On the roles of the brothers in Genesis 37:28

(The above rendering—and footnote—come from the RJPS translation, an adaptation of the NJPS translation. Before accounting for this rendering, I will analyze the plain sense of the Hebrew text.)

As with the NJPS base translation, the general RJPS practice has been to clarify references that are apparent according to Hebrew conventions but would otherwise be opaque or ambiguous in English. (See “Clarified References” in the Preface, p. xix and n. 30.) This is one such instance.

Here the Hebrew text does not spell out the acting subject of the verb וַיִּמְשְׁכוּ “[they] pulled out.” This is a case of what linguists call underspecification. (See, e.g., Frank Polak, “Participant Tracking, Positioning, and the Pragmatics of Biblical Narrative,” in Advances in Biblical Hebrew Linguistics: Data, Methods, and Analyses, edited by Adina Moshavi and Tania Notarius [LSAWS 12; Winona Lake, IN: Eisenbrauns, 2017], 153–72, here 154–157.)

Such underspecification is cognitively motivated. A speaker will tend to produce utterances that are underspecified because an audience is not only capable of “connecting the dots” via inference but also does so automatically. As the linguist Stephen Levinson has pointed out, when it comes to human communication, “the essential asymmetry is: inference is cheap, articulation expensive, and thus the design requirements are for a system that maximizes inference” (Stephen C. Levinson, Presumptive Meanings: The Theory of Generalized Conversational Implicature [Cambridge, MA: MIT Press, 2000], 29). Consequently, as some cognitive scientists have argued, “any effort the speaker makes to express a distinction that could have been inferred is, in effect, wasted effort” (Steven T. Piantadosi, Harry Tily, and Edward Gibson, “The Communicative Function of Ambiguity in Language,” Cognition 122.3 [2012]: 280–91, here 284).

Such underspecification works in cases like this, because the verbal subject is inferable from the context, as a matter of salience and narrative coherence. In the preceding three verses, the text had disclosed the intention of the brothers (except notably for Reuben) to sell Joseph into slavery to the passing caravan (vv. 25–27). Now we learn that the caravan has reached their vicinity, so we expect that they will carry out their dastardly plan. This verse tells us that the expected action does occur. (On the labeling of that caravan in this verse via a changed label, see my separate comment here.)

Crucially, such underspecification is demonstrably a matter of convention in the Hebrew Bible. According to Lénart de Regt’s classification of how biblical narratives conventionally enable the tracking of participant references across clauses, this instance is one of “coreferentiality with a discourse active subject.” De Regt gives the example of 1 Sam 15:27, describing a scene involving King Saul and the prophet Samuel:

וַיִּסֹּ֥ב שְׁמוּאֵ֖ל לָלֶ֑כֶת

Samuel turned to walk away.

וַיַּחֲזֵ֥ק בִּכְנַף־מְעִיל֖וֹ

He [i.e., Saul] seized the corner of his [i.e., Samuel’s] robe.

In that context, Saul was the already-discourse-active subject. True, with the start of this verse, a clause intervenes (shown in purple) that specifies Samuel as its subject, but that clause is in service of the narrator’s ongoing focus on Saul; it does not derail that focus. Hence there is no need to label Saul again by name in the subsequent clause.

De Regt also gives as relevant examples Gen 14:9–11 (where it is the invading kings who continue their conquest, after the local kings have fled); 1 Sam 9:23–24 (where it is Samuel who continues to speak, after the cook has complied with his request); and Ruth 4:1–2 (where it is Boaz to continues to act, after the redeemer has complied with his request). See Lénart J. de Regt, “Anaphoric Accessibility in Biblical Hebrew Narrative: Global and Local Partic­ipant Tracking across Clause Boundaries,” in Ancient Texts and Modern Readers: Studies in Ancient Hebrew Linguistics and Bible Translation, edited by Gideon R. Kotzé, Christian S. Locatell, and John A. Messarra (Studia Semitica Neerlandica 71; Leiden: Brill, 2019), 63–78, here 67–72; idem, Linguistic Coherence in Biblical Hebrew Texts: Arrangement of Information, Participant Reference Devices, Verb Forms, and Their Contribution to Textual Segmentation and Coherence (Piscataway, NJ: Gorgias Press, 2019), 12–13; author manuscript here. Another striking example is Judg 18:2–8, where the five Danite agents continue to be the topic in v. 8, even though the last half of v. 7 has meanwhile described their prey (the people of Laish) at some length.

In this narrative context, if the narrator had indeed stated the subject explicitly, it would have registered as overencoding, that is, as odd. In effect, it would have marked the present narrative development as contrary to expectation—which is the opposite of the intended message. (See, e.g., Steven Runge, “Pragmatic Effects of Semantically Redundant Anchoring Expressions in Biblical Hebrew Narrative,” Journal of Northwest Semitic Languages 32.2 [2006]: 87–104.)

Some commentators instead hold that the text here is intentionally ambiguous or obscure, with regard to who sold Joseph into slavery. Others (Rashbam, Luzzatto) resolve the reference differently, believing that those who pulled Joseph out of the pit were the just-named Midianites—a separate party that came suddenly out of nowhere—who then sold him to the previously mentioned caravan of Ishmaelites, while the brothers were situated too far away to be aware of either event.

The problem with claiming that such construals are the plain sense is that they disregard the conventions of the language, which not only have cognitive priority in the minds of the audience but also yield a straightforward solution and a coherent text in this case. Indeed, given the expectation that was already set up by this point, namely that the brothers would take action as planned, the narrator would have needed to tell the audience about the purported sudden change in direction—such as by saying explicitly “but the brothers did not know.”

In short, RJPS here makes yet another substitution of a label to clarify an important yet ambiguous reference, following its general practice; for a related instance, see my comment at v. 25. Similarly NIV, CEV, NLT.