Cette thèse étudie l'intégration de repères phonétiques dans la reconnaissance automatique de la parole (RAP) continue à grand vocabulaire. Les repères sont des événements à temps discret indiquant la présence d’événements phonétiques dans le signal de parole. Le but est de développer des détecteurs de repères qui sont motivés par la connaissance phonétique afin de modéliser quelques événements phonétiques plus précisément. La thèse présente deux approches de détection de repères, qui utilisent l'information extraite par segments et étudie deux méthodes différentes pour intégrer les repères dans le décodage, qui sont un élagage basé sur les repères et une approche reposant sur les combinaisons pondérées. Alors que les deux approches de détection de repères présentées améliorent les performance de reconnaissance de la parole comparées à l'approche de référence, elles ne surpassent pas les prédictions phonétiques standards par trame. Ces résultats indiquant que la RAP guidée par des repères nécessite de l'information phonétique très hétérogène pour être efficace, la thèse présente une troisième méthode d'intégration conçue pour intégrer un nombre arbitraire de flux de repères hétérogènes et asynchrones dans la RAP. Les résultats indiquent que cette méthode est en effet en mesure d'améliorer le système de référence, pourvu que les repères fournissent de l'information complémentaire aux modèles acoustiques standards. / This thesis studies the integration of phonetic landmarks into standard statistical large vocabulary continuous speech recognition (LVCSR). Landmarks are discrete time instances that indicate the presence of phonetic events in the speech signal. The goal is to develop landmark detectors that are motivated by phonetic knowledge in order to model selected phonetic classes more precisely than it is possible with standard acoustic models. The thesis presents two landmark detection approaches, which make use of segment-based information and studies two different methods to integrate landmarks into the decoding, which are landmark-based pruning and a weighted combination approach. While both approaches improve speech recognition performance compared to the baseline using weighted combination of landmarks and acoustic scores during decoding, they do not outperform standard frame-based phonetic predictions. Since these results indicate that landmark-driven LVCSR requires the integration of very heterogeneous information, the thesis presents a third integration framework that is designed to integrate an arbitrary number of heterogeneous and asynchronous landmark streams into LVCSR. The results indicate that this framework is indeed ale to improve the baseline system, as soon as landmarks provide complementary information to the regular acoustic models.

Tsoi Chun Ho. / Thesis submitted in: June 2005. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2006. / Includes bibliographical references (leaves 125-136). / Abstracts and questionnaires in English and Chinese. / ABSTRACT --- p.i / ABSTRACT (in Chinese) --- p.ii / ACKNOWLEDGEMENTS --- p.iii / Chapter 1 --- INTRODUCTION --- p.1 / Chapter 1.1 --- General Introduction --- p.1 / Chapter 1.2 --- Background of the Present Study --- p.2 / Chapter 1.2.1 --- Teaching English as a Foreign Language (EFL) in Hong Kong --- p.2 / Chapter 1.2.2 --- The Place of Phonics in Hong Kong English Language Curriculum --- p.4 / Chapter 1.3 --- A Response: The Present Study --- p.5 / Chapter 1.4 --- Significance of the Study --- p.6 / Chapter 1.5 --- Organisation of this Thesis --- p.6 / Chapter 2 --- REVIEW OF THE LITERATURE --- p.8 / Chapter 2.1 --- Phonics Instruction --- p.8 / Chapter 2.1.1 --- What is Phonics? --- p.8 / Chapter 2.1.2 --- The Role of Phonics Instruction in the EFL context --- p.14 / Chapter 2.1.3 --- Phonics Instruction in Hong Kong Schools --- p.14 / Chapter 2.1.4 --- Ways of Teaching Phonics in L1 and L2 Environments --- p.18 / Chapter 2.2 --- Task-based Language Teaching and Learning --- p.22 / Chapter 2.2.1 --- Task and Task Types --- p.22 / Chapter 2.2.2 --- Task Types --- p.26 / Chapter 2.2.3 --- Tasks and Comprehensible Input --- p.29 / Chapter 2.2.4 --- TBL and Interactionist Perspective in SLA --- p.30 / Chapter 2.2.5 --- "Tasks, Memory and Accuracy" --- p.31 / Chapter 2.2.6 --- TBL and Affective Factors --- p.34 / Chapter 2.2.7 --- TBL and Context --- p.35 / Chapter 2.2.8 --- Task-based Language Teaching and Learning in the EFL context --- p.36 / Chapter 2.2.9 --- Task-based Learning (TBL) in Hong Kong EFL Curriculum --- p.39 / Chapter 2.2.10 --- Problems of Implementing the Task-based Syllabus in Hong Kong --- p.41 / Chapter 2.3 --- Learning Attitudes and its Measurement --- p.45 / Chapter 2.3.1 --- The Importance of Attitudes towards Language Learning --- p.45 / Chapter 2.3.2 --- Measuring Attitudes --- p.46 / Chapter 2.4 --- Task-supported Phonics Instruction --- p.50 / Chapter 2.4.1 --- Learning Phonics through Tasks --- p.50 / Chapter 2.4.2 --- Traditional versus Task-supported Phonics Instruction for EFL Learners --- p.51 / Chapter 2.4.3 --- Constructing Short Tasks in an HKEFL Lesson --- p.52 / Chapter 2.4.4 --- Techniques of using task-supported Phonics with Older Learners --- p.54 / Chapter 2.4.5 --- Reasons and Limitations of Teaching and Learning Phonics in a Task-supported Way --- p.55 / Chapter 2.4.6 --- A Framework of TSP for EFL Learners --- p.56 / Chapter 2.5 --- Chapter Summary --- p.58 / Chapter 3 --- DESIGN OF THE STUDY --- p.59 / Chapter 3.1 --- Introduction --- p.59 / Chapter 3.2 --- Research Questions and Null Hypotheses --- p.59 / Chapter 3.3 --- Overall Design of the Present Study --- p.62 / Chapter 3.4 --- The Settings --- p.62 / Chapter 3.4.1 --- The Original Setting and the Target Participants --- p.62 / Chapter 3.4.2 --- The New Setting and the Period of Study --- p.64 / Chapter 3.4.3 --- The Participants --- p.65 / Chapter 3.5 --- The Instrument I: Questionnaire --- p.65 / Chapter 3.5.1 --- Attitudinal Battery on Phonics Learning --- p.65 / Chapter 3.5.2 --- Attitudinal Battery on Task-supported Learning --- p.68 / Chapter 3.5.3 --- Adding Items in the Post-test of the Experimental Group --- p.69 / Chapter 3.6 --- The Instrument II: Reading Aloud Test --- p.69 / Chapter 3.6.1 --- Selection of the Texts --- p.69 / Chapter 3.6.2 --- A Brief Analysis of the Texts --- p.70 / Chapter 3.7 --- Pilot Studies --- p.72 / Chapter 3.7.1 --- Pilot Study for the Pre-test Questionnaire --- p.72 / Chapter 3.7.2 --- Pilot Study for the Post-test Questionnaire (TSL version) --- p.74 / Chapter 3.8 --- Data Collection: Description --- p.75 / Chapter 3.8.1 --- Pre-test sessions --- p.75 / Chapter 3.8.2 --- Post-test sessions --- p.76 / Chapter 3.9 --- The Teaching and Learning of the Phonics Classes --- p.77 / Chapter 3.9.1 --- Maintaining the Internal Validity --- p.77 / Chapter 3.9.2 --- Topic Coverage and the Design of the Course --- p.79 / Chapter 3.9.3 --- Designing the Tasks and the Exercises --- p.80 / Chapter 3.9.4 --- Students' Attendance --- p.82 / Chapter 3.10 --- Methods of Data Analysis --- p.83 / Chapter 3.11 --- Chapter Summary --- p.85 / Chapter 4 --- DATA ANALYSIS --- p.87 / Chapter 4.1 --- Introduction --- p.87 / Chapter 4.2 --- Criteria for Excluding Data from my Analysis --- p.87 / Chapter 4.3 --- Results from the Reading Aloud Tests --- p.89 / Chapter 4.3.1 --- Accuracy of the Onsets --- p.91 / Chapter 4.3.2 --- Accuracy of the Vowels --- p.92 / Chapter 4.3.3 --- Accuracy of the Final Consonants --- p.93 / Chapter 4.3.4 --- Number of Syllables --- p.94 / Chapter 4.3.5 --- Omissions --- p.95 / Chapter 4.3.6 --- Summary of the Reading Aloud Results --- p.96 / Chapter 4.4 --- Quantitative Results from the Questionnaires --- p.98 / Chapter 4.4.1 --- Attitudes towards Phonics Learning --- p.98 / Chapter 4.4.2 --- Attitudes towards English and English Learning --- p.101 / Chapter 4.4.3 --- Summary of the Questionnaire Results --- p.102 / Chapter 4.5 --- Qualitative Results from the Questionnaires --- p.103 / Chapter 4.5.1 --- Learners' Feelings towards Phonics Learning --- p.103 / Chapter 4.5.2 --- Things Students Like about the Phonics Course --- p.104 / Chapter 4.5.3 --- Things Students Don't Like about the Phonics Course --- p.105 / Chapter 4.5.4 --- Students' Ideas of Amending the Course --- p.105 / Chapter 4.6 --- Results from Confirmatory Factor Analysis using SEM --- p.106 / Chapter 4.7 --- Chapter Summary --- p.107 / Chapter 5 --- DISCUSSION --- p.108 / Chapter 5.1 --- Introduction --- p.108 / Chapter 5.2 --- Significant Finding: Learners' Improvement in Sounding out Onsets --- p.108 / Chapter 5.2.1 --- Onsets are Sequenced First in the Course --- p.109 / Chapter 5.2.2 --- Onset as a Similar Feature between English and Chinese --- p.109 / Chapter 5.3 --- Significant Findings: Learners' Attitudes towards Phonics Learning --- p.110 / Chapter 5.3.1 --- Attitudes towards Phonics Lessons --- p.110 / Chapter 5.3.2 --- Attempts to Sound Out Unfamiliar Words --- p.111 / Chapter 5.3.3 --- Anxiety Levels in Reading Aloud when Other People are Present --- p.111 / Chapter 5.4 --- An Examination of Findings that Fail to Support the Hypotheses --- p.112 / Chapter 5.4.1 --- Vowels --- p.112 / Chapter 5.4.2 --- Codas --- p.113 / Chapter 5.4.3 --- Omissions --- p.114 / Chapter 5.4.4 --- Learners' Intention to Encourage Others to Learn Phonics --- p.114 / Chapter 5.4.5 --- Learners' Attitudes towards the English Language and English Learning --- p.115 / Chapter 5.5 --- Implications of the Present Study to the Current Model --- p.116 / Chapter 5.5.1 --- A meaningful context is not adequate --- p.116 / Chapter 5.5.2 --- Having lessons at the right time --- p.116 / Chapter 5.5.3 --- Learners' Attendance and Involvement is an Important Factor --- p.118 / Chapter 5.5.4 --- The Attractiveness of the 'Product' Element --- p.118 / Chapter 5.5.5 --- The Contribution of These Implications to the SEM Model --- p.120 / Chapter 5.6 --- Implications of the Present Study to the Current Model --- p.120 / Chapter 6 --- CONCLUSION --- p.121 / Chapter 6.1 --- Summing Up --- p.121 / Chapter 6.2 --- Limitations of the Present Study --- p.122 / Chapter 6.2.1 --- Period of Study and Time for Designing the Course --- p.122 / Chapter 6.2.2 --- The Setting --- p.123 / Chapter 6.2.3 --- Limited Number of Participants --- p.123 / Chapter 6.2.4 --- Infeasibility of Having a Delayed Post-test --- p.124 / Chapter 6.3 --- Recommendations for Future Research --- p.110 / BIBLIOGRAPHY --- p.125 / APPENDICE --- p.122 / Appendix 1 Approval Notice from the CUHK Survey and Behavioural Research Ethics Committee / Appendix 2 Application for Change in Thesis Title / Appendix 3 Invitation Letter Sent to Primary Schools / Appendix 4 Invitation Proposal Sent to Primary Schools / Appendix 5 Prepared Letters for Parents of the Participatory Primary School / Appendix 6 School Notice for Recruiting Students / Appendix 7 Course Outline / Appendix 8 Reading Aloud Tests / Chapter a) --- PowerPoint Slides Used in the Pre-test Session / Chapter b) --- Instructions for the Reading Aloud tests / Chapter c) --- Pre-test Reading Aloud Passage / Chapter d) --- Post-test Reading Aloud Passage / Appendix 9 a) Questionnaire (Pre-test) / Chapter b) --- Questionnaire (Post-test: Control Group) / Chapter c) --- Questionnaire (Post-test: Experimental Group)

The rapid development of corpus-based speech systems such as concatenative synthesis systems for under-resourced languages requires an efficient, consistent and accurate solution with regard to phonetic speech segmentation. Manual development of phonetically annotated corpora is a time consuming and expensive process which suffers from challenges regarding consistency and reproducibility, while automation of this process has only been satisfactorily demonstrated on large corpora of a select few languages by employing techniques requiring extensive and specialised resources. In this work we considered the problem of phonetic segmentation in the context of developing small prototypical speech synthesis corpora for new under-resourced languages. This was done through an empirical evaluation of existing segmentation techniques on typical speech corpora in three South African languages. In this process, the performance of these techniques were characterised under different data conditions and the efficient application of these techniques were investigated in order to improve the accuracy of resulting phonetic alignments. We found that the application of baseline speaker-specific Hidden Markov Models results in relatively robust and accurate alignments even under extremely limited data conditions and demonstrated how such models can be developed and applied efficiently in this context. The result is segmentation of sufficient quality for synthesis applications, with the quality of alignments comparable to manual segmentation efforts in this context. Finally, possibilities for further automated refinement of phonetic alignments were investigated and an efficient corpus development strategy was proposed with suggestions for further work in this direction. / Thesis (M.Ing. (Computer Engineering))--North-West University, Potchefstroom Campus, 2009.

The rapid development of corpus-based speech systems such as concatenative synthesis systems for under-resourced languages requires an efficient, consistent and accurate solution with regard to phonetic speech segmentation. Manual development of phonetically annotated corpora is a time consuming and expensive process which suffers from challenges regarding consistency and reproducibility, while automation of this process has only been satisfactorily demonstrated on large corpora of a select few languages by employing techniques requiring extensive and specialised resources. In this work we considered the problem of phonetic segmentation in the context of developing small prototypical speech synthesis corpora for new under-resourced languages. This was done through an empirical evaluation of existing segmentation techniques on typical speech corpora in three South African languages. In this process, the performance of these techniques were characterised under different data conditions and the efficient application of these techniques were investigated in order to improve the accuracy of resulting phonetic alignments. We found that the application of baseline speaker-specific Hidden Markov Models results in relatively robust and accurate alignments even under extremely limited data conditions and demonstrated how such models can be developed and applied efficiently in this context. The result is segmentation of sufficient quality for synthesis applications, with the quality of alignments comparable to manual segmentation efforts in this context. Finally, possibilities for further automated refinement of phonetic alignments were investigated and an efficient corpus development strategy was proposed with suggestions for further work in this direction. / Thesis (M.Ing. (Computer Engineering))--North-West University, Potchefstroom Campus, 2009.

Orientador: Wilmar da Rocha D'Angelis / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Estudos da Linguagem / Made available in DSpace on 2018-08-25T00:12:51Z (GMT). No. of bitstreams: 1 Ivo_IvanaPereira_M.pdf: 6651068 bytes, checksum: 6997aedfa7a3caf004ca18d20275721a (MD5) Previous issue date: 2014 / Resumo: O resumo na íntegra poderá ser visualizado no texto completo da tese digital / Abstract: Note: The complete abstract is available with the full electronic document / Mestrado / Linguistica / Mestra em Linguística

published_or_final_version / abstract / Linguistics / Master / Master of Arts

The focus of this study was to explore how prosodic position and word type affect the phonetic structure and resulting perceptual identification of Korean stops and fricatives. When there is less contextual information, speakers tend to produce clearer speech. For example, consonants at the beginning of prosodic domains, such as syllables, words or phrases, are known to be more clearly articulated and distinguishable than later-occurring consonants. However, it is not yet clear whether the prosodically conditioned realizations of a segment are perceptually distinctive in continuous speech. In addition, there are few studies examining whether the properties of prosodic domain-initial segments are affected by the information content of words (real vs. nonsense words). The acoustic properties of stops and fricatives were compared across IP, PP and Wd-initial positions both in real and nonsense words. It was found that segments in the higher prosodic domain-initial positions showed enhanced durational properties compared to the lower prosodic domain-initial positions. However, the enhancing strategies were different among phonation types. Relative to lenis and aspirated stops, and lenis fricatives, tense stops and fricatives showed less consistent variation as a function of prosodic position and word type. In the perception study, the identification error rates and reaction time for same-spliced CVs were compared to those for cross-spliced CVs. Korean listeners identified the same-spliced CVs more accurately and faster than cross-spliced CVs. In addition, the distinctive acoustic properties of each prosodic domain-initial position were perceptually distinguished by Korean listeners. Due to relatively shorter duration and less distinctive contrast, the target CVs extracted from lower prosodic domain-initial positions caused more confusion in the identification of target segments. In conclusion, this study provides the evidence that speakers modulate their speech clarity depending on information content. By enhancing phonetic properties and phonological contrast, speakers tend to provide perceptual cues for prosodic positions with less contextual information. / text

This thesis investigates upper secondary students’ usage of scientific and colloquial language through a custom-designed game. The aim of the game is for the participants to, without any time limit, explain chemical concepts using other words than the one written on the playing card so that the partner is able to guess what the target concept is. During the audio recorded game sessions the students showed usage of scientific and colloquial language as well as a blend of these two linguistic resources. The students also used metaphors and helpwords in some measure together with a linguistic resource that seems to border to spontaneous metaphors and helpwords, here referred to as ‘phonetic metaphors’. These phonetic metaphors seem to lack a direct connection to the chemical meaning of the target concept in return for its phonetic connection to the concept. It though seems like the students’ usage of these different linguistic resources may support meaning making processes as well as memorization processes. The thesis concludes with a discussion according whether this game, if used professionally and thoughtfully, aligns with the guidelines for Assessment for Learning specified by the Assessment Reform Group and whether it may function as a tool for assessment for learning in practice.

The purpose of this study is to determine if a phonetic approach to speech improvement has a relationship to reading achievement, sound discrimination, and developmental articulation at the first-grade level.

The aim of this thesis is to examine the integration of anglicisms into French from a phonetic point of view. The theoretical part deals with the delimitation of the term 'anglicism', its place in the French lexis, a comparison of the phonological systems of both languages and the regularities in phonetic adaptation of anglicisms in the target language. The practical part includes an analysis of the pronunciation of chosen anglicisms. It records variation in lexicographical works as well as in real speech and determines the degree of variability, based on recordings of native speakers. The thesis proceeds from the assumption that anglicisms were, are and will continue to be a highly relevant issue. In spite of all the measures aimed against their integration into the French lexis, they have their place in it, often in several pronunciation variants - whether due to the different inventory of phonemes or to sociolinguistic factors. KEY WORDS: anglicism, French, pronunciation, phonetic adaptation of loanwords