The term ‘middle’ in this chapter will serve as a rubric for what is essentially predicate valency-reduction fulfilling a variety of ends, all of which are marked on the verb by the morpheme -ye in form-order class [8], immediately to the right of the verb root. Paradoxically however, not all the uses to which middle morphology is put result in a reduced number of arguments for any given predicate.

As its adjacency to the root morpheme suggests, the morpheme -ye is derivational, creating formally intransitive from transitive verb roots. This morpheme performs reflexive, reciprocal, passive and antipassive operations; it is a ‘patientless antipassive’ and an ‘agentless passive’ morpheme in Dixon's (2002:535-536) typology, and a ‘backgrounding antipassive’ in Foley & Van Valin's (1984:338) exposition.

Starting with Austin (1981:151-157) this function with variations has been recognized in a number of Australian languages. Dixon (2002:206-207, 530-536) has proposed an ‘original’ pan-Australian valency-reducing morpheme *-dharri-, although he is not specific about the sense in which this form could have been original, as either a proto-form in phylogeny or in some other sense. From the first syllable he derives subsequent shapes -*dha-, -dhi-, -ji-, and -yi-. These forms are very widespread: they are found, for example, in the Gulf of Carpentaria, on Cape York and in the Lake Eyre Basin.

Clearly the most distinctive aspect of Worrorra phonology, and one of the most distinctive features of the entire language, is the amount of cluster assimilations or fusion involved in any account of its morphology. The assimilations involved are generally consistent with those seen in other Arafuran or Australian non-Pama-Nyungan languages such as Maung, Ungarinyin and Wubuy; nevertheless the extent and nature of this phenomenon as seen in Worrorra is striking by any account. Observe, for example, the forms kunjungu ‘s/he did/said it’ and nye ‘s/he will do/say it;’ or nyimbu ‘s/he hits her’ and ko ‘s/he hits him.’ These two pairs do not constitute instances of suppletion as might be supposed; there is indeed very little suppletion in the language. Both pairs of forms are regular, non-suppletive (ie predictable) entries in the conjugational paradigms of kuN[]=yi ‘do’ and = bwu ‘hit’ respectively. They may only be understood as regular and predictable entries in such paradigms, however, if analysis at a level of abstraction beyond the phonemic level is undertaken.

Morphophonemic changes occur when morphemes are joined. Morphophonemic analysis consists of generalizations about the patterns in which phonemes occur at or near morpheme boundaries, that is to say that morphophonemic processes operate across morpheme boundaries, not within morphemes. Segments at this level of abstraction take part in processes that are analogous to allophonic alternations: morphophonemic rules differ from phonological rules in the sensitivity they ascribe to phonemes with respect to their morphological as well as to their phonetic environments.

Worrorra kinship nouns are nominal expressions which include reference to the propositus (the ‘owner,’ a structurally dependant NP) as well as to the structural head, the person designated as being in such-and-such a relation to the propositus. Although the grammar of Worrorra kinship nominals has been discussed in §13.1, it is relevant here to provide an inventory of Worrorra kinship terms with their equivalents in the usual anthropological metalanguage. As well as being of interest in its own right, such an inventory can be useful in comparative exercises such as that undertaken by McConvell (1997).

Elkin (1954:49-79, see also Berndt & Berndt 1964:76) identifies the north Kimberley or Ungarinyin kinship system as one of five major types of kinship system found in Australia. It is based on the extended patrilineal family or clan as the basic-level category for the purpose of reckoning relatedness. In this system all members of an extended family are to a certain extent equivalent for kinship-reckoning purposes, and terminology reflects this situation. It is what Elkin calls a ‘vertical’ system, in the sense that entire extended families or clans are egocentrically labelled, with either all generations referred to by a single term, or with different terms used for alternating generations. This situation is evident in Lucich's (1968) exposition, and made explicit in Silverstein (nd 1).

Introduction

Complex predicates, also called verbal compounds, constitute the commonest means of verbal expression in Worora. By text count such predicates are far more frequent than simplex verbal expressions. Complex predicates make use of a large and open word class that occurs in few other grammatical contexts: this is the lexical category that I will refer to as preverbal infinitives or preverbs. Worrorra preverbs are semantically verbal, while possessing syntactic properties of both nouns and adverbs. Complex predication, and the lexical category of preverbs it hosts, together constitute one of the defining features of Worrorra in typological terms, and is a characteristic areal feature of north-western Australia generally (see eg McGregor 2002, Amberber, Baker & Harvey (eds) 2010).

Complex predicates in Worrorra are typically composed of a preverb followed by an inflecting verb, here termed a verb classifier for reasons that will become apparent. Predicates of this type are not uncommon in Australia, where they have traditionally been referred to as compound verbs or verbplus-auxiliary sequences (eg Capell 1976:615 ff). Other terminologies include ‘root,’ ‘verb,’ ‘prefix,’ ‘infinitive,’ ‘particle,’ ‘uninflecting verb’ and ‘transparent event’ for initial elements, and ‘thematizing suffix,’ ‘root,’ ‘auxiliary,’ ‘classifier,’ ‘classifier complex,’ ‘light verb’ and ‘inflecting verb’ for final elements (Dixon 1980:426, Nash 1982, Silverstein 1986, McGregor 1990, 2002 and references therein, and Amberber, Baker & Harvey (eds) 2010).

Pronouns, especially the demonstratives reviewed in Chapter Seven, combine in a variety of ways with each other and with adverbs to produce an extensive pragmatic system with referential, deictic and indexical functions. We have already seen some of these pragmatic devices in the existential interrogatives in §7.4.3, the contrastive forms in §7.5.2 and §7.5.5, and the presentational forms in §7.5.4. In this chapter we will briefly observe an array of configurations that encode a number of important pragmatic categories, and which are in turn important in discourse cohesion.

As discussed in §5.3.1, subject and object grammatical relations are indexed on verbs as prefixes, by way of a set of tokens of the agreement-class of the nouns entering into those relations. This is quite a different state of affairs from signalling grammatical relations on the nouns themselves. Verb morphology merely codes the class of the noun in such-and-such a grammatical relation, and any instance of that class in the linguistic or extra-linguistic environment is a potential candidate for interpretation in that relation. The problem for this kind of grammar is that individuals rather than classes are usually required to be encoded in core grammatical relations. For instance eeja inja ‘man’ and warrala inja ‘boy’ are both members of the masculine singular agreement class, but the coding of that agreement class in a particular grammatical function fails to distinguish between the two nouns themselves.

This chapter is concerned with a description of the morphology and semantics of minimally-extended, fully-finite, active verb forms in canonical or characteristic usage; that is, with the inflexion of morphemes signalling core argument structure, tense and aspect. To achieve this, the formally and functionally unmarked indicative category will be taken as an exemplar. Indicative verb forms denote positive, factual assertions and questions in typical reference-and-predication discourse styles.

Worrorra finite verbs occur in four inflexional categories or form-classes, as follows:

1. (i) Indicative mood, described in this chapter,

2. (ii) Optative mood (§8.1 & §8.2),

3. (iii) Counterfactual mood (§8.3 & §8.4),

4. (iv) Subjunctive mood (§15).

As well, a lexical category consisting of non-finite verb forms (preverbs) is described in §11.2.

Classes of verb-forms are defined by the shapes of their prefix clusters and by the shapes of their tense-marking morphemes, both of which are specific to each form-class. These formal distinctions encode a separate and specific semantics in each case, and in each case the entire formal and functional package will be referred to as a mood. Different modalities within each mood are brought into being by constraints on the occurrence of tense and aspect morphemes in order classes [9] and [13].

Indicatives are unmarked forms, while the other three moods are marked. Indicative forms are therefore simpler than the shapes of the other moods, and so indicative prefix clusters are employed in §5.3 to illustrate the way grammatical relations are indexed. The prefix clusters that uniquely characterize indicative forms are set out in Tables 5.1 and 5.2 in §5.3.2. The other forms signalling this mood are its tense markers, which appear in order-class [9] and are displayed in Tables 5.4 and 5.5 in §5.4.

In Chapter 11 generally, and in §11.5.2 in particular, it was shown how preverbs may frequently enter into hypotactic relations with verb classifiers: in these cases the preverb is embedded in the classifier's core as a complement of the classifier, although the linkage is tighter than that usually envisaged when verbal complementation is considered. In this chapter another, looser, kind of complementation is brought into consideration, one that has been referred to intermittently throughout this grammar, in §4.1.3 (iii), §5.1, §5.3.1, §7.2, §8.2.1, §8.4.3, §9.6.2, §10.6.3 (i), §11.5.2, and §13.2.2 (vii). This construction involves the general-purpose pro-classifier kuN[]=yi ‘do’ in its non-classifier use as a simplex verb. A quick outline of this verb's basic usage, as a simplex verb, has been provided in §13.2.1 (iii), and the reader is referred to that section. The present chapter describes the use of kuN[]=yi as a matrix verb, and more particularly the kinds of clauses that may be subordinated to it, as a matrix. The sentences or speech events embedded under or subordinated to kuN[]=yi will be referred to as complement clauses1.

A number of factors motivate the treatment of the framing verb–framed clause relationship in Worrorra as one of syntactic dependence. Such an analysis is isomorphic with the classifier–preverb relationship described in Chapter 11; in both cases the embedding of an eventive complement is signalled formally by celestial indexation on the verb kuN[]=yi in undergoer position.

A predicate is the pragmatic goal of a speech act; that is to say, a predicate is the reason why someone embarks on an act of predicating-and-referring in the first place. For the purpose of investigating linkage phenomena, clauses may be considered as expanded predicates, and grammatically equivalent to them. In Role-and-Reference grammar the clause is understood to be a layered structure, with an inner level called a nucleus, an intermediate level called a core and an outer layer called a periphery. The nucleus contains the non-finite manifestation of a lexical predicate — a denotative type in Langacker's (1991:51) terms. The core contains the nucleus and its ‘core arguments,’ which are nowhere well defined, but which are considered generally to be the subject and in transitive predicates one or two of its objects, selected on morphosyntactic grounds by access to a privileged position within the clause, and on semantic grounds by reference to a hierarchy of semantic roles (Foley & Van Valin 1984:59). The periphery contains everything outside the core. Also distinguished (ibid: 208) are two relevant categories in the clause, a substantive category called ‘constituents’ and a non-substantive one called ‘operators.’ The operators over a predicate are aspect, directionality, modality, reality status, tense, evidentiality and illocutionary force. These operators are said to have scope over (be an integral part of, make specific reference to) different layers of the clause.