STUDENT NO: 9902414P MA HUMANITIES SCHOOL OF HUMAN AND COMMUNITY DEVELOPMENT / ABSTRACT Rationale: In the early 1990s, the Lidcombe Programme, which is a direct, parentconducted, operant intervention programme for early childhood stuttering, was introduced to therapists. Available data in Australia suggest that the Lidcombe Programme is an effective early communication intervention as it eliminates stuttered speech in the medium and long term and is able to maintain these effects for up to seven years post-treatment (Onslow, Costa & Rue, 1990; Onslow, Andrews & Lincoln, 1994; Lincoln & Onslow, 1997; Jones, Onslow, Harris & Packman, 2000). However, despite the plethora of research in Australia, no South African research appears to have previously been conducted to establish long-term results and maintenance of fluency following the administration of this programme with young children. Purpose: The present study endeavoured to establish long-term results and maintenance of fluency using the Lidcombe Programme with children who stutter. Participants: Nine children were recruited, ranging in age from six to eleven years who had undergone therapy using the Lidcombe Programme during the previous two to six years. The parents of these children also participated in the study. Method: As part of the Lidcombe Programme, the percentage syllables stuttered (%SS) and the severity rating score (SR) were recorded at the time of the assessment and termination of therapy. The researcher evaluated each participant’s speech in two different situations using the measurements of %SS and SR and then compared these scores to the %SS and SR score recorded at the initial assessment and termination of therapy to ascertain whether each participant’s speech had improved, deteriorated or remained consistent. An interview with the child’s parents was also conducted to obtain a qualitative assessment of the parent’s views of their child’s speech since the termination of therapy. Results: The Wilcoxon signed rank test revealed a significant decrease in %SS and SR between pre-treatment scores and scores recorded at the time of the study (p = 0.00195 & p = 0.00195 respectively). Four participants’ %SS obtained at the time of the study compared to immediately after termination of therapy had increased slightly although this difference was not statistically significant (p = 0.4375). All parents reported that they had noted improvement in their child’s speech since the termination of therapy except for one mother who reported that her child’s speech had remained the same. Conclusions: It would seem that the Lidcombe Programme was successful with this particular group of participants as all of the children demonstrated a definite decrease in stuttering following therapy compared to pre-treatment and had maintained the levels of fluency obtained post-treatment to the time of the study. These findings have implications for early communication intervention, training of speech-language pathologists, theory and future research. KEY WORDS: stuttering, Lidcombe Programme, children, severity rating (SR) percentage syllables stuttered (%SS)

Auf der Grundlage psycholinguistischer, deutsch-didaktischer und linguistischer Forschungsergebnisse wurde ein Lesetraining auf Silbenbasis entwickelt, in dem Kindern mit Leseschwierigkeiten explizit Strategien zur visuellen Gliederung von Wörtern in größere funktionale Einheiten (Silben, einzelne Wörter (bei Zusammensetzungen), Präfixe) vermittelt werden. Es wird davon ausgegangen, dass Kinder mit Leseschwierigkeiten Wörter in zu kleinen linguistischen Einheiten erlesen. Das Training basiert linguistisch auf der Schreibsilbe als funktionaler Einheit beim Lesen. Basis des Lesetrainings ist ein von Scheerer-Neumann (1981) durchgeführtes Training zur Verbesserung der Lesegenauigkeit von Grundschulkindern. <br><br> Die empirische Überprüfung des Lesetrainings hatte zum einen das Ziel, die Wirksamkeit des Trainingsprogramms als solches zu überprüfen und zum anderen, ob das Training zu einer deutlich verbesserten Lesegeschwindigkeit und -genauigkeit bei Grundschulkindern der 3. und 4. Klassen führt. Das Training wurde von der Autorin ohne den Einsatz von Hilfskräften mit einer relativ kleinen Zahl von Kindern (N=11) durchgeführt, um so intensiv auf die Bedürfnisse der einzelnen Kinder eingehen und gegebenenfalls flexibel reagieren zu können. Aus organisatorischen und ethischen Gründen wurde auf eine Kontrollgruppe verzichtet. Die Wirksamkeit des Trainings wurde als Prätest-Training-Posttest-Follow-up Test-Design überprüft. Das Training umfasst 18 Einheiten (je 45 Minuten) und fand zweimal wöchentlich statt. Trainingsgruppen bestanden aus einer Dreiergruppe und einer Zweiergruppe; die anderen sechs Kinder wurden einzeln trainiert. <br><br> Der Posttest nach Abschluss des Trainings zeigte deutliche Verbesserungen der Lesegeschwindigkeit und -genauigkeit, die zum Teil auch signifikant waren. Der Follow-up Test drei Monate nach Ende des Trainings zeigte weitere Verbesserungen, die ebenfalls zum Teil signifikant waren. <br><br> Die Befunde sprechen dafür, dass das Trainingsprogramm geeignet ist, die Lesefähigkeit von leseschwachen Grundschulkindern der dritten und vierten Klasse zu verbessern. Die Teilnehmerzahl ist allerdings relativ gering, so dass die Ergebnisse nicht ohne weiteres auf die allgemeine Population von Kindern mit Leseschwierigkeiten übertragen werden können. Eine erneute Durchführung des Trainingsprogramms mit einer größeren Stichprobe und einer parallelisierten Kontrollgruppe, bei der die Kinder der Experimentalgruppe in größeren Gruppen trainiert werden, erscheint deshalb sinnvoll. / A syllable-based training program for reading disabled children was developed on the basis of psycholinguistic, educational and linguistic research results. The program teaches explicit strategies to visually segment words in larger functional linguistic units (syllables, prefixes, single words (in compound words)). <br><br> It is assumed that children with reading difficulties process words in linguistic units that are too small to make use of the linguistic structure of words. The training program, Potsdamer Lesetraining, emphasizes the role of the orthographical syllable as a functional unit in reading development. The program is based on a training program developed by Scheerer-Neumann (1981) that aimed at improving reading accuracy in reading disabled elementary school children. <br><br> The aim of the empirical test was to check if the training program had a positive effect on the decoding speed and accuracy of elementary school children with reading disabilities (3rd and 4th grade). The program was carried out by the author with a relatively small sample of children (N=11) and without further staff. This procedure made it possible to react intensively and flexible to the children's individual needs. A control group was not implemented for ethical and organizational reasons. <br><br> The validity of the training program was tested with a pretest-posttest-follow-up-test design. The program consisted of 18 training units (45 minutes each) and took place twice a week. The training groups consisted of one group with three children, another group with two children; the remaining six children were trained individually. <br><br> The posttests which were carried out immediately after training ended, showed significant effects of training for decoding speed and an increase of accuracy for several subtests. Follow-up tests carried out three months after the training had ended showed further improvement in decoding speed and an increase in accuracy; several subtests revealed significant effects. <br><br> The results indicate that the training program is suitable to improve the decoding speed and accuracy of dyslexic elementary school children in 3rd and 4th grade. A transfer of this conclusion to the general dyslexic population, however, is limited due to the small number of participants. <br><br> A replication of the study with a larger sample of reading disabled children trained in groups of two and three children and a matched control group would therefore seem necessary.

Made available in DSpace on 2014-08-20T13:47:52Z (GMT). No. of bitstreams: 1 Gisele Ramos Lima_Dissertacao.pdf: 9536721 bytes, checksum: d0e79c5fa880ab43b9ffc12bdc70fbe9 (MD5) Previous issue date: 2013-09-16 / The present Master Degree Paper presents an analysis about the types and the recurrence of activities involving syllables, observed in 68 Class Diaries that belonged to literacy teachers from a period between the years of 1973 and 2010. These Diaries belong to the collection from the Research Group HISALES, linked to the Universidade Federal de Pelotas /Faculdade de Educação. The objective of the research is to understand why the activities with syllables are recurring and were kept during the studied period. This study is inserted in the field of history of the literacy, which in Brazil, according to Maciel (2003), groups mainly the problematization around two axis: os métodos de leitura e escrita e os manuais escolares cartilhas ( The methods of writing and reading and the school manuals primer ) (MACIEL, 2003, p. 233). The nature of this research is qualitative (LÜDKE & ANDRÉ, 1986), having as epistemological model the evidentiary paradigm proposed by Ginzburg (2011), that presents as presupposition the search for traces , clues , evidences in the researched documents, aiming to find in the details what is possible to consider as relevant elements to answer the questions from the research. In relation to the results found, among other observed questions, it was possible to contrast the predominance of traditional methods of literacy in the planning registered in the Diaries. In relation specifically to the work with syllables, the most frequent activities worked by the literacy teachers were those that privilege the copy and the memorization of such structure. The study also showed that the divulgation of studies of the Psicogênese da Língua Escrita (FERREIRO; TEBEROSKY, 1985) seems to have had little influence in the planning of great part of the researched Diaries. Such results aim to the need that the activities with syllables must be rethought and resignified, considering the studies about the Psicogênese da Língua Escrita and other studies in the field that reference the writing as a notational system and not a code to be deciphered. / A presente Dissertação de Mestrado apresenta uma análise sobre os tipos e a recorrência de atividades envolvendo sílabas, observados em 68 Diários de Classe pertencentes a professoras alfabetizadoras no período compreendido entre os anos de 1973 a 2010. Esses Diários pertencem ao acervo do Grupo de Pesquisa HISALES, vinculado à Universidade Federal de Pelotas/Faculdade de Educação. O intuito da pesquisa é compreender por que razão as atividades com sílabas são recorrentes e se mantiveram ao longo do período de abrangência. Este estudo está inserido no campo da história da alfabetização, que, no Brasil, segundo Maciel (2003), agrupa principalmente a problematização em torno da alfabetização em dois eixos: os métodos de leitura e escrita e os manuais escolares cartilhas (MACIEL, 2003, p. 233). A pesquisa é de cunho qualitativo (LÜDKE & ANDRÉ, 1986), tendo como modelo epistemológico o paradigma indiciário proposto por Ginzburg (2011), que apresenta como pressuposto a busca de vestígios , pistas , indícios nos documentos pesquisados, objetivando encontrar nos detalhes o que é possível considerar como elementos relevantes para responder às questões da pesquisa. Em relação aos resultados encontrados, entre outras questões observadas, foi possível constatar a predominância dos métodos tradicionais de alfabetização nos planejamentos registrados nos Diários. Em relação especificamente ao trabalho com sílabas, as atividades mais recorrentes trabalhadas pelas professoras alfabetizadoras foram aquelas que privilegiam a cópia e a memorização de tais estruturas. O estudo também mostrou que a divulgação dos estudos da Psicogênese da Língua Escrita (FERREIRO; TEBEROSKY, 1985) parece ter tido pouca influência nos planejamentos de grande parte dos Diários pesquisados. Tais resultados apontam para a necessidade de que as atividades com sílabas sejam repensadas e resignificadas, considerando os estudos sobre a Psicogênese da Língua Escrita e os demais estudos na área da alfabetização que referenciam a escrita como um sistema notacional e não como um código a ser decifrado.

Clear speech is a speaking style that has been shown to improve intelligibility in adverse listening conditions, for various listener and talker populations. Clear-speech phonetic enhancements include a slowed speech rate, expanded vowel space, and expanded pitch range. Although clear-speech phonetic enhancements have been demonstrated across a variety of talkers, only a subset of these changes may be required for listeners to benefit perceptually from clear speech. Furthermore, while current literature has provided some understanding of the phonetic enhancements that are typical of clear speech and the improvements in intelligibility resulting from its use, less is understood regarding how listeners benefit from clear speech. Understanding the mechanisms underlying the effectiveness of clear speech will provide insight into speech processing more generally. To that end, two studies were conducted to investigate the possible mechanisms underlying clear-speech benefits. The first study tests the hypothesis that clear speech reduces the amount of information needed for syllable identification. The amount of information presented to listeners was controlled using a silent-center syllable paradigm, in which various amounts of the center or edge of the syllables were removed. The second study tests the hypothesis that phonetic processing of clear speech requires fewer neuro-cognitive resources than typical, or conversational, speech. An Event Related Potential (ERP) paradigm, focused on the Phonological Mismatch Negativity (PMN) component, was used to compare participants’ neurophysiological responses to conversational- and clear-speech stimuli. Results from the first experiment supported the hypothesis of a clear-speech benefit in partial syllables, although the effect was stronger for some vowels than for others. The second experiment demonstrated differences in the way the brain responds to clear versus conversational speech, for syllables in which only the nucleus varied across stimuli. Match-mismatch differences were found in the MMN and PMN components, while the N600 component was found to respond to style differences.

The primary aim of the present study is to gain an understanding of the translation process involved when translating song lyrics by investigating to what extent 10 translation features occur in a corpus of 12 song lyrics from the musicals CHESS, MAMMA MIA! and Kristina från Duvemåla. Comparing the source texts to the translated texts, taking into account: number of words, syllables vs. words, word-for-word translations, additions/omissions, metaphors, rhymes, reorganization of text, paraphrases and last if there were any untranslated English words kept in the Swedish version – led to the conclusion that the translation strategy of using paraphrases (express something written in other words) was the most common translation strategy used when translating song lyrics. In addition, translating song lyrics also requires a translator who is musical, has good association skills, a large vocabulary and is also very good at playing with words. Taking the findings into consideration it could be said that the word 'translation' should be avoided in reference to the act of transferring the song lyrics of a musical in one language into another language. More apt descriptive phrases for this process would probably be 'text arrangement' or 'interpretation'.

Mon thème de recherche est d’étudier l’influence de l’apprentissage de la musique sur le traitement acoustique et phonologique de la syllabe chez des enfants normolecteurs et dyslexiques. Dans ce but, j’ai conduit plusieurs expériences basées sur l’utilisation conjointe des méthodes issues de la psychologie expérimentale (Temps de Réaction, TRs, et pourcentage d’erreurs, %err) et de l’électrophysiologie chez l’homme (Potentiels Evoqués, PEs). En comparant des enfants musiciens et non-musiciens de 9 ans, j’ai d’abord testé les effets de l’expertise musicale sur les traitements attentif (TRs et %err) et pré-attentif (en utilisant la Mismatch Negativity, MMN) de paramètres acoustiques, fréquence et durée des syllabes, et d’un paramètre phonologique, le Voice Onset Time (VOT; Expérience I). Les résultats montrent que l'expertise musicale améliore les traitements pré-attentif et attentif de la durée et du VOT dans les syllabes et le traitement attentif des variations de fréquence. Dans une seconde étude, j’ai utilisé la MMN pour comparer le traitement de ces mêmes paramètres chez des enfants dyslexiques et normolecteurs. Les enfants dyslexiques montrent un déficit du traitement de la durée des syllabes et du VOT comparés aux enfants normolecteurs. Enfin, dans les deux dernières études, j’ai utilisé la méthode longitudinale pour tester l’influence de l’apprentissage de la musique sur le traitement pré-attentif (MMN) de ces mêmes paramètres chez des enfants normolecteurs (Expérience III) et dyslexiques (Expérience IV). Les résultats de l’Expérience III montrent que 12 mois d’apprentissage de la musique améliorent le traitement pré-attentif de la durée et du VOT chez les enfants normolecteurs. En reproduisant les effets trouvés dans l’Expérience I, ces résultats soulignent que l’avantage mis en évidence chez les enfants musiciens ne résulterait pas uniquement de prédispositions génétiques pour la musique mais serait causalement lié à l’apprentissage musical. Enfin, les résultats de l’Expérience IV montrent que 6 mois d’apprentissage de la musique améliorent le traitement pré-attentif du VOT chez les enfants dyslexiques, suggérant que l’apprentissage musical pourrait être utilisé comme une aide à la remédiation de la dyslexie.Pris dans leur ensemble, ces résultats démontrent une relation forte entre traitements acoustique et phonologique. L’apprentissage de la musique, en améliorant la sensibilité des enfants aux paramètres acoustiques dans la musique et dans le langage (processus communs), augmenterait également leur sensibilité aux paramètres phonologiques associés et permettrait ainsi la construction de représentations phonologiques plus robustes (transfert d’apprentissage de la musique vers le langage). / My research is aimed at studying the influence of musical training on the acoustic and phonological processing of syllables in children with dyslexia and in normal-reading children. To this aim, I conducted several experiments by using methods issued from experimental psychology (Reaction Times, RTs, and error rates, %err) and from human electrophysiology (Event-Related brain Potentials, ERPs)By comparing 9-year-old musician and non-musician children, I first tested for the effects of musical expertise on attentive (RTs and %err) and preattentive processing (by using the Mismatch Negativity, MMN) of the acoustical parameters, frequency and duration, of syllables and of a phonological parameter, the Voice Onset Time (VOT; Experiment I). Results showed enhanced preattentive and attentive processing of syllables’ duration and VOT in musicians compared to nonmusician children. Secondly, I compared the processing of these same parameters in dyslexic and normal-reading children (Experiment II) by using the MMN. Results revealed that children with dyslexia showed deficits for the processing of duration and VOT in syllables compared to normal-readers. Finally, in the last two studies, I used the longitudinal method to test for the influence of musical training on the processing of the same acoustic and phonological parameters of syllables, in normal-reading children (Experiment III) and in children with dyslexia (Experiment IV). Results of Experiment III showed that 12 months of musical training enhanced duration and VOT processing in syllables, thereby demonstrating that the effects of musical expertise shown in Experiment I are not likely to only result from specific genetic predispositions for music but are causally linked to musical training. Finally, results of Experiment IV revealed that 6 months of musical training in children with dyslexia enhanced their sensitivity to VOT processing, suggesting that musical training could be an aid for the remediation of dyslexia.These results highlight the relationship between acoustical and phonological processing. Musical training, by refining the acoustical network responsible for the acoustic processing in music and speech sounds (common processing) also enhances sensitivity to phonological associated features and, consequently, the building-up of more robust phonological representations (transfer of training effect from music to language processing).

Bis auf den heutigen Tag ist das Fach Gehörbildung im Allgemeinen und der Gebrauch von Tonsilben in der basalen Musikausbildung im Besonderen durch unterschiedliche länderspezifische Ausprägungen und Lehrtraditionen geprägt. Dieser Beitrag stellt überblicksartig unterschiedliche Entwicklungen im 18. und frühen 19. Jahrhundert dar, die zwischen Beibehaltung, Reform und Abschaffung der guidonischen Solmisationspraxis wechseln. Anhand von vier Fallbeispielen aus Frankreich, Italien, Spanien und Deutschland wird die Vielfalt der Lösungsansätze aufgezeigt. In Frankreich hat sich zwar bereits im 18. Jahrhundert das absolute Solmisieren durchgesetzt, daneben existierten jedoch mehrere Versuche, die Praxis der relativen Solmisation zu erhalten oder zu reaktivieren, etwa in der ›Méthode GalinParis-Chevé‹ von 1844. In Spanien zeigt sich neben einer lebendigen gregorianischen Praxis mit Mutationen (con mutanzas) eine Vielfalt unterschiedlicher Solmisationssysteme, eine Eigenheit scheint die enge Verbindung mit dem sicheren Erlernen des Schlüssellesens durch die Praxis der Schlüsselsimulation (fingimiento de clave) zu sein. Am Beispiel von Niccolò Zingarelli lässt sich für Neapel die Gleichzeitigkeit alter und neuer Lehrtraditionen nachweisen, dabei wurde das alte System der Hexachord-Solmisation zunehmend von der oktavbasierten Variante der scala composta ersetzt. Im deutschsprachigen Raum spielte im 18. Jahrhundert das Singen und Denken in Hexachorden nur noch eine marginale Rolle, statt der guidonischen Silben wurden die lateinischen Buchstaben, neue Silbenbezeichnungen oder Ziffern verwendet. Die aufgezeigten Methoden konsolidierten und vereinheitlichten sich zwar im Verlauf des 19. Jahrhunderts insbesondere durch die zunehmende Institutionalisierung der professionellen Musikausbildung. Im Zuge der Ausbildung moderner Nationalstaaten wurden die Unterschiede dafür umso deutlicher entlang der Landes- bzw. Sprachgrenzen sichtbar. / To this day, the discipline of ear training in general and the use of syllables in basic music education in particular have been informed by different country-specific characteristics and teaching traditions. This article provides an overview of different developments in the 18th and early 19th centuries, which alternate between the maintenance, reform and rejection of Guidonian solmisation practice. The variety of possible methods is demonstrated by four case studies from France, Italy, Spain, and Germany. In France, fixed-do solmisation was already established during the 18th century, although there were several attempts to maintain or reactivate the practice of relative solmisation, for example in the ‘Méthode Galin-Paris-Chevé’ of 1844. In Spain we find, in addition to a lively Gregorian practice with mutations (con mutanzas), a variety of different solmisation systems, an apparent peculiarity being the close connection with an education of clef-reading through the practice of clef simulation (fingimiento de clave). Solfeggi by Niccolò Zingarelli exemplify the coexistence of old and new teaching traditions in Naples, the old system of hexachord solmisation progressively being replaced by the octave-based variant of the scala composta. In German-speaking countries, the practice of singing and thinking in hexachords was already of marginal importance in the 18th century; instead of the Guidonian syllables, the Latin letters, new syllable names or numbers were used. The methods shown here became consolidated and unified during the 19th century, in particular due to the increasing institutionalisation of professional music education. During the emergence of modern nation states, the differences became all the more visible along national or language borders.

De par leur présence à l’interface entre plusieurs domaines linguistiques (phonologie, morpho-syntaxe, sémantique ou pragmatique), les déterminants ont depuis longtemps suscité l’intérêt des chercheurs en acquisition. On trouve ainsi de nombreux travaux portant notamment sur les contraintes phono-prosodiques, lexico-syntaxiques, sémantiques ou pragmatiques à l’œuvre dans le processus d’acquisition de ces formes, mais peu d’entre eux ont cherché à proposer une analyse combinée de ces facteurs, en tenant compte à la fois des morphèmes adultes et des formes de transition (omissions et fillers). Nous avons donc souhaité, dans le cadre de ce travail, décrire l’émergence des déterminants et de leurs proto-formes en essayant de saisir, dans une perspective multidimensionnelle intégrant à la fois des facteurs distributionnels et fonctionnels, quels facteurs influencent le plus ce processus, et à quelles périodes.Cette thèse repose sur l’analyse d’un corpus longitudinal de quatre enfants, âgés de 1;6 pour le plus jeune lors de la première séance sélectionnée à 3;5 pour le plus âgé lors de la dernière. Nous avons repéré au sein de ce corpus tous les noms produits par les enfants ainsi que les formes les précédant et chaque syntagme a fait l’objet d’une analyse phono-prosodique, lexico-syntaxique, sémantique, pragmatique et discursive.Les résultats montrent d’une part que tous les facteurs distributionnels peuvent - à des degrés différents - influencer la production des formes pré-nominales, et d’autre part, que les usages que les enfants font des formes témoignent d’une sensibilité précoce à certains contrastes pragmatico-discursifs. Toutefois, aucun des facteurs pris séparément n’a permis de rendre compte de l’intégralité des productions des enfants. L’analyse conjointe de ces facteurs distributionnels et fonctionnels a confirmé les tendances dégagées au préalable et a mis en évidence l’influence d’un ensemble de facteurs (et en particulier, phono-prosodique, syntaxique et pragmatique) plus que d’un facteur unique ou prédominant tout au long du processus d’acquisition des déterminants. / Given their presence at the crossroads of different linguistic fields (phonology, morphosyntax, semantics or pragmatics), determiners have for a long time aroused the interest of language acquisition researchers. We thus find several studies dealing with the phono-prosodic, lexical and syntactic, semantic or pragmatic constraints involved in the acquisition of determiners. But only few of them have provided a combined analysis of all these factors, considering both adult morphemes and transitional forms (omissions and filler syllables). Our study aims to describe the emergence of filler syllables and determiners and understand - in a multidimensional perspective examining simultaneously distributional and functional factors - which factors may affect the acquisition process, and over which periods of time.Our study is based on a corpus of four children, aged from 1;6 for the youngest in the first selected session to 3;5 for the eldest in the last session. We first identified all the noun phrases produced in the data and each occurrence was then analysed according to phono-prosodic, lexical, syntactic, semantic, pragmatic and discursive factors.Our results show that each distributional factor may influence, at different degrees, the production of pre-nominal forms and that children’s use of forms shows early sensitivity to pragmatic factors. However, none of the factors taken separately have accounted for all of the children’s productions. An additional combined analysis of both distributional and functional factors has supported the initial trends and highlighted the influence of a series of factors (and in particular, phono-prosodic, syntactic and pragmatic), more than a single and predominant one throughout the acquisition process of determiners.

Cette thèse fait partie du projet RAPSODIE dont l’objectif est de proposer une reconnaissance vocale spécialisée sur les besoins des personnes sourdes et malentendantes. Deux axes sont étudiées : la modélisation lexicale et l’extraction d’informations para-lexicales. Concernant la modélisation lexicale, nous avons étudié les modèles de langage hybrides combinant mots et syllabes, et nous avons proposé une nouvelle approche basée sur une notion de similarité entre mots pour l’ajout de nouveaux mots dans le modèle de langage. Concernant l’extraction d’informations para-lexicales, nous avons étudié l'utilisation des paramètres prosodiques, des paramètres linguistiques ou de leur combinaison pour la détection des questions et des affirmations. Cette détection a comme but de signaler aux personnes sourdes ou malentendantes quand une question leur est adressée / This thesis is part of the RAPSODIE project which aims at proposing a speech recognition device specialized on the needs of deaf and hearing impaired people. Two aspects are studied: optimizing the lexical models and extracting para-lexical information. Regarding the lexical modeling, we studied hybrid language models combining words and syllables, and we proposed a new approach based on a similarity measure between words to add new words in the language model. Regarding the extraction of para-lexical information, we investigated the use of prosodic features, of linguistic features and of their combination for the detection of questions and statements. This detection aims to inform the deaf and hearing impaired people when a question is addressed to them

Denna studie handlar om variation i skriven meänkieli med fokus på variation i ett specifikt drag, h i efterstavelsen (t.ex. saunhaan &gt; sauhnaan 'till bastun'). Tidigare forskning har visat att detta språkdrag påträffas med omfattande variation efter ett särskilt mönster i talspråket (t.ex. Mantila 1992, 2010; Vaattovaara 2009). Syftet med denna studie är att undersöka om, och i vilken omfattning, det mönster som beskrivits för det talade språket också påträffas i skriven text på meänkieli. Studien föranleds av det faktum att meänkieli för närvarande befinner sig i en standardiseringsprocess. h i efterstavelsen i talat språk utgör ett komplext system, som kan uppfattas av vanliga språkanvändare som ”svårt att beskriva”. Hur som helst har tidigare forskning indikerat att användningen av h i talat språk är ganska systematiserat och ett mönster för variationen har utvecklats i en bestämd riktning över tid. Med tanke på denna studie är det av intresse att undersöka om mönstret är liknande i skriftlig form, det vill säga om de enskilda författarna i den undersökta korpusen använder sig av ett liknande system för h i sina litterära verk. Studien har inte som syfte att diskutera språkplanering eller principer för standardisering, utan syftet är att skapa en helhetsbild av språklig variation i användningen av h i skriftspråket. Korpusen består av 43 böcker av 18 olika författare från tre olika decennier (1980–2000). Från varje publikation är de första 5250 orden excerperade om verket innehåller fler ord (barnböckerna, som är kortare och till antalet 18, har excerperats i sin helhet.) Studiens primära fokus besvaras med hjälp av frågorna huruvida det förekommer variation mellan de olika författarna, och vilken typ av variation som förekommer inom de texter varje enskild författare skrivit, även över tid betraktat. Syftet med studien är att avslöja om variationsmönster eventuellt sprider sig, eller om mönstret indikerar lika användningssätt mellan de enskilda författarna. Enligt sociolingvistiska teorier om språklig förändring borde en s.k. undersökning i verklig tid över flera decennier visa huruvida en pågående förändring är på gång. Hypotesen för denna studie (huvudsakligen på basis av Mantila 2010) är, att ett liknande system som påträffas i talspråket också påträffas i skriven form. h i efterstavelsen har en historisk utveckling med många förändringar och är ur ett morfologiskt och fonologiskt perspektiv ett mångfacetterat språkdrag. Studien fokuserar enbart på de sex olika lingvistiska positioner där den metatetiska varianten av h i efterstavelsen kan förekomma, vilket alltid sker i samband med en tonande konsonant exempelvis på följande sätt: 1. Stockholmhiin &gt; Stockholhmiin 'till Stockholm', 2. koulhuun &gt; kouhluun 'till skolan', 3. korjaamhaan &gt; korjaahmaan 'för att städa', 4. Pajalhaan &gt; Pajahlaan 'till Pajala', 5. istumhaan &gt;  istuhmaan 'för att sitta', 6. ymmärtänheet &gt; ymmärtähneet 'har förstått'). Tidigare forskning har ansett denna specifika variant av h med metates vara ”ett varumärke för Tornedalen” och språkdraget har visat ett tydligt mönster för variation i Tornedalen. I denna studie speglas variationen av h‑metatesen i litteraturen på meänkieli mot tidigare forskningsresultat rörande talspråk, där språkdraget undersökts både i svenska och finska Tornedalen. Bengt Pohjanen (med det största antalet utgivningar), tillsammans med tre andra författare har undersökts närmare. Ett intressant resultat är att det förekommer relativt omfattande variation mellan dessa författare, samtliga bördiga från en och samma kommun, nämligen Pajala. I motsats till vad som hävdas i traditionell sociolingvistik stämmer inte här antagandet att författarnas härkomst skulle vara en styrande faktor i författarnas språkliga val. Det allra viktigaste resultatet i studien är att enskilda böcker har under de tre undersökta decennierna visat ett överraskande resultat i fråga om variation under 1980- och 1990-talet, men under 2000-talet har det mönster som påträffats i talat språk systematiserats och tagits i bruk på ett enhetligt sätt i litteraturen på meänkieli mer eller mindre hos samtliga författare. I positionerna 1–4 sker metates systematiskt i mellan 85 – 98 % av fallen, medan den i positionerna 5–6 endast påträffas i 0,8 – 1,4 % av fallen. / This study deals with variation in written Meänkieli, focusing on the variation of one specific feature, h in non-initial syllables (e.g. saunhaan, sauhnaan 'into sauna') which has been found to be rich in variation in spoken language according to earlier research (e.g. Mantila 1992, 2010; Vaattovaara 2009). The aim of this study is to investigate if, and to what extent, the variation patterns found in the spoken variety correspond to literature written in Meänkieli. The study is motivated by the fact that Meänkieli is currently undergoing standardization process. The variation of h in non-initial syllables forms a complex system in spoken language which, as a folk belief, is “difficult to describe”. However, earlier research has indicated that in the spoken language the use of the h is linguistically quite systematically structured and the variation patterns have developed in certain direction across time. In the interest of the present study is to investigate if the pattern is similar in the written form in the chosen corpus, that is, if the individual authors are implementing a similar system in their literary work. The principal aim of this study is not to deal with language planning and principles of language standardization but to form an overall picture of linguistic variation in the use of h in written language. The data consist of 43 books from 18 authors, and from three decades (1980–2000). From each publication, the first 5250 words are extracted into analysis, if the word count exceeds this number (children’s books, 18 included in the corpus, typically do not). The main interest of the current work lies in the question whether or not there is remarkable variation found between the different authors, and what kind of variation may appear within the texts authored by each individual author or across time. The aim of the study is to reveal if the variation patterns possibly scatter, or if the patterns indicate agreement between the individual authors. According to sociolinguistic theories of linguistic change, the so called real time investigation across the decades should indicate if there is an ongoing change. The hypothesis of the research (mainly on the basis of Mantila 2010) is that the variation pattern similar to spoken language is being adopted also in written forms. Since the h in non-initial syllables is both from the point of view of historical development as well as morphological and phonological conditions a very rich feature, this study focuses only on the six linguistic environments concerning the metathetic h variant (e.g. 1. Stockholmhiin &gt; Stockholhmiin 'to Stockholm', 2. koulhuun &gt; kouhluun 'to school', 3. korjaamhaan  &gt; korjaahmaan 'to clean', 4. Pajalhaan &gt; Pajahlaan 'to Pajala', 5. istumhaan &gt;  istuhmaan ' to sit', 6. ymmärtänheet &gt; ymmärtähneet 'have understood'), which always involves a voiced consonant. Earlier research has regarded this particular variant as “the trademark of Torne Valley” which has, on the present Finnish and Swedish side of the area, shown a clear pattern of variation. In this study, the variation of the metathetic h in Meänkieli literature is mirrored against earlier findings from the Torne Valley spoken data. Bengt Pohjanen (with the largest number of publications) together with three other authors have been investigated in more detail. An interesting finding is that there is relatively much variation among authors from one municipality, Pajala. Against the traditional sociolinguistic expectations, the origins of the authors is not a governing factor in linguistic choices. The main outcome of the whole analysis is that during the three decades investigated, individual books have indicated unexpected variation in the 1980s and 1990s, but by the 2000’s the same variation pattern found in spoken language has been more systematically and strongly adopted also into written Meänkieli by more or less all of the authors: in the phonotactic positions 1–4 the metathesis is systematic (85–98 %) while in the positions 5–6 only 0,8 to 1,4 %.