Textualism

Textualism is a family of theories of legal interpretation that are united by their stress on the primacy of the text of the canonical legal text in question. For the textualist, the text's meaning always trumps the intended meaning of the text's author(s).

We reply to the textualists in two stages. In stage one, extending the argument of Chapter 2, we show the impossibility of ‘pure' textualism, that is, the total primacy of text over authorial intent. In stage two, we show that ‘impure' textualists, the only possible kind, are either those who would construct - and hence reauthor - legal rules out of raw materials provided by the original author(s) of the legal texts, or those who are intentionalists but who impose limits on the evidence of intent that interpreters may consult.

4.1.1 The impossibility of pure (intention-free) textualism

Self-described ‘textualists' hold a variety of positions on how one ought to interpret legal texts. Indeed, there does not appear to be any canoni­cal description of textualism. What unites textualists is their stated refusal to consider the intentions of the laws' authors to determine what the laws mean. However, such intention-free textualism is a con­ceptual impossibility. For, as we argue, authorial intentions constitute the meanings of authoritative legal texts; and a charitable reading of textualists' statements would not saddle them with a commitment to intention-free textualism. Nevertheless, self-described textualists often say things that appear to endorse intention-free textualism.

Consider, for example, one of the most famous modern textualists, the late Justice Antonin Scalia, and his discussion of textualism in A Matter of Interpretation (Scalia 1997). His version of textualism seems to have three principal tenets. The first is that a textualist searches for an ‘objectified' intent, the intent an idealized reader who knows the entire body of law would infer from the text of the particular law in question.

The first is the most important tenet, as the other two tenets are its corollaries. The reason for searching for an ‘objectified' intent is, for Scalia, ‘that it is simply incompatible with democratic government, or indeed, even with fair government, to have the meaning of a law determined by what the lawgiver meant, rather than by what the law­giver promulgated'. After all, says Scalia, ‘[i]t is the law that governs, not the intent of the lawgiver'. No matter what the legislators intend, ‘[i]t is only the laws that they enact which bind us' (Scalia 1997, 17). To govern by ‘unexpressed intent' is, Scalia argues, tyrannical in the same sense that Nero's posting of laws high up a pillar was so: people will not be able to make sense of the law if they try to discern the subjective intent of the legislature.

Justice Scalia derives the second tenet from an examination of what judges actually do in practice. If the intent of the legislature mattered, then judges would not apply the rule that ‘when the text of a statute is clear, that is the end of the matter' (Scalia 1997, 16). Likewise, if legis­lature intent were the touchstone for interpreting legislation, judges would not assume that the enacting legislature was aware of all existing laws. Instead, says Scalia, judges would pay attention to the text and the legislative history of the particular statute in isolation, for that is all the legislators likely had in mind.

The third tenet follows directly from the first. For if only objectified intent matters, legislative history, which yields only indications of subjective intent, should not be regarded as relevant evidence (Scalia 1997). Moreover, says Scalia, even if legislative intent were the touch­stone of statutory interpretation, we should not look to legislative history to discern legislative intent.

Justice Scalia's explication of textualism prompts a series of questions. One is whether his version of textualism makes textualism merely a quest for the intentions of the legislature with certain evidence (like legislative history) barred from consideration, or whether it is a version that makes the search for legislative intent completely irrelevant even when that intent is known with certainty? It appears on balance that Justice Scalia adopts the latter approach - actual legislative intent is always irrelevant.^[43] Any uncertainty about whether Justice Scalia is truly an intention-free textualist stems from his use of the concept of ‘objectified intent', which might appear to acknowledge that the intent of some author matters. Nonetheless, Scalia's reference to ‘objectified' intent, when placed in the context of the other statements just quoted, can be taken to mean that he regards the legal rule's text as meaningful apart from the actual author's or authors' intent.

For our purposes, it does not matter whether Justice Scalia or other textualists, such as John Manning (Manning 2001), really are inten­tion-free textualists.^[44] What we want to do is show that such a position is a conceptual impossibility and leave it to others to determine if any self-described textualist actually holds the position we discredit.

4.1.1.1 Argument one: texts cannot declare the language in which they are written

One cannot attribute meaning to marks on a page, to sounds, or to any other medium that might be used as a symbol for communicating thoughts - think of sign language, semaphore, Morse code, smoke signals, and so on - without reference to an author, actual or ideal­ized, who is intending to communicate a meaning through the marks, sounds, or other symbols.

For the same point, consider the following amusingly bewilder­ing statement: ‘I am speaking English, not Schmenglish', which in Schmenglish means ‘I am speaking Schmenglish, not English'. How will an intention-free textualist decide whether this statement is in English or Schmenglish? We do not believe that intention-free textual- ists will have an answer, or at least an answer that does not smuggle in reference to authorial intent.

Our claim is that we must posit the existence of some author if we are to attribute meaning to these statements. If we know that the real author of ‘canard' generally speaks French, we most likely would con­clude that ‘canard' in this context means ‘duck'. If the author usually speaks English, we most likely would conclude that it means ‘fib'. If we are unaware of (or indifferent to) the author's usual tongue (and likely intentions), we might imagine what we would have meant had we spoken the term, imagining ourselves as the authors.

This is not merely a problem across languages. For even within English, these issues will often arise. If someone walks into a restaurant and declares ‘I would like some chips', what is meant by ‘chips'? Once again, we think we should understand ‘chips' by reference to the communica­tive intentions of the speaker.

4.1.1.2 Argument two: texts cannot declare that they are texts

An equally fundamental problem for intention-free textualists is that texts cannot declare that they are texts or even declare which part of the putative text constitutes the text. Suppose the US Constitution that appears in our casebook was actually typed by a monkey. Are the ink marks made by the typewriter keys a text? Unless one posits a hypothetical author with intent to convey a meaning, we think not. Without an author, real or hypothetical, intending to convey a mean­ing through these marks, our seemingly grand Constitution is nothing but a randomly generated mass of inked shapes that merely resem­bles a text. Or suppose a Martian composed the Constitution in our casebook, and that Martians treat what we take to be spaces between letters and words as the actual letters and words and regard what we take to be letters and words as the actual spaces. If that supposition were correct, then the ‘text' in our casebook would be quite different from the text that we assume it is. The text that we assume to exist as the US Constitution is actually no different in kind from meaningless marks made by waves on the beach or by cloud formations in the sky; it is merely the meaningless residue of the Martian's text.

Our simple point is that one cannot look at the marks on a page and understand those marks to be text (i.e., a meaningful writing) with­out assuming that an author made those marks intending to convey a meaning by them. The reason why no one treats the Constitution as a bunch of unintelligible lines and curves is because everyone assumes a particular kind of author for the Constitution, such as its actual drafters (the Framers); those who purported to make it law (the Ratifiers); or the author that would have been assumed by an idealized, contemporary reader of text.

4.1.1.3 Argument three: textual meaning cannot be autonomous from intent - one must always identify an author

The argument in this section builds upon the previous one. Imagine some people who come upon marks on the ground that are shaped like a c, and a, and a t. They begin to debate whether the marks mean ‘domestic tabby cat', ‘any feline', or ‘jazz musician'. They are then told that the marks were made by water dripping off a building. Their debate over meaning should now cease: no actual author, no actual meaning.

Suppose now that they know that a person made the marks. They encounter him and tell him of their debate. He tells them that he never intended to make letters. Rather, he was marking out the contours of patches of a vegetable garden. Again, the debate over meaning ought to cease: no intended meaning, no meaning.

Now suppose that the person did seek to make a word. The people debate the meaning of ‘cat'. The ‘author' then informs them that he was writing an ode to his beloved tabby. That should settle the debate: ‘cat' here means tabby. The alternatives - any feline and jazz musician - are just as much off the table as they were in the previous examples of no author and no intended meaning.

The same point applies to other examples of ‘mindless' texts. If ‘trunk' is produced by an elephant who paints with his trunk, or by legislators each drawing letters randomly from a hat, it is useless to ask whether it means the main axis of a tree, the rear storage compartment of a car, or the nose of an elephant, or even what language it is in. Without an author who intends a meaning, such marks are meaningless. ‘Texts' without authors and intended meanings are not texts; and texts with intended meanings are texts only with respect to the intended meanings.^[46]

4.1.1.4 Argument four: texts can have ‘deviant’ meanings because those meanings are intended

How did ‘cat' come to mean jazz musician? It did so because it was used by some people with the intent that it be understood as refer­ring to a jazz musician. That is ultimately how all words acquire their meanings. And the word ‘cat' meant jazz musician the very first time it was used with such an intention, even before it was listed as a defini­tion in the dictionary. Similarly, if a speaker says ‘Gleeg, gleeg, gleeg', it means what the speaker intended it to mean, even if to others it sounds like nonsense.^[47] And if your aunt is Mrs. Malaprop, and she asks you to make sure the ‘autobahn' is pulled next to the sofa when she comes to visit you - and you know that she intends for you to move the ‘ottoman' - then, if you are a dutiful nephew or niece, you will pull up the ottoman and not attempt to relocate a German highway.^[48]

4.1.2 Impure textualism and the construction of rules by ‘interpreters'

4.1.2.1 The algorithmic textualist

When the conceptual impossibility of pure, intention-free textual- ism is hammered home, textualists frequently retreat to a less pure position. They will admit that we must assume authorial intentions to determine that the marks we are supposed to interpret is in fact a text, that is, an attempt by some rational being or beings to convey a meaningful proposition to others. And they may admit that we must refer to authorial intent to determine the language in which the text is written (or spoken). At this point, however, the textualists claim that we should jettison the search for the speaker's intended meaning and rely solely on the textual (utterance) meaning.

So how would a textualist interpret a canonical legal text? After ascer­taining that it is in, say, English, she would have recourse solely to dictionaries and books that set forth proper grammar, punctuation, and usage. Those, she would argue, tell us what the utterance - the text - means, not what its author(s) intended to mean by the text. And the text is ‘the law'.

The first point to note about this impure textualism is that, if one employs its methodology, ‘the law' is constructed not solely by the authorities who author it but by those authorities together with a ‘mindless' algorithm. The mindless algorithm is a function of the dictionary and grammar book that are used to construct a meaning out of the author's intended symbols and intended language. Your aunt utters the sounds or makes the marks ‘autobahn'; the faithful interpreter runs a German highway through the living room. Your aunt did not mean that, and we know she did not. Nevertheless, the textualist would claim that her text meant that. (We leave aside the problem to which we have previously adverted of individuating languages and distinguishing between a language and idiolects thereof: did your aunt misspeak in English, or did she speak correctly in Malapropenglish, either her particular idiolect of English or perhaps even a separate language?)

Further, many words have several meanings, even in standard English. So the algorithm would have to specify which meaning counts as the meaning of the text when the dictionary delivers more than one possi­ble meaning. For example, the algorithm might say that the first listed meaning of the word is the meaning to be attributed to the text.^[49]

Of course, the textualist's algorithm would have to be a norm higher in authority than the lawmaker whose texts are being interpreted because that algorithm operates as a constraint on the rule-maker's ability to bring about the legal norms that he intends. Textualist canons are to the laws they govern as constitutional rules are to statutory rules, or statutory rules are to administrative or judicial (non-constitutional) rules.

Hypothesize the following textualist algorithm: once marks are deter­mined to be a legal text in English, give them the meaning they would have according to the first meanings given in a particular dictionary and according to particular references for grammar, punctuation, and usage. What would a world governed by textualist algorithms such as this one look like?

For one thing, many legal rules, the intended meaning of which we know perfectly well, would have to be treated as meaningless and thus of no legal effect because they would not parse grammatically or syntactically, or because a word was misspelled and rendered a non- word.^[50] Other laws would have to be given meanings that we know were not the intended meanings and that may be absurd. Scrivener's errors could not be ignored, and in Cernauskas v. Fletcher,^[51] all of Arkansas's laws would have been repealed by some minor law con­taining a scrivener's error. The Seventeenth Amendment would have expired after only six years.^[52] And so on.

Textualists, of course, might argue that legal rule-makers would be very careful in conveying their intended meanings so that they cor­responded to the meanings the textualist algorithms produced. They would argue that over time, scrivener's errors and errors regarding dictionary meanings, punctuation, and grammar would be minimized. Indeed, putting impure textualism in its best light, interpreters might arrive at the rule-maker's intended meaning more often by following the textualist algorithms than by consciously attempting to discover those intended meanings. In other words, algorithmic textualist interpreters, interpreting the words of rule-makers who understand that their interpreters will be algorithmic textualists and, accord­ingly, choose their words (and punctuation marks) with great care, may come closer to the rule-makers' intended meanings over the full array of cases than intentionalist interpreters. That would, of course, ultimately depend on the skills of interpreters at discerning intended meanings and the skills of rule-makers at conveying intended mean­ings in conformance with the textualist algorithms.

Of course, if the textualist algorithms are incomplete, then they will have to be supplemented. If, for example, the algorithms do not elimi­nate an ambiguity regarding whether a word in a legal text means A or B, perhaps because the dictionary treats a single spelling as two different words rather than as one word with several definitions, then the interpreter will have to go beyond the algorithms. Indeed, it will be difficult to devise any algorithm that will completely eliminate ambigu­ity. Does ‘No vehicles in the park' cover a tank used as a war memo­rial, skateboards, and shopping carts? Does the ‘no wild animals' rule we hypothesized proscribe a stuffed grizzly hunting trophy? If we are told not to decide such issues by reference to authorial intentions but rather to decide them solely by reference to dictionaries and so forth, we will be at sea. Likewise, in cases of ambiguity, the natural question for the interpreter to ask is whether the rule-maker meant A or meant B. But it is possible that the rule-maker meant neither A nor B. In such a case, authorial intent clashes with textualist algorithms, and the latter fail to settle the matter.

Perhaps textualist algorithms can be constructed so that no cases such as the one hypothesized can arise. The algorithms will produce a single answer to every interpretive question, and ambiguities will be impos­sible. Moreover, following such textualist algorithms will perhaps, over the run of cases, come closer to mirroring authorial intentions than attempts to discover those intentions directly. That possibility cannot be ruled in or out as a matter of theory. But even if such algorithms were possible and beneficial, the important point for our purposes is this: to the extent the legally effective meaning is a joint product of authorial intentions (in choosing marks in a language) and textualist algorithms, it is in part a mindless production, and at the interpre­tive stage a purely mindless matter. And mindless, mechanical con­struction of meaning for legal texts through such algorithms requires absolutely no judgment and surely no special craft skill possessed only by lawyers. It is not some sort of special ‘legal' reasoning. What does require judgment or skill is determining authors' intended meanings. But such abilities are abilities everyone has to varying degrees, and lawyers have no corner on that market.

Moreover, although impure textualism, unlike pure intention-free tex- tualism, is a conceptual possibility - it is really a form of intentionalism that posits, as the author of the legal text in question, not the actual author(s), but a purely hypothetical author who has perfect command of grammar, punctuation, and usage and always uses, say, the first dictionary definition given - no legal systems of which we are aware employ impure textualism. Clearly unintended absurdities and scrive­ner's errors are everywhere ignored (Alexander and Prakash 2004). So, too, are punctuations that obviously conflict with intended mean­ings. And context, not the order of listed meanings in dictionaries, is resorted to in order to resolve ambiguities. The sign that says ‘Keep off the grass' located in the window of the drug counselor's office has a dif­ferent meaning from that of the same sign located on the lawn outside, no matter which definition of ‘grass' is listed first in the dictionary.

4.1.2.2 Four non-algorithmic textualists

Most self-styled textualists do not argue for the use of mechanical algorithms to produce textual meanings. But, if intention-free textual- ism is an impossibility because interpreters must always have recourse to some author, real or otherwise, what other possibilities are there for textualists? They reject full-blooded intentionalism of the sort in which interpreters gather all the evidence available of the authorially intended meaning, for that position is their foil. Moreover, textualism would not be a significant alternative to intentionalism if it merely claimed that one ought to exclude some evidence of the authorial intent from consideration on grounds of unreliability - for example, some forms of legislative history. No full-blooded intentionalist would endorse use of unreliable evidence of authorial intent.

This leaves textualists with perhaps four possible positions. First, tex- tualists might wish to exclude certain evidence of authorial intent for reasons other than unreliability but otherwise interpret as would an intentionalist. This position is one that we believe is tenable, so long as the norms excluding the evidence of authorial intent are superior in authority to the lawmakers whose laws are being interpreted.

Second, textualists might advocate interpreting laws based on asking some sample of readers - or some median reader - what meaning they believe the actual authors intended. This position is unattractive for a host of reasons.

Third, textualists might advocate interpreting laws based on the intentions that a purely hypothetical construct - an idealized reader - would attribute to the law's author. We find this position, to the extent it differs from the first one, to be quite problematic.

The fourth textualist position would have the interpreter read the text as if it were written by, and thus carried the intended meaning of, an idealized author. We find this position to be perhaps the most prob­lematic of all.

Position one: textualism as rule of law-restricted intentionalism. On this version of textualism, the interpreter seeks out authorial intent, but in doing so refuses to consider certain kinds of evidence thereof, even if reliable. For example, we might have reliable evidence that a law, which appears to be written in standard English, and which can be given a sensible meaning therein, was actually written in non­standard English, or Schmenglish. We could imagine an interpretive norm to the effect that lawmakers will be irrebuttably presumed to use standard English in writing laws. We might tell a rule of law story about the justification for such a norm, such as the need for the general public to know the laws, and so forth, or an indirect consequentialist story that pits getting authorial intentions right in the general array of cases against getting them right in any given case. And we might give a similar rationale for excluding even reliable legislative history - for example, that such history is not generally available or can lead to non­transparent manipulations of the lawmaking process.

This version of textualism is coherent and perhaps plausible. But we would have to know the provenance and authority of the norms excluding various types of evidence of lawmakers' intentions. Because the evidence of authorial intentions excluded by such norms is reli­able evidence, the interpreter will end up in a situation in which the authoritative meaning of the law is different from what the interpreter knows was the meaning intended by the lawmakers. This may not seem to be a devastating criticism; for, in treating statutory and constitu­tional interpretations by courts as having the force of precedent - that is, as binding even if incorrect - the courts countenance a similar gap between authoritative meanings and actual meanings. Moreover, if our interpretive norms exclude certain kinds of evidence of lawmakers' intentions, the lawmakers will legislate in light of those norms, thereby narrowing the gap between the meaning they actually intend and the meaning that they will be deemed to have intended. For instance, if they know their intentions will be interpreted as if they had expressed them in standard English, they will try to use standard English and not Schmenglish in writing the laws. Still, the gap between what the inter­preter knows the lawmakers actually intended and what, according to these norms, the interpreter will deem them to have intended remains a constant possibility under this version of textualism. Indeed, if the indirect consequentialist justification for excluding certain evidence is sound, this gap is no different from the gap between a rule and its underlying justification.

Position two: textualism as man-on-the-street interpretation. Textualists could be seen as advocating interpreting legal texts as would some sample of average members of the public. Such a method might be thought by some to have rule of law benefits, particularly in giving the average citizen clear notice of what the law means. We believe that any such theoretical benefits are largely chimerical because the position faces a devastating problem of indeterminacy.

One aspect of this problem relates to how much background context we ought to provide to the average interpreter. We might take the law to mean whatever it would mean to a collection of people who are provided no context whatsoever - other than, perhaps, that its authors were English speakers and enacted the law on a given date. Then we are back to algorithmic textualism and might as well con­struct a computer program that incorporates datable dictionaries and rules of syntax, grammar, and punctuation and then ask the computer to give us the law's meaning. Or we could ask the ‘median' member of the public what the law means to him or her. But this encounters a daunting problem, namely, that of identifying the ‘median' member of the public. Because there are an indefinite number of dimensions on which one can identify a median member of the public, the concept of a median member of the public is indeterminate. We will derive different authoritative interpretations depending upon the qualities of the median man-on-the-street. Given this indeterminacy of meaning, we view the supposed notice benefits of this mode of interpretation to be largely imaginary. And given that the median member of the public, however designated and however much evidence of authorial intent he is allowed to seek, will be less knowledgeable regarding what the lawmakers meant by a legal text than a judge, it is not at all evident why we would want the latter to defer to the former in the absence of the benefits of determinate advance notice.

Position three: the idealized reader. Textualists frequently invoke the construct of an idealized reader, one who exists contemporaneously with the rule's enactment, and ask how that person would interpret the text. As Judge Easterbrook has put it, textualists interpret language by asking how a skilled, objectively reasonable user of words would have understood the text. Justice Scalia has argued that judges should read the federal statutes as any ordinary member of Congress would have read them, and apply the meaning so determined. But textualists usu­ally go further and consider the context of the statute and background legal conventions. Hence, the idealized reader for these textualists is a lawyer (or at least someone who knows the standard legal conventions) who knows the factual background of the statute's enactment.

Supplying the idealized reader the ‘context' of the statute is, as we have already noted, really a backdoor means of reintroducing the author's intent. Here, we make some different points about the use of the ideal­ized reader. First, the idealized reader will seek clues illuminating the actual author's intent. For when asked to interpret something, people typically seek the actual author's intent as the source of meaning. (Recall the ‘autobahn next to the sofa' example.) So it is possible that textualists, in creating an idealized construct to generate an ‘objec­tive' meaning, have instead merely created an abstraction that filters authorial intention. The more evidence, whether direct or indirect, an idealized reader is given of what an author meant by a text, the more the reader will read the text to mean what the author intended to mean by it.

Moreover, even if the textualist denies that the idealized reader will seek the intent of the actual author, the idealized reader will still have to search for some intent. Again, one must envision an author when­ever one attempts to make sense of text, or even to identify it as a text. Thus, the idealized reader will have to imagine a hypothetical author (or authors). Although there may be certain advantages to treating a text generated by a multi-member body as if one person created it, one of those advantages is not an ability to dispense with the search for intended meaning.

If we assume that the idealized reader imagines one hypothetical author (rather than multiple authors), there may be a benefit to this textualist idealization. With the selection of one idealized author, it becomes much more likely that every statute will have a meaning. After all, the more authors a text has, the more likely it is that there is no shared intent as to the meaning of the text. And for the intentional- ist interested in authorial intent, if there is no intent that is shared by the requisite number of legislators, the text has no authoritative inten­tion to give it meaning and therefore has no meaning. Hence, if one prefers that there be more meaningful legislation, there is an advantage to hypothesizing one author when there are multiple real authors.

There is a potential benefit from requiring that the idealized reader be an average member of the general public. For if the law would be incomprehensible to members of the general public, then in this view, there is no law, even if a well-versed lawyer would be able to tease out some meaning for it. And this approach has the added benefit that people generally might know what the law requires without having to consult with high-priced experts or go to court.

Most modern textualists, however, assume that their idealized reader knows the standard legal conventions and the entire corpus juris. For that reason, the meanings generated by their idealized reader will lack any of the advance notice and rule of law benefits mentioned pre­viously; for average folks are unlikely to know either standard legal conventions or the entire corpus juris. The only benefit secured by modern textualists is, it seems, the avoidance of the problems of multi­ple intended meanings in multi-membered lawmaking bodies.

Of course, this ‘average reader' approach is itself not strictly empiri­cal, nor is it determinate. It requires us to determine how the median reader - not the average reader, because readers and their mean­ings cannot be ‘averaged' - would read a legal text. As we have said, however, the notion of a median reader is indeterminate. To make it determinate, we would have to spell out all sorts of characteristics of the median reader, such as whether he had a median IQ, had a median knowledge of public affairs (which is meaningless, because there is no unitary scale of such knowledge), was of a median age, had a median geographical location (again, a meaningless notion), and had a median education (again, meaningless, because there is no single educational continuum). The idealized reader is, indeed, a radically indeterminate notion. Making it more determinate, without making the idealized reader into someone who knows what we know about the authorial intent, will likely depend on arbitrary stipulations, such as that the idealized reader went to a particular high school and reads a particular newspaper.

Finally, recall that intentionalists must deal with the aggregate intent problem when there are multiple lawmakers, and the concomitant possibility of occasions when there is no dominant authorial intent and hence no dominant meaning. Textualists of the idealized reader type have a mirror-image problem, the problem of a surfeit of mean­ing. That is because the idealized reader whom the textualist stipu­lates, precluded as that reader is from looking to all the evidence of actual authorial intent, might well conclude that a text has two or more meanings that are equally supported by the evidence to which he is restricted. For the textualist, the text then just does have these multiple meanings. There is no deeper metaphysical fact, like intent, of which these multiple possible meanings are merely evidence. The multiple meanings just are the metaphysical fact at issue. The text just means two or more things, however silly or pernicious as a practical matter that is.

Position four: the idealized author. The final possible textualist posi­tion is that legal texts should be interpreted, not by reference to an idealized reader of the text, but by reference to an idealized author. In other words, legal texts should be interpreted to mean what they would have meant had they been authored by this single idealized lawmaker rather than by the one or several actual lawmakers.

The problems with this approach should be obvious. In order for it to yield interpretations, we need to specify the attributes of the idealized author. What language does he speak? Does he always use primary definitions of words, or does he sometimes (when?) use secondary defi­nitions, or technical definitions, or terms of art? Are his grammar and punctuation perfect? How rational is he? How just? And so on. How we construct the idealized author will determine the authoritative interpretation. And the obvious question then is, Why not construct this idealized author to be ideal? In other words, why not, as Ronald Dworkin advocates, ‘interpret' every law to be the best law it can be (Dworkin 1986)? And if we do not get the best possible law by assum­ing the lawmaker is writing in standard English, why not assume the lawmaker is writing in Schmenglish, a language in which the law would be ideal from the interpreter's vantage point? This natural progression leads to what is in effect the reauthoring of the law by the interpreter and constitutes a reductio of the position.

These four positions that textualists could be advocating, given the impossibility of intention-free textualism, seem to exhaust the pos­sibilities. Positions two, three, and four are difficult to defend, even if they can be made determinate. Position one is at least a tenable posi­tion, but it makes it possible that the authoritative meaning of the law will differ from the meaning we know was intended, and intended even by all the authoring lawmakers. Thus, a policy-constrained interpreter of the position one variety will not be a faithful agent of the lawmaker.

Irrespective of their normative attractiveness, none of these textualist positions requires any special form of legal reasoning that is differ­ent from ordinary reasoning. The first three positions merely require ordinary intentionalist reasoning, albeit intentionalist reasoning that is constrained in terms of evidence (position one) or that is hypothetical (positions two and three). Position four, on the other hand, requires purely moral reasoning, though, as we have said, it can hardly be deemed a method of textualist interpretation as opposed to a method of text creation.

4.2