The Phonetics-Phonology-Interfaced Analysis of Prosodic structure in english: focused on syllable weight

Abstract

Since the syllable was introduced as a phonological unit, there have been many discussions on the structure of syllable. Among many theories (e.g., sonority, the CV tier, and mora), we adopt the mora theory in this dissertation to explain syllable weight. In English, there exists the asymmetry between the word-internal coda and the word-final coda. Three main theories on this asymmetry are discussed in this dissertation: word-final C extrametricality, catalectic final syllable, and word-final mora-sharing C. This dissertation argues for the mora-sharing word-final consonant to explain the asymmetry between the word-internal coda and the word-final coda. In previous studies, it has been proven that if a word-final coda shares a mora with the preceding vowel, the duration of the mora-sharing vowel is shortened. However, little attention has been drawn to the duration of the mora-sharing consonant. This dissertation finds that the duration of the mora-sharing consonant is also shortened compared to the non mora-sharing consonant (word-internal coda). For example, the duration of the /t/ in let us is shorter than that of the /t/ in lettuce. It is argued in this dissertation that the duration of the /t/ in let us is shorter compared to the duration of the /t/ in lettuce because the /t/ in let us is sharing a mora with the preceding vowel. Therefore, the hierarchical structure of let us should be different from that of lettuce. This dissertation also argues that the syllable plays a critical role in prelexical classification in English. In order to access the meaning of any lexical unit, a recognizer must be aware of where the unit begins because speech signals are continuous. There are two possible solutions to deciding where the lexical unit begins: matching the arbitrary speech signals to stored acoustic templates and undertaking prelexical classification. To achieve greater efficiency, a human recognizer undertakes a prexical classification of the speech signal. English is known to be a stress-timed language. Therefore, English is sensitive to the stress, some phonological phenomena being more sensitive to the foot than the syllable. For example, stress clash and shift occur at the foot level, not syllable. Nevertheless, it is shown in the dissertation that listeners tend to prefer the syllable in terms of prelexical classification. There is also some evidence presented by some researchers (Cutler et al., 1983; Mehler et al., 1981, 1986; Noriss et al., 1985, 1988 among others) that the syllable functions as a fundamental perceptual unit. For example, both balance and balcony begin with the same three segments bal- . However, ba- is identified faster than bal- in balance because ba- (not bal-) constitutes the syllable. By the same token, bal- is perceived faster than ba- in balcony because bal- is the syllable of balcony. This dissertation also shows that when English native speakers listen to the two-word readings that sound like one word (e.g., mark it sounds like market), they tend to perceive them as the one-word readings (e.g., market).