nÃo hà / Nesta pesquisa, analisamos as caracterÃsticas e mecanismos que tornam a categorizaÃÃo, como processo cognitivo, um Sistema Adaptativo Complexo e as estratÃgias de categorizaÃÃo que atuam na emergÃncia de inferÃncias para a conceitualizaÃÃo da categoria VIOLÃNCIA e a subcategoria VIOLÃNCIA URBANA. Nosso suporte teÃrico para a investigaÃÃo dos nossos objetivos sÃo os pressupostos do paradigma do caos, da complexidade e dos sistemas complexos, conforme delineados por Bertalanffy (1977), Morin (2005), Holland (1995; 1998) e Larsen- Freeman e Cameron (2008; 2012), que propÃem os conceitos de sistemas, complexidade, atratores, espaÃo fase e caracterÃsticas e mecanismos de um Sistema Adaptativo Complexo. No caso do processo inferencial, buscamos amparo teÃrico na Teoria da RelevÃncia, conforme proposta por Sperber e Wilson (1995; 2001), Feltes (1999; 2007), Alves e GonÃalves (2006) e Yus (2008; 2013). Para chegarmos à caracterizaÃÃo da categorizaÃÃo como Sistema Adaptativo Complexo, levamos em consideraÃÃo as propriedades apresentadas por Holland (1995) e Larsen- Freeman e Cameron (2008), buscando ampliar o conceito e explicar a instabilidade do sistema categorizacional à luz da complexidade (MORIN, 1977). Para essa investigaÃÃo com base na complexidade foi necessÃrio ainda incluir o conceito de sistema, espaÃo fase e atratores tÃo caros à abordagem metodolÃgica utilizada. Esse procedimento resultou em uma tipologia de estratÃgias de categorizaÃÃo para anÃlise e explicitaÃÃo de como se aciona os diversos espaÃos possÃveis para conceitualizaÃÃo de VIOLÃNCIA. Escolhemos a categoria VIOLÃNCIA para investigar nosso objetivo tendo em vista a atualizaÃÃo do assunto nos Ãltimos vinte anos e pelos trabalhos com essa categoria realizados por Larsen-Freeman e Cameron, Macedo e Feltes, cujos estudos serviram de base para nossa proposta metodolÃgica. Para verificarmos nossas hipÃteses, utilizamos como desenho metodolÃgico uma pesquisa com observaÃÃo direta e intensiva de 33 categorizadores que responderam a questionÃrios sobre a categorizaÃÃo de VIOLÃNCIA e participaram de protocolos verbais para verificaÃÃo dos mecanismos de inferenciaÃÃo. Os resultados das anÃlises permitem as seguintes conclusÃes: a categorizaÃÃo de VIOLÃNCIA possui propriedades e mecanismos dos Sistemas Adaptativos Complexos, pois os sistemas apresentam no todo e nas partes, variedade dentro de uma estabilidade. Os categorizadores utilizam o processo inferencial para acionar os atratores que levam ao espaÃo fase em que se encontram diversos conhecimentos sobre violÃncia para sua conceitualizaÃÃo de forma estratÃgica. / In this research, I analyzed the characteristics and mechanisms that make categorization, as a cognitive process, a Complex Adaptive System and the strategies of categorization that work in the emergency of inferences for the conceptualization of the category VIOLENCE and subcategory URBAN VIOLENCE. The theoretical support for the investigation of the research aims are the assumptions of the paradigm of chaos, complexity and complex systems, as outlined by Bertalanffy (1977), Morin (2005), Holland (1995, 1998) and Larsen-Freeman and Cameron ( 2008; 2012), who propose the concepts of systems, complexity, attractors, phase space and characteristics and mechanisms of a Complex Adaptive System. In the case of inferential process, I sought theoretical support in Relevance Theory, as proposed by Sperber and Wilson (1995, 2001), Feltes (1999, 2007), Alves and GonÃalves (2006) and Yus (2008, 2013). To get to the characterization of categorization as a Complex Adaptive System, I considered the properties presented by Holland (1995) and Larsen-Freeman and Cameron (2008), seeking to expand the concept and explain the instability of the categorical system based on the complexity theory (Morin, 1977). For this investigation, based on the complexity theory, it was also necessary to include the concept of system, phase space and attractors so relevant to the methodological approach used in this research. Such a procedure resulted in a typology of categorization strategies for the analysis and explanation of how the various feasible spaces for the conceptualizing of VIOLENCE are triggered. The category "VIOLENCE" was chosen for analysis in view of its update status in the last twenty years and the various researches on the subject carried out by Larsen-Freeman and Cameron, Macedo and Feltes, scholars whose studies served as basis for the methodological proposal of this thesis. In order to verify the research hypotheses, a methodological design which involved intensive direct observation of 33 categorizers was used. The participants answered questionnaires about the categorization of VIOLENCE and participated of verbal protocols to verify the inference mechanisms involved in the process. The analyses result allow for the following conclusions: VIOLENCE categorization has the properties and mechanisms of Complex Adaptive Systems because the systems present, in whole and in parts, variation within a stable range. The categorizers use the inferential process to trigger the attractors that lead to the phase space in which diverse knowledge about violence is available for its conceptualization in a strategic way.

Submitted by Irene Nascimento (irene.kessia@ufpe.br) on 2016-06-29T16:53:02Z No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Tese Final - Ilka.pdf: 1372344 bytes, checksum: 39673ba3ef3d0f8d6e44ff59bc66eaff (MD5) / Made available in DSpace on 2016-06-29T16:53:02Z (GMT). No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Tese Final - Ilka.pdf: 1372344 bytes, checksum: 39673ba3ef3d0f8d6e44ff59bc66eaff (MD5) Previous issue date: 2015-08-11 / CNPQ / A compreensão de textos e a consciência metatextual são atividades de natureza linguístico-cognitiva que têm despertado o interesse de estudiosos das áreas da Psicologia, Linguística e Educação.Sabe-se que as pesquisas sobre a compreensão de textos têm se mostrado mais abundantes, ao passo que a consciência metatextual mostra-se uma habilidade menos investigada pelo fato de ter sido sistematizada e definida recentemente. Consequentemente, os estudos que se propõem a investigação da relação entre a consciência metatextual e a compreensão de textos têm sido poucos e um tanto inconclusivos na afirmação quanto ao tipo de relação existente e ao modo como se dá esta relação. E é nesse sentido que a literatura que trata da consciência metatextual indica haver uma controvérsia que versa sobre a existência ou não de relação com a compreensão de textos. Assim, o presente estudo tem por objetivos investigaras relações entre a compreensão de textos e a consciência metatextual, a partir da consideração destas como habilidades multifacetadas examinadas através de diferentes tarefas; examinar o desempenho dos participantes nas tarefas de compreensão de textos e de consciência metatextual, considerando aspectos distintos das duas habilidades e analisar as propriedades psicométricas dos instrumentos utilizados na mensuração da compreensão de textos e da consciência metatextual.Foram investigadas 200 crianças de escolas públicas e particulares, com idades variando entre 7 e 8 anos (do 2º e 3º anos do Ensino Fundamental). Na presente investigação, foi feita a aplicação de quatro tarefas: três tarefas de compreensão de textos, com foco na identificação do tema/assunto principal, na reprodução oral e nas perguntas inferenciais e uma tarefa de consciência metatextual subdividida em itens de natureza epilinguística e metalinguística.Na primeira sessão foram realizadas as três tarefas de compreensão de textos (Tarefa 1 –Identificação do Tema/Assunto Principal, Tarefa 2 -Reprodução Oral de História e Tarefa 3 -Perguntas Inferenciais sobre uma História). A segunda sessão contemplou uma tarefa de consciência metatextual com os itens de natureza epilinguística (itens referentes à identificação do gênero textual, à identificação da incompletude do texto, à identificação de textos com estrutura híbrida e à identificação da ordem de textos) e de natureza metalinguística (itens referentes à identificação de partes presentes de textos incompletos, à identificação das partes ausentes de textos incompletos e à identificação do tipo de estrutura híbrida presente nos textos). Resultados revelaram que a compreensão de textos não se correlaciona com a consciência metatextual, donde depreende-se que o conhecimento requerido no ato da compreensão de textos está apoiado em uma representação mental global do texto, o que não ocorre com a consciência metatextual que requer um conhecimento linguístico mais específico referente à estrutura do texto, resultado que reforça a ideia da inexistência de correlação entre as duas habilidades. Quanto aos instrumentos, estes revelaram boa configuração fatorial e índices satisfatórios de fidedignidade, o que condiz com a robustez dos instrumentos na mensuração das habilidades investigadas. No que tange à comparação entre grupos, somente houve diferença significativa na comparação das tarefas com o tipo de escola, sendo o desempenho dos alunos de escola pública superior ao dos alunos de escola particular nas tarefas de compreensão de textos e de consciência metatextual, o que revela um padrão diferente do que se tem como norma e salienta como necessário o entendimento de como se configura a proposta de Ensino destas escolas, aspecto este relevante, mas que se distancia do que propõe a presente investigação. / Text comprehension and metatextual awareness are linguistic-cognitive nature activities that have aroused the interest of scholars from the fields of Psychology, Linguistics and Education. It is known that research on text comprehension has been more abundant, on the other hand, metatextual awareness appears to be a less investigated ability, mainly because only recently it has been systematized and defined. Consequently, studies that purport to research the relationship between metatextual awareness and text comprehension have been few and somewhat inconclusive in determining the type of relationship and how this relationship occurs. It is in this sense that the literature about metatextual awareness indicates that there is a controversy regarding the existence of a relationship with text comprehension. Thus, the present study aims to investigate the relationships between text comprehension and metatextual awareness, from the consideration of these as multifaceted abilities examined through different tasks; examine the performance of participants in text comprehension and metatextual awareness tasks, considering different aspects of both abilities and analyze the psychometric properties of the instruments used to measure text comprehension and metatextual awareness. Two hundred children, from public and private schools, with ages ranging between 7 and 8 years (2nd and 3rd years of elementary school), were investigated. In the present investigation four tasks were applied: three text comprehension tasks, focusing on the identification of the main theme/subject,oral reproduction and inferential questions, and a metatextual awareness task subdivided into epilinguistic and metalinguistic nature items. In the first session were performed the three text comprehension tasks (Task 1 –Identification of the Main Theme/Subject, Task 2 –Oral Reproduction of a Story and Task 3 –Inferential Questions About a Story). The second session covered a task of metatextual awareness with epilinguistic nature items (identification of the textual genre, identification of the incompleteness of the text, identification of texts with hybrid structure and identification of the order of texts) and metalinguistic (items related to the identification of parts present in incomplete texts, identification of the missing parts in incomplete texts and identification of the type of hybrid structure present in the texts). Results revealed that text comprehension does not correlate with metatextual awareness, surmising that the knowledge required in the act of text comprehension is based on a globalmental representation of text, which does not occur with metatextual awareness that requires a more specific linguistic knowledge concerning the structure of the text, a result that reinforces the idea that there is no correlation between the two abilities. As for the instruments, these showedfactorial configuration and satisfactory levels of reliability, which is consistent with the robustness of the instruments in the measurement of the abilities that have been investigated. With regard to the comparison between groups, there are significant differences when comparing the tasks with the type of school, the performance of public school students is higher when compared to the performance of the private school students in the tasks of text comprehension andmetatextual awareness, which shows a different pattern from what is perceived as standard and highlights the importance of understanding the education proposal from these schools. Although relevant, this aspect distances itself from the present investigation purposes.

Este trabalho está inserido com contexto da modelagem cartográfica e cartografia de síntese dentro do universo dos sistemas de informações geográficas (SIG). Serão apresentadas três metodologias distintas para obtenção de mapas sínteses através de inferências geográficas, são elas: Operações Pontuais de Soma (OPS), Processo Analitico hierarquico (AHP) e Redes Neurais Artificais (RNA). Aqui serão desenvolvidos e apresentados todos os procedimentos técnicos e metodológicos para a obtenção desses produtos cartográficos através do uso de dois softwares: o SPRING (SIG) e o SNNS (simulador de rede neural artificial). Também será apresentada uma discussão sobre a qualidade dos modelos gerados por essas distintas metodologias e a importância do papel do pesquisador na obtenção desses produtos. / This work is inserted in the context of cartographic modeling and mapping of synthesis within the universe of geographic information systems (GIS). We will present three different methodologies for obtaining maps synthesis by geographic inferences, they are: Operations Locations Sum, Analytic Hierarchy Process and Artificial Neural Network. Here will be developed and presented all the technical and methodological procedures to obtain these cartographic products through the use of two software: SPRING (GIS) and SNNS (artificial neural network simulator). It will also be a discussion of the quality of models generated by these different methodologies and the importance of the researcher in obtaining these products

Comprendre un texte, c’est élaborer un modèle de situation, une représentation cohérente qui est le résultat de l’interaction entre les informations textuelles et les connaissances du lecteur (van Dijk & Kintsch, 1983 ; Kintsch, 1988 ; Ehrlich et al., 1993). Cependant, les connaissances du lecteur ne sont toujours pas objectives et attestées (vraies). Ces connaissances peuvent également être des croyances voire des idéologies (van Dijk, 1998). Ce travail a pour objectif de montrer le rôle que peut jouer l’idéologie des lecteurs dans la compréhension de textes. Nous nous sommes intéressés à l’idéologie politique que nous avons mesurée à l’aide d’un test d’association implicite (IAT) que nous avons construit. Une première expérience a mis en évidence que la compréhension des textes est sensible à l’idéologie des lecteurs. En effet, les temps de lecture et les performances des lecteurs dépendent de l’appartenance politique des participants et de la cohérence entre cette appartenance et la nature idéologique des textes. Ensuite, à partir des simulations que permetle modèle Landscape (van den Brook et al., 1996, 1999), nous nous sommes intéressé à la dynamique de la compréhension à travers les inférences de type antécédent causal par rapport à l’idéologie des lecteurs. Les résultats obtenus à partir de trois expériences montrent que l’activation, le maintien et la récupération des inférences en mémoire est sensible à l’idéologie des lecteurs. En somme, ce travail complète les travaux sur la compréhension de texte enmettant en évidence l’importance des croyances des lecteurs, ici leur idéologie politique. / To understand a text, it is to work out a model of situation, a coherent representation which is the result of the interaction between textual informations and knowledge of the reader (van Dijk & Kintsch, 1983 ; Kintsch, 1988 ; Ehrlich and al., 1993).However, the reader’s knowledge are not still objective and attested (true). This knowledge can also be beliefs even ideologies (van Dijk, 1998). This work aims to demonstrate the potential role of ideology readers in reading comprehension. We were interested in the political ideology which we measured using an implicit test of association (IAT) that we built. A first experiment showed that text comprehension is sensitive to the reader’s ideology. Indeed, the reading times and the performances of the readers depend on the political affiliation on the participants and consistency between this membership and ideological nature of texts. Then, starting from simulations which the model Landscape allows (van den Brook and al., 1996, 1999), we were interested in dynamics of comprehension through the bridging inferences compared to the ideology of the readers. The results obtained starting from three experiments show that activation, the maintenance and the retrieval of inferences in memory are sensitive to the ideology of the readers. To sum up, this work supplements work on text comprehension by highlighting the importance of the beliefs of the readers, heretheir political ideology.

Cette thèse s’inscrit dans un double objectif. Il s’agit d’une part d’approfondir notre connaissance des processus de lecture et de compréhension au début de l’apprentissage de la lecture et d’autre part, de proposer des outils d’évaluation et des aides adaptées aux enfants en difficultés de lecture. Pour répondre à ces objectifs, plusieurs études ont été menées. Dans une première série d’études, les prédicteurs précoces de la réussite en lecture sont mis en évidence notamment grâce au suivi longitudinal d’une même population d’enfants depuis la grande section de maternelle jusqu’au CE1. Différents aspects de la lecture sont ici distingués, l’identification de mots écrits et la compréhension, ainsi que différents aspects de la compréhension de textes, la compréhension littérale et la production d’inférences. Un second axe de la thèse s’attache ensuite à développer des outils nouveaux permettant d’évaluer et de remédier aux difficultés de compréhension chez l’enfant. On observe en effet dans le contexte français actuel un manque certain de ce type d’outils. Ce travail de thèse se propose donc de combler ce manque en construisant 1/ une épreuve originale d’évaluation des capacités de compréhension écrite de récits chez l’enfant d’école primaire et 2/ un logiciel de remédiation des difficultés de compréhension pour les enfants apprentis lecteurs/ faibles compreneurs. La validité de ces instruments a été éprouvée auprès de vastes populations d’enfants. / The main objectives of this thesis are to deepen our knowledge about early reading and comprehension processes and to construct new instruments for assessing and remediating comprehension skills in children. First, in a three-years longitudinal study (from kindergarten to Grade 2), early predictors of reading achievement have been examined. Here, two aspects of reading have been distinguished, written words identification and comprehension, as far as two aspects of comprehension, literal comprehension and inference generation. Secondly, we aim to construct new instruments to assess comprehension skills and remediate comprehension difficulties in elementary-school children. Indeed obviously, there is currently a lack of such instruments in France. This second part of the thesis aims at filling this gap. A new test of reading comprehension has therefore been constructed and allows establishing a precise diagnosis of comprehension difficulties endured by some beginning readers. A computer-assisted learning program has then been proposed to less skilled comprehenders. This program specifically targets comprehension skills by stimulating 1/ literal information extraction, 2/ coherence inferences construction and 3/ knowledge-based inferences production. Efficiency and validity of each instrument have been tested with large population of children.

Submitted by Fabio Sobreira Campos da Costa (fabio.sobreira@ufpe.br) on 2016-04-18T12:25:42Z No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissert ALSM BC 27.11.2015.pdf: 1228269 bytes, checksum: 7f394fc3f8e5356e291eb7623627f94e (MD5) / Made available in DSpace on 2016-04-18T12:25:42Z (GMT). No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissert ALSM BC 27.11.2015.pdf: 1228269 bytes, checksum: 7f394fc3f8e5356e291eb7623627f94e (MD5) Previous issue date: 2015-02-10 / A compreensão de um texto envolve um conjunto de habilidades, a principal delas permeia todo o processo de construção do significado de um texto e sua efetiva compreensão e está na capacidade de elaborar inferências. Nesta pesquisa, buscou-se investigar como a compreensão acontece em sujeitos muito jovens e que ainda não sabem ler convencionalmente. Como base teórica desse estudo, adotou-se o Modelo de Construção-Integração proposto por Kintsch, em que as inferências têm lugar de destaque na compreensão de textos. Sendo assim, o presente estudo teve por principal objetivo investigar, por meio de diferentes recursos metodológicos, se as crianças da educação infantil estabeleciam inferências de diferentes tipos (causais, de estado e de previsão) enquanto ouviam um texto (metodologia on-line), no caso, uma história. Além disso, checou-se a compreensão global do texto através de tarefas realizadas após a leitura (metodologia off-line). Participaram desse estudo 60 crianças de 4 a 6 anos de idade, de classe média baixa, matriculadas em uma escola pública da Região Metropolitana do Recife. Foram divididos dois grupos de crianças (Grupo 1, composto por sujeitos da Educação Infantil 2, com idade entre 4 e 5 anos, e Grupo 2, crianças da Educação Infantil 3, com idade entre 5 e 6 anos). Cada criança foi entrevistada individualmente em duas sessões. A sessão 1 envolvia as tarefas 1 e 2. A tarefa 1 consistia na leitura interrompida de um texto narrativo pela pesquisadora dividido em partes, mediante as quais iam sendo feitas perguntas inferenciais (causais, de estado e de previsão). As respostas dadas nesta tarefa foram analisadas por dois juízes independentes, sendo classificadas nas seguintes categorias: Categoria I (não responde); Categoria II (incoerente e/ou improvável); e Categoria III (coerente e/ ou provável). Imediatamente após a audição da história, a Tarefa 2 foi aplicada, solicitando que a criança desse o tema da história ouvida. As respostas das crianças foram analisadas pelos mesmos juízes, de acordo com a seguinte classificação: T1 (não responde), T 2 (inapropriado), T 3 (apropriado/ vago), T4 (apropriado/ preciso). No dia seguinte, foi realizada a sessão 2, que dizia respeito à Tarefa 3, em que a criança era solicitada a recontar a historia ouvida no dia anterior. As reproduções elaboradas pelas crianças foram analisadas pelos dois juízes independentes de acordo com as seguintes categorias: R1 (reproduções desconectadas), R 2 (pouco fieis à historia original), R 3 (limitam-se a eventos de alguns blocos), R4 (reproduções globais), R5 (reproduções completas). De modo geral, comparações entre os diferentes tipos de inferências (causais, de estado e de previsão) mostraram que o tipo de pergunta realizada não influenciou o desempenho das crianças, uma vez que nos três tipos de inferências investigados e nos dois grupos de crianças houve maior frequência de respostas da categoria III. Verificou-se que a idade e o ano escolar não tiveram influência no desempenho das crianças. Considerando o desempenho nos dois grupos na elaboração dos temas para a história ouvida, a concentração de respostas esteve no tipo de tema T3 (apropriado/ vago), não se verificando diferenças entre os grupos para esta tarefa. Observou-se ainda uma grande dificuldade por parte das crianças de ambos os grupos na tarefa de reprodução oral da história. Contudo, as crianças do Grupo 2 demonstraram mais habilidade ao reproduzir do que as do Grupo 1. Uma comparação entre as tarefas demonstrou que as crianças da Educação Infantil tinham mais facilidade para propor um tema adequado e responder perguntas inferenciais do que reproduzir a história oralmente. O cruzamento dos dados de desempenho das crianças na Tarefa 1 com a Tarefa 2 revelou que responder perguntas inferenciais e elaborar tema da história parecem estar relacionadas, pois os participantes que tiveram melhor desempenho na Tarefa 1 também o tiveram em relação à Tarefa 2. Entre as tarefas 1 e 3, observou-se que houve alguma relação entre as duas habilidades, porém não se pode dizer que seja uma relação consistente devido a grande dificuldade na Tarefa 3 por parte dos participantes em geral. Entre as tarefas 2 e 3 não foi verificada relação entre as duas habilidades (indicação do tema e reprodução da história). Conclui-se que as crianças da Educação Infantil demonstraram certa compreensão de textos e que esta compreensão varia de uma tarefa para outra, não sendo uma habilidade que se manifeste igualmente em todas as situações de compreensão que são solicitadas a realizar. / Reading comprehension involves a set of abilities; the main one of these – the ability to elaborate inferences – permeates the whole process of constructing a text’s meaning, leading to effective comprehension. In this research we sought to investigate how comprehension happens in very young subjects that do not yet know how to read conventionally. As the theoretical basis of this study we adopted the Construction-Integration Model proposed by Kintsch, in which inferences play a crucial role in reading comprehension. Thus, the main objective of this present study was to investigate, using different methodological resources, whether pre-school children establish different types of inferences (causal, state, and predictive) while listening to a text (on-line methodology), specifically a story. Furthermore, we checked global comprehension of the text through tasks carried out after reading (off-line methodology). The subjects of the study were 60 lower middle-class children aged 4 to 6 enrolled in a public school in the greater Recife area. They were divided into two groups: Group 1, made up of subjects in pre-kindergarten, aged 4 to 5, and Group 2, of kids in kindergarten, aged 5 to 6. Each child was interviewed individually in two sessions. Session 1 involved tasks 1 and 2. Task 1 consisted of an interrupted reading, out loud by the researcher, of a narrative text divided into parts. At each break, inferential questions (causal, state, and predictive) were made. The answers given on these tasks were analyzed by two independent judges, and classified in the following categories: Category I (no answer); Category II (incoherent and/or improbable); and Category III (coherent and/or probable). Immediately after the reading of the story, Task 2 was applied, in which the children were asked to state the theme of the story they heard. The children’s answers were analyzed by the same judges, according to the following classification: T1 (no answer), T2 (inappropriate), T3 (appropriate but vague), T4 (appropriate and precise). On the following day, Session 2 took place, during which the children did Task 3, in which they were asked to retell the story they had heard the previous day. The reproductions the children elaborated were analyzed by the two independent judges, according to the following categories: R1 (unconnected reproductions), R2 (not very faithful to the original story), R3 (limited to the events of certain parts of the story), R4 (global reproductions), R5 (complete reproductions). Generally speaking, comparisons among the different types of inferences (causal, state, and predictive) showed that the type of question asked did not influence the children’s performance, since in all three types of inferences that were investigated and in both groups of children there was a preponderance of Category III answers. We observed that neither age nor grade in school influenced the children’s performance. Considering the performance of both groups on elaborating the themes of the story heard, most answers were of the T3 type (appropriate but vague). No difference between the two groups was found on this task. We observed, furthermore, great difficulty on the part of children in both groups with the task of orally reproducing the story. However, the children in Group 2 demonstrated a greater ability in this task than those in Group 1. A comparison between the tasks demonstrated that the pre-school children found it easier to propose an adequate theme and answer inferential questions than to reproduce the story orally. Cross-referencing the data on the children’s performance on Task 1 with that of Task 2 revealed that answering inferential questions and elaborating the theme of the story seem to be related, since the subjects who had the best performance on Task 1 also had the best on Task 2. Between Tasks 1 and 3, we observed that there was some relation between the two abilities, but we cannot say it is a consistent relation due to the great difficulty most subjects had with Task 3. Between Tasks 2 and 3, we found no relation between the two abilities (indicating the theme and reproducing the story). We conclude that the pre-school children demonstrated some reading comprehension, which varies from task to task, not being a skill which manifests equally in all comprehension situations they are asked to perform.

<div>Recovering combinatorial structures from noisy observations is a recurrent problem in many application domains, including, but not limited to, natural language processing, computer vision, genetics, health care, and automation. For instance, dependency parsing in natural language processing entails recovering parse trees from sentences which are inherently ambiguous. From a computational standpoint, such problems are typically intractable and call for designing efficient approximation or randomized algorithms with provable guarantees. From a statistical standpoint, algorithms that recover the desired structure using an optimal number of samples are of paramount importance.</div><div><br></div><div>We tackle several such problems in this thesis and obtain computationally and statistically efficient procedures. We demonstrate optimality of our methods by proving fundamental lower bounds on the number of samples needed by any method for recovering the desired structures. Specifically, the thesis makes the following contributions:</div><div><br></div><div>(i) We develop polynomial-time algorithms for learning linear structural equation models --- which are a widely used class of models for performing causal inference --- that recover the correct directed acyclic graph structure under identifiability conditions that are weaker than existing conditions. We also show that the sample complexity of our method is information-theoretically optimal.</div><div><br></div><div>(ii) We develop polynomial-time algorithms for learning the underlying graphical game from observations of the behavior of self-interested agents. The key combinatorial problem here is to recover the Nash equilibria set of the true game from behavioral data. We obtain fundamental lower bounds on the number of samples required for learning games and show that our method is statistically optimal.</div><div><br></div><div>(iii) Lastly, departing from the generative model framework, we consider the problem of structured prediction where the goal is to learn predictors from data that predict complex structured objects directly from a given input. We develop efficient learning algorithms that learn structured predictors by approximating the partition function and obtain generalization guarantees for our method. We demonstrate that randomization can not only improve efficiency but also generalization to unseen data.</div><div><br></div>

Thesis advisor: Henry I. Braun / Sponsored by OECD, PIAAC represents the first attempt to assess adult problem solving in technology-rich environments (PS-TRE) on an international scale that is comparable cross-culturally and cross-nationally. The objectives of this study are to study (1) the distributions of PS-TRE proficiency scores across 14 selected countries and (2) within each country, the associations between PS-TRE proficiency scores and the different formats of adult education and training (AET) participation. Using data on full-time professionals (at least 25 years old) from these countries, propensity score weighting was applied to estimate the associations between the different formats of AET participation and their PS-TRE proficiency scores. To place these estimates in context, parallel analyses were conducted – one with the sample of full-time associates in the 14 selected countries and the other with full-time professionals’ Literacy and Numeracy proficiency scores as measured by PIAAC. The results showed that after controlling for socio-demographic background, occupational categories, use of key information-processing skills (both at home and at work), as well as use of generic workplace skills, no consistent pattern was found across the 14 selected countries. At the individual country level, scattered significant relationships were identified. For example, in Denmark, both formats of AET participation (vs. None) are significantly and positively associated with full-time professionals’ PS-TRE proficiency scores and their probability of scoring in the top quartile of the PS-TRE distribution (p < .01). While in the United States, Formal AET (vs. None) is significantly and positively associated with full-time associates’ PS-TRE proficiency scores and their probability of scoring in the top quartile of the PS-TRE distribution (p < .01). The variations in relationships between the different formats of AET participation and working adults’ skills proficiency across domains and samples indicate the necessity of conducting qualitative research on AET programs in individual countries. Furthermore, to provide recommendations tailored to the specific needs of each country, a fine-grained classification of AET programs based on the OECD guideline was suggested. / Thesis (PhD) — Boston College, 2020. / Submitted to: Boston College. Lynch School of Education. / Discipline: Educational Research, Measurement and Evaluation.

In this research two functional magnetic resonance tomography experiments were conducted to identify core regions of language comprehension processes. The focus of the studies was on inferencing, i.e. the activation of information which has not been explicitly mentioned in a given utterance but which is somehow implied because of general world knowledge. The research strategy was two-fold. First, text materials were used which allowed to isolate inference processes from more basic language processes. Second, two tasks verification in Experiment 1 and recognition in Experiment 2 were assigned to the participants to selectively enhance or attenuate processing at different levels of representation. In both experiments a network of brain areas was found to be active during language comprehension including areas all along the left superior temporal sulcus, the left lateral and medial prefrontal areas, as well as the right anterior temporal lobe and the posterior cingulate cortex. The results of Experiment 1 indicated that the dorsomedial prefrontal cortex was most prominently associated with inferencing in the context of the verification task. As expected, activity in this region was attenuated in Experiment 2 during recognition. No indications were found that the right hemisphere plays a particular role for inferencing as has been suggested by some authors. The results of both experiments are discussed with respect to the neuroimaging literature on language comprehension and with respect to recent approaches to memory systems in the brain particularly the episodic memory system. Finally, a functional neuroanatomical model of inferencing is sketched.

No description available.