Translation and clarity in Swiss multilingual legislation: A qualitative study on textuality

: Debate around clear legal language has become very prolific in recent years. This especially applies to multilingual contexts, where clarity is crucial to ensuring the equivalence of equally binding language versions. Clarity is a multifaceted concept that can be analysed through different lenses. While several studies have tackled clarity from the lexical and terminological perspective, textuality remains a largely under-investigated area. This paper sets out to illustrate that different concepts of text linguistics can be extremely useful for investigating multilingual legislation and its level of comprehensibility. To this aim, a qualitative pilot study was designed. It examines a sample of Swiss federal legislative acts in their Italian, French and German versions and tries to reveal recurrent micro-textual aspects that help illustrate how translation can influence text comprehensibility. It was found that the three language versions often formulate the same legal provision using different linguistic structures. Slight changes, for instance in thematic structure, position of adverbials or use of punctuation, can create a canonical and more comprehensible formulation in the translated texts, without hampering equivalence at the content level. In this respect, it is key that drafters of legislation and translators are aware of the relevance and potential of textuality to ensuring a high level of clarity in multilingual legislation.


Translation and clarity in Swiss legislation
In multilingual countries, such as Switzerland, translation is fundamental to the proper functioning of institutions.Besides making the law accessible to all linguistic communities, translation also helps improve the quality of both lex ferenda and lex lata (Schnyder, 2001, pp. 43-45).The latter case refers to multilingual law as a precious instrument for judges, who can rely on different language versions of the same act to interpret the "law in force"; the former pertains to the different improvements that translation into another language can stimulate at the multilingual drafting stage, thus enhancing the "future law" as well (Egger & Ferrari, 2016, pp. 509-511).The Swiss context is particularly well suited to illustrate these two concepts.In Switzerland, the vast majority of federal legislative acts are drafted in German and some in French; Italian is almost systematically a translation language (Zwicky & Kübler, 2018, pp. 17-21). 1 The availability of trilingual legislative acts seems to be particularly appreciated by the Federal Supreme Court, which often relies on a comparison of the three language versions to correctly understand the legislator's intent (Lötscher, 2009, pp. 379-384;Schubarth, 2001).Furthermore, in case of divergences, it often bases its decisions on the Italian version of a legal act (Burr, 2000), since texts in Italian often feature a high level of clarity thanks to its peculiar position as a translation language (Schweizer, Baumann & Scheffler, 2011, p. 32).The beneficial effect of multilingualism is even more marked at the drafting stage, when translation imposes a pre-interpretation (Flückiger, 2005) of the source legal text and often helps identify ambiguities and shortcomings (FOJ, 2007a, p. 351).These faults can be avoided in the target text, and sometimes also rectified in the source text as well.In other words, the idea of "translation as a catalyst for clear legislation" (Canavese, 2022), before and after the adoption of a legislative act, seems to apply perfectly to the Swiss context.
Although quality in Swiss legislation has been investigated extensively in the last three decades, systematic empirical studies remain scarce.Since 1990, the Swiss specialised journal LeGes 2 has provided a platform for scholars and practitioners from the legal and the linguistic domains to converge and discuss quality-related issues.Moreover, in recent years, some interdisciplinary research projects have been carried out that have shed light on the multilingual law-making process and clarity (Schweizer & Borghi, 2011).However, it was not until very recently that the first corpus-based projects started to appear.One of them, described in Canavese (2019), aims for instance to investigate the level of linguistic clarity of Swiss federal legislation in Italian with LEX.CH.IT, an ad hoc corpus of all Swiss federal acts from the nineteen-seventies to the present day.
Linguistic clarity can be defined as the sum of two properties, readability and comprehensibility (as suggested by Piemontese, 1996, pp. 79-122).Readability, on the one hand, is a quantitative measure and provides a picture of potential hurdles that may jeopardise the reading process.These may concern the shallow level, i.e., the mean sentence and word length, the lexical level, e.g. the use of archaic and difficult words and expressions, and the syntactic level, e.g. the use of convoluted sentences. 3Comprehensibility, on the other hand, is a qualitative property and deals rather with how information is structured and presented in order to be accessible to the intended addressee.In this respect, comprehensibility is closely linked to textuality.A text is comprehensible if the logical links between the information units are easy to grasp and if referents are introduced and referred back to in an intuitive manner, for instance by moving from "given" to "new" information. 1Even if in some cases legislative acts are co-drafted in German and French, the dominant position of German is irrefutable.Despite this, the three language versions are equally binding (see art. 14 para.2, Publications Act of 18 June 2004, CC 170.512) and should therefore be referred to as language versions.In this paper, I sometimes also use source/target versions or translation.Even if these terms are improper, they allow me to refer more easily to French and Italian, which, at the drafting stage, are often the result of a translation. 2Available at https://leges.weblaw.ch/legesissues/2020.html (last accessed 27.10.2020). 3Cf.e.g.Canavese (2021;2022) for some analyses related to lexical readability based on LEX.CH.IT, in which Swiss legislation is analysed also in comparison with other italophone institutional contexts.Other papers on further aspects, such as syntactic complexity, are currently in preparation.Cf. also the author's doctoral dissertation on this topic (Canavese, 2023).
However, this text-based component, which is related to cohesion and coherence, does not suffice to evaluate comprehensibility in all its facets.Comprehensibility is indeed a relational property as well, which varies according to the recipients, their previous knowledge of the subject matter, their ability to make inferences and their (a)symmetry with the text producer (cf.also Christmann & Groeben, 2019;Schnotz, 2000 for a more detailed description of the different components of comprehensibility).Only by taking into account both sides of the coin, i.e., readability and comprehensibility, can we assess the level of clarity of a text.
Even though the textual level is so crucial for comprehensibility, it is often neglected in studies on legal language (as pointed out by Nussbaumer, 2000;Visconti, 2009), most of which deal prominently with lexis and, more specifically, with terminology.Similarly, guidelines on clear legal and administrative acts tend to relegate textuality to just a few indications and rarely address it in a systematic manner.In general, the only aspect of textuality they address is the macrostructural organisation of the act and its division into sections and articles.They rarely discuss, however, the micro-textual level and aspects such as the thematic structure or the position of adverbial phrases and clauses within any given utterance.This is the case, for instance, in one of the main sets of Swiss legal drafting guidelines, the Gesetzgebungsleitfaden/Guide de legislation/Guida di legislazione, which devotes a short section to language and makes few remarks about the textual level. 4Legal drafting guidelines from other countries devote longer sections to language, such as the German Handbuch der Rechtsförmlichkeit (BMJV, 2008) or the Italian Regole e suggerimenti per la redazione dei testi normativi (Gruppo di lavoro per la revisione del manuale regionale di drafting, 2007) and Guida alla redazione degli atti amministrativi (Ittig & Accademia della Crusca, 2011), but their indications on textuality are just as sparse.
An increased awareness of the importance of textuality in legislation has recently arisen in Switzerland.Recent publications have stressed the importance of combining legal theory and text linguistics when drafting accessible legislation (Höfler, 2016;Lötscher, 2011).As illustrated by Höfler (2017;2019), analysing legislative acts from a textual angle can make their function clear and ensure comprehensible thematic structures and connections between propositions, as well as appropriate referential devices. 5These studies largely draw from the direct experiences of the Drafting Committee, thus presenting real-life examples and making it possible to capture not only the result, but also the process of legal drafting.However, they revolve exclusively around the German version of federal legislation and do not address the contribution of translation to textuality and 4 See FOJ (2007a, pp. 375-382) for the French version and FOJ (2007b, pp. 373-388) for the German version of the guide.It suggests, for instance, to strive for coherence, clarity and concision and provides advice on how lexical and syntactic choices can enhance the textual construction of a legislative act.In the new edition of the guidelines from 2019, also issued for the first time in Italian, the section on clear legislative language has been omitted.This section has been made into a separate document and enriched with bibliographical references by the German Section of the Central Language Services of the Federal Chancellery, which published it on its website (https://www.bk.admin.ch/bk/de/home/dokumentation/sprachen/hilfsmittel-textredaktion/gesetzessprache.html,retrieved 02.11.2020).On the same page, a short guide on coherence in legislative texts can also be found. 5These aspects are analysed in great detail in a series of publications.These publications discuss the potential of text linguistics to inform for instance on the use of presuppositions (Höfler, 2014), the types of references to other legal sources (Höfler, 2015) or the topiccomment structure (Höfler, 2016) in order to enhance understanding.comprehensibility.Moreover, they adopt a theoretical angle rather than an empirical, corpus-based one.This paper builds on these previous, invaluable studies and tries to add a further layer to the analysis of textuality and comprehensibility, by considering them from the translation perspective, and to lay the foundations for further empirical investigations.To this aim, an exploratory qualitative study of comprehensibility in its text-based component was designed and carried out based on a small trilingual sample extracted from LEX.CH.IT.

Study design
The main aim of this paper is to show how a range of text linguistics concepts can be successfully applied to the study of multilingual legislation.More specifically, they can be particularly helpful for determining the influence of translation on text comprehensibility.In fact, despite the highly binding context of legal texts, different versions of the same legislative act often turn to different textual structures to convey the same regulation.These differences can sometimes be explained by adopting a contrastive perspective; they can be determined by the intent to resort to the canonical linguistic structures of the target language and to avoid calques, i.e., literal translations that result in unnatural formulations.In other cases, these differences can also be ascribed to the willingness to improve textuality.In many cases, however, both aspects come into play simultaneously and it is not always possible to distinguish them in a clear-cut way.In all three possible scenarios, the result is a language version that expresses the same norm in a comprehensible, and sometimes even a clearer way.For this reason, the contrastive perspective and the intent of improving textuality during translation will be addressed simultaneously here.To provide a complete overview, some cases in which translated texts result in a less clear formulation will also be presented.
This study is of a prominently explorative and descriptive nature; therefore, it will discuss a wide range of aspects of textuality without aiming at analysing them in an exhaustive way.Further in-depth investigations on each aspect will be carried out in future related work.These aspects concern microstructural elements in particular, such as the information structure of utterances and different cohesive devices adopted to ensure comprehensibility.In fact, the macrostructure of a legal text is rigid and does not change between the three language versions.A legislative act is divided using titles, chapters, articles and paragraphs, and translation has no influence on this level of textual construction or, at least, it is not possible to discern what changes were suggested during the translation process once an act is adopted.
Methodologically, a qualitative, text-based approach has been designed to carry out this study.A small sample of Swiss federal acts has been constituted based on solid criteria.The starting point for building the sample was LEX.CH.IT (see section 1).First, only recent legislation (2014 and later) was considered, so as to control for a potential diachronic variable.A minimum number of five texts was set.I discarded both texts that were too short and those that were too long, which would be poor in relevant aspects and too time consuming to process manually, respectively.To do so, I resorted to descriptive statistics and only selected texts with a length falling in the second and third quartile.The most productive legal fields6 were identified, i.e., those that contain at least 10% of all federal acts collected in LEX.CH.IT: citizenship, school, public transport, healthcare, international economic cooperation.After applying these filters, the following acts could be selected: They were retrieved from the Official Compilation of Federal Legislation in their first version and then aligned.The sample consists of a total of around 10,000 tokens per language.
I carried out an interlingual comparison of the legal acts making up the sample in their German, French and Italian versions.Through a manual coding carried out with the help of MAXQDA (Kuckartz and Rädiker, 2019), some relevant differences at the textual level between the three language versions were annotated, extracted, analysed and commented on.This made it possible to capture cases in which one of the three versions of a federal act may have gained or lost in clarity.Starting from a wide range of potential aspects, some of those that were the most frequent and relevant for shedding light on the influence of translation on textuality and comprehensibility were identified.This laid the groundwork for creating, following a corpus-driven logic, a preliminary coding framework, which will be used in further studies to annotate a larger sample of legislative acts.

Conceptual framework
When working on textuality, some conceptual and terminological fragmentation is found in the literature.I will first present the conceptual framework adopted here to analyse textuality, i.e., the "Basel Model of paragraph segmentation" (Ferrari et al., 2008;Ferrari, 2014), before moving on to the results of the study.Attempting a description of this model in few lines is impossible.Therefore, I will only present its underlying assumptions and the main concepts that I will use in my analysis.This highly articulate and complex model allows for the textual analysis of nonliterary texts at the paragraph level.In my view, the concept of "paragraph" as intended in the Basel Model corresponds to that of "article" when working on legislative texts.Moreover, the term "paragraph" is used in legal language to refer to the subsections that constitute an article.When commenting on the examples in the next sections, I will use the term "paragraph" in the accepted legal sense of the word.
The Basel Model provides a conceptual framework to identify the semanticopragmatic units that make up a text and to highlight their connections and hierarchical structure.According to the model, a paragraph is divided in utterances. 7Utterances not only have an illocutionary force, but also fulfil a textual composition function, in that their harmonic combination makes it possible to create a coherent text.Each utterance comprises one or more information units, which divide the semantic contents of the utterance into hierarchically ordered segments.Syntactically, information units correspond to one or more clauses and/or phrases.Three types of information units can be discerned: nuclei, frames and appendixes.
The nucleus is the core of the utterance and is sufficient to convey its illocutionary force; in legal and other specialised texts, it often corresponds (at least) to the main clause.The nucleus can be complemented with optional background information packed in frames and appendixes.Frames, on the one hand, are placed at the beginning of the utterance, and contain, for instance, connectives or adverbial indications and express the settings in light of which the nucleus is to be read and understood.Frames can also be used to ensure that the new utterance is introduced in a cohesive way in the co-text or to introduce a topic.Appendixes, on the other hand, are placed within or after another information unit and expand the content thereof, thus fulfilling a narrower scope than frames.In this respect, one should note that punctuation plays a crucial role in segmenting and hierarchising the content of an utterance.As clearly shown by other works of the aforesaid Basel team (cf.e.g.Ferrari et al., 2018;Ferrari, 2018), in Italian the use of punctuation is determined by communicative and textual reasons rather than by syntactic or prosodic considerations.This also holds true for French, but not for German, in which for instance the use of commas is mainly determined by syntax (Ferrari e Stojmenova, 2015).This explains why it is crucial to take punctuation into account when working on textuality.
Besides the segmentation and hierarchical organisation of the utterance discussed so far, two further levels need to be considered to fully grasp its information structure.The topic-comment level refers to the concept of aboutness in terms of Lambrecht (1994).The topic is to be understood as the textual referent the utterance is about, whereas the comment is the piece of information given about the topic.A last level concerns the degree of cognitive activation of referents (in the sense of Chafe, 1987).A referent can be active, inactive or semi-active in the discursive memory based on whether (and where) it has been introduced or last mentioned in the co-text.
The following example should help the reader grasp the various concepts outlined above.Here, both levels of information structure that will be explored in this paper are annotated.First, background information (i.e., frames and appendixes) and foreground information (i.e., the nucleus, without any underlining) is made visible.Second, the topic-comment structure is marked by adding a subscript "Topic" and "Comment" next to the relevant text chunk.In the first of the two utterances making up paragraph 4, for instance, the topic is the referent the Federal Assembly, whereas the comment corresponds to the rest of the nucleus.
Example 1 Basel Model's main concepts applied to article 4 of the Swiss Higher Education Act (CC 414.20) 1 The ConfederationTopic is responsible for leading coordination of the joint activities of the Confederation and the Cantons within the higher education sectorComment.
[…] 4 The Federal AssemblyTopic may issue decrees enabling the Confederation to fully or partially take over higher education institutionsComment, with the consent of the sponsor, if the said institutions are particularly important for federal activities.Before taking such action, the ConfederationTopic shall hear the Higher Education CouncilComment. 5Under the terms of special legislation, the ConfederationTopic shall make contributions to the Swiss National Science Foundation, to the Swiss Innovation Agency (Innosuisse) as well as to national and international education and research programmesComment.
Moving from the information structure of the utterance and placing the focus upon the interaction between utterances, three main dimensions of textual organisation can be discussed.The thematic-referential level is the most relevant to ensure coherence within a legal text and will therefore be explored in greater detail in this paper.This dimension concerns primarily the thematic progression (Daneš, 1974;Ferrari & De Cesare 2009), i.e., the way in which the topic of a new utterance refers back to previous referents.It also concerns, more broadly, other anaphoric ties within the text.The further two main dimensions discussed in the model, the logical and the polyphonic one, will not be taken into account in this paper and therefore not presented here.

Preliminary findings
In comparing and contrasting the German, French and Italian versions of five Swiss federal acts, a number of articles emerged in which the same provision is formulated using different textual structures.Even if equivalence at the content level is not undermined, these micro-textual shifts from one language version to another contribute to either a more or less comprehensible result, bearing in mind the textbased component of comprehensibility presented in section 1. Due to the qualitative nature of comprehensibility, it is not always straightforward, nor useful, to rank the three language versions on a "clear-unclear" continuum.On the contrary, as the examples discussed below will show, it rarely happens that one of the three versions turns out to be completely unclear.The micro-textual shifts identified in translated legislation mainly contribute to improving comprehensibility, especially in comparison to a hypothetical utterance that more closely follows the German structure.
To better illustrate the main outcomes of this study, a few selected examples taken from the texts under analysis will be reported and commented on.The idea is not to present examples that display extreme and rare phenomena, such as cases of potential divergence or translation errors, because they would provide a distorted image of the legal acts under analysis.Instead, the examples offered in the following pages were carefully chosen in order to capture the most relevant and recurrent phenomena that determine a certain textual autonomy of the three, equally authentic language versions.This section will focus in particular on the topiccomment structure, the use of the passive voice, anaphoric devices, the position of adverbial elements, and the use of punctuation.It will show how all these features are related to textuality and can contribute to fostering comprehensibility.

Topic-comment structure
One of the most frequently observed textual aspects in the sample consists of a different disposition of syntactic constituents in the three language versions.In particular, in each of the five acts analysed, several articles were found that display a divergent thematic structure in at least one of the three versions.Even though they convey the same content, they package information with different topic-comment structures.This is the case in example (2),8 where the three language versions display different syntactic subjects and topics: The choice of different topic-comment structures to express the same content depends on different underlying elements, such as the context, the subject matter of the regulation and the communicative perspective the author wants to adopt (Höfler, 2016, pp. 233-236).In the above example, the French and German versions build the utterance around an inanimate referent, the connecting devices (topic), which are to become property of the infrastructure manager (comment).The Italian version, on the other hand, puts a person at the centre of the utterance, the infrastructure manager (topic), who will become the owner of the connecting devices (comment).Both choices are fully justified and correct as both topical referents were already introduced in the co-text.However, the Italian version seems to be more direct and addressee-centred, as the main stakeholders interested in this legal provision are arguably the infrastructure managers themselves.Responding to readers' expectations and adopting their perspective is a tenet of comprehensibility.Moreover, one should also consider the position of the topic along the animacy scale.Animate subjects, such as the infrastructure manager in the above example, are characterised by a higher cognitive salience compared to inanimate referents, such as the connecting devices.

Passive voice and thematic progression
The use of a different thematic structure in two languages is often connected to the diathesis of the main verb.More specifically, it can depend on the choice between passive and active voice, which is a highly multifaceted aspect, not only in legal texts.First, the contrastive dimension has to be taken into account.In general, it has been noticed that German makes greater use of the passive voice compared to Italian (Bosco Coletsos, 1997, pp. 42-45), while Italian and French prefer the active to the passive voice (Arcaini, 2000, p. 155).Although it is quite difficult to make such generalisations, because even within the same language the preference for active or passive voice is highly dependent on registers, some shifts from the active to the passive voice, and vice versa, could be related to different linguistic traditions.Second, the clarity dimension is to be addressed; avoiding passive sentences whenever possible is often suggested by plain language rules and legal drafting guidelines (FOJ, 2007a, p. 382;Ittig and Accademia della Crusca, 2011, p. 24).Cutts (2013, pp. 63-72) considerations on plain English are applicable to the languages investigated here as well.Indeed, in contrast to passive voice, in active sentences the agent is not only explicit, but also coincides with the "doer", the syntactic subject of the sentence.Third, one should also bear in mind that, depending on text genres, the passive voice is sometimes the best solution from a textual perspective; this holds doubtlessly true for legal genres (Höfler, 2016, pp. 241-243).This section tries to illustrate the interaction between these three dimensions.
Adopting the textual lens, one should consider the importance of the passive voice in shaping thematic progression.A passive sentence, indeed, can be used to refer back to an element that had already been introduced in the co-text, which becomes the topic of the new utterance.This is the case of the Italian and German versions in the following example: Here, the choice of i mezzi necessari per finanziare i provvedimenti/Die Mittel zur Finanzierung der Massnahmen as the topic of the utterance and the necessary legal procedure for allocating them as the comment is fully reasonable from a textual point of view.In fact, the previous articles (7-9) are part of the measures section, in which different cooperation measures are presented in detail.Article 10 is the first article of the financing section, where given information is the existence of such measures, whereas new information is the way in which these measures are financed.A difference between these two versions consists in the explicitation of the agent in the German text (von der Bundesversammlung) and its omission in Italian.This omission was most probably intended and therefore should not be considered a translation error.Indeed, the Federal Assembly is the only authority that passes federal decrees (FOJ, 2007a, p. 241).The Italian version exploits the possibility offered by the passive voice to omit the agent, thus avoiding a constituent that can be considered superfluous from a legal perspective.Doing so results in a shorter, but still precise formulation.The French version, on the other hand, organises the content of the communicative unit in a completely different way.First, it opts for an active sentence.This might be due to an intent to reduce the use of passive voice, determined by the contrastive and/or clarity-related considerations previously introduced.As a consequence, however, the agent not only cannot be omitted, as it is in the Italian version, but it also becomes both the subject of the sentence and the topic of the utterance.Choosing a new, inactive constituent instead of a given, active one (in Chafe's terms, 1987, cf.section 2) as the topic of an utterance might not be the most advisable solution with regard to a clear thematic progression.This example clearly shows to what extent each language version is independent from the others and how different choices might impact comprehensibility.
This example shows that the thematic dimension also has to be analysed in light of the further two dimensions presented above.As previously discussed, plain language rules suggest avoiding the passive voice whenever possible and more verbs in the passive voice are generally found in German.As language specialists, translators are aware of this, so I expected to find a number of examples in which a passive sentence in German was transposed to an active one in the French and Italian versions.Opting for such transpositions in those cases in which a passive sentence is not needed contributes to the overall clarity of the translated text.This is shown in paragraph 1 of the following example: From a textual perspective, the three language versions coincide, the topic being Swiss citizens and the comment being their possibility of submitting an application to any of their cantons of origin.However, these constituents have a different syntactic function: Swiss citizens and the application are, respectively, the subject and the direct object of the active sentence in French and Italian, whereas they are packed as an adverbial prepositional phrase and the subject in the passive German sentence.Here, the use of the passive voice has no textual justification, as the co-text makes it clear enough that the agents are persons with more than one cantonal citizenship.Therefore, it is not necessary to create vagueness, as the (source) German version does, around the person entitled to submit an application.Moreover, German packs two adverbials with different functions in the same utterance, both built around a prepositional phrase introduced by bei.This complication is avoided in the Romance languages, resulting in syntactically less complex, thus more readable sentences.
The use of passive voice in legal language, however, also has other functions that are not directly linked to textuality and that will therefore be discussed elsewhere.One of those is the need for vagueness, which was already mentioned en passant at the beginning of this section.Vagueness represents an inherent feature of legal language (Bhatia et al., 2005).The passive voice makes it possible to omit the agent, and this can prove useful for instance when a provision is laid down but the authority responsible for enacting it will be determined by another legal act.Overall, the analysis shows that, in those cases in which the use of the passive voice is the best choice from a textual or another perspective, translated acts tend to maintain it.In other words, translators seem to deliberately examine each utterance before deciding to make changes at the textual level.

Anaphoric expressions
Coming back to and expanding on the information structure of the utterance at the referential level, another crucial comprehensibility-related aspect is the use of anaphoric devices. 12In the sample that was analysed, the links between anaphoric expressions-not only those in the topical position-and their antecedents are generally clear, and no cases of ambiguity were observed.It was nonetheless interesting to note some articles in which different anaphoric devices were chosen in the three languages.Paragraph 2 of the following example features three different strategies.

Example 5
Continuing Education Act 13 The German text uses a third-person singular personal pronoun as a proforma (sie) to make it clear that the subject of the sentence is the same as the previous paragraph (i.e., the noun phrase die staatliche Durchführung, Förderung oder Unterstützung von Weiterbildung made up by three feminine nouns).This anaphora makes it possible to avoid the repetition of a heavy constituent.The Italian version responds to the same need for compactness by replacing the long subject with a more compact nominal phrase made up of a determiner (tali) and a general noun (attività) referring hyperonimically to the subject.This also implies that the Italian version is more explicit.Also, the German and Italian versions opt for a constant thematic progression, meaning that the first and the second paragraph share the same topic.The French version, on the contrary, chooses a linear progression, where the most salient element within the comment (i.e., the focus) of the first 13 "Art.9 Competition / 1 The organisation, promotion or support of continuing trainings by the State must not hinder competition./ 2 Such activities do not hinder competition if, taking into account the quality, the content of the offer and the specific direction, the continuing training: / a. is offered at prices which at least cover the costs; or / b. is not in competition with private, non-subsidised offers./ 3 Obstacles to competition are permitted provided they are justified by an overriding public interest, are proportionate and have a legal basis."Note that this example contains two further types of annotation in the Italian version, which concern the use of commas and a clumsy formulation in paragraph 3.These aspects will also be discussed in section 3.4.utterance becomes the topic of the second.This has an impact on both the syntactic structure and the device used to create a referential tie to the previous paragraph.Indeed, the French version needs the passive voice to refer back, by means of repetition, to la concurrence, which becomes the topic of the second utterance; also, it creates a link to the content of the first paragraph by a discourse deictic (au sens de l 'al. 1).This use of deixis to introduce intra-and intertextual references, however, is to be avoided whenever possible, as the literature on comprehensible legal language suggests (Höfler, 2015, p. 347).While the semantic equivalence of the three language versions is not undermined, the French formulation arguably tends toward a more bureaucratic style, although the repetition combined with an intra-textual reference might be also seen as a means of explicitness and precision.This example shows once more the difficulty in determining which language version is the most comprehensible due to the highly multifaceted nature of comprehensibility.

Background information and informational hierarchies
The aspects analysed so far concern the informational foreground and, more specifically, the nucleus.Observing the background as well, i.e., adverbial elements contained in frames and appendixes, can provide interesting indications of comprehensibility.Here, I will present some observations on the position of adverbial elements and, closely related to this, the use of commas to create informational hierarchies.

Position of adverbials
As far as the placement of adverbial elements is concerned, the contrastive dimension plays a crucial role.While German tends to pack them, alongside arguments, in the Mittelfeld, Italian and French can place them in a more flexible manner (cf.Blasco Ferrer, 1999, p. 60 andCinato Kather, 2011, pp. 73-84 for a contrastive perspective between German and Italian). 14In the sample under analysis, it emerged that the Romance-language versions of Swiss federal acts take into account these typological differences; for instance, they often do not reproduce the German non-argumental constituent position and move side information into frames and appendixes.This not only contributes to a canonical formulation in each language, but also to comprehensibility, as illustrated, for instance, by example (2) above.In the German version, the adverbial phrase of time ein Jahr nach Inkrafttreten dieses Gesetzes is placed between the main verb and the argument of the sentence.In its Italian and French counterparts, it is moved to the frame and to an appendix in end-position, respectively, so as not to break the continuity of the predicate-arguments structure and to ensure a smoother reading flow.Moreover, one of the functions of the frame is to introduce a condition in light of which the content of the nucleus should be read and understood (cf.section 2).From this perspective, the Italian version seems particularly effective; this article is taken from the transitional provisions section, where the transitional provision should be read in light of its restricted validity period.
The frame is not only used to provide temporal, locative or modal indications, but also to ensure coherence within the text.This is shown by paragraph 2 of example (6).

Example 6
Federal Act on the Electronic Patient Record15  While in German the adverbial phrase of time nach Abschluss der Evaluation appears in its canonical position after the main verb and before the direct object, in French and Italian it is anticipated in the frame.In this way, the frame creates a link to the previous paragraph, which establishes the obligation to carry out an evaluation, thus enhancing the logical construction of the article as a whole.
Sometimes, Italian and French also insert adverbial elements in appendixes in the median position.This is the case for instance in paragraph 2 of example (5).The underlined element modifies the content of the conditioning fact.In German, the noun phrase unter Berücksichtigung der Qualität, Leistung und Spezialität is integrated between the arguments of the conditional subordinate clause, whereas it is expressed through an adverbial clause in French and Italian.This choice of putting some side information in the background helps a reader more quickly identify the main arguments of the sentence and creates a pertinent informational hierarchy.As far as the position of this clause is concerned, one should also note that Italian inserts it between the subject and the verbs, which plain language rules advise against (Cortelazzo & Pellegrino, 2003, pp. 86-88).French anticipates it between the conditional conjunction introducing the subordinate clause and its subject, thus granting the adjacency of the main arguments of the conditional clause.It also introduces a predicate that is absent in German and Italian (remplit les conditions suivantes).This predicate fulfils a cataphoric function and ensures a smoother transition to the list of conditions under which competition is not hindered."Encapsulating" a list through an introductory element turned out to be quite common in the French versions analysed.

Use of commas
As discussed in section 2, the use of punctuation in general, and of commas in particular, is also related to typological considerations.While commas are used in French and Italian to segment the utterance in information units and create informational hierarchies, in German they are almost exclusively governed by syntax.The widespread use of frames and appendixes in Romance-language versions of Swiss legislation, formally introduced with commas, testifies to the fact that each language version employs punctuation that respects its own canonical syntactic and textual structures.
However, some isolated interference phenomena were also spotted.The Italian version of example (5) provides food for thought in this respect.In paragraph 2, letter b, the noun offerte is modified by two attributes, private and non sovvenzionate.At first glance, the comma separating the two adjectives seems to fulfil a textual function.Indeed, it places the second adjective in a new information unit (appendix), thus changing the informational hierarchy of the utterance.An interlinguistic comparison, however, reveals that the presence of this comma is more likely due to a calque from German, where instead the comma between privaten and nicht subventionierten in prenominal position has a mere syntactic function.A similar reflection holds true for the first comma in paragraph 3, which, in this case, does not separate two words, but rather the main and the subordinate clause.Once again, this comma has a syntactic function in German, in which all explicit subordinate clauses must be separated from the main clause with a comma, but is not necessary, and could even be misleading, in Italian.As in the French version, it would be preferable not to put the conditional clause in a separate information unit, hence in the background.In fact, its content belongs informatively to the nucleus of the utterance.The specific cases in which obstacles to competition are permitted is the core element of paragraph 3 and should therefore be integrated in the nucleus, in the comment position.
Another linguistic hurdle that impacts the readability of the same subordinate clause is the use of the verb be (siano) simultaneously in its auxiliary and copulative function.It governs three heads, giustificati, proporzionati and fondati.Cohesion is not at stake here, since the suffix -ati creates a parallelism between the three dependents of the head, thus establishing a morphosyntactic tie.However, while in the first and third cases it is used as an auxiliary verb that forms a compound tense in combination with a past participle, in the second it is used as a copula that links the subject to a deverbal adjective (derived from a past participle).Calques and unusual formulations are one of the main inconveniences of language-contact, translation settings.While lexical and semantic calques and mistakes are relatively easy to spot, it is more difficult to recognise flaws that concern finer-grained aspects such as syntactic inaccuracies and faulty punctuation.This is why they sometimes go unnoticed, even to the trained eye.

Concluding remarks and future research
One might think that the language versions of a multilingual legal act tend toward a word-for-word correspondence.The sample of Swiss federal acts analysed here from a qualitative perspective shows that this is often not the case.In fact, a certain amount of freedom was found in the way in which a norm is expressed in the three official languages, as the different textual aspects highlighted in the previous section show.They point to a common result; in Swiss legislation, legal provisions are often reformulated not only to account for and accommodate divergent grammatical and syntactic structures of each language, but also to build clearer textual patterns, thus ensuring coherence and cohesion.In some cases, for instance, a more literal translation might have produced a grammatically correct and linguistically fully acceptable result, and yet different structures and devices were chosen instead.Some of the aspects discussed might not seem surprising and uncommon in translation.However, one should consider the highly binding context of legal translation and the need for precision and equivalence of the three language versions.Bearing that in mind, even minor changes on the textual level such as changing the topic of an utterance, transposing a sentence from passive to active voice (or vice versa) to improve the thematic progression, opting for a more explicit anaphoric device, anticipating an adverbial element to the frame of an utterance or even just inserting or deleting a comma, gain in significance, especially when analysed from the perspective of comprehensibility.Even if interference phenomena can also sometimes be discerned, the balance seems to tip in favour of translation as an aid rather than a hurdle to clear legislation.
This study provided promising results in terms of extending its scope.16Applying concepts of text linguistics proved to be extremely useful to the analysis of multilingual legislation.The aspects discussed in this paper through a limited number of examples can be used to create a preliminary coding framework to apply to a wider range of legislative acts in order to further consolidate the interpretations presented here, allowing for a higher representativeness and generalisability of the results.In this way, it may be possible to fully understand to what extent translation and, more broadly, the existence of multilingual provisions can be beneficial for comprehensibility and clarity.An important variable that will also need to be considered is diachrony.In Switzerland, the translation sector has undergone major changes over the last few decades and the status of minority languages has improved significantly (Pini, 2017;Egger, 2015).While this study has focused on recent legislation, it will be interesting to take into consideration comparable texts that cover a longer time span in order to account for the evolutions of translation practices throughout the decades.
Ultimately, this study has shown that translation is a prime opportunity to reflect on the textual construction of a legal act.Ideally, such reflections should not be limited to the post hoc evaluation stage, as is the case of this paper.They should, instead, play a central role in the multilingual drafting and translation process.A widespread awareness of the importance of textuality for comprehensibility, but also of the contribution of translation in this respect, is key to drafting clear legislation.More research on the interplay of translation, textuality and comprehensibility would represent a big step forward in the prolific discourse about plain and accessible legal language.