Par erruer: error analysis and the early stages of adolescent foreign language learning

Abstract

Error Analysis has been widely used in studies of second language learning. At the same time, foreign language learning (as opposed to teaching), has largely been ignored as an object of research. The research had three major aims: to examine the potential of Error Analysis in foreign language learning by the development and trialling of a complex new coding schedule for analysing learner errors in French; to provide some descriptive data on the written errors and performance of foreign language pupils over a complete scholastic year; and to explore the nature of foreign language learning strategies used by the pupils studied. The thesis opens with a statement of purpose and method. This is followed by a short discussion of theories, topics and techniques in second and foreign language learning. The development of two forms of the coding schedule used to analyse errors is described and discussed. Recode checks and the development of 'Coding Confidence Levels' are presented. Procedures used in a longitudinal study of errors are then described. This study acted as an important trial for the coding schedules, known as Foreign Language Error Analysis: French (FLEAF). All written French produced by eight pupils in the same school class was collected over the 1978 New Zealand school year (February to December). The group were in their second year of high school French and had an average age of 14 years 5 months midway through the study. Background information is given about the subjects, including results of selected IEA French (Population II) tests administered during the year. Some description of the year's work is given, followed by general and case study analyses of errors. Selected results from both the longer (FLEAF-L) and shorter (FLEAF-S) coding schedules are then tabulated and discussed. Particular attention is paid to variables coding possible explanations for errors. Detailed analyses of errors in word order, negation and gender are offered, along with a review of correct performances for selected aspects. These analyses lead to two preliminary hypotheses which could in part explain the occurrence of errors. One hypothesis relates to the frequency with which pupils had been required to focus through drills on the point at issue, the other to the number of choices available to the pupil at the time of error. Discussion of Error Analysis and pupil strategies is then presented. It is concluded that Error Analysis has a valid role in developing hypotheses for a theory of foreign language learning. Five such hypotheses, suggested by the longitudinal study, are presented. It is argued that these five could all be classified on one of a proposed five levels of 'Operating Procedures' (McLaughlin, 1978a). Selection of a unique cluster of operating procedures would constitute the learning/performance 'strategy' of a particular pupil. These notions are incorporated into a tentative framework for a theory of foreign language learning. A modified 'Principle of Least Effort' (Zipf, 1965), is suggested as a key factor in the early stages of learning a foreign language, and ideas for subsequent research are proposed.