| |
Abstract:
In Japanese, there are two morphologically distinct sets of
adjectival inflections, which are neither semantically nor
syntactically distinguishable (Nishiyama, 1998). The two take
distinct suffixes for all slots in the inflectional paradigm,
including attributive, negative, and non-past
(-i
vs.
-da)
and past
(-katta
vs.
-datta)
tense. We propose that (1) inflections in one set
(-da, -datta,
etc.) apply as defaults when those in the other
(-i, -katta,
etc.) fail; and (2) this dichotomy is best explained by a
memory/rule dual-system computational model.
In a number of morphological paradigms, one set of morphological
transformations applies in limited contexts, while a distinct
default transformation applies when the first set does not. This
pattern has been explained by three competing computational
theories. One posits that all but the most arbitrary
transformations are computed by mental rules (Halle & Mohanan,
1985). The second posits that rules are only descriptive entities,
and that all transformations are computed in associative memory
(Plunkett & Marchman, 1993; Hare & Elman, 1995). The third
proposes a dual-system model, with only the default transformation
computed by mental rules, and others dependent upon associative
memory (Pinker, 1991).
In Japanese adjectival inflection,
-i/-katta/
etc. suffixation is not productive, applies almost solely to
native Japanese words, is constrained by the phonological structure
of roots, and is common among high-frequency words. In contrast,
-da/-datta/
etc. suffixation is highly productive, applies to native and
non-native adjectives alike, does not appear to be constrained by
root phonological structure, and is prevalent among low-frequency
words.
Native Japanese-speaking adults provided past-tense forms of
-i-
suffixed adjectives ("IAs"),
-da-
suffixed adjectives ("DAs"), and IA- and DA-sounding novel forms.
The IA(-sounding) items had a range of IA "neighborhood-sizes"
(number of adjectives with shared root-final moras) but small DA
neighborhoods. The DA(-sounding) items had a range of DA
neighborhood-sizes but small IA neighborhoods. Each item was
presented with the expected(IA(-sounding)
-i,
DA(-sounding)
-da)
and inappropriate (IA(-sounding)
-da,
DA(-sounding)
-i)
suffixes.
Subjects inflected IA(-sounding)
-i
and DA(-sounding)
-da
forms with the expected past-tense
-katta
and
-datta
suffixes, respectively. The IA(-sounding)
-da
forms were
-datta-
suffixed more than the DA(-sounding)
-i
forms were
-katta-
suffixed (p<.05) , even though the IA(-sounding) and
DA(-sounding) items had small and non-significantly different DA
and IA neighborhood-sizes, respectively. IA neighborhood-size
correlated positively with IA(-sounding)
-katta
formation (and negatively with IA(-sounding)
-datta
formation) (p<.01), whereas DA neighborhood-size did not
correlate significantly with DA(-sounding)
-datta
or DA(-sounding)
-katta
formation.
These data suggest a default pattern for Japanese adjectival
inflection, and are best explained by a dual-system model.
References
Halle, M., & Mohanan, K. P. (1985). Segmental phonology of
modern English.
Linguistic Inquiry,
16(1), 57-116.
Hare, M., & Elman, J. (1995). Learning and morphological
change.
Cognition,
56, 61-98.
Nishiyama, K. (1998).
The Morphosyntax and Morphophonology of Japanese Predicates.
Doctoral dissertation, Cornell University.
Pinker, S. (1991). Rules of language.
Science,
253, 530-535.
Plunkett, K., & Marchman, V. (1993). From rote learning to
system building: Acquiring verb morphology in children and
connectionist nets.
Cognition,
48, 21-69.
|
