Emanuel Tov and His Evolving Categories for the Biblical Dead Sea Scrolls

Emanuel Tov is the most well-known textual critic of the Hebrew Bible and for good reason. Under his leadership, thirty-three volumes of the authoritative series Discoveries in the Judaean Desert (DJD) were published in less than twenty years. Before his tenure, only seven volumes were published in nearly forty years. Despite this impressive feat, Tov is probably most well-known for his work Textual Criticism of the Hebrew Bible.

The book is now in its fourth edition, and it is the go-to work for those interested in textual criticism of the Hebrew Bible. One aspect of the fourth edition, among many, worth discussing is his categorization of the biblical Dead Sea Scrolls. A comparison of these categories across his four editions shows a certain evolution in how he views the text of the Hebrew Bible in the Second Temple period.

First Edition (pp 114-117)

QSP	Proto-MT	Pre-Sam	Close to Hebrew of LXX	Non-Aligned
20%	60%	5%	5%	10%

 

Second Edition (pp 114-117)

QSP	Proto-MT	Pre-Sam	Close to Hebrew of LXX	Non-Aligned
20%	35%	5%	5%	35%

 

Third Edition (108-110)

Torah

Proto-MT	Pre-Sam	Close to Hebrew of LXX	Non-Aligned
48%	11%	2%	39%

 

Rest of Scripture

Proto-MT	Close to Hebrew of LXX	Non-Aligned
44%	7%	49%

 

Total

Proto-MT	Pre-Sam	Close to Hebrew of LXX	Non-Aligned
45%	5%	5%	45%

 

Fourth Edition (pp 135-136)

Torah

Proto-MT	MT-SP Block	Pre-Sam	Close to Hebrew of LXX	LXX-SP Block	Non-Aligned
2%	28%	9%	2%	27%	32%

 

Rest of Scripture

MT-Like	MT-LXX Block	Close to Hebrew of LXX	Non-Aligned
22%	16%	5%	57%

 

Total

Proto and semi-MT	MT and LXX Block	MT-SP Block	Pre-Sam	Close to Hebrew of LXX	LXX-SP Block	Non-Aligned
13%	9%	13%	4%	4%	12%	45%

 

Observations
Observation #1: The category, Qumran Scribal Practice, was a group of texts united by a distinctive orthography and morphology. These texts, however, did not share a textual background, so Tov eliminated this category in his third edition (see 3rd ed., 110). The elimination of this category caused the non-aligned and proto-MT categories to increase by 10% each.

Observation #2: The statistics of the proto-MT category and the non-aligned category have changed most drastically. Tov has not been completely clear on the qualifications for the categorization of these manuscripts. For example, he describes a non-aligned text as one that is “not exclusively close to MT, LXX, or SP.” They are inconsistent in aligning with MT, SP, and LXX while preserving unique readings (3rd ed., 109). This description of the non-aligned category lacks precision. At what point does a text morph from being proto-MT or MT-like to being non-aligned?

Despite this lack of precision, it is clear that Tov’s proto-MT, and even his semi-MT, categories are restrictive categories. For a text to fit into these categories, a text must align closely with the MT even regarding its orthography. If a text deviates orthographically from the MT, it most likely becomes non-aligned as in the case of 1QIsaa. The opposite is true with the non-aligned category. It is by definition, extremely broad since these manuscripts are joined together not by a common denominator of shared readings but by the simple fact that they disagree with the other texts (MT, SP, and LXX). Over time, Tov’s proto-MT category has become more restrictive while the non-aligned category has become broader. This detail has led to a dramatic decrease in those texts labeled proto-MT texts and a dramatic increase in those texts labeled non-aligned.

Observation #3: A superficial survey of the above statistics shows that the fourth edition has more categories than the first three. The reason for this is that in earlier versions, texts that were equally close to the MT and SP were categorized as MT. Similarly, texts equally close to MT and LXX were understood as proto-MT (3rd ed., 108). The MT, however, in the fourth edition, is no longer the default text in the categorization of the biblical Dead Sea Scrolls. If a text agrees with the MT and the SP, it is labeled MT-SP. If a text agrees with the MT and LXX, it is labeled MT-LXX. Moreover, Tov now differentiates between proto-MT and semi-MT texts. Proto-MT texts are more closely aligned with the MT than semi-MT texts. Overall, the addition of more categories has led to a further decrease in those texts labeled proto-MT.

Observation #4: Since 1992, more manuscripts are included in Tov’s statistics. For example, Tov understood 4QRPc-e (Reworked Pentateuch) to be non-biblical in the first two editions, but beginning in the third edition, these manuscripts are categorized as non-aligned. The statistics in the fourth edition also include seventeen tefillin that do not seem to have been included in the statistics for earlier editions. Although tefillin are technically non-biblical texts (they are excerpted texts), Tov now includes them in his categorization grid. These details, likewise, have shifted the statistics away from the proto-MT and semi-MT categories.

Tracing the Evolution
Overall, the evolution of the biblical Dead Sea Scrolls in Tov’s categorization grid moves away from proto-MT. The steps contributing to this movement included the increase in categories, the increase in texts now included in the statistics, and the understanding that the MT is no longer the default text in his categories. The elimination of the QSP category led to an increase in the proto-MT category in the third edition, but this type of change has been the anomaly. The evolution of Tov's categories is rather straightforward: for Tov, fewer texts are proto-MT, and more texts are now non-aligned. 

4QPsx: A Poorly Copied Manuscript

Several details about the Dead Sea Scrolls are common knowledge. One of these details is that these manuscripts preserve a certain level of textual plurality previously unknown among Hebrew OT manuscripts. Although this detail is a fact, the nature of this textual plurality is mostly unknown by laypeople and scholars alike. We should be aware that common explanations explain most of this textual plurality. One of these explanations is poor copying. (In a previous blog, I discussed that some of this diversity results from scribes normalizing their biblical texts [such as 4QGenk]).

4QPsx is one manuscript, among many, that was copied poorly. Interestingly, some scholars date this manuscript, which preserves portions of Psalm 89, to 175-125 BC. If this is right, this manuscript is our earliest manuscript available that preserves a psalm; yet, it is an unreliable guide to the Psalter's state, and Psalm 89 in particular, during the second century BC. Several details about this manuscript suggest that this scribe was either an unskilled or a beginner scribe. Varying letter size, inconsistent space between lines, curved lines, cancelation dots to erase a mistake, inconsistent use of final letters, and the inability to space words properly are just a few of these details. (See Skehan’s article “Gleanings” for these and other points). 

4QPsx: https://www.deadseascrolls.org.il/explore-the-archive/image/B-473784

Besides these scribal features, we should consider several other material features of this manuscript that call into question its dependability. This manuscript is unruled, and this reality contributes to some of the mistakes listed above. Indeed, the absence of a writing block made it challenging to space the words, and the lack of horizontal lines caused the scribe to write upwards and downwards at times. The fact that this manuscript is unruled is puzzling since most manuscripts at Qumran were (see Tov, Scribal Practices, 57, 104). More puzzling is that this manuscript was stitched before inscribing these words, which, again, is unusual (Scribal Practices, 34-37). This fact is evident since the scribe wrote around the stitching (“Gleanings,” 441). Thus, these two details suggest that this sheet of leather was previously a handle sheet (a protective piece of leather stitched unto inscribed sheets at a manuscript's beginning and/or end).

The scribal habits and the material features of this manuscript cast serious doubt on the text's reliability. Although some might understand this text as an alternative form of Psalm 89, I suggest that it is a recycled handle sheet (like Skehan): perhaps a scribe's school exercise. Overall, this text contributes to the textual plurality preserved at Qumran but provides no evidence to suggest that the OT existed in a state of fluidity during this time (fluidity meaning that the text had not yet reached its final form). Instead, this manuscript suggests that some manuscripts were copied poorly and, perhaps, some manuscripts are mere school exercises. As already noted, this manuscript may be our earliest manuscript preserving a psalm. This reality reminds us of a vital point of textual criticism: the earlier manuscript is not always better.

*For more information about this manuscript, see my dissertation, A Comparison of the Non-Aligned Texts of Qumran to the Masoretic Text. It can be accessed on ProQuest. You may also view the presentation that I gave at the Sacred Words Conference, where I discuss this manuscript in more detail. 

4QGenk: A Normalized Manuscript

4QGen^k (4Q10) is a poorly preserved manuscript that contains 70 partial words. Among these partial words, there exist three variants (I am not counting differences of plene/defective spelling here). Each of these variants all function to replace an uncommon grammatical form of the MT with the more common form. This common denominator suggests to me that scribe of 4QGen^k has taken liberties with his exemplar, or perhaps the scribe has copied from a manuscript that has taken these liberties, to normalize the grammar of the MT. 

Variants between MT and 4QGenk
Fragment, Line, and Verse	4QGenk		MT
F1:L1 (Gen 1:9)	ותרא	And let it appear	וְתֵרָאֶה	And let it appear
F2:L3 (Gen 1:14)	ולש֯[נים.  ]	and fory[ears]	וְשָׁנִים	and years
F5:L2 (Gen 3:1)	האף֯	Did…really	אַף	Really…

The difference persevered at Genesis 1:14 concerns a preposition, but the difference is minor since the scribe is simply making the grammar of the MT explicit. In the MT, one preposition governs two nouns – “days and years.” Although a preposition can govern more than one noun in a sequence in biblical Hebrew (the reading of the MT), the more typical form is to repeat the preposition after each noun (see GKC §119hh). The scribe has replaced the uncommon form of the MT with the more common form.

The same explanation applies to the addition of the interrogative heh in 4QGen^k at F5:L2 (Gen 3:1). Interrogative statements can be expressed in Hebrew by an added particle (4QGen^k) or by intonation (MT [See GKC, §150a]). Although both forms are possible, interrogative statements are most often marked with an interrogative particle (see GKC, §150c). 

Finally, the fragmentary nature of this manuscript complicates a certain understanding of the last difference. What is certain at F1:L1 (Gen 1:9) is that 4QGen^k reads ותרא while Leningrad reads וְתֵרָאֶה. 4QGen^k has omitted the final heh of the verb ראה. The form of the MT is a jussive (a tense of volition), but the form is uncommon. The typical jussive of a third yod/vav is an apocopated form: that is, the final heh is lost. It is not surprising given the other two variants in this manuscript that the scribe of 4QGen^k provides this form: the more common form.

Tov labels this manuscript as non-aligned, and this categorization may give the impression that this manuscript provides positive evidence that the OT text existed in a state of diversity without any unity. This impression would be wrong. Rather, this manuscript suggests, in my mind, the presence of a stable text existing alongside a diversity of texts. Moreover, this manuscript demonstrates that some of the textual diversity preserved in the non-align category is the result of scribes normalizing uncommon forms. This tendency does not suggest the absence of an authoritative text in the Second Temple period; rather it supports its existence. 

*For more information about this manuscript, see my dissertation, A Comparison of the Non-Aligned Texts of Qumran to the Masoretic Text. It can be accessed on ProQuest. You may also view the presentation that I gave at the Sacred Words Conference.

Is There a Unity amid this Diversity?

Currently, there is debate about the state of the Old Testament text before the second century AD. While many argue that the OT text existed in a state of fluidity – that is, the OT text had not yet reached its final form – some others believe that a stable text existed alongside a diversity of texts. I hold to this latter view, and I identify this stable text as an MT-like text.

Now, Old Testament textual critics cite a variety of evidence when discussing this important topic. Perhaps the strongest evidence of this debate is the Hebrew/Aramaic biblical manuscripts discovered in the Judean Desert. At least two reasons make these manuscripts unique among the other evidence: 1) they are the oldest biblical texts available to scholars, and 2) they are written in the OT’s original languages. These two details compel us, then, to take seriously two further details: nearly half of these manuscripts align closely with the MT while the other half do not. Emanuel Tov labels these latter manuscripts non-aligned texts.

First four columns of 1QIsa^a
http://dss.collections.imj.org.il/isaiah

Thus, in a series of blogs, I am going to discuss the non-aligned manuscripts. I hope to show that these manuscripts are largely secondary and dependent on an MT-like text. This analysis suggests to me that the stable text that existed alongside the diversity of the non-aligned texts is an MT-like text. I hope this creates some intriguing dialogue for the glory of the great God whom these manuscripts bear witness!

Anthony Ferguson on Texts Preserving Psalms from Qumran

Last week, guest blogger, Anthony Ferguson, described his recently defended and completed dissertation on the ‘Non-Aligned’ Texts of Qumran and reported on its major conclusions. I asked Anthony to write another post on the texts preserving Psalms at Qumran, and he kindly obliged. Thank you, Anthony, for your labors, and we look forward to reading more of your work in these areas in the future.

---

11QPs-a (XX-XXIV; Image from Leon Levy Library)

The texts preserving Psalms from Qumran classified by scholars as biblical texts are significant for the fluid/standard text debate because they preserve large-scale differences that designate them in the mind of many scholars as an alternative tradition or edition of the Psalter (e.g., Sanders, Ulrich, and Flint). Contrary to these scholars, many others understand these texts to be secondary to the MT, and not strictly biblical texts at all but liturgical texts (e.g., Talmon, Gottstein, Skehan, and Tov). Despite Tov viewing many of these texts as non-biblical, he still labels them non-aligned. I understand these texts to be based on a Masoretic like text and secondary to the Masoretic Psalter. Thus, I label many of these texts as texts belonging to the Masoretic tradition that contain liturgical alterations.

Types of Differences found among the Texts Preserving Psalms

11Q5 (11QPs-a) generally represents the types of differences preserved in the texts preserving Psalms from Qumran. The major differences preserved in 11Q5 include a different sequence of Psalms, the inclusion of Psalms not found in the Masoretic Psalter, and large-scale additions.

Sanders, the editor of 11Q5 in DJD, argued that 11Q5 represented a genuine Psalter. Sanders’ reasons included the following: his belief that David was the author of the Psalter, the nature of the non-Masoretic Psalms, the presence of large-scale additions such as superscriptions and interjections, and the fact that 11Q6 (11QPs-b) appears to be another copy of the text preserved in 11Q5 (whether 4Q87 represents this same text is debated).

11Q5 as secondary and dependent on the Masoretic Psalter

Contrary to Sanders, I argued that 11Q5 is secondary to and dependent on the Masoretic Psalter. First, I believe that the major differences that distinguish 11Q5 from the MT can reasonably be explained as liturgical adjustments. For example, the different sequence of Psalms can be attributed to liturgical traditions that recognize a stable Masoretic Psalter (see m. Tamid 7:4). Moreover, the inclusion of non-canonical psalms is a phenomenon found in texts that preserve the Masoretic Psalter. The LXX superscription to Psalm 151 demarcates this text as non-canonical, and thus, shows that ancient Jews did not have a fundamental problem with combining canonical and non-canonical psalms together in the same scroll. Last, one should note that there are analogies in later Jewish liturgical texts to many of the large-scale additions preserved in 11Q5. Space permits only a few examples. The liturgical function of the Hallel Psalms (Pss 113-118) during the feast of Booths led to additions in the liturgy. For example, if a slave, woman, or minor answers the reader of the Psalms, they were to repeat what was said. If an adult male answered the reader, he was only obligated to respond halleluyah (m. Sukk. 3:10). Local customs then permit further types of repetition. Moreover, as the Levites walked around the altar, it was known that they would repeat portions of Psalm 118:25: אָנָא ייי הוֹשִׁיעָה נָּא אָנָּא והוֹשִׁיעָה נָּא, “Save now, we beseech thee, O Lord! We beseech thee, O Lord” (m. Sukk. 4:5). Thus, there appears some precedent for dissecting and reworking Masoretic Psalms for liturgical purposes in liturgical traditions that clearly recognized a stable Masoretic Psalter. Overall, it seems that the major differences that distinguish 11Q5 from the MT are liturgically motivated.

Anthony Ferguson on the ‘Non-Aligned’ Dead Sea Scrolls

Last month, I noted the completion of Anthony Ferguson’s dissertation, “A Comparison of the Non-Aligned Texts of Qumran to the Masoretic Text.” In this guest post, Anthony describes his thesis in more detail, and I hope it stirs up some discussion on this crucial topic. Anthony will write one more post on his treatment of texts preserving Psalms, a most significant datum for the entire discussion so be on the look out for it.

---

A preliminary reading of Emanuel Tov’s classification of the biblical texts from Qumran gives the impression of textual fluidity since he labels roughly 35% of the biblical texts from Qumran as non-aligned or independent. Michael Law [p. 79 here] among other scholars use Qumran to argue for a fluid text. He says:

We have seen repeatedly that the Septuagint and especially the Dead Sea Scrolls offer proof that the Hebrew Bible was not fixed before the second century CE and, perhaps more surprisingly, that many readers and users of scriptural texts before then were not bothered about it.

The Qumran texts Law refers to appear to be the non-aligned texts since this category represents the most diverse Hebrew/Aramaic manuscripts of the OT from Qumran. However, Law seems to have misunderstood the nature of the non-aligned category. Not all non-aligned texts are equally non-aligned in Tov’s mind. This fact may not be immediately apparent when one first learns of Tov’s classification grid (Tov labels some texts as proto-MT, LXX, proto-S(amaritan) P(entateuch) while others are categorized as non-aligned.); yet, a brief survey of appendix 8 of Tov’s monumental work Scribal Practices and Approaches Reflected in the Texts found in the Judean Desert confirms it (cf. pp. 332–35 where he categorizes 57 texts [38.77 percent of the biblical-DSS] as non-aligned. A non-aligned text is defined by Tov as a text that is inconsistent in its agreement with the MT, LXX, and SP while preserving unique readings. Cf. also p. 98 in Time to Prepare the Way in the Wilderness: Papers on the Qumran Scrolls). In Scribal Practices, some non-aligned texts are noted with a question mark (?), others with an exclamation mark (!) while others lack notation. As Tov notes, the non-aligned category contains a range of texts that are more or less aligned to the previously known witnesses. Therefore, it seemed pertinent to understand how aligned each of these texts were to the MT in order to illuminate the nature of the category. My thesis offers a major correction to the comments of Law.

The thesis of my dissertation was as follows:

Contrary to Emanuel Tov’s analysis that fifty-five texts from Qumran are exclusively identified as textually non-aligned, a more cautious analysis of each text demonstrates that once the few ambiguous texts are excluded from the category, the remaining texts can reasonably be explained as belonging to the Masoretic tradition.

One should note that I am not arguing that all of the non-aligned texts should be classified as proto-MT texts per Tov’s classification grid; that is, I am not arguing that they are close to Leningrad to the same degree that some other biblical-DSS are (e.g., 1QIsa-b), to the degree that the other Judean Desert texts are (e.g. MasLev-b), or to the degree that the Medieval Manuscripts are. Rather, I say that these texts belong to the Masoretic tradition. Classifying texts according to textual tradition, not simply as texts, is a significant methodological difference that distinguishes my approach from Tov’s. Bruno Chiesa (on p. 266 here) has already insightfully critiqued Tov for classifying texts as mere texts. In short, this classification is too specific to be helpful for the fluid/standard text debate. I agree with Chiesa.

The Non-aligned Category: A Diversity of Texts

I discovered that the non-aligned category is made up of a range of texts that align more or less with the MT. A few examples illustrate this conclusion, and thus, show that this category as a whole cannot be used as evidence of a fluid text. First, 1QIsaa is non-aligned, according to Tov, in the least “meaningful type of deviations, namely in orthography” [see p. 303 here]. An alternative orthographic profile, in my mind, is not germane to the fluid/standard text debate. This criterion should thus be set aside (a sample of Tov’s graphic representation of this text can be viewed online [PDF]). Second, 4QEzeka (4Q73) is labeled non-align by Tov and yet Sanderson (the editor of the text in DJD) notes that the orthography and text of 4Q73 is close to the MT. The text, according to my calculations, sufficiently preserves 214 words (מלכי “my king” being two words), and yet, the orthography only deviates from the MT in four places (DJD, 210) while the text only preserves three certain textual variants when compared to the MT: all are very minor. In the end, this text is actually very close to the MT even regarding orthography. Grounding a theory about the fluidity of the OT text based on a category that includes 4Q73 is untenable. Third, unlike the first two texts that differ from the MT in minor details, 4QLama (4Q111) is one of the non-aligned texts that aligns statistical least with the MT. Moreover, it contains one difference that lacks a typical scribal explanation; yet, with Cross, I argue that a deeper analysis of this text indicates that it is not far removed from the Masoretic text. Even in this instance a general relationship can be traced.

Conclusion

This analysis demonstrates how misleading the non-aligned category is if one hopes to use this category to substantiate the idea that the text of the OT was fluid during the Second Temple period. Some texts are actually very close to Leningrad (e.g., 4Q73) while most texts are close enough to warrant the conclusion of belonging to the Masoretic tradition. At this point, let’s revisit the comments of Law [p. 79 here]. He says:

We have seen repeatedly that the Septuagint and especially the Dead Sea Scrolls offer proof that the Hebrew Bible was not fixed before the second century CE and, perhaps more surprisingly, that many readers and users of scriptural texts before then were not bothered about it.

What does Michael Law mean by a fluid text and which Qumran texts substantiate this claim? My dissertation reasonably demonstrates that the non-aligned texts cannot unequivocally be these texts unless the fluidity Law has in mind concerns expected scribal differences. The fact that these texts generally share a high statistical relationship with the MT and the fact that most of these variants can be attributed to common scribal tendencies (e.g., mechanical errors and interpretation) indicates to me that the majority of these texts can reasonably be understood as belonging to the Masoretic tradition. (The ambiguous texts were 4Q47, 4Q49, 4Q95, and 4Q98g. These exceptions do not prove the theory of a fluid text since these texts may not be biblical. For a discussion of these texts and their biblical/non-biblical status, see the discussion of each text in my dissertation.) This conclusion calls into questions Law’s comments if, of course, he has in mind the non-aligned texts. Regardless of Law’s comments, this conclusion reveals a reasonable unity amongst the diversity of these texts. That unity may be broader than Tov’s classification gird permits, but it is, nonetheless, the tradition of the Masoretic text.

---

Anthony Ferguson finished his Ph.D. at The Southern Baptist Theological Seminary under the supervision of Russell Fuller. He teaches at Gateway Seminary and California Baptist University. His publications include “The Elijah Forerunner Concept as an Authentic Jewish Expectation” in JBL vol 137 and two forthcoming book reviews to feature in SBJT and Presbyterion.

New Dissertation on the ‘Non-Aligned’ Dead Sea Scrolls

1QIsaa, col. XXXIII

Last month, Anthony Ferguson (recent PhD graduate from Southern Baptist Theological Seminary) successfully defended his doctoral thesis entitled “A Comparison of the Non-Aligned Texts of Qumran to the Masoretic Text.” Russel Fuller, Peter Gentry, and Terry Betts were on his committee, and Emanuel Tov served as the external examiner of the thesis.

Thesis and Object of Study

Ferguson defined a non-aligned text according to Tov’s and others’ definition:

[A]s [texts] exhibiting the greatest amount of diversity because they are inconsistent in their agreement with the MT, LXX, and Samaritan Pentateuch while preserving unique readings (p. 1).

In his dissertation, Ferguson analyzed the following fifty-five texts and current scholarship on them, which Tov identified as non-aligned: