--------------------------------------------------------------------------------

READING: Fernando, Ch./Flavell, R. 1981. On Idiom: Critical Views And Perspectives. (Exeter Linguistic Studies, 5). Exeter: University of Exeter, pp. 18-48

--------------------------------------------------------------------------------

CHAPTER II

THE NATURE OF IDIOM

(in: EXETER LINGUISTIC STUDIES, Vol. 5, ON IDIOM - Critical views and perspectives, BY Chitra Fernando and Roger Flavell, University of Exeter, 1981)

2.1 Characterising Idioms

The central problem one comes up against in attempting to define idiom is identifying the property (or properties) which will adequately capture all the idioms in a language while excluding all the non-idioms. A number of approaches to idiomaticity together with various definitions of what an idiom is have already been presented in the previous chapter. If one narrowed down the consideration of idiomaticity to the specific problem of definition it is possible to identify two approaches to what idiomaticity is:

(i) Those scholars who adopt the first approach look on idiomaticity as manifesting the specific character or genius of a language. Their investigations of idiomaticity are directed towards revealing this specific character which is, in effect part of the underlying conceptual design of the language. Such an approach ultimately leads to the nature of cognition itself and therefore has strong psycholinguistic implications. The chief exemplars of this approach to idiomaticity in the Anglo-American tradition of linguistics, Smith (1925) and Roberts (1944), do not carry their investigations to these depths. Their work, already referred to in Section 1.1, simply outlines the cultural preoccupations, the 'world view' implicit in the idioms of English, together with the pecularities of phrasing and other distinguishing features (e.g. non-literalness) that distinguish expressions as idioms. But the main emphasis in such work is on the conceptual design of the language in so far as it emerges through a consideration of idiomaticity rather than on the structural properties of idioms.

(ii) Scholars who adopt the second approach are more structurally orientated and seek to define idiomaticity in terms of one or more structural properties. They are, therefore more selective in their indentification of idioms. The second approach, in addition, enables the linguist to make topological classifications of such idioms on the basis of the properties he adopts as criteria. The majority of the scholars whose work has been examined in Chapter I adopt this second approach which is also the one adopted in this chapter.

2.2 Morpho-syntacitc Composition

Of the five properties listed in Chapter I the fact that an idiom is non-literal functions as the highest common denominator of idiomaticity. As far as definitions encapsulating properties go, Hockett's is the most explicit (Section 1.2)

p. 19

and has been adopted in one form or another by several later scholars. Makkai rejects Hockett's definition on the grounds that it is a catch-all for virtually everything in a language and argues that it is more economical "to use the term idiom only for units realised by at least two [free] morphemes" (1972: 58). We have, accordingly, two conflicting criteria, a conflict that is reflected in the variety of morphological forms that have been identified as idioms: bound forms, single free forms, compounds, phrases and sentences. As far as idiom types go such forms range from proverbs and metaphors to a variety of set phrases including rhetorical questions like Has the cat got your tongue? and social formulae like Long live X. This multiplicity of morphological forms and idiom types gives rise to two major problems:

(i) Should one of the defining criteria of idiom be morphological structure? Should such a structure entail a constraint which would result in a minimal and maximal size for an idiom?

(ii) How should the fact that idiomaticity is a phenomenon lacking in monolithic uniformity be best brought out? There are various types of idiom. If the fact that the meaning of an idiom is not the compositional function of its constituent parts (i.e. an idiom is non-literal) is taken as the highest common denominator of idiomaticity, then obviously the meaning of kick the bucket is much less easily deducible from its constituent parts than long live X which borders on the literal. Yet the latter has some claim to be considered an idiom in terms of the fact that it is a set expression which is institutionalised as such in a given language.

A solution is offered to problems (i) and (ii) by recognition of the fact that idiomaticity is a phenomenon too complex to be defined in terms of a single property.Idiomaticity is best defined by multiple criteria, each criterion representing a single property. If idiomaticity is so defined, certain types of idiom will be seen to possess more distinguishing properties than others. There exist, in other words, varying degrees of idiomaticity correlating with different types or categories of idiom. The adoption of multiple criteria would enable the investigator to filter out the non-idiomatic while retaining all those forms which show one or more of the properties of idiom. Hockett uses a single criterion for defining and identifying idioms: that the meaning of an idiom is not the compositional function of its constituent parts. Makkai uses five: morphological composition, the susceptibility of an idiom to literal interpretation, ambiguity, semantic unpredictability and institutionalisation. The result of using multiple criteria is that Makkai's definition of idiom is more explicit and his identification of what forms are idioms more selective than Hockett's.

p. 20

The foregoing discussion has centred mainly on the question of how the key defining property of idiom - the fact that the meaning of an idiom is not the compositional function of its constituents - relates to the issue of the compositional structure of an idiom. Hockett admits even bound and single morphemes as idioms. We are, however, in agreement with Makkai when he implies that such a classification would obscure the distinctive nature of idiom. An even more pertinent argument is that the majority of bound morphemes in English function predictively as affixes so that even though they may be more than the sum of their component parts in terms of phonemic composition, their signification is predictable at the morphosyntactic level of the language. They lose their idiomaticity in semantic terms and thus lose the defining property which led them to be defined as idioms in the first place. Morphs such as -ceive or -tain which gain their morphemic status from combining with other morphs, re- and de- for example, result in deceive/receive and detain/retain. But none of these morphs have grammatical meaning in themselves, unlike other phonologically similar morphemic affixes. In the case of the morphemic affix de- of de-freese or de-escalate and the morphemic affix re- of re-write or re-analyse meaning may be attached to the bound morphemes in question, namely 'negative' and 'repetition' However, the predictability of such items renders them literal.

In our view idiom is a morpho-syntactic phenomenon, for it is at this level of the language that the key property of idiom, the asymmetry between sense and syntax, manifests itself most unequivocally. On the other hand we also recognise that there are a large number of monomorphemic forms such as gay, croak, bastard, swine, lemon, etc. which have both a literal and a nonliteral signification and which, therefore, have some claim to idiomaticity. The structural cut-off point between the idiomatic and the non-idiomatic is a relatively arbitrary matter since idiomaticity saturates the morpho-syntactic level of a given level of a given language. We regard single items which have a contrasting literal homonym such as those listed above as constituting an indeterminate zone of idiomaticity. In view of such indeterminacy we have chosen to concentrate on those items which are constituted by two or more free morphemes. In other words, we regard the lower structural boundary of idiomaticity as being the compound lexeme (e.g. foxglove, egghead, sweetheart), while the upper structural limit is constituted by the sentence (e.g. Don't count your chickens before they are hatched ).

A stretch longer than the sentence may be rendered idiomatic by its non-literalness, but such mini-discourses do not become idioms but remain nonce-items bound to one specific situational context. Those items that eventually gain idiom status are generally short, easily memorisable items like single words, compounds, phrases and syntactically simple sentences such as

p. 21

the coast is clear, a stitch in time saves nine, a rolling stone gathers no moss, etc. The general tendency is towards deletion and hence reduction. To draw a red herring across the trail is usually shortened to red herring, a bird in the hand is worth two in a bush to a bird in the hand, a rolling stone gathers no moss to X is a rolling stone, etc. The principle of least effort is in operation all the time.

The operational tests most frequently used for establishing the boundaries of idiom are those of removing or replacing the suspected idiom's morphemes one by one (Jespersen 1924, Healey 1968). The removal test does not separate the idiomatic from the non-idiomatic with complete effectiveness since there are literal expressions which do not admit of any such removal either. Thus it is no more possible to remove a single morpheme from She is my daughter(l) or on the grass than it would be from Don't count your chickens before they are hatched or pay through the nose. But the removal test has some validity in instances such as spill the beans (cf. he hasn't got a bean ), dig in one's heels, stoned to the eyeballs, pulling strings, etc., contrasted with red herring(s), (a) skeleton(s) in the cupboard, white elephant(s), etc. The bound plural morpheme in the first set of examples is a necessary constituent of the idiom, but this is not so in the second where it is a variable.

The replacement or substitution test is generally more effective in the establishment of idiom boundaries since the possibility of substitution either converts an idiom into a non-idiom (pay through the nose ® pay heavily ) or indicates the presenceof some non-idiomatic variable. X gave Y a kick in the pants, X paid through the nose for Z. X is a male chauvinist pig, etc. Any attempt to replace any other morpheme in, for instance, X paid through the nose for Z (* paid down the nose, *paid through the eyes, etc.) results in loss of idiomaticity. Hence the minimum form of the idiom is pay through the nose. Similar observations would also apply to the two other examples cited above.

The point that emerges from this section is that idiomaticity is too all-pervasive to be correlated with a specific form of morpho-syntactic structure or the presence or absence of syntactic constraints such as given transformations. As Randolph Quirk states:

The problem of idiomaticity is rather that most phenomena in language respond very well to treatment by the procedures that have evolved for handling 'syntax' on the one hand and 'lexicon' on the other.

p. 22

But languages include entities that are hermaphroditic in sharing the properties of both and responding satisfactorily to the descriptive procedures of neither ... There is nothing analogous to the Olympics requirement that all humans must be 'men' or 'women': you can frankly admit that your world embraces a wide range of linguistic objects for which the idealised generalities of syntax do not fully apply any better than the monemic specifics of lexicology. The task is thus to evolve a linguistic framework that can handle this range of entities .... (personal communication)

This kind of linguistic framework which is capable of handling the intrinsic diversity of idiom is looked at in Chapter III. But we devote the rest of this chapter to examing more closely the features which idioms (2) as diverse in form and type as Don't count your chickens before they are hatched, the coast is clear, keep one'sfingers crossed, spill the beans, red herring, egghead, etc. have, in order to arrive at a configuration of properties which we could regard as defining idiomaticity. Our findings incline us to the view that idiomaticity is the outcome of the interaction of a form, a sense and a situational context.

In conclusion, though we regard the structural composition of an idiom as useful in determining the upper and lower limits of idiomaticity, we have chosen to concentrate on other properties in attempting to formulate a definition of idiom. In trying to separate the more idiomatic from the less idiomatic and from the non-idiomatic, we regard those properties that establish the semantic unity of idioms as more important than those that establish points of grammatical contrast (e.g. transformational constraints). Such properties are of' three sorts: semantic, syntactic and sociolinguistic. We shall take them up in this order in Sections 2.3, 2.5 and 2.6.

2 3 Semantic Properties

The meaning of an idiom is not the result of the compositional function of its constituent parts, i.e. its syntax is non-correlative and the resultant signification non-literal in terms of the referents denoted by these parts. Or, put briefly, an idiom is a lexeme. Here 'lexeme' is used to mean the smallest semantically irreducible unit, whether expressed formally in monomorphemic words or polymorphemic words or expressions, e.g. cat, man, house, antidisestabtishmentarianism, demythologisation, go out with, drop a brick, pass the buck, etc.

(2) Such forms have been accepted as idioms in the consensus of scholarly opinion or in idiom dictionaries.

p. 23

2.3.1 Semantic Structure

Obviously idioms and lexemes are not interchangeable synonyms; single morphemes for instance are lexemes but not necessarily idioms, although idiomatic compounds and phrasal verbs (e.g. foxglove, come down with, etc.) are lexemic. Harold Conklin (1962: 121) comments on this issue:

A full lexical statement (i.e., an adequate dictionary) should provide semantic explanation, as well as phonological and grammatical identification, for every meaningful form whose signification cannot be inferred from a knowledge of anything else in the language. It is convenient to refer to these elementary units as 'lexemes', although other terms have been suggested, for example 'idiom' by Hockett.

An idiomatic phrase, for example spill the beans, generally gives one reading which is the superficial, literal sense derived from the correlative syntax of that phrase. The meaning here is transparently clear, as in sentence (1) below. It is only minimally misleading in less clear-cut contexts in so far that the phrase may be known as an idiom, and the idiomatic meaning expected, whereas in fact the literal reading is to be taken. The other reading which an idiom gives is more confusing in that if an idiom is not suspected then unless the context is particularly clear an unwary reader might be tempted to take the surface reading that derives from the correlative syntax and miss the idiomatic sense. In (2), for instance, a literal reading would be just possible, with some semantic incongruity. In (3) a reader is much more likely to look for an idiomatic interpretation.

(1) Alfred spilled the beans all over the table.

(2) Alfred spilled the beans all over the boarding-house.

(3) Alfred spilled the beans all over the town.

It is with (3) that the semantic opacity of idioms is clearest. There is no discernible link between the literal reading of (1) and the meaning of (3) which is paraphrasable as Alfred revealed secrets to people everywhere in town. Another example we might use to demonstrate the semantic opacity of idioms arising from the asymmetry between syntax and sense is the phrase kick the bucket. The method of killing pigs which gave rise to the expression kick the bucket has long since died out. It is not generally known nowadays that the bucket was (in Norfolk, England) the beam from which a pig about to be

p. 24

killed used to hang by its hind feet. The nervous reflex actions of the pig, after its throat was cut to let the blood drain into a receptacle below, caused it to kick the bucket as it was dying.

What happens in these and other cases is that independent lexemes ( spill, the, beans; kick, the, bucket ) initially are used in a straightforward, literal way such that the resultant meaning of an expression formed with them is simply the sum of the constituent lexemes. Slowly, through various semantic processes, these expressions take on new, idiomatic meanings and at the same time calcify syntactically. At a certain stage of this diachronic process it is no longer possible to talk of lexeme + lexeme giving the full meaning, i.e., a polylexemic construction, but only of one single lexeme, with a sense not obviously connected with that of the original constituent lexemes, i.e., a monolexemic construction.

The meaning of the idiom has now to be expressed by means of a paraphrase. In fact the original lexemes are now positively misleading, and do not carry any independent meaning of their own

(4) The pig kicked the bucket.

(5) John kicked the bucket accidentally, and hurt his toe.

(6) John kicked the bucket and they buried him next day.

In (4) and (5) each item carries its full lexemic value; in (6) only two lexemes are present: John and kicked-the-bucket (='died'). In sentence (6) kicked, the, and bucket are pseudolexemes which make possible an erroneous interpretation of the expression, if it is not recognised that they have no independent meaning of their own but together form a monolexemic construction with the meaning 'die'.

Any attempt, therefore, to give an independent meaning to the individual constituents of idioms, rather than to the whole unit, is open to criticism.

At first sight, the so-called 'idiom variants' appear to throw into doubt this principle of the semantic integrity of the whole idiom. Idiom variants are such pairs or triplets as: bury the tomahawk/hatchet, lead someone a merry chase/dance, not to touch something with a ten-foot/barge pose, blow one's stack/cool/top, hit the hay/ sack, etc. The point here is that, despite difference of register frequency, dialectal usage and so on, the same lexemic sense is involved in each pair or triplet, even though there is variation

p. 25

in the form. There is still no independent meaning given to the idioms' constituents. That idiom variants exist is not a proof against the lexemic nature of idioms; phonemic and morphemic variations do not affect the concept of the phoneme or the morpheme.

A necessary corollary of the view that no constituent of an idiom carries independent meaning is that if a constituent of a suspected idiom can be shown to recur elsewhere with the same sense, then the phrase is at best only marginally idiomatic. Phrasal verbs provide many illustrations of this principle: fill up might be considered idiomatic, until contrasted with the recurrent completive or effected sense of up in eat up, drink up, dry up, etc.; head for is non-idiomatic since for with the sense 'in the direction of' recurs in leave for, set out for, etc.

A further example is blue joke. It might be suspected that blue joke (meaning 'obscene humorous story') is a pure idiom. This is not so since joke can be replaced by a range of other items (e.g., play, film, gag, picture, comedian, etc.), and obviously blue retains its sense of 'obscene' outside the collocation with joke.

Another illustration is hit the high spots. It is reasonable to conclude, after consideration of hit the road, hit the town, hit London (or any (large) town), etc., that hit is often collocated with places, with the connotation of speed and hurry. It has a recurrent meaning of its own. The potential in this and the other examples above for productive, or quasi-productive, patterning rules them out as pure idioms, since the recurrence of the same meaning in different contexts of a constituent of a phrase debars the whole expression from having the status of pure idiom.

2.3.2 The Formation of Idioms

Several attempts have been made to 'generate' idioms (Weinreich 1969; Chafe 1970; Fraser 1970). It seems to us that since the process of 'idiomaticization' lies in diachronic evolution, idiomaticity cannot be adequately explained by generative rules. Neither a given sense or a given syntactic structure by itself constitutes an idiom. Rather it is the regular association of one with the other that is the source of idiomaticity. Such an association is the product of contextual extension. in the everyday situations of communicative use over a period of time. The majority of idioms, it is agreed by idiomatologists (Smith 1925; S.I. Ozegov 1957; Hockett 1958; George Vallins 1960; Healey 1968; Chafe 1970; Makkai 1972), exhibit

p. 26

certain stages in their development. We have already drawn attention to one such association in our brief discussion of kick the bucket (see Section 2.3.1). An individual uses a phrase, which generally includes words referring to everyday concrete objects, figuratively in a particular situation. His interlocutor, then the local speech community, and finally the body of native-speakers accept the metaphorical use into their own speech for its aptness, felicity, picturesqueness or even plain illogic. In fact, Smith (1925: 187-9) considers irrelevance, illogic and absurdity to be important factors in the genesis of many idioms. Pierre Guiraud (1961: 10) similarly says:

la bizarrerie, voire le non-sens, est une source de succes et de survie pour de nombreuses locutions.

In modern English, consider, for instance, the tendency to say You can't have your cake and eat it rather than the logical You can't eat your cake and have it.

Those forms which seem to have no figurative stage are in a minority. In these cases, there is instead a slow hardening, a fixing of the form, through repeated use, usually accompanied or followed by a change in meaning, for example, now and then, now and again, by and large, fancy that, nothing doing, etc. Julio Casares (1950: 222-3) demonstrates a similar process with regard to many Spanish adverbials.

Where there is metaphorical use of the kind referred to above, the phrase in question may undergo the well-known processes of semantic extension, specialisation, etc. which operate in any shifts of meaning. The resulting phrase may become an idiom, with no obvious link with its literal predecessor (which of course in most cases still exists independently). There has also probably been a move from the 'concrete' vocabulary of the literal expression to more 'abstract' meaning. That the concrete and tangible, often in conjunction with an action verb, becomes the abstract and intangible has long been observed in relation to proverbs as well. This is especially relevant since many proverbs have claims to be considered idioms. The analysis of the sources of idioms in Smith (1925), Guiraud (1961) and Ernest Rogivue (1965), to cite just three authorities shows the everyday concrete origin of idioms. From a theoretical standpoint, the theories of Heinz Kronasser (1952) and Ferenc Kovacs (1961) have been verified by Chafe and Makkai. The objectively verifiable statistical evidence on phrasal verbs presented by Makkai in Part Two of his Idiom Structure in English (1972) is particularly convincing proof of 'Kronasser's Law'.

p. 27

At any particular moment, therefore, synchronically there are (i) literal phrases, (ii) metaphors which are clearly connected with the literal, (iii) other more opaque but possibly interpretable phrases, and (iv) opaque idioms. To take an explanatory example, in the second category there probably comes a phrase such as skate on thin ice meaning 'to court danger'. There, the jump in the metaphorical transfer from the literal to the figurative is sufficiently small for the link between the two readings to be easily discernible. The future meaning of the phrase is of course unknowable, but various hypotheses can be made in an attempt to show the processes involved. In the dim and distant future, in a temperate climate, ice, and a memory of it, might totally disappear. In this case skate on thin ice in its idiomatic meaning would be unmotivated and the link between literal and metaphorical broken so that it would now be legitimate to class the phrase in the third, semi-opaque, category. Similarly, if for some reason people should give up skating, the phrase could simply change its meaning: 'to court danger' could become 'to court danger ostentatiously', then 'to be a show-off' which eventually could be used in a wide range of contexts in no way tied to the collocational range of the present-day skate on thin ice: The greatest defect of his character was to skate on thin ice in public. He has a desperate need for love and recognition.

All this is purely speculative. Idioms which are already in the third and fourth categories show verifiable evidence of similar changes in their etymologies. An extreme case in point is the phrase to pull someone's leg. This expression is believed to come from the old practice of pulling the legs of a man in the process of being hanged, to speed his death and so spare his agony. By a long process of semantic change this same phrase now means 'to make gentle fun of'. Quite clearly there is no link between the two expressions synchronically. The idiom is totally opaque. Similarly, proverbs and nursery rhymes can be equally removed from their origins. Ring-a-ring-a-roses, for example, comes from the deaths caused by the Plague, yet is today a children's nursery rhyme. Although a historical explanation of idioms is illuminating in showing the forces at work and in demonstrating why certain idioms are in fact opaque or transparent, the decision as to which category an idiom is to be ascribed to must be taken on the basis of synchronic, not diachronic, evidence. Primary is the present-day status of an idiom; diachronic considerations provide useful secondary corroboratory evidence.

Weinreich (1969: 77-79)totally rejects any attempt to study the meaning relations between literal and idiomatic expressions. To him each case must be an individual historical study and the semantic motivation is something that can only be established a posteriori. He concludes:

I feel that the relation between idiomatic and literal meanings is so unsystematic as to deserve no place in the theory.

p. 28

The point, however, is not to turn everyone into an amateur etymologist, attempting to link historically the literal phrase with the contemporary idiom, but rather to assess how much of a link is perceived to exist synchronically between the idiomatic meaning of a phrase and the cumulative literal meanings of the correlative elements of the expression. To take the example cited earlier, one can skate literally on thin ice; one can skate on thin ice idiomatically and court danger. How clear is the relation felt to be between these two senses? If the answer is that there is no connection, the idiom is opaque and non-motivated. If, as perhaps in this case, there is considered to be a fairly clear link between literal and figurative meanings, then the phrase is semi-transparent and well-motivated. Motivation here is taken in the Saussurean sense of a perceived semantic link between a metaphorical expression and its literal counterpart. The most motivated expressions are those labelled here 'semi-transparent', unmotivated expressions are 'opaque'. This is after Ferdinand de Saussure (1955: 180ff); Stephen Ullmann (1962: 92, and note 2) extends Saussure's original formulation which was applied only to compounds and derivatives, and applies it quite legitimately to figurative language in general.

The usefulness of attempting to make an assessment of the strength of the link between literal and figurative meaning is that the expressions which cluster at the opaque end of the scale can be called 'pure idioms'. In their case the figurative link between literal and idiomatic is broken. In the case of phrases at the transparent end of the spectrum, the metaphor is clearly perceived and still alive and kicking. In these cases it is fairly easy to deduce the 'idiomatic' meaning from the literal constituents. So, despite Weinreich's objections, it is considered here valuable to establish a ranking as an aid to distinguishing between simply metaphorical, figurative phrases and the more truly idiomatic ones.

It has been suggested that there is a continuum from totally transparent to the totally opaque. For practical purposes it is necessary to divide arbitrarily this continuum into certain categories in order to obtain a convenient scheme to assess the degree of motivation of a given expression. The following table gives one suggestion as to how this could be done.

Literal

TRANSPARENT

cut wood, break eggs, a pink shirt, bring in, rely on, etc.

Metaphors

SEMI-TRANSPARENT

skate on thin ice, kill two birds with one stone, add fuel to the fire, etc.

SEMI-OPAQUE

burn one's boats, tarred with the same brush, the boot/shoe is on the other foot, jump down someone's throat, etc.

Idioms

(iv)

OPAQUE

pull someone's leg, pass the buck, foxglove, off the top of one's head, trip the light fantastic, etc.

p. 29

The decision whether a given phrase is an opaque idiom or a transparent metaphor can be rather subjective. When faced with an unknown expression, one has no way of knowing whether it is to be interpreted metaphorically to arrive at the actual meaning, or whether in fact it is a totally unguessable opaque idiom. For instance, in an otherwise excellent practical book on teaching idioms, M.J. Murphy (1968) uses a lesser-known idiom kick up one's heels. This was given to native-speaker informants who were asked what it meant. The consensus was that it meant 'to be rebellious', 'to reject authority', or (idiomatically) 'kick against the pricks'. This is perhaps the most obvious metaphorical interpretation, but it is not in fact correct. Murphy defines it as 'relax and have a good time', similar to 'let one's hair down'. Clearly, when faced with an unknown phrase, the tendency is to give it its most likely metaphorical interpretation. Non-native speakers do exactly the same thing: have cold feet is often classed by them as semi-transparent, meaning 'to be ill' or something similar, rather than the opaque 'to be afraid'.

Even in the case of apparently simple metaphors, such as buttercup, the interpretation is not easy. The compound may be metaphorical, but which metaphor is applicable? If buttercup is unknown, as it may be to a non-native, it could be the flower Ranunculus bulbosus, a yellow bowl of a certain design, a special kind of cake (cf. hot dog ), or a thousand and one potential associations.

Sufficient context will give the meaning, at least in general terms. But the interpretation of metaphors and idioms is only reasonable after the event. The tendency is that the native speaker classes phrases as more transparent than they actually are because of his inbuilt a posteriori knowledge of their actual meaning. Since native speakers appear to have this inbuilt hind-sight, it is suggested that the following scheme to define whether a phrase is simply metaphorical or opaque and so idiomatic should be presented to both natives and non-natives, for one group to act as a check on the other.

Questionnaire 1 A (see Authors' Note for more details) is a battery of phrases given to native speakers as part of a project to assess the degree of figurative motivation implicit in idioms. For each phrase they are asked to decide if there is any figurative reading. If so, they are asked if its literal reading is to them obviously linked with the figurative meaning, whether it is doubtfully or unclearly linked, or whether they cannot see a connection at all. The corpus of idioms would then be distributed under the headings indicated in the above table. Questionnaire 1 B involves exactly the same battery of phrases as in Questionnaire 1 A, but is given by the investigator to non-native speakers. Once they have assessed the transparency or opaqueness of each

p. 30

phrase, the results of the two questionnaires are matched and averaged. Preliminary results show that natives class phrases as much more transparent than non-natives, pointing to the fact that native-speaker hindsight is a considerable force. It is one thing to consider a meaning reasonable and clear with prior knowledge of that meaning, but quite another to deduce the meaning of a form from its constituents as a non-native has to.

The overall results are one answer to the difficulty Ullmann points out of devising a statistical method to determine how far people are aware of the motivation of expressions. Taking caterogies (ii) to (iv) - since literal expressions are not the concern of tests for semantic motivation - phrases which are finally allotted to category (ii) are well-motivated metaphors. Their clearly figurative nature is not in question. Those assigned to category (iv) are clearly idiomatic. The metaphorical link is broken and not perceived even dimly by speakers. Category (iii) represents the grey area in which there is dissension and doubt amongst informants. To some, a phrase found here retains a perceptible metaphorical link with its paired literal counterpart; to others this connection is non-existent, hence the phrase is idiomatic rather than figurative.

The conclusion to be arrived at from the discussion above is that a pure idiom must have constituent elements from which the overall meaning of the whole is not deducible. Such a viewpoint not only raises the issue of the connection between idiom and metaphors (see above) but also a number of others. The relationship between proverbs and idioms, for instance, is best pictured as 2 circles which overlap. While some proverbs range from semiopaque to opaque: don't count your chickens before they're hatched; he who pays the piper calls the tune; he who laughs last, laughs longest, etc., others are quite literal: if at first you don't succeed, try, try, try again; never put off till tomorrow what may be done today; practise what you preach; spare the rod and spoil the child; who chatters to you will chatter of you; children should be seen and not heard, etc.

2.3.3 The Idiom Family

The idea of the 'idiom family' provides some equally troublesome problems. Idiom families can be classified under two main headings, which in practice usually overlap to some extent: the 'formal' idiom family and the 'concept' idiom family. The first of these involves a set of idioms sharing a similar syntactic pattern and at least one lexical item. For instance:

p. 31

HIT BLUE

A hit the sack

hit the hay

B hit the town B blue Joke

hit the high spots blue film

hit the road blue comedian

C hit the bottle C turn the air blue

hit the ceiling scream blue murder

hit the mark blue stocking

hit the jackpot

The second of these types, the 'concept' idiom family, contains pairs such as bury the hatchet and take/dig up the hatchet. Further, white-collar worker (which itself replaced black-coated worker historically) has spawned blue-collar worker and iron curtain finds its Oriental counterpart in bamboo curtain.

The difficulties presented by the 'formal' idiom family can be satisfactorily explained. Examples under A above are idiom variants as discussed previously. Examples under B are semi-idioms, since a recurrent meaning for one of the phrases' constituents can be isolated, as also discussed above. Examples under C share a lexical item and a common structure (in some cases) but, most importantly, subject to a closer analysis, have no semantic links. The 'formal' similarity of the common lexical item and structure is coincidental.

As for the 'concept' family, the patterning inherent in the creation of blue-collar worker and bamboo curtain suggests a degree of productiveness which rules these expressions out from being considered as fully-fledged idioms. It is important to bear in mind the metaphorical nature of these expressions. The new coinings were made diachronically at a time when the original metaphor was still living and available for extension and further application. Metaphors, by definition still 'living', often spawn new ones. However, synchronically, it is a matter of opinion whether these forms are now 'dead' and more idiomatic, or not. It may well be that a person unaware of the etymology of iron curtain ( in fact originally coined in telegrams to President Truman in 1945 by Winston Churchill, and first publicly used by him in a speech at Fulton, Missouri, in 1946) and the political situation, and quite possibly unaware of the phrase bamboo curtain, might consider both expressions to be opaque and inexplicable. These would then be classed as at least semi-opaque idioms. At the present time it is a matter of opinion whether the original metaphor is sufficiently 'dead'.

p. 32

As a general rule it seems true that the 'concept' families are restricted to metaphorical phrases at an early stage of development. From a present-day synchronic point of view, there are the still-productive metaphors in families, and the once productive metaphors which have ossified, or are in the process of doing so, into at least semi-opaque idioms.

2.3.4 Conclusions

We have examined several issues which focus attention on the idiom as a single lexeme that is non-correlative in its syntax and therefore non-literal in terms of its constituents. The most satisfying and sensitive criterion to establish idiomaticity is undoubtedly the semantic one. In the first place, definitions of idioms by linguists, lexicographers and language teachers are nearly always couched primarily in semantic terms. Such a semantically-based definition is intuitively more satisfying for the ordinary non-specialist native speaker. When asked for a definition of 'an idiom', his first thoughts are that it does not mean what the individual words in it would lead you to expect. Semantic considerations underline most of the criteria proposed for idiomaticity. Obviously the transparent-opaque axis is semantic in nature, and the fact that idioms are unanalysable lexemic wholes is a semantic statement.

In our opinion, there can be little doubt of the primacy of semantic criteria in establishing the idiomaticity of any expression.

2.4 Homonymity: A Syntactico-semantic Property

An idiom has a literal homonymous counterpart which complements the fact that its syntax is non-correlative, i.e. a pure idiom constitutes a 'double exposure'.

To Makkai the most important feature of idioms of decoding is that they are expressions which are ambiguous and therefore potentially misleading. The specific term used by Makkai is 'disinform'. Makkai contrasts the disinformation potential of idioms arising from the possibility of "logical yet sememically erroneous decoding" with the 'misinformation' which occurs as a result of "accidentally homophonous forms" having "equally meaningful decodings" in similar environments, for example She bears children signifying both 'carries' and 'gives birth to' (1975: 122). We have already observed in Chapter I that Makkai implies an opposition between idiomaticity and homonymity whereas the disinformation potential of idioms arises, as we see it, from the fact of their possessing homonymous literal counterparts. The relation between homonymity and disinformation is recognised by Weinreich (1969: 43-44). He argues that in terms of normal colloquial language use we would not use the phrase bacon and eggs for packaged bacon

p. 33

and a carton of eggs. Yet this is not an impossibility in terms of the referents denoted by the phrase. It is the legitimacy of this possibility that constitutes for Weinreich, as it does for Makkai, one of the essential features of the pure idiom, though the former fails to carry this criterion far enough, as in the case of by heart, red herring and rain cats and dogs, and thelatter could, perhaps, have explained more fully the usefulness of this feature.

It would appear to us that homonymity is an essential feature of all pure idioms. This would exclude all those constructions which violate truth conditions and, therefore, have no literal counterparts, as well as those which are phraseologically peculiar. The first category includes expressions such as rain cats and dogs, move heaven and earth, throw caution to the winds, jump down somebody's throat, lose one's heart or one's cool, be beside oneself, put one's heart and soul into something, fish for compliments, etc., which are all forms of hyperbole or some other figure.

The second category, set phrases, includes by heart, happy-go-lucky, to kingdom come, by and large, do a (Henry Kissinger/Watergate, etc.). The problem centres on deciding whether one is justified in excluding these two classes of expressions from the class of pure idioms on grounds of their lack of homonymous literal counterparts and, if one does, what the advantages of such exclusion would be.

Certain set expressions manifest what may be termed 'a double exposure' (Henry G. Widdowson, personal communication), i.e. they manifest a non-literal and a literal meaning. Non-literal idiomatic meaning generally arises as a result of a figurative extension. Yet the original situation which is the source of the extension is often unperceived by the average speaker. One way of distinguishing true idioms from metaphorical expressions which may show some of the properties of idiomaticity is to test their disinformation potential, i.e. do they have homonymous literal counterparts which could render them misleading in certain contexts? Since the members of the first category referred to above do not have such homonymous literal counterparts, they fall into a class of set metaphorical expressions related on the one hand to the class of idioms and on the other to the class of fossilised metaphors with which language is riddled, such as the eye of a needle, the arm of-a chair, the head of the family, etc.

Such a test would also exclude similes both explicit and implicit: like the pot calling the kettle black, like a bolt from the blue, like the tail wagging the dog, like a cat on hot bricks, as different as chalk from cheese, shed crocodile(i.e. shed tears 'like a crocodile'), etc. The homonymity criterion also excludes metaphors of the form X is Y: X is: big fish in a small pond, a fish out of water, a wet blanket, a bookworm, an ugly duckling, etc.

p. 34

We would regard a red herring, one of Weinreich's key examples of idiom (1969: 42) as occurring lower down on a scale of idiomaticity than kick the bucket or smell a rat. But it must be admitted that the dividing line between the more and the less idiomatic can be blurred by context. When a red herring is used, for example, as a subject complement, or as an appositional modifier, its import is indisputably metaphorical:

As is so often the case with conspiracies, Mr. Coulson's rationale for the union's actions is a red herring. (The Australian August 1975)

Professor Evans' letter should be treated with a great deal of caution. He criticises Mr. Williams for his "pseudo-statistics" on illiteracy and for ignoring the existence of migrant children in schools, the most vermilion of red herrings. (The Australian September 1975)

To say that something is a red herring is parallel to the assertion that someone is a beast, a wolf, an ass, a chicken, etc., all expressions no one would claim to be idioms of the same sort as kick the bucket or smell a rat. On the other hand, the metaphorical force of a red herring is substantially filtered out by a context like the following:

Red herrings and the Iraki breakfast

But Mr. Whitlam has to talk about these things - any red herring will do -because as he showed us last week, he is unwilling to talk about the Iraki breakfast. He has to divert our attention from the uncomfortable fact: the Iraki breakfast happened .... For, however some sections of the Labour Party allow themselves to be hypnotised by Mr. Whitlam red herrings, the rest of the world still grasps the salient point of the affair - the Iraki breakfast happened. ( The Australian March 1976)

In a typical idiom such as blow one's own trumpet or be in a hot spot the idiomatic and literal meanings are capable of appearing simultaneously as a 'double exposure', a kind of pun, in one and the same context:

The Australian Youth Orchestra should blow its own trumpet much more often on stage and off. After its half empty pre-Rostrum 75 concert at the Sydney Opera House on Wednesday night false modesty is out. It needs more bottoms on seats. ( The Australian August 1975)

Fire chief in hot spot

Canberra's chief fireman, Mr. Jack Mundy, is standing alone between his city and disaster. All his 100 firemen are on strike but if he gets the call he dreads, he'll do his duty and take a fire engine out alone. (The Australian September 1975

p. 35

The second class of items, set phrases, for example happy-go-lucky, referred to above, also deficient in respect of homonymous literal counterparts, consists of a class of expressions which in addition to having some figurative force also shows a phraseological peculiarity. Once again the fact of such expressions not having homonymous literal counterparts enables us to separate set phrases which show two of the properties of idiomaticity, non-literalness and phraseological peculiarity, from the class of pure idioms.

The homonymity of pure idioms serves not only to separate them from various types of semi-idioms, but also renders the idiom the most potentially opaque of all figurative expressions. The simile makes a very explicit comparison overtly marked by like or as ... as: X is like Y: My luve is like a red red rose (Burns). A metaphor which presents X as being Y is more opaque than a simile:

Make me thy lyre even as the forest is (Shelley)

It would appear Western [the bank robber and gunman] is the blacksheep of the family, and his mother and wife appear to be people of the highest integrity. (Sydney Morning Herald August 1976)

Even more opaque is the simile where one talks about Y with no mention of X at all: O wild West Wind, thou breath of Autumn's being (Shelley). The pure idiom, quite apart from the fact that both origin and original figurative connection is lost to the average user is also the most opaque manifestation of figurative language in terms of its individual constituents. Not only is X hidden in Y with no necessarily explicit indication that it is X that is really being talked about, but Y can also function as a purely straightforward, literal statement, thereby reinforcing the fact that the meaning of an idiom is not a compositional function of its constituent parts:

If we had to do some belt-tightening at our end of the School - OK so it had to be done. (C.J., September 1977)

We went there one evening - I twisted Richie's arm - I said he's your brother-in-law too - but they weren't in. (C.J., August 1977)

Asked to justify this statement, though, she pointed out the regularity with which at certain climactic moments in talks and negotiations my viceroys withdrew, packed their bags and came home for consultation. (P. Scott Raj Quartet, Vol.4)

p. 36

A property such as homonymity which results in a given string (or a single word) being subject to both a literal and a non-literal interpretation enables the investigator to categorise idioms, but is in itself an insufficient reason to justify making such a property a criterion of idiomaticity unless it adds to that idiomaticity. We think it does. The fact that smell a rat, scratch each other's backs, etc. have homonymous literal counterparts complements the fact that their meaning is not the result of the compositional function of their individual syntactic constituents. Both pure idioms and those that we place lower down on the scale of idiomaticity (see Chapter III) are non-literal, but the noncorrelative syntax of an idiom is further emphasised by the fact that it functions contrastively with its homonymous literal counterpart. The identical string functions both correlatively and non-correlatively. The literal/non-literal 'two-faced' idiom can be misleading in certain texts. One of the examples given in a questionnaire submitted to a group of informants (see Appendix for details on Questionnaire 2) contained the sentence

We don't live in Nazi Germany and union officials cannot twist the arms of the unionists behind their backs and force them back to work. (Sydney Morning Herald April 1978)

Behind their backs was identified and interpreted differently on the basis of homonymity. Some informants identified twist the arms of the unionists behind their backs as an extended form of twist somebody's arm; the literal specification of the locative goal of the twisting behind their backs simply served to emphasise the idiomatic signification 'coerce', 'bring pressure on somebody'. Others identified behind one's back asan altogether different idiom signifying 'without one's knowledge'. To these informants the signification of the sentence was 'to bring pressure on the unionists without their being aware of what was being done' (i.e. unethically). We also have an attested instance of a child being misled by the literal sense of break a record in The doctor's son has broken a record. In this case the child was puzzled by the fact that in the context of utterance the doctor's son was praised for an apparently 'naughty act'.

An idiom string which also constitutes a figure such as metaphor or hyperbole has no complementary homonymous literal counterpart. The import of such an expression is clearly figurative. Peculiarities of encoding like hell for leather or to kingdom come do not yield any signification at all on the basis of correlative syntax. A pure idiom like scratch someone's back yields a literal signification on the basis of the correlative syntax, but that signification is never the intended one when the expression is used idiomatically.The expression can, of course, be used non-idiomatically, which makes the mismatch between the covert idiomatic meaning and the overt literal one far greater than in the case of figures that are also idioms (X is a dead duck, be born with a silver spoon in one's mouth, etc.) and set phrases ( hell for leather, to kingdom come, etc.).

p. 37

Since pure idioms have literal counterparts they show no special peculiarities of encoding in themselves, though, like all true idioms, they have lexical integrity. In other words, they are non-anomalous in terms of selectional and strict sub-categorial restrictions. To put it differently, the patterns of collocation and colligation that such idioms manifest in their composition are normal and predictable. This is not the case with hell for leather which departs from predictable sub-categorial (colligational) selections or with a storm in a tea cup which departs from selectional restrictions (collocations). Though from a syntactic point of view storm can co-occur with a locative (a storm at sea) the noun the preposition governs in the phrase is normally not of the class of tea cup. Pure idioms are in themselves perfectly well-formed as they must be in order to have homonymous literal counterparts.

The consequence of the two features presented above is that the mismatch between the apparent and the intended signification of a pure idiom is much greater than is the case with forms of figure. In addition, the fact that both figures and phrases manifesting peculiarities of encoding lack homonymous literal counterparts renders such expressions structurally different from what we are calling 'pure' idioms. Being more covert, a pure idiom could also be more deceptive. Since the essence of idiomaticity is an asymmetry between syntax and sense, our argument is that the presence of a homonymous literal counterpart complements such asymmetry both structurally and contextually.

2.5 Syntactic Properties

An idiom is a syntactic unit which manifests lexical integrity. Such syntactic units manifest, accordingly, that "peculiarity of phraseology approved by usage of the language" noted in the Oxford English Dictionary definition. Makkai (1972) identifies this property of an idiom as that of encoding. He cites avec une certaine vitesse, mit einer gewissen Geschwindigkeit, etc. as being idiomatic in French and German respectively, whereas the corresponding English phrase is he drove at 70 m.p.h. Pure idioms are simultaneously idioms of both encoding and decoding. The phrase in a hot spot as in Fire chief in hot spot could mean either 'a difficult situation' or 'a place with a high temperature'. If one applied Makkai's terminology to the phrase, one would say that hot spot carried two options in decoding. Such options would also be present in ordinary cases of ambiguity: Pam enjoys boring old ladies. But Makkai is right when he says that what makes an idiom an idiom is the property (along with others) of a given meaning manifesting itself in an institutionalised "peculiarity of phraseology". What this means is that there is no intrinsic reason why be in a hot spot should signify 'be in a difficult situation' when other encodings

p. 38

( have a coal in one's mouth, be in Alcatraz goal, etc.) could have served equally well. But the institutionalised idiomatic encoding happens to be be in a hot apot and none other. One can, therefore, speak of a "peculiarity of phraseology" in the sense that only one out of many possible encodings correlates with a given meaning Though potential candidates, none of these other encodings (have a coal in one's mouth, etc.) actually function as institutionalised idioms in English for the meaning 'be in a difficult situation'.

2.5.1 The Lexical Integrity of Idioms

The lexical integrity of idioms is not always as inviolate as some linguists imply (Chafe 1970: 46, 47; Gumpel 1974: 1). As already indicated in Section 2.1, parts of idioms can be deleted. Idioms can also be modified by items not part of the idiom and sometimes even the order of the idiom reversed in the service of Wit. But the language-user can go even further in order to achieve surprising and unusual effects. In such instances an element of the idiom may be changed, without its original import being changed at the same time, by substitution, inversion, or deletion.

Substitution:

Give the green light

The ordeal of negotiation is on, the Opera House in Sydney has been booked and all that is needed now is the Red light from Peking to go ahead (The Australian August 1975)

Cry over spilt milk

"Well it's no use crying over spilt uranium," lisped Dr. Hippo, regarding the ruins of the Senior Common Room. (M. O'Connor Half an Hour After)

Reversal of the order of constituents:

Whereas the morphemic constituents of a single word (e.g. unmeaningfully ) cannot be reversed (* fullymeaningun ), the order of phrasal or sentential idioms could be if the userwished to make a point opposite to the one conveyed by the idiom in question:

For the cat to swallow the canary

For a while in Hobart this week it looked faintly as if the canary had swallowed the cat. (The Australian August 1975)

Deletion of part of the idiom:

Since idioms are institutionalised expressions, part of the original can for this very reason be deleted:

p. 39

A bird in the hand is worth two in a bush

This fellow thought that the Professor would drop him like a hot potato so he preferred a bird in the hand ...(B.T., in the context of employment possibilities, October 1975)

Swallow the bitter pill

Jimmy Connors, tennis king of Australia ... has lost his crowns to John Newcombe, Arthur Ashe and yesterday to Manuel Orantes. A bitter pill for someone who also defeated Australians Rod Laver and John Newcombe in challenge matches at Las Vegas earlier this year. (The Australian October 1975)

With no strings attached

We want wage indexation with no strings. (The Australian August 1975)

Some deletion is possible in certain English idioms, but there are also large numbers, particularly those comprising verb with either preposition or particle, where no deletion is possible: see through someone, bring the house down, turn on, put up, step up, etc. The presence of prepositions and particles is one of the salient criteria used by Cowie and Mackin (1975) in identifying verbal idioms for inclusion in the Oxford Dictionary of Current Idiomatic Usage. The possibility of deletion in idioms is largely a matter of use and would vary widely from language to language.

The point must be made that the three sets of examples given above are variations of established idioms and are dependent on knowledge of the original forms for their interpretation. Many of these variations are nonce forms, but others, for example those cited under deletion, are in the process of becoming contracted, as red herring from the original draw a red herring across the trait.

2.5.2 The Transformational Behaviour of Idioms

A clear indication of our assessment of the place of an idiom's transformational behaviour in defining idiomaticity has been given earlier (see Section 1.3). But we do not deny that such behaviour is relevant to an examination of the nature of idiomaticity. We have so far said little about the way in which the syntax of an idiom can be manipulated. It might be useful to first look at some of the constraints on the transformational manipulation of idioms.

p. 40

... no part of the idiom has retained any literal interpretation, if it ever had any ... That is, we maintain that no part of the idiom actually contributes to the semantic interpretation of the expression, once the idiom has been formed. (Fraser 1970: 33)

Hence any transformation which attempts to extract and give meaning to any one element of the total idiom will be non-operative. Substitution, for example, is normally inadmissible, as in sentence (1), notwithstanding the witty exceptions mentioned in Section 2.5.1 above.

(1) *After the exams, they painted the town green.

Similarly, topicalisation, e.g. clefting, is not generally possible as in sentence (2) and (3).

(2) *It was the beans Alfred spilled.

(3) *It was the cat Bernard let out of the bag.

(4) It is your advice Mary has taken to heart.

But NPs within a discontinuous idiom may be clefted, as in sentence (4). Conjunction reduction is, however, inadmissible (sentences 5 to 8).

(5) Mary took heed of John's warning and later took steps to rectify the situation.

(6) *Mary took heed of John's warning and later steps to rectify the situation.

(7) He laid down the law in private, but he laid down the red carpet for her in public.

(8) *He laid down the law in private but the red carpet for her in public.

For NPs there is no pronominalisation, no taking of a restrictive clause, and no gapping:

(9) *Fred kicked the bucket and a year later his wife kicked it too.

(10) *Their own juice that they stewed in was not pleasant.

p. 41

Idioms composed of adjective + noun do not allow predicative usages, nominalisations, the formation of comparatives or superlatives, or modification, whereas this would be possible in a literal use, as Weinreich (1969: 46-49) has demonstrated with examples (11) to (15).

(11a) a hot sun

(l1b) a hot dog

(12a) The sun is hot.

(12b) *The dog is hot.

(13a) the heat of the sun

(13b) *the heat of the dog

(14a) Today's sun is hotter than yesterday 's.

(14b) *Today's dog is hotter than yesterday 's.

(15a) It's a very hot sun today.

(15b) *It's a very hot dog today.

Binomials are generally irreversible (cf. Malkiel 1959):

(16) *He walked fro and to.

(17) *They fought tongs and hammer.

Idioms including a verb in their make-up usually do not allow many transformations that are open to their literal counterpart:

(18a) Steam was let off as the boiler was seriously overheating, then the train moved slowly out of the station.

(18b) *Steam was let off as the employee got out of his boss's office, but it did no good.

Fraser's hierarchy (outlined on p.12 above) which looks at the transformational possibilities of idiom ranges from LO (completely frozen) through L1 (adjunction), L2 (insertion) L3 (permutation), L4 (extraction), L5 (reconstitution) to L6 (unrestricted). Such a hierarchy is useful for classifying idioms syntactically, but it would have greater value for idiomaticity per se if it were possible to establish a syntax-semantics correlation as a complementary basis for the hierarchy. That is, are transparent idioms freer syntactically and opaque idioms more restricted in the transformations open to them?

p. 42

What we attempt to do now is to present particular instances of idiom which are the result of the application of specific transformations in the interests of a witty economy of expression. In other words, the application of the transformation is linked to the achievement of a specific stylistic effect. The evidence shows that there is considerably more flexibility in idiom syntax than the formulations of the constraints on idiom manipulation mentioned earlier in this section would allow.

There is nothing particularly striking from a stylistic point of view in adjunction pure and simple, for example, the gerundive nominalisation transformation (...s ...ing):

Nobody thought his turning over a new leaf would have any effect at all.

But when adjunction is combined with reconstitution (i.e. when the syntactic function of the idiom changes as a result of its being organised into another constituent structure) the result is more striking:

The Government thinks Mr. Kasim's the kind of man who won't try to make excuses for his son turning coat. (P. Scott Raj Quartet, Vol. 3) (from a turn-coat)

Much more striking is the reconstitution of a noun-compound idiom, a dogsbody into a verb with the subsequent adjunction of was . . . ing:

I was dogsbodying for him in a fairly hectic programme. (Raj Quartet, Vol . 4 )

Or penduluming, where pendulum is not only transformed into a participial verb, but also by a process of radical deletion made to substitute for the full idiom swing of the pendulum:

... it's when what you publish is either revolutionary or reactionary that you get this see-saw swing of the pendulum, this jumping on the bandwagon ... if the advice is going to be penduluming if there's such a verb it's not going to be useful to the teaching profession. (Speaker at IATEFL Conference)

Another interesting example of deletion, though less radical, results from the truncation of pour oil on troubled waters:

... as it was I often had to pour oil especially during the time the Brig was pressing for him to call the troops out. ( Raj Quartet, Vol. 3)

p. 43

Other stylistically interesting examples of the transformations an idiom may undergo are ice-breaker and blind-eye approval, both of which show permutation and reconstitution. In fact, both examples show two applications of the reconstitution transformation: the first combines with permutation to transform the object of a verb ( break the ice, turn a blind eye ) into an attributive structure, while the second converts this attributive structure into a compound:

He thinks a party would be a good ice-breaker. (Raj Quartet, Vol. 3)

Sarah had broken the ice in the maternity ward, by picking up a baby ... ( Raj /Quartet, Vol. 4)

He couldn't help wondering whether the evidence of the bicycle had been planted by Merrick or with Merrick's blind-eye approval. (Raj quartet, Vol. 4)

In such instances the natural opacity of an idiom is also accompanied by extreme syntactic condensation.

Insertion, whereby an adjective or an adverb is inserted into the structure of an idiom, can be quite pedestrian, as in

But a penny had just dropped. ( Raj Quartet, Vol. 4)

but livelier in

The delegates talked serious shop deep into the night. ( The Times Educational Supplement April 1973)

... the most vermilion of red herrings. (The Australian September 1975)

Although an examination of their transformational behaviour offers insights into idioms as syntactic objects, it does not throw much light on idiomaticity per se. Not only are transformational constraints insufficient to distinguish the idiomatic from the non-idiomatic, they also fail to distinguish between semi-literal idioms with affective idiomatic adjectives ( my dear lady, that wretched woman, you poor man, etc.), semi-literal idioms with non-affective idiomatic adjectives ( white lie, blue film, red cent, etc.), completely non-literal idioms (white elephant, blue stocking, turn a blind eye; etc.) and peculiarities of usage ( criminal tow, an utter fool, an outright lie, a real hero, drunken quarrels, miscellaneous articles, etc.). All the adjectives in these items share the feature that they can be used only attributively. Predicative use, nominalisations and inflection for degree do not generally apply to these adjectives. On the basis of transformational deficiency alone, a feature, incidentally, which

p. 44

contributes to the status of these items as institutionalised set phrases, they would be classed as idioms of equal status. But they are not of the same standing or type. Criminal law and drunken quarrels, for example, are peculiarities of English which are also fixed collocations in the language, but their opacity springs from a highly condensed form of encoding which disappears if the phrases are expanded: criminal law ~ law that applies to criminals; drunken quarrels ~ quarrels between drunken people. Idioms such as hot dog, blue stocking or white elephant cannot be expanded to reveal their signification in this way. You poor man, etc. has one non-literal constitute and one literal constitute, while others like hot dog or blue stocking are completely opaque.

In conclusion, idioms are at best only a sub-class of all transformationally deficient structures. All idioms are transformationally deficient, but not all transformationally deficient structures are idioms. Therefore any attempt to set up syntactic deviance as the primary norm to establish idiomaticity fails, since it cannot distinguish between idioms and these other transformationally deficient structures.

2.6 Sociolinguistic Criteria

An idiom is an institutionalised expression, i.e."it is approved by the usage of the language." (QED)

Unless an idiom has currency among the members of a specific speech community or a sub-group of such a community for a reasonable period of time it cannot be regarded as institutionalised. By the institutionalisation of idiom we mean the regular association in a speech community of a given signification with a given syntactic unit (a compound, a phrase or a sentence), such that the resulting expression is interpreted non-literally. In other words, part of the phenomenon of idiomaticity is the institutionalisation of an asymmetry between sense and syntax in the case of compound, phrasal and sentential idioms.

The criterion of institutionalisation is useful for separating items which show a peculiarity of encoding and are in addition non-literal from those that are nonce-idioms or predictable collocations. Nonce-idioms can be coined and used between two or three individuals, either for the moment or for a very limited period of time such as a few days. Hockett (1958: 305) has one such example (already cited on p. 5 above) of a wife teasing her husband who incorrectly identifies as blue the colour of a blouse "of one of those indeterminate blue-green shades for which women have a special fancy name." That's a nice shade of blue is then used idiomatically by the wife to indicate her husband's insensitivity to colour.

p. 45

For days thereafter, Mrs. X teases her husband by pointing to any obviously green object and saying That's a nice shade of blue, isn't it? (Hockett, 1958: 305)

Makkai (1972: 161), commenting on Hockett's example, says that outside the very special situational context in which only these two speakers operate such an 'idiom' will have no idiomatic value at all. But an expression like Don't count your chickens before they are hatched is recognised as an idiom even as a citation item because of its currency in the speech community.

Predictable collocations like the clink of glasses or sharing drinks have marginal idiomatic status in terms of peculiarities of encoding, but apart from being literal expressions their degree of institutionalisation as set phrases is also very limited. In Questionnaire 2 only 13% informants regarded the clink of glasses as an idiom and even fewer (11%) regarded sharing drinks as such. In these instances one could connect both the literal and the variable nature of such expressions (one can also say the clink of coins and sharing a meal ) with their low rating. But there are other items which though sharing the structural properties of items that have indisputable idiomatic status in the community, properties such as non-correlative syntax, homonymity and lexical integrity, have only a very restricted currency in the speech community.

The term 'jargon' as used in languagestudies generally means the specialist terminology associated with a particular trade, profession or activity. A number of idioms have a very restricted currency because they belong to the language of a particular profession or academic discipline. Black box is associated with computer technology, Morton's fork with history and Occam's razor with philosophy. Our mini-survey conducted with 50 informants (see Authors' Note for more details) shows that only 26% had heard the phrase Occam's razor (these were all academics) and only one knew its meaning. There are also large numbers of specialised terms associated with activities such as opal mining, speleology, surfing, etc., some of which have claim to be restricted idioms.

Groups such as Citizens' Band Radio constitute clubs formed to conduct an activity, namely private radio operation. The special terminology of CB Radio is, however, better described as a form of 'slang' rather than as a jargon, since it is associated with a group rather than with an activity and differs from one geographically based group to another. Some of the idioms (associated in this case with the Australian CB group) are expressions like pig panic 'driving at reduced speed due to sighting police', concrete ribbon ' interstate highway', ashphalt pilot 'a truck driver', blood box ' an ambulance', ptomaine palace 'a bad

p. 46

place to eat at' and thumb merchant 'one who signals for a lift'. (These examples were supplied by Shirley Brookes, a member of Australian CB Radio, who notes that one of the reasons for the invention of such terms is exclusiveness and secrecy, a distinctive feature of slang.)

Allusions are another type of idiom also restricted in social currency. The two allusions presented to the informants in our survey were the emperor's new clothes (Hans Andersen's The Emperor's Dew Clothes ) and the mystery of life (Victor Herbert's Ah, sweet mystery of life ). The emperor's new clothes, which in a slightly different form also functions as a quotation idiom, has a much wider currency than the mystery of life. 88% informants had either read or heard the expression used, but only 46% said they would use it idiomatically. The mystery of live has a double status. It is a cliche like the wonder(s) of nature, the marvels of science, etc. and in this sort of use could be made plural: the mysteries of life. It also functions as an allusion, generally ironic, to Victor Herbert's romantic song Ah, sweet mystery of life, in which case it has only the form the mystery of Fife. 80% had heard the expression the mystery/ies of life and accepted it as a cliche. Its currency as an idiomatic allusion was much smaller. 36% informants had heard the Victor Herbert song, but only 6% said they would use the mystery of life and then, ironically, as an allusion. One informant claimed that the expression was widely used in the nineteen-forties as an ironic allusion to the Victor Herbert song in obviously non-romantic contexts like sausages: "I wonder what the butcher puts into these sausages - it's a mystery of life." The mystery of life as an idiomatic allusion has a much more restricted currency today than it once apparently did, but it is still used by those who belong to an older generation (we have three attested uses of the mystery of life as an idiomatic allusion in the Sydney Morning Herald of 17th, 18th and 25th May 1978). "Approved by the usage of the language", which forms part of the Oxford English Dictionary definition of idiom, must therefore be interpreted to mean use that may be restricted by factors such as specialised knowledge, group membership and age.

Inclusion in a dictionary sets the seal on the institutionalisation of an idiom. On the other hand., items included in collections of idioms or even dictionaries could decrease in currency while idioms that are not included may be quite widespread. The degree of currency an idiom has in the community varies, as we have already indicated above. Perhaps it might be more accurate to think of institutionalisation in terms of variable degrees rather than as a fixed phenomenon. Idiomaticity ranges from noninstitionalisation which is the case with a private idiom (It's a nice shade of blue, isn't it? ) through slang terms ( blood box, thumb merchant) and jargon (black box, Occam's razor) to those idioms (smell

p. 47

a rat, cross one's fingers) that find a place in dictionaries. The wider the currency of an idiom, the greater the degree of permissible variation by deletion, permutation or addition of elements. Such items in spite of the distortion of their form are still recognisable and interpretable as particular idioms because a given sense is so firmly associated with a given syntactic construction in the language use of a specific speech community.

2.7 Conclusions on the Nature of Idiom

Given below is a summary of the defining features of a 'pure' idiom. Such an idiom (e.g. Bob smelt a rat and decided he wouldn't go)

(i) is semantically not the result of the compositional function of its constituent parts, i.e. its syntax is non-correlative and therefore an idiom qua idiom is not generatable;

(ii) has a literal homonymous counterpart which complements the fact that the syntax of an idiom is non-correlative, i.e. a pure idiom has both a literal and a non-literal 'face' and as such comprises a 'double exposure';

(iii) is a syntactic unit manifesting relative lexical integrity;

(iv) is institutionalised.

Makkai (1972) and Weinreich (1969) select the property of 'disinformation' underlying (ii) as the key feature of' an idiom, while Gumpel (1974) chooses (i). We regard (i) and (ii) as complementary and of equal importance in the definition of pure idiom. Feature (ii) enables us to distinguish idiom from many more related phenomena (figures of speech) than (i). In fact, if Weinreich had applied (ii) more thoroughly to his chosen examples he would not have classified red herring (metaphor), rain cats and dogs (hyperbole), or by heart (set phrase) as pure idioms. If Gumpel had given (ii) consideration she may not have classified How do you do? as belonging along with kick the bucket, etc., but rather as a set phrase like by heart, to kingdom come, etc., all of which show some properties of idiom but because they lack homonymous literal counterparts in terms of synchronic use occur lower down on the scale of idiomaticity. How are you?, also a greeting though used in a different context, on the other hand, is a pure idiom having real disinformation potential. As such it may elicit a logically correct but semantically erroneous interpretation in the form of a response like Perfectly rotten! I think I'm getting the flu. Howare you?, it may be noted, has a wider range of contextual applicability

p. 48

than How do you do? and may function both as a greeting and a genuine request for information: How are you? I haven't seen you for ages (at an accidental meeting) or How are you? (in the sickroom). Our definition of a pure idiom incorporates all the four features listed above:

A pure idiom is a non-literal set expression whose meaning is not a compositional function of its syntactic constituents but which always has a homonymous literal counterpart.

Idiomaticity in general is adequately covered by Sense 3 of the Oxford English Dictionary definition:

A peculiarity of phraseology approved by the usage of the language and often having a signification other than its grammatical or logical one.

In defining idiom we have stressed three features in particular: a non-correlative syntax resulting in non-literalness, homonymity and institutionalisation. Because it is institutionalised and therefore shared, idiom has a powerful cohesive force. It binds the members of a given speech community and the various sub-groups within it together while at the same time serving to distinguish such groups from others. Perhaps what is unique to idiom is the fact that it is both shared and secret: it binds the members of a group together and is a manifestation of their identity, while at the same time keeping out non-members of the group (either within the community or outside it) by virtue of its hidden meaning. The opacity of idiom is a powerful asset in avoiding having to say something 'to the face', a prime motivation for strategies of indirection in social behaviour. The homonymity of an idiom either in toto or in part provides the language-user with the means for word play, which can be on occasion a form of indirectness. It is by virtue of such features as its non-literalness and the homonymity that so frequently characterises it that idiom overlaps with the figures of a language. But we would regard idiom as being essentially distinctive in itself, though it may manifest features of other types of language phenomena such as metaphor and polysemy.

--------------------------------------------------------------------------------

Fernando, Ch./Flavell, R. 1981. On Idiom: Critical Views And Perspectives. (Exeter Linguistic Studies, 5). Exeter: University of Exeter, pp. 18-48

--------------------------------------------------------------------------------