How to Do “Ought” with “Is”? A Cognitive Linguistics Approach to the Normativity of Legal Language

Abstract

The paper addresses the question how descriptive language is used to express legal norms. Sentences we find in legislative acts, i.e. statutes, constitutions and regulations, express legal norms. Linguistically speaking, there are various grammatical and lexical ways of expressing norms, such as imperative mood, modal verbs, deontic verbs, etc. However, norms may also be expressed by descriptive sentences, namely sentences in present or future tense and indicative (declarative) mood (i.e. The minister determines the tax rate). In many civil law countries (including Poland), this is a very common, if not the default, form of expressing norms in legislative texts. Often presented as a legal peculiarity, this phenomenon has yet to draw much academic attention. The normative meaning of descriptive sentences is usually attributed to purely pragmatic factors stemming from our shared assumptions about the legal system. However, a closer look reveals that similar grammatical constructions are ubiquitous in everyday communication and in different languages. We tend to utter various sorts of directives using descriptive sentences (Now we add a spoon of salt to the sauce; credit cards are not accepted). This suggests the possibility for a linguistic (as opposed to exclusively legal) explanation. This paper aims to offer such an explanation. Rather than resorting to formal semantics, so prevalent in legal theory, it borrows from Cognitive Linguistics to reveal the cognitive underpinnings of our surprising tendency to express normativity in descriptive terms. This involves four different, yet complementary, theories. Firstly, the theory of conceptual metaphor by George Lakoff and Mark Johnson which explains the phenomenon in question in terms of the metaphor “ought is is”; with “ought” as the more abstract target domain and “is” as the more concrete, cognitively simpler source domain. Secondly, the theory of speech act metonymy by Panther and Thornburg which presents descriptive legal sentences as referring to various components of the underlying cognitive scenario of obligation. Thirdly, Ronald Langacker’s notion of the virtuality of language as the explanation for non-present, including future, perpetual and directive, uses of the present tense. Fourtly, the notion of normative generics which points to the nominal, as opposed to verbal, structure of descriptive legal sentences as the source of their normativity.

1 Introduction

In the classic collection of lectures, John L. Austin explained how to do thing with words, that is, how words are used to perform various social acts. In another classic paper, John Searle suggested how to derive “ought” from “is”, that is how to overcome the naturalistic fallacy and draw normative conclusions from descriptive statements. This paper may be seen as a spiritual beneficiary of those two famous works. It addresses a much more specific question; namely how descriptive sentences are used to express legal norms through statutory language. To answer this question, it applies theories and concepts from Cognitive Linguistics, a highly popular and influential approach to linguistics that studies language against the background of other human cognitive capacities, such as categorisation, perception, and memory. The choice of Cognitive Linguistics as the “lenses” through which to investigate the topic is justified by two observations. Firstly, this linguistic paradigm remains relatively unknown in legal theory, unlike various competing approaches from traditional linguistics, formal semantics and philosophy of language. Secondly, Cognitive Linguistics seeks to explain various linguistic phenomena in terms of their semantic motivation. This means acknowledging the underlying semantic factors responsible for conventionalised linguistic forms (i.e. the grammar of legal provisions). Such an approach is particularly promising in the context of legal language which has traditionally been studied mostly from formal (i.e. logical) perspectives.

Law is a normative phenomenon. The normativity of law is, without doubt, one of the most exploited topics in legal theory and philosophy of law. There are numerous problems that are discussed under this label, such as: what does it mean that law is normative? Is the normativity of law different from the normativity of something like ethics? Where does the normativity of law come from? Can the normativity of law be derived from social facts? Can the logic of legal norms be constructed? And so on. However, the subject of this paper is not the normativity of law, but rather the normativity of legal language, which is understood as the language of legislative texts, such as statutes, regulations, constitutions, etc. The sentences we find in legislative texts express legal norms. Linguistically speaking, there are various grammatical and lexical ways of expressing norms and legislators do not always opt for the most obvious ones. This topic has been discussed mostly by experts in legislative drafting and translation scholars. Legal theorists and philosophers, on the other hand, do not seem to be very interested in this topic.

Legal philosophy knows countless typologies of legal norms. For the purposes of this paper, it will suffice to intuitively distinguish between obligations (duties), prohibitions and powers (rights). To make the argument clearer and more concise, the article will be generally restricted to obligations (duties), namely norms that make certain behaviour mandatory for their addressees. There are obviously numerous linguistic ways of expressing duty [1]: 424, [2]: 173–181, [3]. Every language provides its own repertoire of grammatical and lexical means. Still, some of them seem to be quite universal. These include:

• a) The imperative mood, i.e. Take an oath, judge!

• b) Performative sentences, i.e. I order/command every judge to take an oath!

• c) Modal verbs, i.e. A judge must/should/shall take an oath.

• d) Deontic nouns, i.e. A judge has an obligation/duty to take an oath.

• e) Deontic adjectives, i.e. Taking an oath is mandatory/obligatory for a judge.

• f) Passive voice constructions, i.e. A judge is required/obliged to take an oath.

Not all of these constructions are used in legislative texts with equal frequency. Quite paradoxically, the only construction that serves almost exclusively to express directives, namely the imperative mood (a), is not used at all by contemporary legislators. Similarly, explicitly performative sentences (b) are rarely if ever used in legislative writing at statutory level. This is not the place to discuss the reasons behind this. Still, it interesting to note that the only two unambiguously normative forms [1]: 474] are effectively excluded from the legislative drafter’s toolbox.

A lot has been said already about other forms of expressing duty, especially the use of modal verbs in various legal and linguistic environments [4,5,6,7]. Still, despite the variety of grammatical and lexical means of expressing duty at the legislator’s disposal, many legal provisions do not feature any of them. Instead, they only feature a verb in the indicative mood in the present or future tense. Consider several examples from Polish statutes:

• 1) When appointed, a judge takes an oath before the President of the Republic of Poland…¹

• 2) A social enterprise employs at least three people…²

• 3) The court orders the forfeiture of items coming directly from criminal conduct.³

• 4) The minister will issue a regulation…

In Polish legal culture, this phenomenon is known as a “seemingly descriptive” or simply “descriptive” form of legal provisions [8,9,10]: 100–107]. In Polish law, it is extremely pervasive. A small empirical study was conducted to determine the actual frequency of this form of expressing obligation. Four of the most recent⁴ Polish statutes were read, marking all sentences expressing obligation. Next, all the sentences using the descriptive form, as opposed to other forms of expressing obligation, were counted and the percentage of uses of the descriptive form was calculated. The results are presented in the table below (Table 1):

Table 1 The descriptive form in Polish legislation – a quantitative analysis

As we see, the descriptive form is by far the most frequently used form of expressing obligation in recent Polish legislation. Even more significantly, it is not some Polish spécialité de la maison. Analogous constructions can be found in statutory texts in other European languages, including Italian, German, French, and Czech. Consider the following examples:

• 5) Italian: Salvo patto contrario, le spese della permuta e le altre accessorie sono a carico di entrambi i contraenti in parti uguali.⁵

• 6) French: Si, dans le cas où l'ouvrier fournit la matière, la chose vient à périr, de quelque manière que ce soit, avant d'être livrée, la perte en est pour l'ouvrier, à moins que le maître ne fût en demeure de recevoir la chose.⁶

• 7) German: Verspricht der Schuldner dem Gläubiger für den Fall, dass er seine Verbindlichkeit nicht oder nicht in gehöriger Weise erfüllt, die Zahlung einer Geldsumme als Strafe, so ist die Strafe verwirkt, wenn er in Verzug kommt. Besteht die geschuldete Leistung in einem Unterlassen, so tritt die Verwirkung mit der Zuwiderhandlung ein.⁷

• 8) Czech: Půjčitel přenechá vypůjčiteli věc ve stavu způsobilém k užívání. Způsobí-li škodu vada věci, kterou půjčitel zatajil, nahradí půjčitel škodu vypůjčiteli z toho vzniklou.⁸

There is abundant literature on the topic of modal verbs, normative language, deontic logic, etc. However, despite its arguably universal character, the phenomenon of the descriptive form of legislative provisions has not drawn much attention from legal theorists and philosophers. There may be several reasons for this. Firstly, it is so ubiquitous that it may simply go under the radar. Secondly, legal scholars tend to focus on the “deep structure” of legal rules, i.e. abstract constructs postulated in their theories, rather than the actual language of statutory provisions, which is often left for linguists to discuss. Thirdly, perhaps the most important factor is that the descriptive form of expressing obligation is not very pervasive in English language and English is the mother tongue of some of the most influential contemporary legal thinkers and the lingua franca of legal theory and philosophy of law. Admittedly, the descriptive form can be found in English legislative texts, but most often they feature the modal verb shall. This creates its own problems, as shall is notoriously ambiguous [11,12,13,14]. Consider the following comparison of different language versions of a single provision from the European Union regulation⁹ presented in Table 2. Although EU legal language has its own peculiarities and may diverge from national legal languages, the comparison is instructive and proves the point. Where the English version uses shall, other language versions use the indicative mood in present tense, without any modal verbs.

Table 2 Language versions of the Article 27.1 of EU’s GDPR – a comparison

The similarities and dissimilarities between different languages in the ways of expressing legal norms provide an interesting subject for a comparative linguistics research. This paper, however, is concerned with a different topic, namely the search for a theoretical explanation for the phenomenon of expressing obligation in the descriptive form.

Clearly, the descriptive form has several advantages. It is generally shorter and it avoids the ambiguity typically associated with modal verbs, not only English shall, but also other modal verbs in different languages. But how can it be explained at a theoretical level? Legal scholars and most linguists point to one obvious explanation – the legal context. Aleksandra Matulewska’s position is symptomatic in this respect: “The normative character of such utterances results from the pragmatic situation. In other words, the statutory instrument is obligatory in its nature.” [4]: 138]. In Polish legal theory, this idea is known as the assumption of the normativity of legal language [10]: 105]. We know that sentences in legislative texts express legal norms, so we read them as such, regardless of their grammatical form. The explanation is thus dumped onto pragmatics, “the wastebasket of the study of meaning,” as George Lakoff once put it [15]: 474].¹⁰

The purely pragmatic explanation, however, overlooks a crucial fact. As it happens, the descriptive form of expressing obligations is not restricted to legal language. In fact, in everyday communication we use a variety of constructions to formulate obligations, requests, instructions, suggestions and other types of directives. In philosophy of language, these are known as “indirect speech acts”. Consider the following examples:

• 9) Could you pass me the salt?

• 10) Won’t you help me with that?

• 11) No more discussion! Tomorrow, you go to work!

• 12) Now, we add a tablespoon of vinegar to the boiling sauce.

• 13) You go through the lights, then turn right.

These sentences (9–13) can be given a normative reading without much cognitive effort, despite the fact that they do not feature any of the grammatical or lexical means listed previously (a-f). These examples are formulated in English, but they can easily be translated into other languages. Many cross-linguistic studies confirm that similar constructions appear in various European languages [2,3,4, 17,18,19], including: Polish, Czech, German, Spanish, Hungarian, Norwegian, Russian, Bulgarian, Macedonian, and Slovakian. In fact, most languages seem to have not one, but several ways of imposing duty in the indicative mood without modal verbs: present tense, future tense, first person, second person, sometimes even third person or impersonal forms, singular or plural, etc. [17].

Two conclusions follow. Firstly, the descriptive form of expressing duty is not a legal-specific phenomenon. It is commonly used in everyday communication and, so cannot be explained by simple reference to the pragmatic context of law. Secondly, the descriptive form of expressing duty is universal, meaning that it can be found in many languages. It is not a local phenomenon characteristic only for Polish, English, or any other European language. These conclusions should encourage us to search the general linguistic explanation of the phenomenon in question. In this paper, it will be sought within Cognitive Linguistics which, as will be argued, offers several plausible explanations for the descriptive form of expressing duty.

2 Conceptual metaphor

One of the flagship theories of Cognitive Linguistics is the conceptual metaphor theory created by George Lakoff and Mark Johnson. According to this theory, a metaphor is not just a rhetorical device, but a fundamental cognitive tool that enables us to think and talk about one thing in terms of another: “The essence of metaphor is understanding and experiencing one kind of thing in terms of another” [20]: 5]. For instance, we are used to talking (and thinking) about scientific theories as buildings:

His theory lacks solid foundations. It is not supported by empirical findings. If confronted with new facts, it may collapse. So far, she has only put together the framework of the theory. She needs to fill some holes in it, or it will fall apart.

Similarly, we talk (and think) about time as money:

I have no more time for you. How did you spend your free time? That traffic jam cost us an hour. I’ve invested a year of my life into this relationship. We have lost so much time. What that worth your time? How much time do we have left? Would you spend some time with me?

Another very prevalent series of conceptual metaphors concerns physical orientation. For example, good is identified with up (high position), whereas bad is identified with down (low position):

The value of the company is increasing. She enjoys a high position in the company. He is at the top level. Let’s discuss how she rose to power. This is the peak of his career. She feels low. She is clearly under his bad influence. He is climbing up in his career. He sunk all the way to the bottom.

These expressions can hardly be considered as poetic additions to our everyday way of speaking. They seem perfectly neutral. We typically talk about increasing values, laying foundations for scientific theories, or losing time. But why do we need metaphors in the first place? The pattern discovered by Lakoff and Johnson is this: scientific theories, time, and goodness are fairly abstract ideas (or domains) that cannot be touched or seen. Generally, we cannot experience them directly with our senses. We may say that they are cognitively distant. On the other hand, buildings, money and spatial relations are cognitively close. We see and use buildings, we touch, count, and spend money, and we spatially orientate our bodies all the time. According to conceptual metaphor theory, abstract domains are conceptualised in terms of specific domains. This is why metaphor is described as a conceptual, rather than merely a linguistic, phenomenon and an important tool of human cognition. In fact, Lakoff and Johnson argue that the human conceptual system is largely metaphorical [17, 20]. This, in turn, stems from the idea of embodied (and embedded) cognition: “the structures used to put together our conceptual systems grow out of bodily experience and make sense in terms of it; moreover, the core of our conceptual systems is directly grounded in perception, body movement, and experience of a physical and social character” [21]: xiv]. This bold claim has attracted severe criticism, especially from more traditionally oriented cognitive scientists [see: 17. Here is not the place to address this heated debate, though it should be noted that the general claim of Lakoff and Johnson, namely that our conceptual system is largely built on unconscious processes and imaginative abilities, has gathered large empirical evidence and forms one of the foundations of the Cognitive Linguistics movement.

Metaphors, understood as conceptual phenomena expressed in language, are ubiquitous in the legal domain. While legal culture obviously makes use of common metaphors, such as the metaphor more is up which was already mentioned. For instance, taxes and penalties may be higher (increased) or lower (decreased). More importantly, every legal system comes with a variety of unique legal metaphors, some of which seem to be quite universal. For instance, obligation is often conceptualised as a heavy object. An obligation may be imposed on a person, subsequently a person may be released or relieved of the obligation. In judicial procedures, one party bears the burden of proof, etc. obligation is also alternatively conceptualised as a bond. Parties are bound and restricted by contracts, contracts may be binding, breached or broken, etc. Metaphors in law exist at various levels, i.e. at the level of official legislative texts, at the level of legal doctrine, at the level of laypeople’s understanding of law, etc. [22]: 193]. They serve various functions, such as: cognitive (they help convey abstract legal concepts), argumentative (they impose structure and coherence on a given discourse), and persuasive (they may be used instrumentally to highlight certain aspects of a given phenomenon and hide others) [22,23,24]. What is more, metaphors can influence how people, including legal decision-makers, attempt to solve complex socio-legal problems and how they gather information to ensure that their decisions are well-informed [25].

As Sylwia Wojtczak argues [22], the conceptual metaphor theory offers a cognitive explanation of the normativity of legal language. The language of legal provisions may be seen as a linguistic realisation of the conceptual metaphor ought is is: “in legal culture <  < what ought to be >  > is expressed and cognized by the agency of <  < what is >  > or <  < what will be >  > ” [22]: 175]. The source domain of this metaphor is the plane of reality (is, Ger. Sein). This is the plane of existence that we live in and which, therefore, is cognitively closer. The target domain is the plane of normativity (ought, Ger. Sollen). We cannot directly experience normativity with our senses. Instead, “normative categories (from the Ought world), abstract in their principle, need to be explained by categories derived from the empirical (Is) world” [26]: 194]. This explains why a sentence: When appointed, a judge takes an oath before the President of the Republic of Poland… is understood as imposing the obligation to take an oath by a judge, not merely reporting the common practice. The activity of doing something (here: taking an oath) is metaphorically mapped onto the target domain of normativity and read as the obligation to do something (here: to take an oath). By the same token, not doing something is read as a prohibition on doing that thing, as in the sentence: Provision X does not apply.”

The ought is is metaphor has further ramifications for legal theory and philosophy of law [22]: 177–191]. For instance, it can be seen as the implicit foundation of Hans Kelsen’s pure theory of law. Many of the crucial aspects of his theory can be explained as manifestations of this conceptual metaphor. This includes, among other things, the ontological duality of Sein and Sollen, as well as the postulated analogy between the empirical relation of causation and the normative relation of imputation (Ger. Zurechnung) [22]: 186–188, [26]. The metaphor in question also offers a plausible explanation of the notorious ambiguity of modal verbs, such as the English must, may, shall, etc. These expressions, which are frequently used to express normativity in legal language, have (at least) two meanings: deontic and epistemic: must means both (deontic) obligation and (epistemic) necessity, may means both (deontic) permission and (epistemic) possibility, and so on [22]: 176–179]. According to the ought is is metaphor, the deontic meanings of modal verbs may be seen as metaphorically derived from epistemic ones.¹¹

3 Speech act metonymy

Another imaginative ability, closely related to conceptual metaphor and extensively researched within Cognitive Linguistics, is conceptual metonymy [27]. Metonymy constitutes a stand-for (indexical) relation within a single domain: “one conceptual entity, the vehicle, provides mental access to another conceptual entity, the target, within the same cognitive model” [28, 29]. Consider the following examples:

• 14) I see some new faces here.

• 15) I was just reading Dworkin.

• 16) Do you know why Putin invaded Ukraine?

• 17) We will not forget Bucha.

• 18) Warsaw strongly supports Ukraine.

Sentence (14) uses the part for the whole kind of metonymy. People’s faces are used to stand for people. The metonymic relation between the source (faces) and the target (people) is based on physical association. Sentence (15) uses a producer for product metonymy. The name of Ronald Dworkin stands for one of his books (or papers) by virtue of his authorship. Sentence (16) uses a controler for controlled metonymy that is very prevalent in public discourse. The name of Vladimir Putin stands for Russia’s army, by virtue of the fact that he is the country’s leader. Sentence (17) uses a place for the event metonymy. The name of the town of Bucha stands for the massacre of Ukrainian civilians and prisoners of war committed by Russian forces in March 2022. Sentence (18) uses another common metonymy of place for the institution. The name of the city of Warsaw stands for the Polish government by virtue of the fact that Warsaw is the capital of Poland and the seat of its government. Note that the metonymic relation between the vehicle (i.e. faces, Dworkin, Putin, etc.) and the target (i.e. people, books, Russian army, etc.) may be based on various associative links (physical, spatial, causal, historical, etc.), but it is not conceptually necessary. We may say that the metonymic relation is contingent and, therefore, defeasible. In addition, the links between source and target may vary in strength based on the conceptual distance between them and contextual factors involved [27]: 240].

Metonymy in sentences (14–18) serves mainly a referential function. However, metonymies may have not only referential, but also predicational function, in example:

• 19) They had to abandon their positions.

• 20) Ukraine was able to resist Russia’s aggression.

In sentence (19), the past necessity to abandon positions (“had to abandon”) may be read as an actually occurring past action. Similarly, in sentence (20) the ability to resist (“was able to resist”) stands for actual successful resistance. Additionally, the words Ukraine and Russia stand for the respective armies, in which cases metonymy serves a purely referential function.

Klaus-Uwe Panther and Linda Thornburg have distinguished yet another type of metonymy operating on the speech act level, namely illocutionary metonymy [27, 30, 31]. It addresses the phenomenon of indirect speech acts, traditionally discussed by philosophers of language within the Gricean framework. According to this traditional approach, the illocutionary force of an indirect speech acts has to be inferred on the basis of the literal meaning of an utterance, the linguistic and extralinguistic context, and Gricean principles [32]. Such an approach has several drawbacks [17]: 112]. For instance, it does not account for the psycholinguistic evidence that participants of conversations are often able to draw from the required inferences without any noticeable cognitive effort [30]: 756]. In order to provide a cognitive linguistic perspective on the issue, some authors have proposed to describe speech acts in terms of cognitive models, semantic frames or scenarios consisting of several components. According to Panther and Thornburg, a simplified scenario of request would look like the following [30]: 759]:

• a) the before: A can do X, B wants A to do X.

• b) the core: B puts A under an obligation to do X

• c) the result: A is under an obligation to do X (A must/should/ought to do X)

• d) the after: A will do X.

When an utterance pertains directly to the core component it is a direct speech act, such as in sentence (21):

• 21) I command you to open the door!

But an utterance may also concern other components thus forming indirect speech acts. For example:

• 22) I want you to open the door

• 23) Can you open the door?

• 24) You will open the door

• 25) Will you open the door?

Sentences (22) and (23) directly pertain to the BEFORE component, whereas sentences (24) and (25) directly pertain to the AFTER component. Nevertheless, these components are still parts of the request scenario and hence sentences (22–25) may stand for the speech act as a whole by virtue of metonymic relations between various components. The scenario may be extended to include, among other things, existential presuppositions, preparatory conditions, sincerity condition and satisfaction condition [30]: 760–761]. However, the metonymical links between various components grow weaker accordingly to the conceptual distance from the core of the speech act: “The more distant a speech act scenario component is from the core, the weaker is its ability to evoke the scenario metonymically” [30]: 761]. For instance, a sentence: There is the door is not likely to be interpreted as a request to open the door, unless some special contextual factor is in play.

Obviously, speech act metonymy occurs in legislative texts. I assume that Panther and Thornburg’s model of request speech act generally applies to legal obligation speech act. Minor modifications are made to better fit the legal context:

• a) (the before: legal subjects can do X, the legislator wants legal subjects to do X.)

• b) the core: the legislator puts legal subjects under an obligation to do X

• c) the result: legal subjects are under an obligation to do X (legal subjects must/should/ought to do X)

• d) the after: legal subjects will do X.

Compared with Panther and Thornburg’s model scenario of request, the legal obligation model does not include a specified singular addressee (A), but plural indefinite addresses: “legal subjects”. This is because legislative acts are typically addressed to indefinite audiences. However, as we will see, this may be linguistically achieved by using singular definite nouns. This can be considered as a sort of part for the whole metonymy in which a member of the category stands for the category as a whole. Additionally, the before component in the legal context is arguably the least significant (i.e. conceptually more distant from the core, to the point where one may actually be tempted to remove it altogether from the model. We will address this issue below.

With that in mind, consider the following examples of legal provisions translated from Polish:

• 26) Until 30 June 2023, it is prescribed [literally: to prescribe + reflexive pronoun się] to cover the mouth and nose with a mask in buildings where medical activities are carried out.¹²

• 27) The employer is obliged to protect the health and life of employees.¹³

• 28) The minister will issue a regulation.

• 29) The minister shall issue a regulation.

Sentence (26) may be considered a direct directive speech act. It concerns the core component and expresses the very act of imposing an obligation (“is prescribed”). This is particularly clear in the original Polish version, which uses the impersonal reflexive construction (to prescribe + reflexive pronoun się). Such constructions, which do not exist in English, are characterised by a very low degree of presupposed level of cooperation and solidarity [17]: 117]. Still, it should be noted that sentences such as (26) are an exception rather than a rule in Polish legal language. Sentence (27), which uses the much more common construction “is obliged to”, concerns the result component of the speech act scenario. Conceptually, the result and core components are closely related. This means that the sentence (27) is very likely to successfully activate the directive speech act scenario and be interpreted as imposing an obligation. Sentence (28) concerns the after component, namely to the expected effect of issuing a regulation. The conceptual link between core and result is still strong, but arguably weaker than in the previous example. Thus, the normative interpretation of sentences such as (28) is not as natural and inevitable as in case of (27). Finally, because of the ambiguity of shall, sentence (29) may be considered as pertaining either to the result or the after component. This applies also to other modal verbs, such as must or should, which tend to share the dual deontic-epistemic meaning of shall. Arguably, the illocutionary metonymy theory provides a plausible framework for describing and explaining this duality [27]: 245].

Note that none of the examples above includes an explicitly indicated imposer of the obligation (the legislator). The sentences are formed using verbs in the third person or impersonal constructions. This is perhaps because of the depersonalised nature of legal communication. The legislator is considered as an institutional entity, not an actual person (or even group of people) and legal obligations are thought to be imposed by the law itself, not by any particular person. An explicit mention of the legislator would introduce subjectivity and, potentially, undermine the authority of law [33]: 63]. It also seems that legal obligations are never expressed by pertaining directly to the before component (i.e. I want/the legislator wants the minister to issue a regulation). This may be seen as a manifestation of the antipersuasive restriction present in the tradition of Western legislation: lex iubeat, non suadeat, non doceat, non laudat [34]: 61, [33]: 92]. Legal provisions are not places to express wishes, desires, or justifications. These may be relevant only in case of some extraordinary interpretive doubts. Typically, when dealing with statutes: “We do not inquire what the legislature meant; we ask only what the statute means” [35]: 419]. An explicit mention of the legislator’s intention behind a given rule could, arguably, encourage a critical dispute and, again, undermine the authority of law.

4 Non-present uses of the present tense

Metaphor and metonymy are just two examples of the imaginative abilities that manifest at the language level. Ronald Langacker, another prominent figure of Cognitive Linguistics, places them under a broader label of “virtuality” or “fictivity” of language. Virtuality, in Langacker’s theory, allows humans to transcend bodily experience that is directly available to human senses and form abstract thought [36]: 524–525]. He claims that “surprisingly much of our linguistic effort goes into the description of virtual entities, even when our main concern is with actual ones” [37]: 78]. The distinction between the actual and the virtual planes should not be confused with similar notions, i.e. the distinction between the real and the fictional planes or between true/false values of a sentence. Consider the following sentences:

• 30) My cousin’s dog is called Lora.

• 31) Jon Snow’s direwolf was called Summer.

• 32) I don’t have a dog.

• 33) Dog is a carnivorous mammal.

Sentence (30) refers to the actual plane, which also happens to be the plane of reality. Sentence (31) is set in the context of a fictional world, namely George R.R. Martin’s fantasy series. Moreover, it is false – in Martin’s novels the name of Jon Snow’s direwolf was Ghost, not Summer. Still, it refers to an actual (although erroneously described) situation within that fictional world. According to Langacker’s theory it is actual, not virtual. Sentence (32) refers to the plane of reality, but the dog that is referred to is virtual. There is no actual dog (real, fictional, or otherwise) that I do not have. Finally, sentence (33) is entirely virtual. Both the dog and mammal refer to virtual, abstract entities, rather than any actual individuals.

The list of the virtual phenomena in language is long and open ended. Apart from conceptual metaphor and metonymy, it includes implicatures, mental spaces, conceptual blending, generalisation, abstraction, quantifiers, imagined vantage point, virtual movement, virtual bounding, representative instance, the conceptualiser invoked by a grounding element and fictive invocation of a speech-act scenario, i.e. rhetorical questions [36, 37]: 524]. It is also crucial to understanding the grammatical category of the present tense, which is our main focus here [37]: 91]. Although Langacker explicitly limits the scope of his findings to the English present tense, the examples used below suggest that they have a much more universal appeal, i.e. they apply perfectly to the present tense in Polish.

The paradox about the English present tense is that it can apparently be used for anything but the description of the present time, i.e. events occurring at the time of speaking. Consider the following sentences:

• 34) *I read a book now.

• 35) I get back from work yesterday and I see you crying.

• 36) Our plane leaves at 12:00.

• 37) Water boils at 100 Celsius degrees.

• 38) Dogs fight with cats.

On the one hand, sentence (34) which refers to the actual present time, cannot be correctly expressed in the English simple present. Instead, a progressive construction has to be used (I am reading a book right now). Sentences (35–38), on the other hand, use the simple present correctly, but refer to events clearly not occurring at the time of speaking. Sentence (35) refers to a past event, (36) refers to a future event, 37) is a timeless statement of general validity, and (38) is a generalisation. Still, Langacker claims that the present tense should indeed be characterised as indicating coincidence with the time of speaking, provided that we understand the role of virtuality in its use [38]: 251].

If taken seriously, the characterisation that the simple present tense indicates coincidence with the time of speaking requires that the speech act and the event described in the speech act co-occur. This is hard to achieve, and for two reasons: (1) the duration of an event is generally not equal to the duration of the speech act (the durational problem) and (2) by the time the event is observed by the speaker, it is already too late to initiate its verbal description (the epistemic problem) [38]: 263]. These problems do not arise with the use of imperfective verbs (I understand this book now) and imperfective constructions, such as progressives (I am reading the book now), because such expressions profile (denote) not events construed as temporary bounded entities, but rather representative portions of events [38]: 259–260]. In addition, these problems do not arise with performatives, instructions and sportscast commentaries, which are typically expressed in simple present:

• 39) *I’m ordering you to open the door.

• 40) I order you to open the door.

• 41) I pronounce you man and wife.

• 42) Now we slice the onion into thin slices.

• 43) Gavi passes to Lewandowski…. and… Lewandowski scores! What a beautiful header!

In case of performative and instruction sentences, such as (40–43), there are no durational or epistemic problems because the speaker has the control over the event he is describing. In the case of performatives (40–41), the event is the speech act itself. In the case of instructions (42), including cooking recipes, the event is performed by the speaker and described verbally at the same time. Moreover, the duration of the event approximately matches the duration of the speech act. In the case of sportscast commentaries (43), the speaker is certainly not in the control over the sporting event. Still, his job is to follow the course of the event as closely as possible and describe it as it unfolds. Again, the duration of particular commentaries (“Gavi passes to Lewandowski”) roughly match the duration of the events themselves, with minor time-lags expected [38]: 263–265].

The durational and epistemic problems are no longer troubling if a sentence refers not to an actual event, but to a virtual one. According to Langacker, this is what happens in various non-present uses of the present tense: “what is being encoded linguistically is not the actual occurrence of events but rather their virtual occurrence as part of noncanonical viewing arrangement” [38]: 267]. Because the event is virtual, the speaker maintains full control over its time and duration of occurrence. The virtual occurrence coincides with the time of the speaking, even if the actual one – which is our real concern – does not.

In the case of the “historical present” (35), the speaker is describing a past event as if he was reading a document or mentally “replaying” a video tape. He changes his viewpoint and mentally puts himself in the past. A similar mechanism explains present tense sentences referring to future events, such as (36). The speaker is describing the future as if reading off from a virtual schedule or plan. The event directly referred to is a mental representation of the actual one and virtually occurs at the time of speaking. Note that this does not apply to events that cannot be scheduled or fully predicted (*It rains at 12:00). Finally, “timeless”, universal sentences such as (37) and generalisations (38) may be explained as read off from virtual documents representing the stable structure of the world.

The non-present use of the present tense also applies to directive speech acts, including those already invoked:

• 44) Now we slice the onion into thin slices.

• 45) We raise our arms and turn our wrists again.

• 46) You go through the lights, then turn right.

• 47) No more discussion! Tomorrow, you go to work!

Sentences (44) and (45) may be read as a cooking recipe and a fitness instruction in which the speakers instructs the hearers. They both involve a virtual actor (“we”) that may be actualised by anybody who chooses to follow the instructions. Similarly, the road directions (46) and the order in (47) feature a virtual actor (“you”). All these sentences describe virtual events that occur virtually as they are “read off” by the speaker. They depict a virtual agent (“we” or “you”) following the instructions and cutting onions, raising arms, turning right, etc. Their directive character stems from the fact that they are supposed to be carried out, in reality, by actual actors. Using the previous terminology of illocutionary metonymy, we may say that these sentences concern the after component of the respective directive speech act scenario, but it is construed as occurring at the time of speaking, not in the future.

As Iwona Góralczyk and Joanna Łozińska observed when conducted a cross-linguistic study of Yoga instructions, the illocutionary force of such directives is actually stronger than in case of typical imperatives, additional contextual factors notwithstanding [18]: 634–636]. This is due to the fact that in imperative constructions (Go to work!), the hearer’s compliance with the directive is not presupposed or automatically assumed. The hearer has the “mental room” to make his decision; he can react to the directive in either a “yes” or “no” fashion. In indicative constructions, on the other hand, the hearer’s immediate and automatic acceptance of the directive is taken for granted. The event is construed as already happening at the time of speaking, leaving little “mental room” left for the hearer to express his attitude toward the directive. In sentences in the first-person plural (44–45), the illocutionary force is weakened by the use of the inclusive virtual actor (“we”). In this way, the speaker and the hearer are conceived as sharing the strain of the pressure and the requested behaviour is conceptualised as a joint activity. Similar conclusions can be found in Rita Brdar-Szabó and Mario Brdar, who characterise directives expressed in the first person plural indicative as involving a very high degree of cooperation and solidarity between the speaker and the hearer, as compared with numerous other forms of expressing directives in the context of cooking recipes [17]: 118].

Langacker’s notion of virtuality in the non-present uses of the present tense applies to legislative provisions, such as:

• 48) When appointed, a judge takes an oath before the President of the Republic of Poland…

• 49) A social enterprise employs at least 3 people.

• 50) A minister issues a regulation.

Each of these sentences construes the prescribed activity (taking an oath, employing at least 3 people, issuing a regulation) as virtually occurring at the time of speaking. By doing this, it imposes an obligation on the actual actors (judges, social enterprises, ministers); an obligation that is presumably accepted by them as there is no “mental room” left for disobedience or expressing reservations. The sentences in legal provisions never use the first person plural, so the illocutionary force of the directive is not softened by conceptualising the object of the obligation as a joint activity. Instead, the third person singular is always used. This results from the institutional framework of legislative acts and the fact that they are targeted towards a broad audience, not individuals. However, this also affects the nature of the directives they express. Directives in the third person are characterised by a very low degree of presupposed cooperation and solidarity and – in this respect – are close to impersonal constructions [17]: 117]. These features fit well with the nature of legal normativity.

5 Normative generics

Yet another example of virtuality in language, discussed by Langacker and many other cognitive linguists, is generalisation. The human ability to generalise, i.e. to extract “the commonality inherent in multiple experiences to arrive at a conception representing a higher level of abstraction,” [36]: 17] is one of the most fundamental mental abilities that form the backbone of language and cognition in general [39, 40]. Here, we are concerned with generality linguistically encoded as generic statements. Generic statements, or simply: generics, express generalisations about classes (kinds, types) rather than individuals. They take different forms in different languages, though there is no known language that has a dedicated, articulated generic operator [41]: 277]. One possible explanation for this fact is that generics, unlike quantified sentences, articulate cognitively default generalisations [9, 39, 40, 42]. Psychological research shows that generic statements, unlike quantified sentences, are perfectly understood at a very early stage of development, i.e. by two-year-olds [2, 4, 5, 9, 27, 30, 31, 39, 43, 44].

In the English language, the three most common constructions used to express generic generalisations are (1) the bare plural, (2) the indefinite singular, and (3) the definite singular. Consider the following examples:

• 51) Tigers have stripes.

• 52) A tiger has stripes.

• 53) The tiger has stripes.

Each of these sentences can be given a generic interpretation, i.e. can be read as referring not to any particular individual or a group of individuals, but to tigers as a kind (species). There are various subtle differences in the distribution of the three forms of expressing genericity (bare plural, indefinite singular, definite singular) that cannot be discussed here [45, 46]. Within Cognitive Linguistics, there are several approaches that seek to explain generic references as a linguistic and cognitive phenomenon. Langacker discusses it in terms of virtuality [37]: 96–97, 47: 211, 36: 527]. Generic statements are characterisations of the world’s essential nature referring to virtual types (51) or virtual representative instances of a type (52–53). Radden, while building upon Langacker’s ideas, describes generics in terms of metonymy instance for type and type for subtype and the conceptual blending of instance and type [41]: 280–281].

One of the crucial characteristics of generics is that they are not necessarily universal statements without exceptions. The fact that there are a number of tigers without stripes (i.e. albino tigers) does not render the sentences [6, 16, 24] false. The truth conditions of generic statements have long puzzled philosophers and linguists. Recent psychological findings suggest that the truth value of generics depends on vast causal knowledge about the world, as opposed to solely statistical data about the prevalence of a given feature [48, 49]. Generics also differ from quantified sentences referring to a number of instances of a respective class, but not to the class itself. For instance, sentence [6] is not equivalent to All tigers have stripes, Some tigers have stripes, Most tigers have stripes, etc. It seems that generic statements are not guided by purely quantitative factors, but instead are sensitive to rich, content-based factors [39]. Consider the following examples:

• 54) Birds lay eggs.

• 55) *Birds are female.

• 56) Ticks carry lime disease.

• 57) Sharks attack swimmers.

Sentence (54) seems intuitively true, as every three-year-old would confirm, although of course only female birds lay eggs. On the other hand, sentence (55) sounds plainly wrong. In fact, however, the percentage of female birds must be greater than the percentage of birds laying eggs. This is because some female birds are immature, some are not fertile, etc. According to Sarah-Jane Leslie, the difference between predicates “lay eggs” and “are female” is that the former describes a characteristic feature of a kind and the latter does not [38, 39]. Laying eggs is a prototypical feature of birds that helps us distinguish the whole category, while being female (or male) is not. The same cannot be said about sentences (56–57). Carrying lime disease and attacking swimmers are not characteristic features of ticks and sharks, respectively. However, these features are strikingly dangerous and threatening to people [37, 39]. Perhaps this is why we tend to accept such sentences even if, in reality, only up to 50% of ticks carry lime disease (depending on the region) and only a tiny percent of sharks attack swimmers. There is also an inverted dependency involved: of all creatures carrying lime disease and attacking swimmers, most are ticks and sharks, respectively.

The above features explain why generics are often used to verbalise stereotypes, including harmful social stereotypes. Some authors even claim that generics provide the typical form of expressing stereotypes [28, 29]. Consider the following examples:

• 58) Dogs fight with cats.

• 59) Italians love pizza.

• 60) A boy doesn’t cry.

• 61) Poles are heavy drinkers.

• 62) Blondes are stupid.

As we remember, Langacker described generics as characterisations of the world’s essential nature. This applies not only to natural kinds (i.e. tigers have stripes), but also to social kinds, such as described in sentences (58–62). By describing social kinds with generic sentences, we presuppose certain essential characteristics about them; we “essentialise” them [42]. For example, sentence (61) does not entail that all or even most Poles are heavy drinkers, as it would be patently false; nor does it entail that some Poles just happen to be heavy drinkers. Rather, it entails that being a heavy drinker constitutes an essential characteristic of being a Pole or – in other words – it locates the source of being a heavy drinker in being a Pole. It may be explained in terms of a pragmatic implicature: “Poles are heavy drinkers by virtue of what it is to be a Pole” [1, 18, 50]. There is psychological evidence that the use of generic sentences facilitates essentialist thinking about social kinds and, in turn, endorses real-life stereotypes [17, 48, 49].

In our context, the most interesting aspect of generics is that some of them have a normative force. This mostly applies to generics referring to social kinds, roles, models, etc. For example, the sentence (60) may be read as communicating a generic proposition that boys do not cry, but also as communicating a norm that boys should not cry, or a deontic statement that if someone is a boy, then he should not cry [7, 28]. Consider other examples:

• 63) A real man provides for his family.

• 64) We share. [i.e. uttered by a mother to her unwilling-to-share daughter]

• 65) Men do not wear pink trousers.

These generic sentences use a variety of grammatical forms (singular – plural, first person – third person), but can all be reasonably read as expressing social norms. Sentence (63) may be interpreted as expressing a norm that being a man requires one to be a provider for the family. This normative effect is strengthened by qualifying the grammatical subject with the evaluative adjective real. Sentence (64), in the evoked context, depending on the prosodic factors involved, may impose a very strong moral obligation to share with others. Sentence (65) was actually uttered by my three-year-old daughter as a commentary to my purchase of a pair of pink trousers. Clearly, it was not a description of the current fashion trends; rather, it was a prescription that because men (in general) do not wear pink trousers, I should not be wearing them.

Normative generics have so far not been the subject of systematic linguistic treatment and there is no commonly accepted explanation of them. Langacker and other cognitive linguists who discuss generic reference in their theories (see also: [41]) are yet to offer a direct treatment of normative generics.¹⁴ Arguably, Langacker’s approach, which characterises generics as statements about the world’s essential nature referring to virtual types and virtual representative instances, is flexible and robust enough to explain normative generics as well. Still, it may be instructive to discuss several competing theories that, although different from Cognitive Linguistics in their fundamental assumptions, address precisely this subject matter:

According to Ariel Cohen, generic sentences can be given two readings: inductive, i.e. expressing descriptive generalisations and normative, referring to a rule or a regulation: “The respective logical forms of the two readings are different; whereas the former reading involves, in some form or another, quantification, the latter has a simple predicate-argument structure: the argument is the rule or regulation, and the predicate holds of it just in case the rule is <  < in effect >  > ” [14, 45]. He observes that, while bare plural generics in English are often ambiguous between these two readings, singular generics clearly favour normative reading [14, 45, 52]. The rules invoked by generic sentences may be of different kinds, including linguistic rules (definitions), but the paradigmatic cases are “conventions, i.e. man-made, explicit regulations” [13, 45], such as (63). The source of the normativity of generics, under Cohen’s account, is found at the level of syntax, namely in the underlying logical form of normative generics.

A competing explanation of the phenomenon in question was offered by Sarah-Jane Leslie. Her approach, as opposed to Cohen’s, may be described as semantic [40]: 133]. She suggested that the key to understanding normative generics is the “dual character” of the concepts they characterise. This seems to apply to some social kind concepts, namely those that “are believed to have primary social roles/functions that are dissociable from the criteria for membership in the kind, in the sense that one can count as a member of the kind without occupying the role in question, and one can occupy the role in question without counting as a member of the kind” [40]: 128]. For instance, the concept of man, as in sentence (63), clearly has a dual character: the descriptive sense (membership criteria) involves biological features, while the normative sense refers to the social ideal associated with men, which presumably includes them being the main providers in a family. A given biological male may satisfy the descriptive criteria, but fail to occupy the required social role. In such a case, one may say that “he is not a real man.” Conversely, the social role may be occupied by a biological female, in which case one may comment, for instance, that “she is the real man in the family.” The normativity of a generic sentence is thus accounted for not by postulating an entirely different logical form underlying the sentence, but rather in terms of the lexical polysemy of the relevant nominal phrase. The “dual character” concept has two related, but distinct senses: descriptive and normative [40]: 119–120]. This explains why one can reasonably say that “he is a man, but not a (real) man,” signifying that a given person satisfies the descriptive criteria, but not the normative ones (or vice versa). Note that the adjective real in this example, as in sentence (63), functions not as an intensifier, but as a modifier. “Real man” is not just a subcategory of “man”. In fact, the extensions of both categories need not overlap at all, because they have quite different criteria of membership [40]: 116].

Yet another theoretical proposal comes from Samia Hesni, who opposes to Leslie’s semantic account and instead explains normative generics in terms of Gricean implicatures [53]. She points out that normativity may be ascribed not only to generic statements about dual character concepts, such as man, woman, or scientist, but also concepts that clearly lack a dual character, such as rock or pharmacist. Given the right pragmatic circumstances, virtually any concept may be used to construct a generic sentence with normative force. This leads her to question the polysemy thesis and to locate the source of normativity of generics at the purely pragmatic level of utterance. Contrary to Cohen, she argues that bare plural generics are more likely to be given a normative reading than indefinite singular generics [17, 54, 55]. Indefinite singular generics, on the other hand, serve a metalinguistic function – they stipulate definitions of respective terms. Under Hesni’s account, sentence (63) should be read as a proposal to restrict the use of man to individuals who provide for their families [52, 55]. Indefinite singular generic statements about social kinds generate normative readings more easily than those about other kinds of concepts (i.e. natural kinds), simply because social kind terms tend to be less well-defined and, therefore, more prone to metalinguistic claims about their proper use [28, 55].

From a Cognitive Linguistics’ perspective, it seems that normative generics should be given a semantic explanation. This makes Leslie’s approach the most plausible one. In my opinion, however, the relation between descriptive and normative readings of a generic sentence requires more than just positing lexical polysemy of the nominal phrase, for reasons so aptly expressed by Hesni (and some others). To put it another way: lexical polysemy should be seen as a result, rather than the cause of a normative reading of generic statements. The cause, perhaps, may be sought with the help of the ideas of virtuality (Ronald Langacker), Idealised Cognitive Models (George Lakoff), and mental spaces (Gilles Fauconnier). Regardless of what is the most appropriate explanation of normative generics, we are now ready to show how they can help us understand the normative character of legislative provisions.

Statutory sentences written in the descriptive form may be read as normative generics. Consider some familiar examples:

• 65) When appointed, a judge takes an oath before the President of the Republic of Poland…

• 66) A social enterprise employs at least 3 people…

• 67) The court orders the forfeiture of items coming directly from the criminal conduct.¹⁵

Firstly, sentences such as (65–67) may be deemed generic statements. Although they use indefinite singular nouns, they do not refer to any specific individuals or institutions, i.e. specific judges, social enterprises or courts. Likewise, they are not universal generalisations referring to all members of respective categories. They cannot be falsified by giving counterexamples, i.e. acknowledging that a particular judge refused to take an oath before the president. This may be explained by the fact that they refer to ideal judges, ideal social enterprises and ideal courts. To put it in Langacker’s terms, sentences (65–67) refer to virtual representative instances of the respective kinds. These representative instances are a part of the structure or essence of the world. The features predicated about them are not just any features, but presumably only essential ones. This is especially true about indefinite singular generics, which, as we have seen, tend to be given definitional reading. All this applies perfectly to our example sentences (65–67) once we acknowledge that we are talking about the legal world, legal definitions and legally essential features. The language of statutes creates a whole new virtual world with its own internal structure. This claim is based on the ideas of cognitive linguists, such as mental spaces and Idealised Cognitive Models, but also resonates with ideas from legal theory. Recently, some legal philosophers have proposed that law is best explained when viewed as a social plan to be collectively executed [56] or a vision of a possible world to be brought into existence by legal agents [43].

Secondly, sentences (65–67) are not only generics, but normative generics. Despite their grammatical forms, they impose obligations upon their addressees: judges, social enterprises and courts. As observed by Sally McConnel-Ginet: “speakers uttering sentences like those (…) are usually urging their addressees to act so as to make the actual world more like an ‘ideal’ world of which these sentences could truly be uttered descriptively” [44]: 273]. For instance, sentence (65) seems to convey not only that taking oath is what judges generally do, but also that if you want to be properly considered a judge, you must take an oath. The mechanism responsible for the normative character of this sentence is aptly described by Leslie as the “connecting principle”: “If one is a member of a social kind, and that social kind has a particular primary role or function, then there is a prima facie obligation to fulfil that role or function, and do so effectively” [40]: 130]. This is also reminiscent of Hesni’s proposal to treat indefinite singular normative generics as stipulating definitions of respective terms. The difference is that the function of legal provisions is not metalinguistic – what is at stake here is not linguistic, but the legal status of a judge.

The case of legal normative generics shows the limitations of Leslie’s polysemic approach. Granted, judge is likely to be considered a “dual character” concept, namely there is a distinctive social role that judges are expected to occupy regardless of their formal qualifications (i.e. being fair and impartial, seeking truth, pursuing justice, etc.). As Lon Fuller once observed: “It is probably well that our legal vocabulary treats a judge as a judge though of some particular holder of the judicial office I may quite truthfully say to a fellow lawyer, <  < He's no judge >  > ” [52]: 122]. However, the theme of the provision in question, which is taking an oath at the time of their appointment, is arguably not part of the social ideal associated with a judge. Moreover, the subjects of sentences (66) and (67), social enterprise and court, cannot be considered “dual character” concepts at all. They are not names of occupations or social roles, but institutions; in the case of social enterprise – an institution created from scratch by the statute. The role of context seems to be crucial. According to Cohen, an indefinite singular sentence “that, in the null context, cannot be read generically, may receive a generic reading in a context that makes it clear that a rule or a regulation is referred to” [45, 52]. There can be no clearer indication that we are dealing with a rule than putting a sentence in a legal authoritative text, such as statute or regulation. Therefore, the legal context is what generates the virtuality of the reference and turns our attention to ideal, rather than actual, judges, social enterprises and courts. Still, the mechanism behind this is by no means legal, it is linguistic and cognitive.

6 Conclusions

We have now discussed four different possible explanations for the normativity of the descriptive form of legal provisions that can be found in Cognitive Linguistics: conceptual metaphor, speech act metonymy, virtual present tense and normative generics. It is important to see them not as competing, but rather complementary. For example, generic reference is possible because of the virtual use of the present tense, which, in turn, can be analysed as a case of speech act metonymy [18]: 628]. Different theories, however, enable us to appreciate different facets of the phenomenon in question. Conceptual metaphor is perhaps the most general explanation, addressing the most metaphysically relevant aspect of expressing norms in descriptive language. Speech act metonymy offers a pragmatic explanation and has the capacity to cover various forms of expressing obligation. The virtual use of the present tense and normative generics seem like two sides of the same theoretical coin. They are both manifestations of virtuality in language. They put the focus on the semantic aspect of the phenomenon.

The analyses presented in this article confirm that the descriptive form of expressing directives is no legal peculiarity. It is a common linguistic practice that pervades various languages in a multitude of social contexts. Accordingly, it requires a linguistic rather than purely legal explanation. The advantage of Cognitive Linguistic approaches, as pursued in this paper, is that they go beyond the purely linguistic (i.e. lexico-grammatical) level and reveal cognitive mechanisms responsible for the phenomenon in question. Admittedly, Cognitive Linguistics is just one – though currently very influential – among many competing paradigms within linguistic theory. Thus, one may object to the analyses and explanations presented in this paper simply by objecting to the basic assumptions of Cognitive Linguistics, such as the non-formal approach to meaning or denial of the modularity of language. Still, the hypotheses presented here can be put to a practical test. Legal linguistics and legal translation studies seem to be particularly fruitful areas of application. As numerous works cited throughout the paper show, the problems of expressing legal norms through the statutory language are vividly discussed in these disciplines, especially in a multilingual context. The findings presented here may offer not only a theoretical explanation for using descriptive language to express normativity, but also practical guidance for drafting and translating legal provisions in accordance with underlying cognitive structures. Eventually, they may also tell us something important about the nature of normativity and, hopefully, the very nature of law.