Legal Statute and Integrational Linguistics: Determining the Meaning of Theft

A paper delivered at the Sociolinguistics Symposium, Bristol UK, April 2000.

1. Socially-embedded linguistics.

In his programme note introducing this colloquium, Roy Harris draws attention to a number of key points: that from the outset of modern linguistics, in Saussure’s concept of langue, a language was assumed to be supported by a corresponding social entity or ‘linguistic community’, but at the same there is no word of encouragement in the Cours for the sociological analysis that this might entail. If Saussure himself was not explicitly hostile to a sociologically-embedded linguistics, his mainstream linguistic descendants by and large surely are, culminating in modern-day “autonomous linguistics”. By contrast with this, everybody speaking at this conference surely subscribes to some form or other of non-autonomous linguistics, a socially-embedded linguistics.

This much is probably common ground for all at this conference. Where differences no doubt emerge is in how best to re-conceptualize the embeddedness of linguistic activity in general human activity, in society, context, situation, and in how best to re-analyze language, communication, culture and society in light of that reconceptualization. Integrational linguistics, or integrational studies of communication, have gained some minor notoriety for being one of the more radical, hyper-critical, and allegedly negative and unconstructive of such revisionist approaches. Those who have looked at integrationist writings and have turned away unpersuaded seem to have decided that not only is the integrational approach extremely critical and deconstructive of standard linguistics, it is needlessly so. In my talk I want to try to show that the deconstruction or demythologization is not needless or gratuitous, but intrinsic to the logic and assumptions of integrationism. I will try to do so by looking at one area of social practice involving people, their language and their behaviour, and looking at it from an integrational point of view and invoking some of the key tenets of the approach.

The social practice I am interested in reviewing today is the working of the legal system and in particular recent and authoritative appeal court deliberations on the correct interpretation of the statutory language defining the crime of theft, as set out in the 1968 Theft Act. My comments will be a blend of recent case-law history, reflections on the assumptions about language and about statutory interpretation that seem to inform senior judges’ methods of argumentation and deliberation, and what–if any–light can be shed by integrationist linguistic thinking on this complex state of social linguistic affairs. But before I turn to the legal complexities, I want to at least mention, however briefly and reductively, some of the integrationist tenets that I take to be relevant to the issues that will subsequently arise.

Perhaps first of those tenets that requires mention is the assumption that, quite routinely, linguistic communication is marked by potential indeterminacy of both form and meaning, that is of both the marks or sounds used and of the sense to be made of them. That potential indeterminacy is, again routinely, locally or situationally resolved: on a case by case or moment by moment basis, particular forms and meanings are postulated as the signifying interactants judge a need for these, and this situated sense-making is firmly guided by the principle of cotemporality, the principle that verbal behaviour is deeply integrated with the nonverbal material with which it is interwoven. And here the verbal and nonverbal components of any interaction might be thought of as analogous not to figure and ground, but rather to the parts for instruments in a musical score. With those founding assumptions, it is hardly surprising that integrationism is deeply antipathetic to representation of a language as, even in idealized theory, a fixed code which in turn facilitates, as the essence of linguistic communication, a replication in the addressee’s head of the very same thought or idea that originated in the speaker’s head before being converted into spoken and heard signals. All of that, for integrationists, amounts to a seductive, compelling, and profoundly mistaken myth about language, which has dominated western linguistics for centuries.

If language is not a fixed faxed code, the irritated linguist asks, how on earth do we communicate with each other, achieve understanding, and develop all the complex practices that the species boasts, from going to the moon to developing microsurgery to inventing the internet to sustaining a jurisprudence around the crime of theft? The integrationist answer begins by insisting that, notwithstanding impressions to the contrary, language is constantly improvisatorily renewed by situated users, who are attuned to such renewal by their reflexive attention to their own behaviour. Purposes and transactions give rise to the particular communicative forms and strategies that people devise and remember and re-use, gradually turning them into norms, routines, conventions, and so-called standard forms. But the situated goals and transactions come first, so that these give rise to similarities and unpredictable differences in the communicative forms and meanings that are situationally ratified. We didn’t get to the moon or develop the internet by following norms, routines, and conventions.

So there is endless scope for linguistic change. At the same time, as is pervasively demonstrated, societies may wish to curb or constrain constant renewal–just as one might object to introduction of abrupt changes in constitutive or regulative rules governing a particular game in the course of playing that game. One of the cover terms used in integrational linguistics to denote the diverse ways in which a culture or subculture may strive to curb improvisation, expel indeterminacy, promote segregation, analysis, and stable nomenclaturism, is codification.

2. Codification.

It is striking to note how powerfully codification came to the fore, in the scholarly cultural life of Britain, in the latter half of the 19th century. In addition to Murray’s dictionary and Gray’s anatomy we might also note Leslie Stephen’s Dictionary of National Biography, and his uncle, Sir James Stephen’s earlier proposed codification of the English criminal law.

Codifications are, seen in a perhaps jaundiced light, a pretence: they pretend that the phenomena categorized and labelled are a stable and unchanging bedrock, as if the domain codified (and by implication the larger world too) comprised just one discrete thing after another, each with its own proper name, without overlaps or residue. At least so they appear when considered in their purest or simplest form; but even with the complexity that hierarchical relations introduce, where something can be codified as belonging at the same time to both a more specific and a broader category, the pretence tendency persists. For it is in the nature of a codification to project the impression that the material treated as if it were part of a code actually is a code. This is the well-known fate of all powerful codifications, from Gray’s Anatomy to the Oxford English Dictionary to Linnaean classification to Quirk et al’s Comprehensive Grammar of English. And in the case of legal codification, set out in accumulating case-law and statute, this is a fate not merely befallen but to a large degree actively sought. A statute such as the Theft Act of 1968 seeks to set out, in clearer and more stable terms than hitherto (for example, in terms that are clearer and more suited to the new times the Larceny Act of 1916) what the crime of theft is and what it is not, and what the criterial components of a theft are. That new act, like every volume of the Oxford dictionary, sends out two contradictory messages: one message is that here you will find a new improved revision and clarification of matters not up to now satisfactorily and authoritatively dealt with, the other is that here you will find the authoritative and definitive and standard description, so authoritative that it should not be thought of as a account but the account. Furthermore, the authority and standardness appears to be performed or enacted in the palpable existence of the Oxford dictionary in all its twenty-plus volumes, written down, a definitive will and testament, signed sealed and delivered. As Roy Harris has commented:

By actually making the stock of words visible in a durable form, the modern dictionary brings powerful support to the notion that linguistic signs have an existence of their own which is independent of that of their users. It also supports the notion that these signs are by nature invariant. They appear to have a fixed form and a fixed meaning, which remain valid regardless of the vicissitudes of day-to-day usage. In this respect dictionaries come to be regarded as having a function that is comparable to the preservation of a standard set of weights and measures. (Harris, 1998: 103)

(Incidentally, it will be interesting to see to what extent lay perceptions of dictionaries and fixity of word meaning, spellings and so on, of the permanence of codifications, will change as dictionaries and the consulting of them go ‘online’, with hotlinks to further references, more rapid updating or emendation of entries, and so on: words visible in not quite the same durable way as in the printed book…)

One of my interests in my topic today, then, is whether a modern criminal statute does something similar: does it articulate an illegal activity in a durable form as if it were a determinate and particular kind of act, captured in its essentials by stable and permanent terms? My answer will be that the statute, like any statute, seems to try to embody such stable permanence, but judges’ shifting interpretations of it belie that effort. [And why should this be a problem? What is the moral of tracking the tortured revisions in the interpretations of the codifications of the Theft Act, beyond recognizing that judges and Parliamentarians must tread carefully?]

3. Dictionaries and Statutes.

The analogy between the seemingly comprehensive monolingual dictionary, of the Oxford English variety, and the book of the law, particularly statute law, stems from the fact that in certain important respects a statute is fully analogous to a dictionary entry. Of both the Oxford Dictionary entry for the word theft, and the 1968 Theft Act, we can say that these purport to supply particular kinds of authoritative guidance, which can be relied upon, as to how the word and the crime, respectively, are to be understood, what they comprise, what they do not include, and so on. Of course there are differences, too: if the Theft Act is like an entry in the dictionary of the law, and to that extent is a proposal like any drafter’s entry, it is a proposal that those to whom it is addressed cannot decline or ignore, and only with effort and expense can those who are seemingly caught by the Act evade its powers to classify and shame.

Just as the monolingual dictionary is used to attempt to limit semantic indeterminacy, so the criminal statute or binding appellate judgement attempts to limit criminological indeterminacies. In Signs, Language & Communication Harris emphasizes that the lexicographer, just as much as everyone else, is involved in a communicational enterprise, making certain assumptions about who reads their dictionary and why, and shaping their dictionary entries accordingly (a similar point is made by Gee, 1999: 28-9, concerning the discourse scientists’ have devised to ‘manage uncertainty’). No lexicographer’s gloss is the semantic truth about the lemma it accompanies, but the lexicographer’s best effort in the given constrained circumstances to best meet the assumed needs of their assumed reader. And, crucially, Harris points out that the lexicographer proposes correlations, as distinct from reporting them; thus they are making signs rather than merely finding them, even if the sign-making is constrained by conventions and expectations.

Harris argues that a dictionary is a communicational process designed to cater for certain perceived communicational needs by means of proposing a reduction of certain kinds of semantic indeterminacy. (Thus a dictionary entry is like a directional sign pointing in the general direction of the Exit: it doesn’t take you to the Exit, and you certainly have to know how to interpret it, but if you do know how to interpret it then the sign helps exclude a lot of wrong directions while giving you a big clue as to the right direction; you’d rather have the sign than not and, besides, it’s hard to think of a situationally more efficient and relatively unobtrusive way of reducing, but not removing fully, uncertainty about how to get out.) If a dictionary purports to reduce certain semantic indeterminacies, something parallel is arguably effected by those who draft and pass a statute. And the judges who interpret and apply such a statute derive assistance from a text that, like a dictionary entry or a signpost, attempts to reduce certain most crucial indeterminacies without ever carrying a guarantee that all possible indeterminacies are therein resolved.

Additionally, Harris reminds that the lexicographer is no neutral go-between, anymore than the stockbroker is a neutral go-between operating between investors and companies: the lexicographer is part of the codifying culture, just as the stockbroker is part of the capitalist world they cater to. It is not in the interests of the lexicographer, for example, for the very idea and practice of lexicography and the use of monolingual dictionaries to be dismissed as if it were astrology or a pseudo-science. By the same token we might suggest that the legal professional is a far from neutral participant, rather a partner in the codifying legal culture, by means of which theft and appropriation and dishonesty and so on are characterized in the particular but shifting ways that they are. One implication of this is that socially-relevant codifications are invariably the work of one social group or another, with its own interests and partisan categorizations. And this goes as much for the law as it does for Scrabble. In Scrabble, where this is played with deference to a Scrabble dictionary, el for example is an English word, being the name for that letter of the alphabet that comes between k and m, even if non-Scrabble players are astounded by this claim.

Why have codifications? Roy Harris has remarked that

The social purpose of a codification is nothing other than to establish the recognition of allegedly ‘invariant’ units and processes of all kinds. (1996:104)

So codification has (as one purpose) the establishment of recognition of invariance. Each of these nominalizations involves a complex, contested effort–in other words, questions of politics, ethics, authority and responsibility are involved in each of these processes–as perhaps can be felt if each of them, as an intervention, is contrasted with its opposite. Thus establishment, which may involve creating and enforcing a pairing of words and things, and a further fixing of each word’s form and meaning, can be contrasted with absence of such efforts to create and enforce and render durable… Recognition, which surely involves publication of the codification, dissemination of the signs manual in ways which secure relevant users’ acknowledgement and even acceptance and adoption of the codification, can be contrasted with not publishing, not disseminating, not seeking and securing recognition of themselves as society’s Guardians or Lawmakers in their reserved area, as Gray and Murray and their descendants have done. And invariance or certainty is of course the enabling assumption of the entire codificatory enterprise. Either there simply is a right word (with a correct spelling and correct meaning) for most things in the world, and it is not merely more efficient but simply more truthful to take counsel on these facts–a ‘reliance on the truth’ view which, I suspect, is tacitly subscribed to a good deal of the time by language users. Or, more sophisticatedly, matching up words (fixed in form and meaning) with things in the world is not underwritten by an uncoveringor recovery of transcendant truth, but is much to be desired on pragmatic grounds of efficiency, reliability, reasonableness, and specificity. This might be called a ‘reliance on the codification’ view. The commonest knock-down justification for this approach is that it is in the interests of those who sponsor or subscribe to that codification–clearly a circular rationalisation.

And indeed codifications by their nature become self-justificatory: once you embark on a codification, it seems that the internal requirements of the process begin to take over, and begin to compel the codifiers to make their product as comprehensive, unambiguous, fixed, incontestable, and so on, as possible. [Just as no-one would prepare or publish a dictionary unperturbed by the possibility that half the entries were vague or error-ridden. And if the published codification does turn out to be blatantly deficient or unsatisfactory, the user tends not to draw the conclusion that the enterprise attempted an impossibility (It can’t be done); the more usual reaction is that thing (which can be done, and indeed has been in the instant case, after a sorry fashion) has been done badly.] I think we see this developing today where new areas of human activity are becoming subject to legal regulation–whether we look at procedures dealing with children charged with criminal offences, or the determination of liability and compensation in medical mistake and negligence cases, or the treatment of patients lacking mental capacity. In such sociolegal areas (and every area of legal codification is always also a social practice) there can be no turning back, nor should there be. But as the legal codifications proceed (with, ironically, their frequent revisions and remakings of what purported to be fixed, certain, and of general application), there is a permanent danger of the codification drawing on only its internal momentum and logic, and becoming less and less responsive to the society it purports to serve. Ironically, it was partly in response to this tendency, that certain terms in the old theft law, the Larceny Act of 1916, were displaced in the 1968 Act; both larceny and conversion were felt to be archaic and lawyerly terms, opaque to the general public, and were replaced by theft and appropriation, these being deemed (by many but by no means all commentators) to be equivalents, respectively.

Essentially, the legal system would like ‘theft’, ‘capacity’, ‘dishonesty’, ‘a building’, ‘in control of’, etc. etc. to carry, indefinitely, context-neutral invariant meanings: if they were truly context-neutral invariants, why ever — or how ever–would a need to change their meanings arise? But the evident reality is that meanings, applications, rulings, turn out to be extensively non-identical, and responsive to change and innovation.

4. What happens in statutory interpretation?

As is widely acknowledged, traditionally there have been three ‘rules’ for statutory interpretation, the literal, mischief and the golden rule, but in practice these have hardly jockeyed for preferment, since in recent decades the literal rule has reigned supreme. From a language-mythological perspective, the literal rule must appear unequivocably as the default option. For the literal rule trusts that, giving the words of a statute their ordinary and reasonable meanings, a coherent account of the Act can be derived. Zander (110) suggests that the ‘most rigorous’ expression of the literal approach came in Lord Halsbury’s remarks in Hilder v Dexter [1902] A.C. 474, where he argued that the draftsman of a statute was the worst person to interpret it, since he would be the most powerfully influenced by what he meant rather than what he had said. On those grounds, because he had himself drafted the statute relevant to the case, he refused to give judgement. While this ingenious line of reasoning invokes the familiar contrast between ‘what was meant’ and ‘what was said’, it is worth emphasizing that it really entails the postulation of two potentially distinct meanings: ‘the meaning of what was meant’ and ‘the meaning of what was said’. It then asserts that the meaning of what was said is there, in the literal objective wording of the statue, autonomous, fixed, but view of it is vulnerable to obscuring or contamination by parti-pris interpreters, especially the author with his exceptional stake in affairs. To proceed otherwise–according to this line of reasoning–that is, to seek to lay bare the meaning of what was meant, is clearly to seek to be guided by intentions–the draftsman’s intentions, or the government’s intentions, or Parliament’s. And to do so, particularly in the spirit of Lord Denning in Magor and St Mellens, might easily involve ‘filling in the gaps and making sense of the enactment’ in ways which are easily criticized as an ultra vires enlargement of the judicial function, ‘making sense of a law’ appearing not very different from roundly ‘making law’. So Lord Simonds, reasserting the norm, censures Denning and reminds ‘the duty of the court is to interpret the words that the legislature has used’. Similarly Denning was in the dissenting minority in the notorious will case, Re Rowland [1963] 1 Ch. 1 (CA), where all hinged on the meaning of coinciding as applied to the deaths of the testator and his spouse. The majority held that the deaths of Dr and Mrs Rowland, who apparently perished when the ship they were travelling on went down, were not sufficiently demonstrably simultaneous. Lord Denning complained that such a construal proceeded from fallaciously assuming that in construing a will ‘It is not what the testator meant, but what is the meaning of his words’ that matters, which he deplored as a nineteenth-century view. But that view, that ‘the meaning of the words’ is sovereign, prevailed; and at this point one is reminded of Hutton’s remark, in his paper on linguistics and the law, that ‘Linguistics is perhaps the “most 19th century” of the academic disciplines’ (Harris and Wolf: 298). In maintaining a fixed-code literalist-textualist interpretive standard, the law upheld in Magor and Rowland could be said to be equally or congruently 19th-century. By contrast, there is little fixed-codism in Lord Denning’s judgement:

You must put on [the testator’s] words the meaning which they bore to him. If his words are capable of more than one meaning, or of a wide meaning and a narrow meaning, as they often are, then you must put on them the meaning which he intended them to convey, and not the meaning which the philologist would put on them. In order to discover the meaning which he intended, you will not get much help from a dictionary. It is very unlikely that he used a dictionary, and even less likely that he used the same one as you. What you should do is place yourself as far as possible in his position, taking note of the facts and circumstances known to him at the time, and then say what he meant by his words… I decline, therefore, to ask myself: what do the words mean to a grammarian? I prefer to ask: What did Dr Rowland and his wife mean by the word ‘coincide’ in their wills?

By contrast the mischief and golden rules involve going behind the words of an Act, and their problematic meanings–or overriding them–to identify the mischief that the Act was intended to remedy, and to interpret the words of the Act so as to effect a cure of the mischief, or–so as to avoid absurdity and inconsistency–to modify the ‘grammatical and ordinary sense’ of the words of an Act. Nevertheless the literal rule has in the recent past remained predominant (Zander, 110), as is only confirmed by the traditional judicial disbarring from consulting Parliamentary debates and the like. This was very much the situation until the late 1980’s.

But it is of great interest that this is no longer quite the case, particularly at the appellate level where law is not merely dispensed, but made or revised. The literal rule has lost some of its former dominance, under pressure from the purposive approach–not unrelated to the formerly marginalized mischief rule–which has taken hold over the last quarter-century (to a degree, but not solely, as a result of the growing accommodation between English law and European union law). And, following Pepper v Hart, there is no longer the absolute exclusion of reference to Hansard and similar Parliamentary records, as there formerly was: that decision permits referencing where a provision that is ambiguous, obscure or absurd on a literal interpretation can be thereby made clear. It is partly in light of these shifts that I will argue that the English judiciary and legal profession are not so largely captivated by the language (or communication) myth as they formerly were. If time allowed, I would want to develop the argument that in the ‘higher’ reaches of the law–in the two English courts of appeal, in the work of leading Queen’s Counsels, and in academic legal circles–fixed-code telementationism is not particularly subscribed to at all. By contrast, law as practised by most high-street solicitors, as dispensed by lay and stipendiary magistrates, and as resolved in the High and Crown courts, arguably remains well-disposed to, and well-served by, a ‘language myth’ theory of linguistic communication.

To re-cap, when the law lords hand down a ruling that Sikhs are, or are not, ‘a racial group’ within the terms of the 1976 Race Relations Act, as in countless other legal ‘hard cases’ that hinge on a nomenclatural issue (what is ‘packed soil’, or ‘mental capacity’, or ‘direct professional responsibility–Hedley Byrne-, or ‘dishonesty’, etc. and do the circumstances of the present case involve an instance of this?), a codificatory and classificatory act entailing both inclusion and exclusion is required from the court. Sikhs must be determined either to be, or not to be, a racial group. And so on. There lies the reflection of a ‘fixed-code’ view of language with, mindful of the principle of equality before the law or the US 14th Amendment’s “equal protection of the laws”, the related assumption that fixed code is the same for all citizens within the jurisdiction.

The law equally, via binding precedent and the hierarchy of the courts, seeks to enforce a homogeneity, in principle, across the entire judicial system, from magistrates up: what the law lords have fixed as code prevails for an indefinite period, until such time as the lords overrule themselves (itself a relatively new notion, only explicitly acknowledged as recently as 1966). This is surely a Saussurean solution to the difficulty of subscribing to homogeneous code-fixity while recognizing that at another time, an equally fixed code–not intrinsically open to ‘development’ or change–may apply.

5. The conflict of interpretations in theft cases: appropriation and dishonesty.

My argument is that the tensions between a literal-rule applying interpretation of statutory language and a much more contextualist, common-sense-minded and purposive/mischief interpretation currently bedevils the courts’ interpretation and application of at least two of the key elements which must be established in a prosecution for theft: appropriation and dishonesty. Matters have particularly come to a head in the recent Court of Appeal case, R v Hinks (1998). In that case, the defendant, a young woman, had struck up a friendship with the alleged victim, a gullible middle-aged man of limited intelligence, as a result of which he withdrew £60,000 from his building society account and deposited it in the defendant’s account. Crucial to the defence case was the claim that the money was a valid gift, and that the recipient of a valid gift could not be guilty of theft. The court rejected this defence and, going further than they had been apparently prepared to in the recent cases of Kendrick and Mazo, they proposed that a gift might be clear evidence of appropriation. Accordingly the defence’s appeal failed. In this way, arguably, a vulnerable individual (and his beneficiaries–other than Hinks) was protected; and clearly there are good policy grounds for such a result.

But the route taken to that result has raised no small disquiet among some commentators, for seeming to create a disruption in the relations between the civil and criminal law, and in particular in relation to the traditional assumption that the theft law is there to protect previously-created entitlements under property law. As a result, as commentators have noted, the Appeal Court has created the seemingly absurd situation in which a valid gift, not voidable for deception, can count as the actus reus, the material action, of a theft. In effect the defendant Hinks was found guilty of stealing her own property. A question of law arising from the case has been sent to the Lords for deliberation, but not the key question, which is whether a perfectly valid civil act (here, a gift) can at the same time be a crucial element in a crime. Critics have argued that in Hinks the court should have relied more on challenging the validity of the gift, on grounds of undue influence, unconscionability, and the donor’s mental capacity, and so on, notwithstanding the view held by some that criminal trials are not the best place to determine such civil issues.

There is a recent history of interpretation of the terms dishonesty and appropriation which has arguably led to this controversial pass, and I need to review some of that history. One stage in that history concerns–in my view–a profound wrongturning on the most integrationally coherent interpretation of the key term appropriation. Appropriation can be said to be one of the key terms in the law of theft since, under the 1968 Theft Act (which superseded the common-law offence of larceny), theft is defined in s.1(1) as having occurred where a person dishonestly appropriates property belonging to another with the intention of permanently depriving the other of it.

Since Gomez (1993), it has been established law that appropriation of any item of property should be interpreted as ‘any assumption of a right of the owner’, regardless of that owner’s consent (particularly in light of some cases where the owner evidently consented to the appropriation–but where, crucially, deception or similar bad faith was involved). In Gomez, with only Lord Lloyd dissenting, the law lords affirmed that the state of mind of the owner–consenting to, protesting at, or unaware of the taking–was irrelevant to appropriation.

Now members of the community who are neither linguists nor legal academics or jurists may be a little alarmed to find that in relation to so fundamental a crime as theft, some of the key terms are continuing to be redefined and reconceptualized, so that what now counts as appropriation is not what the legal system once thought did, and similarly for dishonesty. For linguists and lawyers this is less surprising, but it is interesting to note that these changes would ordinarily be characterized as developments: the law develops, and the language develops (I am thinking particularly of semantic change, and lexical innovation, rather than phonological change), in each case the better to fit developing social requirements. What linguists and lawyers are less enthusiastic about endorsing is the idea that language varies in form and meaning, or that the law varies from case to case in unpredictable ways with each new occasion of use. The implication of this latter account is that a picture of broad development or progress, as distinct from a multiplicity of local differences, is a misleading picture. But this latter account of innumerable local divergences and distinctnesses of form and meaning, without progress, is the one to which the integrational linguistic perspective is more inclined.

One thing I want to argue is that, largely under pressure to find the meanings of theft within the language of the 1968 Act, the appellate courts have struggled to capture an essential characterization by means of careful but strained reinterpretation of dishonesty and appropriation. In so doing they have in relation to the first word had to rely on a term that is both contentious (we can no longer be sure that any two juries or any two defendants will ‘draw the line’ between honest and dishonest at the same place: passing on your unexpired parking meter time to another motorist? Photocopying your income tax form at work?) and vacuous, being applicable to a host of crimes besides theft. And as regards the second term, appropriation, the courts have stripped the term of much of its pertinence, fearful lest ‘consented-to appropriations’ escaped capture by the act. I would argue that it is common ground that a theft is an assumption of one or more of the rights of the owner of a certain property, with intention to permanently deprive, and without valid defence, but signally lacking from present legal formulations is an explicit reference to absence of the informed consent of the owner. Because nothing quite like this phrase appears in the Theft Act, judges guided by the rules of statutory interpretation have understandably declined to read them into the Act; but in avoiding doing so they have placed an excessive interpretive load on the term dishonest. In doing so they have, arguably, embraced a line of interpretation long ago warned against: the draftsman of the 1968 Act wrote to the Larceny Sub-Committee in January 1964 urging “One really cannot have a definition of stealing which relies on the word ‘dishonestly’ to prevent it covering every acquisition of property.” [reported in J. Smith, 1998: 904]. See also D. Elliott [Criminal Law Review, 1982, 395] who proposes that dishonesty is a dispensable concept which, in the law of theft, should be dispensed with.

An approach largely congruent with that proposed here appears in Halpin’s detailed critique of ‘The Test for Dishonesty’ (Halpin, 1996). Like others, he argues that the Ghosh test is unfair or unworkable due to “the absence of a moral consensus within modern society over dishonesty” (292), and he argues that in reality we are limited to just two options in testing for dishonesty:

a wholly subjective approach, which would allow the individual defendant to limit his criminal liability by his own moral standards; or, a legal definition of dishonesty. (292)

Unsurprisingly he prefers the latter, and his proposed general definition of dishonesty for property offences significantly includes

“treatment … of the property of another … without a belief that the other would consent to that treatment if he knew of all the circumstances …”.

By contrast with my stipulation of ‘informed consent’, which is a more objective standard, Halpin opts for the more subjective or defendant-friendly ‘belief [in the owner’s consent]’. (The subjectivity in Halpin’s definition could be curbed by objectivizing the belief: “…without a reasonable belief that the other would consent”.) Clearly my proposed phrasing leaves it open to the judge or jury to decide whether informed consent was secured. But it is interesting to speculate that on Halpin’s definition (the property must be of another, and belief in consent can be claimed) Hinks might well have been acquitted, whereas on mine she would first have had to pass a more objective standard of informed consent, rooted in the facts of the case and the conduct of the parties.

Interestingly, Shute and Horder (1993), taking issue with the conflation of theft and deception that the Lords’ decision in Gomez seemed to licence, offer a conceptual analysis in which theft and deception are distinguished on the grounds of the involuntary ‘having had taken’ in the former, and the voluntariness of the giving (but under false pretences) in the latter: “only theft involves appropriation, and appropriation is essentially an involuntary transfer” (1993: 554). Like me, then, they are inclined to reassert that appropriation relates to the mental element and not merely the observable conduct, even if they prefer involuntary to lack of informed consent.

But the upshot of Gomez and Hinks is that the following are currently valid legal precepts:

  • The donee of a valid gift can be guilty of theft of the gifted property.
  • The alleged victim’s state of mind–e.g., evidence of their informed consent, or of absence of this–has no bearing on determining whether appropriation has taken place.
  • Crucial to a finding of theft is a determination of dishonesty; but where dishonesty is denied and doubtful, even behaviour that is dishonest by ordinary standards may escape criminal censure if the defendant was unaware that the ordinary standard deemed this dishonest (the Ghosh test, said to protect the ticket-less foreign visitor, who has assumed that travel on city buses is free).

This long-running saga of attempted statutory codification (in the 1968 Act and, as in effect a supplement, the 1978 Act), and contested reinterpretations of the key statutory terms dishonest and appropriation suggests to me one thing above all. This is that the judges have created opacity and uncertainty particularly because they have struggled to adhere to the standard principles of statutory interpretation, and particularly the literal rule of statutory interpretation rather than the purposive or mischief approach that has burgeoned in the last twenty years and been granted limited sanction in the important ruling in Pepper v Hart. (Significantly, in his sole dissenting judgement in Gomez, Lord Lowry suggested that the majority decision was not in line with the view of the Criminal Law Revision Committee, whose report formed the basis of the 1968 Act.) There are defensible grounds for the judges deciding that the term appropriation in the 1968 Act’s definition of theft should not be extended so as to imply or entail absence of the owner’s informed consent, and scrutiny of the 380 occurrences of appropriation in the 350-million-word Collins Bank of English suggests that the term is most commonly used in either contexts of revenue and finance allocation, where it collocates with Bill and committee, or in artistic and cultural contexts–cf. appropriation art–where it denotes acts of borrowing, recontextualizing, or quoting of someone else’s art or culture. In short, there are good reasons, on a literal or even a more liberal purposive reading of the Act, for declining to ‘expand’ the meaning of the term appropriation to the point that it can entail ‘absence of owner’s authorization or consent’. Consequently, the term reduces to denoting any kind of taking or interference with property of a kind that only the owner is entitled to perform: any adoption of the rights of the owner. Equally, the diminution of determinative significance of the term appropriation leads to an increased interpretive burden being placed on the only other term in s.1(1) that can be reasonably looked to carry the burden of, as it were, ‘articulating theftousness’, namely the word dishonest. Allowing the requirement that there must be intent to permanently deprive, and that no defences of honest belief in justification or likely permission, the statutory definition boils down to nothing more specific than a usurping of one or more of an owner’s rights dishonestly. As noted above, in the discussion of Ghosh, the courts have directed that in most cases determining whether behaviour is dishonest should be left, without judicial guidance, to the good sense of the jury; in the majority of theft cases, then, the crucial criterion for conviction turns out to be one not defined in the governing statute or case law, but one left to be decided by ordinary reasonable people, in much the same way that they might decide what counts as hot weather, or tall, or wealthy. In a minority of difficult cases, as we have seen, the two-stage Ghosh test has been proposed, with its objective leg and, failing this, a secondary subjective leg at which latter stage the defendant’s protestation that he was unaware that his act was dishonest by societal standards is allowable as a valid defence. On closer inspection it is arguable that the Ghosh guidelines for exceptional cases are no more robust than the reliance on uncodified common-sense involved in run-of-the-mill cases; for by the first stage of the Ghosh test, if the jury decide that the act was not dishonest by everyday standards, then the defendant’s safety-net of the second leg is not even invoked.

What most of the judges felt that they could not do with s.1(1) was to read into it an entirely distinct phrase or criterion that seems absent from its explicit wording, such as ‘without the owner’s informed consent’. And yet that, I would contend, is required so as to better integrate the statute with the cases it appears to be intended to cover. It is also, over many pages of judgement, what Lord Lowry argues is entailed by the term ‘appropriates’ in the 1968 subsection. He first invokes the primary dictionary meaning, which he reports is “to take possession of, take to oneself, especially without authority”, and argues at length that appropriation is a one-sided act, a taking that disregards lack of consent, to be contrasted with section 15 deception, which entails a giving with consent, but fraudulently secured.

But Lord Lowry’s approach was rejected by his fellow judges. And what seems to happen in practice is that dishonestly is tacitly interpreted to mean ‘without the owner’s informed consent’ where the latter criterion lies at the heart of the grounds for prosecution (as, arguably, in the Hinks case). Thus, striving to create a semblance of clarity and consistency, our most senior judges have juggled the two crucial terms given in s.1(1) of the Act

How is an appellate court’s findings, where these concern statutory interpretation of a word or phrase, different from, and similar to, those of one or more lexicographers, working collectively or perhaps independently and preparing different rival publications? And how are the appellate judges proceeding in ways that are different from, and similar to, any party to a casual conversation, who must invoke real-world knowledge, a calculation of what the relevant current context is, in making sense of the sounds and gestures and communicative signs they encounter? Clearly the stakes may be high in appeal court decisions, with money, property, and reputations at risk. By contrast, the stakes in everyday conversations will ordinarily be lower, with lexicography perhaps somewhere in between, with a lexicographer’s lamer or more contentious efforts triggering ridicule from peers and scholars, outrage and denunciation from groups who regard themselves as having been misrepresented.

6. Conclusions?

One conclusion to be drawn from the tangled arguments in recent theft appeal decisions is that judges do not all seem to be adopting the same theory of language, they do not all seem to invoke literalist vs purposive grounds, or text-immanence vs legislative intent, in the same ways, any more than in their judgements do they rely on the same previous cases, preparatory works, academic commentaries, and so on. The language and terms of their judgements, from case to case, is not fixed but changes and adapts to new circumstances: we might say that their discourse lacks the certainty and foreseeability that would cause litigation to wither away, make it pointless.

A related conclusion is that, as Fish has suggested in rebuttal of Dworkin, judges pervasively operate within a practice within which their ways seem natural, almost inevitable, rather than operating by means of, for example, observing a case and a proposed decision ‘from outside’, from where an independent governing theory (endowed with Dworkinian ‘articulate consistency’) can be used to weigh or measure the instant case and ‘derive’ a judgement. Judicial decision-making does not seem to be a practice that judges can step outside of, anymore than we can usefully analyze language from outside it. One of the reigning metaphors of jurisprudence is the scales of justice, and the striking of a balance; the metaphor of striking a balance often seems as close as judges can get to explaining what is meant by something being reasonable. And it is surely significant that in the balance metaphor, various disparate circumstantial considerations are not weighed against any absolute measuring device, to be found of the correct weight or incorrect by virtue or underweight or overweight. In fact, the web of considerations is not weighed at all, but simply or not-so-simply balanced, with no resulting measurement or correctness; and the balancing of one factor is entirely in terms of countervailing factors, with no logical place for any external or ‘third way’ commentary or evaluation. In legal balance-striking, there are only opposites with no positive terms.

The interpretive disputes surrounding appropriation, dishonesty, and owner’s consent in the definition of theft will surely rumble on. Following Pepper v Hart and European practices, purposive interpretation and Parliamentary referencing are likely to grow. Such referencing is, in some respects, an unravelling of codification, like an online dictionary where every entry has hotlinks to further explicatory material, itself likely to be amended or deleted over time. At the same time allowing referencing entails a slight diminution of the appeal lords’ autonomy and powers of construction; some will complain that the rule of law loses some ground not to Parliamentary sovereignty, but to the rule of ministers and commissions. Whether the majority of law lords in a theft case will ever allow referencing of the kind that Lord Lowry and Professor J. Smith would welcome is doubtful (the Act has been being applied for over thirty years now, and criminal law is rather different from tax law).

As I hope I have made clear, my own view is that the notion of ‘lack of informed consent’ should be routinely treated as part of the s.1(1) definition of theft, which could be achieved either by statutory amendment, or by a ratio such as Lord Lowry’s in Gomez, in which he declares the notion an entailment of appropriation, coming to express the majority law lords’ view. But it is in the nature of the situation that no irrefutable logical argument, no appeal to code or system, can be advanced in support of this proposal. Like all legal and communicative determinations, it is a situated and provisional one, my current best effort at integrating a host of diverse factors, including but not limited to treating the 1968 Act as congruent but no copy of the 1916 Act, distinguishing s.1 theft from s.15 deception (where owner’s consent is no defence), maintaining where possible a congruence between the statutory language and everyday discourse about theft, attempting to maintain, thereby, the possibility that ordinary members of the community will understand the rules by which their behaviour is constrained, failing which the rule of law falters, and so on and so forth.

So there are no permanently right answers in the law, even in relation to something so seemingly fundamentally straightforward as theft. And the law lords are all integrationists, even where they imply otherwise. In practice they adopt or disregard dictionaries and commissions and certain words of statutes and clarifications by government ministers as they see fit, mindful of the very real pressures of the instant case, which is always potentially distinguishable. A prior decision of the law lords is binding, but only until a later Lords’ decision revises the law. And quite what the binding ratio decidendi of a case is always remains open to revision. In law as in linguistic communication, past codifications constrain a present meaningful interaction only insofar as those involved in the present interaction allow them to. In the law as in language, integrationism underpins communicational change; and in so doing, it equally underpins sociocultural change.

References

  • Beynon, H. and N. Love. 1984. Mandla and the meaning of racial group. The Law Quarterly Review, 100, 120-136.
  • Bix, B. 1993. Law, Language, and Legal Determinacy. Oxford: Clarendon Press.
  • Evans, J. 1988. Statutory Interpretation: Problems of Communication. Oxford: OUP.
  • Fish, S. 1990. Doing what comes naturally: Change, rhetoric, and the practice of theory in literary and legal studies. Durham. Duke University Press.
  • Goodrich, P. Legal Discourse. Macmillan
  • Halpin, A. 1991. The appropriate appropriation. Criminal Law Review, 426-432 June
  • Halpin, A. 1996. The test for dishonesty. Criminal Law Review. 283-295 MAY 1996
  • Harris, R. 1981. The Language Myth. London: Duckworth.
  • Harris, R. 1998. Signs, Language and Communication. London: Routledge.
  • Hurst, D. J. 1983. The problem of the elderly statute. Legal Studies. 3:1.
  • Hutton, C. 1998. Law lessons for linguists? Accountability and acts of professional classification. Language & Communication, 16:3, 205-214. Reprinted in Integrational Linguistics: A First Reader, eds. R. Harris and G. Wolf, Oxford: Pergamon, 1998, 294-304.
  • Joseph, J. 1995. Indeterminacy, Translation and the Law. In Translation and the Law, ed. by Marshall Morris, 13-36. (American Translators Association Scholarly Monograph Series, 8.) Amsterdam and Philadelphia: John Benjamins, 1995.
  • Love, N. 1985. The fixed-code theory. Language & Communication, 5:1, 1-17.| Reprinted in Integrational Linguistics: A First Reader, eds. R. Harris and G. Wolf, Oxford: Pergamon, 1998, 49-67.
  • Twining and Miers. How to do things with rules.
  • Zander, M. 1994 . The Law-Making Process. 4th edn. London: Butterworths.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s