Dissertations / Theses on the topic 'Lexical and semantic aspects'

Applications to word sense disambiguation, an aspect of semantic interpretation, are used to evaluate the contributions. Disambiguation systems which utilize semantically annotated training data are considered supervised. The algorithms of this dissertation are considered minimally-supervised; they do not require training data created by humans, though they may use human-created data sources. In the case of evaluating a database of common sense knowledge, integrating the knowledge into an existing minimally-supervised disambiguation system significantly improved results -- a 20.5\% error reduction. Similarly, the Web selectors disambiguation system, which acquires knowledge directly as part of the algorithm, achieved results comparable with top minimally-supervised systems, an F-score of 80.2\% on a standard noun disambiguation task. This work enables the study of many subsequent related tasks for improving semantic interpretation and its application to real-world technologies. Other aspects of semantic interpretation, such as semantic role labeling could utilize the same methods presented here for word sense disambiguation. As the Web continues to grow, the capabilities of the systems in this dissertation are expected to increase. Although the Web selectors system achieves great results, a study in this dissertation shows likely improvements from acquiring more data. Furthermore, the methods for acquiring a database of common sense knowledge could be applied in a more exhaustive fashion for other types of common sense knowledge. Finally, perhaps the greatest benefits from this work will come from the enabling of real world technologies that utilize semantic interpretation.; This work investigates the effective acquisition of lexical knowledge from the Web to perform semantic interpretation. The Web provides an unprecedented amount of natural language from which to gain knowledge useful for semantic interpretation. The knowledge acquired is described as common sense knowledge, information one uses in his or her daily life to understand language and perception. Novel approaches are presented for both the acquisition of this knowledge and use of the knowledge in semantic interpretation algorithms. The goal is to increase accuracy over other automatic semantic interpretation systems, and in turn enable stronger real world applications such as machine translation, advanced Web search, sentiment analysis, and question answering. The major contributions of this dissertation consist of two methods of acquiring lexical knowledge from the Web, namely a database of common sense knowledge and Web selectors. The first method is a framework for acquiring a database of concept relationships. To acquire this knowledge, relationships between nouns are found on the Web and analyzed over WordNet using information-theory, producing information about concepts rather than ambiguous words. For the second contribution, words called Web selectors are retrieved which take the place of an instance of a target word in its local context. The selectors serve for the system to learn the types of concepts that the sense of a target word should be similar. Web selectors are acquired dynamically as part of a semantic interpretation algorithm, while the relationships in the database are useful to stand-alone programs. A final contribution of this dissertation concerns a novel semantic similarity measure and an evaluation of similarity and relatedness measures on tasks of concept similarity. Such tasks are useful when applying acquired knowledge to semantic interpretation.
ID: 029808979; System requirements: World Wide Web browser and PDF reader.; Mode of access: World Wide Web.; Thesis (Ph.D.)--University of Central Florida, 2011.; Includes bibliographical references (p. 141-160).
Ph.D.
Doctorate
Electrical Engineering and Computer Science
Engineering and Computer Science

Esta dissertação tem por objetivo descrever o léxico da Língua Portuguesa falada por universitários de Maputo, Moçambique. Sob um viés sincrônico, o método de coleta de dados agrega pressupostos teóricos da Geolinguística e da Sociolinguística variacionista, cujo produto é um Questionário Semântico-Lexical QSL, de modelo onomasiológico, dividido por campos semânticos. As bases teóricas da Geolinguística são fornecidas por Brandão (1991), e as da Sociolinguística são embasadas em Tarallo (2004). O aparato cultural e histórico de Maputo é delimitado pelas considerações de Firmino (2006), Gonçalves (1995), e Mondlane (1995). O público alvo para aplicação do inquérito é constituído de sujeitos que estejam cursando ou tenham concluído o ensino superior, identificados como público universitário, com, no mínimo, 10 anos de vida escolar, divididos equitativamente em homens e mulheres de três faixas etárias (18 a 29 anos; 30 a 44 anos; 45 anos ou mais). A partir desta pesquisa será possível identificar peculiaridades intrínsecas ao processo de formação do léxico dos universitários de Maputo, contribuindo, assim, para a descrição e estudo do português falado em Moçambique.
This dissertation aims to describe the lexicon of Portuguese spoken by university students in Maputo, Mozambique. Under a synchronic bias, the method of data collection combines theoretical principles of Geolinguistics and variational Sociolinguistics, whose product is a Lexical-Semantic Questionnaire - QSL, onomasiological model, divided by semantic fields. The theoretical basis of Geolinguistics are provided by Brandão (1991), and of Sociolinguistics are based on Tarallo (2004). The cultural and historic apparatus of Maputo is delimited by considerations of Firmin (2006), Gill (1995), and Mondlane (1995). The target audience for the survey consists of persons who are completing or have completed higher education, identified as a university students, with at least 10 years of school life, divided equally in men and women, in three age groups (18 to 29 years, 30 to 44 years, 45 years or older). From this research can be identified peculiarities inherent to the formation of the lexicon of students in Maputo, thus contributing to the study and description of the Portuguese spoken in Mozambique.

Cette thèse se propose de spécifier la relation entre les structures motrices et celles du langage en tant que systèmes coopératifs dans la construction du sens. Bien qu'un grand nombre d'études aient mis en évidence que les structures motrices sont impliquées dans le traitement du langage, il est encore difficile de déterminer le rôle de ces structures dans la compréhension. Les théories dites «incarnées» et «désincarnées» débattent de la nature de la représentation du sens des mots en termes de la nécessité des structures motrices pour le langage, en négligeant le fait que les conditions de leur activation n’ont pas été décrites. Des recherches récentes soulignent l’importance de la prise en compte des contextes dans lesquels le langage recrute l'activité motrice. Néanmoins, cette tendance est en contradiction avec les présomptions implicites dans la recherche sur l’interaction langage-motricité, qui se basent sur le modèle «deux-étapes» du traitement sémantique et sur la perspective du «dictionnaire» de la représentation du sens lexical. Dans ce cadre, le traitement du sens des mots est pris comme un processus modulaire. Ce n'est qu'une fois ce processus accompli que le contexte peut influencer la signification. Ces présomptions ont biaisé le débat sur le rôle de l'activité motrice induite par le langage, qui se réduirait à la question de savoir si l'activation motrice doit être considérée comme faisant partie de l'accès lexico-sémantique ou comme résultat de la construction d’un modèle de situation. Or, un grand nombre de travaux ont mis en évidence que le traitement lexico-sémantique et le contexte sont interdépendants. Cette connaissance provenant de la psycholinguistique doit être explicitement intégrée à la recherche sur le rôle de l'activité motrice induite par le langage. Dans un effort pour porter le débat hors de la discussion «lexical vs. post-lexical», cette thèse vise à déterminer les conditions sous lesquelles les contextes linguistiques déclenchent l'activité motrice. Pour ce faire, nous avons testé un nouvel outil qui analyse en ligne les modulations de la force de préhension pendant que les participants écoutaient des mots cibles intégrés dans différents contextes. Nos résultats montrent que quand le mot cible était un verbe d'action de la main et que la phrase focalisait l'action (« John signe le contrat»), une augmentation de la force de préhension était observée dans la fenêtre temporelle associée à la récupération lexico-sémantique. Aucune augmentation de la force de préhension comparable n’a été détectée lorsque le même mot d'action était intégré dans des phrases négatives («John ne signe pas le contrat») ou dans des phrases dont le focus avait été déplacé vers l'état mental de l'agent («John veut signer le contrat»)
The present dissertation was conducted in order to specify the relationship between motor and language structures as cooperative systems in lexical meaning construction. Specifically, this thesis aimed at deepening our understanding of how the linguistic context coordinates the recruitment of motor structures during lexical semantic processing. Although the involvement of motor activity in action-related language comprehension is now sufficiently documented, the specific role that motor structures play in action language processing is still unclear. “Embodied” and “disembodied” theories debate the nature of meaning representation in terms of the necessity of motor structures, neglecting the fact that the conditions of their activation during language processing are not well-described. Very recent research has begun to note the necessity of exploring the context under which words trigger modality-specific cortical activity. However, this trend is at odds with implicit theoretical assumptions that have been made in research on motor-language crosstalk, which are based on the “two-step” model of semantic processing and the “dictionary-like” view of lexical meaning representation. Within such framework, word meaning recognition is taken to proceed in a modular fashion. Only after this process has concluded is the context thought to exert its effects. These assumptions have biased the debate on the role of language induced motor activity. The discussion has been centered on whether motor activation should be considered an integral part of the lexical access process or taken as the result of an ensuing “higher order” operation (i.e., situation model construction). A large body of work evidences that lexical semantic processing and semantic context are far more integrated and interdependent. It seems crucial to integrate this knowledge gained from psycholinguistics into the research on the role of language induced motor activity. In an effort to liberate the debate from the “lexical vs. post-lexical” discussion, this thesis aimed at determining the conditions under which language triggers motor activity. To accomplish these objectives, we introduced a novel tool that analyzes on-line modulations of grip-force while participants listened to specific target words embedded within different types of contexts. Our results show that when the target word was a hand action verb and the sentence focus centered on that action (“John signs the contract”), an increase of grip force was observed in the temporal window classically associated with lexical semantic processing. No comparable increase in grip force was detected when the same action word was embedded in negative sentences (“John doesn’t sign the contract”) or in sentences which focus was shifted towards the agent’s mental state (“John wants to sign the contract”). Our results suggest that the presence of an action word in an

Актуальність дослідження обумовлена постійним інтересом лінгвістів до мовних засобів створення мотивації, вивчення мовного впливу в окремих текстах, необхідністю комплексного вивчення та опису засобів персуазивної комунікації на матеріалі англомовних мотиваційних жанрів. Окрім того, на сьогодні тексти мотиваційного дискурсу англійської мови все ще залишаються недостатньо дослідженими як з лінгвістичної, так і з перекладацької точки зору. Це обумовлює необхідність вивчення мовних особливостей таких текстів та їх перекладу українською мовою. Мета дослідження полягає у вивченні лексико-семантичних аспектів мотиваційної літератури на матеріалі книги Н. Хілла «Think and Grow rich» та особливостей їх перекладу українською мовою.
Актуальность исследования обусловлена ​​постоянным интересом лингвистов к языковым средствам создания мотивации, изучению речевого воздействия в отдельных текстах, необходимостью комплексного изучения и описания средств персуазивной коммуникации на материале англоязычных мотивационных жанров Кроме того, на сегодня тексты мотивационного дискурса английского языка все еще остаются недостаточно исследованы как с лингвистической, так и с переводческой точки зрения. Это обуславливает необходимость изучения языковых особенностей таких текстов и их перевода на украинский язык. Цель исследования состоит в изучении лексико-семантических аспектов мотивационной литературы на материале книги Н. Хилла «Think and Grow rich» и особенностей их перевода на украинский язык.
The study's relevance is due to the constant interest of linguists in the linguistic means of creating motivation, the study of linguistic influence in individual texts, the need for comprehensive research, and the description of means of persuasive communication on the material of English motivational genres. In addition, today, the texts of the motivational discourse of the English language remain insufficiently studied both from a linguistic and translation point of view. The purpose of the study is to study the linguistic and semantic aspects of motivational literature on the material of N. Hill's book "Think and Grow rich" and the peculiarities of their translation into Ukrainian.

У вступі викладено актуальність вибраної теми, описані предмет та мета дипломної роботи. У першому розділі описано саме явище «гендеру» та його дослідження. Розподіл соціуму за ознаками статті, так звана – «гендерна нерівність». Поява в мові слів на відображення або тільки чоловічих ознак – «маскулінності» або слова на відображення «фемінності» - жінки та її ознак. Другий розділ присвячений дослідженню лексико-семантичних аспектів гендерно-маркованих одиниць в сучасній англійській мові. Виявлення концептів «чоловіка» та «жінки», виокремлення їх саме за «гендером», а не за «статтю». Утворення слів, які можуть бути характерними для чоловіка, і для жінки. У третьому розділі зображено перекладацькі аспекти гендерно-маркованих одиниць на прикладі політичних промов як чоловіків-політиків, так і жінок-політиків. Вираження певних особливостей у їх промовах, які характерні тільки чоловікам або тільки жінкам. У висновку узагальнюється виконане дослідження та формулюються підсумки по темі даної дипломної роботи.
The introduction describes the relevance of the chosen topic, describes the subject and purpose of the thesis. The first section describes the phenomenon of "gender" and its research. The distribution of the society on the basis of the "sexus", the so-called "gender inequality". The appearance in the language of words to display only the male characteristics - "masculinity" or words to show "femininity" - the woman and her attributes. The second section is devoted to the study of the lexical-semantic aspects of gender-labeled units in modern English. The discovery of the concepts of "man" and "woman", their separation by "gender", not "sexus". Formation of words that may be characteristic for a man and a woman. The third section depicts translation aspects of gender-marked units in the case of political speeches as male politicians and female politicians. The expression of certain features in their speeches, which are characteristic only of men or only women. The conclusion summarizes the performed research and formulates the results on the topic of this diploma.

Thesis (Ph. D.) -- University of Maryland, College Park, 2004.
Thesis research directed by: Computer Science. Title from t.p. of PDF. Includes bibliographical references. Published by UMI Dissertation Services, Ann Arbor, Mich. Also available in paper.

This study explores the two Japanese temporal connectors mae ni and nai uchi ni that express the notion of ‘before.’ These have been claimed to differ in factuality and certainty (Kuno, 1973) and on pragmatic grounds in the form of speaker attitude (Hasegawa, 2015). Using The Balanced Corpus of Contemporary Written Japanese, this study investigates the veracity of previous findings and aims to further deepen the understanding of what sets these two temporal connectors apart. Rather than in factuality or certainty, they are found to differ in lexical aspect and predicate class tendencies, as well as the ability to express minimal scales that work similarly to negative Horn scales in the case of nai uchi ni. As these are more informative the smaller they are, this also explains the pragmatic aspects that have been identified in previous research.
Den här studien utforskar de två japanska temporala konnektorerna mae ni och nai uchi ni som uttrycker begreppet ’innan.’ Det har hävdats att dessa skiljer sig i faktiskhet och säkerhet (Kuno, 1973) och på pragmatiska grunder i form av talarattityder (Hasegawa, 2015). Den här studien använder sig av korpusen The Balanced Corpus of Contemporary Written Japanese för att undersöka sannfärdigheten i tidigare fynd och ämnar att fördjupa förståelsen om vad som skiljer dessa temporala konnektorer. Studien finner att de inte skiljer sig inom varken faktiskhet eller säkerhet utan istället i tendenser i lexikal aspekt och predikatklasser, samt förmågan att uttrycka minimala skalor liknande negativa Horn-skalor i fallet av nai uchi ni. Eftersom dessa är mer informativa ju mindre de är förklarar detta även de pragmatiska aspekter som identifierats i tidigare forskning.

We report a series of experiments using a new methodology to investigate the relationships between visual and verbal representations and the process of acquiring new semantic associations. Transfer of associative information between stimulus modalities was investigated by training paired associations between novel pictures and novel words. Our results showed that the transfer of associations is a symbolic process, occurring only when participants are aware of the correspondence between the visual and the verbal items afforded by the name relations. We also obtained evidence to suggest that symbolic associations develop more readily from picture associations than from word associations. We argue that this is evidence that semantic knowledge is grounded in perceptual experience. Our most striking result, replicated across experiments, is that transfer of associations between modalities only occurs when subjects have specific conscious awareness about the relationships among associations. This should have implications for cognitive theories of symbolic representation. The methods we developed to expose this phenomenon can be extended to examine those implications more thoroughly. We discuss some of these implications in the terms of competing and complementary cognitive and behavioural theories relating representation to perception and symbols. Dual coding models fit our modality-transfer results more readily than single semantic store models, but neither is well suited for interpreting our awareness results, or for iv discussing perceptual grounding of representation. The models of Deacon and Barsalou both focus on systems of distributed representations grounded in perception; the role of awareness in symbol acquisition in their models is discussed and contrasted with theories from the stimulus equivalence tradition of behaviourist research. From these considerations, we argue that implicit associations underpin symbolic associations, but that semantic knowledge is conscious knowledge about the patterns of association which link representations.

There are now extensive behavioral and neuropsychological evidence to indicate that semantic information of a word can be activated without conscious awareness. However, semantic activation alone may not be sufficient for observing semantic priming effects in masked lexical decision task. In the following study, two tasks were used: lexical decision and semantic categorization. Conscious awareness of the prime was systematically manipulated by varying the duration of the prime and by varying the placement of the mask in the prime-target presentation sequence. Priming effects were observed in the semantic categorization task at prime durations of 42 milliseconds but no semantic priming was observed for the same prime duration in the lexical decision task. However, semantic priming effects began to emerge in lexical decision at the longer prime durations (55 & 69 ms) and under the least effective prime-mask presentation sequences. It is proposed that semantic activation alone is not sufficient for semantic priming effects in the lexical decision task but that central executive involvement is necessary, if only at the lowest level, for facilitatory effects to be observed. Furthermore, no such central executive involvement appears to be required for the semantic categorization task. The priming effects obtained in this task is interpreted in terms of a "decision priming" effect.

This thesis is an investigation of phenomena at the interface between syntax, semantics, and pragmatics, with the aim of arguing for a view of semantic interpretation as lexically driven yet contextually dependent. I examine regular, generative processes which operate over the lexicon to induce verbal sense shifts, and discuss the interaction of these processes with the linguistic or discourse context. I concentrate on phenomena where only an interaction between all three linguistic knowledge sources can explain the constraints on verb use: conventionalised lexical semantic knowledge constrains productive syntactic processes, while pragmatic reasoning is both constrained by and constrains the potential interpretations given to certain verbs. The phenomena which are closely examined are the behaviour of PP sentential modifiers (specifically dative and directional PPs) with respect to the lexical semantic representation of the verb phrases they modify, resultative constructions, and logical metonymy. The analysis is couched in terms of a lexical semantic representation drawing on Davis (1995), Jackendoff (1983, 1990), and Pustejovsky (1991, 1995) which aims to capture “linguistically relevant” components of meaning. The representation is shown to have utility for modeling of the interaction between the syntactic form of an utterance and its meaning. I introduce a formalisation of the representation within the framework of Head Driven Phrase Structure Grammar (Pollard and Sag 1994), and rely on the model of discourse coherence proposed by Lascarides and Asher (1992), Discourse in Commonsense Entailment. I furthermore discuss the implications of the contextual dependency of semantic interpretation for lexicon design and computational processing in Natural Language Understanding systems.

Previous studies show that bilingual infants are slower in developing phonology and tend to experience some difficulties in acquiring some grammatical rules. Furthermore, as compared to their monolingual peers, bilingual infants tend to have less vocabulary. This thesis set out to explore how bilingual infants organise the lexical information in their two languages. Specifically, we examined the lexical-semantic relationships between words within and across languages using a word-to-word priming paradigm. The thesis sought also to uncover any relationship between semantic priming effects and the size of vocabulary. Vocabulary measures such as the BPVS II, the SETK, and the Oxford CDI were used in the experiments, along with an experimental design close to that used in Styles and Plunkett (2009) and Arias-Trejo and Plunkett’s (2009) studies, based on the Intermodal Preferential Looking (IPL) paradigm. The basic design was that the infants were presented with a prime word (e.g. ‘dog’) followed by a target word which was related either semantically to the prime (e.g. ‘cat’) or not (e.g. ‘bus’). Immediately thereafter, we presented two images, depicting the target and a distracter, and monitored the looking times towards the images. In Experiment 1, we tested whether upon hearing related prime and target words, as compared to unrelated pairs, 30 month old monolingual infants preferred to look at the target more than the distracter image. This constituted a benchmark priming effect. In Experiment 2, we examined whether the presence of the target word was necessary for a priming effect to occur. The results demonstrated an effect of semantic priming in the word-word condition (Exp.1) but no semantic priming effect was found in the word-image condition (Exp.2). Experiment 3 investigated, in each of their languages, the semantic priming effect in bilingual 30-month-olds (Arabic-English). The overall result revealed a significant semantic priming effect along with a different pattern in Arabic and English. Experiment 4 was designed to investigate, in a cross-linguistic design, semantic priming in bilingual 18-month-old infants and to address the symmetry between forward (L1-L2) and backward (L2-L1) priming. The overall results showed no semantic priming effect; however, infants showed a non-significant tendency for forward (L1-L2) over backward (L2-L1) priming. As controls for the previous experiments, in Experiments 5 and 6 we examined monolingual and bilingual 18-month-olds to explore whether a priming effect could be obtained only in English. The results showed no semantic priming effect and no strong evidence of a naming effect. All the findings suggested that hearing words activated automatically some other words which, in both monolingual and bilingual infants, were related semantically around 30 months of age, but were not found at 18 months in either population. Despite what is reported often about a delay of language in bilingual children, these findings suggest that, although they show a smaller size of vocabulary in each of their languages when compared to monolingual infants, bilingual infants may build the semantic relationship between words at the same time as their monolingual peers, and with similar word-to-word relations.

Capes, Capes-Cofecub
Esta tese investiga a contribuição semântica dos verbos aspectuais acabar, começar, continuar, deixar, parar, passar, voltar e terminar em português brasileiro. O objetivo geral é contribuir para a análise sobre aspecto e sobre verbos aspectuais em PB e nas línguas naturais em geral, utilizando a Semântica Formal como modelo de análise. Este trabalho defende que os verbos aspectuais são modificadores de eventualidades e por isso contribuem para a expressão do aspecto lexical (Aktionsart), seguindo trabalhos anteriores, como os de Oliveira et al. (2001) e de Laca (2002; 2004; 2005). Ele assume que os verbos aqui estudados se comportam de forma diferente dos verbos de aspecto gramatical como estar e ir, tendo inclusive uma posição diferente na estrutura sintática. Como os verbos aspectuais se relacionam com o aspecto lexical, este trabalho também investiga as propriedades do aspecto lexical presentes no predicado selecionado por um verbo aspectual. Para a seleção dos verbos aspectuais, as propriedades relevantes se encontram no nível do sintagma verbal (VP) e são dadas composicionalmente. Por isso, constituintes como o objeto direto podem alterá-las, permitindo (ou restringindo) a seleção do VP em questão por parte do verbo aspectual. A tese aqui defendida é a de que cada verbo aspectual seleciona seus complementos a partir de propriedades específicas presentes no VP e que estão ligadas à semântica do verbo aspectual em questão. Dessa forma, esta pesquisa defende que as restrições de seleção de cada verbo aspectual podem ser acessadas na entrada lexical do verbo aspectual, sendo possível explicar a seleção de complementos feita por esses verbos. Além disso, defendemos a tese de que a entrada lexical pode apresentar a diferença entre verbos como começar e passar, nas perífrases começar a+infinitivo e passar a +infinitivo, por exemplo.
This thesis investigates the semantic contribution of the following aspectual verbs: começar ‘begin’, continuar ‘continue’, deixar ‘quit’, parar ‘stop’, passar ‘pass’, voltar ‘resume’, and acabar/terminar ‘finish’, in Brazilian Portuguese (BrP). The main goal is to contribute to the discussion about aspect and aspectual verbs in BrP and in natural languages in general, within a Formal Semantics approach. This work treats aspectual verbs as eventuality modifiers, and, consequently, it defends that they contribute to express lexical aspect (Aktionsart) – or situation aspect (Smith 1997) – in accordance with previous works such as Oliveira et al. (2001) and Laca (2002; 2004; 2005). This thesis also proposes that the verbs under discussion behave differently from verbs which contribute to express grammatical aspect – or viewpoint aspect (Smith 1997) – such as estar ‘be’, in the progressive aspect, and ir ‘go’, in the prospective aspect, also assuming that they have different positions in the syntactic structure. Since aspectual verbs are related to lexical aspect, the present research also analyzes lexical aspect features in the predicate which is selected for these verbs. Relevant features for this selection are located in the verbal phrase level (VP) and are given compositionally. Thus, phrases such as the direct object are able to change them, allowing (or not) that an aspectual verb select this VP. This work defends the hypothesis that each aspectual verb selects their complements based on specific features found in the VP, and that these features are linked to the semantics of this aspectual verb. Then, this research defends the idea that the selection constraints of each aspectual verb can be observed in its lexical entry, which explains the selection made by these verbs. Furthermore, we defend the idea that we can verify differences between semantically similar verbs like começar and passar, in their periphrastic construction, from differences found in their lexical entries.

Business Intelligence tools support making better business decisions by analyzing available organizational data. Data Warehouses (DWs), typically structured with the Multidimensional (MD) model, are used to store data from different internal and external sources processed using Extract-Transformation-Load (ETL) processes. On-Line analytical Processing (OLAP) queries are applied on DWs to derive important business-critical knowledge. DW and OLAP technologies perform efficiently when they are applied on data that are static in nature and well organized in structure. Nowadays, Semantic Web technologies and the Linked Data principles inspire organizations to publish their semantic data, which allow machines to understand the meaning of data, using the Resource Description Framework (RDF) model. In addition to traditional (non-semantic) data sources, the incorporation of semantic data sources into a DW raises the additional challenges of schema derivation, semantic heterogeneity, and schema and data management model over traditional ETL tools. Furthermore, most SW data provided by business, academic and governmental organizations include facts and figures, which raise new requirements for BI tools to enable OLAP-like analyses over those semantic (RDF) data. In this thesis, we 1) propose a layer-based ETL framework for handling diverse semantic and non-semantic data sources by addressing the challenges mentioned above, 2) propose a set of high-level ETL constructs for processing semantic data, 3) implement appropriate environments (both programmable and GUI) to facilitate ETL processes and evaluate the proposed solutions. Our ETL framework is a semantic ETL framework because it integrates data semantically. We propose SETL, a unified framework for semantic ETL. The framework is divided into three layers: the Definition Layer, ETL Layer, and Data Warehouse Layer. In the Definition Layer, the semantic DW (SDW) schema, sources, and the mappings among the sources and the target are defined. In the ETL Layer, ETL processes to populate the SDW from sources are designed. The Data Warehouse Layer manages the storage of transformed semantic data. The framework supports the inclusion of semantic (RDF) data in DWs in addition to relational data. It allows users to define an ontology of a DW and annotate it with MD constructs (such as dimensions, cubes, levels, etc.) using the Data Cube for OLAP (QB4OLAP) vocabulary. It supports traditional transformation operations and provides a method to generate semantic data from the source data according to the semantics encoded in the ontology. It also provides a method to connect internal SDW data with external knowledge bases. On top of SETL, we propose SETLCONSTUCT where we define a set of high-level ETL tasks/operations to process semantic data sources. We divide the integration process into two layers: the Definition Layer and Execution Layer. The Definition Layer includes two tasks that allow DW designers to define target (SDW) schemas and the mappings between (intermediate) sources and the (intermediate) target. To create mappings among the sources and target constructs, we provide a mapping vocabulary called S2TMAP. Different from other ETL tools, we propose a new paradigm: we characterize the ETL flow transformations at the Definition Layer instead of independently within each ETL operation (in the Execution Layer). This way, the designer has an overall view of the process, which generates metadata (the mapping file) that the ETL operators will read and parametrize themselves with automatically. In the Execution Layer, we propose a set of high-level ETL operations to process semantic data sources. Finally, we develop a GUI-based semantic BI system SETLBI to define, process, integrate, and query semantic and non-semantic data. In addition to the Definition Layer and the ETL Layer, SETLBI has the OLAP Layer, which provides an interactive interface to enable OLAP analysis over the semantic DW
Les eines d’Intel·ligència Empresarial (BI), conegudes en anglès com Business Intelligence, donen suport a la millora de la presa de decisions empresarials mitjançant l’anàlisi de les dades de l’organització disponibles. Els magatzems de dades, o data warehouse, (DWs), típicament estructurats seguint el model Multidimensional (MD), s’utilitzen per emmagatzemar dades de diferents fonts, tant internes com externes, processades mitjançant processos Extract- Transformation-Load (ETL). Les consultes de processament analític en línia (OLAP) s’apliquen als DW per extraure coneixement crític en l’àmbit empresarial. Els DW i les tecnologies OLAP funcionen de manera eficient quan s’apliquen sobre dades de natura estàtica i ben estructurades. Avui en dia, les tecnologies de la Web Semàntica (SW) i els principis Linked Data (LD) inspiren les organitzacions per publicar les seves dades en formats semàntics, que permeten que les màquines entenguin el significat de les dades, mitjançant el llenguatge de descripció de recursos (RDF). Una de les raons per les quals les dades semàntiques han tingut tant d’èxit és que es poden gestionar i fer que estiguin disponibles per tercers amb poc esforç, i no depenen d’esquemes de dades sofisticats. A més de les fonts de dades tradicionals (no semàntiques), la incorporació de fonts de dades semàntiques en un DW planteja reptes addicionals tals com derivar-hi esquema, l’heterogeneïtat semàntica i la representació de l’esquema i les dades a través d’eines d’ETL. A més, la majoria de dades SW proporcionades per empreses, organitzacions acadèmiques o governamentals inclouen fets i figures que representen nous reptes per les eines de BI per tal d’habilitar l’anàlisi OLAP sobre dades semàntiques (RDF). En aquesta tesi, 1) proposem un marc ETL basat en capes per a la gestió de diverses fonts de dades semàntiques i no semàntiques i adreçant els reptes esmentats anteriorment, 2) proposem un conjunt d’operacions ETL per processar dades semàntiques, i 3) la creació d’entorns apropiats de desenvolupament (programàtics i GUIs) per facilitar la creació i gestió de DW i processos ETL semàntics, així com avaluar les solucions proposades. El nostre marc ETL és un marc ETL semàntic perquè Es capaç de considerar e integrar dades de forma semàntica. Els següents paràgrafs elaboren sobre aquests contribucions. Proposem SETL, un marc unificat per a ETL semàntic. El marc es divideix en tres capes: la capa de definició, la capa ETL i la capa DW. A la capa de definició, es defineixen l’esquema del DW semàntic (SDW), les fonts i els mappings entre les fonts i l’esquema del DW. A la capa ETL, es dissenyen processos ETL per popular el SDW a partir de fonts. A la capa DW, es gestiona l’emmagatzematge de les dades semàntiques transformades. El nostre marc dóna suport a la inclusió de dades semàntiques (RDF) en DWs, a més de dades relacionals. Així, permet als usuaris definir una ontologia d’un DW i anotar-la amb construccions MD (com ara dimensions, cubs, nivells, etc.) utilitzant el vocabulari Data Cube for OLAP (QB4OLAP). També admet operacions de transformació tradicionals i proporciona un mètode per generar semàntica de les dades d’origen segons la semàntica codificada al document ontologia. També proporciona un mètode per connectar l’SDW amb bases de coneixement externes. Per tant, crea una base de coneixement, composta per un ontologia i les seves instàncies, on les dades estan connectades semànticament amb altres dades externes / internes. Per fer-ho, desenvolupem un mètode programàtic, basat en Python, d’alt nivell, per realitzar les tasques esmentades anteriorment. S’ha portat a terme un experiment complet d’avaluació comparant SETL amb una solució elaborada amb eines tradicional (que requereixen molta més codificació). Com a cas d’ús, hem emprat el Danish Agricultural dataset, i els resultats mostren que SETL proporciona un millor rendiment, millora la productivitat del programador i la qualitat de la base de coneixement. La comparació entre SETL i Pentaho Data Integration (PDI) mostra que SETL és un 13,5% més ràpid que PDI. A més de ser més ràpid que PDI, tracta les dades semàntiques com a ciutadans de primera classe, mentre que PDI no conté operadors específics per a dades semàntiques. A sobre de SETL, proposem SETLCONSTUCT on definim un conjunt de tasques d’alt nivell / operacions ETL per processar fonts de dades semàntiques i orientades a encapsular i facilitar la creació de l’ETL semàntic. Dividim el procés d’integració en dues capes: la capa de definició i la capa d’execució. La capa de definició inclou dues tasques que permeten definir als dissenyadors de DW esquemes destí (SDW) i mappings entre fonts (o resultats intermedis) i l’SDW (potencialment, altres resultats intermedis). Per crear mappings entre les fonts i el SDW, proporcionem un vocabulari de mapping anomenat Source-To-Target Mapping (S2TMAP). A diferència d’altres eines ETL, proposem un nou paradigma: les transformacions del flux ETL es caracteritzen a la capa de definició, i no de forma independent dins de cada operació ETL (a la capa d’execució). Aquest nou paradigma permet al dissenyador tenir una visió global del procés, que genera metadades (el fitxer de mapping) que els operadors ETL individuals llegiran i es parametritzaran automàticament. A la capa d’execució proposem un conjunt d’operacions ETL d’alt nivell per processar fonts de dades semàntiques. A més de la neteja, la unió i la transformació per dades semàntiques, proposem operacions per generar semàntica multidimensional i actualitzar el SDW per reflectir els canvis en les fonts. A més, ampliem SETLCONSTRUCT per permetre la generació automàtica de flux d’execució ETL (l’anomenem SETLAUTO). Finalment, proporcionem una àmplia avaluació per comparar la productivitat, el temps de desenvolupament i el rendiment de SETLCONSTRUCT i SETLAUTO amb el marc anterior SETL. L’avaluació demostra que SETLCONSTRUCT millora considerablement sobre SETL en termes de productivitat, temps de desenvolupament i rendiment. L’avaluació mostra que 1) SETLCONSTRUCT utilitza un 92% menys de caràcters mecanografiats (NOTC) que SETL, i SETLAUTO redueix encara més el nombre de conceptes usats (NOUC) un altre 25%; 2) utilitzant SETLCONSTRUCT, el temps de desenvolupament es redueix gairebé a la meitat en comparació amb SETL, i es redueix un altre 27 % mitjançant SETLAUTO; 3) SETLCONSTRUCT es escalable i té un rendiment similar en comparació amb SETL. Finalment, desenvolupem un sistema de BI semàntic basat en GUI SETLBI per definir, processar, integrar i consultar dades semàntiques i no semàntiques. A més de la capa de definició i de la capa ETL, SETLBI té una capa OLAP, que proporciona una interfície interactiva per permetre l’anàlisi OLAP d’autoservei sobre el DW semàntic. Cada capa està composada per un conjunt d’operacions / tasques. Per formalitzar les connexions intra i inter-capes dels components de cada capa, emprem una ontologia. La capa ETL amplia l’execució de la capa de SETLCONSTUCT afegint operacions per processar fonts de dades no semàntiques. Per últim, demostrem el sistema final mitjançant el cens de la població de Bangladesh (2011). La solució final d’aquesta tesi és l’eina SETLBI . SETLBI facilita (1) als dissenyadors del DW amb pocs / sense coneixements de SW, integrar semànticament les dades (semàntiques o no) i analitzar-les emprant OLAP, i (2) als usuaris de la SW els permet definir vistes sobre dades semàntiques, integrar-les amb fonts no semàntiques, i visualitzar-les segons el model MD i fer anàlisi OLAP. A més, els usuaris SW poden enriquir l’esquema SDW generat amb construccions RDFS / OWL. Prenent aquest marc com a punt de partida, els investigadors poden emprar-lo per a crear SDWs de forma interactiva i automàtica. Aquest projecte crea un pont entre les tecnologies BI i SW, i obre la porta a altres oportunitats de recerca com desenvolupar tècniques de DW i ETL comprensibles per les màquines.
(Danskere) Business Intelligence (BI) værktøjer understøtter at tage bedre forretningsbeslutninger, ved at analysere tilgængelige organisatoriske data. Data Warehouses (DWs), typisk konstrueret med den Multidimensionelle (MD) model, bruges til at lagre data fra forskellige interne og eksterne kilder, der behandles ved hjælp af Extract-Transformation-Load (ETL) processer. On-Line Analytical Processing (OLAP) forespørgsler anvendes på DWs for at udlede vigtig forretningskritisk viden. DW og OLAP-teknologier fungerer effektivt, når de anvendes på data, som er statiske af natur og velorganiseret i struktur. I dag inspirerer Semantic Web (SW) teknologier og Linked Data (LD) principper organisationer til at offentliggøre deres semantiske data, som tillader maskiner at forstå betydningen af denne, ved hjælp af Resource Description Framework (RDF) modellen. En af grundene til, at semantiske data er blevet succesfuldt, er at styringen og udgivelsen af af dataene er nemt, og ikke er afhængigt af et sofistikeret skema. Ud over problemer ved overførslen af traditionelle (ikke-semantiske) databaser til DWs, opstår yderligere udfordringer ved overførslen af semantiske databaser, såsom skema nedarvning, semantisk heterogenitet samt skemaet for data repræsentation over traditionelle ETL værktøjer. På den anden side udgør en stor del af den semantiske data der bliver offentliggjort af virksomheder, akademikere samt regeringer, af figurer og fakta, der igen giver nye problemstillinger og krav til BI værktøjer, for at gøre OLAP lignende analyser over de semantiske data mulige. I denne afhandling gør vi følgende: 1) foreslår et lag-baseret ETL framework til at håndterer multiple semantiske og ikke-semantiske datakilder, ved at svare på udfordringerne nævnt herover, 2) foreslår en mængde af ETL operationer til at behandle semantisk data, 3) implementerer passende miljøer (både programmerbare samt grafiske brugergrænseflader), for at lette ETL processer og evaluere den foreslåede løsning. Vores ETL framework er et semantisk ETL framework, fordi det integrerer data semantisk. Den følgende sektion forklarer vores bidrag. Vi foreslår SETL, et samlet framework for semantisk ETL. Frameworket er splittet i tre lag: et definitions-lag, et ETL-lag, og et DW-lag. Det semanvii tiske DW (SWD) skema, datakilder, samt sammenhængen mellem datakilder og deres mål, er defineret i definitions-laget. I ETL-laget designes ETLprocesser til at udfylde SDW fra datakilderne. DW-laget administrerer lagring af transformerede semantiske data. Frameworket understøtter inkluderingen af semantiske (RDF) data i DWs ud over relationelle data. Det giver brugerne mulighed for at definere en ontologi for et DW og annotere med MD-konstruktioner (såsom dimensioner, kuber, niveauer osv.) ved hjælp af Data Cube til OLAP (QB4OLAP) ordforrådet. Det understøtter traditionelle transformations operationer, og giver en metode til at generere semantiske data fra de oprindelige data, i henhold til semantikken indkodet i ontologien. Det muliggør også en metode til at forbinde interne SDW data med eksterne vidensbaser. Herved skaber det en vidensbase, der er sammensat af en ontologi og dets instanser, hvor data er semantisk forbundet med andre eksterne / interne data. Vi udvikler et høj niveau Python-baseret programmerbart framework for at udføre de ovennævnte opgaver. En omfattende eksperimentel evaluering, der sammenligner SETL med en traditionel løsning (hvilket krævede meget manuel kodning), om brugen af danske landbrugsog forretnings datasæt, viser at SETL præsterer bedre, programmør produktivitet og vidensbase kvalitet. Sammenligningen mellem SETL og Pentaho Data Integration (PDI) ved behandling af en semantisk kilde viser, at SETL er 13,5% hurtigere end PDI. Udover SETL, foreslår vi SETLCONSTRUCT hvor vi definerer et sæt ETLoperationer på højt niveau til behandling af semantiske datakilder. Vi deler integrationsprocessen i to lag: Definitions-lag og eksekverings-lag. Definitionslaget indeholder to opgaver, der giver DW designere muligheden for at definere (SDW) skemaer, og kortlægningerne mellem kilder og målet. For at oprette kortlægning mellem kilderne og målene, leverer vi et kortlægnings ordforråd kaldet Source-to-Target Mapping (S2TMAP). Forskelligt fra andre ETL-værktøjer foreslår vi et nyt paradigme: vi karakteriserer ETLflowtransformationerne i definitions-laget i stedet for uafhængigt inden for hver ETL-operation (i eksekverings-laget). På denne måde har designeren et overblik over processen, som genererer metadata (kortlægningsfilen), som ETL operatørerne vil læse og parametrisere automatisk. I eksekverings-laget foreslår vi en mængde høj niveau ETL-operationer til at behandle semantiske datakilder. Udover rensning, sammenføjning og datatypebaseret transformationer af semantiske data, foreslår vi operationer til at generere multidimensionel semantik på data-niveau og operationer til at opdatere et SDW for at afspejle ændringer i kilde-dataen. Derudover udvider vi SETLCONSTRUCT for at muliggøre automatisk ETL-eksekveringsstrømgenerering (vi kalder det SETLAUTO). Endelig leverer vi en omfattende evaluering for at sammenligne produktivitet, udviklingstid og ydeevne for scon og SETLAUTO med den tidligere ramme SETL. Evalueringen viser, at SETLCONSTRUCT forbedres markant i forhold til SETL med hensyn til produktivitet, udviklingstid og ydeevne. Evalueringen viser, at 1) SETLCONSTRUCT bruger 92% færre antal indtastede tegn (NOTC) end SETL, og SETLAUTO reducerer antallet af brugte begreber (NOUC) yderligere med 25%; 2) ved at bruge SETLCONSTRUCT, er udviklingstiden næsten halveret sammenlignet med SETL, og skæres med yderligere 27% ved hjælp af SETLAUTO; 3) SETLCONSTRUCT er skalerbar og har lignende ydelse sammenlignet med SETL. Til slut udvikler vi et GUI-baseret semantisk BI system SETLBI for at definere, processere, integrere og lave forespørgsler på semantiske og ikkesemantiske data. Ud over definitions-laget og ETL-laget, har SETLBI et OLAP-lag, som giver en interaktiv grænseflade for at muliggøre selvbetjenings OLAP analyser over det semantiske DW. Hvert lag er sammensat af en mængde operationer/opgaver. Vi udarbejder en ontologi til at formalisere intra-og ekstra-lags forbindelserne mellem komponenterne og lagene. ETLlaget udvider eksekverings-laget af SETLCONSTUCT ved at tilføje operationer til at behandle ikke-semantiske datakilder. Vi demonstrerer systemet ved hjælp af Bangladesh population census 2011 datasættet. Sammenfatningen af denne afhandling er BI-værktøjet SETLBI . SETLBI fremmer (1) DW-designere med ringe / ingen SW-viden til semantisk at integrere semantiske og / eller ikke-semantiske data og analysere det i OLAP stil, og (2) SW brugere med grundlæggende MD-baggrund til at definere MDvisninger over semantiske data, der aktiverer OLAP-lignende analyse. Derudover kan SW-brugere berige det genererede SDW-skema med RDFS / OWLkonstruktioner. Med udgangspunkt i frameworket som et grundlag kan forskere sigte mod at udvikle yderligere interaktive og automatiske integrationsrammer for SDW. Dette projekt bygger bro mellem de traditionelle BIteknologier og SW-teknologier, som igen vil åbne døren for yderligere forskningsmuligheder som at udvikle maskinforståelige ETL og lagerteknikker.

Finnish belongs into the Finno-Ugric language family, and it is spoken by the vast majority of the people living in Finland. The motivation for this thesis is to contribute to the development of a semantic tagger for Finnish. This tool is a parallel of the English Semantic Tagger which has been developed at the University Centre for Computer Corpus Research on Language (UCREL) at Lancaster University since the beginning of the 1990s and which has over the years proven to be a very powerful tool in automatic semantic analysis of English spoken and written data. The English Semantic Tagger has various successful applications in the fields of natural language processing and corpus linguistics, and new application areas emerge all the time. The semantic lexical resources that I have created in this thesis provide the knowledge base for the Finnish Semantic Tagger. My main contributions are the lexical resources themselves, along with a set of methods and guidelines for their creation and expansion as a general language resource and as tailored for domain-specific applications. Furthermore, I propose and carry out several methods for evaluating semantic lexical resources. In addition to the English Semantic Tagger, which was developed first, and the Finnish Semantic Tagger second, equivalent semantic taggers have now been developed for Czech, Chinese, Dutch, French, Italian, Malay, Portuguese, Russian, Spanish, Urdu, and Welsh. All these semantic taggers taken together form a program framework called the UCREL Semantic Analysis System (USAS) which enables the development of not only monolingual but also various types of multilingual applications. Large-scale semantic lexical resources designed for Finnish using semantic fields as the organizing principle have not been attempted previously. Thus, the Finnish semantic lexicons created in this thesis are a unique and novel resource. The lexical coverage on the test corpora containing general modern standard Finnish, which has been the focus of the lexicon development, ranges from 94.58% to 97.91%. However, the results are also very promising in the analysis of domain-specific text (95.36%), older Finnish text (92.11–93.05%), and Internet discussions (91.97–94.14%). The results of the evaluation of lexical coverage are comparable to the results obtained with the English equivalents and thus indicate that the Finnish semantic lexical resources indeed cover the majority of core Finnish vocabulary.

Thesis (M. A. (African Languages)) --University of Limpopo, 2001
This study explores the lexical semantic representation of break verbs in Xitsonga. Chapter One is the introduction of this study. It describes the statement of the problem. The main aim of this study is described which m bnef is to investigate the form, struc e and interpretation of break verbs in Xitsonga. The significance of this study, which is to highlight the semantic value of break verbs in Xitsonga is discussed. The methodology, lite rature review as well as the theta - theory and its properties have been examined. Chapter Two explores the predicate argument structure. The difference between two lexical representations, which are lexical-syntactic and lexical­ semantic representations was investigated. A brief definition of break verbs as well as the six categories of the verbs of change of state have also been discussed. This chapter also analyses the lexical features of break verbs. Various sub - classes of external arguments and of internal argument are observed. Chapter Three presents the lexical - semantic representation of break verbs where focus is on argument structure, event structure, qualia structure and lexical inheritance structure. Chapter Four deals with the syntactic alternation and selection restriction of break verbs. The difference between transitive and intransitive alterna tions was also highlighted. This chapter also explores some of the different types of verbal alternations, such as ' instrument subject alternation', ' locative alternation' and the ' with/ against alternation' . Chapter Five gives the main conclusion of this study.

In this work, we test two novel methods of using word embeddings to detect lexical semantic change, attempting to overcome limitations associated with conventional approaches to this problem. Using a diachronic corpus spanning over a hundred years, we generate word embeddings for each decade with the intention of evaluating how meaning changes are represented in embeddings for the same word across time. Our approach differs from previous works in this field in that we encode words as probabilistic Gaussian distributions and bimodal probabilistic Gaussian mixtures, rather than conventional word vectors. We provide a discussion and analysis of our results, comparing the approaches we implemented with those used in previous works. We also conducted further analysis on whether additional information regarding the nature of semantic change could be discerned from particular qualities of the embeddings we generated for our experiments. In our results, we find that encoding words as probabilistic Gaussian embeddings can provide an enhanced degree of reliability with regard to detecting lexical semantic change. Furthermore, we are able to represent additional information regarding the nature of such changes through the variance of these embeddings. Encoding words as bimodal Gaussian mixtures however is generally unsuccessful for this task, proving to be not reliable enough at distinguishing between discrete senses to effectively detect and measure such changes. We provide potential explanations for the results we observe, and propose improvements that can be made to our approach to potentially improve performance.

Current models of language processing advocate that word meaning is partially stored in distributed modality-specific cortical networks. However, while much has been done to investigate where information is represented in the brain, the neuronal dynamics underlying how these networks communicate internally, and with each other are still poorly understood. For example, it is not clear how spatially distributed semantic content is integrated into a coherent conceptual representation. The current thesis investigates how perceptual semantic features are selected and integrated, using oscillatory neuronal dynamics. Cortical oscillations reflect synchronized activity in large neuronal populations that are associated with specific classes of network interactions. The first part of the present thesis addresses how perceptual semantic features are selected in long-term memory. Using electroencephalographic (EEG) recordings, it is demonstrated that retrieving perceptually more complex information is associated with a reduction in oscillatory power, which is in line with the information via desynchronization hypothesis, a recent neurophysiological model for long-term memory retrieval. The second, and third part address how distributed semantic content is integrated and coordinated in the brain. Behavioral evidence suggests that integrating two features of a target word (e.g., Whistle) during a dual property verification task, incurs an additional processing cost if features are from different (visual: tiny, audio: loud), rather than the same modality (visual: tiny, silver). Furthermore, EEG recordings reveal that integrating cross-modal feature pairs is associated with a more sustained low-frequency theta power increase in the left anterior temporal lobe (ATL). The ATL is thought to converge semantic content from different modalities. In line with this notion, ATL is shown to communicate with a widely distributed cortical network at the theta frequency. The fourth part of the thesis uses magnetoencephalographic (MEG) recordings to show that, while low frequency theta oscillations in left ATL are more sensitive to integrating features from different modalities, integrating two features from the same modality induces an early increase in high frequency gamma power in left ATL and modality-specific regions. These results are in line with a recent framework suggesting that local, and long-range network dynamics are reflected in different oscillatory frequencies. The fifth part demonstrates that the connection weights between left ATL and modality-specific regions at the theta frequency are modulated consistently with the content of the word (e.g., visual features enhance connectivity between left ATL and left inferior occipital cortex). The thesis concludes by embedding these results in the context of current neurocognitive models of semantic processing.

In this thesis I investigate the syntax and semantics of tense-aspect stem participles in the Ṛgveda, focusing primarily on the data from the earlier books II–VII and IX, seeking to establish a comprehensive and coherent analysis of this category within the linguistic system of Ṛgvedic Sanskrit. In recent literature tense-aspect stem participles are usually treated as semantically equivalent to finite verbs wherever possible, but contradictorily where they differ from finite verbs their adjectival nature is emphasized. I argue that tense-aspect stem participles are a fundamentally verbal formation and can be treated as inflectional verb forms: they are adjectival verbs rather than verbal adjectives. At the same time, however, they constitute an independent sub-category of verb form which is not necessarily semantically dependent on corresponding finite stems. I examine the syntactic and semantic properties of tense-aspect stem participles both in relation to finite verbal forms and their wider syntactic context, formalizing the evidence in the framework of Lexical-Functional Grammar. Consequently I am able to categorize the syntactic and semantic deviations which many participles exhibit in comparison to finite verbal forms. I contend that many such forms cannot be treated synchronically (and sometimes diachronically) as participles, but form distinct synchronic categories. My analysis permits a considerably more refined definition of the category of tense-aspect stem participles, dependent on clear morphological, syntactic and semantic criteria, as opposed to the usual, purely morphological, definition. From a diachronic perspective I argue that the category of tense-aspect stem participles as found in the Ṛgveda more closely reflects an inherited Proto-Indo-European category of tense-aspect stem participles than is usually assumed. I also reconsider theoretical treatments of participial syntax and semantics, and develop a more precise typology of non-finite verb systems which adequately accounts for Sanskrit participles.

Classic studies on ambiguous words (e.g. ‘bark’ – dog/tree) imply that comprehenders’ lexical-semantic representations remain relatively stable across time. However, recent research has shown that a single encounter with a particular word-meaning biases interpretation up to 20 minutes later (“word-meaning priming”), suggesting that representations update to reflect recent experience. Nine experiments in this thesis investigate in detail the effects of recent experience on the comprehension of ambiguous words. Using word association, Chapter 2 replicates the single-encounter subordinate priming effect and shows that this effect is reduced by a subsequent dominant meaning encounter. Three recent subordinate encounters boost priming compared to a single encounter but only when encounters are temporally spaced; massed encounters seem to provide no such boost. Chapter 3 assesses a newly-developed semantic relatedness test of word-meaning availability effects on comprehension, using picture probes. It shows that, compared to word association, semantic relatedness can detect dominance with the additional benefit of testing dominant and subordinate meaning availabilities independently. Chapter 4 shows that this semantic relatedness test can detect single-encounter word-meaning priming and that this effect is driven by increased availability of the primed meaning, not decreased availability of the unprimed meaning. Furthermore, an additional priming boost from three repetitions reflects an increase in primed meaning availability for both massed and spaced repetitions, with an additional decrease in unprimed meaning availability after spaced repetitions only; there was no evidence that massed repetitions reduced unprimed availability. Possible mechanisms are discussed that account for these different repetition priming patterns observed with semantic relatedness and word association tests. The findings suggest that the word-meaning priming effect might be driven by episodic memory and consolidated lexical-semantic representations. Taken together, these experiments confirm that recent experience plays a key role in retuning lexical-semantic representations and can help to refine our theoretical accounts of this important phenomenon.

In this work, we explore the usefulness of contextualized embeddings from language models on lexical semantic change (LSC) detection. With diachronic corpora spanning two time periods, we construct word embeddings for a selected set of target words, aiming at detecting potential LSC of each target word across time. We explore different systems of embeddings to cover three topics: contextualized vs static word embeddings, token- vs type-based embeddings, and multilingual vs monolingual language models. We use a multilingual dataset covering three languages (English, German, Swedish) and explore each system of embedding with two subtasks, a binary classification task and a ranking task. We compare the performance of different systems of embeddings, and seek to answer our research questions through discussion and analysis of experimental results. We show that contextualized word embeddings are on par with static word embeddings in the classification task. Our results also show that it is more beneficial to use the contextualized embeddings from a multilingual model than from a language specific model in most cases. We present that token-based setting is strong for static embeddings, and type-based setting for contextual embeddings, especially for the ranking task. We provide some explanation for the results we achieve, and propose improvements that can be made to our experiments for future work.

Measures of semantic distance have received a great deal of attention recently in the field of computational lexical semantics. Although techniques for approximating the semantic distance of two concepts have existed for several decades, the introduction of the WordNet lexical database and improvements in corpus analysis have enabled significant improvements in semantic distance measures.

In this study we investigate a special kind of semantic distance, called semantic relatedness. Lexical semantic relatedness measures have proved to be useful for a number of applications, such as word sense disambiguation and real-word spelling error correction. Most relatedness measures rely on the observation that the shortest path between nodes in a semantic network provides a representation of the relationship between two concepts. The strength of relatedness is computed in terms of this path.

This dissertation makes several significant contributions to the study of semantic relatedness. We describe a new measure that calculates semantic relatedness as a function of the shortest path in a semantic network. The proposed measure achieves better results than other standard measures and yet is much simpler than previous models. The proposed measure is shown to achieve a correlation of r = 0. 897 with the judgments of human test subjects using a standard benchmark data set, representing the best performance reported in the literature. We also provide a general formal description for a class of semantic distance measures — namely, those measures that compute semantic distance from the shortest path in a semantic network. Lastly, we suggest a new methodology for developing path-based semantic distance measures that would limit the possibility of unnecessary complexity in future measures.

CITE/Language Arts
Ed.D.
In a series of studies using semantic relatedness judgment response times, Jiang (2000, 2002, 2004a) has claimed that L2 lexical entries fossilize with their equivalent L1 content or something very close to it. In another study using a more productive test of lexical knowledge (Jiang 2004b), however, the evidence for this conclusion was less clear. The present study is a partial replication of Jiang (2004b) with Japanese learners of English. The aims of the study are to investigate the influence of the first language (L1) on second language (L2) lexical knowledge, to investigate whether lexical knowledge displays frequency-related, emergent properties, and to investigate the influence of the L1 on the acquisition of L2 word pairs that have a common L1 equivalent. Data from a sentence completion task was completed by 244 participants, who were shown sentence contexts in which they chose between L2 word pairs sharing a common equivalent in the students' first language, Japanese. The data were analyzed using the statistical analyses available in the programming environment R to quantify the participants' ability to discriminate between synonymous and non-synonymous use of these L2 word pairs. The results showed a strong bias against synonymy for all word pairs; the participants tended to make a distinction between the two synonymous items by assigning each word a distinct meaning. With the non-synonymous items, lemma frequency was closely related to the participants' success in choosing the correct word in the word pair. In addition, lemma frequency and the degree of similarity between the words in the word pair were closely related to the participants' overall knowledge of the non-synonymous meanings of the vocabulary items. The results suggest that the participants had a stronger preference for non-synonymous options than for the synonymous option. This suggests that the learners might have adopted a one-word, one-meaning learning strategy (Willis, 1998). The reasonably strong relationship between several of the usage-based statistics and the item measures from R suggest that with exposure learners are better able to use words in ways that are similar to native speakers of English, to differentiate between appropriate and inappropriate contexts and to recognize the boundary separating semantic overlap and semantic uniqueness. Lexical similarity appears to play a secondary role, in combination with frequency, in learners' ability to differentiate between appropriate and inappropriate contexts when using L2 word pairs that have a single translation in the L1.
Temple University--Theses

This thesis examines the various approaches which have been taken to investigate the concrete/abstract word distinction both in normal subjects and in patients who, as a result of brain damage, have an impairment of lexical semantic representations. The nature of the definition task as a tool for assessing the semantic representations of concrete and abstract terms was examined. It was found that definitions for abstract words differed from those of concrete words only in style, not in semantic content. The metalinguistic demands of the definition task therefore make it inappropriate for assessing the semantic representations of concrete and abstract terms in patients with any form of language impairment. The performance of four patients with semantic impairments was examined using a variety of tasks designed to assess concrete and abstract word comprehension. While some of the data can be accommodated within the framework of several theories, no single theory can adequately account for the patterns of performance in all four patients. An alternative model of semantic memory is therefore proposed in which concreteness and frequency interact at the semantic level. Jones' Ease of Predication Hypothesis, which states that the difference between concrete and abstract terms can be explained in terms of disproportionate numbers of underlying semantic features (or "predicates") was also investigated. It was found that the ease of predication variable does not accurately reflect either predicate or feature distributions, and is simply another index of concreteness. As such, the validity of this concept as the basis of theories of semantic representation should be questioned. Models based on the assumption of a "richer" semantic representation for concrete words (e.g.: Plaut & Shallice, 1993) are therefore undermined by these data. The possibility that concrete and abstract concepts can be accessed from their most salient predicates and/or features was examined in a series of semantic priming experiments. It was concluded that it is not possible to prime either concrete or abstract concepts from their constituent parts. Significant facilitation only occurred for items in which the prime and target were synonymous and therefore map onto concepts which share almost identical semantic representations. In summary, it is apparent that no current theory of semantic representation can adequately account for the range of findings with regard to the concrete/abstract word distinction. The most plausible account is some form of distributed connectionist model. However, such models are based on unsubstantiated assumptions about the nature of abstract word representations in the semantic network. Alternative proposals are therefore discussed.

Thesis (B.Sc)--University of Hong Kong, 2002.
"A dissertation submitted in partial fulfilment of the requirements for the Bachelor of Science (Speech and Hearing Sciences), The University of Hong Kong, May 10, 2002." Also available in print.

Thesis (Ph. D.)--University of Oregon, 2004.
Typescript. Includes vita and abstract. Includes bibliographical references (leaves 432-460). Also available for download via the World Wide Web; free to University of Oregon users.

Zugl.: Heidelberg, Univ., Diss., 2008 (leicht abgeänd.).
Originaltitel: Translation-based empirical analysis in semantic domains, Titel der Diss. Originaltitel: Translation-based empirical analysis in semantic domains.

Lexical-semantic relationships between words are key information for many NLP tasks, which require this knowledge in the form of lexical resources. This thesis addresses the acquisition of lexical-semantic relation instances. State of the art systems rely on word pair representations based on patterns of contexts where two related words co-occur to detect their relation. This approach is hindered by data sparsity: even when mining very large corpora, not every semantically related word pair co-occurs or not frequently enough. In this work, we investigate novel representations to predict if two words hold a lexical-semantic relation. Our intuition was that these representations should contain information about word co-occurrences combined with information about the meaning of words involved in the relation. These two sources of information have to be the basis of a generalization strategy to be able to provide information even for words that do not co-occur.
Les relacions lexicosemàntiques entre paraules són una informació clau per a moltes tasques del PLN, què requereixen aquest coneixement en forma de recursos lingüístics. Aquesta tesi tracta l’adquisició d'instàncies lexicosemàntiques. Els sistemes actuals utilitzen representacions basades en patrons dels contextos en què dues paraules coocorren per detectar la relació que s'hi estableix. Aquest enfocament s'enfronta a problemes de falta d’informació: fins i tot en el cas de treballar amb corpus de grans dimensions, hi haurà parells de paraules relacionades que no coocorreran, o no ho faran amb la freqüència necessària. Per tant, el nostre objectiu principal ha estat proposar noves representacions per predir si dues paraules estableixen una relació lexicosemàntica. La intuïció era que aquestes representacions noves havien de contenir informació sobre patrons dels contextos, combinada amb informació sobre el significat de les paraules implicades en la relació. Aquestes dues fonts d'informació havien de ser la base d'una estratègia de generalització que oferís informació fins i tot quan les dues paraules no coocorrien.

The objective of this dissertation was to characterise the relative contribution of genetic influences to individual differences in cross sectional performance and decline of semantic-lexical abilities and to investigate whether these linguistic effects indicating semantic degradation are sensitive indicators of medial temporal atrophy in early Alzheimer's disease and in patients with mild cognitive impairment of amnestic type (aMCI). The effect of ApoE status in the genetic profile of these groups on deterioration of semantic abilities was studied to verify whether there was any relationship between variation in lexical factors and genetic variability. Oral generation of words belonging to two categories (animal and fruits) during a fluency tasks was required. In AD patients there was an effect of genotype but, although strong, this was diluted by the advanced cognitive deterioration and could only be seen as a tendency to be stronger in ε4 carriers. The words produced by the aMCI carriers were significantly earlier acquired than those of non-carriers and controls. These behavioural findings confirmed evidence from other recent studies and showed that a significant proportion of phenotype variability in performance on fluency tasks was influenced by genetic factors. Impairments in semantic tasks in the ε4 allele carrier population might indicate either that individuals who will develop AD never fully develop semantic skills, or that the neuroanatomical substrate of semantic abilities is selective sensitive to the earliest effects of the AD neuropathology. On the basis of this result it seemed reasonable to hypothesise that the presence of the “semantic endophenotype” in people carriyng the ApoE vulnerability mutation might be associated with atrophy in areas early affected by neuropathology due to AD and involved in semantic memory retrieval. Using lexical semantic competency in aMCI carriers as an endophenotype, grey matter volume loss in aMCI ε4 carriers/non-carriers and in controls was compared and the residual volume correlated with allele burden and with age of acquisition values for words produced in a category fluency task. Direct group comparisons showed that carriers had grey matter volume loss which was generally confined to limbic regions and medial temporal structures, and non-carriers had greater atrophy in temporal and parieto-occipital neocortex. aMCI subjects had significantly impoverished lexical semantic output compared to controls, more marked in aMCI carriers. A voxel based correlation analysis showed that greater volume loss in parahippocampal gyrus and thalamus was associated with a tendency to retrieve earlier acquired words in the category fluency task. The results suggest a relatively specific impact of ApoE 4 burden and underline the value of linguistic assessment in preclinical diagnosis. The detrimental role of this mutation found in aMCI individuals was also assessed at the larger stage in the disease process by direct comparisons in minimal to mild AD ε4 carriers/non-carrier patients. VBM comparison analysis confirmed the observation done in the genetically determined aMCI subgroups. AD ε4- carriers showed greater atrophy in mediotemporal structures compared to non-carriers whose grey matter volume loss was more widespread in more neocortical areas. Finally, an age, gender and education based norms for AoA, Typicality and Familiarity was built up in order to create a valid psychometric instrument able to detect and monitor subtle semantic deficits in ApoE ε4 carriers over time.