Abstract Ziem

Mental Spaces, Idealized Cognitive Models, Frames, or Domains? Toward a Unified Theory of Semantic Representations

Starting up from a conceptual approach to linguistic semantics, my talk aims at developing a general framework unifying different aspects of semantic representations. The guiding working hypothesis is that current cognitive-semantic theories disregard representational aspects and therefore violate important methodological principles. Only a unified theory of semantic representation can avoid this deadlock.

Following a non-modular approach to the study of human cognition Cognitive Semantics, in particular, has focused on a wide range of different phenomena during the last 25 years of research. On the one hand it aims at explaining cognitive construal operations which are relevant not only to language processing but also to other cognitive processes such as visual perception. The ability to categorize, to schematize, and to distinguish figure from ground, to mention but a few, are assumed to be such general cognitive operations. On the other hand, this ‘principle of unitariness’ also holds for cognitive structures in which linguistic knowledge is mentally stored and through which linguistic knowledge is retrieved during language comprehension. From this point of view, all kinds of cognitive processes are ultimately based in conceptual entities.

Although there is no doubt about the intrinsic interconnection between cognitive processing and cognitive representations, the focus of current research, however, lies on the study of construal operations. Representational aspects are widely disregarded. As a result, semantic representations are described by means of many different representation structures. The most famous among them are mental spaces (Fauconnier, Turner), idealized cognitive models (Lakoff), domains (Langacker et al.), and frames (Fillmore). Partly, these different conceptions seem to compete with each other; partly, they seem to be complementary to one another. Mental spaces, for example, fill an explanatory gap of semantic descriptions in so far as they explain the status of knowledge aspects (e.g. desires, counterfactuals, beliefs) to which an expression refers; frames, domains, and ICMs cannot handle such data.

The main point is that this ‘division of labour’ essentially contradicts central cognitivistic assumptions. Considering the methodological principle of cognitive economy it would be suitable to have one single representation structure which covers all different semantic phenomena. Furthermore, it is not plausible to delimit the unitary approach of human cognition to conceptualization processes. Why should representation structures be excluded? Rather, a semantic model should be preferred which is explanatory adequate but theoretically more economical. Finally, in contrast to generative approaches, Cognitive Linguists strongly advocate a cognitively realistic theory of language. But to claim that semantic knowledge is stored in four (or even more) different representation structures in long-term memory is not psychologically realistic at all.

In my talk diverse conceptions of semantic cognition such as mental spaces, domains, idealized cognitive models, and frames shall be investigated in terms of both mutual compatibility and explanatory adequacy. The aim is to develop a framework for a unified theory of semantic representations covering all different semantic phenomena each approach claims to explain.