Abstract:

The application of multivariate analyses in text studies is receiving attention from many researchers (Ho and Laman 1997; Arppe 2007; Nishina 2007). Multivariate statistical analysis refers to multiple advanced techniques for investigating relationships among multiple variables at the same time. According to McEnery and Wilson (2001), the common aim of all these multivariate analyses is “to summarise a large set of variables in terms of a smaller set on the basis of statistical similarities between the original variables, whilst at the same time losing the minimal amount of information about their differences”. In relation to corpus-driven studies, it has been discovered that multivariate analyses are a potential statistical analysis to quantify similarities and differences among text types by picturing the relationships visually for further investigation and interpretation. In this study, the correspondence analysis (CA), one of the multivariate analyses, is adopted to investigate statistical similarities and differences across corpora. Focusing mainly on highly frequent function words in the context of English for Electrical Engineering, this study shows how the results of the study of lexical items in specific corpora can provide the basis for EAP/ESP teaching and learning. In the course of this study, the integration of corpus-based approach and CA in text analysis is demonstrated. The findings confirm the potential use of the correspondence analysis, which provides interesting insights into the complex inter-relationship between the corpora, as a tool for language description. CA makes it possible to select significant function words that contribute to the differences between the corpora. The recognition of these function words, as a result, allows further investigation into their lexical behaviour, which include patterns of usage and typical neighbouring words in the specialised texts. As many researchers have suggested (Nattinger & DeCarrico 1992; Willis 1997; Schmitt & Carter August 2000; Nation 2001), common combinations of word forms, lexical phrases and meanings associated with the commonest words economically provide learners with good coverage and they are a key element of fluent language production.