[pt] O presente estudo busca investigar o uso de verbos modais na produção escrita de universitários brasileiros em inglês como língua estrangeira (ILE), contrastando-o com o emprego realizado por alunos universitários americanos e britânicos que têm o inglês como primeira língua. Para tanto, lança-se mão do arcabouço teórico da lingüística de corpus (Sinclair, 1991; McEnery & Wilson, 1996; Hunston, 2002; Berber Sardinha, 2004; McEnery, Xiao & Tono, 2006) e de recursos tecnológicos, especialmente o programa computacional WordSmith Tools (Scott, 1999), para analisar dois corpora. O corpus de estudo consiste em uma fração do Br-ICLE (The Brazilian Portuguese Sub-corpus of the International Corpus of Learner English), coletada em quatro universidades do Estado do Rio de Janeiro, totalizando 51.430 palavras, enquanto o de referência corresponde a uma parte do LOCNESS (Louvain Corpus of Native English Essays), somando 165.135 palavras. Adotando uma abordagem textual à escrita (Hyland, 2002) e uma abordagem estatística para quantificar os resultados, a investigação tem como foco os verbos modais centrais (Biber et al., 1999), a saber, can, could, may, might, must, shall, should, will e would. Os resultados indicam que os falantes brasileiros de inglês empregam menos verbos modais do que seus pares americanos e britânicos. Mais especificamente, nota-se que will e would são usados com freqüência significativamente menor na escrita em ILE de brasileiros. Diferenças entre os modais também são encontradas quando se analisa o padrão de uso destes verbos. Os resultados da pesquisa evidenciam, de certa forma, a diferença na produção escrita dos dois grupos estudados, o que pode contribuir para o ensino de língua inglesa no contexto brasileiro. / [en] The present study aims at investigating and contrasting the use of modal verbs in the writing of Brazilian undergraduates majoring in English as a foreign language (EFL), contrasting it to that of British and American university students who speak English as a first language. To this end, it makes use of both the theoretical background of Corpus Linguistics (Sinclair, 1991; McEnery & Wilson, 1996; Hunston, 2002; Berber Sardinha, 2004; McEnery, Xiao & Tono, 2006) and of technological resources, especially the computer program WordSmith Tools (Scott, 1999), to analyze two corpora. The research corpus consists of a section of the Br-ICLE (The Brazilian Portuguese Sub-corpus of the International Corpus of Learner English) which has been collected in four universities in the city of Rio de Janeiro, totaling 51,430 words, while the reference one corresponds to a part of the LOCNESS (Louvain Corpus of Native English Essays), accounting for 165,135 words. Adopting a textual approach to writing (Hyland, 2002) and using statistical tests to quantify the results, the investigation focuses on central modal verbs (Biber et al., 1999), namely, can, could, may, might, must, shall, should, will and would. Results indicate that the Brazilian speakers of English make less use of modal verbs than their American and British counterparts. More specifically, it is observed that both will and would present significantly lower frequencies in Brazilian EFL writing. Differences among modals are also found when the patterns of use of these verbs are taken into account. Research results show that there are certain differences in the written production of the two groups, which may contribute to the teaching of English in the Brazilian context.

No description available.

Cette thèse se donne pour objectif d’étudier les apories théoriques et épistémologiques liées à la synonymie. Pour l’essentiel, ces problèmes se concentrent dans la faiblesse théorique de la notion par rapport au succès empirique dont elle jouit dans l’usage et les dictionnaires. Cette faiblesse théorique prend racine dans l’histoire de la notion : la synonymie a été définie en premier lieu par Aristote et a subi de nombreuses transformations au cours des siècles. La définition scientifique de la synonymie s’est constituée à partir de la synonymie distinctive des synonymistes. Les notions fondatrices de la linguistique moderne ont été appliquées à la synonymie alors qu’elle a été élaborée avant leur conception. La synonymie se plie mal aux cadres théoriques de la linguistique moderne d’où des apories. L’approche distinctive centre l’étude de la synonymie sur les différences de sens alors qu’elle repose sur des équivalences sémantiques approchées. L’histoire de la notion aide à éclairer les apories et à s’en distancier pour recentrer la notion sur les équivalences puisqu’elles rendent possibles la synonymie et sur le discours qui est le lieu de l’émergence du sens. Cela rompt avec la synonymie réduite à la recherche de différences de sens entre items lexicaux synonymes. Une approche onomasiologique et textuelle est mise en place pour proposer un nouveau cadre d’étude de la synonymie. Cette approche se manifeste par la détection de réseaux lexicaux témoignant de relations d’équivalence qui émergent en contexte. Pour tester cette approche, les réseaux lexicaux du vocabulaire du vice et de la vertu sont étudiés dans des textes du XVIIe siècle traitant de thèmes moraux. / This thesis studies theoretical and epistemological aporia of synonymy. These problems are concentrated in the theoretical weakness of synonymy in comparison with its empirical success both in usage and dictionaries. This theoretical weakness originates in the history of the notion: at first, synonymy had been defined by Aristoteles and was afterwards subjected to transformations during the following centuries. With the beginning of modern linguistics, scientific definition of synonymy had been built from the distinctive synonymy of French synonymists. The fundamental notions of modern linguistics, such as the opposition between language and discourse, had been applied to synonymy whereas it had been elaborated before their conception. Synonymy can hardly be submitted to modern linguistics theoretical frameworks without generating theoretical difficulties. The distinctive approach centers the study of synonymy on semantic differences whereas synonymy is based on approximate semantic equivalence. History of synonymy sheds light on aporia and helps to distance from it and to center the notion on the semantic equivalence because they make synonymy possible and on discourse because that is where sense emerges. This breaks off with the synonymy which is solely based on semantic differences between synonymic lexical items. An onomasiological and textual approach is set up to propose a new scientific framework to synonymy: this approach consists in the detection of lexical networks showing semantic relations appearing in context. To test this approach, lexical networks of vice and virtue vocabulary are studied in seventeenth century French texts treating moral themes.