On modeling loanword phonology in Cantonese : reviewing the uniformity approach

Abstract

 Hong Kong Cantonese is well-known for borrowing numerous English words, many of which were borrowed by phonetic transliteration. Past studies have attempted to describe the patterns of phonetic transliteration using theoretical phonological models. The applications of such models by Silverman (1992) and Yip (1993, 2002) have been cited by a number of later studies in support. However, recent data of Cantonese loanwords seem to cast doubt on the reliability of the proposed models. ⟨山埃⟩ [san55 aj55] (transliterated from English ‘cyanide’) is only one of the many examples of loanwords the said models would fail to account for.

Our data suggest that while there are English phonemes that consistently transliterate to certain Cantonese phonemes (e.g. from [b] to [p]), the transliteration process is far from uniform. For example, [pʰ] in ⟨Pie⟩ [pʰaɪ] transliterates to [pʰ] in ⟨批⟩ [pʰɐj55], but [pʰ] in ⟨Pump⟩ [pʰʌmp] transliterates to [p] in ⟨泵⟩ [pɐm55]. Also, there may be some degree of influence exerted by the orthography on the phonetic forms of loanwords, a factor not considered by the reviewed models. ⟨Minced⟩ [mɪnst] transliterates to ⟨免治⟩ [min23 tsi22] in Cantonese, the first syllable of which carries a tone unexpected of past models; the reading pronunciation of the assigned character, ⟨免⟩ [min23], prevails over the one predicted by past models, suggesting that the transliteration pattern is by no means absolute, such that a rule-based phonological model would surely fail to account for a number of Cantonese loanwords.