<OT> New Posting: ROA-958

[roa at ruccs.rutgers.edu]

ROA 958-0308

Bridging the gap: MiniCorp analyses of Mandarin phonotactics

James Myers <lngmyers at ccu.edu.tw>

Direct link: http://roa.rutgers.edu/view.php3?roa=958


Abstract:
Most phonological research is a form of informal corpus
linguistics, using corpora of preexisting lexical items
(e.g. dictionaries). Though traditional phonological corpus
analyses are not explicitly quantitative, they do rely on
type frequencies to test claims about regularity vs. exceptionali
ty and systematic vs. accidental gaps. This paper introduces
a software tool, MiniCorp (www.ccunix.ccu.edu.tw/~lngproc/MiniCor
p.htm), that attempts to bridge the gap between this traditional
logic (as applied in the Optimality-Theoretic framework)
and truly quantitative corpus analysis. Unlike most OT software,
MiniCorp is not an automatic grammar learner, but rather
it tests the statistical reliability of proposed OT constraints
and their proposed ranking using standard methods in categorical
data analysis (Poisson regression). MiniCorp also includes
tools to simplify the annotation of corpus items. These
features are demonstrated in tests of an analysis of a phonotacti
c pattern in Mandarin.

Comments: WECOL 2007; A4; Doulos SIL
Keywords: corpus linguistics, quantitative linguistics, computational linguistics, Mandarin Chinese
Areas: Phonology,Computation
Type: Conference Proceedings Chapter

Direct link: http://roa.rutgers.edu/view.php3?roa=958