The essay argues that de Man's fabled shift from history to reading and rhetoric--to the rhetoric of reading, as the Preface to _Allegories of Reading_ puts it--was in fact always already a shift _past_ rhetoric and _to_ an other history:  the material history of his last essays. This essay appears in _Legacies of Paul de Man_, a volume of _Romantic Circles Praxis Series_, prepared exclusively for Romantic Circles (http://www.rc.umd.edu/), University of Maryland.

1. "Legacies of Paul de Man" is a most appropriate topic for a publication specializing in romanticism. As is well-known, romantic literature was, for de Man, a privileged locus for asking the question of history (in particular, the question of our history). Indeed, one could say that de Man's thinking of history—in fact, what he in his last essays calls "material history" or "the materiality of actual history" (and what no doubt constitutes one of the most valuable and enduring legacies he has bequeathed to us)—gets produced by his reflection on, and reading of, the romantics. But to say this may seem a bit odd, for de Man's own verdict on this work sounds much rather like the confession of a failure, in particular the failure to arrive at a "historical definition" of Romanticism. Looking back with some misgivings upon the essays collected in The Rhetoric of Romanticism—for him a "somewhat melancholy spectacle" in that it offers "such massive evidence of the failure to make the various individual readings coalesce" (RR viii)¹ —de Man writes that these readings seem always to "start again from scratch and that their conclusions fail to add up to anything" (RR viii). He continues: "If some secret principle of summation is at work here, I do not feel qualified to articulate it and, as far as the general question of romanticism is concerned, I must leave the task of its historical definition to others. I have myself taken refuge in more theoretical inquiries into the problems of figural language" (RR viii). De Man makes the same gesture in the opening sentences of the Preface to Allegories of Reading, and this time formulates the "failure" explicitly as a "shift" from history to reading (and thus to a "rhetoric of reading"): "Allegories of Reading started out as a historical study and ended up as a theory of reading. I began to read Rousseau seriously in preparation for a historical reflection on Romanticism and found myself unable to progress beyond local difficulties of interpretation. In trying to cope with this, I had to shift from historical definition to the problematics of reading" (AR ix)² . That this shift is once again a move from historical definition to the problems of figural language, i.e., to rhetoric, is clear enough in the following sentences of the Preface and their account of a "rhetoric of reading" where "rhetoric is a disruptive intertwining of trope and persuasion." So: de Man's own account of his work on the romantics and Romanticism would seem to indicate, if anything, a turn away from history and to the theoretical problematics of reading, rhetoric, and figural language. Nevertheless, it might be prudent not to take de Man's own remarks about his alleged "failure" too literally. It might be better to take a page from de Man's own book, as it were, and actually read what it is that happens, what takes place, in this alleged "shift"—and its necessity ("I had to shift")—from history to reading and rhetoric. If we do so, it turns out that this shift is in fact already (always already) a shift past the rhetoric of reading and to...history, indeed, to the material history of de Man's last essays. Ironically (and undialectically) enough, the "failed" attempt at a historical definition of Romanticism turns into a certain "success" for de Man's thinking of history.

2. The mechanism and the necessity of de Man's apparent shift from historical definition to the problematics of reading and rhetoric are best legible in his 1967 Gauss lectures on Romanticism and Contemporary Criticism (and not so much in the essays on Rousseau collected in Allegories of Reading, in which the shift has, in a sense, already been completed), in particular the lecture on Heidegger's interpretation of Hölderlin ("Patterns of Temporality in Hölderlin's 'Wie wenn am Feiertage'") and the lecture on Wordsworth (and on Geoffrey Hartman's interpretation of Wordsworth in his 1964 Wordsworth's Poetry) called "Time and History in Wordsworth." The latter is particularly helpful for understanding the shift because it consists of two "layers"—an original "pre-shift" lecture written in 1967 and some "post-shift" passages interpolated into the lecture around 1972 that reformulate the lecture's thematic concerns (death, time, and history) in explicitly rhetorical terms. Nevertheless, the actual push to rhetorical terms and rhetorical reading occurs already in the lecture on Heidegger's interpretation of Hölderlin's "Wie wenn am Feiertage," and it occurs on account of the lack of other terms and the failure of any other reading to do justice to Hölderlin's text. In short, the turn to rhetoric occurs on account of the lack in Heidegger's terms and Heidegger's failure to think the "temporality of poetic form," as de Man puts it, when he comes to interpret Hölderlin's poetry. How so?

3. De Man begins his critical reading of Heidegger in a hopeful vein: Heidegger's "ontological understanding of Hölderlin's key concepts as they are seen to operate within the limits of particular poems" (RCC 56)³ is a promising development that could lead to a "reorientation of literary interpretation toward an ontological understanding." Such an understanding is promising because it "does allow, in principle, for the combination of a sense of form (or of totality) with an awareness that poetic language appears as the correlative of a constitutive consciousness, that it results from the activity of an autonomous subject. Neither American formalist criticism nor European phenomenological criticism has been able to give a satisfactory account of this synthesis: the former had to give up the concept of a constitutive subject, the latter that of a constituted form" (RCC 57). De Man's statement of the advantage of such an ontological orientation is pithy, as it calls to mind his critiques of the American New Criticism, its misunderstanding of the concept of intention, and its consequent reification of poetic form, and his critiques of a phenomenological criticism like that of the Geneva School which ignores questions of form and, in a sense, simply does not read.⁴ Now Heidegger, de Man's argument continues, seems particularly qualified to undertake "this renewal of critical method" even though literary interpretation was not his own academic field. Although Sein und Zeit nowhere deals with literature, except for some passing references, "it does contain insights that can give a more concrete direction to an ontological interpretation of texts" (RCC 57). De Man's statement of these insights amounts to an extremely compact summary of Sein und Zeit. Because it contains in germ everything to come in de Man—including the impetus for the shift from history to reading—it is worth quoting in full:

   Sein und Zeit, indeed, stresses not only the privileged, determining importance of language as the main entity by means of which we determine our way of being in the world, but specifies that it is not the instrumental but the interpretative use of language that characterizes human existence, as distinct from the existence of natural entities. And this interpretative language possesses a structure that can be made explicit. This structure is in essence temporal—a particular way of structuring the three dimensions of time that is constitutive for all acts of consciousness. The main task of any ontology thus becomes the description of this temporal structurization, which will necessarily be a phenomenology of temporality (since it is the description of consciousness) as well as a phenomenology of language (since the manner in which temporality exists for our consciousness is through the mediation of language). One understands that, as the 'purest' form of interpretative language, the one least contaminated by empirical instrumentality and reification, poetic language is a privileged place from which to start such a description. And conversely, one sees that an approach to poetic language that would, by a description of its temporal structure, bring out its interpretative intent, would come closest to the essence of this language, closest to accounting for what Heidegger calls 'das Wesen der Dichtung.' We could thus legitimately expect from the Heideggerian premises a clarifying analysis of poetic temporality, as it is seen to act within the poetic form. (RCC 57-8)

   Although the terms de Man uses to summarize Heidegger may be a bit too phenomenological—too Hegelian phenomenological (for instance, in his use of the word "consciousness")—his account is precise and rigorous. And in its very precision and rigor it presents Heidegger's fundamental ontology with a redoubtable task: namely, to be not only a "phenomenology of temporality"—something Heidegger manages quite well, thank you—but also a "phenomenology of language." This latter half of the task is more difficult and has far-reaching consequences because it necessarily entails, sooner or later, some account of "the manner [my emphasis] in which temporality exists for our consciousness," and that "manner," i.e., the way that language "mediates" consciousness and temporality, may include factors and functions of language irreducible to a hermeneutics of self-understanding, no matter how fundamental the ontology it bases itself on: in short, that "manner" may include the rhetorical dimension of language. Although de Man does not yet put it that way, one could already say that the reason Heidegger's interpretations of Hölderlin's poetry are so disappointing—indeed, so downright wrong, according to de Man—is on account of his inability to read the manner (i.e., ultimately the rhetoric) in which temporality exists for the "consciousness" of Hölderlin's poetic language. In any event, ironically (but, as always, rigorously and consistently) enough, the result is that the great thinker of temporality cannot think, cannot read, the temporality of the poetic form of Hölderlin's poem. In his interpretation of "Wie wenn am Feiertage," Heidegger's misreading consists of his flattening out, levelling, the temporal articulations and tensions of the poem in the service of an apocalyptic pattern. De Man summarizes: "By its gradual widening out from particular physical nature to history, to the gods, and finally to being itself, the poem dramatizes a process of all-encompassing totalization that stretches from the beginning to the end of the text. The progression takes place without discontinuity and moves in one single direction, toward the full disclosure of being. The pattern is apocalyptic, a temporal movement that culminates in a transcendence of time" (RCC 64). The poet, in Heidegger's interpretation, is "someone who stands in the presence of being in the past (when he is waiting for the disclosure), in the present (when it takes place in the heroic acts of history), and the future (when, like the countryman caring for his land, the concern of his work will maintain, for others, a mediate form of contact with being)" (RCC 65). In short, Hölderlin would be an apocalyptic poet, "an eschatological figure, the precursor who, during a period of temporary alienation from being (Seinsvergessenheit), announces the end of this barren time and prepares a renewal" (RCC 65).

4. Now, according to de Man, this interpretation of Hölderlin as an apocalyptic poet is wrong in general and, in the particular case of Heidegger's interpretation of "Wie wenn am Feiertage," it is wrong in specific ways for specific reasons. It is wrong in general because Hölderlin, rather than being an apocalyptic poet, is precisely he who warns against the danger of believing that the poet can accomplish the kind of proximity to being Heidegger sees in the poem. Indeed, the poem instead "cautions against the belief that the kind of enthusiasm that animates a heroic act is identical with the predominant mood of a poetic consciousness" (RCC 67). But more important than the erroneous results of Heidegger's interpretation is the specific way in which Heidegger manages to get things so wrong in his resolute misreading of "Wie wenn am Feiertage" in terms of an apocalyptic pattern. Heidegger is able to flatten out the temporal tensions of Hölderlin's poem in two ways. First, Heidegger ignores and treats as unproblematic a certain "ambiguity of metaphorical reference" (RCC 62) in the poem's opening simile that makes it impossible to decide whether "they," the poets, are like the countryman who goes out to look at his field after the lightning storm or whether they are like the trees that stand exposed during the storm and get blasted by the lightning. The ambiguity is important because its temporal tension is what gets unfolded in the rest of the poem—indeed, it is what constitutes the temporality of this poem's poetic form. And it gets unfolded in terms of the triadic tonal pattern of Hölderlin's theory of the alternation of tones (Wechsel der Töne), as the poem modulates from the "naive" tone of its opening scene, to the "heroic" tone of the heroic acts of history it describes later, to end in the reflective, meditative tonality that Hölderlin calls "ideal." Heidegger cannot read the alternating tones of Hölderlin's hymn because he ignores the poem's Pindaric triadic structure and simply cuts off the fragmentary lines that would have constituted the strophes of the poem's end—that is, makes the poem "whole" by truncating it. So: by glossing over the ambiguity of metaphorical reference in the poem's opening simile and by ignoring the poem's tonal structure and truncating its ending, Heidegger completely disregards its poetic form. And since its poetic form is the temporal structure of the poem's self-understanding, disregarding it means also disregarding the poem's temporal structure. Again, it is a case of Heidegger—the thinker of temporality—not being "Heideggerian" enough! The consequences for Heidegger are clear: in short, there is a flaw in Heidegger's method, as de Man puts it, "that leads to a misinterpretation of Hölderlin as an apocalyptic poet, when Hölderlin's main theme is precisely the non-apocalyptic structure of poetic temporality" (RCC 71). This flaw, de Man concludes vigorously, is "the substitution of ontological for what could well be called formal dimensions of language. The ontologization of literary interpretation, which seemed so promising in the Heidegger of Sein und Zeit, does not mean that literature can be read, so to speak, from the standpoint of being, or from that of a poet who is said to act as a direct spokesman for being. The standpoint can only be that of a consciousness that is ontologically (and not empirically) oriented but that nevertheless remains a consciousness, rooted in the language of a subject and not in being" (RCC 71).

5. But if the results for Heidegger's "method" of his having substituted "ontological for what could well be called formal dimensions of language" are clear, the results for de Man's own developing "method" are more complicated because they are double. On the one hand, what de Man ends up with is a still more thoroughgoing "ontologization" of language and of poetic form than Heidegger's. The main difference would be that whereas for Heidegger the poem's temporal movement takes place "without discontinuity" and moves "in one single direction"—again, according to an apocalyptic pattern whose temporal movement culminates in a transcendence of time—for de Man the poem's temporal structure is one in which "beginning and end come together within the tension of the radical discontinuity that seemed to keep them apart" (RCC 70). In connecting the beginning of the poem with its end, the radically discontinuous temporality of the poem's poetic form nevertheless remains a principle of totalization. Indeed, de Man goes so far as to call it a "hermeneutic circularity" (RCC 71) and to deposit its discontinuous temporality in the structure of being itself. "The principle of totalization is indeed ontological," he writes, "in that it has to be sought in the discontinuous structure of being itself" (RCC 72). So, on the one hand, in his ability to read the discontinuous temporality of Hölderlin's poem—and in his depositing of this discontinuity in the discontinuous structure of being itself—de Man would seem to be more "Heideggerian" than Heidegger. On the other hand, the conclusions of de Man's reading of Heidegger nevertheless go in an entirely different direction and prohibit such a "super-Heideggerian" ontologization of poetic form. Indeed, what can de Man mean by charging Heidegger with having substituted ontological for what could well be called formal dimensions of language—and by saying that the poetic consciousness is rooted in the language of a subject and not in being—and then going on to deposit the discontinuous temporality of poetic form in the discontinuous structure of being itself? The tension—in fact, a certain discontinuity—between ontological and "what could well be called formal" dimensions of language, between a poetic consciousness rooted in being and a poetic consciousness rooted in "the language of a subject," is legible throughout de Man's attempt at an ending for his essay. For what de Man has come up with in his more-Heideggerian-than-Heidegger thoroughgoing ontologization of language and poetic form is "formal structures"—the reversals and discontinuities of which they are capable—that work according to laws different from those of "the structure of being itself." That these "formal" structures are specifically linguistic, indeed already rhetorical, structures is evident, as de Man tries out various names for the "discontinuous element" that constitutes the temporal structure of Hölderlin's poetic form. Adorno's "parataxis" is one possibility, which (parataxis) is linked by Auerbach to what he calls a "figural style." Hölderlin's own term for this discontinuous element is "the caesura referred to in the commentaries on the Oedipus tragedies, which marks a reversal of tone as well as a reversal of time and in which the end reestablishes with the beginning a contact which it seemed to have lost" (RCC 72). De Man's ending needs to introduce these explicitly rhetorical terms—parataxis, caesura, figural style, and others—because the discontinuous temporality his reading of Hölderlin has disclosed is one whose reversals can no longer be accounted for in ontological terms. That de Man's reading of Hölderlin has pushed Heidegger's fundamental ontological terms to their breaking point is especially legible in an almost stuttering formulation de Man uses in trying to distinguish the totalizing yet discontinuous temporality proper to Hölderlin's poetry from an organic unity (like that of Schelling's philosophy of identity) and from a purely dialectical one (like that of Hegel): "Nor is it purely dialectical, in the Hegelian sense," writes de Man, "for time itself, which remains unproblematically forward-directed in Hegel, here becomes itself a discontinuous element of a structure that consists of a series of temporal reversals" (RCC 72). The tortuousness of the formulation becomes apparent if we try to paraphrase it: if time itself becomes itself a discontinuous element of a structure that consists of a series of temporal reversals, then time becomes a discontinuous element of a structure that consists of a series of discontinuous reversals that will never allow us to say how time itself could ever become, or be, itself! In short, de Man's reading of the temporality of poetic form proper to Hölderlin's poetic language has disclosed a discontinuous temporality and structures of reversal and substitution that cannot be accounted for in the terms of Heidegger's fundamental ontology—or even in the rhetorical terms of Adorno and Auerbach insofar as these are still compatible with their fundamentally hermeneutic orientation.

6. The tension or discontinuity in the double ending of "Patterns of Temporality" comes to full fruition in "Time and History in Wordsworth," de Man's veritably "Hölderlinian" reading of Wordsworth. In fact, one could say that the double ending of "Patterns" produces the two layers of "Time and History." De Man's readings of "The Boy of Winander" and the Duddon sonnet can be called Hölderlinian because they consist of a certain "application" of Hölderlin's "caesura" for an understanding of the reversals and substitutions that lie at the basis of both poems and that de Man, in a sense, "re-reverses." In the case of the Boy of Winander, the poem substitutes the death of a third person (the Boy) in the past for the death of the first person ("I") which lies in the future. "Wordsworth is thus anticipating a future event as if it existed in the past. Seeming to be remembering, to be moving to a past, he is in fact anticipating a future. The objectification of the past self, as that of a consciousness that unwittingly experiences an anticipation of its own death, allows him to reflect on an event that is, in fact, unimaginable" (RCC 81). In the case of the Duddon sonnet, the poem substitutes one temporality—a movement that goes from nature to history—for another, more authentic, temporality—a movement that goes from nature to the dissolution of self and the loss of the name. In doing so it reverses middle and end (of the poem) and makes it seem as though the derived, secondary temporality (in short-hand, empirical "history") could contain the more original, authentic temporality of dissolution, mutability, and ceaseless deathward progressing, when, in fact, it is the other way around: the authentic temporality (one clearly based on Heidegger's analytic of Dasein and the finitude proper to it on account of its being fundamentally a being-unto-death) contains "history." In other words, both poems perform a reversal and a substitution that makes the impossible—reflecting on one's own death, history-as-progress overcoming mutability—possible. But already in the first, thematic layer of the lecture, de Man recognizes that the impossibility is made possible only thanks to a certain sleight-of-hand which, already, is clearly a linguistic, indeed rhetorical, sleight-of-hand. In the Boy of Winander, conquering the time, the surmise, that would allow one to reflect on one's own death is possible only as a "fiction" which, "since it is a fiction [...] can only exist in the form of a language" (RCC 82). That this language is necessarily a figural language is legible in de Man's formulations of how it is that this "fiction" can allow one to look back upon, as it were, one's own death: "The poem is, in a curious sense, autobiographical, but it is the autobiography of someone who no longer lives written by someone who is speaking, in a sense, from beyond the grave" (RCC 81) and "it is the epitaph written by the poet for himself, from a perspective that stems, so to speak, from beyond the grave" (RCC 82). It is clear that speaking or writing from beyond the grave is "possible" only thanks to the rhetorical shifts of "in a sense" and "so to speak." Although in the first layer reading of the Duddon sonnet the rhetorical shift is not as marked, the fact that the substitution and reversal—of history and temporality, of the poem's end and middle—are in fact a rhetorical structure is, in a sense, still more explicit, since its reversal and substitution of container and contained, enveloppant and enveloppé, amounts to the very definition of a particular trope, namely metonymy. Indeed, it is no doubt the rhetorical shifts and rhetorical structures of his own language that push de Man to perform a self-reading which forces the thematic readings of the Boy of Winander and the Duddon sonnet to turn into readings "properly speaking"—that is, rhetorical readings in explicitly rhetorical terms—in the second layer of the lecture. In the Boy of Winander, the substitution of a first person subject by a third person subject, the "Boy" for "I," is now said to be based on a "metaphorical substitution," just as in the Duddon sonnet the reversal of "history" (contained) and authentic temporality (container) is said to be based on a "metonymic figure." But this passage, this shift, from a thematic reading and its terms—death, finitude, history, temporality, and mutability—to a rhetorical reading and its terms—metaphor, metonymy, metalepsis—should not mislead us into thinking that the thematic has simply been left behind, surpassed, as though de Man had succeeded in reducing temporality and history to a question of merely tropological substitutions and transformations. If we read his second-layer interpolations with any attention at all, we cannot make this mistake. For it is clear that in the Boy of Winander the "metaphorical substitution" of the first by a third person, of a living self by a dead self, "is, of all substitutions, the one that is, thematically speaking, a radical impossibility: between the living and the dead self, no analogical resemblance or memory allows for any substitution whatever" (RCC 201). In fact, as de Man goes on to say, "the metaphor is not a metaphor since it has no proper meaning, no sens propre" (RCC 201) and could more properly be called "the metonymic reversal of past and present that rhetoricians call metalepsis" (RCC 201). But even to call this reversal a metaleptic metonymy would be claiming to know more than one can about the radically discontinuous nature of this reversal. Just as there can be no analogical resemblance between the living self and the dead self, so there can be no contiguity or juxtaposition, no "next-to-ness," between a dead past and a living present that would allow for a "properly" metonymic substitution. In time the dead self may be "near to" the living self—just as the child, according to a sentence de Man crossed out, "being the father of man [...] stands closer [my emphasis] to death than we do" (RCC 202)—but this proximity has no empirical, phenomenal, thematic existence, and therefore the "metonymy" is a blind, mutilated metonymy—in fact, more of a catachresis than a metonymy. In short, the "metaphorical substitution" is in fact a self-undoing trope that self-deconstructs into the catachrestic imposition of a name. "The poem does not reflect on death," de Man concludes, "but on the rhetorical power [my emphasis] of language that can make it seem as if we could anticipate the unimaginable" (RCC 201). In the same way, the metonymic reversal of the Duddon sonnet, because it "is a rhetorical device that does not correspond to a thematic, literal reality" (RCC 202), also gets undone in what de Man is already able to call (after his reading of Grammatology) a "de-constructive rhetoricity" (RCC 203). This is thematically, literally, understandable: if the poem performs a reversal and a substitution of contained (empirical history) for container (the authentic temporality of dissolution), then its rhetorical device amounts to the equivalent of saying that the water or the wine can contain the glass. And if the "glass" here is the authentic temporality of ceaseless dissolution, then even "properly speaking," as it were, it was not much of a "container" to begin with! In any event, the point is not to dwell on the mechanics and the details of this "de-constructive rhetoricity"—we can read all about it in Allegories of Reading and elsewhere—but rather to insist that already here, at the very pivot of de Man's "shift" to rhetoric and rhetorical terms, the move to rhetoric is already a move past rhetoric, to an awareness that tropological textual models will also not be able to account for what actually happens, what actually occurs, in and as the texts of Hölderlin and of Wordsworth. And, as we can read in Aesthetic Ideology, what actually occurs—that which is truly, materially, historical—is not the textual linguistic model into which the tropological model empties out and passes (for example, of language as performative) but rather the passage, the passing, itself—a break, gap, or discontinuity like the one that cleaves Kant's Third Critique (between the tropological "model" of the mathematical sublime and the performative "model" of the dynamic sublime).⁵ In "Kant and Schiller," de Man does not hesitate to say directly what thinking history as event, as occurrence, ultimately means, and he does it in terms particularly resonant for our discussions of "Patterns of Temporality" and "Time and History": "History is therefore not a temporal notion, it has nothing to do with temporality, but it is the emergence of a language of power out of a language of cognition" (AI 133). In saying starkly that history is not a temporal notion, that it has nothing to do with temporality, de Man draws out the full implications of his 1967 readings of Hölderlin and Wordsworth and their disclosure of reversals and substitutions whose discontinuity is not temporal but rhetorical. What this also means is that de Man's alleged "shift" from history to reading and rhetoric—as one that is also a shift past rhetoric—is in fact a shift from history to history⁶ —a shift whose own discontinuous "passage" or passing "from" and "to" is what happens, what actually occurs, materially historically, in and as "de Man."

7. For a coda it would be good to offer an example or an emblem of de Man's discontinuous shift from history to history. If we are right to call this shift material and historical, then what would be the equivalent of this moment "in the order of language" (AI 89)? In de Man's last essays, this equivalent always turns out to be what he calls the materiality of inscription, the prosaic materiality of the letter. And the double-layered lecture on Wordsworth in fact provides a material inscription that renders the discontinuous shift—from history to reading, from rhetoric past rhetoric, from history to history—vividly legible. In passing from his reading of the "complex temporal structurizations" in the Boy of Winander to the Duddon sonnet, in order to take one further step toward an understanding of "his [Wordsworth's] temporality," de Man in the second layer of the lecture simply inscribes the word "rhetorical" above the word "temporal" (in "temporal structurizations") and the phrase "rhetorical movement" above the phrase "his temporality" (RCC 202), in both cases without crossing out what he had originally written in the first version of the lecture. However legible this shift or passage "from" temporal "to" rhetorical may be, it also remains singularly unreadable and incomprehensible in terms of any narrative that would tell stories of from and to, before and after, or even "first layer" and "second layer." What happens happens "between" the two inscriptions and, as such (i.e., as something that happens), is genuinely, materially, historical—de Man's history and our legacy.