A corpus-based study on deontic modality in Competition Law: Insights from the English-Greek language pair

This study investigates deontic modality, the grammatical category through which legal texts express mainly obligation and permission, in an English-Greek bilingual corpus composed of legislative texts related to European Union (EU) Competition Law. More specifically, the study is based on Biel’s discussion on deontic modality, i.e. deontic obligation and deontic permission (Biel 2014: 158). The analysis of the data is mainly quantitative, while a small-scale qualitative analysis is also carried out when necessary. The results of the study are compared with the specific guidelines proposed by the EU Institutions for English and Greek, i.e. the Joint Practical Guide of the European Parliament, the Council and the Commission for persons involved in the drafting of European Union legislation (2015), the English Style Guide: A handbook for authors and translators in the European Commission for English (2018) and the Greek Style Guide: A handbook for authors and translators in the European Commission for Greek (n.d.), as well as with those of earlier studies on legislative texts.


Introduction
It is widely accepted that EU legal language forms a huge and important area for linguistic research. EU legal texts, such as directives and regulations, carry specific characteristics forged mainly from the circumstances under which they are produced and are directly or indirectly implemented into the various national legislations of the EU's 28 Member States, shaping the behavior of all society members. The present study aims to investigate an aspect of those texts, i.e. deontic modality, in an English-Greek bilingual corpus of EU Competition Law.
Deontic modality was chosen for the present study for two reasons. First, following Hart's (1961: 27) statement about the primary functions of law, which are to impose duty and confer power, legal texts express mainly obligation and permission (Biel 2014: 158). These functions are expressed in language through the grammatical category of deontic modality. The second reason for choosing deontic modality for the present study is the fact that among the grammatical categories, modality varies more than others across languages (Palmer 2001: 18, Biel 2014, thus constituting a robust area for interlingual research. In particular, the present study aims at analyzing aspects of deontic modality in an English-Greek corpus (see section 4). The results of the analysis are compared with the specific drafting guidelines suggested by the EU Institutions, i.e. the Joint Practical Guide of the European Parliament, the Council and the Commission for persons involved in the drafting of European Union legislation (2015) 2 , the English Style Guide (2018) and the Greek Style Guide (n.d.) 3 . As far as the analysis is concerned, it is mainly a quantitative analysis. However, a small-scale qualitative analysis is carried out in order to cast some light into the most interesting patterns of deontic modality. Besides, the value of mixing qualitative and quantitative approaches to corpus research is uncontestable (Hasko 2012: 4761).
The paper is divided into two parts. The first part (sections 2 and 3) provides theoretical insights into the specific characteristics of EU legal texts, the circumstances under which they are produced as well as deontic modality in EU legal texts. In the 2 Joint practical guide of the European Parliament, the Council and the Commission for persons involved in the drafting of European Union legislation (for English). Retrieved from: https://op.europa.eu/en/publication-detail/-/publication/3879747d-7a3c-411b-a3a0-55c14e2ba732. Joint practical guide of the European Parliament, the Council and the Commission for persons involved in the drafting of European Union legislation (for Greek). Retrieved from: https://op.europa.eu/en/publication-detail/-/publication/3879747d-7a3c-411b-a3a0-55c14e2ba732. second part (sections 4 and 5), the methodology and the results of the study are presented and discussed.

Characteristics of the genre of EU legal texts
As pointed out in the introduction, the production of EU texts is a complex procedure. In particular, multilingual texts such as those produced in the EU institutions are tied up with "socially developed intercultures" (Pym 2004: 14). Therefore, as Sosoni (2011: 82) claims, "the broader societal and political as well as intercultural and linguistic framework in which such texts are embedded cannot be overlooked".
EU legal texts have specific characteristics which derive from the idiosyncratic circumstances under which they are negotiated and produced, namely in a multicultural and supranational environment. As a result, they are idiosyncratic, reflecting specific textual features (Sosoni 2012: 76). In this section, we focus on some of the key concepts of EU legal texts: multilingualism and the principle of equal authenticity, hybridity, equivalence and intertextuality.

Multilingualism and the principle of equal authenticity
The importance of multilingualism was recognized right from the beginning of the European Community, the ancestor of the European Union, and is closely related to its policy and purposes. The EU is an economic, political and democratic union of 28 present Member States. One of the most prominent features of the EU is the principle of multilingualism, i.e. that the official languages of all Member States are both official and working languages in the EU institutions and are considered to be equal (Šarčević 2001: 314). The principle of multilingualism is guaranteed by Council Regulation No.1 (1958). Another prominent feature of Council Regulation No.1, which promotes even more the policy of multilingualism, is that the citizens of the EU have the right to communicate with the EU institutions in their official language. The features mentioned above distinguish the EU from other supranational organizations, which use just a few working languages. The supranational, multilingual and multicultural EU has 28 Member States and 24 official languages.
The principle of multilingualism goes hand in hand with the principle of equal authenticity (Šarčević 1997: 64) or plurilinguistic equality (van Els 2001: 311). In fact, no version is more authentic than the other. There is no original and no translation. Once a legal text is published, it is presumed to have the same meaning in all the 24 languages.
The notion of multilingualism is directly linked with translation. In fact, translation is present in almost all the drafting procedures of the various EU institutions. As Robinson (2005: 4-5) aptly observes, the draft and the production of a legal instrument, such as a directive or a regulation, passes through many stages. The first draft is initially written by an expert, and then it is revised before it goes through an approval process -which necessarily involves translation-at the Commission, the Parliament and the Council. There, translations are discussed and amended, while they may go back and forth several times, accompanied at all times by the translated drafts and amendments (Sosoni 2012: 82).

Hybridity
EU law is often described as a hybrid legal system (Cao 2007: 150;Biel 2014: 337;Sosoni et al. 2018: 170, among others). The controversial concept of hybridity was proposed by Trosborg (1997: 145-146), who stresses that hybrid texts are products of intercultural communication in a supranational and multilingual environment where there is no linguistically neutral ground. Schäffner & Adab (2001: 173) point out that hybridity is a result of convergence between linguistic and cultural conventions and has to do with concessions made during intercultural exchange (see Biel 2014: 53). Thus, a hybrid text can be a source text (ST) as well as a target text (TT). This claim seems to be particularly valid in the context of EU discourse (Biel 2014: 53), where the concept of hybridity holds true for STs, TTs as well as EU law itself. Felici (2010: 102) attributes hybridity to the complex drafting procedures that a ST undergoes (see Robinson 2005) and points out that "the final product is a hybrid text, the nature of whose source and original has become more and more blurred".
The special nature of EU language (and EU legal language in particular) has resulted in various names in the literature, such as Eurolect, Eurospeak, Eurofog, Eurojargon and others (Biel 2014: 76).

The concept of equivalence
The principles of language equality and equal authenticity mean that translations are not really translations but language versions. Indeed, even though the EU has the largest translation service in the world and employs both in-house and external translators, the term 'translation' is absent from EU legislation (Felici 2010: 96, Sosoni 2012. However, in practice all language versions are actually translations, and translators have to follow specific guidelines. In fact, these guidelines are close to literal translation strategies, because all language versions have to be similar. Indeed, all language versions are 'mirror images' of each other and should correspond paragraph for paragraph, even sentence for sentence, as far as possible (Sosoni 2012: 84). The ST as a notion ceases to exist as such after its translation and we then speak of 24 equally valid texts.

The concept of intertextuality
The last characteristic of EU texts that we briefly discuss in this section is bound with the concept of intertextuality. This concept postulates that a text cannot exist alone as a self-contained, hermetic whole, since it is shaped by the repetition and transformation of other text structures and a writer is always influenced by texts that he/she has read before and is also the creator of other texts (Sosoni 2012: 83;Gibová 2009: 148, in De Beaugrande, Dressler 1981. Therefore, the final text is influenced by every kind of other, previous texts. This concept is valid for the majority of EU texts, as they are full of references and direct and indirect quotations from other, previously drafted, texts.

Deontic modality in EU legislative texts
As already mentioned in the introduction, EU legislative texts are produced in a very specific way while their 'translation' into other languages should follow specific rules and conventions. In addition, legislative texts must communicate the fundamental functions of law which are to impose duties and confer power (Hart 1961: 27). This means that their main purpose is to oblige, permit, or prohibit citizens to do something or behave in a specific way into the multicultural EU. These concepts are realized through deontic modality, i.e. modals, semi-modals and related patterns that convey obligation and permission (Biel 2014: 158).
According to Palmer (2001: 70), deontic modality is one of two main types of event modality (the other being dynamic modality). With deontic modality, the conditioning factors are external to the relevant individual, whereas with dynamic they are internal (Palmer 2001: 9). Thus, deontic modality relates to obligation and permission emanating from an external source, whereas dynamic modality relates to ability or willingness, which comes from the individual concerned (Palmer 2001: 9-10). Palmer also argues that a comparative study of modality between various distinct languages is very useful (Palmer 2001: 18). English has a formally defined set of modal verbs which express modality.
In legislative texts, the use of the appropriate modal verbs is predicted by the Joint Practical Guide of the European Parliament, the Council and the Commission for persons involved in the drafting of European Union legislation (2015) and the English Style Guide: A handbook for authors and translators in the European Commission for English (2018). However, in Greek the formal status of modal verbs is very much in doubt 4 (Tsangalidis 2004: 231) and the Greek Style Guide (n.d.) and the Joint Practical Guide for Greek (2015) provide drafters and translators with some guidelines on the expression of meanings such as obligation and permission.
The expression of meanings such as obligation and permission is mainly controlled via the various Style Guides and Joint Practical Guides for the official languages of the EU. Their main purpose is to ensure the highest quality of Community legislation with a view to guaranteeing citizens' rights and easing judicial control (Greek Style Guide n.d.: 86). The manner in which an act is drafted should also take account whether or not the act is binding (Joint Practical Guide for Greek 2015: 12). Regulations have direct application and are binding. Therefore, they are drafted in such a way that the addresses have no doubts as to the rights and obligations resulting from them (Joint Practical Guide for English 2015: 11). On the other hand, directives, which are also binding, set out a goal that all EU countries must achieve, but it is up to the individual countries to devise their own laws on how to reach these goals (https://europa.eu/european-union/eu-law/legal-acts_en).
The English Style Guide (2018: 48-50) clearly suggests the use of the modal verb shall in order to express an obligation or requirement in the main clauses of the enacting terms 5 and not of the modal verb must which has prevailed in most English-4 Following Tsangalidis' (2004) discussion on Modern Greek modal verbs, this study focuses mainly on two modal verbs, i.e. πρέπει <prepei>, which means must and expresses mainly obligation, and μπορεί <borei>, which means may and expresses mainly permission. 5 The enacting terms is a very important part of the corpus because they constitute the legislative part of the act and the guidelines are addressed mainly to them. speaking countries. Although shall has fallen victim to the 'modal revolution' under the Plain Language Movement in most jurisdictions (Garzone 2013: 69;Biel 2014: 10, among others), it is still used widely in EU legislation. In addition, shall not is used in order to impose a prohibition instead of must not. Furthermore, for the expression of permission, EU legislation uses the modal may, while for the permission not to do something, need not is used. Finally, EU legislation uses the simple present tense in order to express provisions that have a direct implementation. However, where a provision applies to the future or is contingent on a future event, the verb to be used is shall (English Style Guide 2018: 49).
The Greek Style Guide (n.d.) and the Joint Practical Guide for Greek (2015) also provide some guidelines on drafting and translating legislative texts in Greek. The first feature is that in the enacting terms of legal acts, the simple present tense of the indicative mood is used in order to express an obligation (Joint Practical Guide for Greek 2015: 12, Greek Style Guide n.d.: 63). This is a common feature with French and different from English which uses the modal verb shall in order to express obligation. However, it is stipulated in the Greek Style Guide (n.d.: 94) that all three languages must avoid the future tense wherever is possible.

Corpus design and methodology
In line with the above, the present study focuses on deontic modality in legislative texts related to the European Competition Law. In particular, the study of deontic modality follows a corpus-based approach (Tognini-Bonelli 2001: 65). A corpus is "a large collection of authentic texts that have been gathered in electronic form according to a specific set of criteria" and intends to be "used as a representative sample of a particular language or subset of that language" (Bowker & Pearson 2002: 9). In the case at hand, a bilingual English-Greek corpus (Corpus A) of legislative texts related to EU Competition Law was compiled and analyzed (see Table 1). The texts were downloaded from Eur-Lex database (https://eur-lex.europa.eu/homepage.html) which provides official access to EU legal documents in the EU's 24 official languages. More specifically, Corpus A is composed of articles 101 to 109 of the TFEU, which refer to EU Competition Law, as well as of regulations and directives concerning EU Competition Law. Moreover, a bilingual English-Greek subcorpus (Corpus B) composed of the enacting terms -the legislative part of the acts-of the said regulations and directives was also compiled and analyzed (see Table 2) in order to compare the findings of the research with the specific guidelines proposed by the Joint Practical Guides and the Style Guides for every language. The corpora were analyzed using the AntConc software 6 (Anthony 2019). This software belongs to the third-generation tools according to the classification of McEnery and Hardie (2012: 40).
Regulations and directives were chosen for the present study because they are both types of secondary EU legislation that share a common characteristic, i.e. they both have binding force. Regulations are binding legal acts that apply automatically and uniformly to all EU Member States as soon as they enter into force (https://europa.eu/european-union/eu-law/legal-acts_en). In regulations, rights and 6 The version of the software used for this study is AntConc 3.5.8. (Dev) (Windows) 2019. obligations should be formulated in a straightforward way and addressed directly to the addressees without references to national authorities (Biel 2014: 160). On the other hand, directives are also binding acts and set out a goal that all EU countries must achieve. However, it is up to the individual countries to devise their own laws on how to reach these goals (https://europa.eu/european-union/eu-law/legal-acts_en). As a result, deontic modality forms a very interesting field of study in both regulations and directives.
Following Biel's discussion on deontic modality (Biel 2014), this study focuses on deontic modals, semi-modals and other related patterns 7 identified in the corpora. The quantitative analysis focuses mainly on deontic modals and semi-modals in the English-Greek bilingual Corpus, while the findings are compared with the guidelines given in the Joint Practical Guides and the Style Guides for every language as well as similar studies for other language pairs (see Biel 2014, Gibovà 2011. In addition, a small-scale qualitative analysis is also carried out in modals and patterns of the corpus that are interesting.

Results
This chapter summarizes the findings of the study. The results of the quantitative analysis are displayed in graphs. For the better presentation of the results, the results regarding deontic obligation are displayed firstly (see subsection 5.1), while right after follow the results regarding deontic permission (see subsection 5.2). Moreover, in each subsection (5.1 and 5.2), the graphs from the analysis of the English Subcorpora of Corpus A and Corpus B are displayed firstly, while those from the analysis of the Greek Subcorpora of Corpus A and Corpus B follow. During the discussion of the results presented in the graphs 8 , the findings of the qualitative analysis are also presented. 7 For the purposes of this study, other related pattern refers to every other word or sequence of words apart from modals and semi-modals. 8 For the ease of presentation, the graphs regarding Corpus A are presented in pale grey, while those regarding Corpus B in grey.

The case of English
Obligation modals and semi-modals are very frequent in the English Subcorpus of Corpus A. Graph 1 illustrates the presence of obligation modals, semi-modals and other related patterns in Corpus A, while graph 2 illustrates the frequency of obligation modals, semi-modals and other related patterns in the English Subcorpus of Corpus B.

Corpus B (English Subcorpus-Enacting Terms): Deontic obligation
Frequency per 1000 words The modal verb shall is very frequent in both the English subcorpora. This is in line with the institutional guidelines which concern enacting terms (see section 3) and which specify that obligation should mainly be imposed in English using shall. For this reason, shall appears a lot in the English Subcorpus of Corpus A as well as in the English Subcorpus of Corpus B (see Graph 2) 9 . In addition, the use of other modals expressing deontic obligation seems not to be significant, especially in the enacting terms, something that is also in line with the guidelines of the English Style Guide (2018).

The case of Greek
According to the Greek Style Guide (n.d.: 63), the expression of obligation and requirement in the enacting terms should be expressed with the verb of the main clause in the present tense, while expressions like θα πρέπει <tha prepei> (must) and θα έπρεπε <tha eprepe> (should) should be avoided. Graph 3 summarizes obligation modals and other related patterns in the Greek Subcorpus of Corpus A, while graph 4 the obligation modals and other patterns in the Greek Subcorpus of Corpus B. As we can see, πρέπει/θα πρέπει <prepei/tha prepei> (must) are widely used in the Greek Subcorpus of Corpus A (see Graph 3). However, it doesn't appear as often in Corpus B with the enacting terms (see Graph 4). In fact, a closer look at the WordList and the concordance lines of the Greek Subcorpus of enacting terms (Corpus B) reveals that deontic obligation is expressed with the use of verbs in the present tense of the indicative mood, something that means that the respective rule of the Greek Style Guide (ibid) is also often followed. Last but not least, a significant result is the frequent presence of the Greek noun for obligation, i.e. υποχρέωση <ypochreosi> (see Graphs 3 and 4).

Corpus B (Greek Subcorpus-Enacting Terms): Deontic obligation
Frequency per 1000 words The findings of the quantitative analysis concerning deontic obligation show that the institutional guidelines are often followed. The modal verb shall is used a lot in the English Subcorpora, while in the case of Greek the use of verbs in present tense prevails. However, at this point, a small-scale qualitative analysis was needed in order to re-evaluate the results. For the qualitative analysis, 15 random instances of Greek verbs in present tense were chosen from the Greek Subcorpus of Corpus B and were compared with their equivalents in the English Subcorpus of Corpus B. The findings are displayed in Table 3 and validate that deontic obligation is mainly expressed with the modal verb shall in the English Subcorpus and with the verb in present tense in the Greek Subcorpus.

Corpus A (English Subcorpus): Deontic permission
Permission is mainly expressed in English using the modal verb may, which is the par excellence choice of expressing permission in the English Subcorpus of Corpus A (see Graph 5). The use of may is predicted by the English Style Guide (2018: 55) for the expression of positive permission, namely to give permission to someone to do something. However, as we can see in graph 5, negative permission, i.e. to give permission not to do something, is not expressed mainly with need not as predicted in the English Style Guide (2018: 55). Instead, may not is used. The same applies in the enacting terms of the English Subcorpus of Corpus B (see Graph 6). The modal verb may is very frequent, while need not is not used a lot. Instead, may not is used to express negative permission.

The case of Greek
In the Greek Subcorpus of Corpus A, permission is expressed mainly with the modal verb μπορεί <borei> (3 rd singular) and μπορούν <boroun> (3 rd plural form), which mean may or can. In addition, the modal verb δύναται <dynatai> (3 rd singular) and δύνανται <dynantai> (3 rd plural) also expresses possibility (is able to) and is used mainly in the Greek legal language. Graph 7 summarizes the frequency of μπορεί/ μπορούν and δύναται/δύνανται and other related patterns in the Greek Subcorpus of Corpus A, while graph 8 their frequency in the Greek Subcorpus of Corpus B. As we can see in graphs 7 and 8, μπορεί/μπορούν (<borei/boroun>) is the main means for the expression of deontic permission in both Greek Subcorpora. A closer look at the concordance lines of Corpus B concerning μπορεί/μπορούν reveals that this particular modal is used to grant permission or possibility (e.g. Η Επιτροπή μπορεί = The Commission may or Οι αρμόδιες αρχές μπορούν = Competent authorities may) as well as to deny permission when combined with the negative particle δεν <den> (δεν μπορεί/δεν μπορούν). Another way of expressing permission or possibility is the use of the modal verb δύναται/δύνανται (<dynatai/dynantai>), even though it is not used very frequently in the everyday life, where μπορεί/μπορούν has prevailed (e.g. Η Επιτροπή δύναται επίσης να απορρίψει μια καταγγελία = The Commission may likewise reject a complaint).

Summary of the results and further research
The present study focused on deontic modality in legislative texts concerning the EU Competition Law. For the purposes of the study, an English-Greek bilingual corpus (Corpus A) with legislative texts related to EU Competition Law (mainly regulations and directives) and a bilingual subcorpus containing only the enacting terms of the legislative texts of Corpus A (Corpus B) were compiled and analysed. The analysis of the data was mainly quantitative and was presented in graphs and tables (see sections 4 and 5), while a small-scale qualitative analysis was carried out in order to shed light into a few interesting findings. Although deontic modality is mainly expressed in different ways in English and Greek, there are some points that they have in common. English has a standard system of modal verbs, while Greek uses fewer modal verbs. However, both languages make good use of the simple present tense in order to express obligation, but Greek uses the present simple tense more often for this reason.
As far as the institutional guidelines for both languages are concerned, the results of the study show that the guidelines are often followed in the enacting terms of the legislative texts of the corpora. The institutional guidelines concern mainly, as mentioned above, the enacting terms of legislative texts. The analysis of Corpus B led us to the conclusion that the institutional guidelines are often followed in both languages (see section 5). More specifically, in the English Subcorpora of both Corpus A and Corpus B, the modal verb shall is widely used in order to express deontic obligation (see Biel 2014: 159;Gibovà 2011: 7-8), while other modal verbs, like should, are usually avoided. The simple present tense is also prominent. In addition, modal may is used a lot in order to express deontic permission, something that is also in line with the institutional guidelines. The only divergence from the guidelines was, as mentioned in the subsection 5.2.1., the use of the modal may not instead of need not, which is recommended in the guidelines. In the case of Greek, on the other hand, deontic modality is expressed mainly with the simple present of the indicative mood, something that is in line with the institutional guidelines for Greek. In addition, the modal verb πρέπει/θα πρέπει (must), is widely used for the expression of deontic obligation in Corpus A, but not in Corpus B (enacting terms), something that is also in line with the institutional guidelines. As far as the Greek permission modals are concerned, μπορεί/μπορούν (may) is used a lot in the Greek Subcorpora, while δύναται/δύνανται (is/are able to) doesn't present high frequency 10 . To sum up, we can claim that the results of the study for the language pair of English-Greek show that the institutional guidelines are often followed when drafting and translating an EU legislative text. There are, of course, some divergences (for instance the use of may not to express negative permission), but in general terms the guidelines are followed.
The present study focused on exploring deontic modality in one aspect of legislative texts, i.e. those concerning EU Competition Law, and tried to shed light into it. The study was quite challenging especially because of the nature of deontic modality and the fact that it is expressed differently from one language to another (see Palmer 2001: 18;Biel 2014: 158). The results of the study showed that there is a general convergence between the praxis and the institutional guidelines. However, more studies in this field are needed in order to cast more light in the subject of deontic modality and re-evaluate the results of the study. Finally, the comparison of the results with national legislation of both languages or the comparison between regulations and directives, two different types of binding legal instruments of secondary EU legislation, would be also very enlightening.