72 Anthony E. Backhouse
in the language between Verbals (including verbs and i adjectives) and Nominals
(nouns and na adjectives) (1998: 86-7), although he uses the term 'adjective' in his
labels for both i and na items. Martin (in common with US structuralists start-
ing with Bloch) uses occurrence before the copula as a major criterion, yielding
nouns (including our uninflected adjectives) vs. verbs and i adjectives; this is a
very broad criterion and leads to multiple sub-classification of nouns (into 'pure
nouns', 'adjectival nouns', 'precopular (or quasi-adjectival) nouns', and 'predicable
adverbs') on the basis of further criteria (1975:178-9,782-3).^15 Traditional Japanese
grammar and lexicography recognizes two distinct word classes for i and na adjec-
tives, with no adjectives generally being treated as nouns. Finally, linguists such as
Suzuki (1972) have included all items considered here in a single word class, label-
ling them as type i (our 'inflected') and type 2 (our 'uninflected') adjectives.
To return to questions (a)-(c), the inflected items in (a) show clear differences
in the shape of basic inflections, in lexical derivation, and in syntactic areas such
as combination with auxiliaries and with filler verbs, although negative forms of
verbs stand closer to inflected adjectives on several criteria. The uninflected items
in (b) show high-level syntactic differences in such matters as the ability to head
NPs and the nature of constructions whereby they are modified, and there can be
little justification for regarding these all as nouns; derivational behaviour also dif-
fers widely. With regard to the items in (c), derivational processes are widely shared,
as are many syntactic functions (supported by lexical semantics), with differences
of detail largely reflecting morphological inflection vs. non-inflection. Amalgam-
ation of inflected and uninflected types in the same word class may be seen as
uncomfortable by some, but examples are by no means unusual. Thus, English is
generally regarded as having inflected and uninflected adjectives (for grade), based
at least partly on phonological and morphological factors, with cases of overlap
(cf. Huddleston and Pullum 2002:1582-4). Dixon's (1982) survey of adjective class-
es across languages indicates several other cases: Rotuman has an open adjective
class, of which twelve members inflect for number; seven out of forty adjectives in
Acooli likewise inflect for number; in Swahili fifty adjectives take concordial pre-
fixes of modified nouns, but twenty or so borrowed items (chiefly from Arabic)
take no prefixes; Kiriwinian has a closed class of fifty or so adjectives, of which a
dozen or so obligatorily take classificatory prefixes, a larger number exclude them,
and around twenty optionally take the prefixes; finally, in Dyirbal, one adjective
only ('big') inflects for number (1982: 5, 6, 37, 42-3, 45-6).
Dixon's stance in Chapter i of this volume considers morphology and syntax,
correlating with semantics, and assumes three major classes of simple items. On
the basis of the data analysed here, we conclude that within this framework Japan-
ese adjectives may justifiably be characterized as comprising a large inflected type
(0 which is closed to borrowings in simple stems, and a major open uninflected
(^15) Interestingly, Martin comments that his term 'adjectival noun (i.e. our na adjective) is used for
words with 'special adjectival properties' among 'what looked like predicated nouns', and that 'some
readers may prefer to think of (this) as "nominal adjective'" (1975:30).