THE CORRESPONDENCE AND CLASS-INCLUSION MODELS²

For useful overviews of the psycholinguistic literature on metaphor processing, I refer the reader to Gentner et al. 2001, as well as the special issue of Metaphor and Symbol 16 (Giora 2001).

³

The classification into correspondence and class-inclusion models is not intended to be exhaustive. For example, the graded salience hypothesis (Giora 1997, 1999, 2003) seems to lie outside this dichotomy in that it is more concerned with demonstrating that the processing of language, both metaphorical and literal, is governed by a principle of salience, such that salient meanings (e.g., meanings that are conventional, frequent, more familiar, or enhanced by earlier contexts) are processed first (Giora 1997:183).

AS DISCOURSE TYPES

Broadly speaking, a correspondence model of metaphor assumes that metaphor processing works by systematically mapping relations present in the source domain onto relations in the target. There are, to be sure, a number of specific variants of the correspondence model, differing in terms of whether they assume that the correspondences are created on-line (Ortony 1979) or retrieved from long-term semantic memory (Lakoff 1993, Lakoff & Johnson 1999), and whether processing is restricted to local interactions between particular pairs of terms (Ortony 1979) or extended across large-scale domains (Gentner 1982, 1983; Tourangeau & Sternberg 1981, 1982). But these differences notwithstanding, what makes these variants instances of the correspondence model is the assumption that metaphor works by connecting source and target in a structurally consistent manner so that particular relations between objects and their properties are preserved. As an example, consider the claim that speakers have a set of mappings that relate the target domain of love to the source domain of a journey, mnemonically referred to as the love is a journey metaphor, so that knowledge about the latter is used to generate inferences about the former (cf. Lakoff 1993). Knowledge about the journey domain (e.g., that to keep on moving is a sign of progress) is systematically mapped onto the love domain, leading to a particular understanding of the latter (e.g., a relationship that is not going anywhere is not progressing). Various other expressions such as Our marriage is on the rocks or This relationship is at a crossroads are claimed to be understood in a similar manner, because of correspondences between the two domains.

The class-inclusion model (Glucksberg et al. 1997, Glucksberg & McGlone 1999) accepts that in processing metaphorical language, the source domain does play an important role in influencing how the target, and consequently the entire expression, are understood. But, instead of relying on correspondences between source and target, the class-inclusion model claims that the source domain is merely treated as a prototypical instantiation of a larger, newly created superordinate category, which is then seen to encompass both the source and target. Because it is a newly created category, the superordinate category cannot be directly named, and hence there is a need to use a prototypical member as an exemplar of such a category. Thus, to understand an expression such as My job is a jail, one presumably treats a jail as a prototypical member of the category of “situations that are unpleasant, confining, etc.” (Glucksberg & McGlone 1999:1543). This is a superordinate category which then allows both job and jail to be seen as members. Thus, the key difference between the two models lies in whether metaphor processing works by establishing relational correspondences or by creating a superordinate category.

A number of psycholinguistic experiments have been conducted to test the two models, with evidence being adduced in support of both sides of the debate (e.g. Gibbs 1992. Gibbs & O'Brien 1990. Glucksberg et al. 1992, Glucksberg & McGlone 1999). As a specific example, consider a study by Nayak & Gibbs (1990) in which students were asked to judge the appropriateness of idioms in various contexts. They were given short narratives that used lexical items that either related to an anger is animal behavior metaphor or an anger is heated fluid in a container metaphor, and they were asked to decide which idiomatic expressions more appropriately continued the narratives. Nayak & Gibbs found that an idiom like blew her top was consistently considered more appropriate than bit his head off if the narrative was presented in terms of the heated fluid metaphor, leading them to conclude, according to a later critique, that “readers not only have relevant analogical information available, but that readers also use this information to facilitate idiom comprehension” (Glucksberg & McGlone 1999:1551). However, Glucksberg & McGlone suggest that Nayak & Gibbs's data can be interpreted differently. First, it is possible that the appropriateness ratings do not reflect ease of comprehension; rather, they might simply result from a post-comprehension judgment process. Second, the data might not even indicate deliberate choices on the part of the participants. The ratings might be due to “simple lexical priming rather than stylistic consistency per se,” so that even if there were no conceptual metaphors involved, the semantic association between the lexical items in the narrative and those in the idioms might have influenced the participants' responses. Glucksberg & McGlone therefore conclude that Nayak & Gibbs did not provide any support for the correspondence model.

Because there is at present no consensus regarding the evidence from psycholinguistic experiments, it has been suggested (Gibbs 1999:151; Glucksberg et al. 1992:578) that it might be more fruitful to look at the kinds of circumstances that may favor one model over another.⁴

It has been pointed out by a number of analysts (Caballero 2003, Cameron 1999, Goatly 1997) that the emphasis on metaphor as a cognitive phenomenon (metaphor “in the head”) has, unfortunately, tended to overshadow the importance of approaching it as a discursive strategy (metaphor “in society”). Caballero (2003:147), for example, notes that even in analyses of professional communication that attempt to explore the role of figurative language, little has been said about the part played by metaphor in the management of discourse, with the exception of works such Low 1997, Drew & Holt 1998, and Moon 1998. Caballero goes on to suggest that by approaching metaphor from a “genre vantage,” we may be able to “go beyond the level of lexis (i.e. metaphor's contribution to disciplinary jargon) to place the emphasis on how metaphor may reveal the world-view of a community, fulfill several rhetorical purposes, and contribute to the unfolding of text according to the genre's conventions” (2003:148).

Rethinking the class-inclusion and correspondence models within a discourse-analytic perspective can make an important contribution toward the goals mentioned by Caballero. I do not intend to imply, however, that attempts to establish the psycholinguistic status of the models have been misguided; rather, while waiting for the psycholinguistic dust to settle, these two models, I believe, have much to offer to the discourse analysis of metaphor. Rethinking these models as discourse types governing the use of metaphor means that we are less concerned with the extent to which they exist as processing mechanisms than with the extent to which they are socially institutionalized discursive strategies. This also means that, rather than being concerned with whether one processing mechanism is used in exclusion of another, we can acknowledge that both discourse types are potentially relevant to any given piece of text and talk. The more sociologically interesting question lies in whether one particular discourse type can be considered more dominant,⁵

STRATEGIES OF RECONTEXTUALIZATION AND ACTIVITY TYPES

Linell offers the following definition of recontextualization:

Linell, in fact, goes on to note that metaphor, “which concerns the transfer of meaning across contexts, is a notion which is obviously related to ‘recontextualization’” (1998:155, note 7). I take this observation seriously in this article, because the use of metaphor in discourse necessarily involves utilizing such “transfer of meaning across contexts” in the service of discursive goals, where concepts or even ways of speaking typically associated with one particular domain are being employed in the context of a different domain. Thus, metaphor is not merely related to recontextualization; it is, by definition, a form of recontextualization. But what exactly does metaphorical discourse recontextualize?

As Sarangi reminds us, recontextualization involves an element of decontextualization, since events and entities are separated from “their original loci of occurrence and their indexicals thereby cut off from the elements they had originally indexed” (1998:306). In the case of metaphorical discourse, because a metaphor relates a source to a target, the simple answer is that metaphorical discourse recontextualizes the source, since it is the source that has been imported from some other context (decontextualized) into the prevailing discourse (recontextualized). This is not to deny that the target is also changed by virtue of being related to the source; after all, most metaphor theorists follow Black 1962 in acknowledging that some kind of “interaction” takes place between source and target. But as far as the analysis of discourse is concerned, the target crucially constitutes the ongoing context, and it is the source whose relevance to the target needs to be established. For example, in (4) below, the author describes audience behavior at different concerts in order to explain the properties of subatomic particles. Readers know from the outset that the text is about particle physics, and hence, when an apparently unrelated source (audiences at concerts) is imported into the prevailing context, its relevance to the current discussion needs to be made clear. In this sense, it is the source that gets recontextualized in metaphorical discourse. But, as I show below, the precise nature of this importation – the specific strategy of recontextualization – will differ depending on whether the discourse type involves the correspondence or class-inclusion model.

Furthermore, the recontextualization strategies are themselves very much dependent on the kind of activity type that they are part of. An activity type, according to Levinson, refers to “a fuzzy category whose focal members are goal-defined, socially constituted, bounded, events with constraints on participants, setting, and so on, but above all, on the kinds of allowable contributions. Paradigm examples would be teaching, a job interview, a jural interrogation, a football game, a task in a workshop, a dinner party, and so on” (1992:69, emphasis in original; see also Sarangi 2000). Activity types are important because they draw our attention to the question of how the structure of the activity constrains, in the case of language, the kinds of text and talk that can or should be employed during the conduct of the activity.

A good example of how the activity type can constrain recontextualization comes from Scheuer's (2001; 2003) study of Danish job interviews. Scheuer demonstrates that the “felicitous” job interviews (those that lead to successful outcomes for the applicants) depend significantly on how the applicants are able to recontextualize their lifeworld resources, such as leisure activities and personal interest in topics such as economics and political science, within a professional setting like the job interview (2001:235). According to Scheuer, the interview committees tend to apply a teleological discourse in which “professional fields and careers are goals chosen and consequently aimed at by agents … [so that] (p)rofessional dispositions are constructed as general orientations running through a person's life” (2001:234). Successful applicants were those who were able to present their lifeworld experiences as career-oriented choices, while unsuccessful ones were unable to recast their personal choices as career-oriented.

Scheuer is not particularly concerned with metaphorical discourse. But if, as suggested above, metaphorical discourse is a form of recontextualization, then it would follow that the specific manner in which metaphor is discursively employed might also be constrained by the relevant activity type. Summing up, the framework we have arrived at can be presented thus:

The next section applies these ideas to specific texts and shows that there do in fact appear to be interesting associations between particular activity types and particular models of metaphorical discourse.

TWO DISCOURSE TYPES, TWO ACTIVITY TYPES

Consider the activity of providing practical advice or tips, and contrast this with the activity of explaining concepts that may be fairly technical or abstract. While both may be forms of teaching, they differ in important ways. In the former the goal of the activity is performance-oriented, and the addressee, if things go well, acquires an ability to do certain things. In the latter the goal of the activity is perhaps more intellectual, so that the addressee comes to have a better appreciation of particular ideas. Although both types of activities may make use of metaphors to facilitate their conduct, evidence from the following texts suggests that the metaphors are used in systematically different ways. As an example of the former, I consider the activity of providing practical advice on management; as an example of the latter, the activity of explaining scientific concepts to a lay reader. I begin with the latter.

Explaining concepts in popular science texts

I analyze extracts from two books: John Gribbin's In search of Schrödinger's cat: Quantum physics and reality (1984) (henceforth, Text A), and Richard Dawkins's The blind watchmaker: Why the evidence of evolution reveals a universe without design (1986) (henceforth, Text B). As exx. (1) and (2) show, the authors are clear that, in writing these books, they aim to explain various scientific concepts to a non-expert reader. (1) and (2) serve to frame (Goffman 1974) the activity type that the texts are part of; that is, they are the authors' own statements of “what is going on, what the situation is, and the roles which the interactants [i.e. the authors, and by implication, the readers: LW] adopt within it” (Drew & Heritage 1992:8).

These texts, then, are more concerned with conceptual clarification than with dispensing practical advice. To see how metaphor is discursively employed in these texts, consider (3). In (3), Gribbin tries to explain Rutherford's conception of atomic structure, and in particular the vastness of the space that is present within an atom. To do this, he uses a metaphor of a pinhead situated within St. Paul's Cathedral, and asks the reader to imagine the distance between this pinhead and dust motes located at the far end of the cathedral. As Gribbin himself points out, “The pinhead represents the atomic nucleus; the dust motes are its retinue of electrons.” Thus, Gribbin is explicitly appealing to a correspondence discourse type where the reader is expected to map the pinhead and the dust motes onto the nucleus and the electrons respectively. This is crucial because only by establishing these correspondences can the metaphor do the work intended by Gribbin: to allow the reader to appreciate how much space exists within an atom. Of course, it may be possible, in support of the class-inclusion discourse type, to argue that both the source and the target are simply members of the category “things with lots of space inside them,” but this misses the point of the discourse, which is to convey not merely the fact that a large amount of space is present, but specifically the amount of space that intervenes between an atomic nucleus and its electrons.

In (4), to explain the difference between subatomic particles such as fermions and bosons, Gribbin makes use of differential audience behavior at different concerts. He likens orderly audience behavior to that of fermions, and the rowdier audience behavior at a rock concert to that of bosons. Note that for this metaphor to work, once again, specific correspondences must be established: Each individual audience member corresponds to a particle (either boson or fermion), and specific locations in the concert venue (seats/the area in front of the stage) correspond to particular quantum states.

Finally, in (5), to explain how some kinds of macromutation are not as complex as they may first appear to be, Dawkins describes how an airliner, the Stretched DC8, is built. While Dawkins begins by comparing the building of a Stretched DC8 with the concept of macromutation in general, it soon becomes clear that the choice of a Stretched DC8 has an even more specific motivation. Dawkins uses the snake as an example of macromutation in animals, and here, the lengthening of the fuselage that results in the creation of a Stretched DC8 is specifically compared to the adding of segments to the body of a snake. Such specificity of comparison makes use of the correspondence model: The engineering instructions required for building a Stretched DC8 correspond to the evolutionary changes that led to snakes becoming longer than their ancestors.

Of course, it would be possible, along the lines of the class-inclusion discourse type, to treat both snakes and Stretched DC8s as members of the same superordinate category (whatever this might be), but to do this would be to misconstrue the discursive purpose of the metaphor, which is to establish specific point-by-point correspondences between source and target so that the reader has a better understanding of the details of the latter. The same argument applies to all the other examples from Texts A and B. In the case of (4), for example, a reader who insisted on constructing a single category that contains both behavior at concerts and subatomic particles would be missing the point of the text. Such category construction is not impossible, but surely irrelevant given the activity type.

The metaphors used in the examples above thus tend to make use of the correspondence rather than the class-inclusion discourse type. Notice also that the status of the discourse as metaphorical is foregrounded (Goodwin & Duranti 1992:11) in a number of ways. By “foregrounded,” I mean that the authors sometimes make use of “contextualization cues” (Gumperz 1982, 1992) to signal to their readers that they are appealing to metaphors. This can be seen in the use of the words like imagine (3) and analogous (5). Another sign that the metaphors are being foregrounded can be seen in how careful the authors are to keep the source and target domains as distinct as possible. Keeping the source and target distinct follows from the activity type of conceptual clarification, where the goal is for the reader to better understand the target concept; the source is merely a convenient means toward this end, and it must not become conceptually confused with the target. As mentioned already, in (3), Gribbin uses the verb represents (The pinhead represents the atomic nucleus) to help the reader establish the intended correspondences. This simultaneously serves to remind the reader of the distinction between source and target, since to appreciate that something represents something else is to maintain a distinction between the two. In (4), the use of quotation marks (‘particles’, ‘energy state’) performs the same function of keeping source and target distinct by reminding the reader that these terms are not being used literally. And in (5), Dawkins even uses a rhetorical question with the adjective real (What has this to do with real animals?), thus acknowledging that his discussion of airliners (which are obviously not animals) constitutes a very different topic from that of biological mutation.

Explicit contextualization cues are, however, not always necessary. As Levinson reminds us, the activity type crucially helps “to determine how what one says will be ‘taken’ – that is, what kinds of inferences will be made from what is said” (1992:97). For example, an utterance such as How are you? as spoken by a doctor in a clinical setting would be interpreted by a patient as a request for information about the patient's health. The converse, where a patient opens the interaction with How are you? and the doctor responds with information about his/her (i.e. the doctor's) own health, is not impossible, but unlikely in the context of the clinical setting. Thus, How are you? derives its meaning from the activity type in which it is embedded (Sarangi 2000:2). In the same way, readers of Texts A and B are expecting to be enlightened about the nature of fairly technical concepts, and any discussion of events outside the domain of quantum physics or evolutionary biology will be interpreted as a metaphor intended to contribute toward this end.

Finally, it is worth pointing out that the use of metaphors in Texts A and B is optional and sporadic; it is up to individual authors to decide whether the explanation of a particular concept will benefit from the use of a metaphor. And precisely because the use of metaphor is sporadic, we see a variety of different metaphors being used, depending on the particular concept that an author is attempting to clarify. Dawkins makes use of Stretched DC8s, while Gribbin utilizes cathedrals and concerts. All this follows from the activity type under consideration. Because the texts are concerned with explicating a range of concepts, a variety of metaphors are called upon as and when the authors feel that this will facilitate the reader's understanding. Bearing these observations in mind, we can now move on to texts from a different activity type. As we will see in our discussion of management texts, the metaphors there make use of the class-inclusion rather than the correspondence discourse type.

Practical advice in management texts

The extracts in this section come from two different books on management: Ken Blanchard and Don Shula's The little book of coaching: Motivating people to be winners (2001) (henceforth, Text C), and Chris and Reina Komisarjevsky's Peanut butter and jelly management: Tales from parenthood, lessons for managers (2000) (henceforth, Text D). These books are part of a well-established genre of management texts which attempt to provide insights by drawing upon experience in a variety of areas not traditionally considered part of the corporate world, as the authors themselves acknowledge (see exx. 6 and 7). Thus, in (6), although the authors of Text C acknowledge that football and business are “different,” they claim that “similar” strategies apply “whether it's sports or business.” And in (7), the authors of Text D suggest that what we do “at home” can teach us important things about how to lead adults “in the workplace.” As do (1) and (2) above, (6) and (7) serve to frame the activity type that Texts C and D are part of. Notice that unlike (1) and (2), (6) and (7) emphasize the practical relevance of the information in Texts C and D. (6), for example, asks rhetorically what the reader would be able to “use” in order to win, while (7) points out that its contents are “not theoretical,” but “practical.”

There are clearly metaphors at work in these books; each draws upon domains of experience or understanding that are typically considered distinct from the domain of management in order to provide what are hoped to be unique insights. To see that these texts employ the class-inclusion rather than the correspondence discourse, recall that in the former, the source is treated as a prototype of a superordinate category which is then understood as subsuming both source and target as specific instantiations. Both Texts C and D describe events or activities from the source domain (football and parenthood, respectively), and from these they extract a number of principles that are claimed also to be applicable to the workplace. Hence, each source event/activity is presented as an exemplar of a general principle (i.e., the superordinate category) which can also be realized in the target domain of management. These principles are the practical advice or tips that form the main selling points of the books. The importance attached to these principles is typographically indicated. In each book, the principles appear in bold so that attention is drawn to them. Additionally, Text C allows each principle to occupy a complete page all on its own, while Text D presents its principles as headings. As an example, consider (8):

In (8), the principle being emphasized is Make sure everyone knows what the target is. It presents an activity from coaching – creating practice systems and procedures that made clear to the athletes what was expected of them – to exemplify a superordinate category, realized here as a principle. This principle is presented to the reader as one that is also relevant to the world of management. Thus, both coaching and managing, it is claimed, require the coach/manager to make sure that everyone knows what the target is. This is the class-inclusion discourse type at work. Incidentally, recall that in neither Texts A nor B were any general principles being established.

It is not exactly clear how the principle in (8) is to be applied to the world of management. This is perhaps necessarily so, since unlike the popular science texts, which were intended explain the complex structure of particular concepts, establishing specific correspondences between source and target is not the main point here. Whether or not individual readers find it useful to treat specific coaching activities as analogues of specific management techniques (thus making use of the correspondence discourse type), all readers are expected to go on and appreciate that activities in both football and management can be seen as instantiations of the principle.

Aside from the stating of the principle, there are other linguistic cues that point to the class-inclusion discourse type at work. Because the principle is intended to subsume both football and management activities, the vocabulary makes use of superordinate terms such as target and everyone. Similarly, consider the use of pronouns in the last two sentences of (8). Pronouns and demonstratives in the penultimate sentence clearly refer to individuals in the world of football. I in Everything I did… refers to the coach, while they and that target in so they could aim for that target refer to the players and their goal of winning football games. However, you and do that in the very next sentence, And you do that one day at a time, show a switch to the world of the reader (presumably someone interested in management). This switch helps to blur the distinction between the world of football and that of management, which is exactly what we might expect if the goal is to extract principles of sufficient generality that they apply to both source and target. Again, this is an important contrast with the popular science texts, where keeping source and target distinct was absolutely crucial.

(9) and (10) provide further data in support of the observations just made. With (9), a coaching activity (blowing the whistle) is one way of indicating when the team or a particular player has done something wrong. But whether blowing the whistle specifically corresponds to some analogous activity in management is less important than appreciating that some way of realizing the general principle (There is no such thing as a minor mistake) can and must be found:

In (10), the use of the class-inclusion discourse type is perhaps even more obvious, since the final paragraph makes clear that the principle (The courage to let go) is exemplified not just in the world of parenthood (Whether you are letting go of your child …), but also in pilot training (to let the student fly alone …) and management (or send one of your staff to that all important business meeting alone …). The list is, of course, open-ended; one can imagine a host of different activities where the courage to let go is relevant.

Unlike the metaphors in Texts A and B, the metaphors in Texts C and D are more backgrounded. They are backgrounded because attention is not specifically drawn to their status as metaphors; in fact, in the attempt to highlight the shared general principles, there is a tendency to blur the distinction between source and target domains. And unlike Texts A and B – where a variety of different metaphors are used as and when the authors felt there was a need for them – Texts C and D are each dominated by specific metaphors (with football as the source in the case of Text C and parenthood as the source for Text D) that run throughout the books. Furthermore, it is arguable that this use of metaphor has become institutionalized as a subgenre of management self-help books, so that authors of such books are expected to provide management insights by looking beyond management (that is, via metaphor). For example, in addition to Texts C and D discussed here, there are attempts to extract management tips from the world of music (Zander & Zander 2000), literature (Whitney & Packer 2000), and religion (Briner 1996), among many others. In other words, unlike Texts A and B where authors were free to use metaphors or not depending on whether they thought it would help in conceptual clarification, in Texts C and D, the genre itself demands that the texts be metaphorical.⁶

This is not to imply that the genre that Texts A and B belong to is in any way less institutionalized than that of Texts C and D. Myers (2003:267), for example, suggests that the popularization of scientific ideas is a “routinized social activity that has led to the creation of a number of fairly stable genres.” What is interesting, though, is the way in which metaphor is differently institutionalized in popular science and management texts.