Ciro Cattuto (Centro Enrico Fermi Rome) and Andrea Baronchelli (University of Rome)
The abstract of this atelier will be announced shortly.
Ateliers
Semiotic Dynamics on the Web
Collostructional Analysis (Statistical Linguistics)
Anatol Stefanowitsch (University of Bremen)
Collostructional analysis is a family of quantitative corpus-based methods for studying the relationship between words and grammatical constructions, especially at the semantic level. In this atelier, three basic methods are introduced: (i) collexeme analysis, which aims at investigating the relationship between a given construction and the lexical items that occur in it (cf. Stefanowitsch and Gries 2003); (ii) distinctive collexeme analysis, which aims at investigating subtle semantic differences between semantically and/or functionally near-equivalent constructions (cf. Gries and Stefanowitsch 2004a); and (iii) covarying collexeme analysis, which aims at uncovering semantic and/or cultural frames by investigating the relationship between words occurring in different slots of the same construction (Gries and Stefanowitsch 2004b, Stefanowitsch and Gries 2005). Each method is introduced through a series of step-by-step explanations and hands-on exercises. Some recent extensions of the method will also be demonstrated. In addition, some freely available software applications will be introduced for calculating the statistics required for collostructional analysis.
- Gries, Stefan Th., and Anatol Stefanowitsch (2004a). Extending collostructional analysis: a corpus-based perspective on 'alternations'. International Journal of Corpus Linguistics 9(1): 97-129.
- Gries, Stefan Th., and Anatol Stefanowitsch (2004b). Co-varying collexemes in the into-causative. In: Michel Achard and Suzanne Kemmer (eds.), Language, Culture, and Mind, pp. 225-236, Stanford: CSLI.
- Stefanowitsch, Anatol, and Stefan Th. Gries (2003). Collostructions: investigating the interaction of words and constructions. International Journal of Corpus Linguistics 8(2): 209-243.
- Stefanowitsch, Anatol, and Stefan Th. Gries. (2005). Covarying Collexemes. Corpus Linguistics and Linguistic Theory 1(1): 1-46.
Implementing Language Games on Robots
Martin Loetzsch (Sony CSL Paris) and Joachim De Beule (VUB University of Brussels)
The abstract of this atelier will be announced shortly.
The Co-Ordination Between Language and Categorization
Angelo Cangelosi (University of Plymouth) and Tony Belpaeme (University of Plymouth)
Categorisation is one of the fundamental capabilities of natural and artificial cognitive systems. It allows individuals to "sort out" the complexity of the environment and build a discrete categorical representation of it. At the same time, categorical representations constitute the grounding for the acquisition of language (Harnad, 1990). In this atelier we will look at the interplay between categorisation and language and explore the effects that language has on category learning.
The atelier is organized into three parts.
- Part 1: Categorisation, categorical perception and language. Initially we will review the psychological theories of Categorisation (e.g. feature-based Categorisation, prototype models, exemplar-based theories, connectionism). We will the focus on the role of categorical perception, the phenomenon by which the structure of internal similarity space changes ("warping") during category learning. The role of language in enhancing categorical perception effects will be demonstrated using computational models based on neural networks and cognitive agents (Cangelosi 2005).
- Part 2: Colour Categorisation and naming. Colour cognition is an excellent domain in which to study language and categorisation. Colour categories are to a large extent similar between different cultures, meaning that it is often easy to translate colour words from one language to another. If we want to maintain the position that categories are influenced by language, then the fact that colour categories are nearly universal is disconcerting. We explore a number of theories and present a computational model that explains the observations of psychologists and anthropologists (Belpaeme and Bleys 2005, Steels and Belpaeme 2005).
- Part 3: How machines learn the meaning of words. In order to acquire categories and their linguistic referents, children need to learn the meaning of words. A number of different paradigms exist in computational modelling of lexicon acquisition, and we briefly look into how children manage to learn words for categories, and how this can be implemented in silicon.
- Cangelosi, Angelo (2005). Approaches to Grounding Symbols in Perceptual and Sensorimotor Categories. In H. Cohen & C. Lefebvre (eds), Handbook of Categorisation in Cognitive Science, Elsevier. [PDF]
- Harnad, S. (1990). The Symbol Grounding Problem. Physica D 42: 335-346. [HTML]
- Belpaeme, Tony and Bleys, Joris (2005). Explaining universal colour categories through a constrained acquisition process. Adaptive Behavior, 13(3). In press. [Draft available here.]
- Steels, Luc and Belpaeme, Tony (2005). Coordinating Perceptually Grounded Categories through Language. A Case Study for Colour. Behavioral and Brain Sciences. 28(4):469-529. [Draft available here.]