Historiographic narratives and empirical evidence: a case study

Abstract

Several scholars observed that narratives about the human past are evaluated comparatively. Few attempts have been made, however, to explore how such evaluations are actually done. Here I look at a lengthy “contest” among several historiographic narratives, all constructed to make sense of another one—the biblical story of the conquest of Canaan. I conclude that the preference of such narratives can be construed as a rational choice. In particular, an easily comprehensible and emotionally evocative narrative will give way to a complex and mundane one, when the latter provides a more coherent account of the consensually accepted body of evidence. This points to a fundamental difference between historiographic narratives and fiction, contrary to some influential opinions in the philosophy of historiography. Such historiographic narratives have similarities with hypotheses and narrative explanations in natural science.

1 Introduction

“In history, the demand for an explanation is often interpreted in such a way that the proper answer assumes narrative form” (Dray 1954).¹ A narrative recounts events and circumstance in a (perhaps implicit) time-ordered manner, connecting these events and circumstances by “casual inputs” (ibid, p. 164),² so that earlier events are causally relevant to the later ones (Carroll 2001; Currie and Sterelny 2017). Narratives are also expected to have a beginning, middle, and end (Mink 1978, p. 186; Wise 2011, p. 370) and have a continuously existing (Hull 1975), or unified (Mink 1978) subject.³

When several substantially different historiographic⁴ narratives are offered to account for the same traces of the past, which of them (if any) is to be preferred and for what reasons?

Several writers (Kuukkanen 2015; Roth 2017) suggested that assessment and preference of narratives can only be done comparatively. But what is to be compared, and how is the evaluation to be carried out?

This question is, of course, closely related to a more fundamental matter: What are the essential qualities of a historiographic narrative, or using Hayden White (1980) language, its value? What should it deliver?

Narrativist philosophers of historiography have maintained that while historiographic narratives should refer to “real events” (White 1980, p. 22, 2009, p. 46) and contain true factual statements (White 1984, p. 22), their meaning is allegorical (White 1984) and metaphorical (Ankersmit (1983) and their value has to do with affective qualities such as their appeal to individual imagination (Mink 1978), moralizing power (White 1980), emotional cadence (Velleman 2003), or aesthetic virtues (Ankersmit 1989, 2012).

The question of adjudicating among alternative and conflicting narratives has not received much attention from these scholars. One may conclude from the above that when considering which narrative to prefer one should pay attention to their moral, aesthetic or emotional virtues rather than to epistemic ones. Often, however, they seem to imply that the choice is mostly idiosyncratic:

Just as ‘evidence’ does not dictate which story to construct, so it does not bear on the preference on one story or another” (Mink 1978, p. 145).

We are driven back to moral and aesthetic reasons for the choice of one vision over another as the more “realistic” … we are free to conceive history as we please, just as we are free to make it what we will. (White 1973, p. 433)

Other scholars (e.g., Carroll 2001; Kuukkanen 2012, 2015) maintained that historiographic narratives could be evaluated rationally according to agreed epistemic values. Few attempts were made, however, to see how such narratives are actually assessed and how they come to be accepted or rejected. Partially, at least, this is because treatises in the philosophy of historiography tend to discuss, and use as examples, individual texts written by historians. Whether one accepts that “as soon as one statement is omitted or added we have to do with a different Narrative Substance” (Ankersmit 1983, p. 213; cf. also Ankersmit 2012, p. 94) or allows, with Roth (1989, p. 454), that a few non-critical factual errors do not discredit a narrative, one must conclude (Kuukkanen, forthcoming) that such narratives are “empirically immune.” As for other, non-epistemic, modes of appraisal they can, as we saw, be a matter of idiosyncratic preference.

But a narrative can also be understood as a generalized representation of a series of events, processes, and states-of-affairs that has a distinctive unifying rationale, possibly presented by several narrators in various texts. This is the construal of the term “narrative” used by many historians. One reads, for example, that the narrative proposed by Ashton (1968) and others for the British Industrial Revolution is one of widespread technological advancement that enabled a concomitant increase of the population and the standard of living, while Barca’s (2011) narrative of the same period stresses negative aspects of ecological deterioration and increasing social inequalities.

Recently, such “generalized narratives” (a term coined in Glennan 2010) became a topic of interest for philosophers (though not among the narrativist philosophers of historiography). Generalized narratives were evaluated, for example, in relation to the ancient demography of Tuscany (Wise 2011), the battle of Antietam during the American Civil War (Glennan 2014), the rise of Rome to domination of the Mediterranean world (Berry 2014), and in the comparison of alternative accounts of evolutionary processes (Rosales 2017). In all of these works, the narratives in question are not a particular book or essay but “a candidate explanation of a particular causal trajectory in the past thought to be of interest in its own right” (Currie and Sterelny 2017).

Looking at generalized narratives opens the way to exploration of “narrative dynamics” and in particular of factors that can influence the preference of one narrative over another. That is because generalized narratives—unlike individual texts—can evolve, as people who hold and recount them modify secondary details while retaining the narrative’s general conceptual and causal rationale. When several different generalized narratives that seek to account for the same ensemble of traces are offered, it makes sense to speak of them as competing for acceptance by the relevant community.

My aim, then, is to explore what influences the preference of generalized narratives.

I shall look at a case in which four distinct narratives competed by postulating different events and causal processes in the same spatiotemporal context. The narratives in question describe the transition between the Late Bronze Age and the Iron Age in Palestine, and in particular “the emergence of Israel” there. All four have the same beginning (a system of Canaanite city-states under Egyptian rule), the same ending (territorial states; in particular, the monarchies of Judea and Israel) and a middle plot (they had, in this respect, the distinctive “narrative structure”). Otherwise, they differ considerably, narrating different activities of different protagonists under substantially dissimilar circumstances and processes.

In a nutshell, a powerful and evocative narrative—arguably, one of the best-known ever, that of the conquest of Canaan by the tribes of Israel—gradually receded until it was entirely abandoned. The currently widely-accepted Autochthonic narrative is more complicated, less comprehensive, and certainly less emotionally impressive and morally evocative than all the alternatives. Its only claim to merit is that the events, states-of-affairs, and processes it retrodicts agrees better with currently available evidence.

The role of case studies in philosophical investigations is a matter of ongoing debate (Cf., e.g., Sauer and Scholl (eds.) 2016). Even scholars who are reserved about their probative value, however, usually grant case studies a heuristic (Chang 2011), illustrative and debate-structuring (Currie 2015a), and insight-yielding (Schickore 2011; McAllister 2017) role. On a more positive side, Burian (2001) stated that “case studies can produce findings that cannot be gotten from more abstract ‘armchair’ philosophical work” and Kinzel (2016) stressed the importance of recalcitrant cases for forcing belief revision.

The particular case explored here, which covers the contest between four mutually incompatible narratives over most of the twentieth century, certainly has the “sufficient variation” required by Pietsch (2016) as a condition for generalizability.⁵ Because the debate touched (and still does) upon issues of strong emotional and ideological importance to all involved as well as to the general public, it also qualifies as a “hard” case, sensu Scholl and Räz (2016) and so can, arguably, “demonstrate the power of a principle, and … show that the same principle can plausibly handle a host of similar but less difficult cases” (ibid, p. 77).

I posit that the following result from the case study in this article is generalizable: For a significant class of historiographic narratives, acceptability is a matter of maintaining internal coherence while agreeing with the available, consensually accepted, body of evidence and not at all of their imaginative, moralizing or emotional power. Some general epistemic virtues, like simplicity, scope, and the ability to subsume the narrative under a more general theory, also carry little weight.

Several scholars (Kosso 2001, 2009; Currie and Streleny 2017) emphasized the crucial importance of coherence for the viability of narrative explanations and the constraining power that evidence can have on the coherence of narratives. The case discussed here enables me to look into the dynamics of a struggle between competing narratives: How narratives evolve in the face of changing empirical data and how their degree of success in maintaining coherence influences their “chances for survival” in the competition for acceptance by the relevant epistemic community.⁶

My conclusion is that coherence considerations can indeed determine the fate of historiographic narratives, provided that sufficient relevant evidence is available. The availability of such evidence depends upon the vicissitudes of nature and our technical and methodological capabilities, but the case discussed below indicates that it is neither uncommon nor limited to trivial and uninteresting issues.

I suggest that historiographic narratives that meet this criterion are not unlike hypotheses and narrative explanations in the historical natural sciences and can benefit from recent advances in the philosophy of these disciplines. They are, contrary to assertions by scholars from the narrativist school, fundamentally different from fictional narratives, even when the latter refer to the same subject matter.

Admittedly, the constraining power of coherence considerations is not always sufficient to decide between differing and conflicting narratives. When it is not, several such narratives can be maintained concurrently. These narratives will be underdetermined, and emotional, aesthetic or moral virtues cannot remove this underdetermination.

The next two sections present the case study: A single subject, four narratives, and the meta-narrative of how these narratives competed and fared throughout most of the twentieth century. The fourth section analyses the factors that can influence the acceptance or rejection of generalized historiographic narratives and the fifth examines implications of this case study. Finally, section six discusses the relationship between generalized narratives and particular historiographic texts that exemplify them and the difference between historiographic narratives and fiction and concludes with several general remarks.

2 Case study, part 1: a tale of two narratives

At least since the beginning of the modern era, various scholars have sought to interpret the biblical narratives in historical terms, locating events, places, and circumstances mentioned therein in a recognizable historical context and relating them to knowledge attained from other sources.

Through most of the twentieth century, two schools of thought offered different and conflicting narratives to account for an important part of the biblical texts: The books of Joshua and Judges, which narrate how the tribes of Israel came to be settled in the land of Canaan. Following a general convention, I shall refer to them as the Conquest narrative and the Peaceful Immigration (or simply the Immigration) narrative.

The guiding rationale of the Conquest narrative was that the biblical texts contain a generally correct—though not wholly accurate—account of the relevant period. Thus, an invasion by people with a common origin and unifying faith, followed by a coordinated series of military campaigns, led to the destruction of many Canaanite cities. The invading tribes then entered a lengthy settlement process, interrupted by occasional conflicts with remaining local populations.

Throughout the first six decades of the twentieth century, variations of the Conquest narrative were held by a near-totality of historians and archaeologists of the relevant periods, as well as by the general public. Among the most influential texts that articulated and promoted this narrative were books and articles by archaeologists like William Foxwell Albright (1939, 1940), Ernest Wright (1946), and Yigael Yadin (1965, 1982, 1984).

The Peaceful Immigration narrative conjectured a different scenario: Combining analysis of the biblical texts with analogies to present-days Middle-Eastern Nomads (Bedouins), it described a protracted process of seasonal transhumance by nomadic tribes between the eastern desert and sown Western Palestine, gradually leading to a sedentary settlement. The “heroes” of the immigration narrative, then, were pastoralists rather than warriors (though perhaps with the same charismatic leader, Alt 1936 [1953]). Only later did these tribes coalesce into coalitions that engaged in battles against the Canaanite cities. The proponents of this interpretation were biblical scholars, particularly German ones. The most extensive and influential texts encapsulating the Immigration narrative are those of Albrecht Alt (1925, 1929) and Martin Noth (1938, 1958).

Advocates of the two narratives argued a lot, sometimes respectfully (Albright 1956) and sometimes less so (Cf. Silberman 1993). As long as the debate revolved around the understanding of biblical passages, it resembled inconclusive exercises in hermeneutics. Both sides held that the texts about the relevant period were aggregated from several sources and put in writing only several centuries after the events described in them and thus cannot be expected to be an accurate account of these events. Therefore, scholars had a wide latitude to select textual passages that supported their view or interpret the text in a manner different from its literal content.

Archaeological explorations in Palestine were initiated around the beginning of the twentieth century.⁷ Over time, as empirical results from sites and their strata accumulated, the ability to infer the periods of occupancy of past human settlements and the material culture of their inhabitants from traces found therein advanced through a successive “bootstrapping and scaffolding” (Chapman and Wylie 2016) in which “successive stages of knowledge, each building on the preceding one, are created in order to enhance the achievement of certain epistemic goals” (Chang 2004, p. 45).

The relevant epistemic goals, in this case, included a desire—sometimes explicit (Moorey 1991, pp. 14–23) and often implied—to demonstrate the veracity of the Bible, with particular attention to the Conquest narrative. Initially, archaeological results appeared to be doing just that: Excavations exposed destruction layers (Albright 1934; Yadin et al. 1960) that were deemed to be in the right places (mounds assumed to be the sites of Canaanite cities) and at the right time (the latter part of the thirteenth century B.C.E.), thus substantiating elements of the Conquest narrative.

Other findings, however, were more problematic. For example, excavations at a mound universally believed to be the site of the biblical Ai showed that it was uninhabited during the Late Bronze Age and a thousand years before (Marquet-Krause 1949). After several attempts, it was determined that no sign of the walls of Late Bronze Age Jericho could be found (Kenyon 1954). The last Canaanite town of Lachish was associated with the period of Ramesses II (Tufnell 1958),⁸ making the terminus post quem for its destruction problematic for the conquest scenario, and so on.

It is important to note that though these findings presented grave challenges to the Conquest narrative, they were not disputed by scholars who supported that narrative. The methodology of the archaeological discipline was stable and fine-grained enough that publication of these results resulted in a “heterogeneous, uncoerced and large consensus” about the interpretation of findings and their chronological context, even as disagreement about the general narrative persisted. Such a consensus is the hallmark of shared knowledge (Tucker 2004, Ch. 1), and this indeed is how the disturbing findings were regarded by scholars and laymen alike.

However, acceptance of the results necessitated modifying the Conquest narrative: As Ai was not inhabited, it could not have been conquered. Jericho could not have been the impressive fortified town of Josh. 6, and so forth. To preserve consistency with accepted evidence, the narrations in Chapters 7–8 in the book of Joshua were now assumed (Albright 1934) to be an etiological story told by people of later generations to explain the origin of imposing architectural ruins. The descriptions of Jericho’s mighty walls must have been (Yadin 1965) an exaggeration by later writers, and whatever had existed there had supposedly been eroded or removed by subsequent activity. The belated destruction of Lachish was interpreted (Yeivin 1937) as a result of its remote location, notwithstanding Josh. 10, and so on.

The narrative could sustain these alterations because it did not assume a literal truth of the biblical texts and because other results still appeared to support it. But the necessary modifications strained its coherence: To preserve the narrative’s consistency with the new evidence, its unity was compromised as more and more relevant sites were considered to be exceptions to the narrative’s overall rationale.

A tipping point was reached when a series of wide-area archaeological surveys carried out during the late 1960s and 1970s exposed remains of hundreds of small, unfortified settlements with very simple material culture in mountainous and semi-arid regions of Palestine that were uninhabited during the Late Bronze age. By their very nature, these meager settlements agreed with the conjectured scenario of the Peaceful Immigration narrative. Supporters of the Conquest narrative could appeal to a single obscure passage (Josh. 17, pp. 14–18) in which Joshua commands certain Israelite tribes to settle “in the Mountain,” but this could hardly be a satisfactory account for what was now understood as evidence for the major demographic phenomenon of the Bronze-Iron Age transition in Palestine.

Other discrepancies with the biblical text surfaced during the 1970s and early 1980s. In 1984, Yadin opened a conference talk by noting that he was probably the only one of the speakers still adhering to the Conquest narrative (or as he called it, “the Albright school”). He was right. At that time, the Conquest narrative was discredited in the eyes of most archaeologists, historians and biblical scholars, including many who had strong motivation to support it because of their national-political ideologies and/or religious beliefs. An influential book, From Nomadism to Monarchy (Finkelstein and Na’aman 1994) written by a group of Israeli archaeologists and directed to public audiences epitomized the shift to the Immigration narrative in its title as well as in its content.

3 Case study, part 2: a tale of three narratives

As the Immigration narrative was gaining support, a new one was articulated to challenge it. The Revolt narrative rejected the scenarios of both the Conquest and the Immigration narratives: It submitted that the real “heroes” of the Bronze-Iron transition in Palestine (at least the bulk of them) did not penetrate the land of Canaan from outside, either by military conquest or by immigration. Rather, they were people from the lower strata of the Late Bronze Age Canaanite society who rebelled against its feudal and exploitative regime. Empowered by an ideological-religious zeal derived from the new Yahwistic faith (that was either formed indigenously or adapted from a small group of people who escaped the bonds of slavery in Egypt) this oppressed population united to attack and destroy the cities of their former overlords.

The Revolt narrative for the Bronze–Iron transition in Palestine was formulated by the American biblical scholar George Mendenhall (1962, 1974). It was further elaborated in the formidable book “Tribes of Yahweh” by Norman Gottwald (1979). Its rationale was grounded in sociological theories and original reading of relevant biblical texts, with some auxiliary hypotheses about the circumstances in Bronze-Age Palestine. It offered a novel and cogent narrative for “the emergence of Israel in Canaan.” There was one problem, though: It could not point to physical evidence in support of one of its central claims, namely the mutinous and insurgent character of the early Israelites,⁹ and its expositors actually doubted (Gottwald 1984) if such evidence could ever be found.

It is perhaps for this reason that the Revolt narrative did not gain much support in the disciplinary community or the public audience. But one crucial element of it did, as shown below.

The Autochthonic narrative Visitors to the archaeology section of the Israel Museum in Jerusalem are presented with the following piece of information: “While the biblical story of the Exodus relates that the Israelites came from Egypt, many archaeologists believe that they actually originated [here] in the land. They lived in semi-nomadic groups in the hill country and eventually began to build small, permanent settlements….”

This brief passage summarizes the current opinion of most archaeologists, historians, and biblical scholars. They may disagree over whether the background of the people who lived in the hill settlements was pastoral (Finkelstein 1988b, 1995) or agrarian (Dever 1998, 2003) and on a few other things. But they share the narrative cited above, namely, that the inhabitant of the host of small settlements in previously uninhabited regions of Palestine were Israelites or Proto-Israelites (meaning that their descendants formed the population of the later monarchies of Judea and Israel) and that their demographic sources were mostly indigenous.

Several things contributed to the relatively rapid decline of the Immigration narrative. First, it was pointed out¹⁰ that a regime of long-range nomadism and seasonal transhumance between the deserts and the sown land was not viable prior to the domestication of camels in the Southern Levant, something that happened several centuries after the relevant period. Thus, an essential element of the Immigration narrative had to be modified. Maybe the early Israelites were “donkey nomads” (a concept initially used (Albright 1968) to salvage another biblical narrative—that of the Patriarchs) and still conform to the transhumance model. But reflecting on this, even people who were supportive of the Immigration narrative began to doubt its adequacy (Finkelstein 1988a, p. 307).

Even more substantial was the failure to detect evidence of foreign origin or influence in the material culture of the hill settlements. The relationship between “pots and people” is, as every archaeologist knows, often tenuous. Still, the thousands of potsherds and vessels found at these sites were much closer to Canaanite ware than to any non-local material culture of the period, something that did not cohere well with the Immigration narrative.

Related to the issue of material culture was the all-important one of dating. The fact that surveys and excavations at the hill sites did not produce remains of wares of Aegean and Cypriot origin made it much more likely that these settlements appeared after the upheaval that quenched Mediterranean commerce at the end of Late Bronze Age. This placed the onset of that phenomenon at the twelfth century B.C.E., a date that fits well with a narrative that connects the emergence of the settlements in the hill regions to the decline and eventual withdrawal of Egyptian rule during the twentieth pharaonic dynasty, but not with a centuries-long process of nomadism, transhumance and gradual settlement as conjectured by the Immigration narrative.

Modification of the Immigration narrative could perhaps resolve these difficulties. One suggestion (Fritz 1987) posited a version that had the nomadic Israelite tribes absorbing the material culture of their Canaanite neighbors through a long interim period of co-inhabitance. The lack of imported items in their settlements was hypothesized (Faust 2006)¹¹ to be a result of norms and ideology of communities that rejected the abhorred Canaanite practices. None of these suggestions (actually mutually inconsistent in what they imply for relationships between the early Israelites and the indigenous population), however, follow naturally from the Immigration narrative. Their ad-hoc nature diminished the narrative’s overall unity.

By the late 1990s, most scholars came to support variations of a narrative that avoided the difficulties mentioned above by conjecturing a different scenario, one that combines elements of the Revolt narrative (an indigenous demographic origin) and the Immigration narrative (a gradual and mostly peaceful process of settlement and coalescence). The Autochthonic narrative, in several shades, is currently the consensus opinion. It is the one recounted by archaeologists, biblical scholars, and historians (from diverse origins, affiliations, and ideological affinities) and is making inroads into the perception of the general public.

4 On the rise and fall of narratives

The previous two sections describe how several narratives competed to account for the same ensemble of textual and archaeological traces of the human past. In this competition, there were winners and losers: A once-dominant narrative was abandoned, another enjoyed temporary popularity but later went out of favor, and the current consensus endorses a narrative that was not even imagined (and would perhaps have been considered utterly improbable) until late in the twentieth century.

I am going to show that these profound changes of opinion are attributable to the influence of evidence that was discovered and published during the same period. Recognizing new evidence that conflicted with a particular narrative did not, however, bring about some kind of instant Popperian refutation. Rather, each narrative could be (and was) adapted to the new data by modifying some details while preserving its basic tenets.

Such modifications, however, often came at a cost. Each new piece of evidence, once it was accepted as valid and relevant, constrained the narratives: Depending on the case, some events and states-of-affairs formerly conjectured as part of a narrative appeared well-supported and more secure, some had to be omitted, while others that were not conceived formerly had to be incorporated.

For example, when a mound, generally believed to be the location of biblical ‘Ai, was found to have been uninhabited during the Late Bronze Age, the Conquest narrative had to be modified so that it did not include events described in Josh. 7–8, while surmising some other events (such as an etiological transfer of another conquest to these prominent ruins) to explain the discrepancy. When it was discovered that more than a century separated the destructions of Hazor and Lachish, this constrained the Conquest narrative to abandon at least one of these events as part of the conquest scenario, thus distancing itself from the biblical story that mentions both in short succession. When a multitude of simple settlements dated to the relevant period was exposed in formerly desolated areas, an important element of the Immigration narrative became part of the consensually accepted past scenario, thus buttressing this narrative. However, when these locations failed to produce evidence of a foreign origin of their inhabitants, a lengthy period of early Israelite co-habitation amid the Canaanite had to be assumed to make this absence plausible in the context of the Immigration narrative.

Such modifications are part and parcel of every generalized narrative. As long as they can be coherently incorporated, the narrative is likely to be considered sustainable. As Wise (2011) observed, “interpretation and judgment are always in play, but in general, conviction comes with narrative coherence and empirical adequacy.” But what is it that makes a narrative more or less coherent?

Coherence is frequently mentioned as an essential attribute of a narrative, (Hull 1975; Mink 1978; White 1980; Carroll 2001; Wise 2011; Kuukkanen 2015; Morgan 2017; Currie and Sterelny 2017) but the notion is often left undefined. Obviously, coherence involves consistency. But a set of beliefs can be both logically consistent and incoherent (BonJour 1985, p. 95), for example when its constitutive parts are mutually irrelevant. The additional ingredient necessary for coherence is not a mere relevance, however, but some sort of connectedness:

On my account of the narrative connection, earlier events raise certain possibilities, rather than others. … If ensuing events do not fall into that range and are not recuperated by the addition of causal inputs, the narrative will appear to be incoherent and unintelligible (Carroll 2001, p. 37).

A system of claims coheres to the degree that there is explanatory relevance among various claims. One part of the description should explain why another part is as it is, whether in terms of motives of persons in the past, or of the physical causes at work. (Kosso 2001, pp. 75–76)

The coherence condition requires that one try to maximize the ‘fit.’ That is, to show that there are inferential connections between the elements of data or even that they constitute a unified whole (Kuukkanen 2015, Ch. 7).

I shall use the term unity to describe this combination of relevance and connectedness, or “fit.” Since we are concerned with the dynamics of narratives, we need to consider what can enhance or diminish a narrative’s unity, rather than its absolute value (which is anyway difficult to assess). Appropriate criteria were formulated by BonJour (1985, pp. 98–99)¹²: the coherence of any system of beliefs increases with the number and strength of inferential connections between its elements, and it is decreased when it is divided into unconnected subsets and also when it contains unexplained anomalies in its believed content. Accordingly, the unity and coherence of a historiographic narrative will be jeopardized when it fails to account for evidence that is deemed pertinent to its core scenario (accepting them as “believed anomalies”) or when adding hypotheses that are unconnected to its rationale becomes necessary in order to accommodate such evidence.

Thus, the unity of the Conquest narrative was strained as it had to admit more and more localities as anomalies in order to preserve consistency with the collected evidence, and it became irreparably damaged when it transpired that the primary demographic trend in Palestine of the Late Bronze and Early Iron Age was external to its core “story.” Similarly, when the traces for that demographic trend failed to manifest any evidence for outside sources of their inhabitants or a sufficiently prolonged process, only badly disunified—and therefore, incoherent—versions of the Immigration narrative could be offered.

The dynamics of narratives’ competition, evolution, and replacement was driven by a succession of (in this case, archaeological) evidence, of which Chapman and Wylie (2016) say:

Neither these data nor the evidential claims based on them constitute a self-warranting empirical foundation, and yet they can powerfully challenge and constrain the reconstructive and explanatory claims we project onto the cultural past. …. Material evidence is inescapably an interpretive construct; what it ‘says’ is contingent on the provisional scaffolding we bring to bear. And yet it has a striking capacity to function as a ‘network of resistances to theoretical appropriation’ that routinely destabilizes settled assumptions, redirects inquiry and expands interpretive horizons in directions no one had anticipated.

Contravening evidence cannot disprove a historiographic narrative, but it can necessitate its modification. Such modifications, however, should be reasonably coherent to be sustainable, and as more and more relevant evidence accumulate this can become “an increasingly important, increasingly demanding constraint” (Currie and Sterelny 2017, p. 19).

Our case study shows that an incoherent narrative, even a strongly entrenched one, becomes unconvincing and may be totally discredited, in the sense that even those that were inclined to prefer it change their mind. If another available narrative fares better in this respect it is likely to be adopted, as was the Immigration narrative during the second half of the twentieth century; but can it also happen that a new narrative is created, as was the autochthonic narrative later.

5 Discussion

The case described above also demonstrates what will not determine a historiographic narrative’s fate: Literary attributes—be they its moralizing impulse (White 1980), its potential for allegoresis (White 1984), its “irreducible imaginative preferences or choices” (Mink 1978), its aesthetic qualities (Ankersmit 1989, 2012), or “its power to initiate and resolve an emotional cadence” (Velleman 2003)—may have made this or another narrative attractive to some individuals (especially when it corresponded to their prior epistemic and ideological convictions). But these qualities were not decisive for the narrative’s eventual acceptance when confronted with a stream of evidence that was commonly deemed genuine and pertinent.

Arguably, a much stronger “emotional cadence” is to be found in the story of swift military conquest or of lowly social strata rebelling against oppressive overlords, than in a story of a gradual and haphazard sedentarization of heterogeneous groups amalgamating over generations to form a new identity. And the more emotionally loaded narratives were also more conducive to moralizing: Nothing in the currently accepted Autochthonic narrative compares to the moral message implied in:

Real spiritual progress can only be achieved through catastrophe and suffering, reaching new levels after the profound catharsis… (Albright 1940, p. 310).

or to:

It was the common loyalty to a single Overlord and obligation to a common and a simple group of norms … solidarity of loyalty which was attractive to all persons suffering under the burden of subjection… [who] received deliverance from their bondage… (Mendenhall 1962, p. 74)

However, when the underlining scenarios of these evocative narratives were rendered incoherent (in the first case) or could not be supported (in the second) by the available evidence, they could not be maintained. Their moral message was simply inconsequential to in this respect.

Kuukkanen (2015, Ch. 7) listed five values according to which “one can judge and choose appropriate colligatory¹³ concepts”: Exemplification (the descriptive content of a narrative should exemplify the relevant historical data it subsumes), coherence, comprehensiveness (applicability to “larger amount of historical data on the assumed historical phenomenon”), scope (applicability to other historical phenomena), and originality.

As Kuukkanen notes elsewhere (2012, p. 360), this analysis blurs the distinction between historiographic narratives and scientific hypotheses, since “the same kind of epistemic values are found in historiography as in a number of scientific fields.” Similar epistemic values (or “virtues”) for evaluation and adjudication of scientific theories were offered by philosophers of science: Hempel (1965, p. 117) highlighted clarity and precision, explanatory power, simplicity, and evidential confirmation. Kuhn (1970) cited problem-solving capacity, accuracy, simplicity, self-consistency, and compatibility with other theories. Quine and Ullian (1978) mentioned conservativism, modesty, simplicity, generality, refutability, and precision, and Lipton (2001, p. 95) named scope, precision, mechanism, unification, and simplicity as desiderata of a “lovely” (and conducive to “the best”) explanation.

Considering the historiographic narratives in our case study, however, one notes that some of these criteria for evaluation and selection played only a minor part, or none at all, in their eventual acceptance or rejection. As time progressed and evidence accumulated, the competing narratives tended to be successively less streamlined, more complex, less “elegant” and in this sense, less comprehensible. They also became less comprehensive: The Conquest narrative offered an account not only of the appearance of the territorial monarchies of Judea and Israel in Iron-Age Palestine but also for the disappearance of the city-states order in much of Bronze-Age Canaan. The Immigration and Revolt narratives had no comprehensive story to tell about these destruction events: They merely assumed that they happened in one way or another. The currently dominant Autochthonic narrative shares this trait and, while combining elements from the Immigration (gradual settlement) and the Revolt (indigenous demographic origin) narratives, is silent about the emergence of the Yahwistic faith that was a significant element of all the previous narratives.

Values like “scope,” “generality” or “unification” also carried little weight: The Conquest narrative implied an analogy to peoples’ migrations, the Immigration narrative to transhumance by nomads, and the Revolt narrative to social revolutions. Such analogies, however, tended to persuade only those already convinced and did not contribute much to the change of opinions described above. If the analysis above is correct, coherence considerations, driven by new evidence, played predominant role first in inducing modifications in the narratives and later in judging and choosing among them.

Narratives as hypotheses This tendency to progress from simple to more complex and detailed structure as relevant data accumulate is a common feature of scientific hypotheses in the historical sciences (and probably elsewhere too). Tucker (2004, p. 148) points out that “If a vague but large scope theory attempts to increase its accuracy, it will have to add ad hoc hypotheses to account for the complexities of the evidence and become more complex, or narrow its scope to become more accurate.” Currie (2014, p. 1173) uses narrative explanations for past natural phenomena (“snowball earth” and sauropod gigantism) to demonstrate that in the historical sciences “Some of the time at least, explanations begin with a collection of competing simple narratives, some of which are rejected on empirical grounds, the remainder of which are synthesized into a complex narrative.”¹⁴

The currently consensual Autochthonic narrative demonstrates this process: It synthesizes elements of the Migration and Revolt narratives that preceded it. It is also both more complex (lacking a unifying “driving force” such as transhumance or a social revolution) and narrower in scope (leaving unexplained some traces like the destruction of Bronze-Age cities) than both of them.

Currie (2015b, 2018) and Currie and Sterelny (2017) note also that the requirement of coherence drives research questions, and that some matters acquire evidentiary significance only after a narrative referring to them was formulated. This too is demonstrated in our case study. Evidence for the material culture of the hill settlements, and in particular its foreign or local character, became of paramount importance only after the appearance of these sites in formerly unpopulated regions was understood to be the pivotal demographic process in Palestine of the Bronze-Iron transition and thus crucial to any “emergence of Israel” narrative.

This similarity between historiographic narratives and scientific hypotheses is something that narrativist philosophers of historiography have always objected to:

Nor can we say that narrative form is like a hypothesis in science, which is the product of individual imagination but once suggested can lead to research that can confirm or disconfirm it. The crucial difference is that the narrative combination of relations is simply not subject to confirmation or disconfirmation (Mink 1978, p. 145).

This is why the narrative history can legitimately be regarded as something other than a scientific account of the events of which it speaks. (White 1984, p. 20).

And more recently:

Because of the relation between the historiographical view and the language used by the historian in order to express this view- a relation which nowhere intersects the domain of the past - historiography possesses the same opacity and intensional dimension as art. Art and historiography can therefore be contrasted with science. (Ankersmit 1989, p. 145)

The examples above show that, at least in some cases, the opposite is true: Historiographic narratives are similar to scientific hypotheses in the way they are assessed and adjudicated (as for their relation to art and fiction, see the next section). In other words,

we argue this is just as true for science as it is for history. In this regard, practitioners of human history are methodologically continuous with archaeologists, geologists, cosmologists and palaeontologists [sic] (Currie and Sterelny 2017, p. 18).

I do not contend that the constraining power of coherence requirements is always sufficient to determine the preference between historiographic narratives. In many cases, it will probably not be, for two reasons: First, due to the steady decay of traces left by past events and circumstances there is no assurance that the necessary information will ever be available.¹⁵ Second, a reasonably stable and consensually shared theoretical and methodological basis by which to interpret the available data may be lacking.

These limitations are relevant for human as well as for natural historiographic studies. Unique to the study of human history, however, are obstacles to gaining epistemic access (even with the help of material or textual remains) to the cognitive and intentional stances of past human beings and the reasons for their behavior. Since these factors are essential for understanding many aspects of human history, in particular for answering many “why” (rather than “when,” “what” and “how”) questions, the resemblance between human historiography and historical natural sciences will always be partial and limited.

For example, Kuukkannen (2015, Ch. 9) discusses the unfinished debate about what caused the First World War.

For Ritter, German history in the twentieth century is a classic case of tragedy, and he saw fit to speak about ‘disaster’ and ‘blindness.’ Fischer, by contrast, detected only ‘intent’ and ‘premeditation’ behind the course of events. … And the debate continues.

Ritter and Fischer relied on the same corpus of evidence, and both interpreted it coherently—but differently. This is possible because stable and consensually accepted theories that infer people’s thought, intentions and beliefs from their actions and expressions (unlike those that infer, e.g., periods of occupancy from material remains) simply do not exist. Similarly, while we may have now a more stable narrative about what happened or did not happen during the transition period between the Bronze and the Iron Ages in Palestine, I suspect that issues like the roots of the Yahwistic (or any other) belief¹⁶ will always be only hermeneutically accessible.

When relevant knowledge is lacking, differing and conflicting¹⁷ narratives can be maintained concurrently, but they will all be underdetermined. Emotional or moral elicitation of a narrative cannot overcome this underdetermination.

The case presented here shows, however, that historiographic narratives that are amenable to rational evaluation and adjudication by the combination of empirical evidence and coherence constraints are by no means confined to “trite” (Jenkins 2008) or to “trivial and uninteresting” (Kuukkanen 2015) issues.

6 Conclusion

The previous pages described how a powerful, generally recognized and influential narrative was abandoned, giving way to other narratives that were successively more complex, less comprehensive and less “heroic.” In many respects, this trajectory resembles what often happens to narrative explanations in the historical natural sciences.

Two main conclusions, I posit, can be gleaned from this case study:

• (a) Even though the (generalized)¹⁸ historiographic narrative is underdetermined by evidence, it is constrained by it. When the former changes, either by new discoveries or as the result of a novel analysis, the latter is constrained to changes by adding, modifying or omitting some elements while striving to preserve its core rationale. However, the scope for coherence-preserving modifications is not unbounded.

• (b) When relevant, consensually accepted, evidence exist the narrative that offers a more coherent account of the ensemble of evidence will prevail. The emotional, imaginative or moralizing power that several scholars tend to view as the essence of narratives may sway some people some of the time, but will eventually be immaterial to the narrative’s acceptance. The fact that the case discussed here involves very strong and persistent emotions, beliefs, and ideological convictions reinforces this conclusion.

In this sense, the evaluation and preference of narratives are indeed rational, though they do not necessarily take place according to values like scope, simplicity, generality, etc.

An important corollary of the above is that historiographic narratives are substantially different from fiction. From Mink’s (1978, p. 148) observation that “narrative history and narrative fiction move closer together than common sense would well accept” and White’s insistence on “emplotment” and “fictionalizing” as the gist of historiographic discourse to Ankersmit’s recent (2012, p. 121) assertion that “the difference between the historical text and the historical novel is formal rather than material,” narrativist scholars tended to blur the gap between historiography and fiction. We just saw, however, that evidence can force modification in a historiographic narrative, endanger its coherence and even lead to its total abandonment.

A fictional narrative, on the other hand, will not be so constrained. The inspiring power of “let my people go” is immune to evidence that no mass Exodus of enslaved people from Egypt happened. To use a more recent example, while archaeological and genetic findings have significantly altered our historiographic narrative about the Neanderthals and their interaction with early modern humans, they are entirely irrelevant to our appreciation of the compelling narrative in William Golding’s (1955) novel “The Inheritors,” to its evocative power and, indeed, its alegoresis. Pace Ankersmit, then, the difference between historiographic writing and the historical novel is material and not formal.

By fictionalizing a chapter in history, an author (or a reader) can indeed instill it with meaning. For example, one could emplot the Conquest narrative as an epic victory or a tragic genocide. Both may be legitimate in some allegorical way— and impervious to the question of whether the said conquest ever happened. But the contention (White 1984, pp. 22–23) that the truth of a historical narrative can and should be assessed both factually and emotionally conflates two things that are inherently distinct.

This is not to deny that texts of historiography have a literary dimension and are often evaluated in terms of their artistic value. One may appreciate, for example, Albright’s combination of lucidity and erudition. But this aesthetic dimension is orthogonal to the factual one, whereby the historiographic adequacy of the narrative can be assessed. As Noël Carroll (1990, p. 155) succinctly observed, “the reduction of all narrative to the status of fiction seems a desperate and inevitably self-defeating way in which to grant the literary dimension of historiography its due.”

When a generalized historiographic narrative becomes untenable, so are individual texts that express and exemplify it. This is what happened to the books and treatises about the Conquest narratives cited in the previous pages. Later readers may examine these texts, as I did here, as historical data for another narrative, but not as an acceptable representation of the past.

It can be argued that the case discussed here is atypical because it looks into how competing narratives fared within a professional academic community. I am not sure that scholars and academicians are really unique in the manner in which they react to narratives (and, in fact, the philosophers cited above seldom consider scholars discriminately). Moreover, experience shows that an idea that gains stable scientific consensus will eventually make deep inroads into the general public. Witness, for example, what happened to the heliocentric system and the theory of evolution. The notice about “the early Israelites” in the Israel Museum cited above indicates that a similar process is happening here too.

Finally, the reader has perhaps noted that I avoided the use of the T-word. The truthfulness of historiographic narratives is a controversial subject among philosophers as well as among historians and discussing it would definitely exceed the scope of this work. I venture, however, to submit that even if we cannot safely assume that the currently consensual Autochthonic narrative is true, we are entitled to say that the once-dominant Conquest narrative is false, and that is not a small matter.