A influência da animacidade no processamento de relativas de sujeito e de objeto no Português brasileiro e europeu

Abstract

In this work, we investigate the influence of the animacy feature on the processing of subject and object extraction in relative clauses, taking into consideration Brazilian and European Portuguese data. In order to achieve this goal, a corpus study was performed, as well as five experimental studies using a self-paced experimental technique. Many studies on sentence processing claim there is an asymmetry between object and subject relative clauses. The majority of these studies show that subjects are processed faster than objects in relative clauses. The reasons for this asymmetry, in general, are due to the syntactic complexity of these clauses. Such studies normally do not control for the animacy feature, and the stimuli used in those experiments generally have animate antecedents for both subject and object extraction. Some works, however, have indicated that the animacy feature is a relevant factor not only during sentence processing but also in the distribution of relative clause patterns in natural languages. The hypothesis we assume here is that if the animacy feature really influences the distribution of relative clauses, then the online processing of these structures would be affected by this feature. A corpus study was performed in Brazilian and European Portuguese. The results show that animate subjects and objects in relative clauses do not represent the most usual structure in Portuguese. The animate object is especially rare in the language. Based on this information, experimental studies were conducted. In the two initial experiments, both identical, one in Brazilian Portuguese and the other in European Portuguese, we intended to measure the processing time for both subject and object relative clauses, all of which had an animate feature. From these experiments, we could confirm and corroborate what is presented in the literature concerning this topic: subject relative clauses were indeed processed faster. In experiments 3 and 4, respectively in Brazilian and European Portuguese, four conditions were tested: animate and inanimate subjects, as well as animate and inanimate objects in relative clauses. Significant results were achieved, and they indicate a main effect of the animacy feature. In several measurements, the inanimate object was read even faster than subjects. In the fifth experiment, we applied the self-paced reading technique, but with isolated words. We tested animate and inanimate nouns separately. The results of this last experiment were not significant. This shows that the animacy feature probably does not act when only isolated words are read, but rather this feature becomes relevant when a phrasal structure distribution is presented. In order to analyze our results, we applied Gibson's (2000) Dependency Locality Theory (DLT) as a basic theoretical model. According to this theory, the parser is able to access, among other sources of information, the frequency of distribution of animacy patterns in the language, and it does so in the early stages of the processing. In other words, the frequency of structures with animate and inanimate lexical items was determinant in processing subject and object relative clauses. An asymmetry between these structures was confirmed, but this asymmetry was undone when the animacy feature was properly controlled.