Discussion
Understanding is not simulating: a reply to Gibbs and Perlman

https://doi.org/10.1016/j.shpsa.2010.07.002Get rights and content

Abstract

In this response, I do four things. First, I defend the claim that the action compatibility effect does not distinguish between embodied and traditional accounts of language comprehension. Second, I present neuroimaging and neuropsychological results that seem to support the traditional account. Third, I argue that metaphorical language poses no special challenge to the arguments I gave against embodied theories of comprehension. Fourth, I lay out the architecture of language I advocate and suggest the sorts of data that would decide between traditional and embodied accounts.

Section snippets

What does ELC predict?

Whether the evidence supports ELC depends on what predictions it makes relative to its competitors. G&P charge that I have a mistaken view about what sorts of experiential simulations ELC posits, which in turn leads me to misinterpret the evidence in favor of ELC. In the case of the action compatibility effect (ACE), I argued that it was hard to see why one would expect that sentences describing motion towards oneself would prime a motor response towards oneself, since the natural motion one

Further empirical evidence

As G&P note, recent neuroimaging studies provide converging evidence that seems to support ELC. That there is a link between motor activation and action verbs is indisputable. However, as Willems and Hagoort (2007) say in a recent review, it is difficult to tell whether this shows that motor representations are part of semantic representation proper or part of post-comprehension imagery processes.

Metaphor and nonliteral meaning

I will only make a few brief comments on metaphor and nonliteral meaning. Let me state that I’m not at all committed to the claim that on-line sentence comprehension first involves computing a “context-free, semantic, literal representation” (p. 10). On-line comprehension is always aimed at understanding in a particular context of use, and this purpose is plausibly best served by computing what is said by a speaker making an utterance, rather than the meaning of the sentence (Recanati, 2004).

The structure of the linguistic system

I’ll now briefly state what I take the relationship to be between the linguistic system(s) and the systems responsible for our abilities to produce simulations of various kinds. The TLC architecture is laid out in Figure 1.

Obviously, much detail is being suppressed in this analysis. Language can be either heard or seen, but this figure omits how we determine that auditory or visual input should be interpreted by the linguistic system. Similarly, the internal structure of the language

References (18)

There are more references available in the full text version of this article.

Cited by (1)

View full text