SIL Electronic Book Reviews 2006-003
URL: http://www.sil.org/silebr/silebr2006-003
A Grammar of Tariana, from Northwest Amazonia
Cambridge: Cambridge University Press, 2003. Pp. 730. hardback $170.00, paper $70.00. ISBN 0521826640 (hardback), 0521028868 (paper).
Reviewed by Terry Malone
Tariana is a highly endangered language of the Arawak language family, spoken in the state of Amazonas, Brazil, in the Vaupés river basin. Only 100 of 1500 ethnic Tarianas still actively use their mother-tongue; the rest have shifted to Tucano (Tucanoan), the lingua franca of this region. The author, a professor at the Research Centre for Linguistic Typology at La Trobe University (Melbourne, Australia), has done considerable fieldwork in the Brazilian Vaupés region, including published work on the Arawak languages Baniwa, Bare and Warekena (Aikhenvald 1995, 1996, 1998) and unpublished work on Tucanoan languages spoken in the same region as Tariana.
The grammar is part of the series “Cambridge Grammatical Descriptions”, written in a typological framework, a version which the author, her coworker R.M.W. Dixon and their typological school term “basic linguistic theory”. The author states in her preface that “no attempt has been made to separate pure ‘description’ from theoretical interpretation” (p.xvii). This is indeed the case.
The table of contents is available on the internet1; all the chapter (numeric) subdivisions are reproduced in the printed table of contents (pp. vii-xiv), something which does not always happen in linguistic books and which I heartily appreciate (I intend to use this book, not just review it). The grammar is exhaustive: the author states that “it is far from being the last word on Tariana” (p. xviii), but the endangered state of the language and its thorough coverage in this volume suggest that it is likely to be the last word. Although the author has published a separate volume of annotated Tariana texts (Aikhenvald 1999), she includes five texts at the end of this grammar (pp. 630-670), and a vocabulary listing all Tariana words occurring in the grammar and the texts (pp. 671-680). There is an extensive bibliography (pp. 682-689) and an “index of authors, languages and subjects” (pp. 690-705).
The author begins with a “linguistic profile of Tariana” and then reviews the sociolinguistic situation of Tariana and its speakers in detail (pp.7-24). She provides a detailed evaluation of the competency of each speaker that contributed data to her analysis (pp. 18-24)—crucial when describing a highly endangered language, and something the ordinary field linguist might not think to include. An appendix (pp. 620-629) discusses “the main features of the Tariana dialects”, only three of which are still spoken—also a valuable feature of this grammar.
The chapter on phonology (2: pp. 25-65) includes several matters not always discussed in detail, including “the problem of diphthongs” (2.2), criteria for recognizing phonological words (2.4), detailed discussion of various classes of vowel fusion (2.5), and “H-metathesis”.2 The discussion of the phonological word touches on a thorny problem that most field linguists face: when is a morpheme a clitic, and when is it a suffix? Tariana does not much help the field linguist: clitics can occur between suffixes (see p. 83). The author is more helpful: she includes a chart of distinguishing criteria (p. 42), a detailed discussion of “prosodic classes of morphemes and their properties” (2.6: pp. 53-60) and a diagram of Tariana noun structure (p. 83). Among other things she observes that “some morphemes have “multiple membership”—that is, they can occur as (i) phrasal enclitics and as affixes, or (ii) as phrasal enclitics and as roots” (p. 58).3
Seven chapters (4, 5, 6, 7, 8, 9, 10: pp. 82-202) are devoted to the unusually complex noun and noun phrase morphology of Tariana; chapter 4 (“Nominal morphology and noun structure”) provides orientation and a general introduction to the other chapters. The language is unusual in several areas, including the interaction of a noun class system with a system of numeral and verbal classifiers—along with a system of possessive classifiers. Some categories are double-marked, including some noun classes, case (p.158ff), and number (176ff). One can also mark tense on nouns (p.183ff). The case system is nominative-accusative, but includes remnants of an older split ergative (or agent-patient) system typical of most other Arawak languages of the region.4 There is an abundance of productive derivational suffixes creating nouns which “unlike classifiers…never mark agreement when used as modifiers” (p. 197). Noun class markers function “as derivational suffixes on nouns” (p.200), and such nouns are not always easy to distinguish from compound nouns (p. 202).
The double marking of syntactic functions is more significant than might at first appear: the author observes that it has the effect of “mirroring syntactic structure in morphological structure and allowing the inflection (both case marker and noun class agreement markers) to specify the syntactic environments simultaneously at different levels” (p. 86). This is an intriguing remark, given current discussions concerning the interface between syntax and morphology; there is much data here relevant to this issue (and elsewhere in this grammar).
Chapter 11 discusses “closed word classes”. These include “personal pronouns” (11.1), “specifier articles” (11.2),5 “demonstratives” (11.3), “interrogative-distributive kwa-/kwe-” (‘which?, every’, 11.4), “gestural deictic khi” (11.5), “distributive individualiser napada” (11.6),6 “general indefinite pa:-” (11.7),7 “numerals” (11.8), “quantifiers” (11.9), “connectives” (11.10), and “adpositions” (11.11). The author devotes considerable space to this last category (pp. 222-233). “Morphologically, adpositions can be considered a subclass of nouns; however, etymologically they come from nouns, or from verbs…they are a closed class used as the head of an adpositional noun phrase” (p.222). They function as do prepositions and postpositions in other languages. Some take nominal possessive prefixes and some don’t; there are eight classes which are distinguished morphologically according to their etymological origin.
Eight chapters (12, 13, 14, 15, 16, 17, 18, 19: pp. 234-458) cover verb morphology; chapter 12 (“Verb classes and predicate structure”) provides orientation and a general introduction to the following chapters. Relics of a former agent-patient orientation appear in the verb classes: “most prefixed verbs are transitive, ambitransitive, and active intransitive, while most prefixless verbs are stative intransitive” (i.e. with patient subjects) (p. 234). To help the reader cope with the complex morphology, the author provides two useful diagrams: “Predicate structure in Tariana: Affixes and root” (p. 253); and “Predicate structure in Tariana: Enclitics” (p. 254).
Chapter 13 discusses “valency changing and argument rearranging mechanisms” (chapter 13). These include “passive” (13.1), “reflexive-reciprocal” (13.2), “increasing transitivity” (chiefly the causative) (13.3), an “argument-adding derivation” (13.4), and an “argument manipulation derivation” (13.5). The author alternatively terms the last mechanism “topic advancing derivation”: a suffix -ni occurs with active transitive and active intransitive verbs which “assigns some, but not all, subject properties to a non-subject NP, while still retaining some of the subject properties of the original subject. The author does not consider this “argument advancing mechanism” to be a passive: “Nominalised passives and nominalisations of the topic-advancing voice overlap in their semantics. There is an aspectual difference: a nominalised passive always refers to the result of a completed action, while a corresponding passive participle [i.e. nominalised topic advancing form] refers to the process” (p. 261).8 She includes the passive, but not this “topic advancing derivation” in her summary table on p. 286. She does not mention that the “topic advancing” mechanism could also be a relic of an older agent-patient system.
Remnants of the former agent patient system again rear their head with respect to morphological causatives: intransitive verbs with agent subjects can form morphological causatives, whereas verbs with patient subjects cannot (pp. 266-270); and “morphological causatives can be formed on stative (patient subject) verbs if the meaning of the verb presupposes a state changeable through the intervention of a causer” (p. 268). In contrast, most transitive verbs cannot form morphological causatives (pp. 270-272), but instead are “causativized with causative serial verb constructions” (p. 272). On verbs that do form morphological causatives a “double causative” (signaled by two separate suffixes) indicates a completely affected patient, “ completed and/or intensive action,” or “a definite or a topical patient”. Although the author does not comment, this phenomenon is also suggestive of a former agent-patient system.
Tense and evidentiality (chapter 14), aspect, Aktionsart and degree (chapter 15), mood and modality (chapter 16), and negation (chapter 17) are what one might expect as Amazonian and Arawak languages go: their morphological manifestations are complex, and fascinating. These chapters are thorough, comprehensive discussions of complex phenomena which rarely receive exhaustive treatment in the literature—and a model of description for any field linguist. Throughout these chapters, and indeed, throughout the whole grammar, the author is careful to specify which morphemes can cooccur and which do not cooccur—especially important in analyzing any verb system, and not always thoroughly considered in grammars.
Aikhenvald argues “in favor of the separate existence of ‘visual-based’ and ‘reported-based’ evidential subsystems in Tariana”; the reported-based evidential system “is likely to be inherited from Proto-Arawak” (p. 293). Those familiar with Tucanoan evidential systems will agree with the author’s conclusion that Tariana has calqued its visual-based evidential system from Tucanoan (p. 286; a useful summary table of “the semantics of evidentials” is found on p. 294).9 In her discussion of aspect, Aktionsart and degree Aikhenvald provides useful summary charts (pp. 325, 343, 349, 364) and a valuable discussion of “semantic and syntactic properties of Aktionsart enclitics” (pp. 363-367). Duration and completion are the two basic aspectual categories; the subcategories are all indicated by enclitics. Tariana is especially rich in Aktionsart enclitics. Aikhenvald divides them into five semantic classes, and provides examples showing the use of each one.
Moods discussed include imperatives (influenced by Tucanoan languages), “frustrative”, “intentional” (“marking imminent action…and intention”, p. 383), “apprehensive” (“meaning ‘lest, beware; or else’”, p. 384), “uncertainty”, “conditional”, “purposive” (“normally used to make the predicate of complement clauses; forms marked with purposive can also appear in the predicate slot”, p. 393),10 and “counter-expectation” (“used to mark unexpected action which goes against the speaker’s or the hearer’s expectations”, p. 396). The following classes of negation occur: “general clause and predicate negator”, differentiating between future and non-future predicates; “negative imperative”; “derivational and nominal negation”; “inherently negative verb stems”; “emphatic negative” -ne (p.400).
Chapters 18 “Serial verb construction and verb compounding” discusses Tariana’s complex and unusual system of serial verbs. “More than 70 per cent of clauses in texts and conversation contain serial verbs”, up to seven in a row (p. 424). One normally considers serial verbs and verb compounding to be closely related, the former grammaticalizing into compound verbs, but Aikhenvald argues that in Tariana the two “are independent grammatical processes, each with a grammaticalization and lexicalisation path of its own, and with distinct functions” (p.423).
She further argues that these are distinct from “complex predicates” (chapter 19): “Similar to serial verb constructions, complex predicates are monoclausal, and consist of two verbs in one predicate slot. Unlike serial verb constructions, they may contain an overt marker of syntactic dependency between the two verbs” (p. 449). Complex predicates are single grammatical words, whereas each serial verb is a separate grammatical word; verb compounding involves “verb root sequencing of the contiguous incorporating type” expressing the concepts ‘properly, really’, or ‘associated action’ (pp. 446-7). Complex predicates tend to express a potpourri of chiefly adverbial meanings, whereas serial verbs express aspectual, benefactive, modal, abilitative, causative and similar meanings.11
The author states that chapter 21 “Clause types and other syntactic issues” (pp. 475-514) provides “a background for Chapters 22-24, which deal with dependent clause types” (p. 475). However, Chapter 20 “Participles and nominalisations” (pp. 460–474) provides equally important background, bridging between the chapters that discuss verb morphology and those that concentrate on syntax. The complex interactions between participles (defined as “the predicate of a relative clause”, p. 460), nominalizations, clause types, predicate structures, subordinate clauses, and clause linking devices are another unusual typological distinctive of this language.12 Chapter 20 describes “participles” (20.1), nominalizations and “converbs” (20.2),13 and “other derivational devices” (20.3: these include noun class markers,14 and “topic advancing voice derivations”).
Chapter 21 discusses the structure of noun phrases (21.1) and their coordination (22.2), allowable candidates for predicate heads (21.3), types of clauses (21.4), and grammatical relations (21.5). The section on grammatical relations should be found in every reference grammar, and is one of the many strong points of this grammar: it presents “an overview of the defining properties of grammatical relations” (p. 507). It is of special interest in that once again the remnants of a former agent-patient system appear: transitive (A) and intransitive agent subjects (Sa) and intransitive agent patients (So) share certain properties; some properties distinguish A and Sa from So; some properties separate between all three classes of subjects; Sio verbs (verbs with a single argument behaving similarly to “dative subjects” in other languages) share some properties with all the other subject classes, or with subsets of them, but at the same time show a cluster of unique properties that set them off from the other three classes of subjects.15
At the end of section 20.2 one finds a section comparing “nominalisations and subordinate clauses” (20.2.5). Along with the introduction to the first section of chapter 22, and section 22.4 “subordinate clause and clause linking”, section 20.2.5 is crucial to comprehending the relationships between nominalization and the complex grammar explicated in chapters 22-24. Basically, one must keep in mind “that all subordinate clauses and complement clauses in Tariana can be considered as belonging to the subclass of nominals, since they can be case-marked” (p. 472). Further, “Tariana has a large number of sequencing enclitics which indicate relative tense”; clauses so marked “behave like nominal adjuncts in that they occupy the slot of an oblique argument of the predicate of the main clause” (p. 514). Nevertheless, other than case markers, “subordinate clauses cannot take any other nominal morphology, nor can they be modified or coordinated with NPs or nominalisations” functioning as NPs (p. 514). And last (but not least), “some clauses have a number of nominal properties and behave more like other oblique arguments, or adjuncts, of a clause, while others are less nominal-like…However, no matter how nominal-like subordinate clauses may appear to be, they are still different from nominalisations” (p.534).16
Chapter 22 (pp. 515-536) treats “Subordinate clauses and clause linking.” Some sequencing enclitics on subordinate clauses mark subject switch reference, and others do not (p. 516); these indicate prior or simultaneous relationships. There is a sequential –ka which does not indicate switch reference, marking “subordinate clauses meaning ‘after’, ‘when’, ‘while’ or ‘if’ ” (p. 528). And, “some adpositions can take sentential arguments” (p. 532). The author provides a useful summary table “Properties of subordinate clauses in Tariana” at the end of the chapter. Chapter 23 “Relative clauses” (pp. 537-546) begins with a table comparing main, relative, and subordinate clauses, and then their grammatical properties are discussed. Chapter 24 (547-560)”Complement clauses” describes the classes of complement clauses and “complementation strategies” (these “involve using grammatical mechanisms usually employed for other purposes to mark a clausal NP”, p. 552). The author notes in passing that “serial verb constructions in Tariana often correspond to complement clauses in other languages” (p. 552).
The author includes a chapter on discourse structure (“25 Discourse organization”). Although the chapter suffers somewhat from lack of a clearly defined theory of text structure,17 the author must be commended for having included it at all. The first section (25.1 “Pragmatic basis for constituent order”) reveals her motivation for including this chapter, and is its heart. The chapter basically describes what controls word order—which happen to be pragmatic factors (as is typical of most languages of the Amazonian Basin, no matter what the linguistic affiliation).
One last chapter (26: “Issues in etymology and semantics”, pp. 594-619) discusses a variety of issues which fit nowhere else in the grammar, including Arawak characteristics of Tariana, borrowings from Tucanoan, ‘same’ and ‘different’, generic vs. specific terms, positional verbs (Aikhenvald concludes these are not “classificatory positional verbs”, p.608), and a few remarks on lexical semantics.
Some editorial inconsistencies occur in this book. I noticed some typos (not excessive for a book of this length and complexity).18 Considerable allophonic and other kinds of variation occurs in Tariana, and “most of these variants depend on the age and proficiency of the speaker;” Thus “every example in this grammar records the actual pronunciation by the consultant” (p xxi). Occasionally this policy results in inconsistencies, for example on p. 241 (the same Tariana verb root is spelled two different ways in the text below example 12.29). There are some untranslated Portuguese glosses and terms (see pp. 7, 220, 273). In the preface the author states that “when no English equivalent is readily available, a Portuguese word is used, e.g. abiu-fruit or mucura-rat” in glosses (p. xviii); unfortunately the general term (‘fruit’, ‘rat’ or the equivalent) in English is occasionally omitted, and one is left guessing what the Portuguese term might refer to.
With respect to the exposition, there are occasional instances where definitions or explanations of terminology do not occur when terms are first mentioned (p. 22; pp. 244-5, where the term should have been explained on p. 234). Fortunately, the grammar is comprehensive enough that linguistically competent readers can almost always figure these matters out for themselves, and in the majority of cases the author does provide discussion and explanation.
These inconsistencies are slight scratches on the face of a masterly piece of work. The exposition is rich in summaries, usually in the form of lists, tables, and/or diagrams. This is helpful beyond words, in the course of digesting this analysis of such a morphologically complex and typologically unusual language. Examples are abundant, almost all from texts or natural conversation.19 This is a critical matter, when one is documenting a language on the verge of disappearing, with grammatical differences conditioned by the age of the speaker, and with several distinct dialects. Throughout the grammar the author notes the effects of language obsolescence; she is careful to note differences between older and younger speakers, as well as differences of dialect—another characteristic which enhances the value of this grammar.
A measure of the value and quality of this grammar may be seen in that Cambridge University Press is in the process of producing a paperback edition, to appear in September 2006. Very few technical grammars of obscure languages (especially when spoken by only 100 people out of 1500) appear in paperback, much less one of this magnitude!
Is this grammar useful to the OWL (ordinary working linguist)? Most definitely! It can be used as a model for writing a detailed technical grammar of an unusual language. It is an excellent model for anyone who must write a grammar of a severely endangered language, especially when the grammar is likely to be THE scholarly repository for all human time of that language’s structure. Even though it is not a typical Northern Arawak language, it merits the attention of linguists working with Arawak languages. Due to heavy Tucanoan influence in Tariana, it merits the attention of those linguists who study areal diffusion between unrelated languages. And it could be used in a graduate course concentrating on typological structure and description (along with several other classic grammars of lesser-known languages), as a descriptive model, as a means of teaching advanced linguistic typology, and as source material for relating linguistic description and linguistic theory.
Notes
www.cambridge.org/us/catalogue/catalogue.asp?isbn=0521826640
This is particularly intriguing, in light of unpublished comparative studies which I have conducted on Tucanoan languages. My studies suggest that “H-metathesis” bears investigating in Tucano, Guanano, and Koreguaje (Western Tucanoan). Aikhenvald notes that “H-metathesis is attested in a number of North Arawak languages of the Upper Río Negro” (p.46). “Aspiration floating” is mentioned (pp. 43-4), which seems to be closely related to “H-metathesis”; the author hints that these two phenomena could be related. See also Waltz 2002, who describes mechanisms which result in “postaspiration” in Wanano (Guanano); he speculates that the process has occurred due to the influence of Arawakan languages.
Similar phenomena are staring me in the face as I do field work in a totally unrelated language: I have found suffixes that can act as roots, and phrasal clitics which sometimes act as suffixes.
The author attributes the nominative-accusative case system to Tucanoan influence—a likely possibility
These are not definite or indefinite articles, but instead markers of specificity: “The article can be used with an indefinite noun, provided it refers to an identifiable class of referents which has been introduced in the narrative, and is topical” (p. 205). Further, “modifiers tend to be placed before a noun which is either definite or specific, or which is indefinite but is going to be topical;” data on pp. 205-6, 475-6 suggest that the article and modifiers preceding the noun do not cooccur.
I would be tempted to label this morpheme and kwa-/kwe- as ‘quantifiers’. The author perhaps did not because they take classifiers, and do not behave exactly the same as the group she labels ‘quantifiers’ (11.9); however, there are classifier systems elsewhere in the world in which quantifiers take classifiers.
pa:- is also used for the number ‘one’ (p. 215). Some of its functions as described here are that of a quantifier. It also takes classifiers, and has some other functions, perhaps one reason the author used other terminology to describe it.
The quote here reflects some terminological confusion. The glosses on the accompanying examples helped me figure out what was going on; I insert in brackets the correct terminology for the reader’s benefit. Other confusing terminology may be found on pp. 44 (‘H-metathesis’ and ‘aspiration floating’—see p. 46ff).
See Barnes 1984 for a nice description of a Tucanoan evidential system
Presumably, the author has included this category under “mood and modality” because the corresponding enclitic can mark clauses in the main predicate slot.
As far as I can tell, chapters 15 and 16 do not make reference to the fact that serial verbs can also express aspectual and modal distinctions. Tariana’s morphological complexity pretty well forces one to organize a grammar description according to morphology, as Aikhenvald has done. Chapters 15, 16 and 18 are a case in point: chapter 15 deals with morphemes occurring in two positions following the verb (see the diagram on p. 254); and chapter 16 covers morphemes occurring mostly in one position. The distinctions expressed by serial verbs (chapter 18) seem somewhat different semantically to those described in chapters 15 and 16.
Tucanoan languages sport complex interactions between all of these as well; however, the interactions of Tariana are distinct (even with some calquing from Tucanoan languages).
The author sticks the definition of ‘converb’ within the gloss of a morpheme -ri: ‘ “general non-past nominalisation, converb” ’ (p. 461).
In the exposition these are termed ‘classifiers’.
“They show subjecthood only through ‘the same subject constraint’ in serialisation” (p.511).
One would have wished for a less fragmented orientation to matters so crucial to chapters 22-24, by perhaps combining the summary quotes above (or their equivalent) into section 20.2.5, putting this section at the beginning of chapter 22 and then repeating or referring to the beginning of chapter 22 when necessary. I suspect that the author was trying to emphasize the differences between Tariana’s profusion of clause types, so that the readers could grasp more easily that indeed these are separate grammatical categories.
This could be due to the particular typological theory which the author practices. Except for the first section “25.1 Pragmatic basis for constituent order”, chapter 25 feels like a hodgepodge of discourse-related, albeit valuable comments. One thing that perhaps would help pull it together would be to put section 25.5 at the beginning, rewriting it somewhat so as to describe how the language handles major discourse categories (participants, theme, topic, place, time, contrast, text divisions).
Typos are found on pp. xxii (‘NUM.CL’ is out of alphabetical order), 39, 43 (should be ‘dhala’, as on p. 672), 62, 173 (substitute ‘above’ for ‘in the Position paper’) 129 bottom (Tuc ‘c’ should be ‘k’ in phonemic format), 157, 164, 166, 178 (‘late’ instead of ‘sg.non-fem’), 217-218, 227, 234 (semicolon instead of a comma), 246, 253 (should be ‘diagram’, instead of ‘scheme’—see p. 83), 254 (ditto), 278 (ditto), 309, 413, 440 (second vb missing in ex. 18.51), 507 (should be 5.2.5), 545 (missing a comma), 555.
I only stumbled across one place (p. 305, first full paragraph), where an example or two would have been nice. Elsewhere there were all the examples anyone could ever want.
References
Aikhenvald, Alexandra Y. 1995. Bare. Languages of the World/Materials 100. Munich: Lincom Europa
Aikhenvald Alexandra Y. 1996. Classifiers in Baniwa. Moscow Journal of Linguistics 3:7-33.
Aikhenvald Alexandra Y. 1998. Warekena. In Desmond C. Derbyshire and Geoffry K. Pullum (eds.), Handbook of Amazonian Languages: Volume 4, 225-439. Berlin: Mouton de Gruyter.
Aikhenvald Alexandra Y. 1999. Tariana Texts. Languages of the World/Text Collections 07. Munich: Lincom Europa.
Barnes, Janet. 1984. Evidentials in the Tuyuca verb. International Journal of American Linguistics 50(3):255-71.
Waltz, Nathan E. 2002. Innovations in Wanano (Eastern Tucanoan) when compared to Piratapuyo. International Journal of American Linguistics 68(2):157-215.