Processing of syntactic structures with different frequencies of use: an ERP study Angel Tabullo, Yamila Sevilla, Alberto Yorio; Enrique Segura, Silvano Zanutto y Alejandro Wainselboim Facultad de Psicología (UBA) Instituto de Biología y Medicina Experimental-CONICET, Buenos Aires, Argentina angeltabullo@ibyme.conicet.gov.ar Introduction Classic theories of grammar processing assume that a discrete combinatorial system applies algebra-like rules to parsing (Ullman, 2001; Steedman, 2000; Pinker, 1994; Chomsky, 1957). The P600, a late posterior positive ERP observed after syntax violations or structural ambiguity, (Hagoort et al., 1993; Osterhout and Holcomb, 1992; Osterhout et al., 1994) is considered an index of the processing cost of this system (Kaan et. al., 2000). An alternative hypothesis proposes that syntax structure is represented as a probabilistic mapping of co- occurrences of word categories (Elman et al., 1996; Rogers & McClelland, 2004). Each hypothesis makes different predictions about P600 elicited by grammatically correct, but highly infrequent word combinations. Classic hypothesis predicts that infrequent grammatical syntactic structures will not generate a P600 when compared to frequent grammatical structures, because grammar rules are applied by the system in an all-or-none fashion. On the other hand, probabilistic hypotheses consider syntax violations as a combination of words with null probability of occurrence, predicting that the EEG activity evoked by correct, but highly infrequent word combinations will be more similar to the EEG response to violations than to grammatical and frequent syntactic structures. Therefore, both syntax violations and infrequent grammatical structures should evoke a P600. The aim of the current study was to analyze EEG activity evoked by processing of syntax violations and syntactic structures with varying frequencies of use. Furthermore, its goal is to compare the effects of grammaticality and frequency of use over the P600 component. Results Discussion  Effects of grammaticality and frequency of use were observed on the P600.  The fact that grammatical infrequent sentences generated larger positivities than frequent ones is in agreement with the probabilistic mapping hypothesis. On the other hand, this effect was earlier and more frontally distributed than the P600 evoked by syntax violations. It is possible that this earlier component is reflecting an expectancy violation based on probabilistic mapping of word categories co-ocurrences. This probabilities are affected by frequency of use in speech and written language.  An effect of frequency was also observed for syntax violations: violations of infrequent structures evoked larger positivities than violation of frequent ones.  A typical grammaticality effect was observed, with larger and posteriorly distributed positivities for syntax violations. However this component had a larger latency (1000 ms) than usual. References Chomsky, N. (1957). Syntactic structures. The Hague: Mouton. Elman, J. L., Bates, L., Johnson, M., Karmiloff-Smith, A.,Parisi, D., & Plunkett, K. (1996). Rethinking innateness. A connectionist perspective on development. Cambridge: MIT Press. Hagoort, P., Brown, C.M., Groothusen, J., 1993. The syntactic positive shift (SPS) as an ERP-measure of syntactic processing. Language and Cognitive Processes 8, 439–483. Kaan, E., Harris, A., Gibson, E., Holcomb, P., 2000. The P600 as an index of syntactic integration difficulty. Language & Cognitive Processes 15, 159–201 Osterhout, L., Holcomb, P.J., 1992. Event-related brain potentials elicited by syntactic anomaly. Journal of Memory and Language 31 (6), 785–806. Pinker, S. (1994). The language instinct. How the mind creates language. New York: Harper Collins. Rogers, T. T., & McClelland, J. L. (2004). Semantic cognition. A parallel distributed processing approach. Cambridge: MIT Press. Ullman, M. T. (2001). A neurocognitive perspective on language: The declarative/procedural model. Nature Reviews Neuroscience, 2, 717–726. This work was supported by ANPCyT and University of Buenos Aires.  The results are not congruent with those observed in Pulvermuller (2007), who did not find differences between common and uncommon word strings in the mMMN (Magnetic Mistmatch Negativity), and did observe differences between both common and uncommon word strings and ungrammatical strings. On the other hand, the stimuli types and the time-window of the effects are not comparable to those of our study. ERPs by sentence type Methodology  The sample study was composed of 16 healthy right-handed subjects (18-30 years old), who spoked spanish as native language.  Three verbs were chosen to build the sentences of the experiment. These verbs could be attached to three possible structures: an inflectional phrase (IP), a complement phrase (CP) or a noun phrase (NP). Even though all structures were valid, the verbs were more frequently followed by inflectional phrases than complement phrases in a previous sentence completion task (Sevilla et al., 2008). Therefore, the frequency of use of the verbs + IP was higher than that of CP (see Table 1).  160 sentences were shown to the subjects. The senteces belonged to the following experimental conditions: a) grammatical frequent, b) grammatical infrequent, c) violation frequent (a syntax violation of the frequent sentence type, d) violation infrequent (a syntax violation of the infrequent sentence type).  Subjects were instructed to classify the sentences as grammatically correct or incorrect.  EEG activity was registered through 19 channels (10/20 system). ERPs were synchronized to the onset of the critical word in each sentence, and processed off-line (see figure 1). A late positive component was observed, ranging from 600 ms after the critical word onset in each sentence to 1200 ms. Three time-windows were defined for analysis: 600-800ms, 800-1000ms, 1000-1200ms. The component was modulated by the factors “frequency of use” and “grammaticality”, although the topography and time-course of these effects differed. In an early window (600-800 ms), grammatical infrequent sentences elicited a larger positivity than frequent ones (p = 0.033), particularly in frontocentral sites. On the other hand, syntax violations generated smaller positivities than grammatical sentences in posterior sites (p = 0.035). In the intermediate window (800-1000 ms), infrequent sentences (both grammatical and violations) elicited larger positivities that frequent ones (p = 0.035). In the late window (1000-1200 ms), syntax violations generated a larger positivity than grammatical sentences (disregarding their frequncy of use) (p = 0.001) Grammatical frequent Grammatical infrequent Violation frequent Violation infrequent Grammatical Frequent Grammatical I nfrequent Violation Frequent Violation I nfrequent Verb % Answers IP CP NP Reccomended 57 33 10 Forbade 88 10 2 Allowed 78 20 2  Grammatical  frequent) La decordadora recomendó a las señoras redecorar el living / The decorator reccommended the ladies to redecorate the living  infrequent) El capitán prohibió a los marineros que arrojaran el ancla / The captain forbade the sailors that they dropped the anchor  Violations  frequent) El emperador les prohibió a los ministros asumiendo sus cargos / The emperor forbade the ministers assuming their charges  infrequent) La resolución les permitió a los empresarios que reducir los salarios / The resolution allowed the bussiness men that to reduce the salaries Frequency of use Sentence types used in the experiment 600-800 ms 800-1000 ms 1000-1200 ms Table 1 View publication stats View publication stats