32 R. M. W. Dixon
(3) Manange (Tibeto-Burman; see Chapter 3 below) also has two adjective class-
es. What Genetti and Hildebrandt call Verb-like adjectives' (a class with about
fifty-seven monomorphemic members) may function as intransitive predicate but
lack most of the morphological processes available to verbs. Their 'adjectives' (a
class with about thirty members, which could be referred to as 'noun-like adjec-
tives') may occur as copula complement, like a noun, but cannot be NP head and
have distinct phonotactics. The two adjective classes share properties; for example,
both may directly modify a preceding noun in an NP.
The COLOUR, SPEED, and QUANTIFICATION semantic types consist only of
(noun-like) adjectives, HUMAN PROPENSITY and DIFFICULTY involve only verb-
like adjectives, while DIMENSION, AGE, VALUE, PHYSICAL PROPERTY, and POSITION
include members from both classes. In Chapter 3, Genetti and Hildebrandt pro-
vide a detailed and instructive account of adjectives in Manange.
- Correlations with other grammatical parameters
It is interesting to enquire whether there is any correlation between the type of ad-
jective class found in a language and other grammatical parameters.
One suggestion is reported in Wetzer (1992, 1996) and Stassen (1997). These
authors distinguish between two kinds of languages which they describe, colloqui-
ally, as having 'nouny adjectivals' and 'verby adjectivals'. The 'verby' type covers lan-
guages in which adjectival concepts are considered to be expressed by verbs, and
also those with a separate class of adjectives which share grammatical properties
with verbs; similarly, mutatis mutandis, for 'nouny adjectivals'. They suggest that
languages with 'nouny adjectivals' tend to show a tense system, while languages
with 'verby adjectivals' tend to lack such a system (where tense is defined as, min-
imally, a distinction between past and non-past). They offer an explanation for this.
The referents of verbs are said to be time-varying, while those of adjectives are not.
If adjectives are grouped together with nouns rather than with verbs, then verbs
can show tense, but if adjectives are grouped with verbs, then there will be no tense
specification for this combined class.
There appears to be a degree of statistical support for this generalization, although
there are a considerable number of exceptions. A much more fine-grained study is
needed, with greater attention to the varying grammatical properties of adjective
classes in individual languages.
One recurrent pattern (not mentioned by Wetzer and Stassen) is that if a language
has verbs derived from adjectives, then the adjective is preferred for describing a
fairly permanent property and the verb for referring to a more transient state. This
was illustrated for Igbo by (21-2) in §6.1.
I have observed a quite different correlation—between types of adjective class
and mechanisms for marking the syntactic function of core arguments within
a clause. Following Nichols (1986), we can recognize the following types of lan-
guages:
