The Athapaskan verb is often regarded by linguists as an object of some wonder, and I have frequently been asked how anyone could ever acquire an Athapaskan language, given the extreme complexities of the verb system. In this book, I have tried to come to an understanding of some of the mysteries of the verb, arguing that scope is an important factor in regulating morpheme order in the verb. I hope the reader has enjoyed this journey through the verb, received some satisfactory answers, and, most important, been stimulated and engaged.

In this final chapter, I turn away from ordering and briefly examine several issues that arise from the claims that I have made related to the lexicon and word formation. These include the nature of the lexical entry, models of word formation, the distinction between inflection and derivation, and the nature of morphological change. I do not provide answers to any questions, but intend instead to demonstrate more of the fascination of the Athapaskan verb and the important issues that it raises for the study of morphology.

On the Nature of the Lexical Entry

In this section, I examine the nature of the lexical entry. What is listed in the lexical entry of a typical verb in an Athapaskan language? I pose this question from a theoretical rather than a lexicographic perspective.

This volume consists of a number of detailed case studies of transitivity across a selection of languages – from North, Central and South America, New Guinea, Australia, the China/Myanmar (Burma) border, North-east Africa and the Caucasus. In the introduction we summarize the theoretical presuppositions and parameters, suggest generalizations that can be made on the basis of comparison of the individual studies, and draw attention to useful directions for further research.

§1 describes varieties of predicate arguments and clause types. §2 deals with classes of verbs, and transitivity-encoding devices. Then §3 presents an overview of derivations which change valency. In §4 we discuss derivations which typically reduce valency – passive, antipassive, reflexive, reciprocal and anticausative (plus the ubiquitous middle). §5 considers derivations which typically increase valency – applicative and causative (there is a full discussion of causatives in chapter 2). In §6 we emphasize the need for a holistic approach; every derivational process is likely to have syntactic, semantic and discourse/pragmatic aspects, each of which interrelates with and helps explain the others. §7 briefly discusses the propensities of different groups of verbs – according to their semantics – for taking part in the various derivations that affect valency. In §8 we look briefly at syntactic alternatives to valency-changing derivations; in some languages there may simply be alternative construction types, with no derivational link. §9 gives a short preview of each of the following chapters. Finally, in §10 we suggest a number of fruitful lines for further research.

Introduction

This chapter considers transitivity alternations in Tariana. There are two valency-reducing operations and four valency-increasing causativizing mechanisms; there is also a valency-increasing mechanism distinct from both causatives and applicatives. Transitivity alternations in Tariana provide evidence as to what grammatical relations must be distinguished in this language.

The chapter is organized as follows. I describe typological properties of Tariana in §2. Then, in §3, I discuss valency-reducing derivations. Valency-increasing mechanisms are analysed in §4. Some conclusions and typological perspectives are given in §5.

Typological characteristics of Tariana

Tariana is predominantly head-marking with a few elements of dependent-marking. There are three open word classes – nouns, verbs and adjectives. Relations between word classes and functional slots in a clause are summarized in appendix 5.1. The language has a complicated verb structure, and a large system of classifiers (Aikhenvald 1994a). Constituent order is free, with a strong tendency towards verb-final order. How grammatical relations are marked is discussed in §2.1. Verb structure is discussed in §2.2; and in §2.3 we look at transitivity classes of verbs.

Grammatical relations

Grammatical relations in Tariana are marked in two ways.

Cross-referencing on the verb

Verbs in Tariana divide into two classes: those which take prefixes (cross-referencing, relativizing and negative) and those which do not. Active transitive and intransitive verbs obligatorily take cross-referencing prefixes to mark the A/Sa constituent (following the terminology in Dixon 1994). Stative verbs (So) and verbs of physical states (Sio) do not take any cross-referencing markers. This division roughly corresponds to the morphological distinction between active and stative intransitive verbs inherited from Proto-Arawak.

This volume includes revised versions of ten of the sixteen presentations at the International Workshop on Valency-changing Derivations, held at the Research Centre for Linguistic Typology of the Australian National University, 18–23 August 1997. The position paper for the workshop was Dixon and Aikhenvald's ‘A typology of argument-determined constructions’ (pp. 71–113 of Essays on language function and language type, edited by J. Bybee, J. Haiman and S. Thompson. Amsterdam: John Benjamins, 1997). Contributors were also sent a short paper by Dixon on the semantics of causatives; this was later revised and greatly expanded, and is chapter 2 below.

All of the authors have pursued intensive investigation of languages, most of them rather little-known in the literature. They were asked to write in terms of basic linguistic theory – the cumulative theoretical framework in terms of which most descriptive grammars are cast – and to avoid formalisms (which come and go with such frequency that any statement in terms of them will soon become dated and inaccessible).

We thank all of the authors included here, for taking part in the Workshop, for getting their chapters in on time, for revising them according to recommendations of the editors and of the publisher's referees, and for completing their revisions on schedule.

We are also grateful to Jennifer Elliott, Administrator of the Research Centre for Linguistic Typology, who organized the Workshop with care and flair, coordinated the gathering of papers, and prepared a collated list of abbreviations. To Suzanne Kite, who prepared the indices. And to Kate Brett, our Cambridge editor – sympathetic and yet firm – who played a critical role in getting this volume into its present shape.

Every language or language family has its sets of problems that anyone working on that language/family must have their say about. One of these problems for the Athapaskan family involves the so-called classifier system, a poorly named system of the first type of voice/valency-affecting derivation outlined in Dixon and Aikenvald (1997) – there is overt marking of suppression and addition of arguments. In this chapter, I focus on the commonalities and differences in use of this system across the Athapaskan family.

Background

I begin with some background on the languages and language family. Athapaskan languages are geographically widespread in western North America, falling into three major discontinuous geographic regions. The northern group is spoken in parts of Alaska and in the Canadian provinces and territories of Manitoba, Saskatchewan, Alberta, British Columbia, the Yukon and the Northwest Territories. Three of the languages that I examine here come from the northern group: Ahtna and Koyukon, which are spoken in Alaska, and Slave [slevi], which is spoken in the Northwest Territories, one community in the Yukon, a small area of northern British Columbia, and northern Alberta. The Pacific Coast languages are/were spoken in northern California and Oregon. I draw data from Hupa here. Hupa is the most divergent of the Athapaskan languages covered in this survey. Finally, the Apachean group is found in New Mexico, Arizona, Utah and Colorado in the southwestern United States. I use Navajo as an exemplar of this group. I occasionally bring in other languages as well to illustrate particular points.

In chemistry, valency refers to the capacity of an atom or group of atoms to combine in specific proportions with other atoms or groups of atoms. The French linguist Lucien Tesnire is generally credited with introducing this term to linguistics, where it is used metaphorically for the capacity of a verb to combine with distinct arguments or valents (Crystal 1985). A verb like rain, which has no referential noun phrases associated with it, is said to be zero-place or avalent; a verb like disappear, which takes only a subject argument, is said to be one-place or monovalent; verbs like devour and give are said to be two-place (bivalent) and three-place (trivalent), respectively.

This chemical metaphor has had a pervasive influence in linguistics: causative and applicative morphemes are now described as ‘adding arguments’, while passives and middles are described as ‘suppressing’ or ‘deleting’ arguments, respectively. Entire sections of grammars are devoted to ‘valency-changing’, ‘valency-increasing’ or ‘valency-reducing’ processes, suggesting that the primary function of these grammatical processes is to regulate the number of arguments in clauses.

The chemical metaphor contrasts with an older tradition that distinguishes just two classes of predicates – Transitive and Intransitive – and a category of voice. Passive voice and middle voice are seen within this tradition as altering the ‘point of view’ or ‘centre of interest’ (Jesperson 1924: 167) within a clause rather than applying mathematical operations to it, and causatives and applicatives are sometimes included in, and sometimes excluded from, the traditional range of voice-related phenomena.

This chapter investigates the range of constructions in Ngan'gityemerri that can be thought of as involving shifts in the associated argument structure of verbs. The major contribution of this chapter is to highlight the fact that some languages have at their disposal morphosyntactic mechanisms that encode valency, without necessarily needing morphological derivations to achieve the same kinds of valency shifts that derivations are good for. Ngan'gityemerri's main strategy, a system of complex verb formation, is of the morphosyntactic type. However, it does additionally make use of some minor but genuinely morphological derivations, including a presentative applicative derived from an incorporated bodypart noun.

Preliminaries

There are two parameters around which Australian languages vary. Firstly, most employ suffixes, but in northern Australia there is a large bloc of languages that also use prefixes. These latter ‘prefixing’ languages mostly have A, S and O bound pronominals as prefixes to the verb. Ngan'gityemerri requires the obligatory cross-referencing of core arguments by bound pronominals on the verb in strictly nominative/accusative patterning. However it is unusual amongst prefixing languages in marking A and S as prefixes, but O as a suffix.

Secondly, some Australian languages have many simple inflecting verbs and few complex verbs, while others have few simple inflecting verbs and large numbers of complex verbs. Ngan'gityemerri is of the latter type. One of its most distinctive typological features, shared by many languages in northern parts of Western Australia and the Northern Territory, is that it has only a handful of simple inflecting verbs, and thousands of ‘complex verbs’ made up of two component lexical elements.

Introduction

Motuna is one of eight non-Austronesian, or Papuan, languages from Bougainville, Papua New Guinea. It has several thousand speakers.

Typologically, Motuna is an agglutinative language with considerable morphophonological fusion. It is both head-marking and dependent-marking. NPs are marked by case suffixes. Core case markings are ergative/absolutive (ergative marking is optional in certain environments – see §2(I)). Verbs, kinship terms, classifiers and numerals show extremely complex morphology, with both suffixing and prefixing.

Constituent order tends to be verb-final, with A and O in either order. Any NPs can be left unexpressed if understood from the context.

The major characteristics of the grammar of the language which bear on the main discussion of this chapter are summarized below:

1. A five-term gender system and classifiers coexist in Motuna. Five genders – masculine, feminine, diminutive, local and manner – are distinguished in singular nouns, but these are neutralized in nonsingular number (dual and paucal are marked by diminutive-like agreement, and plural by masculine). The gender of a noun is obligatorily cross-referenced by modifiers within the same NP and/or by the predicate. In addition, most nouns can be classified into one or more of fifty-one semantic types, which can be indicated by classifiers modifying them. Classifiers are combined with numerals, demonstratives, possessive pronouns, etc.

2. Nouns distinguish four numbers – singular, dual, paucal and plural. Personal pronouns distinguish singular and nonsingular number. In 1st person nonsingular there is also an inclusive/exclusive distinction. There is no 3rd person free pronoun, and the demonstrative is used instead in deictic/anaphoric function.

Introduction

Dulong/Rawang is a Tibeto-Burman language spoken on both sides of the China/Myanmar (Burma) border just south and east of Tibet. In China, the people who speak this language for the most part live in Gongshan county of Yunnan province, and belong either to what is known as the ‘Dulong’ nationality (pop. 5,816 according to the 1990 census), or to one part (roughly 6,000 people) of the Nu nationality (those who live along the upper reaches of the Nu River – the part of the Salween within China). Another subgroup of the Nu people, those who live along the lower reaches of the Nu River (in China), speak a language called ‘Anung’ which seems to be the same as, or closely related to, the Kwinpang dialect spoken in Myanmar, so should also be considered a dialect of Dulong/Rawang. Within Myanmar, the people who speak the Dulong/Rawang language (possibly up to 100,000 people) live in northern Kachin State, particularly along the Mae Hka ('Nmai Hka) and Maeli Hka (Mali Hka) River valleys. In the past they had been called ‘Hkanung’ or ‘Nung’, and have often been considered to be a subgroup of the Kachin (Jinghpaw). Among themselves they have had no general term for the entire group; they use their respective clan names to refer to themselves. This is true also of those who live in China, although these people have accepted the exonym ‘Dulong’ (or ‘Taron’, or ‘Trung’), a name they were given because they mostly live in the valley of the Dulong (Taron/Trung) River.

Introduction

This chapter will survey causative constructions in terms of three parameters: their formal marking, their syntax and their semantics. It will also investigate dependencies between the parameters.

I work in terms of a basic theoretical framework. In any language verbal clauses can be categorized as (i) intransitive, with one core argument, in S function; (ii) transitive, with at least two core arguments, in A and O functions (there are subtypes: simple transitive and ditransitive); and (iii) copula, involving two core arguments, in copula subject and copula complement functions (in some languages the copula complement may be omittable). Within a transitive clause, that core argument whose referent has the potential to initiate or control the activity is linked to A function, and that core argument whose referent may be saliently affected by the activity is linked to O function.

A causative construction is sometimes described as involving ‘two events’. Frawley (1992: 159) talks of ‘a precipitating event’ and ‘a result’, and Shibatani (1976a: 1) of ‘a causing event’ and ‘a caused event’. I prefer a quite different characterization – a causative construction involves the specification of an additional argument, a causer, onto a basic clause. A causer refers to someone or something (which can be an event or state) that initiates or controls the activity. This is the defining property of the syntactic–semantic function A (transitive subject).

That is, if a causative construction is formed by derivation, it will involve the addition of a new argument in A function (the causer).