1 Introduction

Previous studies indicate that legal texts belong to the ‘frozen genre’ [1] and that they are textually highly uniformed [2]. In particular, the structure of legal texts is highly conventionalised and predictible and since this is allegedly true of all legal texts it follows that this is a property of legal language as such. Goodrich [3: p. 170], however, warns against such generalisations and points out that “[it] is necessary to recognise the wealth and diversity of the various legal genres, and cautiously admit that it may be problematic to generalise too rapidly from specific features of legal discourse, to a theory of legal language as such” [3: p. 170]. Even if we assume that such a generalisation bears some weight, in reality it does not mean that legal texts are uniformed in the same way. In fact, various studies [4,5,6] demonstrate that legal genres both within the same legal system and language and across different legal systems and languages come up with specific genre structure The objective of the present paper to investigate the textual organisation of the judgments of the Court of Justice of the European Union (CJEU).

CJEU judgments are the official legal decisions issued by the Court of Justice of the European Union. The CJEU is the main institution of the European Union responsible for the uniform application of EU law in EU member states [7]. Its task is to provide an opinion whether EU law has been properly applied and interpreted in individual member states. The Court primarily addresses questions related to the EU treaties, directives, regulations, and other EU legal instruments. CJEU judgments have binding authority on all EU member states, their governments, courts, and administrative authorities. This means that member states are obliged to follow CJEU rulings and adapt their national legal systems accordingly. CJEU judgments are the primary source of EU case law, and they play a crucial role in shaping and establishing legal principles within the EU. The EU case law exists currently in 24 linguistic forms and according to the Article 217 of the Treaty of Rome (the very first regulation issued by the European Council) all official languages of the member states are official languages of the European community [7]. It means that the English version of CJEU judgments is legally equally important as any other language version.

In the following section I will discuss the major principles of genre analysis and in Sect. 3 I will present the data and the methods of analysis used in the present study. Sections 4 and 5 are concerned with the analysis and description of the textual organisation of the English version of CJEU judgments.

2 Genre Analysis and Textual Patterning

Swales famously defined the genre as “a class of communicative events, the members of which share some set of communicative purposes” [8: p. 58]. One of the defining features of a genre is therefore that texts share communicative purposes. The second property of genre is that it is “highly structured and conventionalised with constraints on allowable contributions in terms of their intent, positioning, form and functional value” [9: p.13]. Texts belonging to the same genre are considered to follow consistent textual conventions, and conversely, texts that do not share the same communicative purpose and conventionalised structures belong to different genres. Notice that both properties matter because it is possible that two texts have the same purpose and have a different structural organisation. This is why it is necessary to establish communication purpose at the level of text structure. Swales and Bhatia demonstrate that conventionalised units, which constitute text structure, serve specific purposes not only for the entire text but also within its components. In fact, the general purpose of a text emerges from these specific purposes. The conventionalised units that shape the text structure are referred to as moves or stages within this framework. Occasionally, moves may encompass smaller text chunks known as sub-moves. The approach has been applied to numerous text types, and I will briefly discuss its application to the study of legal cases.

Bhatia [9: p. 136] identifies the following moves and sub-moves in legal cases:

  1. 1.

    “Identifying the case

  2. 2.

    Establishing facts of the case

  3. 3.

    Arguing the case

    1. (a)

      Stating history of the case

    2. (b)

      Presenting arguments

    3. (c)

      Deriving ratio decidendi

  4. 4.

    Pronouncing judgment.”

The description is not based on an empirical and systematic analysis of a specific kind of legal texts and it seems that Bhatia here randomly selected some examples from the UK case law. In the example Bhatia examines a legal case text begins with the name of the case which is given by means of a formula such as “Roles v. Nathan”. Afterwards, a detailed account of facts related to the subject of the case follows. The argumentative part of the judgment contains a section that covers “the history of the case in the courts including the judgment(s) delivered by earlier judges” [9: p. 129]. The central move called “Arguing the case” contains the arguments that serve as a basis for the court’s decision and it is divided into three sub-moves. The final move stating the court's decision is typically very short and highly formulaic.

The genre moves-based approach is advantageous because it allows for the identification of textual organisation in terms of functional units. But, this approach fails to explain the mechanism that governs the relationships between the individual island-like isolated moves. It appears that the only type of relationship that is implied here is linearity. But, according to Bhatia [9], the order of moves in a text is not rigid and strict. It is possible to arrange moves in multiple ways because it's common to deviate from the conventional and expected. Hoey [10, 11] proposes an approach that goes beyond the idea of arranging stages of text development linearly in terms of a flat hierarchy. Following Winter [12, 13], Hoey suggests that texts are organised through discourse patterning. Discourse patterning is based on the reader’s expectations of how a text develops and these expectations operate on both local and global levels [11]. On the local level, the reader's expectations are addressed moment by moment through the immediate relations between clauses or sentences. On the global level, expectations are addressed over longer stretches of text. In the latter case, sentences tend to cluster in groups, forming specific discourse units within a text. The relations between these larger discourse units that do not involve immediately conjoined sentences, are what Hoey refers to as discourse patterns. They constitute patterns since such relations are not isolated occurrences but just as genres they rather tend to be conventionalised in specific cultures, subcultures or linguistic communities. Discourse pattern relations are signalled through lexical devices and “the surface of the discourse… contains sufficient clues for the reader/listener to perceive accurately the discourse’s organisation” [10: p. 33]. Signalling words, therefore, serve as the clues that trigger the reader's expectation of how a text develops. Hoey contents that there a variety of culturally popular patterns of textual organisation and I will provide a summary of the patterns investigated in this paper, which are evident in CJEU judgments. The pattern includes include the Problem–Solution, Question–Answer, Claim-Response, and Gap in Knowledge-Filling patterns.

The Problem–Solution pattern is initiated by a specific aspect of a situation that is problematic in some way. The next step this pattern is Response to Problem and which can be either positive or negative. If Negative Response occurs, this usually means that “the pattern continues to recycle until such as a Positive Result or Evaluation reached” [11: p. 140]. Response which is evaluated in a positive manner is Solution to Problem. Importantly, “[a] Positive Result or Evaluation can always be overridden by an immediately following Negative Result or Evaluation” [11: p. 141].

The Question–Answer pattern begins with a Question and is followed by Response which if negatively evaluated leads to Rejection. Response which is positively evaluated constitutes an Answer to the Question.

Closely related to the Question–Answer pattern is the Claim-Response pattern. This pattern in Hoey's data begins with Claim and is followed by Reason for Claim. If the reason is positively evaluated the pattern is completed and this step constitutes Affirm. Negative evaluation, on the other hand, leads to Denial which is followed by Reason for Denial and then by Correction and eventually Reason for Correction, which eventually completes the pattern.

The Gap in Knowledge-Filling pattern begins with some gaps in knowledge as a part of a situation. Negative evaluation of Responses leaves the gap unfilled, while positive evaluation successfully completes the pattern.

A closer examination of patterning suggests that:

  • Constituting elements form binary relations (e.g. Problem/Solution, Question/Answer);

  • Binary relations are one-directional and asymmetric (Problem precedes Solution, but not the other way around);

  • ConStituting elements form dependency relations (Solution cannot occur without Problem but the opposite is also true);

  • A textual pattern has obligatory and optional elements (an intermediary stage recommendation can occur between the problem and the solution);

  • Patterns can be inserted into other patterns (according to Hoey [11] the Opportunity-Taking pattern may occasionally nest within the Problem–Solution pattern).

Patterning, thus, encompasses a binary and asymmetric interdependence between two indispensable elements, each holding equal significance, and with a predetermined order. The pattern is disrupted if one of the two elements is missing. However, this principle does not extend to optional elements, which are subordinated to the two primary elements. The nesting relations between patterns and optional elements means that the two core elements need not directly succeed each other in a text, resembling the dependency relations in clause structures, where a complement may not always follow the head of a phrase. This flexibility is due to the fact that hierarchical rather than linear relations govern the relationship between elements in a pattern.

One peculiar feature of legal genres, including judgments, is a lower use of anaphoric links than in other kinds of texts, as emphasised as early as in Crystal and Davy [14]. Crystal and Davy also note that legal English typically contains long, complex sentences that are capable of standing alone as complete semantic units. In contrast, casual conversation sentences in English tend to be incomplete, and only in relation to other (incomplete) sentences do they build an information unit [15]. The self-containedness and semantic completeness of sentences in English legal texts are also observable in their layout. In CJEU judgments, this is indicated through paragraphs as the basic organizing units. It is important to note that such paragraphs do not always contain only one sentence; they may consist of more than one long sentence and build complex units. Similar to other types of legal texts, cross-referencing between paragraphs within a judgment is signalled by means of paragraph numbers. All of this suggests that Hoey's discourse patterns approach, which relies on linguistic devices that signal anaphoric relations, may not be suitable for the analysis of the discourse organisation of CJEU judgments. However, patterning in discourse is signalled lexically by means of lexical items that indicate stages of a text development. Unlike anaphoric links which are part of grammatical relations lexical signals are conceptual ad hoc characterisations [16].

What is the connection between discourse patterns and moves? The connection between moves or sub-moves and patterns does not strictly adhere to a one-to-one correspondence. Moves and sub-moves rarely directly indicate patterns; instead, it is more typical to identify one or more patterns within a single sub-move as illustrated in Sect. 5. An integrated analysis of patterns and moves can unveil recurring themes, connections between various segments of the text and the overall information flow. It can also help understand the logical progression of the text and the interconnections between different elements. In this study, both move and discourse patterning analyses will be employed to uncover the underlying structure of CJEU judgments. The move approach is well-suited for the description and analysis of stable features that exhibit less variation in text organisation.

3 Data and Method of Analysis

The analysis combines the following strategies using three sets of data. First, a general distribution of sections which are signalled through headings is investigated using the EUCLCORP [17] which is the largest corpus of EU case law that comprises all the judgments published by the Court from its origin until 2016. This corpus is also used to explore the occurrence of typical fixed multi-word expressions which are subsequently employed in the manual analysis. Second, a sample of 1140 judgments is employed to explore textual interdependence between different sections through a cosine similarity analysis and to investigate the distribution the expressions that signal the textual organisation of individual sections. Finally, a sample of 100 cases is used to explore discourse patterns in the main sections of CJEU judgments. The reason why such an analysis is required is that this kind of patterning cannot be automated at the current stage of research. Instead, a coarse-grained approach is required to explore specific discourse patterns.

I will use the term moves to refer to the sections of CJEU judgments that occur under specific headings. The term sub-moves will be reserved for the stages of text development which will be identified by investigating linguistic signalling devices. The sub-moves are not formally marked in CJEU judgments and they will be identified in the present study by exploring the distribution of recurrent expressions in the EUCLCORP and in the corpus 1140 CJEU judgments in English. CJEU judgments.

The present analysis is concerned only with the English version of judgments. The exploration of flexible and fixed multi-word expressions will play a key role here. Phraseological expressions tend to differ across languages (including legal languages) due to different collocation tendencies [18, 19]. This is why achieving a full correspondence between phraseological expressions across legal languages through translation is exceptionally challenging [20, 21]. In fact, very little is known about the impact of phraseological differences on text organisation in legal texts. What we know is primarily based on a few studies focused only on specific expressions known for their implicit signalling of legal reasoning [22]. The general picture that emerges suggests that despite some lexico-grammatical differences in translation correspondences there is a tendency for such expressions to retain their function in legal reasoning. However, it is unclear whether they also retain their discourse organisation function. While there are certainly divergences in this regard, as indicated in the study of metadiscursive expressions in translation [23], studies concerned with the systemic-functional analysis of thematic development [24, 25] indicate a tendency to retain the same thematic development through translation. And yet one should not too quickly overgeneralise these findings since the studies mentioned explore a well-established category (the Theme-Rheme structure), whereas lexical signalling of specific discourse patterns is typically an idiosyncratic property of individual lexical items. For example, although the English expressions ‘problem’, ‘difficulty’, ‘solution’ or ‘failure’ as well as their German counterparts ‘Problem’, ‘Schwierigkeit’, ‘Lösung’ or ‘Scheitern’ typically signal the Problem–Solution pattern, many other expressions receive a specific textual metafunction only in a context. For instance, the expression ‘deduct [DETERMINER] amount' in the example discussed below in Sect. 5 signals the Problem–Solution pattern, but this does not imply that it and its various translations in other languages (e.g., ‘[DETERMINER] Betrag abziehen’ in German, ‘odbiti iznos’ in Croatian or ‘restar [DETERMINER] importe’ in Spanish) always signal that discourse pattern in CJEU judgments, not to mention other linguistic situations. Considering all of this, it will not be argued that the findings regarding discourse patterning presented here are universally applicable to linguistic versions of CJEU judgments other than English.

4 Moves and Discourse Patterns in CJEU Judgments

The general structure of CJEU judgments has changed over time and the two general periods can be observed. The judgments published between 1955 and 1985 are divided into seven sections with the headings as follows: ‘Keywords’, ‘Summary’, ‘Parties’, ‘Subject of the case’ (referred to below as ‘Subject’), ‘Grounds’, ‘Decision on costs’ (referred to below as ‘Costs’) and ‘Operative part’ (referred to below as ‘Operative’). Some cases published between 1985 and 1999 do not contain ‘Subject of the case’ and ‘Summary’.

The EUCLCORP allows observing specific tendencies in the distribution of these sections. The corpus contains 9434 judgments published in English between 1953 and 2016 and all the sections, except for Keywords, have been annotated with structural XML tags to enable their extraction. Figure 1 below provides an overview of the distribution of sections over six decades by indicating the periods when the sections were absent in CJEU judgments. The data suggests that the structure of the judgments tended to be more consistent in the first three decades than in the most recent period. The section which was most frequently excluded is the ‘Subject of the case.’ In the late 1990s and early 2000s, the frequency of occurrence of the sections ‘Operative Part,’ ‘Parties,’ and ‘Summary’ began to decrease, although there was a period in the mid-2000s when ‘Summary’ made a reappearance in the judgments. The two sections which most consistently occurred throughout the entire period are ‘Grounds’ and ‘Decision on Costs.’

Fig. 1
figure 1

The absence of specific sections in CJEU judgments over time

Full size image

Furthermore, Fig. 2 offers a more detailed representation of how CJEU judgments varied in terms of the contained sections. There is a notable correlation between the occurrence of ‘Parties’ and ‘Operative Part’ as they tended to either occur or to be excluded together. The graph reveals that these sections, along with ‘Summary’ and ‘Subject of the case’ have been especially excluded in recent years, with none of the most recent judgments containing them. Importantly, the decreasing slope on the right-hand side of the plot does not suggest a reintroduction of these sections but the infrequency is rather due to a smaller number of judgments in the corpus. The plot also highlights that the ‘Subject’ section demonstrates the strongest consistency regarding its non-occurrence in the judgments. This overall description provides insight into how the structure of CJEU judgments has evolved over time, demonstrating a systematic variation. While there appears to be a tendency towards a less complex structure, one should be cautious about jumping to conclusions too quickly. The findings imply only a reduced reliance on explicit layout strategies rather than a simplification of the judgments' inherent complexity.

Fig. 2
figure 2

Sections excluded from CJEU judgments over time

Full size image

In terms of their functions, a difference can be made between ‘Keywords’, ‘Summary’, ‘Parties’ and ‘Subject’, on the one hand and ‘Grounds’, ‘Costs’ and ‘Operative part’ on the other. The former serve to familiarise the reader with what a case is about in general terms and building on Labov [26] that macro-move can be regarded as Orientation. The second macro-move is about the Court’s argumentation and decision, and I will refer to it as ‘Judgment’. Let us now consider in some detail each of the moves. It should be also mentioned that most of the sections are divided into enumerated paragraphs, which allows easy reference to specific parts of judgments, but this does not reveal the textual organisation of the judgments. In certain cases published since 2000, ‘Grounds’ is divided into subsections such as ‘Arguments of the parties’ and ‘National legislation’ but these headings are not used systematically.

The label keywords is somehow misleading because the move ‘Keywords’ actually does not contain the most prominent expressions from ‘Judgment’. Instead, the listed expressions are legal concepts which occur in various pieces of EU legislation (e.g. Treaties, Directives, Regulations). The purpose of ‘Keywords’ is to emphasise the legal concepts that are borrowed from other EU legal documents to the present case. The example in (1) shows that the expressions are arranged into lists along with the source information. Occasionally, the same legal concept is repeated more than once in reference to different legal documents.

  • 1. Case: 61968CJ0007.

    1. 1.

      OBLIGATIONS OF MEMBER STATES - FAILURE TO FULFIL - ACTION BY THE COMMISSION BEFORE THE COURT OF JUSTICE - BRINGING THE ACTION - TIMING - DISCRETION OF THE COMMISSION

      (EEC TREATY, ARTICLE 169)

    2. 2.

      FREE MOVEMENT OF GOODS - GOODS - CONCEPT - ARTICLE POSSESSING ARTISTIC OR HISTORIC VALUE

      (EEC TREATY, ARTICLE 9)

    3. 3.

      FREE MOVEMENT OF GOODS - ARTICLES POSSESSING ARTISTIC OR HISTORIC VALUE - CHARGES ON EXPORTS - CHARGE HAVING AN EFFECT EQUIVALENT TO A CUSTOMS DUTY

      (EEC TREATY, ARTICLE 16)

In contrast to what one would expect, the move ‘Summary’ does not summarise the entire case but only reports the Court's decision. There are three strategies that can be observed here. First, an entire sentence is copied from the move ‘Grounds’ which creates specific intratextual relations between the two moves. Occasionally, the original sentence is adjusted such as when lexical items with a cohesive function from a copied sentence are replaced with new items so that the resulting sentence is properly inserted in the running text. This strategy is illustrated in (2) with the copied expressions being highlighted. The second strategy is to write a new sentence by combining various original expressions from sentences that occur in ‘Grounds’, such as (3). The third approach involves recounting the decisions' content by rephrasing the original sentences. The strategies differ in the degree to which the content of ‘Summary’ adheres closely to the literal and exact meaning of the words used in the move ‘Grounds’. The third strategy prioritises the underlying meaning or intent behind the words, rather than being overly fixated on the literal interpretation of the words themselves.

  1. 2.

    Case: 61997TJ0266.

    Move: Grounds

    Move: Grounds

    34 As the Court of Justice has held, Article 90(3) empowers the Commission to determine…

    Move: Summary

    1 Article 90(3) of the Treaty (now Article 86(3) EC) empowers the Commission to determine…

  2. 3.

    Case: 61982CJ0040.

    Move: Grounds

    33. THIS ARGUMENT IS CORRECT, IN AS FAR AS IT IS FOR EACH OF THE MEMBER STATES TO DETERMINE, AND, IF APPROPRIATE, TO ALTER ITS POLICY RELATING TO ANIMAL HEALTH.

    34. HOIVER, THE ARGUMENT OF THE UNITED KINGDOM DISREGARDS THE FACT THAT THE EFFECTS OF HEALTH POLICY ON IMPORTS FROM OTHER MEMBER STATES CANNOT EXCEED THE LIMITS LAID DOWN BY COMMUNITY LAW.

    Move: Summary

    1. ALTHOUGH UNDER ARTICLE 36 OF THE TREATY IT IS FOR EACH OF THE MEMBER STATES TO DETERMINE , AND, IF APPROPRIATE, TO ALTER ITS POLICY RELATING TO ANIMAL HEALTH, THE EFFECTS OF HEALTH POLICY ON IMPORTS FROM OTHER MEMBER STATES CANNOT EXCEED THE LIMITS LAID DOWN BY COMMUNITY LAW.

The similarity overlaps between ‘Grounds’ and ‘Summary’ was compared on a sample of 1140 judgments using a method based on the cosine similarity measure. Cosine similarity is a metric that measures the cosine of the angle between two vectors. In the context of text similarity, each document is represented as a vector in a high-dimensional space, where each dimension corresponds to a unique term. The cosine similarity between two document vectors is calculated, providing a measure of how similar the documents are in terms of their content. The range of n-grams considered during the vectorization process was from 4 to 12. The sections ‘Grounds’ and ‘Summary’ were extracted from the judgments and there were in total 864 files containing ‘Summary’. An overview of the distribution of similarity scores is given below. The data were obtained through the execution of a Python script authored by the author, and the statistical results were generated using the standard statistical packages in R.

figure a

The distribution appears to be right-skewed, with the majority of files having higher similarity scores. The median and mean are relatively close, suggesting a somewhat symmetrical distribution, but the right-skewness is evident from the higher third quartile compared to the first quartile. These results indicate a strong intratextual interdependence between the two sections.

As for the section ‘Parties’, it should be pointed out that the Court typically handles two categories of inquiries received from national courts: (a) a national court seeks the Court's interpretation of EU law; (b) a national court submits a request to the Court for the application of a specific EU law provision. This difference in the types of questions is reflected in the structure of the move ‘Parties’ which may contain as many as seven sub-moves. The following description of the structure of ‘Parties’ is based on a keyword analysis combined with a manual analysis. Keywords which included both individual words and multi-word expressions were extracted from the section ‘Parties’ in the sample of 1140 judgments. After the proper nouns that refer to the parties involved were filtered out a list of key terms was used further for a manual analysis in the sample of 100 files. Some of the most frequent expressions occurring between 700 and more than 2000 times per million words include ruling in the proceedings, for a preliminary ruling in the proceedings application for the annulment, after hearing the oral observations, pending before that court between, failed to fulfil its obligations and gives the following Judgment. The section was extracted using a Python script created by the author and the keyword analysis was carried out using the Sketch Engine tools [27].

The move ‘Parties’ begins with a formula that pinpoints a case using an index (e.g. Case 101/63, Case 307/84, and Case T-88/01). If a case concerns the application of EU law, the move will continue with a list of participants involved in a case. If a case concerns the interpretation of EU law, the second sub-move will provide details about which piece of EU legislation the inquiry directed to the Court is associated with. The subsequent sub-moves are identical for both types of questions. Thus, the subsequent sub-move outlines the questions which is followed by a description of the Court's composition. After that, it is elucidated that the perspectives of all participants and members of the Court are taken into account, followed by the indication of the content of the macro-move ‘Judgment’. The mentioned sub-moves can be referred to as: ‘Identification’, ‘Admissibility of the request’, ‘Participants’, ‘Task for the Court’, ‘the Court’s composition’, ‘Acknowledgment of observations’ and ‘Signalling Judgment’. The following is a schematic representation of the content of ‘Parties’ for the two types of questions (Tables 1, 2).

Table 1 The structure of ‘Parties’ when a case is concerned with interpretation
Full size table
Table 2 The structure of ‘Parties’ when a case is concerned with application
Full size table

The above tables present the original layout of ‘Parties’, reflecting its textual structure in terms of the discussed sub-moves. Standard formulations used in this move are displayed on the right-hand side, with uppercase expressions indicating case-specific information, such as the parties' names. The former are considered constants, while the latter are considered variables in the present study. The variables are populated with specific values; DATE contains specific dates, PARTY contains the names of the parties involved in the case, and so on. The vertical line indicates if more than one formulation is available. Variables contain details that are specific to individual cases. Although there might be differences among individual cases, the above representation reflects the typical textual organisation of ‘Parties’.

According to the data from the corpus of 1140 judgments, the move ‘Grounds; account for 84% of the content of CJEU judgments on average. This calculation is based on a comparison of the number of tokens across different sections. Among all the moves, ‘Grounds’ is the least formulaic, featuring a considerably more complex structure compared to other moves. I will explore its structure in detail in Sect. 5.

The move ‘Costs’, which as we saw above alongside ‘Grounds’ occur in CJEU judgments with the greatest consistency is very short and consists largely of constant elements. The only variables present in this context are those that include the information pertaining to specific EU laws and the names of the participants. Some typical multi-word expressions specific of ‘Costs’ include: matter for that court, the decision on costs, observations to the court, the rules of procedure, step in the action, language of the case and the costs incurred by. The examples in (4) illustrate a typical structure of this move. This representation is based on a combination of keyword analysis and a manual analysis like in the case of ‘Parties’.

  1. 4.

    Version 1: Under A PIECE OF EU LAW the unsuccessful party must be ordered to pay the costs. As the defendant has failed in its submissions it must be ordered to pay the costs.

    Version 2: The costs incurred by PARTY-1 which have submitted observations to the court are not recoverable. As these proceedings are, in so far as the parties to the main action are concerned, in the nature of a step in the action pending before the national court, the decision on costs is a matter for that court.

The move ‘Operative part’ contains the Court's decision, and it is possible here to distinguish between two scenarios: the Court dismisses the application or the Court answers the questions. The first option was observed in only 13 out of 1140 cases in my corpus and (5) illustrates the use of some standard formulations.

  1. 5.

    On those grounds,

    The Court

    hereby:

    1. (1)

      Dismisses the applications;

    2. (2)

      Orders applicant to pay the costs.

More typically, the text of ‘Operative part’ is longer and contains more variation and both constants and variables can be observed. In (6), the variable A NATIONAL COURT is filled with the names of national courts and DATE is the date when the case was received by the Court. The answers given by the Court are contained in DECISION. The number of DECISIONS can vary across cases and in the present dataset it typically falls within the range of one to three.

  1. 6.

    On those grounds.

    The Court

    in reply to the questions submitted to it by A NATIONAL COURT by order of DATE, hereby rules:

    DECISION-1

    DECISION-2…

The content of DECISION has the move ‘Grounds’ as its source. The identical three strategies noted in relation to ‘Summary’ can be observed here as well. i) replication of entire sentences with or without modifications; ii) merging original expressions and producing a new formulation and iii) paraphrasing. (7) exemplifies the use of the first strategy.

  1. 7.

    Case: 61981CJ0104.

    Move: Grounds

    The first paragraph of Article 21 must be interpreted according to its terms and in the light of the objective which it pursues in the system of free trade established by the agreement.

    Move: Summary

    It must be interpreted according to its wording and in the light of the objective which it has in the context of the system of free trade established by the agreement.

A comparison of the cosine similarity indicates a moderate level of similarity between the ‘Operative part’ and ‘Grounds’ as can be seen from the summary of the distribution of scores below. As in the previous case, right-skewness is evident from the higher third quartile compared to the first quartile. However, ‘Summary’ exhibits a higher degree of similarity to ‘Grounds’ than ‘Operative part’, and the nature of the similarity distribution is more balanced there compared to the wider spread and potential influence of lower scores in the dataset comparing the similarity degree between ‘Operative part’ and ‘Grounds’. The difference in the first quartile indicates that even the less similar files in the dataset ‘Summary’/‘Grounds’ exhibit higher similarity than those from the dataset ‘Operative part’/‘Grounds’. The same procedure and tools were used here like above.

figure b

Finally, the move ‘Subject’ like ‘Keywords’ does not contain a coherent text made up of sentences and it has no internal textual structure. However, unlike ‘Keywords’, it does consist of a list of expressions but rather contains a very long noun or prepositional phrase. The purpose of the section is to clarify whether the judgment in question is concerned with the interpretation or with the application of the EU law. The former is signalled by means of the prepositional phrase which comes in a standard format as in ‘On the interpretation of Article, Provisions, or Regulation’. In contrast, the noun phrase indicates that the judgment is concerned with the application of specific EU law provisions, and it occurs in the form of the noun phrase ‘Application for annulment of DECISION’ or ‘Application for a declaration that’. According to the data from the sample of 1140 judgments, ‘Subject of the case’ is more often concerned with the interpretation than with the application of EU law.

From the preceding discussion, the following three conclusions can be drawn. Firstly, CJEU judgments feature a rigid yet intricate text organisation. Secondly, certain moves demonstrate a high degree of formulaicity. Thirdly, the content of certain moves is dependent on the move ‘Grounds’. Based on these results, we can distinguish between two kinds of repetition used in structuring CJEU judgments: structural and lexico-grammatical repetition (see Table 3).

Table 3 Types of repetition used in moves in CJEU judgments
Full size table

The structural repetition is about the genre organisation in CJEU judgments (i.e. the layout and arrangement of moves and sub-moves). Lexico-grammatical repetition involves the use of formulaic expressions both within and across moves and documents, giving rise to intratextual and intertextual connections. Intratextual relations emerge as a result of applying one the three strategies of copying information (discussed above). Such relations are also indicative of dependency relations among moves. As mentioned earlier, the content of the ‘Summary’ and ‘Operative part ‘ is dependent on the content of ‘Grounds’. Furthermore, the content of ‘Keywords’, as well as certain sub-moves in ‘Parties’ and ‘Subject of the case’, is implicitly embedded within ‘Grounds’. This indicates that only ‘Costs’ is a completely independent move in the textual organisation of CJEU judgments. The relations between moves are not of the same kind as those between patterns discussed in Sect. 2. They are not strictly binary and they involve subordination relations in which one element governs the occurrence of subordinate elements. ‘Grounds’ can therefore be regarded as a head that receives other moves as its complements. Besides, ‘Grounds’ unifies the content from different moves. Given the varying extent of the content found in heads and their complements, it is appropriate to differentiate between moves that exhibit a higher or lower degree of parasitism in terms of the information borrowed from external sources. Thus, ‘Summary’, ‘Operative part’ are dependent on ‘Grounds’ whereas ‘Grounds’ exhibits a weaker dependence on ‘Keywords’, ‘Parties’ and ‘Subject’.

Finally, the intertextual relationship is manifested in CJEU judgments through constant formulae which occur in ‘Summar’, ‘Parties’ and ‘Subject of the case’. Intertextual relations are established through lexico-grammatical repetition. As demonstrated in [28] more than 50% of the time the English version of CJEU judgments contain formulaic expressions which were used in previous judgments. Due to the intertextual relationship created through re-use of multi-word expression it can be argued that formulaicity is one of the defining features of CJEU judgments (Table 4).

Table 4 Textual patterns in Background to the dispute
Full size table

5 Textual Development of ‘Grounds’

‘Grounds’ constitute the major part of CJEU judgments both linguistically and legally. This move is made up of two sub-moves I will label as ‘Establishing facts of the case’ and ‘Arguing the case’ due to their similarity with the moves Bhatia identified in his analysis [9]. The former introduces the subject matter of the case and participants and the latter is concerned with questions, observations and the Court’s opinion. This rough outline cannot do justice to the complex text organisation of ‘Grounds’. Unlike in other moves its structure and content cannot be entirely explained in terms of constants and variables and as was mentioned in Sect. 2 headings that signal the containing sections are of more recent origin and they have not been systematically implemented. ‘Grounds’ is linguistically more diverse than any other move and it is also the longest part of a CJEU judgment. This is why in the present paper an approach relying on discourse patterning will be employed. The account provided below combines a corpus investigation of key terms in ‘Grounds’ and lexical diversity with a manual text analysis.

To begin with lexical diversity, the content of ‘Grounds’ has become lexically more uniformed over time. This conclusion is based on the calculation of the type-token ration (TTR) for ‘Grounds’ using the sample of 1140 cases. The values were calculated using WordSmith tools [29]. A lower TTR typically indicates a higher level of repetitiveness, where a smaller set of words is used more frequently. The analysis shows a strong negative correlation (-0.608) and a statistically significant ((Pr( >|t|) < 0.05) inverse relationship between years and TTR. As TTR increases, the predicted years tend to decrease. This pattern suggests that, over the years, the texts have been becoming less lexically diverse or more repetitive. The change in TTR over time and the inclination for 'Grounds' to exhibit less variation is illustrated in Fig. 3 in terms a time series.

Fig. 3
figure 3

Time series showing the change in lexical diversity in 'Grounds'

Full size image

In addition, a correlation analysis based on Spearman correlation coefficient suggests a moderate positive correlation (0.609) between the variable years and the number of words used in ‘Grounds’. This indicates that as the years increase, there has been a tendency for the number of running words in the texts to also increase. In other words, more recent texts tend to have a higher number of running words but this does not increase their diversity but actually leads to their uniformity. While the texts were getting longer over the years, the TTR was decreasing. This suggests that, despite the increase in the number of words (text length), the lexical diversity or richness (measured by TTR) has been diminishing.

5.1 Establishing Facts of the Case

‘Establishing facts of the case’ consists of two parts. The first part summarises what the dispute is about and who is involved. Since the term ‘Summary’ was in use above I will refer to this part as ‘Digest’. The function of the second part is to give a brief history of the case and I will refer to it as ‘Background to the dispute’.

‘Digest’ comes up in two forms depending on whether a case concerns an interpretation or application of EU law. These two forms are represented in (8) and (9). The variables that can be observed here are related to the names of parties, the date when the case was received by the Court and the topic of the dispute described in terms of EU legislation. It should be noted that this is the only part of ‘Grounds’ with structural repetition.

  1. 8.

    By [order|judgment] DATE received at the Court DATE PARTY-1 referred to the Court for a preliminary ruling under A PIECE OF EU LAW questions regarding the interpretation of A PIECE OF EU LAW.

  2. 9.

    By [an application|applications] [lodged|received] at the [Court Registry|Registry of the Court] DATE PARTY-1 brought an action under PIECE OF EU LAW [for the annulment of PARTY-2's decision|for a declaration that by DOING X PARTY-2 failed fulfil its obligations under EU LAW.

According to the data from EUCLCORP, the ‘Background to the dispute’ seems to be of more recent origin. It first appeared in judgments in the 1970s, but it has only been used systematically since 1980. In some judgments it is occasionally included in ‘Arguing the case’.

Despite the change in lexical uniformity the complexity of the textual organisation in ‘Grounds’ has increased. Over time, ‘Background to the dispute’ has grown from just one paragraph to over 50 paragraphs. This is evident in the EUCLCORP, where every paragraph in the corpus has been annotated and enumerated. (10) and (11) are common examples of a one-paragraph long ‘Background to the dispute’ that has the following expressions at the beginning of the first sentence in a paragraph: These questions Ire raised, These questions have arisen, The questions arose, or That order was made. The beginning of ‘Background to the dispute’ has developed since the 1980s into a narrative-like form with the use of the adverbial of time at its beginning as demonstrated in (12). This format has not changed much until today.

  1. 10.

    Case: 61980CJ0279

    2 THE QUESTIONS AROSE IN THE COURSE OF CRIMINAL PROCEEDINGS FOR OFFENCES AGAINST ARTICLE 1 OF THE KONINKLIJK BESLUIT (ROYAL DECREE) OF 10 SEPTEMBER 1970 (STAATSBLAD 410). THAT ARTICLE PROHIBITS THE PROVISION OF MANPOIR WITHOUT AUTHORIZATION FROM THE MINISTER FOR SOCIAL AFFAIRS.

  2. 11.

    Case: 61980CJ0155

    2 THESE QUESTIONS IRE RAISED IN THE COURSE OF A PROSECUTION FOR A CONTRAVENTION OF ARTICLE 5 OF THE GERMAN LAW ON WORKING HOURS IN BAKERIES (GESETZ UBER DIE ARBEITSZEIT IN BACKEREIEN UND KONDITOREIEN), AS AMENDED ON 23 JULY 1969.

  3. 12.

    Case: 61990CJ0196

    4 On 29 January 1980 the vessel was sold to a British company, Minerva Fisheries Ltd...

    5 In the night of 15 to 16 February 1980 the vessel was shipwrecked.

The rest of the 'Establishing facts of the case' section is dedicated to illustrate the circumstances that gave rise to the dispute. Its organisation is indicated lexically rather than through structural repetition or headings, requiring a manual linguistic analysis for examination. I propose here that lexical cues signify relations within discourse patterns, and the structure of this section can be explained in terms of textual discourse patterning. Let us consider one example from case 61989CJ0234. The entire text cannot be reproduced here due to its length and in (13) below I omitted details that provide information about specific aspects of the dispute and kept only the elements necessary to understand discourse patterning. The text is divided into situations which are signalled by means of lexical expressions highlighted below.

  1. 13.

    Case: 61989CJ0234

    Situation 1:...the contract entered into between them [ Mr Stergios Delimitis, formerly the licensee of premises for the sale and consumption of drinks in Frankfurt am Main (hereinafter referred to as "the publican") and the brewery Henninger Braeu AG] on 14 May 1985. … Under Clause 1 of that contract the brewery let to the publican a public house...

    Situation 2: The contract was terminated by the publican on 31 December 1986. … The brewery considered that he still oId it the sum of DM 6 032,15,...

    Situation 3: The brewery deducted that amount from the tenant's deposit which had been paid by the publican.

    Situation 4:The publican challenged the deduction made by the brewery.

    Situation 5: and brought proceedings against it before the Landgericht (Regional Court) Frankfurt am Main in order to recover the sum deducted.

    Situation 6: By a judgment of 10 February 1988, the Landgericht dismissed the action.

    Situation 7: The publican lodged an appeal against the Landgericht's judgment with the Oberlandesgericht Frankfurt am Main...

    Situation 8: [The Oberlandesgericht Frankfurt am Main] considered that it was necessary to ask the Court of Justice for a preliminary ruling on the compatibility of the beer supply agreements with Community competition rules and accordingly referred the following questions to it…

The starting situation is a contract signed between two parties (Situation 1). The second participant (Situation 2) encountered a problem due to the termination of the contract by one of the parties. The latter party resolved the Problem by taking some money from the second participants (Situation 3) but this Solution created a new Problem for the first party. Situation 4 reflects a response by the first party through a Negative Evaluation of the Solution and by creating a Claim (Situation 5) which involved the Landesgerichtshof as a new third participant in the process. The Landesgerichtshof rejected the Claim (Situation 6) but the second participant challenged this Rejection (Situation 7) and at this stage a new participant was introduced. Finally, the fourth participant responded by sending a Question to the fifth participant (Situation 8) which created the Question–Answer pattern and completed the initial Problem–Solution pattern. This example illustrates how nesting of patterns and recycling of the Problem–Solution pattern underlie textual development of ‘Background to the dispute’.

The patterns are lexically signalled in the text. The first Problem is signalled by terminated in Situation 3. The second Problem is implied through the use of the lexical item challenged which indicates that the second participant does not agree with the action taken by the first participant (deduction of money). The lexical signalling of the Claim-Response patternFootnote 1 is slightly more complex. One may oppose labelling Situation 6 as a claim because bringing a proceeding does not seem to be about bringing a claim. However, the utterance (not reproduced here) following the sentence from Situation 6 begins with In support of his claim and the noun claim refers here to bring proceedings. This is illustrative of how signalling function is often contextually dependent as it was mentioned in Sect. 2. The Gap in Knowledge-Filling pattern is signalled by considered that it was necessary to ask the Court of Justice for a preliminary ruling on the compatibility of the beer supply agreements with Community competition rules. The verb consider refers to a mental state and the Gap in Knowledge pattern arises because the Oberlandesgericht Frankfurt am Main does not know how to rule on the compatibility of the beer supply agreements with Community competition rules. Finally, the Question is signalled by the items ask and questions. These textual patterns can be represented schematically by using only the highlighted expressions and by coding participants’ names.

5.2 Arguing the Case

‘Arguing the case’ is made up of two simple sub-moves: ‘Questions’ and ‘Answers’. ‘Questions’ are either individually listed and discussed separately or introduced one after another and immediately addressed. These two sub-moves build the first pattern in ‘Arguing the case’ but there is also the Claim-Response pattern which tends to be embedded with the former pattern. Let us consider in some detail the composition of these sub-moves and patterns.

As mentioned earlier, the reception of questions by the Court is usually indicated in ‘Establishing facts of the case’ while the formulation of the questions occurs in ‘Arguing the case’. Once a question is introduced, it may undergo reformulation, and then the Court deliberates on it alongside the claims presented by the involved parties. Although claims are typically addressed individually, the Court's response is formulated collectively thereby concluding the Claim-Response pattern. The dynamic of the Question–Answer pattern is then resolved in the final part of this move when the Court provides its decision on how the EU law should be interpreted or applied. Let us consider one example here.

Again, only the linguistic expressions that reflect textual development are displayed below in (14). In the original document, the first question is immediately followed by the second one which is then addressed in a paragraph following Situation 5 below. Presently, only the first and the most general claim is represented but in the complete text there are 20 additional claims located between Situation 3 and Situation 4. The expressions that signal patterns are highlighted.

  1. 14.

    Case 62000CJ0208

    • Situation 1: Question: Original questions

      Are Articles 43 EC and 48 EC to be interpreted as meaning that the freedom of establishment of companies precludes the legal capacity, and capacity to be a party to legal proceedings, of a company validly incorporated under the law of one Member State from being determined according to the law of another State to which the company has moved its actual centre of administration, where, under the law of that second State, the company may no longer bring legal proceedings there in respect of claims under a contract. (sic.)

    • Situation 2: Question:Rephrased questions

      By its first question, the national court is, essentially, asking whether, where a company formed in accordance with the legislation of a Member State (`A') in which it has its registered office is deemed, under the law of another Member State (`B'), to have moved its actual centre of administration to Member State B, Articles 43 EC and 48 EC preclude Member State B from denying the company legal capacity, and therefore the capacity to bring legal proceedings before its national courts in order to enforce rights under a contract with a company established in Member State B.

    • Situation 3: Answer:Claim:Defendant

      For NCC and the German, Spanish and Italian Governments, the Treaty provisions on freedom of establishment do not preclude the legal capacity, and the capacity to be a party to legal proceedings, of a company validly incorporated under the law of one Member State from being determined under the rules of law of another Member State, to which that company is found to have moved its centre of administration: nor, depending on the circumstances, do they preclude the company from being prevented from enforcing before the courts of the second Member State rights under a contract entered into with a company established in the second State.

    • Situation 4: Answer: Interpretation of EU Law

      In those circumstances, the refusal by a host Member State (‘B’) to recognise the legal capacity of a company constitutes a restriction on freedom of establishment which is, in principle, incompatible with Articles 43 EC and 48 EC.

    • Situation 5: Answer:Answering questions

      Accordingly, the Answer to the first question must be that, where a company formed in accordance with the law of a Member State (‘A’) in which it has its registered office is deemed, under the law of another Member State (‘B’), to have moved its actual centre of administration to Member State B, Articles 43 EC and 48 EC preclude Member State B from denying the company legal capacity and, consequently, the capacity to bring legal proceedings before its national courts for the purpose of enforcing rights under a contract with a company established in Member State B.

Question is signalled in this text excerpt in Situation 1 by means of a question marker and in Situation 2 lexically by means of the expressions first question and asking. Claim is signalled by the for-construction which ascribes the content of the utterance from Situation 3 to participants involved in the case. The corpus data indicate that this pattern is more often signalled in CJEU judgments by expressions containing the verbs submit, claim or observe such as in The German Government submits; The Italian Government also claims; or The Spanish Government observes. Finally, Answer is signalled by the clause the Answer to the first question must be that. In the above representation, Response from the Claim-Response pattern is part of Answer which means that the same expression can signal the closure of more than one pattern. Responses are lexically signalled through In that regard, it is appropriate to begin by rejecting the arguments; Third, the Court rejects the Spanish Government's argument that. The textual patterns are represented schematically below.

  1. 15.

    Case 62000CJ0208.

    Question: Original question

    1. Are Articles 43 EC and 48 EC to be interpreted as meaning that the freedom of establishment of companies precludes X from doing Y.

    Question:Rephrased question

    22 By its first question, the national court is, essentially, asking whether, Articles 43 EC and 48 EC preclude X from doing Y.

    Answer:Claim:Defendant

    23 For NCC and the German, Spanish and Italian Governments, the Treaty provisions on freedom of establishment do not preclude X from doing Y.

    Answer:Answering question

    94 Accordingly, the Answer to the first question must be that Articles 43 EC and 48 EC preclude X from doing Y.

    The core of the text organisation in this case can be reduced to as much as (16) or an even more abstract fashion as in (17).

  2. 16.

    Case 62000CJ0208.

    Question: Can a Member State deny the company legal capacity and, consequently, the capacity to bring legal proceedings before its national courts?

    Defendant’s claim: They cannot.

    Answer: No, they cannot.

  3. 17.

    Case 62000CJ0208.

    Question: Can A do X?

    Claim: A cannot do X

    Answer: A cannot do X.

‘Arguing the case’ may occasionally contain intratextual repetition as illustrated in (19) which juxtaposes this kind of repetition in Rephrased questions and Answering questions. As this example illustrates, intratextuality establishes strong connections between various elements within the present move. It also highlights how the content of Answer is influenced by the language found in the materials submitted by the parties. In this example, it is evident that the Court's final answer contains expressions originating from both the question submitted by the national court and the claims put forth by one of the participants in the case. There is a lack of comparative research on the linguistic relations between the content of CJEU judgments and the material submitted by the national court and the involved parties and we still do not know to what extent the language of these documents influences the linguistic features of CJEU judgments. Nevertheless, the discussed example indicates that these formulations might play a role in shaping how the Court formulates its answers based on that material. In terms of text development, intratextual relations reveal how the content of one pattern is incorporated into the content of another pattern. In the given example, a portion of the content of Problem migrates from the Problem–Solution pattern into the Question–Answer pattern.

  1. 18.

    ANSWER:Claim:Defendant: For NCC and the German, Spanish and Italian Governments, the Treaty provisions on freedom of establishment do not preclude the legal capacity, and the capacity to be a party to legal proceedings, of a company validly incorporated under the law of one Member State from being determined under the rules of law of another Member State, to which that company is found to have moved its centre of administration: nor, depending on the circumstances, do they preclude the company from being prevented from enforcing before the courts of the second Member State rights under a contract entered into with a company established in the second State.

    QUESTION: Asking questions: Are Articles 43 EC and 48 EC to be interpreted as meaning that the freedom of establishment of companies precludes the legal capacity, and capacity to be a party to legal proceedings, of a company validly incorporated under the law of one Member State from being determined according to the law of another State to which the company has moved its actual centre of administration, where, under the law of that second State, the company may no longer bring legal proceedings there in respect of claims under a contract. (sic.)

  2. 19.

    QUESTION:Rephrased questions: By its first question, the national court is, essentially, asking whether, where a company formed in accordance with the legislation of a Member State (`A') in which it has its registered office is deemed, under the law of another Member State (`B'), to have moved its actual centre of administration to Member State B, Articles 43 EC and 48 EC preclude Member State B from denying the company legal capacity, and therefore the capacity to bring legal proceedings before its national courts in order to enforce rights under a contract with a company established in Member State B.

    ANSWER:Answering questions: Accordingly, the Answer to the first question must be that, where a company formed in accordance with the law of a Member State (`A') in which it has its registered office is deemed, under the law of another Member State (`B'), to have moved its actual centre of administration to Member State B, Articles 43 EC and 48 EC preclude Member State B from denying the company legal capacity and, consequently, the capacity to bring legal proceedings before its national courts for the purpose of enforcing rights under a contract with a company established in Member State B.

As a final remark, I would like to suggest that textual patterns link the communication between the CJEU and national courts. This can be better understood if we explain it in terms of Swale's (2004) notion of genre chains. The concept of a genre chain refers to the sequences of genres typically generated within a specific field of study or professional environment. These genres are interconnected in a chain-like structure, where each genre has a distinct purpose and leads to the creation of the subsequent genre in the chain. For instance, in the realm of academic research, a genre chain might encompass a call for papers, a conference abstract, a conference paper, a revised paper, and a published article. As a general outline, I propose the following nesting of textual patterns that link CJEU judgments with other legal documents.

figure c

The communication between a national court and the CJEU involves an intricate genre chain consisting of various legal documents and correspondence, each with its specific purpose and role in the process. This communication has not been addressed in the previous research in terms of genre chains but we can infer that the following steps are likely to be the key components.

  • Initial Request for Preliminary Ruling (National Court): The genre chain begins when a national court in an EU member state faces a legal question concerning EU law and it initiates the process by sending an initial request for a preliminary ruling to the CJEU. This is when the Problem–Solution pattern begins.

  • Referral to the CJEU (National Court): Following the initial request, the national court may need to provide additional details, context, and legal arguments to support its question. This may involve submitting supplementary documents, legal briefs, and case law references as part of the genre chain. These documents serve to bolster the initial request and provide the CJEU with a comprehensive understanding of the case. The national court sends those details because it is uncertain about the correct interpretation or application of a provision of EU law in a specific case. In this way, uncertainty activates the Gap in the Knowledge-filling pattern. To resolve this pattern the national court formulates a question which is then sent to the CJEU.

  • CJEU Judgment (CJEU): The CJEU, upon receiving the national court's initial request and supporting documentation, evaluates the legal issues presented. The CJEU responds with a judgment that interprets the relevant EU law and provides guidance to the national court. This CJEU judgment is a critical step in the genre chain. This involves creating the Claim-Response other patterns discussed in the previous analysis. At this stage the Question–Answer pattern initiated by a national court is being resolved.

  • Receiving the Decision from the CJEU (National Court): After the CJEU issues its judgment, the national court receives and acknowledges the decision. At this stage both the Knowledge-filling pattern and the Problem–Solution are being closed.

6 Conclusion

The present paper was concerned with the description and analysis of the structure of CJEU judgments. The preceding discussion reveals that CJEU judgments adhere to conventionalised text structures characterised by a high degree of rigidity, which corresponds to the usual expectations about text organisation of a legal text that typically belongs to a frozen genre. The study's contribution lies in providing a comprehensive description of their textual organisation and in explaining what makes the structure of CJEU judgments rigid. The analysis of the occurrence of sections over decades and of textual interdependence between sections revealed dynamic changes in the structure of CJEU judgments. A combination of corpus and manual linguistic analyses showed that the structural rigidity arises from the repetitive occurrence of structural elements, which are realised as moves and moves at the textual level and that serve as constants at the lexical level. Furthermore, recurrent linguistic expressions also enhance textual coherence through intratextuality, a concept that, in contrast to intertextuality has received little attention in legal linguistic studies to date [30]. However, as the present study demonstrates, intratextuality can significantly influence the organisation of texts by facilitating the migration of content across different moves.

Aside from the sections marked by headings, that serve as the basing building blocks of CJEU judgments, the internal structure of CJEU judgments is not transparent. This is particularly the case with ‘Grounds’, which constitute the core of CJEU judgments and that has become very long over time. Paragraph enumeration offers a clear and systematic reference system, facilitating seamless citation and cross-referencing of specific sections and points within the judgments. Moreover, enumerated points serve as clear signposts that direct readers' attention to crucial issues, legal principles, or findings within the judgment. This, in turn, facilitates readers in identifying and concentrating on the case's most significant aspects. But, paragraph enumeration does not reveal the textual development of CJEU judgments and dynamic relations between individual text chunks. Instead, it is the linguistic expressions that function as essential signalling devices which prove instrumental in unveiling the intricate textual development of CJEU judgments. It is therefore evident that there is far more complexity to the textual organisation of CJEU judgments than initially meets the eye and this complexity extends beyond the surface-level information.

The examination of textual development and discourse patterning holds significance in the study of legal reasoning, yet the relationship between these two aspects remains a relatively understudied subject. As demonstrated here, the distinction between the two types of tasks for the court (interpretation or application) is textually reflected in various sections of CJEU judgments, with discourse patterning being lexically signalled. This underscores the need for further exploration into the intricate interplay between textual development and discourse patterning in the context of legal reasoning.