The class-inclusion theory of metaphor: a critical

THE CLASS-INCLUSION THEORY OF METAPHOR: A CRITICAL VIEW Ştefania Alina Cherata* Abstract: The class-inclusion theory of metaphor was proposed by Glucksberg and Keysar (1990), and directed against the view that metaphoric statements of the form X is Y are implicit comparisons, and their interpretation involves a process of feature matching between topic (‘X’) and vehicle (‘Y’). The authors suggest instead that such sentences qualify as explicit class-inclusion assertions, in which the topic is ascribed to a superordinate ad hoc category typified by and named after the vehicle. I will argue that, although superordinate categories comprising both the topic and the vehicle of a metaphoric statement can indeed be constructed in a large number of cases, they fail to provide an adequate theoretical foundation for the interpretation of metaphor. Furthermore, the prototype status attributed to the vehicle obscures the difference between metaphor and certain metonymic patterns in which the name of a prototypical category member stands for the category as a whole. Keywords: metaphor interpretation, mapping, superordinate category, prototype 1. Introduction One of the most prominent theses advanced in metaphor theory connects metaphor with a form of semantic conflict or anomaly translating in category violation1. Thus, a metaphoric sentence such as Man is a wolf (Black 1981 [1954]: 73) seems to place man in the class of wolves, which clearly runs counter to the conventional delimitation of semantic categories. Similarly, the statement My car is thirsty is characterized by category violation in that it attributes a property of an animate entity (‘being thirsty’) to an inanimate object (my car) (Way 1991: 42-43). The need to resolve this tension has posed a major challenge to metaphor theory, and materialized in a number of competing approaches. One notable attempt at dealing with the semantic conflict typically present in metaphor was provided by the so-called comparison view, which treats metaphor as an implicit comparison (cf. Rolf 2005: 2122). According to this theory, a sentence like Man is a wolf can be said to mean roughly ‘Man is like a wolf.’ A series of metaphor-processing models have been devised based on the comparison view. The most rudimentary versions postulate a simple matching of properties between the terms of the implicit comparison (the metaphor vehicle ‘wolf’ and the topic ‘man’)2. In keeping with such models, the metaphoric statement quoted above, * “Friedrich Schiller” University, Jena, . This aspect was brought to the fore of modern metaphor research by Beardsley’s controversion theory (1958). Although it is generally acknowledged that metaphor taken literally involves some form of conflict, it has been pointed out that semantic deviance is neither a necessary nor an exclusive feature of metaphor, but one that is nevertheless present on a large scale (Black 1993 [1977]: 34). 2 I am using the terms “topic” and “vehicle” to refer to linguistic forms (man, wolf), but also to the corresponding concepts (‘man,’ ‘wolf’) and referents (man, wolf). Strictly speaking, metaphor as I understand 1 130 Ş te f a nia Al i na C h er at a Man is a wolf, is interpreted by identifying a set of commonalities between ‘man’ and ‘wolf’ (e.g. ‘being aggressive,’ ‘being reckless’, etc.); these commonalities constitute the ground of the metaphor. Although at first sight they provide an intuitive approach to metaphor processing, matching models are problematic on a number of counts. The class-inclusion theory, developed by Sam Glucksberg and Boaz Keysar in 1990, and refined in later years by Glucksberg and his associates (Glucksberg and Keysar 1993; Glucksberg et al. 1997a; Glucksberg et al. 1997b; Glucksberg 2001; McGlone and Manfredi 2001), was intended as a critical response and an alternative to matching algorithms. The present paper offers a critical discussion of the theory, using as a starting point the versions found in Glucksberg and Keysar (1993) and Glucksberg (2001). I will argue that, despite certain advantages, the class-inclusion model does not provide a viable basis for metaphor interpretation. My argumentation is structured in five main parts. The next two sections outline the context which motivated the emergence of the class-inclusion view. They introduce matching models as a subgroup of metaphor-processing theories (section 2), and call attention to their major weaknesses (section 3). Section 4 summarizes the core tenets of the class-inclusion theory, which is subsequently compared with the matching models it was intended to counter (section 5). Finally, I discuss the role of prototypicality within the class-inclusion view, and its implications for the accuracy of the theory (section 6). 2. Metaphor-processing models Wolff and Gentner (1992) distinguish between three classes of metaphorprocessing theories: matching models, mapping models, and a combination of the two: matching-then-mapping models. According to the first type of approach, metaphor interpretation involves a comparison of topic and vehicle concepts, from which common features are extracted. For example, the metaphoric statement My surgeon is a butcher can be interpreted by recognizing that both surgeons and butchers share the property of cutting flesh (Wolff and Gentner 1992: 504). Moreover, the surgeon to whom the topic refers resembles a butcher in that he performs this operation in a sloppy, imprecise manner (but cf. 5.4.). Mapping models are centred around the phenomenon of property attribution. They postulate the creation of a higher-order category that uses the features of the vehicle as a starting point, and functions as an instrument for the attribution of certain properties to the topic. The sentence My surgeon is a butcher is comprehended by mapping the property of cutting flesh sloppily from the vehicle (‘butcher’) onto the topic (‘surgeon’) (Wolff and Gentner 1992: 504). In matching-then-mapping models, processing begins by aligning vehicle and topic representations to establish common features (e.g. ‘cutting flesh’), and continues with a mapping process, during which further properties of the vehicle that are consistent with it only includes the vehicle; the sentence in which the topic and the vehicle occur will be termed metaphoric statement or metaphoric sentence. The class-inclusion theory of metaphor: A critical view 131 the system of shared predicates are transferred to the topic (e.g. the attribute of sloppiness) (Wolff and Gentner 1992: 504-505). 3. Criticism of matching models The class-inclusion theory of metaphor is a mapping model developed as an alternative to matching theories. In particular, it opposes simple matching models, whose assumptions are objectionable in a number of respects. 3.1 Metaphor vs. simile Perhaps the most obvious weakness of matching models, and of the comparison theory in general, is that they focus on the resemblances between metaphor and sim (...truncated)