Shiur #7d: Nusach Ha-mikra – Accuracy of the Biblical Text

  • Rav Amnon Bazak

D.        "Keri u-Khetiv"


The phenomenon of "keri u-khetiv" variants (where a word is vocalized differently from the way in which it is written) is manifest, to a limited extent, already in Chazal's teachings,[1] and among the Masoretes, who began to note the vocal form in the margins of the manuscript itself, a practice which became very common. Different possibilities have been proposed as to the origins of this custom. Radak offers the following explanation:


"During the first exile, the books were lost or carried away, and the Sages who knew the Bible passed away, and the Men of the Great Assembly, who restored the Torah to its previous honor, found discrepancies in the books and [therefore] followed the majority [of them] to the best of their understanding. Where their conclusions were not decisive, they wrote one option without vowels [thereby leaving room for more than one possibility for vocalization], or wrote on the outside but not on the inside, or wrote the word one way in the text itself, but in a different way on the outside."[2]


According to Radak, then, the phenomenon represents a method of preserving different versions of a word in those instances where the Men of the Great Assembly were unable to reach a final decision. For this reason, Radak throughout his commentary labors to explain the text according to both "keri" and "ketiv,” with the assumption that the latter represents a textual version that existed in the manuscripts that the Men of the Great Assembly worked from.


Abarbanel, in his introduction to the Book of Yirmiyahu, attacks Radak's position in the strongest terms, on multiple grounds. First, he asserts the "argument from faith" against raising the possibility of any doubt as to the correct textual version:


"How can I believe or suggest that Ezra the Scribe found a Book of God's Torah and Books of the Prophets and other works written with Divine inspiration, to contain any doubt or confusion? For a Book of Torah that lacks even a single letter is unfit for use; how much more so [one that contains errors] in 'keri u-khetiv' which come with the Torah."


To this we might respond, as we have already noted, that Chazal themselves testified to their lack of expertise as to the exact text. However, Abarbanel also invokes an "arguments from probability": if indeed the phenomenon arises from some doubt that arose in Ezra's mind concerning the proper version,


"Why, then, in explaining the text, do we always follow the version that is read, rather than the version that is written? If Ezra had any doubt in the matter, why would the vocalization always agree with the 'keri' and not with the 'ketiv'? This itself indicates that in his view the 'keri' (vocalized version) is the correct one, and therefore he vocalized it thus, and not as it is written. And if indeed this was his opinion, then he should have placed the 'keri' version in the text, for this would be the proper variant, in accordance with the vocalization, and the 'ketiv' should have been placed outside!"


One may respond easily to these objections. First, as discussed previously, Ezra did not vocalize (i.e., add vowels) to the text; this was done about a thousand years later, by the Masoretes. Furthermore, the vocalization does not, in and of itself, express any preference for the read version over the written version, since there would be no point in adding vowels to a word which, by its very definition, is not vocalized. Finally, the 'keri' version is not inside the text but rather in the margins because it is not an integral part of the text; it is an instruction as to how it is to be read. It therefore seems reasonable to assume that the 'ketiv' version does reflect the majority of the manuscripts even if it runs counter to what we may have expected.[3]

In any event, Abarbanel's own view is that the 'ketiv' is indeed the correct version, while the 'keri' represents the interpretation of Ezra the Scribe, since the 'ketiv' is more difficult to understand,


"For one of two reasons: Either because the person who wrote these strange words had in mind some secret of the Torah in keeping with his level of prophecy and the profundity of his wisdom…. Or because he who uttered them was not sufficiently precise, either because of insufficient knowledge of the Hebrew language or because of insufficient knowledge of proper writing, such that this emerged from the prophet or the individual speaking with Divine inspiration 'like an error in a royal edict'(Kohelet 10:5)."


In other words, Abarbanel prefers to suggest that the problematic 'ketiv' form may arise from a linguistic error on the part of the prophet, rather than entertaining the possibility of multiple versions of the text. Abarbanel is especially fierce in his explanation of why the phenomenon of 'keri u-khetiv' is particularly prevalent in the Book of Yirmiyahu. He proposes that Yirmiyahu was "young in years when he began prophesying, and was therefore not yet proficient in the ways and rules of language, and the beauty of metaphor; indeed, he said of himself, 'I cannot speak, for I am but a child' (Yirmiyahu 1:6)."


Contemporary scholars have offered other possible explanations for 'keri u-khetiv',[4] but a review of different instances of the phenomenon would seem to indicate that there are different types of keri-u-khetiv, and that no one explanation of the phenomenon covers all instances.


Thus, for instance, in many dozens of cases we see that the differences between the written version and the version that is read relate to normal linguistic phenomena that commonly lead to textual variations. This would seem to correspond to Radak's view that there were different versions in the manuscripts, and in most cases the 'keri' variant, with its vowels, represents the more probable alternative. For instance, there are many instances of variants arising from the graphic similarity between letters such as vav/yud;[5] bet/kaf;[6] or daled/resh.[7] There are also some exchanges that are less common, such as 'daled' and final 'khaf',[8] or even exchanges involving adjacent letters.[9] Other instances involve an inversion of the order of letters in the word,[10] or other such linguistic phenomena.


In contrast, there are instances where the written text (ketiv) displays ancient grammatical forms, and the reading form (keri) replaces them with later forms. Here it makes sense to suggest that the discrepancies do not reflect multiple manuscript versions, but rather deliberate changes,[11] in keeping with Abarbanel's approach. For instance, in ancient Hebrew, the letter 'yud' is a suffix for the feminine second person singular,[12] and in various places the 'keri' version replaces this with the grammatical form more prevalent in Tanakh. Examples include the verses:


"… concerning which you (ואתי [וְאַתְּ קרי]) swore, uttering it also in my hearing…" (Shoftim 17:2);

"What have you (לכי [לָךְ קרי]) in the house… borrow vessels from outside from all your neighbors (שכנכי [שְׁכֵנָיִךְ קרי])" (Melakhim II 4:2-3);

"And you and your children (בניכי [וּבָנַיִךְ קרי]) shall live off the remainder" (ibid. 7).


Similarly, in ancient Hebrew the feminine third-person plural suffix in the past tense was a 'heh' rather than a 'vav', and here too this ancient form is the basis for a 'keri u-khetiv' distinction. Examples include,


"Our hands have not spilled (שפכה [שָׁפְכוּ קרי]) this blood" (Devarim 21:7); 

"… for the ships were wrecked (נשברה [נִשְׁבְּרוּ קרי]) at Etzion-Gaver" (Melakhim I 22:49).[13]


Thus in summary, there are examples of the keri u-khetiv phenomenon that indicate the possibility of linguistic errors and grammatical updates in accordance with Abarbanel’s understanding, but there are also instances of 'keri u-khetiv' which appear to indicate the existence of different textual versions, reflecting discrepancies between different manuscripts, in line with the explanation of Radak.


Translated by Kaeren Fish


[1]The most common example, of course, is the Ineffable Name of God, which is voiced differently from the way that it is written. The Gemara teaches (Nedarim 37b), "The textual reading as established by the Sofrim, and the embellishments of the Sofrim, and the letters that are pronounced but not written, and those that are written but not pronounced – all these were handed down as a law given to Moshe at Sinai." The Gemara then enumerates instances of words that are voiced even though they are not written, and vice versa. [Incidentally, this indicates that the formula “a law given to Moshe at Sinai” is used by Chazal not necessarily in the literal sense, but rather in the sense of an ancient tradition, since the examples cited in the Gemara are from the book of the Prophets and Ketuvim, rather than from the Chumash (as noted by Rabbi Yisrael Lipschitz in his commentary on Yoma, Tiferet Yisrael, 2:12).] Rabbinic literature includes several teachings about "keri u-khetiv" in the more familiar sense – i.e., that the text is written in one form, but read in a different form. See, for example, Sota 42b, pertaining to the keri u-khetiv in Shmuel I 17:6, 23.

[2]See Y. Ofer, "Ketiv u-Keri: Pesher ha-Tofa'a, Darkei ha-Simmun Shelah ve-De'ot ha-Kadmonim Aleiha," Leshonenu 70, 5768, pp. 55-73; Leshonenu 71, 5769, pp. 255-279.

[3]Abarbanel raises further arguments: 1. Sometimes the 'keri' and 'ketiv' distinctions are consistent, such as the word "צביים" which is read as " צְבוֹיִם" (Bereishit 14:2,8; Devarim 29:2. In the first instance the letter 'yud' appears twice even in the 'keri' version, after the letter 'vav.’) How is it logical to suggest that the same scribal error occurred in every instance of this word? 2. How are we to explain the statement in the Gemara (Megilla 25b) that "all descriptions that are written in the Tanakh in explicit terms, are read in a euphemistic way: for example, "… but another man shall lie with her ('ישגלנה/ישכבנה') (Devarim 28:30); 'בעפולים/בטחורים' (ibid., verse 27); 'חריונים/ דביונים' (Melakhim II 6:25);…? Obviously, Radak would respond that this is a different sort of "keri u-khetiv" that has nothing to do with textual variants, but rather reflects the guidelines for how the verses containing such terms are to be read in public.

[4] See Y. Yeivin, Mavo li-Mesora ha-Teveryanit, Jerusalem 5732, p. 50 [= Ha-Masora la-Mikra, Jerusalem 5763, p. 57]. Rabbi M. Breuer, "Emuna u-Mada' be-Nussach ha-Mikra," De'ot 47, 5738, pp. 102-114 [= Y. Ofer (ed.), Shitat ha-Bechinot shel ha-Rav Mordechai Breuer, Alon Shevut 5765, pp. 71-91] proposes that the 'ketiv' represents the tradition of the Sofrim (Scribes) who engaged in the copying of manuscripts in accordance with the customs of earlier generations of scribes, while the 'keri' represents the tradition of the scholars who learned these texts from their teachers. As a result, there were discrepancies between the versions.

[5] See Y. Ofer, "Chilufei vav-yud ba-Mikra ve-Hishtakfutam be-He'arot ha-Masora," Mechkerei Morashtenu 2-3, 5764, pp. 69-84, noting that there are some 315 instances of vav/yud 'keri u-khetiv' variants in Tanakh – almost a third of the total number of such variant pairs.

[6] See, for example, Shmuel I 11:6,9; Shmuel II 5:24; ibid. 12:31

[7]See, for example, Shmuel II 13:37; Melakhim II 16:6; Yirmiyahu 31:39; Mishlei 19:19.

[8] A good example is to be found in the verse, "And he came, and behold, Eli was sitting upon his seat by (יך [יַד קרי]) the wayside, watching…" (Shmuel I 4:13). The logic of the 'keri' version is clear, as Rashi articulates it: "He was waiting by (al-yad) the way". The 'ketiv' is less clear: Radak – who, as noted above, consistently explains both the 'keri' and the 'ketiv', proposes a somewhat forced explanation here: "Meaning, his heart was pounding (makkeh) within him, fearing for the Ark of God which had gone out [to battle], and this is the reason that the text uses the word יך."

[9]A good example involving adjacent letters is to be found in the verse: "And it was, when all the kings of the Emori heard… that God had dried up the waters of the Jordan before Bnei Yisrael until they had passed over (עברנו [עָבְרָם קרי]), their hearts melted and they no longer had any spirit in them before Bnei Yisrael" (Yehoshua 5:1). Here, too, the 'keri' makes more sense, since the verse is part of the narrator's account, which maintains the third person (see Radak). The 'ketiv' reflects the graphic similarity between the pair of letters 'nun-vav' and a final 'mem.’

[10]See, for example, Shmuel I 19:18, 19, 23, 24; Shmuel II 20:14; Melakhim II 2:16.

[11] This does not necessarily mean that at some stage someone amended the text as it appeared in the manuscripts. It may be that the reading tradition was consolidated in a different way from the writing tradition, as is usually the case in a language, and eventually someone added a comment reflecting this (following the introduction of vowelization).

[12] This ancient form has been preserved in lyrical units in various places in Tanakh, such as: "He Who forgives all your sins (עֲוֹנֵכִי), Who heals all your diseases (תחֲלֻאָיְכִי), Who redeems your life (חַיָּיְכִי) from the pit, Who encircles you (הַמְעַטְּרֵכִי) with love and compassion" (Tehillim 103:3-4); "Return to your rest (לִמְנוּחָיְכִי), my soul, for the Lord has dealt with you (עָלָיְכִי) bountifully" (ibid. 116:7).

[13] See further M. Cohen, Ha-Ketiv ve-ha-Keri she-ba-Mikra, Jerusalem 5767.