Compiling a corpus-based list of words commonly mispronounced

Abstract

The inspiration for the paper was professor Sobkowiak’s list of Words Commonly Mispronounced (Sobkowiak, 2001), a collection of over six hundred pronunciation errors that are habitually made by Polish learners of English. The paper explores the ways in which lists such as Words Commonly Mispronounced could be “upgraded” using corpus linguistic tools. The paper describes the results obtained in a previous study (Zając & Pęzik, 2012), whose aim was to compile a corpus-based index of frequent mispronunciations in the speech of Polish learners of English and which used data from the spoken component of the Polish Learner English Corpus PLEC. The paper discusses the list obtained by Zając and Pęzik, describes and evaluates the process of creating the list, and compares the corpusbased index with Words Commonly Mispronounced. The difficulties related to the compilation of lists of common mispronunciations (both corpus-based and “traditional”) are also examined. The general conclusion that can be drawn from the analysis is that employing corpus linguistic tools to examine L2 pronunciation errors may enable one to create a thorough and reliable collection of commonly mispronounced words, which can constitute an effective and powerful tool in pronunciation teaching and learning. At the same time, careful examination of the corpus-based list and the process of its creation reveal that, just as in the case of compiling a list of common mispronunciations using “traditional” methods, creating a corpus-based index of pronunciation errors entails certain problems that need to be addressed when attempting to produce such a list.