According to BDB and HALAT, the verb ערג is attested three times in classical Hebrew: once at Joel 1:20 and twice at Ps 42:2.

(David Stec, author of the volume פ–ס of the DCH, posits a fourth occurrence at Job 24:10. It is there attested, he says, in the present participle עֹרְגִים הִלְּכוּ, which is to be translated “they go about longing.” He notes, however, that עֹרְגִים appears only if the text is not emended to הִלְּכוּ עָרוֹם.‎¹‏ This is unaccountable, for עָרוֹם הִלְּכוּ is already the Masoretic reading. The basis of Stec’s statement is an article from 1974, whose author Arie de Wilde emended עָרוֹם הִלְּכוּ בְּלִי לְבוּש, “they go naked without clothes,” to עֹרְגִים הִלְּכוּ בְּלִי לְבוּשׁ, “though they weave, they go without clothes.”² De Wilde had apparently confused ערג with ארג, “to weave.” David Clines, the general editor of the DCH, later cited de Wilde’s article in his commentary on Job but silently corrected it, making the emended verse read “longing, they go without clothes.”³ Stec then reproduced Clines’ paraphrase in his entry for the DCH on ערג, but apparently mistook the emendation for the Masoretic text. The supposed fourth attestation of ערג is thus a garbled paraphrase of a charitable reworking of a garbled conjecture. It can be safely disregarded.)

In all three genuine attestations, the word takes the singular imperfect form תַּעֲרוֹג (spelt defectively in the latter verse). Here it is in Joel 1:20:

גַּם־בַּהֲמוֹת שָׂדֶה תַּעֲרוֹג אֵלֶיךָ כִּי יָבְשׁוּ אֲפִיקֵי מָיִם וְאֵשׁ אָכְלָה נְאוֹת הַמִּדְבָּר׃

And here in Ps 42:2:

כְּאַיָּל תַּעֲרֹג עַל־אֲפִיקֵי־מָיִם כֵּן נַפְשִׁי תַֽעֲרֹג אֵלֶיךָ אֱלֹהִים׃

The word presents two problems which have vexed commentators for centuries. The first is lexical: the meaning of the verb ערג is unestablished. Some evidence for its sense comes from the context of its attestations. Joel 1:20 is preceded by אֵלֶיךָ יְהוָה אֶקְרָא כִּי אֵשׁ אָֽכְלָה נְאוֹת מִדְבָּר וְלֶהָבָה לִהֲטָה כָּל־עֲצֵי הַשָּׂדֶה. The verse itself is then introduced by גַּם, which suggests that the two clauses אֵלֶיךָ יְהוָה אֶקְרָא and בַּהֲמוֹת שָׂדֶה תַּעֲרוֹג אֵלֶיךָ are to be taken as parallel instances of the same phenomenon. Ps 42:2, meanwhile, is followed by צָמְאָה נַפְשִׁי לֵאלֹהִים לְאֵל חָי, which suggests an idea of urgent desire.

The earliest attested translations of תערוג seem to be guesses informed by these contexts. Joel 1:20 was rendered in the LXX like this:

καὶ τὰ κτήνη τοῦ πεδίου ἀνέβλεψαν πρὸς σέ, ὅτι ἐξηράνθησαν ἀφέσεις ὑδάτων καὶ πῦρ κατέφαγεν τὰ ὡραῖα τῆς ἐρήµου.

And Ps 42:2 like this:

ὅν τρόπον ἐπιποθεῖ ἡ ἔλαφος ἐπὶ τὰς πηγὰς τῶν ὑδάτων, οὕτως ἐπιποθεῖ ἡ ψυχή µου πρὸς σέ, ὁ θεός.

Similarly, the tenth-century grammarian Dunash ben Labrat glossed ערג contextually as “to cry out,” making it a specialized synonym of קרא.

Aquila, for his part, used etymology rather than context to derive the meaning of ערג. He translated תַּעֲרוֹג at Joel 1:20 as ἐπρασιώθη, and כְּאַיָּל תַּעֲרֹג at Ps 42:2 as ὡς αὐλὼν πεπρασιασµένος.⁴ The basis of these translations was the resemblance between תַּעֲרֹג and another rare word, עֲרוּגַה.⁵ (The neologisms ἐπρασιώθη and πεπρασιασµένος are derivatives of πρασιά, “planting bed,” which Aquila had used to translate עֲרוּגַה wherever it appeared.) Later, Jerome—and thus the whole Latin tradition in the case of Joel 1:20—followed Aquila for the translation of these verses.⁶ Probably the word was just as mysterious to all of these pre-modern translators as it is to us.

Scholars since the mid-seventeenth century have tried to establish the meaning of ערג from Semitic cognates. In 1648, Louis de Dieu observed that ዐረገ in Ethiopic and عرج in Arabic both mean “to go up,” and that Ps 42:2 could therefore be translated “sicut cerva ascendit ad decursus aquarum, sic anima mea ascendit ad te.”⁷ In 1967 Edward Ullendorff, believing that he was making an innovative application of his learning in Geʿez, unwittingly recapitulated de Dieu’s conjecture.⁸

Two decades after it appeared, Samuel Bochart threw out de Dieu’s translation on the grounds that a beast goes down, not up, to water.⁹ In its place he cautiously endorsed the opinion of Dunash that תַּעֲרוֹג referred to a deer’s cries, with the reservation that it could not refer only to deer, since at Joel 1:20 it takes all of the beasts of the field for its subject. Bochart also cited the myth of the huntress Argê, turned into a deer by the sun-god, and mused that her name was derivable from ערג.¹⁰

In 1669, Johannes Coccejus speculated that ערג was a derivative or a relative of ערך, with the meaning “to set out [a complaint],” as at Ps 5:4. Hence Ps 42:2 was presumably to be read “as a deer complains for the riverbeds…,” and Joel 1:20 “as the beasts of the field complain …”¹¹

In the same year, Edmund Castell provided a list of cognates to ערג from Geʿez and Arabic. From then on, any scholar could review the variety of meanings associated with both ዐረገ and عرج. Castell himself did not connect any of these meanings to either תַּעֲרוֹג or עֲרוּגַה, and supplied only the traditional glosses desideravit; glocitavit for the former and areola for the latter.¹² A century later, however, Johann David Michaelis mined the cognates in Castell’s entry to formulate a new hypothesis. In the first place, he endorsed de Dieu’s theory that ערג means “ascendit.” Then he posited that עֲרוּגַה was indeed related to ערג, but that it must mean something raised, like a trellis, rather than a planting-bed. To prop up this conclusion he cited the Qurʾanic word مَعرَج, which Castell had recorded as meaning “ladder.”¹³

Such was early-modern learning on the root ערג. Wilhelm Gesenius’ much-cited dictionary entries were no more than résumés of this scholarship. He endorsed de Dieu’s opinion and glossed ערג as “aufsteigen,” but posited that תַּעֲרוֹג is to be taken in a tropical sense of longing, as the LXX had suggested. As for עֲרוּגַה, he deprecated Aquila’s translation “planting-bed” and cited Michaelis’s conjecture with approval.¹⁴

Joshua Blau, departing from the tradition that Gesenius had transmitted, insisted on “to incline” as the basic meaning of ערג.‎¹⁵‏ He observed that the Arabic G-stem verb عَرَجَ, “to ascend,” is only attested in the Qurʾan in the very specialized sense of entering heaven. (The noun مَعرَج is likewise used only in connection with heaven).¹⁶ The word in the sense of ascent was therefore likely borrowed; likely from Ethiopic, in which the ordinary meaning of ዐረገ is indeed “to go up.” Consequently, عَرَجَ cannot be relied on as a straightforward cognate of ערג. The root ع ر ج’s native Arabic meaning, however—and the meaning in which it is properly compared to ערג—is rather “to incline,” a sense which is apparent in the Qurʾanic noun أَعرَج, “cripple”;¹⁷ and in other early Arabic sources in the collateral G-stem verb عَرِجَ, “to limp; to incline,”¹⁸ in D-stem عَرَّجَ, “to turn, to lame,” and in numerous other derivations of the root (most of which can incidentally be found in Castell’s Lexicon). Blau contended from this evidence that Biblical ערג is best understood as a verb meaning “to incline.” The LXX’s translations, moreover, were approximately correct all along, for “to long” is a plausible tropical extension of the verb’s intransitive sense. Like Gesenius before him, Blau was perhaps motivated in the last stage of his argument by a wish to save the appearance of the traditional translations. Below, by contrast, I will argue that there is no objection at all to treating תַּעֲרוֹג as a transitive verb, or even to interpreting it literally.

Putting the lexical problem aside for now, our second puzzle is grammatical. In all three of its attestations the verb ערג takes the form תַּעֲרוֹג, which has been universally interpreted since the LXX as a 3fs imperfect form. So understood, it does not agree with two of its three apparent subjects: plural בַּהֲמוֹת and masculine אַיָּל.

In his commentary on Joel 1:20, Abraham ibn Ezra wrote that singular תַּעֲרוֹג referred to each single animal in a herd, as if its subject were bestiarum quæque. He cited בָּנוֹת צָעֲדָה at Gen 49:22 as an example of the same phenomenon.¹⁹ Wilhelm Gesenius, by contrast, posited that בַּהֲמוֹת שָׂדֶה תַּעֲרוֹג was an instance of a plural subject’s taking a singular feminine predicate, just as non-human plurals do in Arabic.²⁰ As for Ps 42:2, ibn Ezra noted that nouns which usually refer to male animals can sometimes denote females too, like עֵז בַּת שְׁנָתָהּ at Num 15:27.²¹ Eight centuries later, Emil Kautzsch gave precisely the same explanation.²² All of these are serviceable ad-hoc exculpations of their respective verses, but none of them accounts on its own for both problematic instances of תַּעֲרוֹג.

Twentieth-century editors of the Masoretic text saw another way out, offering emendations rather than grammatical explanations. Frants Buhl proposed in his edition of the Psalms that כְּאַיָּל תַּעֲרֹג is haplographic for כְּאַיֶּלֶת תַּעֲרֹג, and thus that the animal in question is a doe after all.²³ All of his successors have followed him, down to Hans Bardtke in the BHS. Thomas Cheyne, by contrast, suggested that תערג in Ps 42:2 had been metathesized from תגער, “cries out.”²⁴

At Joel 1:20, Wilhelm Nowack made the tentative conjecture of בֶּהֱמַת שָׂדֶה for בַּהֲמוֹת שָׂדֶה in order to account for the singular form תַּעֲרוֹג.‎²⁵‏ Otto Procksch retained this emendation for the BH3, and even deleted Nowack’s fortasse. Certainty, however, was no protection from posterity. Karl Elliger, editor for the BHS, recorded the conjecture, but deprecated it in favour of his own emendation תַּעֲרֹגְנָה for תַּעֲרוֹג. Finally, Anthony Gelston did away with both proposals in the BHQ.²⁶ It is just as well that he did so, for there is nothing to recommend Nowack or Elliger’s emendations apart from the comfort of the lectio facilior.

Indeed, all these grammatical explanations and textual emendations are embarrassed attempts to evacuate the scandal of a rare verb that appears in only one form. It is that form, תַּעֲרוֹג, which we must account for: parsimony warns against dodging the problem with piecemeal explanations.

I propose the following solution: ערג is to be treated in the first place as a transitive verb. And תַּעֲרוֹג is not a 3fs, but rather a 2ms verb whose implied subject is in all attested instances “God”; and whose direct objects are אַיָּל, נַפְשִׁי, and בַּהֲמוֹת. This analysis resolves all disagreements between subject and verb. It also lets the Masoretic text stand without modification.

The sense of ערג need not be fixed here. But if we borrow Joshua Blau’s hypothesis that ערג means “to bend” (without endorsing his further conclusions) we can read the verses like this:

Joel 1:20:

Quin et bestias agri ad te torques, nam alvei aquæ exsiccati sunt, et ignis consumpsit prata deserti.

Thou also turnest the beasts of the field unto thee, for the rills of water are dried up, and fire hath consumed the meadows of the wilderness.

Ps 42:2:

Sicut cervum ad alveos aquæ, ita animam meam ad te torques, O Elohim.

As thou turnest a deer unto rills of water, so turnest thou my soul unto thee, O God.

This would be nothing unusual: נַפְשִׁי is very often construed with verbs of spatial manipulation.²⁷

Alternatively, we might attribute to ערג the same tropical sense as חול and torqueo, viz. “to torture”:

Joel 1:20:

Quin et bestias agri de te angis, nam alvei aquæ exsiccati sunt, etc.

Thou also puttest the beasts of the field in anguish for thee, etc.

Ps 42:2:

Sicut cervum de alveis aquæ, ita animam meam de te angis, O Elohim.

As thou puttest a deer in anguish for rills of water, so puttest thou my soul in anguish for thee, O God.²⁸

It might be objected that I have visited the grammatical iniquity of אַיָּל תַּעֲרֹג and בַּהֲמוֹת שָׂדֶה תַּעֲרוֹג upon unsinning נַפְשִׁי תַעֲרֹג, whose traditional interpretation poses no problem of agreement. It is true that the SV interpretation anima mea desiderat is prima facie plausible. But it is no violent innovation to treat נַפְשִׁי תַעֲרֹג as an OV-ordered clause. We find the Psalms replete with such constructions, even if we restrict our selection to the form נַפְשִׁי. Consider, for instance, נַפְשִׁי יְשׁוֹבֵב at Ps 23:3, where the implied second-person subject is God, and where the predicate is placed after the object. Or else Ps 143:3, נַפְשִׁי דִּכָּא לָאָרֶץ; or Ps 86:4,אֵלֶיךָ אֲדֹנָי נַפְשִׁי אֶשָּׂא . With this grammatical possibility in mind, indulge a thought experiment. Consider Ps 16:10: לֹא־תַעֲזֹב נַפְשִׁי לִשְׁאוֹל—but imagine that עזב were an otherwise-unattested root of unknown meaning. I insist that there would be no way of knowing a priori whether נַפְשִׁי were the subject or the object of תַעֲזֹב. We would need to look for all available clues to determine the answer and confess ignorance if none were available. In the case of נַפְשִׁי תַעֲרֹג, however, we do have a clue to the correct syntactical interpretation. It is the very fact that תַּעֲרֹג does not agree in gender and number with the two nouns besides נַפְשִׁי that are traditionally purported to be its subjects, and that it is much more comfortably treated in those cases as a transitive verb that takes those nouns as objects.

Besides their advantages in agreement, my readings also eliminate a serious syntactical difficulty, which is that עַל and אֶל would be awkward prepositional complements for a verb meaning “to long.” Consider Ps 63:1:

צָמְאָה לְךָ נַפְשִׁי כָּמַהּ לְךָ בְשָׂרִי בְּאֶרֶץ־צִיָּה וְעָיֵף בְּלִי־מָיִם׃

And Ps 42:3:

צָמְאָה נַפְשִׁי לֵאלֹהִים לְאֵל חָי׃

The traditional view is that Joel 1:20 and Ps 42:2 are parallel cases in syntax as well as meaning to these two passages. Observe, however, that צָמְאָה and כָּמַהּ both take the preposition לְ, whereas only עַל and אֶל appear in Ps 42:2. In fact, Hebrew verbs of hungering, thirsting, and longing almost always take the complement לְ—not עַל or אֶל—before the thing longed for. If תַּעֲרֹג were to be interpreted as a synonym of צָמְאָה, then its construction with עַל or אֶל would be odd indeed. A seeming exception is Ps 119:20, גָּרְסָה נַפְשִׁי לְתַאֲבָה אֶל־מִשְׁפָּטֶיךָ בְכָל־עֵת. But there, אֶל־מִשְׁפָּטֶיךָ is not an independent complement of a verb, but is rather set within a prepositional phrase that complements the object-nounתַאֲבָה . When the verb תָּאַבְתִּי appears later in the psalm, it does take the complement לְ after all.

The LXX, the Rabbis, and our dictionaries tell us that תַּעֲרוֹג is an intransitive 3f form. But tradition is no substitute for the evidence, which urges us to draw another conclusion.