The role of speech perception in the adaptation of loanwords

The discovery in second language acquisition that a learner’s native language (L1) has considerable influence on the perception of the learned language (L2), has been largely ignored by loanword research. Where the issue of perception is discussed, there is disagreement about the perception of segments; the role of the native phonotactics in perception, however, is agreed to be irrelevant. To gather empirical evidence concerning this issue, the adaptation of Russian initial consonant clusters by native speakers of English in production and perception is examined. The production data (repetition and orthographic representation) show high variability between clusters in terms of the preferred method of adaptation as well as a tendency of adaptations towards greater acceptability in English. In a perception study, claims that speakers analyse the incoming signal phonotactically in terms of their native language has not been confirmed, although there is evidence that Russian and English listeners perceive the cluster sequences differently. 1 BACKGROUND A major part of learning a new language is the acquisition of sounds that are not part of the learner’s native phoneme inventory. In second language acquisition research, the perception of these phonemes has been shown to be influenced considerably by the learners’ native phonology and thus to be a major source of mistakes and difficulties with the phonological system of the learned language. For example, learners tend to project new phones onto phonemes of their first language, e.g. French learners often perceive English [ ] as /s/ (Rochet 1995; for further examples see Best 1995, Flege 1995, Kuhl and Iverson 1995). Research on loanwords, on the other hand, which are also created from a clash of the native and a foreign system, has to date not taken perception into account to a great extent. Its main focus has instead been on the grammatical differences between the two languages and the mechanisms that transform the representation of the input (which is taken to be the word as it is pronounced in the donor language) to the representation of the output (the pronunciation in the recipient language). This is done without examining speech perception, i.e. the source of the input representation in the adapter’s mind. Thus research to date has only been concerned with the boxed part in figure1: