University of Virginia Library

Search this document 


  

expand section 
Textual Criticism and Deconstruction by G. Thomas Tanselle
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 

collapse section 
  
  
  
  
  
  

1

Page 1

Textual Criticism and Deconstruction
by
G. Thomas Tanselle

In recent years there has been an increasing tendency for literary critics to refer to literary works as "texts." In consequence, the term "textual criticism" has become ambiguous, some people regarding it as a synonym for "literary criticism."[1] Traditionally, of course, "textual criticism" has meant the scholarly activity of studying the textual histories of verbal works in an effort to propose reliable texts of those works (according to one or another definition of correctness). This activity has commonly been though of as preparatory to literary criticism—as if it were mechanical or objective, and as if its results were to be unquestioningly accepted as the starting point for interpretive discussion. The products of textual criticism, however, like all historical reconstructions, incorporate judgments and assumptions. Textual critics must try to understand the works they are editing (however one defines "understand") in order to make choices among variant readings and decisions about emendations; they inevitably engage in literary criticism. So-called literary critics, for their part, should recognize—though they rarely do—that they must evaluate the makeup of particular texts in the process of analyzing literary works. Any development that brings "textual critics" and "literary critics" closer together is to be welcomed. The problem with the recent equation of the words "texts" and "works" is that, while it brings the two terms together, it completely ignores the issues that textual criticism has always dealt with.

The distinction between the texts of documents (handwritten or printed, private or published) and the texts of works is basic to textual criticism. The effort to "reconstruct" or "establish" the texts of works presupposes that the texts we find in documents cannot automatically be equated with the texts of the works that those documents claim to be conveying. Literary critics who neglect, or are oblivious to, this point are not commenting on works but on the arrangements of words and punctuation that happen to have come their way in particular documents.[2] The critical movement widely known as "deconstruction" provides an interesting illustration. Those who have written from this point


2

Page 2
of view have been especially concerned with exploring the nature of literature, and it is therefore surprising that they have given so little attention to a matter so fundamental to the definition of reading and the understanding of the medium of literature.

Geoffrey Hartman, in the preface to Deconstruction and Criticism (1979), a collection of essays often thought of as a manifesto of deconstruction, continually uses the word "text" to mean "work." When at the outset, for example, he refers to "the great texts of our culture," he is not thinking of great documents—that is, landmarks in the transmission of historically important works—but of the works themselves. He identifies a central concern of the volume to be the "priority of language to meaning"; but he does not acknowledge the failure of the volume to consider an equally fundamental matter—the logic of accepting any sequence of words as it comes to us. Such acceptance implies a reverence for artifacts; otherwise the words or their arrangement could be freely altered. Yet even an interest in verbal artifacts as historical statements requires a questioning of documents, given the variability, from one document to another, of the texts—the arrangements of words—that purport to represent the same work. In the process of examining how a given word functions in a particular context, those interested in history and those not interested in history must alike ask whether it is the word that ought to be there.

In what follows I shall look at the use of the word "text" in the five essays of Deconstruction and Criticism (with a glance or two at some related writings)[3] and consider the implications of that usage for the arguments they make. Although Hartman says that the book is not "a manifesto in the ordinary sense" and although the five authors do not hold identical positions, they were willing to be published together under the title Deconstruction and Criticism, and we may therefore regard the essays as indicative of the "shared set of problems" (in Hartman's words) that characterizes the deconstructive approach.

I

The first essay in Deconstruction and Criticism is Harold Bloom's "The Breaking of Form" (pp. 1-37), an appropriate opening because (as the title suggests) it focuses directly on an operative principle of deconstruction, stated succinctly in the first paragraph: "in a poem it is not form itself that gleams or sparkles. . . . the lustres of poetic meaning come rather from the breaking apart of form." In the ensuing five pages, Bloom, speaking in Emersonian cadences, utters a powerful statement of the importance, cultural and personal, of what he calls "strong reading"—the activity whereby readers struggle with poems, fight with poems


3

Page 3
to create their own meanings. "Only strength is memorable," he says, for "only the capacity to wound gives a healing capacity the chance to endure, and so to be heard" (p. 5). Therefore "good poems must be combative"; "Reading well is a struggle," a "warfare" in which one fights to establish, during the "encounter," one's own interpretation.

A strong reader of Bloom's essay, excited and invigorated by this opening, will look forward to the development of these ideas in the remainder of the piece—and will, unfortunately, be disappointed. This disappointment has nothing to do with agreement or disagreement: a strong reader who disagrees with the ideas of a strong essay—or, rather, breaks the structure of those ideas apart and re-creates them in a new form—would not be expressing disappointment but rather respect for the intellectual challenge posed by the essay. Such disagreement would indicate the success of the essay, and one would have no reason in those circumstances to be disappointed by it. What would cause disappointment is a fundamental weakness, in its own terms, of the process of thought underlying the essay, a flaw that would render it a weak combatant.

The first hint of such a flaw in Bloom's essay occurs on the sixth page of its printed form, where the word "textual" first appears. Up to this point Bloom has spoken of "poems," "fictions," and "works" but has said nothing about "texts." A reader alert to the difference between works and texts will have wondered whether a strong reading of a poem does not begin with being critical or combative about the constitution of the text of the poem, whether "reading well" does not entail the questioning of the makeup of the text as well as of the meaning of the work; and the reader will have assumed that these matters are to be pursued as the essay develops. But the first reference to "texts" is disquieting: "Gnostic exegesis of Scripture is always a salutary act of textual violence" (p. 6). Out of context this remark would be taken by most readers, I think, to refer to the extracting of meaning from particular sequences of words (that is, particular "texts") found in particular documents. When, however, one reads this sentence in the light of what has led up to it, the meaning of "textual" is not so clear. The violence described earlier occurs in the clash between a "strong" or "creative" reader and a "combative" poem—a poem, a work of literature, not the text of an artifact. If Bloom in this sentence is still talking about the same kind of violence, are we to conclude that he is using "text" as a synonym for "poem" and "work"?[4] If so, then what word, we wonder, will he use to refer to the sequence of words found in a particular document? Or is he not going to recognize the distinction between such a sequence of words and a "work"?


4

Page 4

The disappointing answers to these questions emerge in the passage immediately following (pp. 7-8), in which Bloom discusses his belief that "a strong reading is the only text": "there are no texts," he says, "but only interpretations."[5] Since there manifestly are texts, in the sense of series of words handed down to us on pieces of paper or parchment,[6] he is obviously equating "texts" with "works" and asserting that works have no meanings independent of the interpretations of those who encounter them. But how do we know when we are encountering a work? How do we know that the texts we encounter in documents are reliable representations of works? We may legitimately not care whether they accurately reflect the author's, or anyone else's, intention; but if that is our position, it must be made clear. Because verbal works do not exist in physical form, as visual works do, there is no way of avoiding the question of the relation between texts and works, whatever names we call them by. I am of course using "texts" here to mean specific oral or written sequences of words and pauses (or marks of punctuation); but the issue I am raising is not one of definitions of terms. Bloom is free to use "texts" and "works" synonymously if he wishes, but he then needs another term to refer to what "texts" conventionally denotes; for he is not free, if he hopes to present a coherent argument, to ignore the distinction.

A natural opportunity to pursue this matter presents itself to Bloom at this point, as he takes up Gershom Scholem's work on Kabbalah. Scholem, explaining why "there is no written Torah here on earth," says (as quoted by Bloom): "Everything that we perceive in the fixed forms of the Torah, written in ink on parchment, consists, in the last analysis, of interpretations or definitions of what is hidden. . . . There is no written Torah, free from the oral element, that can be known or conceived of by creatures who are not prophets." How this passage fits Bloom's purpose is obvious. But what he might also have said is that it offers an analogue of the relation between every text and the work it purports to represent. Just as the Torah as work is not fully knowable through any of its written manifestations, so also are the documentary texts of all literary (and other verbal) works but imperfect guides to the works they attempt to transmit. And just as the only Torahs we can have are the interpretive results of the process of analyzing the physical evidence present in documents, so also the only texts that we can possibly have of all verbal works are the ones we construct from the evidence presented to us in documents (or, in some cases, in oral traditions), using our individual judgments to choose among variant readings or to insert new readings. Not only are the meanings of works subject to our interpretation (which may or may not be historically oriented), but also the very constitution of the texts of those works is the product of our interpretive judgment.


5

Page 5

One thinks, at first, that Bloom is going to make this point, for he says that Scholem's statement is true not only of "Text Itself" (Scripture) but also of "all lesser texts." Immediately, however, he places "poem" in apposition with "texts": ". . . true of all lesser texts, of all poems more belated than the Torah." Once again, then, "texts" are works. If, therefore, one substitutes "work" for "text" in his next paragraph, one has no difficulty following, indeed admiring, its ridicule of professors who believe "in the real presence of the literary text . . . in editions, definitive editions, upon which responsible commentaries might be written." Those who believe that works reside in printed books and who unquestioningly accept the texts of any editions, even scholarly ones, as if they were the works themselves, indeed deserve our criticism for misunderstanding the nature of the medium that verbal works employ. But we still have a problem here, for Bloom fails (whether or not the failure is induced by his use of the word "text") to include in his formulation the indeterminacy of texts (arrangements of words) as well as of meanings (interpretations of the significance of words).

He proceeds to say, in commentary on the foregoing, "I only know a text, any text, because I know a reading of it," and elaborates, "I do not know Lycidas when I recite it to myself, in the sense that I know the Lycidas by the Milton." A primary reason for his not knowing "Lycidas" in this sense is that he cannot be sure how close his text of it is to the one Milton intended (or to any of the ones Milton may have intended at various times). But this reason is not what Bloom has in mind. He means that he can only know "Lycidas" through his, or someone's, interpretation of it and that there is no one Miltonic meaning that stands independent of those interpretations. But he does not say how he comes by the particular text he is following or what difference it might make to his interpretation if he followed a different one. He is therefore behaving in a fashion much more similar than he suspects to that of the professors he castigates and pities, for by ignoring the textual question entirely he is tying himself just as uncritically as they to some fixed text (or texts). He is tacitly endorsing the notion that editing is distinct from criticism, failing to see that editing involves critical judgments and that criticism must question the constitution of texts if it is to be concerned with works rather than documents.

Bloom's total neglect of the relation of texts to works—or, put differently, of the mode of existence of verbal statements—undercuts his eloquent arguments and reduces them to incoherence time and again. In the next paragraph, for instance, he offers a prose-poem about words, words that "refer only to other words," words that "will not interpret themselves"—and, finally, "words lying against time" (p. 9). But where


6

Page 6
are the words that lie against time, carrying on "the verbal agon for freedom"? Everything in the paragraph suggests that he is talking about verbal works, verbal constructions, not documents (printed or handwritten) with their multiplicity of errors.[7] In the context of the essay, however, where no distinction between works and texts is made, the passage loses the force that it rightly should have, for readers will be at a loss to know how one locates those words that, lying against time, can serve to occupy the independent mind in its struggle for a sense of freedom. How much richer the passage would be if it resonated with the ironies inherent in every attempt to preserve verbal works: poems survive—to the extent that they do survive—only through remembered, written, or printed versions, all of which must, in one degree or another, be suspect. We constantly encounter texts lying against time, in the documents that come to us from the past, and we are free to decide that the words in those texts are the words we wish to be concerned with. But they cannot automatically be equated with the words that constitute works of poets from the past. Whatever our attitude toward the past, we cannot be moved by Bloom's phrase "words lying against time" because it reflects a naïve oversimplification of the process it aims to help elucidate.

These comments on the first nine pages of Bloom's essay provide, I believe, a way of reading the rest of it. References to "a specific poem" (p. 10), to reading a poem "properly" (p. 16), and so on are simply vague, rather than alive with the intriguing indeterminacies that accompany any effort to locate a verbal work. A place does occur in the essay where a distinction between "text" and "poem" is made, but it is not the one I am making: in discussing the "revisionary ratios" that a poet uses against a predecessor, Bloom says, "As text, a ratio names intertextual differences; as poem it characterizes a total relationship between two poets, earlier and later" (p. 19). The relation of work to documentary text remains untouched.

In the second part of his essay, Bloom examines a "proof-text," Ashbery's "Self-Portrait in a Convex Mirror"; and although he makes extensive quotation from some (unspecified) text of the poem, he never once questions that text as a representation of the work. Presumably he is using the text in Ashbery's 1975 volume of the same title, for one of his quotations includes a line that varies between the 1975 text and the August 1974 Poetry magazine text, and the reading in his quotation is that of 1975 book.[8] Furthermore, the division into verse-paragraphs in the text Bloom is following is the same as in the 1975 book. Bloom says (p. 27) that the first verse-paragraph closes with the line "Affirmation that doesn't affirm anything"—as it indeed does in the 1975 text (p. 70),


7

Page 7
but not in the 1974 text, where there is an earlier paragraph break (53 lines earlier, after the line beginning "Its hollow perfectly:"—p. 69, line 17, in 1975). This matter is of some moment, since the division into six verse-paragraphs is a "happy division" for Bloom, allowing him to associate the six sections with his six-part "apotropaic litany of evasions or revisionary ratios" (p. 23).[9] At another point in the essay, however, his quotation contains a word that differs from both the 1974 and the 1975 texts, raising the question whether he is following some third text, if such a text existed when he wrote.[10] More likely the word in his quotation is simply a typographical error, but one cannot be sure. Nevertheless, this error (if it is one) does illustrate the point that all texts must be questioned if our interest is not in random texts but in the works they are intended to represent.

Just as Bloom's quotations may contain errors, so may the 1974 and 1975 texts that Ashbery authorized.[11] The few differences between them do not seem particularly significant, but one cannot rule out the possibility that some readers would find them significant. Besides, however trivial the differences may be,[12] can we assume that the 1975 readings are always the ones Ashbery wished? Is not the possibility that Ashbery made the changes itself significant? And may there not be other places, alike in both texts, that do not accurately reflect his intention? I am not suggesting that Bloom should explicitly have taken these questions up in his essay; but he should give evidence of having thought about them. As long as he is treating the poem as by a particular person (and he does comment on Ashbery's "career" and allude to other poems by Ashbery), he cannot avoid being concerned with "Self-Portrait" as a work, not merely as a text that is located in a given copy of a printed edition. His not doing so casts doubt on the whole undertaking: the textual variants in this instance may be minor, but the failure to recognize the issues is a major conceptual flaw.

II

Paul de Man's essay "Shelley Disfigured," which follows Bloom's essay in Deconstruction and Criticism (pp. 39-73), exhibits the same blurring of the essential distinction between texts of documents and texts of works. The first endnote might make one think otherwise, for in it de Man identifies the scholarly edition that is the source of his quotations and refers to another piece of textual scholarship.[13] But that note is attached to the second sentence of the essay, and the opening sentence has already revealed de Man's confusion: "Like several of the English romantics' major works The Triumph of Life, Shelley's last poem, is, as is well known, a fragment that has been unearthed, edited, reconstructed


8

Page 8
and much discussed." Up to the word "poem" the sentence is fairly clear: it refers to "major works" and identifies "The Triumph of Life" as a "poem"—that is, a "work" (though whether he wishes us to understand that he considers it a "major work" is not clear).[14] He then states that the poem is a "fragment"—a word that does not necessarily pose a problem, since a work can be regarded (by its author or by others) as unfinished. When, however, he says that the fragment has been "unearthed" and "edited" and "reconstructed," the sentence becomes incoherent. Though "unearthed" is a figurative term here and does not literally refer to removing from the earth, it does surely imply the uncovering, the discovery, of a pre-existing tangible object—a manuscript on which a text has been inscribed. Given the medium of verbal works, one cannot "unearth" such a work but only a physical document containing a text purporting to be the text of a particular work. Thus the "fragment" that has been unearthed is a document, not a "poem"; the meaning of "fragment" required to go with "unearthed" is different from the meaning required by the earlier part of the sentence (in which "Shelley's last poem" is "a fragment"). The next verb, "edited," makes sense if the reader will excuse a slight imprecision, for what has been edited is not literally the document (or "fragment") but the text of the document. The third verb, "reconstructed," is a different matter, however, since the goal of the editorial reconstruction referred to is presumably some version of the work (not the text of some lost fragmentary document)—and therefore the required meaning of "fragment" has swung back to "work" or "poem." That the reader must continue to switch the meaning of "fragment" as the sentence progresses is a sure indication that the writer of the sentence had not carefully considered the relation of literary works to the documents that attempt to transmit them.

In the remainder of the first section of his essay, de Man continues to use the archeological metaphor, speaking of " 'digging in the grounds for the new foundations' "—that is, using "history as a way to new beginnings" (p. 40). But the confusion of the opening sentence undercuts everything else: are the foundations provided by history to be thought of as the texts that survive in documents or as the works that can be derived from those documents? The latter may subsume the former, but the question is nevertheless a central one, for what it really asks is where in the process we conceive individual judgment as entering. De Man's bias is evident when he says that our "curiosity about antecedents has produced admirable philological results," allowing for "the establishment of texts whose unreliability is at least controlled by more reliable means," but that "the questions which triggered all this industry remain more than ever in suspense: What is the meaning of The Triumph of


9

Page 9
Life, of Shelley and of romanticism?" To call the philological work (that is, the scholarly editing, the "establishment" of a text) "admirable" but nevertheless mere "industry" that does not touch on the main issue of "meaning" is to suggest that textual scholarship and literary criticism are strictly separate and that the former—somehow scientific or objective—produces the materials for the latter, which engages in assessment and discrimination. There seems to be no recognition here that even the editor who wishes simply to transcribe the text of a document, to "establish" what in fact is in the document, must try to understand the context of each word and continually makes judgments that reflect one understanding or another of what message is being conveyed. The editor who does not claim to be transcribing a documentary text but who instead aims at reconstructing from it the text intended by its author is clearly making still more judgments. The two goals are distinctly different, but both rely on critical judgment. If textual criticism is thus truly "criticism," literary criticism cannot simply accept its conclusions but must examine them as part of the process of reading.[15] Attempting to understand the "meaning"—in the narrowest and the broadest senses—of a communication from the past cannot be divorced from a questioning of the words that are thought to make up that communication.

De Man's handling of this point is unfortunately just what the reader would expect on the basis of the opening sentence. He actually touches on a much more illuminating line of inquiry when he asks (p. 41), "Is the status of a text [like] the status of a statue?"[16] If he had pursued this question, he might have extricated himself from his confusion, for he would have recognized that the medium of literature, unlike that of sculpture, is not tangible and that no tangible rendering of a piece of verbal communication can be the work itself. Each one is, if you will, an unearthed fragment—or, at least, possibly a fragment, because the wholeness of each document is a matter of conjecture, to say nothing of the wholeness of the work represented by its text. Instead of developing this line, however, de Man brushes it aside: "But there are more economic ways to approach this text [that is, the work called "The Triumph of Life"] and to question the possibility of establishing a relationship to Shelley and to romanticism in general." The economy of his route proves illusory in the end, for the path never again comes as close to a vista that would display the relation of language—the medium of literary art—to the documents that give it physical form.

The functioning of language and reading do form the primary concern of the essay; and one must feel regretful in reading it to recognize how de Man's failure to distinguish works from texts not only weakens his argument but robs him of insights that would have reinforced the


10

Page 10
general direction his thought was already taking. "The Triumph of Life," rendered fragmentary by Shelley's death, provides him with a "mutilated textual model" exposing "the wound of a fracture that lies hidden in all texts" (p. 67). Our understanding of this wound is obscured, however, by his equating versions of works with the texts that survive in documents. This confusion is revealed early in the essay when, in introducing his discussion of the changing role of Rousseau in the poem, he speaks of the "unearthed fragments of this fragment, the discarded earlier versions" (p. 41). Versions of works are works as they stood at particular moments, and version thus exist only in the medium being employed, which in this case is language. One therefore cannot assume that the text of any document, even one in the author's hand, is a faithful representation of a given moment in the evolution of a work, for it may contain not only slips of the pen but also words and phrases already superseded in the author's mind but mechanically copied from a previous document and not altered before the new document was laid aside. To say that the "last available text" of "The Triumph of Life" was "frozen into place by Shelley's accidental death" (pp. 42-43) is true only if one takes "text" to mean "documentary text"—though there would in that case be little point in making the statement, since every documentary text is "frozen into place" by its placement in a physical object, whether or not the person who placed it there has died. But if "text" in this sentence means "text of a work" or "text of a version," as in de Man's usage it is likely to mean, the sentence cannot be true: Shelley's death cannot stop the speculation about what the text of "The Triumph of Life," or some version of it (such as the "last available" one), consists of—speculation that would have been as relevant while Shelley was alive as it is now that he is dead. The text of the work, or of versions of it, is never frozen because it is always the product of our critical judgment.

Over and over again de Man speaks of the power of words, words that "cannot be isolated from the deeds they perform" (p. 49), sometimes expressing himself with eloquence:

And to read is to understand, to question, to know, to forget, to erase, to deface, to repeat—that is to say, the endless prosopopoeia by which the dead are made to have a face and a voice which tells the allegory of their demise and allows us to apostrophize them in our turn. No degree of knowledge can ever stop this madness, for it is the madness of words. (p. 68)
This emphasis on words underscores the necessity of questioning the makeup of whatever sequence of words we are examining; what is left unsaid here is how difficult—in the end how impossible—it is to determine what words we ought to be examining. De Man says, "In Shelley's absence,

11

Page 11
the task of thus reinscribing the disfiguration now devolves entirely on the reader" (p. 67), but his statement does not go far enough: Shelley's absence is irrelevant, and what devolves on each serious reader includes the determination of the text to be read. At the end we are told, "Reading as disfiguration, to the very extent that it resists historicism, turns out to be historically more reliable than the products of historical archeology" (p. 69). This statement would have carried more weight if it had grown out of the paradox of textual historicism: if one is interested in the text of a work as intended at a particular time by one or more of its producers, one will come nearer the goal through a critical reconstruction, based on surviving evidence and one's knowledge and judgment, than through the acceptance of any of the texts that happen to have survived in written or printed documents. This insight is not unique to textual studies, of course: it reflects a basic point of view about how we come to know, or think we know, the past. De Man was clearly attuned to this view, even if he did not think there is a past to know; but he does not, in this essay, reveal any understanding of its application to the establishment of texts. Since the essay is concerned with texts and works, with verbal productions from the past, that weakness is fatal.

It may be worth turning for a moment to an essay not in Deconstruction and Criticism, for de Man's naïve conception of editorial "archeology" shows up in extreme form in his essay on "Heidegger's Exegeses of Hölderlin" originally published in French in 1955 and republished in English in 1983 (in an enlarged edition of Blindness and Insight).[17] Because de Man was content to allow this early piece to be included, with no additional commentary, in a late book, one is perhaps justified in concluding that it may represent his view of textual scholarship throughout his career.[18] He begins the essay with a discussion of editorial matters because the reader, he believes, "must bear in mind the special circumstances of the editing and elucidation of Hölderlin's works" (p. 247). If the astonishing statement he makes soon thereafter were true, the circumstances would be special indeed: "More so than for everyone else, the reliability of his text is all important" (p. 248).[19] The 1906 Hellingrath edition of Hölderlin, the one Heidegger used, has been superseded, de Man points out, by the Beissner edition, and de Man's praise of the new edition fully reveals the thoughtlessness of his approach to textual criticism:

It is one of the great achievements of modern scientific philology. By drawing upon the most proven methods (detailed study of the sources and of historical and biographical references, internal comparative references, syntactic explanations, study of formal metrics, etc.) as well as upon some modern technical processes (study of paper and writing with the aid, I am told, of slides

12

Page 12
of enlargements of the manuscript) Beissner has produced the irreproachable critical edition, something that, in the case of Hölderlin, was at once necessary and most difficult to achieve. (p. 248)
If "critical editing" is "scientific," employing "proven methods" and "technical processes," then why is the result "difficult to achieve"? Perhaps the quantity of data makes the task time-consuming, but that is surely not the kind of difficulty de Man has in mind. The difficulty arises because the techniques listed as examples of "proven methods" and of "modern technical processes" all require judgment: decisions are called for at every turn. What emerges from such a critical process can be "irreproachable" only in the sense that it follows from a responsible way of proceeding. In the same sense a critical essay could be "irreproachable," but it would not necessarily be "correct" or earn universal agreement. De Man's comments show that he does not take seriously the word "critical" in "critical edition" and imagines that editors can provide objectively established texts for literary critics to exercise their judgment on.

Indeed, he praises Beissner for claiming as much. But he immediately sees a drawback to this "prudent philological modesty": Beissner is "forced to leave unresolved a number of issues, including some at the level of the text establishment." So the establishment of texts does, after all, require judgment. Is it not possible, therefore, that some people will disagree with Beissner about which cruxes fall into the resolvable, and which into the unresolvable, categories? Can the literary critic ever afford to be uncritical in accepting the results of a textual investigation? De Man asserts that Heidegger decided textual questions "in the name of the internal logic of his own commentary" (p. 248), and he concludes: "it cannot be denied that the exegete capable of providing a coherent and responsible interpretation has the right, indeed the obligation, to decide according to the conclusions of his interpretation. . . . Everything rests, then, on the intrinsic value of the interpretation" (p. 249). Of course it does—for critical editors as well as "critics"—as long as the interpretation does not contradict any "facts" that can be agreed on. Editors, like everyone else interested in the past, try to establish a framework of facts not subject to interpretation (though they have exercised judgment in arriving at it) and then supplement it with interpretation. The boundary line between fact and interpretation is forever indistinct. In his naïveté about scholarly editing, de Man was not distinguishable from large numbers of his professorial colleagues.

III

Jacques Derrida's contribution to Deconstruction and Criticism (translated from the French by James Hulbert) is by far the longest (pp.


13

Page 13
75-176) and occupies the center of the book through its content as well as its physical position. In it Derrida is characteristically playful, and the game he plays with the title of the piece makes any simple citation of its title inaccurate. The game is a productive one, however, leading directly to Derrida's central point—and also, in my view, illuminating a central flaw in his process of thought. The title as given in the table of contents is "LIVING ON · border lines". At first one might assume that the typographic design of this volume calls for subtitles to be placed in small capitals following a centered dot. But then one notices that the listing for the next essay, the only other one with what appears to be a subtitle, consists entirely of large capitals and has a colon between the two parts of the title. A distinction is being made, and it is made again, though less effectively, in the running titles of the two essays. The running title of the Derrida piece is "LIVING ON: Border Lines", in which the colon rather confuses the issue; but when one sees that the running title of the next essay has the same sized capitals following its colon, one recognizes that the designer is trying to reflect the differing status of the "subtitle" in the two cases. That the words "Border Lines" are not a conventional subtitle is apparent on the first page of the essay. At the head of the essay there is only the title "Living On"; near the foot of the page (above the seventh line up from the bottom) a rule runs across the page, and the text below that rule is headed "BORDER LINES." Those two words obviously form the title of a companion piece of writing that runs along the bottom of each page of text:[20] two related compositions are proceeding simultaneously.

One does not have to read far into the essay, or to know anything about Derrida's other writings,[21] to see what he is suggesting by this scheme. Since the essay concerns the breaking down of borders, limits, boundaries, he is providing the essay with an indistinct border, edging it with a layer of words, related to the essay but moving out from it—a border implying that borders are only links to something else, one border following another, moving ever outward until the whole world is encompassed. The device of a secondary essay running along the lower part of each page raises, however, some problematic questions. Is Derrida saying that his composition is a visual as well as a verbal work? Does he consider readers' constant awareness of a line across the page to be an integral part of what he is communicating? Is he creating for readers the visual experience of finding that, wherever on the page they are looking, there is text on the other side of the line as well? That is to say, if his essay were to be set in type anew for inclusion in another volume of essays, and if "Border Lines" were on this occasion printed as a block of text at the end of the main piece (in the way that footnotes in one edition


14

Page 14
of a work sometimes become endnotes in another), would the meaning of the work be altered?[22] Would the work, indeed, have become a different work? If so, how much of the present arrangement is a part of the visual form of the work? Is it essential to have 27 or 28 lines of type above the division (except where there are inset quotations) and seven below, except on the last seven pages? Must the typeface remain the same? Can one have a visual work in which some of the visual features are not integral to the work?

I assume Derrida would be pleased that his piece gives rise to such questions. He would take them, I imagine, as evidence that his double composition stimulates thinking about the nature of verbal and visual art and illustrates, in this additional way, the indeterminacy of dividing lines. Perhaps it does, but only on a superficial level, for a more considered analysis of the visual aspect of the piece leads one to ask how carefully its implications have been thought through. The questions posed above are of course potentially fruitful, but here the reader raises them out of puzzlement rather than insight. The reader who has previously given thought to the nature of the medium of verbal works will begin to suspect, reading this piece, that the author's use of spatial metaphors springs more from confusion about that medium than from a constructive playfulness. Derrida says that "what used to be called a text" (p. 83) had limits that separated it from the rest of the world; but "in the last dozen or so years" these boundaries have been overrun, so that a "text"—"what I still call a 'text'" (p. 84)—is no longer to be differentiated from "everything that was to be set up in opposition to writing." The idea that verbal works shade into everything else—are parts of the world rather than statements about it—is worth exploring thoughtfully and carefully. But Derrida cannot do the concept justice because his approach to it is incoherent, as an examination of his metaphors of limits and edges will reveal.

This incoherence is manifested in the shifting signification of his references to the physical boundaries of works, for these boundaries are sometimes presented as analogical and sometimes as literal. When, for example, he speaks of "the referential realm outside the frame" (p. 83), he is appropriately using a conventional metaphor to describe what he considers an outmoded way of looking at a verbal work. And he is using this metaphor in the conventional way: the "frame" is the intangible boundary that defines the beginning and ending of a work and has nothing to do with any physical frame that may or may not be present on the pages of the book containing a text of the work. Similarly, when he refers to "all those boundaries that form the running border of what used to be called a text," the "border" he has in mind is clearly not a physical one. "Border" and "frame" are metaphorical, likening something abstract to


15

Page 15
something concrete. When, in the same sentence, he mentions "margins," the word is particularly meaningful, because "margin" is the word, rather than "frame" or "border," generally used to refer to the physical boundaries of texts in books. It is still metaphorical: he is saying that just as pages of texts have margins, so verbal statements or works were formerly thought to have their edges also. The word "margins" is simply a particularly apt synonym for "boundaries" in this context.

This progression of thought is shattered in the next sentence, however, when he describes the new meaning of "text" as "no longer a finished corpus of writing, some content enclosed in a book or its margins, but a differential network, a fabric of traces referring endlessly to something other than itself, to other differential traces" (p. 84). To speak of the "content" of a work being bounded by the physical dimensions of a book is not to engage in a productive analogy but to mix the two levels of the metaphor. In rejecting the notion of "some content enclosed in a book or its margins," Derrida is rejecting something that does not exist. To be sure, there are people who think of verbal works (and works in other media) as separate, in one way or another, from the rest of life; but one cannot effectively refute their position by claiming that what they see as separate is "content enclosed in a book or its margins." Perhaps some confused thinkers do fail to distinguish between verbal works, which are intangible, and the particular texts of those works conveyed to us in physical objects; but Derrida cannot possibly mount a successful argument for the boundary-less nature of verbal works if he shares with some of those he hopes of convert a misunderstanding of the relation between texts and works.

The point that I believe Derrida wishes to make could be more coherently expressed as follows: "Just as books, being physical objects, have edges that mark the limits of the space they occupy, so, in the view of some people, do verbal works (of which the texts in books are evidences) have boundaries that set them off from other experience; but, as against this view, some recent thinkers have come to believe that verbal works are part of an endlessly interconnecting and all-encompassing network." Put in this way, the statement rises above the confusion of those who equate works with physical texts; indeed, it corrects their error in the process of making the more inclusive distinction between those who see verbal works as having boundaries (however they arrived at their position) and those who do not. The physical metaphor can thus enhance the argument; but Derrida's handling of it undermines the argument by revealing a conceptual flaw at the base of it.

Once this problem has manifested itself, we see that it pervades the essay. The next sentence provides another particularly telling instance.


16

Page 16
Derrida says that "the text overruns all the limits assigned to it so far" and defines those limits as "everything that was to be set up in opposition to writing (speech, life, the world, the real, history, and what not, every field of reference—to body or mind, conscious or unconscious, politics, economics, and so forth)." The appearance of the word "writing" here is unfortunate. Derrida is using "writing" to mean "the kind of work normally conveyed in written form," not "the physical presentation of words," although the latter is what is suggested by the juxtaposition with "speech." But surely he would classify as "writing" a lecture delivered without notes: it would be a formal verbal composition, not conversational everyday "speech." His concern, in other words, is not with the texts that appear in particular documents but with the works they reflect; he is not claiming that the conventional view contrasts "life, the world, the real, history, and what not" with printed pieces of paper but with the contents of verbal works. Yet the blemish of using the words "writing" and "speech" is more than a temporarily misleading infelicity; it is an index of his failure to recognize the distinction between texts and works. As Derrida knows, our language will betray us in the end.

A few lines later he points out that what the new approach does is "to transform the world into a library." Again, the physical metaphor reveals more than Derrida imagines, for it brings into the open a shortcoming of his own. The world is of course a "library" in that it is full of physical objects that we can "read": all artifacts can be examined for clues to their own production history and to their role in the lives of their former owners. When read in this way, books can provide information (as analytical bibliographers have shown) about the workings of printing shops and the habits of compositors, proofreaders, and press editors; and books, through their paper, typography, and format, further tell something of the social status of the authors and genres represented. But Derrida is not thinking of reading books in this way; what he means by a library is clearly a collection of works, not recognizing that what a library in fact contains is the material out of which we can attempt to reconstruct works. A library—like the world—contains physical objects; and the inked letterforms that one finds within the special objects in libraries constitute documentary texts, each of which is an attempt to report the text of a work. How successful each book text is in reflecting a work is a matter for investigation and informed judgment. Libraries do not contain authors' works in the same way that museums contain artists' works, since visual artists' works are tangible and can be contained within a physical space, whereas verbal artists' works are not tangible and can be represented physically only by instructions (the accuracy of which must always be questioned) for their reconstitution. If Derrida believes


17

Page 17
that breaking down the barriers between verbal works and the rest of life makes the whole world a library, he is using a metaphor that undercuts what he wishes to say.

The problems so evident in this key passage reverberate throughout the long essay. Even when Derrida talks about variant versions of Maurice Blanchot's L'arrêt de mort (as on pp. 101-102), he concerns himself only with the meaning of "version," with the relation of versions to works, and is not led into the inextricable question of the relation of words on the page to the works they attempt to transmit.[23] One feels that he is approaching the distinction between texts and works when he cites Blanchot's statement that the "'narrative voice'" utters a work from "'the placeless place where the work is silent'" (p. 104), but he does not seize this obvious opportunity. And in the "Border Lines" he focuses on translation: "the problems that I wished to formalize above all have an irreducible relationship to the enigma, or in other words the récit, of translation" (p. 89). Pondering what translation can be thought to accomplish would naturally provide a way of embarking on a consideration of the nature of verbal works; but instead of moving from such works to their texts, Derrida moves in the opposite direction, from works to the thought lying behind them. The idea of "content" that "does not touch the borders of language" (p. 95) takes us away from the concept of work, for a work, as distinguished from thought, is tied to a medium (which can be visual, aural, verbal, and so on). Just how a body of thought, or content, is related to individual expressions of it in different media is a philosophical question worth exploring; but to do so profitably one must understand the peculiarities of the various media. If language is to be one of the media considered, there is no way to avoid dealing with the status of physical texts as intermediaries, as messengers that may at any point give us incorrect information. Derrida's essay is a richly textured meditation, remarkable in its convolutions; but what is more remarkable still is that a piece so full of word-play, so full of twists and turns and bypaths in its exploration of the nature of works, could fail to find its way to a confrontation with the ultimate indeterminacy of the texts (not just the meanings) of verbal works.

IV

Geoffrey H. Hartman's essay ("Words, Wish, Worth: Wordsworth," pp. 177-216), the next one in the volume, makes more reference than the others to specific variants between one text of a poem and another. Hartman appends to the essay a section headed "Text of Poem and Bibliographical Note," in which he prints (with two trivial differences) the text of "'A LITTLE onward . . .'" as it appears in Ernest de Selincourt


18

Page 18
and Helen Darbishire's edition (Vol. 4 of The Poetical Works of William Wordsworth [1947], pp. 92-94). His headnote to the text is "bibliographical" only in the sense of "reference bibliography," for it cites several other critical writings either about this poem or about matters taken up in the essay; there is no discussion of such bibliographical concerns as the publication history of the poem or physical details that might bear on textual decisions. He does mention one textual variant, the allusion to Antigone rather than Dora (line 11) in editions before 1850; and in the body of his essay he mentions such other variants as some of those between the 1805 and the 1850 texts of The Prelude (e.g., pp. 198, 211). Hartman clearly takes the poem under discussion as an utterance of a particular person at a particular time in the past, and he refers to other works by the same person and recognizes the value of textual variants as evidence of a writer's developing thought.[24]

It is odd, therefore, that he does not report all the textual variants in the poem, as recorded in de Selincourt and Darbishire, or comment on the textual policy of that edition. He was not obligated, of course, to present the entire text of the poem at all; but since he chose to do so for the reader's convenience, it is misleading not to have present the variants noted in the edition cited, especially when he refers to one of them. Readers might naturally conclude that there were no other variants.[25] In any case, as long as a "bibliographical note" is provided, it is strange not to have some commentary on the principles that governed the construction of the text that is being accepted. In her preface, Darbishire says, "In following Wordsworth's final text I have made a few corrections, in most cases supported by the manuscripts."[26] Although it does not appear, from the variants recorded in the 1947 edition, that any "corrections" from manuscript are incorporated into the 1947 text of the poem Hartman is discussing, there is one variant passage, identified in 1947 as "MS. corr. to text," that raises a question about the meaning of "final text." It was incumbent on Hartman, in any event, to think about the implications for his argument of following a "final text" incorporating revisions first printed in 1820, 1827, and 1850; but no illumination on this question is afforded by the "bibliographical note."

These matters bear on the primary concerns of the essay. Hartman is exploring Wordsworth's "intertextuality," the presence in his work of passages and ways of thinking influenced by his predecessors, by the "effaced or absorbed memory of other great poems" (p. 179). Going further, Hartman suggests that among the poet's "precursor text(s)" is a "primordial speech act," that indeed the "primordial wish" and the "primordial speech act . . . converge in the vision" (p. 197), that "Creation and response merge" (p. 199). Hartman is seeking the roots of poetic inspiration,


19

Page 19
in terms—growing out of the poem—that link light and sound, seeing and hearing: Wordsworth "brings the speaking darkness to light" (p. 195). The enterprise Hartman is engaged in necessitates thinking about the state of being of literary works. Any meditation on the shift "from visionary voice to visionary text" (p. 202)—a shift that is "part of a vast metaphoric activity identifiable with creative power itself"—must reflect some attitude regarding the mode of existence of verbal works and postulate some relationship between those works and the texts whereby we approach them. Hartman is trying to understand the act of reading in the broadest sense, the "reading" that poets perform and our own "reading" of those poets: "Wordsworth's poem suggests that we must read the writer as a reader" (p. 187). The immediate object of "reading" is "texts" (whether verbal or nonverbal), but how we define "texts" is the crux of the matter and will in turn affect how we conceive of "reading." Hartman's seeming lack of inquisitiveness about the production and status of the de Selincourt-Darbishire text—about how those two editors had read Wordsworth—is indicative of his broader failure to come to grips with the relation of texts to works in his own reading of Wordsworth, which is to say in his interpretation of Wordsworth's reading.

The problem becomes evident as we follow the word "text" through his argument. It first occurs in the opening paragraph, where Hartman discusses the initial image of the poem, the link between Wordsworth walking with Dora and the blind Oedipus being led by Antigone. The poet, Hartman says, is surprised by this association, "the usurpation of that text on his voice, and the anticipatory, proleptic nature of the thought." This linking of "text" and "thought" is revealing, for "text" here is similar to a thought that has been influenced by a reading of a text in the past. The referent of "text" is literally the quotation that constitutes the first two lines of the poem, but the intended meaning is evidently not a specific documentary text but any text of Samson Agonistes—or, perhaps more accurately, the experience of having read these lines of Samson Agonistes (and, in turn, Sophocles) in some text. Shortly thereafter he associates the word "quotation" with "voice" and an echo: imagination, he says, has risen "in the form of a voice, the quotation from Samson Agonistes—echoing back to Oedipus at Colonus" (p. 178). The "quotation," like the "text," refers to a passage in a work, not the words that happen to be in one or another specific text of the work. The question of how we find our way to the work from the signals available to us, of the problematic nature of the connections between texts and works, is not broached.

At least we think we have learned from the opening paragraphs how


20

Page 20
"text" is being used, even if a fundamental question is being overlooked at the same time. And our observation is confirmed by the immediately ensuing occurrences of the word: "a text surfacing in his [Wordsworth's] mind" (p. 180), "a well-known text flashing on the poet's mind" (p. 181), even "a famous text from Wordsworth's own poetry" (p. 183). In all these instances "text" means a passage from a work. But whether it can still mean the same thing in Hartman's conclusion to his third section is arguable: "An unmediated psychic event," he says, "turns out to be a mediated text: words made of stronger words, of the Classics and the Bible" (p. 186). Inner feelings or responses become "texts" to be read, and in the process they turn into actual verbal texts, arrangements of words, which are influenced by previous verbal works that have entered one's repertory of thought patterns. This description of a particular succession of mental events conceives of "text" in at least three ways: as something nonverbal to be interpreted, or "read"; as a passage from a verbal work of the past; and as a specific sequence of words and pauses (or punctuation) that one constructs from the other two kinds of text. The first involves—continuing the metaphor—"documentary" evidence, for one must accept as a given the particular combination of elements one is faced with. The second, though it is recalled as an individual series of words, is not perceived as the text of a specific document, with its own idiosyncrasies, but as the text of a linguistic work, which has its existence in some realm apart from the mundane realities of paper and ink or even of sound waves. The third, the new composition, is—what? Not really a third conceptual type, for how could it be something other than the text of a document or the text of a work abstracted from one or more documents? But which is it? Hartman's vagueness here will be noted only by those who have thought about the difference between texts of documents and texts of works. It is regrettable that Hartman does not appear to be among those, since his interest is in the nature of verbal works and their mediating role in our perceptions: his account, for all its richness, appears naïve as a result of being divorced from this basic distinction.

Recognizing how documents and works differ is essential to an understanding of reading, and much of his essay is concerned with the process of reading. He begins his fourth section by asserting, "The relation of 'text' and 'soul' is the province of a theory of reading" (p. 186). This statement takes on different meanings depending on whether "text" refers to the text of a document or the text of a work. The former connects us to "soul" by showing us the humanity of the producer or producers of the document, whether it is a manuscript in the hand of the author (or scribe) or a printed book set in type by a compositor. Documents always contain evidences of the striving of their producers; but


21

Page 21
the human dramas preserved in them may be a number of steps removed from the minds underlying the works that are reflected (however accurately or inaccurately) in the texts of the documents. The text of a work, on the other hand, links us directly to the "soul" of the author, and presumably it is the reading of texts of works that Hartman has in mind. But where do we find such texts? We can never know when we have found them; all we can do is approach them through the texts of the documents that seem intended to convey them. We will therefore be constantly questioning those documentary texts, for at any point they may be misleading guides. The process of reading leads us along an intricate path in which at each point we must confront the mysteries of physical objects before we can hope to recover the messages they seem to offer us. "Text" and "soul" are indeed connected, but Hartman's account is grossly oversimplified through its failure to address the relation of physical texts and verbal messages.

We may readily agree that authors bring "an inner light" to the "texts" they read and that their reading is "a form of life" (p. 187). But when we are told that the "inner voice also proves to be a text"—in this case "the textual voice of Milton" (p. 188)—are we not bound to aks for a more precise account of where that particular arrangement of words, now in the poet's mind, came from? Must we not confront the process by which one or more texts on paper become transmuted into a text in the mind, a text that is perhaps identical with one of those on paper, perhaps a conflation of them, and perhaps different in various respects from all of them? And what relation do all these texts have to the work as intended by its author? If poetry is "the working through of voices," if a poet experiences a "latent pressure of voices or texts" (p. 191), if "the usurping voice is referred to a specific text" (p. 193), we had better inquire into this process. Even though Hartman uses the phrase "a specific text," he clearly means "a specific work," not "a specific text of a work." When he speaks of "the relation between textuality and referentiality" (p. 193) and says that poetry "textualizes a phantom voice" (p. 194), that it is "echo humanized" (p. 195), and that the poet "brings the speaking darkness to light" (p. 195), he is concerning himself with the creation of a verbal work out of experiences that have themselves been affected by prior verbal works; but he is not even touching the equally profound mystery of where verbal communications have their being, of the ontology of works made of words.[27]

One could argue that it does not matter whether different people are influenced by different texts of Milton (texts as previously read or as created in the mind through the unconscious, but perhaps willful, tricks of memory); all that matters, one might say, is that their responses to the


22

Page 22
world are affected by prior groupings of words.[28] This approach could certainly be defended, but "Milton" would then not be a biographical reference, and the use of the term "Milton" would raise anew the question of the relation between ideas and each separate grouping of words employed to express them—that is, the relation between "referentiality" and "textuality," the very question that Hartman is attempting to explore. A coherent discussion of this question cannot escape the requirement of facing the problems inherent in defining "textuality," in segregating the two strands that make it up. If we are to understand how Wordsworth's experience on Snowdon "includes the poet as reader of a prior and sacred text" (p. 196) and how Wordsworth "has created his own text by a verbal geometry that extends the lines of force in a prior scripture" (p. 197), we have to explore how "lines of force" emanate from specific sequences of words and how varying sequences of words can be related to each other in such a way as to be encompassed by the singular term "scripture."

Hartman apparently uses "text" to mean written, as opposed to spoken, words, for he takes a stand in the debate over which are primary by saying that critics should understand "how texts eclipse voice and speak silence" (p. 207). Yet he seems to see no awkwardness in letting "texts" also mean the "scriptures" or "poems" that transcend the limitation of particular documents. In a key passage of his peroration he asserts that the work of Wordsworth's under discussion is "both a minor poem and a considerable text" (p. 213); but this statement does not mean that he has come around, however belatedly, to distinguishing texts of documents from texts of works. (What would a "considerable" text, in that sense, be?) He is saying that this poem does not conform to his criteria for great art but that it is an interesting example of intertextuality as an "undersong" to intratextuality. The "considerable text" here is a poem treated as a biographical document, without reference to the physical documents that provide our evidence for reconstructing it. Hartman, however, equates the two levels: "There is no authentically temporal discourse, no timely utterance, except by resolute acts of writing. . . . Writing . . . defers utterance of the definitive parole or password—from generation to generation" (p. 207). The actual situation is more poignant: the preparation of written or printed texts gives a verbal work its best chance of surviving; but what survives on pieces of paper consists of the accidents of particular historical moments, and the work that speaks to us across the generations will forever be a conjecture arising from those time-bound, vulnerable objects. Hartman's account of the creative process, and of the relation of literary works to sound and to time, cannot


23

Page 23
but seem thin compared to what it would have been if these fundamental considerations had been a part of it.

V

The final piece in Deconstruction and Criticism, J. Hillis Miller's "The Critic as Host" (pp. 217-253), is explicitly an apologia for deconstruction, taking as its basic metaphor the reciprocal, reversible relationship between parasite and host and offering, through an analysis of Shelley, "an 'example' of the word 'parasite' functioning parasitically within the 'body' of work by one author" (p. 232). Despite the quotation marks around "body," the essay displays no awareness of any problems surrounding the determination of what words constitute the texts of the works to be discussed. Even a reference to a clause in "one of the drafts of the preface" (p. 239) to Epipsychidion does not serve to open the subject of variant wording and what it implies about the nature of verbal works.[29] In the next sentence we are referred to "the poem itself"—the "itself" being presumably intended to emphasize the distinction between poem and preface but also implying, in the context of the essay, that there is no question about what "the poem itself" is.

For Miller, the word "text" is a simple synonym for "work" or "poem." He begins a sentence, for example, by saying, "The poem, like all texts, is . . ." (p. 226). He speaks of critical essays and "the texts they treat" (p. 226), of "such a text" as Shelley's "The Triumph of Life,"[30] of "all the great texts of Western metaphysics" (p. 228), of "any text in Western literature" (p. 248). In describing the parasitical relationship between a poem and earlier poems, he says, "The new poem both needs the old texts and must destroy them" (p. 225); he refers to "the encapsulation in the poem of echoes and references to a long chain of previous texts" (p. 232); and thus, when he comments on "the relation of the poem to previous and later texts" (p. 233), he does not mean "previous and later texts of that poem" but "previous and later poems." The shift from "poem" to "text" is apparently for the sake of verbal variety; if Miller's point were that poets are influenced by particular texts of previous poems, he would have to say something more about the independent existence of individual documents as physical objects, as well as to recognize that the text of the poem under discussion is just as problematical as the texts of those earlier poems. But this issue is obviously not in his mind. When he is totaling the number of appearances of "parasite" in Shelley, he arrives at the figure of seven "if one counts The Daemon of the World and The Revolt of Islam as separate texts" (p. 247): "texts" here can only mean "works." The essays that critics write, being works,


24

Page 24
are also called "texts," as in "the asymmetrical relation of critical text to poem" (p. 224). Some readers, familiar with scholarly editions that present critical texts (texts that incorporate emendations resulting from the editors' critical judgments), may at first think that Miller is posing the question of the relation of such critical texts to the poems or works as they may be thought to exist independently of their transcription in particular documents. This question would perhaps be more interesting than the one he has chosen to discuss; in any case there is no doubt that "critical text" here means "an essay by a literary critic."

In the other essays in this volume the use of "text" to mean "work" produces ambiguity, for "text" sometimes seems to convey two meanings in passages that will not support both. There is no such ambiguity in Miller's essay, where—purely and simply—"text" equals "work." But this lack of ambiguity is no compliment to Miller, for through it his essay reflects so great a removal from any thought of textual matters that the two words are never even tinged by an underlying recognition of such issues. He does provide some endnotes citing the editions from which his quotations are drawn, and in them he twice uses "text" to mean the particular arrangements of words in those editions; "works" could not be substituted for "texts" in these instances. But neither this conventional use of "text" nor his reference to the previous publication of part of his essay "in a preliminary form" causes him to pay attention to the variability of texts. He is free, of course, to use "text" and "work" as synonyms in the body of his essay if he chooses; but what he must then do, and does not, is to find some other way of referring to what has traditionally been referred to by the word "text." That he does not refer to texts in this sense at all has serious consequences for his discussion.

At the beginning, for example, he quotes a comment from an essay by M. H. Abrams, a comment in which Abrams in turn quotes from Wayne Booth. Miller then asks, "What happens when a critical essay extracts a 'passage' and 'cites' it? Is this different from a citation, echo, or allusion within a poem?" And he proceeds to ask whether the relationship between the quotation and the surrounding words is like that between parasite and host, and, if so, which is the parasite and which the host. All these are important questions—indeed, questions that go to the heart of the nature of texts and works. But Miller does not do them justice, since he does not treat the textual basis of them. He does not—to put it another way—give evidence of recognizing that quotations are mini-editions, raising all the questions about "establishing" texts that are raised by editions of the texts of whole works. Whatever "quoting" means is what "preparing an edition" or "determining a text" means. There is a whole spectrum of standards that can be applied to these


25

Page 25
activities, and Miller sensibly mentions "citation," "echo," and "allusion" in this connection. They—like "paraphrase," "influence," and so on—are all forms of reference to other works, forms in which different standards of textual accuracy are appropriate.

But accuracy to what? This is the central question, regardless of which genre of reference one is employing. Echoes, allusions, and paraphrases are not expected to have the same accuracy to something outside themselves that a quotation has; but their tenor, their general drift, their evocative effect, will nevertheless depend on the details of the text of that outside thing. (The point is not altered by the fact that some writers do not wish to be accurate, intending their paraphrases or quotations to be distorted to serve a particular purpose: they are still responding to what they believe some outside text to be.) If writers who quote from other writers (or refer in another way to them) accept unquestioningly the texts of particular editions (or the texts in their minds based on the reading of particular editions), they are actually referring—whatever they may think—only to documents, not to writers; and anyone reading their writings must try to understand their handling of citations as part of the process of understanding what they are saying. The first answer—preliminary to any other—to the question "What happens when a critical essay extracts a 'passage' and 'cites' it?" is that the quoter either understands or misunderstands the relation of texts to works.

In pursuing the question of quoting, however, Miller ignores this issue and moves into an etymological analysis of words (like "parasite") beginning with "para-," which is a "double antithetical prefix," signifying many sets of opposites. Each "para-" word "may seem to choose univocally one of these possibilities," but "the other meanings are always there as a shimmering in the word which makes it refuse to stay still in a sentence" (p. 219). His examination finds in the "little piece of language" that Abrams quotes from Booth (which includes the word "parasitical") "context after context widening out from these few phrases to include as their necessary milieux all the family of Indo-European languages, all the literature and conceptual thought within those languages, and all the permutations of our social structures of household economy, gift-giving and gift-receiving" (p. 223). In this way the quotation grows like a parasite; but the surrounding essay is also parasitical on the earlier one. This relationship, Miller asserts, is true of all quotations or allusions in all verbal works, poems as well as essays. He presents this analysis as "an 'example' of the deconstructive strategy of interpretation" (p. 223), as "a model . . . for the incoherence within a single critic's language" or "within any single literary text" (p. 224). Deconstruction thus uncovers the "equivocal richness" inherent in verbal expression.


26

Page 26

The absence, in this introductory exposition of deconstructive procedure, of any recognition of the difference between texts and works forces one to consider again, perhaps from a new angle, the much-discussed question of the relation of deconstruction to historical study. Textual criticism, as traditionally understood, is a historical enterprise; whether its focus in a given instance is an author's intention or the combined intention of a collaborative group (author, author's friend, publisher, and so on), its goal is the construction of texts of works as intended at some point in the past. The texts in surviving documents form the most important class of evidence for such reconstruction. When, instead of using those texts critically as evidence for textual reconstruction, one accepts them uncritically, one is abandoning any concern with intention. But one may still approach them with a historical interest, for those documentary texts were available to be read in the past and are therefore evidence for examining past readers' responses. One need not, of course, be interested in history at all, in which case one text is as good as another to provide a sequence of words for analysis—though the fact that a particular sequence of words comes from the past in then of no significance, and the sequence could just as well be devised by the critic. These logical possibilities emerge clearly once the text-work relationship is understood; but since Miller does not deal with that relationship, it is not surprising that his position in regard to these basic possibilities is unclear.

Words do have their genealogies, both etymological and associative; and it is not only the words of quotations that are importations into a work but, as Miller understands, all the words making up the work, for all are outsiders with their own histories. If, as a result, works (or individual texts of them) are "incoherent" (cf. p. 224), "unreadable" (p. 226), and "self-subversive" (cf. p. 228) in that they contain the basis for contradictory interpretations, one would assume that this universal condition either (1) eliminates any useful interest in what human beings have attempted to express in words or (2) can itself be eliminated, as a given, in order to focus on what human beings have attempted to say in spite of it. At times Miller seems to be leading toward the former view, as when he cites "the law that language is not an instrument or tool in man's hands, a submissive means of thinking," and concludes, "Language rather thinks man and his 'world,' including poems" (p. 224). This "triumph of language" (p. 233) would seem to make irrelevant any interest in the authorial intention behind certain sequences of words; how the words came to be combined would be of no significance. One text would serve as well for analysis as another—but there would still be textual decisions to make, for there would be no reason why any pre-existing sequence of words had to be kept intact, and critics would be free to alter any words


27

Page 27
they wished. Quoted texts would then be entirely swallowed up in the new context: commentary and subject would completely coalesce.

Such a position would be a tenable one, if carried through in this way. But, despite the similarity of the conclusion to Miller's host-parasite concept, this is not the position Miller takes. It is revealing that, when Miller says, "Language . . . thinks man and his 'world,'" he adds, "if he will allow it to do so" (p. 224). That man has any choice in the matter is an unexpected revelation, and how it can be is not clear. Of course, our world is what we decide it is; but Miller's "law"—"the law that language is not an instrument or tool in man's hands" (p. 224)—is presumably independent of our wishes, and language has had its way with us even if we think we did not "allow it to do so." In any case, contrary to what would seem to be consistent with the framework Miller sets up at the beginning, he continually refers to works and texts as the products of individual human beings, especially in the last three sections of his essay, where he is concerned with "the 'body' of work by one author" (p. 232), Shelley. Why should an analysis of the implications of the word "parasite" in various contexts stop at the boundary separating work by Shelley from other verbal work, unless there is an interest in seeing how this one human mind struggled with the difficulties of expressing ideas in language? If that is indeed the interest, then the focus has shifted from the helplessness of man in "the prisonhouse of language" (p. 230) to the intended meanings of individuals as they attempt to express themselves in spite of the obstacles thrown up by words with ancient heritages. And further, if that is the case, one must try to reconstruct, from the documentary texts that have come down to us, the texts intended by their authors—for there is no point linking two works on the grounds that they are by the same author unless one has made an effort to ascertain the author's intended texts of those works.

Fleetingly Miller seems to sense the impasse he has placed himself in. He asks, "Who, however, is 'Shelley'? To what does this word refer if any work signed with this name has no identifiable borders, and no interior walls either?" (p. 243). It is crucial for his argument to deal with this question, but he does not do so satisfactorily. He does not even explain appositely why Shelley's work (or anyone else's) may be thought to have "no identifiable borders": "It has no edges," he says, "because it has been invaded from all sides as well as from within by other 'names,' other powers of writing." But if those other "names" possess "other powers of writing," they are individuals who have placed their own stamp on intractable language; presumably Shelley's use of them would make them an element in his own personal "power of writing." The real reason that a writer's words could be considered to have no borders is that they are


28

Page 28
simply fragments of language, part of the immense mass of arrangements of words from the past. If we are interested in language as the driving force, and not the individuals who have used it (or been used by it), we have no reason to notice "authors." Miller does not make this point here; he places Shelley's name in quotation marks three times (p. 243), then refers to him for a page or so as "the poet" (which does not avoid the problem), then once more puts the name in quotation marks (p. 245), and then the troublesome question of why he is discussing Shelley as an individual slips away, unanswered.

Miller's essay is thus incoherent in argument, which is a different kind of incoherence from the one he finds in all verbal works. That one may distinguish two levels of coherence or incoherence is implied by Miller's own treatment of Shelley, for he does not regard Shelley as confused, even if Shelley's poetry, containing the contradictions and "blind alleys" (p. 230) embedded in language, must inevitably be "the record of a perpetually renewed failure" (p. 237). Miller's concern is, finally, biographical, a concern with the dignity and pathos of Shelley as a human being who, like all artists, attempted to reproduce in some lasting form outside his mind the sense of life that existed within it. In the case of literature, that lasting form is intangible and can only be brought to our attention through the second-hand reports of documents or of oral recitations; Miller's neglect of this situation is the basis of the incoherence of his essay. Of Shelley's "failure," he says, "The words, however, always remain, there on the page, as the unconsumed traces of each unsuccessful attempt to use words to end words" (p. 237). The words on paper are indeed the evidence of a human struggle—that of the producer of the document (whether author, scribe, or printer) to transmit a text accurately. But the work of poetry is not "there on the page."[31] Miller presents deconstruction as "simply interpretation as such" (p. 230), "analytic criticism as such" (p. 252),[32] because it deals with the fundamental nature of language. His account, however, overlooks something no less fundamental: that verbal works are not coequal with their representations in tangible form. This oversight causes him to confuse the deconstruction of particular texts with the deconstruction of authors' works and, as a result, to write an unpersuasive brief for deconstruction.

VI

A strong case for deconstructive criticism can be made, and anyone who has given careful thought to the implications of the distinction between texts of documents and texts of works could make a stronger one than any presented by the five writers in Deconstruction and Criticism.


29

Page 29
It is remarkable that a group of critics so preoccupied with theoretical matters should have failed to include in their discussions some recognition of the difference between the stationary arts (like painting and sculpture), which use tangible media, and the sequential arts (like literature and music), which use intangible media and which can therefore be preserved only through instructions for their repetition. My comments here do not constitute a criticism of deconstruction but rather a criticism of five essays that, in their various ways, reflect some of the attitudes of deconstruction. What I hope has become clear is that the traditional concerns of textual criticism help to clarify the issues that deconstructive critics must face.

Those who have been called textual critics and those called literary critics are alike in making critical judgments about literary works. All of them must follow, explicitly or implicitly, a theory of literature, and any theory of literature must take into account the implications arising from the intangibility of the medium of literature. One may choose to be interested or uninterested in history, but one must recognize how the medium of literature requires, in either case, that textual decisions be made. Language, like the other parts of what we think of as the world we inherited, can indeed be treacherous and unaccommodating. So, for that matter, is the nervous system through which we hope to assert our existence. But whatever we attempt to express, in one medium or another, is a historical fact, a part of the record of human sentience. Even if language always subverts what we think we have to say, we persist in selecting certain words rather than others, and arranging them in one order rather than another, in our effort to communicate—or even to assert that there is not or should not be such a thing as communication.

Whenever we wish to examine an effort of this kind from the past, we are forced by the nature of verbal transmission to include in our deliberations a questioning of the accuracy of the available texts that claim to represent it, and to accept as well the consequent possibility of altering those texts. When, on the other hand, we do not wish to receive groupings of words as evidence of previous human expression, we are then not inhibited by any historical argument from altering those words or their arrangement so that they seem more pleasing or effective or interesting to us. Either way, we have to consider making alterations in what has been passed down to us; either way, there is no coherent argument for considering inherited texts as inviolable. Critics who fail to question the constitution of documentary texts cannot argue convincingly for either of these approaches, and indeed are likely to confuse the two. Deconstruction raises questions about the role of history in human discourse,


30

Page 30
and it must therefore assess the historical status of preserved texts—the tenuous relation of the texts of artifacts to verbal communication—if it is to offer a satisfying model of human thinking.

Notes

 
[1]

An indication of the new prominence of the word "text" is offered by the announcement in the Chronicle of Higher Education (in an article by Scott Heller on 3 August 1988) that "A faculty committee at Syracuse recently approved a plan to change the name of the major to English and Textual Studies" (p. A16).

[2]

I have discussed these issues at greater length in my Rosenbach Lectures of 1987, published as A Rationale of Textual Criticism (1989).

[3]

A convenient listing of other writings by four of the five authors (all but Derrida) is Wallace Martin's "Bibliography" in The Yale Critics: Deconstruction in America, ed. Jonathan Arac, Wlad Godzich, and Wallace Martin (1983), pp. 203-212. Extensive lists of the literature of the subject are provided by Josué V. Harari at the end of his anthology Textual Strategies: Perspectives in Post-Structuralist Criticism (1979) and by Richard Barney in a supplement to the Reports of the Society for Critical Exchange, No. 8 (Fall 1980). Two useful introductory surveys are Jonathan Culler's On Deconstruction (1982) and Vincent B. Leitch's Deconstructive Criticism (1983).

[4]

In The Anxiety of Influence (1973) he speaks only of "poems"; in A Map of Misreading (1975) he frequently uses "texts" as well.

[5]

In A Map of Misreading he says that "there are no texts, but only relationships between texts" (p. 3).

[6]

Such texts exist, that is, to the extent that we admit the existence of external objects or external simuli.

[7]

That is, what would be considered errors by the author of the work transmitted in a given document or by one or more of the producers of the document.

[8]

Bloom's first inset quotation on p. 30 contains the phrase "on a rock", the reading of the 1975 book text (p. 73, line 24); the magazine text reads "up on a rock" (Poetry, 124 [1974], 252, line 22).

[9]

Even with the extra division, the magazine text would appear to most readers to have only six verse-paragraphs, for one of the divisions in the book text, after "To get to sleep tonight, at least until late" (p. 75, line 3), coincides with a page break in the magazine (p. 253), where there is no typographic indication that a break in the poem also occurs. Does a break in fact occur there? Perhaps Ashbery decided later to make a break there—or perhaps the space in the book text is an error. And could the failure of the book text to have the earlier break (after "Its hollow perfectly") also be an error? Does the poem have six or seven sections, and, if six, where do they occur?

[10]

Bloom's quotation at the bottom of p. 35 contains the phrase "This sample"; but both the 1974 text (p. 256, line 10) and the 1975 (p. 77, line 18) read "The sample". (At several points he cites line numbers, which are not provided in 1974 or 1975 for this 552-line poem.)

[11]

Certainly the word "amazament" in 1975 (p. 74, line 10) is a typographical error (the word is "amazement" in 1974, p. 253, line 8).

[12]

And can one really call the location of paragraph breaks a trivial matter? Or consider the shift from "emerge out of" to "issue from" and the addition of "with Pierre" trivial changes in poetic lines? (In the 1975 edition, these occur, respectively, on p. 74, line 13, and p. 75, line 10; in 1974, p. 253, line 11, and p. 254, line 7.)

[13]

The wording of his note is strange, however: after citing Donald H. Reiman's Shelley's "The Triumph of Life": A Critical Study (1965) as his source, he says, "Together with G. M. Matthews' edition . . ., this edition is authoritative." In fact, his quotations do not correspond in every respect to the Reiman text: all the lines in his quotations are printed flush left, whereas in Reiman alternating lines are indented; no italics appear in his quotations (except at a few points where he notes "emphasis added"), whereas in Reiman


31

Page 31
there are a number of italicized words and phrases; and the word "and" is spelled out and certain American spellings occur in de Man's quotations, whereas in Reiman the ampersand and British spellings are used. At some points de Man's quotations follow Matthews's edition or the first edition (Posthumous Poems, 1824) instead of Reiman's, but without any pattern: thus on p. 54, the capital letter in "Shape" (line 352) and the lower-case initial in "dawn" (line 353) correspond not to Reiman but to both Matthews and the first edition; the lowercase initial of "earth" (line 353) corresponds to Reiman and the first edition, not Matthews; but on p. 52 the capital letter in "Shape" (line 425) corresponds only to Matthews and not to Reiman or the first edition. (Whether or not such words are capitalized is obviously a matter of some significance.) At one point, de Man's text even follows Harry Buxton For-man's 1876-77 edition in a reading that, as Reiman points out (p. 161), "has no foundation": on p. 46, de Man's quotation of line 187 reads "the holes he vainly sought to hide" instead of "the holes it . . . ." Furthermore, there are places where de Man's quotations do not match any text cited in Reiman's notes: e.g., "the" is inserted in line 33, which is quoted (on p. 52) as "O'er the evening hills . . ."; "the" is omitted from line 353, which is quoted (on p. 54) as "Dew on earth . . ."; and "the' 'is substituted for "one" in line 367, which is quoted (on p. 72) as "As the enamoured . . . ." Whether or not one regards these differences as significant is beside the point: de Man says unequivocally, "All quotations from The Triumph of Life are from the critical edition established by Donald H. Reiman." Their presence, however, does reinforce the point that the texts of documents must always be questioned.

[14]

If he does, the sentence, lacking the word "other," is not well phrased (it could begin, "Like several other major works of the English romantics").

[15]

In good scholarly editions, much of the information necessary for assessing the editors' decisions is provided in apparatus. An accurate apparatus could be called "definitive" in regard to the information it aims to record; but no critical text based on it can be called "definitive" because no such text can embody all the responsible judgments that are possible.

[16]

I have placed "like" in brackets because the word in the printed text is "line." It seems unlikely that de Man meant to say "Is the status of a text line the status of a statue?", for a "text line" would seem to mean a unit or building block of a text and would therefore not be parallel with "statue," a whole work. The matter must remain uncertain, however—as, indeed, the constitution of all texts of works is uncertain. This typographical error, if it is that, illustrates the necessity for deciding on the makeup of the text as a part of the act of reading. It also shows how documents can be unreliable witnesses to past intentions, even though we can never know with certainty the precise extent of that unreliability.

[17]

This essay, in a translation by Wlad Godzich, was one of the pieces added to Blindness and Insight (originally published in 1971) for its second edition in 1983 (pp. 246-266). The original essay, "Les Exégèses de Hölderlin par Martin Heidegger," appeared in Critique 100/101 (septembre/octobre 1955), 800-819.

[18]

One must note, however, what he says in the foreword to the enlarged edition: "I have left all these texts exactly as they were first published and have made no attempt to update them or to make them more germane to present-day discussions about literary theory —also when, as is the case in the essay on Derrida, I am myself aware of inadequacies with which I have tried to cope elsewhere" (p. xi). I have not yet discovered a place where he revises the view of editing expressed here.

[19]

In the original French, the statement reads, "pour lui plus que pour tout autre, la correction du texte importe grandement" (p. 802).

[20]

This title in the original French, we are told on p. 82, is "Journal de bord."

[21]

He had earlier dealt with similar matters in, for example, the prefatory section of La Dissemination (1972); trans. Barbara Johnson, 1981).

[22]

At many points in "Border Lines" he calls it a "band" (e.g., pp. 77, 89, 90) and refers to it or the main essay in spatial terms: "a procession underneath the other one" (p. 78), "here, at the foot of the other text" (p. 87), "the upper band" (p. 91), "the procession above" (p. 98).

[23]

In a related passage Derrida comments on the "double narrative" of L'arrêt de mort


32

Page 32
in terms of physical presentation. He lists "binding" as one of the conventionally enclosing features of "the framed unity of the corpus" in literature; this particular verbal work in two parts is being held together, within a physical binding, at "an invisible hinge . . . (the space between the last sentence of the first récit and the first of the second)" (p. 142). In the first of Blanchot's two versions of this work, Derrida says, there was a second blank space separating the second récit from a kind of epilogue; but in the second version there is "only one blank space in the typography of the book, between the two récits," and this space takes on "an even more remarkable singularity" (p. 144). This space occurs "at the bottom of one page and at the top of another"—so that "after the narrator has said, 'What is extraordinary begins at the moment I stop. But I am no longer able to speak of it,' on the next page, the facing page, the other shore, truth enters—thematically, and by name" (pp. 144-145). But if part of the meaning connecting the two récits is being expressed by the page-break that allows two particular passages to face each other on facing pages, the work is then a work of visual as well as verbal art. Derrida does not, however, proceed to examine the problem of treating spatial arrangements as an element in an intangible work, as if verbal works were in fact tangible and existed on paper. If they did, there would be nothing wrong with thinking of bindings, or page-breaks, as unifying elements of such works. Derrida does not seem to see that his questioning of "the entire conventional system of legalities that organizes, in literature, the framed unity of the corpus" (p. 142) is linked with an unquestioning acceptance of a conventional misconception about the location of verbal works.

[24]

Hartman had earlier made a statement about the value of studying revisions, in "Retrospect 1971" prefixed to the 1971 printing of his book Wordsworth's Poetry 1787-1814 (originally published in 1964): "Perhaps Wordsworth never did emerge to an assured sense of self or a decisive poetry. There is something peculiar in the way his text corrupts itself: the freshness of earlier versions is dimmed by scruples and qualifications, by revisions that usually overlay rather than deepen insight. I should have paid some attention to this problem but was more interested, I now see, in the integrity of the mind than in that of the single poem" (p. xvii). Hartman's choice of the verb "corrupts" recalls the standard jargon of textual criticism, in which a textual "corruption" is an unauthorized "reading" (a word or phrase or mark of punctuation) that occurs in a particular documentary text; but Hartman's belief that Wordsworth's "text corrupts itself" refers to something else—to a poem becoming in some ways less satisfactory, in Hartman's view, as Wordsworth revises it. In the process of analyzing revised versions to come to this conclusion, one still needs to assess the texts of those versions to try to eliminate "textual corruptions" (in the traditional sense) from them. Hartman's explanation of his previous neglect of revisions—that he was more interested "in the integrity of the mind than in that of the single poem"—seems not to recognize that revisions form a significant part of the evidence for examining "integrity" of mind, and that therefore one must ask whether the documentary texts of each revision reflect the author's intentions at that stage.

[25]

A footnote on the first page of the essay tells the reader that "the entire text of the poem" appears at the end. The text printed Americanizes the punctuation of line 31 (placing a comma within quotation marks) and modernizes the spelling of "recal" in line 43 (to "recall"), without comment. However, it accurately reproduces "Holy writ" in line 51 (and in the quotation of the passage on p. 181), without noting that in the 1958 printing of the de Selincourt and Darbishire edition (printed "from corrected sheets of the first [i.e., 1947] edition") the initial letter of "writ" is capitalized.

[26]

Hartman cites the edition as that of de Selincourt only, not mentioning Darbishire at all; but she is named on the title page as co-editor, and she signed the preface, which—as in this sentence on the editing—uses the first person singular pronoun. Her reference to "following Wordsworth's final text" is in line with the policy set forth in the first volume of the edition (1940), where de Selincourt says that he follows Wordsworth's collected edition of 1849-50, "the last to appear under his personal supervision."

[27]

Even his few references to versions—e.g., "Wordsworth's manuscript revisions" (p. 196) and the "1805 version" (p. 200)—do not serve to bring up the issue.

[28]

That is, in speaking of "the usurping memory of Milton's text" (p. 204), one is


33

Page 33
emphasizing "memory" over "text"—or "text" as a framework of ideas, not a specific arrangement of words.

[29]

He does cite in endnotes the sources of his quotations; but inaccuracies in quoting provide unintended illustrations of the distinction between works and documents. The most striking question raised by his quotations concerns the word "uncanniest" in his quotation (p. 227) from the Kluback-Wilde translation (1958) of Heidegger's The Question of Being—a word that he proceeds to use several times in his discussion following the quotation. In the original printing of this bilingual edition, published in New York by Twayne Publishers, the word at this point is "strangest" (p. 37); and the translators comment on this translation of "unheimlichste" in brackets in the previous sentence (where the word also occurs): "the word unheimlich here signifies not having a home." Miller cites a later printing of this edition ("New Haven, Conn.: College & University Press, 1958" [the "1958" presumably occurring in the copyright notice]); but, since the currently available printing (with the same imprint that Miller cites) also has the reading "strangest", it seems unlikely that the copy Miller used had a different reading. Perhaps, instead, the change from "strangest" to "uncanniest" in Miller's quotation somehow springs from his knowing the association of "uncanny" and "unheimlich" in translations of and commentaries on Freud. (Cf. Miller's later citation of Freud's "The Uncanny" in his chapter "Wuthering Heights: Repetition and the 'Uncanny'" in Fiction and Repetition [1982], p. 69 and note 6 on pp. 236-237.) For a discussion of Freud's essay on "The Uncanny" and translations of "unheimlich," see Karl Kroeber, Romantic Fantasy and Science Fiction (1988), p. 90 (and the references on pp. 176-177, note 13).

[30]

Three of the five essays discuss this poem, reflecting (as Hartman explains in the preface) "an earlier scheme" to concentrate on Shelley (and presumably on this poem).

[31]

Another instance of this phrase in Miller's writing occurs in the chapter on Wuthering Heights in Fiction and Repetition (1982): "Words no different from those we use in everyday life . . . may . . . go on functioning as the creators of the fictive world repeated into existence, . . . whenever the act of reading those words is performed. The words themselves, there on the page, both presuppose the deaths of that long line of personages and at the same time keep them from dying wholly, as long as a single copy of Wuthering Heights survives to be reread" (p. 72). Such a reference to words "there on the page" also presupposes that the literary work exists on the page; but what happens to be present in a given copy (such as the "single copy" that might be the sole surviving copy) may not in every respect perpetuate the "fictive world" that was in the author's mind. A somewhat more sophisticated reference to words on a page appears in Miller's The Linguistic Moment (1985): the linguistic moment in Wallace Stevens, we are told, is "when language almost emerges as surd, not sign, becomes almost a material substance, almost sounds or marks on a page, almost blanched, drenched, drained of meaning" (p. 14). When words on a page take on this status, they are being approached as visual art; but the issue of what words (or "marks") ought to be there, and what standard should be employed in answering that question, is still completely avoided.

[32]

In The Ethics of Reading (1987), he says, "Deconstruction is nothing more or less than good reading as such" (p. 10).