Exploring Intricacies in English Passive Construction Translation in Research Articles’ Abstracts by Arab Author-Translators

Self-translation of academic texts has received little attention thus far in literature, particularly in terms of how cross-linguistic features are rendered into target language. This study undertakes to examine the various linguistic strategies of rendering English passive structures by Arab academics when they translate their research articles’ abstracts (RAAs) into Arabic. Fifty-one English abstracts with their Arabic translations were collected from Languages, Humanities and Social Sciences journals published in different Arab universities. To analyze the collected data, #LancsBox 4.5 Lancaster University corpus tool was used to identify the English passive structures (208 instances) and to analyze some of their Arabic translations. The most interesting finding is that the Arabic linguistic alternatives diverge from the English passive structures and they include the use of Arabic active verbs, Arabic periphrastic constructions, Arabic passive verbs, and Arabic verbal nouns. The results cast a new light on the use of periphrastic structures. While the literature usually refers to the occurrence of this structure in journalistic Arabic, this study provides evidence of its occurrence in academic texts in almost 22% of the corpus. The increasing use of this strategy is a feature of Modern Standard Arabic as discovered in some corpora. The study supported the argument that Arabic does not avoid passive verb forms in academic discourse but expresses them by using stylistically different strategies.


Introduction
Studies on self-translation were integrated into Translation Studies in the second half of the 20th century (Panichelli-Batalla, 2015). Much of the existing research on self-translation pays particular attention to how it works in literature (Cordingley, 2013). In linguistics research, however, it is very often neglected (Pinto, 2012). This category of translation has recently gained popularity among the non-native English researchers who produce research in English and are required by local journals to translate the abstracts of their manuscripts into their mother tongue. These researchers serve as mediators between "the original text and the readers of the translation" (Chan, 2016, p. 152). They are also considered as "interculture[s]" functioning at the locus of textual intersection and divergence (Hokenson & Munson, 2014). Given the potential differences between distant languages at all levels, it is expected that the translated abstracts are likely to diverge at some level from the source texts written by the same authors commonly known as author-translators.
Authors who write in a language other than their mother tongue are sometimes known as "translingual or multilingual" (Kellman, 2003;Peñalver, 2010, p. 193). Their products in the target language (translators' mother tongue) represent ideal translation because "the author . . . knows his creation as nobody else and has the authority to allow himself shifts in the translation which might not have been allowed by another translator" (Maklakova et al., 2017(Maklakova et al., , p. 1265. It was argued by Grutman & Van Bolderen (2014, p. 329) that ". . . a self-translator legally, intellectually, and morally owns the source text and can have the impression she is less bound by it than another translator." The current study is an attempt to explore the strategies employed by Arab researchers who write research in English and are mandated by editorial policies to translate the abstracts of their manuscripts into Arabic as a pre-requisite for publication in journals published by Arab universities. These journals seek to enhance the international visibility of their publications by accepting to publish some articles in English. In addition, they require the authors to render their English abstracts into Arabic in order to attract local readership. In other words, the self-translation of RAAs contributes to enlarging the size of the journals' academic community (Pezzini, 2003). Most of these journals are interdisciplinary within the fields of humanities and social sciences. The particular concern of this study is exploring how English passive structures are rendered into contemporary academic Arabic. Within self-translation perspective at such microlevel, no study has been traced in the literature that has investigated the use of passive constructions in the academic discourse of research articles abstracts across English and Arabic.

Literature Review
Translation in modern times plays a crucial role in the dissemination of academic texts. This section presents a review of the literature on self-translation of academic discourse and on passive translation, with special focus on studies pertaining to translation of English passive construction into Arabic as used in research articles abstracts. Montgomery (2009, p. 9) refers to this type of translation as "translating one's own writing . . . for the purpose of publication." This view is supported by Pinto (2012), who considers self-translation as an activity commonly used in non-literary writings by scholars who translate their articles to different languages for publication requirements. There is scarcity of publications on self-translation of academic texts and it is considered as nonconventional by some experts. In the same vein, Pisanski Peterlin (2019) argues that the practice of self-translation in academic settings is not sufficiently explored. Her paper addresses the experiences and attitudes of Slovene authortranslators of academic discourse, the key issues they encounter and the contrastive, intercultural, and pedagogical implications. The importance of this study lies in the significant finding that the boundaries between the author and the translator, the original and the translation, the source and the target language texts which are so central to self-translation of academic texts are completely blurred. Alharbi and Swales (2011) published a relevant paper in which they described the similarities and differences between a corpus of Arabic and English abstracts that were written by author-translators. The study found differences between the two languages with more rhetorical and metaphorical features in some Arabic texts and more attention to background information in the English abstracts. Though self-translation was methodologically emphasized by the authors, it was not scrutinized as an independent variable in the study. Escudero and Swales (2011, p. 49) traced convergence and divergence in pairs of Spanish and English research articles abstracts and found that divergent patterns were attributed to collocational differences and beliefs about naturality in Spanish. Interestingly, convergence was found in "the degree to which the text is given agency." In an attempt to highlight the difference between self-translation and other mainstream types of translation, Santoyo (2010) notes that self-translators are free to create a different version of the original text. Since selftranslation takes place between languages with asymmetric relationship in academic contexts (English vs. other languages), it is argued, therefore, that a new type of diglossia is created where languages other than English are increasingly becoming "minority languages in academic settings" (Calaresu, 2011;Pisanski Peterlin, 2019, p. 849).
The linguistic strategies of translating passive construction in general discourse from English into Arabic have received considerable attention in the literature. For instance, Khafaji (1996) investigated the various linguistic strategies an English-Arabic translator employs to translate English passive, and studied the conditioning factors that determine the selection of these strategies. His pioneering study found that passivized verbs in the Arabic text are only 25% of those found in the English source text. Similarly, Farghal and Al-Shorafat (1996) probed into the linguistic strategies and resources which translators from English into Arabic employ when they encounter passive utterances. The study interestingly revealed that English passive constructions are predominantly structure-based, whereas Arabic passive structures are mostly semantics-based and thus utilize a variety of strategies. Khalil (1993) evaluated Arab students' translations of English agentive sentences and found that they do literal translation very frequently, which is not acceptable in Classical Arabic. According to the author, Classical Arabic should be the target language in translation courses, a finding that may not be valid in today's Modern Standard Arabic. On the other hand, Al-Raba'a' (2013) addressed the argument that since translation is a manifestation of language contact between Arabic and English, it can stimulate convergence. The study confirmed that the Arabic speaking monolinguals who learn English as a foreign language use the by-phrase along with the agent in the Arabic passive structures because of the English influence on Arabic. Al-Ali and Alliheibi (2015) attempted to explore whether Arab students would translate the English passive constructions into their corresponding Arabic passive structures or they would employ other alternatives. The detailed findings of their study revealed that Arab student-translators use a variety of processes to realize the transformation.
This survey of the literature on self-translation in academic discourse in general and on the translation of passive from English into Arabic in particular has demonstrated that up to now, far too little attention has been paid to the features of self-translation of Arab English researchers' academic discourse (Al-Ali & Alliheibi, 2015;Alharbi & Swales, 2011). The present study is a modest contribution in this direction, which aims to explore the rhetorical and linguistic preferences used when Arab author-translators render English passive structures into Arabic in research articles abstracts.

Corpus, Method, and Procedures
It can be argued that English passive voice should be translated into passive because it is a semantic category with fixed pragmatic functions that should be carried over to any language in an equal manner. However, this hypothesis is not valid in the case of English-to-Arabic self-translation of academic discourse as this study demonstrates. The current study attempts to investigate and interpret this phenomenon. The corpus of this research consisted of 51 English Research Article Abstracts (RAAs) and their 51 Arabic versions translated by the same articles' authors. To maintain the homogeneity of the discipline, the samples came from language studies including applied linguistics (n = 39), Linguistics (n = 6), literature (n = 3), and translation (n = 3). It was ensured that all writers of these articles are Arabic native speakers by referring to the universities they belong to as academics. These author-translators come from different Arab countries and not necessarily from countries where the source journals are published. The editors receive contributions from different researchers in the Arab World.
The articles that comprise the corpus of this study were published between 2011 and 2018 by different journals affiliated to different Arab universities. The countries these journals and universities are based include Iraq, Jordan, Palestine, Saudi Arabia, UAE, and Yemen. The major data sources are

Journal of the College of Languages (Iraq), Journal of Basra Researches for Human Sciences (Iraq), Jordan Journal of Modern Languages and Literature (Jordan), King Khalid University Journal of Humanities (Saudi Arabia), Umm Al-Qura University Journal of Language Sciences and Literature (Saudi Arabia), Journal of Al-Qadisiya in Arts and Educational Sciences (Iraq), The Arab Journal for Quality Assurance in Higher Education (Yemen), and University of Sharjah Journal for Humanities and Social Sciences (UAE).
In their publication policy, the journals stipulate the submission of Arabic abstracts beside the one in the language of the manuscript. For instance, Jordan Journal of Modern Languages & Literature states in its publication ethics and malpractice statement that "All manuscripts submitted should be accompanied by two abstracts and keywords: one in English and the other in Arabic in addition to the abstract in the language of the paper" (http://journals.yu.edu. jo/jjmll/Publication%20Ethics.html).
In order to explore how Arab author-translators render the passive structures in their English abstracts into their mother tongue, the researcher adopted the Critical Contrastive Rhetoric (henceforth CCR) approach (Kubota & Lehner, 2004). This approach qualitatively encourages reflection on the preferred "discourse patterns of the target language and [evaluates] how these practices might reinforce cultural binaries and assimilation." It also views writing as a social practice involving human agency rather than merely a reflection of cultural thought patterns. CCR emphasizes the role of the evolution of rhetoric and encourages an investigation of how the rhetoric of a specific language assumes new styles as a result of internal and external factors.
The situation in this study is quite novel. The authortranslators are not rendering an L1 text into L2. They are rather translating an academic text, which they have created in their second language (English) into their mother tongue (Arabic). In this case, the source language is English and Arabic is the target language.
The analysis started by tracing all instances of passive structures in the English abstracts. #LancsBox 4.5 corpus analysis tool of Lancaster University was used for this purpose to avoid the trouble of manual identification of the required data. This software is free for non-commercial use (http://corpora.lancs.ac.uk/lancsbox/docs/pdf/LancsBox _4.5_manual.pdf). Using "smart searches" feature of this tool, complex linguistic patterns can easily be found as illustrated in Figure 1. The search for passive structures resulted in 208 instances in 47 out of the 51 abstracts as shown in the shot of search result below. Interestingly, Biber et al. (1999) as cited in Hyland (2005) gave the figure of 18.5 cases per thousand words for passive voice constructions in published academic writing which almost coincides with the corpus of the present study (18.25).
Finding the Arabic equivalents for all English passive constructions was performed manually because #LancsBox 4.5 does not support the same "smart searches" in Arabic corpora. Guided by the search results of the English abstracts corpus, the Arabic translation of each instance was identified and the results of the adopted strategies were classified into five categories. The same corpus tool was used to examine the frequency of one grammaticalized item, the controversial "tamma" which means (completed or finished) that occurred frequently in the Arabic corpus as will be shown in the results section below.

Results
The use of passive voice is a recurrent feature in academic discourse. Pérez-Llantada (2012, p. 66) asserts, "Passive constructions . . . ascribe greater emphasis to the research topic/field than to the researchers themselves, hence contributing to impersonal, drawn-upon-facts reporting of research outcomes." It is also one of the characteristics of abstracts as has been verified in the literature (Pezzini, 2003). However, Arabic seems to behave differently with regard to the use of passive voice in RAAs. El-Yasin (1996) observes that Arabic and English do not exhibit parallel behavior. Arabic tends to use less passive than English does. Rosenhouse (1988, p. 93) maintains, "Arabic is known to avoid passive verb forms . . . English is known to make abundant use of the passive verb, especially in scientific texts." The findings of the present study do not totally agree with this argument. Arabic does not completely avoid the use of passive verb forms. Yet, it does not use it as abundantly as English does in scientific discourse. Therefore, for the English-Arabic translators, this is problematic since they would have to encounter the task of converting a big number of passive constructions in their English source texts into other linguistic alternatives for the purpose of producing appropriate Arabic texts (Khafaji, 1996).
The results obtained from the corpus analysis reveal that the English passive constructions were rendered by employing different alternatives in Arabic by the author-translators in this study. The strategies followed are (a) rendering the passive structures into active in (31%) of the data, (b) use of periphrastic structure with tamma (22%), (c) constructing the passive voice by verb vowel change (19%), (d) nominalization (4%), and (e) creating a new text (23%).

English Passive Voice Rendered to Arabic Active Voice
The translated corpus data exhibited that the author-translators rendered 65 instances (31%) of the English agentless passive instances into Arabic active structures. This strategy involves the reconstruction of an agent from the context for these produced active sentences perhaps to avoid the use of passive structures; or transforming the passive structure into active by transposing "the English passive verbs into Arabic active structure" (Al-Ali & Alliheibi, 2015). The examples below illustrate this alternative. Table 1 that the rhetoric in the two data sets is not the same. The use of agentless passive structures in English abstracts is quite natural and lends an impersonal tone. Arabic academic rhetoric tends to avoid using passive voice abundantly and instead writers employ alternative strategies. In examples 1, 3, and 5, the equivalent structures are active sentences where the agent is explicitly stated which is albaHithan "the researchers," whereas in examples 3 and 6 the agent of the produced equivalent sentences is addirasah "the study." Therefore, the implied agent in English passive voice is expressed explicitly in Arabic rhetoric. Reference to the author-translator's scholarly activities as agents that reveal findings take different terms. The words baHith/baHithan meaning "researcher/ two researchers" or na pronoun suffixed to action verbs to stand for the agent are used 27 times; and the terms diraasah "study" and baHth "research" are used 21 times in this rendering category. Rest of examples in this strategy refer to the survey, ainah "sample," ikhtibar "test," and nataij "results" as the reconstructed agents in the respective Arabic active translations. In example 4, "The addressee is invited to reconstruct the meaning . . ." the agent is implied which is "the poet." However, in the Arabic respective translation, the sentence is reconstructed to become active with an explicit stating of the agent ashshaer "the poet."

Use of Periphrastic Structure With tamma
The second most frequent strategy used by author-translators involves rendering passive constructions into a periphrastic structure consisting of the grammaticalized verb tamma "expressing tense and a sense of action completion" followed by masdar "verbal noun." This alternative option is used in 22% of the corpus. According to Al-Raba'a' (2013), tamma is used to paraphrase passive structures. Abdalla (2018) asserts that tamma is used as a substitute to replace a passive verb in Modern Standard Arabic. The construction of tamma + verbal noun periphrasis is used instead of an internally voweled passive to report the completion of durative or iterative processes where the focus is on the result of the process rather than on the process itself (Holes, 2004). The examples in Table 2 illustrate how author-translators in the Arabic corpus of this study use this structure in their self-translation.
As Table 2 illustrates, all the seven examples from the English corpus data were translated into periphrastic structures. Example 1 "This test has been administered" which follows the English passive rhetoric structure is self-translated as tamma taTbeeq al ikhtibaar ala. tamma functions as an auxiliary verb which indicates the past tense and has the meaning of completion of the following action which is taTbeeq al ikhtibaar that means "the administration of the test." The produced translation, though it sounds passive in meaning, is an active voice structure, according to Arabic grammar, which consists of the delexicalized verb tamma followed by the grammatical subject which is the verbal noun. The same explanation applies to the remaining examples in the corpus. #LancsBox 4.5 corpus analysis software was used for easy identification of this structure in the Arabic translated corpus (See figure 2). When the lemma tamma was searched, the following results were generated. All instances of tamma + verbal noun periphrasis were found to be translations of English passive structures in the source texts.
From the screenshot above, we can see how the search result is presented in concordance lines with the searched word highlighted in the center and some context to the left and right of the target word. The results obtained from this figure show the verb tamma with the verbal nouns it collocates with to the left. It is obvious that all the verbal nouns that follow the verb tamma express a research-related process that is completed such as taTbeeq "application," tadrees "teaching," taTweer "developing" taQseem "dividing," ikhtibar "testing," etc.

English Passive Rendered Into Arabic Passive by Verb Internal Vowel Change
Arabic is a Semitic language with a morphologically rich verbal system. Khafaji (1996, p. 22) states "The passive is performed in Arabic verbs by introducing apophonic vowel changes in the active basic verb forms or by the affixation of certain morphemes." English passive, on the other hand, is formed by using one form of the auxiliary verb "be" followed by the past participle form of the main verb. This English passive translation strategy presented in this section demonstrates that not all English passive voice structures can be translated into Arabic passive constructions. Only 19% of the English passive corpus in the present study was rendered ‫ذاتها‬ "The study also shows that most of the errors in English verb form refers to intra-lingual reasons." 4. The addressee is invited to reconstruct the meaning himself with the help of the context ‫يدعو‬ ‫فإنه‬ ‫الضمنى‬ ‫المعنى‬ ‫الشاعر‬ ‫يستخدم‬ ‫حين‬ ‫بنفسه‬ ‫المعنى‬ ‫إنشاء‬ ‫إعادة‬ ‫إلى‬ ‫المتلقى‬ "When the poet uses the implied meaning, he/she invites the addressee to reconstruct the meaning himself/herself." 5. This is followed by the presentation of some important particles.
‫االدوات.‬ ‫هذه‬ ‫اهم‬ ‫استعرضنا‬ ‫ثم‬ "Then we presented the most important of these particles." 6. Good readers were found to be high achievers in the second language.

‫لديهم‬ ‫الذين‬ ‫الطلبة‬ ‫أن‬ ‫إلى‬ ‫الدراسة‬ ‫توصلت‬ ‫كما‬ ‫اللغة‬ ‫في‬ ‫أفضل‬ ‫نتائج‬ ‫يحققون‬ ‫القراءة‬ ‫على‬ ‫جيدة‬ ‫قدرة‬
‫الثانية‬ "The study found that good readers are high achievers in the second language." into Arabic passive by following the traditional rules of Arabic grammar. Table 3 provides examples of English passive rendered by author-translators into Arabic passive in almost only one fifth of the data while the remaining percentage of the translation that is almost 79% is non-passive. As shown in Table 3, an English passive verb structure that consists of two or three words ((have) + be + past participle verb form) is rendered into one Arabic passive verb formed by internal vowel change. For example, the English passive verb construction has been classified in sentence 1 is translated into Arabic as yusannaf, a derivation from the active form yusannif; in sentence 6, have been employed is translated into Arabic as ustukhdimat, a derivation from the active form istakhdama; and the English passive structure was applied is rendered into Tubbiqat, which is a derivation from the active verb Tabbaqa. In all these examples from the  Arabic translation corpus of abstracts, passivization is created by altering the vowels of the active verb forms.

English Passive Verbs Translated Into Verbless Arabic Structures With a Verbal Noun
This Arabic passive translation strategy is called "nominalization" and it is used in only 4% of the corpus. The English passive verb structure is rendered into a verbal noun which is similar in function to English gerund followed or preceded by the patient experiencing the action. Consider the following examples in Table 4 for some English passive instances with their respective self-translation into Arabic. Nominalization strategy as shown in the above examples is similar to the strategy of periphrastic construction presented earlier in terms of the use of verbal nouns to replace the passive structure. Yet, this strategy does not use the grammaticalized word tamma as part of the translation because the utterance does not refer to an action that is completed but to cognitive activities that may express reference (example 1), prediction (example 2), possibility (example 6), need (example 5), contrast (example 1), etc. This structure is very common in Arabic (Fakhri, 2012) and it is mostly verbless and can be accounted for by the tendency of Modern Standard Arabic to avoid overuse of passive which might make the processing of meaning by readers of academic discourse quite complicated.

Non-Translation or Creating a New Text
It was difficult to find equivalents for 23% (47 instances) of the English passive verbs in the Arabic corpus. The only

Source text
Self-translation 14. The speech act of apology has been classified . . . ‫لالعتذار‬ ‫الكالمي‬ ‫الفعل‬ ‫ويصنف‬ "has been classified the speech act of apology" 15. A test has been constructed.
‫واالقتراحات‬ ‫التوصيات‬ ‫بعض‬ ‫قدمت‬ "have been put forward suggestions" 18. A qualitative method is used to examine to what extent . . . ‫لمعرفة‬ ‫نوعية‬ ‫طريقة‬ ‫استخدمت‬ "is used a qualitative method to identify . . ." 19. Two methodological tools have been employed.
‫للبحث‬ ‫المنهجية‬ ‫األدوات‬ ‫من‬ ‫اثنتان‬ ‫استخدمت‬ "have been employed two methodological tools" 20. The experiment was applied during the first term of 2017 to 2018.
‫األساسية‬ ‫وظيفتها‬ ‫عن‬ ‫فضال‬ ‫اللغة،‬ ‫إستخدام‬ ‫يمكن‬ ‫ضمنيا‬ ‫المعنى‬ ‫عن‬ ‫للتعبير‬ ‫التواصلية،‬ "Using language is possible to imply . . ." interpretation for this gap is that author-translators did not render this portion of the research data in the source language (English). Author-translators probably think they have the freedom of not being strictly loyal to the source text structure as it is their own product and can render it by creating a new text in the target language. Panichelli-Batalla (2015, p. 17) contends, "As with many self-translations, the author makes full use of his freedom to adapt the 'original' text." Similarly, Santoyo (2010) argues that self-translators enjoy the freedom of reconstructing a different version of the original text. For example, in the study of Mohammed and Kadum (2016, p. 1), the passive statements "Several TV interviews are selected and examined on the discourse level, Certain linguistic markers have been used by male speakers, Different aspects have been utilized by female speakers." are missing in the Arabic translation. This observation led the researcher to compare the English source text and the Arabic target text (See Table 5). I found them quite different in form and content. The English abstract is brief and consists of three moves in three sentences with 63 words only. The first sentence presents the aim of the study; the second one briefly presents the method, and the third sentence concisely states the findings. The Arabic translation, however, has 247 words and consists of four moves, theoretical introduction in a full paragraph, introduction, aim, method, and findings. This comparison answered the question why one passive sentence in the source text is invisible in the target text. The author-translator created a new text. In some cases, author translators jump some information for several possible reasons. Consider the following example from the corpus of paired abstracts written by Al-Asadi (2018).
The source text, which is an excerpt of 75 words taken from the English abstract, consists of five sentences (S1, S2, S3, S4, and S5). S3 contains a passive structure. On the other hand, the Arabic version of the extract consists of three sentences that contain only 40 words. These three sentences are the translation of S1, S2, and S5. English S3, which contains a passive construction, and S4 are not rendered in the target text. This is the reason why some gaps in the translation of passive structures exist. It is interesting to notice that this example is in contrast with the previous ones in terms of the length of source and target texts. The variable of self-translation, which allows the writer to decide what to include and what to exclude in the target text, is likely the reason.

Hedging With Passive Structure
A possible reason behind the gap between English passive structures and their Arabic equivalents in self-translation of research articles abstracts is the use of hedging in English and its lack in the Arabic data. There is only one example in the current corpus that can support this claim which is not sufficient to make a generalization. Yet, it can help in interpreting the present data analysis. Hedges contribute to the negotiation of a successful reader-writer relationship and are "a significant resource for anticipating a reader's possible rejection of a proposition and for presenting claims with precision and caution" (Hyland, 2005). Consider the example found in the English corpus.
28. Arabic discourse has been argued to be as hypotactic as English.
This sentence was translated as follows in the Arabic corpus.

‫اإلنجليزية.‬ ‫مثل‬ ‫مثلها‬ ‫الجمل‬ ‫بين‬ ‫اإلسناد‬ ‫على‬ ‫العربية‬ ‫تعتمد‬
Literal translation: "Arabic depends on hypotaxis between sentences in the same way as English does." The sentence in the source corpus uses the passive construction to soften the proposition, make it cautious, and impersonalize the statement. The Arabic corresponding translation does not hedge and it is stated as an established fact.

Discussion
In reviewing the literature, little data is found on contrastive rhetoric studies that explore the linguistic strategies used in the self-translation of passive construction in the English research articles abstracts into Arabic by Arab language

Source text
Self-translation 27. Based on the relevant existing literature, three approaches are still dominant and influential (S1). They are: product approaches, process approaches, and genre approaches (S2). The current writing instruction to teach EFL writing at the Arab tertiary level is characteristically inspired by product approaches (S3). Such approaches have come under serious scrutiny for a number of reasons (S4). The purpose of this paper is to propose the introduction of an integrated process-genre approach as an alternative one in this context (S5).
researchers. The results of this study indicate that translation preferences seem unexpected because they do not agree with the objective of any translation process, which is rendering the meaning of a given text from one language to another. Although passivity is "an important component of the semantic structure of any passive sentence" (Khafaji, 1996, p. 29), it is used in only 19% of the Arabic sub-corpus in this research most likely for pragmatic reasons. Arab authortranslators who translated English passive verbs by Arabic passive verbs in the current study do not follow this strategy with conscious guidance from the typical rule that passive is used when the agent is unknown. In many instances of such preference in the corpus, the agent is obvious, namely the researcher/s. Interestingly, the percentage of this pattern coincides with (Al-Ali & Alliheibi, 2015) study on studenttranslators. Such consistency in results may lead to a cautious generalization that the translation of academic English passive verbs into modern Arabic academic prose, almost only one fifth of the preferences will be the rendering of English passive structures into Arabic passive structures by active verb vowel change.
The translation of English passive verb into Arabic active verb came first in frequency (31%) probably because the agent in empirical research is known "the researcher/s" and in Arabic, passive constructions are typically used when the agent is unknown. Because the abstracts' target audience most likely cannot read the English text, Arab author-translators seem to want to engage them and help them understand the arguments easily. Therefore, they create an agent from the context as an engagement strategy. According to (Hyland, 2001, p. 550), "A central aspect of the writer-reader dialogue involves careful interpersonal negotiations" and this dialog can only be positive if the writer uses linguistic preferences which are familiar to Arabic readers.
Consistent with the literature, this research found that the use of periphrastic structure with the auxiliary verb tamma to replace Arabic vowel melody passive structures is gaining high frequency in Modern Standard Arabic. Although many previous studies emphasize the use of this structure in media and journalistic Arabic (Abdalla, 2018;Holes, 2004;Lundin, 2013;Parkinson, 2011), the contribution of this study is the provision of substantial evidence of the extension of its use to the academic discourse of published research. The findings of this study prove that there is a tendency by authortranslators to use this active grammatical pattern of "tamma or any of its variants + verbal noun" to smoothly engage their readers by using a pattern of high frequency in modern Arabic prose. This phenomenon is supported by a corpusbased query of (arTen Ten) on Sketch Engine corpus manager and text analysis software (https://www.sketchengine. eu/artenten-arabic-corpus/). This Modern Standard Arabic corpus contains 7,475,624,779 words. When tamma was searched, the results showed 4,517,777-word frequency (542.87 per million) in context (See figure 3).
One interesting finding is the creation of a new target text by author-translators. A possible explanation for this has already attributed this linguistic behavior to the influence of self-translation variable and the freedom an author-translator enjoys. However, this freedom may not apply to all cases of untranslated passive structures in this research. Therefore, an alternative interpretation might be writers' expectations of the editors in the Arabic Humanities and Social Sciences research and publication context. Alhuqbani (2013, p. 379) believes, "Arabic journals' publication policy leaves the writing of abstracts at the researchers' disposal." Therefore, Arab author-translators are likely to take advantage of this drawback to structure the translated abstracts in an unconventional manner leaving some details in the source texts untranslated including some passive structures as illustrated in the results section above.

Conclusion
The purpose of the current study was to examine strategies of self-translation of passive structures by Arab academics from English into Arabic in the genre of the academic article abstract from a contrastive rhetoric perspective. The results have revealed important cross-linguistic differences between the paired texts produced by same authors. The 208 instances of structure-based passive voice that comprised the source English corpus were rendered by using four major semantics-based linguistic alternatives, namely active voice, periphrastic structures with tamma, verbal noun, and passive structures constructed by Arabic verb vowel change which represent almost 76% of the corpus. The remaining 24% is divided between the creation of a new text by author-translators (23%), and the use of certainty option to replace English hedging. This study argues that the rhetorical differences in rendering English passivity into academic Arabic, like allnatural languages, is guided by communicative considerations and the avoidance of meaning loss of the translated message since translation is viewed as an act of communication. These findings contribute in several ways to our understanding of how Modern Standard Arabic expresses English passive voice and provide a basis for revisiting the pedagogy of passive in traditional Arabic grammar books. The current books of Arabic as a first, second, or foreign language still introduce passive structurally rather than communicatively. Arabic grammar curriculum designing should be informed by corpus-based findings of studies that examine passive voice as it occurs in natural discourse.

Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
The author received no financial support for the research, authorship, and/or publication of this article.