The Adjective Class
II. Grammatical properties. There is a rough division into four types of adjective class:
4. Use as adverbs
12.9 Semantic overlapping between word classes
accordance with the correlation in (24). In Nunggubuyu, the adjective class has had its grammatical possibilities extended so that it may now accept subject pronominal prefixes, like an intransitive verb (it does not yet directly accept tense and other verbal suffixes; this would be the next step). In Japanese, a new adjective class has been established, which is steadily increasing in size.
Korean is a clear exception to the correlation in (24), being NP-marking and having just one adjective class, which is ‘verb-like’. The speculative scenario just suggested for Japanese could be extended to Korean—supposing that the language originally lacked both Predicate-marking and NP-marking, and had a class of ‘verb-like’ adjectives, but then developed NP-marking. Syn-tactic function is, as in Japanese, shown by particles following an NP, and some of these are thought to have developed recently. (For example, Sohn 1999: 30 mentions that the subject particle ka—which is now a conditioned allomorph of the earlier subject marker i—first appeared in the literature in 1572 ce and may possibly have been a borrowing from the Japanese subject particle ga.)
It may be that extending the Japanese scenario to also apply to Korean is transcending speculation in the direction of fantasy. Like other results in linguistic typology, (24) is a statistical correlation, not a hard-and-fast rule.
There are exceptions to it. Besides Korean, these include Southern Paiute (Uto-Aztecan; Sapir1930–1), which is also NP-marking and shows ‘verb-like’
adjectives. It is, of course, worthwhile according detailed examination to these and other exceptions, to see whether there is an explanation (along historical or other lines). But it is unlikely that every exception will be provided with an explanation, and it is an error to try forcibly to provide one. Some languages just do have a typologically unusual combination of properties, in some area of the grammar.
12.9 Semantic overlapping between word classes
It was mentioned in §1.8 that while each word class has a similar semantic core across languages, there are a number of non-core concepts whose word class membership varies between languages; for example ‘hunger/hungry/be hungry’ can be a noun, an adjective, or a verb.
We have seen that some languages have a small closed adjective class.
However, most languages exhibit a large, open class. For such languages, it is interesting to compare the semantic membership of the three open classes—
noun, adjective, and verb. This will now be attempted for Dyirbal, contrasting it with English.
The adjective class in Dyirbal is large and open, and it is like the noun class in its grammatical properties (very similar to European languages). The semantic contents of the adjective classes in Dyirbal and English are similar;
that is, most adjectives in Dyirbal correspond to adjectives in English, and vice versa.
However, there are a number of small semantic fields for which Dyirbal has adjectives while English has verbs. Some of these are exemplified in (25).
(25) verb in adjective in englishdyirbal
(a) divide ñarri ‘divided up’
split yagi ‘split’
crack gajala ‘cracked’
smash muñi ‘smashed up’
tear gini ‘torn’
fold wujun ‘folded’
(b) gather balmbu ‘gathered together’
heap gurruñ ‘heaped up’
muster guwurr ‘mustered’
(c) lean yulgarra ‘leaning’
(d) marry julbun ‘married’
It will be seen that the terms in (a) deal with related concepts, to do with changing the form of an object; those in (b) have to do with getting together several things. Note that, just as adjectives can be derived from verbs in English (cracked, torn, leaning, etc.), so can verbs be derived from adjectives in Dyirbal—adding -bi-l to form an intransitive and -ma-l to form a transitive stem (for example, yagi-bi-l ‘be split’, balmbu-ma-l ‘gather together’).
There are thus some differences in the ways in which languages divide up semantic space into word classes. Corresponding to adjectives happy, clever, and jealous in English, Fijian has S = A type ambitransitive verbs maarau (-ta"ina) ‘be happy (about)’, vu"u(-ta"ina) ‘be clever (at)’, and vuuvuu(-ta"ina)
‘be jealous (of)’. Corresponding to verbs divide and marry in English, Dyirbal has adjectives ñarri ‘divided up’ and julbun ‘married’.
It is now relevant to ask whether there is any semantic overlap between word classes in individual languages, and—if so—whether different languages show similar or different overlaps. That is, we can enquire whether a given concept may be coded (1) by both verb and noun; (2) by both adjective and noun; or (3) by both verb and adjective.
12.9 semantic overlapping between word classes 101 Looking at English, there are many instances of (1) and (2), but few of (3).
Some examples of verb/noun and of adjective/noun overlap are given in (26).
(26) verb noun adjective noun
(a) hit blow big, small size
go journey new, young, old age
happen event fast, slow speed
(b) arrive arrival long, short length
think thought accurate accuracy
announce announcement happy happiness
The pairs in (a) are non-cognate between word classes. Those in (b) are some of the many examples of nouns derived from verbs and from adjectives (a different form of derivation appears in each example).
There are very few instances, in English, of verb and adjective with similar meanings (where the forms are not related through a productive derivation).
One example consists of fear and afraid, as in:
(27) (a) I fear having to enter the lion’s cage (b) I was afraid when I entered the lion’s cage.
The adjective afraid is generally used to refer to the feeling one gets when one is actually in contact with something scary, while the verb fear tends to be used for a general feeling in connection with something that might happen.
In fact, this particular verb/adjective overlap recurs in a number of lan-guages. For instance, the Australian language Yidiñ has a transitive verb yarNga-n and an adjective munu with very similar meanings to fear and afraid in English (Dixon 1991b: 240, 272). However, not all languages have such an overlap. Describing the Oceanic language Mokilese, Harrison (1976:
150) mentions that there is just one lexeme, mijik, corresponding to both
‘fear’ and ‘afraid’ in English. (Following the Oceanic tradition—see §12.2—
Harrison says that this belongs to the class of stative verbs; however, applying the criteria set forth in the present study, the label ‘adjective’ is appropriate.)
English thus has considerable semantic overlap between the verb and noun classes, and between the adjective and noun classes, but very little between verb and adjective classes. Turning now to Dyirbal, we find exactly the oppo-site situation. Here there is no overlap at all between verb and noun classes, or between adjective and noun classes. Basically, Dyirbal does not have abstract nouns such as ‘journey’, ‘event’, ‘thought’, ‘size’, ‘happiness’, or ‘colour’. One simply has to use the appropriate verb or adjective.
However, Dyirbal does have considerable semantic overlap between the classes of verb and adjective. A sample of these is set out in (28).
(28) verb adjective
(i) ñaju-l ‘cook’ ñamu ‘cooked’
(ii) dadi-l ‘cover’ Nulguñ ‘covered’
(iii) gulba-l ‘block’ gumun ‘blocked’
(iv) wanda-l ‘hang’ burrgaligan ‘hanging’
(v) baNganda-y ‘be sick’ wulmba ‘sick’
There is in fact a clear difference of meaning in each case, with the verb referring to an action, or getting into a state, or being in a state that varies with time, and the non-cognate adjective referring to either a state that is the result of an activity, or a state that is semi-permanent. There is a slightly different semantic contrast for each verb/adjective pair. Taking them one at a time:
(i) The transitive verb ñaju-l refers to the act of cooking; its participle ñajuNu can describe something being cooked a bit or a lot, not enough or too much. In contrast, the non-cognate adjective ñamu means
‘cooked to perfection, ready to eat’.
(ii) The transitive verb dadi-l refers to any sort of act of covering; its par-ticiple dadiNu can describe a blanket over just half a sleeping person.
In contrast, the adjective Nulguñ means ‘properly covered, covered all over’.
(iii) The transitive verb gulba-l can refer to any kind of blocking; its par-ticiple gulbaNu can be used to describe a temporary obstruction across a path. In contrast, the non-cognate adjective gumun refers to some-thing permanently blocked; for example, a road that has been closed off for good, or a road that simply stops at a certain place, never having been constructed any further.
(iv) The transitive verb wanda-l is used to describe hanging something up;
the participle wandaNu can be used of a basket or bucket which has been hung from a hook. The adjective burrgaligan refers to something hanging down; for example, long hair on a person, or bark hanging off a tree (it can also be used to describe something that has been hung up).
(v) The intransitive verb baNganda-y is used to describe feeling sick or ill (or just weary); the participle baNgandaNu refers to someone who is under the weather at present, but is expected to get better. In con-trast, the adjective wulmba refers to someone who is truly sick and is expected to die. (Death is believed to be caused by sorcery, so that using wulmba of a person is saying that a sorcerer has done something to them which will result in their death.)
12.10 summary 103 The kind of overlap between these three major word classes in English and in Dyirbal can be shown diagrammatically:
V N
Adj
V N
Adj Type (a) English Type (b) Dyirbal
It appears that most of the languages of Europe are basically of type (a), like English. It is interesting to enquire what other languages are of type (b), like Dyirbal. Data is hard to come by, since very few linguists provide a detailed description of the semantic characteristics of word classes, let alone investigate the possibility of semantic overlap between classes. There are just a few hints available. For Zuni (isolate, New Mexico), Newman (1968: 66) provides the following examples of overlapping between the verb class and what should probably be recognized as the adjective class:
(29) hemmo Pahha šalu
‘to boil’
‘to buy’
‘to stretch’
woleya Payyo pilla
‘to be boiled’
‘to be bought’
‘to be stretched’
Sohn (2004) offers some illuminating remarks on the semantic overlap between word classes in Korean. He shows that there is overlap between verb and adjective classes involving just native lexemes, whereas noun/verb and noun/adjective overlaps often involve one native and one loan item (the latter from the Sino-Korean stratum of vocabulary).
Overall, one would expect semantic overlap between word classes to be found most commonly in languages which maintain a strict correspondence between word class and functional slot. Dyirbal is of this type—a noun can only function as head of an NP (in predicate argument function) and a verb only as head of a predicate. A language with more fluid class-slot correspon-dences (such as Nootka, briefly discussed in Chapter11 above) might be less likely to feature semantic overlap between its word classes.
12.10 Summary
The label ‘adjective class’ is here used for a word class that:
ris grammatically distinct from noun class and verb class;
r(A) functions either as intransitive predicate or as copula complement and/or (B) modifies a noun in an NP;
rincludes words from some or all of the prototypical adjective semantic types—dimension, age, value, and colour.
In some languages two separate adjective classes can be recognized. Note that although the prototypical adjective class combines functions (A) and (B), in some languages the class has only one of these functions.
There are two broad parameters of variation for an adjective class—size and grammatical properties. Some languages have a small closed class (with anything from a handful to a few score members) whereas others have a large open class (to which new derivations and/or new loans may be added). And adjectives may be similar in their grammatical properties to nouns, or to verbs, or to both, or to neither.
It can sometimes be a tricky matter finding criteria to distinguish ‘verb-like’
adjectives from verbs, or ‘noun-like’ adjectives from nouns. I believe that for every language which is studied in detail, such criteria can be found. Criteria are not always of the definitive ‘yes-or-no’ variety; as Alpher (1991) shows for the Australian language Yir-Yoront (see §12.2), a collection of statistical tendencies can combine to provide a satisfactory grammatical characterization of the adjective class (as opposed to noun and verb classes).
In §12.8, a tentative correlation was established—adjective classes which have grammatical properties similar to nouns tend to be found in languages with NP-marking, while adjective classes with grammatical properties similar to those of verbs tend to be found in languages with Predicate-marking or with neither NP-marking nor Predicate-marking. I suggested, with some sup-porting exemplification, that if a language shifts its Predicate-marking/NP-marking profile, then the orientation of its adjective class is likely slowly to change, to re-establish the correlation.
§12.9 looked briefly at the kinds of semantic overlap between the three major word classes. We saw that English has considerable verb/noun and adjective/noun but rather little verb/adjective overlap, while Dyirbal is almost exactly the reverse, with considerable verb/adjective but no verb/noun or adjective/noun semantic overlap.
12.11 What to investigate
Word classes should be distinguished on language-internal grammatical cri-teria. The label ‘adjective’ is appropriate for that class which (A) functions either as intransitive predicate or as copula complement and/or (B) modifies a noun in an NP. It is extremely likely that this class will include lexemes from the semantic types dimension, age, value, and colour (and, if the class has more than a few dozen members, also fromphysical property, human propensity, and speed).
12.11 what to investigate 105 The linguist should investigate the size of the adjective class—whether it is small and closed, or large and open to the addition of new members.
The morphological and syntactic properties of the adjective class should be studied—whether these are similar to the grammatical properties of nouns, or to those of verbs, or to both, or to neither.
There may be two adjective classes; typically one is grammatically similar to verbs and the other to nouns, but they are linked together through a number of shared properties. An appropriate topic for study, at an advanced stage of analysis, concerns the semantic overlap between noun, adjective, and verb classes (see the discussion in §12.9).
Some of the recurrent criteria for distinguishing adjectives from verbs and from nouns were discussed and exemplified in §§12.5.1–2. A fuller list of features to examine can now be provided. (These can, of course, only be investigated after the full set of morphological and syntactic properties of nouns and of verbs has been established.)
(i) Useful criteria when adjectives are grammatically similar to verbs
1. Can an adjective function (like an intransitive verb can) as head of an intransitive predicate? And can a noun or pronoun (or any other kind of word) or NP also be head?
(a) Do the same morphological processes apply to an adjective as to an intransitive verb in this slot? (TAM marking, pronominal marking, etc.) Or just some of them?
(b) If the same processes apply, are they realized in the same way on verbs and on adjectives?
(c) If a verb may take an auxiliary, may an adjective do so too?
2. Are there any affixes or particles which apply to both verbs and adjec-tives but have a different meaning with the two word classes?
3. Can an adjective be used in imperative mood, in the same way that a verb can?
4. Do adjectives behave like verbs with respect to derivational processes (for example, causative)?
5. What are the transitivity values of predicates that accept adjectives and verbs as heads? (Generally, adjectives only occur in intransitive pred-icates. There may be several subclasses of verbs, each having distinct transitivity potential.)
6. Does reduplication apply to adjectives but not to verbs (or vice versa)?
If it applies to both, does it have the same or different form and/or meaning with the two word classes?
7. If there are nominalization process(es), do they apply in the same way to verbs and to adjectives?
8. Can an adjective modify a noun (and also pronoun?) in an NP?
(a) Directly? (preceding or following the head or either?) (b) Indirectly, with relative clause or other marker?
If a verb may also be modifier, how do its possibilities and marking differ from those of an adjective modifier?
9. Can an adjective be head of an NP? Can a verb also function as head of an NP?
If so, does this apply for a simple NP, or an NP which is a nominalized clause?
10. Does an adjective have the same possibilities as a verb for functioning as head of the predicate in a subordinate clause?
(ii) Useful criteria when adjectives are grammatically similar to nouns 11. When an adjective is modifier to a head noun
(a) Does it agree with the head in gender/number/definiteness/etc.?
(b) If it shows a category in common with the head noun in the NP, does this category have the same realization on noun and on adjective?
12. Is gender/noun class (or classifier) choice a criterion for distinguish-ing adjectives from nouns? (A noun may have just one inherent gen-der/noun class, whereas an adjective may take any gengen-der/noun class, agreeing with the noun it is modifying.)
13. Do adjectives differ from nouns in number marking? In system or in realization? For example, in Berber languages, nouns have irregular number marking but adjectives are all regular.
14. Is there an affix or particle that applies to both nouns and adjectives but has a different meaning with the two word classes?
15. Does reduplication apply to adjectives but not to nouns (or vice versa)?
If it applies to both, does it have the same or different form and/or meaning with the two word classes?
16. Can an adjective make up an NP all by itself? If so, is this best described as an elliptical NP, with head omitted, consisting just of a modifier, or as an NP for which the adjective is head?
17. If an adjective is NP head, can it be modified in the same way as can a noun when it is NP head?
18. If the language has case marking, investigate the rules for its realization on an NP; for example, on last/first/head word, on every word, on words of certain types. Does case go onto adjectives according to this general rule? (For example, if case goes on the last word of the NP, it attaches to an adjective just when it is the last word.) Or does a special case rule apply for adjectives?
12.11 what to investigate 107 (a) If both nouns and adjectives show case, does the same case system
apply for the two word classes?
19. Do adjectives function in the same way as nouns for possession? For example, do they take bound possessive pronouns or other markers of possession?
(a) Can an adjective be (i) possessor, (ii) possessed, within a possessive construction?
20. If a noun can also be a modifier to an NP head:
(a) Which nouns may be modifier?
(b) Do noun modifier and adjective modifier behave in the same way?
For example, if an adjective modifier may be further modified by an adverb (or ‘very’), may a noun modifier also be?
(c) If a noun may have several adjective modifiers, may it have several noun modifiers?
21. Can an adjective modifier be used anaphorically, for a full NP? And can a noun modifier be?
22. Are there alternative constructions for linking a head noun with an
22. Are there alternative constructions for linking a head noun with an