Lexical bundles across sections of applied linguistics research articles

Abstract

The increasing use of computerheld text corpora for the analysis of lexicogrammatical patterns, once unavailable to observers, has allowed researchers to get a better and more precise understanding of language. Corpus Linguistics has also shown to be capable of revealing valuable and detailed features of expressions that play a crucial role in discoursewith linguistic investigations that comprehend not only single words but also idiomatic or nonidiomatic expressions in a given context. The aim of the current research is to investigate 47 word lexical bundles in sections of Applied Linguistics Research Articles, published in English. Lexical bundles, fundamentally defined by frequency (BIBER et al., 1999; CORTES, 2013), are generated with no predefined linguistic categories. Despite the frequency with which they occur, lexical bundles are not idiomatic in meaning and not perceptually salient (BIBER; BARBIERI, 2007, p. 269). Their use, however, has been thoroughly investigateddue to the role those devices play as building blocks of discourse. In order to offer details of intext linguistic variation, a corpus was compiled from 180 articles and split into four subcorpora: Introduction, Methods, Results, and Discussion (IMRD), with more than 1 million words in total. The lexical bundles were individually classified into structural andfunctional categories, including their subtypes, according to previous and established taxonomies (BIBER et al., 1999, 2004; CORTES, 2013; SIMPSONVLACH; ELLIS, 2010). Two different statistical treatments, confidence intervals and the nullhypothesis significance ztest, were adopted as complementary in order to check whether differences acrosssubcorpora were significant. Results show that sections of Applied Linguistics research articles should be treated as separate texts for they display strong distinctions, and grammatical structures may play singular functional roles. The main distinctions between sections entail frequency of devices and a preference for particular lexicogrammatical elements in relation to the pragmatic nature of the section, e.g. fragments with exclusive roles can be related to passive and nonpassive voice forms containing specific verbs. Additionally, the proportion of some structures and functional categories that emerged in the current research challenge generalizations made in previous studies about academic register. Finally, we argue for more research on subregisters with representative and specialized corpora so as to give more insight into academic writing of research articles sections.