Identifying lexical bundles for an academic writing assistant in Spanish

Abstract

This paper presents the process of identifying recurrent multi-word expressions (i.e. lexical bundles) relevant for a writing-aid of academic texts in Spanish. It also proposes a repertory of discourse functions that enables the classification of the candidate word-strings as well as their onomasiological retrieval. This classification into discourse functions is also key in the process of selecting candidates, as those that do not fit in any function will predictably be of little interest in the context of the mentioned writing aid. Through the examination of the resulting data, the study explores the correlation between candidate selection by lexicographers and several association measures proposed in the literature to obtain high-quality lexical bundles, with a view to assess the feasibility of automating this process.