Three dogmas of embodiment: Cognitive linguistics as a cognitive science
Collaborative conclusions: How to exorcise the three dogmas
There can be no doubt that most of the ordinary research tools of Cognitive Linguistics are at the communicative and cultural level of analysis, the level whose undeniable phenomenological components perhaps most lends itself to two standard critiques. First, there is a methodological question about how one is to go about doing the analysis, for example, as to how one knows when a conceptual metaphor is correctly mapped; second, there are legitimate questions as to how empirical such methods are. While the objection that such analyses are not empirical at all can be at least partially rebutted by the argument that the sentences or phrases were empirically gathered from naturalistic contexts, even beginning to answer the first objection is not so easy. The problem is that an appeal as to what counts as a grammatically well-formed sentence, a plausible construal in cognitive grammar, or a correct conceptual metaphor mapping seems to be little more than an introspective gut-check by the practitioners themselves. Of course, one might respond that there is a body of previous work to which a new analysis must either conform, make fine small distinctions from, or challenge and then reorganize satisfactorily. For instance, conceptual metaphors, whether gathered ad hoc (as in Lakoff and Johnson 1980) or systematically gathered from a defined corpus (as in Rohrer 1995), are supposed to yield relatively stable conceptual mappings. However, both in my experience teaching conceptual metaphor analysis in seminars to both advanced students and scholars from other disciplines, and in a careful reading of the conceptual metaphor literature, there appears a wide variety as to how the source and target domains of the same of a conceptual metaphor are named, or as to the exact counterparts or order to the conceptual mappings of the metaphor. In its strongest form, this objection is that such methods are not just non-algebraic, but almost entirely ad-hoc.1
Within Cognitive Linguistics I see my colleagues make three types of responses to these metatheoretical critiques. The first is to ignore such criticisms and continue to do the work, using the methodologies as they are, perhaps continually refining them according to the community aesthetic2 of what counts as a methodological innovation, or perhaps improving them incrementally importing methods from the related social sciences such as inter-coder reliability ratings for corpus analysis, and so forth. A second approach is to confront these questions using a philosophical defense of the importance of sociocultural factors or the irreducibility of consciousness to the methods of Cognitive Linguistics.
For example, Zlatev’s arguments (in press) have the following philosophical shape: first, language is intrinsically a social behavior for communicating; second, language is intentional, in the sense that every word is a communicative act consciously intended by the speaker to share or convey some experience with (or to) the hearer; third, language must be a triadic communicative system, which is to say that the meanings of its ‘signs’ are consciously conventional across speakers in a population of n >= 3 speakers; ergo, all judgments (by either laypeople or linguists) about whether a sentence is grammatical, or whether the grammar cognitively fits the situation, or whether the a term has been used in an appropriate semantic context are intrinsically a conscious act. Other theorists have staked out similar philosophical positions, arguing that it is primarily sociocultural factors of language acquisition that shape spatial relation schemas (for example, Bowerman and Choi 2003; Sinha and Jensen de Lopez 2000), or arguing that the historical context determines metaphor mappings (Geerarts and Grondelaers 1995).
The third response, which I would take to be the standard response of Cognitive Linguistics, is to cite and discuss “converging” bodies of evidence for embodiment from the other cognitive sciences. It is easy to see how such evidence could be “converging” under an eliminativist rubric, because Cognitive Linguistics and the other cognitive sciences would then “converge” on an underlying neurophysiological reality. But methodological pluralism implies that these methodologies are independent and that there is no guarantee that their results can be integrated. If we are not converging on the same ontologically real “material,” as eliminative materialism would have it, then “what” are we converging upon? The answer, I think, is a guiding idea or vision: embodiment. Our convergence is more on a set of theoretical assumptions and methodological biases than any findings.
Recent years have witnessed numerous attempts (Hutchins 1995; Lakoff and Johnson 1999; Gibbs 1994; Clark 1998; Fauconnier and Turner 2002) to retell the history of cognitive science so that that an embodied vision of cognitive science can both incorporate past labors and findings and shape new research questions that might bridge the past with the future. While attempts at such grand unifications are laudable, the persuasiveness of the case for convergence likely depends on whether and how accurately the authors have described the evidence gathered outside their own traditions, and whether the conceptual leaps across these traditions are kept suitably small. If not, we are still faced with the possibility that the “convergence” of the evidence will be construed as a myth and the evidence will “turn out to be very heterogeneous” (Zlatev in press: 19), creating unbridgeable conceptual chasms between rival theoretic monasteries.
Yet there is still more that cognitive linguists can do. It is now, more than ever, particularly important for cognitive linguists to reach out to the other cognitive sciences. If we set out to design collaborative research projects that are deliberately cross-methodological, we will no longer be forced to rely on evidence gathered from disparate traditions. We have already begun to see movements in this direction. For example, within psychology, Matlock, Ramscar and Boroditsky have investigated whether spatial metaphors for time alter our inferences about time (2005). Matlock and Richardson (2004) have investigated whether fictive-motion language induces eye-movements. Amorim, Isableu and Jarraya (in press) have shown that adding a cylindrical “head” to the standard Shepard-Metzler cubes used in mental rotation/mental imagery experiments produces both visual and motoric embodiment effects. Within cognitive neuroscience, Coulson and van Petten (2002) have investigated the neural processing of conceptual blends using event-related potentials (ERPs), while Rohrer (2001b) has investigated whether the sensorimotor cortex responds to both literal and metaphoric body-part sentences using both ERPs and functional magnetic resonance imaging (fMRI). But more explicitly cross-methodological efforts are needed, particularly for testing within experimental developmental psychology (see for ideas Mandler 2005) and in primatological studies of sign language use (see for ideas Tomasello, Savage-Rumbaugh and Kruger 1993 and Fouts, Jensvold and Fouts 2002). Only when Cognitive Linguistics takes its hypotheses to the other cognitive sciences and develops collaborative, cross-methodological studies will cognitive linguistics become not just a listening but a speaking member of the family of cognitive sciences.
1. In this sense, research in computational modeling such as that of Veale (this vol.) can serve as a useful methodological precision check.
2. By “community aesthetic” I mean the constellation of commitments of the scientific community as it is described by Thomas Kuhn in his 1969 postscript to The Structure of Scientific Revolutions (Kuhn 1970).
