Gramatica do Kamaiura, lingua Tupi-Guarani do Alto Xingu (review)

The Amazon basin—an area of great linguistic diversity—is linguistically one of the least known regions of the world. It comprises around 300 languages grouped into over 15 language families and a fair number of isolates. Amazonian languages display dauntingly complicated grammatical structures, providing exceptions to numerous universal statements. These include unusual patterns of split ergativity and complex evidential systems. The vast majority of Amazonian languages are endangered, and many have already become extinct.

Over half of the indigenous languages of Amazonia are spoken in Brazil (see Rodrigues 1986), which covers about half the area of South America and comprises about half its population. About 170 languages are still spoken in Brazil, of which 115 have less than 1000 speakers. Documentation and analysis of these is the matter of urgent priority. And yet there are almost no full, book-length, grammars of indigenous languages published by Brazilian scholars. It is not an exaggeration to say that Lucy Seki’s Gramática do Kamaiurá is the first comprehensive reference grammar published in Portuguese by a Brazilian linguist since J. Anchieta’s (1595) grammar of Tupínambá, a Tupí-Guaraní language once spoken all along the Brazilian coast but now extinct.

Some language families and subgroups are luckier than others. The Tupí-Guaraní branch of the Tupí family is perhaps the best-known and the best-documented. There is hardly any doubt as to its limits and internal classification; a fair amount of its phonology and morphology has been reconstructed (see the summary by Jensen 1999). Kamaiurá, spoken in the Upper Xingu area by about 300 people, constitutes a separate subgroup within the Tupí-Guaraní branch. The Kamaiurá are among the oldest inhabitants of the Upper Xingu, now a place of remarkable linguistic diversity, with 17 distinct indigenous groups (Seki 1999). Whether the Upper Xingu is a linguistic area or not remains an open question—more in-depth studies of individual languages spoken there, especially from Carib and Arawak families, are needed.

This grammar, based on 30-plus years of field research, consists of 22 chapters organized in five parts. The first part contains a detailed overview of linguistic and cultural aspects of the Upper Xingu based on oral accounts and on the few available written sources. Part 2, ‘Syntax’, has an overview of word classes and constituent structure, clause structure, direct and indirect speech, questions, commands, coordination, parataxis, anaphora, possession, negation, and so on. Part 3, ‘Functions and forms’, presents an analysis of nominal and verbal morphology; and Part 4 gives a detailed analysis of the lexicon. The last part is on phonology. Appendices contain 16 pages of fully analyzed texts, a list of lexical items in the examples (c.640), and a list of 83 affixes and clitics with reference to parts of grammar where they are discussed. The list of references is followed by 32 color photos depicting the landscape, traditional activities of the Kamaiurá, S’s fieldwork, and a literacy course. In its overall structure, this book reminds one of Dixon’s (1972) renowned grammar of Dyirbal—which, presumably, served S as an inspiration, a purely Brazilian tradition of grammar-writing being almost nonexistent.

This new grammar incorporates previous work on Tupí-Guaraní languages as its background and builds upon it. The book is typologically oriented, cast in terms of basic linguistic theory (without any reference to transient formalisms). This allows S’s grammar to be read and understood by linguists of any persuasion. And this is why it will remain an accessible model of linguistic argumentation and a source of data and inductive generalizations for many years to come.

For every statement, S provides ample argumentation and exemplification—the total number of examples, from all kinds of textual and conversational data, is over 1500. A methodologically important point implicitly made throughout the grammar concerns the data employed. No doubt, a certain amount...