Linguistic Dating of Biblical Texts
While ancient Hebrew underwent linguistic change, as do languages in general, the biblical texts seem not to reflect this chronology in a way that makes any kind of linguistic dating of the texts possible – in contrast to the consensus prevailing among Hebrew linguists until about a decade ago.
Essay based on Linguistic Dating of biblical Texts. Volume 1: An Introduction to Approaches and Problems. Volume 2: A Survey of Scholarship, a New Synthesis and a Comprehensive Bibliography. BibleWorld. London: Equinox Publishing, 2008.
By Ian Young
Associate Professor, Department of Hebrew
Biblical and Jewish Studies
The University of Sydney
By Robert Rezetko
Assistant Professor, Faculty of Religious Studies
Radboud Universiteit Nijmegen
Honorary Research Associate
Department of Hebrew, Biblical and Jewish Studies
The University of Sydney
By Martin Ehrensvärd
Part-time Lecturer, Department for the Study of Religion
University of Aarhus
Lecturer, Department of Biblical Exegesis
The University of Copenhagen
In the last few years, a challenge has been mounted to the consensus view that Biblical Hebrew (BH) can be divided into two discrete historical periods: Early Biblical Hebrew (EBH) and Late Biblical Hebrew (LBH), or early Hebrew and late Hebrew. The starting point for this challenge was the publication of a volume Young edited with―in the words of one reviewer―the "yawn-invoking title" of Biblical Hebrew: Studies in Chronology and Typology (2003).
EBH, according to the traditional view, is the language of the preexilic or monarchic period, down to the fall of the kingdom of Judah to the Babylonians in 586 BCE. The exile in the sixth century BCE marks a transitional period, the great watershed in the history of BH. After the return from exile in the late sixth century BCE, we have the era of LBH. Thus, EBH developed into LBH. Hebrew biblical texts can, therefore, be dated on linguistic grounds because LBH was not written early, nor did EBH continue to be written after the transition to LBH.
We suggest that following through the logic of this chronological approach to BH actually leads inevitably to the conclusion that all the biblical texts were composed in the postexilic period, which is exactly the opposite of what its proponents have claimed. Now, this may in fact be a conclusion which is congenial to some. But others will not find this agreeable, so we will offer a way out of this conclusion by arguing that the presuppositions of the chronological approach are undermined by the evidence. On the contrary we will argue that the best model for comprehending the evidence is that "Early" BH and "Late" BH, so-called, represent co-existing styles of Hebrew throughout the biblical period. Then we will deal with the objection that Persian loanwords are an irrefutable proof that the chronological approach is correct. Finally we will step back and ask some hard questions about the presuppositions involved in the dating―by linguistic or other means―of the books of the Hebrew Bible.
2. Standard Chronological Hypothesis Leads to Late Dating of Entire Hebrew Bible
It is the work of the great Israeli scholar Avi Hurvitz that has established sounder methodological principles and therefore decisively advanced the study of LBH in recent decades. One of his many important advances is to put to rest older scholars' insistence that "Aramaisms"―or Aramaic-like forms―are necessarily evidence of a late date. Contrast, for example, Otto Eissfeldt's argument regarding Song of Songs―Aramaisms and a Persian word equals lateness―with John Collins, who only mentions the Persian word. It is, therefore, not inappropriate if we concentrate on Hurvitz's methodology and presuppositions here. We illustrate these first of all by several quotes from his early work on the Prose Tale of Job.
Hurvitz argues in his article on the Prose Tale of Job, as he does elsewhere, that the late elements in the text "betray their actual background; and if they are not few or sporadic ― in which case their occurrence may be regarded as purely incidental ― they effectively date a given text." Later, he mentions "the existence of a considerable number of such [late] elements in the Prose Tale..." and concludes: "As far as can be judged from the linguistic data at our disposal, these non-classical idioms ought to be explained as post-classical ― namely, as imprints of late Hebrew ― thus making the final shaping of the extant Prose Tale incompatible with a date prior to the Exile." Thus: "It would appear that in spite of his efforts to write pure classical Hebrew and to mark his story with "Patriarchal coloring," the author of the Prose Tale could not avoid certain phrases which are unmistakably characteristic of post-exilic Hebrew, thus betraying his actual late date."
Thus, according to Hurvitz, despite his best efforts, it was not possible for the author of the Prose Tale of Job to avoid using LBH linguistic features. Here, however, we note a striking fact about the argument. Even demonstrating, using Hurvitz's careful methodology of distribution-opposition-external attestation (as discussed below), that a particular linguistic feature is LBH does not lead to the classification of the text in which it is found as LBH. As Hurvitz admits, LBH linguistic elements are found in EBH texts. Thus, as Hurvitz emphasizes in the quote above, it requires a "considerable number" of such LBH elements in a text before that text can be considered LBH. This is the criterion of accumulation.
In his article on the Prose Tale of Job, Hurvitz identifies seven linguistic features in the 749 words of the Prose Tale as LBH elements and considers this enough evidence to date the work late. Now at this point the uninitiated might be a bit puzzled by the argument. Why, it might be asked, did Hurvitz need to decide that seven linguistic features were enough to date the Prose Tale late? Why, in the first quote, did he need to emphasize that the LBH forms need to be more than few or sporadic? Surely, if these are really "late" linguistic items, the appearance of any one of them should indicate a late date for the passage in which it appears. But this is not the way it works in attempts to date BH writings linguistically.
One of Hurvitz's most important contributions to scholarship is his insistence on a careful methodology. For an individual linguistic item to be considered characteristic of LBH, it must have a distribution among the core LBH books of Esther, Daniel, Ezra, Nehemiah, and Chronicles. Thus, for example, 78 out of 91 occurrences of malkut for "kingdom" are in core LBH books. Hurvitz's basic starting point, which we think is uncontroversial, is that these books are postexilic and therefore their language represents samples of postexilic Hebrew. The other key element in Hurvitz's methodology is that not only must the linguistic element be evidenced in the LBH books, it must exhibit a linguistic opposition; in other words, it must be used in similar contexts as other forms in the core EBH books, especially the Pentateuch and Former Prophets. This crucial step ensures that we really do have variant language, not just linguistic forms that had no opportunity to appear in EBH books. Thus malkut contrasts with other BH words for "kingdom" like mamlakah. Hurvitz has a third criterion, external attestation, which tries to demonstrate that the form really is late by finding whether it occurs in late, mostly post-biblical Hebrew and Aramaic. Thus malkut is used widely in later Aramaic dialects and in Tannaitic literature like the Mishnah. This last criterion, however, promises more than it delivers. Given that the overwhelming majority of extra-biblical Hebrew and Aramaic dates to the postexilic period or later it is virtually inevitable that BH linguistic forms―whether early or late―will be attested in "late" extra-biblical sources. Malkut is also attested in early Aramaic, so it is hard to see how the external evidence proves anything regarding date. We believe, in any case, that the excellent criteria of distribution and opposition are enough to demonstrate that a form is characteristic of the core LBH books.
This is far from being the full story, however. Only a small minority of well-attested LBH linguistic forms are not also found in EBH texts. Thus we have seen that malkut is clearly a characteristic of LBH, occurring 91 times in the Hebrew Bible, 78 of them in the core LBH books, and a further six times in LBH-related psalms and Qoheleth. Yet, still, the remaining seven of those 91 occurrences in the Hebrew Bible are found in core EBH books like Samuel and Kings.
To us this phenomenon raises questions about the chronological approach which are not adequately explored by its proponents. Is malkut actually a "late" linguistic item after all? If so, its appearance in a text should indicate that therefore the text is to be dated late. And if EBH texts which use malkut were dated late, this means late texts need not exhibit an accumulation of LBH features. If against this it is argued that the LBH linguistic feature found in the EBH text is not actually "late" but was also available in an early period, then its value for dating texts "late" is negated. Despite the claims of the criterion of accumulation, to which we will turn shortly, there is no reason to assume that an early author could not produce a text with a clustering of LBH elements if they were available to him. Perhaps, though, it is not the mere availability of LBH features, but their frequency which counts. In the study of the natural development of a spoken language over time, factors such as the date of origin of a specific linguistic feature and the growth in the frequency of its use in the language are indeed important considerations. However, the consensus among scholars of the Hebrew language is that spoken and written varieties of Hebrew were some distance apart in the biblical period. In such a situation, the appearance of forms from the spoken dialects in literary Hebrew is unpredictable and a function of the individual stylistic choices of authors and scribes. The natural development of the spoken language behind the biblical texts is masked by this unpredictability. Issues such as the frequency with which such features occur, or that other texts avoid them completely, are dependent on the stylistic choices of individual authors and scribes, not a reflection of the chronological development of a natural language, as we shall see further below in detail. Or perhaps, finally, the LBH elements in EBH texts are evidence of later textual alteration of the language of the BH books. Proponents of the chronological model have been loathe to invoke this explanation. Once it is admitted that the language of the biblical texts has been changed in scribal transmission, the claim that the language of the current texts is evidence of the date of the original authors is thrown into serious doubt. We will return to this point later in section 5.
These questions aside, it is the phenomenon of the appearance of LBH linguistic items in EBH texts that leads to Hurvitz's final and most important criterion when it comes to dating texts: accumulation. This states that a text can only be considered LBH if it exhibits an "accumulation" of LBH features, identified using the above criteria of distribution and opposition. Unfortunately, nowhere to our knowledge has anyone actually specified how much of an accumulation is necessary for a text to be LBH. Hurvitz's claim that the Prose Tale of Job (749 words in length) is in LBH on the basis of seven LBH features, however, gives us an indication of how he might apply this criterion.
Another difficulty with the criterion of accumulation is that we are not aware of any procedure which has been developed for how such an accumulation should be measured. In response to this problem we developed a simple test of accumulation. Plainly put, this counts how many different LBH features occur in a given stretch of text. Where possible, this stretch of text will be of 500 words length, or to be more precise 500 Hebrew graphic units, so that samples will be comparable. Within this sample we count how many different LBH features there are. We do not count repetitions of the same feature since once an author has demonstrated the possibility of using a particular LBH form, there is no reason why it cannot be repeated as often as the opportunity presents itself.
LBH Features in Biblical and Extra-Biblical Hebrew Texts
(500 Word Samples; Descending Order of Frequency)
|Text||Number of LBH Features|
|1||Ezra 1.1–11; 9.1–10.2a||25|
|2||Daniel 1.1–20; 11.44–12.13||24|
|3||2 Chronicles 30.1–31.3/td>||22|
|6||Qoheleth 1.1–2.9; 6.1–12||15|
|7||Temple Scroll (11QTa) 57.7–59.21||13|
|8||Damascus Document (4QDa) 2, I; 10, I; 11||12|
|10||Community Rule (1QS) 1.1–3.2||9|
|11||War Scroll (1QM) 1.1–2.11a; 2.16–3.6||9|
|12||1 KINGS 22.6–35||8|
|14||1 SAMUEL 13.1–14.9||6|
|15||2 SAMUEL 6.1–20a; 7.1–12||6|
|16||2 SAMUEL 22.1–51 (cf. synoptic Psalm 18.1–51, below)||6 (in 382 words; projected 7.9 in 500 word sample)|
|17||1 KINGS 2.1–29||6|
|19||PSALM 18.1–51 (cf. synoptic 2 Samuel 22.1–51, above)||6 (in 394 words; projected 7.6 in 500 word sample)|
|21||Pesher Habakkuk (1QpHab) 5.3–12.13||6|
|23||GENESIS 24.1–36 (J)||4|
|24||Ben Sira 41.13–44.17 (cols. 3.15–7.24)||4|
|26||EXODUS 6.2–12; 7.1–13; 9.8–12; 12.1–7b (P)||1|
- Core LBH in bold: Esther, Daniel, Ezra, Nehemiah, (non-synoptic) Chronicles.
- Core EBH in capitals and bold: PENTATEUCH, FORMER PROPHETS, PREEXILIC PROPHETS, PSALMS.
The table is very clear. The first surprise is that every sample we have studied includes LBH features. The only difference is the degree of accumulation of them. The core EBH and core LBH books are at different ends of the scale in terms of the amount of accumulation of these LBH features. Thus, while the highest core EBH sample, 1 Kings 22, has eight different LBH features, the lowest LBH sample, Esther 5–6 has 17, more than twice as many as 1 Kings 22, while all the other core LBH samples have yet higher numbers of LBH features.
One fact that is evident from the table is that Hurvitz and other proponents of the chronological approach have underestimated the amount of LBH features in EBH texts. His argument for linguistically dating texts like the Prose Tale of Job to a late period leads, in fact, to the conclusion that all the biblical texts are postexilic.
Recall that, firstly, Hurvitz argues that we know the features of postexilic Hebrew primarily by the distinct features of the core LBH books. Then, secondly, he argues that even when late writers tried to write early Hebrew, they betrayed their late origin by the use of a number of LBH features. This description of the late authors turns out to be a perfect description of "Early" BH: the authors attempted to write in a more conservative style than the LBH authors, yet they repeatedly use LBH features, so as to exhibit accumulations of them. Therefore, if these LBH linguistic features are actually late, Hurvitz's argument logically indicates that EBH is late, postexilic Hebrew by writers attempting (but failing) to write a more archaic or conservative style of Hebrew. The LBH authors in contrast were simply not so concerned to avoid these same linguistic features. The point is that both EBH and LBH authors used the same linguistic forms, just to different degrees. Thus EBH and LBH would turn out to be two styles of postexilic Hebrew, conservative and non-conservative.
3. Unravelling the Presuppositions of the Chronological Hypothesis
So, as we mentioned before, some of us may be quite content to leave all the biblical books in the postexilic period. However, for others of us we will now argue that the logic of the argument that leads to this conclusion is false because many of the presuppositions of the chronological approach are mistaken.
Hurvitz's initial and basic presupposition is that the core LBH books of Esther, Daniel, Ezra, Nehemiah, and Chronicles faithfully reflect postexilic Hebrew. This is a faultless presupposition given the evident dates of all these texts based on their internal references to at least the beginning of the Persian period. The inferences that Hurvitz drew from this initial starting point are, however, all open to question in the light of recent research. These include, firstly, that LBH was the only sort of Hebrew in the postexilic period and that therefore a late author was incapable of writing Hebrew without betraying his lateness, and secondly, that LBH could not be preexilic also. A third presupposition that underlies all linguistic dating work is that the traditional Hebrew MT forms of the books faithfully reflect the language of the original authors. Below (section 5) we will return to this last point.
The primary characteristic of EBH books that marks them apart from the core LBH books is a relatively low accumulation of LBH linguistic features. From the table above it is evident that such a low accumulation is a characteristic also of works doubtlessly composed in the postexilic period. Zechariah 1–8 is, according to the biblical book, a prophet of the early postexilic period. Even more surprising, in light of the expectations created by the chronological approach, is the low number of LBH links in our sample from the book of Ben Sira from the second century BCE and in the Qumran Pesher Habakkuk from the first century BCE. Thus far, in fact, we have not found a non-biblical work at Qumran with a LBH-like accumulation of LBH features. In light of this evidence, we arrive again at the conclusion that LBH was but one style of Hebrew in the postexilic period, alongside EBH.
It is even questionable whether LBH was a style of Hebrew restricted to the postexilic period. Most importantly, dating the language of the MT forms of the EBH books to the preexilic period, as Hurvitz does, indicates that a large number of LBH features were already in existence and available to be used in the preexilic period. Only a very small number of well-attested LBH features are not also found in EBH books. This is a crucial point. Recall the example of malkut. That this is not simply due to postexilic modification of the language of preexilic books―a possibility rejected in practice by Hurvitz―is evidenced by a number of LBH features in preexilic, monarchic era inscriptions. The 500 words of the Arad inscriptions exhibit an accumulation of nine LBH features―more than the accumulation that led Hurvitz to consign the Prose Tale of Job to the postexilic period! In fact, as you can see from the table, the preexilic Arad inscriptions from c. 600 BCE have a higher accumulation of LBH features than Ben Sira and Pesher Habakkuk, sources from the last two centuries BCE. Chronology is clearly not the explanation for these accumulations of LBH features, but rather that some authors have a stylistic preference for them. There is a strong case that many, if not most, LBH linguistic features already existed in preexilic Hebrew. If so, there is no reason why a preexilic author ―like Qoheleth as Young has argued―could not produce a work with an accumulation of LBH features. Thus it is a reasonable suggestion that even in the preexilic period LBH could have been a co-existing style of Hebrew alongside EBH.
Rather than a model suggesting that EBH and LBH are successive chronological phases of the language, which is incompatible with the evidence, a better model sees LBH as merely one style of Hebrew in the Second Temple period and quite possibly First Temple period. Both EBH and LBH are styles with roots in preexilic Hebrew, which continue throughout the postexilic period. "Early" BH and "Late" BH are not different chronological periods, but co-existing styles of literary Hebrew throughout the biblical period. These two general language types, EBH and LBH, are best taken as representing two tendencies among scribes of the biblical period: conservative and non-conservative. The authors and scribes who composed and transmitted works in EBH exhibit a tendency to conservatism in their linguistic choices in that they only rarely used forms outside a narrow core of what they considered literary forms. At the other extreme, the LBH authors and scribes exhibited a much less conservative attitude, freely adopting a variety of linguistic forms in addition to (not generally instead of) those favored by the EBH scribes. Between extreme conservatism (e.g., Zechariah 1–8) and extreme openness to variety (e.g., Ezra), there was a continuum into which other writings may be placed (e.g. Ezekiel).
Within this new model, much of Hurvitz's methodology is still sound. The criteria of distribution, opposition, and accumulation are still valid ways of describing linguistic relationships. It is still a worthwhile task to map the linguistic relationships of the biblical books to each other. Thus it is a significant result that MT Ezekiel has a higher number of links to core LBH books than any other prophetic book. The difference is that our new model simply does not relate such findings immediately to chronology. The relaxing of focus on chronology also allows us to notice other patterns of linguistic relationship that have no relevance to the old chronological model. Thus, scholars have long noted that Deuteronomy has a strong preference for lebab as "heart," while Jeremiah strongly prefers leb. Jeremiah's preference is shared by, among others, Genesis, Exodus, Numbers, Judges and Samuel, and hence is not obviously a sign of "lateness." Among the LBH books, Chronicles and Daniel align with Deuteronomy in preferring lebab, whereas the other LBH and LBH-related books prefer lebab. Because this data cannot be used in support of the chronological theory, it has been generally ignored by language scholars. If enough of such linguistic links are established, we may be able to note other groupings of books beyond just EBH versus LBH. We think that this sort of approach represents the way forward in the study of the linguistic diversity of the Hebrew Bible.
4. Persian Loanwords
Surely the Achilles heel of any non-chronological approach to BH is the distribution of Persian loanwords. We recall Collins' opinion above on the Song of Songs. Even though the argument from Aramaisms has rightly gone out of fashion since Eissfeldt's day, the one related to Persian loanwords is still going strong and has had to bear increasing weight in the current discussion.On the one hand, it is generally considered, as Mats Eskhult has recently put it, that "Persian loanwords...almost unequivocally point to the Persian era" (c. 500–300 BCE). The other side of this argument is also well put by Eskhult: "What deserves to be stressed is that Persian words are notto be found in the Pentateuch at all! If loanwords of Persian origin are considered a strong argument when dating biblical texts, then the lackof every vestige of such loanwords ought to be considered as an important evidence for a date of origin prior to the Persian era."
We believe this approach is wrong for the following reasons. First, Persian loanwords are absent from a number of biblical works that are universally acknowledged as postexilic, such as the books of Haggai, Zechariah, and Malachi set in the early Persian period, but also various other books generally considered postexilic, such as Joel, Jonah, Job, and Ruth. Also in the Persian era, we may point to the existence in Aramaic of two co-existing styles, the conservative Western style, regularly able to avoid Persianisms, and the non-conservative Eastern style, replete with Persianisms. In any case, the Hebrew evidence on its own is enough to indicate that absence of Persian words does not prove that a text is preexilic. In addition, we may note the powerful analogy with the fact that the Qumran authors, even though writing well into the Hellenistic period, were able to avoid using Greek loanwords. Persian and Greek are non-Semitic languages, and hence presumably loanwords from those languages were easy to identify and reject should the author so choose. Therefore, the absence of Persian loanwords in some biblical texts should not be given much weight in arguing for the pre-Persian date of those texts.
It is seldom understood that Persian is one of several related Iranian languages along with, for example, Median. The Medes were an important world power in the preexilic period, showing that Iranian languages were prominent before the Persian period. In fact, various biblical texts contradict the idea that Persians were not in contact with the West in the pre-Persian era. Thus, Ezekiel 27.10, a text internally dated to the end of the preexilic era, refers to Persians serving in the army of Tyre. Ezra 4.9–10 claims that the Assyrian king Ashurbanipal settled Persians in Samaria in the seventh century BCE. Assyrian deportations, in fact, led to significant population movements throughout the Ancient Near East, centuries before the exile of Judah. Media is the second most commonly cited place of origin for deportees, and the Assyrians were famous for moving populations from east to west and west to east. See, for example, the exile of Israelites from Samaria to "the cities of the Medes" in 2 Kings 17.6. Although we seldom possess the full details of the origin and destination of specific deportations, a strong case can be made that Iranians were settled in multiple deportations in the late eighth century BCE to Ashdod, Gath, and further south toward the Egyptian border. In line with this, scholars have identified Iranian names on seventh-century BCE texts from the region. Thus, far from being cut off from Persian and related languages until the era of the Persian empire, people of Persian or other Iranian extraction were probably near neighbors of the kingdom of Judah for a significant part of the preexilic period.
Finally, there seems little understanding of the way Iranian loanwords are identified in biblical texts. It is inevitable in research on loanwords that one must start with presuppositions about which languages are likely to have influenced BH. Thus, it is unlikely that one should turn to Australian aboriginal languages to elucidate difficult words in the Hebrew Bible. In line with the general consensus of biblical scholars, Iranian scholars have typically looked for Persian linguistic elements only in those texts they believed stemmed from the Persian period.
Thus, for example, the Persian word dat ("law") occurs in the MT of Deuteronomy 33.2, but this understanding of the word is generally rejected because it is considered impossible for a Persian loanword to appear in a text as early as Deuteronomy 33. Note, as another example, the discussion by Ran Zadok, a leading expert on biblical names, of the name Arioch in Daniel 2.24–25. He concludes that an Iranian "etymology is accepted here provided that the name in Daniel is not the same as the much earlier 'rywk in Gen. xiv 1, 9." In other words, there is no difference in form between the Iranian word in Daniel and the non-Iranian one in Genesis. It is just that it is impossible for Iranian linguistic elements to appear in "early" biblical texts.
There are no grounds for criticizing the Iranian scholars for their methodology. As we have mentioned, the identification of loanwords is so hypothetical that it is essential that some presuppositions are used as a starting point. The problem comes when scholars working with BH fail to understand the methodology of the Iranian scholars and claim high significance for the fact that there are no indisputable Persian words or names in EBH sources. This, as should now be clear, is nothing more than a circular argument: scholars have not identified Iranian elements in EBH because they have presupposed that EBH is too early to have them; other scholars then claim that since EBH texts do not have Iranian linguistic elements; this demonstrates their early date.
In fact, a preliminary search through the literature has enabled us to compile a list of 12 Iranian words and names in EBH sources. Whether they are really Iranian or not, it is hard to say, given the hypothetical nature of loanwords research, but we have compiled the list to counter the idea that it cannot be done. And we argue that it is very likely that the MT intends us to read the Persian words dat in Deuteronomy 33.2, raz ("secret") in Isaiah 24.16, and peladot ("steel") in Nahum 2.4. So if we take the evidence as it now stands―the usual procedure of Hebrew linguists―then the answer to whether there are Persian elements in EBH texts is unequivocally "yes."
5. Textual Criticism
The issue of loanwords is a convenient stepping-off point for the next issue. They are, in fact, a clear example of what we think is one of the greatest weaknesses of efforts to date BH texts on a linguistic basis. Here we see a prime and essential presupposition of the chronological approach in action.
Two or three Persian words have been enough for some scholars to conclude that the whole books of Song of Songs or Qoheleth cannot date earlier than the Persian period. Let us put this into perspective. The Persian words represent two or three words in texts having 1250 and 2987 words respectively. Given that the Qumran copies of Song of Songs differ from the MT once every six words, or the Qumran Qoheleth once every eight, how confident can we be that the couple of Persian words derive, without change, from the "author" of these works?
Modern readers are accustomed to a book, once published, remaining in the same form. Ancient books, however, according to scholarly consensus, did not maintain a static form, but developed over time. Despite the extremely fragmentary nature of our textual evidence, with no texts dating earlier than about 250 BCE and most of our evidence much later, the Dead Sea Scrolls, the versions of the Septuagint and the Samaritan text of the Pentateuch among others, when placed alongside the traditional MT, provide us with abundant evidence of textual variety. Thus, only a very small percentage of Qumran biblical texts have a very close relationship with the MT. For the parallel text 2 Samuel 22//Psalm 18, David Clines shows that, on average, one in every two words has an attested variant. Leading text critics such as Emanuel Tov and Eugene Ulrich agree that the biblical texts evolved through the production of successive literary editions, as evidenced by the fact that most biblical books have an attested variant edition.
Thus the evidence in our possession indicates a high degree of fluidity of biblical texts in the BCE period. This fluidity is especially noticeable in regard to the language of the biblical texts. At the very least, individual linguistic elements came and went during scribal transmission. The text-critical evidence, therefore, puts a question mark over the whole enterprise of linguistic dating before it has begun, since linguistic dating could only work if the language of the current texts is very close, if not identical to the language of the "original author" of the text being dated. On the contrary, the text-critical evidence indicates that the current linguistic profiles of the biblical books are not only the result of choices made by their authors only but also by later scribes.
According to general scholarly consensus on historical-critical issues and according to the consensus of text-critical scholars, the books of the Hebrew Bible are in their present, final form postexilic, but with elements (generally considered to be substantial) that go back to the preexilic period. Thus we would not expect any BH work to represent purely preexilic Hebrew. The fact that the orthography of all known biblical manuscripts is typologically different and, apparently, later than that of the Hebrew inscriptions would seem to back up this expectation. Textual criticism shows us that after orthography, language was the second most commonly changed element of the texts. The EBH and LBH linguistic forms of biblical books are thus not only the result of choices made by their authors. Later scribes clearly could change the EBH/LBH orientation of a text during its transmission. The classic example is the book of Isaiah in its EBH MT form and its more LBH form in the Qumran 1QIsaa, but many cases exist of linguistically variant forms of the same book, passage, or individual linguistic item. 4QCantb or the Samaritan Pentateuch represent systematic linguistic differences to the MT forms of the books. Often the variant forms represent EBH linguistic forms in contrast to LBH features of the MT. One example is the presence in the MT Pentateuch of examples of the LBH min ("from") separate before a noun without the definite article, all of which are absent from the Samaritan Pentateuch. Note the contrasting tendency of the scribe who added LBH medinah ("province") to EBH 1 Kings 20.14–19 with the one who added EBH anoki ("I") to Ezekiel 36.28 against the LBH tendency of Ezekiel to otherwise exclusively use ani. We must thus see the conservative and non-conservative tendencies applying in different ways and degrees to each individual textual version of each biblical book.
We can now see that the question that needs to be asked is: what are we actually dating? If biblical books were written and rewritten over the generations in the BCE period, then the question of the "original date" when a biblical book was composed is revealed to be anachronistic and irrelevant. The book as a whole was composed over a long period of time. Beyond this there are various things we may try to date such as when do we think the core form of the book came into being or what is the date of the current form of the text we are studying? We can see that these are two separate questions and that we must be very cautious about using the features of the current texts to date a presumed original composition.
To conclude, what we refer to as EBH and LBH are two co-existing styles or tendencies utilized by authors and scribes throughout the biblical period. EBH refers to a tendency to conservatism in linguistic choices, only rarely using forms outside a narrow core of what they considered literary forms. LBH refers to a less conservative attitude, freely adopting a variety of linguistic forms in addition to those favored by EBH authors. Later, scribes could choose to modify the style of a book or passage toward the other style.
The linguistic variety of our biblical texts, therefore, represents the end result of the application of these two writing and editing styles to various editions of the biblical books.
This is not to deny linguistic change in ancient Hebrew. On the contrary, ancient Hebrew obviously did change over time. But the nature of the texts is such that they do not reflect this change in any straight forward way.
Several practical outcomes of the new model for biblical studies in general are clear. First, the many attempts to date biblical books and passages on the basis of linguistic evidence are all based on a misconception of the nature of BH, and hence do not have the evidential value sometimes claimed for them. Second, the new model is more easily compatible with mainstream scholarship in other fields of biblical studies, especially textual criticism.