No description available.

The purpose of this dissertation was to understand and compare the effect of training non-native English Speaking (NNES) learners (N = 480) in two distinct learning environments, (i) traditional face-to-face and (ii) online synchronous hybrid learning (SHL). In the traditional training mode, NNES learners (n = 360) were trained by NNES voice and accent (VANC) trainers in a physical, face-to-face setting. In the second, CAPT+SHL training mode, the NNES learners were trained by NNES VANC trainers with the help of a native-English computer-assisted pronunciation training (CAPT) tool in an online SHL environment. Factor analysis, higher-order factor analysis, hierarchical cluster analysis, and multidimensional scaling yielded a reliable scale, Eddie's Voice Test (EVT). Multiple regression yielded a predictive model between NNES pronunciation and their performance. In addition, the CAPT+SHL training mode produced higher scores on pronunciation and performance than the traditional training mode, suggesting a combination of NES and NNES VANC instructors are more effective in training NNES learners than NNES instructors by themselves. The case study (n = 3) on VANC trainers' perception of CAPT and SHL yielded three themes: (1) challenges with synchronous hybrid learning (sub-themes include physical challenges, social challenges, and cognitive challenges); (2) computer-assisted pronunciation training (CAPT) impact on non-native-English-speaking (NNES) learner pronunciation (sub-themes include self-paced pronunciation learning and pronunciation benchmarking; and (3). SHL as an equitable learning environment.

I föreliggande single subject designstudie har tre personer med thai som modersmål och svenska som andraspråk tränats specifikt i svenskt uttal. Syftet med föreliggande studie var därför att utforma samt genomföra en uttalsintervention vilken delades in i en suprasegmentell och en segmentell modul. Syftet var vidare att undersöka hur suprasegmentella och segmentella förändringar i uttalet påverkade uttalets förståelighet. Uttalsträningen utfördes 14 gånger under en fem-veckors period. Interventionen delades upp i två moduler, en med suprasegmentellt fokus (B1) och en med segmentellt fokus (B2). I den suprasegmentella modulen ingick medvetandegörelse av stavelseuppbyggnad samt träning i svenska betoningsmönster. Den segmentella modulen innefattade träning av konsonantförbindelser, [ɧ] samt /r/. Efter varje tillfälle fick deltagarna repetera 50 ord som valts ut innan studiens början. Dessa ord analyserades utifrån deltagarnas förmåga att korrekt producera suprasegmentella och segmentella aspekter och utgjorde studiens mätpunkter. De suprasegmentella aspekterna bedömdes utifrån uppsatta bedömningskriterier och de segmentella aspekterna bedömdes med hjälp av Percentage Phonemes Correct (PPC). Efter genomförd intervention fick tre utomstående lyssnare skatta förståelighet för att undersöka om samband kunde ses mellan deras skattningar och de valda bedömningsverktygen. Enskilda ord ur ordlistorna valdes ut för att bedömningsverktygen även skulle jämföras mot akustiska mätningar. Resultaten i föreliggande studie visade på att samband kunde ses mellan specifik uttalsträning med segmentellt fokus och förbättrat uttal både vad gäller segmentella och suprasegmentella aspekter. Samma resultat visade den suprasegmentella uttalsträningen ge. Vidare kunde samband ses mellan bedömningen av de segmentella aspekterna utifrån mätpunkternas beräknade PPC och förståelighetsskattningarna. När PPC ökade förbättrades också förståeligheten. Då uttalsträning bör ha som fokus att uppnå god förståelighet hos andraspråksinläraren (Pennington, 1996; Levis, 2005) föreslås att segmentella aspekter bör ingå i uttalsträning. Mellan förbättringarna av suprasegmentella aspekter och förståelighet har inte liknande samband kunnat ses vilket dock kan bero på studiens design. Författarna till föreliggande studie anser därför att suprasegmentella aspekter i uttalsträning inte bör bortprioriteras. / Three native Thai speakers with Swedish as their second language have received special training in Swedish pronunciation in the current single-subject design study. The aim of the current study was therefore to devise and implement a set of pronunciation training exercises, divided into suprasegmental and segmental modules. The aim was also to examine how suprasegmental and segmental changes in pronunciation affected the intelligibility of the pronunciation. The pronunciation training consisted of 14 training sessions during a five-week period. The exercises were divided into two modules, one with focus on suprasegmental aspects (B1) and one with focus on segmental aspects (B2). The suprasegmental module included acquainting subjects with the syllabic structure of the Swedish language as well as training of the Swedish language stress pattern. The segmental module contained training of consonant clusters, the Swedish speech sound [ɧ] and /r/. At the end of each session the participants were asked to repeat 50 words chosen before the start of the study. These words constituted the measuring points of the study and were analyzed according to the participants' ability to produce suprasegmental and segmental aspects correctly. The suprasegmental aspects were assessed according to fixed criteria and the segmental aspects were rated by means of Percentage Phonemes Correct (PPC). After completion of the final training session, three independent listeners estimated intelligibility to determine if relationship existed between these and the chosen assessment tools. Single words were selected in order to compare their criteria-based and PPC-ratings with acoustic correlates. The results of the present study indicate that a relationship can be observed between specific pronunciation training with focus on segmental aspects and improved pronunciation in terms of segmental and suprasegmental aspects. The pronunciation training with suprasegmental focus indicated the same conclusion. Further, a relationship can be seen between the judgment of the segmental aspects based on the measuring points' PPC-rating and the ratings of intelligibility. The intelligibility improved when the PPC increased. The focus of pronunciation training should be to achieve good intelligibility (Pennington, 1996; Levis, 2005); therefore it is suggested that segmental aspects should be included in pronunciation training for native Thai speakers when learning Swedish as a second language. No similar relations have been discovered between the suprasegmental aspects and intelligibility, which, however, may be due to limitations in the design of the study. Therefore suprasegmental aspects should not be neglected in pronunciation training.

This thesis presents a framework for computer assisted language learning using a virtual language teacher. It is an attempt at creating, not only a new type of language learning software, but also a server-based application that collects large amounts of speech material for future research purposes.The motivation for the framework is to create a research platform for computer assisted language learning, and computer assisted pronunciation training.Within the thesis, different feedback strategies and pronunciation error detectors are exploredThis is a broad, interdisciplinary approach, combining research from a number of scientific disciplines, such as speech-technology, game studies, cognitive science, phonetics, phonology, and second-language acquisition and teaching methodologies.The thesis discusses the paradigm both from a top-down point of view, where a number of functionally separate but interacting units are presented as part of a proposed architecture, and bottom-up by demonstrating and testing an implementation of the framework. / QC 20110511

Current computer-assisted pronunciation training (CAPT) tools are too focused on what technologies can do, rather than focusing on learner needs and pedagogy. They also lack an embodied perspective on learning. This thesis presents a Research through Design project exploring what kind of interactive design features can support second language learners’ pronunciation learning of segmental speech sounds with embodiment in mind. ComPron was designed: an open simulated prototype that supports learners in learning perception and production of new segmental speech sounds in a second language, by comparing them to native language speech sounds. ComProm was evaluated through think-aloud user tests and semi-structured interviews (N=4). The findings indicate that ComPron supports awareness of speech sound-movement connections, association building between sounds, and production of sounds. The design features that enabled awareness, association building, and speech sound production support are discussed and what ComPron offers in comparison to other CAPT-tools.

Es wird untersucht, ob das High-Variability-Training (HVT) die Perzeptions- und Produktionsleistungen des Gespanntheitskontrastes deutscher Vokale bei russischspra-chigen Deutschlernenden positiv beeinflusst. Dafür wurde in Anlehnung an bekannte HVT-Studien ein Aussprachetraining speziell für die Vokaldauer konzipiert. Basis ist die Gegenüberstellung zweier phonetischer Lautkontraste als Minimalpaare der fünf deutschen Vokale (A, E, I, O, U). An dem HVT nahmen dreißig fortgeschrittene russischsprachige Lernende des Deut-schen mit muttersprachlich bedingten Fehlern der L2-Vokalproduktion teil, gemäß dem Lernalter (während bzw. nach der kritischen Phase) in zwei Untersuchungsgruppen (RM1, RM2) getrennt. Die Untersuchung verlief in drei Phasen (Pretest, Training und Posttest). Das Training bestand aus acht Sitzungen von jeweils ca. zwanzig Minuten Dauer. Davor und danach wurden die Produktions- und Perzeptionsfähigkeiten ermittelt. Die Testdaten ergeben, dass 73,3 % der RM1- und 80 % der RM2-Probanden die Perzeptionsleistung der deutschen Vokalphoneme signifikant steigerten. In beiden Gruppen stiegen ebenfalls tendenziell die Produktionsfähigkeiten, 66,7 % der RM1- und 60 % der RM2-Probanden erhöhten sie signifikant. Dennoch konnte kein Teil-nehmer eine Vokalproduktion auf muttersprachlichem Niveau erreichen. Gezieltes HVT-Aussprachetraining wirkt sich demnach positiv auf die Perzeption und Produktion des Gespanntheitskontrastes bei russischsprachigen Deutschlernenden aus und ist eine effektive Erweiterung für den Vokalerwerb auch für fortgeschrittene L2-Lernende. Diese Erkenntnis unterstützt die Ergebnisse der Studien von Logan et al. (1991), Wang et al. (1999), Cenoz & Garciá-Lecumberri (1999), Lambacher et al. (2005), Nishi & Kewley-Port (2007), Iverson & Evans (2009), Liatambur & Lai (2011), Iverson, Pinet & Evans (2012), Wong (2012), Rato (2014), Jügler et al. (2015). / This study investigates whether high variability training (HVT) positively influences perception and production of the tension contrast of German vowels in Russian-speaking learners of German. For this task a pronunciation training course was developed especially for vowel length, drawing on established HVT studies. The basis for this is the comparison of two phonetic contrasts as minimal pairs, taken from the five German vowels (A, E, I, O, U). 30 advanced Russian-speaking learners of German with mother tongue interference in L2 vowel production took part in the HVT, separated according to learning age (during or after the critical phase) into two study groups (RM1, RM2). The study took place in three phases (pre-test, training and post-test). The training consisted of eight sessions, each of approx. 20 minutes in length. Production and perception abilities were tested before and after this. The test results show that 73.3% of the RM1 and 80% of the RM2 test subjects significantly increased their ability to perceive the German vowel phonemes. In both groups the production abilities also tended to rise, with 66.7% of the RM1 and 60% of the RM2 test subjects improving them significantly. However, no participant was able to achieve vowel production at native speaker level. This shows that targeted HVT pronunciation training has a positive effect on the perception and production of tension contrast in Russian-speaking learners of German and is an effective extension for vowel development, also for advanced L2 learners. These findings support the results of studies by Logan et al. (1991), Wang et al. (1999), Cenoz & Garciá-Lecumberri (1999), Lambacher et al. (2005), Nishi & Kewley-Port (2007), Iverson & Evans (2009), Liatambur & Lai (2011), Iverson, Pinet & Evans (2012), Wong (2012), Rato (2014), Jügler et al. (2015).

abstract: Much evidence has shown that first language (L1) plays an important role in the formation of L2 phonological system during second language (L2) learning process. This combines with the fact that different L1s have distinct phonological patterns to indicate the diverse L2 speech learning outcomes for speakers from different L1 backgrounds. This dissertation hypothesizes that phonological distances between accented speech and speakers' L1 speech are also correlated with perceived accentedness, and the correlations are negative for some phonological properties. Moreover, contrastive phonological distinctions between L1s and L2 will manifest themselves in the accented speech produced by speaker from these L1s. To test the hypotheses, this study comes up with a computational model to analyze the accented speech properties in both segmental (short-term speech measurements on short-segment or phoneme level) and suprasegmental (long-term speech measurements on word, long-segment, or sentence level) feature space. The benefit of using a computational model is that it enables quantitative analysis of L1's effect on accent in terms of different phonological properties. The core parts of this computational model are feature extraction schemes to extract pronunciation and prosody representation of accented speech based on existing techniques in speech processing field. Correlation analysis on both segmental and suprasegmental feature space is conducted to look into the relationship between acoustic measurements related to L1s and perceived accentedness across several L1s. Multiple regression analysis is employed to investigate how the L1's effect impacts the perception of foreign accent, and how accented speech produced by speakers from different L1s behaves distinctly on segmental and suprasegmental feature spaces. Results unveil the potential application of the methodology in this study to provide quantitative analysis of accented speech, and extend current studies in L2 speech learning theory to large scale. Practically, this study further shows that the computational model proposed in this study can benefit automatic accentedness evaluation system by adding features related to speakers' L1s. / Dissertation/Thesis / Doctoral Dissertation Speech and Hearing Science 2018

This thesis presents tools and resources for the development of applications in Natural Language Processing and Pronunciation Training. There are four main contributions. First, a hybrid grapheme-to-phoneme converter for Brazilian Portuguese, named Aeiouadô, which makes use of both manual transcription rules and Classification and Regression Trees (CART) to infer the phone transcription. Second, a spelling correction system based on machine learning, which uses the trascriptions produced by Aeiouadô and is capable of handling phonologically-motivated errors, as well as contextual errors. Third, a method for the extraction of phonetically-rich sentences, which is based on greedy algorithms. Fourth, a prototype system for automatic pronunciation assessment, especially designed for Brazilian-accented English. / Esta dissertação apresenta recursos voltados para o desenvolvimento de aplicações de reconhecimento de fala e avaliação de pronúncia. São quatro as contribuições aqui discutidas. Primeiro, um conversor grafema-fonema híbrido para o Português Brasileiro, chamado Aeiouadô, o qual utiliza regras de transcrição fonética e Classification and Regression Trees (CART) para inferir os fones da fala. Segundo, uma ferramenta de correção automática baseada em aprendizado de máquina, que leva em conta erros de digitação de origem fonética, que é capaz de lidar com erros contextuais e emprega as transcrições geradas pelo Aeiouadô. Terceiro, um método para a extração de sentenças foneticamente-ricas, tendo em vista a criação de corpora de fala, baseado em algoritmos gulosos. Quarto, um protótipo de um sistema de reconhecimento e correção de fala não-nativa, voltado para o Inglês falado por aprendizes brasileiros.

This thesis presents tools and resources for the development of applications in Natural Language Processing and Pronunciation Training. There are four main contributions. First, a hybrid grapheme-to-phoneme converter for Brazilian Portuguese, named Aeiouadô, which makes use of both manual transcription rules and Classification and Regression Trees (CART) to infer the phone transcription. Second, a spelling correction system based on machine learning, which uses the trascriptions produced by Aeiouadô and is capable of handling phonologically-motivated errors, as well as contextual errors. Third, a method for the extraction of phonetically-rich sentences, which is based on greedy algorithms. Fourth, a prototype system for automatic pronunciation assessment, especially designed for Brazilian-accented English. / Esta dissertação apresenta recursos voltados para o desenvolvimento de aplicações de reconhecimento de fala e avaliação de pronúncia. São quatro as contribuições aqui discutidas. Primeiro, um conversor grafema-fonema híbrido para o Português Brasileiro, chamado Aeiouadô, o qual utiliza regras de transcrição fonética e Classification and Regression Trees (CART) para inferir os fones da fala. Segundo, uma ferramenta de correção automática baseada em aprendizado de máquina, que leva em conta erros de digitação de origem fonética, que é capaz de lidar com erros contextuais e emprega as transcrições geradas pelo Aeiouadô. Terceiro, um método para a extração de sentenças foneticamente-ricas, tendo em vista a criação de corpora de fala, baseado em algoritmos gulosos. Quarto, um protótipo de um sistema de reconhecimento e correção de fala não-nativa, voltado para o Inglês falado por aprendizes brasileiros.

Computer-Assisted Pronunciation Training (CAPT) is becoming more and more popular these days. However, the accuracy of existing CAPT systems is still quite low. Therefore, this diploma thesis focuses on improving existing methods for automatic pronunciation evaluation on the segmental level. The first part describes common techniques for this task. Afterwards, we proposed the system based on two approaches. Finally, performed experiments show significant improvement over the reference system.