A Semantic Model to Study Neural Organization of Language in Bilingualism
Figure 2
Some snapshots of the model response at some instant of the simulations, performed when the L2 word is given as input to the network. The upper panels describe activity in the feature areas, while the bottom panels represent the activity in the lexical area. The L2 word is represented by neuron activity in position 20, 30 of the lexical area, while the corresponding L1 word is represented by activity in position 5, 5. Results of three simulations are given, characterized by a different proficiency of the second language: beginning of L2 training (a), after a weak L2 training (b), and after a prolonged L2 training (c). It is worth noting that, after weak training, the L2 word can evoke the correct object representation in the feature areas (represented by simultaneous activation of four features), but it still requires activation of the L1 word. After prolonged training the L2 word becomes almost completely autonomous from L1.
