Representação fonológica, instrução explícita e produção sonora do espanhol como L2 por aprendizes brasileiros

Abstract

The aim of this work is to investigate the phonological representation of categorical and variable sound patterns in Latin American Spanish as a second language (L2). The categorical sound pattern is represented by the phonemic distribution of the tap /ɾ/ and the trill /ř/ in both word-initial (e.g., rollo [ˈřo.dʒo]) and intervocalic (e.g., caro [ˈka.ɾo] carro [ˈka.řo]) positions of Spanish words. A categorical sound pattern allows inferring the phonological representation in specific contexts, since only one sound manifestation is possible in a single context. The variable sound pattern is represented by the allophonic distribution of /b, d/ both in word-initial (e.g., bota [ˈbo.ta] and duna [ˈdu.na]) and intervocalic (e.g., sabe [ˈsa.ße] and cada [ˈkaða] ) positions of Spanish words. A variable sound pattern does not allow inferring the phonological representation in specific contexts because more than one sound manifestation is possible in the same context. The theoretical framework of this study is based on the premises of the Exemplars Model (JOHNSON, 1997; BYBEE, 2001; PIERREHUMBERT, 2001, 2003) and on the Speech Learning Model (SLM) by Flege (1995). Additionally, methodological and experimental criteria posited by Laboratory Phonology (PIERREHUMBERT; BECKMAN; LADD, 2000) have been considered. The central hypothesis is that sound productions associated with the categorical sound pattern are more accurate than in the variable sound pattern. This is because, based on the Exemplars Theory assumptions, the sounds that make up the categorical sound pattern are more robust, that is, they have less variability and there is a contextual correlation with the L1 sound pattern, as opposed to the sounds that make up the variable sound pattern. The dataset consists of oral production recordings of twenty-two Brazilian native speakers of Brazilian Portuguese (BP) who are Spanish language learners. A categorical analysis was performed, and the levels of the dependent variable were either the presence (1) or the absence (0) of the target sound in both categorical and variable sound patterns. Six independent variables were assessed, namely: (1) explicit instruction, (2) sound pattern, (3) task type, (4) word (5) time of exposure to the target language and (6) individual behavior. Statistical analysis was performed using the MATLAB software (2021) and Pearson's chi-square test (χ2). The following variables are statistically significant: explicit instruction, sound pattern, task type, token frequency and time of exposure to the target language. Statistical tests were not applied on the word and individual behavior variables. In short, results show that data associated with the categorical sound pattern had higher L2 target sound production rates than the variable sound pattern, corroborating the central hypothesis of this work. The results provide important data for the discussion about the phonological representation of students and explicit instruction in the context of learning Spanish as an L2.