7 The Small Adjective Class in Jarawara 197or cooked involves two adjectives (faff and kini) and three verbs (-hata-, -kaha-
and -haw-). And, as mentioned in §5.1, there is also the verb jaff -na- 'be alive, be
raw (not sufficiently cooked)' (which is probably diachronically related to the ad-
jective jati 'new, young'). We have seen that the nominalized forms of the verbs
have different grammatical behaviour from the adjectives.
6 Summary
The small adjective class in Jarawara has just fourteen attested members, four in
the DIMENSION semantic type, two in AGE, three in VALUE, two in PHYSICAL PROP-
ERTY, and three in QUANTIFICATION and QUALIFICATION. The adjectives are non-
verb-like; that is, they cannot function as head of an intransitive predicate. Since
neither nouns nor adjectives take any inflectional affixes, there can be no morpho-
logical similarities between nouns and adjectives.
An adjective may function as modifier to the head of an NP, or as modifier to
an inalienably possessed noun which itself modifies the head. The rule for gender
marking for adjectives in an NP is different from the rule of gender marking on
possessed nouns. An adjective (but not a possessed noun) may constitute a copula
complement.
Present-day Jarawara is head-marking. The first slot in the predicate bears a
pronominal element marking the O argument in a transitive clause; this ends in
-ra if non-singular. The second slot is a pronominal element marking the subject
argument (S in an intransitive, A in a transitive, and CS in a copula clause). An
accusative suffix -ra occurs on NPs in the Paumari language, in the Sorowaha lan-
guage, and in the Jamamadi dialect of Madi, suggesting it should be reconstructed
for proto-Arawa. The -ra suffix on NPs—an indicator of dependent-marking—has
recently dropped out of use in Jarawara (the last old speaker who employed it pro-
ductively died in 2002).
It is possible that an earlier stage was exclusively dependent-marking. The suf-
fix -ra may originally have occurred just on NPs. Then, pronouns (with -ra for
O function) could have developed into obligatory head-marking elements within
the predicate. Later, -ra was dropped from NPs. That is, there could have been a
development from just dependent-marking through a combination of head- and
dependent-marking to just head-marking. The fact that adjectives are non-verb-
like may relate to the past stage of dependent-marking. But this is a highly specu-
lative scenario.References
DIXON, R. M. W. 1995. 'Fusional development of possessed nouns in Jarawara', International
Journal of American Linguistics 61.263-94.- 'Arawa', pp. 293-306 of The Amazonian languages, edited by R. M. W. Dixon and A.
Y. Aikhenvald. Cambridge: Cambridge University Press.