Perception in L2 in a Classroom Environment with L2 Portuguese Chinese Students

The purpose of this study is to contribute to the knowledge on the impact of common European Portuguese (EP) phonetic-phonological processes in second language (L2) learners. It is well established that L2 listening is a complex process, and that the most common difficulties among L2 learners are related to speech segmentation and word recognition. Due to the occurrence of connected speech processes, sounds are altered and the word boundaries can be hard to determine. Vowel reduction within and across word boundaries is usually described as a very frequent process in EP. The reduction of vowels is even more evident in spontaneous speech, e.g. the word 'telefone' (jtifi'foni] in the citation form, 'telephone') can be produced as [t'fon]. The interplay between these processes can be particularly impactful for L2 learners, in word recognition. Furthermore, this correlation is scarcely studied in a classroom setting. The present study explores the impact of vowel reduction and connected speech processes in word recognition tasks from isolated words to continuous speech. Furthermore, it aims to understand the main difficulties that L2 learners, at the intermediate levei B 1, experience dealing with these phenomena. Lastly, it will contribute to understand not only the acquisition of vowel reduction and connected speech processes but also whether L2 learners could cope with them. Therefore, it was designed a set of perception experiments involving these phenomena in increasing degrees of difficulty: single word identification without (i) and with vowel reduction (ii); word identification with simple (iii) and complex connected speech processes (iv). The experiments were conducted in an ecological setting of an intensive Portuguese course, of the intermediate levei B 1, at the University of Lisbon. A contrai group of EP native speakers also performed the experiments. The overall scores revealed a decreasing tendency: (i) 94%; (ii) 65%; (iii) 31 %; (iv) 16%. The results reveal that word recognition is compromised due to ...