Steen Gerard. Finding Metaphor in Grammar and Usage: A methodological analysis of theory and research

Jonh Benjamins Publishing Company, 2007. — 431 p. The focus of the book is on the two-domain approach to metaphor known as conceptual metaphor theory. It was first presented to a wider readership by George Lakoff and Mark Johnson in their influential work Metaphors We Live By (1980) and further developed, e.g., in Lakoff (1987, 1990), Lakoff and Turner (1989), Lakoff and Johnson (1999), and, as the ‘neural theory of metaphor’ in Lakoff (2008). Conceptual metaphor theory stipulates that metaphors involve unidirectional mappings from one conceptual domain(source) into another distinct conceptual domain(target),where, in general, the source domain is assumed to be conceptually more concrete than the target domain. Metaphorical utterances often group into families of conceptually related expressions that are all ‘‘generated’’ by the same underlying conceptual metaphor. It is thanks to Lakoff and Johnson’s and their associates’ pioneering studies that whole systems of conceptual metaphor have been uncovered. Often-cited and discussed examples are the TIME IS MONEY metaphor, instantiated in expressions such as waste/lose one’s time and spend time; the ARGUMENT IS WAR metaphor, in shoot down an argument, defend a thesis, and attack an opponent; and the THEORIES ARE BUILDINGS metaphor, instantiated in expressions such as His theory lacks solid foundations and Language is a window to the mind. Steen argues persuasively that the empirical adequacy of conceptual metaphor theory can only meaningfully be examined within certain research areas, viz. grammar, usage, symbolic structure, conceptual structure, metaphor production, and comprehension. For example, it is one thing to assume a unidirectional mapping between two isomorphically structured domains on the level of symbolic structure (be it grammar or usage), but quite a different thing to hypothesize that the processing of metaphor in production and comprehension, i.e., what Steen refers to as ‘‘behavior’’ in his book, is achieved through a unidirectional ‘‘mental’’ mapping from a source domain into a target domain. In fact, psychological evidence speaks against the view that metaphorical meanings are always retrieved via the activation of literal (source) meanings (see below). One of the strong points of the Steen’s book is that the reader is constantly cautioned against sweeping generalizations, in particular against the systematic transfer of theoretical insights from one research area into another. In what follows, adopting and elaborating some of Steen’s thoughts, I present two brief case studies that illustrate the argumentative pitfalls of overgeneralization. Consider first how metaphor should be analyzed as a phenomenon of grammar (in contrast to usage). The (idealized) distinction between grammar and usage is probably made in one way or another in all standard linguistic frameworks. Steen conceives of grammar as a system of linguistic units that form conventionalized form-meaning pairs. Grammar, in the author’s view, thus encompasses more than just morphosyntax and phonology. Metaphors, as long as they are conventional form-meaning pairs, are an integral part of (lexico)grammar.1 In contrast, the term ‘usage’ refers roughly to what in pragmatic theories is called ‘speaker meaning’ or ‘utterance meaning’ (form-meaning pairings on specific occasions of use). In lexicogrammar, metaphor is especially productive as a means of meaning extension ,i.e., it creates lexical polysemy. There are several methodologically reflected decisions that have to be made by the researcher before a (unidirectional) mapping between a ‘‘basic’’ and a metaphorical sense can be reasonably hypothesized. As an example, also discussed in Steen’s book, consider the meanings of the adjective indefensible (p. 139), for which Steen finds the following two readings in a monolingual dictionary: (a) ‘impossible to defend from criticism’, as in a morally indefensible statement, and (b) ‘impossible to protect against military attack’. These two senses are semantically related. Furthermore, the fact the two readings are listed in a monolingual dictionary is one piece of evidence that they are conventional meanings: they belong to lexico-grammatical system, not to usage. The next step for the analyst is to determine the kind of conceptual relation that obtains between the two senses. Lakoff and Johnson (1980) have proposed that sense (a) is an instantiation of the fairly pervasive ARGUMENT IS WAR metaphor – paralleled by the figurative senses of some other lexical items such as attack, demolish, shoot down, etc. – which can be used both in the context of physical struggle and that of verbal argument. The usual diagnostic for metaphor is to assume a relation of non-literal similarity between the literal sense and the assumed figurative sense. In the case of indefensible, the situation of an indefensible position in a physical fight or in war can be seen as similar (analogous) to the indefensible (intellectual) position of a participant in a debate. According to the Pragglejaz Group,2 one of whose prominent members is Gerard Steen, the basic meaning of the lexical item in question from which the metaphorical meaning is derived tends to be ‘‘more concrete’’, ‘‘related to bodily action’’, ‘‘more precise’’ and ‘‘historically older’’ (pp. 88–89). In the case of indefensible, the assumed basic meaning (source) is indeed more concrete and it is also related to bodily action. It is however hard to establish whether the source eaning is ‘‘more precise’’ than its metaphorical counterpart. As to the relative ages of the two senses, unfortunately, the earliest date of the ‘war’ sense of indefensible cited in the Oxford English Dictionary (OED) is 1569, whereas the abstract ‘argument’sense is attested as early as 1529. The consultation of other sources (e.g., historical corpora) might of course lead to different results regarding the chronological order of emergence of the two senses in the English language, but given the data in the OED one must consider the possibility that diachronically there is perhaps no unidirectional mapping from a more concrete source domain into a more abstract target domain, but simply structural parallelism in how physical struggle (war) and the activity of arguing are conceptualized. Structural parallelism is actually advocated by Ray Jackendoff (1987: 156; 1990: 25–27; 2002: 356–360), who suggests that lexical items, e.g., verbs, can function in more than one lexical field. While Lakoff and Johnson would assume that the ‘change-of-position’ meaning of turn in The coach turned into a driveway is the metaphorical source of the ‘change-of-appearance’ meaning of the verb in The coach turned into a pumpkin, Jackendoff (1987: 156) calls this relation a ‘cross-field generalization’, i.e. a systematic semantic and grammatically relevant set of non-directional correspondences between semantic fields. In other words, Jackendoff does not regard linguistically entrenched correspondences of the above kind as metaphorical mappings at all. When it comes to metaphor processing (in contrast to metaphor as symbolic structure), Lakoff and Johnson’s two domain approach meets an even more serious empirical problem. Their conception of metaphor as an asymmetrical mapping from a source into a target domain suggests that in actual processing the mental point of departure of the interpretation is always the source domain. This view has been challenged from a psychological perspective. For example, Giora (2002) presents experimental evidence that salient metaphorical word meanings, i.e., frequent, familiar, conventional, and prototypical meanings, are immediately accessed, i.e., not retrieved via their literal senses, as Gricean models and also the Lakoffian two-domain approach would suggest. These empirical findings do not necessarily present insurmountable difficulties for the two-domain approach, but they have to be addressed in a principled fashion. To conclude, Steen’s book is one of the relatively rare publications in cognitive linguistics with a truly interdisciplinary perspective: it takes the term ‘cognitive’ in cognitive linguistics seriously. The author displays great expertise in both linguistics and cognitive psychology, and the book is a fountain of knowledge and inspiration for metaphor theorists. An additional asset of the book is its very useful and up-to-date bibliography of 468 references. The subject index is extensive, though without subheadings and cross-references. Regrettably, there is no name index. Given the over 400 pages of densely printed text, it can be quite time-consuming for the reader to retrieve the names of specific authors and their theoretical views. Nevertheless, Gerard Steen has written an important, often provocative, text, which will surely stimulate discussion about the heuristics and empirical validation of metaphor identification and analysis. The author does not so much offer a new theory of metaphor intended to supplant existing models; rather he provides a state-of-the-art account that critically but also constructively seeks to pave the way for future research. Steen is, in general, quite sympathetic to conceptual metaphor theory as developed in the ground-breaking work of George Lakoff and Mark Johnson, but he is quite uncompromising in displaying the many open questions regarding its empirical plausibility as a linguistically and cognitively adequate model of metaphor.