Sentence Understanding Engages Motor Processes

Sentence Understanding Engages Motor Processes Benjamin K. Bergen (Bergen@Hawaii.Edu) Kathryn B. Wheeler (Kwheeler@Hawaii.Edu) Department of Linguistics, 569 Moore Hall, 1890 East West Road, Honolulu, HI 96822 USA Abstract running a mental simulation of a described scene, wherein an understander activates motor representations corresponding to what participants in the scene might do and perceptual representations of images they might perceive. The reasoning behind this is straightforward. Perceptual and motor imagery are known to play critical roles in higher cognitive functions like memory (Wheeler et al. 2000, Nyberg et al. 2001). By extension, they may well be integral to language understanding as well. On this view, viable communication involves the successful evocation of an aligned set of perceptual and motor images in the mind of a hearer (Glenberg & Robertson 1999, Bergen & Chang 2005). Recent evidence from neuroscience suggests a mechanism by which the internal imagination of a scene described by language may be effected. Two studies provide evidence that processing motion language associated with particular body parts also results in the activation of areas of motor and pre-motor cortex involved in producing motor actions associated with those same effectors. Using both behavioral and neurophysiological evidence, Pulvermuller et al. (2001) found that verbs associated with different effectors (mouth, hand, leg) were processed at different rates and in different regions of motor cortex. More recently, Tettamanti et al. (m.s.) have shown through an imaging study that passive listening to sentences describing mouth versus leg versus hand motions activate different parts of pre-motor cortex (as well as other areas, specifically BA 6, BA 40, and BA 44). From a broader perspective, to the extent that evidence is found indicating that imagery plays a role in language understanding, this bolsters an embodied view of meaning, in which the particular experiences a language user has had in their life, in their body, create the substantive basis for language production and understanding, as suggested by a number of authors, like Barsalou (1999), Zwaan (1999), Glenberg and Robertson (2000), Bergen, et al. (2003), Feldman and Narayanan (2004), Bergen and Chang (2005), and Matlock (To Appear), and MacWhinney (In Press). While important and productive lines of research are beginning to gain momentum in the area of mental simulation and language understanding, there has been insufficient focus thus far on the exact linguistic mechanisms that trigger mental simulation. More specifically, there are two questions that have yet to be addressed in previous studies. The first concerns the types of language that trigger simulation and the second concerns the degree of detail involved in simulation. It is of critical theoretical importance to determine how prevalent language-triggered motor simulation is – whether it occurs only when people process language about Processing sentences describing actions performed by the hearer (e.g. You gave Andy a pizza) primes actually performing a motor action compatible with the one described (Glenberg & Kaschak 2002), This result has been interpreted as indicating that processing meaning involves the activation of motor control circuitry to prepare the language understander for situated action. A complementary view (Bergen & Chang 2005) argues that motor control structures are automatically and unconsciously engaged during motion language processing irrespective of whether that language pertains to the understander, because the meaning of action language is inherently grounded in motor control circuitry. To distinguish between these views, we must determine whether motor activity results from processing language about actions performed by and on third persons (e.g. Andy gave Sara a pizza.) Two experiments tested the scope of compatible action facilitation during sentence understanding, where sentences either encoded motion in a particular direction (e.g. John opened the drawer) or using a particular handshape (e.g. Mary grabbed the marble.) Both studies yielded significant Action-sentence Compatibility Effects, demonstrating that whether language is about the understander or not, it engages their motor system. Keywords: sentence processing, mental simulation, motor control, Action-sentence Compatibility Effect Introduction Language use integrally involves perceptual and motor processes, since the production of language requires the control of facial, manual, and other effectors, and language processing begins with the detection of visual, auditory, and other perceptual cues. However, language appears to engage perceptual and motor systems and the neural structures dedicated to them in another, less obvious way. Just as performing imagery about action (Porro et al. 1996, Lotze et al. 1999) and perception (Kosslyn et al. 2001) makes use of action and perception systems, so it now appears that processing language about perceptual or motor content results in the activation of neural structures overlapping with those that would be used to actually perceive or perform the described content. The mental (re)creation of perceptual and motor experiences (among others) in order to deeply understand language goes under the rubric of mental simulation (Barsalou 1999). The notion that language understanding makes significant use of perception and action imagery has been proposed in a variety of contexts (Barsalou 1999, Zwaan 1999, Glenberg & Robertson 2000, Feldman & Narayanan 2004, Gallese & Lakoff 2005, Bergen & Chang 2005). What these various views share is the idea that language understanding entails