Syllables as concatenated demisyllables and affixes

Abstract

A method is proposed to decompose English syllables into phonetically and phonotactically well motivated units, so that the complete inventory for segmental concatenation will contain at most 1000 entries and still reproduce natural allphonic variations. A syllable is decomposed into a syllable core and syllable affix(es). Syllable cores have a general form of (Ci) (Ci)V(Cf) (Cf), where V is a single vowel, and Ci and Cf are consonantal elements where Cf may be a glide (semivowel or vowel elongation). The /sp/, /st/, and /sk/ are treated as simple consonantal elements with a place specification, and the ordering of the elements within the core is strictly governed by the vowel affinity principle [Fujimura, IEEE Trans. ASSP-23, 82–87 (1975)]. The syllable core is divided into initial and final halves—demisyllables each of which can contain only one specification of the place of articulation. The final consonantal elements (such as /s/ in /tæks/) that follow a place-specified true consonant (/k/ in this case) are treated as syllable affixes. These affixes are all apical, and they observe voicing assimilation with respect to the true consonant in the core.