Universiteit van Amsterdam

Archives

Institute for Logic, Language and Computation

Please note that this newsitem has been archived, and may contain outdated information or links.

Four Veni grants 2011 at ILLC

Four young ILLC researchers have been awarded a Veni grant 2011.

  • Dr. Michael Franke: Models of Language Evolution and the Topology of Semantic Space: The Case of Gradable Adjectives

    Evolutionary game theory does well in explaining the socio-cultural evolution of linguistic meaning, and has therefore received considerable attention recently in this context. In particular, the use of signaling games to study the evolution of meaningful communication is thriving. But the meanings that these models give rise to, so to speak, are still fairly basic and lack many interesting properties that human languages show. This research project aims to improve on signaling models in this respect by taking recent and relevant work from linguistic semantics into account. In particular, the project turns towards work on the meaning of gradable adjectives, which is a topic of great current interest within theoretical linguistics. A prominent line of research links the meaning of gradable adjectives to degrees on scales. Differences in meaning between adjectives are explained by postulating different properties of the associated scales. These models from formal semantics form a formidable test suit for models of language evolution. The main research question to be addressed is therefore: can evolutionary game theory account for the linguistically-attested differences in meaning of gradable adjectives, and if so, how? By addressing this question, this project will, on the one hand, contribute to the toolset of evolutionary theorizing about language, and, on the other hand, it will also contribute to the debate about the semantic and pragmatic properties of gradable adjectives by providing a functional explanation for relevant differences in meaning and use of gradable adjectives.

  • Dr. Aline Honingh: Representing Music: A New Basis for Computational Musicology

    Music informatics is an emerging interdisciplinary research area which has arisen from the fields of artificial intelligence, mathematics and music theory. In the last decade, the amount of digital music has increased so enormously that new research questions related to the representation and classification of music have come up, in particular: how can we automatically recognize, classify, and order music?

    Music can be represented in various ways, for example as audio or scores. Many existing representations of music are only suitable for one specific task or application. This research project proposes a new representation of music that can form a basis for many tasks and applications.

    Our proposed representation is based on the concept of interval categories. A particular interval category contains segments of music that are dominated by that particular interval. The representation of interval categories will be enriched with tree structures and temporal information so as to answer the research questions: 1) What is the most adequate integration of pitch, temporal and hierarchical knowledge?, and 2) How can we improve on existing applications in music research with this new representation?

    For the evaluation of our representation, we focus on the following tasks/applications: 1) music classification, 2) measuring melodic similarity, and 3) finding sequential association rules. By integrating low-level and high-level features (such as tree structures) we hope to show that actual progress is possible in music informatics and its applications.

  • Dr. Floris Roelofsen: Interpreting Questions - fine-grained compositional semantics

    Questions play a central role in linguistic communication and scientific inquiry, and have been investigated extensively from linguistic, philosophical, logical, and computational perspectives. Considerable progress has been made, but many crucial aspects remain poorly understood. The recently developed framework of inquisitive semantics (Groenendijk and Roelofsen, 2009) offers a natural starting point for the pursuit of a more comprehensive and precisely articulated account.

    The project will bring innovation on two different fronts. On the one hand, it will develop a richer formal conception of the meaning of questions. Traditional semantic representations of questions are designed exclusively to embody the issue that the question raises. However, questions do more than raising issues. For instance, they may express a bias towards a certain answer (e.g., Isn't Fred married?), or they may 'highlight' a certain possibility, thereby making it available for subsequent anaphoric reference (e.g., Do you need a loan? Then we can help you). We will pursue an account of these and other semantic aspects of questions that go beyond their issue-raising potential. Our hypothesis is that many such phenomena can be given a unified explanation based on the assumption that questions always highlight some of the possibilities that they propose (Roelofsen and van Gool, 2010).

    The second main innovative aspect of the project is that it will pay close attention to differences in surface form and intonation. Why, for instance, do questions with auxiliary negation (e.g., Isn't Ann going?) have a different range of interpretations than questions with internal negation (e.g., Is Ann not going?), and how exactly does the interpretation of disjunctive questions (e.g., Is Ann or Bill going?) depend on intonation. We will identify the relevant syntactic and intonational factors, and formulate a comprehensive compositional interpretation procedure.

  • Dr. Bryan Renne: Evidence-Based Belief Revision

    Belief Revision is the study of how new, possibly contradictory information should rationally affect one's beliefs. This is an active, multi-disciplinary area of study with applications in Logic, Artificial Intelligence, Philosophy, Law, and Economics. Much work in Belief Revision has focused on the so-called "postulate-based" approach, which characterizes the belief change process in terms of a series of statements that say what ought to be the case after a belief change has occurred. While this extremely popular approach has had its share of successes, it has neglected to address the underlying reasoning process an individual might use in actually changing her beliefs. Such a process ought to take into account the uncertain evidence one has at her disposal, allowing her to perform stepwise reasoning about her evidence without demanding infinite cognitive and logical precision as is typically done in Belief Revision theory.

    The aim of this project is to develop a new theory of Belief Revision that describes belief change as a step-by-step evidence-weighing process in which errors might occur but can later be corrected. The theory will combine, adapt, enhance, and otherwise custom-fit ingredients from Dynamic Epistemic Logic and Justification Logic, two fast-growing areas of Applied Logic that present great promise toward this end.

The Veni grant is part of the Innovational Research Incentives Scheme run by the Netherlands Organisation for Scientific Research (NWO). It is one of the most prestigeous grants for young researchers in the Netherlands, providing an impulse to their scientific carreer. Applications were assessed by scientists both from the Netherlands and abroad. The Veni grant will provide the laureates with an opportunity to spend three years developing their research ideas.

See also: http://www.nwo.nl/nwohome.nsf/pages/NWOP_8MQKAT.

Please note that this newsitem has been archived, and may contain outdated information or links.