Lars Brink, Copenhagen, in LexicoNordica 13-2006, Oslo (forthcoming). Phraseological terminology English summary The purpose of this paper is to bring law and order to the phraseological terminology. The best way to summarize the paper, therefore, is to bring my definitions. Expression = Sign (with a meaning) for something which the sender takes responsibility for. (The meaning may be a meaning contribution to a manifestation or a whole manifestation, and the expression may be verbal, non-verbal, established, or unestablished.). ”The fuckfinger”. You saw his. Green letters. Beat it! Established = Already remembered/stored by the sender. Nursery school. Unestablished = Not already remembered/stored by the sender, but improvised. Nursery thoughts. Morpheme = Verbal expression containing only one verbal expression. Clarinet. Word = A group of morphemes referring to each other which can only have their actual semantic relation by standing side by side, although, possibly, separated by other such morphemes. Or a single morpheme not entering such a group. – The common feature, then, is continuity. Greyhound. Now. (This is slightly wrong regarding Danish, cf. amerikansk olie ’castor oil’, but very much wrong regarding English, cf. gold digger in 2 words in a fixed, continuous order. The world will have to wait for my solution.) Polyverbal expression = Expression consisting of two or more words. Grow old. Grow tulips. (Established) phrase = Established polyverbal expression / established utterance. The whole world. Adam and Eve. Go to hell! Thanks! (I am aware that phrase in English, although also established in this meaning, is closely associated with the Chomskyan meaning. Nevertheless I prefer link (cf. Satzglied) for the grammatical concept.) Face meaning = The meaning which an expression has to a person knowing all the rules of a language except the actual meaning variety of that particular expression. ’Kiss my behind!’ Actual meaning = Meaning (as opposed to face meaning). ’What you say is totally irrelevant!’ Predictable meaning = The meaning which follows, according to the rules, from the meaning and syntax of the components of an expression. To steal a potato. A big cigar. Idiom = Phrase with unpredictable meaning. Go bananas. Shut up! Metaphorical idiom = Idiom with an implicit comparison. Steam it up! Collocation = Phrase with a predictable meaning. The whole world. Adam and Eve. Closed collocation = Collocation which excludes a synonymous expression produced according to all other language rules. To put on one’s clothes (**to take on one’s clothes) Open collocation = Collocation which is not closed. A nice fellow. To catch a ball. Transparent = The feature that the actual meaning of an expression – whether it is unpredictable, i.e. non-regular, or not – is easy to connect causally with the face meaning. Don’t let me down! Opaque = Not transparent. Not a Chinaman’s chance. Routine utterance = Utterance phrase the occurrence of which is rather predictable in a certain situation. Thanks! Yes, please! NO SMOKING (signboard) Ritual utterance = Utterance phrase which has to be said in a certain situation. With this I declare the exhbibition open. I, ..., do solemnly swear... Polite phrase = Routine utterance used by courtesy for which the sender, accordingly, does not take full responsibility. Good to see you! Quotation = Utterance which is signalled by the sender to be created by someone else, and for which, therefore, he does not take direct responibility. (To the sender it is a phrase). [As my father used to say:] ”Women and kittens must be pampered.” Familiar quotation = Famous non-anonymous quotation. (Hence phrase). [As the Oracle said:] Know thyself!” Proverb = Quotation put forward by the sender as an old anonymous utterance phrase of universal validity which the sender does not assert (for it is a quotation), but which he normally uses as an argument, thereby presupposing its truth. [It takes a certain amount of courage, of course, but] nothing venture nothing win. Saying = Anonymous quotation phrase which is not a proverb. (Including the parodical proverbs). Time is money. Blondes have more fun. Easy does it. Once bitten twice sly. [About Sylvester Stallone]. Slogan = Saying with a trenchant call. Your Country Needs You! Coca Cola refreshes you best. Allusion = Expression referring to a quotation and forming an integral part of the sender’s speech for which, accordingly, he takes responsibility. This, indeed, was our finest hour. Linguistic cliché = Established linguistic feature (most often a phrase), for which a sender substitutes the normal, straightforward expression (in a situation) in order to achieve a stylistic effect. (The term, of course, is impartial in its scientific use. Observe that the sender’s intention is decisive: What is an effect to someone may be straightforward to another.) Down under. The Big Apple. See you later, alligator! Shiver my timbers. As happy as a sandboy. Construction = Expression with unspecified variables. Sby. gives sth. to sby. Unical = only occurring in one, poly-morphematic expression. Brimstone. Ha’penny. To and fro. [Mother etc.] in law = ’through marriage’. Symptom and signal Let us for once start from the very beginning – with the basic difference between expression and meaning. Phenomena without any form of meaning whatsoever can be referred to as asemantic phenomena. Examples include thunder, sunset, a cough, a hunt. Øhlenschlæger may well have written En sagte Torden dundrer. (A low thunder rumbles.) Hele Norden undrer. (All Scandinavia is surprised.) as nature's signal that Denmark has been blessed with a sacred relic. It is also common to perceive black cats and the baying of dogs as portents of death and destruction. However, in science, we continue to consider thunder as thunder and the baying of dogs as the baying of dogs. This leads us right away to another fundamental difference, which is always ignored by amateurs, i.e. the differentiation between symptom and signal. If my hands shake, I emit a symptom of nervousness, but I do not signal it. Similarly, when Ophelia stands only ten centimetres away while she speaks to Hamlet, it is a symptom that she is in love with him. As a rule, all phenomena are symptoms of something or other. The thunder mentioned is a symptom of electrical tension in the atmosphere, the white garden in the morning is a symptom of the fact that it snowed during the night, and so on. In everyday language these symptoms are called signs of something, as opposed to signs for something, which are signals. Everyday language is far more sophisticated than half-studied dilettantes! The symptoms are completely irrelevant to linguistics, but if, on the other hand, Ophelia tries to communicate to Hamlet her love for him by standing close to him, then we are approaching the linguistic. But as this sign is not established, it does not qualify as proper communication, only an attempt at communication. However, asemantic phenomena can sometimes, as a derived function, become signals, even established signals. An ow! is primarily a symptom, not a signal, but it has also been established as a signal meaning 'Don't do that, it hurts!' And a hesitant sound, an er, which is primarily a symptom of the fact that the person emitting the sound cannot find words, can also function as an established signal: 'It is still my turn to speak. Don't interrupt'. As well as this dual function, there is also the simulated symptom function. Raising both hands up in front of your bowed head serves a basic function as a symptom of despair. However, people are not averse to lying with their hands. It is not beyond the bounds of reason to imagine people adopting this gesture at the funeral of some self-avowedly great leader, even though inside they are actually really quite pleased he is dead. More refined than this distinct deception, one might even refer to it as a lie, is Ophelia's mentioned variant: the symptom is not entirely bogus, but it is sent out simultaneously with an attempt at a signal – for which, incidentally, she does not accept responsibility. In my school days it was quite normal to wear a scout uniform if your troop was meeting after school hours. This was undeniably a symptom, but it also functioned rather effectively as a signal: 'Yes, I am a scout. Please admire my badges!' Similarly, there was once a young woman who undressed in her flat even though I was seated just yards away at another window. Was that simply a symptom of her desire to change clothes, or was she signalling 'You can admire my beautiful body if you want'? Only she knew the answer to that. The scout uniform and the undressing were at any rate unestablished signals. If they had been established, it would not have been possible to simulate a purely symptomatic function. Let us call this non-deceitful transitional form between symptom and signal a true symptom made voluntarily public. When you transmit a voluntary symptom, such as wiping the sweat from your brow in the presence of others, then you are also approving the normal interpretation other people put on the symptom – and the others know it. It is, therefore, not far removed from the signal 'I am hot!' But, as long as you have not signalled this explicitly, you are not taking responsibility for it either. You only accept responsibility for the content of signals. As a consequence, the stripping lady would have been within her rights to act thunderstruck if I had made any approach to her: 'What? Can't I even undress in my own flat?!?' So I just sat back and did nothing. Expression Semantic expressions of any kind are called expressions. These can be verbal expressions, produced using the speech organs, or non-verbal, such as a clenched fist or the opposite – folded, imploring hands. Louis Hjelmslev, who always had to be so original and deep, called them signs, but in colloquial language sign is of course reserved for symptoms (with the preposition of) and non-verbal expressions (with the preposition for), and you must never – never – make terminological exceptions to colloquial language unless there is an explicit point to it. Hjelmslev did not have such an explicit point, and what he called signs ought still to be called verbal expressions. If the expression does not consist of several expressions, it is called a morpheme. In all languages, these morphemes create words. As an experiment, I attempted – albeit incompletely – to define them in Brink 2000. My definition is included above, but we do not need to discuss that here. To do so would require that we hold a separate conference. We just need to be clear that, since phraseology is by definition the study of polyverbal expressions (connections between words), we are sailing with a body in the hold, a word-body, if we have not defined it absolutely. Established expression We are now approaching phraseology. The last two words of this line from a popular Danish children's song, "See the little kitten, isn't it sweet?" undeniably form a polyverbal expression, but it is not established, i.e. remembered and stored, by Danes. Phraseology only concerns itself with established polyverbal expressions. I definitely prefer the term established over lexicalised – both because it is more elegant and because lexicalised is often badly abused, almost becoming an academic buzzword with several competing meanings. Some use it only about, for example, sommerfugl (butterfly), tørklæde (scarf), spille op (play up) and helt i skoven (wide of the mark), as opposed to svømmefugl (aquatic bird), silkeklæde (silk clothes), spille klaver (play piano) and gå i skoven (walk in the forest). In other words, about expressions with a face value or meaning that diverges from the actual meaning; others use it about both. This is extremely unfortunate. What both have in common is that the expressions are remembered and stored, they are not generated by the individual who transmits them. It is this unifying feature that I refer to as established. Unpredictable established expression The peculiar sommerfugl (butterfly) phenomenon is, of course, absolutely fundamental to phraseology, and we will get nowhere without a fixed term for it. I propose therefore – rather straightforwardly – unpredictable meaning, which naturally is not restricted to polyverbal expressions. Both flodhest ("river horse", hippopotamus) and hest (horse), also meaning 'a vaulting horse in gymnastics' and 'a knight in chess', have unpredictable meaning. There is also the major weakness with lexicalised in the latter usage in that it suggests that the expression in question is just as relevant in a lexicon, a dictionary, as the individual words, which of course the most banal of established polyverbal expressions are not. The expression at have for mange penge (to have too much money) is, although established, definitely not as relevant in a dictionary as penge (money). The colloquial term for unpredictable meaning is figurative meaning, but this is not particularly appropriate when no figurative language is involved. We cannot say that sommerfugl ("summer bird", butterfly) has a figurative meaning, partly because the expression has never had any non-figurative meaning, partly because it is not used as a metaphor. Unpredictable meaning, therefore, is a meaning that diverges from the face value/meaning, being the meaning an expression has for the person who is familiar with all the language rules, except the specific variation in meaning of the expression concerned. The face value/meaning of sommerfugl is then 'a bird characterised by something summery'. Similarly, the face value/meaning of slå op (break up) is slå et slag opad (land a blow upwards), while the actual meaning is, of course, bryde et kæresteforhold (break up a relationship). Here, you have to realise that the individual morphemes in their basic meaning do not have any face value/meaning. For hest (horse) in its basic meaning means nothing for the person who does not know its actual meaning, while on the other hand hest (horse) about a piece of gymnastic equipment has the face value/meaning 'four-legged hoofed mammal'. The gymnastic meaning is unpredictable. We notice this clearly when we correct a misunderstanding and say 'No, I am talking about real horses.' We all know the real meaning of heste (horses), stjerner (stars), køer (cows) and negerkys ("negro kisses", a chocolate marshmallow). Unpredictable meaning is unpredictable in relation to the meaning of the components on the top level of analysis, not on a lower level, i.e. at have sommerfugle i maven (to have butterflies in the stomach) has unpredictable meaning, but not because sommerfugl in itself is unpredictable. The colloquial language has a different way of describing the predictable, i.e. the face value/meaning. It is the literal meaning. This ingenious expression, whose own meaning should not be taken literally, is not so useful in science because it never refers to single morphemes. You never talk about the literal meaning of stars (you say real stars). But in linguistics it is, as far as predictability is concerned, immaterial whether we are talking about one or more morphemes, whether we are talking about the face value/meaning of hest (horse/knight in chess) or fra hestens mund (from the horse’s mouth). Both are used metaphorically. Similarly, a loan translation is, as we know, a word whose face value/meaning – and only that – has been translated, e.g. frispark (DK), from free kick (UK). Here, it would seem clumsy to say that the word translates the literal UK meaning because in English the expression never has a face value/meaning. It is in fact unestablished. Not only that, but it is difficult to see precisely what it should be – 'a kick you are not punished for', 'a kick that you do not have to pay for', 'a kick whenever you feel like it’? In my opinion, free kick's face value/meaning is 'a kick, which in some way or other is free'. The person who does not know its actual meaning knows that it is some kind of kick which in some way or other is free; he can work out that the expression is unpredictable but cannot pin down the exact meaning from the face value. It is quite a common situation. A fransk frimærke (French stamp) is a stamp issued by the French state, but a fransk kys (French kiss, which meant the same in my mother's youth as it does today in English) cannot possibly be a kiss given by the French state, an exchange of kisses in France or by Frenchmen, or something similar. Anybody who does not know its actual meaning, but knows all the other rules of language, can assume that it covers some kind of technical sub-category of 'kissing', which in some totally irrelevant way or another is associated with the French, who were in times of yore considered the very epitome of the frivolous. Here too, it is not possible to talk about the literal meaning – because there is none. It is on the other hand perfectly legitimate to talk about the face value/meaning – which is superordinate. We turn now to the question of whether phraseology's subject matter, phrases, covers all established polyverbal expressions, i.e. not just the ones that are absolutely unpredictable, like købe katten i sækken ("buy a cat (UK pig) in a poke') and spille fandango ("play fandango" = UK 'have fun with'), but also the less colourful, the predictable, like Adam og Eva , gå en tur (go for a walk), Se nu dér! (Look at that!), lige akkurat (exactly/just), Jeg ved det ikke (I don't know) and Så er dét a aftale (That's a deal then). These latter expressions are not usually the focus of phraseology, and the reason for this is self-evident – they have no unpredictable meaning. If unpredictability of meaning is to be a precondition for a phrase, then we have to acknowledge that such fixed expressions fall outside the definition. But of course, we must have a term for the whole group of established polyverbal expressions, and the designation I just used does not resemble a term, it is too heavy and resembles what it is – a definition. I will, therefore, join Ken Farø in calling the group (established) phrases. He actually also calls them phrasemes, but I do not advocate that. If an expression ending in -eme does not stand apart and in contrast to a word without -eme, then it is a meaningless parasite, parasitizing on phone–phoneme, graph–grapheme. Indeed, I would even go so far as to say that they directly mislead us into seeing such a contrast. And as long as our terms in general do not end in -eme – in other words, as long as we do not talk of subjectemes, sentencemes, full stopemes, etc. – then there is no good reason to admit them in to the canon of linguistics. 'Yes, but,' you might say, 'isn't phrases a poor choice of term because of the colloquial language's empty phrases?' No, not for that reason, because it is self-evident that we cannot use the term phrase academically in that sense. The colloquial language's empty phrases are spread over at least three linguistic categories: 1) linguistic clichés, which are of course precise, 2) polite phrases, a special category to which we will return in a minute, and 3) particularly frequently used established utterances, such as I love you, which are in no way clichés. In our branch of science there is never any need for the colloquial language's extremely nebulous concept of phrases. And where you are afraid that your audience or readers will confuse the term with the colloquial language's phrase, you can extend it to linguistic phrase, thereby clearly signalling that it is a term. Far worse is the fact that phrase is already used in academic circles as Chomsky's term for part. But I consider this a step backwards. In the European tradition we have the terms part, nominal, subject, predicate and prepositional group. These terms are entirely sufficient. We have no need for Chomsky's phrase, which Chomsky created because American English lacked the specific concept of part (or C. forgot it) which is equivalent to the German Satzglied. A shame for the Americans, certainly, but of course we do not have to adopt their idiosyncrasies. There is, however, a third use of phrase, albeit not very widespread. Those who work in the field of intonation sometimes speak of phrases in the context of sentence-intonation units, in a similar way to the way the term is used in musical theory. This usage is, however, so rare and so clearly different from phraseology's usage that I do not think that its existence justifies the introduction of the misleading and pointless phraseme. We ought not to construct an artificial term using -eme unless that suffix is to mean something specific and significant. And as to the German Phraseologismus (Burger 1998): It may be a lucky term i German, but it has no chances in English and Nordic because of its heaviness. Finally, it should be mentioned that a phrase can sometimes consist of a single word – i.e. when it constitutes a whole utterance, e.g. Thanks! or Out!. These should be established not just as words, but also as utterances. What phrases have in common is not strictly that they consist of multiple words, but that they consist of material which is more complex than merely single words. Idioms Our next term is a term for a phrase with unpredictable meaning, and that can only be idiom. Under this heading we can – just – fit polite phrases: Pleased to meet you, It has been a long time and What are you up to these days?. They are treated in the speech community as if they have full face meaning, but what makes them polite phrases is that we all know that we have to subtract a large chunk of meaning. If a girl I have long fancied says Nice to see you Lars!, you would expect me to be thrilled, but I am not. Most of the linguistic clichés fit too, i.e. the established expressions which involve an opting out of the normal, direct expression expected of the situation in favour of an intended stylistic effect, e.g. Den gamle have (the old garden) about Tivoli Gardens – if you do not know the actual meaning, you have to believe that it is an old garden that has just been discussed. Or take the example of the homeowner who wants to discuss a few things with his neighbour over the hedge, and at some point says Let's move to the next point in the agenda. When this expression is used in the parliament it has its full face meaning and is therefore not an idiom, even though it is extremely conventionalised. The latter I refer to as a routine utterance, like the auctioneer's Any advance?, the dance teacher's And stop! or the sign No Smoking. These are formed according to rules (lexical and syntactic) and are therefore not idioms. This is not the case for the non-routine utterance Passér gaden! (Move along!), which is idiomatic. Idioms overlap correspondingly with actual ritual utterances, i.e. ones used in particular situations, such as the Aaronite blessing (The Lord bless thee, and keep thee...). They can be idiomatic or closed collocations, but they are, naturally, always established phrases. In the loose old lexographic terminology, idiom normally covered neither particle verbs nor proverbs, and certain lexicographers resisted letting idiom cover these. But I think the reason for this was that they did not welcome overlapping terms. Where particle verbs and proverbs are rather precise, idiom was equivocal, and it was generally held that various phrases had to be either/or. However, if unpredictability of meaning is a relevant criterion – and it is of course – then we need to have a term for all established unpredictable word links, and that can only be idioms. We have to face up to the fact that while certain particle verbs like throw up are also idioms, others like throw down are not, just as certain proverbs are idioms, e.g. Never judge a book by its cover, and others are not, e.g. Lykken er ikke gods or guld (Happiness never depends on success). Only then will we have everything in place. For the difference between regular predictability, in terms of meaning, and unpredictability is equally relevant, whether the expression consists of a sentence, a verbal link or something else. And if idioms are allowed to include whole sentences – which everybody does allow – there would be no reason to exclude proverbs. Collocations So if unpredictable phrases are to be idioms, what about predictable ones? There is a certain tradition of calling them collocations. And indeed it is an excellent tradition, as long as it is upheld. According to its Latin roots, collocation means 'place together', which implies a "pure" collocation, but since we are not so adept at Latin these days, there is nothing to stop it being used exclusively for phrases. As Farø has so brilliantly pointed out, there are two critical types of different collocations, which I (cf. Clausén & Lyly) will call open and closed collocations. An open example would be to walk on water. It does not exclude either to go for a walk on the water or to walk on the surface of the water or anything else with the same truth value (allowing, of course, for nuances of style). Here there are no problems for either sender or recipient; when you know the individual words and the general syntax of the language, the construction is self-explanatory without any need for further rules. On the other hand, Adam and Eve cannot be called anything else – not Adam plus Eve nor Adam as well as Eve, and since Adam is the elder, and a head and shoulder taller than Eve, it sounds unnatural to say Eve and Adam. Similarly, consider for example Han måtte trække sin cykel hele vejen ("He had to pull his bike the whole way"). The meaning is entirely predictable, the recipient has no problems, but the sender has, in that he has to use this phrase in Danish even though skubbe sin cykel (UK = push his bike) would be just as natural. In Germany you push a bike, and make a journey (eine Reise tun) – entirely predictable, but in Denmark this expression is old-fashioned. Here, you foretager (carry out) a journey, we laver (make) food, we binder (tie) a knot and begår (commit) a murder. The meanings are entirely predictable but the constructions are not per se. They are closed, indeed so closed that Danish children have had problems with them for the last couple of generations, during which usage of lave (make/do) has increased. Young people laver ("make/do") burglary and laver ("make/do") a mugging – far too many of them, in fact. In doing so, they transform closed collocations to open ones, as they are easier to deal with. Let us just summarise the most important points above in the following table, which I really wish we could all stick to – because we notice terminological Babylon at every single conference. A lot would be gained if we could just agree on this: PH R A S E S Collocations Idioms Metaphorical and non-metaphorical idioms An important dividing line among idioms is the one between the non-metaphorical and the metaphorical. The latter suggests an implicit comparison, and in my opinion it is best that they are named, quite simply, metaphorical idioms. They cannot be called iconographic or visual idioms, since there has to be an implicit comparison. Sort som kul (black as coal) and som et skib uden ror (like a rudderless ship) are as iconographic as you could want, but their explicit comparison means they are not metaphorical. As per the definition, the metaphorical idioms always have transparency (see below) between face value and actual meaning, but they are not unique in that respect. The non-metaphorical holde mund (literally 'hold your mouth' = UK 'shut up') and spytte blod (spit blood) have this attribute too. Since many idioms are metaphorical, it may be beneficial to identify some that are not. I have already mentioned holde mund. Another one is 117 gange (for the 117th time), which is Danish for 'an incredibly large number of times’. Funnily enough, the Swedes say femti-elva (fifty and eleven). Along similar lines, where we say et par stykker (literally 'a couple of pieces' = UK 'a couple'), the Spanish say cuatro (four)! In Spanish, a word with a meaning; in Danish, an idiom. I could also mention all the (more or less) opaque idioms, which given their opaqueness cannot be metaphorical, e.g. [Hun] har det med at [spise sæbe] ([She] has a habit of [verb]), or [Noget] er [nogen] meget om at gøre (It takes many [somebody] to do [something]). Transparency As previously mentioned, idioms have different degrees of transparency. The collocations are all by definition transparent, but the idioms are transparent only when it is easy to see a causal link between their components and the actual meaning (irrespective of whether this relationship is regular or not). At falde på halen (literally 'to fall on your backside over' = UK 'to really admire') is obviously unpredictable in its meaning, but we can naturally connect the components to the face value meaning and see this as a comparison with the actual meaning. In the idioms holde mund (literally 'hold your mouth' = UK 'shut up') or give sig i kast with (throw yourself into) there is no form of comparison; if you think about language you will notice that mund (mouth) can be understood as a natural expression for 'tale' (speak) and kast (throw) as a 'sudden beginning'. There are very few completely opaque phrases (although there are degrees of transparency/opaqueness, cf. Farø 2003). I have found that skyde papegøjen (literally 'shoot the parrot' UK = 'strike it lucky'/'hit the jackpot') and ikke stikke op for bollemælk (originally literally 'would not work at pitching the hay for boiled milk and dumplings' = UK 'wouldn't work for peanuts', later changed meaning slightly to something along the lines of 'not a quitter') – are both on their way out of the language for that very reason. Integrated and non-integrated idioms Certain idioms can be confused with individual words in special meanings, namely, the non-integrated as opposed to the integrated. In an integrated idiom the whole expression carries meaning; no single word has predictable actual meaning. In the non-integrated idiom certain words have predictable actual meaning. Metaphorical idioms are integrated: glide (slide), ud (out), ad (of), tangenten (the tangent) have no independent predictable meaning in the well-known idiom (to go off on a tangent); they combine to create the actual meaning. In the non-metaphorical idiom slå ud (hit out = UK 'knock out'), slå (hit) keeps its normal meaning; only ud (out) is unpredictable. But it is a phrase because the link is established, has at least two words and contains an unpredictable meaning/syntax. Similarly, slå ud (as in break out/develop a rash) is non-integrated, but here it is the other way around – slå is the unpredictable word, while ud (out) has its normal directional sense. On the other hand, slå op (break up (of people)) is integrated: neither word contributes anything independently and predictably to the meaning 'end a relationship'. Phrase and construction It is important to stress the difference between phrase and construction. A phrase is precisely a quite definite phrase, an explicit expression, if I might put it like that. A construction is a form that has not been filled in. At give ngn. ngt. (to give somebody something) is an established construction for everybody who speaks Danish, but it is not a phrase. At give julegaver (to give Xmas presents), on the other hand, is a phrase – it is an established polyverbal expression with a predictable content as per the component rules, in other words a collocation. Similarly, at give bagerbørn brød (literally 'to give baking dough to the baker's kids' = UK 'carry coals to Newcastle') is an established polyverbal expression with unpredictable content – in other words, an idiom. Phrases can be included in a construction, as in [Nogen/Subjekt] giver bagerbørn brød ([Somebody/Subject] carries coals to Newcastle), but the unfilled place in the construction does not belong to the phrase. All my attempts at definition should be understood as stipulative and appropriate clarifications of colloquial language or academic tradition. As such, they cannot all be razor-sharp. Language has, unlike atomic particles, all sorts of teasing little transitional cases. Therefore, it can be a virtue to leave space for a certain degree of borderline uncertainty. Allow me to give an example of a transitional case. I once wrote 'Min lille datter kan Møllehave udenad' ('My little daughter knows Møllehave inside out'). It was a deliberate, conscious ambiguity that recalls two phrases: 1) At kunne [noget] udenad (UK = to know [something] off by heart) and 2) At kunne [nogen] udenad (UK = to know [somebody] inside out). The former has to be called a collocation, as the components retain their normal meaning. According to the rules you put in a subject instead of [something], in this case the subject means 'Møllehave's texts' (cf. Jeg læser Bodelsen (I read Bodelsen)). There is nothing unpredictable here, even though the word udenad (off by heart) is unpredictable in itself. The second phrase, at kunne [nogen] udenad (to know [somebody] inside out), is a metaphorical version of the first one – you compare a person, e.g. Møllehave, with a text containing what that person says, and then claim that somebody knows this text off by heart, in other words is able to recite it ad nauseum. Metaphorical? Well yes, but it lies just on the borderline, because mentioning the man instead of his oral hobbyhorses is more reminiscent of pars pro toto than of a normal comparison. Another difficult case is the phrase ikke sandt? (UK = isn't it? – used at the end of a sentence). There is no doubt that it is a phrase. But is it an idiom? That would require unpredictability of meaning, and I doubt anybody was ever uncertain about the meaning the first time they heard it. The words sand (literally = true) and ikke (literally = not) make sure of that. It think it is too pedantic to say that the meaning does not correspond 100% with the regular sub-meanings of the components. The actual meaning is rather 'I presume you agree with me so far?' and a regular, unestablished expression such as Er dét ikke sandt? (Is that not true?) would also mean the same thing. But the syntax is the key. What appears to be a simple ellipse is in reality a unical ellipse. We can't just add ikke morsomt? (not funny?), ikke løgn? (not a lie?), ikke trist? (not sad?) or any other random combination of [adjective/noun] and not to the end of a Danish sentence. Therefore, we are dealing with an idiom – solely on the basis of the syntax. And what is critical here is not that people are so linguistically talented that they can in practice predict the meaning, but rather that the meaning of the phrase is not formed according to the rules. Idiom stress Finally, I would like to draw attention, perhaps superfluously, to the fact that Danish marks out a lot of idioms phonetically. If we take the metaphorical (and therefore transparent) idiom at tage tyren ved hornene (to take the bull by the horns), there is a loss of stress in the verb, but the only reason for that is because the expression is an a idiom! If it wasn't, it would be **'tage 'tyren ved 'hornene (‘take ‘the bull by ‘the horns). It can't be called that, not even if it is about a real bull, because – as a certain man wrote in 1976 – as a rule of thumb it is the case that idioms, which have arisen as idioms, cannot be used in predictable meaning. To do so would associate to the idiom and reject a corresponding unestablished expression. My son's name is David, but I will never seriously be able to say 'hvor David købte øllet' ('where David bought the beer' ) If the idiom on the other hand was transmitted from an already existing established expression, e.g. løbe af stablen (literally 'run from the stocks' = UK 'launch s.th'.), then we could still use, naturally, the expression in its old sense. We also have idiomatic stress marking in 'ikke ligge på den 'lade side (literally 'don't lie on the barn side' = UK 'don't take it easy'). According to the rules, the stress should be 'ikke 'ligge på den lade 'side', but you can't say that in Danish. Other examples in Danish include stå with hatten i hånden (stand with your hat in your hands) and føle [nogen] på tænderne (feel something in your teeth). I have also come across the idioms stå på hovedet (stand on your head) and ligge på ryggen (lie on your back). Are they really unpredictable idioms? Yes. Because that head and that back can only be the subject's. In our house we sometimes play a game called 'Døde Heste' ('Dead Horses'). For example, I will say to my daughter 'Lie on the sofa', to which she can reply 'Yes but David is lying there!' I then say 'So lie on the back [of the sofa]!' Naturally, I don't say 'Lie on your back!' Die Sprache ist ein Wunder, das sich nie auswundern läβt. (Language is a never-ending wonder.) Bibliography Brink, Lars: Ordforrådets omfang og vækst. P. 109-175 i "Børnenes sprog – sprogene omkring børn". Hans Reitzel. Kbh. 1979 Brink, Lars: Kongen af Danmark s bolsjer. P. 35-42 i Nydanske Studier 26-27. Dansklærerforeningen . Kbh. 2000. Burger, Harald: Phraseologie. Eine Einführung am Beispiel des Deutschen. Neuburg 1998. Jämförelsekonstruktioner i Svensk konstruktionsordbok… . P. 75-85 i Clausén, Ulla: Nordiske Studier i Leksikografi 6. Nordisk forening for leksikografi. Torshavn 2003. Farø, Ken: Ordsprog i nutidsdansk: funktioner og problemer. P. 38-64 i Danske Studier. C.A. Farø, Ken: Reitzel 2003. Omkring det grønne bord: Dansk-tysk idiomatik og sprogforandring. P. 51-75 i Studier i nordisk 2002-2003. Selskab for nordisk filologi. Kbh. 2004. Hansen, Erik: Den sproglige kliché. P. 5-23 i Danske Studier. Akademisk Forlag 1979. Hansen, Erik & Jørn Lund: Sæt tryk på. Syntaktisk tryk i dansk. Kbh. 1983