| Summary: | The purpose of the present study is to test whether symbol grounding for action verbs occurs in entrenched native verb forms, and whether they transfer to novel verb forms that are acquired as explicit translations of existing verbs. The entrenched and novel verbs were referred to here as L1 verbs and L2 verbs respectively, and were used as analogs of meanings in first and second language learning. Symbolic grounding was investigated by observing behavioral data in lexical decision tasks, and scalp electrophysiological signals using 128-channel EEG data.
The present study used different kinds of action verbs (e.g., verbs of foot movement, such as kick or jump, and verbs of hand movement, such as swipe or grab) and abstract verbs, such as learn or plan. Previous researchers have provided empirical evidence showing that when action verbs are accessed in reading, there is concomitant activation of primary motor and/or somatosensory cortex. The established relationships of action verbs in L1 and their sensorimotor groundings as a reference were used to determine successful transfer of groundings of L1 words to L2 words. By observing the responses to the L2 words that are acquired through symbolic manipulation without perceptual or bodily experiences and examining, it can be determined if they produce similar neural activations as in those found in L1 words, and we can test whether the symbol-grounding-transfer occurs in part or in whole, given this minimal learning context.
The behavioral measure was a lexical decision task where the participants respond to meaningful words (foot-related verbs or hand-related verbs) with two kinds of response modalities (button press with a finger or foot pedal press with a foot). Although either facilitation (foot verb to foot movement and hand verb to hand movement) or interference (Foot verb to hand movement and hand verb to foot movement) effect of action verbs was expected, the results showed that the participants consistently responded faster to the L1 English verbs than to the L2 verbs and responded faster with finger-pressing button box than foot-pressing pedal. However, at the slowest response times condition, the condition of foot-pedal pressing to L2 words, the facilitation effect of Foot related verbs was observed. The response times of foot pedal pressing to L2 Foot-related verbs were significantly faster than both L2 Hand-related verbs (p=.003) and abstract verbs (p=.005) at the paired t-test. This result is consistent with the research hypothesis and provides partial evidence supporting the assumption that the Foot-related action verbs have close link with sensorimotor cortex associated with foot movement and reading those verbs will facilitate corresponding body movement.
The three kinds of EEG data analysis methods were used in the current study: Event Related Potential (ERP) component analysis, EEG topographic analysis, and EEG source localization with low resolution brain electromagnetic tomography (LORETA). The ERP components were used to examine the effect of language (L1 vs. L2) and lexicality (Word vs. Non-word) in terms of amplitudes and temporal points of ERP components. The EEG topographic analysis and EEG source localization with LORETA are methods for spatiotemporal analysis, which provide information on intracranial neural activations that are sources of scalp electric signals.
When ERP components of the montaged electrodes placed on the central area of the scalp (vertex and neighboring 14 electrodes) were examined, the P3 component for L1 (at around 390~400ms) reached its peak earlier that that of L2 (495~505ms). Topographic analysis results that compared topographic maps created by different verb groups provided evidence that different configurations of the neuronal activations were created by the verb groups representing body movements of different body parts. In addition, by means of the source localization analysis with the LORETA, the differential neural activations at sensorimotor cortex were observed when the brain activations elicited by L1 Foot related and Hand related verbs were compared. At both temporal windows, early (126~134ms) and late (318~322ms), the regions of the sensorimotor cortex associated with Hand movement were activated significantly more by the Hand related verbs than Foot related verbs. In order to test Harnad's symbol grounding transfer hypothesis, the neural activations at the sensorimotor cortex elicited by L2 Foot verbs and the L2 Hand verbs were examined in comparison with those of the abstract verbs. Significant larger activations were elicited by both of the action verbs over the abstract verbs. To summarize, the current study provided the neurophysiological evidence on the symbol grounding at L1 word and the symbol grounding transfer at L2 words by exhibiting links between the regions of sensorimotor cortex and L1 and L2 action verbs in terms of differential neural activations elicited by the verb groups.
|
|---|
