 |
Tutorials
-
Language Tutorial: Prosody-Syntax Interface in a Typological Perspective
Caroline Féry
The course will investigate the role played by prosody in language processing. Why do we prosodically phrase and intonate speech, even when syntax and semantics provide an unambiguous meaning? Clearly, prosody is much simpler than these latter components, and needs less constituents. But still, an impressive amount of information lies in the prosodic structure of utterances. In the first part of the course, prosody and intonation (categories, theories, etc) will be introduced.
Prosody can be realized or silent. While reading silently, we project an unmarked prosodic structure on sentences. When this unmarked prosodic structure is incompatible with the syntax, garden-path effects are the result, and processing difficulties (like increased reading times) arise. In the second part of the course, we will discuss the issues around garden-path effects: syntactic and semantic ambiguities and why certain readings are preferred over others. Languages use prosody in different ways. This is especially true in the way prosody is used in the expression of information structure, like focus and topic. Some languages, like English, German, and many other European languages use prosody to signal information structure, beside syntax. But other languages do not accent prominent constituents, but rather move them syntactically or identify them with the help of morphological markers. In the last part of the course, typological variation in the way information structure is marked will be discussed.
Literature and references:
- Theories
Gussenhoven, Carlos (2004) The Phonology of Tone and Intonation. Cambridge University Press.
Ladd, D.R. (1996): Intonational phonology. Cambridge: Cambridge University Press.
Pierrehumbert, J. (1980) The phonology and phonetics of English intonation. MIT Diss.
- Processing
Cutler, A./Dahan, D./van Donselaar, W. (1997): Prosody in the comprehension of spoken language: A literature review. Language and Speech 40/2, S. 141-201.
Fodor, J. D. (2002b): Psycholinguistics Cannot Escape Prosody. Proceedings of the Speech Prosody 2002 Conference. Aix-en-Provence, S. 83-88.
Steinhauer, K./Alter, K./Friederici, A. (1999): Brain potentials indicate immediate use of prosodic cues in natural speech processing. In: Nature Neuroscience, 2, S. 191-196.
Watson, Duane/Gibson, Edward (2004): Making Sense of the Sense Unit Condition. In: Linguistic Inquiry 35.3, S. 508-517.
- Information structure and typology
Gussenhoven, Carlos (1983): Focus, mode and the nucleus. In: Journal of linguistics 19, S. 377-417
Jun, Sun-Ah, Ed. (2005) Prosodic Typology. The Phonology of Intonation and Phrasing. Oxford University Press.
-
Logic Tutorial: The modal mu-calculus
Yde Venema
Process theory studies the behavior of programs. Of particular importance is the specification of *ongoing* behavior, which is characterized by properties like safety (something bad never happens), liveness (something good eventually happens). Process theory is one of the most important application areas of modal logic. What characterizes these logics is the use of fixed point operators --- this is needed to capture the ongoing nature of the behavior under study.
The tutorial will offer a general introduction to one of the most important process logics: the modal mu-calculus. This logic also has some very attractive (meta-)logical properties. In particular, it strikes a remarkably good balance between expressive power and computational behavior. The topics that we will cover include: syntax and semantics, some basic fixed point theory, evaluation games, the connection with automata theory, and some logical properties such as finite model property and decidability.
-
Computation Tutorial: Grammars, Old Georgian, and Annotation (working title)
Aravind Joshi
Starting backwards, I would like to use the last slot for discussing some aspects of discourse structure, transition from sentence to immediate discourse--all this in an introductory manner. Then I will describe the annotation project for annotating discourse connectives and their arguments, etc. for the I million word Penn Treebank corpus-- a project that I am currently involved in. So this is sort of empirical work.
The first two hourly talks would be about some formal aspects of grammars leading up to the tree-adjoining grammars and related grammars (in particular categorial grammars).
In the second talk I will use some time to describe how some Old Georgian data has played a role in a controversy about some formal claims about the growth of sentences. Since these arguments involve some Old Georgian data of Bader, it might be interesting and amusing(!) to the local audience. Of course, my knowledge of OG is limited to this data!
Invited lectures
-
Giovanna D'Agostino (University of Udine)
Bisimulation Quantifiers and Fixed Points over Modal Logics
Fixed points of monotone operators can be added to modal logics to extend their expressive power: e.g. in the case of basic modal logic K, the resulting extension is the well known and intensively studied modal mu calculus. In the case of logics based over transitive frames, fixed points can be expressed by means of the so called bisimulation quantifiers. In this lecture we give a brief introduction to bisimulation quantifiers starting from their relation with uniform interpolation. We then compare fixed points and bisimulation quantifiers over some specific classes of frames, showing as the latter may be used in some cases as honest alternatives to fixed points, but also considering examples in which bisimulation quantifiers behaves very wildly.
-
Matthias Baaz (University of Technology, Vienna)
Skolem functions: the Hidden Concepts in Proofs
-
Leo Esakia and David Gabelaia(Georgian Academy of Sciences)
Reincarnations of the Standard Provability Predicate: a modal consideration
We are going to discuss certain "intrinsic" reincarnations of the standard Provability Predicate in Peano Arithmetic PA that are of special interest in connection with the study of Provability Logic. All special reincarnations of standard provability predicate proposed for consideration here are metamathematical predicates distinct from the standard provability, yet strong enough to satisfy Hilbert-Bernays Derivability conditions.
Our reincarnations are internally definable and need not be introduced as an additional structure. The purpose of our talk is to discuss certain observations arising from a study of such reincarnations of the standard provability predicate. Namely:
- What are the implications of such reincarnations?
- What can we say about modal logic of such "non-standard" provability predicates?
Note that our study has opened more questions than it has answered.
-
Frans Groen (University of Amsterdam)
Distributed Collaboration and Decision Making in Uncertain Situations
In future intelligent systems will be deployed in real-world situations, for instance as service robots, transportation systems, in exploration of hazardous environments, in homeland security and disaster scenarios. The societal and economical benefits of such systems is huge, while at the other hand still important research questions have to be answered before such systems can be applied. Research questions involve the distributed perception of the real-world by artificial agents using sensors, the communication and fusion of that information with human actors who use their senses to perceive the world and how to obtain a local situation awareness of the real world. Decisions have to be made based on this information and collaborative actions to be taken. Some of those actions can be taken fully autonomously, other actions should be in human control.
In this talk we will focus on aspects of distributed collaboration and decision making in a multi-agent setting. We will discuss methodologies such as coordination graphs, which enable real world multi-agent systems to cooperatively solve a task. Partial Observable Markov Decision processes (POMDPSs) can be used for planning in uncertain environments. The POMDP framework allows for successfully handling uncertain situations, using a belief state of a robot. Solving a POMDP in an exact fashion is an intractable problem. A recent line of research on approximate POMDP algorithms involves the use of a sampled set of belief points on which planning is performed. Distributed Bayesian Belief Networks enable to take decisions in uncertain situations. The robustness of these networks and the possibility of detection of faulty links will be discussed.
How to evaluate these real-world systems is not an easy question. Simulation is certainly useful in this respect, but real comparisons require the deployment of systems in real world scenarios. It has been discerned that international challenges may play an important role in those evaluations. Examples of such challenges are RoboCup with Robots playing soccer, Robot Rescue with robots trying to find victims after a disaster and the Darpa Grand challenge of autonomous driving in unstructured terrain. RoboCup and disaster scenarios will be used as illustration of the methods discussed.
-
Robin Hirsch (Computer Science, University College London)
TBA
-
Igor Melcuk (Montreal University)
TBA
-
Vladimir Borschev and Barbara H. Partee (University of Massachusetts)
Diathesis and Referentiality: Interacting Factors in the Russian Genitive of Negation Construction
This contribution concerns the interrelationships among (i) diathesis alternations, (ii) shifts in the lexical semantics of the verb in different diathesis frames, and (iii) the distribution of "strong" and "weak" NP interpretations in different diathesis frames, with particular application to the Russian Genitive of Negation construction.
-
Henk Verkuyl (Utrecht University)
Georgian as the testing ground for (Western) theories on tense
The Georgian tense system appears to be very complicated by the so-called screeve system. Its complexity can be dealt with by carefully eliminating all sorts of factors that do not seem to belong to the proper tense domain. But what is the proper tense domain? Is it a Reichenbachian sort of structure (i.e. a system based on a 3x3-partition) or a Te Winkelian one (i.e a system based on a 2x2x2-partition)? Or do existing tense theories fail to account for Georgian. These are questions that will be dealt with by showing how far Western theories "get" when we apply them to Georgian.
-
Michael Zakharyashev (Department of Computer Science, King's College)
TBA
-
Ede Zimmermann (Johann Wolfgang Goethe Universität Frankfurt)
TBA
|
