January 2003
·
6 Reads
this paper, depends on the incorporation of a knowledge of phoneme sequence constraints. Phoneme sequence constraints are based on a knowledge of phoneme sequcuces which do not occur word-interr, ally: for example, since there are no words which end in/m g/ ad since/m g l/does not occur word-internally, a word boundary must occur after/m/ (Lamcl & Zue, 1984). Harrington, Johnson & Cooper (1987) showed that word boundary CVC sequences ore often excluded word-internally in monomorphemic words fithe pre- and post-vocalic consonants are similar: thus,/s N V N/(N = aasal),/C I V l/,/f V p/,/g V ld,/z V ,/,/sh V sh/ are all exchlded, or are at least extremely rare, word-internally in British English Received Pronunciation (rip). ]in the study discussed below, we extend the investigations of Lamel & Zue t (1984) and l [arrington et al. (1987) by developing an algorithm for the automatic identification o' word boundaries fi'om such sequences ia a contiuuous speech recogniser