An integrated acoustic and phonological investigation of weak syllable omissions

Abstract

This dissertation is an in-depth study of weak syllable omissions from polysyllabic words, a phenomenon seen in three quite different English-speaking populations: young children with normally developing language, older children with specific language impairment (SLI), and adults with types of non-fluent aphasia. The omissions result in forms such as nana for banana, or raffe for giraffe. I first review the main theoretical accounts of weak syllable omissions, which implicate input (perception), production, or grammar (phonology). After concluding that perceptual accounts are inadequate, I propose four studies based on production and grammar accounts of omissions by the populations of interest. Specifically, the studies ask two questions: Are syllables truly deleted? And what is the mechanism behind syllable omissions? Three hypotheses address these questions. The Structure Reduction Hypothesis states that syllables are truly deleted, whereas the Generic Trace and Variable Trace Hypotheses hold that omitted syllables leave a measurable trace. On the Generic Trace Hypothesis, the trace is of generic length, independent of the subsyllabic information, while on the Variable Trace Hypothesis, variation in subsyllabic length affects the trace. In three experiments, children with normal language and SLI were asked to repeat sentences that included words with strong-weak stress patterns and weak-strong-weak patterns, which children reduced to strong-weak outputs. Acoustic analyses of the non-reduced and reduced outputs revealed that children left an acoustic trace in the latter, in the form of durational lengthening. Lengthening did not vary based on the subsyllabic content of the omitted syllable, supporting the Generic Trace Hypothesis. In addition, the child studies and a word repetition study of adults with aphasia revealed certain phonological factors bearing on the rates of syllable omissions. Based on the four studies, I propose a preliminary model of weak syllable omission that contains factors triggering omissions, multiple paths from the phonology to the phonetic realization, and a developmental continuum of strategies used by children to reach the adult target. In concluding, I suggest that instrumental acoustic analyses are a crucial component to any study addressing variation in children's productions, and I suggest implications of the research for current discussions of the phonetics-phonology interface.