The Routledge Handbook of Translation and Pragmatics

Contrastive approaches to pragmatics and translation

Svenja Kranich

Introduction

Contrastive approaches to pragmatics and translation take as starting point the study of pragmatic contrasts between the source language and the target language conventions, and on this basis, they identify the way translators handle these contrasts. The focus of contrastive pragmatic approaches to translation studies is thus to find out how translators handle the respective two “linguacultures”¹ divergent ways of “doing things” in texts. While general contrastive approaches to translation often focus on contrasts of the two language systems (e.g. the differences in the structure of word fields, the differences in word order etc.), contrastive pragmatic approaches are interested in differences in pragmatic conventions (e.g. the degree of subjectivity or directness expected in a text).

The most well-known representative of this field is Juliane House (e.g. 1997, 2007, 2015), whose wide-ranging studies on English–German pragmatic contrasts and translations have led to the formulation of five dimensions of communicative contrasts:

1 indirectness – directness

2 person-orientation – content-orientation

3 addressee-orientation – self-orientation

4 implicitness – explicitness

5 verbal routines – ad-hoc-formulation (with the first member of the pairs being typical of English, and the second member being typical of German discourse).²

According to her findings, one of the most notable differences between English and German pragmatics lies in a greater focus on the interpersonal domain of language characteristic of English discourse, be it spoken or written, compared to German discourse, which concentrates more on the ideational function, i.e., rather on content than on the interaction with the addressee. Covert translations, i.e., translations aiming at communicative, rather than formal equivalence of texts, tend to make adaptations to target language norms to some extent, but at the same time, they carry over pragmatic features of the source language text, thereby introducing variation into the genre (cf. House, 1997, 2015).

As House, as a main representative of this approach, focuses on English–German contrasts and their impact on translations, this chapter will also exhibit a focus on English–German pragmatic contrasts, because of the large body of work produced by House herself or inspired by her work (cf. e.g. House, 1997, 2007; Baumgarten, 2007, 2008; Böttger, 2007; Baumgarten & Özçetin, 2008; Becher et al., 2009; Becher, 2009; Kranich, 2011, 2016, to name but a few), while work on other language pairs following a clear contrastive pragmatic approach is more scarce (but cf. e.g. Böttger & Bührig, 2007; Amouzadeh & House, 2010; Kranich & Zhao, 2016). What unites these studies is the belief that in order to interpret target texts and to evaluate translations appropriately, it is necessary to have a solid understanding of the pragmatic contrasts that exist in a given genre between texts from the source language community and from the target language community.

Pragmatics is generally seen as the area of linguistics that deals with “people’s intended meanings, their assumptions, their purposes and goals, and the kinds of actions (for example, requests) that they are performing when they speak” (Yule, 1996: 4). Clearly, these issues are not limited to what people do when they speak. Writers also have intended meanings, assumptions, purposes and goals, and strive to perform certain actions via their communicative activity (e.g. to inform, to convince etc.). The way they do this can be assumed to differ, on occasion quite drastically, between different cultures, as can the way they take into account their addressee (i.e., the reader of the text). In written communication, there obviously is no direct interaction between people, but authors build into their text a simulated interaction between a “writer-in-the-text” and a “reader-in-the-text” (cf. Thompson & Thetela, 1995), and the way they do this shows interesting contrasts between different cultures. To be aware of these contrasts in communicative conventions is, in turn, of utmost importance for successful translation in most translation contexts, as these contrasts will often make adaptations necessary to the target audience’s expectations regarding genre conventions.

The remainder of this chapter is organised as follows: section 1 presents some important areas dealt with by contrastive pragmatics approaches to translation and summarises some of the most important findings. Section 2 delimits the field of contrastive pragmatic approaches to translation from other common approaches in translation studies. Section 3 offers an overview of data resources, common methods and methodological challenges. The final section highlights several potentially fruitful avenues for further research.

1 Overview of key areas in the investigation of pragmatic contrasts and translation

Theoretically, there are almost endless ways in which pragmatic conventions of two linguacultures may differ from each other in relation to certain aspects, and therefore one could envisage numerous potential areas where the investigation of pragmatic contrasts and their impact on translation could be of interest. If one takes a closer look at the state of the art in the field, however, one will notice that there are certain more global or “macro-pragmatic” contrasts, pertaining to the whole make-up of texts and characterising a genre or even text production strategies in general in a culture, which have been brought to light in various contrastive pragmatic and translation studies. In this section, I have chosen three of these “macro-pragmatic” contrasts and will summarise some interesting and representative findings that have been brought forth concerning them.

1.1 Contrasts in degrees of subjectivity and addressee-orientation

Many studies have highlighted differences between linguacultures in the degrees of subjectivity and addressee-orientation expected in texts from a given genre. Some of the examples discussed in the previous sections can be subsumed under this category, such as the investigation of the use of first person pronouns. Baumgarten’s (2008) findings on first person pronoun use in popular scientific writing show that American English texts tend to make more frequent use of the first person plural pronoun we than German texts of the same genre use wir, especially of reader-inclusive uses of the pronoun. The contrast between English and German non-translated popular scientific magazine articles thus resides in the greater subjectivity and addressee-orientation of the English conventions (confirming the general tendencies described in House, 1996, 1997). English–German translations clearly show an influence of the English conventions (cf. Baumgarten, 2008).

Similar findings can be seen in Amouzadeh and House’s (2010) results on first person pronouns in English and Persian academic texts. First person pronoun use in non-translated Persian texts and even in spoken discourse is uncommon, as excessive self-reference is perceived as self-centered behaviour by Persian speakers (Amouzadeh & House, 2010: 68). In English–Persian translations, however, first person pronoun use is quite frequently found: first person plural pronouns thus occur between ten times (in earlier translations) and four times (in recent translations) as often as in non-translated academic writing (cf. Amouzadeh & House, 2010: 70). Similar to the English–German translations investigated by Baumgarten (2008), there is evidence of English text-pragmatic norms shining through in the target texts.

Findings on epistemic modal expressions may also be linked to the more global contrast of subjectivity/addressee-orientation on the English side vs. content-orientation on the German side. Epistemic modal markers refer to the degree of likelihood that is attributed to the proposition’s truth – most often attributed by the speaker, so they are inherently typically subjective (cf. Palmer, 2001). Where epistemic modal markers are used as a resource to open up the discursive space and to leave the addressee room for their own opinion (cf. White, 2003; White & Sano, 2006), they also represent a means of creating more addressee-oriented texts. The fact, then, that epistemic modal markers are more typical of English than of German texts (cf. Kranich, 2011, 2016) shows the greater English tendency towards subjectivity and addressee-orientation; while the fact that the English–German translations apply a mix of adoption and adaptation strategies allows us to conclude that the translations introduce more variation into the German text norms, representing texts in German that to some extent follow Anglophone conventions.

The translation of tourism brochures also supports the view that genre norms vary across cultures with respect to subjectivity and addressee-orientation and that translators tend to apply a mixture of adaptation and adoption (shining through). Thus, Mason (2004) shows that French and Spanish tourism brochures are rather characterised by a more content-oriented, impersonal style, whereas English tourism brochures tend to prefer a more personal, more addressee-oriented style. Both shining through and adaptations to target language preferences are found in the translations (Mason, 2004: 165–169).

1.2 Contrasts in explicitness/implicitness of language use

Linguacultures furthermore differ with respect to the amount of information stated explicitly in discourse and the amount of information addressees have to infer from context and world knowledge. As House (1996, 1997, 2015) has shown, German rather tends towards explicit information whereas English rather tends towards the implicit site of the spectrum. She relates this to Hofstede’s (1980) cultural contrast of “uncertainty avoidance” (a concept which has attracted some criticism recently, cf. e.g. McSweeney [2002], but still seems like a useful tool) (cf. House, 2008b). According to Hofstede (1980), cultures tend to differ in the importance they accord to the avoidance of uncertainty. Hofstede (1980) sees German culture as ranking rather high on the dimension of uncertainty-avoidance, which leads to a tendency to state matters more directly and to a greater prominence of rules, detailed instructions, timetables, and precisely defined topics in everyday life (cf. House, 2008b: 571). This contrast between English and German is reflected in the English–German translation practice of producing more explicit verbalisations than present in the source text, as for example in the translation of children’s literature, where the translators add information that readers of the source text would have to infer, and in the translation of film titles from English to German, where the German translation, unlike the English source title, tends to give away the main plot in the title (cf. House, 1997, 2004).

Explicitation (i.e., making something explicit in the target text that is only implicit in the source text) has also been claimed to be a universal of translation, i.e., a process that occurs due to the translation process, regardless of the translation direction (cf. Baker, 1996; more information on “translation universals” can be found in section 2). However, in recent studies, the specific language pair as well as the translation direction have been shown to play an important role. Thus, Behrens’ (2005) study of the Norwegian connector dermed “thus” in translations showed that the Norwegian–German translations almost always translate the connector, while Norwegian–English translations left around every fifth occurrence untranslated, thereby reducing explicitness in cohesive marking and adapting the text to the greater preference for implicitness in English compared to Norwegian. Becher’s (2009, 2011) findings on English–German and German–English translations show a clear impact of the translation direction: the translations from English into German contain a significantly higher number of explicitations than the translations from German to English, while implicitations are more common in translations from German to English. In both translation directions, however, explicitation is a much more common phenomenon than implicitation, which allows one to assume that explicitation, while not a universal, is a robust tendency in translation – albeit one that is clearly influenced by pragmatic contrasts between source and target language.

1.3 Contrasts in the use of repetition and variation

In this section, we move away from central concerns of pragmatics (interaction, implication) to focus on an issue that might be argued to be more pertinent to the domain of stylistics. However, it makes sense to treat this issue here, as many studies interested in the contrasts described previously have also studied the issue of repetition and variation. The reason is that the two areas (contrastive stylistics and contrastive pragmatics) are sometimes rather closely related in actual discourse: if your main aim is to describe the facts as precisely as possible (because of your linguaculture’s norms of uncertainty avoidance, content-orientation and explicitness), then it is reasonable to assume that you will vary your lexical choices, always searching for the most fitting term. If, on the other hand, your language norms tend more towards addressee-orientation and are not so much concerned with the highest possible degree of explicitness, you might favour repetition, as it can serve a rhetoric purpose and thereby help you to interact with the addressee (cf. e.g. Böttger and Bührig’s [2003, 2007] findings on translation of the repetitive use of we believe into German and French).

Findings by Hansen-Schirra and colleagues (2007) and Neumann (2013) show that the German texts in the CroCo corpus (described in section 3) are characterised by a higher type-token-ratio, i.e., by more variation, than the English texts in the corpus. In this respect, adaptation to target language conventions typically occurs, as is also evident from results by Kranich (2016: 87f.): English–German translations show considerably more variation than the English source texts – sometimes even more than comparable non-translated texts in German. The reason is most likely that the ideal of variation, the stylistic requirement of avoiding repetition is so strongly engrained in the German concept of what constitutes a good writing style that over-adaptation occurs in translation. This is different in other language pairs: Musacchio (2005) discusses a similar contrast between English and Italian, with the Italian writing style tending towards the avoidance of repetition. While adaptation strategies are used, shining through also occurs in her data (cf. Musacchio, 2005).

2 Delimiting the field

2.1 Contrastive approaches to translation vs. other approaches to translation

What is specific about the contrastive approach to translation? Translation studies, as the studies referred to in the previous section, have shown time and again that translation typically exhibits an interplay between adoption (“shining through”) and adaptation (cf. e.g. Toury, 1995; Teich, 2003). That means, to some extent translators carry over features they find in the source text, even though these features might not be standard usage in comparable, non-translated target language texts (adoption), and to some extent they make changes to the text in order to adapt it to target language usage norms (adaptation). Translators rely on knowledge gained in their translator training, on intuitive knowledge from their bilingual and bicultural competence, as well as on reference works when making these adaptations. As researchers, we wish to find out whether these changes are appropriate and truly reflect differences in e.g. genre expectations between source and target language readers, or whether they are not, because, for instance, the translator’s intuition about stylistic contrasts was inaccurate. In order to do so, knowledge about the relevant contrasts between source and target language usage in non-translated texts is necessary – which is what a contrastive approach aims to establish.

The notion that translations show an influence of the contrasts between source and target language norms (i.e., that they contain adaptations and adoptions) is a basic assumption necessary to consider contrastive approaches fruitful. If, by contrast, one considers translations “a third code”, as Frawley (1984: 168) puts it, one would assume that contrasts between source and target language conventions are less relevant and that instead the reason translations differ from non-translated comparable texts in the target language has to do with the translation process itself, leading to universal, non-language-pair-dependent properties of translated texts, such as explicitation, simplification, normalisation, etc. (cf. Baker, 1996: 180–184; Laviosa-Braithwaite, 1998). This view is, however, typically not regarded as adequate in contrastive approaches to translation (cf. e.g. House, 2008a; Becher, 2010a, b), while in other works, it is simply not seen as the main reason for the observable differences between translated and non-translated text (e.g. Kranich, 2016: 8–10). Instead, contrastive approaches focus on the differences between source and target language conventions and take them as crucial for determining the reasons why translated texts might contain different features from comparable non-translated texts. As could be seen from the studies on explicitation summarised above, it is clear that the source language and target language and their respective conventions are important factors in determining to what extent explicitation and its reverse, implicitation, occur.

In my own understanding of translation studies, contrastive insights constitute a necessary step in the investigation of translation. In order to evaluate shifts in translation appropriately, one needs to know whether they constitute “shining through” (Teich, 2003) of source language features, or rather adaptation to different target language norms, and only if neither of these two explanations seems plausible, does it seem permissible (in view of the most economic explanation, also known as Occam’s razor) to understand them as translation-inherent shifts, i.e., as having been caused by the translation process itself (cf. also Becher, 2010a).

To illustrate this, take a study of the supposed translation universal “explicitation” which investigated the use of the complementiser that after reporting verbs (e.g. He said he loved her vs. He said that he loved her) (Olohan & Baker, 2000). Arguably, the latter variant is more explicit. Olohan and Baker (2000) investigated this phenomenon and found that the use of that in these contexts is notably more common in translated English (based on the Translational English Corpus [TEC]) than in non-translated English (based on a comparable sample from the British National Corpus [BNC]). They conclude that this finding provides evidence for the accuracy of the explicitation hypothesis. However, they do not consider at all to what extent source language norms could have influenced the results. The TEC contains English translations from a wide variety of source languages. Among them are languages like Spanish and Portuguese where the connective que is not, like English that, optional after the verbs that mean “say”, but is required by the grammar (cf. Saldanha, 2008: 22). So the TEC, which only makes it possible to consider target texts without checking the corresponding passages in the source texts, does not allow one to see clearly what is happening in the translations: is the over-occurrence of that after reporting verbs in the translated texts due to translation-inherent explicitation and independent of source and target language norms? Or is it rather a case of shining through (i.e., translators see a connective in the source text and are thus likely to use connective that in the target text)? One needs to know the contrasts between source and target language and study the precise translation relations in order to find out (cf. also Becher, 2010, 2011).

2.2 Contrastive pragmatic approaches vs. other contrastive approaches to translation

As stated above, the focus of contrastive pragmatic approaches to translation studies is on pragmatic issues, i.e., to find out how translators handle the respective two linguacultures’ divergent ways of “doing things” in texts, and not on contrasts arising from differences between the two language systems.

To illustrate the fact that this may sometimes not be completely straightforward, take the example of the use of personal pronouns. Contrasts in frequencies of personal pronouns between texts from two different languages can either be a consequence of systemic contrasts or of pragmatic contrasts. The most relevant systemic contrast in the use of personal pronouns lies in the presence or absence of pro-drop rules. If language A is a pro-drop language, e.g. one that allows pronouns to be deleted in subject position when the referent is retrievable from the context, and language B is a language that does not allow pro-drop, then we will necessarily see differences in pronoun frequencies in texts from language A and from language B. Also, we will see that translations contain fewer or more pronouns, depending on translation direction: when translating from B (non-pro-drop) to A (pro-drop), pronouns will tend to be omitted in certain contexts, where they do not seem natural. When translating from A to B, they obligatorily need to be added for grammatical reasons. An example is provided in (1).

1 SpO: El médico me dijo que debo hacer más ejercicio.

ET: The doctor told me that I must exercise more.

(linguee.es)³

The Spanish original only uses the first person singular form of the modal verb deber, i.e., debo, without a personal pronoun. In English, one cannot render this sentence as *The doctor told me that must exercise more, because of the differences in the grammatical systems.

However, different frequencies of personal pronoun use may also point to pragmatic, rather than systemic contrasts. For example, if a corpus of German translations from English contains more first person pronouns than German comparable texts, this cannot be explained by grammatical contrasts between the two languages, as both require personal pronouns and do not allow pro-drop. Rather, pragmatic contrasts are likely to play a role. In this particular case, the relevant contrast is between a greater person-orientation of English discourse, whereas German discourse tends to be more content-oriented (cf. House e.g. 1996, 2015). Hence, first person pronouns tend to occur more often in English texts than in German texts, and the translators to some extent adopt the English norms, leading to English–German translations with higher frequencies of first person pronouns than those found in comparable non-translated German texts (cf. Baumgarten, 2008). The following example shows a translation that does not imitate the pronominal use of the source text: where the source text contains two instances of we, the target text has none of wir.

2 EO: We know the short-term side effects from experience with HIV-infected patients, but we know almost nothing about the long-term consequences of using HIV-fighting drugs in people who may not in fact harbor the virus.

GT: Aus den Erfahrungen mit HIV-infizierten Patienten sind die unmittelbaren Nebenwirkungen der verwendeten Anti-HIV-Mittel bekannt (Bild 1), doch weiß man leider so gut wie nichts über die langfristigen Folgen bei Menschen, die das Virus möglicherweise gar nicht beherbergen. (POP 1999–2002)

Back-translation: The immediate side effects of the anti-HIV drug used are known from experience with HIV-infected patients (image 1), but unfortunately, one knows almost nothing about the long-term effects in human beings who possibly do not harbor the virus at all.

In the English source text, the author places himself in the group of doctors concerned by the development described and thus creates a rather personal form of communication. The text thus exhibits a more interactional profile. The German translation is quite different from this. Here, the author is no longer part of a group designated by a first person plural pronoun. Instead, the situation is described impersonally, from an “objective” perspective. Two different translation strategies are used here to this end: the first occurrence of we is avoided through the use of a passive construction, and the second occurrence is translated by the impersonal pronoun man (“one”). These translation choices clearly suggest that the translator of this passage has made a conscious effort to de-personalize the text for his German audience, supporting the notion of robust tendencies in communicative preferences between English (tendency towards more person-oriented discourse) and German (tendency towards more content-oriented discourse) (cf. also Baumgarten et al., 2004: 91f., 94f.; Probst, 2009: 114–173; Kranich, 2016: 24–27).

If we now wished to investigate to what extent the person-orientation of Spanish and English discourse differs from one another, a mere quantitative analysis of the occurrence of pronouns would not be of great help, since we would not know whether a lower occurrence of personal pronouns in Spanish is only due to the pro-drop feature of the language, or whether it additionally reflects a difference in pragmatic conventions. This could only be achieved through careful, manual analyses of relevant instances.

This example is simply meant to illustrate that the systemic differences between source and target language must nevertheless be taken into account when studying pragmatic contrasts, in order to formulate adequate research questions, choose appropriate corpus searches, and be aware of potential challenges in interpreting the findings. We will look at further potential methodological challenges in the context of the following section.

3 Data, methods and approaches in contrastive pragmatics and translation

3.1 Which data to use? Corpora in contrastive pragmatics and translation studies

Present-day contrastive translation studies is unimaginable without corpora. Since the rise of corpus linguistics in the 1980s, the field has boomed and expanded, and more and more corpora have become available. Large monolingual corpora that can serve as reference corpora for contrastive and translation studies, such as the COCA (Corpus of Contemporary American English, ~ 560 million words) (cf. https://corpus.byu.edu/coca/ and Davies [2012] for a description), have become available, as well as parallel corpora containing aligned versions of source texts and target texts, such as the CroCo Corpus (cf. Hansen-Schirra et al. (eds.) (2012) for a description), which contains English and German translated and non-translated texts.

A corpus make-up like CroCo’s can be considered ideal, containing source texts in language A, their translations into language B, source texts in language B, their translations into language A, and additionally reference corpora for both languages. The source texts and their translations come from a broad spectrum of text types, which have been selected for their translational relevance (i.e., text types where translation is common, such as manuals, letters to shareholders, fiction, popular science and travel brochures). This allows a multitude of research designs. Contrastive studies can be carried out, comparing the source texts from language A with the source texts from language B to see what norms are found in non-translated texts in the two languages. Translation relation studies can also be carried out, i.e., one can search for a term and all its translations in the data. Since the corpus is aligned, one can retrieve all hits in the source language together with their translations, or one can search for specific expressions in the target texts and see which source text expressions triggered them (cf. e.g. Hansen-Schirra, 2011; Neumann, 2013 for CroCo-based studies of interest to researchers in contrastive pragmatics and translation).

3.2 Semasiological approaches to contrastive pragmatics and translation studies

A classic corpus-based approach to a research question is semasiological (taking linguistic forms as its starting point). The researcher decides which linguistic forms or specific constructions would be interesting to investigate and conducts a computerised search for these in the corpus of his/her choice. To take an example, a researcher might be interested in a corpus-based investigation of the use of the first person plural pronoun in English and German (we, wir), since, as we noted above, these pronouns are sensitive to pragmatic contrasts based on a greater preference for person-orientation in the English texts and a greater preference for content-orientation in the German texts. In order to tackle a research question like “Are there differences between the use of we and wir in English and German texts, and how are they handled by translators” in a corpus-based manner, one could start by establishing contrasts between English and German use by using the non-translated source texts. Then one could check how many of the occurrences of we are actually translated by German wir and how often by other constructions, and by which means German source text use of wir is rendered in English in the opposite translation direction. Furthermore, the opposite perspective could be investigated, taking a look at occurrences of we in English target texts to see how many of them are triggered by wir in the corresponding source text, and doing the same for the German target texts to check how many of the wir instances one finds there are results of translating we. With this approach, one can reach a complete overview of contrasts between English and German usage of the item, as well as of translation strategies (cf. Baumgarten’s 2008 investigation of we/wir in popular science writing).

The motivation for looking at we/wir in the previous example is rather straightforward, I think, from what has been explained in the preceding section. Often, however, it can be difficult to decide what exactly to retrieve in a corpus-linguistic approach. If a broader communicative contrast is the focus of investigation, such as the dimensions of contrast proposed by House (1996, 2015: 88–92), e.g. directness vs. indirectness, then careful decisions will need to be made with regard to which linguistic expressions would be adequate to single out in order to determine how such contrasts can be quantitatively verified. That means, first analysing previous studies and/or samples of the data to gain a solid idea of which linguistic items and constructions are clearly related to the pragmatic contrasts under investigation. Thus, a first step would be to identify which constructions are used to be more direct or more indirect, leading perhaps to a focus on directness/indirectness in requests, a topic which has been rather well-researched from a contrastive perspective (e.g. by Blum-Kulka & Olshtain, 1984; Blum-Kulka, 1987; Blum-Kulka et al., 1989). These classic studies provide a scale of directness-indirectness as well as examples of the different strategies, e.g. the most direct strategy is termed “mood-derivable” and is characterised by the use of an imperative verb form (e.g. Leave me alone!) (cf. Blum-Kulka & Olshtain, 1984: 202). Conventionalised indirect strategies, on the other hand, are characterised by the use of modal auxiliaries (e.g. Could you clear up the kitchen, please? or Would you mind moving your car, please?, cf. Blum-Kulka & Olshtain, 1984: 202). If a researcher wished to conduct a corpus search to see how often direct requests and how often conventionalised indirect requests are used, then, a search can be conducted for verbs in the imperative form and for modal verbs as a category, if the corpus is tagged for parts of speech. If the researcher has an untagged corpus, elements such as exclamation marks could be searched as the typical punctuation after an imperative, and the individual forms of modal auxiliaries, such as can, could, would and so on, to retrieve the conventional indirect requests. The frequencies of more direct and more indirect requests in texts of language A and language B can then be established. Based on the contrasts thus established, checks can be made as to whether translations from language A to language B (and vice versa) follow source language norms or whether they adapt the directness level to target culture conventions.

What the researcher chooses to count may have considerable impact on the results, as the following studies of hedging (i.e., markers that serve to express propositions with less than full force, often as a means of mitigation) illustrate. Contrastive studies of the phenomenon of hedging in academic texts have yielded completely opposite results: on the one hand, Markkanen and Schröder (1989) and Clyne (1991) find hedging to be more typical of German than of English academic writing. On the other hand, Kreutz and Harres’ (1997) results on a subset of the data used by Clyne disconfirm this: their close manual analysis leads them to conclude that “[t]he German texts show very few hedges over all” (Kreutz & Harres, 1997: 189). The difference in results is based on their definitions of hedging: Markkanen and Schröder (1989) and Clyne (1991) included a broader spectrum of linguistic markers, notably different types of impersonalising constructions (such as agentless passives), in their concept of hedging, whereas Kreutz and Harres (1997) had a more restricted definition: they only counted as hedging those markers that clearly lead to a weakening of the proposition’s force, such as epistemic modal markers. The approach by Kreutz and Harres (1997) seems more fruitful to me in terms of gaining insights into the way a particular pragmatic/cultural contrast may be realised in different textual strategies. Thus, Hofstede’s (1980) concept of “uncertainty avoidance” or House’s (1997) contrast of greater content- vs. greater addressee-orientation would be reflected in different amounts of hedging in the way the category is delimited in Kreutz and Harres (1997). The inclusion of impersonal constructions produces a less clear-cut picture, because their use rather relates to contrasts in the expected degree of person- vs. content-orientation. This example serves to highlight that the choice of linguistic markers to be analysed in a given study should be cautiously checked for its adequacy with respect to the general pragmatic contrast under investigation.

3.3 Onomasiological approaches to contrastive pragmatics and translation studies

A further possibility – one that is often neglected in corpus-based approaches – is to follow an onomasiological approach, i.e., a research design that takes a particular meaning or a particular function of language as its starting point. An example can be found in Kranich’s (2011, 2016: 103–163) study of epistemic modal markers. These studies examine the use of epistemic markers in English and German non-translated texts and in English–German translations, which was assumed to differ based on House’s (1996) dimension of directness vs. indirectness and content- vs. addressee-orientation. If discourse is generally more indirect and more addressee-oriented, as in English when compared to German, the use of epistemic modal expressions as hedges can be assumed to be more common. Compare There might perhaps be a problem to There is a problem: both might be uttered in the same context, but the first one states the bad news in a more indirect and a more face-saving, hence more addressee-friendly way.

To find out whether English texts and German popular scientific texts differ systematically from each other in this respect, it would seem most appropriate to attempt to capture all epistemic modal expressions in the corpus, i.e., all expressions that have the function to modify the degree of certainty attributed to the likelihood of the proposition’s truth. A variety of linguistic expressions can fulfil this function: modal verbs (such as may, might), modal adjectives (such as it is possible) and adverbs (e.g. probably), as well as lexical verbs (e.g. it seems) and longer lexical constructions (e.g. es ist noch nicht abzusehen “it is not yet foreseeable”). A semasiological approach, searching for a list of pre-determined items, would not be able to capture all of these elements and would thus provide an incomplete picture of the field of epistemic modal marking. This lack of completeness of the findings would be particularly dangerous in the context of contrastive studies and could lead to incorrect conclusions: it is easily imaginable that in texts from one language, the easily searchable expressions from a closed class (such as modal verbs) dominate in the field of epistemic modal marking, and in texts from the other language, more varied, lexical expressions are often used. Merely conducting a computerised corpus search of the closed class elements, would thus lead to a skewed picture. In our example, this is actually the case: in English popular scientific articles, it is in particular modal verbs that are used for the expression of epistemic modal meaning, whereas in German, there is a greater variety of elements (including modal adverbs, but also more individual lexical constructions) that are used for this function. Thus, if only the closed class elements are searched with a focus on the modal verbs, the result would make it seem as if English used epistemic modal markers with an enormously higher frequency than German. Kranich’s (2016) investigation did indeed show that epistemic modal markers are more common in the English than in the German texts, but this tendency would have been over-represented if the approach had been to simply search for the modal auxiliaries in both languages. Interestingly, the translations showed shining through in terms of frequency, that is, English–German translations contained more epistemic modal markers than German non-translated texts, approaching almost the frequency of the English source texts. The choice of linguistic category to express epistemic modal marking, by contrast, exhibited adaptation to target language norms, by using fewer modal verbs than the source texts, and more of the modal adverbs and lexical expressions common in German texts of the genre.

3.4 The combination of insights from semasiological and onomasiological approaches

Onomasiological approaches have the benefit of allowing us to see the complete picture. They have the downside that they necessitate a manual, close-reading approach to the text in order to retrieve all relevant instances. Thus, large amounts of data cannot be handled in this way. A good solution is to combine onomasiological and semasiological approaches. For the onomasiological approach, a smaller sample of the whole corpus data can be used. Going through the sample manually, the researcher can gain an overview of the linguistic markers and constructions used for the expression of a particular function. This can be followed by a semasiological approach, consisting of computerised corpus searches in larger databases of the most typical representatives of a particular functional field, making it possible to retrieve sufficient amounts of data to allow meaningful quantitative analyses.

To come back to the study of epistemic modal markers: while the proportion of modal expressions in English and German differed, the study made it clear that modal verbs were a frequently chosen category to express epistemic modal meaning. A corpus search of all modal verbs in the English source texts in the parallel, aligned and tagged Popular Science Corpus (for a description, cf. Kranich, 2016: 18f.) allowed an analysis of the use of modal verbs in the English source texts and of their translations into German. The results of this study showed that English–German translators most often translated the English modal verb by a German modal element expressing the same degree of certainty, but not infrequently chose a different word class (e.g. may > vielleicht “maybe”). This means that in terms of the distribution of modal elements across word classes, adaptation strategies were common. Concerning the modal strengths of the elements (i.e., whether they express mere possibility or high probability), shining through was common, as translators most commonly kept constant the modal strength they found in the source text. This leads to the picture that was also brought to light by the onomasiological study described above, namely that English–German translations differ from German non-translated texts in terms of the distribution of modal elements across high and low modal strength: where German non-translated texts prefer elements of high modal strength (such as wahrscheinlich “probably”), English–German translations resemble the English originals in commonly using elements of low modal strength (such as könnte “could”, vielleicht “maybe”) (cf. Kranich, 2016: 145–148). However, the semasiological corpus study showed that at the same time, translators do apply a “cultural filter” (cf. House, 1997), i.e., make adaptations to the texts that take into account pragmatic contrasts between the two languages: in 12 per cent of cases, English modal verbs are not rendered by any modal element in the German translations, i.e., the translations contain a plain indicative; in a further 5 per cent to 13 per cent (depending on the time-frame of the translations), a modal expression that conveys higher modal strength is chosen (cf. Kranich, 2016: 148–153). These translation choices indicate that translators do make adaptations to the overall frequency of epistemic modal expressions as well as to modal strength that go in the appropriate direction. Based on the study using the onomasiological approach, it is clear, however, that they do not go sufficiently far to produce target texts that resemble non-translated originals in the target language, which are characterised by even fewer modal expressions overall and fewer expressions of low modal strength. Had only an onomasiological study been conducted, attempts at adaptation brought to light in the semasiological, translation-relation study would likely remain obscured. Had only the translation relation been considered, it is likely that the effects of the adaptation strategies would have been overrated and the massive shining through that takes place nevertheless would have been missed. This example may therefore serve to highlight the benefits of combining both approaches. Onomasiological approaches can also serve as a good starting point for an investigation of texts for which no appropriate parallel corpus exists. Manually going through source texts, target texts and comparable non-translated texts from the target language allows the researcher to get a good overview of linguistic markers characteristically employed for a certain function and to determine which of these markers would be fruitful to count and classify to produce both qualitative and quantitative findings.

Concluding remarks

The findings reported on in the preceding sections show the fruitful nature of contrastive pragmatic approaches to translation. Where pragmatic contrasts exist between source and target language, the investigation of translation practice promises to bring to light interesting findings of either adaptation or shining through, most often of a combination of both.

As indicated by the brief overview of key research findings discussed above, a lot of the research in this paradigm has focused on pragmatic contrasts between English and German and the way these are handled in English–German and, to a lesser extent, German–English translation, which can be seen as a consequence of the influential nature of House’s work in the field. A greater variety of language pairs could be investigated using similar principles. Comparisons between different language pairs could bring forth more in-depth insights on the role of different factors in determining whether more adaptation or more shining through can be found in the translations. Factors of interest here would for instance be the prestige of the source language, the degree and length of contact between source and target language, the degree of standardisation of the target language and the typological proximity between source and target language (cf. Kranich, 2014; Kranich & Zhao, 2016).

Some studies that have been cited also discuss to what extent the pragmatic innovations and variations in the translations gain acceptance in the target language community, triggering changes in genre conventions. Both Baumgarten (2008) and Amouzadeh and House (2010) find a certain increase of first person pronouns in more recent, non-translated texts in German and Persian respectively, presumably due to the influence of translations from English into these languages, where first person (plural) pronouns become common first. Kranich and colleagues (2012) discuss to what extent popular scientific texts in German have undergone changes due to impact of innovations first introduced by translations from English. Although studies on the impact of translations on target language norms are not numerous (cf. Kranich, 2014 for an overview of some exceptions), the phenomenon of translations containing pragmatic shining through and having an impact on the development of pragmatic norms in the target language is far from rare, especially in genres where translation is common and occurs from a source language that carries a certain prestige in the target culture (such as English in present-day Western societies). Further studies would seem very desirable, especially on translations from English into other languages, since the prestige of English makes such an influence especially likely.

Finally, research on the impact of shifts in the translations on reader attitudes has hardly begun. Previous studies have brought to light, as we have seen, the fact that the frequency of hedges differs between texts translated from English and non-translated texts in popular science writing in German (cf. Kranich, 2011, 2016: 127–153). But what does this mean for the reception of these texts? Research by Crismore and Vande Kopple (1997) has shown that the use of hedges has positive effects on the transmission of controversial ideas. Texts with hedges were shown to foster a more positive development in attitude towards potentially controversial ideas than un-hedged versions of the same text. Additionally, the authors of texts with many hedges were perceived as friendlier. The same study showed, however, that it can be detrimental to reader attitudes when the use of hedges is perceived as excessive: readers then doubt the competence and the credibility of the author. In the extreme, this could mean that German target text readers of texts translated from English perceive the authors as nice, but not very competent and reliable, because of an overuse of hedging. The effect of pragmatic contrasts and of the shining through of source language pragmatic norms in translated texts thus constitutes another promising avenue for further research.

Notes

1 The term linguaculture (originally coined by Friedrich [1989: 307]) highlights the connection between language use and cultural background. When discussing pragmatic contrasts, it is somewhat misleading to speak of differences between languages, as the differences often do not hold between use of the language everywhere, but in a particular speech community. Thus, most of the insights on English–German contrasts are actually contrasts between Anglo-American English and Standard German as used in Germany, and the same contrasts may not hold at all in the same way between, for instance, Indian English and Standard Austrian German. Rather, the pragmatic contrasts and contrasts in communicative styles hold between specific cultures and the way in which language is used in them within its social context. The term linguaculture is meant to underline this fact.

2 The existence of the contrast has been corroborated by numerous studies (cf. Kranich, 2016: 29–66 for an overview). The differences hold both between American English and German (as used in Germany) and between British English and German in Germany, and seem to also hold, based on one study by Grieve (2010), between Australian English and German. Other standard and non-standard varieties of English and German are still underinvestigated in contrastive pragmatics.

3 Examples provided in this entry come from the website Linguee, which can be searched for translations of particular lexical elements in a variety of language pairs, as well as from the corpora created in the Research Centre on Multilingualism (SFB 538, University of Hamburg, 1999–2011), the Popular Science Corpus (POP), the Mixed Business Corpus (MixB) and the Letters to Shareholders Corpus (LeSh) (described in more detail in Kranich, 2016: 17–21), whose creation was made possible through generous funding from the German Research Foundation (DFG). This support is hereby gratefully acknowledged.