This dissertation investigates segment insertion through two contextualized loanword adaptation processes: intervocalic nasal and coda [m] adaptations, with corpus and experimental data. My research focuses are: 1) to identify the phonological environments of segment insertion in the target adaptation processes in my corpus data, and propose explanations for the insertion patterns in Standard Mandarin loanwords; 2) to verify my arguments with experimental data; 3) to examine whether bilingualism affects target segment adaptations; and 4) to discuss what linguistic mechanism controls the adaptation processes.Nasal insertion in Mandarin loanwords is considered an "unnecessary repair" since both English and Mandarin allow a syllable containing consonant-vowel-nasal. However, the existing loanwords show that nasal insertion has strong correlation to the prenasal vowel quality, and the primary word stress location in the source language. To trigger nasal insertion, the English prenasal vowel has to be lax (vowel type condition) and must bear the primary word stress (stress location condition), e.g. Denis → [tan.ni02D0.s02790329]. Nasal insertion rarely occurs when the prenasal vowel is tense or a diphthong, e.g. 02C8Lina → [li02D0.na02D0]. Variable adaptation occurs when the prenasal vowel is [0259], e.g. 02C8Tiffany →[ti02D0.fan.nei] 0303 [ti02D0.fu02D0.ni02D0]. Low back [0251] never triggers nasal insertion, e.g. Ca02C8bana → [kha02D0.pa02D0.na02D0]. I propose that adaptors are sensitive to the fine acoustic cues of the prenasal vowel and stress syllabification pattern. I argue that the "unnecessary repair" is necessary for acoustic cue mapping so the input is perceptually similar to the output.Vowel epenthesis is argued to fix illicit coda [m] in Mandarin, but it is related to syllable location and the following consonant type. It occurs in word-medial and word-final coda positions. It never occurs in homorganic environments with a prenasal lax vowel, e.g. Columbia → [k026402D0.lun.pi02D0ja02D0]. However, with a prenasal diphthong or tense vowel, vowel epenthesis still appears. Variable adaptation is present when coda [m] is in word-medial position followed by an obstruent, e.g. Camden → [kha02D0.mu02D0.t0259n] 0303 [kh0259n.tun]. I propose that vowel epenthesis takes place to preserve all the segmental information, although the output may be perceived as less similar to the input due to an excess vowel. Repairing with [n]/[n] in homorganic environments not only preserves all the underlying features but also leads the output to be more perceptually similar to the input. Both modification methods suggest that coda [m] adaptation is motivated by phonological grammaticality, and acoustic cues play a relatively minor role.33 Mandarin monolingual and 24 Mandarin-English bilingual speakers participated in the experiments. Test item structures followed the corpus generalizations. For intervocalic nasal adaptation, the results from both groups are similar to each other and to the corpus data patterns: participants were sensitive to prenasal vowel quality. Similar results from the two groups further confirm that nasal insertion is more phonetically driven. For coda [m] adaptation, the results from the bilingual participants follow the Preservation Principle, even in homorganic lax vowel environments. The monolinguals repair the coda [m] through epenthesis and nasal place change by chance. I suggest that the coda [m] adaptation process is phonologically driven.This dissertation demonstrates that nasal adaptations in Standard Mandarin are contextualized. Two adaptation processes with essential differences in grammaticality are controlled by different linguistic mechanisms. Intervocalic nasal adaptation supports the perception-based loanword model, whereas coda [m] adaptation supports the phonology-based model.
Read
