CONICET | Buscador de Institutos y Recursos Humanos

Classic theories of grammar processing argue for a discrete combinatorial system that applies algebraic-like rules to parse the sentence structure. The P600, an ERP elicited by syntax anomalies, is thought to reflect this processing cost. An alternative account suggests that syntactic structure is processed as a probabilistic mapping of co-occurrence between word classes. Each hypothesis makes a different prediction: classic theories predict that infrequent grammatical combinations of words will not evoke a P600, while probabilistic theories predict that infrequent grammatical combinations will be processed similarly to ungrammatical ones, consequently expecting a P600 for both events. This study compared possible frequency and grammaticality effects in ERPs during syntax processing. Three verbs were employed, followed by two structures with different frequencies of use: a) Infinitival Object Clause (IOC, 80%), or b) Finite Object Clause (FOC, 15%). Four types of sentences were shown: 1) grammatical frequent (GF): NP-V-IOC, 2) grammatical infrequent (GI): NP-V- FOC 3) syntax violation of GF 4) syntax violation of GI. A widely distributed late positivity was observed in the 600-1200 range. Within 600-800 ms, the positivity was larger for syntax violations at posterior sites (p = 0.035), and larger for GI than for GF sentences (p = 0.033). The frequency effect between GI and GF was more frontally distributed than the grammaticality P600. Within 800-1200 ms, only the syntax violations yielded a significant P600 (p = 0.001). Therefore, in addition to the P600 elicited by ungrammatical events, a similar (although shorter) positivity was evoked by grammatical infrequent sentences. This positivity could be the result of expectancy violations based on probabilistic mapping of words.