Abstract

This is a quantitative SFL-based study that is aimed at evaluating an Arabic translation of an English comparative religion text with respect to the realisation of tactic and logico-semantic relations. The evaluation is conducted against the source text and a reference corpus of Arabic non-translations, or original texts, from the same genre. Based on the proposed criteria for determining types of taxis and logico-semantic relations in Arabic discourse, relative frequencies of the tokens of those relations are calculated in isolation and as intersections. The main findings show interesting similarities and differences between the TT, ST, and non-translations concerning distributions of tactic and logico-semantic relations. Specifically, the TT was found to follow a division of labour between hypotaxis and parataxis that is closer to the English ST than to the Arabic non-translations. The results also suggest that the construal of expansion relationships in the TT is incongruent with the TL or TL genre conventions. This incongruency was attributed to the translator’s literal approach to translation at the clause nexus/complex level.

Keywords: systemic functional linguistics, logico-semantic relations, taxis, translation, ArabicEnglish translation

1. Introduction

Arabic has often been described as a language that favours coordination over subordination. A number of studies contrast Arabic and English with respect to coordination and subordination, including Aziz (1989), Reid (1992), Hamdan and Fareh (1999), Othman (2004), and Fareh (2006). The tendency of Arabic to make more use of coordination is a main conclusion in these studies. Other studies with the main focus on Arabic conjunctions or connectives also conclude that coordination is favoured over subordination (e.g. Fareh 1998, Al-Batal 1985). A recent study by Dickens (2017) analyses aspects of the pervasiveness of coordination over subordination in Modern Standard Arabic. Dickens argues that Arabic favours coordination linguistically, textually and rhetorically. Linguistically, Arabic makes abundant use of such conjunctions as و [/wa-/ and] and ف [/fa-/ so, and, then], and there is a high possibility of backgrounding coordinated clauses.[1] Textually, Arabic makes extensive use of (near-) synonym repetition and chained coordination, favouring coordination over subordination. Finally, Arabic is also known to make rhetorical use of coordination, as in hypernym-hyponym repetition and associative repetition (as in السفينة وطاقمها [/ʾal-safīnatu wa- ṭāqamuhā/ the ship and its crew].[2]

Relevant translation research based on systemic functional linguistics (SFL) is scarce. Fattah (2010) investigates clause complexing and conjunctive explicitation in a specially compiled corpus of Arabic translations and comparable non-translations. Focusing on causative and concessive conjunctive markers, Fattah finds lexicogrammatical evidence of explicitational shifts in selected target texts, confirming findings of earlier studies on explicitation. In a similar study, Othman (2019) compares cause construal in a translation from Arabic into English, both with the source text (ST) and respective non-translations. The analysis shows shifts in the realisation of cause-effect relationships across different metafunctional spaces.

In this study, we adopt an SFL perspective to evaluate an Arabic translation of an English text from the genre of comparative religion, first against a sample of non-translations from five Arabic books from the same genre and then against the ST. The aim is not to prove or refute the pervasiveness of coordination in Arabic writing, as this has been shown to be true in the literature (see above, but see also Farghal 2017, who challenges this claim). Nor does this research aim to draw up relations between comparative religion on the one hand and tactic and lexico-grammatical relations on the other hand; this specific genre is used here for the following reason. One of the authors of the current research was commissioned to evaluate the target text (TT) in terms of naturalness of expression, mainly at the textual level. To provide a research-based evaluation, this author decided to compare the TT not only with the ST, but also with a reference corpus of similar texts originally written in the target language (TL). This line of research (e.g. Hansen-Schirra 2003, Matthiessen 2015) is basically quantitative in nature, whereby original texts, or non-translations, are analysed to identify patterns based on frequencies and distributions. The patterns that may emerge from the analysis are seen as the product of conventions or style appropriate to the genre of Arabic comparative religion writings and can thus be regarded as a benchmark for evaluating genre-relevant translations. In fact, the current research can be a first step in an ambitious project aimed at identifying norms and conventions in different genres and text types, which can be used in translation assessment and teaching.

The investigation concerns the two logical systems of taxis (parataxis and hypotaxis, which roughly relate to the traditional notions of coordination and subordination) and logico-semantic relations of expansion (elaboration, extension and enhancement; see Section 2).[3] The current study is thus different from previous ones not only in its theoretical framework (i.e. SFL), methodology (i.e. corpus-based), and approach (i.e. genre-based translation studies), but also in method, which considers two lexicogrammatical systems (i.e. taxis and logico-semantic relations) both in isolation and in combination (i.e. the two taxis modes are investigated in combination with types of logico-semantic relations), as well as other aspects relevant to the realisation of taxis and expansion relations. In short, the research aims at answering the following questions:

1. To what extent is the TT different from respective non-translations, and the ST with respect to realisation of taxis and logico-semantic relations?
2. If the TT is significantly different from the ST with respect to realisation of taxis and logico-semantic relations, what factor(s) may have caused this difference?

The paper proceeds as follows: Section 2 is a brief introduction of taxis and logico-semantic relations from the perspective of SFL. In Section 3, criteria are proposed for the identification and classification of taxis modes and logico-semantic types in Arabic. After the data and methods (Section 4), Sections 5 and 6 provide the analysis and discussion and Section 7 the study conclusions.

2. Taxis and logico-semantic relations

In SFL, the clause conflates several strands of meaning, or metafunctions: ideational, interpersonal, and textual. The experiential mode is related to the content or ideas and is realised by the system of transitivity (i.e. the configuration of the clause comprising Participants, Process, and Circumstances). The interpersonal metafunction is concerned with the relations between the addresser and addressee. Interpersonal meanings are enacted in grammar by the systems of mood (i.e. indicative or imperative) and modality (e.g. probability, usuality, temporality). The textual metafunction is concerned with the distribution of information in the clause and is realised by the Theme and Information systems (Halliday & Matthiessen 2014). In addition to these metafunctions, there is also the logical mode of the ideational metafunction. This mode is related to relations between ideas and is realised by taxis (i.e. hypotaxis and parataxis) and logico-semantic relations (or the meanings that join clauses together, i.e. elaborating, extending, enhancing). Although the metafunctions are all simultaneously instantiated whenever language is used, the primary interest in this paper, given space limitations, is in the systems of taxis and logico-semantic relations. These systems determine the relationships between clauses and belong to the logical mode of the ideational metafunction.

According to SFL, units of every rank may form complexes by means of expansion. For example, a clause simplex may be linked to another clause simplex through some logico-semantic relation to form a clause complex.[4] When a clause complex consists of more than two simple clauses, each single linkage is referred to as a clause nexus.

Taxis refers to the degree of interdependency between one clause and another. In a paratactic nexus, the two clauses are treated as being of equal status, each constitutes a proposition in its own right and could thus be tagged; e.g. Kukul pulled out the arrow, didn’t he? and headed for the river, didn’t he? (Halliday and Matthiessen 2014: 438). For such constructions with the status of equal, there is a closely agnate version where the two clauses are not brought together structurally in a clause complex but rather form a cohesive sequence; i.e. with a full stop between clauses (Ibid: 458). For example, the paratactic construction in Kukul crouched low to the ground and moved slowly can be readily rephrased as Kukul crouched low to the ground. He moved slowly. Alternatively, the two clauses forming a nexus could be treated as being of unequal status, where only the main clause constitutes a proposition in its own right and can thus be tagged, or queried (Ibid: 440); e.g. Kukul headed for the river, didn’t he; did Kukul head for the river? Clause complexes may also be formed by a mixture of parataxis and hypotaxis.

The clause simplexes making up a clause complex are referred to as ‘primary’ or ‘secondary’. In a paratactic clause complex, the primary clause is the one that comes first (initiating). In hypotaxis, on the other hand, the primary clause is the independent (dominant) one, and the secondary clause is the dependent, regardless of the order in the clause complex. Figure 1 below illustrates this and shows the SFL notations used for each. In this paper, we will be using the terms ‘primary’ and ‘secondary’.

Figure 1: Primary and Secondary clauses in a clause nexus

Within clauses complexes, we can find sub-complexes that are sequenced (as in Figure 2) or nested (as in Figure 3)

Figure 2: Clause complex with a linear sequence

Figure 3: Clause complex with nesting

Logico-semantic relations comprise two main types: projection and expansion; only the latter is investigated here.[5] Within the general category of expansion, there are three subtypes: elaborating, extending, and enhancing. These are introduced here and illustrated with relevance to Arabic in the next section. SFL notation is also shown.

Elaboration (=) is a logico-semantic relation of expansion, where a clause or group restates, specifies, comments on, or exemplifies the meaning of another. In the clause complex John didn’t wait; he ran away, the simplex he ran way elaborates on he didn’t wait by restating its meaning; 1^ =2.

In the extension (+) type of expansion, the extending clause or group provides an addition, a replacement, or an alternative. In the clause complex John ran away, whereas Fred stayed behind, the first clause is extended by the adversative information in the second; α^ +β.

Enhancing (x) is a relationship of expansion through which a clause qualifies another with some circumstantial feature of time, place, manner, cause, or condition; for example, Because he was scared, John ran away; β^ xα.

SFL looks at taxis and logico-semantic relations in full-ranking and embedded complexes, the latter are not included in the current analysis. According to Halliday and Matthiessen (2014: 493), within embedded clauses, the distinction among the three categories of elaborating, extending and enhancing, as found in parataxis and hypotaxis, is less foregrounded.

3. Taxis and logico-semantic relations in Arabic

3.1. Methodological problems and proposed criteria

Research analysing Arabic text is typically faced with the problem of defining the notion of sentence in Arabic (see, for example, El-Shiyab 1990, Khafaji 2001). Determining sentence boundaries in Arabic is not as straightforward as in English because punctuation marks are not strictly rule-governed as they are in English. This means that the clause simplex, clause complex, and cohesive sequence are not so clearly marked in Arabic writings. To make things even more difficult, Arabic sentences are often introduced by connectives, particularly /wa-/. This Arabic particle could serve as a logical conjunction (i.e. between clauses) or as a cohesive conjunctive (i.e. between sentences or sentence-/paragraph-initially). According to Fattah (2010), this makes it difficult to determine with certainty whether a potentially freestanding simple clause or clause complex is a member of a coordinate structure or an independent sentence.

Another problem with relevant research is indeterminacy regarding the classification of Arabic conjunctions as paratactic or hypotactic markers. Waltisberg (2006: 468) claims this indeterminacy renders it “at least hazardous to analyse Arabic conjunctions prima facie as coordinating or subordinating”. In fact, this has always been a problem for research because traditional grammars of Arabic, even those written in English, do not deal with conjunctions, particularly subordinating ones, in a separate section; they are mainly discussed in connection with different types of sentences, such as conditional sentences and circumstantial clauses (Kammensjö 2011). Worse still, classifying conjunctions as coordinating or subordinating sometimes tended to be based on their English counterparts (Ibid).

Given the aforementioned problems, investigating taxis and logico-semantic relations in Arabic is more challenging than in English. In this research, we use the syntactic and semantic criteria proposed by Al Kohlani (2010) for determining sentence boundaries and an informal operational means adopted by a number of scholars working on Arabic. Those scholars have found it practically useful to determine sentence boundaries based on intonational features of sentences when read out aloud (see Dickens 2010 and several others cited therein). On the other hand, because the focus here is on investigating clause linkage, the analysis within sentences will be based on a count of clause nexuses or combinations within clause complexes. After all, the aim is to decide whether the translation follows the respective genre’s conventions of clause linkage.

Al Kohlani (2010) defines a sentence as “an independent unit whose components are bonded syntactically and semantically in order to perform a communicative meaning that contributes to the overall rhetorical purpose of the text” (Ibid: 201). A sentence candidate, she argues, “must be omissible leaving behind no non-sentences” and must express a complete thought, as well as serve a specific rhetorical function, such as clarifying a foregoing sentence or piece of discourse (Ibid: 194-9). Together with the read-out-aloud strategy (used in Dickens 2010), this criteria seems to be adequate for our purpose; because we are mainly interested in clause linkage, the statistics will be of clause nexuses within clause complexes.

However, even at the level of clause complex, or clause nexuses within complexes, it will still be difficult to determine the taxis mode, at least because some Arabic markers have several semantic functions (see examples below). This is why an attempt is made to provide clear criteria for singling out paratactic and hypotactic constructions.

To identify paratactic constructions, we propose the following parameter:

Paratactic constructions, but not hypotactic ones, can be broken down into structurally correct, fully propositional juxtaposed clauses separated by a full stop or a semicolon, either with or without a conjunction. The conjunction can be the one used in the original construction or one of a closed set of clearly paratactic conjunctions that construes the same logico-semantic relationship between the conjoined clauses, and either conjunction can take place sentence-initially or even paragraph-initially.

Examples (1) and (2) are illustrative.

In (1), the conjunction كما [/kamā/ and also] marks a relationship of extension: additive, and can thus be replaced with /wa-/. Note also that breaking down this nexus into two disjoined clauses separated by a full stop will result in a cohesive sequence comprising two structurally correct, fully propositional juxtaposed clauses. In such juxtaposition, the conjunction can be retained (in which case it becomes more textually than logically operative), or dropped (which could impact the text cohesiveness).

Here again the conditions stated in the proposed parameter are satisfied: (1) Breaking down the construction would result into structurally correct, fully propositional juxtaposed clauses. (2) The conjunction can be (i) retained after the full stop, becoming more textual than logical, (ii) dropped altogether, or (iii) replaced with /fa-/, one of the main paratactic conjunctions in Arabic.

In addition to this parameter, particularly in the case of indeterminacy, we propose using SFL’s trinocularity perspective. SFL adopts a system perspective on language in which any type of phenomenon or descriptive category can be viewed and interpreted from a trinocular perspective. In Halliday’s words (2008: 141),

When we are observing and investigating language, or any other semiotic system, our vision is essentially trinocular. We observe the phenomenon we want to explore – say, the lexicogrammar of language – from three points of vantage. We observe it from above, in terms of its function in various contexts. We observe it from below, in terms of its various modes of expression. And thirdly, we observe it from its own level: from within, or from roundabout, according to whether we are focussing on the whole or some of its parts.

These three perspectives correspond to different strata. In our case, since we are looking at the lexicogrammatical realization of taxis and logico-semantic relations, the perspective from below corresponds to the stratum of expression (morphological and phonological realization of meaning). The perspective from roundabout corresponds to the stratum of lexicogrammar (what other options the option at hand contrasts with and the paradigmatic choices associated with those options). The vision from above corresponds to the stratum of semantics and looks at what each category realizes or how it relates to meaning. This latter perspective is given priority in functional grammar, according to which form follows function and the meaning of an expression decides its realization. From above we also consider context, which is often helpful in determining the type of categories being investigated. In describing the grammar, we can work from above downwards: how contextual choices activate semantic choices, which activate the lexicogrammatical ones, which activate morphological and phonological ones. We can also start from below: how morphological and phonological realizations construe lexicogrammatical choices, which construe semantic choices, which construe contextual ones (Hasan 2009).

In this research, the trinocularity perspective is adopted to determine the taxis mode or the logico-semantic type in clause nexuses/complexes that prove difficult to analyse on the basis of the proposed parameter. This is particularly useful in cases of indeterminacy that result from, for example, the use of a multifunctional conjunction, such as /wa-/.

In (3), the logico-semantic relationship between the conjoined clauses is marked by the conjunction /wa-/. Based on the view from below, it could be argued that this construction is paratactic and the logico-semantic relationship is an additive one, since the /wa-/ is prototypically a paratactic marker of extension and there is no other covert indication of a different type of logico-semantic relationship. However, when we look at the instance from within, it becomes clear that this is an instance of hypotactic enhancement; the paradigmatic choices here pertain to the structure of the circumstantial clause in Arabic (الحال Haal). Finally, the view from above also suggests enhancement, particularly temporal, since the co-text/context involves a circumstantial relation of time. In short, the construction above is a hypotactically related clause complex denoting a relationship of enhancement: time.

3.2. Arabic paratactic and hypotactic markers

The main and most frequent Arabic conjunctions that fulfil the proposed parameter include a closed group of particles listed in all traditional grammars of Arabic as paratactic conjunctions: و [/wa-/ and], ف [/fa-/ so, then, and], ثم [/thumma/ then, later], أو [/ʾaw/ or], بل [/bal/ rather], أم [/ʾam/ or], and لكن [/lākin/ but]. Two of the closed-set conjunctions, namely /wa-/ and /fa-/ are clitic conjunctions that could serve as paratactic and hypotactic conjunctions, as well as cohesive/textual conjunctives, and both can mark several logico-semantic relations. Because of the multiple senses of the /wa-/ and /fa-/, they have received most attention in research on Arabic conjunctions; see for example: Saeed and Fareh (2006), Dendenne (2010). According to Saeed and Fareh (2006), Arabic /wa-/ can mark relationships of addition, contrast, concession, comment, simultaneity, sequence, and resumption, while the /fa-/ can encode concession, reason, result, sequence, and explanation.

The /wa-/ is the conjunction with the highest frequency of use. Its frequency of use and the wide range of functions it may serve cannot be reproduced in English (Cantarino 1975: 12). It is mainly a paratactic extension conjunction of the additive subtype, but it is not unusual to find instances of /wa-/, or /wa-/ prefixed to a pronoun, introducing elaborating clauses. The /wa-/ can also function as a hypotactic conjunction, where it introduces a circumstantial clause, or realizes the semantic meaning of accompaniment. In short, the /wa-/ results in paratactic constructions except when it introduces circumstantial clauses of Haal or accompaniment. When used sentence-initially or paragraph-initially, the /wa-/ is cohesively rather than logically operational.

The /fa-/ is another paratactic conjunction with a very high frequency of use. Dickens (2017) refers to the /fa-/ as another ‘and-type’ coordinator, which contrasts with the one English ‘and’ coordinator. Like the /wa-/, the /fa-/ can be used cohesively or logically; for example, it is cohesive when used at the start of a paragraph, or when it indicates a “slight shift in topic” (Ryding 2005: 410), in addition to other cohesive functions (see Cantarino 1975: 20–34). Logically, the /fa-/ serves as a paratactic conjunction.

There are other frequently used conjunctions that satisfy the proposed parameter and therefore result in paratactic constructions. These includeإذ [/idh/ since/as], كما [/kamā/ also], and أي [/ʾay/ that is]; see Fattah (2010) for a fuller list. To the list of paratactic conjunctions, we can also add cohesive conjunctions and conjunctive phrases that can function as paratactic conjunctions when used inter-clausally, also based on the parameter above, including the possibility of replacing such conjunctive phrases with those in the closed group. An example is the concessive conjunctionغير أن [/ghayra ʾanna/ however]. See Example (4).

In this clause complex, the first clause is potentially free-standing, which is conjoined to the second clause by means of the concessive conjunctionغير أن [/ghayra ʾanna/ however]. This conjunction, Fattah (2010) argues is paratactic as it could potentially replace the clearly paratactic conjunction لكن  [/lākinna/ but] in most situations where the latter is serving a concessive paratactic function, even where it is paragraph-initial. That is, both can take place sentence-initially, or even paragraph-initially and can link freestanding, independent clauses and longer stretches of discourse. This also applies to other synonymous conjunctions such as بيد أن [/bayda ʾanna/], على أن [/ʿalā ʾanna/], and إلا ان [/ʾillā ʾanna/]. All these conjunctive phrases could link paragraphs and potentially freestanding clauses. This contrasts with conjunctive phrases such as بالرغم من أن [/bi-rraghmi min ʾanna/ although], which could not be preceded by a full stop or be paragraph-initial, and are thus regarded as hypotactic conjunctions. See below.

Some textual conjunctives frequently occur in combination with /wa-/, as in ولذلك [/wa-li-dhālika/ and therefore], and وبالتالي [/wa-bi-ttāli/ and consequently]. Following Fattah (2010: 99), the /wa-/ in such combinations does not seem to make any unique logico-semantic contribution to the clause sequence, although it provides a stronger link between the conjoined clauses. In other words, the /wa-/ in such instances does not mark the logical relation between the conjoined clauses, but it marks a paratactic mode of taxis.

Although the proposed parameter, in addition to the trinocularity vision, could be sufficient for identifying taxis mode, we think it would add to the robustness of the research procedure if we elaborate on hypotactic conjunctions. In general, Arabic hypotactic constructions are easier to identify than paratactic ones, although the former make use of a much larger group of conjunctions. In this research, instead of providing a conclusive list, which is neither possible nor necessary, we will classify hypotactic conjunctions into five main types, in addition to the circumstantial /wa-/ in Haal clauses.

I – Non-defining relative clauses

II – Conditional clauses

III – Conjunctive phrases ending with the complementizer أنّ [/ʾanna/ that], provided such phrases are not prefixed with /wa-/ and cannot be used sentence or paragraph initially, such as بالرغم من أن [/bi-rraghmi min ʾanna/ In spite of the fact that.

IV – Adverbial clauses (as the term is used in traditional grammar)

Such clauses are mostly enhancing; they are introduced with words that mark time, place, manner, and cause, for instance: طالما [/ṭālamā/ as long as]; حالما [/ḥālamā/ as soon as], etc.

V – Clauses initiated with non-finites, prepositions, and prepositional phrases that explicitly mark the logico-semantic relations, e.g.بغية [/bughyata/ for the purpose of]; بهدف[/bihadafi/ for the goal of]; خشية أن [/khashyata ʾan/ for fear of]; لكي [/li-kay/ in order to]; من أهمها[/min ʾahammihā/ the most important of which].

4. Data and method

The data of this study comprise random samples from five Arabic books from the genre of comparative religion, in addition to a sample from a translated book from the same genre and its ST. Each sample consists of 200 clause nexuses. See Table 1.

Table 1: Sources of investigated samples

As mentioned earlier, since a comma, rather than a full stop, is commonly used as an end punctuation mark, the unit of analysis used here is the clause complex and the clause nexuses within those complexes. To determine boundaries, whether of complexes or simplexes, we applied the syntactic and semantic criteria proposed by Al Kohlani (2010), as well as the read-out-aloud strategy (used in Dickens 2010). Following this procedure, we were able to make a distinction between simplexes, complexes and cohesive sequences. Instances of simplexes or complexes ending with a full stop did not constitute a problem. The process of identifying and classifying relevant instances in terms of taxis and logico-semantic relationships was based on the criteria explained above.

Because the sampled books are only available as hard copies, the analysis of the following distributions was conducted manually:

(1) Distributions of paratactic and hypotactic nexuses;

(2) Distributions of the three types of logico-semantic relations across hypotaxis and parataxis

Chi-square test, or Fisher Exact test for numbers below five, were used to determine statistical significance. The tests involved comparing the TT first with the non-translations and then with the ST.

5. Analysis

5.1. Paratactic and hypotactic nexuses

Table 2: Distribution of nexuses across taxis in the non-translations

Table 2 shows the numbers of nexuses in the Arabic non-translations. The data indicates roughly similar distributions between hypotactic and paratactic nexuses in the five Arabic samples, which clearly reflects a preference for parataxis, similar to conclusions in previous literature (see Section 1). This division of labour in construing tactic relationships can be seen as the typical distribution for non-translations in this genre and provides a benchmark for evaluation of translated texts. Thus, a translated text can be considered to follow typical genre conventions if it exhibits a fairly similar division of labour in its realisation of tactic relationships.

Table 3: Distribution of nexuses across taxis in the non-translations, TT, and ST

Table 3 compares the distributions of paratactic and hypotactic nexuses in the Arabic non-translations with those of the ST and TT. It can be gleaned from the table that the TT, compared to the non-translations, features a completely different distribution, with parataxis and hypotaxis roughly equal. A chi-square test of independence (using the values for non-translations and TT as a 2 X 2 contingency table) returned a result of X² = 33.708; p < 0.05. This statistically significant difference indicates that hypotaxis is over-represented and parataxis is under-represented in the TT compared to the non-translations. Another chi-square test involving the ST and TT gave a result of X² = 4.923; p < 0.05, which is statistically less significant than that found for the TT and non-translations. This means that the TT follows a division of labour between hypotaxis and parataxis that is closer to the English ST than to the Arabic non-translations. It is more hypotactic and less paratactic than conventionally expected.

5.2. Logico-semantic types across taxis

Table 4 shows the distributions of the three types of logico-semantic relations of expansion across hypotaxis and parataxis in the non-translations, TT and ST. Starting with the non-translations, a total of 1000 nexuses cited are accounted for. It can be clearly seen that elaboration is more typically realised as paratactic than hypotactic constructions (81 percent and 19 percent, respectively). The clear skew in frequency here could be attributed to the high frequency of the Arabic paratactic conjunction /fa-/, which is commonly used to mark elaboration.

Table 4: Distribution of logico-sematic types across taxis in the non-translations, TT, and ST

Extension was also found in favour of parataxis over hypotaxis, with a notable 98 percent for paratactic extension and two percent for hypotactic extension. The conspicuous skew here is in fact expected because extension relationships in Arabic are mainly signalled with the paratactic conjunction /wa-/, which was also cited prefixing other paratactic extension conjunctions, such as ولكن [/wa-lākin/ and but].

On the other hand, enhancement, as Table 4 shows, is more frequently realised in hypotactic than in paratactic constructions (72 percent and 28 percent, respectively). Enhancement citations included almost all types of hypotactic enhancement conjunction; e.g. لـ [/li/ to], لأنّ [/li-ʾanna/ because], حتى [/ḥattā/ so that, until], إذا [/ʾidhā/ if], etc. Frequently used hypotactic conjunctions also included non-finites, or reduced relative clauses (e.g. نتج عنها [/nataja ʿanhā/ leading to]), in addition to instances of the Arabic لأجله المفعول  [/ʾal-mafʿūl li-ʾajlihi/ benefactive object], such as أن خشية [/khashyata ʾan/ for fear of].

We now need to compare the frequencies for the non-translations with those found for the TT and ST (all listed in Table 4 above).

The most evident difference between the non-translations and the TT is manifested in the contrasting patterns and markedly different distributions of elaboration relationships, with the non-translations mainly paratactic (81 percent) and the TT mainly hypotactic (67 percent). The chi-square test result of these contrasting patterns was X² = 22.272; p < 0.05. Another chi-square test, of the counts for the TT and ST, returned a non-significant difference (X² = 1.575; p < 0.05). More precisely, the counts for elaboration in TT are similar to those in the ST but markedly different from those in the non-translations. With the TT being considerably more hypotactic and less paratactic than expected, it could be said that the construal of elaboration in the TT is incongruent with the conventions of TL genre.

The statistics also show a significant difference between the non-translations and the TT in construing extending relations across parataxis and hypotaxis. Although the non-translations and the TT are both more paratactic than hypotactic, the test result (X² = 8.164; p < 0.05) suggests that the TT still differs from respective TL non-translations in that it makes notably more use of hypotaxis in construing extending relationships (seven percent in the TT but two percent in the non-translations). Another chi-square test involving extension in the TT and ST returned non-significance (X² = 0.655; p < 0.05). The results of these two tests suggest that the construal of extension relationships in the TT is closer to the ST or SL conventions than to the TL or TL genre conventions, in that the TT is less paratactic in realizing extension than it should typically be.

In the case of enhancement, although the non-translations and the TT are both more paratactic than hypotactic, there is a statistically significant difference (X² = 7.252; p < 0.05) in construing enhancing relations across parataxis and hypotaxis. The distributions in the TT and ST also reflect a different division of labour, evidenced by a significant difference (X² = 3.921; p < 0.05). Since both results are significant, with the one involving the TT and ST less significant, we could say that TT is slightly more similar to the ST than to the non-translations.

6. Discussion

As mentioned above, the TT was found to follow a division of labour between hypotaxis and parataxis that is closer to the English ST than to the Arabic non-translations; it is more hypotactic and less paratactic than conventionally expected. This result carries implications of unnaturalness at the logical/ textual level, most likely due to interference from the ST/SL. In fact, a very strong factor affecting tactic distributions in the TT seems to be the translator’s inclination to a literal approach of translation at the clause nexus/complex level. Most of the ST hypotactic constructions are rendered literally into Arabic, as in example (5), which in both the ST and TT comprise a primary clause (i.e. Christianity has no choice) hypotactically expanded with secondary clauses (i.e. to prove …Judaism and but to be …Christ).

In a few cases, where a shift is obligatory or necessary for a more natural construction, the translator rendered a hypotactic construction into a paratactic one, as in example (6), where the non-finite clause (i.e. providing … leap) is rendered into an independent clause introduced with the paratactic conjunction إذ [/idh/ since/as].

The following is an instance where the translator opted for a more natural rendering.

In this example, the English sentence is a clause complex, with the secondary purpose clause (to describe …godhead) enhancing the primary clause (The otherwise …metamorphosed). In Arabic, the same hypotactic construction is used, but the translator also made a shift at rank level, rendering the conjoined noun group (and ultimately godhead) as a clause by repeating the verb (describe). This is completely natural in Arabic and actually very common (Hassan 2015:145), given that repetition is a main device for linguistic cohesion and rhetorical force in Arabic writing (Johnstone 1991:131).

The percentage of paratactic constructions in the TT could have been lower had the translator not sometimes opted for conjoining juxtaposed simple clauses into paratactic constructions, as in Examples (8) and (9) below.

Example (8) is a case of disjunction in English, but the two sentences are in a relation of elaboration, which is left implicit. The translator chose to link the two simplexes into a clause complex, using the Arabic /fa-/ as an explicit marker of elaboration. In Example (9), the simplexes are linked with the coordinator /wa-/, which brings out the implicit extension relationship in the ST.

In short, it can be concluded that the skewed division of labour between hypotaxis and parataxis in the TT is mainly attributed to the translator’s literal approach to translation at the clause complex rank. The limited cases of shifts from hypotaxis to parataxis were mostly obligatory or made to improve naturalness and avoid awkwardness.

7. Conclusions

This research investigated taxis and logico-semantic relations in an Arabic translation of an English text from the genre of comparative religion by comparing it with the ST and a sample of non-translations from the same genre. The aim was to evaluate the TT vis-à-vis congruency with registerial norms and conventions, which would result from analysing the non-translations. To this end, the study adopted an SFL perspective and a corpus-based methodology to study taxis and logico-semantic relations both in isolation and in combinations. An attempt was made to provide criteria that would help address problems relating to identifying sentence boundaries and inconsistency in the use of punctuation marks, as well as indeterminacy with respect to the classification of Arabic conjunctions as paratactic or hypotactic. The unit of analysis was the clause complex but the count was of clause linkage relations in clause nexuses. A parameter was proposed for classifying conjunctions in terms of taxis, which in short states that only paratactic constructions can be broken down into structurally correct, fully propositional juxtaposed clauses separated by a full stop or a semicolon, either with or without a conjunction. In cases of indeterminacy, SFL’s trinocularity perspective was also suggested as a means for determining the taxis mode and the logico-semantic relations in Arabic nexuses (See Section 3).

The analysis of the original Arabic texts from the genre of comparative religion confirmed results in previous research with respect to the pervasiveness of paratactic over hypotactic constructions in Arabic writing in general. See for example Al-Batal 1985; Aziz 1989; Reid 1992; Fareh 1998; Hamdan and Fareh 1999; Othman 2004; Dickens 2017. The TT, on the other hand, was found to include roughly even distributions of the two taxis modes, suggesting overuse of hypotactic relations relative to genre conventions and therefore possible interference from the SL. With respect to the construal of logico-semantic relations across taxis, comparing the TT to the ST and non-translations indicated incongruency in realisation of all logico-semantic types, with elaboration being the most incongruently realised. Further research could address shifts in taxis and logico-semantic relations in terms of structural complexity. According to Halliday (2009), “it is usually assumed that hypotaxis adds more to the structural complexity than does parataxis.”

According to Halliday and Matthiessen (2014), quantitative patterns can be related to qualitative properties of the system as a whole or of specific genres and registers within the system. In a sample comprising 6,832 clause nexuses in spoken and written texts from a fairly wide range of registers, Halliday and Matthiessen (446–7) found that parataxis and hypotaxis are roughly equally frequent, expansion is considerably more frequent than projection, and enhancing relations account for almost half of all instances while elaborating and extending relations have almost equal shares of the other half. In our genre-specific sample of non-translations, the patterns found conform with the preferences of Arabic writing in general. Similar preferences, could be hypothesised with relevance to specifc genres, but this can only be confirmed through quantitative examinations of large corpora from different genres or registers. This line of research is in agreement with the SFL understanding of a language as an assemblage of registers (Ibid: XV).

The skewed division of labour between hypotaxis and parataxis in the TT was attributed to the translator’s literal approach to translation at the clause complex rank, hence the marked prevalence of hypotaxis over parataxis. There are, however, instances of shifts that have contributed to the ratio of parataxis. These included literal renderings of English paratactic constructions, obligatory shifts, and optional shifts made to avoid awkward structures, in addition to instances of simplexes linked into paratactic nexuses.

Both types of renderings, whether literal (non-shifts) or adaptations (shifts), could affect not only the structural and textual organisation of the text, but also its overall rhetorical purpose. For instance, literally rendering two English semantically related simplexes as two simplexes in Arabic would be a deviation from normal usage unless the disjunction is rhetorically motivated. Similarly, rendering an English hypotactic nexus with a foregrounded secondary clause into a paratactic nexus in Arabic would impact the textual surface of the text and may eventually affect its rhetorical force. In this context, research, ideally text type-/genre-based, is direly needed to study rhetorical implications of structural and textual shifts and non-shifts in taxis and logico-semantic relations. After all, “rhetorical purposes impose their own constraints on how a sequence of sentences becomes a text” (Hatim 1997: 32). For an informative relevant discussion, see Hatim (1997: 102–3), where he illustrates “the need on the part of the translator to be extra vigilant when switching from a literal mode to a freer one or vice versa”. Using two instances of the subordinator whether drawn from different text types, Hatim shows how opting for the same form in Arabic reflects a flawed interpretation of the intended function. See also Bazzi (2009) on the analysis of news media and war reporting discourse. In the context of thematic structuring of information, Bazzi (161–2) cites an example where an instance of but in an expository text is thematically foregrounded to express judgment.

The research did not aim to draw up relations between comparative religion texts and tactic and lexico-grammatical relations, although such an investigation could yield interesting results on the relation between genre/text type and expansion types. For example, Smith and Frawley (1983) suggest that genres differ in how conjunctive they are as well as in the types of conjunctions they prefer. In religious texts, for example, we can expect heavy use of negative additive and causal conjunctions. Similar research into different genre conventions could be very helpful in translation assessment and teaching.Given scope and time limitations, as well as the dataset size, further research could make use of a larger corpus to produce more generalizable results. Also, given the difficulties encountered with respect to delimiting boundaries of sentences and clause complexes, due to extended lengths of sentences and inconsistencies in punctuation marks, and since the count was of clause nexuses, rather than complexes, the validity of the results might be arguable. The argument is plausible, and therefore, the authors still think there is dire need for a more systematic theory-based approach to the delimitation of boundaries of sentences or clause complexes in Arabic. However, the effect of this limitation may be alleviated when we look at the current investigation from the vantage point of statistical frequency of conjunctions; that is, through basing the classification on the counts of paratactic and hypotactic conjunctions between nexuses, rather than on the number of complexes or sentences.

Further research could also look at distributions of hypotactic nexuses vis-à-vis the sequence of primary (α) and secondary (β) clauses, in order to explore implications of shifts in clause sequencing on the overall text cohesiveness in terms of theme-rheme and given-new information. Such a study could compare source texts, target texts, and reference corpora for distributions of unmarked sequences vs. marked sequences; i.e. the primary clause preceding the secondary clause α^β or the other way round. Relevant to this, Dickens (2010: 1129) cites diachronic research comparing Classical Arabic with Modern Standard Arabic that has found that MSA makes far greater use of fronted (thematic) adverbial clauses, which could indicate a partial shift from coordination to subordination in sentence structuring. This shift, Holes (2004: 265-6) explains, could also be seen as an influence from foreign textual models, mainly English, particularly in media writing and some modern literature. Holes explains that “MSA has shown a tendency toward the use of complex sentences”, which he ascribes to the use of lexical phrases such as ‘in accordance with’, ‘regardless of the fact that’, etc. Comparing MSA with older styles of prose writing, Holes (Ibid) claims that logico-semantic relationships are more easily deducible with such lexical phrases than with a few general-purpose particles that allow a variety of interpretations, which refer to the closed group of paratactic conjunctions mentioned above. Relevant citations in the current research included على اعتبار أن [/ʿalā iʿtibāri ʾanna/ on the grounds that] and على أساس أن / [/ʿalā ʾasāsi ʾanna/ on the basis that], among others.

The results of this research can be seen as a contribution to the study of taxis and logico-semantic relationships in Arabic translations and non-translations, since the literature has so far mainly focused on the issue of parataxis vs. hypotaxis from the vantage point of contrastive linguistics. The research can also claim a methodological contribution, manifested in the operationalization procedure proposed for the classification of constructions with relevance to taxis mode and logico-semantic relationships. Further research could make use of the proposed criteria to study translations and non-translations from other genres and registers, which could potentially lead to the identification of genre/register-based quantitative profiles of patterns, providing benchmark data for research in translation studies and contrastive linguistics.

About the author(s)

Waleed Othman received his PhD in Translation Studies from the University of Birmingham and is currently assistant professor at the University of Petra, Jordan. His research interests include register-/genre-based translation studies, corpus-based translation studies, discourse analysis, and systemic functional linguistics. His most recent publications are in the Journal of Functional Linguistics (2020) and Meta (2020).

Dima Al Qutob received her PhD in linguistics from the University of Exeter and is currently assistant professor at the University of Petra, Jordan. Her research interests include translation studies and sociolinguistics.

Email: [please login or register to view author's email address]

•  Previous article 
•  Next article