Effects of orthography presentation and loanword frequency on L2 speech shadowing

It has been demonstrated that a speaker aligns their speech production with a model talker when shadowing model speech stimuli. The current study explores the Japanese English learners’ shadowing of the American English vowels /æ, ɑː/. Our specific interests are in the effects of orthography presentation and loanword frequency on the formant values of the two vowels Japanese English learners produce when shadowing model speech stimuli. Twenty-two Japanese English learners participated in a shadowing experiment, in which they shadowed native English speech stimuli while and without seeing written forms, and, following the shadowing experiment, they subjectively rated how often they use the English words as loanwords in daily Japanese speech. Our results suggest that the outcome of their imitation during the shadowing session is influenced by the presence of written forms. In particular, it was found that Japanese English learners produce /ɑː/ with more Japanese-like formant values when seeing written forms than without seeing written forms, and they produce /æ/ with more English-like formant values when seeing written forms than without seeing written forms. In addition, we found weak evidence that shadowed tokens are influenced by loanword frequency. These findings are discussed in terms of implications for episodic memories, category activation, spreading-activation, and production biases.