4. Constructivism

It has been recognized before that historiography poses its special epistemological problems, although these were usually seen as problems within rather than about an epistemological approach to historical knowing. Oakeshott, Collingwood, and Goldstein argued that the truth of statements about the past can never be verified conclusively, since the past no longer exists. Consequently we can never compare the actual past to the statements the historian has made about it. Collingwood tried to solve the problem by saying that the historian "reenacts" the past in his own mind and thus makes the past contemporary with himself so that he can make verifiable, true statements about it.^[13]

More relevant to our present purpose, however, is the way Oakeshott and Goldstein have attempted to deal with the epistemological problem.^[14] The idea is that the past itself can never be an ingredient in the process of acquiring historical knowledge or in historical discussion, since the past by its very nature can no longer be observed. The past no longer exists and thus cannot be a proper object of investigation. We have at our disposal only the traces the past has left us in the form of documents, inscriptions, paintings, buildings, and so forth. Consequently, all we have are constructions produced by historians on the basis of these traces (that is why the term constructivism is used for describing the position of Oakeshott and Goldstein). Even the word re constructivism would be out of place since it suggests a parallelism between the past itself and the historian's reconstruction of the past that can never be verified. So, constructivism empha-

[12] P. Gay, Style in History , London, 1975 (see Gay's introduction and conclusion).

[13] Van der Dussen has emphasized that Collingwood's "reenactment" theory was born from epistemological considerations. See W. J. Van der Dussen, History as a Science: The Philosophy of R. G. Collingwood , The Hague, 1981, 143ff.

[14] M. Oakeshott, Experience and Its Modes , London, 1978, chap. 3. L. J. Goldstein, Historical Knowing. London, 1976.

sizes, as did the narrativist position sketched in the previous section, the autonomy of historiography with regard to the past itself, and that is why constructivism demands our close attention.

In order to outline the constructivist position with more precision, Goldstein distinguishes between the infrastructure and the superstructure of historical writing. The superstructure is historical narrative itself, the linguistic structures we find in history books or in articles in historical journals. The infrastructure comprises the totality of methods and techniques employed by the historian in the course of his journey from his first acquaintance with the historical documents, et cetera, to the ultimate production of the superstructure (e.g., palaeology, numismatics, chronology, etc.).^[15] According to Goldstein, the superstructure of historiography has not altered noticeably since the days of Thucydides, whereas all progress in historiography was due to evolutions and new developments on the level of the infrastructure. Due to these developments, progress proved possible in historiography, and when by common consent a piece of historical writing is judged to be better (or worse) than another, this can always be explained by looking at their infrastructures. It is this infrastructure, and not the correspondence with a no-longer-existing historical reality, on which decisions regarding the acceptability of the historiographical constructions produced by historians are based.

Several objections have been leveled at constructivism. Oakeshott's constructivism was criticized by Meiland on the ground that he confused "knowledge that p " with "evidence for p. " Oakeshott rejected the possibility of historical knowledge, since he required of "knowing that p " that which is indeed true of "evidence for p ": namely, that its object is given there and then. According to the standard analysis of " A knows that p, "^[16] however, this statement implies: 1) "A believes that p "; 2) " p is true"; and 3) " A has evidence for p " and this means that there is a difference between "knowing that p " and "having evidence that p ."^[17] "Evidence that p " is always evidence for "knowing that p " and may therefore not be confused with it.

Most often, constructivism is attacked on the basis of the same arguments that can be used against verificationism. Verificationism is a theory concerning the meaning of statements: according to it, the meaning of the statement that p is equivalent to the meaning of those statements capable of verifying that p. It need not surprise us that the rejection of verificationism is the most obvious point of departure for a criticism of constructivism as defended by Oakeshott and Goldstein: both verificationism

[15] L. J. Goldstein, Historical Knowing , London, 1976, chap. 5.

[16] See, e.g., A. C. Danto, Analytical Philosophy of Knowledge , Cambridge (Eng.), 1968, 73.

[17] J. W. Meiland, Scepticism and Historical Knowledge , New York, 1965, 41-63.

and constructivism demonstrate a shift from the statement itself to the evidence we have for verifying the truth of the statement. A good example of the criticism of constructivism along these lines can be found in an article by P. H. Nowell-Smith.^[18] He accuses Goldstein of confusing the reference of a statement with its verification. The referent of a statement is the (historical) state of affairs the statement is about; the verification of the statement is the evidence we have for its truth. The difference between the two will need no amplification. If, however, reference and verification are identified, the result is the idea that historians never refer to the past itself but only to the evidence they have for verifying statements about the past. And that, in fact, is the position Goldstein wishes to defend.

However, a constructivism not of the infrastructure but of the superstructure is not subject to such criticism. In order to sustain this claim, let us first answer the question as to what such a brand of constructivism would look like. The superstructure is a linguistic construction consisting of many singular statements about the past. Each of these singular statements describes the past, so we might initially suppose that the historian's narrative is also a description of the past. This, however, is not satisfactory. Let us take two historical narratives on roughly the same topic (e.g., the French Revolution) and let us assume, furthermore, that both contain only true descriptions of this part of French history. Nevertheless, in such situations, it often happens that historians still prefer one narrative to the other. We have two options. First, we might maintain that such a preference is unfounded since both historical narratives are descriptively unexceptionable. This option is, however, in conflict with all we know about historiography and about historical discussion. According to the second option, the historian's narrative as a whole has a descriptive capacity of its own which we take into account when we are comparing two narratives (on, e.g., the French Revolution). But if we wish to put it this way, we must be able to make sense of the suggestion that there is some correspondence between narrative and the past: only if such a correspondence exists can we decide upon the descriptive merits of the two historical narratives. However, this idea of a correspondence between the two historical narratives and historical reality is a redundant one and does not clarify anything about how we decide upon the relative merits of the two narratives. For, the controversy between these two narratives on the French Revolution cannot be settled by simply establishing (in the way this can be done for singular statements) which one corresponds best with the past. There is not, in addition to the two historical narratives, a third thing—that is, an objective yardstick—to measure the correspondence between

[18] P. H. Nowell-Smith, "The Constructionist Theory of History," History and Theory, Beiheft 16 (1977): 1-28.

each of the two narratives and the past itself: narratives are all we have.^[19] The actual past may provide us with arguments for preferring one historical narrative to another, but in historiographical discussion it is never compared with narratives in toto in the way we can compare reality with singular statements in order to establish their truth or falsehood. Since the actual past is only an argument and is never conclusive in settling historiographical debate, the idea of a correspondence between a historical narrative and the actual past will get us nowhere if we want to understand the narrative writing of history. At most, we could say that each historical narrative is an attempt or proposal to define, in a specific case, the correspondence between language and historical reality. But by doing so we have defined correspondence in terms of historiographical adequacy instead of explaining the latter in terms of the former (and that would have been the only compelling argument for the introduction of the notion of a correspondence between [part of] the past and the historical narrative as a whole).

To conclude, whether we see historical narrative as a conjunction of statements or as a whole, in neither case can we meaningfully speak of a correspondence between historical reality and historical narrative. Constructivism, as a theory on the autonomy of narrative with regard to the past, is right in discouraging our belief in a correspondence between historical language and reality. The previous argument shows what is right and what is wrong in constructivism. Constructivism, as it was defined by Oakeshott and by Goldstein, is a theory concerning the statements of the historian's narrative. However, in order to avoid objections like those of Meiland and of Nowell-Smith, constructivism should be interpreted as a theory on the historian's narrative as a whole. Goldstein's superstructure, narrative, is a linguistic construction built of many individual singular statements. Better than any other term could possibly do, the term constructivism reflects the fact that it is the historian's task to build these linguistic constructions whose logical characteristics cannot be reduced to those of its constituent components.

This constructivist interpretation of historiography also gives us an answer to the question of how language is used by the historian: the historian uses language (i.e., individual singular statements) in order to construct a narrative. In an imprecise way, we might say that singular statements are used to express knowledge (about the past). But this is impre-

[19] In a similar vein, Mink writes:

the alternative ia to abandon the remnant of the idea of Universal History that survives as a presupposition, namely the idea that there is a determinate historical actuality, the complex referent for all our narratives of "what actually happened," the untold story to which narrative histories approximate. (L. O. Mink, "Narrative Form as a Cognitive Instrument," in R. H. Canary and H. Kozicki, eds., The Writing of History , Madison, 1978, 148.)

cise, since such statements, in fact, are knowledge (of the past). And nothing can be used to express what it is; what we wish to attain by means of x is different from x itself. At most, we could say that singular statements are used to express truth, for the true statement is not truth itself. Since we have accorded to historical narrative a status apart and different from that of the singular statement, we are permitted to say that according to the constructivism advocated here the historian uses language in the proper and true sense of the word.