a1 University of Pittsburgh
This is a report on a computerized experiment with a miniature linguistic system, consisting of five morphological rules and a lexicon of 98 words. Two hypotheses derived from the literature in cognitive psychology and psycholinguistics were tested: that explicit-deductive learning would be better than implicit-inductive learning for straightforward (“categorical”) rules, and that implicit-inductive learning would be better than explicit-deductive learning for fuzzy rules (“prototypicality patterns”). Implicit-inductive learning was implemented by pairing sentences with color pictures; explicit-deductive learning was implemented by means of traditional grammar rule presentation, followed by picture-sentence pairing. The findings were in the expected direction for both hypotheses, but only the first one could be confirmed through statistically significant results.
(Received August 26 1994)