Linear distance modulates P600 amplitude for gender agreement: An ERP study in Norwegian

Former ERP studies have shown that linear distance (number of intervening words between two agreeing elements in a sentence) modulates P600 amplitude for gender agreement violations. However, these studies have primarily focused on across-phrase dependencies, making it difficult to tease apart the effects of linear distance from those of structural distance (number of intervening phrases between two agreeing elements in a sentence). Norwegian, an understudied language in gender processing literature, permits manipulation of linear distance while holding structural distance constant. We tested 36 native Norwegian speakers with EEG, comparing three violation types: noun-suffix (within-word), determiner-noun (within-phrase), and noun-predicative adjective (across-phrase). Our aims were to: 1) assess processing of within-word gender violations, which, to our knowledge has not previously been done; 2) examine the effect of linear distance on within-phrase gender agreement by comparing processing of noun-suffix and determiner-noun dependencies; and 3) establish an ERP baseline for Norwegian gender processing. All three violation types elicited P600 responses, indicating repair/reanalysis processes. Noun-predicative adjective violations additionally produced a left anterior negativity, and determiner-noun violations showed a late frontal negativity. The noun-suffix P600 demonstrates that within-word gender violations recruit similar repair mechanisms as between-word violations. Moreover, the noun-suffix P600 was larger than the determiner-noun P600, likely due to either weaker constituent links resulting from longer distance, or longer distances reducing repairs due to a cost-benefit tradeoff. The findings from this study show that 1) within-word violations engage the same reanalysis processes as between-word violations, 2) within-phrase gender agreement processing is sensitive to linear distance, and 3) Norwegian exhibits similar ERP correlates of gender processing as those found for other languages.