Browsing by Author "Erik Thomas, Committee Chair"

Now showing 1 - 3 of 3

The Emergence of Hispanic English in the Raleigh Community: A Sociophonetic Analysis
(2004-04-01) Carter, Phillip Martin; Mark Darhower, Committee Member; Walt Wolfram, Committee Member; Agnes Bolonyai, Committee Member; Erik Thomas, Committee Chair
Though the recent influx of native Spanish speakers to the Southeastern United States has caused sociolinguists to pay closer attention to Hispanic English, most studies have focused their attention on the adaptation of segmental features, leaving rigorous examinations of suprasegmental features vastly underrepresented. Although some studies have commented on prosodic differences between Spanish, English, and dialects of English influenced by Spanish, most of these have relied on impressionistically based observations and have avoided systematic, quantifiably based examinations. Nevertheless, Ramus et al. (1999) were able to show quantifiable differences between Spanish and English, firmly classifying the former as more syllable-timed and the later as more stress-timed. The development of the Pairwise Variability Index (PVI) by Low and Grabe (1995) provides a method for examining the degree of stress-timing or syllable-timing in a given linguistic variety. Fought and Fought (2003) used PVI to show that bilingual Chicanos in California were more syllable-timed than the adjacent English-speaking community, though only for the first five syllables of an utterance. This thesis study examines the Spanish and English of adolescent bilinguals in Raleigh, NC and applies the PVI method in order to a) report empirically quantifiable differences between the two systems b) determine the rhythmic nature of Hispanic English and c) explore possible influences of southern American English on the Spanish of immigrants to the Mid-Atlantic South. As expected, findings show a range of rhythmic productions that is best represented on a continuum, where Spanish is located on one endpoint, the English of native monolinguals on the other, and the English of Hispanic immigrants somewhere in between. This analysis provides further insights on the bilateral affects of Spanish-English contact situations. The nature of prosody as a substrate feature in emerging varieties of Hispanic English in the Mid-Atlantic South is also considered in this description.
In Search of Ethnic Cues: The Status of /ae/ and /epsilon/ and Their Implications for Linguistic Profiling
(2005-04-25) Grimes, Drew; Agnes Bolonyai, Committee Member; Walt Wolfram, Committee Member; Erik Thomas, Committee Chair
Historically, the study of African American English (AAE), perhaps the most thoroughly documented language variety in the United States, has been centered on syntactic and morphological characteristics. The phonological features that have been discussed are generally limited to consonantal characteristics such as consonant cluster reduction; thorough studies of the vowel systems of African American speakers have only begun to appear in earnest in the literature since the 1990s. Similarly, systematic studies of dialect perception, despite a long tradition in experimental phonetics, have only sporadically been employed in sociolinguistics. This thesis begins to fill both of these gaps by (1) reporting results from two ethnic identification experiments conducted in North Carolina and by (2) reporting data collected from a comparative analysis of two vowels as produced by NC black and white speakers. The vowels examined were /æ/ and /ε/, which are both reported to be raised by African American speakers. The perception experiments required a process of instrumental resynthesis of recorded speech to generate stimuli approximating a continuum of values between two dialectal vowel variants. This process, used for both /æ/ and /ε/, is based on the source-filter theory of vowel production and entails the use of the phonetics software Praat. Using LPC analysis, Praat can synthetically extract a 'filter' from a recorded vowel sound, thereby leaving an approximation of the speaker's unfiltered glottal source. This 'filter' can then be manipulated to represent different formant values and used to refilter the previously generated source signal. With this procedure, a range of experimental stimuli were created for /æ/ and /ε/, representing a continuum of vowels between the black and white variants for each. Results from experimental subjects who heard the stimuli demonstrate that for both /æ/ and /ε/, there was a significant correlation between vowel height and ethnic identification: the stimuli representing a higher vowel position were perceived more often to have been produced by an African American speaker than the stimuli representing a lower vowel position. Furthermore, this correlation seems to be stronger for /æ/ than for /ε/. The data generated by the production study corroborate these perception results. In the production study, F₁ measurements were taken from African American (AA) and European American (EA) pronunciations of the vowels /æ/ and /ε/. These measurements came from field recordings of North Carolina speakers from the North Carolina Language and Life Project archives. The results from this study show a statistically significant difference in the mean AA height and the mean EA height for both of these vowels; furthermore, the difference between the two means for /æ/ was much greater than the difference for /ε/. The impetus for this ethnic identification approach to the analysis of dialect perception comes from the reality of linguistic profiling. Potential landlords and other real estate professionals use this type of racial discrimination to deny housing opportunities to minorities inquiring over the telephone about property availability. The use of such practices is a documented reality; however, many people reject the notion that speakers' ethnicities can be identified based only on their voices. It is hoped that the empirically generated perception data presented here will provide incontestable evidence for the plausibility of linguistically based discrimination practices, and thereby help to solidify the case for legal defendants who have experienced this type of discrimination.
Regional /o/ in North Carolina: A Cartographic Analysis of a Feature of the Southern Vowel Shift
(2007-08-01) Sellers, James Clark; Walt Wolfram, Committee Member; Erik Thomas, Committee Chair; Jeffrey Reaser, Committee Member
This thesis examines sociophonetically the variation of ⁄o⁄ in North Carolina. Though often overlooked in terms of its diagnostic status, it is one of the most symbolic vowels in terms of a range of social and ethnic variables. Previous research by Kurath and McDavid (1961), Thomas (1989), and Thomas and Wolfram (2002) has found that ⁄o⁄ fronting was occurring in the late 18th Century in the northeast coastal region while in the Piedmont region ⁄o⁄ was farther backed. The centralization and lowering of the nuclei of back vowels, including ⁄o⁄, is considered to be a part of the Southern Vowel Shift (Labov 1991). The glide of ⁄o⁄ in the Southeastern US has become fronted more recently (Labov, William, Yaeger, and Steiner, 1972; Thomas 1989; Thomas 2001). Is /o/ in North Carolina conforming to the traditional specifications of the Southern Vowel Shift? Does its conformity vary regionally? How did the change spread across North Carolina? Using acoustic analysis, this paper attempts to answer these questions by looking at the speech of European Americans from twelve sites scattered throughout North Carolina. These sites roughly represent the major dialect regions of the state in an attempt to map the fronting and raising of the nucleus and glide of ⁄o⁄ regionally. Previous research has shown that the nucleus of ⁄o⁄ in North Carolina cities is currently more fronted and lowered, in accordance with the Southern Vowel Shift, than it was in the past. However, the results of this comparative analysis show that this is not universally true for North Carolinians and that the trajectory of change for ⁄o⁄ varies regionally throughout the state. For example, ⁄o⁄ fronting was previously less advanced in the western part of the state but young speakers from that area are now showing a high degree of ⁄o⁄ fronting while speakers from areas like Hyde County, a coastal area, have younger generations showing increasingly backed nuclei for ⁄o⁄. State wide, however, the range of possible values narrows among the younger speakers while the older speakers are much more varied. This variation could indicate that the possible vowel space for ⁄o⁄ is becoming more static as the feature has undergone a change and is now settling into a vowel space that is shared statewide. This study focuses on a single feature and includes both a large quantity of subjects engaged in conversational speech and a broad-based geographic analysis with a representative sample of subjects for each region. This representation allows for an in depth analysis of each speech community so that instead of representing each community as a homogenous group, intra-community variation is accurately represented in a way that allows for regional comparison. For example, the younger generation in Hyde County can be divided up into those who have a fronted glide for ⁄o⁄ and those who do not. This study looks at correlations between site specific variables and represents these correlations geographically allowing for a regional evaluation of intra-community variation. By incorporating geography into sociolinguistic inquiry this study contextualizes intra-community variation within the larger regional dialect, showing how macro- and micro-variables need to be factored into variation analysis.