ERP decoding shows bilinguals represent the language of a code-switch after lexical processing

Background. For decades, research using ERPs has revealed how and when comprehenders respond to unexpected linguistic material. For example, N400 effects often occur after hearing an unexpected word, e.g., I like my coffee with cream and salt. N400 effects can also occur after hearing an unexpected switch into another language or code-switching, e.g., I like my coffee with cream and azúcar (sugar in Spanish). In a recent study, Yacovone and colleagues (2019) tested whether or not these two N400 effects are functionally distinct. To do this, they used spoken English stories with target words that varied in language (English, Spanish), contextual fit (Strongfit, Weak-fit), or both. They reasoned that, if there were two distinct N400 effects, the weak-fitting code-switches would result in an additive effect. Results indicated that all weak-fitting conditions (regardless of language) elicited the same N400 effect. The strong-fitting code-switches, however, only elicited N400 effects in their most predictable contexts (see Figure 1). After initial lexical processes, all Spanish words elicited a late positive complex (LPC) and all weak-fitting words elicited a sustained negativity. Given these findings, the authors concluded three things: 1) codeswitches do not elicit a unique N400 effect; 2) listeners can predict a particular lexical item (not just semantic features) in highly predictable contexts; and 3) bilinguals only notice that a word is in another language after the N400 time window—thus, after lexical processing.