Title Promoter Affiliations Abstract "Variation and change in constructional semantics" "Timothy Colleman" "Department of Linguistics" "The project is situated at the intersection ofconstruction grammar, historical linguistics, and variationistsociolinguistics On the basis of extensive qualitative and quantitative corpusresearch into the semantic properties of Dutch argument structureconstructions across regional, stylistic, and diachronic varieties of thelanguage, the project aims to identify and analyze patterns of intralingualvariation and change in constructional semantics." "Sabbatical Dirk Speelman: Distributional semantics meets Digital Humanities" "Dirk Speelman" "Quantitative Lexicology and Variational Linguistics (QLVL), Leuven" "Recent methodological advances in corpus-based distributional semantics have made it possible to study semantics in new ways and, above all, on a scale that was previously unimaginable, both within earlier theoretical/descriptive and more applied research. Whereas extensive empirical linguistic research previously had to make do with the study of semantics with a rather limited set of instruments, certainly compared to what was available to study the formal side of language, and the study of semantics was forced to remain manual and small-scale, now a long-awaited catch-up has finally been made and semantics could finally gain a full place in large-scale empirical linguistic research. Having devoted myself in the last decade to deepening, refining and applying this new approach to primarily lexicological research — a logical starting point — the time has come to broaden and explore its potential in other domains. In this sabbatical period focused on resourcing, study and exploration, I want to thoroughly explore and prepare broadening in three possible directions. In all three cases, these are domains in which advanced corpus-based methods are already used intensively, but for the time being mainly for the study of more formal aspects of language, without (or only rarely with) incorporation of scalable operationalizations of semantics, an addition that can nevertheless represent an important enrichment. The first domain is the corpus-based study of the use of constructions (in a language-based theoretical framework). The second domain is that of applied textual analysis in a Digital Humanities context (with a view to the study of concepts in literary and historical texts). The third domain is that of the analysis of communication and language use in social media (with a view to opinion mining and the study of the dissemination of information, disinformation and misinformation)." "BITSHARE: Bitstring Semantics for Human and Artificial Reasoning" "Hans Smessaert" "Formal and Computational Linguistics (ComForT), Leuven, Brain and Cognition, Declarative Languages and Artificial Intelligence (DTAI), Centre for Logic and Philosophy of Science" "Bitstring semantics, which was recently developed in logic, is a formalism for exploring the meaning relations between different expressions. Because it operates in a thoroughly local fashion, this formalism is able to achieve a good balance between high expressivity and low mathematical/computational complexity. The overarching goal of the BITSHARE project is to explore new applications of bitstring semantics in other disciplines that are concerned with meaning and reasoning, viz., philosophy, linguistics, psychology and computer science. Our target applications include some of the most important research topics in these disciplines today, such as scalar implicatures, the semantics/pragmatics interface, collaborative knowledge extraction, and probabilistic causal models. Through these well-chosen target applications, the project will demonstrate that bitstring semantics offers a shared representation formalism for a broad, interdisciplinary community of reasoning researchers, in which the insights from different disciplines can be expressed and combined with each other." "Modality and provability from the perspective of non-deterministic semantics." "Joke Meheus" "Department of Philosophy and moral sciences" "In this project, we will use non-deterministic semantics to study various modal notions. We will be particularly interested in logics that have a coherent provability reading, such as informal provability, intuitionistic provability, and various other logics validating axiom T. The main interest is the relations between the perspectives offered by non-deterministic semantics and by Kripke's semantics, as well as by the standard semantics offered for provability-related logics. Fleshing out the non-deterministic perspective of these types of modalities not only will lead us to a new understanding of these modalities, but also may shed new light on the limitations of the non-deterministic semantics itself. This in turn will allow us to adequately assess the philosophical aspects of non-deterministic semantics and also to form a fair comparison between these semantics and other competing semantics such as Kripke's and truth-maker semantics. The ultimate aim of this project is to adequately situate the non-deterministic semantics in the context of provability-related logics, both technically and philosophically. These investigations will allow one to better understand the limitations of the finitely-many valued approach offered by non-deterministic semantics, especially towards semantical paradoxes like the paradox of informal provability or the paradox of naive validity. On top of that, developing uniform semantics for all of these modalities will result in a framework where one can easily dissect these notions and sketch new philosophical solutions to some of the mentioned paradoxes." "SfS++: Efficient & Robust Structure from Semantics" "Tinne Tuytelaars" "Processing Speech and Images (PSI), Czech Technical University in Prague" "In the context of this CELSA project, we see the opportunity to extend the work performed in the context of the SfS project in several directions, building on the specific expertise of CTU and the synergy of such collaboration." "Mapping metatext. On semantics of connectors, nature of conjunction and non-correspondence of metatexts of various languages on example of metatextual units marking opposition in Polish, Russian, Bulgarian, English and Dutch" "Anna Kisiel" "Functional and Cognitive Linguistics: Grammar and Typology (FunC), Brussels Campus" "Metatext in the presented study is understood as a group of language units that refer to our ways of speaking about our speaking and about ourselves as speakers in contrast to the units that are used for speaking about the world around us. The study concentrates on providing semantically and structurally anchored base for delimitating metatextual groups. It also aims at verifying the hypothesis on such division providing various results in different languages (it does not seem likely that two or more languages have the same metatextual resources). Such an approach, if proven right, will enforce a fundamental change in comparative linguistic concerning metatextual units." "Understanding the Multi-Modal World: Towards Multi-Modal Semantics, Information Search and Retrieval." "Marie-Francine Moens" "Informatics Section" "As more and more data nowadays are present in more than one modality (e.g., text, vision, video) besides text-only data, there is a pressing need to transfer the research from unimodal text environments to multi-modal environments, and to jointly model information that organically comes from multiple modalities. The focus of this project is on textual and visual modalities. One essential step of the project includes learning representations of multi-modal data based on both textual and visual/perceptual inputs. As one part of this project, I propose to work towards multi-modal and cross-modal models of information search and retrieval based on these joint representations (e.g., retrieving images that are relevant to an issued textual query and vice versa, or retrieving information based on a multi-modal query comprising both textual and visual information). Another part of the project tackles the recently started initiative on multi-modal semantics, where the goal is to learn semantic concept representations from both linguistic and visual input. I plan to investigate whether the additional inclusion of visual information may lead to improved models of semantic representation, with an emphasis on cross-lingual models of semantic similarity and association. For instance, different languages use different words to refer to the same concept which may be expressed by the same image, e.g., elephant(EN)-olifant(NL)-slon(HR) are words in three different languages that refer to the same image of an elephant. Therefore, I plan to study whether shared visual input may help to further bridge the lexical chasm across different languages." "Semantics and syntax of the verb sentir: a comparative Romance study" "Renata Enghels" "Department of Linguistics, Department of Romance languages (other than French)" "This project fits within the research into the possibly universal polysemy of perception verbs, but concentrates especially on a very complex but neglected element. More particularly, the project aims at the systematic analysis of the semantic and syntactic characteristics of sentir in Spanish and in comparative perspective with other Romance languages (French, Italian, Portuguese, Catalan)." "Search in art image collections based on color semantics" "Koen VANHOOF" "Business Informatics" "One of the most felicitous analogies for presenting the existing semantic gap in area of Content-Based Image Retrieval (CBIR) can be found in ""The Hitch-Hiker┌s Guide to Galaxy"" by Douglas Adams. In this story, a group of hyper-intelligent pan-dimensional beings demand to learn the ""Answer to Life, the Universe, and Everything"" from the supercomputer Deep Thought, specially built for this purpose. It takes Deep Thought 7½ million years to compute and check the answer, which turns out to be ""42"". The efforts of covering the semantic gap in CBIR are turned to avoid these misunderstanding between human perceiving and the ways of communications and computer manner of low-level representations. As it is mentioned in [Castelli et al] the user questions in image search are partitioned into three main levels: Low level ├ this level includes basic perceptual features of visual content (dominant colors, color distribution, texture pattern, etc.). Low-level queries and analysis can support the retrieval of art images in order to seek some specifics or common characteristics between artists, schools or movements. Intermediate level ├ this level forms next step of extraction from visual content, connected with emotional perceiving of the images, which usually is difficult to express in rational and textual terms. The visual art is area, where these features play significant role. Typical features in this level are color contrasts, because one of the goals of the painting is to produce specific psychological effects in the observer, which are achieved with different arrangements of colors. High level ├ this level includes queries according to rational criterions. In many cases the image itself does not contain information which would be sufficient to extract some of the characteristics. For this reason current high-level semantic systems still use huge amount of manual annotation. Different features' levels imply different ways for communication between the user and the CBIR system. When a system uses low-level properties such as color percentages, color layout, and textures, (see e.g. QBIC, developed by IBM) the queries do not need to be described in words. When working with such systems, the user can select a sample image and the system returns all images which are ""similar"" to it. For systems, which operate with high level features, only choosing a sample or drawing a sketch and search similar characteristics is not sufficient, even because such system has to ""know"" which of characteristics are targeted by the user. There are two mutually connected tasks in this area: ┘ Defining features and terms, which present certain effect or criterion and describing correlation between defined concepts; ┘ Finding appropriate algorithms for generating metadata, which alone or in combination with present terminal features and terms will allow improved image search as well as proposing adequate methods and tools for establishing belonging of a sample to same concept. In order to present properly concepts and their correlation between low and intermediate levels as well as the connections to the high level, every system usually creates its own dataset. This allows implementing the specific elements of the used methods and tools." "NEPHOLOGICAL SEMANTICS: Using token clouds for meaning detection in variation linguistics" "Dirk Geeraerts" "Quantitative Lexicology and Variational Linguistics (QLVL), Leuven" "The increasing importance of corpus data in linguistics creates a need for appropriate methods for retrieving semantic information from corpora. In the project proposed here, existing computational methods of distributional corpus semantics are further developed in the form of a meaning detection approach based on token clouds, i.e. clusters of distributionally similar attestations of words or expressions in a multidimensional vector space. The first phase of the project has a methodological orientation, focusing on the finetuning of such a 'nephological' method for detecting linguistic meanings in corpus data. In the second phase of the project, the method is put to use in two descriptive research lines: lectometrical research into the relationship between language varieties, and variationist grammar research."