ABSTRACT
Phonology is replete with theories and principles governing the process of syllable
segmentation. However, with a myriad of conflicting theories available, a major
problem facing researchers wishing to apply the syllable unit in models of speech
processing is in selecting which of these theories to use in their definition
of the syllable. In this thesis, an attempt has been made to judge the merits
of the various principles and theories of syllabification applied to French,
a syllable-timed language, by comparing the theoretical predictions of these
theories with the experimental findings of a series of psycholinguistic syllable
segmentation and perception experiments.
A number of factors have been found to influence syllable segmentation. These
include, the nature of intervocalic consonant clusters and singletons, with
preference given to minimal legal syllable onsets, excepting OBLI clusters,
which are tautosyllabic. The aperture of the vowel at the nucleus of a syllable,
with an open vowel inducing closed syllables, and vice versa. Also, orthographic
bias, which, at the onset of literacy can influence syllabification decisions
in metalinguistic tasks. In addition, examination of the differences between
the syllable onset and offset detection revealed that, if there is a legal syllable
onset before the first consonant of an intervocalic consonant cluster/singleton,
then there will be a high degree of ambisyllabicity for this segment.
These findings suggest that listeners are taking advantage of a number of separate
cues when segmenting speech into syllables. However, even when factoring these
effects into the segmentation responses of subjects, there is still considerable
variability in subject segmentation responses. By implementing the factors found
to influence syllabification thus far, it is possible to suggest a set of preference
rules which can predict where the boundary is likely to be located, and also
to predict in which situations the location of the syllable boundary is likely
to be most ambiguous.