Summary When systemic differences between languages preclude the possibility of a truly “faithful” translation, the translator may choose to preserve either the form of the original, that is its syntactic structure, or its function, which includes not only the meaning but also stylistic effects such as the degree of markedness. Faithfulness to the original form may result in a different degree of markedness in translation, or even in a reinterpretation of the structure, with a concomitant change in meaning. On the other hand, the preservation of the original function by introducing structural devices congenial to the target language may obscure the original cohesive devices, including figures. Since the balance between form and function is particularly delicate in poetry, this paper examines the choices between them in two English translations of “Belo”, a free verse poem by the contemporary Slovenian poet Dane Zajc. $e central issue discussed is the ordering of constituents, which is dominated by different principles in the two languages. Since both form and function play a vital role in poetry, such an analysis cannot yield a formula for each type of translation problem; nevertheless, it can locate the points of divergence in the source and target languages, as well as use the translators’ choices to draw conclusions about the nuances of acceptability in the target language. Povzetek Kadar prevajalec zaradi medjezikovnih razlik ne more ohraniti »zvestobe« izvirniku, se lahko odloči, da bo poustvaril ali obliko izvirnika, torej njegovo skladenjsko zgradbo, ali pa njegovo funkcijo. Slednja ne zaobsega le pomena, marveč tudi slogovne učinke, kakršna je zaznamovanost. Zvestoba izvirni obliki lahko v prevodu ustvari drugačno zaznamovanost; možno je celo, da isto zgradbo v drugem jeziku tolmačimo povsem drugače, tako da pride tudi do spremembe v pomenu. Če pa skuša prevajalec ohraniti izvirno funkcijo in zato uporabi zgradbo, ki je bolj v duhu ciljnega jezika, utegne s tem zabrisati izvirna kohezivna sredstva s figurami vred. Ravnovesje med obliko in funkcijo je zelo občutljivo v poeziji, zato v pričujočem članku preučujem odločitve za prvo ali drugo v dveh angleških prevodih pesmi “Belo”, besedila v prostem verzu, katerega avtor je sodobni slovenski pesnik Dane Zajc. Osredotočam se na zaporedje členov, ki ga v angleščini uravnavajo druga načela kakor v slovenščini. Ker sta za poezijo temeljnega pomena tako oblika kot funkcija, takšna analiza ne more ponuditi stereotipnih prevajalskih rešitev; lahko pa ugotovi, na katerih področjih se jezika razhajata, in na osnovi prevajalskih odločitev sklepa na odtenke sprejemljivosti, ki jih imajo v ciljnem jeziku posamezne zgradbe. DOI: 10.4312/elope.1.1-2.173-194 Systemic differences between languages present a challenge to translators, since they rule out the possibility of a truly “faithful” translation. $e available choices may be narrowed down to two: the preservation (1) of the syntactic structure of the original, or (2) of its function, with the latter subsuming not only the meaning but also stylistic effects such as the degree of markedness. Faithfulness to the original syntactic structure may result in a different degree of markedness in translation, or even in a different interpretation of the clause structure, which entails a change in meaning as well. On the other hand, the preservation of the original function by employing structural devices congenial to the target language may obscure the original cohesive devices. $ese, however, play a vital role in establishing text types such as poetry. $is paper will examine the two options as adopted in two English translations of a contemporary Slovenian poem in free verse, concentrating on the issue of word order, which is dominated by different principles in the two languages. Word order in Slovenian is guided by two principles: the fixed word order principle (i.e. the “grammatical principle”, according to Firbas 1992, 118) and the free word order principle. $e former determines the order of phrase constituents, the position of clitics, and the sequence of clitics if several occur together. $e latter, on the other hand, obtains at the level of clause elements and is based on functional sentence perspective (FSP), namely the way in which the speaker or writer organises a message according to what he or she wishes to communicate. Generally speaking, this organisation is determined by two factors, which often (but not always) coincide: the given (old, retrievable) or new status of the information conveyed, and the dichotomy between what the message is about and what is actually said about its topic. $e less informative part of the message is often called theme and the most informative rheme. According to the “FSP linearity principle” (ibid., 118), elements are arranged in a linear progression from the most thematic to the most rhematic ones. In Slovenian, then, marked word order is one which deviates from the grammatical principle and/or the FSP linearity principle (i.e. if a rhematic element precedes a thematic one). In English, on the other hand, the influence of the FSP linearity principle on word order is much smaller, the dominating principle being the grammatical one. According to Quirk et al. (1992, 50), the unmarked sequence of clause elements in English is: (adverbial adjunct) 1 – subject – (adverbial adjunct) – predicator – (object) – (object) – (subject / object complement) – (one or more adjuncts). As evident from this scale, adjuncts are relatively mobile, but even many adjunct types are subject to various restrictions. While adjuncts of 1 place or time serving as scene-setting devices often occur in initial position, others, such as manner or direction adjuncts, or those required by the predicator as complements, normally occur in final position (Davis, 1996). Word order in English is therefore marked if it flouts the grammatical principle. $e use of poetry as material for this analysis is based on the fact that, while certain types of text primarily require the preservation of the original function in translation, the choice between form and function is much more delicate in works of literature. A poem is established as such precisely through the specific textual devices of grammatical and lexical cohesion employed. $e intratextual relationships formed by these devices are particularly important in free verse, which lacks other “poetic” markers, such as metre and (usually) rhyme. $e poem selected is therefore an instance of free verse, “Belo” by the contemporary Slovenian poet Dane Zajc. Both translations discussed are the works of English native speakers, Michael Biggins and Erica Johnson Debeljak. Of course the paper cannot presume to establish the motives of either the poet or the translators for selecting a particular structure. What it can and attempts to do objectively, however, is examine the texts in their own right and offer explications which are substantiated by textual evidence. T ext 1: Belo zunaj barvastega sveta kot zid stojiš pred našimi očmi zunaj sluha čuta in okusa si ko te srečamo se zmedejo koraki oči zaplešejo v očesnih ležajih so misli pest suhe travice in vemo: pot je uročena pot je napačna zavožena je pot za hip v prostoru osvetljenem belo nevidno vidimo si kmalu kmalu zapustimo svet zvokov barve in okusa kmalu kaplja med sestrami kapljami brez zvoka brez občutka kapljastega bomo padli na svoje bele razlaščene kosti kmalu T ext 2: Whiteness (translated by Michael Biggins) beyond the multi-colored world you stand like a wall before our eyes beyond hearing, touch and taste you are when we meet you our steps falter our eyes reel in their sockets thoughts become a bunch of dry grass we know: the path is determined the path is mistaken one second and in a floodlit expanse white unseen we see you are soon very soon we’ll leave the world of sounds, color and taste soon a drop among our sister drops soundless senseless of our dropness we’ll fall onto our white repudiated bones soon T ext 3: White (translated by Erica Johnson Debeljak) outside painted light like a wall stands before our eyes outside hearing feeling and taste you are when we meet our steps become confused our eyes rotate in their sockets our thoughts are a fist full of dry grass and we know: our path is bewitched our path is mistaken is lost our path for an instant the place is illuminated white invisible yet seen you are soon soon we will leave the world of sounds colors and tastes soon only raindrops among other raindrops without sound without feeling like drops we will fall on our dispossessed bones soon 3.1 A structural analysis of the first stanza reveals two items of contrastive interest. $e first is a structure in T ext 1 from which both translations diverge, providing two different restructuring solutions. For the most part, however, the original form is maintained throughout the stanza. $e other structure meriting discussion is thus one which is preserved in both translations but results in a highly marked effect and potentially in a different interpretation in English. $e structure which is changed in T exts 2 and 3 is the one corresponding to kot zid in line 2, i.e. an adverbial adjunct of similarity, or possibly manner, since the two types of adjuncts merge when they “cooccur with verbs used dynamically” (Quirk et al. 1992, 1075). $e other structure is the placement of a complement, i.e. an obligatory adjunct of place (zunaj sluha čuta in okusa, line 3), before a copular verb (si, line 4). $e structure of the first stanza can be represented for the three texts as follows (the passages to be discussed are in bold type): T ext 1 adverbial adjunct of place (zunaj barvastega sveta) – – adverbial adjunct of similarity or manner (kot zid) – predicator with an implicit subject (stojiš) – adverbial adjunct of place (pred našimi očmi) – – adverbial adjunct of place (zunaj sluha čuta in okusa) – – predicator with an implicit subject (si) T ext 2 adverbial adjunct of place (beyond the multi-colored world) – – subject (you) – predicator (stand) – adverbial adjunct of similarity or manner (like a wall) – adverbial adjunct of place (before our eyes) – – adverbial adjunct of place (beyond hearing, touch and taste) – – subject (you) – predicator (are) T ext 3 adverbial adjunct of place (outside painted light) – – adverbial adjunct of similarity realised by a clause: subordinator (like) subject (a wall) predicator (stands) adverbial adjunct of place (before our eyes) – – adverbial adjunct of place (outside hearing feeling and taste) – – subject (you) – predicator (are) All three versions begin with a sentence-initial adjunct of place, which represents unmarked word order both in Slovenian and English. Optional place and time adverbials, i.e. those not required as complements by the predicator, often express a setting for the whole clause or sentence and thus convey background information, which is of thematic rather than rhematic status. In consequence, they are likely to appear at the beginning of a Slovenian clause, and even the English word order in this case proves susceptible to functional sentence perspective. Huddleston et al., for example, observe that (optional) adjuncts may occur in initial position unhindered by the pragmatic constraints imposed on complements (2002, 1372), and Davis likewise lists optional place and time adverbials among those which may appear in initial position in English (1996, 62). $us the three texts correspond in regard to both the structure and the level of (un)markedness of the introductory line. $is adjunct of place, however, is followed in Text 1 by an adjunct of similarity or manner, which is succeeded in its turn by the clause predicator. It is here that the translations diverge: while Text 2 changes the order of elements, placing the predicator with its subject in front of the adjunct, Text 3 opts for a complete rearrangement, which results in a partial change of meaning as well. From a structural point of view, Text 3 introduces an adverbial clause of similarity, subsuming under it the original main clause predicator and (at least) one adverbial adjunct of place; as a consequence, the syntactic structure of the whole stanza is reduced to one main clause consisting of three adverbial adjuncts – of place, similarity, and again of place (the ambiguous status of this last element will be discussed later) – and, finally, the subject and predicator. $e reason for the change in T ext 2 may be attributed to three principles in the English language. Firstly, adjuncts of similarity, as well as those of manner, are described in Quirk et al. as “predication” (as opposed to “sentence”) adjuncts (1992, 1074–5, 1110), which, except “for rhetorical purposes”, typically occur in final position in English (ibid., 1074). $is position is also described as the most appropriate for English manner adjuncts by Davis, who refers to them as predicate (as opposed to sentence) modifiers (1996, 62). $us T ext 2 replaces a word order that would appear marked in English with a neutral one, one which would have the same effect as the Slovenian text. Of course the above tendency does not preclude the possibility of manner adjuncts appearing in initial position in English. In fact, the findings of the corpus analysis in Gómez-González (2001) suggest that these adjuncts may rank highly among instances of preposing 2 . $e most typical preposed element in the corpus is found to be the circumstantial adjunct; of its subtypes, the most common is the adjunct of accompaniment, e.g. of condition, while the second place 2 is occupied by the adjunct of place or time (ibid., 233). $e frequency of manner adjuncts is not given; however, the fact that they are mentioned together with the three most frequent types (i.e. accompaniment, temporal, and spatial adjuncts) as a subtype of the circumstantial adjunct, and, moreover, in a context where the latter is described as “the prototypical function fulfilled by fronted #emes” (ibid., 231–2), suggests that they are frequently to be found among preposed, or fronted, elements. Nevertheless, both translators have opted for other solutions, which may have been influenced by two other principles. Unlike Slovenian, English is susceptible to the “principle of weight distribution”, which discourages the accumulation of too many elements (Davis 1996, 66), as well as to the “end-weight principle”, according to which longer elements are reserved for the end of the clause (Gómez-González 2001, 32–3). $e redistribution of two initial adjuncts which results in one of them occurring in the final position (which is congenial to its semantic category as well, and more typical of adjuncts generally) operates in accordance with it. Here, however, a question arises. Is the position of the similarity, or manner, adjunct really unmarked in Slovenian? According to Firbas, manner adjuncts 3 usually carry new information and thus have high information value (1992, 53); since the order of Slovenian clause elements reflects functional sentence perspective, such adjuncts should appear in final (rheme) position for the word order to be stylistically neutral. $is objection may be countered by two arguments. According to Jug-Kranjec, it is a common device in works of literature to employ word order unwarranted by the context, but even in such cases the initial element represents the theme, although the latter is selected arbitrarily (1981–82, 39). $is word order is “occasional” rather than “usual”, but still in the unmarked sphere because it follows the progression from theme to rheme (ibid.). (A comparable device in English is that of equipping items expressing new information at the beginning of a narrative with the definite article, thus presenting them as familiar and creating an in medias res effect, cf. Firbas 1992, 40.) $erefore the word order in the Slovenian text can be explained as unmarked, although unmotivated by the context and selected for the sake of a particular effect. Moreover, the relative information value of the adjunct itself may be weakened by the presence of the place adjunct pred našimi očmi, “before our eyes”. As noted by Firbas, an adverbial which is not an obligatory complement of the verb but stands in a looser relationship with it has a potentially lower information value; in such cases, it is often its position in the sentence that marks its rhematicity or thematicity (1992, 50–1). While the adjunct of similarity or manner is optional, the adjunct of place has a strong link with the predicator stati, “stand”, in Slovenian and English. Accordingly the relative information value of the former may be weakened enough for it to gain the status, and thus position, of theme. It may be concluded that the word order of the Slovenian version is unmarked, and that this stylistic feature is preserved in T ext 2 by means of reordering the clause elements in accordance 3 with three principles operating in English: the usual final positioning of similarity and manner adjuncts, the principle of weight distribution, and the end-weight principle. A more complex analysis, on the other hand, is required for T ext 3. T exts 1 and 3 begin to diverge on the same structure (the similarity adjunct) as T exts 1 and 2; however, it appears at first glance that the restructuring in T ext 3 not only fails to adapt the text to the unmarked English word order principles but violates some of them even further. $e new predicator is still preceded by a similarity adjunct (the newly introduced clause of similarity); the number of adjuncts in initial position is still at least two, as in Text 1 – or, according to one possible interpretation, even three (see below); and there are none in final position. If the whole stanza is taken into account, the position of the similarity adjunct is thus still marked, in contrast to the unmarked choice in Text 1. If, however, the stanza is approached not as a compact syntactic unit but as a progression of lines, each constituting its own unit of meaning, the solution in Text 3 does conform to the unmarked word order in English. $ere are several arguments in favour of focusing on each line as a unit of meaning in its own right. $e poem discussed here is an instance of free verse, where the observation of line divisions is particularly important, since these form the only clues to the structure of a text (Kmecl 1995, 64, 66). In fact, the independence of free verse lines is so strong that it results in “syntagmatisation”, a process by which the words occurring in one line come to be understood as an autonomous unit of meaning, irrespective of grammar (Bjelčevič 1998–99, 37–8). $e tension between the structure formed by grammatical relations and the one formed by line divisions may have the effect of widening the range of associations and interpretations of a passage, as shall be demonstrated for the Zajc poem as well. Apart from the theoretical observations adduced above, it is therefore also the ambiguous status of several lines and phrases in this particular poem that argues for a line-by- line analysis in addition to the grammatical approach. If the line is therefore taken as a unit, the restructuring (but not reordering) of the elements in line 2 results in a sequence which corresponds to the unmarked English word order: the subordinator is followed by the subject, predicator, and an adverbial adjunct of place. $us both translations preserve the stylistic value of the original, one by reordering and the other by restructuring it. $e other structure in the first stanza to be discussed is the sequence constituting lines 3 and 4. $is passage displays an identical structure in all three texts: a prepositional phrase realising an adjunct of place in line 3, followed by a finite form of the verb “be” in line 4. However, the differences between the principles operating in Slovenian and English result in a different stylistic effect and even slightly different possibilities of interpretation for the two English texts. In terms of sentence structure the Slovenian version is ambiguous, admitting the following two possibilities: (1) $e adjunct of place belongs with the verb “be” (si), combining with it to form the second coordinate clause in the stanza. $e emerging pattern is that of a copula preceded by an obligatory adjunct, i.e. a complement. Since obligatory context-independent adverbials are rhematic (cf. Firbas 1992, 50), the adjunct must represent the rheme of the clause, and as such it typically occurs in final position. Nevertheless, its failure to do so in this case does not mean that the word order is marked. Since this sequence contains a clitic (si), it does not fall in the domain of the FSP-based free word order but is subject to the fixed word order rules. A clitic normally occurs after the first clause element, or after a part of the predicator (T oporišič 2000, 675); since si duly occurs after the adjunct representing the first clause element, the word order is to be interpreted as unmarked. (2) $e adjunct of place belongs with the first clause (the one with stojiš as its predicator), while the verb “be” forms a clause on its own. In this case “be” is used as a full lexical verb denoting existence, its isolation contributing to an intense concentration of meaning and a corresponding rhetorical effect. $is interpretation actually offers itself first when the text is read for the first time; the realisation that the adjunct could be paired with the second finite verb comes only when the end of the stanza has been reached. At this point, however, the recipient is likely to revise his or her assumption in favour of interpretation (1). $ere are several arguments speaking for the latter, if the customary structuring of texts is taken as a guideline: unlike interpretation (2), it is stylistically neutral, and it implies a relatively even distribution of weight among the two clauses (two lines are allotted to each), while (2) assigns three lines with four adjuncts to the first clause and a single word to the second. Nevertheless, the second, stylistically marked interpretation remains a valid option, particularly in the context of poetry. $e notion of the verb’s isolation is reinforced by the syntagmatisation effect (the verb is alone in its line), as well as by a parallel with the third stanza, whose final si can also be understood to stand on its own (but see discussion). Both possibilities are thus fully acceptable in Slovenian. An application of these interpretations to English yields different conclusions: Ad (1). $e possibility of an obligatory adverbial adjunct preceding the copula appears to be very slight in English: according to Davis, obligatory adverbials are predicate modifiers and thus normally occur in post-verbal position (1996, 62). $is possibility is even reduced when the adjunct forms no link with the preceding context and is unequivocally rhematic, as is the case here. Kirkwood, who treats obligatory and optional adjuncts of place together, notes that adverbial preposing is “not possible or not usual in English”, although a prepositional phrase in initial position is more acceptable than “here” or “there” (1970, 108–9). He only addresses one example of a preposed rhematic adjunct, where rhematicity is signalled by the intonation nucleus; the others are themes. A similar view is to be found in Quirk et al. (1992), who allot a separate section to the preposing of obligatory adjuncts, subject and object complements. $e acceptability of sentences containing such preposing without accompanying inversion is questioned; they are further described as “bathetic or misleading”, or as examples of “bad style” (ibid., 1379–80). $e remark that they are misleading because they perspective the focus towards the verb in final position, whereas the real focus is usually the subject (ibid.), reveals that the structures discussed are understood to function as themes. 4 $us even the (more frequent) preposing of thematic adjuncts is found to be questionable, whereas the possibility of rheme preposing is not considered at all. Huddleston et al. (2002) likewise deal with the preposing of complements (including obligatory adjuncts) separately, but they consider both “non-focus” and “focus” complement preposing. $ese correspond to the fronting of theme and rheme respectively. In contrast to the views expressed by Kirkwood (1970) and Quirk et al. (1992), neither process is labelled as unusual; it must be added, though, that all the examples of preposed adjuncts, thematic (Huddleston et al. 2002, 1376) as well as rhematic ones (ibid., 1381), are realised by prepositional phrases, which are regarded as more acceptable by Kirkwood as well. However, there are pragmatic restrictions – according to Huddleston et al., even a preposed rheme must have some link with the preceding context: “In focus complement preposing the preposed constituent serves both as a link with the prior discourse [underlined by N. G.] and as the value of the variable in the discourse-old 5 open proposition expressed by the rest of the clause” (ibid., 1381). In other words, a rhematic complement may only be preposed if, in addition to providing new information such as further specification, it is anticipated in the preceding context and thus conveys old information as well. An example is: “I promised my father – on Christmas Eve it was – to write home at my first opportunity” (ibid.); the preposing of the complement is said to be made possible by our knowledge that “events occur at particular times” (ibid., 1382). $e quality of old information, however, is not present in the Zajc poem; thus the preposing is irregular, or at least heavily marked, in contrast to the unmarkedness of T ext 1. Ad (2). $e place adjunct belongs to the preceding clause (the first coordinate clause in T ext 2, and the subordinate clause of similarity in T ext 3). Although the resulting text organisation is rather marked, as explained above in regard to T ext 1, it is syntactically more congenial to English than the adjunct-plus-copula interpretation. In fact, the oddity of the latter in English might well result in its not being recognised at all by a reader unfamiliar with the original text, so that interpretation (2) would remain not only the first but the only choice. $e original ambiguity is thus potentially lost in T exts 2 and 3. T o recapitulate, the preservation of the same structure in this case leads to the same interpretation being neutral in Slovenian but highly marked in English, as well as to a potential reduction in the number of interpretations. 4 5 3.2 An analysis of the second stanza shows that the three texts contain an identical sentence structure, namely an adjunct of time followed by three coordinate main clauses. In all three texts, this structure is also characterised by unmarkedness: due to the “setting” function of temporal adjuncts (Davis 1996, 62), these are common in initial position in both Slovenian and English. However, there is a difference between Text 1 on the one hand, and Texts 2 and 3 on the other, at the level of clause elements. It relates to word order and the accompanying stylistic effect. T ext 1 adverbial adjunct of time (ko te srečamo) – – main clause 1: predicator (se zmedejo) subject (koraki) – – main clause 2: subject (oči) predicator (zaplešejo) adverbial adjunct of place (v očesnih ležajih) – – main clause 3: predicator (so) subject (misli) subject complement (pest suhe travice) T ext 2 adverbial adjunct of time (when we meet you) – – main clause 1: subject (our steps) predicator (falter) – – main clause 2: subject (our eyes) predicator (reel) adverbial adjunct of place (in their sockets) – – main clause 3: subject (thoughts) predicator (become) subject complement (a bunch of dry grass) T ext 3 adverbial adjunct of time (when we meet) – – main clause 1: subject (our steps) predicator (become) subject complement (confused) – – main clause 2: subject (our eyes) predicator (rotate) adverbial adjunct of place (in their sockets) – – main clause 3: subject (our thoughts) predicator (are) subject complement (a fist full of dry grass) $e order of elements in the Slovenian text is exceptional in the first and third clauses, and neutral in the second. $e neutral FSP-based order is for the subject to precede the predicator, since the former (also when context-independent) usually has less information value than the context-independent verb. In the terminology of Firbas, “[t]he quality bearer carries a lower degree of C[ommunicative] D[ynamism] than ... an irretrievable quality” (Firbas 1992, 9). $e first clause therefore violates the principles of free word order. 6 $e third clause, on the other hand, which contains a form of the verb “be” – a clitic – in initial position, runs counter to the fixed word order principle as well, since clitics normally appear after the first element. $e order of this clause cannot be explained by the presence of the initial temporal adjunct as the first element, because the influence of the adjunct has been interrupted by the second coordinate main clause with its neutral word order. $e first and third clause in Slovenian thus display uncommon (and, in the case of the latter, marked) word order, whereas the second is neutral. By comparison, both translators into English have opted for unmarked word order in all three clauses. $is has consequences not only for the style but also for the iconicity of the text. While the Slovenian text could be said to reinforce the notion of vertigo described in the passage through its use of “confused” word order, and through a regular alternation of the latter with neutral order (the “confused” order predominating), the translations convey no such image. 3.3 !e third stanza displays three items of contrastive interest. $e first is a structure with unusual word order in Slovenian (clause 3a), which is omitted altogether in Text 2, while the attempt at imitation in Text 3 results in a different structural interpretation. $e second is the beginning of main clause 2 in Text 1 (the sequence of a temporal and spatial adjunct), which is again restructured in both English versions. And finally, the third is the ambiguous structure of lines 3–5, rendered even more vague by the lack of punctuation and the syntagmatisation effect, and its (likewise ambiguous) translations. 6 T ext 1 main clause 1 (in vemo:) – – direct object main clause 1a: subject (pot) predicator (je) subject complement (uročena) main clause 2a: subject (pot) predicator (je) subject complement (napačna) main clause 3a: subject complement (zavožena) predicator (je) subject (pot) – – main clause 2: adjunct of time (za hip) adjunct of place (v prostoru osvetljenem (belo? nevidno?)) (direct object: belo nevidno? nevidno?) predicator with an implicit subject (vidimo) – – predicator with an implicit subject (si) T ext 2 main clause 1 (we know:) – – direct object main clause 1a: subject (the path) predicator (is) subject complement (determined) main clause 2a: subject (the path) predicator (is) subject complement (mistaken) main clause 3a: OMITTED – – verbless main clause 2 (one second) – – main clause 3: coordinator (and) adjunct of place (in a floodlit expanse (white? unseen?)) (direct object: white unseen? unseen?) subject (we) predicator (see) – – subject (you) – predicator (are) T ext 3 main clause 1 (and we know:) – – direct object main clause 1a: subject (our path) predicator (is) subject complement (bewitched) main clause 2a: subject (our path) predicator (is) subject complement (mistaken) main clause 3a: predicator (is) subject complement (lost) subject (our path) – – main clause 2: adjunct of time (for an instant) subject (the place) predicator (is) 7 subject complement (illuminated (white? invisible yet seen?)) – – main clause 3: (subject complement: white invisible yet seen? invisible yet seen?) subject (you) predicator (are) $e first structure to be discussed is clause 3a and its translations. In the Slovenian text, clauses 1a, 2a, and 3a are identical with respect to clause elements, consisting of pot as the subject, the copula je as the predicator, and a subject complement. Clauses 1a and 2a display parallel (unmarked) word order as well; this parallelism, however, is interrupted in the third coordinate clause. $e change in word order cannot be accounted for by FSP because pot, which appears in the rhematic, final position, has been mentioned twice and unmistakably conveys old information. On the other hand, the change readily admits of several explanations as a rhetorical figure, which results in the string of coordinate clauses beginning and ending with the same word. For example, the inversion may signal the completion, “wrapping up”, of the string, or the circularity may reinforce the description of feeling lost. In any event, the sequence is uncommon. $e two translations address this device in radically different ways. Whereas T ext 2 simply omits the entire clause, so that only the two parallel clauses are preserved, T ext 3 attempts to convey 7 the changed order. $e resulting structure, however, displays the sequence predicator – subject complement – subject, which is neither preposing (in this case the SC would come first) nor inversion, but a structure not only marked but grammatically irregular in English. Attempting to impose an acceptable interpretation on it, a reader would probably take the entire predicate, “is lost”, as an instance of asyndetic coordination with the predicate of clause 2a, while the final “our path” would be understood as an expressive repetition of the subject already mentioned in 2a. In this case the parallelism of clauses 1a and 2a would be lost, since clauses 2a and 3a would be interpreted as forming one clause containing a coordinate predicate. Even if the coordinate predicates are interpreted as the coordination of two clauses, with the subject of the second ellipted (Quirk et al. 1992, 948–9), this ellipsis nevertheless links them more closely. $us the second and third clause would be perceived as forming a more tightly knit unit, whereas the first would stand alone. $e conclusion is that the English structure, if taken as an imitation of the original, is not only marked but irregular; if interpreted independently, on the other hand, it results in a different stylistic effect. $e second structure where both translations diverge from the original is the sequence of a temporal and spatial adjunct in line 3 of the stanza. T ext 2 introduces two clauses instead of one, expressing the first adjunct as an irregular verbless clause, whereas T ext 3 opts for a complete rearrangement, not only of the line but of the entire clause. $e head of the phrase realising the erstwhile spatial adjunct becomes the subject; the verbal elements are realigned in keeping with the new situation, so that the participle originally realising a modifier (osvetljenem) appears in the predicate as the SC or as part of the predicator (“is illuminated”), while the original predicator vidimo is changed to a passive construction, “seen you are”, or – in another interpretation – reduced to a participle, “seen”. $e changes in the two texts suggest that an accumulation of (initial) adjuncts is not congenial to English. $e resulting attempts to avoid it implement the principle of weight distribution operative in English but not Slovenian (Davis 1996, 66). Moreover, the transformation of an adjunct into the subject in Text 3 reflects the well-attested tendency of the English language to make the element in initial (thematic) position the grammatical subject. $is is a means of avoiding a clash between two word order principles, i.e. the grammatical and the FSP linearity principles (Firbas 1992, 119–20). $e relation between English and Slovenian in this respect is treated e.g. by Klinar (1996), who addresses the frequent transformation of Slovenian adverbial adjuncts realised by prepositional phrases into English subjects. $e change in Text 3 is thus revealed to be a common device of bridging the differences between the two languages. A by-product of this change, however, is a redistribution of the prominence allotted to the clause elements. While prostor osvetljeni (“the illuminated place”) only sets the scene in T ext 1, being a spatial adjunct, it is foregrounded in T ext 3 as the subject; also the actions expressed by the verbs are shown in a different light, since the restructuring of the past participle osvetljenem as part of a finite verbal form in English endows it with prominence and an absolute temporal dimension, whereas the reverse process is applied to vidimo. $e changes in T ext 3 therefore result in a change of clause element prominence and even meaning. $e third issue is the ambiguity of the last three lines, which can be established for all three texts. T ext 2 closely follows the original and thus offers similar possibilities of interpretation. By comparison, T ext 3 is ambiguous as well but completely restructured, so that the interpretations differ from those applicable to T exts 1 and 2. In all three texts, structural ambiguity is caused by the sequence belo nevidno and its translation equivalents (“white unseen” in T ext 2, “white invisible” plus “yet seen” in T ext 3). $e analysis of this element in its turn determines the analysis of the – likewise ambiguous – final element, si (“you are”). $e sequence belo nevidno admits the following three interpretations in Texts 1 and 2: (1) Both words function as modifiers in the nominal phrases beginning with prostoru osvetljenem and “a floodlit expanse”. $e difference is that in T ext 1, they can only be understood as adverbs, and thus modifiers of the participle, whereas in T ext 2 they function as non-restrictive postmodifiers to “expanse”. If both words are understood to modify the nominal phrase, the transitive finite verb (vidimo / “we see”) remains without a direct object; this slot may then be filled by the final clause (si / “you are”). $is interpretation establishes a parallel with the beginning of the stanza, where the predicator of the main clause is likewise realised by a finite form in the first person plural, and complemented by clauses lacking a marker of subordination. ($e absence of a colon between vidimo and si is no counterargument, given the general lack of punctuation in the poem.) If belo nevidno / “white unseen” is construed as a constituent of the nominal phrase, the word order is marked both in Slovenian and English. In English the two modifiers function as an afterthought, which is a less basic structure. $e Slovenian version, by contrast, is unmarked at the nominal phrase level because a participial clause in modifying function typically occurs to the right of the headword (T oporišič 2000, 561–2). Within this participial structure, however, the order is marked, since an adverbial of quality such as belo or nevidno would normally precede the participle or adjective (ibid., 565). $us both texts employ marked word order, albeit at different levels. (2) $e second interpretation is that only the first word (belo / “white”) is attached to the preceding nominal phrase, whereas the function of direct object to vidimo and “we see” is implemented by nevidno, “unseen”. $is interpretation is in fact the more likely of the two mentioned so far, since the etymological relatedness but semantic opposition between nevidno and vidimo suggests a rhetorical figure, thus linking them as a unit. In this case the direct object slot is filled and si / “you are” is an entirely independent clause. (It functions fairly independently even if realising the direct object to vidimo / “we see”, which is due not only to its lack of subordination markers but also to the syntagmatisation effect.) $e marked character of belo / “white” when attached to the preceding phrase has been discussed above; moreover, the position of nevidno / “unseen” as a direct object merits discussion in both texts as well. In Slovenian, where the criterion for (un)marked word order is FSP , the sequence of a context-independent object occurring before the predicator violates the FSP linearity principle, since, according to Firbas, “context-independent objects... exceed the verb in C[ommunicative] D[ynamism] irrespective of sentence position” (1992, 42). However, this word order can be explained as the literary device of arbitrary theme selection, which means that the sequence is rhetorically coloured but unmarked. $e English text, on the other hand, is marked because it violates the grammatical word order principle (the sequence subject – predicator – object). $e unusualness of such a construction in English is supported also by the findings of Gómez-González, whose corpus shows that objects and complements realised by noun phrases are the least common instances of preposing (2001, 232). $us neither version displays the basic word order, but the degrees of markedness vary. (3) Both words are the direct object of vidimo / “we see”, while si / “you are” is independent. $is interpretation is supported most strongly by the syntagmatisation effect, since it concurs with the distribution of words across lines. Again, the position of the context-independent direct object before its predicator is nonbasic in both languages, for the reasons discussed above. Moreover, the order within the Slovenian phrase prostoru osvetljenem is now marked: the isolated participle no longer functions as a participial clause but rather as an adjectival modifier, which should occur to the left of its headword according to the fixed word order rules (T oporišič 2000, 558). $e English version, using the participle as a premodifier, is accordingly unmarked. Ambiguity is present in T ext 3 as well. T o begin with, there is the question whether “illuminated” is to be taken as part of a passive form and thus subsumed under the predicator, or as an adjectival subject complement following a copular predicator. $is issue, however, does not pertain to the present contrastive discussion; for practical reasons, “illuminated” will be regarded as a subject complement, on a par with the following adjectives. $e analysis will focus on items which can be compared across all three texts, namely the elements corresponding to the sequence belo nevidno and si. It must be noted that the sequence subsumes more elements in T ext 3 than in T exts 1 and 2: “seen” can be coordinated with “white” and “invisible”, and the meaning strongly advocates its being paired with “invisible” into one unit, “invisible yet seen”. An analysis of “white” and “invisible yet seen”, then, reveals that both or one of the units may be linked either to the clause beginning in line 3 or to the one in line 5, as is the case also in T exts 1 and 2. However, they represent different clause elements: instead of being modifiers or direct objects, the units are always to be understood as subject complements, linked either to “is” in line 3 or “are” in line 5. If taken together with “is”, one or both units constitute a coordinate subject complement with “illuminated”; if both are taken with “are”, they form a coordinate SC on their own. $e former structure with the complement following the predicator has unmarked word order (as opposed to the first interpretation for T exts 1 and 2); the latter, on the other hand, is marked. $e preposing of a subject complement without accompanying inversion in fact appears to be heavily marked in English, especially when the element represents the rheme. Kirkwood (1970), for example, does not address this structure at all; a discussion can be found in Huddleston et al. (2002, 1375), but limited to the section on non-focus, or theme, preposing, whereas rheme preposing is not considered. Quirk et al. likewise only address the preposing of thematic subject complements, questioning its acceptability and labelling it as “bad style” and “misleading” (1992, 1379–80). $e only one of my references to address the possibility of preposing rhematic subject complements is Firbas (1992, 46–9). Discussing context-independent subject and object complements, he refers to the aforementioned passages from Quirk et al. and corrects them, pointing out that, in the case of a preposed context-independent complement, the focus would not be on the subject but on the complement – in other words, he presents the possibility that a preposed subject complement may be the rheme. He concurs in viewing the structure as awkward, but adduces an example from Katherine Mansfield which he considers free of this fault (ibid., 48). It must be noted, however, that the Mansfield text is highly literary and thus stylistically marked; moreover, the greater acceptability of the structure may be partly influenced by the context. Since the sentence occurs with two others which share the same communicative purpose (the description of flowers), even if not the same structure, it may be this parallelism that renders it less conspicuous. All this evidence suggests that the preposing of rhematic subject complements is well-nigh irregular, or at least heavily marked; this is also corroborated by the corpus analysis of Gómez-González, mentioned above, where objects and complements realised by noun phrases are found to be the least frequent types of preposing (2001, 232). $e implication of the different analyses for the status of “you are” is the same as with T exts 1 and 2: if all the adjectival words are subsumed under the preceding clause, it stands alone, if not, it gains a preposed subject complement. However, there is a difference concerning the overall text organisation: since “you are” is no longer preceded by a finite verbal form in the first person plural and cannot function as its direct object, there is no parallelism with the beginning of the stanza. 3.4 !e fourth stanza comprises two main clauses. $e first presents no translation problems, since it is structurally parallel and unmarked in all three texts, beginning with an adjunct of time as the setting. All three texts also contain an anaphora of kmalu / “soon” at the beginning of both main clauses and in the fifth stanza, which consists only of this line. (T ext 2 differs from the other two in subsuming the last line under the fourth stanza.) $e second clause, on the other hand, shows four structures of interest to be discussed below. $e first, which is ambiguous, offers almost the same possibilities of interpretation in all three texts. $e second concerns the different translational approaches to a nominal phrase with marked word order. $e third is a verbless structure which is heavily marked in Slovenian but unmarked when preserved in English. $e last issue is the position of the predicator, which is again identical in the three texts, but marked in Slovenian and unmarked in English. T ext 1 main clause 1: adjunct of time (kmalu kmalu) predicator with an implicit subject (zapustimo) direct object (svet zvokov barve in okusa) main clause 2: adjunct of time (kmalu) ?verbless clause (kaplja med sestrami kapljami (brez zvoka brez občutka / kapljastega?)) (adjunct of manner: brez zvoka brez občutka / kapljastega) predicator with an implicit subject (bomo padli) adjunct of direction (na svoje bele razlaščene kosti) T ext 2 main clause 1: adjunct of time (soon very soon) subject (we) predicator (’ll leave) direct object (the world of sounds, color and taste) main clause 2: adjunct of time (soon) adverbial adjunct realised by a supplementive clause (a drop among our sister drops / (soundless senseless / of our dropness?)) (adjunct realised by a supplementive clause: soundless senseless / of our dropness) subject (we) predicator (’ll fall) adjunct of direction (onto our white repudiated bones) T ext 3 main clause 1: adjunct of time (soon soon) subject (we) predicator (will leave) direct object (the world of sounds colors and tastes) main clause 2: adjunct of time (soon) adverbial adjunct realised by a supplementive clause (only raindrops among other raindrops (without sound without feeling? / like drops?)) (adjunct of manner: without sound without feeling (like drops?)) (adjunct of similarity or manner: like drops) subject (we) predicator (will fall) adjunct of direction (on our dispossessed bones) $e first issue to be addressed is the syntactic role of brez zvoka brez občutka / kapljastega, the original structure of which (i.e. two prepositional phrases) is preserved in T ext 3 but restructured as two adjectival phrases in T ext 2. Due to their parallelism, the two units are assumed throughout the discussion to be coordinated despite the lack of an overt coordinator. T exts 1 and 3, which have the same structure, admit the following two analyses: (1) $e sequence is the second postmodifier to kaplja / “raindrops”, the first being med sestrami kapljami / “among other raindrops”. T echnically speaking, the sequence could modify sestrami kapljami / “other raindrops” as well, but the focus on “we” throughout the text makes this interpretation very unlikely. (2) $e sequence is an adjunct of manner; as such it would be understood as marked in English (on the normally post-verbal position of manner adjuncts, see Davis 1996, 62) but unmarked in Slovenian. It is true that, according to the FSP linearity principle, the Slovenian adjunct would be expected to follow the predicator, seeing that a manner adjunct normally introduces important new information and is thus more informative than the predicator (Firbas 1992, 53). However, this choice may be explained as the literary device of arbitrary theme selection, and as such rhetorically coloured but unmarked. T ext 2, although employing adjectival phrases, lends itself to two interpretations as well. $e first is identical to (1) for T exts 1 and 3 (the sequence is a coordinate non-restrictive postmodifier to “a drop”, or, technically speaking, even to “our sisters drops”). $e second interpretation in its turn is similar to (2), suggesting the role of an adjunct. In this case, however, the adjunct is not one of manner but is realised by a supplementive clause. 8 As a result this version is unmarked, since initial position is very common for supplementive clauses in English (Quirk et al. 1992, 1125). $e conclusion is that T ext 3 preserves the phrase and clause structure of the original but possibly acquires markedness as a consequence, whereas T ext 2 deviates from the structure but preserves an unmarked character. Within the sequence discussed above, the Slovenian phrase občutka / kapljastega has marked word order, since the fixed word order rules require an adjectival modifier to precede its 8 headword. $is arrangement, which would be marked in English as well, is avoided in both translations and replaced by unmarked paraphrases. $e use of these, however, carries still new implications. In Text 2 the lack of structural markedness is compensated for by striking word choice (“dropness”). Text 3, on the other hand, introduces a new structural ambiguity, reinforced by the syntagmatisation effect, which may result in a different meaning; moreover, this other possible interpretation again employs marked word order. In Slovenian, the adjective kapljastega unmistakably modifies the noun in the preceding line, as shown by its agreement with the latter. $e prepositional phrase, on the other hand (“like drops” in Text 3), may likewise be understood as a postmodifier to the nominal headword (“feeling”), but it could also realise an adjunct of manner referring to “we will fall”, in which case the word order (with a pre-verbal manner adjunct) would be marked. $e line division, which follows the layout of the original, actually tips the scales in favour of the adjunct interpretation, so that the replacement of the adjective with a prepositional phrase may be said to change the meaning. $e third item of interest is the verbless sequence kaplja med sestrami kapljami (brez zvoka brez občutka / kapljastega). In Slovenian, its syntactic role presents some problems of classification. On the one hand, it could realise what Toporišič labels povedkov prilastek (“predicative modifier”), namely part of a semantically complete clause which attributes a nonobligatory feature to – in this case – the subject phrase (Toporišič 2000, 618). $e examples adduced, however, suggest that the structure in Zajc is very rare, since none in the relevant section (ibid.) correspond to it entirely. If the predicative modifier is in initial position and not introduced by a conjunction, as in the Zajc passage, it is never exemplified by a nominal phrase but only by adjectival ones, as in Še star je bil zmeraj vesel. If realised by a nominal phrase, it follows the verb if lacking a subordinator (Jure se je od vojakov vrnil cel možak), while in initial position it is always introduced by kot (Kot otrok sem se rad igral v potoku). $e possibility of an initial nominal phrase without an introductory kot is not exemplified in this section at all. Another section does list a structure parallel to kaplja med sestrami kapljami etc. (Otrok sem / si bil zmeraj vesel; ibid., 608), albeit without analysing it in terms of clause elements; the scarcity of such examples, however, implies that it is highly uncommon. Another possibility would be the role of samostalniški polstavek, i.e. a “nominal verbless clause”. However, judging by the example Sosedov sin, fant od fare, je pa res vzor mladeniča (T oporišič 2000, 633), such a clause should normally occur after the subject. Whichever interpretation is adopted, the position of the structure thus appears to be heavily marked in Slovenian. In the English texts, on the other hand, the original structure is preserved, yet its status is unequivocal and unmarked: it is a verbless supplementive clause realising an adjunct, which often occurs in initial position. $e sequence in T ext 3 does appear the slightly less usual of the two, possibly because of the introductory “only”; nevertheless, the fact remains that the structure which is marginal in Slovenian is perfectly common in English. Finally, the position of the predicator is marked in Slovenian, since bomo is a clitic and thus due to appear after the first clause element (kmalu kmalu in this case), not the second or even third, if brez zvoka brez občutka kapljastega is taken to realise a separate adjunct. By contrast, the – identical – position of the predicators in the translations is unmarked. Since both form and function play a vital role in poetry, no uniform solution to translation problems can be expected even from one translator, let alone two. $us the analysis of poetry translations cannot yield a prescribed formula for each recurring problem. $e value of such comparisons, however, lies in detecting the common points of divergence from the original, which may then be established as fields of contrast between the source and target languages. Moreover, the translators’ choices to preserve either form or function in each case reflect the delicate nuances of acceptability in the target language, which are lost in routine translating. [ ] ó á Á [ ]