This dissertation examines word-prosody of ciTonga, a Malawian Bantu language spoken by lakeshore people of northern Malawi. It is argued that the real word-prosody in this language (and perhaps many Bantu languages) revolves around the idea of Strong Accent Constituency, power relations between segments, syllables and between lower and higher prosodic categories as determined by Universal Guidelines such as SONORITY, FINALITY, EDGENESS and PROSODIC HIERARCHY as well as constraints which favour language- or context-specific Strong Accent Constituents (PROSODIC STEM, ACCENT FOOT, STEM-SYLLABLE1, PENULTIMATE SYLLABLE or FINAL SYLLABLE). Tone and prosodic morphemes such as Minimal Prosodic Words and Reduplicative Prosodic Morphemes also seem to be heavily regulated by Strong Accent Constituency. The empirical bases are three speech styles found in ciTonga (Nkhata-Bay Variety) namely, formal, common, and elderly speech styles. It is one of several understudied and endangered languages in Malawi. This study therefore is in line with one of the goals of the University of Malawi's Centre for Language Studies, where this candidate serves as a member, which is to prioritize research activities on such languages. The candidate is a native speaker of ciTonga and , as such, he is primary source of most of the data. Other methods such as elicitation and focus group discussions were conducted with informants not only to get to the bottom of the matter, but to also understand social issues underlying language variation. The dissertation has been presented in 6 chapters. Chapter 1 presents introductory remarks. These include the problem statement, a note on methodology, summary of findings, theoretical precedents, and, finally, organization of the dissertation. Chapter 2 presents basic facts about the language under study. These include language classification, previous works on ciTonga, speech sounds, the syllable, tone, as well as nominal and verbal morphology. Chapter 3 presents a proposal for the theory of Strong Accent Constituency. It presents the data on vowel and consonant deletion which motivates this theory analysis. Then attempts are made to account for the facts in terms of stress-accent theory and Downing's (2006b) Morpheme-Based Templates Theory both of which are found to be slightly problematic to account for segment deletion and preservation patterns exhibited in ciTonga. Finally, the chapter introduces the theory and accounts for the facts in terms of Strong Accent Constituent Theory. Chapter 4 presents formally the theory of Interaction between Tone and Strong Accent Constituents. The chapter presents the data on tone assignment in basic verbs, simple past tense verbs and present progressive aspect verbs which motivate this type of theory analysis. Attempts are then made to account for the facts in terms of tone alignment theory (as argued by Mtenje 2006), autosegmental accent (Clements and Goldsmith 1984) and 'pitch-accent' or accentual properties of tone in Bantu languages (as hinted upon by Downing 2004). All these theoretical perspectives are found to be slightly inadequate to account for tone distribution patterns in ciTonga. On the other hand, a theory based on Interaction between Tone and Strong Accent Constituents is shown to account for the facts slightly better. Chapter 5 presents formally proposals for Strong Accent Constituent-Based Templates as a theory of morphology-prosody interfaces in ciTonga and perhaps many other Bantu languages with a Strong Accent Constituent system. It presents the data on general phonological words, Minimal Prosodic Words and Reduplicative Prosodic Morphemes. It then reviews two competing theories in the literature within the Generalized Templates Theory namely, the Prosodic Hierarchy-Based Templates Theory and the Morpheme-Based Templates Theory both of which have a goal to account for morphology-prosody interfaces. Both these theories are shown to be slightly inadequate to account for parameters exhibited by phonological words in ciTonga. On the other hand, it is suggested that a theory of Strong Accent Constituent-Based Templates may account for the facts slightly better. Chapter 6 summarizes and concludes the dissertation.

In this dissertation, I examine the effects of lexical accent on the perception of intonational prominence in Japanese. I look at how an F0 accent peak is perceived relative to another flanking F0 peak in the same utterance with respect to perceived intonational prominence. Through four experiments, I show that the lexical prosodic structure plays a significant role in the perception of intonational prominence. I first show that two distinct perceptual processes are at play in the perception of relative perceived prominence in Japanese: accentual boost normalization and downstep normalization . Accentual boost normalization normalizes the accentual boost of an accented word. In this process, the extra F0 boost assigned by a lexical accent does not count as part of the F0 peak's excursion that contributes to the perceived prominence of the F0 peak. I demonstrate that when an accented word and an unaccented word are perceived as having the same prominence, the accented word has a higher F0 peak value than the unaccented word does. Downstep normalization compensates for the production effect of downstep, a pitch range compression phenomenon after a lexical accent. Experiments show that for an F0 peak to be perceived as having equivalent prominence to a preceding F0 peak, the second peak is always lower in F0 when the first word is accented than when it is unaccented. This suggests the existence of a perceptual process that normalizes the effect of downstep. I then examine the nature of accentual boost normalization and downstep normalization and show that they refer to two distinct types of lexical accent property when they are applied. One is the phonetic F0 contour shape that is characteristic of accented words. The other is the phonological lexical accent information that is uniquely specified for accented words. The experimental results show that the perceptual effects of the normalization processes are seen when only the phonological lexical accent information of a word is present with its F0 contour shape being ambiguous as well as when the same word is acoustically manipulated into different F0 contour shapes.

The objective of this dissertation is to develop speech processing methods that enable without altering their identity. We envision accent conversion primarily as a tool for pronunciation training, allowing non-native speakers to hear their native-accented selves. With this application in mind, we present two methods of accent conversion. The first assumes that the voice quality/identity of speech resides in the glottal excitation, while the linguistic content is contained in the vocal tract transfer function. Accent conversion is achieved by convolving the glottal excitation of a non-native speaker with the vocal tract transfer function of a native speaker. The result is perceived as 60 percent less accented, but it is no longer identified as the same individual. The second method of accent conversion selects segments of speech from a corpus of non-native speech based on their acoustic or articulatory similarity to segments from a native speaker. We predict that articulatory features provide a more speaker-independent representation of speech and are therefore better gauges of linguistic similarity across speakers. To test this hypothesis, we collected a custom database containing simultaneous recordings of speech and the positions of important articulators (e.g. lips, jaw, tongue) for a native and non-native speaker. Resequencing speech from a non-native speaker based on articulatory similarity with a native speaker achieved a 20 percent reduction in accent. The approach is particularly appealing for applications in pronunciation training because it modifies speech in a way that produces realistically achievable changes in accent (i.e., since the technique uses sounds already produced by the non-native speaker). A second contribution of this dissertation is the development of subjective and objective measures to assess the performance of accent conversion systems. This is a difficult problem because, in most cases, no ground truth exists. Subjective evaluation is further complicated by the interconnected relationship between accent and identity, but modifications of the stimuli (i.e. reverse speech and voice disguises) allow the two components to be separated. Algorithms to measure objectively accent, quality, and identity are shown to correlate well with their subjective counterparts.

This study investigates the attitudes of standard accented Turkish speakers towards non-standard Kurdish accented speakers of Turkish. Given the fact that there are strict language policies in Turkey, this paper analyzes the effect of such standard language ideologies on listener attitudes using a mixed-methods design. The study included 50 Turkish participants with ages ranging from 19 to 51. Participants completed a survey with 21 questions and could volunteer to also participate in an interview. The survey asked about biographical data, evaluations of various speakers and ratings of accents of Turkish. Using a matched guise technique in the survey, a Kurdish accented speaker was recorded both in standard and non-standard accented Turkish. To explore possible differences based on age, survey responses were divided into groups with young adults (under 30) and adults. For qualitative data, 13 respondents were interviewed to explore their language ideologies. The results show that the Kurdish accented speaker received the lowest scores among all the speakers in the survey and were perceived negatively in all categories such as pleasantness, correctness and educatedness When the standard accent was attained by the same speaker, the ratings increased. The attainment of the standard accent also effected the identification of the speaker as respondents identified the Kurdish speaker as “Kurdish” when they heard the non-standard and “Turkish” when they heard the standard accent. During the interviews, One nation-one language ideologies which was promoted in the country and the standard language ideologies have been observed.

Elicited imitation (EI) assessment has been shown to have value as an inexpensive method for low-stakes tests (Cox & Davies, 2012), but little has been reported on the effect L2 accent has on test-takers' ability to understand and process the test items they hear. Furthermore, no study has investigated the effect of accent on EI test face validity. This study examined how the accent of input audio files affected EI test difficulty as well as test-takers' perceptions of such an effect. To investigate, self-reports of students' exposure to different varieties of English were obtained from a pre-assessment survey. A 63-item EI test was then administered in which English language learners in the United States listened to test items in three varieties of English: American English, Australian English, and British English. A post-assessment survey was then administered to gather information regarding perceived difficulty of accented prompts. A many facet Rasch analysis found that accent affected item difficulty in an EI test with a separation reliability coefficient of .98—British English being the most difficult and American English the easiest. Survey results indicated that students perceived this increase in difficulty, and ANOVAs between the survey and test results indicated that student perceptions of an increase in difficulty aligned with reality. Specifically, accents that students were “Not at all Familiar” with resulted in significantly lower EI test scores than accents with which the students were familiar. These findings suggest that prompt accent should be carefully considered in EI test development.

No description available.

No description available.

The issue of identity has generated a significant amount of research in recent years. In this literature review, the relationship between learner identity and accent is explored, specifically the learner’s cultural affiliation and identity and the link to his/her accent acquisition in a second or foreign language. Social networks, motivation, L1 use, socio-cultural knowledge, discrimination and power relations, and anxiety are all shown to affect parts of the learner’s cultural identity, which, in turn, may influence his/her accent in the L2. / text

abstract: The speech of non-native (L2) speakers of a language contains phonological rules that differentiate them from native speakers. These phonological rules characterize or distinguish accents in an L2. The Shibboleth program creates combinatorial rule-sets to describe the phonological pattern of these accents and classifies L2 speakers into their native language. The training and classification is done in Shibboleth by support vector machines using a Gaussian radial basis kernel. In one experiment run using Shibboleth, the program correctly identified the native language (L1) of a speaker of unknown origin 42% of the time when there were six possible L1s in which to classify the speaker. This rate is significantly better than the 17% chance classification rate. Chi-squared test (1, N=24) =10.800, p=.0010 In a second experiment, Shibboleth was not able to determine the native language family of a speaker of unknown origin at a rate better than chance (33-44%) when the L1 was not in the transcripts used for training the language family rule-set. Chi-squared test (1, N=18) =1.000, p=.3173 The 318 participants for both experiments were from the Speech Accent Archive (Weinberger, 2013), and ranged in age from 17 to 80 years old. Forty percent of the speakers were female and 60% were male. The factor that most influenced correct classification was higher age of onset for the L2. A higher number of years spent living in an English-speaking country did not have the expected positive effect on classification. / Dissertation/Thesis / Appendix C / Appendix E / Appendix F / Ph.D. English 2013

L’élégie chez Rousseau rassemble les époques, elle rend le sens étymologique [l’élégie thrène], celui de l’époque romaine [l’élégie d’amour] et le sens moderne [l’élégie tendre de l’amitié] où on parle plutôt de tonalité élégiaque. Elle rappelle à plusieurs égards le genre classique illustré principalement par Tibulle, Properce et Ovide. Rousseau semble emprunter des motifs à l’héritage antique sans pour autant s’y assujettir. Le ton de la plainte ou du chant est dicté par des personnages vertueux. L’élégie chez Rousseau échappe au stéréotype et à la « spécialisation » : elle n’est pas dans le sillage du modèle pour lequel elle fut réputée, et dont la matière est purement funèbre ; elle n’est pas, également, un paradigme de l’élégie romaine qui s’est associée presque exclusivement à l’expression de la passion amoureuse, elle n’est pas limitée à l’expression tendre de l’amitié. Elle n’est pas seulement une esthétique. L’élégie chez Rousseau est l’ensemble de toutes ses manifestations. C’est l’expression d’une attitude. / Rousseau’s elegy deals with the themes of the lament for the death, of the absence and of the loss of one beloved. Rousseau seems to be influenced by the latin poets such as Tibulle, Propers and Ovide. Nevertheless, he is distinguished from them by the way, with which he treats the motifs. These are submitted to the nature of the characters in the novel of Julie ou la Nouvelle Héloïse. In addition, Rousseau uses these motifs in different other contexts. Then, elegy becomes a source of "tools" to say an idea or a point of view. Rousseau’s elegy is more than a statement of complaints or a sad love story: it is an art of writing.