Processing problems in L2 listening comprehension of university students in Hong Kong

Abstract

This research is motivated by the scarcity of investigation into L2 learners' online processing problems in listening comprehension, in particular those relating to L2 proficiency. It aims to identify the online processing problems encountered by Hong Kong students of differing L2 proficiency and whether these problems can be overcome by compensatory schema use. Two questions are addressed: 1) Given the same listening situation, do high and low proficiency students experience the same online processing problems? 2) Does success in compensatory schema use vary according to L2 proficiency? Four data collection methods (semi-structured interviews, questionnaire survey, partial transcription, and introspection) and both quantitative and qualitative data analyses are employed. Exploratory semi-structured interviews indicate that apart from the problems of unfamiliar topics, vocabulary and accents, students report experiencing what might be called an "acoustic blur" during L2 listening. They hear streams of English speech sounds but cannot identify the key words. Even when they think they have identified the words, these words often do not fit the context. The questionnaire survey was designed using the data gathered from the semi-structured interviews. The survey results show that practice frequency has a significant effect on students' perception of overall listening difficulty; the more often the practice, the less often listening problems are perceived. Slightly more than one third of respondents have almost no listening practice in English outside class. L2 proficiency impacts significantly on students' perceptions of processing problems such as those caused by unfamiliar topics and vocabulary, and failure in sound segmentation and word recognition. In their self-assessment of word recognition, low proficiency students' word recognition in an average lecture and TV program is less than 50% and 40% respectively, whereas that of proficient students is 80% and 65%. These figures mirror those obtained in the partial transcription study. The partial transcription study investigates students' online processing problems and compensatory schema use in an experimental setting. Results demonstrate that low proficiency students transcribe 50% or less of the words in two texts successfully, whereas high proficiency students' success rates are near 80%. Of all types of transcription errors, "chunk" errors are the most prevalent for all proficiency groups. This confirms that listeners' online processing problems are spread across a string of two or more words. The two major categories of chunk errors are perceptual and lexical/syntactical in nature. The perceptual errors frequently arise as a result of the perception of "acoustic blur" and over-reliance on phonetic cues. The near preservation of syllable and stress patterns in errors indicates that some L2 listeners use a prosodic processing strategy when listening to English. Unfamiliar lexical items affect the processing of neighbouring words for low proficiency students, who either neglect what follows or retrieve familiar words that phonetically resemble the target items. The phonetically based errors show an attempt to apply phonological knowledge but a lack of or a failure to apply linguistic schemas (knowledge of the language and co-text). Successful compensatory schema use is quite rare; the rates are low regardless of L2 proficiency, especially when they are compared with those of LI listeners. These contrasts suggest that L2 processing is far from interactive. Furthermore, extra-linguistic schema use (use of world knowledge that lies outside the text) is uncommon. Some findings of the supplementary introspective study are consistent with those found in the partial transcription study and questionnaire survey, while others are uniquely the results of introspection. Students tend to blame fast speech or lack of concentration for their online processing problems. Furthermore, low proficiency students appear to use extra-linguistic schemas as a compensatory strategy more often than proficient students do, probably because they lack resources at the linguistic level. Conclusions suggest that LI interactive listening models may only be partially relevant for L2 listening comprehension. They also suggest a possible proficiency threshold for making reasonable guesses during the L2 listening process, as well as a threshold for overall understanding in terms of the percentage of words recognized. The partial transcription method has broken methodological new ground for studying L2 listening problems. It is reliable and has the potential as a diagnostic tool for testing and teaching. Implications for both L2 listening teaching and English-medium teaching across the curriculum are discussed.