Summary Phraseology is seen as one of the key elements and arguably the most productive part of any language. %e paper is focused on collocations and separates them from other phraseological units, such as idioms or compounds. Highlighting the difference between a monolingual and a bilingual (i.e. contrastive) approach to collocation, the article presents two distinct classes of collocations: grammatical and lexical. %e latter, treated contrastively, represent the focal point of the paper, since they are an unending source of translation errors to both students of translation and professional translators. %e author introduces a methodology of systematic classification of lexical collocations applied on the Slovene-English language pair and based on structural (lexical congruence) and semantic (translational predictability) criteria. Key words: contrastive lexicology, contrastive phraseology, collocation, lexical collocation Povzetek Članek predstavlja frazeologijo kot najbolj tvoren in zelo pomemben del jezika in se znotraj le-te posebej posveča kolokacijam. Slednje loči od ostalih frazeoloških enot, kot so idiomi ali zloženke. Izpostavi razliko med enojezično in dvojezično (kontrastivno) obravnavo kolokacij, znotraj kolokacij pa razloček med slovničnimi in leksikalnimi kolokacijami. Osredini se na obravnavo leksikalnih kolokacij, ki predstavljajo tako študentom prevajalstva, kot tudi poklicnim prevajalcem neusahljiv vir prevajalskih problemov. Na podlagi empiričnih raziskav in analize študentskih napak predstavi metodologijo sistematičnega razvrščanja kolokacij na podlagi stukturnih in semantičnih kriterijev oz. strukturne kongruence in prevodne predvidljivosti. Ključne besede: kontrastivna leksikologija, kontrastivna frazeologija, kolokacija, leksikalna kolokacija DOI: 10.4312/elope.7.2.57-73 Lexicology has traditionally been seen as the study of words, as the following entries in some of the prominent monolingual English dictionaries clearly show: “the science of the derivation and signification of words: a branch of linguistics that treats of the signification and application of words” (Merriam-Webster 2003); “the study of the meaning and uses of words” (Longman Dictonary of English Language and Culture, Summers 2005). %ese views, however, are to be understood in the broader sense of the term word: indeed, it would be more appropriate to broaden the definition with, say, “and anything functioning as words in texts”. %is definition brings us closer to the concept of lexeme (or lexical item) as it is perceived in current lexicology: “(...) lexemes may be single words (crane, bank), parts of words (auto-, -logy), groups of words (the compound blackbird and idiom kick the bucket), or shortened forms (flu for influenza, UK for United Kingdom” (McArthur 1992, 600). Phraseology is by definition interested in lexemes that are made up of “groups of words”, which is the most vibrant and productive part of any lexicon. Old and new words as well as word combinations compete daily to gain a gradually wider and wider acceptance among speakers of a language; some are rejected and some survive. Only relatively recent studies have shown the overall amount of phraseology in texts. Although early estimates of the phraseological share in texts were as high as eighty percent (Sinclair 2000), further corpus-based studies brought the number down and suggest that about fifty-five percent of texts are accounted for by phraseology (Erman and Warren 2000). %e pervasiveness of what Sinclair (1991, 110) has called the idiom principle (as opposed to the open-choice principle) in language is a fact which is in stark contrast to the traditional views of only few decades ago (i.e. those preceding computational linguistics) that saw lexicon as a pool of mostly single words and a restricted number of idioms, which are ruled by grammar alone. %e role and impact of phraseology in foreign language teaching and learning have received due recognition in the last two decades. With the advent of corpus linguistics, the overwhelming presence of phraseological patterns in texts suddenly became visible and as a direct result phraseological units (other than idioms that have been taught even before language corpora became available) are now considered valuable building blocks of a learner’s L 2 vocabulary. Nation (2006, 449) points out that L 2 word combinations which are not “parallel” to the combinations in L 1 deserve special attention and have to be learned, which complies with the underlying methodology of approaching L 2 vocabulary contrastively as applied in this paper. %ere are many (often overlapping) views of what the scope of phraseology is; some are quite broad in setting the ground to be covered (e.g. Cowie 2001), and others take a narrower approach (e.g. Moon 1998). Where the above authors differ from one another is principally in the delimitations they set between what they perceive as the basic types of units of phraseology and in the terminology they use to name the units – these range from “gambits, prefabricated routines/patterns, prefabs, chunks, holophrases, semi-fixed patterns, formulaic phrases, routine formulae, (semi-) fossilized phrases, lexicalized sentence stems to set/fixed/frozen expressions” (Gabrovšek 2005, 166). In the last three or four decades, the universe of phraseology was seen as revolving around three kinds of word combinations: collocations, compounds and idioms, and in case of English with the additional class of phrasal verbs. While English idioms and phrasal verbs have received abundant lexicographical treatment (cf. Jarvie 2009; Flavell and Flavell 2006; Parkinson and Francis 2006; Sinclair 2002; Rundell 2005), this was not the case with collocations, which were hardly a lexicographer’s favourite, with only a handful of publications (cf. Benson et al. 1986; Lea 2002). On the other hand, regardless of what denomination is applied to these multi-word lexical units (hereafter simply referred to as word combinations), what all of them appear to have in common are the two criteria according to which they are distinguished from one another and classified, viz. their structure (syntactic fixedness) and their meaning (semantic [non-]opacity or compositionality). %e present paper will focus on word combinations that are semantically transparent, i.e. collocations, for the following reasons mostly: first, their ubiquity, and second, their importance in foreign language learning at advanced/proficient levels. Compared to idioms (both partial and full), which are exceedingly rare in the current text production (Moon 1998, 85), collocations are found in practically all utterances, but go largely unnoticed by L 2 addressees precisely due to their semantic transparency (Handl 2008, 48). However, this feature cuts both ways – while in most cases collocations pose few problems in L 2 decoding and are readily understood by learners, they are virtually invisible to the learner’s eye (unlike idioms, where the addressee’s stream of thought comes to an abrupt end if decoded literally), and therefore so difficult to learn in a traditional way. Learners may go as far as to develop an overcautious reaction to idioms, so even when the translation involves only the most predictable translation equivalents and can be translated quite literally (e.g. počivati na lovorikah – to rest on one’s laurels, opeči si prste – to burn one’s fingers), they are blocked: simply because they have identified a word combination as an idiom, they make no attempt at translating it word-by-word. %ey know better than that and because idioms are tricky, so they were told over and over again, the key to their problem is not to be found in straightforward translation, but rather in specialized dictionaries. What is needed, it has been argued (Jurko 1997, 79), is for students at intermediate level (secondary school students, typically after having received four to six years of L 2 education) to be made aware of collocational restrictions that pose formidable problems in encoding, so that their L 2 mental lexicon can accommodate more complex and web-like relationships between lexical items (Aitchison 1987, 72). In the following chapters the notions of structure and meaning, will prove very important and provide the basis of comparison for the contrastive analysis of Slovene and English lexical collocations. Collocations need no special introduction to anyone even remotely interested in languages, let alone students of translation and practising translators. Due to their pervasiveness they are a vital factor of idiomaticity of any learner’s spoken or written text production. In our analysis we will rely on the basic terminology of the field as introduced by Hausmann (1989) and Benson (1986). Hausmann introduced the terms base and collocator, respectively, and according to him they are the constituent parts of every collocation. He sees the base as the semantic nucleus (or the meaning bearing element, the entity spoken about) of the collocation, and the collocator as a modifier of the base, i.e. as the element describing the base in more detail. For instance, in the noun-verb collocation a mosquito bites it is a mosquito that we are talking about and not entities that can bite in general, which makes it the base of the collocation. %e verb bites is the collocator then, providing more information on what it is that mosquitoes do 1 . Benson introduced two categories of collocations, viz. grammatical and lexical ones. Grammatical collocations consist of a dominant word (noun, adjective, verb) and a preposition or grammatical structure (cf. Benson 1986, ix-xxiv for a detailed classification), while lexical collocations contain two (or more) lexical words. %ese mostly binary combinations typically “consist of nouns, adjectives, verbs and adverbs” (ibid., xxv) and are the subject of this analysis. %ere is one important methodological issue that has to be resolved before continuing: the monolingual approach and the contrastive enterprise differ significantly in their respective methodologies and goals. For the purposes of this paper suffice it to say that the monolingual concept of collocation is often based solely on corpus-derived statistics (e.g. Kjellmer 1994), while the contrastive view of collocations primarily takes into account their meaning and structure interlingually, as well as relevant corpus-derived data on frequency of co-occurrence. %e main objection to the monolingual approach to collocations is the frequently observed practice of including a sizable amount of free combinations (Bahns 1996), resulting from the fact that it is impossible to draw a clear dividing line between the two kinds of word combinations on the basis of frequency of co-occurrence alone. As a case in point, consider the following list of collocates of the English adjective marked as compiled by Kjellmer (1994, 1060): MARKED marked arteriosclerosis marked by marked contrast marked difference marked with be marked had marked is marked is marked by was marked was marked by were marked has been marked %e results of the monolingual approach as shown above undoubtedly yield important insight into the intricacies of word combinability of the English language. For instance, readers learn that marked can precede names of medical conditions, and can be used with the prepositions by and with, etc. However, the fact that it can be used with the verb be in various forms (i.e. be, is, was, were, had been) seems of little use to, say, a translator into English who needs to find the right translation of the Slovene word combination znatno izboljšanje – in English marked improvement. As will be shown in more detail below, even combined with a contrastive filter, frequency of co-occurrence will not always yield useful results: the Slovene collocation izgubiti potrpljenje has a matching English expression to lose patience, and both collocations have a relatively high frequency of occurrence in their respective corpora (FidaPlus and Corpus of Contemporary American, aka. COCA). However, the pair of collocations is of little contrastive pedagogical value, because the L 1 -into-L 2 traslation equivalent is completely predictable, or in Nation’s terms the L 2 collocation is “parallel” to that inL 1 . Note, on the other hand, that frequency of occurrence can be the decisive factor in discriminating synonymous collocations, i.e. when two or more seemingly identical word combinations exist in L 2 , as in e.g. the Slovene bratovska ljubezen, for the translation of which there are two competing roughly synonymous adjectives, viz. brotherly and fraternal. %e respective frequencies of brotherly love and fraternal love in COCA are 160 and 6, which clearly makes the translator’s or EFL teacher’s choice much easier. An alternative to dealing with the notion of collocation within the framework of one language is a contrastively conceived approach that would take into account the most prominent features and patterns of vocabularies of both languages of a given language pair, which is also the key principle behind the analysis presented here. T wo kinds of sources of collocations were used in the survey; they served as the basis of the underlying empirical work: a) dictionaries: the following available monolingual dictionaries of English and Slovene collocations, respectively, were random-sampled: #e BBI Dictionary of English Word Combinations (Benson et al. 1986), the Oxford Dictionary of Collocations for Students of English (Lea 2002) and Vezljivostni slovar slovenskih glagolov (Žele 2008). %e randomly chosen collocations were then translated into Slovene and English, respectively. b) bilingual lists of Slovene and English collocations resulting from BA theses of students of the Faculty of Arts of the University of Ljubljana (Bradač 1996; Štupnikar 2006; Krašovec 2007), as well as my own privately compiled glossary of contrastively interesting lexical items that triggered many encoding translation errors in Slovene BA and MA students of translation. All word combinations stem from authentic contemporary Slovene and English texts and were checked in reputable corpora of Slovene and English (FidaPlus and COCA, respectively). As noted above, all parameters presented below will be discussed in terms of either structure or meaning. %ese parameters are involved in translation of lexical collocations from Slovene into English and are partly based on empirical research found in Jurko (1997). %ere is one important caveat to the methodology introduced in the following paragraphs: frequently the listed Slovene-English pair of collocations is not the only possible translation equivalent, nor is it argued that it represents the best or most suitable translation equivalent. It is, however, a legitimate and idiomatic translation chosen for various reasons or motives that do not fall in the scope of our topic. %at said, we will also leave aside the (often problematic) fact that each collocation is studied and listed in isolation and not in context, where it naturally appears. In most cases it is precisely the context that provides the clues which make a translation equivalent either more or less suitable. On the basis of above sources a list of Slovene and English collocations and two-word compounds alongside their translation equivalents was compiled. No principled distinction was made between lexical collocations and compounds, because to the encoding translator compounds behave very much like collocations and frequently feature difficulties that are identical to those observed in collocations. For simplicity’s sake both collocations and compounds will be referred to as collocations hereafter. %us a pool of about 500 collocation pairs was created and analyzed according to the parameters specified below. While the analysis would undoubtedly benefit from a higher number of collocations, the number of 500 was seen as substantial enough to yield reliable insight into contrastively relevant patterns of Slovene and English collocations. %e terminology adhered to in this analysis was introduced by Marton (1977, quoted in Bahns 1993, 58), by drawing a distinction between what he called cases of lexical congruence (where the structure of a word combination in one language is identical to that in the other), and cases of lexical non-congruence (all other cases where any kind of structural change is present). Not surprisingly, lexical non-congruence between Slovene and English turned out to have many faces, which significantly differ from one another in terms of morphology and syntax, as well as their complexity. Five identifiable kinds of lexical (non-)congruence were recorded in our pool and will be described in more detail below: - word-class shifts - lexical expansion / reduction - change of grammatical number - word order - no structural change: lexical congruence. %is is the most frequently observed type of lexical non-congruence and occurs when either the base or the collocator (or both) belong to one word-class in the Slovene collocation and to another in its English equivalent. Given the systematic differences between the two languages in this respect, this was expected. In highly inflected Slovene the demarcation lines between individual lexical word- classes are very clear. Each word-class has its morphological markers (mostly suffixes) that make it easily distinguishable from the others. %is situation is markedly different from that in English, mostly owing to the fact that English is only poorly inflected. Although this may not be a fitting place to discuss a topic that clearly belongs to word-formation, its implications for the Slovene translator and/or student of translation are far from negligible. Indeed, even advanced students of translation are frequently at a loss when the most elegant or idiomatic translation into English involves the so-called conversion. %is feature allows many English words to belong to several word classes without adding any affixes (e.g. text can be used as a noun or a verb). T o put it simply, the trouble is that students are looking in the wrong place: to translate a noun they look for an equivalent among nouns, when they should also be checking items they have stored as verbs, but can be (and are!) as well used as nouns. A case in point is the unassuming Slovene občutek za žogo, which is in students’ translations almost invariably rendered as the strange sounding *feeling for the ball/*ball feeling, yet is best translated into English as ball feel. What are the reasons for this oversight? One of the strongest factors at play here is arguably the purely verbal character of feel in the mental lexicon of Slovene learners, which is very likely adopted in the early stages of EFL acquisition and is based on a one-to-one mapping of L 2 lexical items onto the established L 1 lexical framework. %us the English lexeme feel is mapped as a verb only in the Slovene learner’s lexicon, which makes subsequent expansion of the lexeme’s word class affiliation very difficult. While Slovene translation students hardly ever fail in decoding nominal instances of feel in texts, it is in encoding tasks that they frequently make errors of the feeling-feel type, arguably due to the strict fencing of word classes in Slovene that they have mapped onto English. Back to our topic, in our corpus word-class shifts in lexical collocations were found to consist of several sub-types, distinguishable from one another in terms of the specific type of shift. On that basis the following two-fold division has been proposed: - word-class shift I: Slovene adjective rendered as an attributively used noun in English; - word-class shift II: all other cases of word-class shift. Let us now take a closer look at the above sub-types, each featuring typical examples. %e most frequently observed case of lexical non-congruence was dubbed word-class shift I: it occurs when in the Slovene word combination an adjective is used, and in the English one a noun in attributive position (also called noun premodification). In Slovene adjective-plus-noun collocations the adjective is mostly used as the collocator, and the noun as the base. %e same relationship is maintained in English: the attributively used noun performs the function of the collocator, while the head-noun is the base. Slovene English collocator base collocator base nadurno delo overtime work oglasna deska notice board kazenski proctor penalty area cestna varnost road safety poulični izgredi street riots avtobusna proga bus line destilacijska buča distillation flask plinski gorilnik gas burner Note that in encoding the required task of translating a Slovene adjective into an English noun is mastered by most learners almost automatically. However, caution is advisable in cases where English makes a systematic distinction between a premodifying noun and a derived adjective, e.g. wool – woolly, silk – silky, milk – milky, fruit – fruity, etc. In these pairs the premodifying noun has the meaning “made of the [noun]”, while the adjective has the meaning “resembling or having the quality of the [noun]”. Both English noun and adjective, however, are rendered by a single Slovene adjective, as in the following examples: volnena kapa – wool cap; volnen bas – woolly bass svilena kravata – silk tie; svilen glas – silky voice mlečni izdelki – milk products; mlečna tekočina – milky liquid sadni sok – fruit juice; sadni okus (o vinu) – fruity taste All other cases of word-class shift were classified as word-class shift II. In this mixed-bag category of word-class shifts the most frequently recorded change is that of the collocator from a Slovene adjective to a prepositional of-phrase in English. Other cases that make up the word-class shift II category include shifts from Slovene adverb or adjective to English verb (often in the s.c. –ing form), Slovene noun to and English of-phrase and others. Please note that in the tables below there are cases where the collocator does not precede the base: in all such occurrences the collocation will be given in full in a separate column. Slovene English collocator base collocation collocator base izpitna komisija board of examiners of examiners board ptica ujeda bird of prey of prey bird tovorna žival beast of burden of burden beast krivo pričati to commit perjury akcijski radij striking distance žvečilni gumi chewing gum nalepiti tapete [na zid] Ø to paper [the walls] temeljito razmisliti [o čem] to give [sth] careful consideration to give careful consideration %e basic criterion for this type of lexical non-congruence is the number of lexical items that make up the equivalent collocations in Slovene and English, respectively. Grammatical items (e.g. English articles, the particle to) were left out of the analysis, as they were seen as parts of either the base or the collocator. Since lexical expansion and reduction are reversible relationships, depending on the observer’s point of view, we have decided to look at them from the perspective of the Slovene translator into English. In terms of justifying this admittedly arbitrary choice, it is after all in encoding that most collocation-related lexical problems occur, so taking Slovene as the source language and English as the target seems reasonable. With this in mind, lexical expansion will be dubbed the situation in which the English collocation uses more lexical items than its Slovene counterpart, and vice versa, all collocations with fewer items on the English side will be termed lexical reduction. In most cases of lexical expansion we are dealing with an English hyphenated compound that is in Slovene rendered as an adjective. Another frequent feature of expansion are all collocations that contain English phrasal verbs, which are in Slovene expressed by one-word verbs. %e latter are often very poorly represented in English texts produced by Slovene translators and/or learners (Drstvenšek 1998, 107), and should accordingly be given more stress in advanced English vocabulary training. Apparently translators frequently rely on the assumption that a one-word entity in L 1 will be rendered as such in L 2 as well, leading to awkwardness or lower idiomaticity of their translations. Slovene English collocator base collocator base honorarna zaposlitev part-time employment samica [o zaporu] solitary confinement neprebojni jopič bullet-proof jacket zaspati (=predolgo spati) late/in sleep biti zaprepaden to be taken aback naročiti se [na kaj] take out a subscription [to sth] maršal field marshal nasesti ladja to run aground ship Cases of lexical reduction predominantly consist of two syntactic patterns: one, a reflexive verb in Slovene becomes in the English translation a simple verb, and two, a Slovene prepositional phrase is rendered as a noun-plus-noun combination in English. Another source of reduction is an adjective-plus-noun combination in Slovene, which is translated as a solid-spelled compound in English. Slovene English collocator base collocator base odpraviti se na ekspedicija to launch an expedition ustvariti si družino to start a family morska bolezen seasickness bojna ladja warship preiti v ilegala to go underground za piškote model cookie cutter mahati s krili beat the wings pritisniti na sprožilec to pull the trigger All pairs of Slovene and English collocations involving a change of grammatical number fall into this category. A change that hardly ever occurs naturally in translation is the change of grammatical number. Instances of grammatical translation errors involving the pluralization of a Slovene noun in singular (witness the so-called “century” type of error, e.g. Slovene v 13. in 14. stoletju and English in the 13th and 14th *century) are well known and dealt with in the realm of contrastive grammar analysis, however, besides those there are also lexical contrastive issues that have so far received less coverage. Slovene English collocator base collocator base obratovalni čas business hours dvovišinska bradlja asymmetric bars jutranja telovadba morning exercises kurja polt goose bumps brez izgubljenega niza zmagati in straight sets to win zbrano delo collected works spreletavati srh to give somebody the creeps Only four occurrences of the plural-to-singular lexical incongruence were recorded in our sample of collocations. %is particular form appears to be much less frequent than the preceding one; however, this discrepancy might be caused by the relatively limited size of our sample. %e situation calls for extensive testing on bigger samples when larger parallel corpora become available. Slovene English collocator base collocator base razviti jadra to set sail morski sadeži seafood medeni tedni honeymoon študentski nemiri student unrest %is particular class of lexical incongruence was recorded only rarely and occurs when the word order of the collocator and the base in Slovene is the reverse of that in English. Most cases of word order change were recorded with Slovene word combinations consisting of two juxtaposed nouns in the nominative case, which is a rather closed set of items and thus not a very productive pattern in contemporary Slovene. Slovene English base collocator collocator base angel varuh guardian angel ptica pevka song bird država članica member state država kandidatka candidate state država gostiteljica host country pes čuvaj guard dog brat dvojček twin brother Slovene collocations that when translated into English retain their word-class membership, number of constituent parts as well as their grammatical number are called lexically congruent collocations. %is type of collocation turned out to account for approximately a quarter of all collocations in our sample, which is good news for the Slovene translator, of course, as far as structure is concerned. Slovene English collocator base collocator base neplačani dopust unpaid leave glavna atrakcija main attraction varovalna barva protective color nastaviti uro Set a clock neprevozna cesta impassable road tratiti čas to waste time ponarediti denar to forge money iti na dieto to go on a diet Our treatment of semantic parameters will focus on two features encountered in translating Slovene texts into English: one is translational unpredictability, and the other is divergent translation equivalents. While there are fewer parameters to examine, they are in most cases quite difficult for translators to cope with and generally represent a greater challenge than structural changes analyzed above. What is translational unpredictability? %e issue might be clear to all translators on an intuitive level, because we, translators into non-mother tongue, are all guilty of committing errors when it comes to translating collocations. Errors occur when translators rely on the hypothesis of transferability (Bahns 1993, 61) simply because they are familiar with the meaning and syntactic patterns of both constituents of the word combination, but are oblivious of their co-selection restraints. A quick example is the Slovene divji kostanj which the unsuspecting translator is likely to translate as *wild chestnut, just because s/he finds both words “easy”. So, even as a pre-scientific concept, translational unpredictability is hard to define, as it is clear that it is highly subjective and will mean various things to various people. On the other hand, when it comes to scientific research methods, subjectivity should be ruled out as a matter of principle. T o provide a firmer ground for comparison of our analysis, we have turned to the largest Slovene-English dictionary available (Grad and Leeming 1996) and to the most recent one (Zaranšek 2006). Although admittedly quite different in terms of size and scope, the two dictionaries represent a combination of the traditional approach to Slovene bilingual dictionaries witnessed in the former, and the modern corpus-based approach featured in the latter. So, in order for a collocation to be labelled unpredictable, its base or collocator or both had to be either omitted or treated inadequately in both dictionaries. Slovene English collocator base collocator base srečno naključje lucky break izvesti akrobacijo to do a stunt risalni blok sketch pad naravna bogastva natural resources beg možganov brain drain destilacijska buča distillation flask adventni čas Advent season morski ježek sea urchin Slovene English collocator base collocator base preklicati embargo to lift an embargo polniti akumulator to charge a battery debela Berta big Bertha enojajčna dvojčka identical twins dvojajčna/-i dvojčka/-ici fraternal twins dolgotrajna bolezen lingering disease zasenčene luči (avtomobilske) dipped headlights klubska mizica coffee table Slovene English collocator base collocator base inbus vijak hex bolt dvovišinska bradlja asymmetric bars zaključena družba private party gasilni aparat fire extinguisher mrtvi kot blind spot voditi evidenco to keep a record violinski ključ treble clef častni krog victory lap Readers may notice differences in the above examples in terms of varying degrees of difficulty involved in the Slovene-English translation. Also, the consequences of approaching the translation unsuspectingly, i.e. treating the Slovene word combination as a free combination and as a result producing a nonexistent English collocation, will vary from case to case. %e impact of such poorly translated collocations seems closely related to which element of the collocation is translationally unpredictable, e.g. - the collocator in the pair klubska mizica – coffee table; - the base in the pair morski ježek – sea urchin; - both the base and the collocator in the pair mrtvi kot – blind spot. T o assess these differences, which are reflected in their respective difficulty involved in the process of correct translation, let us take a brief look at incorrect translations that are not only possible, but likely to occur in Slovene learners’ texts: - *club table: the addressee will know that the subject of the conversation is some sort of table, and although s/he might not be sure what kind of table, it seems unreasonable to assume that this specific lack of information caused by the wrong collocator would cause communication problems; - *sea hedgehog: in this case the addressee is very likely to require additional information, as they would be misled by the wrong base of the collocation. Since the base is the semantic nucleus of the collocation, an error in its translation gives the addresee a wrong point of reference, puts her/him in a wrong picture in a manner of speaking. Although some context is provided by the collocator sea, this can hardly be considered helpful to decode the collocation as a whole, as it is semantically too broad. - *dead angle: if confronted with a word combination like that the addressee is bound to be clueless as to the intended meaning of the syntagm, so the speaker’s failure to provide neither the collocator nor the base of the correct collocation is practically certain to cause a communication breakdown. In this type of semantic change either a Slovene polysemous base or collocator triggers several translation equivalents in English collocations. As in 4.1, depending on which part of the Slovene collocation is polysemous, i.e. the base or the collocator, all collocations were classified accordingly. %is type of relationship is actually a special sub-type of translation unpredictability (4.1), and a particularly difficult one for the Slovene encoding translator. Cases, where a single Slovene base is translated into a variety of English bases, each depending on the specific collocator, were classified as divergent base translation equivalents. Slovene English collocator base collocator base jetniška celica prison cell plinska celica gas chamber telefonska celica telephone booth teroristična celica terrorist cell razredni boj class struggle neizprosni boj relentless fight smrtni boj deadly combat jedrski napad nuclear attack spolni napad sexual assault epileptični napad epileptic seizure napad kašlja of cough fit %is is another sub-type of translation unpredictability, where a polysemous Slovene collocator is translated into a host of English collocators, most of which will be unpredictable from the Slovene translator’s point of view. Slovene English collocator base collocator base strasten ljubimec passionate lover strasten kadilec heavy smoker strasten bralec avid reader umetna noga artificial leg umetno cvetje fake foliage umetno zobovje false teeth umetno oko glass eye umetno gnojilo ø fertilizer umetno usnje imitation leather umetni diamant synthetic diamond %e above classification of Slovene and English interlingual relationships in the field of lexical collocations is conceived contrastively in that it takes into account lexical patterns in the two languages and focuses on translation procedures that are required to produce idiomatic English translations of Slovene lexical collocations. T wo general kinds of parameters that are at play in translation were identified and subsequently contrasted: those related to structure and those having to do with meaning. On that basis the required changes involved in translation were studied. %e analysis yielded several classes of collocation pairs, each with its distinctive type of change of either structure or translational predictability. However, the classes should by no means be seen as an exhaustive list of all possible procedures that make up the translation of lexical collocations. It should rather be seen as a practical simplification, hopefully not over- simplified, of a highly complex realm of lexical collocations as seen from a Slovene translator’s point of view. %e picture painted by the examples featured in the above tables is certainly nicer than the linguistic reality, but the choice seemed justified by its classroom value. Above all, one has to be aware that while only relatively clear-cut examples are included in the tables, very often the difficulty in translating a given collocation is multiplied by a combination of two or more types of the neat classification presented above. Another important caveat is that although all data has been corpus-proven, future parallel-corpora-driven research is bound to bring more fresh and reliable data on the subject, which will undoubtedly have a profound impact on the typology and classification of interlingual relations in Slovene and English lexical collocations.