Three dogmas of embodiment: Cognitive linguistics as a cognitive science
A brief history of embodiment in Cognitive Linguistics
Recent formulations of the embodiment hypothesis
In their recent collaboration in Philosophy in the Flesh, Lakoff and Johnson have turned much of their attention away from investigating the socio-cultural dimension of embodiment and toward investigating how the physical and neural substrates of the body shapes language and mind, although they would certainly argue for the importance of continued research on the cultural and social dimensions.1 Lakoff and Johnson currently formulate the embodiment hypothesis:
The embodied-mind hypothesis therefore radically undercuts the perception/ conception distinction. In an embodied mind, it is conceivable that the same neural system engaged in perception (or in bodily movement) plays a central role in conception. That is, it is possible that the very mechanisms responsible for perception, movements, and object manipulation could be responsible for conceptualization and reasoning. (Lakoff and Johnson 1999: 37–38)
In this version of the embodiment hypothesis is that the same neural mechanisms which are responsible for ‘lower-level’ activities like perception and movement are taken to be essential to ‘higher-level’ cognitive abilities, namely to our reasoning and conceptualization. Thus, on their view Lakoff and Johnson argue “that the very properties of concepts are created as a result of the way the brain and body are structured and the way they function in interpersonal relations and in the physical world” (1999: 37). The way these properties are created is by means of conceptual metaphors which project cross-domain image-schematic patterns, which in turn are drawn from the more specific structures within visual perception, locomotion, object manipulation, and so on.
In their view, most of the evidence they have gathered earlier is at the ‘top levels’ of investigation– studies on categorization, metaphorical structuring, historical semantic change, gesture and so on. They argue that what remains to be done is the project of establishing the specific neural and physiological bases of image schematic conceptualization. In Philosophy in the Flesh, however, they directed their attention primarily to the neurocomputational modeling of language rather than its neurophysiological bases. For example, Regier (1992, 1996) has investigated how spatial relations terms such as up, down, above, etc. can be learned by structured connectionist networks that utilize low-level schematizations which have plausible neural analogues in the neuroanatomy of visual perception. Together with related research (Bailey 1997; Narayanan 1997) that approach to the neurocomputational modeling of language eventually came to be known as the Neural Theory of Language (Lakoff and Johnson 1999: 569–583; Feldman and Narayanan 2004).
Another important, if underspecified, facet of the embodiment hypothesis appears Philosophy in the Flesh as a new theory of “primary metaphor” (Lakoff and Johnson 1999: 45–59). Primary metaphors are posited as resulting from repeated sensorimotor “primary scenes” in a child’s early experience. They are acquired first as conflated linguistic wholes and only later differentiated into source and target domains (Grady and C. Johnson 2002).2 In older children and adults, primary metaphors are supposedly combined into complex conceptual metaphors via conceptual blending (Grady 2005a, 1997). While the attempt to integrate developmental issues is laudable, there are a number of unresolved questions in the theories of primary metaphor and conflation, including the uncertain and complex relationship between image schemas and primary metaphors– two theoretical constructs that are both supposedly preconceptual and underlie conceptual metaphor mappings (see Grady 2005b: 45–50.). In contrast to the developmental evidence work on image schemas (see summaries in Rohrer 2005: 174–175, Mandler 2005), there is also a dearth of experimental support for both the experiential conflation followed by the emergence of primary metaphors hypothesis and for the notion of primary metaphors as atomic units of meaning that combine into complex conceptual metaphor. Future studies, however, may remedy such deficiencies.
4. For example Johnson and Rohrer (in press) discuss the social and cultural dimensions of embodiment in detail, arguing that they are continuous with the physical and neural substrates. Virtually all such attempts to place embodied cognition within an evolutionary context begin with the physical and neural and trace their way up to the social and cultural. This habitual mode of presentation is perhaps a continuing source of the mistaken dogma that embodiment leads to eliminativism.
5. The conflation hypothesis was arrived at by Christopher Johnson analyzing language acquisition in a well-known corpus of a single child, Shem (Johnson 1997, 1999). Single subject data, while empirical, is of questionable value standing alone. The conflation hypothesis is also at odds with many known facts of the initial development of neural structures; however, it is possible that at a later stage of cognitive development children might find it difficult to differentiate image schematic features which are well-integrated in the underlying cross-modal maps. If found, such evidence would certainly lend credence to Zlatev’s (in press, 2005) hypothesis that language socialization is required for image-schematic thought.
