Research reportReferential processing in the human brain: An Event-Related Potential (ERP) study

- We describe an ERP experiment on the processing of long-distance dependencies.
- LAN was elicited by the 2nd element in both syntactic and referential dependencies.
- These 2 types of relationships thus appear to pose similar processing challenges.
- Results suggest LAN may index referential/syntactic “back association” processes.

A substantial body of ERP research investigating the processing of syntactic long-distance dependencies has shown that, across languages and construction types, the second element in such configurations typically elicits phasic left anterior negativity (LAN). We hypothesized that these effects are not specific to syntactic dependencies, but rather index a more general cognitive operation in which the second (dependent) element in sentence-level linguistic long-distance relationships triggers a process of association with the first element. We tested this hypothesis with straightforward referential dependencies, comparing pronouns with proper name antecedents to those without, and proper names with and without preceding co-referring pronouns. We predicted phasic LAN effects in response to the second referential element in both comparisons, but observed them only in response to pronouns with antecedents; no differences were observed between responses to proper names with and without preceding co-referring pronouns. We argue that LAN effects observed at the pronoun index the cognitive operations necessary for the association of a pronoun with its antecedent, on which it depends for its reference. Similar but non-identical responses were elicited by the main clause verb following the gap position in object relative clause constructions compared to coordinate clause controls in an orthogonal manipulation. LAN effects were thus elicited by the second dependent element in both construction types, suggesting that long-distance syntactic and referential dependencies pose similar processing challenges. These findings help to clarify the cognitive processes indexed by anterior negative responses to associated dependent elements in a variety of language contexts.