Non-Native Listeners’ Representations of Within-Word Structure Takashi Otake*, Kiyoko Yoneyamat and Hideki Maki$ *Department zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA of English, Dokkyo University, I-I Gakuen-cho, Soka, Srritama, 340 JAPAN jDeparttner]t of Linguistics, The Ohio State Uni~~ersity, 222 Oxley Hall, Coluntbus, ON 43210 $Department of Japnnese Studies, Salem-Teikyo lJni\tersify, Salem, WV 26426. Abstract: This study investigated how non-native listeners learning Japanese as a second language represent within-word structure of Japanese spoken words, focusing on two kinds of phonological units, syllables and moras. Three groups of American subjects, whose experience with Japanese varied, were tested with an explicit segmentation task. The results showed that irrespective of learning experience, they preferred syllable-based representations. However. it was also found that there was a tendency for the learning cxpericnce to facilitate sensitivity to moras. zyxwvutsrqponmlkjihg INTRODUCTION Human listeners may form conscious representations of potential within-word structure in terms of certain phonological units. It is uncertain whether there exists any universal phonological unit to represent the structure of words. In order to explore this question, we have conducted some experiments with an explicit segmentation task, asking listeners to find a viable within-word division point. Our earlier study examining monolingual listeners of Japanese and English with their native materials showed that Japanese listeners may be sensitive to several phonological units to represent word-internal structure, while English listeners were sensitive only to syllables (1). Our more recent work investigating monolingual listeners of Japanese and English with both native and foreign materials showed that both listener groups chose the same phonological unit to represent foreign materials, suggesting that the way that human listeners represent within-word structure of foreign input may be monolingual by nature (2). In this respect, it is language-specific as reported in speech segmentation studies (3). Now another question is whether foreign language learning experience enables learners to develop alternative phonological units. The present study examined whether American-English speakers learning Japanese could be sensitive to moras to represent within-word structure of spoken Japanese, comparing three groups of subjects whose Japanese learning experience varied. METHOD Mnteriuls: Materials consisted of 50 stimuli and 100 fillers. Stimuli were either 4 or 5 mora words, each of which had a nasal as a coda in the second syllable from the word onset (e.g. burmko). All the words were either nouns, verbs or adverbs, which could be represented either in kana and kanji. S&&fs: Subjects were 90 native American English students who were taking Japanese courses at Ohio State University, Salem-Teikyo University or Brigham Young University. They were divided into three groups, depending upon their Japanese learning experience: Beginning, Intermediate, and Advanced. Beginning and intermediate subjects learned Japanese for less than 1 year and for between 2 and 4 years, respectively. Advanced subjects had lived in Japan for a period of one and a half years to two years as missionaries. No one reported any hearing difficulties. AI1 subjects received a small payment for their participation. Procedure: Materials were recorded on tape. Each word was repeated twice. They were presented to subjects through headphones. Subjects were tested in groups of up to 5 in a quiet room. They were provided with a transcript of the words written in Roman characters in order of presentation. They were instructed to mark, for each word, the second viable division from the beginning of words on the transcript. They were instructed first to listen to the word, then to decide upon their preferred segmentation, and only then to look at the transcript. The words were presented every two seconds, allowing subjects no time to reflect over possible alternative responses. 2067