| Abstract: | In Japanese, English CVC monosyllables usually show consonant gemination after obligatory vowel epenthesis (e.g. 'put' becomes 'putto'). The katakana syllabary, which is a good reflection of pronunciation, allows us to study very quickly how a number of native speakers and learners at various levels handle novel loanwords. We show that, while learners do not geminate at as high a rate as native speakers do, they improve over years of study. More interestingly, learners use another strategy, namely vowel lengthening (e.g. 'puuto'), to represent these items, a compensatory strategy, we believe, related to their perception of the proper number of morae to be rendered in the output. We show how Broselow and Park's (1995) account of mora conservation will not handle the complexity of these data, particularly learner performance in the gemination of unstressed syllables, and we provide a variable account rather than one which suggests that parameters are set to a native speaker, learner, or mixed setting. Additionally, we show the surprising influence of gender in some areas of learner performance, a reflex, we believe, of the type of male students more typically registered in Japanese language classes at the university level. |
