The Design and Implementation of a Framework for the Analysis of Lexical Errors in the Compositions of Students of English as a Second Language

Picot, Anthony John (2025) The Design and Implementation of a Framework for the Analysis of Lexical Errors in the Compositions of Students of English as a Second Language. Doctoral thesis (PhD), Manchester Metropolitan University.

Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (2MB)

Abstract

This dissertation seeks to revisit the potential of Error Analysis with a specific focus on lexical errors which will inform second language teaching and learning. The dissertation attempts to devise a new framework and guidance for analysing lexical errors, as a systematic and reliable method of identifying and categorising lexical errors would benefit SLA researchers, English Language teachers and learners (James, 1998; Hemchua and Schmitt, 2006). A replication of one of the most recent and potentially strongest Lexical Error Analysis (LEA) studies, Hemchua and Schmitt (2006), was completed and similarity was found between sets of results, giving promise to the concept and replicability of LEA. Other previous frameworks for LEA, namely Dušková (1969), Corder (1973), Richards (1971), Zimmerman (1986), Zimmerman (1987), Meara and English (1987), Lennon (1991), Zughoul (1991), Engber (1995), James (1998), Hemchua and Schmitt (2006) and Llach (2011), were tested and analysed, for their strengths and weaknesses in terms of their ease of use and the depth of analysis that they provide, before a new LEA framework was devised (NewLEAF Version 1). The new framework was then tested: interrater agreement for error identification and categorisation was investigated in two studies. The first, which showed promising quantitative results for consistent identification and categorisation between raters, also sought qualitative feedback on using the framework and accompanying guidance when it was employed by six highly qualified and highly-experienced EAP university teachers. Following this feedback, refinements were made to the framework and guidance to produce NewLEAF2, which was then tested on 41 participants. These participants comprised a mixture of Linguistics and TESOL staff and students. Following less encouraging results in terms of error identification and categorisation, further refinements were made to produce NewLEAF3, which was then used to analyse 20 scripts produced by Greek learners of English. It was found that the framework and guidance proved to be easy to use in that there were no issues in error identification, no uncategorisable errors, no dual categorisation issues, and a satisfactory spectrum of lexical error types. Results of the final analyses showed that there were far more semantic errors than formal errors: phrase errors were the most common followed by preposition and verb errors. Although further work may be required to improve the framework in terms of potential issues with inter-rater agreement in the areas of error identification and categorisation, the improved depth of analysis and ease of use provided by the new framework offers considerable advantage to learners and teachers of English.