Fixed Expressions and Idioms in English: A Corpus-Based Approach (review)

Rosamund Moon, noted Cobuild lexicographer (e.g. Moon 1995, Sinclair et al. 1987), continues her contribution to the field of phraseology with this book based on her 1994 doctoral thesis (University of Birmingham). It provides a compendium of scholarship, an attempt to use, critique, and expand computational research, and a meticulous analysis of a full data set of phrases. As M is careful to point out, it is preliminary, calling for additional research.

M sets out to describe in detail a corpus of 6776 FEIs (fixed expressions and idioms) appearing mainly in the eighteen-million-word Oxford Hector Pilot Corpus (OHPC), and supplementarily in sources such as the Bank of English (323 million-word corpus at M’s press time). No claim of comprehensiveness for English FEIs is made. M’s approach is descriptive, to ascertain the character, use, and behavior of FEIs. Using the cover term FEIs is an important decision based on the vexing nature of earlier attempts to define idioms and fixed expressions. The opaqueness of the acronym relieves readers of facing the conundrum that, as she shows, fixed expressions are often not totally fixed.

Chs. 1–3 concern background, theory, problems, and the field in general. Chs. 4–10 present specific analyses of her data. Ch. 11 is a brief conclusion. Ch. 1 (1–25), ‘Introduction and background’, reviews important earlier works, brings consensus to a messy set of concepts, and establishes the direction of the study. Bibliographically, M presents an impressively deep review of scholarship subdivided into various approaches, including largely unavailable Russian research.¹ Section 1.3 provides a classification, summary, and evaluation of ‘phraseological models’, scholarly approaches to the study of FEIs. Final evaluation for M concerns generally whether a particular approach’s findings are ‘borne out by real [i.e. corpus] evidence’ (17) as opposed to those based on intuition: ‘One of the most serious flaws in syntax-based [i.e. generative] models of FEIs is that many are based on intuition and non-authentic data’ (16). For example, though impressed by lexicalist approaches that accord status to ‘phraseological units other than pure idioms’ and that recognize the ‘range of types of fixed or semi-fixed collocation’, M warns that such research ‘may prove ultimately to be no more than an abstraction unless it can be shown to describe effectively the phenomena observed in real data’ (14).

Insistence on ‘real data’, i.e. examples of language collected in megacorpora, however, necessitates M’s declaration that her groundbreaking investigation is preliminary and suffers from the lack of using corpora massively larger than the 18 million word OHPC. Readers who witnessed the paradigmatic shift from descriptive-based structuralism to intuitive-based generativism half a century ago will hope that her discomfort with generative approaches is likewise joined by an admission that collections of ‘authentic data’ pose their own flaws because of the gap between competence and performance. We will return to this point.

Especially enlightening in Ch. 1 is discussion of the insufficiency of defining FEIs based simply on compositionality or noncompositionality. FEIs fall on a continuum between noncompositional and compositional groups of words. Her criteria for categorization of a phrase as an FEI include the process of ‘institutionalization’, the word group’s lexicogrammatical fixedness (‘formal rigidity’), and its noncompositionality (to some degree). M creates three macrocategories (or idiomaticity types), sorting the messy mix of FEIs according to problems that might lead a lexicographer to afford a given expression its own dictionary entry as opposed to letting a dictionary user look up each of the words to ascertain meaning (19). Lexicogrammatic concerns give rise to the category anomalous collocations, pragmatics problems to formulae, and problems of semantics to metaphors, with each major FEI category having three or more subcategories. For example, formulae subdivides into four subcategories including simple formulae, sayings, proverbs, and similes. The range of items includes expressions requiring fine shades of understanding, e.g. alive...