, 249–251, 253–275, 277–289, 291 cleft 11, 17, 22–23, 95, 97, 99, 102, 166, 197, 208, 253–254 clitic 80, 103, 164, 221, 249, 258, 260, 273, 288 close 6, 17, 25–26, 42, 63, 71, 77, 142, 188, 196, 243 code 22, 25, 28–29, 63, 73, 76, 186, 188 cognate 64–65, 80, 145, 189, 211, 214, 220 collection 29, 31, 277 com 4, 32–33, 61, 102, 105, 159, 185, 190, 197–198, 200–201, 204, 217–218, 224–225, 247, 249, 275, 287 combination 85, 94, 101, 189, 227, 279 combine 58, 93, 140, 227, 281 comitative 11, 155, 191, 198–200, 216 comment 135, 142, 161, 185–186, 192, 244, 249 comp 26, 43, 67, 70, 106–107, 144–148, 152, 211, 230, 236, 238 complement 19, 49, 53–54, 56–57, 66, 70, 75, 82, 86, 89, 91, 114, 163, 181, 199, 202, 230–231, 234–235, 240, 249, 266–267, 269, 272–274, 283, 290 complementizer 11–12, 24–25, 27, 36, 48–49, 55, 128, 131, 161–163, 165, 167, 176–177, 179, 181, 196, 238 complex 5–6, 46, 48, 55, 57, 59, 79, 83–84, 98, 102, 115, 127, 130, 145, 171, 178, 185, 189–190, 204, 210, 219, 221, 223, 264, 267 component 99, 139, 272 compound 252, 258, 265 compounding 66, 258 con 38, 67–69, 77–78, 86–87, 89, 96, 102, 229–230, 234, 236–240, 242–244, 246, 283

Index 

concept 6, 97, 116, 186, 202, 264 concord 11, 34, 40, 120–121, 137, 164–166, 170, 185, 189, 192–196, 209 condition 6, 57, 68, 76, 199, 216, 227 Congo 5, 7, 17–18, 33, 75, 81, 83, 102, 105, 110, 135, 159, 187–188, 223, 225, 274, 283, 290 conjunction 119, 210, 212, 265, 267, 269 connective 12, 29, 40, 66, 77–78, 86–87 consonant 11, 120, 136, 140, 165, 209, 227 constituent 12, 25, 45, 47–48, 64–65, 68, 70, 76–80, 95, 99–100, 108–109, 117, 122–123, 125, 130–131, 141, 143, 171, 185, 190, 193, 208, 225–226, 228, 234, 237–241, 244–245, 247, 249, 253–254, 258–260, 264–266, 268–269, 271, 273 constraint 26, 28, 52, 55–57, 62, 135, 157, 192, 280, 288 construction 9, 17, 20–21, 24, 27, 30, 63–69, 71, 73, 75–81, 83–84, 86–87, 94–95, 97–99, 101–102, 107–109, 113, 117–119, 124–125, 128–129, 132, 137–138, 142, 144, 147–148, 157–159, 161, 166, 168, 173, 179–181, 183, 185–187, 189–193, 195–197, 199, 201–205, 207, 209, 211, 213, 215, 217–224, 227, 229–230, 237, 242, 247, 253–254, 257–259, 264, 268, 273, 277, 279–281, 283, 286, 289 cont 24, 142–143, 147–148, 150, 152 contact 5, 17, 25–26, 185, 188, 213, 221 content 9–10, 22–23, 25, 50, 100, 142, 144–145 context 19, 21–22, 27–28, 30, 42, 57, 59–60, 63, 66, 75, 77, 97–98, 100, 116, 121, 123, 142–145, 147, 150, 156, 166, 168, 194, 208, 222, 229, 235, 241, 244–245, 253–254, 258, 260, 262–263, 267, 271, 274 contour 109, 113, 189 contradict 73, 277–278 contrast 24, 28, 34, 42, 44, 49, 51, 84–85, 87, 89, 91, 94–95, 98, 100–101, 109, 115, 119, 123, 126, 189, 203, 234, 258, 260, 263, 278 contrastive 40, 166, 190, 246 cop 28, 42, 73, 111–112, 116, 119–122, 124, 127, 129, 166–167, 197–198, 200, 204, 250, 253–257, 259, 263–264, 267, 269, 271

 279

copula 12, 111, 151–152, 197, 220, 250, 254–255 correlative 107, 125, 277 correspond 39, 93, 114, 173, 222 CP 12, 49–51, 53–54, 56–57, 99, 167 CV 40, 77, 105, 120, 131–132 data 4–5, 35, 45, 63, 78, 80, 87, 91, 93, 95, 97, 119–120, 132, 155–157, 161–162, 185, 218, 222, 249, 260, 273–274, 277, 281 database 5, 275, 290 declarative 34–35, 163, 165, 170, 204 def 25, 58–60, 163–169, 171–178, 180, 263, 268 definite 12, 26, 34, 39, 42, 45, 48–49, 57–60, 63, 73, 75–77, 80, 86, 111, 133, 163, 193, 209–211, 235, 289 definiteness 15, 20–21, 34–35, 41–42, 44, 57, 59–60, 71, 73, 75–77, 115, 161–163, 166–168, 176, 179, 181, 193, 210–211, 281 definition 64–65, 139 definitizer 12, 25–26, 123, 126 defocused 12, 225, 227, 230–232, 234, 237 demonstrate 35, 43, 46, 55, 78, 137, 150, 153, 255 demonstrative 9, 11–14, 19–23, 27–28, 30, 33–37, 39–65, 68, 71–80, 82, 85, 87, 91, 93–95, 97, 121–122, 137–139, 164–165, 193–194, 209–211, 213, 233, 236–237, 240, 245–246, 260–261, 264–265, 267–268, 271–272, 274, 279–280, 284–285, 290 dependency 37, 48, 60, 105, 283 dependent 12, 15, 39, 105, 148, 168, 191, 227, 230–232, 234, 237 derivation 49–51, 56, 66, 102, 108, 252, 258, 263, 270, 283 derive 48, 51, 65, 140 description 5–6, 18, 21–22, 26, 64, 78, 81–82, 84, 93, 102, 110, 132, 159, 161, 186, 222, 224, 247, 281, 283, 286, 290 Det 193, 209, 273 determiner 12, 15, 32, 45, 48, 59–61, 76, 78–79, 92–93, 95, 111–112, 115, 128, 133, 161, 163–165, 167, 174–175, 179, 181–182, 193, 198, 209–211, 213, 223,

280 

 Index

249, 260–263, 265, 267–269, 273–274, 286–287 dialect 15, 26–27, 30, 33, 35, 48–49, 76, 83–84, 136, 138, 185, 188–189, 192, 195–196, 214, 220, 249 diminutive 12 direct 25, 80, 89, 91–92, 97–98, 100–101, 109, 118–119, 124, 137–138, 162, 171–172, 179–180, 226, 228, 234 discontinuity 225, 240, 243–245, 247 discourse 9, 20–21, 26, 58, 67, 72, 76, 82, 99, 103, 123, 142, 158, 168, 225, 227, 229, 231, 233, 235, 237, 239–241, 243, 245, 247, 261, 285, 287–289 displacement 152, 204 dist 145, 217, 244 distal 12, 22, 27–28, 30, 43, 57, 139, 145, 209–211, 213, 260, 265 distinction 21–22, 28–29, 34, 40–41, 55, 77, 99, 105, 107, 109, 121–122, 139, 142, 149, 189–190, 193, 203, 206, 218–219, 222, 227, 249, 274, 281 distribution 17, 20, 27, 30, 33–34, 37, 39, 42–45, 48, 53–54, 56–58, 60, 70, 73, 86, 101, 161, 173, 179–181 diversity 5, 7, 17–18, 29, 31–32, 159, 223, 277, 281, 286, 290 effect 6, 43, 52, 57, 84, 87, 94, 99–101, 161, 176–177, 179–182, 207, 218, 285 enclitic 11, 13–14, 210, 221, 226–227, 231, 233–234, 280 encode 14, 28–29, 77, 95, 97, 119, 122–123, 125, 189, 191, 215, 227, 237, 242–243, 277 endangered 249, 259, 275, 288 Ewondo 64–65, 70, 73–74, 80, 82, 126, 167, 289 example 7, 18–30, 37–38, 40, 42–43, 46–47, 51, 53, 55, 58, 63, 65–75, 77–78, 85, 101, 106, 108–109, 111, 123, 127, 136–138, 140, 142–143, 145, 148, 150–151, 157–158, 162, 165–172, 176–180, 188, 191–194, 196, 199, 201, 203–205, 207, 209, 211, 216, 221, 227, 230, 232, 242, 250, 252–255, 257–259, 261–271, 278, 280 exist 29, 50, 78, 99, 115, 158, 201, 218, 242–243, 253, 255–256, 265, 270

expression 27, 31–32, 70, 100, 111, 142, 144, 149, 158–159, 163, 205, 212, 240, 242, 286, 290 extension 11–13, 15, 21, 245, 251, 268, 273, 275, 290 family 5, 33, 241, 277 Fang 64, 73–74, 76, 78, 187–188 feature 20, 28, 31, 51–52, 57–60, 73, 85, 94–95, 97–98, 100, 105, 111, 113–115, 127, 130–132, 135, 139, 179, 221, 249, 260, 264, 269, 274–275, 285 field 6, 15, 31, 76, 108, 159, 185, 192, 227, 247, 286, 289 final 12–14, 24, 26–29, 31, 59, 64, 71, 116, 135, 142, 149, 154, 166, 168, 171, 192–193, 202, 204–205, 208–209, 221, 225, 227, 231, 233–235, 238 find 20, 30–31, 39, 68, 75, 90, 107–108, 110–111, 139, 148, 163, 178, 186, 203, 210, 221, 278 finding 6, 39, 132 form 9, 12–15, 17–25, 27–28, 30–31, 33–34, 36, 39, 45, 47, 59, 63–67, 69–71, 73–81, 84–87, 89, 91, 93–101, 111, 113, 119, 122–124, 131, 135–143, 146–149, 151–154, 159, 163, 166, 169–173, 176, 178, 189–190, 193–196, 198, 203–204, 210, 212, 217, 220, 225–227, 230–234, 238, 243, 251–253, 255, 257, 260–261, 263–265, 271, 274, 278–279, 281, 285, 290 formation 25–26, 60, 62, 81, 83–85, 92, 94, 98–99, 102–103, 122, 133, 137, 146, 161–163, 200, 222, 277, 282, 287, 291 function 5, 20–25, 28–30, 41–42, 47, 70, 72–74, 83, 85–91, 93–94, 97–101, 109–111, 113, 121–123, 126, 131, 143–146, 152, 163, 167, 171–172, 191, 197, 200, 213–214, 216, 220, 225, 227–228, 232–235, 237–238, 242–247, 251–253, 258, 260–264, 267–268, 271, 273–274, 279–281 functional 48, 61, 66, 88, 93–95, 97, 99, 103, 114, 164, 286, 289 fut 89, 96, 189, 199–200, 210–211 future 12, 24, 29, 42, 60, 142, 189–190, 238

Index 

generalization 39, 99, 114–115, 181, 221, 277–279, 281 genitive 11–12, 28, 38–39, 44, 66, 68, 109, 112–113, 124, 127, 174, 200, 209, 216, 225–227, 230–231, 233–234 give 7, 19–20, 24, 26, 37, 83, 89, 92, 95, 100, 138, 143, 153, 155–157, 164, 166–168, 172, 186, 195, 199–200, 204–205, 212, 217–218, 228, 239, 241, 243, 250, 256, 261, 268 gloss 64–65, 189, 199, 222 group 5, 17, 20, 23–24, 27, 31, 33, 35, 42, 57, 59, 64, 77, 83–84, 110, 161–163, 165, 167, 180–181, 187–188, 194, 222, 257 hand 5, 30, 85, 93, 149, 168, 173, 207, 213–214, 217, 219, 222, 235, 260, 281 head 5, 13, 17–22, 24–25, 27, 29–30, 33–38, 45–51, 53–55, 57, 59–60, 62–66, 68, 73–74, 76–81, 84–93, 95, 97–101, 105–107, 115–116, 119–120, 122, 135–136, 139–140, 143–148, 150, 152–154, 156, 158, 164, 167–168, 171, 180–181, 186, 189, 191–198, 201–202, 204–205, 207–212, 214–217, 222, 225–230, 232–235, 237, 240–246, 250, 261, 264–267, 269–271, 274, 277, 279–281, 283, 288, 291 hierarchy 14, 21, 26, 30, 68, 92, 114, 124, 154–155, 222, 277–278 high 5, 11, 20, 24, 36, 38, 65, 70–71, 74, 78, 85, 113, 116, 122, 136, 140–141, 146, 149, 179, 191, 195, 209–210, 252, 257 history 75, 213 hold 6, 25, 46, 49, 93, 109, 141, 152, 279 host 51, 114, 221 house 53, 68, 72, 79, 99–100, 111–113, 117, 119–120, 124, 127–128, 150, 157–158, 177–178, 189, 195, 200, 202, 212, 218, 227, 231, 239, 246, 255, 259, 264–265, 269, 273 immediate 27, 238, 262 immediately 70, 80, 94, 131, 137–138, 168, 176, 190, 194, 204, 210, 235, 280 imp 71, 146–147, 228, 232, 234, 239 imperative 13, 138, 147 imperfect 13, 113–116, 122

 281

imperfective 13, 25, 141, 143, 147, 189, 203, 218, 226, 231–232, 234, 249–253, 255–256, 258, 263, 265, 274 impersonal 13, 130 indefinite 13, 42–46, 50, 59, 77, 111, 130, 193 independent 42, 79–81, 136, 226, 231–232, 278 indf 230, 234, 241, 246, 263 indicate 18, 47, 93, 97, 101, 114–115, 120, 122–123, 135, 138, 140, 143, 156, 158, 180, 191–192, 195, 214, 221, 225, 231–233, 245, 279 indirect 13, 30, 89, 91–92, 97–98, 100–101, 118–119, 124–125, 137–138, 172, 179, 226, 232, 249, 267–268, 272, 274 inf 21, 65, 67–68, 70–73, 144, 203, 230, 232, 234–236, 238, 240–242, 244, 246, 250, 255–257, 263, 265, 270 infinitive 13, 29, 64, 87, 137, 149, 232, 234–235, 238, 250, 257, 279 infix 137 information 4–5, 7, 23–24, 31–32, 40, 81, 87, 91, 93, 95, 100, 115, 122–123, 139, 142–143, 147, 153, 159, 168, 178, 186, 188, 193, 207, 223, 227–228, 231–232, 235–236, 243–247, 261, 286, 290 initial 6, 34, 58, 81, 132, 138, 165, 193, 210, 212, 214, 225–226, 228, 237, 244–246 innovation 59, 213, 221 instance 18, 22–23, 28–30, 40, 50, 58, 66, 84, 88, 95, 98, 114, 139 instrumental 30, 66, 90, 108, 162, 191 INT 13, 238 interest 6, 19, 45, 60, 162 interrogative 97, 133, 137–138, 142, 218, 289 ipfv 27, 190, 193, 197–199, 201, 210–211, 213–215, 217–218, 220, 230, 232, 236, 241–244, 246, 252–254, 256–257, 263, 265, 267, 270 irr 190, 193, 195, 199–201, 204–206, 230, 234, 236, 239–240 item 93, 95, 196 lack 24, 27–28, 40, 50, 56–57, 59, 92–93, 105, 116, 129, 221, 225 language 5–7, 9, 17–34, 40, 42, 48–49, 51–52, 57–64, 66, 73–84, 86, 88,

282 

 Index

97–98, 101–103, 105–111, 113–115, 117–118, 122, 129–133, 135–136, 139, 141, 154–155, 158, 161–163, 165, 167, 169–172, 174–183, 185–191, 193–197, 199, 201–203, 205, 207, 209, 211, 213–227, 230, 233, 240, 243, 245–247, 249, 251, 259–260, 273–275, 277–291 level 56, 59, 62, 113, 137, 182, 185, 189, 192, 220–221, 288 Libreria 181, 283 linguist 73, 159, 192, 247, 277, 289 linguistic 5–7, 21, 31, 62, 78, 82–84, 102–103, 105, 107, 132–133, 182–183, 223–224, 226, 249, 277, 281–282, 285, 287–290 live 66–67, 71, 100, 105, 136, 234–235, 245 loc 24, 37, 46, 48, 69–70, 118, 121, 126–127, 145–146, 150–151, 196–200, 202, 206, 215, 217, 220, 231, 235–236, 239, 241, 255, 262, 266–267, 269 location 24, 35, 146, 149–151, 178, 215, 225, 227, 237 locative 13, 24, 41, 66, 69, 88, 90–92, 119, 121, 125–126, 137, 144, 149–150, 152, 173, 198–200, 202, 215–216, 226–227, 232, 241, 244, 258, 265, 267 make 6, 19, 22, 26–30, 40, 42, 73, 78, 85, 91, 94–95, 101, 115–116, 131, 139, 142, 147, 161, 163–164, 168, 189–190, 201–204, 210–214, 217, 221, 229, 232, 240–241, 255, 258, 271, 278 marker 9, 11–13, 15, 17, 20–22, 25–30, 35, 40–42, 44, 57–60, 63–65, 67, 69, 71, 73, 75–81, 86, 89, 98–100, 105, 110, 113–117, 119–122, 127–129, 138–139, 144, 147, 161–169, 171, 174, 176–179, 181, 189, 195–196, 198–199, 203–204, 211, 213–216, 219, 221, 225–228, 230–231, 233–234, 237, 239–240, 243–247, 249, 252–253, 255, 259–274 meaning 11, 23, 25, 29, 58, 66, 75–76, 94, 138, 145, 196, 201, 232, 242, 253, 258, 261, 263–264, 267 modal 114, 181, 227, 242, 251, 283 modification 21, 24, 29, 57, 61, 63, 65, 73–74, 76–77, 88, 113, 117, 141, 171, 284

modifier 20–21, 24, 33–37, 39, 41–42, 45, 57–58, 68, 72, 75–82, 87, 110–111, 114, 143, 186, 189, 192–193, 207–209, 211, 227, 250, 252–253, 264, 279–281, 289 modify 7, 40, 51, 72, 80, 85, 105, 111, 115, 122, 139–140, 152, 162, 178, 209, 212, 233–234, 241, 256–257, 261, 265–266, 270 mood 13–15, 22, 114, 137, 147, 190, 203, 247, 251, 274 morpheme 12, 20, 25, 60, 64, 105–106, 108, 110–111, 113–116, 118–127, 129, 131–132, 140, 162, 195–196, 220, 222, 238, 283 morphology 18–19, 31, 35, 64, 75, 79, 98, 102, 110, 114, 159, 162, 189–191, 194, 223, 249–250, 255–256, 258, 262, 265, 274, 283, 286, 290 morphosyntax 61, 93, 110, 159, 247, 285, 289 mouth 143 movement 26, 34, 48–53, 55, 61, 91, 99–101, 161–162, 179–182, 279, 284, 286, 288 narrative 13, 130, 147, 226, 244, 254 negative 12–13, 31, 75, 158, 219–220, 226, 239, 251, 255–256, 258, 265, 274, 278 nom 146, 229, 236, 239, 246 nominal 13, 20–21, 24, 29, 31, 34–35, 37, 41–42, 49, 55, 61, 64–66, 68, 73–80, 82, 110–111, 113, 120, 131–132, 139–140, 143–144, 146, 163, 165, 180, 186, 191–195, 197–202, 204–212, 214–216, 219, 249, 278–279, 284, 286, 289–290 noun 5, 7, 11, 13–14, 18–27, 29–32, 34–44, 46–51, 53, 57, 59, 61–63, 65–66, 71–82, 84–95, 97–102, 105–111, 113–116, 119–120, 122, 125, 127, 130–131, 135–141, 143–150, 152–154, 156, 159, 163–164, 170, 181, 185–186, 189, 191–196, 199–202, 207–210, 212–215, 217, 222–223, 225–230, 232–245, 247, 250, 260, 264–265, 269, 274–275, 277, 279–281, 283, 285–287, 289–290 np 49, 77, 109, 119, 122, 153–155, 158, 226, 236, 238, 243, 246, 253, 258–259, 262, 265, 271 number 18, 24, 29, 40, 63–65, 70–72, 75–77, 83, 98, 114, 119–120, 128, 131, 135–137,

Index 

141, 156, 158, 162, 170, 185, 207, 249, 258, 261, 264–265, 269, 274, 279 numeral 36, 41, 137, 139, 193, 209 object 13–14, 18, 21, 26, 29–30, 35–40, 43–44, 52, 65, 75, 87–89, 91–92, 97–101, 107–109, 114, 117–119, 123–127, 137–138, 145, 153–154, 156–158, 162, 168, 170–173, 175, 179–180, 190, 198–200, 205–208, 210, 214–216, 223, 226, 228, 230, 232, 235, 238, 249–250, 253, 256–257, 261, 266–268, 270, 272, 274 oblique 14, 21, 26, 30, 37, 39, 91–92, 98, 128, 138, 157–158, 168–173, 175, 179, 223 observation 25, 45, 48, 87, 98, 100–101, 176, 203, 278 occasion 66, 260 occurrence 42, 116, 148, 161, 171, 175, 218, 231, 253, 259 operation 49–50, 62, 99, 110, 182, 288 operator 9, 15, 33–37, 39, 41–61, 99, 113 opposition 22–24, 74, 102, 283 option 50–51, 66, 79 order 19, 23, 29, 34, 50–51, 62, 76, 79, 81–82, 97, 100, 105–106, 110–111, 114–117, 119, 132, 136–137, 139, 141, 181, 186, 190, 192–193, 204–205, 208–209, 212, 219, 221, 225–228, 244, 250–251, 258, 261–263, 266, 269, 275, 283, 290 origin 63–64, 66, 75, 78, 80–81, 213 parameter 17, 214, 247, 278 part 20, 28, 31, 33, 45, 63, 79, 81, 84, 100–101, 138, 162, 176, 178, 186, 189, 203, 207, 221, 226, 231–232, 237, 246, 252–258, 261, 263–271 participant 11, 15, 100–101, 158, 168, 227, 230, 235, 243–247 participial 64, 87, 255 participle 14, 28, 87, 106, 249, 251–258, 263, 265–266, 268, 274, 278 particle 15, 113–114, 128, 162, 167, 179, 181, 190–192, 196, 212–213, 218–219, 221–222, 251, 265, 283 pattern 18–23, 31, 44, 57, 60, 65, 71, 74, 79, 83, 86, 94, 101, 110, 115–116, 124, 127,

 283

131–132, 181–182, 190, 210, 218, 226, 228, 254, 258, 283, 287 periphery 60, 93–94, 181, 183, 283, 289 phase 42, 75, 193 phrase 7, 12, 14–15, 18, 21, 24–26, 29–32, 34, 36–39, 41–44, 46–47, 49–51, 57, 59, 61, 77, 79–82, 85–90, 92–95, 97–103, 107–110, 115, 125, 127, 130, 132–133, 135–136, 139, 142, 146, 150, 154, 159, 162–163, 181–182, 185, 191, 193, 198–199, 202, 208–209, 212–213, 215–216, 224–226, 228, 232–233, 237, 239–240, 246–247, 250, 255–257, 275, 279–280, 282–283, 286–287, 289–290 place 23, 26, 30–31, 38, 49, 66–67, 71, 90–91, 96, 100–101, 148, 157, 166, 173–174, 192–193, 198–199, 209, 213, 215, 230, 237–241, 243–245, 252, 265, 272 plural 11, 14, 91, 97, 105, 111, 119, 131, 137–138, 145, 150, 227, 260–261 POL 14, 238 position 21, 26–27, 30, 34–38, 40–41, 45, 47–51, 53, 56–59, 61, 63–65, 68, 70, 73–75, 77–80, 85, 91, 93, 97–98, 101, 105, 107–110, 117, 120, 124, 130, 138, 146–147, 152, 161–162, 164, 166, 168–169, 171, 176–177, 179–181, 190, 204–206, 208, 211, 214, 219, 221, 228, 230–231, 233–234, 237, 240–241, 244–246, 266–267, 272, 279–280, 284 possess 33, 270 possession 68, 78, 98, 250, 259, 269 possessive 14–15, 21–22, 35–38, 41, 43, 55, 58, 66, 68, 72–73, 75–77, 89, 91–92, 97, 111–112, 119, 127–128, 132, 137–139, 143, 157, 181, 193–194, 209, 212, 232, 250–251, 259, 269, 272, 274, 284–285 possessor 21–22, 30, 38, 44, 66, 68, 70, 77, 89, 91–92, 156–157, 168, 174–175, 179, 200, 209, 216, 226, 230, 233, 249–251, 258, 269, 272, 274 possibility 19, 38, 80, 106, 144, 161–162, 179, 181, 190, 192, 194, 203–204, 206, 218, 237, 250, 279 postposition 191, 215, 226, 228 postverbal 21, 27, 70, 114, 190, 210–215, 219, 222, 279–280

284 

 Index

potential 6, 14, 25, 49, 57, 163, 237, 242–243, 251, 274 pragmatic 25, 46, 109, 142–143, 156, 205, 207–208, 232 predicate 14, 39, 75, 86, 109, 111, 232, 234, 255, 259, 261 prefix 19–20, 22, 34–36, 39–41, 43–45, 48, 57, 59–60, 63–65, 70–71, 74–75, 77, 111, 120, 131, 136–138, 140–141, 148–149, 151, 153, 156, 165, 189, 194–196, 209, 226, 281 prep 87–92, 94–96, 99, 195–196, 198–199, 202 preposition 14, 22, 34, 37–39, 65–69, 91, 98–99, 107–108, 113, 119, 121, 125–127, 138, 154–155, 172–173, 191, 199, 215, 226, 230, 238, 267–268, 272–273 present 14–15, 23, 25–26, 34, 39–40, 45, 47, 56, 59, 70, 72, 94, 110, 115, 126, 130, 136, 141, 147–148, 152, 156, 162, 171, 185–186, 188, 196–197, 199, 206–207, 209, 211, 221–222, 251, 259, 269, 277 prf 23, 106, 110, 113–131, 146, 150–152, 195–196, 200 primary 5, 20–21, 26–27, 30, 39, 63, 137, 153–154, 157–158, 258, 274 pro 26–27, 67, 129–130, 211 process 9, 57–58, 79–80, 83–85, 87–89, 91, 93–95, 97, 99, 101, 103, 108, 111, 185, 190, 204, 251, 273, 281 pronominal 22, 30, 64–66, 75, 91, 95, 97, 99, 103, 125, 138–139, 162, 164, 288 pronoun 5–7, 11–15, 17–18, 21–22, 25–28, 30, 33–39, 41, 47, 53, 55, 62–70, 72, 75–77, 79, 89, 91–93, 95, 97–102, 105, 108, 111, 117, 121–122, 125, 128, 135–140, 143–145, 149, 154–158, 161–165, 168–172, 175, 177, 179–183, 186, 190–191, 193, 197, 199, 207, 209–212, 214, 216, 220, 222, 225, 228–231, 233, 235, 237, 240–241, 245–246, 249–250, 260, 263, 267, 269, 271–272, 274, 277–279, 284, 288–289 proposal 49, 51, 53, 55–57, 223 prosody 83, 100–101, 161–162, 175, 181–182, 224, 285 proverb 130, 237

quantifier 14, 111–112, 158, 180, 233, 237 question 12, 14, 19, 22–23, 25, 29–30, 48, 51, 55, 67, 96–97, 102, 122–123, 125, 154, 170, 176, 226, 254, 277, 280 RC 5, 13, 15, 50–51, 89, 135, 139–148, 150, 152–154, 156, 158, 193, 196, 209, 221, 227–229, 233, 235, 237–240, 242–246, 249–250, 252–253, 256, 258–261, 263–267, 269–274 receive 47–48, 51, 53, 56–57, 168 recipient 137–138, 154, 191, 228–229, 232 reconstruction 53, 81, 161, 179–182, 285, 288 red 26, 85, 112, 116, 194, 236 reduplication 190, 204–207 ref 90, 99, 230–231, 233–236, 238, 245–246 reference 6, 15, 17, 35, 63, 71, 88–90, 94, 100–102, 130, 138, 145, 147, 163, 189, 207, 222–223, 263, 283 referent 5, 21, 64, 72, 76–77, 81, 95, 109, 135, 139, 143, 145, 152–153, 158, 201, 246 referential 11, 15, 23, 26, 32, 64, 75–76, 82, 85, 93, 97, 135, 143, 153, 158, 188, 230, 235, 244–245, 287, 289 reflex 39, 64, 70, 73–75, 80, 125 region 19, 25–26, 33, 84, 223 relation 24–25, 63–64, 66, 73–74, 76, 109, 135, 138–140, 142–143, 153–158, 186, 190–192, 200–201, 215, 278–279 relative 5–7, 9, 11, 14–15, 17–31, 33–39, 41–73, 75–81, 83–95, 97–103, 105–111, 113–125, 127, 129–133, 135, 137–141, 143–144, 152–154, 156, 161–169, 171, 173–183, 185–186, 190–194, 196–225, 227–238, 240–247, 249–250, 252–257, 260–275, 277–289, 291 relativisation 62–63, 66, 69–70, 103, 287 relativizer 15, 18–28, 30, 97, 105, 113, 130–131, 135, 137, 139–140, 143–152, 158, 185–186, 192, 194–197, 209–215, 218–219, 221, 225, 228, 277, 279–280 relator 15, 63, 65, 78, 80, 123–124, 127–128 remark 19, 27–28, 84, 101, 171, 246, 274 require 22, 27, 30, 38, 40, 43, 46, 49, 76, 78, 91, 94, 97, 100–101, 108, 115, 117, 125, 144, 146, 197, 215, 217, 220, 227, 274

Index  restriction 17, 31, 33–34, 49, 52, 58, 205, 216, 218, 222 restrictive 14, 17, 24, 28–29, 45–47, 51, 53–54, 63, 71–73, 77, 83, 87–88, 100–101, 109–110, 115, 133, 139, 142–143, 175–176, 178, 227, 235, 240, 247, 249, 256, 260–261, 263–265, 271, 274–275, 284, 289 resumption 33–35, 38, 43–45, 47, 52, 55–57, 154, 162, 168, 171–172, 175, 179, 181, 267, 283 resumptive 15, 17, 21–22, 26–28, 30, 33–34, 37–39, 43, 47, 54–55, 60, 63–66, 68–69, 89, 91–93, 97–101, 117, 125, 127, 135, 154–158, 161–162, 164, 169, 171, 173–174, 181, 186, 210, 216, 249, 269, 272, 274, 277–278 reveal 18–19, 23, 110, 132 right 6, 29, 36, 39, 54, 87, 94, 100, 105, 149, 162–163, 166–167, 181, 209, 233, 250, 274 role 18, 21, 24, 27, 29, 34, 64, 79–80, 84, 93, 111, 144, 162, 168, 181–182, 191, 201, 208, 210, 213–214, 217, 258, 280, 286 root 19, 111–112, 136, 140, 148–149, 151, 196, 226, 251–252, 254, 257 rp 25, 65, 164–165, 167, 169, 171–172, 174–176, 178 rule 99, 108, 122, 162, 222, 232, 274, 279 sample 277, 281 second 28, 37, 46, 51, 56–58, 65–66, 70, 72, 74, 77, 79, 81, 86, 138, 145, 149–150, 154–156, 171, 185, 190, 211, 213, 220, 222, 229, 231–232, 234, 240, 242, 245–246, 251, 253, 257, 263, 270 semantic 5, 22, 24–25, 30, 42–43, 48, 57–60, 62, 91, 97–98, 100–101, 109, 123–124, 139, 142–143, 145, 156, 158, 182, 196, 201, 210, 220, 222, 227, 232, 237, 288 sentence 11, 14, 22, 34–35, 42, 55, 70, 89, 94–95, 99, 101, 108–110, 116–117, 125, 132, 138, 153–154, 163, 166, 170–171, 192, 197–198, 201–202, 205, 208, 213–214, 223, 226, 229–231, 236–238, 240–241, 244–245, 249–251, 253–255, 257, 259, 261–263, 265–269, 271, 273–274, 280

 285

sequence 11, 86, 181, 209, 262, 283 shift 250, 254 signal 123, 131, 147, 228, 233, 246 similarity 17, 19, 26, 65, 84, 86, 93, 97, 99, 103, 161–162, 181, 186, 214, 220, 246, 289 situation 64, 74, 79–80, 84, 142, 145–146, 152, 173, 175, 203, 206–207, 213, 216–217, 222, 244, 261, 278, 280–281 slave 146, 150 Southern 62, 70, 82–83, 103, 136, 159, 223, 289–291 space 21, 72, 76–77, 100, 153, 203, 227, 237 speak 71, 268 speaker 5, 7, 12, 35–36, 40, 43, 63–65, 67, 69, 71–72, 83–84, 97, 118, 122–123, 128, 136, 142–143, 158, 162, 164, 166, 168, 185, 188, 197, 199, 226, 231, 247, 249, 259, 262–263, 271, 273, 280 specifier 47–50, 53, 99 specify 95, 101, 145, 148, 152 stem 24, 77, 149, 151, 189, 203, 209, 231, 252, 255, 257–258, 260, 268 step 49, 74, 80 strategy 5, 17, 19–21, 23, 26–29, 33–35, 37–39, 43–45, 47, 52, 54, 63–64, 66, 68–70, 72, 81, 86, 94, 97–98, 101–102, 105, 107, 110, 115–118, 132, 135, 153–158, 161–163, 165, 171, 180, 191, 210, 213, 219–220, 222–223, 225, 228, 246, 249, 271–272, 274, 277–278, 281, 283 strength 6, 237, 242–243 stress 171, 262 structure 5–6, 17–20, 23–24, 29, 31–32, 34, 46, 49–50, 54–57, 60–61, 79, 81, 83–88, 90, 93–97, 99, 101–102, 105, 110–111, 114–116, 118, 130–132, 135, 139, 144, 147, 158–159, 174, 181, 183, 188, 192, 201, 208, 213, 216, 223, 225–226, 228, 235, 242, 249–252, 259, 263, 266–269, 274–275, 283–286, 288–290 subgroup 5, 7, 17–18, 25, 77, 162 subject 7, 12, 15, 19, 21–22, 26, 29–31, 34–35, 39, 43, 47, 52–53, 55, 61–63, 68, 75, 81, 87–88, 91–92, 98, 105–106, 109, 113–114, 116–119, 121–122, 124–125, 129–130, 136–138, 146, 152–156,

286 

 Index

161–162, 164, 168–172, 176–181, 190, 197, 199–200, 204–207, 210, 214, 216, 219, 222–223, 226, 228–229, 231–232, 234–235, 241, 254, 256, 263, 266–268, 270, 272, 274, 278, 285, 289 subordinate 17, 22–23, 27, 64, 105, 139, 156, 163, 165, 217, 237, 240, 261–264 subordination 23, 133, 224, 289 subordinator 15, 25, 29, 140, 210, 212, 214, 263, 271 subsection 163, 166, 168–169, 172 subtype 48, 107, 279 suffix 86, 137, 141, 146, 150, 191, 226–227, 231–235, 245, 249–255, 257–258, 260, 265, 269, 272–274 suggest 17, 22, 29, 45, 48, 57–58, 73, 75, 83–84, 86, 93, 97, 100–101, 137, 176–177, 202, 213, 273, 280–281 summary 19, 39, 60, 84, 135, 147, 158, 196, 260 survey 220, 249 syntactic 20, 34, 37, 39, 42–43, 45, 47, 58, 61, 63–64, 72–73, 79, 81, 84, 88, 91, 97–98, 101–103, 109–110, 115, 124, 132, 135, 138, 156, 158–159, 161, 180–181, 201, 210, 222, 224, 227, 234, 246–247, 261–263, 265, 280–281, 283–284, 286, 289 syntax 5, 9, 32, 35, 56, 61–62, 82, 88, 97, 99–100, 102–103, 110, 114, 133, 142, 161–163, 165, 167, 169, 171, 173, 175, 177, 179, 181–183, 250, 261–264, 274, 283–284, 286–291 system 32, 34, 40, 98, 111, 137–138, 159, 185–186, 188–190, 194, 205, 209, 224, 226, 287, 290 TAM 15, 63, 70–71, 107 tense 14–15, 22, 27, 29, 36, 70–71, 81, 162–163, 171, 189, 203–204, 208, 218–219, 224, 238, 251 tone 11, 13, 19–20, 24, 34, 36, 38, 40–41, 64–65, 70–71, 74, 77–78, 85, 105, 113–117, 121–122, 128, 136–137, 140–141, 146, 149, 151, 159, 163–164, 166, 175, 185, 189, 191, 194–195, 199, 203, 209–210, 218, 231, 233–234, 252, 257, 273–274, 290

topic 6, 15, 19, 29, 31, 42, 47, 60, 102, 122–123, 158, 185, 227, 244, 247, 263, 285 trigger 72, 78, 178 type 5, 18, 25, 31, 36, 52, 58, 63, 66, 75–76, 78–79, 86, 88, 97, 99–101, 107, 115, 123, 125, 127, 129, 137, 147–150, 162, 164, 181, 186, 190, 192, 195–199, 203–204, 207, 214–215, 221–222, 227, 238, 244, 250, 252–253, 256, 266, 271, 277, 279–281, 283 typical 20, 30, 33–34, 60, 73, 83–84, 93, 95, 97–98, 101, 110, 203, 226, 250 typological 19, 60, 105, 135, 186, 214, 220, 247, 271, 280–281 typologically 25, 29, 52, 60, 63, 79, 81, 186, 216, 281 typology 5, 30, 81, 102, 107, 132–133, 159, 222–224, 247, 250, 277–278, 282–283, 285–287 universal 31, 61, 102, 132, 135, 158–159, 213, 223, 282, 286 utterance 67, 72, 92, 101 value 74, 190 variant 77, 79, 144, 199, 212, 260 variation 9, 11, 17, 19, 21–23, 25, 27, 29–31, 39, 42, 60, 71, 74, 84, 98, 100, 105, 135, 139–140, 147, 152, 158, 186, 214, 224 variety 21, 25, 27, 30, 64, 80, 163, 165, 170, 186, 188, 192–194, 196, 204, 213, 279 verb 11–13, 15, 17, 19, 21–25, 27–32, 34, 36–37, 39, 63–66, 70–71, 75, 81–82, 84–89, 91–101, 103, 113–114, 116–117, 121–122, 124, 129, 136–138, 141–143, 145, 147, 149, 152–153, 159, 163, 168, 189–190, 193, 197, 202–206, 208, 210, 212–213, 218–221, 223, 225–232, 234–235, 237–240, 245–246, 249–255, 257–258, 260, 262, 264–270, 274, 278–279, 286, 289–290 village 140, 146, 152, 173–174, 186, 188, 213, 225, 227, 236, 239, 244–245, 258 vowel 12, 69–70, 74, 78, 84, 86, 112, 116, 120–121, 131, 136, 165, 175, 185, 189, 191, 198–199, 209, 221, 227, 231, 233–234, 251–252, 260, 274

Index  word 19, 25, 27, 29–30, 34, 42, 50–51, 66, 74–76, 79, 81, 106, 109–110, 117, 123, 136–137, 141–142, 145, 150, 166, 171, 177, 181, 194–196, 199, 204–205,

 287

209, 214, 217, 219, 221, 226, 233–234, 237–239, 242, 250, 252–254, 257–258, 260, 262–263, 266, 269–270, 274, 278, 280–281, 283