出版物 ｜Sophia Linguistica
Yee-ping WONG
The present study aimed to investigate how two different types of oral corrective
feedback, input-providing recast and output-pushing prompt, would differentially impact
the perception of Japanese vowel length contrasts by Cantonese-speaking learners in
Hong Kong. 54 learners were assigned to three groups and received three hours of taskbased
instructional treatment online. Perception measures included (i) a comparatively
spontaneous test measuring sensitivity in which listeners were required to identify the
contrasts and process the meaning of sentences simultaneously; (ii) an identification
test measuring boundary point and boundary width in manipulated continuum of vowel
duration. Results were compared with the baseline data of 15 native Japanese listeners in
a pre-test, post-test, delayed test plus generalization test design. The results showed that
giving a prompt had somewhat limited effect on boundary point in the post-test and the
generalization test. By contrast, the recast group thrived with significant improvement
in the sensitivity measure in the delayed test. It is apparent that recast is more effective
for trained familiar items in the long term. These results could be attributed to the fact
that recast provides both negative and positive evidence for learners’ interlanguage
development.
