|Title:||Connected Speech Processes and Lexical Access in Real-Time Comprehension||Add Dissertation|
|Author:||Mercedeh Mohaghegh||Update Dissertation|
|Email:||click here to access email|
|Institution:||University of Toronto, Department of Linguistics|
|Linguistic Subfield(s):||Phonetics; Phonology; Psycholinguistics;|
|Abstract:||Connected speech can entail variability in the production of speech sounds. This can in turn create ambiguity at the lexical level. For example, the word bean in ‘bean box’ can sound like beam due to the phonological process of coronal-to-labial place assimilation. Previous studies have shown that listeners can perceptually compensate for place assimilation, but the extent of compensation, as well as the factors that guide this process, are a topic of ongoing debate.
This thesis explores listeners’ compensation for place assimilation from several novel perspectives. One specific concern involves potential differences across the sound classes affected by place assimilation (nasal versus oral stops), and whether these sound classes show similar patterns of compensation when the triggering phonological context (i.e., a word-initial labial consonant following the assimilated sound) is or is not present. A second issue is the extent to which effects of compensation are observed for these sound classes in the early moments of word recognition, and a third consideration is whether the English lexicon is structured in a way that minimizes confusion. An acoustic analysis, two word identification experiments, and two experiments using an eye tracking methodology combined with a priming paradigm are used to examine the production and perceptual processing of unassimilated and assimilated word-final nasal and oral stops. A corpus analysis is also conducted to explore the structure of the lexicon with respect to the cases where place assimilation might result in lexical ambiguity.
I demonstrate that complete compensation for assimilation may be less likely to occur than previously assumed. However, the phonological context information clearly plays a role in compensation for assimilation even when the degree of assimilation is very strong. Further, the results of priming manipulations suggest that the precise nature of compensation may vary across sound classes. Finally, the structure of the lexicon seems to reflect the potential for confusion that results from coronal-to-labial assimilation in nasal and oral stops. Together, the findings suggest that, in addition to general auditory processing, inferential mechanisms and the statistical patterning of sounds within the lexicon play important roles in facilitating the recognition of assimilated words.