Article 37

Introduction

Translation is rarely a transparent transference of propositional content from one language system to another. Even when a translator strives to reproduce the source text as closely as possible, there are systematic and regular ways in which the target text differs from the source – differences that cannot be easily attributed either to individual style or the specific nature of translation. One of the most well-documented of these divergences involves cohesion—the network of grammatical and lexical devices that provide coherence to a text, with conjunctions occupying an important location in this network (Halliday & Hasan, 1976, p. 4). Corpus-based research by several researchers also report greater density of overt conjunctive markers in many translated texts than their own source texts, or comparable native language (target text) texts (Blum-Kulka, 1986, p. 21), which new kinds of similar evidence found to support and additive endowment to expliciation hypothesis may be explained as acquires were among the most well-cited aspects.

The current paper approaches conjunction density as an empirical window into a more general theoretical issue: whether translationally cause any changes to the typical frequency of cohesive devices in a target language and if so, when. The answer to this question is important for at least three reasons. It directly affects the theoretical status of explicitation as a translation universal, which has been claimed, qualified and disputed in approximately equal measure ever since Blum-Kulka’s initial proposal (Baker 1996: 180 Becher 2010: 2)). Secondly, it has also a practical relevance regarding translation quality assessment and pedagogy because an unmotivated extension of connectives is viewed by target readers as the mark of a text being too much under the influence of its source even if each addition is locally justified (Øverås 1998: 559). Third, for a language pair (English and Arabic) in which the rhetorical conventions of the source language favour coordination over subordination, conjunction density is therefore an even more complex diagnostic to interpret: it is possible that additional connectives in the English TT result from universal translation effect or simple source-language interference or the combination of both (Othman & Al Qutob 2022, p. 3).

Here we do not present a novel empirical corpus study, but rather a critical theory-building synthesis of the literature on explicitation and cohesion/conjunctive relations in translation as it pertains specifically to what is known (and what remains properly unresolved) about the Arabic–English pair. The structure of the paper is as follows. In section 2, we introduce the theory starting from Blum-Kulka’s initial hypothesis all the way to Halliday and Hasan’s explanation of cohesion & conjunction and Klaudy’s well-known four-way typology for explicitation. This section (Section 3) focuses on conjunctions as a privileged marker of density and ways to measure that density. The Arabic–English pair, the focus of section 4, follows and contrastive-rhetoric findings on parataxis versus hypotaxis are reviewed. Finally, section 5 presents an analytical framework for its future testing on a corpus-based level. We cover mechanism in Section 6, implications for practice and training in Section 7, before concluding the paper in Section 8.

Theoretical Framework

2.1 The Explicitation Hypothesis

The explicitation hypothesis was established by Blum-Kulka (1986) in a comparison of cohesion and coherence shifts in a small corpus of literary translation. At the heart of Blum-Kulka’s claim is that the kind of interpretation a translator has to make of a source text typically results in a target text that is more redundant, i.e. it contains greater numbers of overtly marked cohesive ties than does the original. Most important, Blum-Kulka did not frame this as a hypothesis regarding the stylistic choices of individual translators but as a hypothesis about translation as a cognitive and communicative act: A translator is the most diligent reader that any text will ever have while also possessing in their mind an idea of a different set of readers to whom they must shape the text; they are therefore structurally predisposed to delineate links that could safely be presumed by the source author (Blum-Kulka, 1986)

The hypothesis was subsequently generalized far beyond the literary corpus on which it had been first tested in the decades that followed. Recently, however, the theory of explicitation as one unitary phenomenon has been reformulated by Klaudy (2008) and has more specifically and elaborately been presented in Klaudy and Károly (2005) through family of shifts that function at various levels of obligation. Among a small number of alleged translation universals — including simplification and normalisation — Baker (1996) placed explicitation, highlighting the potential for corpus methods to test such claims on a practical rather than anecdotal or impressionistic basis (Baker, 1996: 180). This corpus-based turn is what eventually rendered conjunction counting a tractable, replicable empirical method rather than interpretive judgement on its own.

2.2.Cohesion Theory and Conjunctive Relations

Blum-Kulka’s hypothesis has modern theoretical vocabulary, but it borrows heavily from Halliday and Hasan (1976), who remains the reference point for most later work on cohesive shift in translation. Reference, substitution or ellipsis conjunction and lexical cohesion (Halliday & Hasan 1976) are five types of cohesive ties which Halliday and Hasan identified. The other four marks are distinguished from conjunction in that they direct a reader on how to retrieve a referent, while conjunction directs a reader to the logical or rhetorical relation—additive, adversative, causal or temporal—that holds between one stretch of text and the next (Halliday & Hasan, 1976). Since conjunctive relations, even if not explicitly marked as such (with conjunctions or punctuation), can always be recovered inferentially, it is the choice between marking them overtly and leaving them implicit that makes explicitating and non-explicitating preferences embersastically visible for the translator.

This is actually why conjunctions, the 5th category of cohesion, have received disproportionate attention in explicitation studies. Reference and ellipsis are tightly constrained by each language’s grammar and allow the translator little in the way of options, while lexical cohesion — largely a matter of degree (since it is always at least near-synonymy) — can be reliably quantified. In contrast, conjunctions are a relatively restricted and finite class, they are often optional in that the underlying logical relation remains intact upon their dropping as there is no formal marker of this relationship, and logical relations can be compared directly between source-sourced propositions or followed across two languages since most logical relations if not specific lexical items expressing them tend to be cross-linguistic (Baker, 1996:183). This is why conjunctive density has effectively become a proxy for cohesive explicitness.

2.3 Typologies of Explicitation

Klaudy (1998 cited in Klaudy, 2009) distinguishes between obligatory, optional, pragmatic and translation-inherent explicitation in his typology. Structural conflicts, such as when one language is less tense-explicit than another (Baker & Aisha, 2009), require mandatory explicitation of time expressions in translation to ensure the target text is grammatical. The next two categories of conjunctions are cases where the same proposition can be expressed with either an explicit connective or left implicit, that is to say they are motivated strategically not grammatically (which bounds the definition in optional explicitation), with conjunction addition belonging to this category as most occurrences are stylistic pragmatic preferences in text-building. The cause of pragmatic explicitation is the differences in background knowledge that can be presumed from source and target readerships. The most theoretically contested of the four explicitation types, translation-inherent explicitation is assigned by Klaudy to the translation process itself, irrespective of a particular SL−TL pair (Klaudy, 2009).

According to Becher (2010), the category of translation-inherent explicitation forms the weakest link in the typology since a large number of identified explicitation instances can instead be attributed to source-language-specific or target-language-specific motivation rather than any process-inherent universal. This critique is immediately pertinent to the current topic, because it suggests that a higher density of conjunctions observed in a translated text alone cannot constitute evidence for a translation universal; however, unless simpler explanations grounded in both relative grammatical and rhetorical convention —in our case what could be called (contrary to common assumptions) great differences between Arabic (the source language, which is where the data originate from) and English (the target language)— are ruled out beforehand by the researcher(s), he/she/they need not proceed further with such an argument and must proceed much more cautiously than before —as we see as particularly difficult in Section 4 on Arabic & English translations.

3. Conjunctions and Cohesive Density

In this literature, “density” refers to the normalised frequency of conjunctive markers per unit of text (usually expressed as occurrences per 1K words) instead of a raw count since the latter is not comparable across texts with different lengths (Baker, 1996). Natural density of a text is operationalised in comparable-corpus designs as the density measured in a reference corpus of texts written natively in the language so far and balanced by genre, register, and period. In particular, the explicitation hypothesis posits (in the conjunction specific version) that for any translated text the normal baseline density of overt conjunctive markers will be greater than in a comparably cohesive non-translated – or those where every cohesive resources is taken into account (Blum-Kulka, 1986)

Marco (2018) recommends treating connectives not just as a frequency variable but, rather as an indication which need to be interpreted in conjunction with the taxonomic category of the relation signaled — because additive connectives like “and” are not cognitively or stylistically equivalent to causal or adversative connectives like “because” and “however.” Research that pools all conjunction types into a single density figure risks obscuring genre- and register-specific patterns: legal and institutional texts, for example, already have an excess causal and conditional connectives before translation struggles (Chaalal, 2018).

Another layer, derived from Murtisari (2016, p. 68), is that explicitness itself a graded notion rather than simply on-off: incorporation of a connective does not imply that it will make the text more explicit in all relevant senses; but if it makes an already recoverable relation redundant merely, this item is qualitically different and it has its own processing implications for the reader exactly as well. The key idea presented by Séguinot (1988) in an early response to Blum-Kulka, i.e. the possibility of using conjunction density as a clean measure of explicitation, is still one of the more lasting theoretical problems due to the distinction between implicitness and redundancy becoming established.

A connected methodological issue is the observation that not all connectives are potential evidence of explicitation. An additive marker like “and” is strongly grammaticalised, so that it carries little independent processing cost, while the addition of a causal or concessive connective (e.g., nevertheless and consequently), which are low in frequency, constitute a much more open signal of intention on the part of the translator. Consequently, any density-based measure that does not track conjunctive tokens in a subtype-sensitive way runs the risk of obscuring those distinctions that are most significant for testing the explicitation hypothesis in theoretical terms; this is why counting by subtype—and treating an aggregated total as a threat to lower validity—are treated as minimum methodological requirements at (5).

4. Arabic-English Contrastive Rhetoric

The Arabic–English pair may thus be a particularly informative, and especially challenging, test case for the explicitation hypothesis as applied to conjunctions in view of the fact that Arabic has been independently described in comparative-linguistic literature (Aziz, 1989; Fareh, 1998) as a language that astro-pg but parataxis: the coordination of clauses through connective such as وَ) wa-, “and”) and فَ )fa-, “so, then”), whereas hypotactic subordination is comparatively more prominent in English expository prose — cf. Dickens & al., 2017 cited Othman & Al Qutob, 2022 p. In fact, Al-Batal (1985) demonstrated the role of connectives in modern expository Arabic in detail and argued that chained coordination and near-synonyms repetition is not just a stylistic ornamentation in Arabic; it forms one of the central devices of textual cohesion on its own.

This contrastive-rhetoric literature creates conflicting readings of conjunction density in translation with sign languages, Arabic directionality also matters. And when translating from Arabic into English, a translator who translates the source text’s dense coordination with literal parallelism is likely to produce a target text relative to normative expectations of English expository language that reads as heavily and unidiomatically over-connected, or “close”-copying rather than explicitation per se. And in a cyclical manner, Othman and Al Qutob’s (2022) recent systemic-functional analysis of the effects of Arabic to English translation on degree of hypotaxis and parataxis across textual genres demonstrated just such an imbalance between translated texts as compared with non-translated comparable language texts, attributing this not only to a more literal transfer at the clause complex level from text-to-text, but also refuting the finding of universal explicitating tendency (Othman & Al Qutob 2022).

The reverse path — English into Arabic — yields a different empirical picture. On the one hand, Chaalal applied his theoretical framework for conjunctions to a corpus-based study of 15163 segments in Arabic and English (English as the source text) between 1990–2013 from United Nations with a finding of an overall broader use of additive, causal, and temporal connectives in Arabic than what was prototypically found in the English source texts concerning explicit cohesion despite the broad comparability between taxonomies describing relations expressed by conjunctive relations across languages (Chaalal, 2018). Given the pre-existing tendency of Arabic to coordinate within its own milieu, this result is consistent with (at least) two explanations that are analytically nearly indistinguishable without a properly-matched comparable corpus: this may represent an exegetical explicitation layered onto top of Arabic’s base preference for parataxis; or it could simply be translators working in the English source material directly towards higher connective density expected in any passage of expository writing by an Arabic reader (Fattah, 2010, cited in Othman & Al Qutob, 2022(

Khalil’s contrastive analysis, based on a comparative study of the relative frequency of use of coordination and subordination respectively in Arabic and English, compares parataxis and hypotaxis (the contrasting roles they play as discourse devices) concluding that this usage frequency difference is not sufficient to account for the fact that their hads may perform different rhetorical functions in each language context, so that an orthodox one-one substitution of connectives rarely suffices _ necessitating judgements about the desired effect being sought by contrasts. The current implication is that any claim about conjunction-density shift in Arabic–English translation needs to be compared with a target-language reference corpus for the particular translation direction concerned, rather then contrasting it against one monolithic notion of what constitutes a natural density of connectivity across the language-pair as a whole.

5. Analytical Criterion of Conjunctions Density

The previous sections suggest that testing the explicitation hypothesis in this way is methodologically limited and should be done with extreme care greater than merely counting and comparing connectives across a single translated work. This paper (1) builds upon the comparable-corpus paradigm developed by Baker (1996) and further refined in other corpus-based translation studies, (2) proposes a three-part analytical framework meant to direct future empirical work on this question while being particularly relevant for typologically distant pairs such as Arabic and English.

The first one is the normalized frequency count on the conjunctive markers for all four subtypes as defined by Halliday and Hasan (1976) — additive, adversative, causal, and temporal since these four clusters of development have been shown to behave differently across genres when considered as a unitary component (Marco, 2018). The other component is a parallel-corpora alignment of the target text to its source text, which permits classifying each connective in the target text as being an addition, a retention, an omission or shift in logical-relation type (which follows closely Becher’s 2011 classification scheme of translation connectives regarding translator motivation). The third is a comparable corpus baseline, comprised of pairs of non-translated target-language texts as closely matched for genre, register, publication period and (where applicable) translation direction against which normalised frequencies from the translated text might be compared ((Overås 1998)).

The combined application of these three components enables a researcher to determine at least three logically distinct patterns that a raw frequency count would conflate: 1) a genuine explicitation effect, in which the connective density of the translated text exceeds both its own source text and the relevant target-language baseline; 2) again an interference effect from the source language, in which connective density for the translated text more closely tracks patterning in terms of connectivity from EITHER source language than with norms for the same dimension reported for texts formed directly by target language (TL); or finally 3) adjustment toward a conventionalized register — especially implicating whatever surface-level effects of such comparison contrast grammatical feature among one given translator’s choices across several samples within shared generalized discursive space. The first of these three patterns is, for Blum-Kulka’s original sense, evidential for explicitation in so far as only a design like this can separate it reliably from the other two.

6.Mechanisms Driving Conjunction-Density Shift

A few (not necessarily exclusive) explanations as to why when translating, translators often add connectives even in cases where the source text does not require them. The first is a processing-based account where the translator’s so-close reading of the source text renders logical relations only weakly cued in the original maximally salient to her or him, and therefore over-signalled for a readership who has not completed similar interpretive work (Blum-Kulka, 1986, p. 21; Pym, 2005). The second is a risk-avoidance account, according to which translators prefer to firmly resolve ambiguity at the cost of explicitness (because they know that the source account author’s style rather than their own error are more likely to be blamed when an ambiguous utterance is misinterpreted by a target reader) rather than preserve unique implicitness that may not get reconstructed by a target reader (Seguinot, 1988).

A third, cognitively oriented account regards explicitation as a side-effect of pressures imposed by the translation task in real time: littérateurs must decode source text and encode target text at the same time, which tilts their preferences toward whichever of any two grammatically available options is more explicit (Becher 2010). A fourth case, specific more to the Arabic–English pair than to explicitation as a general translation process is source-language interference of type (Othman and Al Qutob 2022): as having source language rhetorical convention that tends to lay dense networks of coordinating connectives so that literal or source-oriented translation endows the translated text with high connective density but is not explicitation in the sense of being related to a general tendency within target languages.

When considered together, these narratives of conjunction-density shift in translation represent the view that rather than being mono-causal, their over-determination is perhaps better described as non-made objectual: independent pressures are comprehensively pushing towards greater overt connectivity, which may help explain why the effect has been shown to be so consistently durable across multiple language pairs and why its theoretical explanation remains firmly contested (Murtisari 2016). As we have seen in Section 4, finds evidence from Arabic–English translation research that these pressures do not exert themselves universally—that is, their relative strength depends on translation direction and genre—and certainly why the hypothesis cannot be formulated as an unconditional cross-linguistic universal.

It should also be mentioned that these mechanisms are not stable properties of a language pair but can change depending on genre and register within one language pair. Text types like the institutional and legal ones studied by Chaalal (2018), have conventionally high connective density already in Arabic, and English for reasons that stem from drafting convention rather than translation e.g. compression tends to reduce the observable gap between translated and non-translated connective frequencies. By contrast, literary and journalistic prose are typically less constrained in their allowable connective density margin, so while the direction is consistent with either Blum-Kulka’s (1986) original findings from -then- contemporary-style literature or Chaalal’s (2018) institutional-text findings, the magnitudes of these two studies are not directly comparable.

7. Translation practice and evaluating quality

As regards translator training, the above discussion indicates that increasing connective density should not be taught as a universal explicitating strategy but rather as a genre- and register-sensitive choice to be considered with respect to target-language norms rather than source-text expectations alone. Since Othman and Al Qutob’s (2022) suggest that such literal clause-level transfer is a readily detectable source of unidiomatic over-connectivity in student and professional translations alike, trainee translators working out of Arabic in particular stand to gain from explicit instruction on when a source-language coordinating connective should be retained, subordinated, or simply dropped in the English target text.

Conjunction density is also a truly quantifiable and corpus-comparable tool for evaluating how well natural-sounding sentences flow, but it must be assessed with the context of an appropriate target-language reference corpus rather than impressionistic judgements based solely on intuition about what counts as “natural” in terms of connectivity. This corresponds to Baker (1996, p. 186) original argument that corpus methods represent a shift for translation scholars from anecdotal claims about translated language to falsifiable, quantifiable ones—a claim one may indeed be almost tempted make with regard to conjunctions since they are so perfectly countable.

Finally, editors and revisers of Arabic-to-English translations should be on the lookout for the particular failure mode identified by Othman & Al Qutob (2022): an over-coordinated English target text not as a result of explicitation in Blum-Kulka’s sense but because of too direct an interlingual transfer of clause-level coordination from the Arabic source. It follows, in such cases we should not just delete connectives without catering for the potential ill-cohesion of an under-cohesive target text. Rather, disjointed clause chains deserving to be rendered as strings of sub-ordinate or sentence-initial structures more normally associated with English expository prose, and which preserve the logical relations that must have formed part of what the source author intended in using parapraxes plus complements will arguably require substitution in order to bring this aspect of connective profile back into line with target-language genre norms.

8. Conclusion

Outline support for the claim that translation raises levels of conjunctive markers relative to their natural baseline target language proportions is sincere, but the strongest formulations of the explicitation hypothesis go considerably farther than these conditional claims. While conjunctions have a privileged status in this debate because they are countable and comparable across languages, as well as being closely connected with Halliday and Hasan’s (1976) theory of cohesion— the very properties that make them an appealing proxy for explicitness also mean one should perhaps be cautious about misinterpreting a straightforward frequency increase as evidence for a translation universal when source-language interference, genre convention or target-normalisation might provide just as parsimonious an explanation.

This might be particularly clear-cut for the Arabic–English pair itself, due to Arabic’s own well-known preference for coordination of clauses (as mentioned in Section 4), meaning that an increased connective density in a translated text could be attributable to at least three different mechanisms here that can maintain the same observable outcome. This will necessarily involve future empirical work in parallel with the three-part analytical framework proposed in Section 5 and across both translation directions, combining the deep investigation of parallel-corpus alignment as a basis for comparing explicitation measures with genre-matched comparable-corpus baselines, before we can confirm, qualify or reject the explicitation hypothesis between this remote pair of languages at all — let alone at the same level of confidence that has been reached for more closely related European language pairs. In the meantime, we advocate treating conjunction-density shift as a conditional tendency whose strength and even direction is language-genre-direction specific, rather than as proof of a translation universal.

References

Al-Batal, M. (1985). Coherence function of connectives in a contemporary expository Arabic text (Doctoral dissertation unpublished). University of Michigan, Ann Arbor, MI.

Aziz, Y. Y. (1989). AN ARABIC ENGLISH COMPARATIVE GRAMMAR. Mouol, Yehya and Singaravel, V. (2007) Sarbast Smea Sleen A Policy Proposal for Okers Republican by the University of Mosul Press, Moselsel

Baker, M. (1996). Future challenges in corpus-based translation studies. In H. Somers (Ed.) Terminology, LSP and translation: Studies in language engineering in honour of Juan C. Sager (pp. 175–186). Amsterdam, Netherlands: John Benjamins.

Becher, V. (2010). Disowning the level of explicitation “inherent” to translation: The argument is against a dogma of translation studies. Across Languages and Cultures, 11(1), 1 – 28.

Becher, V. (2011). Connective adds the insight — When do translations need to be added? A corpus-based study. Target, 23(1), 26–47.

Blum-Kulka, S. (1986). Translation: Changes in cohesion and coherence J. House & S. Blum-Kulka (Eds.) In C. van der Gucht, Interlingual and intercultural communication: Discourse and cognition in translation and second language acquisition studies (pp. 17–35). Tübingen, Germany: Narr.

Chaalal, I. (2018). Conjunctive cohesive devices in Arabic United Nations texts and their English versions: a corpus based study. Passerelle, 7(1), 87–96.

Dickens, J. (2017). Syntactic properties of coordination and subordination in Modern Standard Arabic. Journal of Semitic Studies 62 (1) (2017): 187–212.

Fareh, S. (1998). Modern Standard Arabic and the functions of و Journal of Pragmatics, 30(3), 303–312.

Halliday… wffHHALLIDAY, M. A. K., & HASAN, R. (1976). Cohesion in English. London, England: Longman.

Klaudy, K. (1998). Explicitation. In M. Baker (Ed.) 88088, Routledge encyclopedia of translation studies (pp. 80–84). London, England: Routledge.

Klaudy, K. (2009). The asymmetry hypothesis in translatology. Paper presented at the Business and Social Science Conference, Sydney. These assumptions are developed further in this essay in a tribute to Eugene A. Nida, who died some years ago; see Hermans (2014) [Hermans, T. Local publisher, Brussels (Belgique) : Les Éditions du Hazard

Klaudy, K., & Károly, K. (2005). Implicitation in translation: An empirical case of operational asymmetry. 13 PagesPosted: 22 Mar 2016 Last revised: 31 Jan 2017 Abstract This paper proposes a new approach for cross-linguistic and cross-cultural studies of what factors shape frameworks of categories (meanings) in human languages (and cultures).

Marco, J. (2018). A comparable and parallel corpus study of connectives as markers of explicitation in literary translation Target, 30(1), 87–110.

Murtisari, E. T. (2016). Axe explicitation in translation studies: A long way of an equally elusive concept. Translation & Interpreting, 8(2), 64–81.

Othman, W., & Al Qutob, D. (2022). Taxis and logico-semantic relationships in English-Arabic translation inTRAlinea, 24. Retrieved from https://www.intralinea.org/archive/article/2591

Overas, L. (1998). Hunting for the third code: A normative investigation of literary translation Meta, 43(4), 557–570.

Pym, A. (2005). Explaining explicitation. In Proceeding K.Károly & Á.Fóris, (Eds.) In A. Pym & G. Sofer (Eds.), New trends in translation studies: In honour of Kinga Klaudy (pp. 29–34). Budapest, Hungary: Akadémiai Kiadó.

Seguinot, C. (1988). Pragmatics and the explicitation hypothesis. TTR: Traduction, Terminologie, Redaction, 1(2), 106–113.