INTRODUCTION
What constitutes an idiom is a very important yet difficult initial question for all those interested in the study of idioms, including those in the business of learning or teaching idioms (Cornell, 1999; Fernando, 1978, 1996; Grant & Bauer, 2004; Liu, 2003; Moon, 1998; Tabossi & Zardon, 1993; M. M. Wood, 1981). A fairly extensive reading of publications on idioms will reveal that the definition of idiom often varies considerably from scholar to scholar. For some scholars, and in a broad sense, the term is rather inclusive, covering, among other things, all fixed phrases, clichĂ©s, formulaic speeches, proverbs, slang expressions, and, at the extreme, even single polysemic words. For example, Cooper (1998), Hockett (1958), and Katz and Postal (1963) have included as idioms individual words, especially those used metaphorically such as the word âweighâ in the phrase âweigh a decision.â Yet for other scholars, and in a more restrictive use, âidiomâ is a much narrower term, referring only to those âfixed and semantically opaque or metaphoricalâ expressions such as the bucket or spill the beans (Moon, 1998, p. 4); for a very few, the concept of âidiomâ even excludes metaphorical idiomatic expressions (Grant & Bauer, 2004). Confronted with these diverse definitions of idiom, one will have to agree with Moon (1998) that â[i]diom is an ambiguous term, used in conflicting waysâ (p. 3). What makes the term ambiguous is, of course, the fact that idioms are âmultifaceted objectsâ that are ânot only complex, but also in many ways elusiveâ (Tabossi & Zardon, 1993, p. 145). For such a complex and elusive concept, perhaps no single definition can be adequate or even possible, certainly not for scholars whose research interests vary significantly.
This chapter will first review the major theories and practices concerning the definition of idiom so as to provide the reader with a basic understanding of what scholars in general believe constitutes an idiom.1 It will then briefly explore the perspective of second language learners on idiom definition. Addressing the issue from the learnersâ perspective is necessary because learnersâ criteria for identifying idioms may sometimes differ from those of researchers due to the learnersâ unique interest in idioms: to learn to understand and use them. The chapter will end with a discussion of the major approaches and systems that scholars have employed in classifying idioms.
DEFINING IDIOMS: DIVERSE VIEWS
Of the scholars who have dealt with the issue of defining what constitutes an idiom, Hockett (1958) appears to espouse the broadest definition. To him, any language element whose meaning cannot be deduced from its structure is an idiom, including units as small as morphemes (e.g. work, ed, tele, phone, class, and room) and as large as clauses (e.g. Whatâs up?). Because Hockett is perhaps the only person who has considered individual morphemes as idioms, it is important that we understand his reasons for doing so. In his theory, work, ed, class, and room are each idioms because one cannot deduce the meaning of each of the morphemes from its structure. On the other hand, the words worked, biology, and classroom are not idioms because one can easily determine the meaning of each of these words by looking at its two composing morphemes or idioms. Work-ed means the past tense of work, biology refers to âlife studyâ or the study of life, and class-room signifies the room for classes. In fact, in Hockettâs theory every morpheme is an idiom except âwhen it is occurring as a constituent of a large idiom, since a morpheme has no structure from which its meaning could be deduced.â (1958, p. 172). Hockett illustrates his point as follows:
Thus new is an idiom in She wants a new hat, but not in Iâm going to New York, because it is part of the larger idiom New York. New York, in turn, is an idiom in the preceding sentence but not in The New York Times or The New Yorker, since in the latter expressions New York occurs as part of larger idioms. (p. 172)
However, if we use Hockettâs definition we will soon discover that there are simply too many language items that will be labeled idioms. Hence, his definition of idiom is too broad to be of much practical value in idiom research, learning, and teaching.
While Hockett appears to be the only scholar who has considered individual morphemes as idioms, there are a few who believe that individual words consisting of polymorphemes such as greenhouse and telephone should be classified as idioms. Katz and Postal (1963) and Makkai (1972) are representatives of this group. To Katz and Postal (1963), â[t]he essential feature of an idiom is that its full meaning, and more generally, the meaning of any sentence containing an idiomatic structure, is not a compositional function of the meanings of the idiomâs elementary grammatical partsâ (p. 275). In other words, any linguistic structure (including polymorphemic words) whose meaning is not the compositional meaning of its constituent parts is an idiom. On such a definition, greenhouse is an idiom because its meaning as a place for nursing plants is not the composite meaning of the morphemes green and house. Similarly, telephone is an idiom because its meaning as a device for long-distance talk does not really come from the composite meaning of the morphemes tele (far) and phone (sound)âthat is, a âfar soundâ does not equate to a âlong-distance talk deviceâ (although one may infer the meaning from the context in which the two-morpheme word is used). In contrast, words like unsafe and overestimate are not idioms because the meaning of each is indeed the compositional meaning of its constituents. For example, the meaning of ânot safeâ comes directly from the combined meanings of its two constituent morphemes, un and safe.
Of course, according to Katz and Postalâs definition of idiom, not only can polymorphemic words be idioms, but so too can phrases whose meaning is not the composite meaning of their elementary structures, such as kick the bucket. Katz and Postal differentiate these two types of idioms, though. They call the idioms made up of polymorphemic words âlexical idiomsâ and those consisting of multiple words âphrase idiomsâ (Katz & Postal, 1963, pp. 275â276). In their theory, one that Katz further elaborated in an article in 1973, the two types of idioms differ in two important aspects. First, they differ in structure, with lexical idioms belonging to âthe lowest syntactic categories (noun, verb, adjective, and so on),â and the second type falling into âhigher syntactic categories (phrases, clauses, and sentences)â (Katz, 1973, p. 360). Second, the types of idioms also differ in the way they are stored in a speakerâs language system. Lexical idioms, like all other individual words, are stored in a personâs lexicon, but phrasal idioms are registered differently in an idiom list because these idioms may have two possible meanings, with one being the composite or literal meaning of its syntactic elements (kick the bucket = âstrike the bucket with oneâs footâ) and the other being the idiomatic meaning, one that is not derived from its syntactic elements (kick the bucket = âdieâ). Using a generative-transformational approach, Katz and Postalâs identification of âphrase idiomsâ relies largely on the test of whether a phrase is productive and transformation-permissible. Phrase idioms are not productive or transformation-permissible. For example, âput oneâs foot in the mouthâ and âeat oneâs wordsâ cannot be turned into passive without losing their idiomatic meaning. This use of generative-transformational approach is the trademark of Katz and Postalâs study and it makes their analysis rather formal.
Makkai (1972) also considers some polymorphemic words as idioms, but, unlike Katz and Postal (1963), Makkai believes that only polymorphemic words consisting of at least two free morphemes, such as blackmail, may qualify as idioms. In other words, polymorphemic words made up of only one free morpheme plus one or more bound morphemes, such as telephone (tele as a bound morpheme or an affix and phone a free morpheme), cannot be classified as idioms because there are adequate morphological rules for decoding first the bound morphemes and then the words containing the morphemes. For example, knowing the meaning of the affix tele, one may fairly easily predict the meaning of a word containing it, such as telescope. In other words, the meaning of such a polymorphemic word is deducible from its constituent parts. In contrast, the meaning of an idiom is not deducible from its elementary parts because, Makkai (1972, p. 120) argues, these elementary parts so put together can âpotentially misleadâ or âdisinformâ a listener or reader. For instance, the meaning of blackmail as âextortion by threatsâ is not derived from the meanings of the two free morphemes, black and mail. So it seems that Makkaiâs and Katz and Postalâs definitions are really the same in the sense that both insist that the meaning of an idiom cannot be derived from its individual components. Yet they disagree on whether the meaning of a word consisting of one bound and one free morpheme can be derived from its components, with Katz and Postal saying yes and Makkai claiming no. Thus, in terms of polymorphemic idioms, Makkaiâs definition appears narrower than Katz and Postalâs. Yet Makkaiâs overall notion of idioms is much broader than Katz and Postalâs, as will be shown below.
Applying Lambâs (1962) stratificational grammar, Makkai renders a very formal study of idioms. He identifies two major types of idioms: idioms of encoding (also called phraseological idioms) and idioms of decoding (also known as semantic idioms). The former refers to stable collocations peculiar to a language. An example cited in his discussion is drive at 70 m.p.h. rather than with 70 p.m.h., as is used in French. These expressions are called idioms solely for their phraseological peculiarities (Makkai, 1972, pp. 56â57). The latter type, idioms of decoding, on the other hand, comprise expressions that possess a non-literal meaning such as red herring and come up. These idioms are all potentially misleading or have âdisinformation potential,â as can be seen in the two examples just mentioned. Each of them has both a literal and a figurative meaning: red herring= (1) a type of fish that is red, and (2) a phony issue; and come up = (1) move to a position as in âcome up to the front,â and (2) occur or be mentioned as in âthe issue came up at the meeting.â All idioms of decoding are, of course, idioms of encoding because they are also rather fixed and unique in composition. Because of their meaning opacity and structural uniqueness, idioms of decoding are the focus of Makkaiâs study.
Like Katz and Postal (1963), Weinreich (1969), another important scholar on the topic, also employed the principles of generative-transformational gr...