ABSTRACT
A review of phonological syllabification theory reveals considerable controversy,
with a number of conflicting theories put forward to explain this process. In
this study the performance of five, French specific, syllabification algorithms
were compared and contrasted both against each other, using lexical analysis,
and against human syllable boundary placement, using a metalinguistic syllable
repetition task. Lexical analysis revealed substantial, practical, differences
in the application of algorithms, with disagreements rising along with consonant
cluster complexity. Results from the syllable repetition task showed differences
in participant's syllabification consistency due to experimental condition,
that is, syllable onset or offset detection, and the consonant cluster used
in the stimuli. Comparison between the predictions of syllabification algorithms
and human segmentation show greater agreement for algorithms based upon phonotactic
regularities than sonority. Furthermore, segmentation by maximising the length
of syllable onset, practised in most algorithms, was not reflected in our results.
Instead participants preferred single consonant onsets, apart from the case
of obstruent liquid clusters, which are considered as a single indivisible unit.