Why recognize an adjective class?

possibilities for direct modification of a noun; and the two word classes behave differently with respect to reduplication (see Post 2008, who provides a list of ten properties in which adjectives differ from verbs).

The previous chapter commenced with a quotation from Milner (1956: 10), maintaining that there are no lexical classes at all in Fijian; in his grammar, every kind of lexeme was combined together in a class of ‘bases’. But we showed in §11.8 that classes of noun and verb must be distinguished for proper statement of reduplication processes and for word-class-changing derivations.

In addition, an adjective class must be recognized for proper description of NP structure, further word-class-changing derivations, the conditions for occurrence of certain predicate modifiers, choice of classifier, and so on. The appendix to this chapter provides an account of the necessity for recognizing classes of noun, verb, and adjective in Fijian.

In these and all other cases an adjective class can be recognized, and it is useful. We need now to explain why this is so.

12.2 Why recognize an adjective class?

As stated in §1.1, linguistics has two interwoven components—description and theory. The description of a language is framed in terms of basic linguistic the-ory, choosing from the available roster those categories and construction types which are relevant and useful for the language under study. The theory itself is made up of interrelated inductive generalizations based on good-quality descriptions. As each new language is described, it will throw up significant features which lead to the refinement, revision, or extension of part of the theory.

The recognition of an ‘adjective class’ in the grammar of a particular language (as for every other category) is justified on two grounds—(a) its usefulness and explanatory power within that grammar, and (b) its relation to the general typological theory.

(a) Utility in description. Unlike many formal theories, basic linguistic theory does not consist of a list of components which every grammar must include. What it does, instead, is provide a range of theoretical tools and a pool of conceptual categories, each of which may be utilized in the grammar of a particular language if it fulfils a useful role there in description and explanation. As an illustration, for some languages it is not appropriate to recognize anything which could be felicitously called a complement clause construction. A distinction between derivational and inflectional processes is most helpful in some instances, but there are quite a few languages for which the distinction is simply inapplicable. Basic linguistic theory does not require

that distinct classes of noun and verb be recognized for each language. Our conclusion in the previous chapter that it is appropriate to identify these two classes in every language is based on empirical investigation, rather than it being a theoretical postulate. It is the same with the adjective class.

The inventory of categories and construction types recognized in basic linguistic theory provides the fieldworker with an idea of what to look for in a new language. For example, a few decades ago little had been published on the grammatical category of evidentiality. As a consequence, for a number of languages with a system of evidentiality distinctions the category was not recognized. Now that evidentiality is fairly well understood and well described (its typical content, and patterns of cross-linguistic variation), a student work-ing on a previously undescribed language will be on the lookout for the category. And if it occurs they will be in a position to describe it accurately, and within a cross-linguistic perspective.

There is never just one point of justification for an analytic decision in linguistics. It is always the case that a number of criteria come together—and reinforce each other—to define a category. This category will then play a role in explanation. This is as true for the adjective class as for any other feature of a grammar.

Saeed (1999: 104–9) recognizes a smallish class of about forty-two adjectives in Somali (Cushitic branch of Afro-Asiatic, Somalia, Ethiopia, and Kenya).

Criteria for recognizing adjective as a class distinct from noun include: adjec-tives ‘do not occur with suffixed determiners’ and ‘they do not have inherent number and gender’. They differ from verbs in that they ‘may mark plural agreement with a nominal head by reduplication, which does not occur with this function in verbs’. Having established an adjective class, it has further properties within the grammar. For example, only adjectives can occur as complements of the copula verb yahay ‘be’, being positioned between satellite clitics and the verb. Adjectives may then fuse with the present tense form yahay (for example wanaagasán ‘good’ plus yahay ‘is’ gives wanaagaányay).

In addition, adjectives enter into comparative constructions with adposition ká ‘from’ as marker of the standard of comparison (see §3.23).

For every language which has been thoroughly studied from this point of view, once an adjective class has been recognized, it does play a significant role in the grammar. That is, there is never just one property which serves to identify this or any other word class; there are always—at the least—several.

(b) Role in theoretical generalization. As said before, every new gram-matical description is likely to provide feedback into the make-up of basic linguistic theory. But, in order to achieve this, the description must be framed within the general theoretical matrix. In §6.1 ‘Requirement for consistent

12.2 why recognize an adjective class? 67 analysis’, three languages were contrasted. It was shown that, under one method of analysis, an adjective class would be recognized for all three lan-guages, whereas, if the same criteria were applied in a different order, only language3 would be assigned a distinct adjective class. The discussion there can now be extended.

Consider four languages whose adjective classes show different grammat-ical properties—types (a)–(d) described under II in §12.1. They can be dia-grammed, with grammatical similarities and differences between word classes modelled by spatial distance:

(a) noun adjective verb

(b) noun adjective verb

(c) noun adjective verb

(d) noun adjective verb

In language (d), adjectives have grammatical properties different from those of nouns and verbs, so that a distinct adjective class must be established.

However, alternative analyses are available for languages (a) and (b). In (a), the adjective class has similar grammatical properties to the verb class. We could either:

(i) Say that adjectives constitute a subclass of a combined verb-adjective class (conveniently called just ‘verb class’). Or:

(ii) Say that adjectives are a separate class, noting that their grammatical properties are similar to those of verbs.

In language (b), the adjective class has similar grammatical properties to the noun class. The same two analyses are available, mutatis mutandis. Either:

(i) Say that adjectives constitute a subclass of a combined noun-adjective class (conveniently called just ‘noun class’). Or:

(ii) Say that adjectives are a separate class, noting that their grammatical properties are similar to those of nouns.

If one were interested only in the description of a single language, either alternative would be equally good, the difference being pretty much termino-logical. But if the linguist is interested in relating their grammar to a general theoretical framework, then the consequence of choosing one alternative over the other is immense. Under analyses (ii) all of languages (d), (a), and (b) have a major word class ‘adjective’. As will be shown below, the three adjective classes will have similar functional properties and semantic content.

Were analysis (i) to be followed, only language (d) would have a major word class ‘adjective’. It would be possible—but both complex and unnecessary—to try to relate the adjective class in (d) to a subclass of verbs in language (a) and to a subclass of nouns in (b). Analysis (i) would greatly impede the task of comparing languages and working towards a simple and elegant general theory of language structure.

And what about language (c), where adjectives share significant grammat-ical properties with both verbs and nouns? If analysis (i) were extended to language (c), we would have to say that adjectives are simultaneously a subclass of noun and of verb. There would just be two major word classes, noun and verb, with overlapping identity. Again, the approach followed in analysis (i) has complex and unnecessary consequences.

For every language that has been closely examined, an adjective class can be recognized, although for languages of types (a) and (b) there is an alternative analysis as ‘subclass of verbs’ or ‘subclass of nouns’. If the linguist is interested not only in description of their language but also in the continued refinement of the general grammatical framework in terms of which grammars are writ-ten, then analysis (ii) is the alternative to follow.

It is interesting to briefly survey how ‘adjective class’ has been treated in the past. Both the ancient grammar of Sanskrit by P¯an

˙ini and the early grammars of Greek and Latin—which began the western tradition—failed to make any distinction between noun and adjective. It was only at about1300 ce, in the scholastic grammar of Thomas of Erfurt, that the criterion of gender was invoked—each noun has one inherent gender, whereas an adjective has no gender in itself but may show any of the genders, by agreement with the noun it relates to. On the basis of the European languages they knew, it became the accepted doctrine among linguists that adjectives are a class with similar morphology to nouns, differing from nouns in terms of gender possibilities.

Indeed, it appears that Jespersen (1924: 72) considered this to be the only crite-rion. Since Finnish has no genders he inferred that in this language adjectives could not be distinguished from nouns. There are, in fact, a fair number of other relevant criteria in Finnish—only nouns (not adjectives) take possessive suffixes, and only adjectives (not nouns) take comparative and superlative suffixes.

Australian languages are like the languages of Europe in that adjectives have very similar morphological possibilities to nouns. Some languages have noun classes (similar to genders) and this is accepted as a viable criterion. But for languages without this aid, it is often said that there is no separate class of adjectives. It is instructive to consider the implications of this position. If a language has a category of gender, then it will have a class of adjectives. If

12.2 why recognize an adjective class? 69 it loses gender, then presumably it loses adjectives as a separate word class.

If it then redevelops gender marking, it will regain an adjective class. Such a scenario is surely unacceptable.

In a classic study, Alpher (1991: 22–6) investigates the basis for recognizing a class of adjectives in Yir-Yoront, an Australian language which lacks noun classes/genders. There is no obvious clear-cut criterion to distinguish adjec-tives from nouns, the two types of word having virtually the same morpho-logical and syntactic properties. Alpher is, however, able to suggest five fairly subtle properties in which nouns and adjectives differ. One he labels ‘grading’:

‘Both “nouns” and “adjectives” occur with postposed morr “real, actual, very”.

With common nouns, morr has the sense “actual present-day”, as in kay morr

“the present-day (steel) axe”, or “real and not imaginary”, as in warrchuwrr morr “real woman (not one in a dream)”. With “adjectives” susceptible of grading, however, morr means “very”: karntl morr “very big”, wil morr “very bitter”. Such adjectives, moreover, can be quantified with adpositions like mangl “a little”, as in mangl-karntl “a little bit big”, wil-mangl “a little bit bitter”;

common nouns lack this possibility.’

The modern discipline of linguistics has been centred on the study of European languages, and is generally undertaken by speakers of European languages. There has, as a consequence, arisen the idea that if a language has an adjective class, then it should be similar to the adjective class in European languages; that is, functioning directly as the modifier of a noun in an NP, acting as copula complement, and showing morphological categories similar to those of nouns (number, case, etc.), quite different from the categories applying to verbs (tense, aspect, mood, etc.).

This has undoubtedly played a role in the failure to recognize an adjec-tive class for languages in which adjecadjec-tives show a rather different profile, functioning as head of an intransitive predicate (rather than as copula com-plement), and having some of the same morphological properties as verbs.

There is an oft-repeated tradition of saying that in Chinese ‘all adjectives are verbs’ (see, among many others, Hockett1958: 223; Lyons 1968: 324–5; Li and Thompson1981: 141; Schachter 1985: 18). This lacks insight. In an important study, Xu (1988) demonstrates a range of criteria for recognizing adjectives to be a separate word class in Chinese. For example, adjectives and verbs show different syntax when modifying a noun within an NP, have different aspectual possibilities when functioning as intransitive predicate, take differ-ent derivational possibilities. In addition, reduplication has differdiffer-ent semantic implications for the two word classes; see examples (8–9) in §12.5.1.

Even when a linguist does provide criteria for distinguishing adjectives from verbs (in a language where adjectives can function as intransitive predicate), there is often an unwillingness to use the label ‘adjectives’, simply because

these adjectives are so different in grammatical properties from the familiar kind of adjective occurring in European languages. A term like ‘descriptive verb’ may be used instead (for example, Seki1990, 2000, on Kamaiurá, Tupí-Guaraní branch of Tupí family, Brazil). Oceanic languages typically have an adjective class similar in grammatical properties to the verb class. Buse (1965), writing on Rarotongan, called them ‘statives’, and this label (or ‘stative verbs’) has become institutionalized in Oceanic linguistics.

In summary, there has been a tendency for linguists working on a language of type (a)—where adjectives have similar grammatical properties to verbs—

to opt for analysis (i), treating adjectives as a subclass of verbs (or else simply saying that all adjectival concepts are realized as verbs), whereas linguists working on a language of type (b)—where adjectives have similar grammatical properties to nouns—appear to be more open to recognizing ‘adjective’ as a distinct word class. Anyone maintaining that adjectives are a type of verb in Chinese should be consistent and also say that adjectives are a type of noun in Spanish—analysis (i) in each instance. But, as stated above, if the grammar is to be oriented towards the general typological theory which aims to model human language ability, then alternative (ii) should be followed in each case.

Every language is then accorded an adjective class (on a par with noun and verb classes), the properties of which will now be discussed.