Contrary to what has been done to date in the hybrid field of natural language processing (NLP), this doctoral thesis holds that the new approach developed below makes it possible to semi-automatically detect semantic changes in digitised, OCRed, historical corpora. We define the term semi-automatic as “making use of an advanced tool whilst remaining in control of key decisions regarding the processing of the corpus”. If the tool utilised – “topic modelling”, and more precisely the “Latent Dirichlet Allocation” (LDA) – is not unknown in NLP or computational historical semantics, where it is already mobilised to follow a priori selected words and try to detect when these words change meaning, it has never been used, to the best of our knowledge, to detect which words change in a humanistically-relevant way. In other terms, our method does not study a word in context to gather information on this specific word, but the whole context – which we consider a witness to a potential evolution of reality – to gather more contextual information on one or several particular semantic shift candidates. In order to detect these semantic changes, we use the algorithm to create lexical fields: groups of words that together define a subject to which they all relate. By comparing lexical fields over different time periods of the same corpus (that is, by mobilising a diachronic approach), we try to determine whether words appear over time. We support that if a word starts to be used in a certain context at a certain time, it is a likely candidate for semantic change. Of course, the method developed here and illustrated by a case study applies to a certain context: that of digitised, OCRed, historical archives in Dutch. Nevertheless, this doctoral work also describes the advantages and disadvantages of the algorithm and postulates, on the basis of this evaluation, that the method is applicable to other fields, under other conditions. By carrying out a critical evaluation of the tools available and used, this doctoral thesis invites the community to the reproducibility of the method, whilst pointing out obvious limitations of the approach and propositions on how to solve them. / Doctorat en Information et communication / info:eu-repo/semantics/nonPublished

Semantic change is a phenomenon that has been subject to a lot of research during the past few decades. However, a large part of the existing research has been conducted with the goal of finding causes of change or creating typologies in order to classify different types of changes. The present study has been conducted with the aim to examine how a specific extra-linguistic factor has affected a select few words. The extra-linguistic factor that is the focal point of this study is the rapid technological change during what is known as the digital revolution. This essay explores how the digital revolution has affected the uses of four words: bug, web, mouse,and cloud. The first part of this study was a collocational analysis of these four words. The results of the collocational analysis indicated that changes occurred during certain time periods. A closer context analysis was performed for each of the words on the time period during which a semantic change was suspected to occur. The findings of this essay are that all the examined words have gotten new technological meanings during the past 70 years, thus exhibiting semantic widening. All four words are currently polysemous words. Replacive change, meaning a change in the primary meaning of a word, is only apparent in web. The remaining words keep their primary senses throughout the examined period, and the new senses are added as periphery senses. The trends in the usage of these words indicate that it is possible that more of them will undergo replacive change, however, it is too early to tell.

Der Einsatz von automatisierten quantitativen Methoden in den Sozialwissenschaften gewinnt stetig an Bedeutung. Dies hat zum einen mit der rasant wachsenden Menge und Verfügbarkeit digital vorliegender Daten zu tun. Zum anderen erlauben es innovative automatisierte Ansätze, Ergebnisse zu produzieren, welche durch qualitative Arbeit allein nicht möglich wären. Die Implementierung innovativer Algorithmen zur Anwendung quantitativer Verfahren bedarf jedoch eines großen Maßes an Wissen im Bereich der Programmierung sowie der Funktionsweise der anzuwendenden Methoden. Da dieses Expertenwissen aber nur in den wenigsten Fällen in rein sozialwissenschaftlichen Projekten vorhanden ist, ist es notwendig, andere Lösungsmöglichkeiten zur Anwendung automatisierter quantitativer Verfahren in den Sozialwissenschaften zu nutzen. Lediglich die Bereiche der Computational Social Science sowie die Digital Humanities stellen Forschungsbereiche der Sozialwissenschaften dar, welche als Vorreiter bereits Erfahrungen im Umgang mit automatisierten quantitativen Verfahren aufweisen. Eine mögliche Lösung für den breiten Einsatz von automatisierten Verfahren in den gesamten Sozialwissenschaften ist die Erstellung und Anwendung von Text-Mining-Infrastrukturen, die speziell für den Einsatz in den Sozialwissenschaften ausgerichtet sind. Diese erlauben es Sozialwissenschaftlern, mit einer vergleichsweise geringen Einstiegshürde aktuelle Verfahren und Forschungsansätze der Bereiche Text Mining und Machine Learning auf ihre eigenen Forschungsfragen und Daten anwenden zu können. Damit diese Infrastrukturen aber auch tatsächlich einen deutlichen Mehrwert für den Sozialwissenschaftler darstellen, müssen verschiedene Anforderungen erfüllt werden. Diese teilen sich auf in generelle an Software gestellte Forderungen wie beispielsweise Skalierbarkeit und Performanz sowie in spezifische Anforderungen für die Anwendung in den Sozialwissenschaften. Zu diesen speziellen Anforderungen zählt die Möglichkeit des Umgangs mit verschiedenartigen Datengrundlagen. In dieser Arbeit wird der Fokus auf textuelle Daten gelegt, wobei auch diese sehr große Unterschiede in ihrer Charakteristik und damit in deren notwendiger Verarbeitung aufweisen. Es werden darüber hinaus drei Schlüsselanforderungen identifiziert, die für den Einsatz inden Sozialwissenschaften essentiell sind. Die erste Schlüsselanforderung beschreibt die generelle Ausrichtung einer Text-MiningInfrastruktur als generische Plattform, welche durch die Merkmale von Anpassbarkeit, Erweiterbarkeit sowie der Möglichkeit des Exportes von Ergebnissen an die zahlreichen zum Teil sehr diversen Forschungsfragen der Sozialwissenschaften assimiliert werden kann. Die zweite Schlüsselanforderung stellt die Notwendigkeit, qualitative und quantitative Forschungsdesigns durch die Implementierung von dafür vorgesehenen Interfaces vereinen zu können, in den Vordergrund. Beide Forschungsansätze können auf diese Weise voneinander profitieren. Zuletzt wird noch die Bedeutung von schwachen Signalen als Forschungsgrundlage in den Sozialwissenschaften hervorgehoben. Für alle drei dieser Schlüsselanforderungen als auch die übrigen abgeleiteten Anforderungen an eine Text-Mining-Infrastruktur für den Einsatz in den Sozialwissenschaften werden mögliche Implementierungen und Lösungsansätze präsentiert. Dies geschieht zum einen durch die Beschreibung des Designs und der Entwicklung genau einer solchen Text-Mining-Infrastruktur am Beispiel des interactive Leipzig Corpus Miner. Es werden notwendige Abwägungen bezüglich verschiedener Implementierungsstrategien und Softwaredesignentscheidungen, welche zur Umsetzung der gestellten Anforderungen notwendig sind, erläutert. Zum anderen wird ein Maß zur Quantifizierung von diachronen Kontextänderungen in der Form der Kontextvolatilität vorgestellt. Das Maß wird im Laufe der Arbeit zur Detektion und Analyse schwacher Signale in textuellen Daten eingesetzt. Im letzten Teil der Arbeit werden die realisierten Umsetzungen der Schlüsselanforderungen am Beispiel verschiedener durchgeführter Projekte aufgezeigt. Die wichtigsten Beiträge dieser Arbeit sind damit zum Ersten eine Aufstellung spezifischer Anforderungen an Text-Mining-Infrastrukturen für den Einsatz in den Sozialwissenschaften. Zum Zweiten wird darauf aufbauend ein mögliches Design einer daraus resultierenden Forschungsumgebung detailliert erläutert. Den dritten Beitrag dieser Arbeit stellt die Weiterentwicklung der Kontextvolatilität als Verfahren zur Detektion schwacher Signale in diachronen Daten dar.

The presented study focuses on the processes involved in the formation of words in contemporary English slang. The aim of the study is to determine which patterns of word building are the most frequent, also compared to the general word stock of contemporary English, based on the analysis of a representative sample of contemporary slang expressions. In addition, the study attempts to determine if there are any phenomena that have been given little or no attention so far. The study consists of two parts. The theoretical part describes the phenomena that are relevant to the scope of the research and the study in general. The first section of the theoretical part includes a description of slang, also pointing out the differences in the treatment of slang between the English-speaking and Czech linguistic tradition. The second section contains a description of the source of the data used in the research, namely, the open Internet repository of slang expressions Urban Dictionary, which is one of the most widely used slang dictionaries on the Internet. The third section provides an overview of word- formation processes that are productive in contemporary English and relevant to the source material of slang lexemes. They include affixation, compounding, conversion, shortening, semantic change, borrowing or...

This essay examines how the sense of the word parent has developed and possibly changed during the 19th and 20th centuries. The hypothesis is that father was the most common meaning in the early 1800s and that by the end of the 20th century it had changed into having a more general sense, denoting all caregivers of a child. The research has been performed as a corpus study, looking at and analyzing corpus data in the Corpus of Historical American English (COHA) from three different decades – the 1820s, the 1900s, and the 1990s. The word parent was analyzed in 100 samples from each of the three decades by studying the expanded contexts of the word in COHA, and categorizing the perceived meaning into one of seven definitions. The results show that father was the most common sense in the 1820s, while origin was the most frequent meaning in the 1900s. Last but not least, in the samples from the 1990s, either as sense had the highest frequency. Occurrences are analyzed both by decade and by source type. The results indicate that one should be mindful about making assumptions about meaning based only on knowledge of the sense as used in current discourse. Any text should be read and understood in context while taking historical circumstances into account. The definition of parent has changed, both in dictionaries and in the public mind, and there are signals that changes in the legal definition of parent are also to be expected.

The object of this study is to analyse the lexemes and phraseological units that constitute the semantic fields employed in naming four abstract domains, greed, thrift, generosity, and extra­vagance that make up the ethical concept <Attitude to Ownership> in German. On the assump­tion that ideas are accessible to us through the lexicalised items of a language, recent theories in the field of semantic analysis and conceptualisation were applied to the source material. In each domain key words were identified and their definitions in modern and historical dictionaries were analysed. Various dimensions of meaning, which proved to be inherent in the lexical items, emerged from this analysis. The oppositions a/o (action directed to others vs. to oneself), right/wrong (virtues vs. vices) and too much/ too little vs. the ideal mean were established as central. To achieve a more precise description of meaning tentative explications of cognitive levels were proposed. By means of these the underlying ideas, as they were reflected in the lexical units, could be described. The analysis showed greater variation and expressivity in words, idioms, and proverbs referring to the two vices compared to the virtues. Furthermore, a diachronic study produced evidence of semantic and conceptual changes. On the basis of such observations conclusions could be drawn about changes in the ethical system. The data derived from a contrastive corpus analysis of the German and Swedish key words showed numerous similarities as well as some conspicuous differences in the conceptualisation and valuation of attitudes pertaining to the four abstract domains. Moreover, the key words denoting the two virtues showed a clear domination in frequency, indicating that these are more central conceptual categories in today's society than the vices. An ongoing shift in meaning could be established for the key words naming the latter. Applying modern theories of metaphor and metonymy the experiential basis of meaning and thought was explored, showing that the structures forming the ethical concepts studied in this work are grounded in experiences of a physical and socio- cultural nature. The metaphorical concept ILLNESS emerged as a common source domain for the two vices, while the PATH- concept was shown to form the basis of metaphors expressing the o-virtue but not the a-virtue. Among the numerous métonymie concepts HAND proved to be a characteristic of all four domains. / digitalisering@umu

<p>Some people might regard the language of hip-hop as being crude, sexist and inappropriate. Nevertheless, hip-hop culture can also be considered as one of the underground sources of word-formation and language change in mainstream English. Young people have always been a source of language variation and lexical innovation whether we like it or not.</p><p>This essay is focusing on three words frequently used in hip-hop lyrics, namely pimp, queer and chick. The aim of the study is to analyse the semantic change on these words as they are used in hip-hop music and find out whether hip-hop culture has had an influence on the mainstream usage of these words.</p><p>The method used was to study the usage and frequency of these words in hip-hop lyrics and analyse whether there was a similar use in a large diachronic newspaper corpus of American English. The material was predominately found in The Original Hip-hop Lyrics Archive (ohhla.com) and The TIME Magazine Corpus (corpus.byu.edu).</p><p>The results of this study show examples of semantic change in both hip-hop lyrics and mainstream American English and it is concluded that hip-hop culture has an influence on mainstream language above all among the youth.</p>

Some people might regard the language of hip-hop as being crude, sexist and inappropriate. Nevertheless, hip-hop culture can also be considered as one of the underground sources of word-formation and language change in mainstream English. Young people have always been a source of language variation and lexical innovation whether we like it or not. This essay is focusing on three words frequently used in hip-hop lyrics, namely pimp, queer and chick. The aim of the study is to analyse the semantic change on these words as they are used in hip-hop music and find out whether hip-hop culture has had an influence on the mainstream usage of these words. The method used was to study the usage and frequency of these words in hip-hop lyrics and analyse whether there was a similar use in a large diachronic newspaper corpus of American English. The material was predominately found in The Original Hip-hop Lyrics Archive (ohhla.com) and The TIME Magazine Corpus (corpus.byu.edu). The results of this study show examples of semantic change in both hip-hop lyrics and mainstream American English and it is concluded that hip-hop culture has an influence on mainstream language above all among the youth.

MASLIŠ, Martin. Isolated nominal forms in Ancient Greek. Prague, 2019. Master's thesis. Charles University. Faculty of Arts. Institute of Greek and Latin Studies. Supervisor Ján BAKYTA. The submitted thesis is a study of isolated nominal forms in Ancient Greek undergoing semantic change. Synchronically isolated lexemes are defined as forms which cannot be classed with their respective cognates by the speakers due to formal disparities. This leads to increased opacity which in turn hampers the mapping of meaning onto the forms during the acquisition phase. Accordingly, a hypothesis that isolated forms are susceptible to semantic change is tested. An argumentation is developed based on the assumed information structure in the mental lexicon of speakers and the mechanism of word meaning acquisition. This process allows for reanalysis on the part of speakers which results in semantic change from the diachronic point of view. Lexemes which have been selected according to the proposed research method are organized in a dataset and their occurrences textually analysed. A synthesis aims to describe the mechanism of semantic change for the transparent as well as isolated forms. Keywords: mental lexicon, derivational family, isolated forms, semantic change, deetymologization

Cette thèse vise à élucider les mécanismes du changement sémantique en diachronie courte (ou micro-diachronie) dans des corpus. Pour comprendre, analyser et modéliser la dynamique de ces changements et poser les jalons d’un traitement dynamique du langage, le corpus est segmenté en une série de périodes temporelles d’un mois. Ce travail utilise le modèle ACOM (de H. Ji), qui s’inscrit dans le paradigme des Atlas Sémantiques, un modèle géométrique de représentation du sens basé sur l’analyse factorielle des correspondances et la notion de cliques. Les questions de traitement statistique du langage et du sens, de modélisation et de représentation sont traitées conjointement aux questions d’ordre linguistique, psychologique, et sociologique, dans la perspective d’une analyse multidisciplinaire unifiée, telle que conçue par les sciences cognitives. Une démarche de détection et d’analyse du changement sémantique est proposée, accompagnée d’études de cas qui portent à la fois sur de la détection large et sur des détails précis, proposant différent niveaux de granularité. Le changement sémantique est traité comme un déploiement de la polysémie d’une part, et comme une conséquence des modes de communication liés aux médias actuels et à la diffusion de ceux-ci. Linguistique, sociologie et sciences de l’information se rencontrent dans l’étude de la fabrique de sens nouveaux et de mots nouveaux. L’analyse des réseaux sémantiques des termes étudiés montre la réorganisation constante des sens dans le temps et en capture quelques aspects fondamentaux. Les études de cas portent notamment sur le terme « malbouffe », sur le changement sémantique de l’élément de composition « bio- » et sur le glissement de sens observé pour le terme « mondialisation » par rapport à son quasi-synonyme « globalisation ». Un prototype informatique a été développé pour permettre le traitement de ces études et d’études futures. / This doctoral thesis aims at elucidating the mechanisms of semantic change in short diachrony (or micro-diachrony) in corpus. To understand, analyze and model the dynamics of these changes and lay the groundwork for dynamic language processing, the corpus is divided in a series of time periods of one month. This work uses H. Ji’s ACOM model, which is an extension of the Semantic Atlas, both of which are geometrical models of meaning representation based on correspondence factor analysis and the notion of cliques. Language and meaning statistical processing issues as well as modeling and representation issues are dealt with in conjunction with linguistic, psychological and sociological aspects from a holistic multidisciplinary perspective, as conceived by cognitive sciences. An approach of detection and analysis of semantic change is proposed along with case studies which deal both with large scale and precise detailed phenomena, therefore offering several levels of granularity. On the one hand, semantic change is dealt with as the deployment of polysemy in time, and on the other hand as a consequence of communication methods related to the media and the diffusion of such methods. Linguistics, sociology and information sciences all contribute to the study of the making of new meanings and new words. The analysis of the semantic networks of the studied items show the constant reorganization of meanings in time, and captures a few fundamental aspects of this process. The case studies focus primarily on the French term malbouffe (“junk food”), and on the semantic change of the element of composition bio-, as well as on the connotational drift of the French term mondialisation compared to its near-synonym globalisation (“globalization”). A prototype has been developed for these case studies as well as future studies.