Notre étude est dédiée au développement d’un protocole d’évaluation du langage à visée clinique. Nous examinons les spécificités du traitement du langage oral à travers le filtre des fonctions langagières actualisées par la prosodie. Système composite par excellence, largement négligé tant dans l’étude du traitement du langage que dans les évaluations orthophoniques, la prosodie est cependant en étroite relation avec les composantes du langage et réalise par ce biais l’ensemble des fonctions structurantes du langage. Deux fonctions linguistiques de la prosodie sont ici testées en compréhension : une fonction syntaxique, sous-tendue par le marquage accentuel des frontières de syntagmes réalisé par un accent final et un accent initial, et une fonction pragmatique impliquant la fonction de focalisation initiée par un accent emphatique. Après avoir procédé à la normalisation de notre protocole auprès d’un groupe de sujets contrôles, notre évaluation a été intégrée à la prise en charge longitudinale de deux patients atteints de gliomes de grade II. Dans ce cadre spécifique, nous avons adapté nos épreuves au contexte de la chirurgie éveillée. Les capacités langagières des patients ont été testées en condition pré-opératoire afin d’observer l’impact cognitif de la tumeur et en condition post-opératoire afin d’évaluer les conséquences de l’exérèse. Les résultats soulèvent des questionnements concernant la spécialisation hémisphérique de la prosodie. Ils ont aussi permis de mesurer la sensibilité de notre protocole. L’étude exploratoire réalisée en condition peri-opératoire nous a permis de soulever les contraintes inhérentes à la méthode de stimulation électrique directe. / This study is dedicated to the development of an exploratory clinical protocol for language evaluation, where prosody is seen as the core system for assessing spoken language processing in a more comprehensive way than is usually the case in clinical studies. Indeed, prosody is usually largely neglected in studies on language processing, as well as in speech therapy and neuropsychological evaluations. However, prosody is tightly linked to the different components of language, and plays a central role in linguistic structuring. Two linguistic functions of prosody are tested here in language comprehension: 1) the syntactic function of prosody, where both final and initial accenting help delimit syntagmatic grouping ; 2) the pragmatic function of prosody, where prosodic focus brings salience to a specific part of discourse. The protocol was first normalized on a control group of healthy participants. We then ran the evaluation in a peri-surgical context on two patients suffering from grade II glioma. The protocol had to be adapted to the context of awake surgery. The language capacities of the patients were tested first in pre-surgical context in order to assess the cognitive consequences of the brain tumor, then in post-surgical context to evaluate the consequences of the tumor removal. Results raise questions regarding the hemispheric specialization of prosody, and more globally regarding the localization of language processing. The other interesting outcome of this research is to propose improvements for the evaluation of language capacities in a surgical context using direct brain electric stimulation.

In the L2 learning process, prosody is among the most determinant linguistic features for a fully pragmatic competence. Additionally, this ability is often crucial for the disambiguation of certain structures and thus also crucial for perceptive acceptability. Italian prosodic configurations dramatically vary through the diatopic dimension. Nevertheless, certain prosodic structures and patterns can still be perceived as effective and acceptable. With these perspectives, we explored the oral acceptability in L2 of advanced Swedish learners of Italian through perceptive judgments by natives. Despite the lack of acoustic analysis and the limitation in the size of the sample of our study, the results showed a strong indication for a regional variation of the perception by the Italian natives, not only considering the production of learners but also considering the natives themselves. Finally, we discuss some of the didactic implications which may be considered particularly useful for native teachers in L2 contexts.

Cette thèse porte sur des analyses acoustiques et prosodiques du français à partir de grandes masses de données orales illustrant différents styles de parole (préparée et spontanée). Nous nous sommes intéressées aux attributs acoustiques et prosodiques qui pourraient caractériser la prononciation. En français, de nombreuses erreurs de reconnaissance automatique de la parole (RAP) sont dues à des mots fréquents homophones. Pour ces mots, la solution correcte dépend du modèle de langage. Une classification automatique (CA) a été effectuée pour discriminer deux paires homophones (‘et/est’ et ‘à/a’) par des propriétés acoustiques et prosodiques. Les résultats de la CA ont montré que le paire ‘et/est’ était plus dissociable. La CA par des attributs prosodiques et inter-segmentaux (15 attributs) s’est avérée aussi performante que celle utilisant la totalité des 62 attributs. Un test perceptif a été également effectué pour vérifier si les humains utilisaient eux aussi ces paramètres. Les résultats ont suggéré que des informations acoustiques et prosodiques pourraient être utiles pour effectuer un choix correct de mots dans des structures syntaxiquement ambigües. Ensuite, nous avons examiné des propriétés prosodiques globales aux niveaux du nom et du syntagme nominal. La comparaison entre mots lexicaux et grammaticaux a montré que la fréquence fondamentale (F0) montante et l’allongement vocalique de la dernière syllabe caractérisent les mots lexicaux, par opposition aux mots grammaticaux. Ainsi, le profil de F0 moyenne d’un syntagme nominal de longueur n pourrait être différent de celui du nom avec une valeur de F0 basse au début du syntagme. Les profils prosodiques peuvent être utiles pour localiser frontières de mots. Les résultats de ce travail pourront servir à localiser le focus et les entités-nommées par des classifieurs discriminants, et de manière plus générale à améliorer les techniques de localisation des frontières des mots pour la RAP. / This thesis focuses on acoustic and prosodic (fundamental frequency (F0), duration, intensity) analyses of French from large-scale audio corpora portraying different speaking styles: prepared and spontaneous speech. We are interested in particularities of segmental phonetics and prosody that may characterize pronunciation. In French, many errors caused by automatic speech recognition (ASR) systems arise from frequent homophone words, for which ASR systems depend on language model weights. Automatic classification (AC) was conducted to discriminate homophones by only acoustic and prosodic properties depending on their part-of-speech function or their position within prosodic words. Results from AC of two homophone pairs, et/est (and/is) and à/a (ton/has), revealed that the et/est pair was more discriminable. A selection of prosodic and inter-phoneme attributes, that is 15 attributes, performed as good results as with 62 attributes. Then corresponding perceptual tests have been conducted to verify if humans also use acoustico-prosodic parameters for the discrimination. Results suggested that acoustic and prosodic information might help in operating the correct choice in similar ambiguous syntactic structures. From the hypothesis that pronunciation variants were due to varying prosodic constraints, we examined overall prosodic properties of French on a lexical and phrase level. The comparison between lexical and grammatical words revealed F0 rise and lengthening at the end of final syllable on lexical words, while these phenomena were not observed for grammatical words. Analyses also revealed that the mean profile of a n length noun phrase could be different from that of a n length noun with a low F0 at the beginning of a noun phrase. The prosodic profiles can be helpful to locate word boundaries. Findings in this thesis will lead to localize focus and named-entity using discriminative classifiers, and to improve word boundary locations by an ASR post-processing step.

Esta tese de doutorado A entoação do dialeto caipira no Médio Tietê: reconhecimento, características e formação faz parte do Projeto ExProsodia, em auxílio ao PHPP (Projeto História do Português Paulista), que objetiva analisar o português paulista falado no interior do estado. Sendo um dos trabalhos que está dando embasamento a ele, esta pesquisa subsidia os estudos do idioma, trazendo, além do exame da entoação, um breve histórico do Brasil e das cidades interioranas paulistas e portuguesas, de modo a entendermos como foi estabelecido o dialeto caipira nessas localidades. Tendo isso feito, avaliou-se a entoação de falantes do Médio Tietê por meio de análise automatizada promovida pelo ExProsodia. Seguindo a linha de pesquisa de Ferreira Netto (2008), examinou-se a fala dos informantes a partir da análise de f0, particularizando dela o tom médio (TM) e o tom final (TF). Para isso, foram selecionados 40 entrevistados, sendo 28 de sete cidades que compõem o Médio Tietê (Santana de Parnaíba, Pirapora do Bom Jesus, Porto Feliz, Tietê, Capivari, Itu, Piracicaba), e 12 de três cidades do norte de Portugal (Braga, Bragança e Vila Real), além de 10 gravações (5 vozes masculinas e femininas) de jornalistas do programa SPTV da emissora de televisão Rede Globo - tomados como fala neutra (BARBOSA, 2002, p.36) para serem usados como controle. Em cada cidade mencionada, buscaram-se 4 informantes, sendo dois homens e duas mulheres com idades iguais ou superiores a 60 anos, de baixo ou nenhum grau de escolaridade e moradores vitalícios de suas regiões. De seus depoimentos foram colhidas 5 frases para observação, totalizando o valor de 220 dados. Desta forma, por meio de aplicativos especializados (SFS e Audacity), e com especial uso do ExProsodia® para geração de valores para o cálculo estatístico, foi possível verificar que o TM e o TF do dialeto caipira apresentam, em diversas análises, características comuns ao do controle e ao dos falantes de Portugal; ademais, observou-se que a finalização apresentada pelos informantes brasileiros e portugueses tendem a ser menos acentuada, caracterizando um traço plagal na fala. Os mesmos resultados também foram observados com os estudos de Costa e Baz (ambos de 2011), em que foram analisadas as falas de índios guatós e guaranis paraguaios. Sendo assim, este estudo pode gerar algumas conclusões: primeiramente, a finalização autêntica é uma variação entoacional, já que a plagal não é uma característica própria de uma região, logo, como ela é encontrada nos falares de pessoas com baixa ou nenhuma escolaridade, entende-se que este seria um traço comum da fala, o qual o dialeto caipira também mantém. Além disso, como o caipira apresenta o TM e o TF similares ao do controle, demonstra-se que seus falares possuem inflexões como em outras falas, não sendo plano ou igual; e essas considerações atualizam a informação de Amaral (1955, p. 45) sobre esse assunto. / This doctoral dissertation aims at analyzing the Portuguese dialect (caipira) spoken in the region of Middle Tietê, on the Brazilian state of São Paulo. Entitled The intonation of the caipira dialect in the Middle Tietê: recognition, characteristics and formation, this thesis is part of the Project ExProsodia in aid to PHPP (Project History of Paulista Portuguese), providing groundwork to other studies of Portuguese dialects. Besides examining aspects of intonation, this work offers a concise historical analysis on how the caipira dialect was established in Brazil, São Paulo and the Portuguese countryside. The empirical material consisted of speeches from 40 interviewees in 7 cities of the Middle Tietê region (28 interviewees) and 3 cities of the Northern Portugal region (12 interviewees). Additionally, 10 recordings of journalists from a major TV news broadcasting in São Paulo (SPTV by Rede Globo) were adopted as controls, based on the assumption of neutral speech (BARBOSA, 2002, p.36). Four interviewees were selected per city (2 men and 2 women), with the inclusion criteria of being at least 60 years old, having low or no formal education, and being ligelong resident at the surveyed areas. Five phrases were analysed per individual, amounting to a total of 220 data samples. The intonations of these interviewees were processed through specialized computer applications (SFS and Audacity) and the special use of an automated analytical tool called ExProsodia®, which statistically discriminated the various tones of the caipira dialect. Following Ferreira Nettos research (2008), speeches were analyzed by selecting the midtones (TM) and the final tones (TF), in the so-called f0 analysis. Common features were observed among the Portuguese interviewees and the Brazilian controls, whereas the phrasal finalization of Brazilian and Portuguese interviewees tended to be less marked, and featured by a plagal trace speech. Similar results were observed by Costa (2011) and Baz (2011), where the lines of Paraguayan Guarani Indians and Guatós Indians were analyzed. This thesis concludes that the authentic finalization is an intonational variation, because the plagal finalization is a region-independent feature occurring in the dialects of individuals with low or no formal education. Hence, the plagal finalization is suggested to be a common trait of speech, also kept by the caipira dialect. Furthermore, as the caipira dialect presents TM and TF similar to the control, it features inflections which are neither plane or equal; such finding warrants an update on Amaral\'s information on the subject (1920, p. 56).

Made available in DSpace on 2016-03-22T17:27:13Z (GMT). No. of bitstreams: 1 cloris.pdf: 3209158 bytes, checksum: 8cac72e9d1c9c3d5ac6d85e12b8d44d4 (MD5) Previous issue date: 2013-12-16 / This study is affiliated to the Discourse Analysis of pecheuxtian tradition, theory which does not conceive meanings in its literal way, not even subjects as the origin of the saying, since they are within the ideology domain and are affected by the unconsciousness functioning. It is through such a perspective that we analyze the juridical discourse of the defense in a trial by jury. On the one hand, this discourse is subordinated to the rules of Law which induce to the illusion of freedom and equality among individuals by using laws, rituals and specific discursive mechanisms. Such mechanisms translate the supposed neutrality and objectivity of the text of the law, which seeks, through the "literalness" of signifiers and the determination of rights and duties inherent to social life, to bequeath to the individual a supposed security. On the other hand, given the need of argument in favor of the defendant of the case at hand, the defense s discourse invests itself of a subjectivity (represented) in a game of making believe. This game is affected by the affective-prosodic-discursive memory, theoretical notion whose claim is to consider, in the interdiscursivity dimension, the effects of prosody that are irrevocably linked to affectivity and, therefore, subjectivity. In this context, it grows the importance of the question concerning the "truth" and the "lie", fundamental focus of this study, which has sought to conceive them in their discursive dimension. For such, we have initially taken a journey through different areas of knowledge towards the re-significance of the mentioned question, considering the assumptions of Discourse Analysis. Having done this theoretical demarcation, we sought, in the analysis, by observing the prosodic materiality, to understand the effects of meaning, derived from a struggle discourse, whose purpose is to prove the defendant s innocence. The analyses have revealed the functioning of the discursive processes involved in the argument of the defense, showing, through the prosodic elements, the effects of "truth" and "lie" that have been made present / O presente trabalho filia-se à Análise de Discurso de tradição pecheutiana, teoria que não concebe os sentidos de forma literal nem os sujeitos como origem do dizer. Isso porque se situam no domínio da ideologia e são afetados pelo funcionamento do inconsciente. É nessa perspectiva que se analisa o discurso jurídico da defesa em um tribunal de júri. Por um lado, esse discurso subordina-se às normas do Direito as quais provocam a ilusão de liberdade e de igualdade entre os indivíduos, utilizando, para tanto, leis, rituais e mecanismos discursivos específicos. Esses mecanismos traduzem a pretensa neutralidade e objetividade do texto da lei, o qual busca, através da literalidade dos significantes e da determinação de direitos e deveres inerentes ao convívio social, legar, ao indivíduo, uma suposta segurança. Por outro, dada a necessidade de argumentação a favor da ré do processo em pauta, o discurso da defesa investe-se de uma subjetividade (representada) num jogo de fazer crer. Esse jogo é afetado pela memória afetivo-prosódica-discursiva, noção teórica cuja pretensão é considerar, no âmbito da interdiscursividade, os efeitos da prosódia que se encontram irremediavelmente ligados à afetividade e, portanto, à subjetividade. Nesse contexto, avulta em importância a questão da verdade e da mentira , preocupação fundamental deste trabalho que procurou concebê-las em sua dimensão discursiva. Para isso, fez-se, inicialmente, um percurso por diferentes áreas do conhecimento até ressignificá-las, considerando os pressupostos da Análise de Discurso. Feita essa delimitação teórica, buscou-se, nas análises, mediante a observação da materialidade prosódica, compreender os efeitos de sentido derivados de um discurso de embate, cuja finalidade é a comprovação da inocência da ré. As análises evidenciam o funcionamento dos processos discursivos envolvidos na argumentação da defesa, mostrando, através dos elementos prosódicos, os efeitos de verdade e de mentira que se fazem presentes

La Dépression Post-AVC (DPAVC) constitue une des complications psychiatriques les plus fréquentes dans les suites d’un Accident Vasculaire Cérébral (AVC) avec des conséquences majeures en termes de récupération fonctionnelle et de qualité de vie. Il est nécessaire de disposer de critères permettant une identification précoce des patients à risque afin de leur proposer des aides adaptées. L’objectif de ce travail de thèse est d’étudier certains facteurs de risque et de protection de la DPAVC avec une approche en vie quotidienne chez des patients présentant un AVC peu invalidant. Dans une première partie, nous présentons l’étude du rôle exercé par le soutien social et la routinisation sur la sévérité des symptômes dépressifs trois mois après l’AVC par la méthode « Ecological Momentary Assessment ». Dans une seconde partie, nous explorons le rôle de la voix comme marqueur précoce de la DPAVC. Nous avons mis en évidence que la perception du soutien social initial influence l’intensité des symptômes dépressifs et les activités trois mois après. D’autre part, nous avons observé une dynamique temporelle unidirectionnelle entre l’augmentation des routines et la survenue de symptômes dépressifs plus intenses. Enfin, les analyses vocales nous ont permis d’identifier des marqueurs de DPAVC avec des modifications longitudinales de la fréquence fondamentale ainsi que des marqueurs précoces (shimmer et ruptures de voisement). Au-delà des facteurs cliniques traditionnellement reconnus, ce travail a mis en évidence de façon écologique de nouveaux marqueurs de risque de DPAVC dont la détection pourrait permettre d’élaborer de nouvelles stratégies de prévention et de prise en charge / Post-stroke depression (PSD) is one of the most common psychiatric complications following stroke with detrimental consequences in terms of functional recovery and quality of life. Its identification and treatment at the earliest stages of care remains a clinical challenge. The aim of this thesis is to study risk and protective factors for PSD based on data collected in the contexts of daily life. In the first section, we present a study of the role played by social support and routinization on the severity of depressive symptoms three months after stroke by Ecological Momentary Assessment (EMA). In the second section, we examine the voice as an early marker of PSD. We observed that patient perceptions of better support quality, and not quantity, immediately following mild stroke, are associated with better behavioral and emotional outcomes three months later. In addition, we observed a unidirectional pattern of association between mood and routines, whereby the occurrence of routines was associated with a higher level of depressive symptomatology over subsequent hours. Finally, voice analysis allowed us to identify PSD markers with longitudinal changes in fundamental voice frequency and discriminant analysis demonstrated that initial voice breaks coupled with shimmer are strongly predictive of subsequent PSD. Beyond commonly-recognized clinical risk factors such as stroke severity, these ecological investigations identified new markers for PSD whose detection could lead to new strategies for prevention and care

Submitted by Alisson Mota (alisson.davidbeckam@gmail.com) on 2015-06-01T20:18:55Z No. of bitstreams: 1 Dissertação - Suzana Pinto do Espírito Santo.pdf: 4456051 bytes, checksum: 1ad6e7775573c65052d04175f44a5897 (MD5) / Approved for entry into archive by Divisão de Documentação/BC Biblioteca Central (ddbc@ufam.edu.br) on 2015-06-01T20:50:39Z (GMT) No. of bitstreams: 1 Dissertação - Suzana Pinto do Espírito Santo.pdf: 4456051 bytes, checksum: 1ad6e7775573c65052d04175f44a5897 (MD5) / Approved for entry into archive by Divisão de Documentação/BC Biblioteca Central (ddbc@ufam.edu.br) on 2015-06-01T20:53:57Z (GMT) No. of bitstreams: 1 Dissertação - Suzana Pinto do Espírito Santo.pdf: 4456051 bytes, checksum: 1ad6e7775573c65052d04175f44a5897 (MD5) / Made available in DSpace on 2015-06-01T20:53:57Z (GMT). No. of bitstreams: 1 Dissertação - Suzana Pinto do Espírito Santo.pdf: 4456051 bytes, checksum: 1ad6e7775573c65052d04175f44a5897 (MD5) Previous issue date: 2014-03-06 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / This study aims to analyze the behavior of speakers of intonation city Maués at Amazon. This is a study linked to AMPER project, which seeks to describe the prosody of the Romance languages and, therefore, uses a corpus of declarative and interrogative sentences total. The sentences of the corpus may have 10, 13 and 14 members, of which the last two feature types with adjectival or prepositional phrase extension, respectively, to the right of the verb, as in "The bird likes the drunk grandfather" or "The bird likes Renato Venice ". Overall, work with a corpus of 396 sentences in which the pre-core and intonational separately core were observed. We describe, for these regions accent, melodic settings of the fundamental frequency, duration and intensity in order to identify the prosodic feature of sentences. The data are graphed automatically generated with the help of manuals and sripts etiquetagens through the Praat software. Our results showed that in the core region, the declarative mode, the initial movement is low, followed by an increase in the frequency syllable stressed and subsequent decline in post-tonic. In the interrogative mode data revealed that the attack syllabic part of a high frequency, followed by fall and subsequent rise in the post-tonic syllable. But beyond that, what caught our attention in relation to the contours exposed in the description was the behavior of the final vowel of the utterances, ie, the erasure process by which passed with no attendance record this frásica position. While for the duration showed no obvious values that leave the distinction between declarative and interrogative sentence, given the proximity of the measures set out in the histograms, nevertheless, in most cases, the questioning had higher measures, especially in stressed syllables. As regards intensity, interrogative exhibited higher energy than in stressed syllables declarative. In male speech that distinction is not very explicit, since the average intensity values are very close, whereas in female speech, occurred more explicit records for the amounts of energy. / O presente trabalho tem como objetivo principal analisar o comportamento entonacional dos falantes da cidade de Maués, no Amazonas. Trata-se de um estudo vinculado ao projeto AMPER, que busca descrever a prosódia das línguas românicas e, para isso, utiliza um corpus de sentenças declarativas e interrogativas totais. As sentenças do corpus podem ter 10, 13 e 14 vogais, sendo que os dois últimos tipos apresentam sintagma com extensão adjetival ou preposicionado, respectivamente, à direita do verbo, como em “O pássaro gosta do bisavô bêbado” ou “O pássaro gosta do Renato de Veneza”. No total, trabalhamos com um corpus de 396 frases, nas quais foram observados o pré-núcleo e o núcleo entonacionais, separadamente. Descrevemos, para essas regiões de acento, as configurações melódicas da frequência fundamental, duração e intensidade, a fim de identificarmos a característica prosódica das frases. Os dados são representados em gráficos gerados automaticamente com o auxílio de sripts e etiquetagens manuais através do software Praat. Nossos resultados mostraram que, na região nuclear, na modalidade declarativa o movimento inicial é baixo, seguido de um aumento de frequência na sílaba tônica e posterior declínio na pós-tônica. Na modalidade interrogativa os dados revelaram que o ataque silábico parte de uma frequência alta, seguida de queda e posterior subida na sílaba pós-tônica. Porém, além disso, o que nos chamou atenção em relação aos contornos expostos na descrição foi o comportamento da vogal final dos enunciados, isto é, o processo de elisão pelas quais passaram, com ausência de registro de frequência nesta posição frásica. Enquanto que para a duração não apresentaram valores que deixassem evidentes a distinção entre a frase declarativa e interrogativa, haja vista a proximidade das medidas expostas nos histogramas, apesar disso, na maioria dos casos, a interrogativa obteve medidas maiores, sobretudo nas sílabas tônicas. No que se refere à intensidade, as interrogativas exibiram energia mais elevada nas sílabas tônicas do que as declarativas. Na fala masculina essa distinção não ficou bem explícita, pois os valores médios de intensidade são muito próximos, enquanto que na fala feminina, ocorreram registros mais explícitos quanto aos valores de energia.

Esta dissertação trata da queda de sílaba em limite de palavra no falar da cidade paulista de Capivari, observando-se em quais contextos esse processo fonológico pode ocorrer e em quais nunca ocorre. Para realizar essa análise, são considerados três níveis fonológicos, quais sejam: segmental, prosódico e métrico. A queda de sílaba é tratada como dois tipos distintos de redução fonológica: na elisão silábica, há dessemelhança entre as consoantes das sílabas envolvidas; com relação à haplologia, as consoantes das sílabas são iguais ou semelhantes ? a diferença entre elas está apenas no traço [sonoridade]. Pôde-se constatar, pela análise de contextos segmentais, que tanto a elisão silábica quanto a haplologia são processos regidos pelas mesmas regras, uma vez que ambos os processos se dão no mesmo ponto de consoante para que possa haver efetiva queda de sílaba. E, da mesma forma, esses processos fonológicos podem ocorrer em qualquer nível da hierarquia prosódica. Isso evidencia que a elisão silábica e a haplologia são de fato o mesmo processo fonológico. Finalmente, a análise métrica aponta as mesmas propriedades para a elisão silábica e para a haplologia no que diz respeito ao uso desses processos para otimizações rítmicas / This work deals with syllable deletion in word boundary in Capivari speech, a city from the countryside of São Paulo state. It was checked in which contexts this phonological process can occur and in which it can never occur. To fulfill this analysis three phonological levels were considered: segmental, prosodic and metrical. Syllable deletion is dealt as two distinct types of phonological reduction: in syllable drop, the syllable\'s consonants are distinct from each other; in haplology, the consonants are equal or alike - the difference between them is only the feature [sonority]. By the analysis of segmental contexts, it has been proved that both syllable drop and haplology are types of processes which are governed by the same rules, for both of them must have the same consonant place node, so that there can be in fact syllable deletion. In the same way, both the phonological processes may occur in any prosodic level from which it is evident that syllable drop and haplology are actually the same phonological process. To verify if both the processes are rhythmically used in the same way it has been pointed out by metrical analysis that syllable drop and haplology present the same properties as regards to optmization

Esta tese de doutorado A entoação do dialeto caipira no Médio Tietê: reconhecimento, características e formação faz parte do Projeto ExProsodia, em auxílio ao PHPP (Projeto História do Português Paulista), que objetiva analisar o português paulista falado no interior do estado. Sendo um dos trabalhos que está dando embasamento a ele, esta pesquisa subsidia os estudos do idioma, trazendo, além do exame da entoação, um breve histórico do Brasil e das cidades interioranas paulistas e portuguesas, de modo a entendermos como foi estabelecido o dialeto caipira nessas localidades. Tendo isso feito, avaliou-se a entoação de falantes do Médio Tietê por meio de análise automatizada promovida pelo ExProsodia. Seguindo a linha de pesquisa de Ferreira Netto (2008), examinou-se a fala dos informantes a partir da análise de f0, particularizando dela o tom médio (TM) e o tom final (TF). Para isso, foram selecionados 40 entrevistados, sendo 28 de sete cidades que compõem o Médio Tietê (Santana de Parnaíba, Pirapora do Bom Jesus, Porto Feliz, Tietê, Capivari, Itu, Piracicaba), e 12 de três cidades do norte de Portugal (Braga, Bragança e Vila Real), além de 10 gravações (5 vozes masculinas e femininas) de jornalistas do programa SPTV da emissora de televisão Rede Globo - tomados como fala neutra (BARBOSA, 2002, p.36) para serem usados como controle. Em cada cidade mencionada, buscaram-se 4 informantes, sendo dois homens e duas mulheres com idades iguais ou superiores a 60 anos, de baixo ou nenhum grau de escolaridade e moradores vitalícios de suas regiões. De seus depoimentos foram colhidas 5 frases para observação, totalizando o valor de 220 dados. Desta forma, por meio de aplicativos especializados (SFS e Audacity), e com especial uso do ExProsodia® para geração de valores para o cálculo estatístico, foi possível verificar que o TM e o TF do dialeto caipira apresentam, em diversas análises, características comuns ao do controle e ao dos falantes de Portugal; ademais, observou-se que a finalização apresentada pelos informantes brasileiros e portugueses tendem a ser menos acentuada, caracterizando um traço plagal na fala. Os mesmos resultados também foram observados com os estudos de Costa e Baz (ambos de 2011), em que foram analisadas as falas de índios guatós e guaranis paraguaios. Sendo assim, este estudo pode gerar algumas conclusões: primeiramente, a finalização autêntica é uma variação entoacional, já que a plagal não é uma característica própria de uma região, logo, como ela é encontrada nos falares de pessoas com baixa ou nenhuma escolaridade, entende-se que este seria um traço comum da fala, o qual o dialeto caipira também mantém. Além disso, como o caipira apresenta o TM e o TF similares ao do controle, demonstra-se que seus falares possuem inflexões como em outras falas, não sendo plano ou igual; e essas considerações atualizam a informação de Amaral (1955, p. 45) sobre esse assunto. / This doctoral dissertation aims at analyzing the Portuguese dialect (caipira) spoken in the region of Middle Tietê, on the Brazilian state of São Paulo. Entitled The intonation of the caipira dialect in the Middle Tietê: recognition, characteristics and formation, this thesis is part of the Project ExProsodia in aid to PHPP (Project History of Paulista Portuguese), providing groundwork to other studies of Portuguese dialects. Besides examining aspects of intonation, this work offers a concise historical analysis on how the caipira dialect was established in Brazil, São Paulo and the Portuguese countryside. The empirical material consisted of speeches from 40 interviewees in 7 cities of the Middle Tietê region (28 interviewees) and 3 cities of the Northern Portugal region (12 interviewees). Additionally, 10 recordings of journalists from a major TV news broadcasting in São Paulo (SPTV by Rede Globo) were adopted as controls, based on the assumption of neutral speech (BARBOSA, 2002, p.36). Four interviewees were selected per city (2 men and 2 women), with the inclusion criteria of being at least 60 years old, having low or no formal education, and being ligelong resident at the surveyed areas. Five phrases were analysed per individual, amounting to a total of 220 data samples. The intonations of these interviewees were processed through specialized computer applications (SFS and Audacity) and the special use of an automated analytical tool called ExProsodia®, which statistically discriminated the various tones of the caipira dialect. Following Ferreira Nettos research (2008), speeches were analyzed by selecting the midtones (TM) and the final tones (TF), in the so-called f0 analysis. Common features were observed among the Portuguese interviewees and the Brazilian controls, whereas the phrasal finalization of Brazilian and Portuguese interviewees tended to be less marked, and featured by a plagal trace speech. Similar results were observed by Costa (2011) and Baz (2011), where the lines of Paraguayan Guarani Indians and Guatós Indians were analyzed. This thesis concludes that the authentic finalization is an intonational variation, because the plagal finalization is a region-independent feature occurring in the dialects of individuals with low or no formal education. Hence, the plagal finalization is suggested to be a common trait of speech, also kept by the caipira dialect. Furthermore, as the caipira dialect presents TM and TF similar to the control, it features inflections which are neither plane or equal; such finding warrants an update on Amaral\'s information on the subject (1920, p. 56).

For people without speech, voice output communication aids are an assistive technology––but can also be restrictive: whilst Text-To-Speech synthesis can say anything, it affords little choice of how this is spoken. An absence of nuanced tone of voice can inhibit social interaction. This research explores this profound but relatively overlooked issue in augmentative and alternative communication through the lens––with the sensibilities and skills––of interaction design. Tone of voice is such an elusive and intangible quality: difficult for even phoneticians to define, let alone AAC users and carers to discuss in the context of their everyday lives. Therefore the activities of design exploration and design practice have been employed to visualise tone of voice, in order to catalyse new conversations, through two original research projects: Six Speaking Chairs, curated with Andrew Cook, is a collection of interactive artefacts that illustrate alternative models of tone of voice developed by academics and practitioners as diverse as sociolinguists and playwrights;Speech Hedge, created with the assistance of Ryan McLeod, is a visualisation of how someone might interact with nuanced tone of voice using a conventional communication aid in combination with an interface on a smart phone. Audience responses to each project have illuminated the perspectives from which laypeople conceive of tone of voice, challenging the conventional emotional model that dominates speech technology in favour of something more complex and heterogeneous. In order to reconcile such complexity with simplicity of use, design principles have been distilled that could inspire future user interfaces but also inform further research. This research has been published and presented within different academic fields, including design research, interaction design and augmentative and alternative communication.