Learnability, teachability, and argument structure: Adult Japanese learners' acquisition of the English dative alternation

This dissertation presents data from three tasks--novel verb elicited production, picture description, and acceptability judgments--to shed light on how adult Japanese learners of English represent the dative alternation. The experiments were designed within the framework of Pinker's (1989) semantic structure theory, according to which the complex constraints on argument structure alternations are fundamentally semantic, and are learnable largely on the basis of childrens' sensitivity to two levels of semantic criteria. These critieria are assumed to have their effect by means of broad-range and narrow-range lexical rules that operate on the semantic structures of verbs.; Predictions about adult L2 learner behavior in relation to the criteria were derived from Bley-Vroman's (1989) Fundamental Difference Hypothesis, which claims that linguistic principles not instantiated in an adult learner's L1 should not be available for acquiring the L2. Applied to adult native speakers of Japanese, the prediction was that the leaners would have access to the broad-range but not narrow-range semantic distinctions in their acquisition of the English dative alternation.; The results of the experiments clearly corroborated the learners' access to the broad-range rule, which limits the alternation to sentences in which the recipient is a potential possessor of the transferred object, but consistent results were not yielded for the narrow-range rules, which putatively act on small subclasses of closely related verbs. On two tasks, the learners demonstrated more knowledge than expected, but on the third they showed very poor semantic discrimination.; A fourth experiment was done to explore the teachability of semantics-syntax correspondences. One group of learners was given instruction on the semantic criteria underlying the dative alternation, while a second group was given comparable instruction on the locative alternation. A control group was given no relevant instruction. Although no statistically reliable differences were found, the results were in the expected direction. The group with specific instruction on the dative alternation became gradually more accurate, followed by the group with less specifically usable information on a related alternation, followed by the control group, whose judgements remained stable over the three data collection points.