ABSTRACT
It is argued that a rote-and-analogy paradigm, in which knowledge is organized in terms of locally defined pattern sets, is better able to account for mastery of the count-mass distinction by speakers of English than the traditional rule-and-category paradigm. Three experiments test for the type of categorical knowledge posited by the rule-and-category paradigm. In Experiment 1, participants in a concept formation task were unable to learn a rule based on the count-mass distinction. In Experiment 2, participants' performance, although improved, was still poor when they were given explicit instructions about the distinction. In Experiment 3, participants' performance levels were quite good when they were asked to make local judgments about the use of specific determiners (much and many). These results favor the rote-and-analogy paradigm over the rule-and-category paradigm.
