by Joseph Habib


The great textual critic Dominique Barthélemy (1921–2002) lamented the fact that scholars of the time were too quick to render a difficult passage in the Masoretic Text as corrupt or unintelligible. The Masoretic Text refers to that reading tradition of the Hebrew Bible which the Masoretes of Tiberias worked to codify from the period of about 500 C.E.–900 C.E. Barthelémy had come to realize that the solution to some of these issues were to be found in a deep understanding of Biblical Hebrew grammar, syntax and style, an understanding which Barthélemy found in none other than the Bible translations and commentaries of R. Saadya Gaon (d. 942 C.E.) and the early-medieval Karaites (ca. 850–1099 C.E.), both of whom wrote in Judaeo-Arabic. Barthelémy’s description of these scholars’ work reflects his high esteem for them:

“…[T]here are two groups of sources of the greatest import to which we will frequently turn … First are the commentaries of those literal interpreters who are both contemporaries and compatriots of the Masoretes to whom we owe the model codexes of Cairo, Aleppo, and Leningrad. These include the commentaries of Daniel al-Qumisi and Saadya Gaon … But the most valuable for us are those of Yefet ben Ely … Yefet, who lived in Palestine, wrote a commentary in Arabic on all of Scripture … [H]is copious commentary testifies to a number of exegetical traditions that preceeded him. It is thus likely one of the best witnesses of literal exegesis that would have vividly in mind those who fixed the vocalization and accents of the classical Tiberian text.”[1]

To my knowledge, these sources remain relatively unknown to those outside the specialised field of Judaeo-Arabic literature in general, and to those within the fields of critical study of the Bible in particular. In this post I offer a few examples where these medieval scholars provide unique insights into the Biblical text. These insights also seem to reflect certain developments within Tiberian reading tradition as well as clarify otherwise difficult texts.

Some Examples

1 Samuel 3:3 וְנֵ֤ר אֱלֹהִים֙ טֶ֣רֶם יִכְבֶּ֔ה וּשְׁמוּאֵ֖ל שֹׁכֵ֑ב בְּהֵיכַ֣ל יְהוָ֔ה אֲשֶׁר־שָׁ֖ם אֲר֥וֹן אֱלֹהִֽים׃ “The lamp of God had not yet gone out, and Samuel was lying down in the temple of the Lord, where the ark of God was” (English Standard Version), contains the problematic image of Samuel sleeping in the temple. Early on in the Jewish tradition, this issue was resolved by interpreting that Samuel was in the courtyard of the tabernacle along with the Levites, not in the Holy of Holies itself.[2] Later commentators noticed the accent sign athnach (which looks like an upside-down “v”) underneath the word שֹׁכֵ֑ב “lying down.” The accent athnach marks the main division of a verse.

Figure 1 1 Samuel 3:3 in one of the oldest and most prestigious Tiberian Bibles, the Aleppo Codex—from around 930 C.E. (open access). Notice the athnach underneath the word שֹׁכֵ֑ב.

This placement of the athnach therefore rendered the phrase וּשְׁמוּאֵ֖ל שֹׁכֵ֑ב “and Samuel was lying down” as a parenthetical statement so that the lamp—not Samuel—is in the Holy of Holies. According to this interpretation, the verse could now be translated as, “And the lamp of the Lord had not yet gone out (Samuel was lying down) in the temple of the Lord where the ark of God was.”[3]

This tension is evident in sub-traditions of the Tiberian reading tradition. The Tiberian tradition generally prevailed throughout the Middle East, North Africa and, later, Iberia and most of Europe. However, some Medieval manuscripts which have their provenance in these areas preserve minor differences in their accentuation and vocalization. These differences testify to the existence of sub-traditions within the general framework of the Tiberian tradition. The sub-traditions evidently contained readings through which the official Tiberian tradition could not penetrate, as is the case with 1 Samuel 3:3.[4] In some of these sub-traditions, יְהוָה was punctuated with atnach so that the entire phrase וּשְׁמוּאֵל שֹׁכֵב בְּהֵיכַל יְהוָה “And Samuel was lying down in the temple of the Lord” is unambiguously grouped together, viz., “The lamp of God had not yet gone out, and Samuel was lying in the temple of the Lord—where the ark of God was.”

Figure 2 Variant accent pattern of 1 Samuel 3:3, preserved in some Medieval manuscripts. See Christian D. Ginsburg, Prophetae Priores (London: British and Foreign Bible Society, 1911), 139 (open access). Notice the athnach (upside-down “v”) is now under יְהוָֹה “the Lord.”

This same accentuation pattern is found in a copy of Yefet’s commentary on 1 Samuel, where the verse is written out for comment.[5] Of course, there is no way of knowing whether the verse was copied like this in Yefet’s original autograph, or if this is just a scribal addition. In any case, Yefet interprets this verse such that the lamp and/or Samuel could be “in the temple.” On the basis of 1 Samuel 3:15 וַיִּפְתַּ֖ח אֶת־דַּלְת֣וֹת בֵּית־יְהוָ֑ה “And [Samuel] opened the doors of the house of the Lord” and 1 Chronicles 9:19 which mentions שֹׁמְרֵ֥י הַסִּפִּ֖ים לָאֹ֑הֶל “keepers of the thresholds of the tent,” Yefet concluded that the tabernacle at Shiloh was surrounded by a wall on the outside of the outer court. He deduced this since the outer court of that tabernacle was a curtain; a curtain can have neither a door nor a threshold. 1 Samuel 3:15 and 1 Chronicles 9:19 must therefore refer to this wall. The lamp, Yefet concludes, illuminated the space between this wall and the outer curtain of the tabernacle complex. He then adds, “The expression בְּהֵיכַל יְהוָה ‘in the temple of the Lord’ either relates to the lamp or to Samuel. So, that place is called the Temple of the Lord.”[6] Samuel could therefore also occupy the space where the ark of God was, since the whole complex was called the “Temple of the Lord.” For Yefet, then, it did not matter whether the lamp, or both the lamp and Samuel were בְּהֵיכַל יְהוָה “in the temple of the Lord.” Both fit the facts of Scripture. Such an understanding frees the verse of any potential blashphemous imagery. If Samuel’s occupying the same space as the ark of God can be accomodated, any reading tradition which adopts any such interpretation would then be realized in the accentuation pattern of the sub-traditions that divide the verse (with athnach) at יְהוָה “the Lord” (as in Figure 2). This was most likely the more simple and straightforward reading. Yefet attests to this proposition. Further to all that Barthelémy stressed above, Yefet took the whole of Scripture into consideration. Recall also that earlier Jewish literature contained such an interpretation (see above). These facts suggest that the accentuation preserved in the sub-traditions, which divided at יְהוָה “the Lord,” may have been earlier than the accentuation attested in the Masoretic Text.

Yefet’s interpretation above did not reflect the division of the accents of the official Tiberian Masoretic Text. In my PhD research, I have found this generally to be case for most of the early medieval Karaites. By contrast, the data suggest that the translations and commentaries of Saayda Gaon generally do reflect the syntactic divisions of the recieved Tiberian tradition. One example of this is the clause in Genesis 49:9 מִטֶּ֖רֶף בְּנִ֣י עָלִ֑יתָ, where Jacob is blessing his son Judah. Both the LXX and the Targums translate the word בְּנִ֣י “My son” as if it was vocative thus making it the subject of the following verb, “From prey, O my son, you have gone up.” Yefet, a tenth-century Karaite commentary called the Talkhīs (“Abridgement”), a Karaite active in the second half of the tenth century named David ben Abraham al-Fāsi and an eleventh-century Karaite named ʿAli ben Sulayman all interpret the verse in this manner.[7] The problem with this interpretation is that the phrase is punctuated such that מִטֶּ֖רֶף “from prey” is separate from what follows and therefore groups בְּנִ֣י עָלִ֑יתָ together. According to this division, it is difficult to read בְּנִ֣י as vocative.[8] Saadya, however, translates the first half of this verse as, “O Judah, you will be the cub of a lion because you saved my son from death” (emphasis mine).[9]

Figure 3 Saadya’s translation of Genesis 49:9, in Joseph Derenbourg, Version Arabe du Pentateuqe de r. Saadia Ben Iosef al-Fayyoúmi, vol. 1, OEuvres Complètes de. r. Saadia Ben Iosef al-Fayyoûmi (Paris: E. Leroux, 1893), 77 (open access). The reading here is the same as that of the RNL manuscript (see note 9).

This translation reflects a division similar to that of the accents since Saadya separated the verb and its object (בְּנִ֣י עָלִ֑יתָ) from the prepositional phrase (מִטֶּ֖רֶף) which modifies the verb phrase. It also makes better sense within the narrative since it was Judah who suggested to sell Joseph rather than leave him in the pit to die (Gen. 37:26).[10] Saadya therefore offers one way to understand the reading tradition as it stands.

One final example. In Genesis 3:22, God says הֵ֤ן הָֽאָדָם֙ הָיָה֙ כְּאַחַ֣ד מִמֶּ֔נּוּ לָדַ֖עַת ט֣וֹב וָרָ֑ע “Behold, the man has become like one of us, knowing good and evil.” The word אַחַ֣ד “one” has construct vocalization which tells the reader that it is paired with the following word, viz., “one of us.” In the Babylonian tradition, however, the vocalization of אחד in this verse is equivalent to אֶחָד in the Tiberian tradition; that is, it is not in construct with what follows.[11]

Figure 4 The non-construct vocalisation of אֶחַד in Babylonian pointing from Exodus 25:19 (T-S N 283.11). According to Yeivin (see note 11), this is how the consonants אחד are pointed in Genesis 3:22. (An image of that manuscript was not available to me.) In the Babylonian tradition, the vowels are above the letters, not below as in the Tiberian tradition. I thank the Syndics of Cambridge University Library for permission to reproduce this image.

The Karaite al-Fāsi knew this tradition of the “Iraqis” (עראקיין) and explains its interpretation:

“The Lord Almighty said, ‘Behold mankind has become as one. The means to obtain knowledge of good and evil belong to him.’ The meaning is that mankind has become like one. He has become without parallel in his judgement. The free will to know/learn good and evil belong to him. So, lest he also exercise his free will and take from the tree of life…”[12]

Yefet also mentions this same interpretation as one possibility among many others known to him, again confirming part of Barthelémy’s statement above.[13] This evidence strengthens the possibility that the same interpretations which gave rise to the various reading traditions among the masoretes were known in Karaite circles.


The Judaeo-Arabic Bible translations and commentaries of Saadya Gaon and the early medieval Karaites are important sources for those engaged in critical study of the Hebrew Bible. As knowledge of this exegetical literature grows, it seems to confirm Barthélemy’s general attitude towards these sources. The Karaites and Saadya Gaon certainly had contact with the Tiberian masoretes. It is difficult to imagine that this contact would not affect their translations and exegesis to some degree. In any case, the evidence suggests that there is no reason to not make the utmost use of these sources for a precise understanding of the Masoretic Text as it stands.

Joseph Habib is entering his third year of PhD studies at the University of Cambridge under the supervision of Prof. Geoffrey Khan and co-supervision of Prof. Tamar Zewi (University of Haifa). The topic of his research is ‘Accents, Vocalisation and qere/ketiv in the Bible Translations and Commentaries of Saadya Gaon and the Early Medieval Karaites’. This work is made possible thanks to a generous contribution from the Valler Doctoral Fellowship granted by the University of Haifa’s Department of Biblical Studies and Jewish History.


[1] Dominique Barthélemy, Studies in the Text of the Old Testament: An Introduction to the Hebrew Old Testament Project, trans. Stephen Pisano et al. (Winona Lake: Eisenbrauns, 2012), 593.

[2] This interpretation may be found in the Babylonian Talmud (Kiddushin 782:b), Rashi’s commentary and in the Targum.

[3] The role of athnach here is recognised by David Qimḥi (1160–1235), Jedidiah Solomon ben Abraham Norzi (1560–1626) who wrote the masoretic commentary Minchat Shai, and Rabbi Meir Libush ben Yechiel (Malbim; 1809–1879).

[4]  For more on these sub-traditions, see Menaẖem Cohen, ‘תת-שיטות של הטעמה טברנית “חוץ-מסורתית” ושיעור תפוצתן בכתבי-יד של מקרא בימי הביניים [Subsystems of Tiberian “Extramasoretic” Accentuation and The Extent of Their Distribution in Medieval Biblical Manuscripts]’, Lĕšonénu 51, no. 3/4 (1987): 188–206.

[5] See The National Library of Russia, St. Petersburg [RNL] Ms. EVR ARAB I 156 fol. 18v ln. 1 (fourteenth century).

[6] RNL Ms. EVR ARAB I 1313 fol. 37r lns. 1–3. See fol. 36v ln. 6ff for the remainder of the comments for this verse. Italics indicate Hebrew words embedded within the Judaeo-Arabic text.

[7] Yefet, the Talkhīs and ʿAli see here a prophecy referring to David’s future victories (since David was from the line of Judah); al-Fāsi interprets מִטֶּ֖רֶף בְּנִ֣י as a subordinate causal clause, “From eating the prey, O my son, you have gone up and you have grown up.” For Yefet’s commentary see RNL Ms. EVR ARAB I 28 fol. 83r ln. 4–83v. ln. 8. For the Talkhīs, see RNL Ms. EVR ARAB I 4785 fol 190r. ln. 18–fol. 109v ln. 13. For al-Fāsi see Skoss, Solomon L. The Hebrew-Arabic Dictionary of the Bible Known As Kitāb Jāmiʿ al-Alfāẓ (Agrōn) of David Ben Abraham al-Fāsī the Karaite (Tenth Cent.). 2 vols. Yale Oriental Series Researches 20. New Haven: Yale University Press, 1936, vol. 2 p. 397, for ʿAli ben Sulayman see RNL. Ms. EVR ARAB I 2029 fol. 97r lns. 17–19.

[8] For the vocative sense, one would expect the vocative noun to be grouped with what comes before, such as אֵלֶ֥יךָ יְהוָ֖ה אֶקְרָ֑א “To you, O Lord, I call” (Joel 1:19).

[9] תכון יא יהודא כשבל אסד לאנך כלצת אבני מן אלקתל. See RNL Ms. EVR. II C 1 fol. 113r lns. 7–8.

[10] The verb עָלִ֑יתָ in this case must be reinterpreted as transitive (most likely as Hiphil). Cf. Ibn Ezra’s commentary on this verse who also suggests this as a possibility, but for a different reason.

[11] See Israel Yeivin, מסורת הלשון העברית המשתקפת בניקוד הבבל [The Hebrew Language Tradition as Reflected in the Babylonian Vocalization], 2 vols (Jerusalem: The Academy of the Hebrew Language, 1985), vol. 2 p. 1107.

[12] Skoss, Kitāb Jāmiʿ, vol. 1 p. 61.

[13] See RNL Ms. EVR ARAB I 26 fol. 51vff. Also found in the British Library Or. 2462 fol. 96rff; National Library of France Ms. Hebr 277 fol. 131rff.; RNL Ms. EVR ARAB I 142 fol. 24rff.; Institute of Oriental Manuscripts, Russia Ms. B 51 fol. 171vff.

  1. Emilio Spedicato

    See Also al masudi about variances between samaritan he
    Breve and septuaginta…. Especially in chronologies… I am presentato translating into italian samaritan text as in english versione by tsedaka

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.