[mæd/mæt]: análise dinâmico-complexa do desenvolvimento da duração vocálica por aprendizes paraibanos de inglês

Abstract

This dissertation aims to present a process analysis (Lowie; Verspoor, 2015; 2019; Yu; Lowie, 2019) of vowel duration development in minimal pairs ending in ‘voiced/unvoiced’ English plosives (such as in [mæd/mæt]), through analyses of the individual trajectories of fourteen native Brazilian speakers from the state of Paraíba who learn English as an additional language. To do so, the present study is grounded on Complex Dinamic Systems Theory (De Bot; Lowie; Verspoos, 2007; Larsen-Freeman; Cameron, 2008; De Bot, 2017; Larsen-Freeman, 2015, 2017; Lowie, 2017; Albuquerque, 2019; Lowie; Verspoor, 2019; Yu; Lowie, 2020; Hiver; Al-Hoorie, 2020, Schereschewsky, 2021 among others), in order to observe the possible influence of explicit instruction regarding the distinctions involving the length of the vowel preceding the target segments. Learners were divided into two different groups. One of the groups received explicit instruction about phonetic-phonological aspects of English (Experimental Group – 11 participants), and the other group did not receive explicit instruction (Control Group – 3 participants). Based on a longitudinal design with an eleven-month temporal window span and a fortnightly time scale, totalizing 24 data collection points, we conducted a study in an A-B-A format (cf. Hiver; Al-Hoorie, 2020), with an intermediate period of pedagogical instruction (between Data Collections 9 and 16 only for the Experimental Group), in which eight sessions of formal and explicit instruction were carried out about some phonetic-phonological aspects of the English language, in order to obtain favorable results for the distinction of vowel duration preceding voiced and unvoiced plosive segments in the non-native language. Two different methods of empirical investigation were carried out (peak analyzes and change-points analyzes), in order to discuss, based on the results of the relative duration values of the averages and the minimum and maximum values of vowel durations in each data collection preceding voiced/unvoiced plosives in English, in each of the 24 collection sessions, aspects such as complexity, variability, non-linearity and emergence of additional language development. The peak analyzes (cf. Van Dijk; Verspoor; Lowie, 2011), with Monte Carlo simulations, carried out with 10.000 interactions, presented significant increasing and decreasing performance peaks in all members of the Experimental Group, suggesting a possible influence of the period of explicit instruction. On the other hand, one member from the Control Group did not present significant performance peaks, while the others, despite presenting significant ascending and descending performance peaks, did not reach the expectations regarding the behavior of vowel duration in ‘voiced/unvoiced’ English plosives, because despite the variability, the trajectories followed parallel paths throughout the data collections, demonstrating that these three learners did not distinguish between the pairs of analyzed segments. The Change-point analyses (cf. Taylor, 2020), using the Change-Point Analyzer software (Taylor Enterprises, 2020), detected changing patterns in all of the members of both groups. These findings demonstrate that learners are likely to experience different forms of development, given the variability presented by each one. However, amid such variability, it was possible to find similar trends among the strategies used by the participants.