Phraseology

There has been widespread interest over the last three decades in the use of multi-word prefabricated expressions (e.g. in a nutshell, if you see what I mean. This research argues that a good portion of the language we use every day is composed of prefabricated expressions, rather than being strictly compositional (see, e.g., Pawley and Syder 1983, and Nattinger and DeCarrico 1992; also see the reviews of earlier research in Howarth 1996; Wray and Perkins 2000; and Wray 2002). Such multi-word sequences have been investigated under a variety of labels and definitions, including “lexical phrases,” “formulas,” “routines,” “fixed expressions,” “prefabricated patterns” (or “prefabs”), n-grams, and “lexical bundles.” Regardless of label, however, these studies share a focus on how words combine into more and less fixed combinations. Most early studies were primarily theoretical in nature, comparing the various perspectives and approaches to multi-word units, proposing new frameworks for analysis, and calling for further research. Hakuta (1974), Yorio (1980), Pawley and Syder (1983), Weinert (1995), Howarth (1996, 1998a, b), and Wray and Perkins (2000) are good examples of this type. Beginning in the 1990s, most research on phraseological patterns has been empirical, utilizing corpus analysis. Weinert (1995: 182) identifies two basic issues for such research: the best way to define and identify fixed multi-word units, and analysis of the discourse functions that these multi-word units perform. While there have been dozens of empirical studies carried out since then, these issues are still two of the most important considerations motivating current studies. With respect to the first issue, two general approaches have been employed to identify and analyze important multi-word units: corpus-based and corpus-driven (see Tognini-Bonelli 2001: 84-87). In corpus-based studies of formulaic language, the researcher pre-selects multi-word expressions that are perceptually salient or theoretically interesting, and then analyzes the corpus to discover how those expressions are used (e.g. Moon 1998).