Parallel Corpora and ST Analysis: EU Regulations on Immigration in the Specialised Translation Classroom

Starting from the assumption that “if corpora are to play a role in the translation professions of tomorrow, it is important that they impact on the education of the students of today” (Bernardini & Castagnoli, 2008, p. 40), this study endeavours to show how translation corpora of parallel texts (in English and in Italian) can be used in a Specialised Translation Master’s degree classroom. The point of departure is to examine parallel or aligned texts (originals and their translations) taken from the various European Union websites available for citizens to read and consult. The corpora currently being gathered include a variety of text typologies ranging from legal documents (such as regulations or directives), to administrative documents (such as White and Green papers) or informative texts (such as leaflets, brochures or web texts) (Felici, 2010, p. 101), all of them dealing with migration and asylum issues. The various types of documents are analysed quantitatively and qualitatively in class according to the following three main methodologies: Corpus Linguistics (Stubbs, 1996), Genre Analysis (Swales, 1990; Bhatia, 1993) and Systemic Functional Linguistics (Halliday, 1985; Halliday & Matthiessen, 2004). Classroom experiences in the past few years have shown that building ad hoc corpora (or do-it-yourself corpora as coined by Krüger, 2012, p. 514) for classroom consumption is a valuable and precious learning tool which enables students to hone their practical skills in the resource gathering process and consequently in the translation process itself. This study focuses on an EU regulation and shows how a lesson in class can be conducted with students at Cagliari University, Italy. The aim is to get students to work on the quantitative aspects along with the more qualitative linguistic elements of the ST (Source Text) in order to obtain greater awareness and understanding of professional translator strategies used by the professionals in the translation agencies of European Union institutions. From an educational and academic point of view, the linguistic and contrastive analysis of certain features of the ST along with the investigation of specialised terminology associated with contexts of migration and asylum and their equivalents in the TT (Target Text), have so far provided the Master’s students at Cagliari University with useful insights and sound knowledge of the linguistic characteristics of legal and institutional discourse in both English and Italian.

based on parallel corpora, i.e., source language texts in language A and their translated version in language B (Baker, 1995, p. 230), since this is a proven technique for students to identify equivalences, to name just one of the benefits in using texts in this way. The work is organised around three main steps, which are complementary to the actual task of translation: ST analysis, which represents the focus of the present study, is based on corpus analysis, genre analysis and linguistic analysis through Systemic Functional Linguistics. These three methods form the essential background and framework aimed at having students gain awareness of the translation choices adopted by professional translators in the Target Text (TT henceforth), i.e., in Italian. At this point, it is useful to remember that the variety of text types in the EU and "the collective and multilingual drafting process do not support the idea of individual authors and translators. Just as the writers of the documents often remain anonymous, the translators are necessarily unidentified […] To the outside world, EU translators are, and probably will remain, relatively invisible" (Koskinen, 2001, p. 298). Nevertheless, the translated texts represent an example of professional work upon which students can reflect and on which they can draw for future translations of the same type.
The parallel texts (i.e., ST and its translation) considered here are taken from the Multilingual/Bilingual Display of the EUR Lex website, which includes aligned texts used daily in class work. Giordano and Piga (2021), Biel (2010), Bernardini andCastagnoli (2008) andpreviously Zanettin (2001) have shown how parallel corpora are precious resources for discovering well-established translation equivalents of certain legal terms and expressions. Additionally, they can raise students' awareness of professional translation strategies. It is commonly recognized that specialised translation requires a sound knowledge of the two languages, Source Language (SL henceforth) and Target Language (TL henceforth), a profound knowledge of the subject field and firm familiarity with certain legal concepts in both the SL and the TL.
The present investigation considers the English version as the ST, since European documents are mostly and firstly drafted in English and French, as underlined by Felici: "[…], the Commission is the institution that does most of the drafting and translation work and, for practical reasons, this takes place in English and French, usually referred to as 'the procedural languages', though at some other times in German. These are, in fact, the languages in which official documents have to be provided before they can be adopted at a meeting of the Commission. Equal status for all the other languages is only ensured at a later stage by authenticated translations" (Felici, 2010, p. 102). The author adds that, "However, the use of two 'pseudo' source languages is not clear-cut and homogeneous because they both might be used within the drafting of the same document" (Felici, 2010, p. 102).
The qualitative linguistic analysis is a well-established framework taken from, among others, Brusasco's Approaching Translation (2013), which guides the students through the recognition of certain stylistic and pragmatic features, as well as register and genre conventions typical of the text under examination. The qualitative analysis is complemented by a quantitative one.
This study provides an example of the linguistic analysis carried out in the context of translation teaching courses and consequently may be of interest both for translation instructors and for would-be professionals in the field of translation at a European level.

Literature Review
The topic of EU translations and their analysis using parallel corpora, as employed in translator education and training, has been explored by Scarpa (2006 and2008), Bernardini and Castagnoli (2008), Felici (2010), Biel (2010), Pearson (2014), and Scarpa (2020). Bernardini and Castagnoli explain that "[…] the more obvious aspect of corpus work (producing and interpreting concordances, sorting results, inferring semantic preferences and prosodies from collocate sets, building small corpora etc.) within reflective activities that require students and professionals to keep (or start) asking questions about aspects of language and text that corpora can bring to light. Indeed, successful corpus work requires first and foremost an inquisitive frame of mind, a critical attitude and an ability to detect patterns, and only secondarily (some) technical skills" (Bernardini & Castagnoli, 2008, p. 44). The authors add that "[t]he browsing of target language corpora" provides "the translator with an inventory of attested "units of meaning", i.e., conventional ways of expressing specific meanings and performing specific functions in the relevant text type/variety within the target language (TL)" (Bernardini & Castagnoli, 2008, p. 39).
Similarly, Biel argues that " […], parallel corpora allow trainees to study translation techniques, establish equivalents and recurrent solutions proposed by various professional translators rather than to rely solely on the instructor's experience and intuition" (Biel, 2010, p. 13).
Already in 2006 and 2008, Scarpa underlined the importance of using parallel corpora to obtain translation strategies from the observation of how professional translators concretely solved difficulties in past translations. Through concordances and phraseological units, parallel corpora constitute a particularly useful database to be used in order to find the translating routines of the most recurrent phraseology. In 2020, Scarpa clarified that " [b]ilingual (and multilingual) parallel corpora are segmented and 'aligned' so that the linguistic units (such as sentences) in one subcorpus in language A correspond to those in language B. In professional specialised translation, parallel corpora are used in terminology management to create termbases and in statistical machine-translation (MT) tools such as Google Translate and Bing Translate" (Scarpa, 2020, p. 128).
The literature review on linguistic theories and the ST analysis framework that is the object of this study (and will be introduced in the Section about Corpus and Methodology) are based first and foremost on Systemic Functional Linguistics (Halliday, 1985;Halliday & Matthiessen, 2004;Martin, 1991). According to the scholars cited, a clause is a unit in which meanings of three different kinds are combined, each expressing one kind of semantic organization: Field (Clause as representation); Mode (Clause as message); Tenor (Clause as exchange). These contextual aspects are those which may be found in the text and also emerge through the analysis of Jakobson's functions (Jakobson, 1960). Seminal studies on genre analysis such as Bhatia (1993) and Swales (1990) are also considered, inasmuch as they provide useful insights into the pragmatic functions of language in specific domains. The taxonomy developed by Hyland (2005) distinguishing between interactive features, which guide readers through a text, and interactional features, which are used by the writer to comment on content material, account for the metadiscoursive elements in the ST and their transposition in the TT. Cohesion helps students understand the texturing of the text by looking at the various grammatical and lexical networks of relations between and among the several elements within the texts (Baker, 1992(Baker, −2011. Coherence is also looked at in order to understand how the text creates and conveys meaning (Baker, 1992(Baker, −2011. These methodologies go along with corpus linguistics and the quantitative analysis of the texts dealing with immigration. Theories are based on Baker and McEnery (2005), Manca (2015), Grego and Vicentini (2015), Al Fajri (2017) and Piga (2020 and. Almost all the theories cited here, which account for both the qualitative and quantitative linguistic analysis, contribute to forming a framework for ST analysis, as suggested by Brusasco (2013). Identifying and recognizing stylistic and pragmatic features provides a guide for the students; it also enables them to become familiar with the register and genre conventions typical of the text type under scrutiny, i.e., an EU Regulation.

Corpus and Methodology
The text under scrutiny in this study is an EU Regulation on immigration (REGULATION (EU) 2019/1240 of the European Parliament and of the Council of 20 June 2019 on the creation of a European network of immigration liaison officers) in both its English and Italian versions. The quantitative analysis of the ST in English (henceforth REGENG2019) and of some of the aspects in the parallel Italian version (henceforth REGITA2019) are the starting point from which to approach the document and its translation, with the aim of recognizing the translation choices favoured by official translators in the European Union.
The concordance software Sketch Engine (developed by Lexical Computing Limited in 2003) is the corpus manager used for the quantitative analysis. Table 1 shows some basic statistical data such as the number of words, tokens and sentences in the two versions.  Vol. 11, No. 5; counted just once. There are 6,862 types in the text under scrutiny. The TTR (type-token ratio) is calculated as follows: The result of the type/token ratio can be a number between 0 (zero) and 1 (one). If the value obtained is closer to 1 (one), it means there is rich vocabulary content in the text. If the value obtained is closer to 0 (zero), it means that the vocabulary is fairly poor. If the value is equal to 1 (one), it means that the text is made up of just hapax legomena (the singular hapax legomenon indicates a word occurring only once in a given corpus). The value obtained by the calculation in both the English and the Italian versions is near to 1 (see Table 2), which indicates that the vocabulary in both texts is rather rich and varied and this accounts for the lexical variety of both texts (Cresti & Panunzi, 2013).
Sketch Engine also helps students look at vocabulary (which expresses content) and pinpoint the most frequent nouns, verbs and adjectives in the texts in order to carry out an analysis of the lexicon and its occurrences in the documents. Additionally, the corpus manager software helps investigate the metadiscursive features present in the text and calculate how frequently they are used. A thorough perusal of the concordance lines enables students to look at the occurrences and verify whether the features found in the texts are actually working as metadiscursive elements, so that extraneous examples can be disregarded.
Along with these statistical data and the quantitative analysis, students in the Translation Master's course classroom are invited to search and analyse the ST for linguistic characteristics and features listed in the ST analysis framework in Table 3. This framework used in class follows and builds on the model of analysis suggested by Brusasco (2013) and is based on major Linguistics theories and methodologies previously covered in the literature review. It is important to underline that corpus analysis and metadiscursive features are not considered in Brusasco's text analysis. The author herself specifies that "(g)iven the number of variables involved in the act of translation, one has to be aware that there is no hard and fast rule, just as there is no ultimate target text.
[…] Therefore, the procedures illustrated […] are meant as guidelines that can undoubtedly be perfected, just as each translation proposed is but one possible rendering among many" (Brusasco, 2013, p. 59). It could be argued that the corpus used for the analysis is too small, in the sense that the Master's students will be working on a single text and its translation in Italian. However, the text under scrutiny does represent a clear example of the types of ad hoc corpora that Bhatia, Langton and Lung (2004) suggest using in order to apply corpus analysis to texts within the context of legal language. The scholars state that small genre-based corpora are much more useful than large corpora because of the several linguistic peculiarities of legal language (among others, conservatism, crystallized formulae, formulaic form-function correlations, binomials and trinomials, etc.). Therefore, small legal corpora, and in this case parallel corpora of an English document and its Italian version, are equally efficient and reliable from both a quantitative and qualitative perspective. Students are trained to understand that claims and generalisations cannot be made about language from the analysis of a single text, but findings can certainly be representative of the language samples under investigation, as explained by Biel (2010).

Quantitative Analysis
Following López Rodríguez (2016, p. 92), the authors are firmly convinced that "corpus linguistics can facilitate the identification of patterns in text, as well as "rare" instances which may indicate creative uses of language". Firstly, the instructor in the Master's course can suggest search structures to retrieve in the English text, so that students can gain some insights into the translation strategies in the do-it-yourself or ad hoc corpus. Alternatively, students can be invited to read the aligned texts thoroughly at least once in order to grasp the meaning and the context of discourse and then be prepared to analyse them in depth through the concordance software first, and through linguistic qualitative analysis later.  Vol. 11, No. 5; "doing-and-happening" (Halliday & Matthiessen, 2004, p. 181) where the unfolding of the process is projected towards the Goal, whose benefits involve another participant, the "Beneficiary", i.e., the one to whom or for whom the process is said to take place. In Excerpt (11) the "Beneficiary" is semantically denoted by "the third Countries", where the Union through the deployment of immigration officers is committed to protecting human rights and the fundamental freedoms of all immigrants, regardless of their migratory status.
Excerpt (11) Article 1. This Regulation lays down rules to enhance cooperation and coordination among immigration liaison officers deployed to third countries by Member States, the Commission and Union Agencies, through the creation of a European network of immigration liaison officers. (REGENG2019) "Tenor identifies interpersonal dimensions of the event, i.e., the roles played by the participants in the interaction and the social relations between them" (Brusasco, 2013, p. 60). As far as Tenor is concerned, students are asked to analyse how the social relationship between the writer/speaker and the reader/listener is constructed i.e., in terms of power (a)symmetry or equality, and what the relational values are in the vocabulary and in the linguistic elements in terms of mood, modality, speech acts, performativity, constativity, etc. Looking at the Regulation in terms of its relational function (cfr. Halliday, 1994), power asymmetry seems to give way to modalized assessment of equality. As previously noted (Giordano & Piga, 2021), the basic principle seems to be this: the 'commands' are more projected towards negotiation and dialogue. In the text under scrutiny, 'commands' are realized by the modulated clauses, which give more discretion to the addressees. The presence of conditional mood in some way facilitates the potential for negotiation between the EU and its Member States. Instead of the fixed value of the imperative commands 'do it', the text uses the modulated obligation 'you should do' instead. These 'conditional' variants provide a range of a more delicate way of commanding. In very broad terms, the illocutionary force in both the Italian and the English version of the EU Regulation seems to be changing from a 'command' to 'a conditional-indication'. Accordingly, the power relation between the participants in the text is far more equal than asymmetrical (Giordano & Piga, 2021).
"Mode refers to the function of the text in that specific event and includes both the medium through which language is manifested […] and its genre […]" (Brusasco, 2013, p. 60). As regards Mode, namely the structuring of the clause as message, the text is mainly organized in declarative sentences in which the typical pattern at a microtextual level is one in which the Theme is conflated with the Subject, both in a metonymical form, as in the following:

This Regulation lays down rules to enhance cooperation and coordination among immigration liaison officers deployed to third countries by Member States, the Commission and Union agencies, through the creation of a European network of immigration liaison officers. (REGENG2019)
In other cases, the Theme is expressed in a typical generic form of a common name to define experts, representatives of national councils or EU institutions officials: Excerpt (13) Immigration liaison officers need to process personal data to facilitate the proper implementation of return procedures, the successful enforcement of return decisions, and reintegration where relevant and possible. (REGENG2019) In addition to relatively 'local' semantic relations between clauses and sentences, students are also asked to look at and identify more 'global' and higher-level semantic relations over longer stretches of text. One very common example in this EU Regulation is the problem-solution relation (Hoey, 2001), which is built around the following pragmatic-semantic relationship. The 'problem' is the need, the necessity or the challenge that the Union and its Member States have to face in terms of strengthening the policy in the field of migration. The 'solution' is then represented by the effective response of the EU aiming to achieve an all-inclusive move aimed at ensuring an efficient organization of migration flows: Excerpt (14) The

Propositional Meaning Vs. Expressive Meaning
The difference between propositional meaning and expressive meaning lies in the character that words have in the text under scrutiny, in this case an EU Regulation. As Brusasco reminds us (2013, p. 96), most words convey propositional meaning, i.e., what a word refers to in the world. Baker (1992Baker ( −2011) explains that "(t)he propositional meaning of a word or an utterance arises from the relation between it and what it refers to or describes in a real or imaginary world", while "expressive meaning relates to the speaker's feelings or attitude rather than to what words and utterances refer to".
As can be noticed even from a cursory glance at Figure 1 above, the words with propositional meaning in this context are immigration, officer, liaison, union, regulation, member, country, council and so on. Words in Italian conveying propositional meaning in the same context are immigrazione, funzionario, collegamento, unione, regolamento, etc. The absence of adjectives might be said to account for poor expressive meaning in the Regulation, though this is in fact not the case. As shown in Figure 5, qualifying adjectives are used to express the writers' opinion and evaluation; some examples are relevant, appropriate, necessary, possible, effective and fundamental. Examples of qualifying adjectives carrying opinion and evaluation in Italian are efficace, fondamentale, possible, pertinente, etc. Consequently, the writers' thoughts and feelings are expressed using the emotive function, which we will look at more closely in the section on Attitudinal markers.

Grammatical Cohesion
At this point students are asked to analyze the grammatical cohesion of the text, i.e., features indicating the relationship that exists between linguistic expressions, exemplified through substitution, ellipsis and conjunctions, etc.
In the mechanism of substitution, one element is replaced by another. In English, the most common substitution particles are do, one and the same, but all auxiliaries in short answers, which by themselves provide for the omission of the main verb, also fall into this category; these grammatical elements are usually used to avoid repetitions in the text, unless they are specifically desired by the author. In the Regulation, no cases of substitution are found, full lexemes are repeated in order to avoid ambiguity.
Ellipsis, on the other hand, implies the omission of a certain element in the sentence, i.e., it is completely removed. However, it should be emphasized that since ellipsis is an omission, there is no particular element (as in the case of reference) that tells the addressee where to retrieve the missing information, but it is the grammatical structure itself that indicates what element or elements are appropriate to fill that particular "blank space" or "slot" (Baker, 1992(Baker, −2011.
In the text under scrutiny, no relevant cases of ellipses are found; moreover, although they are occasionally present, they do not seem to be used deliberately by the author. Indeed, the presence of omissions is not due to any particular strategy, but rather to an inherent feature of the English language, i.e., the tendency to omit certain elements to avoid repetitions that would burden the fluidity of the text as a whole. In reality, it is the conscious choice of 'non-omission' that is the strategic choice of lawmakers, aimed in this case at highlighting certain constituents in the text or to give it a meticulous regularity.
The mechanism of conjunction implies-as the word itself suggests-the use of conjunctions to connect sentences, propositions and paragraphs. Unlike reference, substitution, and ellipsis, this category does not require the recipient to retrieve missing information, but signals the way the sender wants the reader (or listener) to relate what is about to be said to what has been said previously (Baker, 1992(Baker, −2011. For the purpose of this Master's lesson on translation, and according to the ST framework provided, students have to analyse four main types of relationships that serve to give cohesion to the text: -Additive: and, or, also, furthermore, etc.; -Adversative: but, yet, however, etc.; -Causal: so, since, for, because, for this reason, etc.; -Temporal: then, next, in conclusion, after that, on another occasion, an hour later, finally, at last, firstly, secondly, thirdly; In the Regulation most of them fall into the additive 'and' category, as shown in the section about Transition markers. An example is the following Excerpt (15). The less frequent conjunctions belong to the causal category, with only some occurrences of since (2 occurrences).

third countries of return are not often subject to adequacy decisions adopted by the Commission under Article 45 of Regulation (EU) 2016/679, and have often not concluded or do not intend to conclude a readmission agreement with the Union or otherwise provide for appropriate safeguards within the meaning of Article 46 of Regulation (EU) 2016/679. (REGENG2019)
The additive conjunction also is found 5 times in the Regulation, as will be shown later in this paper. In the Excerpt below, also modifies a temporal clause.
Excerpt (16) Article 2 Definitions For the purposes of this Regulation, the following definitions apply: (1) 'immigration liaison officer' means a liaison officer designated and deployed abroad, by the competent authorities of a Member State, or by the Commission or by a Union agency, in accordance with the respective legal basis, to deal with immigration-related issues, also when that is only a part of their duties. (REGENG2019)

Lexical Cohesion
Lexical cohesion refers to the role of vocabulary choice in organizing the various connections within a text. Although a lexical element does not possess a cohesive function in itself, it establishes a (semantic) relationship with other elements present in the text; consequently, this type of cohesion occurs whenever a given element recalls a previous one, at a semantic level. In this part, students are mainly encouraged to scrutinize reiteration and collocation.
Reiteration-as the name suggests-implies "a repetition of an earlier item, a synonym or near-synonym, a superordinate" (Baker, 1992(Baker, −2011). In the Regulation under investigation synonyms or near synonyms are not present and the repetition of full lexemes is preferred in order to avoid ambiguity. Conversely, superordinate elements are frequently present in the text, The Union, the Members States, etc.
As regards collocations, Baker (1992Baker ( −2011) explains them as words which have the tendency to co-occur regularly in a given language, or in this case in a given text. Typical collocations found in this text are: immigration liaison officers (81 occurrences), Steering Board (40), Member States (36); third countries (33), personal data (26), Union agencies (21), migratory flows (7), European network (5), external borders (4) and so on. All of them concur to make the text a coherent whole and account for the general meaning and topic covered in the Regulation. In the same context, it is possible to note in the Italian version the following collocations: i funzionari di collegamento incaricati dell'immigrazione, comitato direttivo, Stati membri, paesi terzi, dati personali, etc.

Coherence
As for coherence, the ST framework invites students to look at the network of conceptual relations that underlie the surface text and, in so doing, to identify the reader's expectations and experience of the world. In other words, students have to discuss how the text creates meaning and makes sense to a potential reader. Here, students will not be concerned with networks of cohesive devices, in which surface relations link words and expressions in a text, but with the way utterances are used pragmatically in the communicative situation in the context of a Regulation. In the text under scrutiny the absence of pronominal reference and the employment of direct repetition of full lexemes creates a continuity of sense and meaning that makes the text directly accessible even to a layman reader.

Metadiscourse
On the basis of Hyland's classification (2005), the various metadiscoursal features and their functions can be analysed following the indications in the ST analysis framework provided.

Interactive Features
Interactive features such as transitions, frame markers, endophoric markers, evidentials and code glosses deserve particular attention as they are deemed to be as thoroughly and systematically representative of the EU regulation genre.
To begin with, transition markers are "mainly conjunctions and adverbial phrases which help readers interpret pragmatic connections between steps in an argument" (Hyland, 2005, p. 50). As can be seen from Table 4, the most frequently used transition marker is the conjunction and, which can be employed either to separate two or more words or as an additive conjunction to separate coordinated clauses, as explained in the Section about grammatical cohesion. Also is present only 5 times and therefore 4 times. Students thus become aware that the text under scrutiny contains relatively few conjunctions, as confirmed by the analysis of grammatical cohesion above. ijel.ccsenet.org International Journal of English Linguistics Vol. 11, No. 5;  According to Hyland (2005, p. 51), "(f)rame markers signal text boundaries or elements of schematic text structure". They can sequence the text through numbered or bulleted orders, they can label stages, announce goals and shift topic. As a matter of fact, clauses and sentences in the Regulation are juxtaposed in ordered and numbered sections or Articles and this is one of the reasons why there are so few coordinating or even subordinating conjunctions or transition markers.

THE EUROPEAN PARLIAMENT AND THE COUNCIL OF THE EUROPEAN UNION, […]
HAVE ADOPTED THIS REGULATION: Article 1 Scope

2) This Regulation is without prejudice to the responsibility of Member State authorities, the Commission and Union agencies for defining the scope and assignment of tasks and reporting lines of their respective immigration liaison officers, and to the tasks of immigration liaison officers within the framework of their responsibilities under Union and national law, policies or procedures, or under special agreements concluded with the host country or international organisations.
While there are no frame markers with the functions of labeling stages or announcing goals in the text, frame markers used to shift topic, e.g., regard is found 14 times through the corpus, most of all in the expressions having regard, as regards, with regard and regarding.  Vol. 11, No. 5; Per quanto riguarda l'Islanda e la Norvegia, il presente regolamento costituisce uno sviluppo delle disposizioni dell'acquis di Schengen ai sensi dell'accordo concluso dal Consiglio dell'Unione europea con la Repubblica d'Islanda e il Regno di Norvegia sulla loro associazione all'attuazione, all'applicazione e allo sviluppo dell'acquis di Schengen (9) che rientrano nel settore di cui all'articolo 1, lettere A ed E, della decisione 1999/437/CE del Consiglio (10). (REGITA2019) Endophoric markers are expressions which refer to other parts of the text. The expression In this Regulation is found 4 times and there are additional references to other subparagraphs and articles in the Regulation itself, as shown in Table 6. Excerpt (20) Considering the priorities of the Union in the area of immigration and within the scope of the tasks of immigration liaison officers as defined in this Regulation and in accordance with Article 1(2), the Steering Board shall carry out the following activities… (REGENG2019) Drawing from Thomas and Hawes (1994, p. 129), Hyland defines evidentials as "metalinguistic representations of an idea from another source" (Hyland, 2005, p. 51). One of the most frequent structures in the Regulation in the English version is the evidential in accordance with: Table 7 shows that this feature is found 33 times. Table 7 also shows the translation into Italian: the adverb conformemente is the most frequently used to translate in accordance with, followed by expressions such as in conformità, ai sensi, a norma and so on (as already found and discussed in Giordano & Piga, 2021). Another evidential is to be found in (REGENG2019) and this is pursuant to which appears 4 times and is translated a norma and ai sensi in (REGITA2019) (see Table 8). Code glosses "supply additional information, by rephrasing, explaining or elaborating what has been said, to ensure the reader is able to recover the writer's intended meaning" (Hyland, 2005, p. 52). In the Regulation under examination here, as defined is used 5 times and means is used 4 times to provide certain definitions which apply to the document.

Excerpt (21)
Joint deployment of liaison officers should be promoted with the aim of enhancing operational cooperation and information sharing between Member States, as well as to respond to needs at Union level, as defined by the Steering Board… (REGENG2019)

Interactional Features
When dealing with interactional features, Hyland includes hedges, boosters, attitude markers, self-mentions and engagement markers.
Hedges are devices which indicate the author's decision to recognize and tentatively admit potential alternative viewpoints and voices and thus "withhold complete commitment to a proposition" (Hyland, 2005, p. 52). Hedges highlight the subjectivity of a position by allowing information to be presented as an opinion rather than an objective fact and therefore open that position to negotiation. These different scales of obligations are defined as "modulation" (Halliday & Matthiessen, 2004, p. 147); in this sense, modal auxiliaries, whether deontic or epistemic, can be interpreted as hedges whenever they express a certain degree of vagueness (Hyland, 2005). Given the common propensity of lawmakers to avoid ambiguity and subjective interpretation, preventing obscure areas where possible, the role of should in prescriptive legal texts, in which 'what they say is law', is expected to be reasonably constrained and restricted (cfr. Williams, 2005). However, as shown in Giordano and Piga (2021), the deontic discretionary force of the modal should occurs pervasively in EU Regulations. Conversely, together with the occurrences of should conveying "modulation" (Halliday & Matthiessen, 2004, p. 147), namely showing the presence of intermediate possibilities, the use of shall accounts for the presence of items that avoid ambiguity and normally convey an unmistakably prescriptive dimension to verbal constructions.
Another modal students are invited to look at and which plays a major part in EU Regulations is may. In this Regulation may appears 12 times. In EU Regulations, as in legal discourse in general, one of the primary meanings of may is that of conferring discretionary power and mild obligation, either directly or indirectly: Excerpt ( Excerpt (26) Gli Stati membri possono altresì convenire che i loro funzionari di collegamento incaricati dell'immigrazione si ripartiscano taluni compiti tra loro, sulla base delle competenze e della formazione.
However, in subordinate clauses, unlike the modal shall, which generally adds no particular shades of meaning to the verbal expression other than emphasizing its polarity and legalese nature (Williams, 2005), may in certain circumstances expresses modal values that do not seem to be directly connected to discretionary power or prohibition but rather to epistemic possibility (Palmer, 1990, p. 50). Consider Excerpts (27) and (28) Excerpt (27) Since Therefore, students are guided by the instructors in the understanding of the use of the modal may which expresses deontic discretionary power when it occurs in the main clauses, and epistemic possibility when it occurs in subordinate clauses.
In the ST analysis framework provided to students, only attitude adjectives are taken into consideration. In the corpus under scrutiny, interpersonal attitudinal adjectives are present in both the form of 1) positive adjectives of the attributive type, such as effective in Excerpt (29), and 2) of the predicative type, as in possible in Excerpt (30) below.
Excerpt (29) To ensure the effective implementation of all aspects of Union policies on immigration, consistent dialogue and cooperation should be pursued with key third countries of origin and transit of migrants and of applicants for international protection. (REGENG2019) Excerpt (30) Despite the extensive efforts of the Union in cooperating with the main countries of origin of illegally staying third-country nationals subject to an obligation to return, it is not always possible to ensure that such third countries systematically fulfil the obligation established by international law to readmit their own nationals. (REGENG2019) Moreover, it is also important to point out that in the Regulation under scrutiny, expressions of attitude have the tendency to be prosodically coupled throughout the clause. Some attitudinal adjectives tend to be toughened in binary constructions with other attitudinal adjectives forming accumulations and enumerations that create rhetorical figures, all contributing to the same meaning-function. By adding (quasi)synonyms, intensifiers and the like, the aim is that of ideally doubling-up the interactional meaning-function of the metadiscourse. In the following excerpts, the doubling of the adjectives serves to enhance the prosody of the noun group: Excerpt (31) […] migratory flows originating from or transiting through the third country, including where possible and relevant, the composition of migratory flows and migrants' intended destination. (REGENG2019) Excerpt (32) […] flussi migratori provenienti dal paese terzo o in transito sul suo territorio, compresa, ove possibile e pertinente, la composizione dei flussi migratori e la destinazione prevista dei migranti; (REGITA2019) Excerpt (33) Immigration liaison officers need to process personal data to facilitate the proper implementation of return procedures, the successful enforcement of return decisions, and reintegration where relevant and possible.
Excerpt (34) I funzionari di collegamento incaricati dell'immigrazione devono poter trattare dati personali per facilitare la corretta attuazione delle procedure di rimpatrio, l'esecuzione effettiva delle decisioni di rimpatrio e il reinserimento ove opportuno e possibile. (REGITA2019) As well as enhancing the prosody of the text, the build-up of such "clusters of attitude markers" (Hyland, 2005, p. 150) amplifies the positive tone of the relational meaning-function of the text, thus helping to create a convincing EU discourse that aims to establish shared attitudes and values with readers and draw them "into a conspiracy of agreement so that it can be difficult to dispute such judgement" (Hyland, 2005, p. 150).  Alongside the notable frequency of effective, possible and relevant, there are also 14 occurrences of the word appropriate. Appropriate is another example foregrounding the specific judgement of authors. In Excerpt (35), the lawmakers establish their subjective and personal claim about practices and procedures concerning the "safeguards for the transfer of data", increasing once again the explicit evaluation and the personalized attitude of the text. Boosters are words which "allow writers to close down alternatives, head off conflicting views and express their certainty in what they say" (Hyland, 2005, p. 52). Therefore, through the use of boosters, the authors of the Regulation manage to narrow down the field of options, conveying certainty in their arguments. Even if they present several alternatives, they take a position on only one. As shown in Excerpt (37) The establishment of a robust mechanism that ensures better coordination and cooperation among all liaison officers dealing with immigration issues as part of their duties is essential in order to minimise information gaps and duplication of work and maximise operational capabilities and effectiveness.
Excerpt (38) L'istituzione di un solido meccanismo che assicuri un migliore coordinamento e una migliore cooperazione tra tutti i funzionari di collegamento incaricati dell'immigrazione nell'ambito delle loro funzioni è fondamentale per ridurre al minimo le lacune nelle informazioni e la duplicazione dei lavori e per ottimizzare le capacità operative e l'efficacia. (REGITA2019) Excerpt (39) […] it is not always possible to ensure that such third countries systematically fulfil the obligation established by international law to readmit their own nationals. (REGENG2019) 4.8.5 Self-Mentions "Self-mentions refer to the degree of explicit author presence in the text measured by the frequency of first-person pronouns and possessive adjectives" (Hyland, 2005, p. 53). At this point, students are asked to pinpoint which metadiscoursal devices explicitly indicate the writer's presence and engagement in the text. As regards regulations in general and the Regulation analysed here in particular, such personal engagements are, as expected, entirely absent. This is because members of the EU Parliament, as well as members of the Council, downplay their personal role by electing to adopt a less intrusive or personal style suggesting that the correct procedures and authority of rules and regulations are laid down in the norms of the discipline and thus cannot be altered by any individual. However, in order to highlight the reproduction of the rules and regulations, a 'self-mention' rhetorical stance of the cause-effect metonymy is commonly employed, as can be seen in the two Excerpts that follow: Excerpt (40) In accordance with the principle of proportionality, as set out in that Article, The metonymy This Regulation is found 38 times in the Regulation under scrutiny.

Engagement Markers
"Engagement markers are devices that explicitly address readers, either to focus their attention or include them as discourse participants" (Hyland, 2005, p. 53). In this section, students are asked to identify and analyse devices that refer to the various ways authors bring readers into the discourse in order to better relate to them. One of the most explicit way in which readers are brought into the text is, for example, the use of personal pronouns (you, inclusive or exclusive we, etc.). Since in this specific case students have to analyse a Regulation and Regulations claim authority, this sense of communality addressing the reader from a position of confidence, ijel.ccsenet.org International Journal of English Linguistics Vol. 11, No. 5;2021 124 setting-up collegiality and dialogue, is not present in the text under scrutiny.

Aspects of Specific Discourse
When dealing with specific discourse, especially in the context of a translation classroom, students have to become aware of the use of certain specific structures which are to be found in the text under analysis. Some examples of specific discourse have already been noted, i.e., in accordance with and pursuant to. Another example of legal discourse is the expression without prejudice, which appears 4 times in the ST. In order to understand how official professional translators render this structure in the Italian version, students undertake an intensive reading of the Regulation, in order to identify the various translations of the phrase in the parallel text and therefore the choices made by the translators. In this way, they discover that the 4 occurrences of without prejudice are rendered in 4 different ways, with fatte salve, lascia impregiudicati, fatto salvo and non pregiudica, as shown in Table 12 below and the Excerpts that follow, taken from the aligned texts. Excerpt (43) is the translation of (42), and (45) is the translation of (44). The without prejudice clause acts as a safeguard in the presence of potentially damaging or incriminating statements (Alcaraz & Hughes, 2002, p. 166), and in the Excerpts shown here it always preserves and upholds the responsibilities and the competences of Member States.
To conclude, Alcaraz and Hughes (2002, pp. 165−167) underline that in building up two-way repertoires of equivalents in the field of law, translators might usefully consider the various textual arrangements, the specific collocations, adverbial phrases, but also functional vocabulary used to interconnect the major linguistic elements or meaningful words in a text, such as adverbs, conjunctions and prepositions. This proves to be particularly relevant in the context of a Translation classroom, since students who want to become professionals in the field of legal translation necessarily have to understand how, starting from a deep and accurate ST analysis and working with original texts, to systematize lexical and grammatical items in order to develop the contrastive vocabulary and equivalent structures to be used in the TT.

Conclusions
The study has presented one of the possible ways in which ST analysis can be conducted in a Translation Master's course with students who are preparing for the profession. It is both an academic and a professional training context in which parallel texts are used in order to understand the translation choices opted for by ijel.ccsenet.org International Journal of English Linguistics Vol. 11, No. 5; professional translators in the context of the European Union and to be able to replicate them when dealing with similar specific texts or working in analogous environments at European or world level. The ST analysis is the result of a combination of quantitative measurements and qualitative investigations of some of the most important structures in the texts, at a lexical and grammatical level, but also at a syntactic and textual level, beyond the sentence frames. In this way, students not only obtain a greater understanding of legal and specific linguistic configurations, but also gain a profound awareness of how professionals operate in the context of translations aimed at the European public at large. In the meantime, they increase and expand their knowledge of and familiarity with the most relevant linguistic theories and methodologies which can be applied to texts in the field of translation work.