Hebrew Dictionaries and the Field of Text Criticism
A printable version of Parts 1-3 of this demonstration is available here.
Text-critical research is on a new footing thanks to eGELS and databases like The Parallel Aligned Hebrew-Aramaic and Greek Texts of Jewish Scripture (hereafter: CATSS) and those of the Comprehensive Aramaic Lexicon (CAL). Their integration into the Libronix system makes them that much more useful. Commentary series with strength in the area of text criticism, such as ICC, WBC, JPSTC, and Hermeneia, are also available in Libronix format. A combination of background knowledge and intuition are irreplaceably important in the search for text critical discussions of interest relative to a particular study. At the same time, the value of keylinks across multiple resources is immense. A resource designed specifically for the work of text criticism merits development, but I would digress too much if I discussed the matter in this post.
The student of ancient Hebrew fools herself if she thinks that MT as it stands is a serviceable basis for serious study of the language or the texts it transmits. MT is a point of entry, not a final destination. The texts MT contains were copied many times before becoming a part of it. A goal of text criticism is the identification of inadvertent errors and conscious changes introduced into the text over the course of time.
A foray into the field of text criticism is an essential component of a serious word-study. In this section, sixteen אכל occurrences of text-critical interest are reviewed. In ten cases, MT is upheld. In six cases, MT is held to reflect a miniscule corruption involving a single or single cluster of syllables or letters. וְאַ became וַיַּ in Ps 81:17 via assimilation to the preceding line. וַאֲכָלֻם became וְאֹכְלֵם and כְּלָבִים became כְּלָבִיא in Hos 13:8 via assimilation to the preceding context. אַך לוֹ became אָכְלוּ וַ in the course of Ps 22:30’s transmission. יֵאָכֵל בְּדַו became יֹאכַל בַּדֵּי by assimilation with the following line in Job 18:13. By aural misapprehension, אוֹכִיל לוֹ becameאוכיל לא in Hos 11:4-5. וּלְאֹכֶל became וְלֶאֱכֹל via assimilation perhaps to the stress pattern and syllabics of וְלַאֲשֶׁר preceding in Gen 47:24. In two cases, arguments in favor of reading a form of אכל outweigh those opposed, but not decisively: וַאֲכַלֵּם/וְאֹכְלֵם in Exod 32:10 and אֲכֶלְךָ/אֹכַלְךָ in Exod 33:3.
To be sure, scribal errors have sometimes become the foundation for a rich history of interpretation. The same goes for conscious changes made at the micro and macro levels. They are retainable for specific purposes. But if an interpreter desires to know what the text looked like at a point in time prior to the intromission of said errors and changes, she will avail herself of text critical methods.
Text criticism has always been an aspect of the work of serious exegetes. Rashi, ibn Ezra, Kimchi, Luzzatto, and more recently, Driver, Greenberg, Craigie, and Fox, to cite a few prominent examples, sometimes propose a text at odds with MT, with or without the support of ancient witnesses. Of course, the proposals of those who seek to restore the received text to a more pristine state are themselves subject to evaluation.
Inclusion and summary evaluation of text-critical suggestions are traditional features of dictionaries of biblical Hebrew. eBDB is the most helpful in this regard, but is now outdated. DCH claims to abstain from evaluation, but the very fact that it omits mention of most past proposals is a form of evaluation. I discuss eHALOT’s text critical choices below. Sad to say, available dictionaries fail to describe the procedure followed whereby text-critical proposals were collected and evaluated and included or omitted. Glaring omissions of important past text-critical suggestions are evident in every one of them. Two examples may illustrate.
MT Ps 81:17 reads as follows:
וַֽ֭יַּאֲכִילֵהוּ מֵחֵ֣לֶב חִטָּ֑ה וּ֝מִצּ֗וּר דְּבַ֣שׁ אַשְׂבִּיעֶֽךָ
And he fed them with the fat of wheat,
and from the rock I sated you with honey
The line recalls Deut 32:13-14 and like Ps 81 as a whole, is best understood against the background of the traditions about Israel’s transit through the wilderness preserved elsewhere in ancient Hebrew literature. However, the past reference implied by MT is out of place in context. The change from third to first person of the speaking subject midway through the line, furthermore, is jarring. 81:17 is the conclusion of the entire psalm, and one expects it to relate to the lines which precede it. It seems best to understand v18 as the conclusion of a subunit that begins in v. 14, the protasis of a condition; the apodosis takes up all of vv. 15-17. In that case, v. 17 is best understood as a divine promise, and slightly emended: read וְאַאֲכִילֵהוּ for וַיַּאֲכִילֵהוּ. [fn  Cf. 81:11. The translation then is: “And I will feed him with the fat of wheat, // from the rock I will sate you with honey.” The enallage of object is retainable; it occurs often in ancient Hebrew literature. One might have expected BDB, HALOT, and DCH to note the emendation, without giving it credence if such was the thought, but they do not.
MT Hos 13:8b reads as follows:
וְאֹכְלֵ֥ם שָׁם֙ כְּלָבִ֔יא חַיַּ֥ת הַשָּׂדֶ֖ה תְּבַקְּעֵֽם
There like a lioness I will eat them,
a wild animal will rip them to pieces.
The conclusion to a complex simile that begins in 13:7, the difficulties of MT 13:8b are not evident in a number of translations, which render the last stich as if it read כְּחַיַּ֥ת הַשָּׂדֶ֖ה אֲבַקְּעֵֽם (e.g., REB: ‘like a wild beast I shall rip them up’). OG Hos 13:8 read ואכלם as וַאֲכָלֻם ‘and they will eat them,’ the more difficult reading, because it presupposes a change of subject. וְאֹכְלֵם כְּלָבִיא ‘I will eat them like a lioness’ in MT follows easily after אֶפְגְּשֵׁם כְּדֹב ‘I will encounter them like a bear’ of the preceding line, but is not necessarily correct. If one emends כְּלָבִיא to כְּלָבִים, a slight change, the concluding line of 13:7-8 ends strongly, without the inconcinnities of MT. The translation then is: ‘there the dogs will eat them, // a wild animal rip them to pieces.’ Duhm, Rudolph, BHS, Mays, NJPSV, and Borbone agree on the proposed text. eBDB, eHALOT, and DCH fail to mention the proposal’s existence.
Proposed emendations like the above warrant consideration. It is true, of course, that most past attempts at restoring supposedly corrupt passages in MT to their original form belong in the trash heap of history. But that is not the same thing as saying that they all do.
eHALOT in my view proposes to alter MT more often than is called for, as scholars have been wont to do from time to time. The opposite extreme is now common, whereby alteration of MT is foresworn on principle.
Psalm 22:30 and Job 18:13
eHALOT lists ten cases of אכל as reported in MT “to be read as” something else. eHALOT is right, it seems to me, in two instances: Ps 22:30a, where we should read, with BHS, Craigie, NRSV, and NAB: אַך לוֹ יִשְׁתַּחֲוּוּ כָּל־יְשֵׁנֵי־אֶרֶץ ‘to him shall all who sleep in the earth bow down’ in place of אָכְלוּ וַיִּשְׁתַּחֲוּוּ כָּל־דִּשְׁנֵי־אֶרֶץ ‘all the fat ones of the earth shall eat and prostrate themselves’; and Job 18:13, where we should read (similarly, Driver, Dhorme, NRSV, and Clines; for דַּו*, cf. גַּו from √גוי; and the Canaanite noun dw in the phrase y’kl dw in KTU 4.767=TT 433, if that is the correct reading) יֵאָכֵל בְּדַו עוֹרֹו ‘his skin is eaten away by disease’ in place of יֹאכַל בַּדֵּי עוֹרֹו ‘he eats the tendons of his skin.’ A compelling defense of MT אכל is offered by Driver pro 1 Sam 1:18; Greenberg pro Ezek 33:27, Allen pro Ps 105:35; Clines pro Job 34:3; Fox pro Qoh 5:16; and Tov (Septuagint, 142) pro 2 Chr 30:22. In Deut 32:13, the reading sustained by eHALOT assimilates to context and is probably derivative. In Prov 31:27, the proposed emendation is odd. It has failed to attract support and does not deserve mention in a work of reference.
It is a drawback that lexica fail to register construals alternative to their own in a consistent fashion. The case of אוֹכִיל in Hos 11:4 may illustrate.
The form is unique and occurs in a difficult passage. As Macintosh points out, אוֹכִיל was understood as a noun by Aquila, Symmachus, and Theodotion, and by Ibn Ezra and Kimchi. OG Hosea has δυνήσομαι αὐτῷ ‘I will prevail with him.’ Whetherאוכיל לא (MT) or אוכל לו stood in its Vorlage is not known, but the translation reads as if it were the latter (so eGELS, example 2.c. in the Introduction, II. B). On this construal, אוכ(י)ל is a verb derived from יכל. Targum Jonathan, Peshitta, and Jerome, on the other hand, derive אוכיל from אכל. With BHS, eHALOT, NJB, REB, and NRSV, the text to prefer is probably לוֹ אוֹכִיל, in which אוֹכִיל is taken as equivalent to אַאֲכִיל. A truly helpful lexicon in a case like this would note and reference alternative construals. In the case at hand, at a minimum that would mean noting that α σ θ construe אוכל as a noun.
eBDB, eHALOT, and DCH do well to alert the reader to difficult and possibly corrupt passages which have exercised the ingenuity of scholars in the past. The signaling of cases in which MT attests to one thing and one or more versions attest to another is especially welcome. eHALOT does so in the case of Qoh 5:16, but not in the case of Deut 32:13; 1 Sam 1:18; Ezek 33:27; or 2 Chr 30:22. E.g., OG Ezek 33:27 translates as if לאכלה was in its Vorlage, the reading eHALOT prefers. The fact deserves note.
In an ideal world, every instance of lexicographical significance in which an ancient version appears to have worked from a base text at odds with MT, or interpreted the same consonantal skeleton in a different way, would be noted in a reference lexicon of biblical Hebrew. eGELS is a step in the right direction. Its relevant notes, however, are incomplete and not always reliable. E.g., in OG Mal 3:11, באכל is translated εἰς βρῶσιν ‘for food.’ The translator read בְּאֹכֶל. A note to this effect is lacking in eGELS. In OG Ps 100 [=MT 101]:5, אתו לא אוכל is translated τούτῳ οὐv συνήσθιον ‘with him I would not eat’ (translation following NETS). The translator read אוֹכַל (spelled the same way in MT Ps 50:13), not אוֹכֵל as eGELS has it. This last case deserves note. It is not surprising that Mowinckel (91) and NEB follow OG and Peshitta’s lead in Ps 101:5 and read אוֹכַל against MT אוּכַל. ‘With him I will not eat’ fits the larger context (vv. 2, 6-7). But MT’s ‘him I will not endure’ is stronger and fits the immediate context better (vv. 4-5).
וְאַרְבַּ֣ע הַיָּדֹ֡ת יִהְיֶ֣ה לָכֶם֩ לְזֶ֨רַע הַשָּׂדֶ֧ה וּֽלְאָכְלְכֶ֛ם
וְלַאֲשֶׁ֥ר בְּבָתֵּיכֶ֖ם וּלְאֹ֥כֶ֯ל לְטַפְּכֶֽם׃
mt וְלֶאֱכֹל via assim perhaps to the stress pattern and syllabics of וְלַאֲשֶׁר
‘and the [remaining] four parts will serve as field seed for you and as your food and for those in your house, and as food for your children.’
WHM 4.2 and AFAT mistakenly parse לְאָכְלְכֶם as a suffixed infinitive construct. eHALOT, DCH, and WIVU parse it correctly as a suffixed noun. DCH notes a proposed emendation of לְאָכְלְכֶם to לָכֶם לֶאֱכֹל. Vocalizing ולאכל as וּלְאֹכֶל alleviates the text’s difficulties in less invasive fashion.
The clause contains a string of ל-introduced items in coordination. היה sq. ל pred. + ל. pers. structures occur often in biblical Hebrew, in which the predicates consists of nouns and/or noun phrases (see eBDB היה II. 2 f). It is natural to construe accordingly here, with both לאכלְכֶם and לאכל read as nouns rather than infinitive constructs. היה sq. Inf. c. ל structures are also well-attested, but not + ל. pers. (see eBDB היה III. 5 b). MT לֶאֱכֹל may have arisen via assimilation to preceding לאכלְכֶם understood as an infinitive construct (*לַאֲכָלְכֶם; cf. Targum Onkelos (וּלמֵיכַלכֹון) [against CAL; but note CAL's
It is rare that MT’s vocalization is less probable than a conceivable alternative. But MT לֶאֱכֹל in this verse counts as an example.
Exodus 32:10 and 33:3
According to a note in A Dictionary of Biblical Hebrew (DBH), a masoretic note associates with the “East” the reading פֶּן־אֹכַלְךָ in Exod 33:3. There is something to be said for this vocalization, and for reading וְאֹכְלֵםin place of וַאֲכַלֵּם in Exod 32:10. In both cases, Yahweh is the subject, who is elsewhere spoken of as a devouring fire (Exod 24:17; Deut 4:24; Isa 33:14; etc.). His anger is said to be about to חרה ‘burn’ against his people. For Yahweh then to say that ‘that I may devour them’ and ‘lest I devour you’ would be vivid and harsh. The harsh word, according to Exod 33:4, is precisely what got the people’s attention. To be sure, the verb כלה is securely attested in 33:5 and is elsewhere used with God as subject in equivalent contexts (e.g., Num 16:21; 17:10; 25:11; Josh 24:20; Jer 14:12; Ezek 22:31; Job 9:22). But on two occasions the verbs אכל and כלה co-occur (Jer 10:25; Hos 11:6). It is possible that Exod 33:3-5 also contains a co-occurrence. It is easier to imagine original וְאֹכְלֵם and אֹכַלְךָ being assimilated to the more common וַאֲכַלֵּם and אֲכֶלְךָ in Exod 32:10 and 33:3, respectively, than vice versa. The chief obstacle to this line of reasoning is the weak to non-existent attestation of וְאֹכְלֵם and אֹכַלְךָ in ancient witnesses. On the other hand, that may only go to show that the assimilation was ancient and almost universally received.
In the next post, eight passages which according to some but not all exegetes contain the verb אכל are discussed.
 An error of audition, with assimilation to the third person of the preceding line. The error is probably ancient. The versions, in my view, do not attest to the suggested reading, though S and T appear to in part. As translators tend to do, S and T render difficult texts more approximately than easier ones, based on the logic of the whole as they understood it. For a defense of MT, with misgivings, see Delitzsch. The first to propose the emendation seems to have been Wellhausen. It was adopted by Gunkel, Kraus, BHS, Alonso Schökel, and Hossfeld and Zenger; RSV, NASB, NRSV, NAB, and NJB. NIV and ESV silently emend וַ to וְ before the verb. I propose a ban on silent emendations in another post.