O fenômeno do code-blending em surdos oralizados bilíngues bimodais de libras/português: implicações para a arquitetura da gramática

Abstract

The phenomenon of code-blending presents itself in the possibility of simultaneous production of languages of different modalities, which is provided precisely by the availability of different articulators for the production of each language (QUADROS et al., 2013). This is an interesting linguistic phenomenon, since the investigation about the production of bimodal bilinguals has the potential to contribute to discussions about the architecture of grammar at its most varied levels. Inserted in this debate, this work investigates the production of code-blending from data collected from videos made available on the internet by 3 oralized deaf collaborators, signers and bimodal bilingual speakers of Portuguese and Brazilian Sign Language (Libras). In quantitative terms, a total of 183 utterances were collected, and in 154 of them code-blending was detected. This quantitative result points to a preference for utterances containing simultaneous realization, rather than monolingual utterances. From an empirical point of view, we seek to describe and investigate the properties of the code-blending data collected. Therefore, the data were classified based on the typologies proposed by Van den Bogaerde and Baker (2005, 2008) – considering the adaptations of Lillo-Martin et al. (2016) – and by Donati and Branchini (2013). In qualitative terms, we found in our corpus code-blending formations of the following types: (i) Sign-base code-blending, (ii) Speech-base code-blending; (iii) independent code-blending; (iv) mixed code-blending; (v) full code-blending and (v) codeblending with syntactic calque. It is important to point out that, for the most part, the productions found correspond to code-blending of the Sign-base code-blending type, which is characterized by the production of a complete utterance in sign language, with some simultaneous oral words, evidencing the preference of the collaborators of this research by Libras as the dominant language. In turn, from a theoretical point of view, we discuss proposals developed within the scope of generative models, more specifically, the Minimalist Program (CHOMSKY, 1995, 1995), Distributed Morphology (HALLE; MARANTZ, 1993; MARANTZ, 1997) and the Language Synthesis Model (LILLO-MARTIN, QUADROS and CHEN PICHLER, 2016). The generative framework has proven to be interesting for discussions of the code-blending phenomenon for several reasons. One of them is its modular organization, which allows investigations regarding the existence of one or two distinct syntactic derivations for simultaneous production. Another interesting point is that, by separating the syntactic computation from the processes of phonological realization, the generative framework open perspectives for the discussion of the linearization processes involved in these productions. In theoretical terms, we argue in favor of a model compatible with the bases of the Distributed Morphology and the Language Synthesis Model. More specifically, we propose that codeblending data can be better explained from a constructionist model, in which formal features are articulated only by the Syntax component. Furthermore, the code-blending data points to the advantage of the Subspecification property of Vocabulary Items, avoiding the need to specify a double set of lexical entries with items specified for sign language properties and others specified for oral language properties. In addition, we point out, in terms of the grammar architecture design, to the possibility of a bifurcation of the syntactic structure in the PF branch for different channels in languages of different modalities, preventing competition between the elements of these two languages. Finally, we propose that, for Vocabulary Insertion purposes, such a process should occur for all nodes generated by the Syntax in at least one language, without the need, however, for mandatory completion in both modalities.