Loanword adaptation of Japanese consonants in Kavalan

This paper is a first attempt to investigate loanword adaptation of Japanese consonants in Kavalan. Based on first hand data, the paper shows that when a Japanese consonant is adapted as the closest Kavalan consonant, the manner features are of higher phonological weight and are more faithfully retained than the place and voicing features. It is shown that for adaptations involving a change in the place of articulation, the change is generally minimal and confined within the same major place feature. This is except for the [φ] > [h] and the [ç] > [h] mappings which involve a change from [Lab] and [Cor] to [Phar]. It is argued that although the [φ] > [h] and the [ç] > [h] mappings may seem to suggest that loanword adaptation in Kavalan is based on phoneme-to-phoneme mappings, the adaptations of Japanese affricates and the word-medial nasal codas suggest otherwise; the change is in fact driven by place markedness. Together with the fact that native Kavalan phonotactics also plays a role in shaping the non-native sounds perceived, this paper argues that Kavalan loanword adaptation is best accounted for by a loanword theory which takes into account both perception and phonology.