Abstract
This chapter investigates how polysemy can be effectively taught in the classroom by assessing the Projection Model, which predicts that the teaching of marked forms facilitates the learning of unmarked forms. Although the Projection Model has only been assessed in limited areas so far (i.e., relative clauses by Eckman, Bell and Nelson (1988); possessive pronouns by Zobl (1985)), it has great potential in the areas of second language (L2) teaching and learning (Shirai 2012). The present study extends the Projection Model to the domain of semantics by testing it with the polysemous particle de in L2 Japanese. The participants in this study were 62 native Mandarin speakers in Taiwan and 51 native Korean speakers in Korea, who had been studying Japanese for periods ranging from 3 months to 9 years and 9 months. They were divided into two proficiency groups based on their length of study. A pre-test-/post-test design was employed to assess the effectiveness of instruction treatments. The pre- and post-tests consisted of acceptability judgments of sentences that included the eight meanings of de (location, extent, instrument, means, manner, cause, purpose and duration of time) and other particles used as fillers (e.g., ni, o). Participants were randomly assigned to one of two instruction treatment groups: (1) those instructed only on marked meanings such as duration of time and purpose and (2) those instructed only on unmarked meanings such as place and means. Marked and unmarked meanings were determined by the pre-test scores; namely, difficult meanings were classified as marked items. Instructional treatments consisted of two metalinguistic tasks: (1) naming an appropriate meaning of de (e.g., location or cause) and (2) choosing a meaning that a particular example of de denotes. Results show that generalization to the untaught meanings was only observed among the learners who were taught marked items. For Korean speakers, both higher and lower-level groups showed marginally significant improvement from pre-test to post-test on untaught meanings (paired t-test, p = 0.056, p = 0.082), although for Mandarin speakers, only the lower-level group showed marginally significant improvement (p = 0.071). None of the groups who were taught unmarked items showed significant improvement (p-level ranging from 0.347 to 0.687). These results suggest that exposure to marked/difficult exemplars may facilitate the learning of unmarked/easy ones in the domain of polysemous grammatical categories. They also indicate that the amount of input or exercises for each item is one of the key factors for teaching polysemous items effectively.
