“A Relationship Worth Fighting For”: A Response to “Isn’t It Time to Break Up with Linguistic Dating?”

Linguistic dating in Biblical Hebrew research remains a valid scholarly tool despite complications arising from composition, redaction, transmission, and methodological challenges. Although weaker or overly circular approaches deserve criticism, careful, data-driven analysis of Masoretic, non-Masoretic, and extrabiblical Hebrew still reveals meaningful diachronic patterns and can thus contribute to historical periodization.

See also Diachronic Diversity in Classical Biblical Hebrew (Cambridge, 2024).

By Aaron D. Hornkohl
Associate Professor of Hebrew
University of Cambridge, Faculty of Asian and Middle Eastern Studies
March 2026

I wish to thank Robert Rezetko, Ian Young, Martin Ehrensvärd, and Martijn Naaijer (henceforth RYE&N) for their response[1] to my latest book, Diachronic Diversity in Classical Biblical Hebrew.[2] I appreciate the collegial, if critical, spirit of their engagement with my work and value the refinements that their criticisms have prompted over the years. As RYE&N’s article, though prompted by my book, is not so much a review as a summary of their years-long critique of the broad approach and methods that I adopt, I respond here first with some sympathy, then with some criticism of the critique, and finally with justification for a valid approach to Hebrew diachrony and linguistic periodization in the face of their criticism.

Sympathy with the Critique

Not all approaches to linguistic periodization are equally valid. Avi Hurvitz (z"l), widely regarded until his recent death as the foremost authority on Late Biblical Hebrew (= LBH), spent a career encouraging scholars to take note of Hebrew diachrony and, just as importantly, to employ a robust methodology when doing so. He regularly stressed the need for a repeatable procedure based on objective criteria: distribution, opposition, extrabiblical confirmation, accumulation.[3] Much of RYE&N’s criticism rightly finds fault with intuitive and impressionistic stabs at linguistic periodization, which draw questionable conclusions from dubious data. Regrettably, however, they often fail to discriminate between such problematic approaches and more compelling ones.

Data-driven Approaches

A refreshing commonality of RYE&N and many who work in the framework of linguistic periodization is the conviction that arguments should be rooted in data. But data points are only as good as the lens through which they’re interpreted. Throughout their review and the ensuing comments, RYE&N make repeated reference to their previous studies. This is legitimate. Individually and in various combinations, RYE&N have written voluminously against research in Hebrew diachrony and linguistic periodization in the tradition of W. Gesenius, S. R. Driver, E. Y. Kutscher, and A. Hurvitz. Among other strategies, they buttress their arguments with statistical presentations, which give the appearance of objectivity and unassailability. These are often, though, fatally flawed by various forms of misunderstanding. Some such defects have previously been noted in scholarly reviews of RYE&N’s research. RYE&N have sometimes responded directly to the criticism, and other times modified their method. Since, however, they continue in “Isn’t It Time to Break Up with Linguistic Dating” to refer to the results of their past studies as probative, with no mention of how past criticism might undermine their validity, it seems necessary to rehearse points of critique here as well as to update the list. The following catalog is by no means exhaustive, but is representative.

            RYE&N contend that no biblical book, whatever its date of composition, is free of late features and that all books pattern as late according to accepted procedures. For this claim they rely heavily on their ostensibly objective statistical counts of late linguistic elements, which appear to demonstrate comparable rates of late language use in a variety of both early and late sample texts.[4] It is crucial to note, however, that the statistical procedures in question have been roundly criticized.[5] Scholars have complained that they count features, but ignore frequency (i.e., tokens); fail to exercise sufficient discrimination in the selection of features and in the identification of relevant cases; and utilize non-representative sample sizes.

            In their 2014 book, Rezetko and Young answer some of the criticism of their 2008 statistics by arguing that these are simply the results of applying the standard Hurvitzian method.[6] But this is imprecise. By their own admission, Young, Rezetko, and Ehrensvärd “follow a loose definition of LBH features,” accepting “any feature cited by an authority as LBH provided that it occurs in more than one core LBH book (including… Qohelet).”[7] Emblematic of this overinclusive approach to data is the disparity between the 372 lexical entries in Young, Rezetko, and Ehrensvärd’s list of late linguistic features and the mere eighty in Hurvitz’s Concise Lexicon.[8] Second, whereas Hurvitz’s notion of accumulation weighs both features and frequency, as already noted, Young, Rezetko, and Ehrensvärd ignore tokens,[9] meaning that their methodology cannot distinguish between elements genuinely characteristic of a text or period and sporadic or anomalous phenomena.[10] The point is that the critique of the standard Hurvitzian method and results does not actually focus on Hurvitz’s methods and data, but on caricatures thereof.

            In a 2016 article prompted by Hurvitz’s 2014 Concise Lexicon of Late Biblical Hebrew, Rezetko and Naaijer tacitly acknowledge some of the group’s previous methodological oversights by incorporating several of their critics’ suggestions. They now count features and tokens and they limit items checked to those found in Hurvitz’s Concise Lexicon. The resulting article presents copious data in clear tabular form, allegedly demonstrating that no LBH or Transitional Biblical Hebrew (= TBH) composition uses LBH features more than their respective Classical Biblical Hebrew (= CBH) alternatives. Yet they seem unaware of a critical weakness in their study. Their major conclusion is that, according to a “Projected CBH–TBH–LBH” scheme, the core LBH books plot as CBH, or as TBH more proximal to CBH than to LBH, and not as LBH. They explain: “the scheme essentially reflects in a basic way what many, including Hurvitz in his Lexicon, seem to believe about BH: CBH writings generally use early variants” and “LBH writings generally use late variants.”[11] The problem is that this seemingly straightforward summary does not, in fact, reflect the view of Hurvitz or of other experts in Hebrew diachrony. While we do hold that CBH writings generally use early variants, we do not hold that LBH writings generally use late ones, i.e., use late variants more frequently than classical variants. This may or may not obtain with individual features in the relevant compositions. Our claim is, rather, that LBH compositions use late variants more than CBH compositions do. Rezetko and Naaijer thus labor under false premises. Their graphs show interesting variation in the frequency of late features among LBH books, but since they exclude graphs for CBH books—which, given strict adherence to Hurvitz’s methods and features, would certainly demonstrate smaller accumulations of LBH features than LBH books—they prove only that LBH is not as they wrongly imagined it to be. They fail, conversely, to substantiate their central claim, i.e., that the data undermine the established typological difference between CBH and LBH compositions.

            In a similar vein, the group maintain that the core LBH books are singularly redolent of post-exilic style, arguing that no Dead Sea Scrolls (DSS) composition shows a concentration of late features as high as the core LBH books.[12] This is an exaggeration—non-biblical DSS compositions show greater and lesser concentrations of Second Temple Hebrew, some comparable to those of the core LBH books—but even where true, in no way negates the standard diachronic paradigm. Though non-biblical DSS compositions vary in their accumulations of late features, all do exhibit late accumulations that surpass those found in Masoretic CBH and Iron Age II epigraphy. The fact that certain Judean Desert compositions have more classical profiles than others and even than the core LBH books merely reveals a predictable range among writers who all partook, in different ways, of a distinctively late Second Temple linguistic reservoir.

            Still on the topic of the DSS, RYE&N sometimes appear guilty of a category error: specifically, focusing on late biblical features to the exclusion of late extrabiblical features, an oversight particularly problematic when assessing the chronolect of non-biblical material, such as the extrabiblical DSS compositions. The diagnostically late features of Masoretic LBH are only part of a larger array of late features, many of which go unattested in the Hebrew Bible. Consider, by way of example, Young’s discussion of the Great Isaiah Scroll from Qumran (1QIsa-a). We all agree on the linguistic uniqueness of this scroll, consisting of a classical text contemporized with both generally late and specifically Qumranesque features. It is thus telling when Young concludes “I was wrong!” for thinking that 1QIsa-a “throughout reflects systematic linguistic variants that change its linguistic profile from EBH to LBH, comparable to the core LBH books of Esther, Daniel, Ezra, Nehemiah and Chronicles,” eventually culminating in his epiphanic declaration “[1QIsa-a] is not in ‘Late Biblical Hebrew.’”[13] The humility involved in disclosing such a scholarly realization is refreshing and commendable. Yet it is worth emphasizing that the original misconception was in no way representative of the views of experts in Hebrew diachrony, who have long known that the periodization of extrabiblical sources must take account of both late biblical and late extrabiblical linguistic features. 1QIsa-a’s departures from the more classical type of BH found in Masoretic Isaiah (and, for that matter, in [1QIsa-b]) extend beyond late biblical features. Compare Parry’s aptly entitled “Late Hebrew Forms in [1QIsa-a],” wherein he surveys “thirty categories of linguistic elements in 1QIsa-a that reflect LBH, QH [= Qumran Hebrew], or MH [= Mishnaic Hebrew].”[14]

            Far from minor or superficial defects, the above issues severely weaken, qualitatively and quantitatively, major planks in RYE&N’s customary critique. Their principal counterargument is basically reduced to two alternative hypotheses, one extreme and the other lacking any unequivocal documentary evidence: (1) that there’s too much compositional, redactional, and textual “noise” to perceive any trustworthy linguistic patterns of diachronic value and (2) that the ability to compose passable CBH was common among Second Temple scribes.

MT-centrism?

Perhaps counterintuitively, one of the cardinal principles of sound diachronic research on Masoretic BH is going beyond the MT to include non-Masoretic biblical and early and late extrabiblical evidence. Indeed, it is precisely a lack of acquaintance with any form of Hebrew beyond that found in the MT that arguably leaves mainstream biblical scholarship insensitive to Hebrew diachrony. For only via familiarization with the early extrabiblical Hebrew of the Iron Age II inscriptions and the late evidence of Second Temple sources—from Qumran and the Judean Desert, Ben Sira, the Samaritan Pentateuch, Rabbinic literature, and the ancient transcriptions, not to mention Aramaic and cognate evidence—is one equipped with the comparative data to situate Masoretic BH in its historical context and to periodize its component chronolects. It is revealing that higher and lower criticism routinely involve forays outside the MT, but ancient Hebrew continues to be studied in a single manifestation—Masoretic BH—as if it were a historical isolate. From this broader historical perspective, one sees that CBH (in all its manifestations: Masoretic, Samaritan, DSS) generally profiles along with Iron Age II epigraphy (notwithstanding important secondary orthographic, phonological, and grammatical developments), whereas LBH (in its various manifestations: Masoretic, DSS) patterns as one among many Second Temple varieties.

            Given the extensive use of non-Masoretic and extrabiblical language data just described, it is not entirely clear what RYE&N mean when they decry the privileging of the MT among specialists in Hebrew diachrony. What I suspect they mean is reliance on the MT for anything in the way of diachronically reliable linguistic evidence. This is sometimes couched in rather unfortunate pronouncements, e.g., their summary of my position vis-à-vis the MT as “the assumption that it has remained largely unchanged since 800 BCE.” I don’t think this does justice to my position.

            Throughout my book I routinely entertain the possibility that secondary developments, the reality of which I readily admit, might be responsible for observed patterns in linguistic alternation. Consider, by way of example, the following, from my chapter on orthography:

No text in the Masoretic Bible is characterised by spelling conventions as regularly defective as those of the Iron Age Hebrew inscriptions…. This means either that the earliest biblical texts were written later than the inscriptions or, alternatively, that their orthography, once more defective, was updated over the course of their literary and textual development and transmission.

Evidence for the latter alternative is forthcoming from several DSS versions of biblical texts.[15]

I then trace orthographic development in various Masoretic and non-Masoretic biblical sources and attempt to show where the tendency for defective versus plene spelling in the extant Masoretic Torah, whether primary or secondary, might fit into this development in comparison to other parts of the Masoretic Bible. I do, it is true, hesitantly suggest 800 BCE as a dividing line between the CBH of the Torah and the rest of CBH (more on this below). Even so, the claim that I view that date as a sort of cut-off point for development is an oversimplification repeatedly contradicted throughout my book. I nowhere claim immutability, because I don’t need to. I contend only that enough early detail can be made out to sustain the argument that the established CBH profile can be roughly divided into sub-chronolects approximately mappable onto real-world historical periods.

Different Approaches to Bad Data

One of RYE&N’s alternative hypotheses is that the apparently chronological nature of the CBH–LBH divide is itself secondary, a result of the vagaries of compositional development and scribal transmission. This hypothesis reflects a fundamental difference in approaches to what we all can agree is a problematic body of evidence. While I embrace such data, including cruxes and secondary developments, as opportunities for exploring diachronic possibilities, RYE&N insist that the high rate of compositional and textual “noise” precludes reliable testimony. In their view, the variations on which linguistic periodization rests involve grammatical and lexical details that are precisely the sorts of phenomena that regularly fluctuate during redaction and transmission. I do not dispute this. There’s no shortage of evidence of secondary development, some of it extensive. My contention is that there remains sufficient reliable evidence for the tracing of approximate diachronic contours. Put differently, I reject the notion that the obscuring of linguistic detail was so thorough that it rules out the possibility of fruitful diachronic interrogation.

            Crucially, this need not remain some battle of theories in a thought experiment. It can be tested. Indeed, the conviction that reliable linguistic data of diachronic import can be recovered from transmitted texts is not an assumption, nor is it based on intuition, optimism, or dogma. It is grounded in evidence. Detailed, text-based studies repeatedly demonstrate that, despite secondary developments, BH as reflected in the MT was not uniformly leveled to the point where late has become indistinguishable from classical. This does not mean that linguistic periodization is always feasible. RYE&N tend to focus on inner-biblical and MT–DSS parallels where no edition seems to exhibit a demonstrably later linguistic profile, e.g., MT Samuel versus [4QSam-a]. But such cases merely prove the persistence of largely classical profiles despite secondary development. The very fact that scholars consistently perceive late linguistic assemblages in late texts securely periodized on non-linguistic grounds is testament to at least some level of practical preservation of diachronically interrogable detail.

            The alternative view borders on conspiracy theory: is it more credible to hold, as I do, that diachronically significant distribution patterns survived secondary activity or, as they do, that such patterns are the result of redaction and transmission?[16] Where cases of orthographic, linguistic, textual, or compositional cruxes involving diachronically meaningful features are actually attested in manuscripts, these must be thoroughly investigated—indeed, diachronically sensitive studies routinely do this, as such cruxes often prove illuminating from multiple perspectives—but it would be gratuitous—and anti-scientific—to give up the whole enterprise of Hebrew diachrony and linguistic periodization in the face of the dubious blanket claim that the minority of secondary developments that actually touch on diachronically meaningful features render the data uniformly untrustworthy. Such a sweeping assertion is simply not borne out by documentary evidence. Rather, it is itself a hypothesis, and an extreme one at that, not universally accepted among text critics.[17] There is no denying the reality of literary growth and textual fluidity and pluriformity in the development and transmission of biblical texts, but the assumption of an evidentiary situation as dire as that painted by RYE&N requires a leap from genuine documentation to a radical extrapolation therefrom. This runs counter to the spirit of enquiry, because it preemptively blocks investigation.

Circularity?

One often hears that linguistic arguments for periodization necessarily rest on “preconceived notions of what writings are already early and late, so the entire linguistic dating paradigm is circular.”[18] While perhaps typical of some approaches—which do merit criticism—circularity is not a necessary ingredient in linguistic arguments for periodization. For what it’s worth, here’s my application of the standard Hurvitzian approach.

            I take as axiomatic only that the core LBH books as well as late extrabiblical material (Rabbinic Hebrew, Ben Sira, some of the non-biblical DSS, Second Temple Aramaic) are late and that Iron Age II epigraphy is early. I am agnostic—i.e., I initially suspend judgement—about the periodization of everything else. Utilizing an inventory of characteristically late linguistic features amassed on the basis of Hurvitz’s procedure, I check texts for accumulations of late features. I am confronted with compositions displaying conspicuous concentrations of late features and texts lacking such concentrations. Leaving aside, for the sake of argument, liminal and other problematic constellations of features, I contend with two prototypical varieties of BH. However I choose to label them, CBH (or SBH or EBH) and LBH (or PBH),[19] I am responsible to account for the differences (note: though RYE&N often grapple with this matter, it goes largely unnoticed in the “mainstream literary-critical scholarship” they defend). There are numerous factors at work (dialect, register, idiolect, genre, scribal intervention), but the most obvious one, and the one with the greatest explanatory power, has long been recognized as chronology.

            Now, as far as I understand the debate, there is not much disagreement on the association of LBH with the Second Temple period—LBH is by all accounts one manifestation among several representing the broader category of Second Temple Hebrew, the distinctive features of which—biblical and non-biblical—show up in some non-Masoretic biblical sources (e.g., 1QIsa-a, the Samaritan reading tradition) and a great deal of extrabiblical material (DSS, Ben Sira, Rabbinic literature). The association of CBH with the First Temple period, on the other hand, is debated. It is not, however, a baseless assumption, but a hypothesis that effectively accounts for a large body of facts. The hypothesis is based in part on similarities between CBH and Iron Age II epigraphic Hebrew (and differences between LBH and Iron Age II epigraphic Hebrew) and in part on relative chronology as manifest in typology (LBH variants are usually developmentally later): despite partial updates to CBH spelling, pronunciation, and grammar in line with LBH conventions, CBH’s basic structure, e.g., grammar and lexicon, typologically precedes that of LBH. In other words, CBH looks older than LBH. Fundamentally, then, the diachronic hypothesis—whereby the difference between CBH and LBH is linked to real-world chronolects—is a useful narrative that plausibly accounts for much of the linguistic variation in question. What’s more, one’s confidence therein is often reinforced by additional evidence and internal consistency—the more one finds late features confined to late texts, the more convinced one becomes that their absence from texts of unknown date is indeed a consequence of chronology.

            There is nothing circular here. Rather, we build a rubric on the basis of what is known and uncontested using a rule-governed procedure, assess against that rubric, and construct a plausible narrative to explain the resulting patterns. Thus performed, the diachronic approach is a working hypothesis, and one that deserves its fair shake in the marketplace of ideas alongside other hypotheses, whether alternative or complementary. Such an impartial assessment, sadly, remains a desideratum. Many scholars remain un(der)informed on Hebrew diachrony and linguistic periodization and RYE&N’s assessment, in various ways, is neither objective nor fair, but seems driven by the conviction that the scholarly status quo should not be upset. Indeed, one of them has written explicitly: “For me, the central issue is that linguistic dating cannot be reconciled with mainstream literary-critical scholarship….”[20]

Different Notions of Integration

RYE&N call for an integrated approach to historical linguistics and dating biblical texts. So do I.[21] But we clearly have different views on the nature of such integration. I contend that, despite a scholarly lineage that goes back to Gesenius’s 1815 Geschichte,[22] arguments that incorporate linguistic typology are regularly absent from mainstream biblical scholarship. This is arguably due to the specialized linguistic focus of the arguments and to the necessity of venturing beyond Hebrew in its Masoretic guise to non-Masoretic biblical and extrabiblical Hebrew and cognate evidence. I contend that publications in the vein of RYE&N’s opposition encourage and reinforce this ignorance and neglect, which results in segregation rather than integration. If anything, RYE&N’s stance may be construed as a call to disregard “troublesome” data patterns and to cancel “inconvenient” interpretations in defense of what they more than once call “scholarly consensus.” Crucially, though, this “scholarly consensus” (such as it is) was reached without much attention to Hebrew diachrony. Further, the claim that diachronic studies are incompatible with accepted scholarly views is an overstatement. Linguistic periodization frequently confirms conventional theories or furnishes support for one alternative over another in ongoing debate. True integration would involve biblical scholars of all stripes regularly reckoning in their research with diachronically significant linguistic evidence, even if doing so might mean acknowledging its opposition to conclusions reached via other avenues of enquiry. That seems far more intellectually salutary than jettisoning an entire scholarly approach because a majority of the “mainstream” largely unfamiliar with the data and procedures are uncomfortable with its results.

A Positive Example of Integration

Looking forward, it is useful to consider a positive case of integration, for example, Benjamin Suchard’s online review of my book,[23] which, though disputing various of my points, accessibly integrates historical, literary, and linguistic evidence. Suchard takes issue not so much with the theory and methods that I employ, as with the persuasiveness of specific examples and with the absolute chronology that I tentatively propose. With little in the way of Hebrew epigraphy between 600 and 200 BCE, there are few extrabiblical controls to help us verify when Iron Age II Hebrew gave way to the various forms of Second Temple Hebrew that one encounters. Some tie this transition to the Babylonian Captivity (6th century BCE), others to the Restoration (5th century BCE), but, with a dearth of external evidence from precisely this period, it may not be unreasonable to delay this some. But if so, how far? The problem is basically one of compellingly situating the proposed BH chronolects in the space remaining before the DSS—which preserve both CBH and LBH biblical books with the same chronolectal profiles they have in the MT. There is much to consider here. And while it certainly goes beyond language, it would seem unwise to exclude linguistic considerations from the discussion.

Late Writers and Classical Style

The question of the persistence of Iron Age II Hebrew raises the thorny issue at the heart of RYE&N’s second hypothesis: that the ability among late writers to produce passable CBH was commonplace. My own evolving position is as follows: in view of the extant evidence of concentrations of late features in all compositions acknowledged to be late on non-linguistic grounds, the late ability to compose long stretches of flawless CBH must be considered exceptional rather than normative. Thus, the longer a purely CBH composition is, the more implausible its late composition should be deemed. Of course, holding that successful late classical imitation was an exception rather than an impossibility is still incompatible with the view espoused by RYE&N, namely, that any CBH material could conceivably have been written in the Persian or Hellenistic periods. It likewise conflicts with the sheer number of “mainstream” theories that date the composition of large sections of CBH material to the 5th century BCE and beyond.

Conclusion: Linguistic Periodization as One among Many Complementary Approaches

Given the limited scope and problematic nature of the extant evidence, all the avenues of enquiry that comprise biblical studies are necessarily exercises in plausibility rather than certainty, where scholars attempt to construct explanatory narratives on the basis of data sets that practitioners in other fields might consider laughable. Faced with such an epistemological challenge, it is not just acceptable, but indeed desirable that experts in various subfields should each contribute their specialist data, methods, and conclusions. While this might be disparaged as privileging one’s own data, it can alternatively be viewed as reducing variables lest a provisional model become unwieldily complex. Then there can be integration: assessing the relative plausibility of alternative explanatory narratives vis-à-vis existing narratives and synthesizing new narratives, as appropriate. RYE&N treat linguistic periodization as a threat to “mainstream scholarly consensus.” But, just like language, scholarly consensus is not static, but constantly evolving. I see the role of experts in Hebrew diachrony as that of adding one more useful implement to biblical scholars’ toolkits in support of the various sorts of research they undertake.

            But it’s a two-way street. Hebrew diachrony in the service of linguistic periodization can only go so far. It leaves many questions only partially answered, especially when it comes to liminal cases, anomalous linguistic profiles, and sub-chronolects. And, of course, it must be used together with literary and textual criticism as well as various exegetical approaches, since biblical scholars are rightly interested in more than just datation.

            Will neglect, friction, and disagreement continue to characterize the relationship between biblical studies and Hebrew diachrony? Possibly. Is that grounds for a breakup? Surely not! The strongest and most fruitful partnerships are those in which differences are either accepted and appreciated or reconciled. We can still enjoy a long and fruitful relationship, but this will require a commitment, at the very least, to keep listening and weighing the other side’s views, rather than prematurely concluding that our differences are necessarily irreconcilable.