MIXED MESSAGES: THE HETEROGENEITY OF HISTORICAL DISCOURSE

MIXED MESSAGES: THE HETEROGENEITY OF HISTORICAL DISCOURSE STEVEN G. CROWELL ABSTRACT If, as many historians and theorists now believe, narrative is the form proper to historical explanation, this raises the problem of the terms in which such narratives are to be evaluated. Without a clear account of evaluation, the status of historical knowledge (both in itself and in all those social, political, and other contexts in which appeal to historical explanation is made) remains obscure. Beginning with the view, found in Hayden White and others, that historical narrative constitutes a meaning not reducible to the factual content it engages, this essay argues that such meaning can arise only through a synthesis of cognitive and normative discourses. Narrative combines heterogeneous language games in such a way that neither appeal to truth content nor to justice suffices to decide the question of which of two competing historical explanations is, as a whole, superior. Examining in critical detail the opposed positions on this issue articulated by two recent theorists Frank Ankersmit ( narrative idealism ) and David Carr ( narrative realism ) the paper concludes that the debate between those who hold that historical narratives should be judged in essentially cognitive terms and those who hold that they should be judged in essentially political terms cannot be resolved and that a philosophical view of historical narrative that is neither realist nor idealist needs to be developed. I. INTRODUCTION: MEANING AND THE PROBLEM OF EVALUATING HISTORICAL WRITING How are interpretations to be evaluated? Where there is interpretation, as there is in any attempt to understand the past, questions about the validity of the interpretation arise, as do those about the standards such as truth, objectivity, richness, political efficacy, or morality in terms of which such validity may be assessed. In this essay I take up some problems connected with the evaluation of interpretations, focusing specifically on the issue as encountered in the writing of history. It is not at the level of the sentence or the fact that the question of evaluation becomes interesting. We are fairly clear about what constitutes the truth of statements and how such truths can be established. Even if the special mediations that characterize the evidence for true statements in historical writing pose serious problems not encountered in some other disciplines, the most interesting problems in historiography do not concern a kind of skepticism that would impugn our ability to establish any truths about the past. Nor is theory the pri-

THE HETEROGENEITY OF HISTORICAL DISCOURSE 221 mary unit of evaluation in historical writing (in the sense in which philosophers of science explore how we decide between rival theories). Very little theory in the strict sense of systems of sentences deductively governed by laws seems to be important in the work of historians. Should one adopt the looser use of the term theory according to which any conceptual framework, whether organized according to strict laws or not, is a theory, the resulting analogy between the horizon (framework) problem in historical hermeneutics and the idea of the theory-ladenness of facts in science will be too weak to illuminate what is at stake in evaluating historical interpretations. The unit of evaluation must be sought in the form appropriate to historical knowledge, and that, according to a rather established tradition, is narrative. 1 If the unit of evaluation is the historical narrative as a whole, what sort of (principles of) evaluation can and should be brought to bear on it? Before turning to the topic of narrative structure the mixed messages of my title I should stipulate something of my understanding of interpretation. Current usage seems to occupy a continuum between two opposed senses. First, in a very broad sense exemplified by Gadamer s philosophical hermeneutics, interpretation designates an element in any understanding whatsoever. This universal claim of hermeneutics seemed to Habermas, in his debate with Gadamer, illegitimately to preclude the possibility of a genuine social science. 2 Second, an extremely restricted sense invokes interpretation only when one encounters something puzzling, a potential misunderstanding. 3 Here interpretation resembles the deciphering of an unknown script. Arising from developments largely informed by phenomenology, the former view, while hardly ignoring the issue of language, has never abandoned appeal to notions like consciousness and experience. The latter view, on the contrary, has found resonance among those who, taking the linguistic or semiotic turn in philosophy, prefer to see interpretation as a relation between codes rather than between a subject and a code. Though my own sympathies lie in the phenomenological direction, I grant that at the present time the term historical experience denotes a problem and thus cannot uncontroversially support a philosophical account of how historical interpretations are to be evaluated. I shall thus adopt an idiom closer to the second 1. As Hayden White puts it in Narrativity in the Representation of Reality, The Content of the Form: Narrative Discourse and Historical Representation (Baltimore, 1987), 24, when historiography was transformed into an objective discipline, it was the narrativity of the historical discourse that was celebrated as one of the signs of its maturation as a science of a special sort. A concise history of the narrativist position can be found in F. R. Ankersmit, The Dilemma of Contemporary Anglo- Saxon Philosophy of History (originally published in History and Theory, Beiheft 25 [1986], 1-27) and The Use of Language in the Writing of History, both in History and Tropology: The Rise and Fall of Metaphor (Berkeley, 1994). 2. Cf. Jürgen Habermas, Zur Logik der Sozialwissenschaften (Frankfurt, 1967); Habermas, Der Universalitätsanspruch der Hermeneutik ; and H. G. Gadamer, Replik ; and Gadamer, Rhetorik, Hermeneutik, und Ideologiekritik, all in Hermeneutik und Ideologiekritik, ed. K. O. Apel et al. (Frankfurt, 1971). 3. This view, stemming from Schleiermacher, was revived and modified by Paul Ricoeur in his attempt to mediate the Gadamer/Habermas debate. See Hermeneutics and the Critique of Ideology, in Hermeneutics and the Human Sciences, ed. John B. Thompson (Cambridge, Eng., 1981).

222 STEVEN GALT CROWELL view of interpretation, in part because it permits discussion of the problem of heterogeneity in light of the semiotically framed positions of H. White, F. Ankersmit, and J.-F. Lyotard, but also because its allows for a clear formulation of the concept of heterogeneity itself, as the mixing of discursive forms or genres. The argument will show, however, the extent to which satisfactory grasp of the problem of evaluating historical narrative may require going beyond a purely semiotic framework. Let us consider the notion of an utterance as a distinct speech-act (for example, a description, command, question, request, prayer, promise, and so on). 4 In common-sense terms, interpretation is what produces an understanding of the act, one that becomes manifest in the auditor s response. To evaluate an interpretation thus amounts to deciding whether a given response is appropriate; to judge whether an interpretation is valid is to judge (in Wittgenstein s terms) whether the auditor knows how to go on. On what basis is that decided? Here the notion of a language game can get us further. Utterances are not linguistic atoms but are moves in games whose rules (usually unformalized) authorize specific sorts of responses (utterances, behaviors) as proper. This authorization is itself determined in light of the point the end, aim, or stakes of the game. Thus to interpret a given descriptive utterance as a move in the game of cognition (whose end or point is to establish what is real) is to grasp it as necessarily linked to another (possible) utterance, an ostensive statement in which the evidence establishing its truth is given. Only if I so grasp the description would it be proper for me to challenge it; the same utterance as part of a different game (say, in a work of fiction) will be misinterpreted if I challenge its factuality. The propriety of an interpretation, its validity, thus turns on the rules of the language game being played, rules that determine what counts as a move. This allows us to see an important fact about historical narrative, namely, its essential heterogeneity. By this I mean the following: if language games can be seen as genres of discourse which embody rules for appropriate linking of diverse speech acts (for example, description and ostention), narrative is a kind of discourse that essentially links or synthesizes different language games, or genres, themselves. And if that is so, a question arises about the rule it uses to synthesize games whose own rules may have nothing to do with each other and whose aims or stakes may in fact conflict. My argument is that historical narrative (as opposed to fictional narrative) necessarily involves links between (at least) two heterogeneous language games or discourses, each with its own aim the cognitive and the normative and thus poses the difficult philosophical problem of determining, first, the stakes of this sort of discourse, so that, second, we can 4. Since lengthy forays into the philosophy of language are out of place here, I offer these remarks as a simplified, idealized, and (I hope) relatively uncontroversial framework for discussing the real issue, namely, narrative meaning as such. My approach shares something with that of Jean-François Lyotard, The Differend: Phrases in Dispute, transl. Georges Van Den Abbeele (Minneapolis, 1988), but does not depend on the details of his theory.

THE HETEROGENEITY OF HISTORICAL DISCOURSE 223 see what a proper standard for evaluating it might be. That there is a problem here is shown by the fact that arguments over the evaluation of historical narrative tend to polarize between those who defend purely cognitivist criteria and those who advocate the primacy of political or moral ones. My claim is that this combination of analytically distinct (and so possibly conflicting) genres of discourse in which a distinctively narrative meaning arises is involved in all historical explanation. In historiography the issue of validity in interpretation comes to the fore once it is recognized that neither the annals, nor the chronicle, nor the research report is the adequate form of historical knowledge; that is, that historical explanation demands a moment of synthesis in which facts are linked according to an order that is not itself reducible to the structure of fact. As Hayden White puts it, events must be not only registered within the chronological framework of their original occurrence but narrated as well, that is to say, revealed as possessing a structure, an order of meaning, that they do not possess as mere sequence. 5 Replacing earlier attempts to ensure the scientific status of historical knowledge by appeal to covering-law models of explanation, narrative theory recognizes that the principles of such knowledge derive from the hermeneutics of meaning, that the synthetic order pertaining to the formulation of historical knowledge has more in common with fiction than with physics. 6 Hence (as White notes, recalling a point made by Louis O. Mink) the puzzle of such knowledge is that assessment of the truth-value of statements made in a historical narrative still does not provide us with any way of assessing the content of the narrative itself. 7 Here the heterogeneity of historical discourse poses its challenge to philosophical understanding. If, as Mink argued, a historical narrative claims truth not only for each of its individual statements taken distributively, but for the complex form of the narrative itself, 8 one must ask how meaning is constituted by this complex form, and (as I shall argue) to do this is to recognize that neither cognitive nor moral/political standards can fully capture what is at stake in historical narrative. The narrative form has its own point, or stakes, and it is with an eye toward that that a standard of evaluation must be uncovered. To adapt the title of Hayden White s collection, this is to seek the content of the form. Beyond what can be established cognitively, he writes, the narrative form demands a closure which is the vehicle of a moral meaning. 9 Having said that, howev- 5. White, Narrativity in the Representation of Reality, 5. 6. This is not to say that something like causal connections are not established in historical narrative. I would argue, though, that explanations such as The rise in bread prices led to the collapse of the distribution system derive their explanatory force primarily from their satisfying fit in the narrative, that is, from the narrative as a whole, rather than from exemplifying causal laws. I cannot go into this complicated question here, nor the question of how to evaluate those historical discourses which do make explicit use of putative economic or psychological laws. 7. Hayden White, Narrative in Contemporary Historical Theory, Content of the Form, 45-46 (originally published in History and Theory 23 [1984], 1-33). 8. Louis O. Mink, Narrative Form as a Cognitive Instrument, quoted by White, ibid., 46. 9. White, Narrativity in the Representation of Reality, 20.

224 STEVEN GALT CROWELL er, one may not conclude (as it sometimes appears that White is willing to conclude) that such narratives are most appropriately evaluated in moral or political terms. 10 In what follows I hope to show that political criteria, though certainly applicable, remain just as one-sided as do cognitive ones in assessing meaning in historical narrative. II. HETEROGENEITY AND NARRATIVE FORM We may divide the field of narrative theories into the robust and the ascetic. Robust views draw upon the extensive tradition of narrative theory in literature (beginning with Aristotle s Poetics) and attribute to the genre several quite specific features for example, that it traces a development, availing itself of potentia; that it orders the flux of time by establishing beginnings, middles, ends; that these elements adumbrate an overarching form, the plot, which distributes their weights and determines their significance; and so on. 11 Because robust views theorize narrative in terms of the way meaning is constituted in stories and tales, the presence in them of heterogeneous forms of discourse is obvious. But lest it appear that my argument turns on some contingent feature of the robust view, I will also consider a more ascetic approach (in section III). In F. R. Ankersmit s Narrative Logic all associations with the belles-lettres and with the story-telling kind of historiography should be avoided ; nothing is to be borrowed from the theory of narrative as a literary genre. 12 Diachrony, plot, and the like are treated as contingent features found in some historical accounts, but the narratio as such is nothing but a coherent system of sentences in which the statements that make it up serve a double function, namely, to describe a state of affairs and to construct a point of view. 13 Because this approach allows for historical narratives that would consist exclusively of singular descriptive statements, it denies that narrative meaning necessarily involves heterogeneous (cognitive and normative) discourses, and thus holds that cognitive criteria are adequate for assessing historical explanations. Nevertheless, I will argue that the key term in this theory point of view in fact implicates the normative order that narrative logic seeks to exclude as irrelevant to evaluation and points toward a dimension of our interest in history 10. In The Politics of Historical Interpretation: Discipline and De-Sublimation, Content of the Form, 80, for example, White argues, against Vidal-Naquet s inclination to consign the Zionist interpretation of the Holocaust to the category of untruth, that in fact, its truth, as a historical interpretation, consists precisely of its effectiveness in justifying a wide range of current Israeli political policies. This linking of truth with effective political justification is precisely the sort of confusion whose cause narrative heterogeneity I try to analyze in the present essay. 11. In literary theory the robust view is well represented by Frank Kermode s The Sense of an Ending: Studies in the Theory of Fiction (London, 1966). In philosophy, the most comprehensive expression is Paul Ricoeur, Time and Narrative, transl. Kathleen Blamey and David Pellauer, 3 vols. (Chicago, 1984-1985). 12. F. R. Ankersmit, Narrative Logic: A Semantic Analysis of the Historian s Language (The Hague, 1983), 12. 13. Ibid., 26-28.

THE HETEROGENEITY OF HISTORICAL DISCOURSE 225 that is better captured by more ontological approaches. Hence I examine one such, David Carr s attempt to ground historical narrative in the constitution of meaning at the level of lived experience, to see whether it yields a principle for evaluating narrative as a whole (section IV). Regrettably, Carr s narrative realism also elides the heterogeneity of historical writing and, like narrative logic, offers only an insufficient cognitive criterion. The lesson here is that one may retain the term truth as the criterion of narrative evaluation only if it is taken in a non-standard sense (as, for example, in Heidegger s conception of Erschlossenheit disclosedness, aletheia) which, since it concerns not facts but meaning, is cognitivist in name only. Thus the task of clarifying a distinct standard for evaluating the mixed messages of narrative historical knowledge remains unfulfilled or so I shall conclude. How, then, are heterogeneous language games subordinated to an emergent meaning by way of narrative form? We may begin with White s characterization of the distinctive aim or point of the narrative synthesis, namely that it endows reality with a form or formal coherence, gives it the odor of the ideal, a meaning, the completeness and fullness of which we can only imagine, never experience. 14 Frank Kermode describes this function as humanizing the world s contingency : narrative addresses the question of how to do justice to a chaotic, viciously contingent reality, and yet redeem it. 15 Lyotard makes the same point: The narrative function is redeeming in itself. It acts as if the occurrence, with its potentiality for differends, could come to completion, or as if there were a last word. 16 Narrative constitutes meaning by coming to an end and is thereby redemptive in two senses. First, it redeems or idealizes reality, salvaging (or forming) a meaning out of chaos. Second, it redeems the names and dates it mentions as one redeems a promissory note, gives them a fixed identity by fitting them proleptically into a completed context of significance. This function is already the mark of narrative heterogeneity, since the demand for closure, for redemption of the names, does not arise in the cognitive language game, in which confirmation of descriptives by ostensives (by evidence) is always understood as until further notice. Cognitive discourse thus resembles the incomplete narrative form of the chronicle. As White puts it, we cannot say, surely, that any sequence of real events actually comes to an end. 17 For this reason, purely cognitive discourse cannot bestow meaning, and meaning cannot be evaluated in purely cognitive terms. What else, then, is involved in the constitution of narrative meaning? White hints at the answer when he argues that to give meaning by coming to the end is to possess a principle for assigning importance or significance to events, which (he claims) is to identify a social center by which to charge them with ethical 14. White, Narrativity in the Representation of Reality, 20, 21. 15. Kermode, Sense of an Ending, 145. 16. Lyotard, The Differend, 151. 17. White, Narrative in the Representation of Reality, 23.

226 STEVEN GALT CROWELL or moral significance. 18 If this is true, at stake in any narrative is not only the redemption of names but equally the establishment of the right or authority of the narrator to speak as a meaning-giver, to pass judgment on the events. In White s terms, all historical narrative is informed by the moral authority of the narrator. 19 Since Thucydides, historians have often tried to establish such authority by appeal to superior knowledge of the facts, but things are not that simple. It is one thing to establish the reality of certain events by way of evidence and quite another to authorize oneself as a judge of their significance by pronouncing a certain kind of narrative closure. A philosophical problem is posed by historiographical practice, since the canons of cognition do not reach far enough to establish the validity of the historian s story. If we try to be more specific about how narrative accomplishes its meaninggiving function, we can say that in addition to the genre of cognitive discourse, all historical narrative includes the distinct genre of normative discourse. This amounts to the claim that evaluative terms are ineliminable from historical discourse if it is to succeed in giving meaning to (or finding meaning in) events. If one adopts a robust view of narrative, it is relatively easy to show that evaluative terms are not eliminable traces of subjectivity but essential to the way narrative constitutes meaning. 20 To characterize an action as evil, a development as decline, an invention as a boon (or to employ figures in which such judgments are sedimented) is to interpret the sentences describing such things as moves within a normative language game whose rules are other than those of the cognitive, but without which there is no narrative (as opposed to causal) making sense of the facts. 21 Whether this sort of normative discourse will prove to be eliminable in more ascetic views of narrative must be considered later. But first, a more precise characterization of the normative genre must show that, because its rule does not link descriptives with ostensives, its stakes must be other than truth. The paradigm of normative discourse is, of course, moral discourse. When I say it is wrong to tell a lie or the unequal treatment of women is unjust, it might appear that these statements are simple descriptives functioning as such in the cognitive language game. However, the actual pursuit of moral argumentation suggests that this appearance is misleading: the descriptions cannot be cashed in by ostensives (evidence); the wrongness or injustice cannot be pointed to; even 18. Ibid., 11, 14. 19. Ibid., 21. 20. Hayden White s work on figurative language can be seen as showing this. See, for example, Historicism, History, and the Figurative Imagination, Tropics of Discourse (Baltimore, 1978), 101-120. But in addition to figure and trope, historical narrative meaning unavoidably avails itself of a certain prescriptive language, as I try to show below. 21. There is, of course, a sense in which terms like decline or boon need not imply any judgment. For example, decline might simply mean that fewer goods were exported and boon that more crops were harvested. But it seems to me that if these words are so understood, they will be unable to supply the kind of meaning which is at stake in the narrative. The significance, importance, and so on of the decline in exports or the increase in the harvest will not yet have been established.

THE HETEROGENEITY OF HISTORICAL DISCOURSE 227 if all the facts are agreed upon, it is still possible reasonably to dispute the judgment. It is therefore necessary to look for a different language game (with different rules and aims) which govern their functioning. One way of looking at such statements is as reformulated commands Do not steal or Do not engage in unequal treatment hence as deriving from the language game of prescription whose point is obedience (compliance or refusal). When, instead of obeying or refusing, I question the authority of the command ( Why shouldn t I? ), the game is altered, giving rise to the normative statement, Because stealing is wrong. But how is the authority of this statement established? Recalling White, we may say that it presupposes the moral authority of the narrator, a thing generally established by a certain relation to authority (social, political, theological, and so on) per se. 22 Normative statements thus appear as narrative transformations of prescriptions; they aim to legitimate certain obligations by referring them to the authority (at bottom, self-authorizing) of the addressor of the narrative. Stealing is wrong means Among us it is a norm that one shall not steal, where the us, even if it appears to refer to a given group, state, or culture, can be established only by the statement s being acknowledged as normative by auditors of the narrative (in that they identify with it). 23 Thus the normative genre, the narrativization of the language game of prescription, aims not at truth but at justice and reflects, as Lyotard puts it, the idea of a certain humanity. 24 Perhaps an example will get us further. Some historical narratives are constructed around normative terms in obvious ways: Marxist narratives of history as the story of the exploitation of labor, bourgeois liberal history as the story of moral progress, romantic narratives as plotting the growth and decline of things, and so on. Here also belong certain current versions of feminist historiography in which one finds not merely something like a history of women appended to other existing narratives, but instead under the idea of patriarchy a wholly new point of view on the past, an interpretation of that past as the story of the oppression of women by men. Importantly, such a narrative is in principle independent of any new facts. It need not be based on any new discoveries, nor does it necessarily lead to any (though it might). Rather, it is a new way of interpreting the already existing facts. If that is so, however, cognitive criteria will be insufficient for evaluating this narrative in the face of a competing, extensionally (factually) equivalent, one. What else is at stake, then, and what sort of evaluation is appropriate to it? To see, let us unpack the heterogeneity at work in the notion that distinguishes the feminist narrative, namely oppression. To say that women (proper names, dates supplied) were oppressed is 1) to describe certain ways in which women were treated (descriptions reflecting the obvious and not so obvious hierarchies, constraints, and so on) and 2) to judge that You shall not (ought not to) treat women 22. White, Narrativity in the Representation of Reality, 19. 23. The importance of this point will be seen below in my discussion of Carr, 242. 24. Lyotard, The Differend, 136, 155-157.

228 STEVEN GALT CROWELL that way. This last distinguishes the narrative of oppression from a competing narrative which accepts the descriptions but judges that You shall (ought to) treat women that way. 25 To argue that women felt themselves to be oppressed, and that therefore the judgment is wholly descriptive, will not succeed in removing the heterogeneity. To begin with, the narrative that explains in terms of oppression will not want to restrict its scope to those women who in fact felt oppressed, and it will properly draw upon some version of the false consciousness thesis. But even if one did so restrict the scope, the most one could demonstrate is the reality of the feeling of being wronged, not the wrong itself. Does this mean that there are no true judgments about oppression about being oppressed, as opposed to feeling oppressed? That is a central question posed by the heterogeneity of historical discourse. Lyotard suggests that the aim of such judgments within historical narrative is not truth but justice. The prescriptive, Do not treat women that way, is narratively embedded in judgments that imply the normative statement, It is wrong to treat women that way, which in turn can be understood as asserting that Among us (or we decree) it is a norm that women not be treated that way. In other words, in every narrative an addressor authorizes the prescription turns it into a norm and thereby petitions addressees to identify themselves as in community with the author. The normative, writes Lyotard, constitutes a community of addressees. 26 Should one wish to argue that the community is already constituted outside the narrative and that therefore we should be able to establish the authority of the norm cognitively, one must explain how, apart from the narrative, the specific community at issue is to be identified. The narrative genre presents us as established by the narrative, the community for whom the norm signifies justice and can establish a meaning for the names and dates found in the narrative. To support this point, consider the alternative (hypothetically extensionally equivalent) male chauvinist narrative which holds that the hierarchies and constraints described by the feminist do not signify oppression but rather proper order, the nature of things. The we constituted by the normative in this narrative can hardly be said to be the same as the one in the feminist s narrative, yet the actual addressors and addressees may well inhabit the same political space and time. Thus conflicts between these narratives cannot be adjudicated on the basis of their respective cognitive merits (that is, that one more truly describes some pregiven community). One might try to tell a third narrative which somehow encompassed both, such that their contradictory constituencies (and the disputed meanings they assign to their names) could be taken up proleptically into 25. Could a narrative constructed around the notion of oppression be taken to reveal a positive normative judgment perhaps a Nazi history of the Reich s treatment of the Jews? As commonly understood, oppression means an unjust exercise of power or domination; hence such a narrative would be possible only if the object of domination were first morally demonized such that the injustice of the dominating behavior was neutralized. But then oppression seems to me the wrong word to use. 26. Lyotard, The Differend, 143.

THE HETEROGENEITY OF HISTORICAL DISCOURSE 229 the story of a larger, more inclusive community. But this narrative would still be the story of a particular We. Or one might appeal to non-narrative protocols of meaning-adjudication (rational argumentation, say, but also sheer agon, politics ) to establish which narrative, which interpretation, is better. 27 How is the question of justice to be adjudicated? Is it possible to construct, on purely historical grounds, a meta-narrative that discerns, proleptically, the contours of a just community? Or can one propose a non-narrative adjudication by means of philosophy conceived as a kind of meta-cognition that establishes the norms of a universal humanity and so grounds the legitimacy of prescriptions in (transcendental) truth? Or are we left with politics, that agonistic space in which, because it acknowledges only the reality of power, norms can be conceived only decisionistically? Having reached the point where such questions arise, I should emphasize that my claim is not merely that historical narratives employ the heterogeneous regimens of the cognitive and the normative a fact that needs little demonstration but rather that the specific sort of meaning we have found to be at stake in historical narrative (one that establishes a significance and identity, through closure, for the names invoked in the narrative) is irreducible to either, that is, is a genuine synthesis which is other than the sum of its analytically distinct parts. Thus, even though the establishment of facts does not depend on the historian s values, it is still not possible to assess the validity of historical narrative by focusing on its cognitive dimension while consigning the normative to the extrinsic use to which the narrative is put in political and other contexts. Conversely, while there is no such thing as a purely factual narrative, this does not mean that narrative validity lies exclusively in its moral or political consequences. But is it indeed true that there is no purely factual narrative? It might be objected that heterogeneity has appeared irreducible only because we have taken our point of departure from robust views of narrative form found in thinkers like Kermode, White, and Lyotard. If a more ascetic approach can account for the constitution of historical meaning while excluding all normative, non-cognitive moments, then we have not succeeded in showing that cognitive criteria do not suffice for the evaluation of historical discourse. Let us turn, then, to one such approach. III. AN ASCETIC APPROACH TO NARRATIVE: ANKERSMIT S NARRATIVE LOGIC F. R. Ankersmit proposes a theory of historical narrative which denies the essential heterogeneity of historical discourse. While acknowledging that historical 27. This alternative is at issue in Habermas s and Apel s various proposals for a Diskursethik. The fundamental conviction behind their views is well expressed by Apel: to appeal outside narrative to the genre of philosophical justification is to assume that only in this language game [philosophy] can all claims to validity, that in other language games may be exposed as debatable, be redeemed or refuted in a sense that is not merely conventional. The appeal thus presupposes that the language game of argumentative discourse is foundational. K. O. Apel, Pragmatic Philosophy of Language Based on Transcendental Semiotics, Selected Essays, ed. Eduardo Mendietta (Atlantic Highlands, N.J., 1994), I, 245.

230 STEVEN GALT CROWELL writing may involve various discursive forms (commands, questions, evaluations, and so on), he argues that historical meaning can be generated, in principle, exclusively on the basis of (ideally) true, singular statements. 28 Historical narratives, and so the explanations they embody, require no theoretical statements, no universal generalizations, and above all no normative statements. The narratio (Ankersmit s term for this core of statements) constructs an interpretation of the past for which the singular true statements are arguments (evidence), and while the arguments cannot be said to demonstrate the truth of the interpretation (for reasons we shall come to), evaluation of the interpretation takes place exclusively in cognitivist terms. If this view of how historical meaning is constituted is correct, there are no heterogeneous standards of evaluation, no diverse stakes (such as truth and justice) in the genre itself. So is it correct? As a way of managing this large question, I shall try to show the presence of a kind of heterogeneity without which a thesis central to this view that the narratio provides an order that transcends a mere collection of sentences cannot be redeemed. 29 What makes Ankersmit s view so challenging is its formalism. Its ascetic rejection of the relevance of the poetics of narrative for historiography stems not merely from the linguistic turn in philosophy which he shares with Lyotard and White, but within that, from a logicism according to which the philosophy of history consists solely in the logical analysis of historical discourse. For Ankersmit it is too much to say, with Kermode, that narrative slakes the need to speak humanly of a life s importance in relation to the abyss of time. 30 This would tie narrative to a hermeneutic philosophy of human interests, whereas Ankersmit believes that current philosophy of history and of the social sciences makes far too many material assumptions about the nature of human beings, assumptions that belong to the domain of history and the social sciences themselves and not to philosophy. 31 Thus, where the Diltheyan tradition sought to ground the possi- 28. Ankersmit, Narrative Logic, 27-28. 29. Since Ankersmit s recent work introduces notions such as historical experience and representation that might seem incompatible with the central theses of Narrative Logic and since in fact he seems to contrast his new aestheticist approach with his earlier transcendentalist one ( Introduction, History and Tropology, 28) some justification for concentrating on the earlier position is necessary. First, the new position exists so far only as a series of essayistic hints, suggestive analogies, and often brilliant aperçus, but not in the systematic form found in Narrative Logic. In the latter, then, one has a better chance of uncovering answers to a variety of specific questions. Second, that Ankersmit later comes to proclaim something like the end of narrative history (ibid., 18-19) is, given his early ascetic view, not so much a repudiation of Narrative Logic as a restatement of the idea that stories are not necessary to historical meaning. Finally, it does not appear that the later essays advance our understanding of the standard in terms of which one ought to evaluate historical writing. For they oscillate between a theory of representation, on the one hand, that, in rejecting epistemology (and so also metaphor, narrative, and modernity ), seems to reject the very question of assessment of adequacy (though in fact it undermines only the appropriateness of assessment in terms of correspondence to reality a position already reached in Narrative Logic) and, on the other hand, a postmodern aestheticism that never raises the question of evaluation as it actually appears in aesthetics, namely the question of the standards by which we judge the success or failure of a work of art. 30. Kermode, Sense of an Ending, 4. 31. Ankersmit, Narrative Logic, 57.

THE HETEROGENEITY OF HISTORICAL DISCOURSE 231 bility of historical knowledge in a certain congruence between the historian and his or her object (insofar as the historian s self-understanding is said materially to enable an understanding of the meaning of past events), Ankersmit argues that this congruence is merely formal, that is, that the historian s self and the aspect of the past he or she understands are both narrative substances. 32 Our question thus becomes whether historical meaning (interpretation) can plausibly be understood in abstraction from the poetics of genre and from philosophical hermeneutics. Ankersmit s understanding of historical narrative is based on the claim that the singular true statements in a narratio have a double function: while describing facts, they also construct narrative substances (interpretations), that is, points of view on the past. Terms like Renaissance, Patriarchy, and Holocaust name such narrative substances, a term chosen because these images or pictures... are things, not concepts. 33 Though coming into being through statements, narrative substances are not texts which include statements as constituent parts, but rather, thanks to the double function of statements, simple entities that include the statements as properties. 34 Being substances, narrative substances do not designate or refer to the past or to anything else. In certain cases, the same term (say, Napoleon ) can name a narrative substance and a real individual or narrative subject, but in most cases narrative substances either have no names or, if they do (as with The Enlightenment or The Cold War ), sentences in which those names appear in the subject position refer to the narrative substance and not to some piece of the past. Ankersmit devotes considerable effort to explaining why some terms can name both narrative substances and narrative subjects while others cannot, and something of his reasoning needs to be considered in order to appreciate his claims for narrative idealism. A baby view of the world is posited in which no identities are pregiven. 35 Gradually, by a process in which the construction of narrative substances plays a crucial role, intensional types or concepts that is, terms with stable criteria for re-identifying what falls under them come into use. Where we have such intensional types (as in the case of sortal terms like tree or human being ) we can speak of things existing in reality, that is, apart from the narrative substances in which they may be individuated for me in my idiolect. Where we are dealing with narrative substances for which no such inten- 32. Ibid., 193. Ankersmit sees his project as a transcendental one in the tradition of Rickert s neo-kantianism, which he criticizes as insufficiently formal (ibid., 83, 94). The linguistic turn allows him to develop a specifically narrative idealism in which the transcendental subject of earlier neo- Kantianism becomes dispensable: the I in statements such as I am in pain is the name of a narrative substance and not of a thing in extra-linguistic reality (ibid., 187). It is not clear to me whether Ankersmit s recent appeal to Gadamer s notion of historical experience amounts to a rejection of this formalism, such that material views about human beings are once more admissible in the philosophy of history. 33. Ibid., 100. 34. Ibid., 100-101. 35. Ibid., 155-156.

232 STEVEN GALT CROWELL sional types have been generated as in all cases where we have to do with something historically particular (such as the Renaissance ) what we mean can only be specified by means of a complete enumeration of all its properties. 36 Thus Napoleon can be identified as a kind ( human being ) by means of sortal predicates and can also be individuated (by a narrative substance), whereas the Renaissance can only be individuated. 37 Now this view is certainly open to challenge, but it is more to our purpose to note that the narrative substance (interpretation) constructed by the individual factual statements is a proposal to see a whole range of unspecified aspects of the past in light of a certain idea, to see the past as something. 38 Narrative substances illuminate or reveal the past in particular ways, but just for this reason they cannot be said to be true or can be said to be true only in a non-standard sense, which should be avoided in philosophical discussions. 39 According to standard truth-functional logic, a statement is true or false depending on whether its terms refer. Because the narratio as a whole does not refer, then, it cannot be true as a whole. Thus one cannot evaluate competing narratives about the Renaissance by saying that one is truer, or better corresponds, to the Renaissance. What then would make one adopt one point of view over some other? Here one might expect some of the problems of evaluation we encountered earlier, based on the heterogeneity of narrative, to reappear in Ankersmit s deliberation, but that is not the case. For example, Ankersmit argues that though the historian s values may play an important role in suggesting points of view, these values play no constitutive role in the narrative substance. 40 There is no heterogeneity in the core of historical discourse; hence we need no criterion of evaluation that would take such heterogeneity into account. Instead, one narrative substance is superior to another if its scope is greater. The best, the most adequate or the most objective narratio out of a set of competing narratios around a historical topic is the narratio of which the scope beyond its descriptive content has been maximalized (other things being equal). 41 Scope, in turn, is understood as 36. Ibid., 118. 37. For this reason, too, it will be impossible to treat terms like Renaissance as though they were theoretical terms in science, that is, terms whose reference is guaranteed by their function in the whole system. For such terms (for example, quark ) are essentially sortal, possessing reidentification criteria, whereas the names of narrative substances are proper names. 38. Ankersmit, Narrative Logic, 217. 39. Ibid., 77. Heidegger s notion of Erschlossenheit would be an example of such a non-standard sense of truth, as would the very similar view of Hayden White when he writes that [o]ne can produce an imaginary discourse about real events that may not be less true for being imaginary, suggesting that maybe the question of narrative in any discussion of historical theory is always about the function of imagination in the production of a specifically human truth. Narrative in Contemporary Historical Theory, Content of the Form, 57. Ankersmit continues to avoid such a non-standard conception in the later essays, for example, in The Use of Language in the Writing of History, History and Tropology, 95. 40. Ankersmit, Narrative Logic, 241. Cf. the rejection of the pragmatist proposal for evaluating narratives in light of their contribution to a better future, ibid., 31-35. 41. Ibid., 238.

THE HETEROGENEITY OF HISTORICAL DISCOURSE 233 the number of facts the narrative substance allows us to grasp beyond the ones stated in its construction. One point of view discloses a richer view of the past than another: [A]ll the states of affairs described by the statements that can be meaningfully related to the statements of the narratio together constitute the scope of the narratio. 42 We must examine the qualifiers in these passages ( other things being equal and meaningfully related ) in a moment; for now we may simply note that fertility and not truth is our criterion for deciding upon the relative merits of narratios. 43 Yet if scope (the criterion of the best narratio) is understood as number of facts, fertility is itself ultimately defined in terms of truth. Thus the philosopher of history, qua logician, can apparently distill from the phenomenon of historical meaning a criterion based on the cognitive aim of getting facts into view. Without impugning the importance of the scope criterion, it is not obvious that it exhausts what the philosopher of history can or should say about the issue of evaluating historical narrative. For one thing, if, as suggested above, the feminist and male chauvinist narratives may be entirely equivalent in scope, then by that criterion they are equally good as historical explanations. It might be said that we will choose between them on other grounds. But that is to admit that two interpretations so vastly different in meaning as these are not different as history, or else one will need some way to link the relevant new criterion to the point of view, or meaning, constituted in the narrative. This may become clearer if we consider the possibility of a clean separation of the scope criterion from other elements in the construction of a point of view. Ankersmit acknowledges that the sort of normative statement we analyzed above on the example of oppression does show up in historical discourse, but he denies that it belongs to the interpretation in the logical sense. 44 Instead, it is a judgment upon the interpretation. England went downhill after the war means 1) that a narrative substance has been constituted from factual statements that propose a pessimistic point of view England lost prestige, England declined in GNP, England lost its colonies and 2) that I judge This [narrative substance] is an acceptable assessment of part of the past. 45 Hence no normative claim is made, so long as the idea of an acceptable assessment can be understood wholly in cognitive terms. But can it? Here, two points can be made. First, a statement such as The fifth century in Athens saw an increase in the oppression of women does not seem to be analyzable along the lines suggested by Ankersmit, that is, so as to avoid invoking the heterogeneous genre of normative claims. It is a judgment on the justice of certain types of behavior, not a judgment on the adequacy of the assessment of the past composed of statements describing that behavior. To the extent that such a judgment contributes to the meaning of the narrative, then, to evaluate it as an 42. Ibid., 222. 43. Ibid., 223. 44. Ibid., 102. 45. Ibid., 179.

234 STEVEN GALT CROWELL acceptable assessment of the past will go beyond consideration of its scope and will require some stand on the question of justice. Second, in places Ankersmit seems to acknowledge the difficulty, if not impossibility, of a clean separation between normative and descriptive contributions to the narrative whole. Given his claim that narrative substances consist of singular true sentences, it is surprising to read that [h]istorical points of view... may often be inextricably tied up with ethical or political values such that many narratios lose their internal consistency when robbed of the political values which structure them. 46 If so, the scope criterion cannot be isolated. Wishing to hold heterogeneity at bay, Ankersmit describes historical narrative as the trait d union between the descriptive and the normative whereby on the one hand we have a set of descriptive statements, on the other a course of action is recommended. 47 Yet if narratios may lose their internal consistency when robbed of the political values which structure them, why should we think that any narratio ( acceptable assessment of part of the past ) with the power to recommend a course of action could be composed entirely of true singular statements? To see the problem, assume that normative statements can be eliminated from the narrative so far as logic is concerned. How then is the discourse able to define a point of view (or meaning) at all? Because this is to be accomplished solely through singular true statements no theories, no evaluations, no special pleading the point of view must derive from the choice of facts mentioned in the narrative. 48 Thus the feminist historian will describe particularly stunning instances of asymmetrical treatment of men and women, while the historian of decline will note statistics and describe conditions that constitute strong arguments (in Ankersmit s terms) for the point of view he or she wishes to propose, since they adumbrate (suggest? evoke?) a meaning that ex hypothesi the past itself does not possess. 49 The difficulty especially if one limits oneself to logic is to explain how and why certain statements and not others have this power. Ankersmit claims that in the narrative such statements function like metaphors. Metaphors, on his view, state a description and suggest a way of looking at things. 50 Statements like Athenian women were considered the property of their husbands resemble Stalin was a wolf in that they are at once descriptive and suggestive, inviting us to see a whole bunch of things not described in them in their light. Not every statement is equally powerful in this regard, and narratios in which weak statements of fact predominate approximate the non-narrative condition of research reports. This is an intriguing proposal for understanding the genesis of narrative meaning, but it remains in certain respects problematic. For one may still ask how sin- 46. Ibid., 242, my emphasis. 47. Ibid. 48. Ibid., 138. 49. For the argument against narrative realism the idea that the past itself has a narrative structure see ibid. 79ff, esp. 86-87. 50. Ibid., 209-216.