Summary Everybody is talking about collocational analyses these days… Despite recent advances in the monolingual sector, the bilingual environment has not yet come under close scrutiny. It is especially the adjective-noun combinations that have become the focus of attention when it comes to contrastive phraseological studies. Adjectives in particular are subject to semantic tailoring and it is important to bear in mind that (predictable) interlingual lexical one-to-one occurrence, such as the English starless night and the German sternlose Nacht, is a mere exception rather than the rule in the bilingual adjective-noun state of affairs. Factors that have to be considered are (non-) compositionality in contrastive multiword units, like barefaced lie – faustdicke Lüge (‘a lie as thick as a man’s fist’), and metaphorical extensions, like haushoher Favorit – hot favourite (*house-high favourite) as well as structural differences in the two languages in question, like (at) short notice – kurzfristig. Key words: phraseology, contrastive linguistics, translation studies, lexicography Povzetek Kljub najnovejšemu napredku na enojezičnem področju, pa dvojezično okolje še ni bilo podvrženo natančni analizi. Ravno pridevniško-samostalniške zveze so v centru pozornosti večine kontrastivnih frazeoloških študij. Še posebej pridevniki so podvrženi semantičnemu krojenju, zato se je treba zavedati, da so (pričakovani) medjezikovni leksikalni pojavi, kot sta angleški starless night in nemški sternlose Nacht, bolj naključje kot običaj v dvojezičnih pridevniško-samostalniških pojavih. Dejavniki, ki jih je treba upoštevati, so (ne-)sestavljivost kontrastivnih večbesednih zvez, na primer barefaced lie – faustdicke Lűge, metaforične razširitve, na primer haushoher Favorit – hot favourite, kakor tudi strukturne razlike med dvema jezikoma, na primer (at) short notice – kurzfristig. Ključne besede: frazeologija, kontrastivno jezikoslovje, prevodoslovje, leksikografija DOI: 10.4312/elope.2.1-2.57-64 At first sight we might believe that English and German adjective-noun combinations will not cause any major troubles, when regarded interlingually, in a bilingual framework, since both of the languages in question have adjectives in attributive position preceding a complex noun phrase where the adjective modifies the noun. But, why is bread, in English, when it is no longer fresh, stale and not dry or old – as it would be in German? And why is that we can use sour, or sauer, as it were, for milk which is no longer fresh and has a bad taste in both English and German? Well, this paper will certainly not give you any answers to the ‘why’, but will try to make you understand that there is more to it than simply translating languages word-by-word. We know that words do not co-occur freely in a language, and this problem of acceptable, yet only sometimes expected, word combinability shows even more strongly when we look at it from the contrastive angle. But first things first. +e study of words and the company they keep dates back as far as the 19 th century; however, it was only in the 1980s that, with John Sinclair, research in lexical co-occurrence relations got a fresh input (Gabrovšek 2003, 186). With the advent of the computer, increased storage capacities and data processing speed have made it possible to deal with large amounts of data. +us large text files of naturally occurring language, so-called corpora, can be analysed for the unrandomness of lexical patterning producing typical or natural-sounding speech (cf. Hanks 1998, Sinclair 1984, Gabrovšek 2003). Sinclair (1991, 109–10) asserts that the organisation of language can best be described by distinguishing between two underlying principles. +e open-choice principle, on the one hand, says that language consists of sequences of words which follow grammar rules of the respective language. +e idiom principle, on the other hand, accounts for the fact that “the choice of one word affects the choice of others in its vicinity” (Sinclair 1991, 173) on a level other than grammar, namely on the level of phraseology. According to the idiom principle, “users have available a large number of semi-preconstructed phrases that represent single choices” (Gabrovšek 1998, 114). Unfortunately, there is, in the field of phraseological study, still no common descriptive approach, and going into detail concerning the different concepts behind those semi-constructed phrases, or multi-word units, such as collocations, idioms, clichés, formulae, proverbs, etc., would go well beyond the scope of this paper (cf. Cowie ed.1998 and Moon 1998). +e classification of broad phrasal categories suggested by Gabrovšek (1998, 121–22) seems to fit best the needs for a later contrastive analysis. His basic classes are compounds, idioms, collocations and free combinations. Free combinations, in line with Sinclair’s open-choice principle, are sequences of words that follow rules of grammar and syntax, and whose elements allow for free substitution. +ey are the least cohesive type of word combinations. We can easily form free combinations such as beautiful flower and beautiful painting or yellow flower and red flower. Collocations are loosely fixed, typically and frequently co-occurring word combinations. +ey are semantically transparent, i.e. their meaning can be derived from the denotative meaning of their elements. Taking up the adjective from the examples above, beautiful typically collocates with female human beings while it is usually not accepted with reference to male human beings – where we would use handsome. As Bolinger and Sears (1981, 55) put it, there is “no reason, as far as dictionary definitions of words are concerned. We don’t say it because we don’t say it”. Idioms are relatively fixed and semantically opaque word combinations, i.e. their overall meaning is not the sum of the denotative meanings of their constituent parts, e.g. in give somebody the red carpet we are not actually giving someone a carpet which is red, but rather we ‘give special treatment to an important visitor’. Compounds are completely frozen combinations of two (or sometimes even more) words. +ey are referred to technically as endocentric when the semantic head is inside the combination, i.e. the meaning of the whole combination characterises the determinatum, while they are exocentric when the semantic head lies outside the combination and they denote an unknown variable – the connection of which can at best be traced diachronically. Examples of endocentric compounds are wet suit (= a special type of suit), prime minister (= a special type of minister) and magic carpet (= a special type of carpet); by contrast, combinations like lame duck, white elephant, wet blanket and grey matter are called exocentric compounds as they do not denote ‘a duck which is lame’ or ‘a blanket which is wet’ etc., but ‘a person or business which needs help’ (= lame duck), ‘something that is completely useless although it might have cost a lot of money’ (= white elephant), ‘a person who spoils the joy of others’ (= wet blanket) and ‘intelligence or brain’ (= grey matter). Note that there are “bound to be overlaps between the categories” (Moon 1997: 48) and that phraseology has to be seen as “clines of lexical relations” (Gabrovšek 2003, 185) rather than fixed categories. Contrasting two languages is like dealing with two sides of a coin; on the one side, you decode (= comprehend) a foreign written or spoken text, while on the other side, you encode (= produce) such a text. And it is particularly in the encoding process, whether e.g. from English into German or vice versa, that problems can arise. Collocations are accepted passwords to a native-like fluency in a language (Cop 1991, 2776) and it is, therefore, crucial to take this knowledge of typically co-occurring words into account rather than substituting individual words when rendering collocations in another language, thus encoding (Baker 1992, 53). Consider, for example, the German haushoher Favorit. +e literal translation into English would be house-high favourite – a combination which might well be understood by the English native speaker, but that could likewise cause laughter since, in English, the person who is most likely to win in a competition is either the odds-on or the red-hot favourite. It is interesting to note, though, that both haushoch and red-hot are adjectival compounds in which the base adjective is intensified: haushoch is not only high but ‘as high as a house’ and red-hot is not only hot, but, by analogy to red-glowing iron, ‘so hot that it glows red’ – but going into detail concerning semantic images would carry us too far. +e collocational error in this example is due to L1 interference when the speaker, in the encoding process, directly translates an L1 collocation word-by-word into the L2 (Gabrovšek 1998, 127). Likewise, when encoding from English into German, a dry voice, when literally translated, would be trockene Stimme, ‘a voice which is not wet’, while the correct German collocation would be kalte Stimme, literally ‘cold voice’. On the other hand, we can also find English and German collocations which are direct translation equivalents of each other, like starless night and sternlose Nacht or heavy scent and schwerer Duft. Direct translation equivalents account for the fact that, although the collocational ranges of L1 and L2 items might differ, there are certain overlaps to be found which result in the easiest- to-encode or easiest-to-decode category of interlingual collocations: predictable and lexically congruent collocations (cf. Heliel 1990, 131). Predictability, in the light of a contrastive analysis, means that two collocations are semantically predictable, as are wet nappy and nasse Windel. Moreover, these bilingual collocations show lexical congruence since they exhibit the same structure (cf. Gabrovšek 1998, 129). Examples for unpredictable, yet lexically congruent collocations in English and German include barefaced lie and faustdicke Lüge (‘a lie as thick as a fist’) or heavy smoker and starker Raucher (‘strong smoker’). Note that the unpredictable element in adjective-noun collocations is almost exclusively the adjective. +e third contrastive category of collocations is predictable but lexically non-congruent ones, like the English collocation main entrance which is rendered into German by Haupteingang. Both parts of the German compound are predictable (main = haupt and entrance = Eingang), yet Haupteingang is a single-word item, an endocentric compound, while its English equivalent is a collocation. Unpredictable and lexically non-congruent collocations are a class which needs special attention in the contrastive light. According to Bahns (1997, 108), there is no such thing as non-equivalence since these collocations would fall under the category of idioms. Is it as easy as that? +e English collocation fitted carpet has an unpredictable and lexically non-congruent counterpart in German, T eppichboden (‘carpet floor’); however, there is no trace whatsoever of idiomaticity. Likewise, there is the German collocation blinder Passagier, which has to be translated into English as stowaway, a compound which is totally unpredictable and lexically non-congruent. Taking a closer look at blinder Passagier, we shall certainly find that the combination refers to a type of passenger – though semantically transparent, does blind in this combination really mean that the passenger is blind? In German, someone is a blinder Passagier, when they ‘hide and travel without a ticket, especially on a ship or in a plane’. +e meaning of blind in this combination is so restricted that it can only occur in the combination with Passagier and is thus figuratively used for hidden – a meaning which it would never have when regarded in isolation 1 . Additionally, the English stowaway is an exocentric compound; it is also idiomatic since its meaning cannot be derived from the denotative meaning of its constituent parts. Note that contrastive studies of English and German compounding could fill libraries and they will not be dealt with in detail in this paper. It seems as if some collocations, i.e. collocational meanings, are more restricted than others: while heavy in heavy rain is semantically tailored to the noun (Allerton 1984), i.e. “trimmed” from all the denotative meanings heavy might have (Gabrovšek, private communication), blind in blinder Passagier is figuratively used, having this particular meaning only in combination with Passagier. +is finding suggests the need for a further sub-classification of the term collocation for the contrastive analysis of English and German adjective-noun collocations. Cowie et al. (1983) propose a distinction between open collocations and restricted collocations, where the latter are characterised by the “literal sense” of one of their constituents, while the other one is “figurative” – a criterion that applies perfectly to blinder Passagier. Restricted collocations are also called semi-idioms. +is lends further support to the notion of clines of lexical relations (cf. 3.) where one cannot say that a given combination “is either a collocation or an idiom” but rather “more idiomatic or less idiomatic”. +e obvious thing to do, in contrastive terms, would be to say that restricted collocations, like blinder Passagier, are indeed idiomatic, and non-equivalence, i.e. unpredictability and lexical non-congruence, is proof of this since it is primarily in idioms that the structural and semantic images are different in the bilingual framework because they are language-specific (Gabrovšek 1998, 127), e.g. once in a blue moon (from blue moon – which is the second full moon within a month, a phenomenon which is rather seldom to be observed) vs. alle Jubeljahre einmal (a Jubeljahr is a special year in the Catholic Church which recurs every 25 years and in which sinners are shown indulgence). +en again, take naked eye, for example, where naked meaning ‘without optical devices’ is restricted to this combination. According to the hypothesis stated above, there can only be non-equivalence in contrastive terms. However, the German translation, freies Auge, is lexically congruent with its English equivalent; it is, moreover, partially semantically predictable since eye is a direct translation equivalent of Auge. Only the adjectives are unpredictable since frei and naked is not a translation pair which would spring readily to mind, as would free for frei and nackt for naked. +e same holds true for (in) broad daylight; it is idiomatic in the sense that broad as an intensifier is only used in this combination (Gabrovšek, private communication). +e German equivalent is (am) hellichten T age. We could not possibly equate broad with hellichten (which is, above all, actually only used in that combination, i.e. it does not exist otherwise) – while it is also an intensifier, nor can we translate daylight with T ag (‘day’); still, structural congruence is there. Unfortunately, dry wine is also restricted in the sense that dry, as in ‘not sweet’, is idiomatic. But it is 100% predictable and lexically congruent in German: trockener Wein. Summing up, adjective-noun collocations, in contrastive terms, are far from easy to handle; there are really no easy-to-find-rules or classifications concerning translational patterns. While predictable and lexically congruent collocations would seem to be the least problematic ones, one could not say that, generally speaking, predictable and lexically non-congruent collocations are easier to deal with than are unpredictable and lexically congruent collocations, restricted or not – for English and German at least, since many collocational problems can only be identified for a specific language pair (Gabrovšek 1998, 112). +e German collocation, akademische Viertelstunde, for example, can be directly translated into French as quart d’heure académique. +ere is, however, no English equivalent of it and one can only convey this meaning by an explanatory gloss such as ‘the standard 15-minute delay in beginning university lectures’ (Gabrovšek 2003, 189). +is lexical gap can be referred to as collocational mismatch (Gabrovšek, private communication). Another such example is the more idiomatic English collocation dry country – ‘a country where alcohol is prohibited’. For both French and German there is a mismatch in this case, so that we have to render the collocation as an explanatory post-modified noun phrase, i.e. une region où l’alcool est prohibé and ein Land, in dem Alkohol verboten ist. Finally, false friendship is also a recurrent problem in dealing with collocations in the contrastive framework. False friendship is due to L1 interference and occurs when there is a pseudo-direct equivalent in the L2. Directly translating the German trockenes Land into English would result in dry country. While this collocation exists in English, as we have just seen, it does not denote an ‘arid region’ – as the German collocation does. False friendship and other collocational errors can be explained by the hypothesis of transferability (Bahns 1993, 61): Quite in contrast to what they do as regards idiomatic phrases, speakers generally seem to rely on a one-to-one equivalence of collocations in the interlingual light and it is precisely this hypothesis of transferability which underscores the need for bilingual collocation dictionaries which will help the user master restrictions on collocability of lexical items (Gabrovšek 1998, 120) as well as show them typical combinations – something which non-native speakers cannot, by any means, be aware of and which can definitely not be explained by any rule, as this paper tried to show.