Keywords

1 Background

Text corpora of various kinds, among them those composed of original texts in one language and their translations in another, play a significant role in Contrastive Linguistics (CL) as well as in Translation Studies (TS) and have “the potential to bring the two fields even closer together” (Granger, 2003, p. 25). CL has traditionally seen corpora as essential (cf. Filipović, 1973), and its revival over the last three decades has often been attributed to the development of various electronic tools, in particular computer corpora as one of the two key factors (the other one being globalization; cf. Gómez-González & Doval-Suárez, 2005; Granger, 2003; Ramón García, 2002). The new context has affected the discipline in various ways, including, as Ramón García (2002) notes, “a shift in the focus of CL from linguistic systems in a more abstract sense to more specific issues in language use” (p. 395). The great potential of corpus-based research in the field of TS has also been identified by many researchers, who have discussed opportunities as well as limitations (among early examples are Altenberg & Aijmer, 2000; Baker, 1993, 1995; contributors to META 43(4), 1998). Ramón García (2002) thus states that “[e]lectronic corpora have been used extensively in TS since the beginning of the 1990s and their importance is such that they have meant a turning point in the discipline” (p. 401). Corpora have been seen as useful not only in relation to machine translation and terminology but also in studies exploring issues such as translation universals, language of translation, translator style, translation ideology, translation strategies and methods, equivalence, cognitive processes in translation, etc. (cf. Altenberg & Aijmer, 2000, p. 27; Lan, 2014; Lewandowska-Tomaszczyk, 2012b; Mikhailov & Cooper, 2016, pp. 16, 184–192).

The two disciplines do not only share the need to make use of corpora but they are closely intertwined in a more general sense, as was noted already in the decades preceding the appearance of electronic corpora (cf. Ramón García, 2002, p. 397), as well as later on. Hoey and Houghton (2001) describe the bidirectional relationship between Contrastive Analysis (CA) and translation by pointing out that “(…) the translation of specific pieces of text may provide the data for CA (…). CA may provide explanations of difficulties encountered in translation” (p. 49). Just as they strongly state that “Translation as a source of data for CA is strictly unavoidable”. (ibid.), Ramón García (2002) concludes that “(…) any type of approach to translation from a descriptive corpus-based perspective must take into account some kind of contrastive aspect. (…) CL is thus a basic ingredient of TS” (p. 403).

The present study wishes to identify some insights in the realm of either CL or TS that can be gained from focusing on a particular linguistic unit, namely the Swedish pronoun man, and its Croatian translation equivalents in a translation corpus. The possibility to use one and the same corpus as a basis for conclusions within both of the two closely connected disciplines seems particularly welcome in cases such as this, where lesser-taught languages and lesser-studied language pairs are involved.

2 Translation Corpora

The corpus used in the present study consists of original texts in Swedish and their translations into Croatian and it is termed here a ‘translation corpus’, in line with e.g., Altenberg and Aijmer (2000, p. 16). In the literature such corpora are referred to as either ‘translation corpora’ or ‘parallel corpora’ (or both, as in e.g., Lewandowska-Tomaszczyk, 2012a, p. 5; Ramón García, 2002, p. 401). Granger (2003) has suggested that the choice of the name may depend on the affiliation of the researcher (‘parallel corpora’ in the TS framework, ‘translation corpora’ in CL) but that such usage “is not entirely consistent either” (pp. 19–20).

Translation corpora offer good grounds for cross-linguistic research for a number of reasons. They provide a large number of pairs of related units in different languages, which enables comprehensive and nuanced conclusions about cross-linguistic correspondences. The fact that for one source-language unit a relationship is typically established with a number of different target-language renderings does not only “make it possible to discover cross-linguistic variants” (Altenberg & Granger, 2002, p. 9) but it also sheds light on different meanings and different functions of the units thus enabling their better understanding within their respective languages. Even if the exact nature of the relationship established cannot be unequivocally defined—which in itself is a point of great interest for researchers—the fact that two units are seen as establishing translation equivalence in numerous occurrences justifies conclusions about their correspondence and invites its careful interpretation. For TS researchers corpora are a rich source of authentic translation solutions which enable them to “captur[e] the distinctive features of the translation process and product” (Granger, 2003, p. 22). Another advantage of translation corpora, no less significant for being obvious, is that they make research more objective by providing “access to a large number of interpretations besides the linguist’s own introspective judgement” (Aijmer, 2007, p. 33).

While they undoubtedly present a most useful tool, translation corpora have their limitations, and conclusions based on their evidence must be made with caution. Granger (2003), when discussing limitations with regard to corpus-based approach in general, focuses on two such circumstances, namely the limited availability of corpora (still true today of large translation corpora for many language pairs, including Swedish-Croatian) and “the corpus predilection for form-based research” (pp. 22–23). While these two may influence the decision on the kind of study to be conducted, there are other characteristics of translation corpora that may influence the quality of the study and the value of its findings. They stem from the very nature of translation process and translation product, and primarily have to do with the language of translations and the elusive nature of ‘translation equivalence’. When pairs of linguistic units are analysed one must bear in mind that different language systems will possess “what can be called equivalents not of an identity but rather of an approximation type” (Lewandowska-Tomaszczyk, 2012b, p. 4). Researchers using translation corpora must also recognize that translations tend to be influenced by the source text (so much so that a ‘law of interference’ has been proposed by Toury, 1995, p. 275), and that this results in phenomena that may be very difficult to spot (e.g., lexical frequencies different from those in original texts in the same language, as has been observed by Gellerstam, 1986 and others). Translation scholars have identified other common features of the language of translated texts, not source-inspired, e.g., selection of linguistic units in line with what Toury (1995) has termed ‘the law of growing standardization’ (p. 267). Awareness of these, as well as of the intricate nature of translation equivalence (including the very basic distinction between translation equivalence and contrastive correspondence, cf. Ivir, 1997) may prove essential in securing sound conclusions based on evidence from translation corpora.

3 Swedish Pronoun Man and Its Croatian Equivalents

3.1 Man

Descriptions of the pronoun man in grammars of the Swedish language show that its usage is by no means simple. It is often classified as an indefinite pronoun (Bolander, 2005; Holmes & Hinchliffe, 2003; Ljung & Ohlander, 1971/1982) or, more specifically, as a quantitative indefinite pronoun (Josefsson, 2009/2011, who remarks (p. 82) that the Swedish Academy’s Grammar also classifies man as quantitative). Hultman (2003/2008) describes it as a generalizing quantitative pronoun but also points to its anaphoric use “when the speaker/writer cannot or doesn’t want to make his/her reference precise” (p. 122).Footnote 1 In a similar way the generalizing function is implied in other grammars (e.g., Bolander, 2005, p. 128), often explicitly or implicitly connected with remarks about the pronoun, in itself singular, appearing with plural complements in colloquial Swedish (cf. Bolander, 2005, pp. 127–128; Holmes & Hinchliffe, 2003, p. 211; Hultman, 2003/2008, p. 121).

The fact that three qualifiers (indefinite, generalizing, quantitative) are applied to one and the same pronoun need not be considered curious (‘indefinite’ has been interpreted as subsuming ‘general’, cf. Kordić, 2002, p. 49) but man has other, less expected, usages. The following two are characteristic of the colloquial language and are not as regularly described in reference books but they have been noted in some: man can be used by the speaker to refer to him or herself (all the monolingual dictionaries of Swedish listed at the end record such usage and Holmes and Hinchliffe (2003) describe it “as a polite or mildly ironical substitute for jag [“I”]” (p. 149)). They also describe the use of man instead of du (“you, sg.”), as a “familiar term of address with a touch of ironical politeness and formality” (ibid.), as does SAOB (entry: man). Even without having mentioned the latter two uses of man, Ljung and Ohlander (1971/1982) warn that one has to be careful in translation and point out that the pronoun has a number of equivalents in English: one, you, we, they, people, etc. (p. 130).

3.2 Croatian Equivalents

A very obvious candidate for the Croatian equivalent of man is the substantive čovjek (“man”). The difference in the word class can be neglected in view of the Swedish pronoun’s etymology (developed from the substantive man, cf. SAOB: man) and other obvious similarities [cf. Giacalone Ramat and Sansò (2007) who focus on “indefinite man-constructions” in which “the subject position is filled by (an element deriving etymologically from) a noun meaning ‘man’”. (p. 95)]. In her study of the Croatian substantive, Kordić (2002) notes that “[s]ome nouns have a very general meaning, similar to that of pronouns” and that this is “primarily true of the noun čovjek” (p. 49).

Articles for the entry man in Swedish-Croatian dictionariesFootnote 2 also quote čovjek as an equivalent along with several others, complemented with an explanation (“indicates an unspecified person”) and examples of usage, whose Croatian translations include other equivalents than the ones quoted. In the two print editions consulted, ljudi (pl. form of čovjek, “people”) is the only other equivalent, while Lexin online lists netko (“someone”) and neki (“some”), with čovjek quoted third and ljudi not appearing at all. Translations of the usage examples contain, in addition to čovjek, the so called se-passive construction as an equivalent to man: “man räknar med en rekordskörd i höst – računa se s rekordnom žetvom na jesen” (se-passive construction is composed of the reflexive pronoun se and the active form of the finite verb. The Croatian sentence thus reads [A] record harvest in [the] autumn is counted with. rather than the Swedish source’s One counts with a record harvest in [the] autumn).

A cursory look at several web sites offering online Swedish-Croatian dictionaries has not identified any additional equivalents to man. In other words, altogether five lexicographic equivalents have been established: čovjek, ljudi, netko, neki, se-passive. Considering the various meanings of man described above, the list can hardly be exhaustive, and a good way to complete it is certainly by analyzing translation equivalents of the Swedish pronoun in a translation corpus.

4 Aims and Material

In view of all that has been said, a double aim was envisaged for the study: firstly, to compare the translation equivalents of man in a translation corpus to the list of lexicographic equivalents, assess the latter in that light and, if it proves justified, suggest changes to the list; secondly, to see whether the set of Croatian translation equivalents established in the corpus offers any other, if only preliminary, insights, in particular regarding the usage of the pronoun man and the nature of the equivalence established.

When discussing limitations of the corpus-based approach, Granger (2003) observed that “The dearth of parallel [i.e., translation] corpora accounts for the relatively large number of cross-linguistic studies which are corpus-based in the sense that they rely on authentic texts rather than introspective methods, but not in the more usual sense of computer corpus-based” (p. 23). The corpus compiled for this study is electronic but it has had to be of the “do-it-yourself kind” (cf. Mikhailov & Cooper, 2016, p. 39). It consists of six Swedish novels and their published Croatian translations (five translators) as well as 105 standard pages of Swedish non-literary texts of different types and their translations into Croatian taken from the Masters theses (the Swedish-Croatian translation part) by five graduate students of Swedish/translation stream at the University of Zagreb. The Swedish subcorpus exceeds 410,000 words and it includes 998 occurrences of the pronoun man in its subject form (the object form en and the possessive form ens have not been included in the analysis at this stage). The two subcorpora have been aligned, the Swedish one searched for pronominal man and the equivalents in the Croatian subcorpus established. Since some Swedish sentences containing man were left out in translation, 988 pairs of sentences have eventually been analysed. The aims set for the study have required that the analysis be both quantitative and qualitative.

5 Procedure and Results

The first stage of analysis has shown clearly that man presents a translation problem in Swedish-Croatian translation which cannot be solved by choosing from among a few obvious equivalents. On the contrary, three groups of solutions emerged based on the size of the segment for which an equivalent could be established. In the Croatian sentences that have an equivalent to man at the word level, 23 different equivalents appear, either pronouns (19) or nouns (4). Among the nouns, three are indeed specific nouns, namely čovjek (“man”), ljudi (“people”) and osoba (“person”), while the fourth stands for the category ‘noun’, i.e., different nouns that were used when the reference of the Swedish man got explicated in the translation, based on the context. An example is

Sw. Man kom igång med att hurra lite och ställde sig upp (…) (“One started (…)”)

Cro. Gosti su počeli nazdravljati ustavši od stola (…) (“[The] guests started to cheer having risen from the table (…)).

The second group of translation solutions includes three verbal constructions which actually correspond to the combination of man and the finite verb in the source sentence. These are the already mentioned se-passive, as well as impersonal verbal constructions and infinitive:

Sw. Sådant bara vet man, det går inte att förklara. (“One just knows such (…)”)

Cro. Takve se stvari jednostavno znaju, to se ne da objasniti. (se-passive; “Such things are simply known (…)”);

Sw. Man fick vara glad åt det lilla. (“One had to be happy (…)”)

Cro. Treba biti zadovoljan i s malim. (impersonal verb, subject not expressed; “[One] has to be happy also with [the] small”);

Sw. De visste inte hur man överlevde i fiendelägret. (“(…) how one survived (…)”)

Cro. Ne znaju kako preživjeti u neprijateljskom kampu. (infinitive; “They don’t know

how to survive (…)”).

The third set of solutions involves equivalence only at the sentence (clause) level, with translations that either have a different subject or for some other reason need no word-equivalent to man. Such sentences are often constructed through modulation or use of Croatian idiomatic language:

Sw. Jag försöker skrika, men det är svårt om man har tungan fastklistrad. (“(…) if one has the tongue glued up”)

Cro. Pokušavam vikati, ali to je teško kad je jezik zalijepljen. (modulation; “I try to scream but that’s difficult when [the] tongue is glued up.”)

Sw. Vad säger man? (“What does one say?”)

Cro. Je l’ tako? (idiomatic language; “Is [that] so?”).

As the next step, the equivalents of the first kind (23 altogether) were grouped in order to match the usual description of the functions of man, disregarding differences that seemed irrelevant in that context. The various indefinite pronouns were thus lumped together (svi “all”, svatko “everybody”, nitko “nobody”, mnogi “the many”, tko “who”) to match the “indefinite” or “indefinite quantitative” description of man, with the exception of the pronouns that are quoted in Croatian dictionaries, which were kept as a separate group (neki “some”, netko “somebody”). At first glance it might seem that the same could be done with personal pronouns of the same person since the number does not change the relevant functions (e.g., speaker/addressee). That is, however, not the case both because of the function of man (replacing jag and du, rather than the plural pronouns of the same person) and because of the fact that some Croatian personal pronouns have, along with their prototypical referential function, an impersonal use and it is exactly the number that is relevant in that respect (only 2nd person singular and 1st and 3rd persons plural can have that function, cf. Silić & Pranjković, 2005, p. 318). Personal pronouns posed another dilemma in that they sometimes appear as an explicit one-word equivalent for man but are more commonly implicit in sentences with no explicit subject and the grammatical person marked in the finite verb (the Croatian syntax allows for unexpressed subjects). The corpus thus contains e.g.,

Sw. Får man prova? (“May one try?”)

Cro. ”Smijem i ja probati?” (“May I try, too?”, subject “I” expressed)

as well as

Sw. Man måste tänka på pengarna! (“One must think about the money!”)

Cro. Moram misliti na novac. (“[I] must1st sg think about (…)”, unexpressed subject, the verb form indicating 1st person singular).

Since it is the meaning of the person and number that is essential here rather than the way in which it is expressed, examples of both kinds were taken to belong to the same category. Eventually, eighteen different categories of translation equivalents were thus obtained and their respective shares in the overall number (i.e., 988) calculated.

The first point of interest was to see how common the equivalents established in the dictionaries were within the set of translation equivalents. As can be seen in the next graph (Fig. 1), taken together, they account for less than a half of all the translation solutions (42.1%). In the 988 translated sentences, the se-passive construction has been used 340 times (34.4%), the word-equivalent čovjek 66 times (6.7%), pronouns netko or neki altogether 10 times (1%). In 572 sentences (57.9%) equivalence has been established in a way not mentioned in the dictionaries.

Fig. 1
figure 1

Share (%) of lexicographic equivalents of man (se-passive, čovjek, netko/neki) versus other translation solutions in the corpus

Full size image

In the next step the heterogeneous category ‘other’ was broken up and the share in the corpus of each of the eighteen categories mentioned above was established (Fig. 2).

Fig. 2
figure 2

Share (%) of the various translation equivalents in the corpus

Full size image

At this stage it has become clear that the Croatian se-passive construction is the most common translation solution in the corpus (340 occurrences). Following it are the 2nd person singular markers (personal pronoun ti or finite verb forms, 174 occurrences) and the 3rd person plural markers (pronouns oni, one, ona or verb forms, 108 occurrences). The next three categories all have a share of between 5 and 7% in this corpus: čovjek (66), equivalence established through the use of idiomatic language (59), and the markers of 1st person plural (personal pronoun mi or appropriate verb forms, 56 altogether). Following these is the category ‘impersonal verb’ (35), in which an impersonal verb is used in translation, with the subject neither expressed nor implied. Slightly less frequent in this corpus is the use of the 3rd person singular markers (pronouns on, ona, ona/verb forms, 31). Two very different types of solution occur 24 times each: equivalence achieved at the sentence/clause level through modulation and by using an indefinite pronoun other than netko or neki (if the two groups of indefinite pronouns were not kept separate, which has been done here only in order to enable an assessment of lexicographic equivalents, the overall share of indefinite pronouns in the corpus would be 3.4%). The markers of the 1st person singular (personal pronoun ja or the appropriate verb forms) occur 19 times as translation equivalents and are followed by the infinitive verb form (14) and the two remaining categories involving lexicographic equivalents i.e., the indefinite pronouns netko or neki and the noun ljudi with 10 occurrences each. The remaining four types of solutions have been used less than 10 times.

6 Discussion

A comparison of the results of the quantitative analysis and the articles for the entry man in the consulted dictionaries shows that the latter do not fully reflect the variety of uses of the Swedish pronoun. Notwithstanding the fact that many different circumstances determine the lexicographic presentation, some changes in what has been established as the standard set of lexicographic equivalents and explanations are warranted. The following seems relevant:

  • the se-passive construction has proved to be the most frequently used translation solution, almost twice as common as the one following it, while its dictionary status is only minor (not specifically referred to at all, just appearing in translations of the usage examples);

  • while the category ‘2nd person singular’ has the second largest share in the corpus, it is not mentioned in the dictionaries at all. As was indicated above, its common use in Croatian translations has to do with the fact that it is not only used in reference to the addressee but it can also express impersonal meaning (cf. Silić & Pranjković, 2005, p. 318), as in

Sw. Rädsla är något man lever med ensam. (“Fear is something one lives with.)

Cro. Strah je nešto s čime živiš sam. (“Fear is something with which [you] live2nd sg alone.”).

The impersonal use of the 2nd person singular, as well as of the 1st and the 3rd person plural, may be considered ‘covered’ by the dictionary explanation “indicates an unspecified person”. However, the corpus also includes examples of the 2nd person with specific reference to the addressee, as in

Sw. “Hörrudu din lille fjant!” fräste han. “Plötsligt är man så jävla kaxig! Akta dig ditt lilla äckel!” (“All of a sudden one is (…)”)

Cro. - Čuj, ti mali glupane! - zaurlao je. Odjednom si tako bezobrazan! Samo se pazi, govnaru mali! (“Listen, you little jerk!, he shouted. All of a sudden [you] are2nd sg so cheecky.”; both texts are a direct address).

The same is true of the 3rd and the 1st person plural and none of that is accounted for in the dictionaries, not even with usage examples.

  • the possibility for man to refer to the speaker is not mentioned in the dictionaries either, while it is acknowledged in Swedish reference books and the category ‘1st person singular’ appears as the chosen translation solution 19 times in the corpus. In other words, no element in the dictionary articles provides an explanation for sentences such as the following:

Sw. - Får man vara med? (“May one participate/join in?”)

Cro. - Mogu li igrati s vama? (“May1st sg [I] play with you?”)

  • the noun čovjek is rightly quoted as a lexicographic equivalent but this is not always done in a way which would make it obvious that it is its generalizing use that corresponds to man;

  • the pronoun netko is one of the two most common indefinite pronouns among the translation equivalents of man. The other one is, however, not the pronoun quoted in the dictionaries (neki, only one occurrence in the corpus) but nitko (“nobody”) with seven occurrences.

With regard to the second aim of this study—to check whether the compiled translation corpus can offer any insights into other phenomena, linguistic, contrastive linguistic or those that might be of interest for translation studies—several observations can be made. As has already been stated, multiple interpretations of a linguistic unit, in this case man, contained in a translation corpus expose the unit’s various meanings and uses. In the light of the results obtained in this study, it would seem interesting to check how often the ‘indefinite pronoun’ man actually has a known referent and whether that circumstance should be made more prominent in its description. Such a usage prompts translations categorized here as ‘other nouns’ or those grammatical persons that normally do not imply an impersonal meaning, but it is also manifest in many cases where modulation has been applied as well as in a number of translations involving the three grammatical persons that may but need not be used impersonally. It would seem that the overall share of such translation solutions is not negligible. On thother hand, one might easily come to an erroneous conclusion if the translation equivalents were taken to necessarily be truly identical to the source unit in focus. The impression obtained through this study suggests that Croatian equivalents of man often do not preserve its distancing function and that they resolve the (deliberate) ambiguity with regard to its referent. Any conclusions about the use of man would therefore have to be substantiated in a monolingual corpus or in another way. The comments on the interpretation of the referent and the equivalence established are illustrated by the following:

Sw. Och bjuder man Pia måste man bjuda Eva-Lena, nästan skriker Juha.

(“And if one invites Pia, one must invite Eva-Lena, Juha almost shouts.”)

Cro. A ako pozovemo Piju, onda moramo pozvati i Evu-Lenu – Juha sad gotovo viče (…)

(“And if [we] invite1st pl Pia, then [we] must1st pl also invite Eva-Lena – Juha is almost shouting now”, “we” refers to Juha and his friend with whom he is planning a party, no generalized interpretation is possible)

Sw. - Så ska man få det kastat i ansiktet också, sa modern. (“So one shall get it/that thrown in the face too, said the mother.”)

Cro. “Sad me i za to kriviš“, rekla je majka. („Now you blame me for that too“).

Further research could also try to establish whether the disambiguating translation equivalents in the corpus may be taken as resulting from the translator’s choice only and thus indicating that translators find it more important to produce a referentially unambiguous translation than to preserve the generalizing, distancing or a similar quality.

7 Conclusion

In line with the view of translation corpora as a valuable research tool in both Contrastive Linguistics and Translation Studies, the study presented here identifies some potential uses of a translation corpus of the do-it-yourself kind, such as is still necessary for many language pairs. The Swedish-Croatian translation corpus has been analysed in order to get a more comprehensive understanding of the various ways in which the Swedish pronoun man can be rendered in Croatian and to assess the standard lexicographic presentations of the pronoun in bilingual dictionaries in that light. The qualitative analysis required towards that goal has indicated that the corpus could be fruitfully used to pursue other research topics, as different as e.g., relating individual Croatian equivalents to a particular function of man on the one hand and assessing translators’ preferences or the nature of translation equivalence on the other.