Processamento da correferência na posição de sujeito e de objeto por brasileiros falantes de inglês como L2

• Ano de defesa: 2017
• Autor(a) principal: Barbosa, Matheus de Almeida
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal da Paraíba; Brasil; Linguística e ensino; Programa de Pós-Graduação em Linguística; UFPB
• Palavras-chave em Português: Correferência intersentencial; Bilinguismo; Paralelismo estrutural; Intersentential coreference; Bilingualism; Structural Parallelism; CNPQ::LINGUISTICA, LETRAS E ARTES::LINGUISTICA
• Link de acesso: https://repositorio.ufpb.br/jspui/handle/123456789/12049

One of the first theories that tried to understand the use of anaphoric expressions was the Centering Theory (GROSZ et al., 1983). Based on it, Gordon and colleagues (1993) investigating the processing of anaphoric expressions through pronouns and repeated names found a phenomenon, the Repeated-name penalty (RNP), and justified this result based on the Centering Theory, saying that pronouns are more easily processed because they are natural choices to make reference to previously mentioned entities in comparison with repeated names. Chambers and Smyth (1998) observed the same phenomenon, but considering the structural parallelism as a key factor in explaining the results. They found that when antecedent and anaphora were in the same structural position, the reading times of the sentences was facilitated. They also found that the RNP happened when the anaphora was in the object position as long as the parallelism was respected. Studies with bilingual populations were led by Cho (2010) and Gadelha (2012). Cho (2010) did not find the RNP, and explained these results by attributing it to a possible overload of working memory in such individuals. Gadelha (2012) found the RNP in the reading times of advanced level participants, but not to intermediate level participants. She also justified that as an overload of the working memory. In this dissertation, we seek to understand how the processing of anaphoric expressions, pronouns and repeated names in syntactic positions of subject or object, happens for Brazilians speakers of English as L2 at the intermediate and advanced levels of proficiency. Still on the syntactic position of anaphoric expressions, we seek to understand how the structural parallelism affects the processing of the anaphora. Participants were initially divided into two groups according to their language proficiency in L2 through the VLT - Vocabulary Levels Test - (NATION, 1990) with a time limit of 10 minutes as proposed by Souza and Soares-Silva (2015). We conducted a self-paced reading task, which measured the reading time of the critical segment in addition to the reading times and correct answer rate of the probe word at the end of each item. The results showed a main effect for the level of proficiency in the direction of faster reading times by advanced speakers than intermediate ones, and a main effect for the variable structural parallelism, with the anaphora in sentences without structural parallelism being read faster than those with parallelism. There was no interaction between variables. Regarding intermediate level subjects, there were no significant differences in the type of anaphora. Regarding advanced level subjects, the results showed a statistically significant difference in sentences with parallel anaphora, in accordance with the repeated-name penalty. There was no such penalty when the anaphora was not parallel to its antecedent. About the probe word, there was no statistical difference in correct answers rate, despite the advanced participants having taken less time to accomplish this task. Generally, the results of the advanced level participants are in dissonance with the findings of the Shallow Structure Hypothesis (CLAHSEN & FELSER, 2006), that is, such participants showed a preference for the structural information when identifying the entity that was being referred in the discourse. Besides that, the results suggest a tendency to being favorable to the Interface Hypothesis (SORACE, 2011), which states that structures that are processes in the interface of syntax and other cognitive domains display some divergence in the processing of bilingual advanced speakers of L2 and native speakers of such L2.