The aim of this work is to describe and to analyse one type of anaphoric nominal syntagma found in the articles of French sports newspapers. The work is divided in two parts theoretical and practical. The theoretical part describes the basic concepts of text linguistics. Then it focuses on the concept of anaphora and nominal syntagma with brief reference to journalistic-style writing. In the practical part, an analysis of anaphoric nominal syntagmas and the description of their semantico-pragmatic relations is carried out.

No description available.

No description available.

published_or_final_version / abstract / Modern Languages and Cultures / Master / Master of Philosophy

This thesis provides an experimental approach to abstract anaphora resolution in Turkish written discourse. The core of this work consists of identifying various manifestations of abstract anaphoric expressions &ndash / bu vs. bu durum, bu olay, bu is, bu ger&ccedil / ek (bu as the bare abstract object anaphor vs. bu+label abstract anaphors)- in Turkish discourse, and investigating whether any difference is observed in their processing. To this end, two offline experiments are conducted with human subjects, and the results indicate that label anaphors, compared to the bare anaphor bu, have a tendency to disambiguate the antecendent in some cases.

This thesis provides an experimental approach to abstract anaphora resolution in Turkish written discourse. The core of this work consists of identifying various manifestations of abstract anaphoric expressions &ndash / bu vs. bu durum, bu olay, bu is, bu ger&ccedil / ek (bu as the bare abstract object anaphor vs. bu+label abstract anaphors) - in Turkish discourse, and investigating whether any difference is observed in their processing. To this end, two offline experiments are conducted with human subjects, and the results indicate that label anaphors, compared to the bare anaphor bu, have a tendency to disambiguate the antecendent in some cases.

Pronoun resolution normally takes place without conscious effort or awareness, yet the processes behind it are far from straightforward. A large number of cues and constraints have previously been recognised as playing a role in the identification and integration of potential antecedents, yet there is considerable debate over how these operate within the resolution process. The aim of this thesis is to investigate how the parser handles multiple antecedents in order to understand more about how certain information sources play a role during pronoun resolution. I consider how both structural information and information provided by the prior discourse is used during online processing. This is investigated through several eye tracking during reading experiments that are complemented by a number of offline questionnaire experiments. I begin by considering how condition B of the Binding Theory (Chomsky 1981; 1986) has been captured in pronoun processing models; some researchers have claimed that processing is faithful to syntactic constraints from the beginning of the search (e.g. Nicol and Swinney 1989), while others have claimed that potential antecedents which are ruled out on structural grounds nonetheless affect processing, because the parser must also pay attention to a potential antecedent’s features (e.g. Badecker and Straub 2002). My experimental findings demonstrate that the parser is sensitive to the subtle changes in syntactic configuration which either allow or disallow pronoun reference to a local antecedent, and indicate that the parser is normally faithful to condition B at all stages of processing. Secondly, I test the Primitives of Binding hypothesis proposed by Koornneef (2008) based on work by Reuland (2001), which is a modular approach to pronoun resolution in which variable binding (a semantic relationship between pronoun and antecedent) takes place before coreference. I demonstrate that a variable-binding (VB) antecedent is not systematically considered earlier than a coreference (CR) antecedent online. I then go on to explore whether these findings could be attributed to the linear order of the antecedents, and uncover a robust recency preference both online and offline. I consider what role the factor of recency plays in pronoun resolution and how it can be reconciled with the first-mention advantage (Gernsbacher and Hargreaves 1988; Arnold 2001; Arnold et al., 2007). Finally, I investigate how aspects of the prior discourse affect pronoun resolution. Prior discourse status clearly had an effect on pronoun resolution, but an antecedent’s appearance in the previous context was not always facilitative; I propose that this is due to the number of topic switches that a reader must make, leading to a lack of discourse coherence which has a detrimental effect on pronoun resolution. The sensitivity of the parser to structural cues does not entail that cue types can be easily separated into distinct sequential stages, and I therefore propose that the parser is structurally sensitive but not modular. Aspects of pronoun resolution can be captured within a parallel constraints model of pronoun resolution, however, such a model should be sensitive to the activation of potential antecedents based on discourse factors, and structural cues should be strongly weighted. / Pronomenauflösung erfolgt normalerweise scheinbar mühelos und ohne bewusste Anstrengung. Jedoch ist die Verarbeitung von pronominalen Referenzen aus linguistischer Sicht ein hochkomplexer Prozess. Durch unterschiedliche wissenschaftliche Studien wurden bereits zahlreiche Faktoren ermittelt, die bei der Pronomenauflösung eine Rolle spielen, allerdings herrscht weitgehend noch keine Einigkeit darüber, wie genau diese Faktoren die Verarbeitung von Pronomen beeinflussen. Ziel dieser Dissertation ist es zu untersuchen, wie der Leser/Hörer mit Pronomen umgeht, denen mehrere Antezedenten zugeordnet werden können, um zu verstehen, welche Rolle bestimmte Informationsquellen in der Verarbeitung von Pronomen spielen. Besondere Beachtung findet dabei, wie strukturelle Eigenschaften sowie Informationen aus dem vorangegangenen Diskurs für die Suche nach einem passenden Antezedenten benutzt werden. Die angewandte Untersuchungsmethode der vorliegenden Dissertation ist Eye-tracking during reading, ergänzt mit verschiedenen offline-Fragebögen. Die Experimente erforschen die Rolle der folgenden Aspekte in der Verarbeitung von Pronomen: Prinzip B der Bindungstheorie (Chomsky 1981; 1986), Koreferenz und Variablenbindung laut der Primitives of Binding Hypothese (Reuland 2001, Koornneef 2008), Antezedentenreihenfolge im Satz, und Diskursstatus des Antezedents. Obwohl es zeigt sich, dass der Hörer/Leser sensibel für subtile Veränderungen in der syntaktischen Konfiguration ist, wie z.B. für die Reihenfolge der Antezedenten im Satz und für den Diskursstatus des Antezedenten, gibt es keinen Nachweis dafür, dass Variablenbindung zeitlich vor Koreferenz erfolgt. Einige Aspekte der Auflösung pronominaler Referenzen können in einem parallel constraints model erfasst werden, allerdings sollte so ein Modell strukturelle Informationen stark gewichten und sensitiv sein für die Aktivierung potenzieller Antezedenten aufgrund von Diskursfaktoren.

This thesis presents a knowledge-poor pronoun resolution system for Turkish which resolves third person personal pronouns and possessive pronouns. The system is knowledge-poor in the sense that it makes use of limited linguistic and semantic knowledge to resolve the pronouns. As pronoun resolution proposals for languages like English, French and Spanish, the core of the system is the constraints and preferences which are determined empirically. The system has four modules: sentence splitting, pronoun extraction, forming the list of candidate antecedents and determination of the antecedent. It takes a Turkish text as input and rewrites this text with the considered pronouns replaced with their proposed antecedents. In order to compare the success rate of the system, two different baseline algorithms are implemented. The original system is tested against these baseline algorithms on two sample Turkish texts from different sources. Some suggestions to improve the success rate of the system and to extend the domain of the system are also presented.

Orientador: Edson Françozo / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Estudos da Linguagem / Made available in DSpace on 2018-08-15T22:20:33Z (GMT). No. of bitstreams: 1 Godoy_MahayanaCristina_M.pdf: 590425 bytes, checksum: 7516dd531bfc57e40497f21b437a02ad (MD5) Previous issue date: 2010 / Resumo: A resolução da anáfora conceitual traz questões interessantes para a pesquisa psicolingüística, pois abre a possibilidade de investigar como um pronome plural pode ser resolvido a partir de uma expressão antecedente singular, como em "O batalhão foi vacinado. Eles foram para a África". De acordo com Landman (1989), grupos como "batalhão" podem ter uma denotação plural quando recebem uma predicação distributiva (e.g., foi vacinado), mas também podem ter uma denotação singular quando são predicados coletivamente (e.g., foi reorganizado). Em outras palavras, enquanto o predicado seleciona a pluralidade de membros que compõem o batalhão no primeiro caso, no segundo, o mesmo batalhão é predicado como instituição, i.e., como uma entidade singular. Com base nessas considerações, é razoável assumir que a leitura coletiva ou distributiva determinada pelo predicado pode influenciar a resolução pronominal, desde que se considere a saturação anafórica como produto de expectativas geradas pelos leitores acerca de quais referentes têm chances de serem mencionados em partes subseqüentes do texto (Kehler et al., 2007). Para testar esta hipótese, elaboramos dois experimentos. No primeiro deles, investigamos se as leituras coletiva ou distributiva de expressões como "pelotão" poderiam criar expectativas sobre como os sujeitos dariam continuações para sentenças na voz passiva. Os participantes deveriam completar sentenças como "Para viajar ao exterior, o batalhão foi vacinado" apontando seu agente. Os resultados indicam que a continuação para predicados coletivos era preferencialmente um agente singular, enquanto para os distributivos o agente dado tendia a ser plural. Em um segundo experimento, os sujeitos liam sentenças com anáforas conceituais cujo antecedente era predicado coletiva ou distributivamente. O tempo de leitura do pronome nessas duas situações foi comparado, apontando para um tempo significativamente maior quando o pronome seguia uma predicação coletiva. Ambos os resultados, portanto, confirmam nossas hipóteses, indicando que os predicados de termos de grupo podem gerar expectativas que influenciam a resolução da anáfora conceitual / Abstract: Conceptual anaphora resolution brings interesting questions to research in psycholinguistics, for it provides the possibility of investigating how a plural pronoun can be solved from a singular antecedent, as one can see in "The army division was vaccinated. They went to Africa". According to Landman (1989), groups such as "the army division" can have a plural denotation when it receives a distributive predicate (e.g., was vaccinated), but it can also have a singular denotation when its predicate is a collective one (e.g., was reorganized). In other words, while the predicate selects the plurality of members that compose the army division in the first case, in the second one the same army division is predicated as an institution, i.e., as a singular entity. From these considerations, it is reasonable to assume that collective or distributive readings determined by the predicate can influence pronoun resolution, if one consider anaphora resolution as a product of the expectations generated by the readers about which referents are more likely to be mentioned in subsequent text (Kehler et al., 2007). In order to test this hypothesis, we ran two experiments. In the first one, we investigated whether collective or distributive readings of expressions similar to "the army division" would create expectations on how subjects provide continuations for passive sentences. The participants should complete sentences such as "In order to go abroad, the army division was vaccinated" with their agents. The result indicates that the continuation given to collective predicates was preferably a singular agent, while the continuations provided to distributive predicates tended to be plural agents. In a second experiment, subjects read sentences with conceptual anaphoras whose antecedent was predicated in a collective or distributive way. The pronoun reading time in these two situations was compared, showing a significant longer time when the pronoun followed collective predication. Therefore, both results corroborate our hypothesis, indicating that the way groups are predicated can generate expectations that influence conceptual anaphora resolution / Mestrado / Linguistica / Mestre em Linguística

Toutes les informations présentes actuellement sur le web représentent une source d‘informations colossale, qui s‘enrichit de jour en jour. L‘analyse automatique de ces informations, qui sont plus souvent non-structurées, constitue un véritable enjeu économique et scientifique. La résolution des anaphores nominales s‘inscrit dans la structuration des informations grâce à l‘identification du lien entre des groupes nominaux, elle permet de simplifier des tâches à différentes applications : la traduction automatique, le résumé ou l‘extraction automatique d‘information, le data mining etc. Dans ce contexte, le travail que nous avons mené dans cette thèse évoque différentes méthodes de résolution des anaphores nominales de deux types : infidèles et associatives. En nous fondant sur divers aspects autours de la notion d‘anaphore nominale et des notions de voisinage comme la résolution d‘anaphores pronominales, la résolution de coréférences ; en combinant des méthodes existantes avec des outils et des ressources disponibles pour la langue française, notre travail s‘attache à trois modules : module de prétraitement du corpus, module de résolution des anaphores nominales et le module d‘évaluation. Au module de prétraitement, les ressources lexicales sont constituées et mobilisées grâces aux analyses au niveau linguistique des anaphores nominales. La plateforme Unitex est le principal outil utilisé à cette étape. Pour les anaphores du type infidèle, nous avons utilisé deux méthodes différentes : la première mobilise des ressources lexicales simples avec les entrées de groupes nominaux uniquement ; la deuxième mobilise des ressources plus élaborées (les entrées de groupes nominaux et verbaux). Pour les anaphores associatives du type méronymique, nous nous fondons sur la théorie des classes d‘objets afin de décrire le type de relation anaphorique établie Résumé 17 entre l‘expression anaphorique et son antécédent. Les ressources utilisées pour ce type d‘anaphore sont ainsi divisées hiérarchiquement selon les classes et les domaines. Le module de résolution est l‘étape de décision, nous nous basons sur le calcul du poids de saillance de chacun des antécédents potentiels pour sélectionner le meilleur candidat. Chaque candidat peut avoir différents facteurs de saillance, qui correspond à sa probabilité d'être sélectionné. Le poids de saillance final est calculé par le moyen pondéré des poids de saillance élémentaires. Les facteurs de saillances sont proposés après les analyses syntaxiques et sémantiques du corpus. L‘évaluation de notre travail constitue un vrai enjeu à cause de la complexité de la tâche, mais elle nous permet d‘avoir une vue globale sur nos méthodes de travail. La comparaison des résultats obtenus permet de visualiser l‘apport de chaque paramètre utilisé. L‘évaluation de notre travail nous permet également de voir les erreurs au niveau du prétraitement (l‘extraction des syntagmes nominaux, des syntagmes verbaux…), cela nous a permis d‘intégrer un module de correction dans notre système. / In order to facilitate the interpretation of texts, this thesis is devoted to the development of a system to identify and resolve the indirect nominal anaphora and the associative anaphora. Resolution of the indirect nominal anaphora is based on calculating salience weights of candidate antecedents with the purpose of associating these antecedents with the anaphoric expressions identified. It is processed by twoAnnexe317different methods based on a linguistic approach: the first method uses lexical and morphological parameters; the second method uses morphological and syntactical parameters. The resolution of associative anaphora is based on syntactical and semantic parameters.The results obtained are encouraging: 90.6% for the indirect anaphora resolution with the first method, 75.7% for the indirect anaphora resolution with the second method and 68.7% for the associative anaphora resolution. These results show the contribution of each parameter used and the utility of this system in the automatic interpretation of the texts.