This dissertation discusses an understudied phenomenon in Brazilian Portuguese (BP): sluicing, a phenomenon in which the tense phrase (TP) of a structure headed by an interrogative pronoun (-QU) is elided, leaving only the explicit -QU. Merchant (2001) argues that sluicing involves –wh movement and deletion in the phonological form of TP, and proposes the following generalization: a language L will allow preposition stranding in sentences with sluicing iff L allows this stranding in simple sentences. Almeida and Yoshida (2007) argue that the Brazilian Portuguese distorts this generalization as it allows preposition stranding in sluicing, but not in simple sentences. In contrast, Rodrigues et al. (2009) proposes that there are two processes of sluicing in Brazilian Portuguese: regular sluicing and pseudosluicing, with preposition stranding occurring only in these structures. This dissertation investigates the proposal of Rodrigues et al. (2009), leading a grammaticality judgment experiment and using experimental methodology in order to verify deletion relations between preposition and type of structure. The results support the proposal of Rodrigues et al. (2009), and point to a semantic identity between the elided constituent and its antecedent. Our findings suggest that: the Brazilian Portuguese data does not distort Merchant s generalization; and, semantic identity is sufficient for licensing elision processes. This phenomenon is investigated in light of an integrated proposal between generative linguistic theory and psycholinguistics, a proposal that has characterized the research being developed at the Laboratory of Psycholinguistics and Language Acquisition (LAPAL) at PUC-Rio.