When words get physical: evidence for proficiency-modulated somatotopic motor interference during second language comprehension

When words get physical: evidence for proficiency-modulated somatotopic motor interference during second language comprehension Nikola Vukovic (nv254@cam.ac.uk) Department of Theoretical and Applied Linguistics, University of Cambridge Abstract New theories of cognition posit an intimate link between higher cognitive processes and the sensorimotor areas of the brain. In a reaction time-based translation task, second language (L2) speakers responded to action verbs using a microphone or a response pad. A significant interaction among Response Modality, Verb Type, and Proficiency indicated that more proficient L2 speakers took significantly longer to respond with their hands to previously seen hand- related verbs, but not mouth-related ones. Conversely, responding using a microphone led to slower latencies in the case of mouth-verbs, but not hand-verbs. Amidst virtually exclusively monolingual research on embodied cognition, the current study provides evidence that reading L2 action verbs selectively interferes with subsequently performed manual or verbal responses, suggesting that semantic representations of these verbs are distributed over neural substrates underlying action execution. The role of proficiency and experience in language comprehension is discussed. Keywords: embodied cognition, second language, semantics Introduction Ever since the cognitive turn in psychology, the human mind has been likened to a computer, and the essence of our mental life envisaged as the outcome of complex calculation over abstract symbolic elements. In this traditional framework, human cognition is defined as modular, with distinct components operating on information independently and autonomously (Fodor 1998). In contrast to this perspective, recent years have witnessed the strengthening of theoretical paradigms which posit thought as being grounded in experience and sensation. These have become known as Embodied Cognition theories. According to Embodied Cognition, human concepts are not amodal transductions of sensory data, but are instead grounded in sensory-motor processing itself. Much evidence for embodied cognition comes from studies of language processing. Within this new framework, language comprehension is thought of as grounded in, and intimately linked to, neural resources used in action, perception, and emotion (Barsalou 2008). By now there exists a wealth of behavioural research which supports the claim that language comprehension and action execution are subserved by common neural resources. For example, Glenberg and Kaschak (2002) demonstrated an Action-Sentence Compatibility effect, where judging the sensibility of sentences which implied a movement towards or away from the body (You gave Andy the pizza vs. Andy gave you the pizza) facilitated congruent arm movements. In a similar sentence sensibility judgement task, Zwaan and Taylor (2006) found that participants were significantly faster to perform manual rotation of a response knob when the rotation (clockwise or anti-clockwise) was congruent with the meaning of previously presented sentential material (turn the volume up/down). Bergen et al. (2003) asked participants to manually verify names of pictures representing actions, and found that response times were slower when they had to reject actions performed with the same (vs. different) effector. In a different study, Buccino et al. (2005) had participants listen to and judge the concreteness of sentences using hand or feet responses. They observed interference effects pointing toward the conclusion that verbally presented action sentences activate the motor system. These studies suggest that language and action are highly interconnected and that processing action language functionally involves activation of motor representations in the brain. Moreover, this interaction is differentially articulated as facilitation or inhibition, based on the temporal relationship between stimulus and response (Boulenger et al. 2006). The findings from behavioural studies outlined above find additional support in the neurosciences, where experiments have shown interdependence between cognition and simulation of motor and perceptual states. There have by now been numerous studies which demonstrate that semantic processing of a word activates distributed and diverse networks of sensory and motor information (Farah, McClelland, 1991; Damasio 1990; Caramazza et al., 1990). For example, processing the name of an action engages the motor area which is active during performing that same action (Martin et al., 1995 p. 649-652.). Hearing a word activates auditory associations (Pulvermuller et al 2006), and action-related words elicit cortical activation in the motor areas of the brain, even when the participants are not aware of hearing the word (Pulvermuller et al. 2005). Intriguingly, comprehension of action words does not only reliably activate the motor cortex, but does so in an effector- specific i.e. somatotopic manner: face, arm, or leg words activate the corresponding parts of the motor system in the central and precentral cortex which control face, arm, or leg movements (Buccino et al 2005, Hauk et al 2004). Taken together, behavioural and neuroimaging data strongly support predictions and claims of embodied approaches to language and cognition. Semantic representations of words are not amodal or entirely symbolic, but seem to utilise the same neural resources involved in action execution, and it is these strong links which are made apparent in the interactions outlined in the studies above. Some proponents of disembodied and symbolic approaches to cognition, however, have raised concerns