The past two decades have seen a substantial increase in the availability of numerical data that individuals are faced with on a daily basis. In addition, research uncovering the multiple facets of statistical reasoning has become increasingly prominent. Both gender differences and the effect of experience or training have emerged as two key factors that influence performance in statistics. Surprisingly, though, the combined effects of these two variables have not been studied. This gap in understanding the joint effect of gender and experience on statistical reasoning is addressed in the present dissertation with six studies. In Study 1 (N = 201), participants with various levels of experience in statistics were asked to complete the Statistical Reasoning Assessment (SRA; Garfield, 2003). Although the performance of both genders improved with experience, the gender gap persisted across all experience levels. Multiple measures of individual differences were used in a confirmatory structural equation model. This model supported the idea that differences in statistical reasoning are not uniquely a matter of cognitive ability. In fact, gender was found to influence statistical reasoning directly, as well as indirectly through its influence on thinking dispositions. In Studies 2 (N = 67), 3 (N = 157), and 4 (N = 206), the role of stereotype threat was examined as a potential cause of the persisting gender gap in statistics, and value affirmation was tested as an intervention to overcome stereotype threat. Despite the fact that many women believed negative stereotypes about the ability of women in statistics, value affirmation had no significant impact on performance. To help explain this lack of effect, and in keeping with the results of the structural equation model suggesting a multi-pronged approach, efforts were turned towards a different (and potentially richer) cognitive factor. Specifically, mental representations were explored to help shed light on the root causes of those conceptual understanding differences in statistics. In Studies 5 and 6, gender differences in mental representations of statistical features were examined using a categorization paradigm. In Study 5 (N = 219), extending some of the key findings in Studies 1, 3 and 4, it was established that two courses in statistics are necessary to create a significant difference in the quality of mental representations of statistical concepts. More importantly, Study 6 (N = 208) demonstrated how constraining the task format particularly benefits women in that the quality of their reasoning significantly improved, where that of men was equal across tasks. Theoretical and practical implications of these findings are discussed.

The past two decades have seen a substantial increase in the availability of numerical data that individuals are faced with on a daily basis. In addition, research uncovering the multiple facets of statistical reasoning has become increasingly prominent. Both gender differences and the effect of experience or training have emerged as two key factors that influence performance in statistics. Surprisingly, though, the combined effects of these two variables have not been studied. This gap in understanding the joint effect of gender and experience on statistical reasoning is addressed in the present dissertation with six studies. In Study 1 (N = 201), participants with various levels of experience in statistics were asked to complete the Statistical Reasoning Assessment (SRA; Garfield, 2003). Although the performance of both genders improved with experience, the gender gap persisted across all experience levels. Multiple measures of individual differences were used in a confirmatory structural equation model. This model supported the idea that differences in statistical reasoning are not uniquely a matter of cognitive ability. In fact, gender was found to influence statistical reasoning directly, as well as indirectly through its influence on thinking dispositions. In Studies 2 (N = 67), 3 (N = 157), and 4 (N = 206), the role of stereotype threat was examined as a potential cause of the persisting gender gap in statistics, and value affirmation was tested as an intervention to overcome stereotype threat. Despite the fact that many women believed negative stereotypes about the ability of women in statistics, value affirmation had no significant impact on performance. To help explain this lack of effect, and in keeping with the results of the structural equation model suggesting a multi-pronged approach, efforts were turned towards a different (and potentially richer) cognitive factor. Specifically, mental representations were explored to help shed light on the root causes of those conceptual understanding differences in statistics. In Studies 5 and 6, gender differences in mental representations of statistical features were examined using a categorization paradigm. In Study 5 (N = 219), extending some of the key findings in Studies 1, 3 and 4, it was established that two courses in statistics are necessary to create a significant difference in the quality of mental representations of statistical concepts. More importantly, Study 6 (N = 208) demonstrated how constraining the task format particularly benefits women in that the quality of their reasoning significantly improved, where that of men was equal across tasks. Theoretical and practical implications of these findings are discussed.

Ce travail a comme source une réflexion sur les problèmes liés à la complexité de lasémantique adjectivale. Puisque l’adjectif constitue le noyau de cette thèse, nous ledéfinissons et faisons ainsi ressortir ses caractéristiques, notamment la souplesse dont il faitpreuve. Nous envisageons la modification du nom par l’adjectif, puis nous considérons laplace de ce dernier vis-à-vis du substantif recteur ainsi que les changements de sens entreantéposition et postposition.Chemin faisant, notre réflexion porte non seulement sur la notion de sens mais aussisur la construction du sens que nous considérons comme un processus dynamique. Nousmettons en avant l’activation de l’information ainsi que les problèmes rencontrés lors de laconstruction du sens d’expressions Adj-N et N-Adj. La polysémie lexicale constitue un despoints sur lesquels nous nous attardons puisque ce phénomène est au centre de la langue.Nos recherches s’intègrent dans le cadre de la Grammaire Cognitive deR.W. Langacker (1987, 1991, 2008). Nous étudions donc l’adjectif de même que lacombinaison Adj-N/N-Adj à la lumière de cette théorie. Notre raisonnement nous amène ànous questionner, entre autres choses, sur la représentation sémantico-conceptuelle associéeaux adjectifs à sens multiples.Nous postulons que toute expression linguistique est associée, dans l’appareil cognitifdes locuteurs-auditeurs, à une structure conceptuelle d’informations organisées. Notreréflexion s’oriente vers la représentation des items lexicaux à sens multiples en GrammaireCognitive. Le modèle des réseaux schématiques (Schematic Network Model) permet de rendrecompte des différents sens d’un lexème ainsi que des liens les unissant. Sur la base destravaux de D. Tuggy (1993), nous proposons de répartir les représentationssémantico-conceptuelles des mots à sens multiples le long d’un continuum allant del’homonymie au vague, en passant par la polysémie et la multi-facialité. Enfin, nousproposons une analyse des adjectifs à sens multiples pauvre, cher et curieux, toujours dans lecadre de la Grammaire Cognitive. Pour chacun d’entre eux, nous établissons unereprésentation sémantico-conceptuelle et nous donnons les régularités sémantiques liées à laconstruction du sens lors leur combinatoire avec un substantif. / This work is concerned with the problems which surround the complexity of adjectivalsemantics. As the adjective is at the heart of this thesis, we firstly seek to define it and, in doingso, we bring to light its characteristics, and in particular its flexibility. We consider theadjective when used in noun modification before turning to the place of the adjective withregards to the noun and, finally, considering the changes in meaning between adjectives in theanteposition and postposition.Inevitably, our reflection concerns not only the notion of meaning but also meaningconstruction,which we consider as a dynamic process. We focus upon the activation of theinformation as well as the problems linked to meaning-construction of Adj-N and N-Adjexpressions. We pay particular attention to lexical polysemy; this phenomenon being centralto language.The framework for our research is that of R.W. Langacker’s Cognitive Grammar(1987, 1991, 2008). We, thus, study the adjective as well as the Adj-N/N-Adj combinations inlight of this theory. Our reasoning brings us to question, among other things, the semanticconceptualrepresentation associated with adjectives which offer multiple meanings.We postulate that any linguistic expression is associated, in a speaker-hearer’scognitive device, with an abstract structure of organized information. Our reflection turns tothe representation of lexical items which offer multiple meanings in Cognitive Grammar. TheSchematic Network Model is used to allow us to take into account different meanings of alexeme as well as the links between them. On the basis of D. Tuggy’s works (1993), wesuggest distributing the semantic-conceptual representations of words with multiple meaningsalong a continuum going from homonymy to vagueness, and including polysemy andmultifaciality. Finally, we offer an analysis of the adjectives pauvre, cher and curieux whichhave multiple meanings, once again within the framework of Cognitive Grammar. For each ofthese adjectives, we establish a semantic-conceptual representation and we offer the semanticregularities connected to meaning-construction when these adjectives are combined with anoun.

The present dissertation reports the results of a multi-disciplinary experimental study, which combines psycholinguistic and cognitive methodologies in order to achieve two broad objectives. The first objective is providing a comparative psycholinguistic analysis of the expression of motion events in three languages, namely Turkish, English and French, taking Talmy&lsquo / s verb-framed language vs. satellite-framed language typology (Talmy, 1985) as the framework. The second one is investigating the relationship between linguistic representation and conceptual representation by taking motion events as the testing ground. In order to pursue these two lines of inquiry, five complementary tasks are conducted on three groups of adult subjects. The results of the first two tasks, the language production task and the language comprehension task, verify the Talmyan typology experimentally by showing sharp differences between the data obtained from native speakers of typologically different languages (English vs. Turkish and French), as well as remarkable similarities between the data obtained from native speakers of typologically similar languages (Turkish and French). On the other hand, the remaining three non-verbal tasks, the categorization task and the two eye-tracking tasks, present valuable insights into the nature of conceptual event representation by revealing a uniform pattern across languages. This latter result is inconsistent with the renowned linguistic relativity hypothesis (Whorf, 1956) / however in line with the universalist view (Jackendoff, 1990, 1996), which suggests that conceptual event representation is language-free and independent of the linguistic encoding preferences of different languages.

La representació conceptual en terminologia: l'atribució temàtica en els bancs de dades terminològiques, tesi doctoral defensada el desembre de 2005 a l'Institut Universitari de Lingüística Aplicada de la Universitat Pompeu Fabra, revisa la noció d'àrea temàtica i d'adscripció temàtica dels termes a través de la seva representació en els BDT; demostra, amb un treball empíric, les insuficiències d'aquesta representació, tant des del punt de vista de l'adequació als principis teòrics de la terminologia, com des del punt de vista de l'eficiència dels bancs per respondre a les consultes que reben; proporciona una explicació del funcionament de la variació conceptual de les unitats considerades homònimes en els BDT, i elabora una primera proposta de representació alternativa que dóna resposta a algunes de les insuficiències detectades. / La representació conceptual en terminologia: l'atribució temàtica en els bancs de dades terminològiques, tesis doctoral defendida en diciembre de 2005 en el Institut Universitari de Lingüística Aplicada de la Universitat Pompeu Fabra, revisa la noción de área temática y de adscripción temática de los términos a través de su representación en los BDT; demuestra, con un trabajo empírico, las insuficiencias de esta representación, tanto desde el punto de vista de la adecuación a los principios teóricos de la terminología, como desde el punto de vista de la eficiencia de los bancos para responder a las consultas que reciben; proporciona una explicación del funcionamiento de la variación conceptual de las unidades consideradas homónimas en los BDT, y elabora una primera propuesta de representación alternativa que da respuesta a algunas de las insuficiencias detectadas. / Conceptual representation in terminology - the attribution of subject field to specialized terms in terminological databanks is a Ph.D. read in December 2005 at the Institut Universitari de Lingüística Aplicada, Universitat Pompeu Fabra. This dissertation revisits the notion of subject field and the principles and criteria of thematic attribution of the concepts stored in TDB. The empirical work carried out in this dissertation has shown up the weaknesses of such issue, both from the standpoint of complying with the theoretical principles of terminology as well as that of the efficacy of databanks to answer the queries received. Furthermore, an explanation of conceptual variation of units considered homonymous in BDT is provided. Finally, a draft offering an alternative representation is put forward to rectify this situation.

Au cours de la dernière décennie, plusieurs études ont montré que le traitement des mots est influencé par leurs caractéristiques orthographiques, lexicales et sémantiques. Les valeurs élevées des variables sémantiques, telles que la familiarité conceptuelle, l’imageabilité, la concrétude et l’expérience sensorielle, sont généralement associées à un traitement plus rapide et plus précis des mots lors d’une tâche de décision lexicale sur présentation écrite. Cependant, les normes relatives à l’une des variables sémantiques cruciales, comme la familiarité conceptuelle, et à une autre nouvelle variable, celle de la force perceptuelle d’une modalité spécifique, ne sont pas disponibles pour une large base de données francophones. Cette thèse présente des normes de familiarité conceptuelle, de force perceptuelle visuelle et auditive pour 3 596 mots en français. Ces normes ont été collectées auprès de 313 participants franco-canadiens, âgés entre 18 et 35 ans. Des statistiques descriptives et des analyses corrélationnelles ont été conduites sur ces normes et sur d’autres variables subjectives déjà publiées (âge d’acquisition, imageabilité, concrétude) et objectives (nombre de lettres, fréquence objective). Les analyses de corrélation ont révélé la nature sémantique de nos variables. Plus spécifiquement, les estimations de la force perceptuelle visuelle et auditive ont montré une relation forte avec les variables sémantiques, d’où l’importance du rôle de l’expérience perceptuelle dans la représentation conceptuelle des mots. Nous avons ensuite démontré l'utilité de ces nouvelles normes, de familiarité conceptuelle et de force perceptuelle, en évaluant leur contribution aux temps de réponse dans une tâche de décision lexicale. En utilisant des analyses de régression par étapes, nous avons démontré l’apport indépendant de la familiarité conceptuelle et de la force perceptuelle au-delà d’autres variables psycholinguistiques bien connues, telles que la longueur des mots, la fréquence, l’imageabilité, l’âge d’acquisition et la concrétude. Nos résultats procurent des données probantes en français sur la connaissance des caractéristiques psycholinguistiques et leurs impacts dans le traitement des mots. Les banques de données développées dans ces études constituent un apport significatif sur les études futures qui éclaireront davantage l'interaction entre les systèmes linguistique, sémantique et perceptuel. Ces études aideront les chercheurs à sélectionner des stimuli en français avec des caractéristiques psycholinguistiques spécifiques pour des expériences dans lesquelles la familiarité conceptuelle et la force perceptuelle doivent être prises en compte. Ces normes auront également un large domaine d’utilisation, comme la recherche sur les sciences humaines de la santé et la création de matériel didactique et des tests d’évaluation du langage. / Over the past decade, research has shown that word processing is influenced by the orthographic, lexical, and semantic features of words. High values of semantic variables, such as conceptual familiarity, imageability, concreteness and sensory experience, are usually associated with faster and more accurate word processing as, for instance, measured with a written lexical decision task. However, norms for crucial semantic variables, such as conceptual familiarity, and another new variable of modality specific perceptual strength, are not available for a large set of French words. This thesis presents norms for conceptual familiarity, visual and auditory perceptual strength for 3,596 French words. We collected these norms from approximately 300 healthy participants, all French-Canadian native speakers, between 18 and 35 years of age. Descriptive statistics and correlational analyses were conducted on these norms and on other known subjective variables (age of acquisition, imageability, concreteness) and objective variables (number of letters, objective frequency). Correlation analyzes revealed the semantic nature of our variables. More specifically, values of visual and auditory perceptual strength have shown a strong relationship with semantic variables. This highlights the importance of the role of perceptual experience in the conceptual representation of words. We then demonstrated the usefulness of these new norms of conceptual familiarity and perceptual strength by evaluating their impact on latencies in a written lexical decision task. Using stepwise regression models, conceptual familiarity and perceptual strength demonstrated independent contribution to latencies, beyond the contribution of other well-known psycholinguistic variables, such as word length, frequency, imageability, age of acquisition and concreteness. The databases developed in these studies are a meaningful contribution to future studies that will shed further light on the interaction between the linguistic, semantic and perceptual systems. These studies will help researchers to select French stimuli with specific psycholinguistic characteristics for experiments in which conceptual familiarity and perceptual strength must be taken into account. These norms will also have a broad area of use, such as research in human health sciences and the creation of teaching materials and language assessment tests.