Mok & Hawkins Syllabifications of the /st/ cluster and vowel-to-vowel coarticulation in English Peggy Mok* & Sarah Hawkins# *Department of Linguistics and Modern Languages, The Chinese University of Hong Kong; peggymok@cuhk.edu.hk #Department of Linguistics, University of Cambridge, U.K.; sh110@cam.ac.uk Introduction This paper investigates how different syllable boundaries involving the /st/ cluster in English affect vowel- to-vowel (v-to-v) coarticulation. Although several models of coarticulation offer accounts of vowels and consonants in different syllable positions (e.g. Articulatory Phonology (Browman & Goldstein, 1988; 2000) and ‘carrier’ models of coarticulation (e.g. Öhman, 1966)), the effects of syllable structure on v-to-v coarticulation remain poorly understood. Most studies on v-to-v coarticulation only deal with one syllable type, namely open syllables (though see Modarresi, Sussman, Lindblom & Burlingame, 2004). However, notwithstanding the lack of a clear definition of a phonetic syllable, many studies show that syllable onset and coda are different acoustically, articulatorily, typologically and perceptually. Acoustic studies indicate that onset consonants are longer and exhibit stronger cohesion with tautosyllabic vowels than do coda consonants (e.g. Byrd, 1996; Sussman, Bessell, Dalston & Majors, 1997). Articulatory studies show that syllable onset and coda consonants coordinate differently with the vowels, and that gestures for onset consonants are stronger and more distinct than those for coda consonants (e.g. Browman & Goldstein, 1988; Krakow, 1999). Onset consonants are more frequent in the world’s languages, and are more distinguishable than coda consonants in noise (Redford & Diehl, 1999). All this suggests that syllable onset and coda consonants have different coordination with the vowel, which might therefore be expected to affect v-to-v coarticulation. The /st/ cluster in English can be syllabified in three ways: onset /#st/, heterosyllabic /s#t/ and coda /st#/ (where # denotes a syllable boundary). It was hypothesized that the onset /#st/ should allow the least v-to-v coarticulation because onsets are stronger and more stable, followed by the heterosyllabic /s#t/ and the coda /st#/ because coda consonants are the most variable. The /st/ cluster was chosen for investigation because it is homorganic, thus reducing conflicting influences of intervocalic consonants on formant transitions. Method Six native speakers of Southern British English (two male four female) were recorded reading sequences of two real monosyllabic words involving the /st/ cluster embedded in carrier phrases. Three sets of materials were collected: onset /CV 1 #stV 2 C/ (e.g. bar steed), heterosyllabic /CV 1 s#tV 2 C/ (e.g. pass teat) and coda /CV 1 st#V 2 C/ (e.g. past east). Three vowels, /ɑ/, /i/ and /u/ were used in all possible combinations. Focus stress was induced on the non-target syllables by the carrier phrases (Not a x x , but a x x again.). Altogether 324 sequences were used (3 syllable structures × 3 target vowels × 3 context vowels × 2 stress positions × 6 repetitions). F1, F2 and F3 frequencies were measured from LPC spectra (25 ms hanning window, supplemented by DFT spectra) at several temporal locations: 1) close to the offset of periodicity in V 1 or onset of periodicity in V 2 for anticipatory and carryover coarticulation respectively; 2) 26 ms after the /t/ burst (during the aspiration), in order to measure carryover coarticulation at a comparable place in the articulatory trajectory during heterosyllabic /CV 1 s#tV 2 C/ sequences. All formant frequencies were normalised before being submitted to statistical analysis. Intervocalic durations were also measured. Results and discussion Results show that heterosyllabic /s#t/ sequences had the longest intervocalic duration presumably because they contain a syllable/word boundary. The intervocalic durations for onset /#st/ and coda /st#/ are generally Laboratory Phonology 11 95 LabPhon11 abstracts edited by Paul Warren Wellington, New Zealand 30 June - 2 July 2008 Abstract accepted after review