Матричне профілювання семантики фразово-дієслівних емотивів у Британському національному корпусі

Abstract

У дисертації розроблено нову концепцію для побудови тегової когнітивно- семантичної матриці фразово-дієслівних емотивів у Британському національному корпусі. Укладено ініціальну докорпусну (лексикографічну) картотеку фразово- дієслівних емотивів за 8 профілями: “Інтерес”, Радість”, “Подив”, “Горе”, “Гнів”, “Відраза”, “Страх”, “Провина” з подальшою верифікацією реєстру заданих одиниць у доменах БНК; побудовано базові граматичні моделі фразово-дієслівних емотивів для 8-х когнітивно-семантичних субматриць на основі програмного забезпечення Just The Word БНК; обчислено кількісно-статистичні показники для кожного субматричного тега (профілю емоцій) та виявлено когнітивно-семантичні зв’язки між емотивними дієсловами, їхніми формами й фразовими дієсловами в доменах БНК; виміряно залежність швидкості пошуку фразово-дієслівного емотиву від кількості й частоти його вживання в усній і писемній частинах БНК; сконструйовано комп’ютерну модель когнітивно-семантичної матриці фразово- дієслівних емотивів на основі отриманих результатів пошуку цих одиниць у доменах БНК за профілями 8-х тегів.

В диссертации разработана новая концепция для построения теговой когнитивно-семантической матрицы фразово-глагольных эмотивов в Британском национальном корпусе. Составлена инициальная докорпусная (лексикографическая) картотека фразово-глагольных эмотивов за 8 профилями: “Интерес”, “Радость”, “Удивление”, “Горе”, “Гнев”, “Отвращение”, “Страх”, “Вина” с дальнейшей верификацией заданных единиц в доменах БНК; построены базовые грамматические модели фразово-глагольных эмотивов для 8 когнитивно- семантических субматриц на основе программного обеспечения Just The Word БНК; измерены количественно-статистические показатели для каждого субматричного тега (профиля эмоций) и выявлены когнитивно-семантические связи между эмотивными глаголами, их формами и фразовыми глаголами в доменах БНК; установлена зависимость скорости поиска фразово-глагольного эмотива от количества и частоты его употребления в устной и письменной частях БНК; сконструирована компьютерная модель когнитивно-семантической матрицы фразово-глагольных эмотивов на основе полученных результатов поиска этих единиц в доменах БНК по профилям 8 тегов.

The thesis develops a new concept for the construction of the cognitive and semantic matrix of phrasal verbs denoting human emotional states in the British National Corpus. For the first time, a new definition of phrasal verbs is suggested as of unique cognitive and grammatical constructs of “a verb + a postpositive component” in the English language that reflect a grammatical pattern of their conceptually combinatorial construction based on conceptual integration mechanisms, being both innovative and conservative. Its innovativeness is in the selective metaphorical design of the conceptual structure using initial spaces and operations of involving background knowledge, while its conservatism lies in the formal and grammatical expression of a phrase formation according to the rules of the English grammar. To measure the capacity of these entities denoting human emotional states, a comprehensive technique for computer modeling of tag and cognitive and semantic submatrices (involving the BNC software) of all verbal designations capable of expressing those states has been developed. In the ВNC, an initial corpus catalogue of verbal designations for human emotional states has been compiled according to 8 cognitive and semantic tag submatrices: “Interest”, “Joy”, “Wonder”, “Grief”, “Anger”, “Disgust”, “Fear”, and “Guilt”. Basic grammatical models of verbal designations for human emotional states based on the Just The Word software for each cognitive and semantic submatrix have been constructed. Lexical statistics for verbal designations and their forms have been estimated for each emotion profile and submatrix tag; their cognitive and semantic links with phrasal and verbal complexes have been identified based on the concordance of the BNC Web software. The dependency of the retrieval rate of the phrasal verbal complex with the semantics of emotions on the frequency of its use in the BNC domains has been measured. A computer model of the cognitive and semantic matrix of phrasal verbal complexes of human emotional states has been constructed with the Sketch Engine and Visual Thesaurus software based on the BNC search results for major archisemes of 8 tags. If we consider 8 basic human emotions such as “Interest”, “Joy”, “Wonder”, “Grief”, “Anger”, “Disgust”, “Fear”, and “Guilt”, which are verbalized by verbal markers, they can be represented as profiles of the general cognitive and semantic matrix of EMOTIONS verified in the BNC domains. The analysis of domains for verbal designation of each basic emotion as cognitive and semantic submatrices, which is performed based on the BNC software, enabled revealing the frequency and productivity of verbal semantics of emotional states in general and phrasal-verbal complexes in particular. Especially productive was the Sketch Engine corpus system due to the availability of tools in it that, implement the method of lexical semantic clustering, the classical method of distribution statistical analysis, and the method of combinatorial syntagmatics. These specifically include such tools as Collocations – the automatic search for collocations (the so-called broad context (domains) when potential links between units under study throughout the corpus are measured); Word sketch – the automatic search for collocations (within a given verbal formation – its morphological syntactic mini-model / formula template; it is the tool that makes a list of the most frequently established collocations calculated on the basis of the logDice statistical measurement separately for each morphological syntactic formula, as well as estimates the total number of such collocations / combinations throughout the corpus); Thesaurus – to establish systemic relationships between a sampling of language units; Clustering – to group the selected verbs denoting human emotional states, their forms and phrasal complexes as thesaurus units into 8 semantic clusters followed by the profile construction; Sketch diff – to identify similarities and differences in mechanisms and methods of combinatorial compatibility of word pairs. Taken as a whole, these and other procedural techniques have made it possible to reconstruct cognitive and semantic links between the units under study along with quantitative and statistical estimation of their productivity for each profile and domain. Further use of the conceptual integration method contributed to the construction of corpus tagged submatrices of verbal designations for human emotional states with all their 8 lexical statistical profiles. The corpus tagged submatrices are modeled as cognitive and semantic submatrices that look like tables where each emotion is a model of a detailed pattern represented by criteria and indicators in specific columns of the table.