An Acoustic and Perceptual Study of Vowel Nasality in Taiwanese

The contrast between nasal vowels and oral vowels is phonemic in Taiwanese, however, their distributions are not entirely contrastive. For example, nasal nuclei do not occur when the coda is a nasal. It is analyzed that historically the nasal feature of the nasal coda spreads leftward to the nucleus, when the vocalic nasalization is rephonologized, the nasal coda is lost. Therefore, it has been generally concluded that there is no phonemic nasalization in the vowel preceding nasal codas in Taiwanese (Ang, 2002; Li, 1990; Lien, 2000; Wang, 1999). On the other hand, as a universal phenomenon constrained by our speech mechanism, phonetic nasalization can take place in the vicinity of nasal consonants (Cohn, 1990; Ladefoged, 1996). The issue under debate then is whether nasal vowels and oral vowels in nasal contexts are truly distinguishable, or the constraint of their distributions is simply phonological (Chung, 1996). Syllables like CṼN and ṼN are illegal in Taiwanese. To account for this phonologically, Chung posited a nasal domain where onsets and nuclei form one domain and codas form one domain (see (1)). By the percolation principle, if one segment in a domain is nasal, then the other segment in the same domain has to be nasal as well. Only one nasal domain is allowed tautosyllabically in Taiwanese, and that is why the syllable types above are not allowed. Ang (2002) challenged Chung’s division of a syllable into two domains with evidence from transyllabic nasal spreading, which presupposes the nasal spreading from nucleus to coda. He used ranking of constraints in Optimality Theory to account for nasal spreading anomalies which were not well explained by previous phonological rules. However, as suggested by Chung, to validate these phonological patterns, we need to resort to acoustic studies to see to what extent phonemic nasalization (i.e., nasal vowels) and phonetic nasalization (i.e., nasalized oral vowels) are distinguished. That is, if indeed the nasality between nasal vowels and nasalized oral vowels is categorically different to native speakers, we would be able to maintain the analyses in phonology literature. If not, the vowel in VN may well be treated the same as a “true” nasal vowel.