Jump to the main content block
Tsing Hua Journal of Chinese Studies
ISSN 0577-9170; DOI 10.6503/THJCS


An OT Analysis of Nasal Percolation in Taiwanese

Vol. 31 No. 3   9/2001  


An Ot Analysis of nasal Percolation in Taiwanese


ANG Uijin









Key words

Taiwanese, nasal spreading, OT analysis, inner-syllbic percolation, trans-syllabic percolation, dialectal variation


   This paper examines the nasal spreading, an interesting phenomenon in Taiwanese, where a nasal feature spreads throughout the syllable when it is specified in the nucleus, but which cannot spread beyond the coda where the nasal feature is specified. Many argument have been published for explaining the asymmetry of the nasal spreading in Taiwanese, such as rightward spreading theory (Li 1992), leftward spreading theory (Wang 1997), downward spreading theory (Ang 1996), and percolation theory (Chung 1996). All of these theor ies are based on a rule-based analysis, which cannot explain all such nasal spreading behavior in Taiwanese, and this is the motivation for this paper’s reanalysis based on the constraint-based analysis and OT model.

    The downward spreading theory is elegant but cannot explain cross-syllable spreading, where the nasal feature spreads upward to the nucle us of unstressed syllables. The percolation theory is insightful. But Chung divideda syllable into two domains, [CV] and [-C], which cannot cover the nasal spreading from nucleus to coda. In this paper, the nasal domains are redefined in a hierarchical structure, and it is argued that the nasal feature specified in nucleus percolates throughout the syllable that is m- commanded by nucleus, and that nasal specified in coda percolates only in the domain that ism-commanded by coda.

    Under this head-government hypothesis, three constraints are set as follow, (1) two constraints of alignment: A LIGN (N,L), A LIGN (N,R) to define nasal domains: (2) two percolation constraints: C-V N.H. and *-V C to ensure the nasal feature spreads throughout but not beyond its own domain, (3) two faithfulness constraints: S URVIVE N AS and N AS D EP to identify the nasal realization from underlying to surface, and (4) an opacity constraint * ç - to inhibit voiceless obstruents from being nasalized. Then these constraints are ranked in their cruciality, and many candidates of different forms are valuated in tableaux respectively, whereby ill forms are filtered out and the unique well form is elected. The OT analysis of Taiwanesenasal percolation supports the argument that all constraints are violable, and that for a well-formed one, minimal violation is acceptable instead of total lack of violation.

    The second part of this paper deals with the nasal spreading across syllables, where a nasal feature spreads only rightward but not leftward, for this directionality , Chung 1996 argued that it is ‘spreading’ but not ‘percolation’. But it is difficult to explain why a same nasal feature ‘percolates’ within a syllable, but ‘spreads’ across a syllable. It is argued consistently in this paper that trans-syllabic spreading is essentially also percolation., that is ,a stressed syllable opens its right boundary barrier to followed unstressed syllables, which get incorporated into the stressed syllable along with the percolated nasal feature from it.

    Trans-syllabic constraints are almost the same as those for inner syllabic percolation, except those about the expanding of percolation domain to an XP, and about the opacity of obstruents and stressed syllables. So it is necessary to replace the right hand alignment constraint of obstruents by *ç, and to add an opacity constraint for stressed syllables : *σ’.

     The last part of the paper deals with variations in nasal spreading, which have been disregarded, or regarded as exceptions by tradition al linguists. In fact, a dialectal variation or a strange phonetic realization can be explained by different ranking of constraints in OT theory. In this paper, all changes in ranking have motivation, which have caused different people to adopt different str ategies to avoid generating ill- formed syllables in order to obey more highly ranked constraints, or to accept sporadic ill forms. For example, for the Japanese loanword ‘manga’ (cartoon), some people pronounce [baη31 ga11] to avoid violating the *-v C constraint, but some people pronounce [maη31 ga11] to obey S urvive N AS constraint to conserve the underlying Japanese source-pronunciation, in which [mâη] is an ill-formed syllable in Taiwanese.



Author: ANG Uijin
Genre: Article
Click Num: