Monday, March 04, 2019

Deuteronomy 33:2 in Textual and Linguistic Perspectives

In my last post, I introduced the book How Old Is the Hebrew Bible? Essentially, the book argues that diachronic/historical linguistics explains the changes we see in the language of the Hebrew Bible from Classical Biblical Hebrew to Late Biblical Hebrew. The authors (Hendel and Joosten) included a chapter on Textual History and Linguistic History in which they show how the two disciplines work together. In particular, a text’s history can show scribal mistakes, modernizations, and textual growth, and diachronic analysis may explain some of these changes.

At the beginning of the chapter, they provide the “famous example” from Deut 33:2, with which I want to interact in this post:

Ketiv: ֹמִימִינוֹ אשׁדת לָמו “From his right hand ‘sdt for them”
Qere: ֹמִֽימִינוֹ אֵשׁ דָּת לָמו “From his right hand fire was a law/there was a fiery law for them”

The ketiv אשׁדת is difficult to understand; perhaps its meaning was lost. The qere reveals a different word division, but דָּת (“law”) raises a question for the historical linguist. The word is only otherwise attested in LBH (e.g. Esther and Ezra) and this would be the only case of a Persian loanword in the Pentateuch. But the authors conclude that the linguistic difficulty is superficial, for the text is problematic. The MT itself indicates that דָּת is only one possibility, since it is only the qere, the way the text is read. The ketiv is אשׁדת, a reading admittedly “that is hard to understand” (in fact the lexicon is still puzzled over its meaning). The authors then cite the Septuagint (ἐκ δεξιῶν αὐτοῦ ἄγγελοι μετ᾽ αὐτοῦ “at his right hand angels with him”) and the Peshitta (ܡܢ ܝܡܝܢܗ܂ ܝܗܒ ܠܗܘܢ “from his right, he gave to them”), “neither of which seems to reflect the qere“ (p. 48). The authors then conclude:
The qere may reflect a late midrashic interpretation of an earlier text that was at some point no longer understood. The expression אֵשׁ דָּת (fire of the law) does not represent the earliest text of the verse, and as such does not provide a solid basis for historical linguistics (p. 48).
Now, the authors may well be right in their conclusion and they may have a ready defense for their thesis in this instance, but I suggest they have too simplistically solved this problem. The fact is the qere does have early support in the Jewish revisers:
Aquila: ἀπὸ δεξιᾶς αὐτοῦ πῦρ δόγμα αὐτοῖς (“from his right, fire was an ordinance for them”)
Symmachus: …πυρινὸς νόμος… (“...fiery law...”)
Jerome’s Vulgate also reflects this reading (in dextera eius ignea lex) but probably depends on Symmachus for it. Both Jewish revisers are earlier sources than the Peshitta, and their reading of the text reflects an earlier tradition. I’m not saying Aquila, Symmachus, and the qere preserve the more original text. Rather, Aquila and Symmachus show the antiquity of the qere אֵשׁ דָּת which may or may not be the original text. “The late midrashic interpretation” of the authors would now have to be pushed back to the period of some of our earliest evidence for the text.

Here is the tension: is textual criticism guiding linguistics or linguistics guiding textual criticism? What is intriguing here according to the authors is that the qere reflects a later midrash, and part of the leverage for this conclusion is that דָּת would be the only Persian loanward in the Pentateuch, a very important datum, to be sure. We have a good example here of both disciplines attempting to inform one another. The problem now is that there’s more evidence for the antiquity of the qere than the authors presented, and therefore, the TC question has to be reopened.

The ketiv אשׁדת is unintelligible at present and the error of word division could go in either direction, which leaves room for the qere to be the more original text (“fiery law” could be an original, poetic description for the theophanic giving of the law on the mountain; Exod 20:18; Deut 4:15). If the ketiv is the more original, then we avoid the diachronic problem as the authors have framed the matter. If דָּת of the qere is the more original reading, then we have an isolated instance of a word from a later period in an otherwise CBH text. This isolated instance would not date the text of Deuteronomy to a later period because one instance does not satisfy the “Criterion of Accumulation.” Rather, one late word in an otherwise early text might simply be chalked up to a later scribe’s modernization. The qere could represent early updating of the text or the more original text (added late? or דָּת is an older word than we thought?). The ketiv is a rather difficult, if not impossible (corrupted?), text. Perhaps, the LXX and Peshitta attempted to render a corrupted text, which then gives appearance of a difficult one?

We won’t solve the matter today, but I found this example to be interesting for the authors’ case of diachrony because the text history remains unclear. In any case, Hendel and Joosten provide a fascinating and helpful contribution to the ongoing dialogue on the history and formation of the Hebrew Bible.


  1. John, you say, "the error of word division could go in either direction."

    It *could*. But the question is which direction is more likely. And if the qere were primary (though, per both your position and Hendel and Joosten, still not the original reading of Deuteronomy) and the kethiv were secondary, why would the Masoretes preserve the unintelligible secondary reading as the kethiv and the relegate the intelligible primary reading to the qere?

    It seems to me that a scribal conservatism that was so stubborn that scribes would preserve what they found in their exemplars, even if unintelligible to them, rather than make it intelligible by adding a space between two letters is the natural explanation for this.

    The fact that the qere reading is, as you seem to agree, an LBH corruption in an otherwise CBH text, and thus a secondary reading, gives additional support to the conclusion that the primary reading is the attested unintelligible one we have in the kethiv, as opposed to an unattested intelligible one.

    I agree that Aquila should be mentioned in the discussion, and that his version gives us an earlier terminus ante quem for the qere than we would otherwise have. But I don't think this is enough to alter the basic argument. And the Old Greek that Hendel and Joosten mention is a good counterpoint to this datum, since, whatever its Vorlage was here, it was probably something other than the qere, and at least could have been the kethiv.

    1. Eric, I don't have time for a protracted conversation today, though I wish I did. Three comments:

      (1) It's a word division error. It could be accidental or intentional.

      (2) The qere was used in a number of ways but a clear one is to preserve the ketiv no what the state of its text. In Deut 33:2, Q adds a space between two words. In Lamentations 4:3 (K כִּי עֵנִים) it takes the space away in the margin (כַּיְעֵנִים). So I don't expect them to tamper with it no matter its state.

      (3) LXX's "angels" could be an attempt at rendering the ketiv (clearly difficult, if not impossible) or an interpretation of the qere, for what does "law is fire" mean? The Deut translator could have rendered this phrase by making recourse to the text where the angles are spirits and his ministers are a burning fire (Ps 104:4), as clearly the Psalms translator did at a slightly later time.

      I'm generally in agreement with the book's presentation of the matter of diachrony as a whole. I'm not sure how the method works in the details like these ones. I could see one who believes the Hebrew Bible is entirely from the Persian period point to this instance as good evidence for that conclusion. That is, the qere is the original text and it can only come from this time due to דָּת. Thus, it would be the original text, either from the Persian period or the word appeared earlier than what's in the linguistic record.

      I'm not try to over problematize here. It just seemed to me that more needed to be addressed.

  2. The Qr reading is not young, yet it remains difficult to see the word דת in the Pentateuch on diachronic grounds (even one Persian word in that corpus is hard for me envisage), and there is at least one other reading (the Kt) which is not young either and which, being difficult, might explain the Qr (though the particulars of the case are certainly muddy). It is a difficult unit of variation, but I think we can say that one reading (or one conjecture?) that is unlikely to be original is the Qr. (This is assuming that the Pentateuch was not worked over by some from the Persian period or later; suffice it to say here I personally don't see a lot of evidence in favor of that view.)

    1. Right, Stephen. See my comments to Eric above. Thanks for weighing in here.

    2. If you have the time: How strong is the evidence for יהב in the Peshitta? CAL and BHQ seem to prefer a reading without it. I don't have access to a copy of the Leiden edition.

    3. Yeah. That's a good question. I don't have access to Leiden at the moment. I don't see the variant listed in CAL.

  3. This comment has been removed by the author.

  4. Formation of the Hebrew Bible? Deformation is more accurate. This is why our reality is deformed repeatedly in 3rd 4th Generations.

  5. Actually you could ask the question, @“What is the meaning of the redaction of Duet 33:02?” Firstly you need to narrow down the Septuagint to it sources with respect to the Four Main Uncials. Which of the Four have this passage, and those that do exactly what do they say? From there, assuming the passage is uniformly “at his right hand angels with him” you could formulate the open question, @“Does this imply the the overriding intent of the Redactionist ((the Liar)) is to send the Angels to Hell?” If so, what would that mean? When a lie is told, what by the lie told is revealed about the intent of the Liar?

  6. Dear John, thanks for interacting with us at such a high level of discussion. The reference to Aquila and Symmachus is nice: it shows that the qere was already known by the second century CE (which is still late in our account). Perhaps you are right that the word(s) you focused on could be a Persian-period addition to the text of Moses' blessing. What we tried to show in the book is that language history and textual criticism should interact more than they do. Any single example can be explained in various ways, but globally diachronic linguistics and textual criticism can often reinforce one another.

    1. Thanks for your reply, Jan. I do find the overall project very interesting and look forward to reading more.

      Though we can't know for certain, I do wonder whether Aquila reflects a reading tradition from the second temple period. In other words, I would find it difficult to accept that he invented the qere of this verse in the early part of the second century.

      Thanks for your interaction and your work.

    2. Fair enough. But if Deut 33 is archaic BH it may go back to the second millennium BCE, there may still be as much as 800 years between the composition of the blessing and the second-temple reading tradition witnessed in Aquila.

    3. Yes, true. And that date would make it quite possible for אשדת to be the original reading with a lost meaning. It's an interesting problem to be sure. Thanks for your comments.

  7. שדת ancient egyptian word means (fire, flame, burnt offering)