Abstract
<jats:title>Abstract</jats:title> <jats:p> The use of corpora and corpus tools in the identification and analysis of multi‐word units helped improve traditional approaches to the study of formulaic language and brought about new methods that allow for the empirical investigation of this type of expression. Although <jats:italic>collocations</jats:italic> and <jats:italic>multi‐word collocations</jats:italic> have been studied in linguistics for centuries, the use of corpus‐based methodologies allowed for empirical approaches that complement or replace the more intuitive approaches used in the past. Advancements in computer processing and the more recent development of software for linguistic analysis allowed for the identification of corpus‐driven expressions such as <jats:italic>n‐grams, lexical bundles</jats:italic> , and <jats:italic>phrase‐frames</jats:italic> . These new constructs could not have been identified without the development of approaches that focus on the corpus and leave aside any preconceptions or intuitions about the type of multi‐word expressions that frequently occur in a given register. This approach complements the more traditional corpus‐based approach to multi‐word expressions that rely on previously identified expressions to check their occurrence in a corpus. The study of these different types of expressions helps describe registers better from a multi‐word perspective creating a more complete formulaic profile of the register under study. </jats:p>