Why Judicial Formalism Is Incompatible with the Rule of Law

draft for discussion purposes (29/08/2016) Marcin Matczak University of Warsaw [email protected] WHY JUDICIAL FORMALISM IS INCOMPATIBLE WITH THE RULE OF LAW ABSTRACT When a judge follows the letter of the law, her judgment may be considered blinkered by the man in the street. Legal professionals, however, would classify the judgment as formalistic. From a theoretical perspective, formalistic decision-making limits the number of premises on which a judge may base a verdict. It asks the judge to focus on the literal meaning of the legal text and to disregard other interpretative premises, like the purpose or function of the law, legislative history or – in civil law jurisdictions – previous court decisions. Formalism as an art of limiting judicial choices is perceived by many as fully consistent with the rule of law. It seems to both allow the curtailment of interpretive discretion and to ensure fidelity to the will of the lawmaker. This contrasts with an all-things-considered approach, where the premises for judicial decision-making seem unlimited, discretion encouraged, and the will of the lawmaker ignored. In this paper I show that the apparent compatibility between formalism and the rule of law is based on a particular assumption as to the nature of legal language: that this language is criterial in the sense that in order to understand it, one needs to rely on dictionary definitions understood as sets of criteria. This assumption is misguided, and its flaws are revealed by theoretical advances in the contemporary philosophy of language. Specifically, semantic externalism demonstrates that the meaning of language cannot be discovered merely by reading dictionaries; rather, it requires the investigation of the linguistic practices of a particular communicative community, and an insight into the history and function of individual legal terms. Because the nature of legal language is different from that assumed by the formalists, the compatibility between formalism and the rule of law collapses. With such a distorted perspective of the characteristics of legal language, formalism cannot ensure fidelity to it. This paper shows that judicial decisions based on applying definitions are very often surprising to the law’s addressees; this contradicts one of the main tenets of the rule of law, namely, the predictability of court verdicts. As a consequence, the rule of law requires a different, moderately non-formalistic approach to legal interpretation. Within this approach, judges can make decisions based on a broader scope of interpretive premises and by doing so ensure a better level of predictability. 1 draft for discussion purposes (29/08/2016) 1. INTRODUCTION This article offers a critical assessment of the method of applying law traditionally referred to as judicial formalism. By and large, formalism rigidly follows the letter of the law, even if the outcome appears unjust or if it is at odds with common sense. In the philosophy of law, formalism is a theory of the application of law that encompasses several theories of interpretation, including textualism and originalism1. The defining feature of formalism is the belief that by limiting the number of available interpretive premises, the number of interpretative choices for lawyers will be similarly restricted. This restriction may take different forms: in the case of textualism it equates to acknowledging that out of many possible readings of the legal text only the “plain” one is correct2; in the case of originalism it consists of limiting the number of available meanings of the legal text to the original one3. Other forms of formalistic limitations to the interpretive premises exist: formalism does not take into consideration the purpose of legislation, nor does it accept application of general principles to the case at hand. Rather, it promotes the application of so-called “bright-line” and “most-locally applicable”4 rules, even if the result of their application cannot be reconciled with those general principles. Judicial formalism is not a purely theoretical construct but one constantly intertwined with our daily lives. We encounter it while pursuing actions in courts or authorities and often experience disappointment when faced by the specific, formalistic rationality of bureaucratic modus operandi. This disappointment is fuelled by frustration when our case is seemingly decided contrary to common sense. Such decisions, we are often told, are dictated by the so-called letter of the law, an approach which disenchants those people dealing with courts or authorities who expect their decisions to be dictated by justice. Disenchantment with formalism has a long lineage: indeed, as far back as 1BC Cicero occasioned the legal maxim Summum ius, summa iniuria, “The highest law, the 1 Bix (2004) defines formalism as a group of opinions about legal interpretation that includes both textualism and other opinions accepting as correct interpretation consistent with the original legislative intentions. 2 Textualism supports so called “plain meaning” (D’Amato 1991, passim; Ross 1995, passim), understood as an acontextual meaning resulting from the semantic autonomy of language, i.e. the capacity of language to convey meanings independently from the speaker’s intent. See Schauer (1988), p. 527 3 See Solum (2008) 4 Schauer (1988), p. 522 2 draft for discussion purposes (29/08/2016) highest injustice” (Cicero 2008, I, 10, 33): rulings rigidly sticking to the letter of the law often have little to do with justice. What is the origin of this disconnect between the letter of the law and justice? It is not necessarily error or malice on the part of the decision maker: serious arguments support the application of formalism in the decision of legal issues. These arguments include enhancing the predictability and stability of law, and curtailing judicial discretion. Arguments of this sort lead formalists to claim that a judge’s role is not to implement justice but to let the law speak for itself, without correcting it, even if it produces worse results than common sense would prescribe. In line with this position, formalists maintain that an optimal legal procedure entails reduction of the number of interpretive premises guiding the legal decision, manifest in a refusal to factor in values of legal culture other than the text of law5. This argument is clearly attractive, as formalism is widely believed to be in line with the idea of the rule of law6. This paper challenges the alleged compatibility between formalism and the rule of law. The primary reason is not merely that rulings adhering to formalistic methods are unjust; rather, they are simply not compatible with the idea of the rule of law because they disregard the nature of legal language. The nature of legal language requires interpretation that does not reduce the scope of interpretive premises to the acontextual meaning of the text of law – the central tenet of formalism. In what follows I first outline the main feature of formalism and present the structure of the argument on the alleged compatibility between formalism and the rule of law. In the second part of the paper I draw on semantic externalism to challenge the formalist view of the nature of legal language. I conclude that fidelity to legal language and the realisation of the rule of law ideal require a moderately non-formalistic, nonreductionist approach to legal interpretation. 5 Schauer (1988), p. 520 6 See Scalia (1997), p. 25: “Of all the criticisms leveled against textualism, the most mindless is that it is “formalistic.” The answer to that is, of course it’s formalistic! The rule of law is about form”). 3 draft for discussion purposes (29/08/2016) 2. WHAT IS JUDICIAL FORMALISM? Traditionally, a group of lawyer-academics active at the end of the 19th century, mainly from the Harvard Law School, are considered as representatives of classic formalism7. Classic formalists advised judges to avoid the role of a political actor and expected courts to adopt a “deductive” approach to adjudicating8. The latter consists in mechanically or automatically moving from general categories or concepts to conclusions, without taking into account political, moral or practical aspects9. Classic formalists believed that law should fulfil three requirements: first, it must be determinate, its judgments following from the application of norms to facts; second, it must be systematic, creating a coherent structure of relatively abstract concepts and principles; and thirdly, it must be autonomous, deriving its norms only from legal sources rather than from the contestable claims of religion, philosophy, or political economy10. Early critiques identified five properties of classic formalism11: a) the autonomy of law – a belief that a solution to legal issues must be searched for only within the realm of legal materials without referring to any sources outside law; b) textualism – a belief that a leading source of legal arguments is legal text, and not, for instance, purposive interpretation of legal regulations; c) conceptualism – a belief that (i) legal concepts (such as ownership or freedom) may be identified through an analysis of legislation or court judgments, and (ii) detailed rules that will be a basis for judicial dispute resolution can be logically derived from those concepts; d) originalism – a belief that the legislative history or the original understanding of a legal text by readers at the time when it was drafted 7 Cox (2003), p. 57-58. 8 Duxbury (1995), p. 9. 9 Bix (2004) 10 Grey (1999) 11 Cox (2003) and Grey (1999) 4 draft for discussion purposes (29/08/2016) has a decisive role in solving interpretive doubts related to ambiguity in a legal text12; e) “geometrisation” of the process of law’s application – a belief that the outcome of law’s application is an internally consistent, systematic and rationalised law. The legal system from this standpoint is similar to Euclidian’s geometric system, with theorems and axioms, in which an answer to a legal question may be given by referring to the logic of the system and that answer is therefore perfectly deductive and always certain; this element of classic formalism was also referred to as “mechanical jurisprudence”13. In addition to its mechanical nature, formalism in its more contemporary form (so called “new formalism”) “by distinguishing the literal mandates of the most locally applicable legal norm from some arguably better result reachable by considering a wider range of factors, reinforces the systemic isolation, or closure, of the legal system”14. This results in a “narrow” understanding of a legal issue to be resolved; formalism limits the area for interpretation, and hence denies that in a specific case there are other reasonably possible interpretations than that based on the wording of a legal text 15. The contrary position is that in applying laws, one should take into account all available arguments, and not be limited to selected arguments, in particular to the linguistic understanding of a legal text. Formalism does not deny that there is a choice in how we interpret a given rule; rather, it considers its own purpose to be to limit such choice16. In fact, formalists intend to close so-called “escape routes”: ad hoc applied interpretative manoeuvres that aim at avoiding undesirable effects of the application of law resulting from the mechanical application of a literal rule to a given case17. Typical examples of escape routes are 12 This paper does not focus on the originalist branch of judicial formalism. Some of its arguments however, especially those concerning formalism’s criterial approach to ordinary meaning, can be relevant with regard to original meaning as it is defined by the originalist. 13 Pound (1908) 14 Schauer (1988), p. 522. 15 Schauer (1988), p. 512. 16 Schauer (1988), p. 521 17 Schauer (1988), p. 519. 5 draft for discussion purposes (29/08/2016) relying on legislative history or relying on the intended meaning of the legal text18. Formalist dogma is to close all such escape routes by reducing the number of available interpretive premises. The justifications given for thus restricting choice are: firstly, the legal system does not incorporate rules operating as escape routes; and second, that the person making a decision does not have the right to create a rule that would serve as an ad hoc escape-route19. The possibility of creating those routes would increase the scope of judicial discretion to make choices motivated by individual preferences rather than the law itself, which judges are not empowered to do20. Hence, the focus on literal meaning removes the judge’s ability to depart from the plain meaning of a statutory act and thereby their ability to engage in judicial lawmaking21. One prevailing feature seems to emerge from different accounts of formalism22: the reductionism of interpretive premises. This consists in a focus on a particular factor in the interpretive process: on the text, for instance on the plain meaning of a legal regulation and the omission of all other factors – such as the purpose or function of the interpreted provision, moral or political principles (Sunstein23, Pildes24, and others25). The reduction of interpretive premises is considered to be a primary factor in securing formalism’s compatibility with the rule of law. This compatibility is discussed in the next section. 3. JUDICIAL FORMALISM AND THE RULE OF LAW One of the primary reasons for the popularity of formalism is its apparent consistency with the rule of law. It is a popular belief among lawyers that relying strictly 18 Formalism in one of its meanings is “what others call ‘textualism’ – an assertion that statutes and constitutions should be interpreted according to their plain meaning and not (where this would be different) according to their underlying purposes or their drafters’ intentions (Bix, 2004). 19 Schauer (1988), p. 520. 20 Eskridge (1990), p. 648 21 Eskridge,(1990), p. 656 22 In the literature one can encounter several other accounts of formalism, understood as: a refusal to accept that there is a freedom of choice in resolving cases taking into account the semantic interpretation of a legal rule (Hart 1961, p. 124-130); an excessive reliance on the canonical language of written rules (Llewellyn 1962, p. 183-188); the opinion that mechanical application of rules is fair (Kennedy 1973, p. 351, 355); an artificial narrowing of available interpretative choices, (Tushnet 1985, p. 1502, 1506-1507). 23 Sunstein (1999), p. 639 24 Pildes (1999), p. 612 25 Michelman (1999), p. 936, Kennedy (1976), p. 1687 and Huhn (2003), p. 309 6 draft for discussion purposes (29/08/2016) only on the legal text and omitting other interpretative premises ensures the implementation of the rule of law and not the rule of people 26. The rationale for this position is that a legal text is the most transparent interpretative premise whereas referring to extratextual premises involves a possibility of selective choice (because they are so numerous) and misunderstanding (because they are vague). Both formalism and the principle of the rule of law aim to limit the impact of the individual human factor on the application of law and to give priority to an autonomously understood legal text that incarnates the law27. The sources of this ideology go back to “Montesquieu’s contentions that in a republican political system an act of law is the act of sovereign public will, restraining arbitrary judgement of every person, including a judge. Democracy must create clear and precise laws to limit the judge’s arbitrariness”28. The compatibility of formalism with the principle of the rule of law is further corroborated by various attributes of formalist adjudication; these include the effective limitation of interpretative freedom (discretion) of lawyers29, and the shielding of law from the impact of judges’ political, moral and other individual preferences30. By doing so, formalism deals with lawyers’ concerns regarding political choices being made by judges31, which may differ from the choices of a democratic majority32. Formalism also seems to promote the predictability of the process of law’s application, and hence legal certainty – primary elements of the rule of law33. Through its focus on the plain meaning of a legal text, formalism appeals to the intuition that when interpreting we must reach out to the most tangible evidence of legislators’ expectations and to a legal material that is most available to the general public. Only a text accepted in a legislative process is a law, and the conviction of members of parliament as to the purpose or function of the law is not. Since a text is the source of knowledge of a law, then in a sense it would be unfair to rely in interpretative decisions 26 Burton (2004), p. 141 27 See Scalia (1997), passim. 28 Kozak (2002), p. 78. 29 Schauer (2008), p. 433-434 30 See Craig, Hodder-Williams (1996), p. 2 and Craig (2003), pp. 92-111. 31 Barak (2005), p. 54 32 Eskridge (1990), p. 667 33 Schauer (1988), p. 549 7 draft for discussion purposes (29/08/2016) on other bases, in particular, on legislative history as an indicator of the lawmaker’s intention (non quod voluit sed quod dixit): first of all, premises such as the lawmakers’ intention or the purpose of rules are not readily accessible to the general public; and second, they are subject to differing interpretations. Deriving a legal interpretation from a basis other than the legal text leads to the imposition of obligations on citizens of which they have not had a chance to know before taking their actions. Such decision will clearly be surprising to the public34. Involving premises other than publicly communicated text into the decision-making process may even be considered as the creation of obligations ex post facto, which clearly contradicts the rule of law. Formalists’ claims are strong arguments that the process of applying laws should be a rule-based decision-making rather than a particular reasoning in which the principles underlying the rule are directly applied to the case at hand. The former equates to deferring to the person who created the rule, the latter is an independent decision by the interpreter. The reduction of interpretive premises to those derived from the text of the rule only promotes the rule-based decision-making. and is more compliant with the rule of law. Contrary to this, a decision based on extra-textual premises, including the purpose of the rule, makes it possible for the interpreter to engage their personal preferences and prejudices in the decision-making, and as such is more akin to the rule of men than the rule of law. To summarise, formalists strongly believe that they are faithful to the rule of law because their approach is focused on the lawmaker’s language, attempting to interpret it and decode the rules. As such, the backbone of judicial formalism is its fidelity to the language of law. If one is able to prove that formalism fails to secure this fidelity, its compatibility with the rule of law collapses. Below I demonstrate that as a method of application of the law, legal formalism fails to correspond with crucial semantic features of the language of law. As such, formalism does not do justice to the linguistic character of law and thus is incompatible with the rule of law, which – as shown above – is rooted in the language and depends on it. In what follows I apply a new kind of semantic analysis to the text of law with the purpose of justifying an anti-reductionist claim, namely that the reduction of premises used in the interpretive process is a misguided strategy if one attempts to be faithful to legal text. The nature of language as seen from the perspective of recently developed 34 Sunstein (1999), p. 654 and Adler (1998), p. 248 8 draft for discussion purposes (29/08/2016) semantic theories, (collectively known as “semantic externalism”35) requires a different approach which consists in a moderate extension of the interpretive premises. Considerations provided in the next part of the paper will lead to the conclusion that formalistic claims regarding the necessary reduction in number of interpretive premises cannot satisfy their descriptive or normative ambitions. As regards the former, they fail to deliver on promises because they paint an inaccurate picture of the interpretive process. They also fall flat as normative claims because they lead to grave and unjustified oversimplifications of interpretation, resulting in incorrect conclusions. Worst of all, those oversimplifications lead to surprising interpretative results, thereby undermining one of the pillars of the rule of law: the predictability of judicial decisions. 4. JUDICIAL FORMALISM AND THE LANGUAGE OF LAW Two visions of language The question of meaning in general, and the question of the meaning of legal text in particular, can be approached in two ways. The first, based on the descriptionist theory of meaning, assumes that the meaning and reference of terms are contingent upon both a set of criteria a speaker has in mind and a fit between an object in reality and that set of criteria36. In this approach (called semantic internalism) the speaker’s intent controls how the term refers to reality. When the speaker uses a certain term, she has in mind certain criteria that constitute its meaning; to determine what the speaker refers to, the receiver must identify an object that satisfies those criteria. This approach makes use of definitions as the main vehicles of conveying meaning. As definitions are sets of criteria an object must fulfil to be covered by the definition, interpretation within this approach consists in matching extra-linguistic objects (e.g. things, human beings and circumstances) with the criteria specified in the definitions. The second approach to meaning, called semantic externalism, departs from the descriptionist, criterial theory towards one which declares that “meanings just ain’t in the head” (Putnam 1975, pp. 215-271). In its quest for meaning, semantic externalism 35 See Kallestrup (2012) 36 See Kripke (1991), p. 25-29 9 draft for discussion purposes (29/08/2016) abandons the criterial approach and turns to the usage of specific words in social practice, which – owing to an enduring tradition of standard usage of terms – determines which objects and situations occurring in the extra-linguistic reality are referenced. The meaning of a word is identified based on the nature of reality to which that word refers, i.e. on the basis of the features of external reality systematically indicated by that word. Semantic externalism differs from its internalist counterpart in many ways, and some of those differences are particularly relevant for legal theorists. First, according to semantic externalism, it is not the intention or knowledge of the speaker that determines which elements of reality the speaker’s words refer to. The reference is rather determined by the tradition of those terms’ usage. A word or sentence refers to a state of affairs to which a particular group of people traditionally (i.e. for a relatively long time and in a typical way) have referred. Hence, semantic externalism shifts the focus away from the author and focuses instead on her words and the linguistic practice of the communicative community that created and interprets the text, as well as the reality referenced within37. The speaker’s intended meaning is no longer relevant for interpretive purposes. What replaces this focus on intention is one on the relationship between words and reality at the particular instances of this word’s usage in the past. Second, meaning is no longer understood as a description (Mill, Russell) or a cluster of descriptions (Searle). To determine the meaning of a word one does not need a definition, understood as a set of criteria which must be fulfilled by the referent. What must be determined is the set of properties or states of affairs to which this word referred in the past. The ability to refer to those properties and states of affairs constitutes “the proper function” (Millikan, 1984) of terms and sentences, a crucial element of meaning within the externalist framework. Third, semantic externalism is focused on the particular instances of language usage. The language is perceived as a living practice, and not as an atemporal, static structure. In other words, the externalist approach assumes that pragmatics prevails over semantics: it is the use that influences the meaning and not the other way round. 37 Bix (2003, p. 287) criticises semantic theories stemming from Kripke and Putnam for prioritising the “word meaning” at the expense of “speaker’s meaning.” He finds such an approach at odds with the essential feature of law, the lawmaker’s power to change the meaning of words in the legislative process. I do not see this criticism justified, as the lawmaker’s power to choose the words constituting legal text, not to change their meaning, seems to be the essential feature of law. 10 draft for discussion purposes (29/08/2016) The definition-based approach versus the paradigm-based one Formalism draws on semantic internalism, not semantic externalism. As a consequence, it perceives legal concepts as criteria-based concepts. Ultimately, formalism considers the interpretation and application of law to be a process of subsumption (qualification of facts according to criteria), the outcome of which is that facts are qualified according to whether or not they fulfil criteria established in definitions. This approach underweighs the context in which the language is used, especially linguistic practices and traditions in the speaker’s community. It also neglects the fact that in using language we do not apply definitions to objects and situations, but rather seek analogies between previous uses of a word or phrase and the current use. Semantic externalism rejects a criteria-driven analysis and instead examines the history of linguistic practice of a particular community in order to find meaning; by doing so it steers clear of mechanistic, subsumption-based reasoning. As Stavropoulos points out, it is precisely the point of the externalist framework “that determining the content of a concept is not a mechanical exercise, but is a complex-theoretical-procedure shot through with evaluative judgments (Stavropoulos 1996, p. 10).” The starting point for the externalist critique of formalism is the latter’s conceptualisation of meaning, borrowed from semantic internalism. Semantic externalism rejects the notion of meaning defined as a set of characteristic features attributed to extra-linguistic objects allowing one to either classify the given object as a designator of the term or disqualify it if some features are missing (Kripke 1991). Kripke argues that meaning is not constituted by the compliance of the properties of the object that the speaker refers to and those making up the content of the term (so-called clusterof-properties or –descriptions, making up a definition). It is false to believe that for one object to be a designator of a certain term it must satisfy the majority or even all conditions prescribed in the definition38. According to semantic externalism, while identifying objects to which the speaker refers, one must establish not the object the speaker had in mind39, but the object to 38 Kripke (1991), p. 91-92. 39 See Hughes (2006), p. 36; See also Kripke (1979), p. 6-27. 11 draft for discussion purposes (29/08/2016) which the term refers, on the basis of “the most coherent explanation of name or termusing practice40”. This “term-using practice”41 starts with a so-called “original baptism”42 and continues with subsequent uses, which are called “multiple groundings” (Devitt 1981: 138). For Kripke, “original baptism” is the first naming of the object. This first naming consists of description or simple indication (as in the baptism of a child). Evans43 indicates that it is not only the first naming that is significant for the meaning of the given word; subsequent individuals refer to the object in the causal chain by invoking its name, thus, as it were, grounding its reference. A similar conviction that subsequent uses of a term influence its meaning is present in Millikan (1984) in her idea of historical chains of language usages − “lineages”44. In Millikan’s theory, linguistic meaning has been shaped in the process of repeatable co-occurrence of words (sentences) and states of affairs. A name co-occurs with a person, a noun co-occurs with a thing, a predicate cooccurs with a given attribute. The co-occurrence can be physical (words and things or qualities are present at the same time and place), or of a historical-causal nature (as in Kripke-Putnam semantics, in which our current reference extends back to the first use of the word). Multiple groundings or lineages are the main mechanisms of how the tradition of linguistic uses influence meaning. If, as Millikan puts it, meaning is the proper function of linguistic tools45, to understand what this function is one needs to trace back the instances of previous uses. Language users usually do this already during language acquisition, which consists in observing instances of a term’s use and grasping what elements in the surrounding reality the term picks out. Later, the tracing of the tradition takes place in the learning process, when one makes oneself familiar with texts and other utterances from the past, developing a knowledge of how they have referred to the world. Multiple groundings and lineages are theoretical notions covering what lay 40 Stavropoulos 1996, p. 8 41 Kripke (1991), p. 106 42 Devitt (1981), p. 26 43 See the discussion in Mallon (2015), p. 11 44 Millikan (2005, p. 38): “The phenomenon of public language emerges, I believe, not as a set of abstract objects, but as a real sort of stuff in the real world, neither abstract nor arbitrarily constructed by the theorist. It consists of actual utterances and scripts, forming crisscrossing lineages”. 45 Millikan (1984), p. 28 12 draft for discussion purposes (29/08/2016) people call “practice”; they are similar to Wittgensteinian language games46. The more instances of a particular linguistic behaviour one observes, the better one knows the practice, and the better one is equipped to repeat the behaviour correctly in the future, i.e. to correctly link the words with the world. When applied to legal language, the concept of multiple groundings and lineages changes the traditional perception of how the meaning of legal terms is shaped. Upon completing an utterance in the language of law (e.g. provision or norm), the lawmaker applies the terms of ordinary language, which, according to semantic externalism, already have a chain of multiple groundings by virtue of being used in linguistic practice of the given communicative community47. Contrary to a common belief amongst legal philosophers, the lawmaker is not the original reference-fixer, the baptiser48. In enacting a new law, the lawmaker includes into the legal language words that have been used before, along with their histories (their lineages). If this were not the case, communication via legal text would be impossible: no lawmaker can effectively communicate with newly coined words alone. Naturally, in the language of law some terms acquire specific legal meaning right at the moment they are incorporated into it, for example by passing specific legal definitions that modify the ordinary meaning of the word. But mostly, terms used in legal text are adopted from ordinary language as they are, unchanged. With time, terms used to express the law gain an increasing number of groundings both in the domain of ordinary language and the domain of legal language. This is so because the communicative community uses those terms while referring to the reality in which it functions. In legal discourse, court rulings and administrative decisions are examples of multiple groundings. Opinions expressed in legal doctrine also belong here, as lawyers employ legal terms and refer them to reality. 46 Wittgenstein (1958) 47 In this paper I assume that the original area of application of semantic externalism, namely proper names and natural kind terms, may be extended to include fields of social practice where one uses theoretical concepts, and law is undeniably such a field. The scope of this paper prevents me from proving the validity of the assumption made. However, the viability of extending semantic externalism to cover concepts other than proper names or natural terms has been demonstrated in Stavropoulos (1996, p. 6768) where he treats at length semantic externalism’s relevance for the language of law, and Michael Devitt (1981, p. 199) who suggests that causal theory of reference can be applied to terms other than proper names and natural terms, for example to theoretical concepts. See also Burge (1979) and Brink (1984). 48 See Stavropoulos (1996, p. 46) who suggests that lawmakers are original reference-fixers, i.e. that they perform a naming ceremony by using legal terms in the legal text. 13 draft for discussion purposes (29/08/2016) Within this picture of how legal language operates, the task of the interpreter differs significantly from the one proposed by semantic internalism and judicial formalism. In those approaches, interpretation is a deductive process which involves forming a definition and applying it to the external world. Instead of such a “definitionbased” application of general terms, semantic externalism proposes a “paradigm-based” application of the terms49. In this model, interpretation is an inductive process of tracing previous instances of using a particular term or phrase and forming a theory of what is the proper function of the term or phrase. To form this theory, the interpreter must identify typical properties that are picked out by the term when it is used. This cannot be done by simply reading a definition. Instead, it requires immersion in the contexts of particular uses from the past and finding a common denominator among them. A new approach to plain meaning Semantic externalism requires lawyers to change their approach to the plain or ordinary meaning: from one based on definitions to one based on lineages. A useful example of a lineage-based approach has been provided by Recanati (2004) in the form of “semantic potential”, understood as an ability to refer to some set of features in the world50. Recanati distinguishes between “source” and “target” situations: the “source situations” are all the situations in which a term has been used in the past51; the “target situation” is the current situation in which a term is to be used. The task of the speaker is to assess whether the use of the word in the target situation is appropriate as compared to its use in the source situations. The set of source situations provides the speaker with some comparative material to perform this task. To assess whether the term should be applied in the target situation, the speaker first needs to assess if and to what extent the features of that situation resemble the features of the source situations. 49 “Howard Wettstein distinguishes ‘definition-based’ from ‘paradigm-based’ stories about the application of general terms. On the paradigm-based story, one is ‘exposed to a certain number of cases, and . . . perhaps corrected on a number of occasions on the application of the term, one [then] gets the feel for what is to count as a genuine application of the term, somewhat like the way one gets the feel for how to serve in tennis’ (Jackson, 1998, p. 65). 50 Recanati (2004), p. 141 51 Recanati (2004), p. 143. Although Recanati does not use her terminology, the chains of the source situations resemble Millikan’s lineages. 14 draft for discussion purposes (29/08/2016) The speaker’s second task is to find out which similarities between the source and target situations are relevant and which are not. Take the term “dangerous tool” used in some criminal law regulations. The previous applications of that term allowed one to form a definition which would surely involve some criteria concerning the shape of the tool (e.g. hardness, sharpness), some affordances (e.g. being able to shoot with it). If one faces a target situation in which a criminal uses a knife to threaten a victim, the criteria of the definition are sufficient: the target situation is similar in many aspects to the source situations, so the decision on whether to apply the term “dangerous tool” to the knife is an easy one. The same goes for situations involving the use of a gun. If, however, one is presented with a target situation in which a robbery has been committed by a kung-fu master who used only his bare hands to threaten his victim52, a more difficult question arises: is the target situation similar to the source situation in which “dangerous tool” has been used? It may happen that no available definition allows for including bare hands within the term, as no one ever faced a situation similar to the target situation. Nonetheless, the target situation is similar to the source situations in some relevant sense: the bare hands of the kung-fu master fulfil the same role as a knife or a gun, namely dangerous implements used to trigger the victim’s fear and encourage them to give their belongings to the robber. The definition itself is inadequate but our knowledge of source situations can be used to decide whether the application of the term to the target situation is appropriate. Within the context of his analysis, Recanati makes use of the idea of open texture53. The open texture is a flaw inherent to definitions understood as sets of criteria formed on the basis of previous applications of a particular term. The proper use of the term in the subsequent situation depends on whether that situation fulfils previously formed criteria. The problem is that the previous situations in which the term has been used had many features, only some of which a definition can capture. When next applied to the target situation, the definition may be found inadequate if it does not contain the features relevant from the perspective of the target situation. To diminish the negative impact of the open texture, Recanati proposes not to use definitions at all as 52 This is an actual case heard by the Polish Supreme Court 53 Recanati (2004), p. 143. Originating from Waismann, this is the same concept as that used by HLA Hart in “The Concept of Law”. 15 draft for discussion purposes (29/08/2016) intermediaries between the source situations and the target situation54. Instead, he proposes that the target situation should be compared with a raw set of source situations and the language user should decide on the resemblance between the target situation and the source situations on the basis of all features to which the term earlier referred, i.e. on the basis of the full semantic potential of the term55. According to Recanati, in novel and difficult cases we make use of our knowledge of previous situations to assess how to apply the term to the target situation. We compare the two and eliminate those features that are not relevant and select those which are. The set of features selected in the case of one target situation may be different from the features we select in another; the selection depends on the function the target situation has for us. Millikan’s idea of lineages and proper function may be of help here. The set of sources situation in which a term has been applied because a set of features having a particular function existed in those situations forms a lineage. The same word can have several lineages as it may be used with regard to a different set of features or to the same set of features but having a different function for us. Take again the kung-fu-master case. Any definition of “dangerous tool”, formed as a common denominator of the features of all source situations in which that phrase has been used previously will be only a selection of the features. It is probable that one of the features selected to the definition would be a feature of “being an implement”. That feature is not present in the target situation: human hands are not implements in the literal sense. Thus, the target situation does not fall under the definition of the “dangerous tool”. Yet judges deciding that case found the hands of the kung-fu master to be dangerous tools because they compared the target situation with all relevant features of the source situation, irrespective of whether they were included in the definition. One of the features was the ability of the “tool” to harm or the fact that the perception of the “tool” can trigger fear in a robbery victim. Recanati’s approach to literal meaning is paradigm-based, not definition-based. It assumes the definition-based approach is too reductive: any definition may provide speakers with only a small number of relevant features of the source situations and there is no guarantee that the selected features will be relevant for a novel target situation, as in the example of the dangerous tool. Therefore, Recanati proposes to 54 Recanati (2004), p. 143 55 Recanati (2004), p. 147 16 draft for discussion purposes (29/08/2016) expand the number of features to take into consideration instead of reducing them. Reduction petrifies the language and makes it unable to react to a new target situation. The application of a word to the situation may be justifiable because it resembles the source situations; however, it is excluded because that resemblance does not fall within the features selected by a single definition. The result is inconsistency: a term should be applied to a situation but it is not. Recanati’s approach of expanding the number of features interpreters take into consideration allows them to be more consistent and precise in applying terms to reality. Combining Recanati’s concept of ordinary meaning with Millikan’s idea of lineages provides a robust externalist theory of language. This theory can be used to criticise the reductionism of interpretative premises advocated by the proponents of judicial formalism. 5. CRITICISM OF REDUCTIONISM OF INTERPRETATIVE PREMISES Traditionally, the criticism of reductionism of interpretative premises has been undertaken from two positions: a utilitarian and a contextual one. Utilitarians oppose the reduction of interpretive premises because such reduction brings about an inferior result in the process of applying the law when compared to an all-things-considered approach56. An all-things-considered approach would include, for instance, the legislative purpose and intention57, a change in the social environment since the time of enactment, the individual situation of the person to whom law is to be applied, and finally, the consequences of interpretative decisions. Utilitarians emphasise that allthings-considered legal thinking is closer to practical reasoning, the principle of which is to maximise the amount of data in the decision-making process rather than reduce it58. As a result of such maximisation, the final decision is optimal. The utilitarian argument doesn’t undermine the main formalist line of argument, namely, the need to be faithful to the language of the law, because it simply doesn’t address it. Formalists do not treat the outcome or utility of the outcome as the main criterion to assess the appropriacy of judicial decisions. They are ready to accept a sub56 Schauer (1991), p. 113 57 Michelman (1999), p. 937 58 Eskridge and Frickey (1990), p. 348 17 draft for discussion purposes (29/08/2016) optimal outcome, one inferior to that which would be achieved in a particular case by taking all aspects relevant for a given matter into consideration. This inferiority is a cost they are willing to pay for a standardisation which ensures the similarity of all decisions undertaken in similar cases. Therefore, formalists argue that being faithful to a rule means that the person making an interpretation should not have any scope to alter suboptimal outcomes where they arise59. In other words, the individual attempt to optimise a decision may endanger a rule’s standardisation function. A better argument is provided by the contextualists, who criticise the reduction of the interpretive premises not from a utilitarian but from a linguistic point of view. Contextualists (e.g. Barak60 and Fish61) claim that understanding a legal text is not possible without taking into consideration interpretative premises derived from the context in which that text was developed or in which it is interpreted. Consequently, they argue that context-based interpretative premises (e.g. legislative history or a change of circumstances) should be treated as fully legitimate interpretive premises. Formalists are acontextualists. Their main argument for reducing interpretive premises derives from the nature of linguistic meaning; they believe that meanings are acontextual sets of criteria, abstracted from particular situations of language use. This is not to say that all formalists are literalists. The authors supporting formalism acknowledge that “the literal or dictionary definitions of words will often fail to account for settled nuances or background conventions that qualify the literal meaning of language”62 and observe that “no mainstream judge is interested solely in the literal definitions of a statute’s words”63 The criticism of the definition-based approach discussed here is not directed against literalism. It rather targets an assumption that one is able to identify the ordinary meaning of a word by analysing criteria set forth in its definition. As Christopher Hutton writes: “One way in which courts feel they can access the ordinary language user’s point of view is through dictionary definitions”64. What I argue is that definitions cannot give access to the ordinary meaning because they are 59 Schauer (1991), p. 100 60 Barak (2005), p. 26 61 Fish (2005), p. 629 62 Manning (2003), p. 2393 63 Nelson (2005), p. 376 64 Hutton (2014), p. 44 18 draft for discussion purposes (29/08/2016) abstract generalisations and therefore artificial65. Hutton argues that a convincing response to the accusation of the generalisation can be that rules are abstract in the sense that they have generalised focus (Hutton 2014, p. 45-46). However, I argue that this response is not convincing at all – the generalisations we make must be always understood as rooted in particular contexts from which they have been derived. Without doubt, each rule set down in the legal text is a form of generalisation66. This generalisation consists of criteria which a fragment of reality must fulfil to be covered by the definition. The formalist’s problem is that they believe the criteria set forth by the generalisation are immutable in the sense that they are independent from the particular state of affairs to which they are applied. In the kung-fu example, the formalistically understood rule covering “dangerous tools” would oblige the tool to be an implement and as such exclude the application of the rule to hands. This is wellintentioned: the main task of generalisations is to confine the risk of individual error in applying the law to a particular case. Without such generalisations, the individual would be forced to decide the case on an ad hoc basis, using the rationale for which a legal rule has been set, but not applying the rule itself. But where do generalisations come from? Legal provisions contain general terms that are abstractions from experience: we experience particular situations and abstract some set of features common to those situations. The term “tool” encapsulates all experiences of tools, the term “dangerous” summarises all experiences of situations in which a risk or threat is involved. A set of features we extract from such situations is usually expressed as a definition; e.g. the definition of a tool consists of a set of features abstracted from our experience of tools. Within this approach to generalisations, interpretation consists of applying the definition to new situations, events and features. According to this picture, to use language is to make deductions: i.e. to apply a general set of uniform criteria to particular cases. That definition-based approach to language can be contrasted with a paradigmbased one. As we have seen, in order to form a definition and apply it to a new situation, we need to have an experience that led to the formation of the definition: the definition is thus a bridge between a previous experience and a current one. Within the paradigmbased approach, using language is not a question of applying definitions, but rather of 65 A similar argument is presented in Harris and Hutton (2007, p. 212) 66 Schauer (1991), p. 17 19 draft for discussion purposes (29/08/2016) comparing a situation to which the language is to be applied with previous situations to which it did apply. The use of a term with reference to a new situation is justified by its similarity to the situation to which the term had been applied at the previous instance(s) of its use. The paradigm-based approach is a kind of analogy-based reasoning. The problem with the definition-based approach is that its interpretation of language detaches itself from experience, and thereby detaches itself from the real world. The approach focuses on the criteria selected as a result of experience. Those criteria become autonomous from the experience, they live their own lives. As we have seen in the kung-fu master example, the definition of a “dangerous tool”, including the general terms like “implement” does not allow hands to be qualified as dangerous tools. And yet we believe that qualifying hands as dangerous tools makes sense. In order to treat the hands of a kung-fu master as dangerous tools, one needs to go beyond the definition, to the real context of the previous uses of the term “dangerous tool”. By analysing the contexts of the previous uses of the term, one can find the similarity between those uses and the target use, even if the target situation does not fulfil some of the criteria listed in the definition (hands are not implements). By eliminating the use of definitions as intermediaries to interpretation, the paradigm-based approach avoids the reductionism of its competitor. While the latter relies on a limited number of features selected by the definition, the former makes use of the full set of features covered by past uses of a term. Instead of limiting the number of interpretive premises, the paradigm-based approach expands their number significantly. The crucial term for the paradigm-based approach is “similarity”: the key interpretative decision involves finding similarities between target and source situations. But what does it mean for these two situations to be similar? It means they share a feature which led previous users to apply the term in a particular case. In our example of a kung-fu master, the interpreter decides to apply the term “dangerous tool” to the target situation because she realises that the reason for using the word previously was that dangerous tools are things that can harm people and force them to succumb to violence; the fact of tools being implements was not relevant. The paradigm-based approach is a similar approach to language use as the one proposed by Millikan in her theory of linguistic lineages. As we recall from the previous parts of the article, it is exactly the existence of a feature that leads users of a language to apply a term again and again to particular situations which creates a lineage of a term’s 20 draft for discussion purposes (29/08/2016) uses. The recurrence of this feature has induced community of speakers to reuse the same term to signal that feature’s existence, and this reuse of the term has endowed the term with a proper function of signalling that feature’s occurrence. Because the theoretical analysis of language may seem too abstract for lawyers involved in practical, day-to-day cases, in the next section two real cases are analysed to show the explanatory power of the theoretical framework proposed by the theory of lineages and the paradigm-based approach. 6. HOW THE PARADIGM-BASED APPROACH WORKS IN PRACTICE: TWO CASES The purpose of discussing these cases is twofold. First, I aim to show the advantage of the paradigm-based, historical analysis of language practice over the definition-based analysis in real-life cases. Second, I argue that cases solved with the formalistic, definition-based approach result in incongruity: a feeling on the part of the law’s addressee that the solutions are strange and surprising. Lawyers’ reaction to such incongruity is very often to quote a Latin proverb “Dura lex sed lex” – “Harsh law but it is the law”. My conclusion is that the “harshness” of the law arising from the formalistic rule is an adverse effect of taking a non-natural, definition-based approach to language, not a genuine feature of the law per se. When that interpretative approach is replaced with a paradigm-based, externalist one, the incongruity disappears, and with it the sense of harshness. If one analyses language as a historically developed, context-sensitive practice, the judicial decision to apply statutory terms according to their ordinary meaning does not have to result in surprise. To the contrary, the decision can be at once commonsensical and legally justified. Nix v. Hedden The first case illustrating how the lineages work is Nix v. Hedden. In this “tomatoas-fruit case”, as it is known, the US Supreme Court faced the quandary of how to qualify tomatoes in the light of tax tariffs that differ with regard to fruit and vegetables. If tomatoes were qualified as fruit (as they are from a botanical perspective) they would be subject to lower tax than when qualified as vegetable (which they are from a culinary 21 draft for discussion purposes (29/08/2016) perspective). The latter approach was that taken by current practice. The case triggered a wave of legal and general discussions on how to solve this puzzle and it still causes disagreement. When taken from the externalist perspective, however, Nix v. Hedden nicely depicts the usefulness of the paradigm-based approach and the concept of lineages. The issue at the heart of Nix-Hedden can be best observed through the prism of lineages. The dilemma the judges faced was based on the existence of two lineages of using the word “fruit”. The first one, “culinary”, is a tradition of using “fruit” within culinary contexts. The instances within the culinary lineages include situations in which previous users applied “fruit” with a proper function which has been to cover the properties of things like cherries, strawberries or avocados. These properties include typically (but not exclusively): the property of being sweet or the property of being elements of desserts. Tomatoes, which from this perspective are rather similar to cucumbers, onions and pumpkins, have not been covered by the word “fruit” in this lineage. The other lineage, the “botanical” one, is a tradition of using the word “fruit” in a different, specialised context of botany, in which technically fruit is a part of a plant that forms itself from a flower and contains seeds. Within this context the proper function of “fruit” has been to cover the property of being a particular stage in the development of plants. The crucial legal question in Nix v. Hedden was within which lineage the term “fruit” had been used in the relevant tax legislation, and, consequently, if tomato should be covered by this term. It is quite clear that tomato had been covered by the word “fruit” within the “botanical” lineage (a tomato is technically a fruit because it develops from a flower and contains seeds), and not covered by the word “fruit” within the “culinary” lineage (within the latter it has been covered by the term “vegetable”). The Court rightly decided that the use of “fruit” in the tax legislation was an instance of the culinary lineage, as it is a more general one and there no indications that the term “fruit” was used in the narrow, botanical lineage. After all, one taxes objects with an eye how they are used in regular life, not how they develop biologically67. In terms of semantic externalism the two lineages allow for two theories explaining on how 67 Excise tax, for instance, is widely use to burden the sale of luxury products, like perfumes, or products (like tobacco and alcohol) whose use the government discourages. On the other hand, sometimes lower rates of VAT are introduced for products whose use is supported by the government (e.g. books or newspapers). 22 draft for discussion purposes (29/08/2016) the word “fruit” has been used. Each theory presents an explanation by finding a common denominator for a particular practice of using the word. The co-existence of the two lineages and two theories caused the ambiguity the Court faced in Nix v. Hedden. As the case shows, the task of interpretation is to trace the lineages and to decide in which one the interpreted term has been used. Approaching the Nix v. Hedden case via Recanati’s terminology, the court’s application of the word “fruit” to tomatoes is a target situation. To assess the appropriateness of this application, one needs to trace back the source situations in which the word “fruit” has been previously used. The semantic potential, shaped by the history of using the word “fruit” includes both the source situation in which the feature of sweetness and the feature of “being-developed-from-the flower” occurred. As the features have different functions for us, however, two functional lineages of using the word “fruit” exist: a culinary and a botanical one. Those lineages are sets of source situations in which different features with different functions have been chosen by speakers68. As we have seen, legal interpretation within the externalist paradigm consists in deciding whether the target situation (in the Nix v Hedden case, exporting tomatoes) sufficiently and in relevant aspects resembles the source situation, which was the use of the term in the legal text (“exporting fruit” in the tax legislation). The court decided that the lawmaker used the word “fruit” in the tax legislation to invoke the series of the previous source situations in which the proper function of the word “fruit” was to signal the occurrence of the features connected to taste (like sweetness) and culinary use (“dessertness”). As the target situation (tomato) does not have those features of sweetness and dessertness, the tomato cannot be named “fruit”. The judges in Nix v.. Hedden had then to take two decisions: which set of source situations to choose and whether the target situation resembles the source situations within the chosen set. The former decision was crucial: the judges decided that the original use of the word “fruit” in the tax tariff was an instance of a “culinary” use, not a “botanical” one. By doing so, they determined that the right set of source situations with 68 A lineage is a kind of consistent linguistic practice of naming the situations in the world with the same words if those situations include the same relevant features. Polysemy occurs when a word is used in several lineages, and is thus applied to several sets of situations. Some polysemies are obvious: like the word “bank” in English, with its two clearly separated lineages: the financial one and the one applied to parts of rivers. The case of fruit is a case of polysemy, as well, albeit not so obvious 23 draft for discussion purposes (29/08/2016) which to compare the target situation is the culinary one, not the botanical one. Once chosen, the culinary set of source situations provided the judges with a clear set of features, like sweetness and “dessert-ness” that cannot be found in the target situation (tomato), so the application of “fruit” to “tomato” was found to be incorrect. Interpretive controversies like that in Nix v. Hedden cannot be resolved with the use of definitions alone, and the Court was well aware of that69. Like “fruit”, other legal and non-legal terms acquire their meaning not from dictionaries but from the practice of using them in particular contexts. As with the tomato-as-fruit example, every legal interpretation has to look into the past to discover the proper function of terms and phrases. The “normal” or “typical” meaning is not a “plain meaning” but the typical, proper function the terms and phrases acquired by being used in particular contexts in the past. The problem with formalism is conspicuous when one imagines a different decision in Nix v. Hedden, i.e. to decrease the tax rate for tomatoes because they are technically (i.e. according to the definition) fruit. Such a decision would appear incongruous, given that the practice had been to treat tomatoes as vegetables for tax purposes. What is the source of this incongruity? In practical terms, it is that the proposed use of a word (i.e. tomato as fruit) is not typical but idiosyncratic. In technical terms, the proposed use does not fit in with the word’s lineage: tomato is not a fruit in the most common tradition of using this word. The tomato-as-fruit case provides an example of overlooking the distinctions between lineages – or ignoring the very existence of those lineages. A more common name for such practices is “dictionary shopping”, where definitions are chosen for the sake of expediency. It is easier to cherrypick a definition favouring one’s preferred outcome than to analyse whether that definition is aligned with previous uses of the word in question. If the lawmaker uses a term in one lineage and the judge uses the same term in a different lineage, it is clear that confusion will arise and incongruity will follow: treating 69 Hence, in Nix v. Hedden the Court openly diminished the role of definitions in understanding legal text: “There being no evidence that the words "fruit" and "vegetables" have acquired any special meaning in trade or commerce, they must receive their ordinary meaning. Of that meaning the court is bound to take judicial notice, as it does in regard to all words in our own tongue, and upon such a question dictionaries are admitted not as evidence, but only as aids to the memory and understanding of the court.” (emphasis added). 24 draft for discussion purposes (29/08/2016) tomatoes as fruit would have been surprising. It would also have been unpredictable. And yet, predictability is one of the most salient features of the rule of law. Thus it appears that judicial formalism, with its tendency to apply dictionary definitions artificially and mix the lineages actually endangers the rule of law by endangering the predictability of judicial decisions. Smith v. United States As the analysis of Nix v. Hedden has shown, the source of incongruity in judicial interpretation can be an insensitivity as to which lineage the interpreted term belongs. It is, in other words, an insensitivity to the context in which words and phrases are used in linguistic practice. Definitions are generalisations and as such they may be applied in detachment from the contexts in which the defined words have been formed. This phenomenon is conspicuous in another court case, Smith v. United States, in which the judges interpreted the phrase “to use a firearm” and considered applying it to the situation in which the perpetrator exchanges a bag with a firearm in it for drugs. An interpretive decision that exchanging the bag with a gun constituted an instance of “using a firearm” would make it possible to impose a harsher punishment on the perpetrator. Unlike Nix v. Hedden, the Court in Smith intuitively reached for a dictionary definition of the verb “use”, instead of focusing on typical applications of the phrase70. The problem is that, within the theory of lineages, it is not words but phrases and sentences that refer to the world. The reason for this is the nature of language as applied to reality: sentences correlate with situations, and particular words, which correlate with elements of situations, can be the building blocks of many sentences and thus appear in many lineages. Situation A, in which “a firearm” is used, and situation B, in which anything else is used, can be contextually completely different, yet the verb “to use” appears in both. Therefore, understanding “to use” according to its dictionary definition, i.e. as abstracted from those completely different situations, is not justified 70 The Court explicitly cited the following definitions of “use”: “ Webster's defines "to use" as "[t]o convert to one's service" or "to employ." Webster's New International Dictionary 2806 (2d ed. 1939). Black's Law Dictionary contains a similar definition: "[t]o make use of; to convert to one's service; to employ; to avail oneself of; to utilise; to carry out a purpose or action by means of." Black's Law Dictionary 1541 (6th ed. 1990).” (…) Petitioner's handling of the MAC-10 in this case falls squarely within those definitions.” 25 draft for discussion purposes (29/08/2016) within the theory of lineages. In the case, however, the judges applied a general definition of “to use”, detaching it completely from the typical situations within which the phrase “to use a firearm” is generally employed. The dictionary definition of “to use” is a distillation of all the lineages including this word, and as such is multifarious. The theory of lineages requires a focus on particular instances of using language, and there exist no particular instances of using “to use”. In other words, with one exception, there are no particular situations in which one simply “is using”.71 However, there are particular instances of “using a firearm” that form lineages entirely distinct from others such as “using a bag” or “using a car”. If we analyse the lineages of particular words making up the phrase “using a firearm”, the number of possible lineages for each word grows significantly72. This can be seen especially in the case of the verb “to use” taken separately, not as a part of the phrase. The number of meanings of “to use” is enormous. As in its transitive form “use” is always accompanied by a noun, the source situations of “to use” must be taken separately, but with their accompanying noun. In the “using-a-firearm” case, judges analysed the word “using” separately its accompanying noun. The semantic potential of “using” is much wider, as it encompasses all the source situations in which “to use” occurred, with all the possible nouns it can occur with. As nouns modify the meaning of verbs with which they occur73, each use of the verb “to use” with a different noun creates 71 This one exception is a colloquial and intransitive use of the verb and represents a separate lineage; e.g. “My brother is using”. It can be found here: http://www.urbandictionary.com/define.php?term=using 72 That the definition-based approach of the court is too broad is the main topic in Judge Scalia’s dissent to Smith v. New York: “The Court begins its analysis by focusing upon the word "use" in this passage, and explaining that the dictionary definitions of that word are very broad.” (…). “It is, however, a "fundamental principle of statutory construction (and, indeed, of language itself) that the meaning of a word cannot be determined in isolation, but must be drawn from the context in which it is used. (…) That is particularly true of a word as elastic as "use," whose meanings range all the way from "to partake of" (as in "he uses tobacco") to "to be wont or accustomed" (as in "he used to smoke tobacco").” 73 Recanati (2004). As the author points out, drawing on Pelczar’s work:, “the word ‘get’ takes on different senses – denotes different relations – depending on what fills the second argument-place of the relation. Thus ‘Smith has got the virus’ means that he has contracted the virus, while in ‘I need to get some eggs’ ‘get’ has what Pelczar calls the acquisitional sense. (…) The same remarks can be made with respect to the word ‘cut’, whose meaning is modulated differently in ‘John cut the grass’ and ‘John cut the cake.’ As Searle points out, if I ask John to cut the grass, and he cuts it the way one cuts a cake (or if I ask him to cut the cake, and he cuts it the way one cuts grass), he has not complied with my request; he has not done what I asked him to do”. (p. 133). A similar remark is made by Judge Scalia in his dissent: “Just as adding the direct object "a firearm" to the verb "use" narrows the meaning of that verb (it can no longer mean "partake of"), so also adding the modifier "in the offense of transferring, selling, or transporting firearms" to the phrase "use a firearm" expands the meaning of that phrase (it then includes, as it previously would not, nonweapon use)” 26 draft for discussion purposes (29/08/2016) a different lineage. Thus, “to use a firearm” has a different lineage than “to use a valuable thing as a means of payment”. The whole phrase “to use a firearm” has a history of being used in the typical situations in which a person has used a gun for shooting or threatening. Those have been the situations in which one have been using a firearm in a typical way, according to the purpose for which guns are manufactured. Naturally, there must have been situations in the past in which one used a gun for a strange, atypical purpose: to hammer a nail or to open a beer. It seems unlikely, however, to assume that in those atypical situations the ordinary meaning of the phrase “to use a firearm” was to indicate the possibility of using it in that atypical way. The possibility of that atypical use must have been signalled differently from a typical source situation in which one uses a firearm. No one expects a person to shoot a nail or a bottle in those situations, so additional linguistic information must have been provided, either explicitly or implicitly, for allowing the hearer to deduce that a non- typical use is being proposed. It could have happened by adding a phrase “using the firearm” (“use a firearm AS A HAMMER”, “use a firearm AS AN OPENER”) or by a behaviour which was communicatively obvious in a given context (e.g. by handing over a non-loaded gun to a person struggling to open a bottle). In those atypical cases, “using a firearm” has had another proper function, and hence a different meaning. As a consequence, those atypical situations constituted a different lineage of different linguistic expressions (e.g. “use a firearm AS AN OPENER”). Finally, it can be stated that the lineage of the phrase “using a firearm as a firearm” is stable and clearly refers to using a firearm in a typical way. Interpreting it in a context-insensitive way, as the Court in Smith v. New York did, is an interpretative mistake. It is strange to claim, as formalists do, that a plain or ordinary meaning of a particular phrase consists not in how this phrase is typically used, but in the sum of all possible uses of all phrases in which the interpreted word has been used (e.g. understand “to use” as it has been used in any phrase including “use”, and not as it has been used in the phrase “to use a firearm”). The definition of “to use” is too broad to be meaningful – it is a generalisation that is simply too general. 27 draft for discussion purposes (29/08/2016) The interpretative decision in Smith v. New York is surprising, as is rightly stressed in Judge Scalia’s dissent74. The reason is that the Court interpreted the phrase “to use the firearm” in an unnatural way, i.e. in a way which was insensitive to the natural context of this phrase75. The incongruity of the decision results from the acontextual approach taken by the Court to the ordinary meaning of this crucial phrase and this in turn was due to the Court assuming a definition-based, not a paradigm-based, approach to language. Incongruity of the definition-based approach Both cases discussed above demonstrate how easily a judge may assume an interpretation that is completely surprising and incongruous. The incongruity results from a simple fact that the application of a word or phrase to a new situation (the target situation) seems unjustified, as the target situation is not similar in a relevant aspect to previous situations to which the word or phrase at hand has been applied. Applying the word “fruit” to tomatoes is incongruous, because tomatoes, as they are used for culinary purposes, do not resemble the typical objects we call fruit: cherries or avocados. Claiming that handing over a bag with a gun in it is using a firearm seems strange, as this situation does not resemble a typical situation we have been calling “using a firearm” in the past (like shooting or threatening with a gun). The incongruity of the formalist approach is a fatal flaw because it puts the predictability the rule of law requires out of reach. Legal language is a branch of common language and as such should be used and interpreted in a way consistent with 74 Scalia: “To use an instrumentality ordinarily means to use it for its intended purpose. When someone asks, "Do you use a cane?," he is not inquiring whether you have your grandfather's silver-handled walking stick on display in the hall; he wants to know whether you walk with a cane. Similarly, to speak of "using a firearm" is to speak of using it for its distinctive purpose, i.e., as a weapon. To be sure, "one can use a firearm in a number of ways," ante, at 7, including as an article of exchange, just as one can "use" a cane as a hall decoration - but that is not the ordinary meaning of "using" the one or the other. 1The Court does not appear to grasp the distinction between how a word can be used and how it ordinarily is used. (…) It is unquestionably not reasonable and normal, I think, to say simply "do not use firearms" when one means to prohibit selling or scratching with them.” (…) “The Court cannot avoid "anomalous applications" when it applies its anomalous meaning of "use a firearm" in other contexts” 75 Scalia: “We are dealing here not with a technical word or an "artfully defined" legal term, [508 U.S. 223, 245] cf. Dewsnup v. Timm, 502 U.S. 410, 423 (1992) (SCALIA, J., dissenting), but with common words that are, as I have suggested, inordinately sensitive to context” 28 draft for discussion purposes (29/08/2016) common language – i.e. in a contextualised, lineage-based way. If it is not, those who are governed by the law will be surprised how their understanding of language differs from that of judges. To minimise the divergence between legal and common language, legal interpretation needs to abandon the artificial definition-based approach to meaning, which can give rise to incongruity in judicial decisions and unpredictability in the application of law. Unpredictable judicial decisions cannot promote the rule of law. If one is to treat seriously the formalists’ argument on the importance of sticking to language and language alone in legal interpretation, one must ask what concept of language they are talking about. Should we stick to the artificial language of definitions or to the real language of everyday use? If legal interpretation is to be predictable, it must be sensitive to the history and function of the terms we use, just as everyday use is. An outline of such an approach to legal interpretation is proposed in the final part of this paper. 7. A NON-FORMALISTIC APPROACH TO LEGAL INTERPRETATION As we have seen, semantic externalism posits an anti-dictionary thesis: the definitions we encounter in dictionaries are to help us to imagine typical contexts in which a word is used, but they cannot substitute for those contexts. According to semantic externalism, a plain meaning is not a dictionary meaning, but rather one resulting from the “normal” use of a phrase or a sentence, not a particular word. To identify this meaning, one needs to trace a lineage of this phrase or sentence use, and this requires looking into past, not simply into a dictionary. This process is inductive, not deductive. It is not definition but paradigm-based. Normal use consists of using a phrase or a sentence in a typical context, i.e. in the presence of particular facts, things and people. Let us look from the perspective of the theory of lineages at how legal interpretation proceeds within the paradigm-based approach. As noted earlier, the crucial interpretive decision is to find relevant similarities between the target situation (the new situation in which the term is to be used) and the source situations (the previous situations of the same term’s use, the members of the lineage). Is the target situation involving “tomatoes” similar to previous situations involving “fruit”? Is the target situation of bartering the gun for drugs similar to the previous situations in which 29 draft for discussion purposes (29/08/2016) the phrase “using a firearm” has been uttered? These are the questions the interpreter needs to answer. According to the paradigm-based approach, the proper application of a legal term to a new situation requires two decisions to be made: (1) deciding in which lineage the term at hand has been used in the legal text, and (2) deciding whether the target situation resembles the source situations which form the lineage. Let me take each decision in turn, as they are usually taken in legal interpretation. Take for example the word “bank”. If one is to decide if the target situation is suitable for using the word “bank”, one must first choose between its two main lineages: the financial one and the geological one. A judge prohibiting construction of an additional floor on a bank building is surely mistaken if he or she applies to that situation a provision from environmental law prohibiting any construction to be carried out on a bank. The root of the mistake is clear: one cannot compare the target situation with a source situations taken from a different lineage. A similar mistake could have been made in the tomato-as-fruit case – namely to qualify tomatoes as fruit even if the term “fruit” has been used by the lawmaker within the culinary lineage, not the botanical one. The mistake was made in the using-a-firearm case by interpreting the word “using” too broadly, as it is understood in lineages other than that of “using a firearm”. The identification of the correct lineage requires first an analysis of how the subject term was used by the lawmaker. How do we establish to which lineage the lawmaker’s use belonged? One of the best tools to do so is to analyse the context of that use, namely the legislative history. This is comprised of utterances referring to the legal act contained in legislative texts, reports of parliamentary committees, protocols of plenary sessions, etc. All these utterances constitute concurrent uses of terms that are the same as or similar to terms occurring in the text of law. All these utterances concern the scope of matters regulated in the act, social issues to be remedied by the legal act, and so on. By analysing these utterances, an interpreter can establish what the problem was that the lawmaker attempted to solve by enacting the law, and what were the specific topics the lawmaker addressed in the parliamentary discussions. In other words, the analysis of the legislative history helps the interpreter to identify the elements of reality to which the lawmaker referred to in the legal text, and by doing so establish the possible lineages within which the interpreted term was used. The use of the term “bank” within a parliamentary discussion on water resources may 30 draft for discussion purposes (29/08/2016) not be decisive in ruling out the financial lineage, but a deeper analysis of the legislative context can be a key factor in determining the appropriate lineage. Similarly, the analysis of the legislative history in the tomato-as-fruit case reveals that in its original context the term “fruit” referred to culinary situations rather than botanical ones. Judicial formalism, especially its textualist branch, generally refuses to use legislative history in legal interpretation76. It claims that analysis such as that described above would go beyond the language of the statutory text, which it perceives as complete and inviolable. This claim is misguided: if the analysis of legislative history goes beyond anything, it goes beyond the semantics of the text, but only to arrive at its pragmatics, i.e. to the real context in which the language is used. Semantic externalism perceives this pragmatics as a fully legitimate part of linguistic analysis of the statutory text. In other words, if meaning is perceived as having a close relation to the reality to which it refers, analysing that reality cannot be treated as non-linguistic. Legislative history is the real context in which the lawmaker used the term the interpreter is to apply; formalism’s proposal to remove legislative history from the set of legitimate sources of knowledge about language use seems to be an unjustified reductionism of interpretative premises. Once one is sure that both the instance of using the legal term by the lawmaker and the planned use of that term by the interpreter (i.e. its application to the case at hand) will be instances of the same lineage, the second interpretative decision can be taken. It resides in finding similarity between the target situation and the source situations. Traditionally, the key source situation to which the target situation is compared is the lawmaker’s use of the subject term. However, as Recanati pointed out, the semantic potential of a term is shaped by all the source situations in which the term has been used. Thus the theory of lineages encourages interpreters to expand the number of source situations and to include other instances of using the term, including those precedent, contemporaneous and subsequent to that of the lawmaker. Those instances include, for example, other court verdicts in which the term was applied to similar facts, and the uses of the term which took place in legal literature, both before and after the lawmaker’s use. The lineage is a chain of all public uses of a term and, as I indicated earlier, the lawmaker’s use is by no means definitive for the purposes of legal interpretation. 76 See Nelson (2005), p. 361 31 draft for discussion purposes (29/08/2016) The argumentation for expanding the number of interpretive premises by including other court verdicts and the opinions expressed in the legal literature seems convincing within the theoretical framework of semantic externalism. Yet it is not accepted by the formalists77. From the formalist perspective, such expansion may be criticised in those jurisdictions in which court decisions are not formal sources of law (for example in the continental legal systems). As to the opinions expressed in legal literature, despite their wide use in legal argumentation, the last formal recognition of the communis opinio doctorum within the legal system took place in ancient Rome and the first German Reich. The intention of this paper is not to argue for a return to those days but to admit that the analysis of legal language cannot exclude other and related uses of legal terms. These latter constitute legitimate comparative material for legal interpretation if we agree that the meaning of the language is produced by living linguistic practice rather than fossilised dictionary definitions. As we have seen, the dispute between judicial formalism and a non-formalistic approach to language again revolves around the issue of reductionism of interpretive premises: where formalism wants to minimise the number of these premises, the counter-proposal wants to increase it, by embracing legislative history, similar court verdicts and doctrinal opinions as sources of linguistic information. The formalist, definition-based approach tends to limit the number of features that enter the definition and, as a result, are considered while applying that definition to a new case. In opposition to that, the paradigm-based approach, by resigning from the intermediary function of the definition, allows the interpreter to assess the similarity between the target and the source situations on the basis of the whole set of features present in the source situations, i.e. on the basis of the whole semantic potential of the term. The increase of features present in the source situations that are considered by the interpreter is another anti-reductionist characteristic of the approach to language proposed in this paper. The non-reductionist thesis that I am defending concerns not only the quantity of the interpretative premises but also their quality. The paradigm-based approach to interpretation allows the interpreter to consider both the formal and the functional similarities between previous and current instances of language use. As the “dangerous tool” example shows, the hands of a kung-fu master are dangerous tools within the 77 Schauer and Wise (2000), passim 32 draft for discussion purposes (29/08/2016) meaning of the Criminal Code, because they are dangerous in a functional sense, even if they lack some physical features present in the previous source situations (like being an implement). This functional approach to interpretation is absent in judicial formalism; indeed, functional and purposive interpretations are methods formalists criticise as nonlinguistic and prone to advancing judicial discretion78. Switching the perspective from semantic internalism and its criteria-based approach to meaning to semantic externalism, with its focus on the relationship between language and real features in the real world, reveals that judicial formalism’s reluctance to use functional and purposive considerations is entirely unjustified. Linguistic terms have a function, which is to pick up relevant features from reality. We use linguistic terms for the purpose of signalling that those features occur. No approach to interpretation that aspires to be an effective tool of decoding ordinary meaning can neglect those functional and purposive aspects of language. To summarise, legislative history, other judicial rulings, and views expressed in legal literature, as well as the functional and purposive aspects of legal language can all be treated as elements that help in analysing the meaning of legal terms. An interpretation guided by historical, comparative, or legal literature perspectives goes hand in glove with the rule of law, especially with the claim that provisions and rules are essentially of a linguistic nature. This, in turn, unseats formalism, along with its textual perspective, as the exclusive method of interpretation compatible with the rule of law. Critics of non-formalism who fear excessive discretionary powers being handed to lawyers or judges have nothing to fear. Paradoxically, the non-formalist approach to legal interpretation can ensure a better predictability of judicial decisions and a stricter curtailment of judicial discretion by broadening the scope of interpretive premises judges base their decisions on. Although it is not reductive, the theory presented here curtails interpretative discretion and secures predictability thanks to its reliance on lineages. To claim that a word has a particular meaning is to reason that it acquired that meaning as a result of past uses. Thus, an important element of the theory of lineages is deference to past uses that makes idiosyncratic uses unlikely79. In this sense enrooting a typical meaning of a 78 As Horwitz (1977, p. 254) argues, formalism equates to a refusal to admit that law has instrumental functions. Pildes (1999, p. 612) defines formalism as an adherence to the rules without consideration of their purpose (R. Pildes 1999, p. 612). 79 Rauti (2012), p. 1-2 33 draft for discussion purposes (29/08/2016) legal term in the tradition of its use has binding power, like the one depicted by Dworkin in his chain novel metaphor: our current uses of language are at the same time inspired and bound by previous uses80. The non-formalist approach to interpretation, derived from the externalist positions in the philosophy of language, sees language in a natural way: as a historically determined practical tool for co-ordinating human behaviour, not as a set of abstract, criterial definitions, artificially separated from the real world. The latter approach, i.e. that championed by legal formalists, can result in surprising interpretive results because it departs too far from the way people use their language. As such, it undermines the rule of law instead of promoting it. 8. CONCLUSION The American Realists’ critique mentioned at the start of this paper did not dispatch judicial formalism, nor did later ones; in fact, as Weinrib points out, formalism has been killed many times within the last two hundred years but still refuses to remain dead81. One reason that criticism proved unsuccessful is that it rarely attempted to cut formalism from its main life-support system: its purported compatibility with the rule of law. Formalists believe that theirs is the only theory of interpretation that secures the rule of law, a corollary of which is that any criticism of formalism endangers this founding ideal of Western legal systems. In order to be effective, any criticism of judicial formalism must start by decoupling formalism from the rule of law, as I have attempted to do in this paper. What was missing in the criticism of formalism was the argument that competing approaches to the application of law better pursue the principle of the rule of law than formalism itself because they are more faithful to legal language. In order to see this, one needs to analyse legal language from the position of semantic externalism. By revealing a more sophisticated nature of legal language, this new position makes clear that the interpretation of legal texts need not be reductionist. To the contrary, legal interpretation must make use of the richness of our linguistic history and practice if it is to do justice to the language used by the lawmaker. To be interpreted faithfully, legal 80 Dworkin (1986) 81 Weinrib (1988), p. 951 34 draft for discussion purposes (29/08/2016) language does not actually require a fully all-things-considered approach, but it cannot be restricted to a merely formalistic one. A moderately non-formalistic, non-reductionist approach to legal interpretation seems more justified. BIBLIOGRAPHY Adler, A.N. (1998) Can Formalism Convey Justice? - Oaths, 'Deeds' and Other Legal Speech Acts in Four English Renaissance Plays, “St. John's L. Rev.” 72, p. 248. Barak, Aharon (2005) Purposive Interpretation in Law. Princeton and Oxford: Princeton University Press. Bix, Brian (2004) A Dictionary of Legal Theory, Oxford University Press Bix, Brian (2003) “Can Theories of Meaning and Reference Solve the Problem of Legal Determinacy.” Ratio Juris 16[3]: 281-295. Brink, David O. (1989) Semantics and Legal Interpretation. Further Thoughts, Canadian Journal of Law and Jurisprudence, Vol. II, No. II, July. Burge, Tyler (1979) Individualism and the Mental. Midwest Studies in Philosophy 4: 73–122. Burton, S. (2004) Judging in Good Faith, Cambridge University Press, Cambridge 114, p. 141. Cicero, Marcus Tullius (2008) On Obligations: De Officiis. Translated by Patrick Gerald Walsh. Oxford: Oxford University Press. Cox P.N. (2003) An Interpretation and (Partial) Defense of Legal Formalism, “Indiana Law Review", 36. Craig, P. R. Hodder-Williams (1996), Judges and Politics in the Contemporary Age, Bowerdean, London Craig, P. (2003) Constitutional Foundations, the Rule of Law and Supremacy, “Public Law” 2003, SPR D’Amato, Anthony (1991) “Counterintuitive Consequences of ‘Plain Meaning’.” Arizona Law Review 33: 529-579. Devitt, Michael (1981) Designation. New York: Columbia University Press. 35 draft for discussion purposes (29/08/2016) Duxbury, N. (1995) Patterns of American Jurisprudence, Clarendon Press, Oxford Dworkin, Ronald Myles (1986) Law’s Empire. New York: Harvard University Press. Eskridge William Jr. (1990) “The New Textualism.” UCLA Law Review 37: 621-691. Eskridge, William Jr. and Philip Frickey (1990) “Statutory Interpretation as Practical Reasoning.” Stanford Law Review 43: 321-384. Evans, Gareth (1996) “The Causal Theory of Names.” In The Philosophy of Language, Aloysius P. Martinich (ed.), 271-83. Oxford, New York: Oxford University Press. Fish, Stanley (2005) There Is No Textualist Position, “San Diego L. Rev.”, 42, p. 629. Grey, Thomas, (1999) The New Formalism, “Stanford Law School Working Papers” Harris, Roy, Hutton, Christopher, (2007), Definition in Theory and Practice: Language, Lexicography and the Law, Bloomsbury, London-New York. Hart, Herbert (1961) The Concept of Law, Clarendon Law Series Horwitz, M. J. (1977) The Transformation of American Law 1780-1860, Harvard University Press, Cambridge, Mass. Hughes, Christopher (2006) Kripke, Name, Necessity and Identity. Oxford: Clarendon Press. Huhn, W. (2003) The Stages of Legal Reasoning: Formalism, Analogy and Realism, “Viii. L. Rev.” 2003, 48, p. 309 Hutton, Christopher (2014) Word Meaning and Legal Interpretation: An Introductory Guide, Palgrave Macmillan Jackson, Franck (1998) From Metaphysics to Ethics: A Defence of Conceptual Analysis Oxford University Press Kallestrup, Jesper (2012), Semantic Externalism, Routledge Kennedy, Duncan (1973), Legal Formality, “J. Legal Stud.”, 2 Kennedy, Duncan (1976) Form and Substance in Private Law Adjudication, “Harv. L. Rev.”1976, 89 Kozak, Artur (2002) Granice prawniczej władzy dyskrecjonalnej, Kolonia Limited, Wrocław 36 draft for discussion purposes (29/08/2016) Kripke, Saul (1979) “Speaker’s Reference and Semantic Reference” In Contemporary Perspectives in the Philosophy of Language, ed. Peter French, Theodore Uehling and Howard Wettstein. Minneapolis: University of Minnesota Press. Kripke, Saul (1991) Naming and Necessity. Oxford: Basil Blackwell. Llewellyn, K.N. (1962) Jurisprudence: Realism in Theory and Practice, University of Chicago Press, Chicago Mallon, Ron (2015), Social Construction and Achieving Reference NOUS 00:0 (2015) 1–19 Manning, John F. (2003) The Absurdity Doctrine, 116 Harv. L. Rev. 2387, 2408, p. 2393 Michelman, F.I. (1999) A Brief Anatomy of Adjudicative Rule Formalism, “U. Chi. L. Rev.” 1999, t. 66, nr 3 Millikan, Ruth G., (1984) Language, Thought and Other Biological Categories, MIT Millikan, Ruth G., (2005) Language: A Biological Model, Oxford University Press Nelson, Celeb (2005) What is Textualism?, Virginia Law Review, Vol. 91: 347 Pildes, Richard H., (1999) Forms of Formalism, „University of Chicago Law Review" 66. Pound, Roscoe, (1908) Mechanical Jurisprudence, “ColLR” 8. Putnam, Hilary (1975) “The Meaning of Meaning.” In: Mind, Language and Reality. Cambridge: Cambridge University Press. Rauti, Antonio Multiple Groundings and Deference The Philosophical Quarterly Vol. 62, No. 246, January 2012 Recanati, Francois (2004) Literal Meaning, Cambridge University Press, Cambridge Ross, Stephen (1995) “The Limited Relevance of Plain Meaning.” Washington University Law Quarterly 73: 1057-1067. Scalia, Antonin (1997) A Matter of Interpretation, Princeton University Press Schauer, Frederic (1988) Formalism, 97 Yale L.J. 509 Schauer, Frederic (1991) Playing by the Rules. A Philosophical Examination of RuleBased Decision-Making in Law and in Life. Oxford: Oxford University Press. 37 draft for discussion purposes (29/08/2016) Schauer, Frederic (2008) Formalism: Legal, Constitutional, Judicial, in: The Oxford Handbook of Law and Politics, ed. Whittington, Kelemen, Caldeira, pp. 428-436 Schauer, F., Wise V. J,, (2000) Nonlegal Information and the Delegalization of Law, Journal of Legal Studies, Vol. 29, No. 1, Pt. 2, January Solum, Lawrence (2008) Semantic Originalism, Illinois Public Law Research Paper No. 0724 Sunstein, Cass. R. (1999) Must Formalism Be Defended Empirically, „University of Chicago Law Review", 66 Stavropoulos, Nicos (1996) Objectivity in Law. Oxford: Clarendon Press. Tushnet, Mark (1983) “Following the Rules Laid Down: A Critique of Interpretivism and Neutral Principles”. Harvard Law Review 96: 781-824. Tushnet, Mark (1985) Anti-Formalism in Recent Constitutional Theory, “Mich. L. Rev.” 83 Weinrib, Ernest (1988) Legal Formalism: On the Immanent Rationality of Law, „Yale L.J.”, 97 Wittgenstein, Ludwig, (1958) Philosophical Investigations, (G. Anscombe trans. 3d ed.) 38