30th - AEDEAN, Huelva 2006 | Conferences in history

Program Structure / PLENARY SPEAKERS

Prof. Dr. Dirk Geeraerts (University of Leuven)

«Metonymy as a Prototypical Category»

The purpose of the present talk is to discuss the proper treatment of metonymy in the context of Cognitive Linguistics. For the linguist members of Aedean, the talk will highlight a current debate within Cognitive Linguistics. For the literary scholars in the audience, the talk may be interesting as an illustration of the way in which recent developments in linguistic theory shed a new light on concepts (like that of metonymy) that are equally important in linguistics as in literary semantics.

The discussion of metonymy in the context of Cognitive Linguistics has so far concentrated on the definition of metonymy as a shift of meaning within one domain or domain matrix (Kövecses 2002:145). It is gradually becoming clear, however, that this definition relies too much on the vague notion of «domain (matrix)» to be fully operational; see Taylor (2002) or Ruiz de Mendoza & Otal Campo (2002) among a number of critical voices. Various linguists (among them Feyaerts 1999, Riemer 2001 and Croft & Cruse 2004) now seem to agree that it is unwise to use identity versus difference between the semantic domains involved as a basis for the differentiation of metaphor and metonymy.

A simple return to the older definition of metonymy in terms of «contiguity» (see Ullmann 1962) is, however, precluded, because a unitary definition of metonymy in terms of contiguity is as problem-ridden as a unitary definition in terms of domains or domain matrices. Rather, it would seem that a non-unitary definition of metonymy is called for. And the obvious way of constructing such a non-unitary definition in Cognitive Linguistics is to use a prototype-theoretical model of categorization. It will be argued, then, that the concept of «metonymy» is a prototypically structured one with spatial contiguity in the centre of the category.

The empirical basis of the paper consists of a list of widely accepted metonymical patterns (next to the prototypical SPATIAL PART & WHOLE, the list includes patterns like TEMPORAL PART & WHOLE, LOCATION & LOCATED, ANTECEDENT & CONSEQUENT, CHARACTERISTIC & ENTITY, CAUSE & EFFECT, MATERIAL & OBJECT etc.). The list was compiled to a large extent on the basis of the pre-structuralist literature on diachronic semantics (Paul 1880, Nyrop 1913, Waag 1901, Esnault 1925), which, descriptively speaking, is still the richest source on varieties of metonymy.

The analytical part of the paper argues that these metonymical types can all be related to the prototypical core of spatial part-whole contiguity. The relations between the derived types and the core case (and between the extended types among each other) can be plotted against three dimensions: «strength of contact» (going from part-whole containment over physical contact to adjacency without contact), «boundedness» (involving an extension of the part-whole relationship towards unbounded wholes and parts), and «domain» (with metaphorical shifts from the spatial to the temporal and the categorial domain). As may be expected in a prototypically structured category, these dimensions interact in various ways.