But even Empson at his best is difficult, hard to follow or impossible to keep up with, and so he is tiring. A few pages at a time, for most readers, may be fine. Much more than that, and the need to justify Empson's entire way of operating may seem more and more appropriate. Sooner or later, because of the strenuousness of his methods, and no matter how much one from time to time admires them, the question of their justifiability will arise. When it does, the relevance we demand may be of two different kinds. One of these involves the issue of whether what Empson is doing ought to be done at all, but even though that is the larger problem, I want to postpone it in favor of a more local one: how far ought a professional critic to go in "close reading" a text? What ought to be the limits of applied imaginative association? Such questions will be all the more urgent, once again, if one has been distressed at the far-fetchedness of many contemporary close readings: what is to prevent someone from saying just anything about a poem?

The terms in which Empson himself addresses this question demonstrate that it is not nearly so problematic for him as it has become for us. He assumes not only that there are, or will be, limits to the range of association applicable to a piece of literary language but also that our act of reading is itself an effort to discover what these limits are. We shall naturally want to observe, for instance, what "particular section of the English language is being used": "Evidently such a section is defined by its properties rather than by enumeration, and so alters the character of the words it includes" (4). This determination of what "vocabulary" we are confronting will simultaneously indicate to us what responses seem outside the limits of a particular text: "The process of becoming accustomed to a new author is very much that of learning what to exclude in this way" (5). And further:

. . . I have considered the "implications" of sentences so far mainly as what they take for granted, as what must already be in mind if they are to be suitable. The stock example of this is, "Have you stopped beating your wife?", which claims to know already that it has been your habit to do so. A complementary sort of implication may be defined: what must not be in mind if the sentence is to be suitable, what it leaves vague, or is not thinking about, or does not feel. The negative here assumes you might expect this particular thing to be in mind, because otherwise you would not have thought of it as an implication. You might think it lessened the importance of a negative implication that one is only conscious of it if its assumption may chance to be questioned, and most people are conscious that they, therefore, can to some extent impose what they assume.
(28-29)

Besides his rare intelligence and his even rarer honesty, it is one of Empson's most remarkable characteristics that an issue like this never becomes theoretically problematic or causes him seriously to doubt his methods; his frequent self-questioning, in fact, usually turns into a further investigation of the text in question. From the passage I have just quoted, for instance, he continues by showing how a literary process may make use of "negative implication" to do something, to achieve an effect, to fulfill a purpose. He does not for a moment doubt, either, that a relatively single-minded purpose is always to be inferred.

Thus there is in Empson's account of the functioning "ambiguity" more generally none of the anxiety about the radical undecidability of significances that we encounter these days. He is clear that when we are reading a piece of literature we are dealing with a "unit":

Evidently all the subsidiary meanings must be relevant, because anything (phrase, sentence, or poem) meant to be considered as a unit must be unitary, must stand for a single order of the mind. In complicated situations this unity is threatened: you are thinking of several things, or one thing as it is shown by several things, or one thing in several ways. A sort of unity may be given by the knowledge of a scheme on which all the things occur; so that the scheme itself becomes the one thing which is being considered. More generally one may say that if an ambiguity is to be unitary there must be "forces" holding its elements together, and I ought then, in considering ambiguities, to have discussed what the forces were, whether they were adequate. But the situation here is like the situation in my first chapter, about rhythm; it is hard to show in detail how the rhythm acts, and one can arrive at the same result by showing the effects of the rhythm upon the meaning of words. (234)

Of the grounds for "unity" presented here, the first one — that something is unitary when it is "meant to be considered as a unit" — will seem the most dangerous to a modern theorist, especially when he notes how often Empson declares how a line or a phrase must be considered "intended." For much of our recent thinking holds that any recourse to some a priori intention to define the limits of meaning for a text amounts to a jump from the frying pan into the fire, that it is merely a shift from one sort of text to another, even more undecidable one. But there is a much more persuasive and compelling reason than this for questioning the capacity of authorial intention to govern the meanings or establish the unity of a given expression, which is that intentions conceived as anterior to expression and internal to a writer's or speaker's mind are not describable as verbal phenomena. Philosophical and phenomenological investigations of our verbal behavior over the last several decades have shown that the inferences concerning "intentions" in speaking situations are characteristically drawn from the external, expressed features of those situations, and therefore that describing them in psychological form is misleading. It is not, therefore, that a shift to the internal "text" of intentions is improper, but that it is illusory: there is no such text.

But here it must be said that in his actual critical practice Empson seems always within the bounds as we nowadays mark them. His frequent appeals to how a phrase or line is "meant," to how a passage is "supposed to be taken," are always, either initially or eventually, derived from the details of the expression itself, from his fine perception of the particular semantic and syntactic properties of the text against his extraordinary awareness of the background of linguistic possibilities, his knowledge of usage itself. His interest in authorial intentions, like his fondness for alluding to the mental or emotional conditions of his authors, thus does not indicate any need for extra-textual verifications of his readings. Most often its seems motivated rather by a desire to personalize the writer, so as to get closer not to him but to the writing he has done: "Milton is extremely cool about the matter; one is made to sit with him pleasantly in the shade, all day long, needing no further satisfaction; it is delightfully soothing to feel that the devil is all the time falling faster and faster" (12). His personal and even psychological references to authors are in this way usually, if not always, allegories for the way it feels to read their writing, ways of making vivid his encounter with their expressions.

Still, however charming such allegories may be, or however right the perceptions they focus may seem, because they are not really appeals to intentions for the purpose of determining meaning, they leave the theoretical justification for Empson's more extravagant readings up in the air, and the source of the putative unity of the text undetermined. "A sort of unity," he says in a passage I have already quoted, "may be given by the knowledge of a scheme on which all the [conflicting implications or ambiguities] occur; so that the scheme itself becomes the one thing which is being considered." Taken simply, this is to say that one may recognize a context that holds textual disparities together. (In a more complicated way, of course, it suggests that even the context itself may be under scrutiny.) Such a recourse is logical enough to our contemporary thinking. If the way to take a text is not to be fixed by reference to some anterior intention, it must be inferred from the expressive circumstances of that text. If meaning is not intentional, it must be contextual.

But as we know, there is a difficulty here as well. Current theorizing has served notice that as convenient as "context" may seem for deciding questions of meaning, it is not reliable. Jonathan Culler provides a usefully simplified account of the problem when he says that "one could . . . identify deconstruction with the twin principles of the contextual determination of meaning and infinite extendibility of context" (On Deconstruction, Ithaca, NY, 1982, p. 215). This "extendibility" — Derrida's "unsaturability" — means, as Culler specifies earlier in his account, that context does not provide us certainly with the limits that we may require for determining significances, just because it is "boundless":

Context is boundless in two senses. First, any given context is open to further description. There is no limit in principle to what might be included in a given context, to what might be shown to be relevant to the performance of a particular speech act. This structural openness of context is essential to all disciplines: the scientist discovers that factors previously disregarded are relevant to the behavior of certain objects; the historian brings new or reinterpreted data to bear on a particular event; the critic relates a passage or a text to a context that makes it appear in a new light. Striking instances of the possibilities of further specification of context, Derrida notes, are the displacements permitted by the notion of the unconscious . . . .

Context is also unmasterable in a second sense: any attempt to codify context can always be grafted onto the context it sought to describe, yielding a new context which escapes the previous formulation. Attempts to describe limits always make possible a displacement of those limits . . . (123-124)

The first of these two "senses" seems to me the only one worth taking up, and the short answer to it is simply that — as the phrase "in principle" may reveal — it constitutes a misunderstanding of what context is, and how it works, in actual verbal practice. For in such practice context is anything but a matter of principle. To recognize a context is to recognize a particular and historical frame within which some inferences are probable and others are not. What "historical" means here, furthermore, is, on the one hand, that contexts are remembered — they are specific verbal situations that one has learned — and, on the other hand, that contexts like all acts of expression are temporary: the limiting power, the bounds that they provide, are for the moment only. To exempt contexts from these temporary conditions, to consider them theoretically and hypothetically, is to cease to consider them as contexts which of course alters the status of the rules that they can supply, and very much exaggerates their capacity for manipulative modification.

The "theoretical" imagination of context, then, entails its own difficulty, for it is only in theory that an extension of context can occur anytime and instantaneously, only in hypothesis that one need only think about he possibility of such changes to bring them about, and only in the abstract that contexts are "unmasterable." But what seems actually unmasterable here is the deconstructionist's appetite for controlling all hypothetical possibility, and his complacent implication of all our linguistic behavior in his inevitable failure to do so. For contexts, like language more generally, are not mastered by abstract dictation. Mastery in relation to verbal behavior, on the contrary, is much more provisional and particular than "theoretical," and necessitates a responsive disposition to recognize and to learn. The best formulation of the situation would seem to be Wittgenstein's: "All knowledge depends upon acknowledgment."

But even if contexts are not so untrustworthy as deconstructionists would have them, there remain questions as to how in practice they may be established and recognized. Culler's remarks about the way the "scientist" or the "historian" changes his mind slide right over all the particular ways that context might work in situations of discovery, over what actual imaginative sequences — with concomitant alterations in the status of context — these situations might involve. But his line about the critic who "relates a passage or a text to a context that makes it appear in a new light" — because it alludes to an apparently even vaguer and more variable set of cases — may make us ask again how far that critic should be allowed to go. What happens when one is operating not at the obvious center of a reading but at the less decidable fringes? What governs the range of imaginative association in such cases? What allows Empson to draw out the most dramatically heterogeneous inferences and still maintain that they are extensions of the text and thus governed into "ambiguities," when current theorists would claim them as ungovernable deconstructions of that text? We may understand this better by considering Empson's account of the seventh type of ambiguity, "which occurs when the two meanings of the word, the values of the ambiguity, are the two opposite meanings defined by the context, so that the total effect is to show a fundamental division in the writer's mind" (192). This would seem to be fertile soil for growing deconstructions, but as he proceeds we see why it is not:

One might say, clinging to the logical aspect of this series, that the idea of "opposite" is a comparatively late human invention, admits of great variety of interpretation (having been introduced whenever there was an intellectual difficulty), and corresponds to nothing in the real world; that -a.b is contrary to a for all values of b; that words in poetry, like words in primitive languages (and like, say, the Latin altus, high or deep, the English let, allow or hinder), often state a pair of opposites without any overt ambiguity; that in such a pair you are only stating, for instance, a scale, which might be extended between any two points, though no two points are in themselves opposites . . . . Or one might admit that the criterion in this last type becomes psychological rather than logical, in that the crucial point of the definition has become the idea of a context, and the total attitude to that context of the individual.

A contradiction of this kind may be meaningless, but can never be a blank; it has at least stated the subject which is under discussion, and has given a sort of intensity to it such as one finds in a gridiron pattern in architecture because it gives prominence neither to the horizontals nor the verticals, and in a check pattern because neither colour is the ground on which the other is placed; it is at once an indecision and a structure, like the symbol of the Cross. (192)

It is clear that for Empson something that "is at once an indecision and a structure" is quite a special case: his allusions to gridiron and check patterns show just how strict a balance must be achieved before he will allow it as such. And the contrary procedures of the deconstructionist sort, by which the "oppositions" in a text are discovered much more readily, are further illuminated by his suggestion that the idea of opposition itself is an artificial one conscripted to deal, or to evade dealing, with any "intellectual difficulty." Even more enlightening is a connection we may draw between his remarks about the check pattern's color and ground and the deconstructionist conception of the relation of particular expressions and their contexts to general linguistic possibility: "Wittgenstein's suggestion that one cannot say 'bububu' and mean 'if it does not rain I shall go out for a walk,' has, paradoxically, made it possible to do just that" (Culler, 124). For Culler, here, imagining that it is possible to say something is equivalent to saying it; reflecting upon verbal potentialities is enough to wake them up into life, into participation in an active linguistic sequence or exchange. But the possibilities that language holds are not present in particular speaking situations in this way. To suppose that they are is like thinking, in Empson's analogy, that all patterns are check patterns.

But of course they are not, and backgrounds do not mix so freely as this with the colors of figures upon them, and we should have to be very interested indeed in the idea of opposites to perceive foreground and background as so frequently intercommunicative and problematic. Or, as Empson suggests a moment later — though affably and without my misgivings — we should have to be interested in Freud, for whom the unstated "opposite" of an expression was so often more vivid than the expression itself. The essential justification, for Freud and Freudians, for enlivening the background of expressive behavior into a status equal and opposite to it is of course the invention of the "unconscious" as a constantly alternative context that doubles and divides the sense of this behavior. I have argued elsewhere that there is no sense to Freud's arguments on this point, and the unconscious so conveniently conceived is not really an unconscious at all. But here I want only to repeat that I consider deconstructionist problems with context, undecidability, and the like attributable in large measure to Freudian thinking.

Without this fascination with the opposite, and the resulting reversal of background potential and foreground acts of expression, some meanings, it seems a relief to say, will be more probable than others, and some contexts strong enough in their bounds to rule some implications in and others out. But still there may remain the most difficult cases, and Empson does not shy from them:

An ambiguity of the fifth type occurs when the author is discovering his idea in the act of writing, or not holding it all in his mind at once, so that, for instance, there is a simile which applies to nothing exactly, but lies half-way between two things when the author is moving from one to the other. Shakespeare continually does it:

Our Natures do pursue
Like Rats that ravyn downe their proper Bane
A thirsty evil, and when we drinke we die.
(Measure for Measure, I.ii.)

Evidently the first idea was that lust itself was the poison; but the word proper, introduced as meaning "suitable for rats," but also having an irrelevant suggestion of "right and natural," and more exact memory of those nowadays phosphorous) poisons which are designed to prevent rats from dying in the wainscot, produced the grander and less usual image, in which the eating of the poison corresponds to the Fall of Man, and it is drinking water, a healthful and natural human function, which it is intolerable to avoid, and which brings death. By reflection, then, proper bane becomes ambiguous, since it is now water as well as poison.

Ford is fond of the same device, possibly from imitation:

Giovanni. Now, now, work serious thoughts on baneful plots;
Be all a man, my soul; let not the curse
Of old prescription rend from me the gall
Of courage, which enrolls a glorious death:
If I must totter like a well-grown oak,
Some undershrubs shall in my weighty fall
Be crushed to splits; with me they all shall perish.

('Tis Pity, V.iii. end.)

Gall is first used as "spirit to resent insults," the bitterness which is a proper part of the complete man. ("We have galls": Othello, IV.iii.93.) By the next line galls have suggested oak-galls (the reactions of an oak to irritations), and the idea of proper retaliation is transferred to the power of falling on people, whether they are guilty of wrongs against it or not. But in between these two definite meanings the curious word enrolled seems a blurring of the focus; he is thinking of his situation itself, rather than either metaphor, and keeping up the metaphorical language rather as a matter of form.

A glorious death may be enrolled on the scroll of fame, so that the word could stand by itself; or, looking backwards, one may gain strength for a glorious death by being bathed in, sustained by, a spurt of bitterness, so that gall has been rent (now with opposite consequences) from its boundaries in the orderly mind, by being rolled in, or round about by, gall; or, looking backwards, it may be the oak itself which rolls down, both to death and upon its victims. You may say this is fanciful, and he was only looking for a word containing the letter "r" which kept up the style, but in that case it is these associations which explain how that particular word came into his mind. I do not claim that one should admire this turgid piece of writing merely because it is explicable. (155-56)