A Corpus-Based Analysis of Deontic Modality of Obligation and Prohibition in Arabic/English Constitutions

. It is argued that legal language should be formal, precise and clear to avoid ambiguity and/ or misunderstanding. As rights and duties are communicated through modals, clarity and precision in drafting and translating them is crucial. Otherwise, there is a possibility of conveying loose messages in the source text or different and/or inconsistent messages in the target text. However, the drafting of Arabic modal expressions does not follow clear guidelines, and their translation differs from one translator to another. This paper investigates how deontic modality of obligation and prohibition is used in The Leeds Annotated Parallel Corpus of Arabic-English Constitutions in comparison to The Leeds Monolingual Corpus of English Constitutions . More specifically, the paper presents a classification of these modal ex pressions and investigates the different lexical variants expressed in a Corpus of Arabic Constitutions. The paper uses corpus-based tools to analyse the different lexical forms used for deontic modality of obligation and prohibition in Arabic and how they are rendered into English. Results of such analysis are compared to a non-translated Corpus of English Constitutions to find out whether the deontic meaning of the modals is comparable to the set of deontic modals used in the constitutions originally drafted in English. The corpus-based analysis gave a detailed classification of a variety of modal expressions used in the Arabic Corpus. It also showed that the translation of deontic modals of obligation and prohibition from Arabic into English is influenced by the source text lexical variations; however, the corpus techniques employed in the study managed to capture some comparable modals in both corpora.


Introduction
Many Arabic linguists who investigated the Arabic modal system (Wright 1967, Suleiman 1999, Badran 2001, Badawi et al. 2004, Abdel-Fattah 2005, Ryding 2005, Eades 2011, El-Farahaty 2015, among others) concluded that Arabic does not have a standard modal system the same way English does. This is due to the complexity of the formal grammatical structure of the modal system, in addition to several features which are inherent in the modal system in general such as indeterminacy and ambiguity and context-dependent semantic meaning (Leech & Coats 1980, von Fintel 2006, and Abdel-Fattah 2005 3 . Translation of deontic modals in the legal domain has been the subject of interest of a number of researchers including Biel (2014), Jaskot & Wiltos (2017) who investigated English and Polish and Satthachai & Kenny (2019) who investigated English and Thai. Translation of English deontic modals into Arabic in the legal genre has received little attention by a few researchers including (El-Farahaty 2015), but translating them into English has received almost no attention as yet. One crucial issue noted in Abdel-Fattah (2005) is that translating Arabic modals follows the stylistic preferences of the translator, a matter that will have a direct impact on the accuracy of the modal meaning and the consistency of translating them. Also, Abdel-Fattah's (2005) and El-Farahaty's (2015) studies have been applied to a small scale corpus, and due to the absence of authentic corpora, none of these or previous studies identified a well-defined list of deontic modal expressions and their equivalents in English. Hence this paper aims to fill this research gap in legal discourse in general and in legislative drafting in particular. More specifically, this paper aims to investigate how rights and commitments are expressed through deontic modality of obligation in translated and non-translated corpora (i.e. The Leeds Annotated Parallel Corpus of Arabic-English Constitutions, shortened to Parallel English Corpus (PEC) and The Leeds Monolingual Corpus of English Constitutions (shortened to Monolingual English Corpus (MEC).
To achieve this aim, the following research questions will be investigated: 1) How are deontic modals of obligation and prohibition expressed in the Arabic Corpus? 2) What are the translation techniques used to convey the deontic meaning of these modals in the Parallel English Corpus (PEC)? 3) How can deontic modality in the translated parallel corpus be compared to the Monolingual English Corpus (MEC)?
To answer the above research questions, a corpus-based investigation will be undertaken that will analyse the frequency of different variants of Arabic deontic modals in the corpus of Arabic constitutions. The parallel corpus investigation analyses the different techniques of translating this set of deontic modals into English to test the precision of translating them. Using corpus tools and techniques in data analysis, we can identify salient features which may not be evident by casual observation.
Results of such analysis will then be compared to a non-translated corpus of English constitutions to check if the deontic meaning of the modals adheres to the set of deontic modals used in the constitutions originally drafted in English. The paper will be structured as follows. Section two briefly discusses the use of modals in Arabic legal discourse/legislations and gives brief background information about the translation of Arabic constitutions into English. Section three explains the concept of deontic modality in general and in English and Arabic legal discourse in particular. Section four introduces the data and methods used. Section five discusses the corpusbased analysis of deontic modals in the (AC) and the translation of these modals in the (PEC). Section six provides the outcomes of a comparison to the comparative MEC followed by the conclusion and ideas for further research.

Drafting Legislations in the Arab World
Legal concepts are commonly clear and meaningful, but words are the carrier of meaning, and the linguistic choice complements the intended meaning of a particular legal concept (Bani Dhiyab 2011: 14). Hence, "legislative drafting manuals can contribute to coherent and consistent law drafting" (The Organisation for Economic Co-operation and Development/OECD). Drafting of legislative texts, constitutions being one category of them (Cao 2007) has been widely discussed by scholars and law professionals, law institutes, government bodies from different parts of the Arab world. Sabra 2008, Bani Dhiyab 2011, Al-Ganainy 2012, Suteu & Draji 2015, Al-Mu'min 2017, 2000, Sadun & Butshasha 2017, among others, gave general guidelines about language drafting that seem to echo general rules of drafting western legislations such as clarity, preciseness, accessibility, comprehensibility, legal certainty (European Union: Legislative Drafting Guide). Although these authors and institutions write the manuals/rules related to their national legal systems, they all agree on similar legislative drafting pitfalls such as vagueness and complexity; inconsistent use and lack of rules on punctuation marks; unjustified repetition which leads to confusion and result in legal gaps; or even simple linguistic errors which existed in the drafts and were not corrected in the final printed version of a certain law as well as legal errors (in the form of absence of a legal term/concept which will impact on the legality of the ruling) that need to be corrected.
Drafting traditions in Arab countries could differ from a country to another. For instance, the legal system in Saudi Arabia is based on Islamic Law, mirroring the language and legislation of the early centuries of Islam. On the other hand, the laws in Egypt, Syria and Lebanon are based on Napoleonic civil law; there is no one uniform method of drafting Arabic constitutions as it is unique to each country and each era within this country. Therefore, some guidelines have been proposed to ensure some kind of uniformity of drafting Arabic legislations. For instance, Al-Ganainy (2012: 6-7) and Bani Dhiyab (2011: 120-7) highlighted some principles about the language and style of drafting Arabic legislations. These include: making sure that the structure of the legislation is clear and all relevant parts are included in their appropriate place, drafting clear simple short sentences which identify who does what to whom; sentences should be joined by punctuation marks; using the conjunctions 'and'/ 'or' carefully and appropriately; avoiding vagueness and the use of synonyms of the same lexical term; using the active rather than the passive voice; using positive rather than negative sentences; using singular to mean plural and masculine to mean feminine unless there is a need to mention the feminine; using the imperfect (simple present tense) rather than future to convey obligation; using the Gregorian calendar unless mentioning the Hijri calendar is required (as in Saudi Arabia). All these rules were stressed by The OECD 4 in addition to stressing the need to avoid gender-specific terms in drafting legislations. None of these manuals, however, included details about deontic modals and the expressions used to express permissibility and obligation. Sabra (2008: 92), referred to a few examples about modality as in the case of using the positive structure of modality rather than the negative when possible so a drafter should not write: Any person below the age of 18 is not allowed to vote But rather the drafter should write: Everyone who has reached the age of 18 or over is allowed to vote 4 This report (issued in August 2019) "looks at the role of legislative drafting manuals in enhancing legislative and regulatory quality. It presents experiences and good practices from both OECD and MENA countries in the preparation and application of these manuals" (OECD: online). Although it focuses on Egypt, its recommendations could be applicable to other countries in the MENA region. Another assessment report of the Palestinian legislative drafting manuals was issued by the OECD in 2011. For more information, please refer to this document: https://www.oecd.org/mena/governance/50402734.pdf.
Consequently, there should be some standards in the constitution drafting process which guarantee the involvement of all parties in the drafting process, be them "civil society groups, political parties, religious leaders, workers' unions, and the general public to engage in meaningful deliberation" (Moustafa 2012: 3). The only available manual which was published by Interpeace in 2011, then issued in its Arabic form in 2012 is the Constitution-making and Reform -Options for the Process (Brandt et al. 2011). The handbook covers all the possible processes involved in drafting a constitution, structure and language is one of them. While some of the Arabic constitutions drafted or amended after the Arab Spring may have benefited from this manual, some others must not have followed it. In all of the manuals referred to above, there is no reference to the specifics of writing. For example, there are no guidelines for explaining the appropriate set of deontic modal expressions to be used.
Legal professionals have criticised the drafting and the language of some Arabic constitutions and their translations. An example to give here is the Iraqi Constitution: "The language used was incredibly weak and problematic, raising concerns about the ability of the body that drafted or translated the constitution" (Jawad 2013: 18). He (ibid.) comments on the use of modality in article 46 which starts with the words "La yakoon (not to be), whereas the correct legal term is La yajoze (may not)" (see also Sabra 2008: 92-3). Allawi, a senior Iraqi politician, noted that "the constitution was utterly alien, in construction and phraseology, from the Arabic language and the Iraqi experience" (Allawi 2007: 222). The possible justifications one can give here about the above is the many challenges that faced the drafting of this Constitution, the heated debates and the turmoil that surrounded it and the lack of consensus on different aspects of it (e.g. women rights issues). (Iraq Democracy Information Centre 2006: 1-2).

Drafting legislation in the EU: the modal auxiliaries
Most English-speaking countries and European countries have used clear guidelines about modality, which are discussed in detail in Krapivkina (2017ː 312-3). For example, Indiana Drafting Manual gives the following specific details about the concepts of modalsː To create a right, say is entitled to. To create discretionary authority, say may. To create a duty, say shall To create a condition precedent, say must To negate a right, say is not entitled to To negate discretionary authority, say may not To negate a duty or a mere condition precedent, say is not required to. To create a duty not to act, say shall not (cf. Krapivkina) Deontic modals, 'shall' and 'must' are heavily discussed in the literature of legal drafting and legal translation. According to Biel (2014: 341) and Garzone (2013: 70), "the English Style Guide (2012: 35-36) requires EU drafters to use certain modals (e.g. shall, may) but not others (e.g. must and should), the latter being a less binding or weak form of obligation". However, Garzone also proved that the use of 'shall' has declined over the past four decades "with a more dramatic change recorded between 1989-1990 and 2005-2006" (2013: 70). According to the Parliamentary Counsel Office on the principles of clear drafting (online), "although 'shall' is used to impose a duty or a prohibition, it is also used to indicate the future tense. This can lead to confusion. 'Must' should be used in preference to 'shall' because it is clear and definite, and commonly understood". Garner (2012: online) also argues that "in most legal instruments, shall violates the presumption of consistency…which is why shall is among the most heavily litigated words in the English language". Alternatives to the translation of shall include "must, the simple present, the semi modal-is to and the imperative" (Garzone 2013: 75-78).
From this brief review, 'shall' or 'must' are used interchangeably due to drafting styles and conventions in different countries and whether they are used in enacting or non-enacting terms 5 . For example, UK drafters, due to the Plain English campaign, have opted out of shall and replace it with must, which is perceived as simpler, more direct and devoid of legalese. The EU English Language services guidelines use 'must' for all documents but not in legislations which are drafted by lawyers/legal drafters. They do not follow the language services guidelines, and they use 'shall' in drafting legislations (Strandvik European Commission's Directorate-General for Translation (ECDGT), Oct 2018) as 'shall' is considered the 'daily bread' of the lawyers (Mellinkoff 1963: 13) 6 .

Modality in English and Arabic
English modal auxiliaries, their criteria, and typologies, as well as the concept of modality, have been subject to exhaustive research by English linguists (von Wright 1951, Hoffman 1976, Leech & Svartvik 1975, Coats 1983, Quirk et al. 1985, and Palmer 1990. In more recent years, English linguists such as (Hoye 1997, von Fintel 2006 discussed other expressions (adverbs, e.g. perhaps; nouns, e.g. possibility; adjectives, e.g. necessary) which could function as the deontic modality of obligation (Eades 2011: 284-85) 7 . Scholars differ among themselves to reach one standard typology 8 . Nevertheless, most of them agreed on two basic types of modality: epistemic and non-epistemic (deontic and dynamic) 9 . The study of English modals is both structural and semantic according to Palmer (1990), but recently linguists tend to use "purely meaning-based definitions of modality without recourse to structural criteria for the purposes of cross-linguistic comparison" (Eades 2011: 284).
Modality in Arabic is not as well-defined as it is in English. In other words, "the English modal system does not have a defined grammatical counterpart in Arabic" 5 For more information on what is used in enacting and non-enacting terms, see Biel (2014) and Seracini (2015) on modals in UK English and EU English. 6 It was challenging to gather information about the process of translating Arabic constitutions into English due to the large number of constitutions involved and difficulty of identifying the translators. Alternatively, we used two corpora (monolingual and parallel) to explore the deontic modality in legal discourse in an attempt to identify the critical differences between the translated and non-translated texts in this respect. 7 For examples on these categories refer to Eades (2011: 84-85). Refer to Coats 1983 and Palmer 1990 for more information. (Eades 2011: 287). The modal system in Arabic mostly follows a semantic system or rather a "pragmatico-semantic category" (Al-Karooni 1996: 76). They are labelled modal expressions by Anghelecu (1999: 130) because they can be lexical verbs, particles, prepositions, or prepositional phrases. Like any other modal system, modality in Arabic is surrounded by "questions of conceptuality, ambiguity, and indeterminacy" (Abdel-Fattah 2005: 31) 10. One point to highlight here is that "in modal meaning, several contexts interact e.g. tense, aspect, intentionality (in addition to extralinguistic factors) which increase the degree of indeterminacy" (Abdel-Fattah 2005: 35). In addition to that, Arabic modals are inherently weak, and the dynamics of power relationships in English and Arabic are very different (Harris et al. 1997). One obvious example to give is the meaning of the modal (Abdel-Fattah, 2005: 39).
For the above reasons, translating modals from English into Arabic and vice versa is not an easy task, and when translating legal texts, caution is required to avoid possible mistranslations and misinterpretations. This is attributed to the notion of "indeterminacy" of modal auxiliaries English (Coats 1983), "where the specific set of modal auxiliaries in English can open possibilities for more than one interpretation, their Arabic counterparts rarely do" (Badran 2001: 48). Also, the fact that translators follow their stylistic preferences in translating modals into Arabic (Abdel-Fattah 2005) and these choices lead to possible looseness of meaning and changes the primary intended meaning of the original text. Abdel-Fattah (ibid.: 39) gives an example of the interchangeable use of the verb (yanbaghi, should), (yajib, must) and (yata c ayn, have to): "Because there is no usage evidence where the meaning is clearly not epistemic, it is a matter of stylistic choice especially between and . Personally, I prefer though is not incorrect". In fact, translating modal auxiliaries, in general, should rather depend upon "careful consideration of macro-textual factors (context, register, text-type, author's intention, etc.) in both the ST and TT" (Eades 2011: 286).

Deontic modality and legal discourse
The language of the law is communicated through the notions of obligation and permission, which are realised by the grammatical category of deontic modals, semi-modals and related patterns (Biel 2014: 158). Legislative texts, e.g. directives, regulations and constitutions communicate fundamental functions of the law (to oblige, to permit and to prohibit), hence impose duties and confer rights (Hart 1961: 27). These notions of deontic modality vary a lot across languages and are considered a beneficial area of interlingual research (Palmer 1990 andBiel 2014).
The term deontic is concerned with "language as action" (Palmer 1986: 121). It concerns what is possible, necessary, permissible, or obligatory, given in a body of law or a set of moral principles or the like (von Fintel 2006: 2). Deontic modals are essentially "performative" (Palmer 1990: 69), as they express "the possibility or necessity of acts in terms of which the speaker gives permission   -Faharaty, H. y Elewa, A. Estud. trad. 2020, 10: 107-136 (e.g. may, can) or lays an obligation for the performance of actions at some point in the future (e.g. must) or make a promise or threat (e.g. shall)" (Hoye 1997: 43, Palmer 1990. There are two main types of deontic modality: obligation and permission. The legislature, through the use of deontic modals of obligation, "instructs the addressee to do X or outlines the obligations to which the addressee must conform" (Trosborg 1997: 45). Obligation is expressed mainly by the modals 'shall' and 'must' as well as (should, is/are to, have/has to, need to, ought to). Deontic modals of permission in legislations give the legislature an authority to grant a certain body the permission to perform an act (ibid.: 48). They are classified into (may, can, could, might, need not) 11 .
It is challenging to categorise the modal system in Arabic due to "the absence of linguistic corpora of usage" (Abdel-Fattah 2005: 34). He (ibid. 41-2) gave a list of deontic modals which is not finite and different from other lists suggested by other linguists such as Badawi, Carter & Gully (2004: 394-421) and the typology of Arabic deontic modals discussed in a small corpus of legal documents by El-Farahaty (2015: 118-138). This study attempts to derive a more holistic list of modal expressions supported by The Leeds Annotated Parallel Corpus of Arabic-English Constitutions. It focuses on "duty-imposing" deontic obligation, but "power-conferring" deontic permission (Biel 2014: 340) will be the topic of a separate study.

Corpus and methods
The study uses two corpora, one parallel corpus and a monolingual comparative corpus to complement the analysis by comparing translations to non-translations (Baker 1993(Baker , 1995(Baker , 1996Biel 2014), and to provide some techniques of advanced comparative analysis (cf. Biel 2014). The two types of corpora complement one another in examining "how the same content is expressed in two languages" (Aijmer & Altenberg 1996: 13). Parallel corpora, in its own right, may fail to reflect cross-linguistic contrasts, because translated corpora cannot avoid the effect of deviation from the ST (i.e. translationse) (Baker 1993: 243-5, McEnery et al. 2006. Therefore, to rely solely on parallel corpora could be misleading; "parallel texts are inappropriate because the filter of the translator and the translation strategies get in the way, and/or L2 texts may end up with a text structure very similar to that of the L1" (Iørn Korzen & Morten Gylling 2017: 39). Using the original English corpus in this paper could be regarded a tool for triangulation to compare the results obtained from the parallel corpora to find out whether the use of deontic modals is comparable to that of the Monolingual English Corpus. The MEC is almost three times bigger than the English sub-corpus in the parallel corpus and would more likely yield reliable results about the use of deontic modals in legal discourse. This method of triangulation is called "within-method corpus triangulation," i.e. it still employs the quantitative research methods of corpus linguistics (Malamatidou 2018). Therefore, after analysing all the instances of deontic modals in the parallel corpus, MEC is consulted to provide another source of evidence 11 For more information and examples, refer to Palmer (1990: 69-73) and El-Farahaty (2015: 68-70).
for the typical use of deontic modals in English and to see whether the translated corpus is comparable.
PEC consists of 219,212 Arabic words (243,486 tokens) and 206,318 English words (238,165 tokens). The original corpus includes the most current ratified constitutions of 19 Arabic countries and includes a sub-corpus of Preambles 12 . The corpus was collected from the World Intellectual Property Organisation (WIPO) and CONS-TITUTE. It was then annotated and uploaded on Sketch Engine 13 . MEC (557,086 English words/ 677,056 tokens) was compiled 14 to provide some techniques of advanced comparative analysis (cf. Biel 2014). It comprises the up-to-date revised versions of the constitutions of eight countries whose formal language is English, the earliest revised version was Australia in 1985, and the latest revised version was the USA in 2016. These countries are Australia, Canada, Ireland, New Zealand, Singapore, South Africa, United Kingdom 15 , and the United States of America. The corpora were compiled, annotated and uploaded on Sketch Engine (Kilgarriff et al. 2014) which is the toolkit used for the quantitative analysis. It noteworthy that there is no consistency in the use of English across the above listed diverse regions, as demonstrated by Corpus Linguistics studies that there are different "world Englishes", hence the limitations of the MEC corpus for "techniques of advanced comparative analysis".
The methodology is divided into two phases: quantitative frequency analysis of the different variants of deontic modal expressions of obligation in the Arabic Corpus. By undertaking this frequency analysis, this paper introduces a typology of modal expressions in Arabic constitutions that can be indicative of the various modal expressions used in other Arabic legislative documents. The second phase will be qualitative analysis of the techniques used to translate Arabic modals into English modals. The aim of comparing the original Arabic text to its English translation is to analyse the techniques of translating the wide range of modal expressions identified in the first phase of analysis and see patterns of discrepancy, i.e. if translated modal expressions are either a stronger or weaker modal expression 16 , or if "they are placed on a different scale representing a different system of modality" (Badran 2001: 52). Then, a comparison to the MEC will be undertaken.

Analysis of deontic modality in the Arabic corpus
As discussed in section 4, Arabic expresses deontic obligation through the present tense (imperfect), certain lexical verbs, nouns, prepositions, particles and special phrases. These will be introduced in detail in this section. 12 For more information on the original corpus, see Brierley & El-Farahaty 2019. 13 The corpus was annotated in 2018 and uploaded in Sketch Engine by Abdulrahman Alosaimy as research assistant (and then a postgraduate in the School of Computing, University of Leeds) and now a lecturer at the Al Imam Mohammad Ibn Saud Islamic University, Riyadh, KSA. 14 This corpus was collected and uploaded in Sketch Engine by Abdulrahman Alosaimy (see footnote above). 15 The UK does not have a codified constitution, and the version used in this corpus is available on the Constitute Project Website. This version is compiled from: "texts collected from legislation.gov.uk. This complete Constitution has been generated from excerpts of texts from the repository of the Comparative Constitutions Project and distributed on constituteproject.org". (Constitute Project): https://www.constituteproject.org/constitution/United_Kingdom_2013.pdf?lang=en 16 For more information about the modal strength in translation, refer to Biel (2014) and Satthachai & Kenny (2019).

Deontic modality of obligation
Tables 1 and 2 introduce the list of Arabic modal expressions of obligation in the Arabic Corpus. n stands for the total size of the corpus, m for the calculated counts per million and freq for the frequency of the search term.  Table 1 gives the whole list of modal expressions of obligation searched in the Arabic Corpus. This list comprises positive verbs of deontic obligation, and verbs in the past form and one modal expression. The overall count of these forms is (248) the most frequent of them is the positive present form of the modal verb  1 2 3 4 5 (ta c ayana, be incumbent upon) (1). Deontic modality is also expressed through adjectives and nouns such as the concepts under discussion in table 2 below:   wājib (duty/obligation) (67). Despite the high frequency of these adjectives and nouns, some of them denote obligation whereas the majority of them are vague and indeterminate. They are used to lay a moral duty, not an obligation. Although drafted constitutions should be concise and intelligible, the use of adjectives such as necessary and appropriate is common in legal discourse, and usually, criteria of evaluating these terms are not specifically clear (Engberg & Heller 2008, El-Farahaty 2016.

Deontic modals of prohibition in the Arabic corpus
Deontic modality of prohibition "may be considered as an obligation to refrain from performing or acting. Therefore, the exponents of prohibition are usually the exponents of obligation followed by not or other forms of negative sentences" (Matulewska 2010: 82). More specifically, "'may not' is used in English to express general prohibitions and 'shall not' to express criminal prohibitions" (Rosenbaum 2007: 23). In the Arabic Corpus, prohibition is expressed by both positive lexical verbs and negative verbs (see table 3). The overall count of these verbs in the corpus is (608), the most frequent of which is the negative verb (yajūz, may). It denotes both forcible prohibition and lack of permissibility. In another small scale study, El-Farahaty (2015) found that it is the most frequent modal of prohibition. The second frequent modal is  (lā yaḥiq, is not entitled to) (6) and  (yajib c allā, must not) (2), the latter negating the proposition not the verb, so it is not the direct negative of the positive verb

Discussion of modal frequency and deontic modality in the Arabic Corpus
The number of modal expressions given in table 1 and 2 does not reflect the actual modality in the Arabic Corpus. Modality is represented throughout the whole corpus, almost in every article by a wide variety of lexical verbs as a form of "constitutive statements which may serve laying down the law" (Trosborg 1997: 50), in the imperfect which is used in Arabic to express mood. It denotes that this action has already started and yet to be finished (El-Farahaty 2015: 121). These verbs are normally translated to the modal verb of obligation 'shall'. The "latter is legally binding" (Tiersma 1999: 106), and it guarantees that the act will happen (El-Farahaty 2015: 121). A quick search into the PEC we retrieved (4695) occurrences of 'shall' in its positive form and (224) in its negative form (shall not) that express prohibition. This high frequency of 'shall' confirms that constitutive statements including the nominal declarative statements have been translated into 'shall' in most of the cases 17 . This can be tested statistically to figure out the significance of the high frequency of 'shall' as discussed below and examined further in Table 7. By way of illustration, the following excerpt shows that all the nominal declarative statements are translated into 'shall' plus the verb.
. Algeria shall be a People's Democratic Republic. It shall be one and indivisible. Article 2 Islam shall be the religion of the State. Article 3 Arabic shall be the national and official language. Arabic shall remain the official language of the State (Constitution of Algeria) It is noteworthy to mention that, although the word 'shall' occurred 4,695 times, only 130 times are used as deontic obligation verbs, after manually refining the hits. 17 It is noteworthy to mention that nominal declarative statements are not particularly the standard/common way of imposing obligation. Arabic uses the declarative imperfect verbs to convey obligation.
It was used for different senses such as assertion/intention (e.g. I shall be loyal to the king. As to the use of 'shall' for future references, it is not common today, and if it is used, it follows the pronoun 'I' and 'we' according to the Oxford English Dictionary 18 . That is why deontic modals do not occur with 'I' and 'we' in order to avoid the overlapping sense (El-Farahaty 2015: 69). This is why the bigram 'I shall' is not captured in the MEC at all, while 'we shall' occurs only once in an example derived from an English translation of the Magna Carta that dates back to 1215.
We have granted to be holden within this our Realm, as much as appertaineth to Us and our Heirs, we shall observe; and all Men of this our Realm, as well Spiritual as Temporal, as much as in them.
On the other hand, the PEC contains no instances of 'we shall', while 'I shall' occurred 12 times. Another form of deontic modality of obligation, which is not included in the above lists is the preposition      ( c la, on), introduces the sentence to give the meaning of an imperative form. In this case, it is classified in Arabic grammar books as a preposition functioning as a verb like the following example derived from PEC.  The frequency analysis presented in the tables above showed that some deontic modals are not frequent in the Arabic Corpus under investigation. One example is the modal verb of obligation  (min al-wājib, it is obligatory), etc., the former happened once, and the latter two did not happen at all. It is known in Arabic that the perfect carries no mood and it expresses events that "either actually completed or regarded as such" (Badawi, Carter & Cully 2004: 62)   (wajaba, must) (40) occurred in the Arabic Corpus. These past forms continue to be used in legal texts and they mean the same as the imperfect form, but if the imperfect forms of the same verb are used in legislative legal drafting, drafters should not use the past form.

but one frequent obligation modal verb is
In terms of prohibition, there was no hit of the modal expression (yajib, must) which is the most frequent modal of obligation in the Arabic Corpus. It is worth noting that  (yajūz, may), is used by Islamic jurisprudence to mean prohibited or not desirable. It is frequent in modern legal discourse and in formal legal settings to regulate behaviour, i.e. to stipulate the rules of the law, because "most modals are polysemous and their mean-18 https://www.lexico.com/en/definition/shall. ing often changes when they are negated" (Biel 2014: 340). For example, 'may not' is used in English for a negative command: "This additive may not be used in foods (prohibition)" (Biel 2014: 341). However, it is not as strong as 6 7 8 9 10 (lā yajib, must not), or 7 8 9 10 (yajib allā, must not) and can be vague.

Translation of Arabic deontic obligation into English
In this section, the translation of deontic modality of obligation into English in the PEC will be analysed. Due to the massive amount of data, in this section, we will focus more on the most common deontic modals and their translation into English. Translators have used an even wider variety of modals in translating the wide variety of the modal expressions used in drafting the Arabic constitutions into English. However, as noted earlier, the MEC could be used for triangulation to account for variation in translation strategies. (yajib, must/is incumbent upon) has been translated into 'shall' (84) and 'must' (46) (with the highest probability score of 352.69 and 193.14 per million respectively). This high frequency of the modal 'shall' as an equivalent of 1 2 (yajib, must/is incumbent upon) is a clear evidence of the target language (TL) impact, e.g. exaggeration of TL features (Baker 1993) or gravitational pull hypothesis (Halverson, 2007). It is noteworthy that 'must' was used in the case of procedural obligation as in the following example: . . Any person who is arrested must be informed of the reasons for his arrest and his rights, and may not be incarcerated in front of the administrative authority except by an order of the competent judicial authority (Consitition of Syria) Translators have also opted for other forms of modality such as the use of the semi-modals (are to, have to) (Palmer 1990ː 25), lexical verbs (is required), nouns (the duty) and (should):  (wajaba, past) denotes weaker obligation as it is used "for recommendations" (Manual for Drafting International Labor Office (ILO) instruments 2007). Also, the English Style Guide (2012: 35-36) requires EU drafters to use certain modals (e.g. shall, may) but not others (e.g. should), the latter being a less binding or weak form of obligation (Biel 2014: 341, Garzone 2013. Other lexical verbs denoting obligation are the imperfect (present) verb    ( yaḥiq, is entitled to/has the right to). Examples of these two verbs are given below: Congregations shall have the right to establish and maintain their own schools for the education of their own members provided that they comply with the general provisions of the law… (Constitution of Jordan) The President of the Republic shall pass the laws approved by the People's Assembly. He might also reject them through a justified decision within one month of these laws being received by the Presidency (Constitution of Mauritania) The law specifies the cases where a counsel for defense must represent an accused person (Constitution of UAE) In case of flagrante delicto, the Council must be notified of the measures taken against the offending Member; where the Council is not in session, such notification should be made at the first subsequent session (Constitution of Qatar) . El-Faharaty, H. y Elewa, A. Estud. trad. 2020, 10: 107-136 123 -Faharaty, H. y Elewa, A. Estud. trad. 2020, 10: 107-136 (muqadas, sacred). The noun  lā yajūz, may not/it is not allowed), the direct equivalent of which in English is either 'shall not' or 'may not' has been translated into many different ways (see table 6 below) but the top three categories are the negated proposition initiated by 'no'/ 'none' plus ('shall', 'may', or 'can'); 'shall not', 'shall only be' and 'cshall be prohibited'. It has also been translated to less frequent modal verbs and expressions such as 'cannot', 'it is not permissible', 'may only be', 'neither … nor' 'may/shall …', 'under no circumstances may …', 'can only', 'it is forbidden', 'it is not allowed', 'it is impermissible', and very rarely 'must not' which, according to Bázlik & Ambrus (cf. Krapivkina 2017: 310), is the same equivalent of 'shall not'. (lā yajūz, may not/it is not allowed) in the above examples vary; in 16 and 17 it is a conditional prohibition, i.e. "the prohibition to perform only in specific circumstances" (Matulewska 2010: 83), but it is translated into a negative proposition plus the modal verb ('may' in example 16 and 'shall' in example 17) plus the passive. In example 19, it is translated into 'cannot' which functions to instruct through law "what cannot take place" (Trosborg 1997: 48). Example 17 is translated into a negative proposition plus the modal verb "might", a modal verb that is not common in the PEC, it just occurred few times in the "Constitution of Syria".
The lexical verb of strong prohibition (yaḥẓur, forbids/ prohibits/prevents) is translated in different ways throughout the corpus: positive 'shall' collocated with the lexical verbs of prohibition 'shall be prohibited', and 'shall be forbidden', 'shall not' or negative command 'no… shall', as well as other less frequent forms such as the declarative 'is prohibited', 'it is not permissible', 'may not' and 'no… may'. Examples of these include: First: Any entity or program that adopts, incites, facilitates, glorifies, promotes, or justifies racism or terrorism or accusations of being an infidel (takfir) or ethnic cleansing, especially the Saddamist Ba'ath in Iraq and its symbols, under any name whatsoever, shall be prohibited (Constitution of Iraq) Article 34 The accused is innocent until proved guilty in a Court of Law which shall guarantee him the necessary sureties to exercise his right of defense. The accused shall not be bodily or mentally injured (Constitution of Kuwait) .
No individual, organisation, entity, or group shall be allowed to create military or quasi-military squadrons, groups or organisations (Constitution of Egypt) Slavery and slave trade in every form is prohibited. No person shall be held in slavery or servitude (Constitution of Sudan) . Laws may not contain any provisions that provide immunity to any administrative decision or action or against judicial review (Constitution of Palestine) The examples given above show the use of strong prohibition in Arabic in cases where the topics deal with the judiciary, slavery, terrorism but the translation itself varies where 'shall' was used in most of the cases but in few cases 'may not' was also used. Although it is used to account for prohibition (Biel 2014), it is not as strong as the use of 'shall not'" which is used as an equivalent of the same verb in the rest of the examples discussing similar topics. 'May not' can also be ambiguous "in spite of the fact that the positive form of 'may' has a different meaning from 'must' and 'shall', the negative form is the same. The negative form is used to express prohibition, something that is forbidden, and that is why this form should normally be avoided. Therefore may+not is used in the sense of limiting the rights or the powers of a provision" (O'Shea 2015: online). 1 2 3 4 (lā yamna c , does not forbid/ prohibit/ prevent) are the second frequent positive modal of prohibition (15). The three forms have been translated into 'may + prohibit/be forbidden)', 'shall + prohibit/be forbidden', and 'it is forbidden' 'no… shall', and no … may'. Examples 26 and 27 have the same modal verb of prohibition in the passive form, but they are translated differently into 'may not' (26) and into the declarative passive structure (it is forbidden) (27).
The law shall prohibit monopoly and unfair competition practices (Constitution of Algeria) Membership of the government and of the Assembly of the People's Representatives may not be combine. (Constitution of Tunisia) It is forbidden in the domains of Defense and of National Security (Constitution of Mauritania)

Deontic modals in the monolingual corpus
The classification of modals is based on Biel's (2014) analysis of the EN Acquis (regulations and directives). However, more modals such as 'will', 'will not', have been searched in this corpus and have been added in the table below (this corpus also included more instances of 'shall'). It occurred 4,825 times in the corpus (with a probability score 20 of 7, 126.44 per million) in both positive, (4,054) to convey obligation, and negative forms (642), 'shall' plus negative (129), to convey prohibition. 'Must' is the third most frequent obligation modal (1,697/ 2,506.43 per million) in both the positive (1,625) and its negative form (72). 'Might' (50/ 37.84 per million) occurs mostly in its positive form (48) and (2) instances only in the negative form, while 'Should' occurs (133/ 196.438 per million) in the positive form only.
One can also notice that there is a wide gap between the frequencies of some items in the two corpora, but the difference between the observed frequency and their expected values 21 can be compared using Loglikelihood (LL) 22 statistical test as shown in Table 7 below. The probability score or expected value is attained by normalising the observed values (the actual occurrence of the hits) of two different sets of numbers. By normalising the output counts, we can see how probable that X would occur in any text irrespective of the data size. The formula for calculating normalised counts is as follows: (number of occurrences of the hit/length of text) x 1m). 21 See the footnote above.

22
Loglikelihood statistical test is used to spot the most significant items in the two corpora regardless of their unequal size. In other words, it can show how significant the result is.
(yajib, must) did not occur in the Arabic Corpus. It is often replaced by (lā yajūz, may not/it is not allowed), one of the most frequent modals of prohibition which denotes more than one meaning, forcible prohibition and lack of permissibility. Deontic modal verb of obligation Must is to/are to should have/has the righ is/are entitled to ought to need to have to/has to Total (yajib, must) was translated into English as 'shall and 'must', the latter was used to denote forcible procedural obligation where the former is used in abstract general obligation. Translation of prohibition in the PEC was varied using a wide range of 'shall' in negative structures and almost no mention of 'must not'. Lexical verbs ex-132 El-Faharaty, H. y Elewa, A. Estud. trad. 2020, 10: 107-136 pressing strong prohibition such as 7 8 9 10 (yaḥẓur, forbids/ prohibits/prevents) were translated into strong prohibition in English but in some cases into 'may not' which conveys weaker prohibition.
In comparing the frequency of the modal expressions in the PEC and MEC, to answer the third question, we found that 'shall' is a most common modal verb in the drafting of monolingual constitutions and an equivalent of constitutive verbs in the Arabic Corpus despite the difference in the size of both corpora. This can be attributed to TL impact, exaggeration of TL features, i.e. translators are known to opt for more formal variants. Translation of deontic modals of prohibition varies in the PEC in comparison to the MEC, but both have used the most common modals of prohibition: 'shall not', 'may not' and 'no … shall'. 'Must not' was represented in the MEC and almost not existent in the PEC.
In spite of the variation of the modal auxiliaries in Arabic, and the variation of their translation into English, the common modals used in the MEC have been used in most of the cases. However, there was some variation in translating the same modal in Arabic (into English). This variation may be context-dependent, but also depends on the Constitution under discussion and the style used in each Constitution (something beyond the scope of this study) By the publication of this corpus and this research, we hope this analysis would be helpful for translators and drafters alike. The qualitative analysis of both the Arabic Corpus and the parallel translations showed that many articles (topics) have recurred in almost all the constitutions, some of them used the same wording. Nevertheless, the translation of these articles varied from one Constitution to another due to lack of one general unified guidelines and lack of coordination between governments and translators. The current study has only focused on the translation of deontic modals of obligation and prohibition due to the limited capacity in this paper; a possible future endeavour will be to analyse the deontic modals of permission. Another possible useful study will be investigating the deontic modals of obligation and permission from English into Arabic. It would also be useful to compare the modal expressions and their translation in the current and older versions of the constitutions.