File(s) not publicly available
Epenthetic vowel quality in loanwords: empirical and formal issues
journal contribution
posted on 2023-06-08, 09:26 authored by Christian UffmannThis paper discusses vowel epenthesis in loanwords from both an empirical and a formal linguistic perspective and argues that epenthesis patterns are more complex than usually assumed with respect to vowel quality. A statistical analysis of several large loanword corpora in the languages Shona, Sranan, Samoan and Kinyarwanda reveals that the quality of the epenthetic vowel results from the complex interaction of three distinct processes, vowel harmony, local assimilation to the preceding consonant and default insertion. Claims that a default (unmarked or perceptually least salient) vowel is usually inserted are thus challenged. The results of the statistical exploration of loanword corpora are formalized in an Optimality-Theoretic approach which makes crucial reference to autosegmental representations. The relative ranking of constraints against feature insertion and against spreading decide over the preferred strategy in a language. These constraints are scalar, accounting for the preference of different strategies in different environments, and grounded in universal markedness and prominence hierarchies, thus formalizing the observation that across corpora, certain types of spreading are preferred while others are dispreferred. The role of perception is argued to play only a minor role in vowel epenthesis, given the high amount of variation both across and within languages which is not sufficiently explained by notions of salience or perceptibility alone
History
Publication status
- Published
Journal
LinguaISSN
0024-3841External DOI
Issue
7Volume
116Page range
1079-1111Pages
33.0Department affiliated with
- English Publications
Full text available
- No
Peer reviewed?
- Yes
Editors
C Uffmann, M KenstowiczLegacy Posted Date
2012-02-06Usage metrics
Categories
No categories selectedKeywords
Licence
Exports
RefWorks
BibTeX
Ref. manager
Endnote
DataCite
NLM
DC