Pistas prosódicas na desambiguação de sentenças coordenadas no PB

Abstract

This study investigates the use of Intonational Phrase (IP) boundary cues on syntactic processing by Brazilian Portuguese (henceforth, BP) adult speakers. Several studies conducted in the psycholinguistic subarea called Sentence Processing discuss to what extent the initial decisions of the parser – human language processor – would be guided by strictly syntactic information, or also by other linguistic or extra linguistic information. For this, the comprehension of the structurally and temporarily ambiguous sentences – garden-path sentences – has been tested using different techniques and experimental tasks, and experimental results suggest that adult speakers use prosody as a source of information in sentence processing (e.g. KJELGAARD & SPEER, 1999; CLIFTON et al, 2002; MILLOTTE et al, 2007; DEDE, 2010; FONSECA, 2012). In order to contribute to this discussion, we chose a little-investigated structure in our area of studies, specially, in BP – sentences with coordinated DPs, for which ambiguity is undone by prosody ([The judge intimates João]IP [and Pedro and Lucas await (for) the final result]IP – [The judge intimates João and Pedro]IP [and Lucas awaits (for) the final result]IP). We use the phoneme restoration technique masking the number marking on the verb, in order to check if PB native speakers recognize a simple or conjoint subject based on IP boundary cues. We conducted two experiments using the visual word choice task (STOYNESHKA et al, 2010), and two experiments were applied using the sentence-image verification technique, with attention manipulation resource. This procedure allowed us to verify the recovery of a stimulus in working memory and the comparison of linguistic and visual stimuli in the comprehension of a sentence. Our working hypothesis is that the prosodic information can help the mental construction of the meaning of a sentence. Results suggest that the parser takes into account the prosodic information, since the participants' answers are compatible with the intonation of each stimulus.