Cette thèse s’intéresse aux réseaux de confusion comme représentation compacte et structurée des hypothèses multiples produites par un moteur de reconnaissance de parole et transmises à un module de post-traitement applicatif. Les réseaux de confusion (CN pour Confusion Networks) sont générés à partir des graphes de mots et structurent l’information sous la forme d’une séquence de classes contenant des hypothèses de mots en concurrence. Le cas d’usage étudié dans ces travaux est celui des hypothèses de reconnaissance transmises à un module de compréhension de la parole dans le cadre d’une application de dialogue déployée par France Telecom. Deux problématiques inhérentes à ce contexte applicatif sont soulevées. De façon générale, un système de dialogue doit non seulement reconnaître un énoncé prononcé par un utilisateur, mais aussi l’interpréter afin de déduire sons sens. Du point de vue de l’utilisateur, les performances perçues sont plus proches de celles de la chaîne complète de compréhension que de celles de la reconnaissance vocale seule. Ce sont ces performances que nous cherchons à optimiser. Le cas plus particulier d’une application déployée implique de pouvoir traiter des données réelles et donc très variées. Un énoncé peut être plus ou moins bruité, dans le domaine ou hors-domaine, couvert par le modèle sémantique de l’application ou non, etc. Étant donnée cette grande variabilité, nous posons la question de savoir si le fait d’appliquer les mêmes traitements sur l’ensemble des données, comme c’est le cas dans les approches classiques, est une solution adaptée. Avec cette double perspective, cette thèse s’attache à la fois à enrichir l’algorithme de construction des CNs dans le but d’optimiser globalement le processus de compréhension et à proposer une stratégie adéquate d’utilisation des réseaux de confusion dans le contexte d’une application réelle. Après une analyse des propriétés de deux approches de construction des CNs sur un corpus de données réelles, l’algorithme retenu est celui du "pivot". Nous en proposons une version modifiée et adaptée au contexte applicatif en introduisant notamment un traitement différencié des mots du graphe qui privilégie les mots porteurs de sens. En réponse à la grande variabilité des énoncés à traiter dans une application déployée, nous proposons une stratégie de décision à plusieurs niveaux qui vise à mieux prendre en compte les spécificités des différents types d’énoncés. Nous montrons notamment qu’il est préférable de n’exploiter la richesse des sorties multiples que sur les énoncés réellement porteurs de sens. Cette stratégie permet à la fois d’optimiser les temps de calcul et d’améliorer globalement les performances du système / The work presented in this PhD deals with the confusion networks as a compact and structured representation of multiple aligned recognition hypotheses produced by a speech recognition system and used by different applications. The confusion networks (CN) are constructed from word graphs and structure information as a sequence of classes containing several competing word hypothesis. In this work we focus on the problem of robust understanding from spontaneous speech input in a dialogue application, using CNs as structured representation of recognition hypotheses for the spoken language understanding module. We use France Telecom spoken dialogue system for customer care. Two issues inherent to this context are tackled. A dialogue system does not only have to recognize what a user says but also to understand the meaning of his request and to act upon it. From the user’s point of view, system performance is more accurately represented by the performance of the understanding process than by speech recognition performance only. Our work aims at improving the performance of the understanding process. Using a real application implies being able to process real heterogeneous data. An utterance can be more or less noisy, in the domain or out of the domain of the application, covered or not by the semantic model of the application, etc. A question raised by the variability of the data is whether applying the same processes to the entire data set, as done in classical approaches, is a suitable solution. This work follows a double perspective : to improve the CN construction algorithm with the intention of optimizing the understanding process and to propose an adequate strategy for the use of CN in a real application. Following a detailed analysis of two CN construction algorithms on a test set collected using the France Telecom customer care service, we decided to use the "pivot" algorithm for our work. We present a modified and adapted version of this algorithm. The new algorithm introduces different processing techniques for the words which are important for the understanding process. As for the variability of the real data the application has to process, we present a new multiple level decision strategy aiming at applying different processing techniques for different utterance categories. We show that it is preferable to process multiple recognition hypotheses only on utterances having a valid interpretation. This strategy optimises computation time and yields better global performance

In this research, we trace the sociolinguistic profile of the speakers from the state of Alagoas concerning the variation of the verbs "ter" and "haver" in existential constructions, in order to analyse how that variation occurs and verify if the alternance of those verbs, in the alagoana speech, reflects either a process of stable variation or a change in progress. For the development of this work, we appeal to the Theory of Linguistics Variation (LABOV, 2008[1972]) which deals the variation and linguistics change and includes the variable use of the language in its social context. In this way, after the delimitation of the dependent variable and of the independent variables selected as potentially relevant in the variation under analysis, namely, animacity of the internal argument, nature of the internal argument, verbal tense, gender, age and scholarity, we stratify our sample and we did the gathering and transcription of data, as well as, the analysis and codification of all existential constructions formed with the verbs present in the corpus. For the quantitative analysis, we use the software VARBRUL, which delimited the variables statically nonsignificant and the variables statically significant in variation under study, showing not only that there is variation “ter” and “haver” existentials in the community studied and that the uses of “ter” existential is greater than the uses of “haver” existential, as well as that such variation is conditionated by the factor groups scholarity, verbal tense, age and nature of internal argument, leading us to argue that, in the alagoana speech, such variation reflects a process of change in cours in the direction of “ter” existential. / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Nesta pesquisa, traçamos o perfil sociolinguístico dos falantes alagoanos em relação à variação dos verbos ter e haver em construções existenciais, com o intuito de analisar como essa variação ocorre e de verificar se a alternância desses verbos, na fala alagoana, reflete um processo de variação estável ou de mudança em progresso. Para o desenvolvimento deste estudo, recorremos à Teoria da Variação Linguística (LABOV, 2008[1972]) que trata da variação e da mudança linguística e contempla os usos variáveis da linguagem em seu contexto social. Dessa forma, após a delimitação da variável dependente e das variáveis independentes selecionadas como potencialmente relevantes na variação em estudo, a saber, animacidade do argumento interno, natureza do argumento interno, tempo verbal, sexo, faixa etária e escolaridade, estratificamos nossa amostra e fizemos não só a coleta e a transcrição dos dados, mas também a análise e a codificação de todas as construções existenciais formadas com os verbos ter e haver presentes no corpus. Para a análise quantitativa dos dados, utilizamos o programa computacional VARBRUL, que delimitou as variáveis estatisticamente não significativas e as variáveis estatisticamente significativas na variação em estudo, mostrando não só que há variação ter e haver existenciais na comunidade estudada e que o uso de ter existencial é bem maior do que o uso de haver existencial, como também que tal variação é condicionada pelos grupos de fatores escolaridade, tempo verbal, faixa etária e natureza do argumento interno, levando-nos a argumentar que, na fala alagoana, tal variação reflete um processo de mudança em curso na direção de ter existencial.

El procesamiento del lenguaje natural es un área de aplicación de la inteligencia artificial, en particular, del reconocimiento de formas que estudia, entre otras cosas, incorporar información sintáctica (modelo de lenguaje) sobre cómo deben juntarse las palabras de una determinada lengua, para así permitir a los sistemas de reconocimiento/traducción decidir cual es la mejor hipótesis �con sentido común�. Es un área muy amplia, y este trabajo se centra únicamente en la parte relacionada con el modelado de lenguaje y su aplicación a diversas tareas: reconocimiento de secuencias mediante modelos ocultos de Markov y traducción automática estadística. Concretamente, esta tesis tiene su foco central en los denominados modelos conexionistas de lenguaje, esto es, modelos de lenguaje basados en redes neuronales. Los buenos resultados de estos modelos en diversas áreas del procesamiento del lenguaje natural han motivado el desarrollo de este estudio. Debido a determinados problemas computacionales que adolecen los modelos conexionistas de lenguaje, los sistemas que aparecen en la literatura se construyen en dos etapas totalmente desacopladas. En la primera fase se encuentra, a través de un modelo de lenguaje estándar, un conjunto de hipótesis factibles, asumiendo que dicho conjunto es representativo del espacio de búsqueda en el cual se encuentra la mejor hipótesis. En segundo lugar, sobre dicho conjunto, se aplica el modelo conexionista de lenguaje y se extrae la hipótesis con mejor puntuación. A este procedimiento se le denomina �rescoring�. Este escenario motiva los objetivos principales de esta tesis: � Proponer alguna técnica que pueda reducir drásticamente dicho coste computacional degradando lo mínimo posible la calidad de la solución encontrada. � Estudiar el efecto que tiene la integración de los modelos conexionistas de lenguaje en el proceso de búsqueda de las tareas propuestas. � Proponer algunas modificaciones del modelo original que permitan mejorar su calidad / Zamora Martínez, FJ. (2012). Aportaciones al modelado conexionista de lenguaje y su aplicación al reconocimiento de secuencias y traducción automática [Tesis doctoral no publicada]. Universitat Politècnica de València. https://doi.org/10.4995/Thesis/10251/18066 / Palancia

Mastery of basic concepts is an academic building block for preschool children in early education programs. Research shows that understanding basic concepts (e.g. top, under, fast, now, all, behind, full and short) is important for academic success and higher order thinking. Experts in the field of concept acquisition agree on six strategies for teaching basic concepts. These strategies include: using positive examples and negative examples, highlighting critical features of concepts through continuous conversion, isolating the concept, the order in which the examples are presented, and teaching generalization. This study investigated the extent to which nine preschool teachers of children with hearing loss used four of the six strategies (using examples, non examples, continuous conversion, and isolating the concept) during a 20-minute lesson in which a new basic concept was taught. Results indicated that teachers do well with using examples to teach basic concepts, but they lack sufficient use of the other three strategies for teaching basic concepts.

No description available.

<p dir="ltr">This dissertation presents a analysis of negation in a spoken language, i.e., Khuzestani Arabic (KhA), and a sign language, i.e., Sadat Tawaher Sign Language (STSL). STSL emerged naturally without any intervention such as deaf education after a man lost his hearing around sixty years ago in a small village named Sadat Tawaher located in southwestern Iran. After this incident, the deaf person's family came up with a gestural system to communicate with him. Despite the fact that everyone in Sadat Tawaher, including the deaf person's family, speaks KhA, I hypothesized that KhA and STSL possess different grammatical ways to express negation. Data gathered using signed productions, story-telling, and grammaticality judgments clearly showed that negation is preverbal in KhA but sentence-final in STSL. </p>

La tesi fornisce uno studio empirico relativo agli elementi linguistici caratterizzanti il parlato faccia-a-faccia e il parlato filmico americano, due domini conversazionali solitamente detti differire in termini di spontaneità, essendo il primo generalmente descritto come la quintessenza del linguaggio parlato (in quanto totalmente spontaneo) e il secondo come non-spontaneo (essendo scritto-per-essere-parlato) e, quindi, non adatto a rappresentare l'uso generale della conversazione. Entrambe le analisi (i.e. quella multi-dimensionale, che offre una panoramica generale dei due domini presi in considerazione, e quella più specifica relativa al comportamento linguistico dell’espressione you know) basate su esempi autentici tratti da corpora dimostrano che, nonostante quanto venga generalmente descritto dalla letteratura a riguardo, conversazione faccia-a-faccia e conversazione filmica hanno molti tratti in comune e confutano l’idea che il linguaggio filmico non possa essere rappresentativo dell'uso generale della conversazione. / The present dissertation examines empirically the linguistic features characterizing American face-to-face and movie conversation, two domains which are usually claimed to differ especially in terms of spontaneity. Natural conversation is, indeed, considered the quintessence of the spoken language for it is totally spontaneous, whereas movie conversation is usually described as non-spontaneous, being artificially written-to-be spoken and, thus, not likely to represent the general usage of conversation. In spite of what is generally maintained by the literature, both the Multi-Dimensional analysis and the micro-analysis of the functions of you know based on authentic data retrieved from corpora show that the two conversational domains do not differ to a great extent and thus confutes the claim that movie language has “a very limited value” in that it does not reflect natural conversation and, consequently, is “not likely to be representative of the general usage of conversation”.

This work has taken the region of Qwaqwa as a case study. Through this study, the researcher attempted to join in the debate around language varieties that occur as a result of contact between different language groups. To achieve this objective, the factors that have an impact on Sesotho spoken in the Qwaqwa area and, in particular, in schools, have been assessed. The researcher provides a broad and general picture of the language situation and patterns of language use in the Qwaqwa area. A brief overview of the geographical description, historical background and economic development of Qwaqwa is given. Some of the linguistic phenomena that play a role in language variation in this area fall under the spotlight. Language contact, i.e. language and dialect, regional and social dialect, the use of language and the impact of language contact between languages is discussed. Patterns and the extent of language contact and the resultant effects of interference, codeswitching and borrowing as well as the processes and points at which these processes occur are identified. The major similarities and relationships between spoken and written Sesotho, as used by learners in Qwaqwa schools, is highlighted with the discussion focussing on the linguistic description of the similarities and relationships between the two forms. / African Languages / (M.A.(Afican Languages))

The purpose of this thesis was to build a specialised corpus of Banatian Czech. The theoretical section describes the historical and language situation in the Banat area focusing on its development since the 1990s to the present day. This is followed by a presentation and description of specialised corpora focusing on a corpus of spoken Czech. Building a specialised corpus named BANAT is documented in detail (method of collection, processing and transcription of records, building of corpora and publishing). The analysis of possessive pronouns, negative pronouns and adverbs and enclitics confirms a statistically significant difference between Banatian Czech and the present state of Czech language (in all observed phenomena except enclitics). Keywords Banat, Banatian Czech, Bigr, corpus BANAT, corpus building, Czech language abroad, pronouns, specialised corpora, spoken language

A conversação é fonte de grande parte de nossa noção de ordem social e, além disso, apresenta sua própria ordem e manifesta um senso próprio de estrutura. Ao interagirem através da conversação, as pessoas recorrem a procedimentos regulares através dos quais tornam explícitas as marcas de planejamento de seu discurso e as estratégias de gestão dos turnos de conversação, das relações estabelecidas e da interação verbal em si. A língua, portanto, não pode ser considerada simplesmente um código de comunicação, mas sim um meio através do qual atuam sujeitos que manifestam suas ideias sobre o mundo real e sobre seu próprio mundo interior, expressam seus papéis sociais e sua personalidade em um determinado contexto e, ainda, por mecanismos metalinguísticos, estabelecem relações de coesão e coerência nos seus próprios textos através dos vínculos que a linguagem verbal estabelece com ela mesma. Nesse processo, os marcadores discursivos (MD) exercem papel de extrema importância porque possibilitam e fazem funcionar mecanismos de atuação e manifestação dos sujeitos da enunciação. Entretanto, uma análise atenta dos materiais utilizados para o ensino de Português- Língua Estrangeira (PLE) no Brasil revelará que as situações de conversação a que são expostos os alunos pouco se utilizam desses elementos e processos típicos da interação verbal e acabam por prender-se à sintaxe do português culto escrito. Como resultado ao contrário de materiais didáticos de outros idiomas como o inglês, o espanhol e o francês, por exemplo os materiais de PLE incorrem na falha de não equiparem os estrangeiros com ferramentas tais que os possibilitem interagir com falantes de português em situações reais de conversação. Tendo como arcabouço teórico a Análise da Conversação, este trabalho apresenta a importância da presença dos processos da conversação no ensino de língua estrangeira e propõe quatro níveis de exploração dos MD com vistas a construir nos estrangeiros aprendizes de português uma real competência comunicativa e sociocultural. / A great deal of our sense of social order stems from conversation, which in turn has its own order and works according to its own rules of structure. As they interact through conversation, people resort to regular procedures to produce explicit marks of discourse planning along with strategies for managing turn-taking, established relationships and oral interaction itself. Therefore, language should not be considered merely as a communication code, but rather a means through which subjects can play an active role while manifesting their ideas about the real world and their own inner world, and expressing their social roles and personality within a given framework. These subjects also engage in metalinguistic mechanisms to create cohesion and coherence relationships as oral language refers to itself in discourse. Discourse markers play a fundamental role in this process because they enable subjects to act and manifest their own selves while producing utterances. However, a close look at the materials currently used for teaching Portuguese as a Foreign Language (PFL) will show that the conversation instances to which PFL students are exposed seldom use these elements and processes so typical of oral interaction. As opposed to other foreign language materials such as English, Spanish and French, PFL materials tend to insist on forms that reflect educated written Portuguese syntax. As a result, PFL material fail to provide foreigners with useful tools to interact with Portuguese speakers in real conversation. This paper highlights the importance of exploring conversation in foreign language teaching on the basis of Conversation Analysis theories. It also suggests four levels in which discourse markers can be explored in order to enable PFL students to build up effective communicative and sociocultural competence.