CONICET | Buscador de Institutos y Recursos Humanos

Several studies have found a positive ERP component (P600) around 600 ms after the presentation of a syntactic violation in natural languages, and artificial grammars with semantic content (Osterhout & Holcomb 1992; Friederici, Steinhauer, Pfeifer, 2002;),. It has been proposed that P600 reflects an expectancy violation in the analysis of sequenced structures (Patel, 1998), disregarding the semantic content of the stimulus. Furthermore, if P600 indexes expectancy violation, the component should be modulated by the frequency of appearence of the structure, as more frequent structures would be associated with a greater expectancy. Our objectives were 1) to determine the neural correlates of artificial grammar learning, based on transitional probabilites between pseudowords, without semantic content 2) to analyze P600 modulation by frequency of appearence of syntactic structures. 19 right-handed subjects (20-32 years old) were trained by exposure to grammatical sentences, presented audiovisually. The grammar had two posible syntactic structures, the frequent structure was presented in 2/3 of the training examples. During test, subjects had to clasify new sentences as correct or incorrect, with half of them presenting syntactic violations. A bilateral positive P600-like component was found for incorrect sentences( p = 0.002). Furthermore, a significant frequency effect was found in left hemisphere channels (p = 0.001). Results support the hypothesis that P600 indexes expectancy violations in the analysis of structured stimulus sequences. As this component is also elicited by linguistic stimuli, we propose that domain general statistical mechanisms are recruited for syntactic processing in natural languages.

enviar mensaje