Contrastive Analysis of Concessive Conjunctions in Translated and Non-translated Arabic Texts: An Exploratory Study

This study seeks to contribute to addressing a gap in theory-driven corpus-based research focused on the so-called translation specific features (TSF) in Arabic translated texts. It provides a contrastive Systemic Functional Linguistics (SFL)-informed analysis of concessive/contrastive connective markers in a selected comparable corpus made up of translated and non-translated Arabic texts. This area of corpus-based research has been mainly driven by an interest in the linguistic features distinguishing translated from non-translated texts. The characteristic feature of the present study is the fact that it is based on a comparable corpus of translated and nontranslated texts written by the same authors in more or less the same genre. Based on a comparison of concordance data, the study will highlight some interesting patterns of difference in the types and frequencies of concessive conjunctions used, as well as ‘explicitating’ and ‘upgrading’ tendencies between the two components of the corpus. Viewed from an SFL perspective, some such differences do not seem to be triggered by the English source texts involved or dictated by contrastive linguistic requirements but rather by the translation process itself.


INTRODUCTION
Since the nineties, corpus-based translation studies have traditionally focused on authentic parallel corpora composed of source texts and their corresponding target texts. Largely informed by corpus linguistics, this area of translation research has been mainly driven by an interest in the linguistic features that distinguish translated texts in general from non-translated texts, regardless of the source or target language. Deploying the techniques of corpus linguistics, researchers engaged in corpus-based analysis of translated texts have observed and posited certain features or tendencies which seem to be distinctive of the language of translation, as opposed to non-translated texts in the same language, regardless of the language pair involved.
The use of comparable corpora was suggested by Baker (1996) as a resource for investigating such features, where a comparable corpus consists of two separate collections of texts in the same language, one of which is composed of original texts in the language in question while the other consists of translations in that language from a given source language or languages. Both components are meant to be in the same language and comparable in domain and register. The goal of this novel approach, especially when used in conjunction with the more usual parallel approach, is to identify or rather isolate any translation-specific patterns, or the distinctive features of translated text per se, that are not attributable to the source or target language systems. A comparable investigation of this kind, given an appropriate corpus, should provide some insight into the translation process itself as well as the individual translator's translational behaviour, i.e. 'the translator's fingerprints' (Aijmer and Lewis 2017: 3), particularly in the unique case where the comparable corpus studied is composed of collections of translated and non-translated texts produced by the same person, as is the case in this exploratory study.
Work which has adopted and demonstrated the strength of this approach includes Braithwaite (1995), who shows that translated texts tend to be 'simpler' than non-translated texts in the same language as reflected in the lower type-token ratio and lower lexical density of the former. Similarly, in a study based on a comparison of concordance data from two corpora, Olohan and Baker (2000) conclude that the use of the that-connective is far more frequent in the translated corpus than the non-translated corpus. An investigation of the specific linguistic properties of translated texts requires a robust theoretical model to provide the necessary linguistic and textual operationalizations for the complex phenomena involved. Such a model is offered by systemic functional theory, which is powerful and rich enough to provide a principled account of translation specific features (Teich 2013).
This paper seeks to contribute to addressing a gap in theory-driven corpus-based research focused on the so-called translation specific features (TSF) in Arabic translated texts. Our study thus falls within the domain of descriptive translation studies, focusing specifically on any distinctive properties of translated as opposed to non-translated Arabic texts, an area of research which could provide insights for translation teaching, contrastive linguistic studies, language contact and change (Teich 2013). We engage in a quantitative and qualitative analysis of concessive/contrastive conjunctive markers in a specially compiled comparable corpus. There are two distinctive aspects of this study which set it apart from similar work in the field of comparable corpus-based analysis. First, the study adopts a Systemic Functional Linguistics (SFL)-informed approach for analysing concessive/ contrastive connective markers in the compiled corpus. The second characteristic and novel feature of the present study is the fact that it is based on a comparable corpus of Arabic translated and non-translated texts written by the same authors in more or less the same domain and register.
Thus, the aim of this exploratory study is to identify and seek to explain salient differences in the use of concessive/ contrastive conjunctive devices in the texts involved. More specifically, this paper seeks to identify any discernible patterns of difference in the use of concessive/contrastive conjunctives in Arabic translated and non-translated texts produced by the same writers and belonging to the same domain, and the extent to which those patterns could be attributed to or associated with explicitation. For this purpose, concordance outputs for those conjunctive markers in both corpora are subjected to a close qualitative and quantitative analysis in pursuit of any consistent or recurrent differences between the Arabic texts in terms of the deployment of the conjunctive markers themselves and any relevant concomitant structural patterns. Viewed from a systemic functional perspective (Halliday and Matthiessen 2014), any patterns or tendencies suggested by those differences are assessed to see if they are triggered by the source texts of the translated corpus, dictated by contrastive linguistic requirements or indeed attributable to the individual translator's translational behaviour or style, given the unique feature of this comparable corpus.

THE DATA
The corpus on which this study is based (see Table 1) is made up of two subcorpora: a parallel component composed of two English source texts written in the domain of history and philosophy, with a total word count of 248,922 words, and their Arabic translations. One of the Arabic target texts (TT1H) was produced by the Egyptian literary writer, novelist and educationalist, Muhammad Farid Abu Hadid (1893Hadid ( -1967 while the other (TT2M) by the well-known writer, intellectual and professor of philosophy Zaki Naguib Mahmoud (1905Mahmoud ( -1993. The second subcorpus is a comparable one including, in addition to the above Arabic target texts, full Arabic non-translated titles authored by the same two translators and belonging to similar domains (history and philosophy). This non-translated corpus comprises six titles, three by each translator/author, totalling 358,862 words (listed in full in Table 1). The selection of these titles was largely determined by availability and comparability to the translational corpus in terms of genre and register. However, when drawing comparisons or looking for any distinctive patterns or trends, it is important to recognize the inevitable internal imbalances in the composition and sizes of full-text corpora involved in any corpus-based study as well as their implications for the interpretation of findings. Indeed, as Baker (2004: 171) notes, such inevitable imbalances 'are not specific to corpus-based studies', but are rather inherent in 'any attempt to look for similarities and differences', where aspects of comparison 'can never be totally balanced in every respect'.
As indicated in the introduction, the contrastive analysis in this paper will be focused on concessive/contrastive conjunctions, which is fairly limited and amenable to automatic analysis using concordance software. However, with the use of an untagged Arabic corpus, a significant amount of manual sorting and analysis is still needed to 'clean' any crude concordance output extracted from an untagged Arabic text. Thus, we will conduct an overall quantitative analysis of the common concessive conjunctives in the comparable corpus, presenting their overall frequencies and percentages in the two Arabic subcorpora and highlighting any significant patterns of variation. We will then focus on the frequency and distribution in the comparable corpus of some features and patterns emerging from the analysis of concessive conjunctives.

GLOBAL STATISTICS
The overall frequency and distribution of the identified Arabic concessive conjunctive markers across the Arabic comparable corpus are set out in Table 2 1 . The figures listed in the table are based on the concordance output for the pre-identified list of concessive markers, some of which could be polyfunctional or multivalent conjunctives. Except ‫لكن‬ (but), whose concordance output also includes non-concessive occurrences, the counted occurrences of the listed markers are mostly concessive or contrastive, i.e. involving an element of counter-expectancy or contrast.
As shown in Table 2, translated texts are generally making more frequent use of concessive conjunctives than their corresponding non-translations. This observed trend is more pronounced in the case of Mahmoud's translations and non-translations (1.17% vs. 0.80%) than Abu Hadid's (0.99% vs. 0.59%). It is also noteworthy that the total number of occurrences of concessives is higher in Abu Hadid's translation (1157) than in his non-translations (814), even though the former is 14.42% smaller in size than the latter. As will be explained in some detail later in this paper, this ALLS 10(4):28-41 higher frequency of concessives in the translated corpus could be attributed to the correspondingly high frequency of concessive/adversative markers in the source texts ST1B and ST2R (Table 3).
A similar pattern of distribution of concessive conjunctives across the comparable corpus is exhibited in relation to the top five most frequent concessive markers in the entire corpus, as shown in Table 4. It is clear from the table that, except the weaker conditional concessive ‫وإن‬ (even if), all these conjunctives are more common in Mahmoud's translations than his non-translations. Similarly, except ‫أن‬ ‫,غير‬ all the top five concessives are more common in Abu Hadid's translation than his non-translations. Indeed, Abu Hadid's translation seems to make up for the lower frequency of ‫أن‬ ‫غير‬ by its heavier reliance on the almost identical ‫أن‬ ‫,على‬ which is twice as common in Abu Hadid's translation (0.10%) as it is in his non-translations (0.05%).
Interestingly, all the top five concessive markers, except the conditional concessive ‫وإن‬ (even if), are paratactic. We will have more to say about the tactic patterns of concessive conjunctives in the next section, but suffice it to say now that the only conjunctive among the top five concessives to be significantly more common in the non-translations than the translations as a whole is the conditional concessive ‫وإن‬ (even if). In fact, the conditional concessive markers, which are arguably weaker than the prototypical concessive markers, are collectively less common in the translations than the non-translations in general. This is highlighted by their much lower proportion relative to the total number of concessive markers in the translated texts, as illustrated by Table 5. Thus, the translated corpus can be said to favour stronger concessive markers and disfavour weaker conditional ones.
The lower frequency of conditional concessives in the translations seems to be consistent with an overall predilection for stronger concessive conjunctives in the translations compared with the non-translations, as indicated, for example, by the preference in the translations for the longer, and arguably stronger, conjunctive group ‫أن‬ ‫من‬ ‫الرغم/بالرغم‬ ‫على‬ (in spite of (the fact) that…) instead of the shorter ‫أن‬ ‫مع‬ (with that…), which is more common in the non-translations 2 (see Table 2).
As Table 3 above shows, the paratactic conjunctive group ‫ذلك‬ ‫و/مع‬ (in spite of that), which includes the text reference item ‫ذلك‬ (that) as a Complement in a prepositional phrase, is more common in the translations than the non-translations. This is also confirmed by Table 6, which includes the other concessive conjunctive Adjuncts with a text reference item in the comparable corpus, which shows that those conjunctive groups are twice as frequent in both translated texts than their corresponding non-translations. In fact, this also seems to be consistent with a wider trend emerging from an analysis of the parallel corpus, where the text reference item ‫ذلك‬ (that) seems to be frequently deployed in upgrading and tactic explicitating shifts in the translated subcorpus (see Fattah, 2010Fattah, , 2016Fattah, and 2018. We will come back to this issue when we examine some features of relevance to upgrading and expansion. In the following Section, we will compare the distribution of hypotactic and paratactic concessive conjunctives in the comparable corpus to see if there are any salient differences or patterns that could be attributed to, or resonate with, shifts in interdependency (from hypotaxis to parataxis or vice versa) in the parallel corpus.

INTERDEPENDENCY PATTERNS
Previous work on this corpus (see for example Fattah 2018) has revealed optional shifts from hypotaxis to parataxis, which were observed to be consistently more common in Abu Hadid's translation (TT1H). That this is a trend in the translated texts seems to be confirmed by Table 7, which shows that the frequency of paratactic concessives is much higher in TT1H (0.77%) and TT2M (0.80%) than Abu Hadid's non-translations (NTH) (0.46%) and Mahmoud's non-translations (NTM) (0.49%) respectively. Mahmoud's translation of Russell (1946Russell ( /1995 (TT2M) has the highest frequency of both paratactic and hypotactic concessives, but the difference between translations and non-translations is much higher in the case of parataxis (0.31%) than in the case of hypotaxis (0.09% and 0.02% for TT1H/NTH and TT2M/ NTM respectively). These differences seem to be consistent with a predilection for parataxis in the translations observed in the parallel corpus analysis, although this bias may be vitiated or obscured by other factors. The relative proportions of hypotactic and paratactic concessive conjunctives also seem to suggest such a paratactic tendency in the translations, though this could well have been influenced by the corresponding proportions in the source texts, as illustrated in Table 8.
A close look at Table 8 reveals a definite shift in the tactic balance in favour of parataxis in Mahmoud's translation (TT2M) compared to his non-translations, which, in theory,   could be attributed to an influence of the source texts. But there is little evidence of any such influence in Abu Hadid's translation (TT1H), whose relative tactic proportions seem to be almost identical with their counterparts in his non-translations (NTH). It seems probable, however, that the apparent lower frequency of concessive conjunctives in Abu Hadid's translation (TT1H) relative to its source text (ST1B (Butler, 1902(Butler, /1978) is due to logico-semantic shifts into other non-concessive paratactic conjunctives, which are not included in the counts, a reflection perhaps of a higher predilection for concessive/adversative conjunctives in English than Arabic texts. For example, in a random sample of 150 concordance lines from the concordance output for but (824 instances in total) in ST1B-TT1H, only 92 instances of but (61.33%) have been translated as paratactic concessive conjunctives. The remaining 58 instances (38.67%) have all been rendered as paratactic nexuses linked by non-concessive conjunctives or none at all (2 instances), as shown in Table 9. As noted earlier, another possible indicator of this overall paratactic tendency is the higher frequency of conjunctive Adjuncts with a text reference item.
A similar pattern of preponderance of paratactic concessive/contrastive conjunctions can be observed in Table 10, which shows the Arabic equivalents of but in a random sample of 150 concordance lines from the concordance output for but in ST2R-TT2M (857 in total). Thus, the paratactic equivalents (concessive and non-concessive) in the sample amount to 145 out of 150, i.e. approximately 95%.

REINFORCEMENT
The analysis of concordance lines of some English and Arabic conjunctives in the corpus has also revealed a tendency to add a semantic component of reinforcement, foregrounding or exclusiveness in the immediate co-text of the conjunctive in question; a feature which will be collectively referred to as 'reinforcement'. Reinforcement of the con-cessive relation could manifest itself as the deployment of a particular rhetorical construction or device denoting exclusiveness or emphasis, or the use of optional correlative conjunctions. It has also been observed that reinforcement shifts seem to be more common in Mahmoud's translation (TT2M) than Abu Hadid's. Obviously, it would be difficult to assess these manifestations of reinforcement in the comparable corpus without linking them to some fairly specific orthographic clues that can be investigated automatically using a concordancer. For this reason, optional correlative reinforcement is relatively easier to investigate, relying, as it does, on the use of conjunctive combinations. The overall frequency of the most common concessive correlative conjunctives in the comparable corpus is set out in Table 11.
As the table shows, concessive correlatives are more common in the translations, especially Mahmoud's, than the non-translations. A closer look at the correlative use of the concessive conjunction ‫أن‬ ‫إال‬ (however) in Mahmoud's translations and non-translations reveals another interesting pattern; as shown in Table 12, it co-occurs with a weaker conditional concessive in approximately 74% of its correlative instances in Mahmoud's non-translations, as opposed to 44% in his translations. By contrast, ‫أن‬ ‫إال‬ occurs in combination with a strong concessive ‫أن(‬ ‫من‬ ‫الرغم‬ ‫على‬ notwithstanding that), where it is even more redundant, in 43% of its correlative instances in Mahmoud's translations as opposed to 15% in his non-translations.
Similarly, other reinforcement elements seem to be considerably more frequent in the translated corpus. Among these is the frequent use of an emphatic modal Adjunct denoting certainty, such as the prepositional phrase ‫شك‬ ‫بغير‬ (without a doubt). An analysis of the concordance output of the keyword ‫شك‬ (doubt) in negative constructions denoting a modality of high probability reveals that these emphatic modal constructions are indeed more frequent in the translated texts as shown in Table 13.
In fact, a closer analysis of the concordance output for such constructions reveals that 'explicitly objective' modal constructions denoting certainty (in the sense of Halliday and Matthiessen (2014: 679) seem to be much more common in the translated than the non-translated corpus as illustrated by Table 13.
Almost all the explicitly objective constructions listed in Table 6-13 involve an embedded expansion or rank-shifting, where the modalized proposition is realized as an embedded clause separated from the modality, the latter being 'propositionalized' and thereby thrown into relief, hence the rein- forcement. Thus, 'X will certainly happen' [ '(it is) certain that X will happen'. Note that this embedded expansion relies on the use of the binder or complementizer ‫نأ‬ (that), which is heavily involved in an overall tendency towards 'clausalization' (Fattah 2018). Similar explicitly objective constructions involving an embedded expansion seem to be markedly more common in the translated texts. Just like the above modal realizations, such explicitly objective constructions have one thing in common: they all involve the writer explicitly expressing his judgement or assessment of an assertion in the form of a 'substantive proposition' (Halliday and Matthiessen 2014: 679), e.g. ‫أن‬ ‫السهل‬ ‫من‬ (it is easy that), ‫أن‬ ‫الممكن‬ ‫من‬ (it is possible that), ‫أن‬ ‫المتوقع‬ ‫من‬ (it is expected that), ‫أن‬ ‫العدل‬ ‫من‬ (it is fair that), ‫أن‬ ‫الواضح‬ ‫من‬ (it is clear that), etc. Thus, they generally take the form of a relational clause with the Attribute being realized by a prepositional phrase and the Carrier by an embedded clause introduced by the complementizer ‫أن‬ (that): Attribute ‫من[‬ (of) + definite verbal noun/adjective] + Carrier ‫أن[‬ (that) + clause]; for example: (1a) English ST2R: Presumably he is thinking of such things as numbers.
Here, it would have been equally possible for the trans-lator to opt for a less pronounced, i.e. implicit, realization of the objective modality, as in the ST, by using such variants as ‫يقصد‬ ‫قد‬ (he may mean), ‫يقصد‬ ‫ربما‬ (perhaps he means) or expressing the modality as a process ‫يجوز‬ (maybe). Another possible manifestation of reinforcement is the seemingly optional use of the foregrounding construction ‫امأ‬ (as for)...‫ـف‬ (then) as a thematic device for enhancing the prominence, markedness or contrast of the Theme. As Table 15 shows, the conjunction ‫امأ‬ is twice as frequent in TTH as it is in NTH, and one and a half times more frequent in TT2M than NTM.
Another form of reinforcement exhibited in the comparable corpus is the more frequent use of intensifiers such as the pre-modifier ‫اللهم‬ used to reinforce the subtractive or exceptive sense of the particle ‫إال‬ in ‫إال‬ ‫اللهم‬ (which may be rendered 'only except/unless' or 'except/unless indeed'), thus denoting that 'the exception is something very rare' (Lane 1863(Lane /1984. Similarly, intensification could take the form of a Cognate (Absolute) Accusative nominal group introduced by the determiner ‫كل‬ or ‫تمام‬ ('all' used as an emphasizer as in 'all confused', or 'full') and functioning as a Qualifier or circumstantial Adjunct of Degree, e.g. ‫االختالف‬ ‫كل‬ ‫مختلف‬ (different all the difference); ‫يعتمد‬ ‫االعتماد‬ ‫كل‬ (relies all the reliance); ‫الصراحة‬ ‫كل‬ ً ‫صريحا‬ (frank all the frankness). As Table 16 reveals, these two types of intensifiers are markedly more frequent in the translated texts than their respective non-translations.  It is arguably yet another manifestation of reinforcement that the assertive or emphatic use of the aspectual-modal particle ‫,قد‬ with or without the proclitic conjunctions ‫و‬ (and) and ‫فـ‬ (then; therefore; so; that is), or the emphatic proclitic ‫,لـ‬ is drawn upon much more heavily in the translated than the non-translated texts, as clearly illustrated by Table 17. It would be beyond the scope of this study to examine in any depth the vexed question of the functions and uses of this seemingly aspectual-modal marker in Modern Standard Arabic 4 . For the purpose of this study, we will assume without further discussion that, when used in con-junction with a verb in the past (perfect) tense, the verbal particle‫قد‬ has an assertive or emphatic function (cf. Baker, 1992: 135) confirming or stressing the occurrence of the process denoted by the verb 5 . We will also assume, quite plausibly, that the marker ‫لقد‬ is even more emphatic than ‫قد‬ by virtue of the additional emphatic proclitic ‫.لـ‬ It is obvious from Table 17 that the particle ‫,قد/لقد‬ when used in conjunction with a past (perfect) verb, is twice as common in the translated texts as it is in the respective non-translations written by the same translators. As the table also shows, the overall frequency of these particles exhib-

Contrastive Analysis of Concessive Conjunctions in Translated and
Non-translated Arabic Texts: An Exploratory Study 37 its an author-specific variation, where they are considerably more frequent in Abu Hadid than Mahmoud's texts. However, what is interesting here is the strong tendency exhibited by the translated texts to use such assertive/emphatic particles. There is an even stronger tendency to use the more emphatic ‫لقد‬ in Mahmoud's translation, where it is approximately 3.3 times more common than it is in Mahmoud's non-translations. By contrast, this distinctive pattern of predominance of the assertive/emphatic ‫قد‬ in the translated texts is not exhibited in the case of modal ‫قد‬ used in conjunction with the present (imperfect) to denote possibility. In fact, the pattern is even reversed about Mahmoud's translation (TT2M) and his non-translations (NTM), where ‫قد‬ is almost six times as frequent in NTM, as illustrated in Table 18.
In addition to the above reinforcement features discussed above, comparable corpus analysis furnishes other interesting contrasts between the translated and non-translated texts involving 'text reference', which is discussed next.

TEXT REFERENCE
As noted earlier, text reference, in the sense of Martin (1992: 139) 6 , seems to be frequently involved in some explicitating shifts observed in the translated texts, notably paratactic shifts, upgrading and reinforcement (see Fattah 2016 and2018). In the context of concessive conjunctives, for example, the use of the demonstrative ‫كلذ‬ (that) as text reference item in a paratactic conjunctive group has been found to be more frequent in translated than non-translated texts ( Table 6). The common thread that seems to be running through such explicitating instances involving the de-ployment of text reference is that a demonstrative reference item (commonly ‫كلذ‬ (that)) is deployed in the repackaging or expansion of a clause into a clause complex or in transforming a hypotactic clause complex into a looser paratactic one. This paratactic transformation may take the following form for example: X ‫ألن‬ (because) Y → X ‫و‬ (and) Z ‫ذلك[‬ (that x ) ‫ألن‬ (because) Y] Thus, the demonstrative pronoun ‫كلذ‬ (that) is used to contract or encapsulate an entire clause (X) (or a part thereof) into a participant (or a part thereof) in another relational clause (Z), with the paratactic additive ‫و‬ (and) being used to connect the two clauses. Similarly, a clause with multiple circumstantial elements may have one of them shifted or expanded into an additional conjoined (usually relational) clause where the matrix clause (or its process) is contracted into a text reference item in order to enable this kind of expansion, as in the following example: he kissed his wife too lovingly in the day time, and before his daughter → he kissed his wife too lovingly ‫كلذو‬ (and that) was in the day time and before his daughter.
This expansive use of text reference is by no means confined to causal relations. Here is another example involving an implicit concessive logico-semantic relation, which was made explicit in the translation by dint of text reference: (2a) English ST2R: ||| It is also on record || that the trade of Tinnîs with Irak alone amounted to between 20,000 and 30,000 dinârs yearly || before it was crushed by vexatious tariffs. ||| (2c) English back-translation: ||| It is also reported || that the trade of Tinnîs with Irak alone amounted to between 20,000 and 30,000 dinârs yearly, || ‫ذلك‬ ‫ولكن‬ (but that) was before heavy tariffs crushed it. ||| Note how the implicit concessive relation in (2a) is made explicit in the translation (2b), where a projected hypotactic clause complex is converted into a paratactic one mediated by the conjunction ‫ولكن‬ (but), with the before-clause being construed as an Attribute in a circumstantial relational clause, where the Carrier is realized as an anaphoric text reference item, i.e.
X before Y (α ^ β) → X ‫ولكن‬ (but) Z ‫ذلك[‬ (that x )was before Y (1 ^ 2) Interestingly, an analysis of the concordance lines for ‫وذلك‬ (and that), used as a demonstrative pronoun (i.e. Head rather than a Modifier in a nominal group), reveals that it is indeed more common in the translated than the non-translated texts, especially in the case of Abu Hadid, as shown in Table 19.
It would also seem that the translated texts in general, but Abu Hadid's in particular, exhibit an overall propensity for text reference, which is most commonly realised in Arabic by the singular demonstrative ‫ذلك‬ (that) and ‫هذا‬ (this). In fact, phoric elements in general turn out to be remarkably more frequent in Abu Hadid's translation than his non-translated texts, while the reverse is true in the case of Mahmoud's translations and non-translations, as shown in Table 20, which sets out the frequency and distribution of the main reference expressions in the comparable corpus. However, the overall frequency of the demonstrative ‫و/فـ/ذلك‬ ('that' whether on its own or prefixed by the proclitic conjunctions ‫و‬ or ‫)فـ‬ seems to be consistently higher in Abu Hadid's translation (1.53%) and Mahmoud's translations (0.81%) than Abu Hadid's non-translations (0.83%) and Mahmoud's non-translations (0.61%) respectively. A possible explanation for this is the more frequent use of this demonstrative in-text reference as highlighted above.
The relatively high frequency of ‫كلذ‬ (that) in the translated texts may be partly attributable to the fact that it is arguably more explicit, and perhaps less ambiguous, than the other devices commonly used for text reference, namely the singular masculine demonstrative ‫اذه‬ (this) and the singular masculine pronominal clitic ‫هـ‬ (it).

SUMMARY OF FINDINGS AND CONCLUSION
In this paper, we conducted a quantitative and qualitative analysis of Arabic concessive conjunctives, as well as other potentially explicitating features in our compiled corpus. The purpose of this investigation was to uncover any consistent or recurrent differences between the Arabic translated and non-translated texts in the corpus with regard to concessive conjunctions and clause combining patterns as well as other potentially explicitating features in the translated corpus.
The results of the comparable analysis suggest a more frequent use of concessive conjunctives in the translated than the non-translated texts, which is attributable to their higher frequency in the English texts compared with the Arabic texts belonging to more or less the same genre. A similar pattern of distribution of concessive conjunctives was observed in relation to the top five most frequent concessive markers in the comparable corpus, with the order of frequency maintained among the individual texts, i.e. NTH < TT1H and NTM < TT2M. This seems to suggest that Arabic texts have a lower propensity for concessive conjunction than their comparable English texts, at least as far as this kind of genre is concerned. This finding echoes Basil and Hatim's (1997, 111) observation that Arabic in general exhibits a preference for through-argumentation rather than counter-argumentation, which is primarily mediated by concessive conjunctives.
A significant qualitative difference was also noted in relation to the types of concessive conjunctives used in the translated texts compared with their corresponding non-translations. The translated texts seemed to exhibit a certain propensity for stronger concessive conjunctives as opposed to the weaker conditional concessives, which were found to be more common in the non-translations. A notable example of this tendency is the preference in one of the two target texts for the longer, and arguably stronger, conjunctive group ‫أن‬ ‫من‬ ‫الرغم/بالرغم‬ ‫على‬ (in spite of (the fact) that…) instead of the shorter ‫أن‬ ‫مع‬ (with that…), which is more common in the non-translations. In so far as this difference could not be consistently attributed to the respective source texts, it could be suggestive of an explicitating tendency manifesting itself as an exaggeration of the concessive logical relation between clauses.
The analysis of the comparable corpus also seems to confirm an apparent preference for or shift in the direction of parataxis, although this paratactic tendency is not uniformly exhibited by the two translated texts. The comparable analysis revealed that the frequency of paratactic concessives is much higher in TT1H (0.77%) and TT2M (0.80%) than NTH (0.46%) and NTM (0.49%) respectively. Moreover, the paratactic gap between TT2M and NTM, as well as between TT1H and NTH, (0.31% in both cases) is much larger than the hypotactic gap (0.02% and 0.09% respectively). The relative proportions of hypotactic and paratactic concessive conjunctives also seem to suggest such a paratactic tendency in the translated texts, though this could be partly attributed to the corresponding proportions in the source texts.
The comparable analysis also revealed a reinforcement tendency, especially in Mahmoud's translation (TT2M), which could be regarded as a form of emphatic explicita- tion (Fattah, 2016). This involved the addition of a semantic component of emphasis, foregrounding or exclusiveness in the translated text. Reinforcement features uncovered in the translated corpus included the more frequent use of double or triple concessive correlatives and 'explicitly objective' modal constructions, especially those denoting high probability. Such constructions, which were found to be nearly four times as common in the translations as they are in the non-translations, involve embedded expansion or rank-shifting through the use of the binder or complementiser ‫أن‬ (that), which is heavily involved in an overall tendency towards 'clausalization' (Fattah, 2016 and2018). Other manifestations of reinforcement emerging from the comparable analysis include the higher frequency of the foregrounding construction ‫أما‬ (as for)...‫فـ‬ (then) in the translations, especially Abu Hadid's, which can arguably be a possible form of textual explicitation manifesting itself as an exaggerated thematization tendency. Some intensifiers, such as the exceptive pre-modifier ‫اللهم‬ and the emphatic determiner ‫,كل‬ were also found to occur more frequently in the translations. Another notable manifestation of reinforcement observed in the comparable analysis is the substantially more frequent use in the translations of the assertive or emphatic aspectual-modal particle ‫,قد‬ with or without the proclitic conjunctions ‫و‬ (and) and ‫فـ‬ (then; therefore; so; that is), or the emphatic proclitic ‫,لـ‬ when used in conjunction with a verb in the past (perfect) tense. It was also observed that the distinctive pattern of predominance of the assertive/ emphatic ‫قد‬ in the translated texts is not exhibited in the case of modal ‫قد‬ used with a verb in the present (imperfect) tense to denote possibility.
Finally, the comparable analysis demonstrated the heavier use of the text reference expression ‫ذلك‬ (that) in various combinations in the translated texts, which was frequently associated with explicitating shifts in parallel corpus analysis (see for example Fattah, 2018), notably paratactic shifts, upgrading and reinforcement. Thus, the analysis of the concordance output for the demonstrative pronoun ‫وذلك‬ (and that) revealed a higher frequency in the translated than the non-translated texts, especially Abu Hadid's, which seems to suggest that upgrading and paratactic shifts are considerably more common in Abu Hadid's translation than Mahmoud's. In fact, the comparable analysis also showed that the overall frequency of the demonstrative ‫و/فـ/ذلك‬ is consistently higher in the translated corpus. This could be attributable to the more frequent use of this demonstrative in-text reference.
While some of the above findings seem to resonate with those of similar studies in other language pairs (see for example Fabricius-Hansen 1996; Steiner 2008; Hansen-Schirra et al 2012), the conjunctive and structural patterns emerging