Experimentally testing the epenthetic bias in loanword adaptation

In loanword adaptation, epenthesis is the favored way to make non-native sound sequences pronounceable, over other options like deletion or substitution (Paradis and LaCharité, 1997). This epenthetic bias is also apparent at the phonetic level, such as the phonologization of excrescent bursts and vocoids as full vowels (Kang, 2003; Davidson, 2007). It is possible that loanword status in and of itself induces this bias, whether or not the source form of interest would be illicit in the speaker’s native sound system. Weinberger’s (1994) Recoverability Principle suggests that second-language learners prefer to preserve or insert sound material due to less awareness of what may be expendable while retaining word recoverability. The epenthetic bias may therefore hold even for sound sequences available in a speaker’s native language but which they consider to be embedded in a word from a foreign language. The current study tests this prediction. Listeners transcribe nonce words manipulated along a [CCVC]-[CəCVC] continuum in which, crucially, both ends of the continuum are licit in their native language. Surrounding speech is manipulated between two framings of the nonce word as either an unfamiliar word in the native language or a word from a foreign language to test whether the latter framing induces a preference for transcription. This shines light on the phonetic roots of a common phonological pattern and how contextually mediated these are.