Pistas prosódicas e o processamento de sentenças ambíguas do tipo

Aline Alves Fonseca

This dissertation arose from questions about the ambiguity in syntactic constructions like NP1-V-NP2-attribute as in The boy hugged the friend sweating (O rapaz abraçou o amigo suado), which can happen in Brazilian Portuguese. When this ambiguity occurs it refers to the syntactically ambiguous attachment of attribute, which can be locally attached to NP2 or non-locally to NP1. Two principles lead the parser in the sentence processing: Late Closure and Minimal Attachment. The Late Closure principle is applied to the above-mentioned construction in which the local attachment is considered default. This attachment preference was tested by Magalhães and Maia (2006) in silent and loud reading experiments in three modes: segmentation after the VP; segmentation after SP2; no segmentation. It was expected that the segmentation would cause a prosodic break and consequently a new prosodic phrasing which could be associated to the syntactic structure creating a non-local attachment of the attribute, thus supporting the Implicit Prosody Hypothesis (IPH) by Fodor (1998, 2002). However, the strength of the Late Closure psychological principle was confirmed in the experiments and there was a preference for the local attachment even in the segmented reading conditions which forced subjects to choose the non-local attachment. Nevertheless, after acoustic analysis of loud non-segmented reading, researches realized that in readings in which the non-local attachment was chosen, the subjects used some prosodic cues which help them have such interpretation. From this observation we searched for cues which confirmed that the prosody can lead the parsing, thus contradicting the expected syntactic principle in sentence processing like NP1-V-NP2-attribute. To corroborate these findings, production and perception experiments were carried out with spontaneous and manipulated prosody in order to verify at what extent prosodic cues can lead the listeners choice to a non-default attachment, in this case, instead of a nonlocal attachment. Besides having questionnaires, the perception tests were also monitored like the experiments done with the DMDX software for psycholinguistic studies. The obtained data ratify that when the prosody is used emphatically, it can undoubtedly direct the interpretation of sentences like the ones used in this work, that is, it can influence the listeners to interpret these sentences as non-default by attaching the attribute non-locally. Among the studied prosodic cues, the F0 raise in NP1 and in the attribute was the most remarkable one to the listeners, that is, the sentences which had this prosodic cue were mostly interpreted as non-local attachment. When the prosody is not emphatic, it seems that the structure processing is only Good Enough as proposed by Ferreira et al. (2002, 2007) following the syntactic principle of Local Attachment. Several studies in different languages have been carried out in an attempt to prove the viability of IPH (Implicit Prosody Hypothesis). This work also seeks evidence for the use of prosodic information in processing and therefore intends to contribute to the improvement of experimental prosodic studies in relation to psycholinguistic studies.

Pistas prosódicas e o processamento de sentenças ambíguas do tipo "SN1-V-SN2-Atributo" do Português Brasileiro

AUTOR(ES)

DATA DE PUBLICAÇÃO

RESUMO

ASSUNTO(S)

ACESSO AO ARTIGO

Documentos Relacionados