Individual Differences in Grammatical Error Processing

Abstract

Years of research using electrophysiology to study language processes have yielded important discoveries about language-specific event-related potentials. However, most of these studies examined grand averages and population differences and avoided looking at individual differences. More recently, a few studies have shown that variability exists between individuals in their ERP responses to grammatical errors and that this variability may be related to cognitive and linguistic abilities. The goal of this study was to replicate the variability found previously and to investigate its relationship with measures of memory and cognitive abilities. We tested Ullman’s Procedural-Declarative model of language (Ullman 2001, 2004, 2005) by examining the relationship between procedural memory ability and the ERP response. We also investigated the possibility of gender differences in processing grammatical errors. Adult native English speakers read English sentences while event-related potentials (ERPs) were recorded. The relative magnitudes of the N400 and P600 to grammatical errors were compared to calculate a response dominance index (RDI) for each subject. We examined the correlation of the RDI with measures of declarative memory, procedural memory, verbal working memory, and processing speed. We also investigated differences in the RDI between men and women and between those with a first-degree left handed relative (familial sinistrality, or FS+) and those without (FS-). We replicated the finding that individuals vary on their dominant response (P600 vs N400) to grammatical errors. We found no relationship between RDI and any of the predictor variables, but did find a difference between men and women in the size of the N400 effect to syntactic errors. This does not lend evidence to the Declarative-Procedural model of language, which may relate to its strong dependence on data from individuals with language disorder. We suggest that the gender difference indicates that men are more variable in their processing of grammatical anomalies and relate this back to the variation seen across studies of the P600 effect. These results also provide strong support for the feasibility of examining individual differences using ERP data, even in populations previously thought to be homogenous.