The main aim of this thesis is to begin to extend phonetic distance measurements to the domain of intonation. Existing studies of segmental phonetic distance have strong associations with historical linguistic questions. I begin with this context and demonstrate problems with the use of feature systems in these segmental measures. Then I attempt to draw strands from the disparate fields of quantitative historical linguistics and intonation together. The intonation of Belfast and Glasgow English provides a central case study for this. Previous work suggests that both varieties display nuclear rises on statements, yet they have never been formally compared. This thesis presents two main hypotheses on the source of these statement rises: the Alignment hypothesis and the Transfer hypothesis. The Alignment hypothesis posits that statement rises were originally more typical statement falls but have changed into rises over time through gradual phonetic change to the location of the pitch peak. The Transfer hypothesis considers that statement rises have come about through pragmatic transfer of rises onto a statement context, either from question rises or continuation rises. I evaluate these hypotheses using the primary parameters of alignment and scaling as phonetic distance measurements. The main data set consists of data from 3 Belfast English and 3 Glasgow English speakers in a Sentence reading task and Map task. The results crucially indicate that the origin of the statement rises in Belfast and Glasgow English respectively may be different. The Glasgow statement nuclear tones show support for the Alignment hypothesis, while the Belfast nuclear tones fit best with the Transfer hypothesis. The fundamental differences between Glasgow and Belfast are the earlier alignment of the peak (H) in Glasgow and the presence of a final low (L) tonal target in Glasgow and a final high (H) target in Belfast. The scaling of the final H in Belfast statements suggests that the transfer may be from continuation rather than from question rises. I then present a proposal for an overall measure of intonational distance, showing problems with parameter weighting, comparing like with like, and distinguishing between chance resemblance and genuine historical connections. The thesis concludes with an assessment of the benefits that intonational analysis could bring to improving segmental phonetic distance measures.

Spelling is a tool by which one records his thoughts and ideas; therefore it is a vital part of life. To fulfill its task successfully, spelling must be accurate. Spelling is that tool by which the happenings of the past are revealed to the present and are preserved for the future. For any individual who attempts to transfer his thoughts and words by symbols onto paper, correct spelling is a prime essential. It follows, then, that to develop perfect habits of spelling in order that perfect transcriptions of thoughts might be made is the duty of the teacher. This duty has been attempted by teachers for many generations. But it is an established fact that the goal has not been reached, for there is a stupendous number of misspellings in the written work of students in high schools. Many methods have been advanced for correcting this incompetence in spelling; when these were tried, they have failed to secure the coveted goal. In some instances the cure has aggravated the disease. Successful abolishment of this handicap baffles the teaching profession. In a course in American pronunciation recently conducted at North Texas State College, the teacher presented the fact that there are factors in the English language which tend to become stumbling-blocks to the attainment of perfect spelling. For the first time it became evident to this writer that certain elements within the language might be the cause of the spelling problem. Therefore, the readings for this thesis were undertaken for the purpose of finding the logical causes for poor spelling habits and with the hope of discovering a workable remedy.

Esta dissertação tem como objetivo principal realizar uma primeira descrição do sistema fonético e fonológico da língua idaté, falada no Timor Leste na cidade de Manatuto. Esta análise foi feita por meio de um levantamento de dados com falantes nativos residentes no Brasil e com as coletas realizadas em pesquisa de campo no Timor Leste. Primeiramente foi realizada uma transcrição dos sons da língua de acordo com o IPA (Alfabético Fonético Internacional) e foi elaborado um inventário do sistema fonético-articulatório. Depois da análise fonética foi realizada uma análise fonológica para descrever como o sistema de sons está organizado visando verificar quais são os sons distintivos, juntamente com seus traços, quais são os alofones, os processos fonológicos, uma descrição da estrutura silábica e considerações sobre os traços prosódicos. A língua idaté ainda não possui uma ortografia oficial e também quase não há estudos lingüísticos sobre ela. Sabe-se que o estudo aprofundado de uma língua possibilita o desenvolvimento das pesquisas linguísticas e também contribui para o fortalecimento da identidade cultural de um povo. Este fator ainda se torna mais importante em relação a línguas pouco estudadas e que ainda não possuem nenhum registro escrito, pois com o tempo, muitas delas podem ser extintas sem terem sido nem registradas. / The main goal of this dissertation is to give a first description of the phonetical and phonological system of the language Idaté, spoken in the city of Manatuto in East Timor. This analysis was made with data elicited from native speakers of Idaté living in Brasil and with the research in field work in East Timor. First a transcription of speech sounds of the language was given, using the IPA (International Phonetic Alphabet) and an inventory of the phonetic system of Idaté was made. After this phonetic analysis, a phonological analysis was made to describe how the system of sounds is organized, being aimed to check which are the distinct sounds, as well the features, the allophones, the phonological processes, a description of the syllabic structure and the prosodic characteristics in the language. Idaté does not have an official orthographic system yet and there are not many linguistic studies about this language. It is known that a deep study about a language allows the development of linguistic research and also helps to strengthen the cultural identity of the people. This factor is more important related to languages that have few studies about and that have not written records, since such languages may die without being previously analyzed.

A entoação pode ser decomposta em componentes estruturadoras (finalização e sustentação) e, também, semântico-funcionais (foco/ênfase e acento lexical). A proposta assumida pelo grupo de pesquisa ExProsodia, baseada nos estudos de Xu e Wang (1997), permite a análise isolada de cada uma das componentes de F0. Neste estudo tratamos das componentes estruturadoras, especificamente, do estudo da finalização (F). O trabalho tem como objetivo, primeiro, analisar por meio de testes de percepção o que os sujeitos falantes do português brasileiro (doravante PB) julgam ser um final de frase assertiva. Depois, é nosso propósito analisar se os dados perceptivos fornecidos pelos ouvintes coincidem com aqueles disponibilizados pela análise automática obtida por meio da rotina ExProsodia®. É nosso intento, ainda, verificar se, no que tange à percepção dos procedimentos de segmentação frasal com propósito de finalização, é observada a sequência finalizadora proposta por Ferreira Netto (2008), acrescida da hipótese de Consoni et al (2009) e Consoni (2011). Este estudo visa integrar a coletânea de trabalhos desenvolvidos pelo Grupo de pesquisa ExProsodia - Análise automática da entoação na fala de língua portuguesa, linha de pesquisa na área de Percepção de Prosódia sob a orientação do Prof. Dr. Waldemar Ferreira Netto. Elaboramos dois testes. No primeiro teste foram apresentadas nove frases (três frases sem manipulação e seis manipuladas). As frases não manipuladas são, ao longo deste trabalho, mencionadas como frases modelos. As seis frases manipuladas foram subdivididas em dois conjuntos de três frases cada: a de frases monotonais e as frases manipuladas de modo que entre o tom médio (TM) e o tom de finalização (TF) houvesse uma distância de 7st. Nomeamos as primeiras frases como frases neutras e as segundas, frases hipóteses. No segundo teste foram apresentadas seis frases (somente as frases manipuladas, ou seja, as frases neutras e as frases hipóteses). Cada vez que o participante ouvia uma frase era solicitado a este que indicasse se a frase, de acordo com a sua percepção, estava ou não finalizada. Foram dadas duas opções de resposta aos ouvintes: sim, a frase está finalizada ou não, a frase não está finalizada. Os testes foram aplicados em 20 ouvintes de ambos os sexos, com idade entre 18 e 55 anos, todos com escolaridade entre nível médio e nível superior. Os resultados do teste 1 indiciam que as porcentagens dos ouvintes que consideram as frases neutras como não finalizadas sempre foram altas e mantiveram-se, em sua grande maioria, na casa dos 80%. Ademais, o produto da análise das frases hipóteses corrobora os dados dos trabalhos já citados. (CONSONI; FERREIRA NETTO, 2008; CONSONI et al., 2008; CONSONI et al., 2009 e CONSONI, 2011; FERREIRA NETTO, 2006, 2008; ROSA 2009). A análise do teste 2 fora efetuada por meio da teoria da detecção de sinais (TDS) (GREEN; SWETS, 1966; MACMILLAN; CREELMAN, 2005; ADBI, 2007). Ao fazer uso da TDS, pretendemos estimar a força da sensibilidade perceptiva de cada participante do teste 2. O valor dessa força é dada pelo cálculo do dee-prime. Os resultados do teste 2 apontaram dee-primes de três categorias diferentes, a saber: alto; médio e baixo. As respostas da grande maioria dos participantes situaram-se entre uma facilidade extrema e média de discriminar as frases finalizadas e as não finalizadas. / The intonation can be decomposed into structuring components (finalization and support component) and also semantic-functional components (focus / emphasis and lexical stress). The proposal assumed by ExProsodia rechearch group, based on the studies of Xu and Wang (1997), it allows separate analysis of each component of F0. In this study we treat structuring components, specifically, the study of finalization (F). The study aims, first, to analyze through perception tests which Brazilian Portuguese (henceforth PB) speakers think that is the ending of declarative sentences. It is our intent also verify if the finalization sequence proposed by Ferreira Netto (2008), plus the assumption of Consoni et al. (2009) e Consoni (2011) is observed. And lastly, our purpose is to inspect whether perceptual data match with those provided by automatic analysis obtained by routine ExProsodia®. This study aims to integrate a collection of work done by the research group ExProsodia - Automatic analysis of intonation in the speech of the Portuguese language, research line in the area of Perception of Prosody under the guidance of Prof. Dr. Waldemar Ferreira Netto. We developed two tests. In the first test, nine sentences were presented to the Brazilian Portuguese listeners (three sentences without manipulation and six manipulated). Non-manipulated sentences, throughout this work, as mentioned as models sentences. The six manipulated sentences were subdivided into two groups of three sentences each one: the monotonous sentences and manipulated sentences. The last one is manipulated so that between the medium tone (MT) and the finalization tone (FT) there is a distance of 7 semitones. We named first sentences as neutral sentences and the others as hypothesis sentences. At the second test, six sentences (only manipulated sentences, i.e. neutral sentences and hypothesis sentences) were presented to Brazilian Portuguese listeners. Each time the participant heard a sentence, he was requested to indicate whether that sentence, according to their perception, or was not finalized. Two response options were given to listeners: yes, the sentence is finished or not, the sentence is not finished. 20 listeners of both sexes, aged between 18 and 55 years, all with education among middle and senior level have participated of the test. The test results indicate that percentages of listeners who consider neutral sentences as unfinished sentences were always high and remained, mostly at around 80%. Moreover, the product of the analysis of hypothesis sentences corroborates data of the work already cited. (CONSONI; FERREIRA NETTO, 2008; CONSONI et al., 2008; CONSONI et al., 2009 e CONSONI, 2011; FERREIRA NETTO, 2006, 2008; ROSA 2009). The analysis of the test 2 has carried out by the signal detection theory (SDT) (GREEN; SWETS, 1966; MACMILLAN; Creelman, 2005; ADBI, 2007). By making use of the TDS, we intend to estimate the strength of the perceptual sensitivity of each participant of the test 2. Results showed dee-primes in three different categories, namely: high; medium and low. The responses of the majority of participants were between one extreme and mean facility to discriminate the finished and unfinished sentences.

The purpose of this study was to compare the effectiveness of the two instructional strategies for six phonic word-attack skills objectives and to focus on the following specific questions: (a) When student teachers use a basal-reader-oriented strategy (Treatment I) for teaching phonic word-attack skills, what will be the effects on elementary pupil achievement of these skills? (b) When student teachers use a systematic instructional strategy (Treatment II) for teaching phonic word-attack skills, what will be the effects on elementary pupil achievement of these skills? (c) How will the effects of these strategies compare? Written criterion-referenced phonic word-attack skills pre-tests were administered to approximately 110 third and fourth grade pupils. The validity of the criterion-referenced tests was judged by a team of reading specialists from North Texas State University. The reliability coefficients of the tests ranged from .57 to .93 and all were significant at the .01 level. This report concludes that when elementary pupil achievement of phonic word-attack skills is used as the criterion for student teacher effectiveness then training in the conscientious application of systematic instructional procedures incorporating research validated learning principles is a more effective procedure than requiring student teachers to follow the recommended procedures in basal readers. It also concludes that the ability of student teachers to affect the learning of phonic word attack skills by elementary students increases with practice; however, this ability appears to increase at a greater rate for those trained in systematic instructional procedures than for those who are trained to use the basal reader.

Esta dissertação tem como objetivo principal realizar uma primeira descrição do sistema fonético e fonológico da língua idaté, falada no Timor Leste na cidade de Manatuto. Esta análise foi feita por meio de um levantamento de dados com falantes nativos residentes no Brasil e com as coletas realizadas em pesquisa de campo no Timor Leste. Primeiramente foi realizada uma transcrição dos sons da língua de acordo com o IPA (Alfabético Fonético Internacional) e foi elaborado um inventário do sistema fonético-articulatório. Depois da análise fonética foi realizada uma análise fonológica para descrever como o sistema de sons está organizado visando verificar quais são os sons distintivos, juntamente com seus traços, quais são os alofones, os processos fonológicos, uma descrição da estrutura silábica e considerações sobre os traços prosódicos. A língua idaté ainda não possui uma ortografia oficial e também quase não há estudos lingüísticos sobre ela. Sabe-se que o estudo aprofundado de uma língua possibilita o desenvolvimento das pesquisas linguísticas e também contribui para o fortalecimento da identidade cultural de um povo. Este fator ainda se torna mais importante em relação a línguas pouco estudadas e que ainda não possuem nenhum registro escrito, pois com o tempo, muitas delas podem ser extintas sem terem sido nem registradas. / The main goal of this dissertation is to give a first description of the phonetical and phonological system of the language Idaté, spoken in the city of Manatuto in East Timor. This analysis was made with data elicited from native speakers of Idaté living in Brasil and with the research in field work in East Timor. First a transcription of speech sounds of the language was given, using the IPA (International Phonetic Alphabet) and an inventory of the phonetic system of Idaté was made. After this phonetic analysis, a phonological analysis was made to describe how the system of sounds is organized, being aimed to check which are the distinct sounds, as well the features, the allophones, the phonological processes, a description of the syllabic structure and the prosodic characteristics in the language. Idaté does not have an official orthographic system yet and there are not many linguistic studies about this language. It is known that a deep study about a language allows the development of linguistic research and also helps to strengthen the cultural identity of the people. This factor is more important related to languages that have few studies about and that have not written records, since such languages may die without being previously analyzed.

This study examined the effect of context on the association between formant trajectories and tongue and lip kinematics in the American English diphthongs /aɪ/ and /aʊ/. Seventeen native speakers of American English had electromagnetic sensors placed on their tongue and lips to record kinematic signals that were time-aligned with the corresponding acoustic recording. Speakers produced the diphthongs in isolation, in a single word rVl context, in a phrase hVd context, and in a sentence context. Kinematic data and the F1 and F2 trajectories were extracted from the middle 50% of each diphthong production. To allow direct comparison of signals with different units of measurement, all data were converted to z-scores. The z-score records were plotted together on common axes. For each tracked sensor from each diphthong production, an absolute difference between the kinematic and acoustic variables was calculated. Average z-score difference sums were calculated for each speaker's /aɪ/ and /aʊ/ production in each context, and this measure was called the Acoustic Kinematic Disparity Index (AKDI). A repeated measures ANOVA was used to test for main context effects on the AKDI, with concurrent contrasts to test for differences between the baseline (isolated diphthong) condition and the more complex phonetic contexts. The results revealed that context has a significant impact on acoustic and kinematic relationships. The sentence context resulted in the highest number of significantly different AKDI values when compared to the isolated condition, the single word rVl context resulted in the second highest number, and the phrase level hVd context resulted in the least differences. These findings suggest, therefore, that more complex phonetic contexts have a greater effect on the acoustic and kinematic relationship. These findings imply that caution is warranted in relying on acoustics to draw inferences about articulator movements in complex phonetic contexts. These results further indicate that the investigation of sounds produced in one context does not necessarily allow a straightforward generalization to other contexts.

Data-Driven Pronunciation Generation for ASR Maria Obedkova In ASR systems, dictionaries are usually used to describe pronunciations of words in a language. These dictionaries are typically hand-crafted by linguists. One of the most significant drawbacks of dictionaries created this way is that linguistically motivated pronunciations are not necessarily the optimal ones for ASR. The goal of this research was to explore approaches of data-driven pro- nunciation generation for ASR. We investigated several approaches of lexicon generation and implemented the completely new data-driven solution based on the pronunciation clustering. We proposed an approach for feature extraction and researched different unsupervised methods for pronunciation clustering. We evaluated the proposed approach and compared it with the current hand-crafted dictionary. The proposed data-driven approach could beat the established base- lines but underperformed in comparison to the hand-crafted dictionary which could be due to unsatisfactory features extracted from data or insufficient fine tuning. 1

<p><p><p>This study aimed to identify some of the strategies students used when pronouncing unfamiliar words. Questionnaires were handed out to 94 students in the 9th grade in a medium-sized Swedish town. In addition, two teachers and 13 students were interviewed. The results indicate that the students had acquired some basic knowledge about the English sound system from phonetic training in their past education. However, there seemed to be a tendency among the students to use the trial-and-error strategy to a larger extent than using tools such as phonetic transcription in order to figure out the pronunciation of a word. The results also show that the teachers did not teach planned lessons on pronunciation, but instead it was more common that they responded to errors made by students. In conclusion, the results show that the students' knowledge in pronunciation in general was limited. In addition, there seemed to be a connection between the way the students and the teachers approached pronunciation and the student's ability to solve pronunciation issues.</p><p> </p><p> </p><p> </p></p></p>

How do infants learn the sounds of their native language? Do they need to use general-auditory or language-specific mechanisms to make sense of the distributional nature of their phonetic input? To answer this question, this study investigated the neural correlates of phonetic discrimination in monolingual and bilingual infants (2-6 and 10-14 months) and adults using a new lens afforded by functional Near-Infrared Spectroscopy (fNIRS) neuroimaging. All participants heard syllables phonetically contrastive in their native English and Hindi (non-native) in an oddball paradigm while being imaged with fNIRS. Age comparisons of infant brain activation in multiple sites revealed that left Broca‟s area showed a developmental decline in response to native-language experience only. Bilateral STG showed robust recruitment at both ages in response to both stimulus languages. These findings were robust across monolinguals and bilinguals. Together, the results suggest that all infants use neural tissue predisposed for linguistic-phonetic processing in early life.