臺閩語的音段錯誤研究

Abstract

以「臺閩語口誤語料庫」的2,687筆口誤為依據，本文提出音段錯誤的語言實證並重新檢驗三個相關的音韻議題，包括塞擦音的底層表式、鼻音延展的方向性、及語言干擾。第一，口誤驗證支持Clements（1998）的觀點，塞擦音的底層是帶摩擦徵性[+strident]的塞音，因此塞擦音呈現底層與塞音同調而表層與擦音同調的邊際效應。第二，口誤驗證鼻音徵性是音節層次的自主音段，可在音節內或跨音節雙向鼻音延展，反映了共時音韻限制，如聲母韻核間的鼻音和諧以及韻核韻尾間的隔閡，是歷時發展的結果。第三，語言干擾導致的口誤反映音節內部組成之間不同的結構緊密度。從涉及語碼混雜的口誤，韻母的借音頻率最高，聲母最低。此外，在聲母、韻母和聲調之間，前二者的結構較為緊密。

Based on 2,687 errors from the “Taiwanese Speech Error Corpus,” this study provides external linguistic evidence of segment errors to reexamine three phonological issues, including the underlying representation of affricates, the directionality of nasality spreading and language interference. First, as proposed by Clements (1998), the underlying representation of affricates is testified to be a stop with [+strident] feature which is also seen in fricatives, which results in the edge-effect of affricates for patterning phonologically with stops and phonetically with fricatives. Second, nasality has been proved to be an autosegmental feature spreading bi-directionally within or across syllables, which suggests that synchronic phonotactic constraints, such as nasality harmony in the Onset-Nucleus domain and barrier constraint in the Nucleus-Coda domain, result from diachronic development. Third, phonological blending triggered by language interference reveals structural affinity between subsyllabic elements. From errors of code mixing collected in this corpus, the rhyme and the onset are respectively the most and the least susceptible to replacement. Among the three major subsyllabic compositions, namely the onset, the rhyme, and the tone, the former two hold closer affinity.