Rozprawy doktorskie na temat „LEXICON EXPLORATION”

Ce travail de thèse représente une tentative de rationalisation du lexique espagnol en accordant la priorité au signifiant. Inspiré notamment de Pierre Guiraud, nous y avons établi à la suite de Maurice Molho, Didier Bottineau, Georges Bohas ou Dennis Philps, que les mots peuvent ne référer que par la sollicitation d’une partie de leur forme, partie détectable structurellement. Nous avons pris en considération l’ensemble des capacités qualitatives (phones, formant, idéophones, graphèmes, segments) et quantitatives (duplications, répétitions, inversions, homophonies / homographies) de la forme des mots. Nous ne nous sommes donc pas limité à un support sémiologique particulier mais avons conçu le lexique, du fait de sa complexité, comme un organisme de signifiants / signifiés où chaque élément peut entrer en compte pour donner lieu à une motivation (interne ou externe). Nous avons nommé cet élément la saillance, car il s’agit d’une unité résultant d’une focalisation, d’un choix formel pour référer à telle ou telle idée. Nous avons également remarqué, notamment dans des cas d’« homonymie », que plusieurs parties pouvaient être sollicitées et que chacune permettait de renvoyer à un sens distinct. La consubstantialité du signe est donc, de notre point de vue, un principe sauf. La « synonymie » ainsi que la « polysémie » ne sont donc pas non plus des notions pertinentes car chaque terme renvoie d’une manière qui lui est propre à un sens donné. Enfin, nous avons proposé une application, avec des critères similaires, à des énoncés dit « poétiques » où plusieurs actualisations parfois insolites apparaissent mais toujours permises par le langage, par le signifiant
This thesis represents an attempt of rationalization of the Spanish lexicon giving priority to the signifier. Inspired in particular by Pierre Guiraud, we have established in the wake of Maurice Molho, Didier Bottineau, Georges Bohas or Dennis Philps, that the words can refer by the request of a portion of their shape, detectable structurally. We considered all the qualitative capacities (phones, submorphems, graphemes, segments) and quantitative (duplications, repetitions, reversals, homophonies / homographies) forms of words. We did not thus limited to a particular semiological support but conceived the lexicon, because of its complexity, as an organism of signifiers / signifieds where every element can be important to give rise to a motivation (internal or external). We named this element salience, because it is about a unity which results from a focus, from a formal choice to refer any particular idea. We also noted, especially in cases of "disambiguation", which many parties could be sought and each allowed to refer to a distinct meaning. The consubstantiality of the sign is thus, of our point of view, a pertinent principle. The "synonymy" as well as the "polysemy" are not therefore either relevant notions because every term sends back in a way which is appropriate for its to a given meaning. Finally, we proposed an application, with similar criteria, in "poetics" utterances where several unusual actualizations appear but always permitted by the language, the signifier

Cross-lingual word embeddings are an increasingly important reseource in cross-lingual methods for NLP, particularly for their role in transfer learning and unsupervised machine translation, purportedly opening up the opportunity for NLP applications for low-resource languages.  However, most research in this area implicitly expects the availablility of vast monolingual corpora for training embeddings, a scenario which is not realistic for many of the world's languages.  Moreover, much of the reporting of the performance of cross-lingual word embeddings is based on a fairly narrow set of mostly European language pairs.  Our study examines the performance of cross-lingual alignment across a more diverse set of language pairs; controls for the effect of the corpus size on which the monolingual embedding spaces are trained; and studies the impact of spectral graph properties of the embedding spsace on alignment.  Through our experiments on a more diverse set of language pairs, we find that performance in bilingual lexicon induction is generally poor in heterogeneous pairs, and that even using a gold or heuristically derived dictionary has little impact on the performance on these pairs of languages.  We also find that the performance for these languages only increases slowly with corpus size.  Finally, we find a moderate correlation between the isospectral difference of the source and target embeddings and the performance of bilingual lexicon induction.  We infer that methods other than cross-lingual alignment may be more appropriate in the case of both low resource languages and heterogeneous language pairs.

Cette thèse s’inscrit dans le cadre des travaux de révision de la structure de la racine sémitique en général, et arabe en particulier. Dans l’introduction, nous avons entamé une relecture critique des efforts des grammairiens arabes en quête du seuil minimal du sens, et qui ont abouti à poser la racine trilitère comme étant ce seuil ultime et inanalysable. Ce choix a été fait malgré tous les signes d’instabilité que comporte ce concept - tels que son incapacité à expliquer la réversibilité de l’ordre des consonnes et leur variation phonétique indépendamment du sens. C’est un choix synchronique anhistorique qui exclut la notion de temps, et émane d’une conception révélationniste (tawqîf) du langage. Au XIXe siècle, l'évolutionnisme darwiniste, étendu à la philologie, met à mal cette conception figée de la racine. En intégrant la notion de temps, nombre d’Orientalistes, suivis par quelques philologues arabes de l’époque, ont montré à travers une approche comparative que la racine trilitère est une forme évoluée d’une base primitive bilitère (ou monosyllabique). Depuis, plusieurs explications ont été proposées pour la formation de la racine trilitère à partir d’une base bilitère (croisement, incrémentation, affixation). Certains linguistes comme Hurwitz ont essayé d’identifier et de systématiser par déduction les éléments ternaires et leurs valeurs sémantiques. Dans notre travail, qui s'inscrit dans le cadre de la théorie des matrices et des étymons, nous démontrons à travers l’analyse synchronique de près de 1000 items que la racine trilitère est analysable en termes d’étymons bilitères et de créments, ou affixes. Les éléments affixables ou incrémentables à la base sont phonétiquement les mêmes (gutturales, sonantes, labiales, nasales) et corrélés souvent aux mêmes valeurs grammaticales (factitif, statif, moyen) ou sémantiques (l’intensif). La troisième position de la racine est la position privilégiée dans ce processus d’affixation/incrémentation. La racine trilitère n’est donc pas le seuil minimal du sens, et s’avère réductible à une base biconsonantique rendue trilitère grâce à un segment crément ou affixe. Ceci peut avoir un effet sur notre conception du lexique arabe, désormais réorganisable autour des bases bilitères soit abstraites (les traits phonétiques), soit concrètes (les étymons primitifs), qui sont à leur tour transformables en radicaux trilitères grâce à une liste préalable de créments et affixes spécifiant la signification primordiale véhiculée par la base bilitère
This thesis is part of the revisionist work on the structure of the Semitic root in general, and the Arabic root in particular. In the introduction, we present a critical review of the efforts of Arab grammarians in their quest of a minimum linguistic threshold associated with meaning, a quest that resulted in establishing the triliteral root as the ultimate unanalysable unit. This choice was made despite the many obvious shortcomings of this theoretical framework, such as its inability to explain the reversal of the order of consonants, or their phonetic variation (regardless of its meaning). It is an anhistorical, synchronic choice that excludes the notion of time, and finds its roots in a revelationnist (tawqîf) linguistic framework. In the end of the nineteenth century, the extended Darwinist theory has undermined this static conception of the root. By integrating the notion of time, a number of Orientalists - followed by some Arab philologists of that period - showed, through a comparative methodology, that the triliteral root has evolved from a primitive monosyllabic (or biconsonantal) root. Since then, several explanations have been proposed for the formation of the triliteral root from a biliteral base (crossed bilateral roots, affixation of formative increments or determinatives).Some linguists, such as Hurwitz, tried to identify and to systematize by deduction the ternary elements and their semantic values. In this work, which is carried within the framework of the Matrix and Etymons Theory , we demonstrate through the synchronic analysis of nearly 1,000 trilateral items that the triliteral root is analyzable in terms of biliteral etymons and of separable increments or affixes added at the beginning (prefixation), the middle (infixation) or the end (suffixation) of bilateral bases.The characteristic elements that can be affixed or incremented on the base are phonetically similar (gutturals, sonorants, labial, and some dentals) and correlated often with the same grammatical (factitive, stative, reflexive or middle voice) or semantic (intensive) values. The third position of the root is the position favored in this process of affixation / incrementation. Thus, the triliteral root is not the minimal threshold of meaning, and can be broken down to a biconsonantal base, which became triliteral thanks to an incremental or affixal segment. These findings may affect our perception of the Arabic lexicon, which can now be rearranged around biliterals bases, either abstract ( i.e. phonetic features), or concrete (i.e. historical primitive etymons) that are in turn convertible into triliteral radicals through a preliminary list increments and affixes that specify the primary meaning conveyed by the bilateral base

Ce travail de thèse représente une tentative de rationalisation du lexique espagnol en accordant la priorité au signifiant. Inspiré notamment de Pierre Guiraud, nous y avons établi à la suite de Maurice Molho, Didier Bottineau, Georges Bohas ou Dennis Philps, que les mots peuvent ne référer que par la sollicitation d'une partie de leur forme, partie détectable structurellement. Nous avons pris en considération l'ensemble des capacités qualitatives (phones, formant, idéophones, graphèmes, segments) et quantitatives (duplications, répétitions, inversions, homophonies / homographies) de la forme des mots. Nous ne nous sommes donc pas limité à un support sémiologique particulier mais avons conçu le lexique, du fait de sa complexité, comme un organisme de signifiants / signifiés où chaque élément peut entrer en compte pour donner lieu à une motivation (interne ou externe). Nous avons nommé cet élément la saillance, car il s'agit d'une unité résultant d'une focalisation, d'un choix formel pour référer à telle ou telle idée. Nous avons également remarqué, notamment dans des cas d'" homonymie ", que plusieurs parties pouvaient être sollicitées et que chacune permettait de renvoyer à un sens distinct. La consubstantialité du signe est donc, de notre point de vue, un principe sauf. La " synonymie " ainsi que la " polysémie " ne sont donc pas non plus des notions pertinentes car chaque terme renvoie d'une manière qui lui est propre à un sens donné. Enfin, nous avons proposé une application, avec des critères similaires, à des énoncés dit " poétiques " où plusieurs actualisations parfois insolites apparaissent mais toujours permises par le langage, par le signifiant.

Ce travail de thèse a pour but d’explorer les aires cérébrales impliquées dans les traitements syntaxiques et sémantiques des phrases en IRMf chez des adultes sains. Nous avons observé des effets d’adaptation lexico-sémantique dans la plupart des aires temporales et frontales impliquées dans le traitement des phrases. Nous avons trouvé plusieurs régions montrant des effets d’adaptation à la répétition de la forme des mots, du sens des mots et de la structure argumentale de la phrase. Mais aucun effet clair d’adaptation syntaxique n’a pu être observé. Enfin, l’implication des régions temporales supérieures antérieures et postérieures gauches, des régions frontales inférieures gauches et du putamen gauche dans la construction des constituants d’une phrase a été montrée. Ces résultats suggèrent qu’aucune aire cérébrale n’est impliquée dans la représentation de l’arbre syntaxique complet, alors que des aires sont impliquées dans la construction des constituants de la phrase

In order to recognize a written word, the relative positions of its component letters must be encoded. Ultimately, this information must be precise enough to distinguish between anagrams such as causal and casual while retaining enough flexibility to recognize elehpant as elephant. The lexical decision experiments reported here used a more dramatic version of transposed letter priming than has previously been reported in order to identify the constraints on this flexibility. In light of the observed data, several current models of letter position coding were evaluated and suggestions for future models were proposed. The first goal of this research was to determine the degree of flexibility in word recognition in terms of how many transposed letters can be tolerated in the input. Reliable priming was observed throughout the experiments when as many as six of the eight letters had been transposed (most ps < .01). However, Experiments 5 and 6 identified the limit of this flexibility, in that fully transposed primes did not activate their target entries. The second goal was to identify letter position effects, or differences in the importance of various letter positions in lexical access. Experiments 1-4 supported Jordan et al.'s (2003) claim that the exterior letters of a word are the most crucial. Stronger priming was derived from primes with correctly placed exterior letters and transposed interior letters than from the reverse case. Support was also found for Inhoff et al.'s (2003) claim that a word's initial letters are more important to lexical access than later letters (Experiment 7). Overall, a trend of decreasing importance from left to right was observed, with the possible exception of the final letter. The observed data were compared to the predictions made by the BLIRNET model (Mozer, 1991), Grainger & van Heuven's (in press) open bigram coding scheme, the SOLAR model (Davis, 1999), and the SERIOL model (Whitney, 1999). This enabled us to identify particularly effective and problematic approaches to letter position coding. Finally, it is proposed that a visual word recognition system with two parallel, complementary processing streams best describes the data.

Bien que les symptômes moteurs soient prédominants chez les individus atteints de la maladie de Parkinson, les troubles cognitifs font aujourd'hui partie intégrante du spectre symptomatique de la maladie. Depuis peu, des troubles du langage ont été rapportés. Cette étude s'inscrit dans ce contexte et a pour objectif d'explorer l'influence de facteurs cognitivo-linguistiques sur la motricité des patients parkinsoniens. Pour cela, nous avons comparé trois tâches mettant en jeu trois types de motricités différentes chez quatre groupes de sujets : des sujets sains jeunes et âgés ; des patients parkinsoniens avec médication et d'autres patients parkinsoniens sans médication. Ces trois tâches avaient comme caractéristique principale de comparer des mots et des pseudo-mots. La 1ère tâche était une tâche de décision lexicale, la 2ème une tâche de réponse verbale et la 3ème une tâche d'écriture. Dans les 3 tâches, les stimuli étaient vus ou dictés. Nos résultats confirment que les patients parkinsoniens sans médication sont plus lents à réagir que les contrôles âgés. Néanmoins, ce ralentissement ne résulte pas de l'akinésie classiquement décrite. Ces patients parkinsoniens sans médication présentent en effet un trouble auditif majeur et un ralentissement cognitif dans les situations qui nécessitent un traitement lexical. Enfin, ces mêmes patients ont des difficultés à inhiber des processus automatiques qui viennent interférer et ralentir l'exécution de leur tâche motrice. Ces déficits semblent s'estomper sous traitement. Notre étude met ainsi en évidence l'existence de déficits cognitifs qui retardent l'initiation de la réponse motrice des patients parkinsoniens sans médication
Even though the dominant symptom of Parkinson's disease (PD) is motor impairment, cognitive impairment is currently also considered an important symptom. Recently, language impairment has been reported in PD as well. The present study follows up on recent advances in PD research, and aims to explore the influence of cognitive-linguistic factors on motor control in PD. To this end, we compared three tasks, each of which relies on a different type of motor control. We tested four groups of participants: healthy young participants, healthy elderly participants, PD patients on medication, and PD patients off medication. In all three tasks, the primary comparison was between responses to words and pseudo-words, which were presented visually or auditorily. The first task was a lexical decision task, the second a verbal response task, and the third was a handwriting task. Our results show, in line with previous studies, that off-medication PD patients respond more slowly than healthy control participants. However, this slow-down does not result from akinesia, a well known symptom of PD. Instead, off-medication PD patients show auditory impairment and cognitive slowing in situations that require lexical processing. Moreover, these patients have an additional deficit in inhibiting automatic (lexical) processes, which interfere with the motor task. All of these deficits seem to be reduced by medication. Therefore, our study shows clear evidence for cognitive deficits in PD. These cognitive deficits slow the initiation of a motor response in off-medication PD patients

La dominanza dell’emisfero cerebrale sinistro (ES) sul destro (ED) nell’ambito della lettura è ampiamente riconosciuto nell’ambito delle neuroscienze cognitive. Tuttavia, non è ancora chiaro se questa dominanza sia basata sulla completa cecità verbale dell’ED (dominanza assoluta dell’ES), o sulla presenza di rappresentazioni più deboli o minori in numero nell’ED rispetto all’ES (dominanza relativa dell’ES). In linea con la prima ipotesi, leggere dovrebbe essere impossibile se il “sistema delle rappresentazioni visive delle parole” nell’ES è lesionato e sia le connessioni afferenti ed efferenti di quest’area con il resto dell’ES sono interrotte dalla lesione. Inoltre, secondo questa ipotesi, le differenze emisferiche all’interno dei compiti di lettura lateralizzata sono spiegabili in termini di una maggiore sensibilità dell’ED rispetto all’ES a fattori di elaborazione ortografici pre-lessicali. Secondo l’approccio che propone una dominanza relativa dell’ES per la lettura e che suggerisce l’esistenza di abilità lessicali più limitate nell’ED rispetto all’ES, la lettura dovrebbe essere possibile anche in caso di lesione e disconnessione del “sistema delle rappresentazioni visive delle parole” nell’ES, sebbene limitata a parole frequenti e/o concrete. In linea con questo approccio, le differenze emisferiche in compiti di lettura lateralizzata sarebbero spiegabili da variabili semantico-lessicali. Le predizioni di questi modelli sono state testate attraverso uno studio comportamentale e di disconnessione strutturale in una paziente con Alessia Pura, e tramite due studi di lettura con campo visivo diviso (condotti su soggetti destrimani e destrimani e mancini, rispettivamente) in cui fattori lessicali e pre-lessicali sono stati manipolati. La presenza di abilità di lettura residue nel caso di lesione del “sistema delle rappresentazioni visive delle parole” dell’ES e di una sua disconnessione con il resto dell’ES, insieme all’evidenza che un effetto lessicale può spiegare l’effetto di campo visivo/emisfero all’interno di un compito di lettura lateralizzata, sono state considerate come prove a favore dell’ipotesi di una dominanza relativa dell’ES, in linea con la proposta che esista un lessico limitato/debole nell’ED. Uno studio computazionale condotto per simulare lo sviluppo di rappresentazioni ortografiche nei due emisferi suggerisce che rappresentazioni ortografiche più deboli nell’ED rispetto all’ES -che danno luogo a tale vantaggio relativo dell’ES sull’ED- potrebbero essere dovute a un inefficiente consolidamento di conoscenze ortografiche nell’ED.
The dominance of the left cerebral hemisphere (LH) over the right one (RH) for reading is widely recognized in the domain of cognitive neuroscience. However, it is still not clear whether such dominance is underlaid by a complete word blindness of the RH (absolute LH dominance) or by poorer/weaker lexical representations in the RH than in the LH (relative dominance). According to the first account, reading should be impossible when the visual word form system of the LH is lesioned and both afferent and efferent connections with the rest of the LH are interrupted by the lesion. Also, according to this account, hemispheric differences in lateralized reading in healthy subjects are explained by greater sensitivity of the RH than the LH to pre-lexical orthographic processing factors. According to the framework advocating a relative LH dominance for reading and suggesting poorer lexical orthographic abilities of the RH than the LH, reading should be possible even in case of lesion and disconnection of the LH reading system, although limited to frequent and/or concrete words. Accordingly, hemispheric differences in lateralized reading in healthy subjects should be explained by lexical-semantic factors. The predictions of these models were tested by means of a behavioral and structural disconnectome study on a patient with Pure Alexia, and two divided visual field reading studies (conducted on healthy right-handed and left-and right-handed subjects, respectively) in which pre-lexical and lexical factors were manipulated. Evidence of residual reading abilities in case of a LH visual word form system lesioned with both afferent and efferent connections with the rest of the LH being interrupted, together with evidence of a lexical effect accounting for visual field/hemisphere differences in lateralized reading supported the view of a relative LH dominance, according to the idea of the existence of a poorer/weaker orthographic lexicon in the RH. A computational modelling study conducted to simulate the development of orthographic representations in the two hemispheres suggested that weaker orthographic representations in the RH than in the LH -giving rise to such relative LH advantage for reading- could be due to inefficient consolidation of orthographic knowledge in the RH.

L’objectif de ce travail de thèse est de voir dans quelle mesure la motivation participe à l’engagement des processus attentionnels pour la sélection et le traitement des informations lors d’une activité de recherche visuelle. Lorsque la recherche visuelle est orientée par un but (comme une consigne) le comportement visuel d’exploration est modifié (sélection et traitement des informations). La recherche d’information suppose également un engagement de l’individu pour atteindre le but fixé en mettant en place des comportements adaptés. Cet engagement traduit une motivation qui est ici appréhendée au travers de la théorie de l’assignation d’objectif (Locke & Latham, 1990, 2002; Latham & Locke, 2007). Une tâche de recherche visuelle d’une cible spécifique, nommée par sa catégorie sur-ordonnée parmi des distracteurs de catégorie sémantique différente, associée à une consigne spécifique (de vitesse, de précision ou de faire de son mieux) ont été utilisées pour répondre aux objectifs de recherche. Les objectifs des travaux présentés ici étaient de (1) rendre compte de l’effet de l’assignation d’objectif sur le comportement de recherche, (2) étudier la relation entre l’objectif assigné et la spécificité de la tâche et (3) observer l’effet du contexte sémantique sur l’assignation d’objectif. Les trois expériences menées ont permis de mettre en évidence une relation entre la motivation et l’attention sur le comportement visuel de recherche. Ce travail de thèse ouvre sur des perspectives de recherche dans des situations plus écologiques comme les sites web ou les jeux vidéo
The aim of this study was to explore the link between motivation and attentional processes in regards to selection and information processes during visual-search task. When the visual search is goal driven (i.e., with an instruction) visual search behavior is changed (selection and information processes). During visual search, the individual need to be committed in order to reach the assigned goal by using a strategic behavior. This commitment could be defined as motivation as described in the Goal Setting theory (Locke & Latham, 1990, 2002; Latham & Locke, 2007). A visual search task was designed in which participants had to find a specific word, denominated by its super-ordinated category, among distractors. A specific instruction (speed, accuracy or do your best) was added in order to specify the goal of the search. The aims of this work were to (1) reveal the effect of the goal setting on search behavior, (2) study the relationship between the goal setting and the characteristic of the task and (3) observe the effect of semantic context on goal setting. The three experiments allow us to observe a relationship between motivation and attention on visual search behavior. This research concludes on the possibilities to lead more experiments on ecological environment like websites or video games

L’any 2002 es constata que el pacte constitucional de 1978 s’ha esgotat: si Catalunya anhela més autogovern, l’única via és la reforma estatutària. Així comença la reforma de l’Estatut d’autonomia de Catalunya, vigent des del 9 d’agost de 2006. Aquesta tesi ha estudiat quantitativament i qualitativament (amb Lexico3, Coocs i R) el procés d’aprovació de l’EAC 2006 a partir dels textos dels grups parlamentaris i de les versions consensuades en seu parlamentària. La primera part de l’anàlisi caracteritza lexicomètricament totes les parts del corpus. La segona, inclou un estudi qualitatiu de les formes específiques més significatives. Hi hem observat principalment quatre tipus de canvis, que s’expliquen per l’adequació de l’EAC 2006 als factors següents: • als criteris lingüístics i estilístics dels Serveis d’Assessorament Lingüístic del Parlament; • als criteris tècnics dels Serveis Jurídics del Parlament; • a la negociació política; • als criteris lingüístics, estilístics i jurídics de les Corts Generals.
In 2002, it became clear that the 1978 Spanish constitutional agreement was obsolete. Statutory reform represented the only path to greater Catalan self-government. Hence, the process to reform the Catalan Statute of Autonomy (EAC: Estatut d’autonomia de Catalunya) began; the EAC came into force on August 9, 2006. This thesis presents a quantitative and qualitative study (using Lexico3, Coocs and R) of the EAC drafting and approval process based on the texts produced by parliamentary groups and consensus texts agreed upon in parliament. The first part of the analysis offers a lexicometric characterization of the entire corpus. The second part includes a qualitative study of the most significant forms. Four major types of changes were observed. Amendments aimed to rewrite the 2006 EAC based on: • the linguistic and stylistic criteria of the Catalan Parliament’s language services unit; • the technical criteria of the Catalan Parliament’s legal services units; • political negotiation; and, • the linguistic, stylistic and legal criteria of the Spanish Parliament.

Sentiment analysis is a method for examining data in a particular kind of text and identifying its opinion-based content. It is also known as emotion mining or emotion extraction. These days, online communication tools like Twitter, Facebook, YouTube, and others are bringing a lot of zeal into people’s lives. People discuss it and express their opinions or feelings. In this review article, We decide to focus on opinion mining, also known as emotion evaluation, which is a subset of machine learning and online data mining. This article presents the findings of a study that combined multiple Machine Learning(ML) and Lexicon exploration techniques. To conduct an assessment study and validate the estimated quantity of the existing composition, the data are examined. Fu ture researchers will be better equipped to understand the existing beginnings of the arrangement of possibility evaluation as a consequence. Emotion recognition in conversations (ERC) is a field of study and technology that focuses on identifying and understanding emotions expressed in conversations. The goal is to automatically detect and categorize emotions in text-based conversations, such as emails, chats, and social media posts. ERC is interdisciplinary, drawing on fields such as Natural Language Processing (NLP), psychology, and computer science. It has appli cations in areas such as customer service, mental health, and market research. There is currently no database of emotional conversations with more than two speakers per dis cussion in regional Indian languages including Tamil, Telugu, Malayalam, and Bengali. I suggest the Dataset comprise roughly 2000+ utterances from 800+ conversations from the Indian film series Bahubali in order to fill this gap. Each speech will have textual Modalities and be tagged with emotion, personality, and sentiment labels.

碩士
國立清華大學
語言學研究所
93
This study examines the constructional idioms in Chinese within the framework of lexical semantics. In general, there are various ways in which Chinese conceptualizes an event involving two lexical entities. A typical example is chi doufu, a certain V+O set of the verbal ‘eat’ and nominal element ‘beancurd’. The idiomatic interpretation of the idiom ‘came on to’ is not deducible from its component parts. The major questions we attempt to answer are: What are the characters of idioms in language? What is the relationship between the constituents of an idiom? How do they reflect on the mapping that lies within the domain of mental among the multi-layers of grammar? How are they associated with conceptual schema? Do their information structures bear any kind of information, e.g., the semantic, syntactic and pragmatic properties? Our goal is to accomplish the tasks for which the idiomatic meaning of an idiom is being constructed in the lexical semantic structure. The organization of this thesis is as follows. Chapter I gives a brief introduction to previous studies of idiomatic expressions and an overview of the thesis. Chapter II presents the theoretical background. Chapter III analyzes lexical boundaries of idioms in order to further the interaction between combinations of semantic (or discourse) functions. The idioms under discussion are classified into three categories, namely, lou majiao ‘to betray the pot to the roses’, chi doufu ‘to take liberties with’ and dai lumao ‘being a cuckold husband’. We address their conceptual structures and thematic relations by setting out the basis of the idiosyncratic information of their meanings. Chapter IV is the kernel core of the thesis: it deals with the specification of argument structure and constructional template of idiom chunks of events in a given society, involving participants that include the communicator and the addressee. We primarily extract authentic data from corpora to validate the proposal suggested here. Chapter V lays out the compatibility between idioms and four other Chinese constructions: the genitive construction, the passive construction, the ditransitive and pivotal constructions. Chapter VI elucidates the semantic extension in idioms from the literal sense to idiomatic meaning with two cognitive mechanisms: metaphor and metonymy. For example, the central or prototypical reading of the idiom dai maozi ‘to wear hat’ is ‘to put a label on’. We assume here the metaphor as IDENTITY IS HAT. It is taken as a source of insight into personal status of common judgments in the culture. We show that most of the intertwined network can be explained by the account of family resemblance. Lastly, Chapter VII concludes that our purpose is to make considerable claims about the degree of idiomaticity. After all, it should be clear from this investigation that we are in need to take idioms as constellations of semantic roles since we aim to understand more fully the coverage of linguistic universals.

Thesis ( M. A. (Translation Studies and Linguistics)) -- University of Limpopo, 2017
The study explored the restandardization of Sepedi with the aspiration of including Khelobedu dialectical lexicons in the standard form. The standardization of Sepedi, unlike the case of Shona, excluded many of its dialects from the process, thus, left Khelobedu speakers outside of this medium and later subjected them to learn it in schools, putting them at a point of disadvantage academically. Very few studies have been conducted around this term restandardization. This study is mixed method in approach and sequential in design. Data is collected via self-administered questionnaires and face-to-face interviews using an interview guide. A total of 20 participants from four villages in the Mopani District made up a sample for the quantitative data collection phase, while four participants who are Language practitioners by profession made up the qualitative phase of the study. The findings of the study reveal that dialect speakers do not have much confidence in their dialectical variety. They still believe that English and Sepedi are mediums of development and progress. Although restandardization according to the language practitioners is said to possible, PanSALB still has a lot to do in terms of developing Indigenous Languages in South Africa.

This study investigates undergraduate students’ vocabulary size, and their use of formulaic language. Using the Vocabulary Levels Test (Laufer and Nation 1995), it measures the vocabulary size of native and non-native speakers of English and explores relationships between this and course of study, gender, age and home language, and their academic performance. A corpus linguistic approach is then applied to compare student writers’ uses of three high-frequency verbs (have, make and take) relative to expert writers. Multiword units (MWUs) featuring these verbs are identified and analysed, focusing on delexical MWUs as one very specific aspect of depth of vocabulary knowledge. Student and expert use of these MWUs is compared. Grammatically and semantically deviant MWUs are also analysed. Finally, relationships between the size and depth of students’ vocabulary knowledge, and between the latter and academic performance, are explored. Findings reveal that Literature students had larger vocabularies than Law students, females knew more words than males, and older students knew more than younger ones. Importantly, results indicated a relationship between vocabulary size and academic performance. Literature students produced more correct MWUs and fewer errors than Law students. Correlations suggest that the smaller students’ vocabulary, the poorer the depth of their vocabulary is likely to be. Although no robust relationship between vocabulary depth and academic performance emerged, there was evidence of an indirect link between academic performance and correct use of MWUs. In bringing together traditional methods of measuring vocabulary size with an investigation of depth of vocabulary knowledge using corpus analysis methods, this study provides further evidence of the importance of vocabulary knowledge to academic performance. It contributes to debates on the value of a sound knowledge of high-frequency vocabulary and a developing knowledge of at least 5000 words to academic performance, and the analysis and quantification of errors in MWUs adds to our understanding of novice writers’ difficulties with these combinations. The study also explores new ways of investigating relationships between size and depth of vocabulary knowledge, and between depth of vocabulary knowledge and academic performance.
Linguistics and Modern Languages
D. Litt. et Phil. (Linguistics)

In many international schools where English is the language of learning and teaching there are large percentages of students whose first language is not English. Many of these students may have low vocabulary levels which inhibits their chances of taking full advantage of their education. Low vocabulary levels can be a particular problem for students in mainstream classes where fluent English speaking teachers are using English to teach content areas of Mathematics, Science and History. Not only do students have to comprehend the low-frequency, academic and technical vocabulary pertaining to the subject, but they also need to know the higher frequency vocabulary that makes up general English usage. If students’ vocabulary levels fall too far below the vocabulary levels with which their teachers are speaking, then their chance of comprehending the topic is small, as is their chance of succeeding in their subjects. This study has two broad aims. Firstly, I have set out to assess the English vocabulary levels of students at an international school where English is the language of learning and teaching. The majority of students at this school do not have English as their first language. The second aim of this study is to explore the vocabulary profile of the teachers’ spoken discourse at the research school. By gaining a better understanding of the nature of teacher discourse – specifically the percentage of high, mid and low-frequency vocabulary, as well as academic vocabulary that they use – English as a Second Language (ESL) teachers will be in a stronger position to identify what the vocabulary learning task is and be able to assist students in reaching the vocabulary levels necessary to make sense of their lessons. This study revealed a large gap between the generally low vocabulary levels of ESL students and the vocabulary levels spoken by their teachers. As a result the need for explicit vocabulary instruction and learning is shown to be very important in English medium (international) schools, where there are large numbers of students whose first language is not English.
Linguistics and Modern Languages
M.A. (Applied Linguistics)