| |
Abstract:
Within the literature on inflectional morphology, controversy
exists about whether or not regular forms are stored in the
lexicon. Pinker (1991) posits that inflectional morphology in all
languages is subserved by a dual-route system: irregular production
("sing/sang") is based on lexical memory whereas regular forms
("walk/walked") are generated through the application of a symbolic
rule. By contrast, single-route -in particular, connectionst models
posit a uniform basis for irregulars and regulars. If regulars are
generated via a non-lexical route, then standard effects of
lexicality such as, e.g., frequency effects in retrieval, or
similarity-based generalization to nonce words, should be present
for irregular, but not regular forms.
These predictions have been repeatedly confirmed (Prasada and
Pinker, 1993; Clahsen, H., Eisenbeiss, S. and Sonnenstuhl-Henning,
I., in press). Specifically, they were born out experimentally for
nonce word generalization by Prasada and Pinker (1993) (but see
Lee, 1996).
We present two new experimental studies in which participants'
regularization of noncewords for the English past tense was
affected both by their degree of similarity to known regulars and
by the frequency of phonologically similar regulars. Moreover, we
found a frequency by similarity interaction. Specifically, where
phonological distance is small, regularization increases with the
frequency of the phonologically close regular verbs. At greater
distance, however, increased frequency gives rise to decreased
degrees of regularization.
Thus our results contradict findings such as Prasada and Pinker
(1993). At the same time, however, the interaction between
similarity and frequency provides a potential explanation of why
similarity and frequency effects have been missed in other
studies.
On a theoretical level, our results are explicable only if
regular forms are stored in the lexicon. Thus, they provide direct
evidence against the dual-route model. Furthermore, although our
results are compatible with schema-based approaches in spirit, they
are also not predicted by current versions, such as Bybee (1995),
where regular productivity is based on an abstract /ed/-schema.
Bybee, J. (1995) Regular morphology and the lexicon.
Language and Cognitive Processes,
10, 425-455.
Clahsen, H., Eisenbeiss, S and Sonnenstuhl-Henning, I (in press)
Morphological structure and the processing of inflected words.
Theoretical Linguistics.
Lee, B. (1996) On the processing of regular and irregular
inflections: the symbolist-connectionist debate revisited. In
Koster, C. and Wijnen, F. (eds)
Proceedings of the Groningen Assembly on Language
Acquisition,
University of Groningen, 7-9 Sept, 1995.
Pinker, S. (1991) Rules of language.
Science,
253, 530-535.
Prasada, S. and Pinker, S. (1993) Generalization of regular and
irregular morphological patterns.
Language and Cognitive Processes,
8, 1-56.
|
