Esta tese propõe um recurso léxico de verbos com anotação de papéis semânticos, denominado VerbLexPor, baseado em recursos como VerbNet, PropBank e FrameNet. As bases teóricas da proposta são interdisciplinares e retiradas da Linguística de Corpus e do Processamento de Linguagem Natural (PLN), visando-se a contribuir para a Linguística e para a Computação. As hipóteses de pesquisa são: a) um mesmo conjunto de papéis semânticos pode ser aplicado a diferentes gêneros textuais; e b) as diferenças entre esses gêneros se destacam no ranqueamento dos papéis semânticos. O desenvolvimento do VerbLexPor se apoia em dois corpora: um especializado, com mais de 1,6 milhão de palavras, composto por artigos científicos de Cardiologia de três periódicos brasileiros; e um não especializado, com mais de 1 milhão de palavras composto por artigos do jornal popular Diário Gaúcho. Os corpora foram anotados com o parser PALAVRAS, e as informações de sentenças, verbos e argumentos foram extraídas e armazenadas em um banco de dados. O VerbLexPor tem 192 verbos e mais de 15 mil argumentos anotados distribuídos em mais de 6 mil sentenças. Observou-se que o corpus do Diário Gaúcho privilegia uma sintaxe direta e pouco uso de voz passiva e adjuntos, enquanto o corpus de Cardiologia apresenta mais voz passiva e um maior uso de INSTRUMENTOS na posição de sujeito, além de uma menor incidência de AGENTES. Foram realizados também alguns experimentos paralelos, como a anotação de papéis semânticos por vários anotadores e o agrupamento automático de verbos. Na tarefa de múltiplos anotadores, cada um anotou exatamente as mesmas 25 orações. Os anotadores receberam um manual de anotação e um treinamento básico (explicação sobre a tarefa e dois exemplos de anotação). Usou-se o cálculo de multi-π para avaliar a concordância entre os anotadores, e o resultado foi de π = 0,25. Os motivos para essa concordância baixa podem estar na falta de um treinamento mais completo. A tarefa de agrupamento de verbos mostrou que a sintaxe e a semântica são igualmente importantes para o agrupamento. Este estudo contribui para a área de Linguística, com um léxico de verbos anotados semanticamente, e também para a Computação, com dados que podem ser consultados e processados para diversas aplicações do PLN, principalmente por estarem disponíveis nos formatos XML e SQL. / This dissertation aims at developing a lexical resource of verbs annotated with semantic roles, called VerbLexPor, and based on other resources, such as VerbNet, PropBank, and FrameNet. The theoretical bases of this study lies in Corpus Linguistics and Natural Language Processing (NLP), so that it aims at contributing to both Linguistics and Computer Science. The hypotheses are: a) one set of semantic roles can be applied to different genres; and b) the differences among genres are shown by the ranking of semantic roles. The development of VerbLexPor has two corpora at the basis: a specialized one, with more than 1.6 million words, composed by scientific papers in the field of Cardiology from three Brazilian journals; and a non-specialized one, with more than 1 million words, composed by newspaper articles from Diário Gaúcho. The corpora were analyzed with the parser PALAVRAS, and sentence, verb and argument information was extracted and stored in a database. VerbLexPor has 192 verbs and more than 15 thousand arguments annotated with semantic roles, distributed among more than 6 thousand sentences. We observed that Diário Gaúcho has a more direct syntax, with less passive voice and adjuncts, while Cardiology has more passive voice and more INSTRUMENTS for subjects, and fewer AGENTS. We also conducted some parallel experiments, such as semantic role labeling with multiple annotators and automatic verbal clustering. In the multiple annotators task, each of them annotated exactly the same 25 sentences. They received an annotation manual and basic training (explanation on the task and two annotation examples). We used multi-π to evaluate agreement among annotators, and results were π = 0,25. Reasons for this low agreement may be a lack of a thoroughly developed training. The verbal clustering task showed that syntax and semantics are equally important for verbal clustering. This study contributes to Linguistics, with a verbal lexicon annotated with semantic roles, and also to Computer Science, with data that can be assessed and processed for various NLP applications, especially because the data are available in both XML and SQL formats.

Esta tese propõe um recurso léxico de verbos com anotação de papéis semânticos, denominado VerbLexPor, baseado em recursos como VerbNet, PropBank e FrameNet. As bases teóricas da proposta são interdisciplinares e retiradas da Linguística de Corpus e do Processamento de Linguagem Natural (PLN), visando-se a contribuir para a Linguística e para a Computação. As hipóteses de pesquisa são: a) um mesmo conjunto de papéis semânticos pode ser aplicado a diferentes gêneros textuais; e b) as diferenças entre esses gêneros se destacam no ranqueamento dos papéis semânticos. O desenvolvimento do VerbLexPor se apoia em dois corpora: um especializado, com mais de 1,6 milhão de palavras, composto por artigos científicos de Cardiologia de três periódicos brasileiros; e um não especializado, com mais de 1 milhão de palavras composto por artigos do jornal popular Diário Gaúcho. Os corpora foram anotados com o parser PALAVRAS, e as informações de sentenças, verbos e argumentos foram extraídas e armazenadas em um banco de dados. O VerbLexPor tem 192 verbos e mais de 15 mil argumentos anotados distribuídos em mais de 6 mil sentenças. Observou-se que o corpus do Diário Gaúcho privilegia uma sintaxe direta e pouco uso de voz passiva e adjuntos, enquanto o corpus de Cardiologia apresenta mais voz passiva e um maior uso de INSTRUMENTOS na posição de sujeito, além de uma menor incidência de AGENTES. Foram realizados também alguns experimentos paralelos, como a anotação de papéis semânticos por vários anotadores e o agrupamento automático de verbos. Na tarefa de múltiplos anotadores, cada um anotou exatamente as mesmas 25 orações. Os anotadores receberam um manual de anotação e um treinamento básico (explicação sobre a tarefa e dois exemplos de anotação). Usou-se o cálculo de multi-π para avaliar a concordância entre os anotadores, e o resultado foi de π = 0,25. Os motivos para essa concordância baixa podem estar na falta de um treinamento mais completo. A tarefa de agrupamento de verbos mostrou que a sintaxe e a semântica são igualmente importantes para o agrupamento. Este estudo contribui para a área de Linguística, com um léxico de verbos anotados semanticamente, e também para a Computação, com dados que podem ser consultados e processados para diversas aplicações do PLN, principalmente por estarem disponíveis nos formatos XML e SQL. / This dissertation aims at developing a lexical resource of verbs annotated with semantic roles, called VerbLexPor, and based on other resources, such as VerbNet, PropBank, and FrameNet. The theoretical bases of this study lies in Corpus Linguistics and Natural Language Processing (NLP), so that it aims at contributing to both Linguistics and Computer Science. The hypotheses are: a) one set of semantic roles can be applied to different genres; and b) the differences among genres are shown by the ranking of semantic roles. The development of VerbLexPor has two corpora at the basis: a specialized one, with more than 1.6 million words, composed by scientific papers in the field of Cardiology from three Brazilian journals; and a non-specialized one, with more than 1 million words, composed by newspaper articles from Diário Gaúcho. The corpora were analyzed with the parser PALAVRAS, and sentence, verb and argument information was extracted and stored in a database. VerbLexPor has 192 verbs and more than 15 thousand arguments annotated with semantic roles, distributed among more than 6 thousand sentences. We observed that Diário Gaúcho has a more direct syntax, with less passive voice and adjuncts, while Cardiology has more passive voice and more INSTRUMENTS for subjects, and fewer AGENTS. We also conducted some parallel experiments, such as semantic role labeling with multiple annotators and automatic verbal clustering. In the multiple annotators task, each of them annotated exactly the same 25 sentences. They received an annotation manual and basic training (explanation on the task and two annotation examples). We used multi-π to evaluate agreement among annotators, and results were π = 0,25. Reasons for this low agreement may be a lack of a thoroughly developed training. The verbal clustering task showed that syntax and semantics are equally important for verbal clustering. This study contributes to Linguistics, with a verbal lexicon annotated with semantic roles, and also to Computer Science, with data that can be assessed and processed for various NLP applications, especially because the data are available in both XML and SQL formats.

Esta tese propõe um recurso léxico de verbos com anotação de papéis semânticos, denominado VerbLexPor, baseado em recursos como VerbNet, PropBank e FrameNet. As bases teóricas da proposta são interdisciplinares e retiradas da Linguística de Corpus e do Processamento de Linguagem Natural (PLN), visando-se a contribuir para a Linguística e para a Computação. As hipóteses de pesquisa são: a) um mesmo conjunto de papéis semânticos pode ser aplicado a diferentes gêneros textuais; e b) as diferenças entre esses gêneros se destacam no ranqueamento dos papéis semânticos. O desenvolvimento do VerbLexPor se apoia em dois corpora: um especializado, com mais de 1,6 milhão de palavras, composto por artigos científicos de Cardiologia de três periódicos brasileiros; e um não especializado, com mais de 1 milhão de palavras composto por artigos do jornal popular Diário Gaúcho. Os corpora foram anotados com o parser PALAVRAS, e as informações de sentenças, verbos e argumentos foram extraídas e armazenadas em um banco de dados. O VerbLexPor tem 192 verbos e mais de 15 mil argumentos anotados distribuídos em mais de 6 mil sentenças. Observou-se que o corpus do Diário Gaúcho privilegia uma sintaxe direta e pouco uso de voz passiva e adjuntos, enquanto o corpus de Cardiologia apresenta mais voz passiva e um maior uso de INSTRUMENTOS na posição de sujeito, além de uma menor incidência de AGENTES. Foram realizados também alguns experimentos paralelos, como a anotação de papéis semânticos por vários anotadores e o agrupamento automático de verbos. Na tarefa de múltiplos anotadores, cada um anotou exatamente as mesmas 25 orações. Os anotadores receberam um manual de anotação e um treinamento básico (explicação sobre a tarefa e dois exemplos de anotação). Usou-se o cálculo de multi-π para avaliar a concordância entre os anotadores, e o resultado foi de π = 0,25. Os motivos para essa concordância baixa podem estar na falta de um treinamento mais completo. A tarefa de agrupamento de verbos mostrou que a sintaxe e a semântica são igualmente importantes para o agrupamento. Este estudo contribui para a área de Linguística, com um léxico de verbos anotados semanticamente, e também para a Computação, com dados que podem ser consultados e processados para diversas aplicações do PLN, principalmente por estarem disponíveis nos formatos XML e SQL. / This dissertation aims at developing a lexical resource of verbs annotated with semantic roles, called VerbLexPor, and based on other resources, such as VerbNet, PropBank, and FrameNet. The theoretical bases of this study lies in Corpus Linguistics and Natural Language Processing (NLP), so that it aims at contributing to both Linguistics and Computer Science. The hypotheses are: a) one set of semantic roles can be applied to different genres; and b) the differences among genres are shown by the ranking of semantic roles. The development of VerbLexPor has two corpora at the basis: a specialized one, with more than 1.6 million words, composed by scientific papers in the field of Cardiology from three Brazilian journals; and a non-specialized one, with more than 1 million words, composed by newspaper articles from Diário Gaúcho. The corpora were analyzed with the parser PALAVRAS, and sentence, verb and argument information was extracted and stored in a database. VerbLexPor has 192 verbs and more than 15 thousand arguments annotated with semantic roles, distributed among more than 6 thousand sentences. We observed that Diário Gaúcho has a more direct syntax, with less passive voice and adjuncts, while Cardiology has more passive voice and more INSTRUMENTS for subjects, and fewer AGENTS. We also conducted some parallel experiments, such as semantic role labeling with multiple annotators and automatic verbal clustering. In the multiple annotators task, each of them annotated exactly the same 25 sentences. They received an annotation manual and basic training (explanation on the task and two annotation examples). We used multi-π to evaluate agreement among annotators, and results were π = 0,25. Reasons for this low agreement may be a lack of a thoroughly developed training. The verbal clustering task showed that syntax and semantics are equally important for verbal clustering. This study contributes to Linguistics, with a verbal lexicon annotated with semantic roles, and also to Computer Science, with data that can be assessed and processed for various NLP applications, especially because the data are available in both XML and SQL formats.

In the field of Natural Language Processing, word-formation is under-resourced comparing to inflectional morphology. Moreover, the existing resources capturing word-formation differ in many aspects. This thesis aims to review existing language resources for word-formation across languages and to unify them to a common data structure and file format. Basic notions of word-formation are followed by a review of existing language resources and their comparison in both quantitative and qualitative aspects. In the core part of the thesis, the harmonisation process is presented. Design decisions on the unification procedure are presented, and the selection of the resources to unify is described. The resources are unified to the rooted tree data structure and stored in a lexeme-based file format, which is already used in DeriNet 2.0. The procedure applies supervised machine learning model and the Maximum Spanning Tree algorithm. While the model scores word-formation relations, the MST algorithm uses the scores for identifying the rooted tree structure in each word-formation family. The resulting collection of harmonised resources covering 20 European languages was published under the title 'Universal Derivations' (UDer).

Le dictionnaire LVF (Les Verbes Français) de J. Dubois et F. Dubois-Charlier représente une des ressources lexicales les plus importantes dans la langue française qui est caractérisée par une description sémantique et syntaxique très pertinente. Le LVF a été mis disponible sous un format XML pour rendre l’accès aux informations plus commode pour les applications informatiques telles que les applications de traitement automatique de la langue française. Avec l’émergence du web sémantique et la diffusion rapide de ses technologies et standards tels que XML, RDF/RDFS et OWL, il serait intéressant de représenter LVF en un langage plus formalisé afin de mieux l’exploiter par les applications du traitement automatique de la langue ou du web sémantique. Nous en présentons dans ce mémoire une version ontologique OWL en détaillant le processus de transformation de la version XML à OWL et nous en démontrons son utilisation dans le domaine du traitement automatique de la langue avec une application d’annotation sémantique développée dans GATE. / The LVF dictionary (Les Verbes Français) by J. Dubois and F. Dubois-Charlier is one of the most important lexical resources in the French language, which is characterized by a highly relevant semantic and syntactic description. The LVF has been available in an XML format to make access to information more convenient for computer applications such as NLP applications for French language. With the emergence of the Semantic Web and the rapid diffusion of its technologies and standards such as XML, RDF/RDFS and OWL, it would be interesting to represent LVF in a more formalized format for a better and more sophisticated usage by natural language processing and semantic web applications. We present in this paper an OWL ontology version of LVF by demonstrating the mapping process between the data model elements of the XML version and OWL. We give account about its use in the field of natural language processing by presenting an application of semantic annotation developed in GATE.

Le dictionnaire LVF (Les Verbes Français) de J. Dubois et F. Dubois-Charlier représente une des ressources lexicales les plus importantes dans la langue française qui est caractérisée par une description sémantique et syntaxique très pertinente. Le LVF a été mis disponible sous un format XML pour rendre l’accès aux informations plus commode pour les applications informatiques telles que les applications de traitement automatique de la langue française. Avec l’émergence du web sémantique et la diffusion rapide de ses technologies et standards tels que XML, RDF/RDFS et OWL, il serait intéressant de représenter LVF en un langage plus formalisé afin de mieux l’exploiter par les applications du traitement automatique de la langue ou du web sémantique. Nous en présentons dans ce mémoire une version ontologique OWL en détaillant le processus de transformation de la version XML à OWL et nous en démontrons son utilisation dans le domaine du traitement automatique de la langue avec une application d’annotation sémantique développée dans GATE. / The LVF dictionary (Les Verbes Français) by J. Dubois and F. Dubois-Charlier is one of the most important lexical resources in the French language, which is characterized by a highly relevant semantic and syntactic description. The LVF has been available in an XML format to make access to information more convenient for computer applications such as NLP applications for French language. With the emergence of the Semantic Web and the rapid diffusion of its technologies and standards such as XML, RDF/RDFS and OWL, it would be interesting to represent LVF in a more formalized format for a better and more sophisticated usage by natural language processing and semantic web applications. We present in this paper an OWL ontology version of LVF by demonstrating the mapping process between the data model elements of the XML version and OWL. We give account about its use in the field of natural language processing by presenting an application of semantic annotation developed in GATE.

L’enseignement-apprentissage du lexique spécialisé chinois est un chemin semé d’obstacles. Pour les apprenants non natifs, les combinaisons lexicales spécialisées (CLS) (L’Homme, 2000) soulèvent des difficultés syntaxico sémantiques et représentent ainsi un défi majeur dans l’acquisition de compétences lexicales. On recense, toutefois, peu de propositions méthodologiques pour résoudre ces difficultés dans la littérature consacrée à la didactique du chinois sur objectifs spécifiques (COS) (Q. Li, 2011). Dans cette recherche, nous nous attachons à explorer de quelle manière une méthode de description lexicale basée sur une représentation sémantique et syntaxique assiste les apprenants non natifs dans la résolution des problèmes lexicaux soulevés par les CLS. Notre thèse vise à concevoir une méthode de description des CLS en vue de la résolution de difficultés lexicales par les locuteurs non sinophones. La méthode mise au point est appliquée à l’élaboration du dictionnaire CHINOINFO, une ressource lexicale chinois-français portant sur le domaine de l’informatique. Cette ressource s’adresse aux apprenants francophones du chinois. L’objectif secondaire de notre thèse consiste à évaluer l’efficacité du CHINOINFO auprès des apprenants francophones qui reçoivent une formation de chinois dans un établissement universitaire au Québec ou en Chine. Notre recherche fait appel à des notions empruntées à trois cadres théoriques. Premièrement, la Lexicologie explicative et combinatoire (Mel’čuk et al., 1995) nous sert d’appui théorique pour fonder la description des CLS sur la représentation sémantique du lexique spécialisé. Deuxièmement, notre démarche de collecte et d’analyse des CLS est guidée par l’approche lexico sémantique à la terminologie (L’Homme, 2020a). Enfin, nous nous appuyons sur l’approche cognitive en didactique des langues secondes (Chastain, 1990) pour envisager une présentation structurée des connaissances lexicales. Notre démarche méthodologique s’est déroulée en trois phases. Nous avons d’abord assemblé un corpus spécialisé chinois pour en extraire un échantillon de CLS et les renseignements permettant de les décrire. L’analyse des données collectées à partir du corpus nous a amenée à anticiper trois types de difficultés syntaxico-sémantiques soulevées par les CLS : 1) distinguer les acceptions d’un polysème dans différentes CLS; 2) différencier les sens distincts de CLS de forme identique; 3) choisir les cooccurrents appropriés d’un terme. À la deuxième phase, nous avons mobilisé différentes stratégies pour décrire les propriétés syntaxico-sémantiques des CLS. Une méthode descriptive intégrant les solutions proposées a ensuite été appliquée à la création du CHINOINFO. Cette ressource en ligne répertorie 91 termes fondamentaux du domaine de l’informatique, pour lesquels nous avons encodé au total 282 termes reliés et 644 CLS. La structuration des données au sein des articles s’est largement inspirée de l’adaptation du DiCoInfo (Observatoire de linguistique Sens-Texte, 2022) à un dictionnaire d’apprentissage (Alipour, 2014). Différents moyens techniques ont été mis en œuvre pour assurer la convivialité de la ressource. La dernière phase de notre recherche consiste en une expérimentation comparative visant à évaluer l’efficacité pédagogique du CHINOINFO. Nous avons fait passer un test lexical à deux groupes d’apprenants francophones, soit le groupe contrôle (GC) et le groupe expérimental (GE), en leur proposant un nombre d’outils de référence. Seul le GE a eu accès à CHINOINFO. Nous avons aussi collecté, au moyen de questionnaires de sondage, le profil des participants ainsi que leur appréciation sur le test et les outils de référence proposés. De manière générale, l’analyse comparative des résultats du test lexical montre que le GE a mieux réussi à résoudre les trois types de difficultés soulevées par les CLS. Les participants étaient plutôt satisfaits de l’organisation du test. Le GE a eu moins de difficultés à réaliser le test puisqu’il se sentait mieux outillé pour trouver des éléments de réponses aux questions du test par rapport GC. Le GE s’est exprimé favorablement quant à l’utilité du CHINOINFO pour résoudre les problèmes lexicaux dans le cadre de notre expérimentation. Pour conclure, les résultats de notre analyse fournissent des indices sur l’apport du CHINOINFO en tant qu’une ressource d’apprentissage des CLS, ce qui laisse entrevoir l’intérêt de la méthode de description lexicale que nous avons proposée dans un contexte pédagogique. / The teaching and learning of Chinese specialized lexicon is a path strewn with obstacles. For non-native learners, specialized lexical combinations (SLCs) (L’Homme, 2000) raise syntactic and semantic difficulties and thus represent a major challenge in the acquisition of lexical skills. However, there are few methodological proposals to solve these difficulties in the literature devoted to the teaching practice and applied research of Chinese for specific purposes (Q. Li, 2011). In this research, we explore how a lexical description method based on semantic and syntactic representation assists non-native learners in solving lexical problems raised by SLCs. This thesis aims at designing a method for describing SLCs to help non-Chinese speakers solve lexical difficulties. The proposed method is applied to develop CHINOINFO, a Chinese-French dictionary of computer science and information technology terms. This lexical resource is designed for French-speaking learners of Chinese and can also be used as a writing tool for language professionals (translators, technical writers, and proofreaders), as well as professionals in this field. The secondary objective of this thesis is to evaluate the pedagogical effectiveness of the developed resource among French-speaking university students. This research draws on concepts derived from three theoretical frameworks. Firstly, the Explanatory and Combinatorial Lexicology (Mel’čuk et al., 1995) provides theoretical support for founding the lexical description on the representation of semantic features of the specialized lexicon. Secondly, the collection and analysis of SLCs are guided by the lexical-semantic approach to terminology (L’Homme, 2020a). Finally, we draw on the cognitive approach to second language didactics (Chastain, 1990) to explore the effective ways to organize and present the descriptive information of the specialized lexicon. Our methodological approach was carried out in three stages. We started by assembling a specialized Chinese corpus to extract a sample of SLCs and their descriptive information. The analysis of the data collected from the corpus led us to anticipate three types of syntaxico-semantic difficulties raised by SLCs: 1) distinguishing polysemes in different SLCs; 2) identifying, in a given context, the meaning of a lexical combination that is syntactically ambiguous; and 3) selecting appropriate co-occurrents for a term. In the second stage, we deployed different strategies to describe the syntaxico-semantic features of SLCs. Subsequently, a descriptive method that incorporates the proposed solutions has been applied to the creation of CHINOINFO. This online lexical resource contains 91 basic terms related to computer science and information technology. For these terms, we encoded a total of 282 related terms and 644 SLCs. The organization of content in the entries has been largely inspired by the conversion of DiCoInfo (Observatoire de linguistique Sens-Texte, 2022) into a learner’s dictionary (Alipour, 2014). We used various techniques to make the resource user-friendly. The final stage of our research consists of a comparative experiment to evaluate the pedagogical effectiveness of CHINOINFO. We had two groups of French-speaking learners, the control group (CG) and the experimental group (EG), take a lexical test by providing them with several reference materials. Only the EG had access to CHINOINFO during the test. We also collected the information about the participants' learning profile and their appreciation of the test and the proposed reference materials. Overall, the comparative analysis of the test results shows that the GE succeeded better in solving the three types of difficulties raised by the SLCs. The participants were quite satisfied with the organization of the lexical test. The EG encountered less difficulty in answering questions during the test since they felt better equipped to find elements of answers in the reference materials than the GC. The EG commented favorably on the utility of CHINOINFO in solving lexical problems. To conclude, the results of our experiment provide clues about the pedagogical interest of CHINOINFO as a SLC learning resource, which suggests the relevance of the lexical description method we proposed in a pedagogical context.