A Narrativist Revival?

Up till the 1980s narrativist philosophers of history were mainly interested in the cogni-tivist dimension of historical narrative. With Hayden White this interest was exchanged for an exclusive preoccupation with the literary aspects of the historian’s narrative representation of the past. However, it may seem that a revival of pre-Whitean narrativist philosophy of history is at hand. Two recent books suggest as much: one by Chiel van den Akker published in 2018 and one more by Paul Roth that came out in 2020. Obviously, a narrativist revival can take two different forms. It may aim at providing pre-Whitean narrativism with a more up-to-date philosophical basis or at guiding it into new directions. It will be argued in this review-essay that the book by Roth mainly does the former, whereas the book by Van den Akker does both.


Introduction
We all know the story. After WWII debate in philosophy of history began with the discussion of the covering law model, discussion then moved on to Collingwood's re-enactment theory and, next, to Danto's and Mink's narrativism and the constructivism of Leon Goldstein.1 One more new phase began with Hayden White's Metahistory of 1973. White's name is often mentioned in one and the same breath with the narrativists mentioned just now. But for two reasons one may have one's doubts here. White was more interested in how history manifests itself in literature, film, art, psychology and politics than in professional historical writing2 and, furthermore, he appealed to literary theory when analyzing narrative, whereas Danto, Mink and Goldstein took philosophy of language as their guide. It is true, White enriched historical thought with a plethora of wholly new questions that were investigated by him with an unparalleled brilliance and originality. This may both explain and justify his predominance in historical thought for almost half a century. But a price had to be paid for this. The issues Danto, Mink and Goldstein had put on the agenda were now laid to rest and it seemed ever less likely that anyone would feel tempted to address them again.3 But the good news is that a narrativist revival seems to be at hand. A first sign was Jouni-Matti Kuukkanen's Postnarrativist Philosophy of Historiography of 2015. Admittedly, the book's title was meant to suggest a moving beyond narrativism. Nevertheless, such attempts always remain indebted in one way or another to what they propose to leave behind.4 In fact, Kuukkanen himself never denied the debt he owed to narrativism. Next, in 2018 and 2020 two profound studies were published returning to the narrativist thesis itself: Van den Akker's The Exemplifying Past and Roth's The Philosophical Structure of (Ithaca/London: Cornell UP 1987), L. Goldstein, Historical Knowing. (Austin: University of Texas Press, 1976); though Goldstein is not a narrativist in the strict sense of the word, the philosophical structure of his historical thought comes quite close to that of the narrativists). 2 'Metahistory was perhaps the last occasion on which White invited professional historians to change their mode of realism. In the 1980s White came to the sobering insight that a nonironic realism is more likely to be found outside the historical discipline. Encouraged by the rather enthusiastic reception of his tropology among students of literature, White began to put his hope more in novelists and film directors than in historians. Whereas dispel some persistent worries about the reliability of historical knowledge, such as relativism.
Van den Akker takes Davidson as his guide when exploring the other mind problem. The relevant part of Davidson's theory of interpretation is shown by him to rest on three assumptions only: 1) we have to know the truth-conditions of the utterance expressing a speaker's belief, 2) we should assume that the speaker holds his or her beliefs to be true and 3) we should be able to relate the belief to other presupposed but unmentioned beliefs.7 We might wonder why Davidson included the third condition. The explanation is as follows. Suppose we read that the Greeks believed that the earth is flat. We may then jump to the conclusion that this belief of the Greeks was false. But that's going too fast. We should ask ourselves, what the Greeks had in mind when saying that the earth is flat. Perhaps it was meant to say no more than how the earth's surface presents itself to us in daily experience. Now, this question can only adequately be answered against the background of a number of truths shared by the Greeks and us. Obviously so: as long as our Greek interlocutor comes up with statements we regard as false, our perplexity will only increase. Only shared truths can help us out. Generally speaking, we can only identify a belief as false against a backdrop of many true beliefs.8 It follows that there is a curious asymmetry between falsity and truth: falsity stands out against a background of truths, but truth does not stand out against a background of falsities (the problem with Trump, so to speak).
Finally, this imaginary debate between our Greek interlocutor and ourselves can only work if we can be sure that the second condition mentioned above is fulfilled, hence, if the speaker believes himself to be telling the truth (even though he might be proven wrong in the end). We have, therefore, no choice but to hope/expect that our interlocutor intends to speak the truth. Davidson speaks in this context of 'the principle of charity' -though Van den Akker would prefer to speak of 'the principle of humanity' .
It is of the greatest importance to see that the recognition of falsehood or false belief as accounted for by Davidson by no means precludes the understanding of false belief by the interpreter of it (as Giuseppina d'Oro and Serge Grigoriev mistakenly argued). If someone refuses to cross mountains because he believes that devils live there, we will condemn the belief as false, but have no difficulty in understanding it. For one thing, how could we condemn a belief as false if we don't understand it? In such a case we can only suspend judgement. Moreover, as Van den Akker goes on to say: 'the three basic conditions of understanding others guarantee that the belief is understood relative to a large set of unmentioned related and rational beliefs. The root of d'Oro's and Grigoriev's misunderstanding of Davidson's ideas is, I think, his claim that interpretation is directed at maximizing agreement. How to maximize agreement with someone who refuses to cross mountains because he fears devils?'9 That, indeed, might not be easy. But, again, it presupposes rather than excludes understanding. This issue is of so much importance since it suggests where Davidson's theory of interpretation is superior to many variants of hermeneutics. Hermeneutics is often unclear and/or unduly pessimistic (think of the hermeneutic circle) about what understanding is and what it can achieve. Therefore the uncertainties of hermeneutics often provoke a proclivity to relativism: what it fails to grasp is then surrendered to prejudice and arbitrariness. Davidson's three conditions for understanding, on the contrary, give a clear and consistent definition of what understanding is, how it can be achieved and what its results will look like.
This does not mean that Davidson's theory answers all questions, nor that all of traditional hermeneutics should be rejected. For example, unlike Davidson, Collingwood focuses not on statements but on the thoughts of individuals in the past. Statements are context-free, whereas thoughts are not. This may already give an idea of where Davidson and Collingwood will differ: the historian has to find out about the 'thoughts' of a historical actor whereas this is not part of Davidson's program. For Davidson the first condition of understanding an utterance/statement is that we know its truth-conditions. But what is then still left out is what is presupposed by the utterance -and that belongs to the context in which it was made. For example, we may want to know why someone uttered a certain statement. The question is irrelevant from Davidson's point of view, whereas it is, as Collingwood argued, at the centre of the historian's interest. We should therefore consider Collingwood's re-enactment a most welcome supplement to Davidson's theory of language, and vice versa.
A moment ago we already observed Davidson's anti-relativist stance. But Davidson has a far more powerful argument against relativism -an argument, moreover, attacking one of the main sources of relativist interpretations of historical writing. It has often been said that each language, both of past and present, has its own 'Weltanschauung' ,10 its own way of interpreting the world. ) Hence, the idea is that each natural language determines how we will see and experience the world and it does so 'from top to bottom' , so to speak. No part of reality is excluded from how it is interpreted by the language we happen to use. A no less paradigmatic formulation of the spell cast by language on the relation between word and world is the so-called Sapir-Whorf hypothesis. Sapir defined it in 1929 as follows: 'It is quite an illusion to imagine that one adjusts to reality essentially without the use of language and that language is merely an incidental means of solving specific problems of communication or reflection. The fact of the matter is that the "real world" is to a large extent unconsciously built upon the language habits of the group. No two languages are ever sufficiently similar to be considered as representing the same social reality. The worlds in which different societies live are distinct worlds, not merely the same world with different labels attached (…). We see and hear and otherwise experience very largely as we do because the language habits of our community predispose certain choices of interpretation' . (Sapir 1958(Sapir [1929 bits of knowledge that characterize one person have no true counterparts for the subscriber to another scheme. Reality itself is relative to a scheme: what counts as real in one system may not be in another' .13 Davidson rejects conceptual schemes -either one conceptual scheme that is shared by us all (like Kant's categories of the understanding) or the kind of conceptual scheme in the plural that Sapir and Whorf had in mind. (Indeed, the notion of conceptual schemes often stands for languages, both natural and artificial.) If, then, these conceptual schemes are abandoned, there is nothing for truth and reality to be relative to -and that means the end of conceptual relativism. Van den Akker summarizes Davidson's argument as follows: Davidson also claims that there is no such thing as 'uninterpreted reality, something outside all schemes and science' . Perhaps a conceptual relativist would immediately object that his position is precisely that there is no such thing as uninterpreted reality, the data of sensation, or the passing scene. He claims after all that reality is relative to some conceptual scheme and what is real according to one culture's or epoch's scheme is not in another.
But take good notice of what goes on here. Now the conceptual relativist nota bene says himself that there can be no such thing as an uninterpreted reality! Davidson and Van den Akker can now head-in the ball: It is rather easy to see why this objection fails. When the notion of uninterpreted reality is the necessary counterpart of the notion of conceptual scheme, and the conceptual realist rejects the former, then he cannot hold on to the latter either, and with that to the idea that reality and truth are relative to it.14 In one sentence, if you abandon the idea of an uninterpreted reality, you will also have to abandon relativism. The dichotomy/dualism of scheme and content -presupposing there to exist, on the one hand an uninterpreted reality or content and, on the other, a scheme for interpreting it -then goes overboard as well.15 As Van den Akker put it: 'the dualism of scheme and empirical content, 13 Van den Akker, op. cit., 39. 14 Van den Akker, op cit., 40. 15 Van den Akker concludes his discussion of the conceptual scheme with an exposition of Davidson's claim that truth and translatability are indissolubly linked together -the implication being that however much two languages may differ the speakers of these language can nevertheless, in principle, communicate meaningfully with each other on "of organizing system and something waiting to be organized", as Donald Davidson famously argued, "cannot be made intelligible and defensible"' .16 The rejection of the dichotomy of scheme and content is a recurring theme in Van den Akker's book and I would not hesitate to see it as one of the main pillars under all of his arguments in it.17 For example, it supports Van den Akker's categorical dismissal of the literary approach to historical writing as advocated, for example, by Hayden White. Think, in this context, of the tropological grid White had proposed in Metahistory. The four tropes, modes of argument, emplotment and ideological implication defining that grid function here as a formal scheme determining what content(s) are possible in historical writing. Indeed, especially suggestive here is the similarity between Kant's categories of the understanding and White's tropology: both are the transcendental conditions that make knowledge possible -either knowledge in general (Kant) or historical knowledge (White). And in both cases there is an empirical content (sensory perception or historical data) waiting to be organized by these schemes. Indeed, Kantianism is the prototypical example of philosophies separating scheme and content and we can expect to find the dichotomy as soon as the smell of Kantianism is in the air.18 Moreover, as Van den Akker emphasizes, in history narrative is not a literary artefact (White) but a cognitive instrument.19 Taking all this together he concludes 'that the vocabulary of literary theory is of no help in understanding the problem of narrative truth' .20 As we saw a moment ago Davidson had also included 'points of view' in his definition of conceptual schemes. But elsewhere he is more tolerant with regard to the former: 'different points of views make sense, but only if there is a common coordinate system on which to plot them: yet the existence of a condition that they aim to speak the truth. Obviously, this is one more argument against relativism. Van  common system belies the claim of dramatic incomparability' .21 Hence, the difference between the conceptual schemes and points of view is, basically, that whereas the former have the pretension to exclude each other totally, there always is some common ground in the case of points of view. Think of a table that is seen from different points of view. On the one hand we have then different points of view of that table -e.g. one may see the table from the left, the right or from above, but on the other (and wholly apart from them) there is the table itself. But these points of view do not conflict with each other in the way the relativist is in the habit of saying that different conceptual schemes conflict with each other. Points of view and the table are all situated in one and the same three-dimensional space and geometry accurately defines how they are related and how a table will be perceived from a certain point of view. And though the way the table is seen will differ from different points of view, these do not conflict with each other. So again no solace for the relativist; moreover, a rational decision can be made about which point of view gives us the best idea of the table in question.22 Finally, Van den Akker insists that the 'other mind' problem as analyzed by Davidson should be sharply distinguished from the quite different problem of 'other periods' . It is mainly thanks to the clarity brought by Davidson to the 'other mind' problem that we now know these two problems to be widely apart and why hermeneuticists were wrong when not clearly differentiating between the two. To put it succinctly, whereas the other mind problem deals with the problem of what we can say about the actions of a historical agent from a perspective accessible to him, the 'other period' problem focuses on what one may say about an agent's actions from a later perspective. This is of especial importance for the writing of history since what can be said about a historical agent's actions from a later perspective normally outweighs by far what can be said about them from the perspective of the contemporary. Just think of the doctrine of the so-called 'unintended consequences of intentional 21 Van den Akker, op. cit., 44. 22 Bearing in mind Davidson's observation 'if we cannot intelligibly say that schemes [i.e. the schemes that points of view have in common (F.A.)] are different, neither can we intelligibly say that they are one' (Van den Akker, op. cit., 45) one might say that points of view go as far into the direction of conceptual schemes as one might legitimately go. It is of all the more interest, therefore, that the notion of the point of view was used two and a half centuries ago already by Johann Martin Chladenius (1710-1759) for a clarification of the nature of historical writing. He did so in his Allgemeine Geschichtswissenschaft of 1752, which one may well see as the first work on critical or analytical philosophy of history. In his theory of the 'Sehe-punckt' (i.e. point of view) he argued, too, that differences of points of view do not necessarily exclude agreement about truth and that the real question is what point of view reveals more about the past than other(s). human action' .23 History essentially is the science of the ex post facto. Historical wisdom always is a wisdom after the event; as Hegel famously declared: 'erst in die Dämmerung beginnt die Eule der Minerva ihren Flug' . Put differently, historical reality comes into being when the past no longer exists -which has the peculiar implication that historical reality is a reality that never was experienced by anybody dead or alive. It deploys itself in a limbo between the past and the present in which no one ever lived.24

The Metaphysical Status of Historical Events
Let's begin with Arthur Danto's well-known argument about the 'Ideal Chronicle' (IC) and the 'narrative sentences' . He defined the latter as follows: The class of descriptions I am concerned with refer to two distinct timeseparated events, E-1 and E-2. They describe the earliest of the events referred to. (…) 'The Thirty Years War began in 1618' refers to the beginning and to the end of the war, but it is about the beginning of the war.25 The momentous insight here is that we can formulate true statements about a time T in the past (and everything that happened before it) that could not be formulated at T -and as we saw at the end of the previous section the formulation of those kinds of statements is the very raison d'être of all writing of history. Consequently, the problem of other minds is of relevance for the historian only if it surfaces within the larger framework of that of other periods -and that may be often or never. But apart from this so very fruitful notion of the 'narrative sentence' Danto was less revolutionary when it came to narrative as such and narrative 23 'Every step and every movement of the multitude, even in what are termed enlightened ages, are made with equal blindness to the future; and nations stumble upon establishments, which are indeed the result of human action, but not the execution of any human design. (…) It may be affirmed of communities, that they admit of the greatest revolutions where no change is intended, and that the most refined politicians do not always know whither they are leading the state by their projects' . explanation (in fact, he modelled the latter on the CLM).26 It was left to Louis Mink to fill this lacuna -and he did so brilliantly. In the first chapter of his book Roth argues that three theses can be discerned in Mink's writings: 1) the non-standardization thesis, 2) the non-detachability thesis and 3) the nonaggregativity thesis. According to the first thesis there always are many ways of describing a historical event and there is no such thing as a standard or complete description -as ordinarily is the case in the sciences (cf. Kuhn's notion of normal science). The second thesis claims that in historical narratives the conclusions reached can never be detached from the narrative itself. Commenting on typical historical terms such as novelty, development and growth Mink observed that '"singular explanatory statements" are thin instruments indeed for dealing with such phenomena' .27 The conclusion of a narrative can never be reduced to a singular explanatory statement, whether that statement is part of the narrative or not. Roth adds three theses of his own to those of Mink -two of a methodological character -those of the logical formlessness of historical narrative and that of its evaluative intractability -and a third, metaphysical one on whether anything in objective reality corresponds to the language on events. Quod non, as we shall see. According to the first claim the nature of narrative historical explanation is not immediately clear from its logical form and according to the second this must be a handicap when it comes to their evaluation. Roth demonstrates both methodological worries to be ill-founded, since both presuppose 'that precisely one logical form is appropriate to explicating the notion of explication' .29 This assumption is rejected by Roth when discussing narrative explanation (see section 6). So let's turn to the third metaphysical claim.
In order to argue his case that statements about events (such as: 'visits home, heartbeats, a first kiss, the jump of an electron from one orbital position to another) do not correspond to an objective reality, Roth requires us to return to Danto's notion of the Ideal Chronicle containing a complete description of all events up till a certain time T. One problem immediately presenting itself is 'that events may be sliced thick and thin: a glance may be identified as an isolated event or as an instance in an event' . And then he puts his cards on the table: 'what the unit-event is, depends on the telling of it. Given the instructions to record "everything that happens, as it happens", the problem is not that there is too much for an Ideal Chronicler to record: the irony is that there are no things in the abstract to be recorded. An Ideal Chronicler never gets started because there are no ideal events to chronicle' (Roth's italics).30 He has the following argument for this provocative claim. The specification of an event's identity conditions will never overcome the problem of its underdetermination by the available evidence. Put differently: requiring 'ideal events' (i.e. candidates for the Ideal Chronicle) to be events (or objects) independent of our events (or objects) positing a scheme of things -is basically at odds with what we know about the relation between evidence and theory. Hence, even if the set of identity specification of an event is as elaborate as one might wish to have it, even then the realist inference would be unjustified. And Roth concludes: 'events simpliciter cannot be shown to exist; they are not known to be of nature's making rather than our own. Events exist only by proxy. This is why one cannot presume that there are any ideal events for our 29 Roth, op. cit., 24. Though Roth declares on other occasions that the writing of history 'is subject to the same conditions and caveats that apply to any theory of empirical knowledge' (Roth, op. cit., 48). 30 Roth, op. cit., 29.
Journal of the Philosophy of History (2021) 1-25 erstwhile chronicler to chronicle: knowledge of events is restricted to happenings isolated under descriptions provided by interested parties' .31 So we must deny to events a metaphysical status and reject the idea that any 'real' thing corresponds to them, however precise the specifications of their identity may be.32

Indeterminacies and 'Changing the Past'
In the previous section we encountered Arthur Danto's so important insight that what happened after an event E taking place at time T will enable us to formulate true statements about E whose truth was necessarily unknown to people living at T. And this is an ongoing process: in 1702 we will already know more about 1700 than in 1700, in 1800 more than in 1702, and so on. times be discovered in time T by the historian -'kinds' that were still unknown to the people living at T -and that may be appealed to in order to explain the past in hitherto unknown ways. One may need not only think here of typical historical notions like 'the bourgeois(ie)' or 'the Industrial Revolution' , for Hacking insists that notions like homosexuality, multiple personality disorder or suicide also have a history of their own that is, further, retroactively projected onto persons in the past that are now believed to have been homosexual or to have suffered from a multiple personality. In this way we can be said to continuously change the face of the past by redescribing, rethinking or refeeling it. This is why both Van den Akker and Roth characterize the relationship between the past and what is said about it by the historian as 'indeterminate' . Though it must be added that this 'indeterminacy' is far from arbitrary: in fact, each historical narrative can be read as explaining why it must be as it is. According to Roth the 'indeterminacy' thesis he derived from the writings by Hacking, Danto, Mink and others raises the issue of realism versus antirealism. He relates realism to the belief that there should be (or have been) an immutable and determinate past whereas the anti-realist rejects that belief.35 It is not immediately clear why a reality that is not 'determinate' could not count as reality -think of mental states or even physical objects like cloudsand much will depend on how the notion of 'determinacy' is unpacked; but let's grant this point to Roth. Next, realism is defined (with Crispin Wright) as: 'the assumption that the passage of time should have no part in determining our conception of what states of affairs may coherently be believed as possible' .36 What is true is timelessly true. Anti-realism (as defended by Michael Dummett) takes seriously the fact that we are immersed in time and that we can only describe the world as it is now. Here again, one may have one's reservations since these definitions leave room for considerable overlap between the realist and the anti-realist.37 Anyway, for Roth realism and antirealism are the sole metaphysical options, (idealism is not mentioned by him) and he is convinced that a lot depends on the choice we make here. He mentions a few examples to illustrate his point, such as the issue of verification by perception and implication -though without being very explicit about how these fare under the regime of either realism or anti-realism. 35 It should be noted that accepting Danto's narrative sentences are compatible with realism if the sentences connected by them are believed to be seen as being about an immutable and determinate past. 36 Roth, op cit.; 47. 37 Much will depend on whether one will be prepared (or not) to recognize as statements about the real world correct inferences of what is given to us in the here and now.
But with one exception, namely the 'problem of the reality of the past' . This issue came up already at the end of the previous section and it was then concluded that we must deny events a metaphysical status. Or, as Roth provocatively puts it: 'in my revised view nothing answers to "the Past"' .38 And even more provocatively, a moment before he had said that his account of 'the reality of the past, I argue, proves to be no more (or less) problematic than our account of any other aspect of reality, and so historical claims ought to be treated as subject to the same conditions and caveats that apply to any theory of empirical knowledge' .39 The writing of history is an empirical discipline like any other. Now that both realism and anti-realism have been shown to fail since they require the framework-independent conception of reality rejected in the previous section, Roth turns to Nelson Goodman's 'irrealism' . Apparently Roth expects his readers to be familiar with the term since he doesn't clarify it. Nevertheless, he might be too optimistic here and since the term is an unhappy one that can badly mislead, it may be helpful to say just one or two words about it. Goodman rejects the idea that there exists only one 'version' of the world to which all others can, or should be reduced -he thus asks the rhetorical question 'how do you go about reducing Constable's or James Joyce's world-view to physics?' .40 Nor can any of these 'versions' be 'tested by correspondence with a world independent of all versions' .41 Irrealism recognizes that we live in a world of 'multiple realities' (Cassirer) and that it would be preposterous to require that they should or could all be taken up into some 'reality of a higher order' comprising them all. So 'irrealism certainly does not hold that everything or even anything is irreal, but sees the world melting into versions and versions into making worlds, finds ontology evanescent, and inquires into what makes a version right and a world well-built' .42 At the same time, though truth cannot be defined by agreement with the world, a 'version' of the world can be 'taken to be true when it offends no unyielding beliefs and none of its own precepts' .43 These quotes from Goodman may suffice to explain why irrealism is the alternative to realism and anti-realism Roth is looking for: his indeterminacy thesis can be deftly summarized with the claim that we know the past only in terms of 'irrealist versions of the past' . Roth turns to Leon Goldstein's constructivism for answering the question regarding what theory of historical knowledge respects the irrealist version of the past. This should not surprise. It is essential to Goldstein's constructivism that history should be regarded as 'an artefact of disciplined disciplinary imagination' .44 Goldstein's well-known argument is that the past is no longer there and the only thing we have is the evidence left to us by the past in the here and now. The implication is that the past is an artefact in the sense that it is a construction -and emphatically not a re-construction -by the historian guided by the methods applied in historical practice for how to make sense of the evidence. 'What counts as evidence, and for what it counts, turns out to be a product of the practice of inquiry as informed by the use of predicates (past and present)' .45 How we conceive of the past is, therefore, determined by the historian's perspective and not by that of any observer in the past.46 The affinity Goldstein's constructivism has with Danto's argument about narratives sentences and with Roth's own irrealism will need no further clarification.
Roth and Van den Akker both endorse the thesis of the 'indeterminacy of the past'; they both warmly agree with Mink's observation that 'the past is not a determinate realm of unchanging reality' . But from there they travel different paths. Whereas Roth explores the irrealism vs (anti-)realism issue, Van den Akker moves on to that of the logical structure of historical narrative. In fact, this parting of ways is typical of the difference in orientation of the two books. Roth is interested less in developing new theories of historical writing than in a philosophical fine-tuning of those we already have. Van den Akker's book does both: it revolutionizes narrativism and achieves as much by making use of the most sophisticated and up-to-date philosophical instruments. To mention a telling example, Roth pays no attention to the following passage in one of Mink's essays, while it is the point of departure for one of Van den Akker's most revealing dissections of historical narrative. After having stated that there is 'no complex referent for all our narratives of "what actually happened", the untold story to which narrative histories approximate' , Mink goes on to say: fact, of the sort to which so much historical research is dedicated, remain unaffected. But it does mean that the significance of the past is determined only by virtue or our own disciplined imagination. In so far as the significance of past occurrences is understandable only as they are locatable in the ensemble of interrelationship that can be grasped only in the construction of narrative form, it is we who make the past determinate in that respect (my italics).47 Van den Akker infers two absolutely crucial conclusions from this quote. To begin with, the insight that the indeterminate past thus is determined by the narrative that is written about it ex post facto. Van den Akker even goes as far as to agree with David Weberman's assertion that narratives have 'ontological force' .48 But that is not yet all. Referring to the quote's first sentence he reminds his readers of the distinction between historical research and historical writing; hence, between the establishment of historical fact on the basis of the evidence the past left us and the historian's narrative making use of the results of historical research. It has been pointed out ad nauseam that putting things this way is too simplistic and that we get a more realistic picture when we exchange it for a continuous interplay between research and writing, and one in which the two are, so to speak, in a permanent conversation with each other. True, but of more interest is the following. Without rejecting the distinction, Van den Akker warns of two mistakes it may occasion. In the first place, it is suggestive of a determinate past providing the historian with the content which is subsequently given form in and by his narrative -thus resuscitating again the form/content fallacy. Next, historical writing is more than merely presenting the results of historical research: it is a way of thinking, a cognitive instrument. And in contrast to Roth Van den Akker agrees with Mink that we wilfully blind ourselves to what is of interest in historical writing when refusing to recognize where it differs from the sciences.
Recognizing that the writing of history is a cognitive instrument of a quite special kind will also enable us to see what is right and wrong in the distinction between historical research and historical writing. The fact that the statements about the past resulting from historical research are (often) integrated in a historical narrative suggests a continuity between historical research and historical writing. This is often seen as proof that no clear-cut distinction can be made between the two. Nevertheless, Van  narrative may mention the same event, even its description may be graphically or sonically identical, containing the same marks in ink or pixels or producing the same noises in the same order; the truth-conditions, hence the meaning of these statements will differ, for as part of the truthconditions of statements is a narrative there belongs a reference to the historical theses of that narrative, and such reference obviously is lacking in a chronicle consisting of events descriptions' .49 Van den Akker's claim is not hard to explain. Let the results of historical research be expressed by the statements p 1 , p 2 , p 3 , … p n . As I have explained elsewhere the corresponding narrative then does not have the form p 1 , p 2 , p 3 , … p n : a mere conjunction of statements is no narrative. A narrative only comes into being if p 1 , p 2 , p 3 , … p n are recursively read as a definition of narrative N a ; hence, as {(N a is p 1 ), (N a is p 2 ), (N a is p 3 ), … (N a is p n )}.50 Now, it will be clear that the meaning of the statements p 1 , p 2 , p 3 , … p n differs from that of the statements (N a is p 1 ), (N a is p 2 ), (N a is p 3 ), … (N a is p n ): the former are about the past whereas the latter are about a narrative. Van den Akker is, therefore, right when stating that there is an unbridgeable gap between the statement as the result of historical research and exactly that same statement if being part of a historical narrative. We remain blind to this since the narrative requires us to read it as {(N a is p 1 ), (N a is p 2 ), (N a is p 3 ), … (N a is p n )}, though it persistently omits the 'N a is …' with which each sentence begins since that phrase is clearly redundant: no reader of narrative N a will ever feel tempted to see its sentences as the properties of another narrative than N a and there is, therefore, no need to endlessly repeat this phrase 'N a is …' . The foregoing demonstrates that it is not enough to merely speak about 'narrative' in our analysis of it but absolutely imperative to grant it a logical status of its own, next to that of proper names, subject-and predicate-terms, sentences, theories, and so on. As long as one fails to do so all reflection on narrative inevitably remains a groping in the dark. It would be like trying to count without having numbers. In the argument above the relation between historical research and historical writing could only be clarified thanks to the logical entity indicated with the symbol N. The same is true of the relation between statements of fact and statements in a narrative position and between statements about historical entities and the narratives about these entities (a no less fertile source of confusion in historical writing itself as in philosophy of history). Moreover, this is where Van den Akker's book once again favourably 49 Van den Akker, op. cit., 67. 50 And this operation must be carried out on all of a narrative's sentences. As Van den Akker puts it: 'all sentences of a narrative determine the content of a narrative, though some sentences will be more important than others' . Van den Akker, op. cit., 66. compares to that of Roth. It would be ridiculous to accuse Roth of seeing no difference between a narrative and a mere conjunction of true statements. Nevertheless, he doesn't argue the issue. He speaks about a narrative's statements and narratives themselves without analyzing the relationship between them. It is true, Van den Akker doesn't use the symbolization proposed here either, but his argument about historical research and historical writing runs parallel to the shorter and simpler argument presented here by applying it.51 Lastly, as we have seen, both Van den Akker and Roth emphasize the indeterminacy of the past; on several occasions Roth even speaks with manifest satisfaction of the historian's capacity 'to change the past' .52 As long as this phrase 'changing the past' is understood as a 'retroactive alignment' (Van den Akker) or a 'retroactive redescription' of part of the past (Roth) -recall Danto's narrative sentences -no one will object to it. But using that phrase is not without its dangers. For it might make readers forget that we cannot change the facts about the past. We cannot make the hand of Balthasar Gerard's tremble in order to prevent him from shooting Prince William I of Orange on 10 July 1584. Nor can we make Hitler stay just a little longer in the Bürgerbräukeller in München on 8 November 1939 -with the result that he would then have been blown into smithereens by Johann Elser's bomb. We may describe and interpret these events in literally every conceivable way we like, we may categorize the objects in the world in Roth's 'kinds' in any way we prefer, but no amount of redescribing or rewriting the past will wipe this kind of historical fact from the historical record. In this sense we cannot change the past.53

Explanation (Roth) and Exemplification (Van den Akker)
Roth begins his account of how historians explain the past as follows: 'The view defended here will be that narrative explanations explain narrative sentences This claim is not a philosophical truth but a law of physics. It has nothing to do with the philosophical debate on verificationism, naturalism, holism, and so on. Questioning it in a meaningful way will compel us to discuss the merits of e.g. the second law of thermodynamics or of quantum physics. Any other approach can only result in pseudo-scientific humbug.
(i.e. an explanandum expressible as a narrative sentence). In particular, I show why only a narrative can explain some events formulated as narrative sentences' .54 Let me give an example of what Roth apparently has in mind: Consider in this regard the following remark by Raul Hilberg, 'the destruction of the German Jews was thus no accident. When in the early days of 1933 the first [German] civil servant wrote the first definition of 'non-Aryan' into a civil ordinance, the fate of the European Jews was sealed' . Somewhat more prosaically, Hilberg's statement may be worded as a narrative sentence: The Holocaust began in 1933.55 The example makes clear why Roth found narrative sentences of interest from the perspective of narrative explanation: the narrative sentence 'the Holocaust began in 1933' ties together an earlier event (the first definition of 'non-Aryan' in 1933) to a later one (the Holocaust which no one could know of in 1933) and suggests the/a link between the two. Roth's argument is new, ingenious and attractive and it undoubtedly deserves our closest attention. Yet, I have my reservations. Firstly, it's somewhat puzzling that narrative explanations should explain narrative sentences and not the past. My main worry is, however, that if, as Roth suggests, narrative explanations explain narrative sentences, they surely explain an only very minor part of the past in question and, moreover, that they do so in a way that leaves open many questions. Obviously, someone being told that 'the Holocaust began in 1933' will be clueless as to its exact meaning. The narrative sentence just mentioned is the briefest possible summary of Hilberg's book rather than the book's explanandum. Put differently, this narrative sentence is a brief formulation of the thesis defended in a book of almost 800 pages and that thesis is hopelessly lost without all the information presented to the reader in these 800 pages. This is why Hilberg was not content with anything less than these 800 pages. And, more generally, why historians are, unlike scientists, in the habit of writing long books. Hence, the real explanation of the Holocaust offered by Hilberg is the 800-pages-long narrative we will find in the book entitled The Destruction of the European Jews and not in the summary of its main thesis expressed by a lapidary narrative sentence.
The role Roth assigns to narrative sentences is all the more remarkable since it is hard to reconcile with Mink's writings on historical narrative, for which Roth has so much respect. We need only recall that Mink entitled one of his The notion of 'exemplification' is -indeed -Van den Akker's counterpart to Roth's theory of historical explanation. But before getting to it a brief methodological comment must be made. Van den Akker distinguishes between semantics (using terms such as reference, meaning, (historical) truth, narration, historical thesis, exemplification) and pragmatics (using terms like signifying, exemplifying and claiming to be true). Furthermore, Van den Akker insists that his own approach is pragmatist rather than semantical. He follows here Robert Brandom's inferentialism. I shall not enter deeper into this and mention it only in order to make clear that when Van den Akker argues that historical theses are neither true or false, he should not be accused of irrationalism; the idea simply is that for the pragmatist they can only be claimed to be true or false.57 Now, then, Van den Akker's theory of exemplification. He defines it as follows: 'This theory holds that the past as represented in some narrative exemplifies the historical thesis of that narrative. As such the theory explains how 56 Mink, op. cit., 185. 57 Van den Akker complicates the issue by stating that historical theses are not true or false semantically since in terms of semantics they can only be exemplified or fail to exemplify (Van den Akker, 73). So one wonders whether exemplification (of failing to exemplify) is the semantic counterpart of the pragmatist's 'claiming to be true or false' . But why is this so? One gets the impression that this distinction between semantics and pragmatics causes confusion rather than clarity. the past receives its meaning' .58 Note, that narrative and historical thesis do not express a meaning assumed to be present in the past; the 'movement' is, rather, the reverse: the past exemplifies a historical thesis and is in that sense subservient to it. As he had put it elsewhere: 'narratives do not reflect past reality: past reality reflects the narrative' .59 This grants to the historian's narrative a certain 'autonomy' with regard to the past itself -which is explained by Van den Akker as follows: 'this autonomy of the historical narrative is due to the fact that the historian can see in the behaviour and attitudes of individuals something that the individuals themselves and their contemporaries could not possibly see' .60 An example given by Danto may help to understand Van den Akker's intentions. In her book on the Thirty Years War C.V. Wedgwood observes that the battle-cry shouted by the soldiers of the Catholic League during the Battle of the White Mountain in 1620 was 'Sancta Maria' whereas that shouted in the first Battle of Nördlingen in 1634 was 'Viva España' . And she concludes that a primarily religious conflict had changed into a political conflict. Crucial is that Wedgwood may well be right about this whereas in all likelihood none of the parties involved in this stage of the war intended the change or was even aware of it. It is her thesis that this change took place; and the historical reality she refers to in her narrative 'exemplifies' , or 'illustrates' her thesis or her claim that this is what happened. And, obviously, the credibility of her thesis or claim increases when more data can be found in the past itself exemplifying it.
An interesting problem is raised by Van den Akker's exposition. In the previous paragraph it was argued that this change took place but not in the individuals living at the time -it was said they were probably not even aware of it. How to account for this all too apparent absurdity? How could something present in the past itself possibly remain wholly hidden to the people living then? Van den Akker's explanation runs as follows: what we should make clear is the distinction between referring to individuals to illustrate or exemplify social change and referring to individuals in order to provide evidence for social change. I propose to draw this distinction thus. If reference to the attitudes and behaviour of individuals is made to exemplify change, then a change in society is at stake that only became apparent in retrospect, while if reference is made to the attitudes and behaviour of individuals is made to provide evidence of social change, then the identification of an aspect of social reality is at stake, 58 Van den Akker, op. cit., 102. 59 Van den Akker, op cit., 73. 60 Van den Akker, op. cit., 112.
proving that society was changing in the eyes of witnesses of that change (Van den Akker's italics).61 In brief, what the historian finds in the archives may be used to exemplify and as evidence of social change. Thus one may say that Locke's Two Treatises on Government is evidence of Locke's liberal outlook but not of the history of liberalism. In the latter case we should say that Locke's book exemplified a certain phase in the history of liberalism. Two questions remain, though. Firstly, can the distinction also be applied to individuals (as the Freudians amongst us would probably argue)? Secondly, how does it relate to the semantic versus pragmatics distinction? Can exemplification give us only claims to truth, whereas using material the past has left us may give us truth? Next, Mink had argued that the conclusions (historical theses in Van den Akker's terminology) reached by the historian are ingredients in the historian's narrative argument itself and cannot be detached from it. It is most intimately interwoven with the whole of the text. Whereas Roth accepts Mink's nondetachability doctrine while rejecting it in practice, Van den Akker remains strictly faithful to it. Even more so, he even lifts his theory of exemplification, stating that historical theses are exemplified rather than justified by the historical evidence to the level of the historian's narrative in the sense that the latter can be said to exemplify the historical thesis it defends. And, indeed, the claim is almost a tautology: clearly, all that the historian writes in his narrative is meant to support the thesis the historian defends in it. All that fails to satisfy this condition is redundant and should be left out. Respecting this restriction on narrative length is absolutely indispensable precisely because each historical narrative has an innate tendency to grow as long as Danto's 'Ideal Chronicle' .

Conclusion
I hope this review-essay has made clear that the two books discussed here are of the highest quality. In fact, they contain many more valuable insights which I left out with a bleeding heart because I decided to focus on the most important only.62 Since the two books cover roughly the same ground comparing both is the obvious thing to do. 61 Van den Akker, op. cit., 108. 62 As for Van den Akker, I omitted from my exposition of his book its last chapter dealing with Danto's philosophy of art. Far harder was the decision to leave out from my account his chapter on representation. The chapter's main conclusion is that all representations The first thing to come to mind is that while the argument in both books is relentlessly technical (in Van den Akker's book even more so than in the one by Roth), technical precision is meant to serve different purposes in both. Roth uses it to argue claims not necessarily peculiar to historical writing (such as the metaphysical status of events or the irrealism of the past). Even more so, two of the seven chapters of his book are on Kuhn and of more interest for the philosopher of science than for the philosopher of history and the better half of one more is an exposition of Roth's methodological naturalism.63 Finally, he sometimes claims to have addressed some issue of importance for philosophy of history, whereas the reader will look in vain for where he actually does so.64 This is different with Van den Akker. All through the book the focus remains steadfastly on the issue of narrativism. It will not come as a surprise, therefore, that Van den Akker's book-reader will learn more about the narrativism of historical writing than Roth's book-reader.65 This is certainly not meant to belittle the great merits of Roth's book. For example, his argument about the metaphysical status of events is a lasting contribution of great value to philosophy of history. But this will already be less so with his advocacy of the irrealism of the past. The irrealist quietism that Roth borrowed from Goodman stand for what they represent. Next, some representations take the place of what they represent and function as their substitute. And, finally, some representations express something about what they represent. This is the case with artistic and historical representation. 'Expression' and exemplification are cognate notions: only if a expresses c about b, b exemplifies c (Van den Akker, op cit., 95). But since representation is not a truly indispensable link in Van den Akker's chain of argument I felt justified to leave it out of my account. 63 The other half discusses some of the absurdist claims of postmodern philosophy of history of some three decades ago. 64 For example, in the Conclusion Roth writes: 'as argued in other chapters (especially 4, 5 and 6), if an event can be explained only narratively, then evaluating that explanation will typically have to be done comparatively, i.e., relative to a competing narrative' (Roth, 143). I believe I read Roth's book carefully enough; nevertheless, I would be hard pressed to tell where he addressed this momentous issue in these three chapters, satisfactorily explained why this should be so, how this comparison is carried out in practice and, last but not least, what are the standards that will then be decisive. Nor is the lemma 'comparison' in the index. 65 I agree, therefore, with Zammito's observation about Roth's book: 'the ideal audience of this work, then, turns out to be philosophers -and, indeed, philosophers of scienceand the lapidary assertion, 'History is a science' (145), makes science the crucial problemterm in this work, not history' . J. Zammito, The Philosophical Structure of Historical Explanation, Notre Dame Philosophical Reviews, June 16, 2020. Apparently Zammito had hoped Roth to tell us more about history. It is, therefore, somewhat amazing that Zammito focuses in his review almost exclusively on Roth's two chapters on Kuhn that have little, if anything, to offer to the philosopher of history.