1.0. Introduction

1This study takes a close look at five fragments of ‘Common Bibles’ from the Cairo Genizah, a category of biblical text that encompasses probably the majority of Hebrew Bible fragments found there. The texts are analysed on a textual and linguistic basis to see what they reveal about the phonetics of the Tiberian reading tradition in the Classical Genizah Period (the end of the tenth to the mid-thirteenth centuries CE) and the fidelity with which they follow that tradition. Common Bibles, I argue, provide a further glimpse into the phonetics of Tiberian Hebrew in this period, as their producers did not always adhere to the strict letter of the written Tiberian tradition, either through choice or ignorance, and the results reveal more about how the users of the text were pronouncing their Hebrew than the correct application of Tiberian graphemes would ordinarily allow. For instance, the substitution of vocalic shewa by a different vowel sign will reveal how the shewa was being pronounced, something normally hidden behind the inscrutable two dots of the sign itself.

2.0. The Corpus

2‘Common Bible’ is the term proposed by Colette Sirat in her Hebrew Manuscripts of the Middle Ages (2002) as one category of a fourfold division of the extant Hebrew Bible manuscript codices from the Muslim lands of the Middle Ages (Sirat 2002, 42–50). The full list is as follows: (a) Great Bibles, fully vocalised and cantillated, with Masoretic notes; (b) Common Bibles, ‘more modest’, usually without masora magna, but “they always have the vowel and cantillation signs”; (c) Bibles with translations; (d) the Bible with Arabic translation and translator’s commentary (e.g., Saʿadya’s Tafsīr or, for Karaites, the commentary by Yefet ben ʿEli). The recent book by David Stern, The Jewish Bible: A Material History (2017), talks about three “distinct generic types of Hebrew Bible” in the Middle Ages, “the Masoretic Bible, the liturgical Pentateuch, and the study Bible,” which categories overlap, but not in contradictory fashion, with Sirat’s (Stern 2017, 88–90). Of relevance too is an earlier study by Goshen-Gottstein (1962) of the range of extant Hebrew Bibles found in the United States, one of the first to attempt to classify the types of biblical manuscript in the Genizah. He distinguishes ‘study codices’ from ‘listener’s codices’ (Goshen-Gottstein 1962, 36–44). His former category is differentiated from Great (Masoretic) Codices by an absence of Masoretic notes, indicating they are “not meant for ‘professional’ usage or to serve as an exact model,” while his latter, the ‘listener’s codices’ (which he estimates form about a half of the Elkan Nathan Adler Collection at the Jewish Theological Seminary, on which he bases his analysis) were meant for “everyday use” and were “not written in order to please future hunters of variant readings and do not represent in any way… any hidden stream of tradition” (Goshen-Gottstein 1962, 38–40). He chose the term ‘listener’s codex’, because these texts in his view supported the congregation in its listening, not its reading, and were “little more than ‘hearing aids’” (Goshen-Gottstein 1962, 40–41). Despite having been written more than half a century ago, Goshen-Gottstein’s assertive impressions are still some of the more insightful on the subject, and the whole article, despite its parochial-sounding title (‘Biblical Manuscripts in the United States’), is a useful one.

3In his survey, Stern (2017, 88) asserts that “the surviving codices are only a fraction of the Hebrew Bibles that once existed, and we do not know how representative they truly are.” While this is arguably true when looking at the Jewish world at large and over time, such is the scale of the biblical manuscript inventory in the Genizah Collection (more than twenty-five thousand pieces in the Taylor-Schechter and Lewis-Gibson Collections in Cambridge alone) that we can be quite confident we have a sense of the biblical landscape at least as it relates to the Eastern Mediterranean in the High Middle Ages (equivalent to the Classical Genizah Period).

4Given all this, particularly that we can see the extent of the inventory and how the different types of biblical manuscript sit in it, I see no harm in adapting these categories to suit the differing kinds of analysis that should be done on them. For my current purposes, I am most interested in fragments with Tiberian vowels that are prone to deviation from the standard orthography and vocalisation of the text. These are going to be found mostly among Goshen-Gottstein’s ‘listener’s codices’, but can also sit among his ‘study codices’. I think it most effective therefore to take Sirat’s broad definition of Common Bibles, i.e., excluding those with translations, commentaries, etc., but further exclude all with Masoretic notes, since these are, by definition, going to be less fruitful in significant deviations from the written tradition (which is what the masora is there to guard against!). This gives us a clear and handily unambiguous distinction between what we can call Masoretic Bibles and a broad category of Common Bibles: those that have Masoretic notes and those that do not.1 For current purposes, therefore, a Common Bible preserves some or all of the biblical text in an extended form (i.e., not including collections of biblical verses for liturgical or homiletical purposes, but including collections of hafṭarot readings); it should not have the masora, in the form of Masoretic notes (masora parva and magna), but may have varying amounts of the rest of the panoply of the Tiberian Masoretic apparatus: vocalisation, cantillation signs, parasha and seder markers, demarcated parashiyyot, and qere/ketiv notations. It happens that Bibles of this type are often of a smaller format, and may be written on parchment or on paper, but the current study is not concerned with the codicological categorisation of Common Bibles, only with their value in the analysis of the Tiberian tradition that they transmit.2

5The Common Bible, under any form of categorisation, has not been the focus of much research. Palestinian and Babylonian-vocalised manuscripts from the Cairo Genizah have been investigated at length, and those with Extended Tiberian too, though to a lesser extent.3 Those, on the other hand, with ‘ordinary’ Tiberian vocalisation have probably been viewed as insufficiently interesting to be worthy of close analysis: the Tiberian is either seen as poorly executed, and therefore too debased a form to be relevant to the study of the tradition itself (hence the appellation ‘vulgar’ sometimes applied to them), or the manuscripts are viewed as too far removed in time from the Masoretic era, from the core Ben Asher tradition. Israel Yeivin, in his Introduction to the Tiberian Masorah (translated by E. J. Revell, 1980) discusses the Bibles of the Cairo Genizah and touches on these points: “Most are fragments of ‘vulgar’ texts, some without Masorah, without accents, with many extra vowel letters, and so on… . MSS written after 1100 contain, as a rule, little of interest to the study of the standard tradition and its development… . They do, however, contain much of value to the study of the development of the tradition up to the time of printing, and also for the study of the pronunciation of Hebrew in different periods and localities” (Yeivin 1980, 30–31). I agree wholeheartedly with his last point, that these manuscripts—though without limiting it to those written after 1100—can be of great interest for the study of the pronunciation of Hebrew, and, in particular, of the pronunciation of the Tiberian tradition as practised by the disparate congregations who made up the Jewish community of Fusṭāṭ, or from further afield, whose discarded manuscripts ended up in the Genizah Collection. In support, I enlist a further assertion from Moshe Goshen-Gottstein (1962, 41) about his ‘listener’s codices’: “This freedom in copying out these texts is of vital importance for our understanding of Hebrew reading traditions and linguistic habits.”

3.0. Scope of the Analysis

6Classification of Tiberian Hebrew Bibles generally relies on a text’s degree of adherence to the standard Masoretic Text, as exemplified in Codex Leningrad (Russian National Library Евр. I B19a) or the Aleppo Codex (Ben Zvi Institute). This is not a new idea, and indeed can be traced at least as far back as Maimonides, who belittled the copies of the Bible in circulation in his day, comparing them unfavourably to the Tāj, which he described as corrected by Ben Asher himself (Mishne Tora, Hilkhot Tefillin, Mezuza ve-Sefer Tora 8.4). We now identify this manuscript with the Aleppo Codex, the production of which was “the great event in the history of the Tiberian Bible text” (Goshen-Gottstein 1963, 86). Such textual perfection is not, however, a useful yardstick to employ when examining the Common Bible on its own terms. While some may have been copied by practised hands from reliable precursor texts, many, as will be shown below, have no such aspirations of rigid adherence to Tiberian norms, let alone Ben-Asherian perfection. Some were used to practise or learn the copying of the text, others to practise or learn Hebrew itself; some were used for recitation, or for learning the text of a hafṭara or festival reading; others perhaps served as ‘lap’ Bibles, books to be held to follow the readings in the service, either for utility alone or as signs of status.

7There is more to be written about the production, ownership and use of Common Bibles, but this is not the focus of the current study. For the moment, I suggest just that as a category it encompasses both user-produced codices (i.e., owned and used by those who originally wrote them), which are probably the majority of the fragments, as well as those created by third parties— relatives, friends, professional or semi-professional scribes. As will be seen, some of these Common Bible fragments are of the highest quality in terms of their production, whereas others are definitely at the ‘barely good enough’ end of things.4

8Given that Common Bibles are so numerous, their value should be self-evident: they form a large body of evidence for ordinary Jewish engagement with the text of the Hebrew Bible in the Middle Ages. But beyond their interest as a cultural artefact of popular religion, their textual value, too, is considerable. That is not to say that they have great importance for textual criticism of the biblical text in its traditional sense. Their frequent departures from the consonantal Masoretic Text can usually be explained by error or analogy, and it is less likely that they somehow preserve ancient or alternate streams of textual transmission.5 They do, though, have a real and unique value for the history of the Hebrew language. Our sources for the pronunciation of Tiberian Biblical Hebrew in this period are few: Masoretic treatises (and the successor works of the medieval grammarians and the more linguistically conscious commentators) and the Karaite transcriptions of the Hebrew Bible into Arabic script. There are not many more sources than those that point to the linguistic reality of Tiberian Hebrew at the end of the first millennium. Among the huge variety of Common Bibles, however, particularly those at the more home-made, budget end of the scale, are those which do not follow the accepted norms of spelling and vocalisation. They provide rare glimpses of how Hebrew was pronounced in the home and synagogue of the High Middle Ages.

9To demonstrate this value, and to present some of the range of Common Bibles preserved in the Genizah, I have selected five different manuscripts from the Additional Series of the Taylor-Schechter Collection. No small selection from the huge Additional Series, which contains around fifteen thousand pieces of biblical manuscript, can be completely representative of the inventory at large, but the aim is to give a sense of the different types of Common Bible, as well as show their potential interest for the transmission of the Tiberian tradition. To that end, all the manuscripts selected have Tiberian vocalisation and some have cantillation too. Below, I analyse each from a textual and linguistic standpoint, focusing the analysis on the phonetics behind the orthography and vocalisation. There is no detailed palaeographic or codicological description. In general, the majority of manuscripts in the Taylor-Schechter Collection come from the High Middle Ages, and most were probably produced in Egypt for and by the congregation who used the Synagogue of the Palestinians in Fusṭāṭ. A substantial number of Genizah manuscripts fall outside those temporal and geographical limits, but I have not chosen any that are clearly late (fourteenth century onwards) or obviously ‘foreign’ (such as in Spanish or Yemeni hands, frequent interlopers in the Genizah). The manuscripts featured here are more likely to be from the period between the end of the tenth and the middle of the thirteenth centuries, and are likely to have been produced in Egypt, Syria-Palestine, or eastern North Africa.

10The fragments under analysis are all from Cambridge University Library’s Taylor-Schechter Collection of Genizah fragments. All were catalogued (in very terse fashion) in Davis and Outhwaite’s (2003) catalogue of the Additional Series, but have otherwise not been published.

T-S AS 44.35, a bifolium of Lamentations
T-S AS 68.100, a leaf of Psalms
T-S AS 53.90, a leaf of Kings and Ezekiel
T-S AS 5.144, a leaf of Leviticus
T-S AS 59.215, a bifolium of Proverbs

4.0. Codex of Lamentations, T-S AS 44.35

4.1. Description

11The Cairo Genizah manuscript Cambridge University Library T-S AS 44.35 is a small-format paper bifolium containing continuous text from Lam. 2.13–18; 3.51–4.2. As the gap between the content of the two folios suggests, it was probably from a copy of the whole book, rather than just an excerpt. Lamentations is read in the evening service of Tishʿa beʾAv ‘the ninth of Av’ (Elbogen 1993, 108), and individual copies of the book or of all the Megillot together can be found in the Genizah. T-S AS 44.35 is fully furnished with Tiberian vowels, but there are no cantillation signs, and no masora.6 The divine name, in the form of the Tetragrammaton, is written in full. Consonants and vowels are in the same ink and, most likely, the same hand.7 The text does not seem to have been ruled, and the left-hand margin is quite ragged, though there are some line fillers and elongated letters. Perhaps these are more for effect than actual utility. The writing fills most of the page, with minimal space left for margins. The execution of certain letter shapes and vowel signs is unusual: most notably qibbuṣ is often reversed, with the three dots sloping up from left-to-right.

12The system of Tiberian vocalisation used in T-S AS 44.35 is idiosyncratic, but appears to behave consistently within its own rules, as far as these can be discerned. The most obvious feature of the vocalisation is that silent shewa is usually not marked unless it falls under one of the bgdkpt consonants, where it probably serves to mark that the consonant has spirant, i.e., fricative, pronunciation. Vocalic shewa is frequently replaced by pataḥ. Full vowels occur in place of ḥaṭefs. Dagesh (lene or forte) is absent, as is rafe.8 No dot distinguishes the consonants śin and shin. Sof pasuq ( ׃ ), as part of the consonantal text, occurs at the end of verses; maqqef, as part of the accentuation system, is not used.9 The vowel u, regardless of length, is usually marked with a digraph or

4.2. Consonantal Text of T-S AS 44.35

13T-S AS 44.35 does not slavishly follow the Masoretic Text, although there are sufficient defective forms to show some awareness of and fidelity to the basic consonantal form of the text. For instance, ‘my enemy’ (Lam. 3.52) is defective in the fragment and the MT.10 Where the form is plene in the MT, at Lam. 2.17, the fragment is too: ‘enemy’. However, there are numerous differences, with the following plene spellings all defective in the MT:

‘passers-by’ (Lam. 2.15); ‘you said’ (Lam. 3.57); ‘you have redeemed’ (Lam. 3.58); ‘and shake’ (Lam. 2.15);11 ‘beauty’ (Lam. 2.15), etc.

14The reverse occurs rarely in T-S AS 44.35; only the following defective forms are plene in the MT:

‘your enemies’ (Lam. 2.16); ‘How? ’ (Lam. 4.1)

15There is obviously a greater tendency towards the use of matres lectionis, but not a complete departure from the consonantal tradition behind the MT.

16The precedence of the oral tradition over the consonantal can be seen in the frequent ellipsis or replacement of quiescent ʾalef, where the text presents a more phonetic, rather than historical, spelling, e.g.,

‘he will heal’ (Lam. 2.13); ‘their head’ (Lam. 2.15); ‘and not’ (Lam. 2.17); ‘my head’ (Lam. 3.54); ‘I called’ (Lam. 3.55) with ʾalef added above the line; ‘do not fear’ (Lam. 3.57); ‘at the head of’ (Lam. 4.1)

17The spelling of the MT’s (Lam. 2.14) as corrected above the line with או and written correctly as אוש on its second occurrence in the verse, also reflects the more phonetic impulses of the scribe, confusing the two homophonous consonants.12 Similarly ‘their thoughts’ (Lam. 3.60), where the ב was inserted only as an afterthought, is probably symptomatic of the same confusion.

18The substitution of the Tetragrammaton twice, in Lam. 2.18 and 3.58 (written the second time , where the MT on both occasions has ינדא , similarly underlines the oral nature of this transcription, suggesting that it was not copied from a written exemplar, but taken down from memory or from dictation.

19Beyond the interchange of vowel letters, T-S AS 44.35 shows two minor consonantal differences from the MT:

‘the whole earth’ (Lam. 2.15); ‘from my cry’ (Lam. 3.56)

20The text follows the MT qere with ‘your captivity’ (Lam. 2.14). There is an obvious dittography in ‘the best’ (Lam. 4.1), where the scribe recognised their error and did not vocalise the repeated word.

21The evidence of the consonantal text of T-S AS 44.35 is that the scribe who produced it, though possessing familiarity with the general spelling conventions of the MT, certainly did not meticulously following a Masoretic Vorlage. The more phonetic elements, in particular the ellipsis of quiescent alef, show the pervasive influence of the reading tradition, that is, of the oral recitation, which tends often in the fragment to override the spelling conventions of standard Biblical Hebrew.

4.3. Shewa in T-S AS 44.35

22Further evidence of the influence of the oral component in the text’s composition can be seen in its approach to marking the shewa sign, sparsely used in the text. Where shewa occurs on non-bgdkpt consonants and is silent in the MT, no sign is written, e.g.,

‘we have swallowed’ (Lam. 2.16); (pausal, MT ‘I am cut off’ (Lam. 3.54); ‘their thoughts’, with pataḥ in place of the MT’s vocalic shewa (Lam. 3.60)

23While no standard Masoretic codex of the Bible follows this practice, the occasional elision of silent shewa can be found even in the best manuscripts. The Aleppo Codex, for example, exhibits at least three words where the naqdan, possibly Aharon ben Moshe ben Asher himself, has forgotten to write silent shewa, for instance on ‘inside him’ (Job 20.14; Yeivin 1968, 16). However inadvertent it was in that meticulously vocalised manuscript, the elision of silent shewa in T-S AS 44.35 can nevertheless be seen as the natural culmination of an understandable tendency to ignore or forget a ø vowel.

24In contrast, silent shewa is often marked in the fragment on vowelless bgdkpt consonants, where possibly its primary purpose was not to indicate the ø vowel, but to mark the fricative pronunciation of the consonant, e.g.,

‘passers-by’ (Lam. 2.15); ‘to my soul’ (Lam. 3.51) with pataḥ for vocalic shewa; ‘you have pleaded’ (Lam. 3.58); ‘judge! ’ (Lam. 3.59); ‘the lips of’ (Lam. 3.61)

‘this’ (Lam. 2.15); ‘the perfection of’ (Lam. 2.15); ‘burdens’ (Lam. 2.14); ‘the best’ (Lam. 4.1)

25This extended use of shewa is perhaps most evident when it occurs on the final consonant of a word:

26Though contrary to standard Tiberian practice in the marking of the sign, this still accords with Tiberian pronunciation, where word-final shewa is usually silent (Khan 2013a, 100).13

27There are a number of exceptions in the application of these apparent rules by the scribe of T-S AS 44.35. Silent shewa is not always marked on vowelless fricatives:

‘they clap’ (Lam. 2.15); ‘you came near’ (Lam. 3.57); ‘they are poured out’ (Lam. 4.1); ‘streets’ (Lam. 4.1); ‘as jars’ (Lam. 4.2)

28But given that this is an informal reworking of their system, we should not expect the same rigour as that exhibited by the Masoretes.

29Shewa also occurs occasionally on vowelless non-bgdkpt consonants, for instance ‘vain’ (Lam. 2.14), a rare case of complete fidelity to the historical MT spelling, but more unexpectedly on ‘the joy’ (Lam. 2.15),14 ‘the day’ (Lam. 2.16) and ‘from a pit’ (Lam. 3.55). These are all, as above, on a final vowelless consonant. Medially, the text shows variance in regard to mem, however, with ‘they said’ (Lam. 2.16), showing shewa, but‘they have made an end’ (Lam. 3.53) and ‘your name’ (Lam. 3.55) both eschewing it.

30Shewa on ‘do not hide’ (Lam. 3.56) similarly marks what is a silent shewa in the MT. Given the text’s general approach to shewa, the use here probably serves to underline that the ʿayin is vowelless [ʔal taʕˈleːm]. No shewa occurs on vowelless ʿayin in ‘you have heard’ (Lam. 3.61), however.

4.4. Pataḥ for Shewa in T-S AS 44.35

31T-S AS 44.35 usually puts pataḥ where we find a simple vocalic shewa in the MT. This is in accordance with the Tiberian pronunciation tradition’s rendering of vocalic shewa as a short [a], equivalent in quality to a pataḥ (Khan 2013a, 98). The scribe does not use ḥaṭef pataḥ for this purpose as no ḥaṭefs occur in the fragment at all:

‘Jerusalem’ (Lam. 2.13) not pausal in the MT, the fragment has qameṣ for MT’s pataḥ; (MT ‘that I may comfort you’ (Lam. 2.13); (MT ‘virgin of’ (Lam. 2.13); ‘your prophets’ (Lam. 2.14); ‘and not’ (Lam. 2.14); ‘your captivity’ (Lam. 2.14); (MT ‘Jerusalem’ (Lam. 2.15), showing pataḥ for the MT’s pausal qameṣ; ‘the perfection of’ (Lam. 2.15); ‘and not’ (Lam. 2.17); ‘the daughters of’ (Lam. 3.51); ‘your ear for my relief’ (Lam. 3.56); ‘their thoughts’ (Lam. 3.60); ‘their rising up’ (Lam. 3.63); ‘their hands’ (Lam. 3.64); ‘in anger’ (Lam. 3.66); ‘the heavens of’ (Lam. 3.66); ‘children of’ (Lam. 4.2); ‘the hands of’ (Lam. 4.2)

32In every case in the fragment where the standard Tiberian pronunciation of shewa is equivalent to a short [a], the scribe uses a pataḥ rather than a shewa. In a text that is not emulating the MT to a great degree, it should not be a surprise, given that the chief distinction between shewa and pataḥ is morphophonological and not phonetic (shewa cannot, under most circumstances, form a syllable in Tiberian Hebrew). This distinction was evidently of little significance to the writer of this manuscript.

4.5. Shewa before Yod or Guttural in T-S AS 44.35

33Where shewa occurs before yod in the MT, T-S AS 44.35 has a ḥireq: ‘on the day’ (Lam. 3.57), [biˈjoːm]. This reflects the standard Tiberian pronunciation of shewa before yod as a short [i] (Khan 2013a, 98). It is also a feature that is found, although with great inconsistency, in Palestinian-vocalised manuscripts (Revell 1970a, 90; Heijmans 2013a, §3f).

34Before a guttural, shewa is pronounced with the quality of the vowel following the guttural (Khan 2013a, 98). This is reflected in the fragment in ‘to turn away’ (Lam. 2.14), which ignores the technicalities of syllable structure and prefers qameṣ to shewa, [lɔhɔːˈʃiːv].

4.6. Shewa on the First of Two Identical Consonants in T-S AS 44.35

35Unless adjacent to another shewa or under a geminated consonant, shewa in the middle of a word is usually silent in the standard Tiberian reading tradition (Yeivin 1980, 277; Khan 2013a, 99–100). Masoretic treatises, including Aharon ben Moshe ben Asher’s Diqduqe haṭ-Ṭeʿamim, present a number of exceptions to this rule, one of which is when the shewa occurs on the first of two identical consonants after a long vowel (Dotan 1967, I: 115– 16 [§5]; Yeivin 1980, 280–81). In many cases these are marked with ḥaṭef pataḥ in the Aleppo Codex and occasionally in Leningrad.15 In similar fashion T-S AS 44.35 reflects the vocalic nature of this shewa, but as we might expect by now, a full pataḥ is used in preference to a ḥaṭef, ‘affects’ (Lam. 3.51), [ʕoːlaˈlɔː].

4.7. Ḥaṭef in T-S AS 44.35

36There are no ḥaṭef signs in T-S AS 44.35. A full vowel is used in place of ḥaṭef wherever it occurs in the MT, e.g.,

‘that I may comfort you’ (Lam. 2.13); ‘according to the work’ (Lam. 3.64); ‘your curse’ (Lam. 3.65)

4.8. Differences in Vowel Quality in T-S AS 44.35

37In Lam. 2.14  ‘and they have seen’, the fragment replaces both the MT’s ḥaṭef segol and segol with ṣere, apparently giving [vay (y) eːḥeːˈzuː]. This can be seen more widely throughout the text: it frequently replaces Tiberian segol with ṣere, particularly in the nominal ֶֻֿ ה - ending and especially in segolate forms, e.g.,

‘I shall compare’ (Lam. 2.13); (MT ‘I shall make equal’ (Lam. 2.13); ‘way’ (Lam. 2.15); ‘the land’ (Lam. 2.15); ‘that we hoped for’ (Lam. 2.16); (MT ‘his word’ (Lam. 2.17); ‘which’ (Lam. 2.17); ‘old’ (Lam. 2.17); ‘horn’ (Lam. 2.17); ‘a stone’ (Lam. 3.53); (MT ‘I call you’ (Lam. 3.57); ‘their reproach (Lam. 3.61); ‘fine gold’ (Lam. 4.1); ‘they are considered’ (Lam. 4.2); ‘earthen vessel’ (Lam. 4.2)

38Segol is replaced by ṣere in both stressed and unstressed syllables. However, segol is not avoided altogether in T-S AS 44.35: ‘their mouth’ (Lam. 2.16) and ‘this’ (Lam. 2.16) both retain segol. Furthermore, it is found in place of the MT’s ṣere on two occasions: ‘he has raised’ (Lam. 2.17); and ‘do not hide’ (Lam. 3.56). In both cases the vowel exchange is on a guttural ( ע , ה ) in a verbal form, once each on an unstressed and a stressed syllable. The construct noun ‘the work of’ (Lam. 4.2) preserves the MT’s ṣere. In general, the frequent interchanges and evident confusion are suggestive of the influence of the Palestinian pronunciation of Hebrew, i.e., the Sefardi-Palestinian reading tradition, where the two vowels e and ɛ have merged (Henshke 2013b). If this is the case, then we should also expect to see evidence of a merging of the vowels a and ɔ.

39Confusion between a and ɔ in T-S AS 44.35 is mostly found in pausal forms, where the text substitutes pataḥ for the MT’s pausal qameṣ:16

‘Jerusalem’ (Lam. 2.15); ‘he devised’ (Lam. 2.17); ‘he pitied’ (Lam. 2.17)

40Rather than a general merging of the vowels, this may instead reflect a loss of distinct pausal forms in the recitation that sits behind this fragment, although we do find pausal qameṣ in accordance with the MT at Lam. 3.54  ‘I am cut off’. This is a major, verse-final, pause, though, whereas the previous examples were all mid-verse (i.e., at atnaḥ) or minor pause (at reviaʿ), and perhaps therefore elided through lax recitation. The qameṣ in non-pausal ‘Jerusalem’ (Lam. 2.13), however, points at a greater degree of confusion in the scribe’s pronunciation. Similarly, the ō vowel in [jaroːʃɔːˈlɔːyim] might suggest some phonetic overlap between u and o, such as can also be found in Palestinian Aramaic pronunciation (Yahalom 1997, 18). However, the ū vowel is retained in all other cases, even in the same word when it occurs two verses later Lam. 2.15), and therefore a scribal lapse, due to the casual nature of the work, is more likely.

4.9. T-S AS 44.35 in Conclusion

41Consonantally, the Lamentations manuscript deviates from the MT in its plene orthography and particularly in its frequent ellipsis of quiescent ʾalef. The substitution of the Tetragrammaton for the MT’s ינדא on two occasions suggests that it may not have been copied from a Vorlage at all, but produced from dictation. Its vocalisation diverges greatly from that of the MT, but in a consistent, logical manner. Indeed, for a fragment that looks very casual in its execution—the work of an individual for their own purposes—the text is very consistent in its vocalisation. The shewa is sparsely used and serves a secondary purpose of marking the fricative pronunciation of bgdkpt consonants. Vocalic shewa is replaced with pataḥ in most circumstances, with ḥireq when preceding yod, and with a full vowel before a guttural. The ḥaṭef is ignored entirely as an irrelevance. A more significant divergence from standard Tiberian is found in vowel quality, with a slight blurring of the distinction between, respectively, the u and o, a and ɔ, and, to a much greater degree, e and ɛ vowels. It could be ascribed to the influence of a background Palestinian reading tradition, an example of Palestino-Tiberian vocalisation, but is only consistently apparent in the e/ɛ vowels.

42In other respects, the pronunciation reflected in the vocalisation accords with that of the standard Tiberian reading tradition. This includes even the more potentially problematic renderings, such as the correct pronunciation of the first of two identical consonants after a long vowel. The absence of cantillation signs might suggest that either the correct cantillation was well known to the user of the book or else it was irrelevant to its intended use. The absence of dagesh calls for an explanation. The use of silent shewa apparently to mark the fricative pronunciation of the bgdkpt consonants means that dagesh would serve a purpose only of indicating consonantal length. That it is not used at all suggests that the length of consonants, like the length of vowels, was not of primary interest to the creator of this fragment and may not have been discernible in their reading of it.

5.0. Codex of Psalms, T-S AS 68.100

43Cambridge University Library T-S AS 68.100 is a fragment on parchment containing Ps. 119.72–92, with stichometric spacing of the text, sof pasuq at the end of verses, and the Tetragrammaton written in full. It is vocalised and cantillated, though the silluq accent is not marked. There is no evidence of additional Masoretic paratext. Rafe is used on the bgdkpt letters and there are some (musical) gaʿyot. The vowels and accents are written in a different ink and with a different pen from those of the consonants. It has the appearance of a leaf from a good quality codex, the work of at least two hands, a scribe (responsible for consonants and sof pasuq) and a vocaliser (vowels and accents), though not a full Masoretic Bible.17

44Vocalisation aside, the most noticeable difference from the MT is in the use of accents: silluq is absent, and the prepositive disjunctive deḥi occurs regularly for reviaʿ mugrash and once for conjunctive merkha. While the ellipsis of silluq is a feature shared by Extended Tiberian manuscripts, the swapping of reviaʿ mugrash for deḥi is not (Díez-Macho 1963, 22–24). The lack of silluq may therefore be seen as a general feature of non-Masoretic manuscripts, a symptom of a tendency towards the loss of inessential elements (after all, sof pasuq is already present to mark the last word of the verse), rather than a specific pointer of this text’s affinity with the Extended Tiberian family of manuscripts.18

5.1. Consonantal Text of T-S AS 68.100

45Consonantally, the text of T-S AS 68.100 is in accordance with the MT, e.g., it shares the defective forms ‘let them come to me’ (Ps. 119.77) and ‘let them be ashamed’ (Ps. 119.78). The only exception is ‘may they return’ (Ps. 119.79), where both Aleppo and Leningrad have the plene spelling.

5.2. Pataḥ for Shewa in T-S AS 68.100

46Like the Lamentations manuscript, T-S AS 68.100 does not use ḥaṭef vowels. It also prefers pataḥ to the MT’s simple shewa, suggesting an uncoloured pronunciation of vocalic shewa as [a], e.g.,

‘your kindness’ (Ps. 119.76); (MT ‘let them come to me’ (Ps. 119.77); ‘they subverted me’ (Ps. 119.78), without dagesh forte; ‘against those who pursue me’ (Ps. 119.84); ‘and I will keep’ (Ps. 119.88); ‘your word’ (Ps. 119.88)

47As several of the examples above show, silent shewa is usually not represented in T-S AS 68.100, e.g.,

‘your kindness’ (Ps. 119.76); (MT ‘for your deliverance’ (Ps. 119.81); (MT ‘your word’ (Ps. 119.88); ‘and I will keep’ (Ps. 119.88); ‘in my affliction’ (Ps. 119.92)

48The shewa sign is used in T-S AS 68.100 for a vocalic shewa occasionally, e.g., under an initial consonant:

‘those who fear you’ (Ps. 119.79); ‘in your laws’ (Ps. 119.80)

49In ‘for my comfort’ (Ps. 119.76), ‘your mercies’ (Ps. 119.77), and ‘you will act’ (Ps. 119.84), its occurrence on the guttural could be ambiguous, since in the standard Tiberian system simple shewa under חהא " ע is silent, never vocalic (Levy 1936, 21* and גכ ll. 12– 14). Given, however, that the shewa sign is not used for silent shewa elsewhere in T-S AS 68.100, we should on balance consider it vocalic here too, marked in contravention or ignorance of the standard Tiberian practice. Compare ‘that I may live’ (Ps. 119.77), where the ø vowel of ḥet is not marked.

50Given this, the shewa under nun in (for MT ‘and they made me’ (Ps. 119.73) is probably intended to be vocalic. This is in keeping with the Tiberian rule that shewa under the first of two identical consonants following a long vowel is vocalic. The shewa gaʿya under waw, for the MT’s pataḥ gaʿya, is reflective of the interchangeability of the two a vowels, shewa and pataḥ, in this fragment (a further example is noted below). It represents only graphic divergence from the MT’s practice. The pataḥ under the yod, however, shows a clear difference from the MT, as it reflects a pronunciation of the MT’s silent shewa as vocalic here [vaːyḵoːnaˈnuːniː] > [vaːyaḵoːnaˈnuːniː] (the gaʿya is a minor gaʿya, i.e., lengthening a closed syllable). This, in a complex multi-syllable word, however, is the only example in the fragment of a clear divergence in pronunciation from the standard Tiberian tradition.

5.3. Ḥaṭef in T-S AS 68.100

51Where the MT would use ḥaṭef pataḥ, e.g., for vocalic shewa under gutturals, T-S AS 68.100 can use a pataḥ, e.g.,

‘give me understanding’ (Ps. 119.73); ‘that’ (Ps. 119.85); ‘but I’ (Ps. 119.87); ‘your servants’ (Ps. 119.91)

52Or the fragment uses a simple shewa, e.g.,

‘for my comfort’ (Ps. 119.76); (MT ‘your mercies’ (Ps. 119.77); ‘you will act’ (Ps. 119.84)

53Further evidence for the vocalic pronunciation of the shewa sign in general in T-S AS 68.100 can be found in ‘and it stands’ (Ps. 119.90), where simple shewa not only stands for a vocalic shewa under the ʿayin, but also substitutes, in the form of shewa gaʿya, for the MT’s pataḥ gaʿya under the conjunction—another minor gaʿya.

5.4. Shewa before guttural in t-s as 68.100

‘according to your promise’ (Ps. 119.76); ‘let my heart be’ (Ps. 119.80); ‘and I will keep’ (Ps. 119.88), with silent shewa unmarked and a pataḥ for MT vocalic shewa; ‘in my affliction’ (Ps. 119.92)

54Where vocalic shewa immediately precedes a חהא " ע consonant, T-S AS 68.100 substitutes a full vowel, e.g.,

55The pronunciation represented by this combination of vowels accords with the realisation of shewa before a guttural in the Tiberian tradition, which matches the quality of the vocalic shewa to that of the guttural following it, unless the shewa itself sits under a guttural (Yeivin 1980, 281–82; Khan 2013a, 98–99).

5.5. T-S AS 68.100 in Conclusion

56Altogether a different kind of manuscript from the first example, the Psalms fragment has been carefully produced, probably by two different hands. Consonantally, it is very close to the MT of Leningrad and Aleppo, with only one minor divergence. In accents, it diverges slightly, with a greater use of deḥi and the absence of silluq. Vocalically, it preserves the standard Tiberian phonology, with only one minor syllabic difference in the word (Ps. 119.73). This is revealed particularly through the apparent free substitution of simple vocalic shewa with pataḥ as well as through the substitution of various contextually conditioned shewa vowels (e.g., before gutturals) with the corresponding full vowel sign. The naqdan of this fragment was wholly familiar with the Tiberian reading tradition.

6.0. Hafṭara Lectionary, T-S AS 53.90

57Cambridge University Library T-S AS 53.90 preserves the text of 1 Kgs 3.25–28 and Ezek. 37.18–21. A torn paper manuscript, it shows no ruling, and the left-hand margin is kept only irregularly, with no elongation of letters or line-fillers. It is partially vocalised: on recto, 1 Kings has only a few words with Tiberian vowel signs; on verso, Ezekiel is almost completely vocalised. There is no cantillation. The vocalisation is in the same ink as the consonantal text, most likely the work of the same hand. Dagesh, and the śin and shin dots are not marked, though there is an occasional rafe. Sof pasuq is used at the end of a verse. The Tetragrammaton is written in full.

58The fragment contains two hafṭarot according to the annual reading cycle of the Torah, for the parashot Miq-qeṣ ( ץקמ , no. 10, Gen. 41.1–44.17), with its hafṭara from 1 Kgs 3.15–4.1, and Wayyiggaš ( שגיו , no. 11, Gen. 44.18–47.27), with its reading from Ezek. 37.15–28. On recto there is a partially preserved rubric before the start of the hafṭara: שגיו ] וילא ריטפמ [ ... זחיב ] לאק [‘ [ “And he approached] him” one concludes [with the reading in Eze] kiel’. The rubric confirms that the fragment is a lectionary of prophetic readings, although its original extent—whether it covered just a small number of texts, or was part of a more comprehensive work—cannot now be determined. The casual nature of its construction suggests the former.

6.1. Consonantal Text of T-S AS 53.90

59There are a few corrections in the fragment. The whole top line on recto (1 Kgs 3.25 from תאו יצחה to 3.26  לא ךלמה יכ ) appears to be an addition, perhaps in a different hand, and ורמכנ ‘they yearned’ (1 Kgs 3.26) is represented only by כנ in the right-hand margin. On verso, the scribe spotted the error ךל and crossed it through before writing the correct form ךילא ‘to you’ (Ezek. 37.18).

60The text exhibits a tendency towards more matres lectionis than are found in the MT:

‘my lord’ (1 Kgs 3.26); ‘and make them’ (Ezek. 37.19)—both are defective in the Aleppo and Leningrad codices

61But some MT defective forms are retained, e.g., בתכת (MT ‘you will write’ (Ezek. 37.20). The interrogative אלה ‘will you not?’ (Ezek. 37.18) is defective in the fragment, but plene in the MT

62וירבח ‘his fellows’ (Ezek. 37.19) follows the MT’s qere. At 1 Kgs 3.27 the fragment has לא והתימת ‘do not kill him’ for the MT’s This reading is probably influenced by the phrase earlier in 1 Kgs 3.26.19

6.2. Vocalisation of T-S AS 53.90

63Most of 1 Kings is unvocalised, perhaps because it was a familiar text that posed little difficulty in its reading. The addition of a qibbuṣ to ‘do not kill him’ (1 Kgs 3.26) is understandable, since the 2mpl verb is written defectively, as in the MT. But the vowels on ‘in front of the king’ (1 Kgs 3.28) appear superfluous, given the commonplace nature of the words. From this point on, however, the text is mostly vocalised.

64No dagesh, forte or lene, is written, even in the fully vocalised portion of the text, e.g.,

‘your people’ (Ezek. 37.18); ‘speak’ (Ezek. 37.19)

65Rafe, however, is occasionally used to mark the spirant pronunciation of bgdkpt consonants, e.g.,

66 ‘to do’ (1 Kgs 3.28)

6.3. Ḥaṭef in T-S AS 53.90

67The fragment eschews ḥaṭef signs completely, preferring pataḥ in every case where we would expect ḥaṭef-pataḥ:

68 ‘is it not? ’ (Ezek. 37.18); ‘I’ (Ezek. 37.19); ‘which’ (Ezek. 37.19); ‘and make them’ (Ezek. 37.19); (MT ‘on them’ (Ezek. 37.20)

6.4. Shewa before Yod in T-S AS 53.90

69Although shewa, both vocalic and silent, is used in the fragment, e.g., ‘and the tribes of’ (Ezek. 37.19), on the three occasions in the text that it directly precedes yod, a more phonetic transcription with ḥireq occurs:

70 ‘in the hand of’ (Ezek. 37.19); ‘in my hand’ (Ezek. 37.19); ‘in your hand’ (Ezek. 37.20)

71This pronunciation of shewa before yod as an i vowel is reflective of Tiberian pronunciation (Khan 2013a, 98), if not the practice of standard Tiberian vocalisers. It is quite frequent in non-Masoretic Bible texts from the Genizah, as can be seen from its use in T-S AS 44.35 above. Manuscripts with Palestinian vowel signs, too, can place a Palestinian i vowel before yod, where standard Tiberian would have a shewa, though inconsistently (Revell 1970a, 90; Heijmans 2013a, §3f).

6.5. Differences in Vowel Quality in T-S AS 53.90

72The vocalisation exhibits a small number of qualitative differences from standard Tiberian pronunciation, with pataḥ occasionally replacing segol in the environment of the gutturals:

‘Ephraim’ (Ezek. 37.19); ‘one’ (Ezek. 37.19)

73However, ‘the wood’ (Ezek. 37.19) shows that a distinction between segol and ṣere is maintained. ‘behold’ (Ezek. 37.19) has e in place of i in a closed, unstressed syllable, a pronunciation found in the Palestinian vocalisation tradition (Heijmans 2013a, §3d), but possibly also reflecting the common realisation of closed, unstressed /i/ as a central vowel [e] in various Sefardi reading traditions, such as Baghdad, under the influence of the Arabic vernacular (Ya‘aqov 2013; Shatil 2013).

6.6. T-S AS 53.90 in Conclusion

74The fragment is a small paper hafṭara lectionary, only partially vocalised and with no cantillation, a more casual piece of work than the preceding examples. Dagesh is entirely ignored, perhaps indicating a disinterest in consonantal length, but the use of rafe shows the fricative versus plosive distinction is probably maintained. Ḥaṭef vowels are generally avoided, and ḥireq is used for shewa before yod. The interchange of some vowels could be indicative of a different background pronunciation from the Tiberian, but they mostly reflect the lowering of the ɛ vowel in the guttural environment.

7.0. Writing Exercise, T-S AS 5.144

75Cambridge University Library T-S AS 5.144 contains the text of Lev. 18.11–23 and 18.25–19.3, written on both sides of a parchment leaf. There is no evidence of ruling. The text includes Tiberian vowels and cantillation signs, and verse endings are marked with sof pasuq. There are no further Masoretic signs. The Tetragrammaton is abbreviated. Given the divine abbreviation, the lack of masora and the fact that the text of Leviticus is the most frequently used book of the Bible for learning to write Hebrew (Olszowy-Schlanger 2003, 65), the fragment is probably a writing exercise.

76Dagesh is not used, either forte or lene, nor does mappiq occur where it is expected (which is frequently in this part of Leviticus), e.g.,

ַ ‘your mother’ (Lev. 18.13); ‘his wife’ (Lev. 18.14); ‘her nakedness’ (Lev. 18.15); ‘her daughter’ (Lev. 18.17)

77An erroneous mappiq appears in ‘a woman’ (Lev. 18.19).

7.1. Consonantal Text of T-S AS 5.144

78There are only a small number of differences from the MT in the consonantal text of the fragment, two plene forms for the MT’s defective, and one defective for the MT’s plene:

‘her sister’ (Lev. 18.18); ‘in her life’ (Lev. 18.18); ‘your neighbour’ (Lev. 18.20)

79Elsewhere the text remains close to the MT in the use of matres lectionis, e.g., ‘abominations’ (Lev. 18.20).

80There is one uncorrected error, an ellipsis of םכיהלא in Lev. 18.30 after the divine name. The construct תשא is corrected to השא in Lev. 18.17.

81The text of Leviticus included in the fragment contains a number of parashiyyot, i.e., paragraph breaks, of which only one is marked in the fragment: there is a space after the end of Lev. 18.30, which is a parasha petuḥa in the MT (Leningrad) and in Maimonides’ list in the Mishne Tora (Tefillin, Mezuza ve-Sefer Tora 8). Although the fragment is torn, causing the loss of the rest of the line, an open paragraph should start on the next line. The next line, however, begins with the final word of Lev. 19.1. Therefore the fragment does not follow the usual method of noting an open paragraph, and either treats it as a closed paragraph (parasha setuma), which would leave a space within a line, or just leaves an indeterminate amount of space without strict adherence to the usual medieval format of the open paragraph. No space is left for the closed paragraphs (setumot) at Lev. 18.15, 16, and 17. Similarity to the layout of the MT is therefore more superficial than rigorous.

7.2. Ḥaṭef and Shewa in T-S AS 5.144

82The text of Leviticus in T-S AS 5.144 uses only full vowels and shewa, with no ḥaṭefs. Pataḥ and segol always substitute for their ḥaṭef equivalents, e.g.,

‘her sister’ (Lev. 18.18); ‘your neighbour’ (Lev. 18.20); ‘your God I am’ (Lev. 18.21); ‘you will do’ (Lev. 18.26); ‘who’ (Lev. 18.27)

83Full vowels also sometimes occur in place of simple (vocalic) shewa:

‘in impurity of’ (Lev. 18.19); (MT ‘and they will be cut off’ (Lev. 18.29)

84This includes the use of a pataḥ for vocalic shewa under the first of two identical letters following a long vowel:

‘in your midst’ (Lev 18.26)

85Conversely, shewa twice occurs in place of pataḥ, again demonstrating the pronunciation of shewa as an a vowel:

‘when you defile’ (Lev. 18.27); ‘speak’ (Lev. 19.2)

86Unexpectedly, shewa and ḥireq interchange in (for MT ‘your brother’ (Lev. 18.16), though the repetition of the same form later in the verse preserves a ḥireq,

7.3. Differences in Vowel Quality in T-S AS 5.144

87T-S AS 5.144 exhibits some variation from the MT in the interchange of ɛ and a vowels. Forms of ‘nakedness’ regularly have pataḥ in place of segol under the ʿayin, e.g.,

‘nakedness’ (Lev. 18.15); ‘her nakedness’ (Lev. 18.15)

88Pataḥ occurs similarly under ʾalef in ‘the native’ (Lev. 18.26). All of these presumably reflect the lowering of ɛ under חהא " ע . An oddity, perhaps reflecting an uncertainty over ɛ and a, is found in ‘you will reveal’ (Lev. 18.14). Occurrences of similar forms show pataḥ in each case, however: (Lev. 18.13); (Lev. 18.15). In (Lev. 18.15), damage obscures the vowel under the preformative, so it is unclear whether this is a morphological variant, or whether the shewa substitutes for pataḥ in a closed syllable. A further case of ɛ for a in a non-guttural environment occurs in (for MT ‘in impurity of’ (Lev. 18.19).

89There is variation in the vocalisation of conjunctive waw before the labial מב " ף consonants:

‘and her daughter’ (Lev. 18.17); but ‘and from among your offspring’ (Lev. 18.21)

90Conjunctive waw before shewa has no vowel:

‘and you shall keep’ (Lev. 18.30)

91Perhaps the student was flummoxed at this point.

7.4. T-S AS 5.144 in Conclusion

92We can question the competence behind the production of this fragment: it is probably a writing exercise, rather than a Bible proper. There are indeed a number of errors. But it does display also, to varying degrees, some of the trends found in the other fragments described above: the redundancy of dagesh, the replacement of ḥaṭef with the equivalent full vowel, and pataḥ in place of simple shewa. Where it differs from the others is in a more frequent interchange of ɛ and a vowels, mostly in the environment of gutturals. This may be best ascribed to the writer’s status as a language learner and is possibly influenced by their Arabic vernacular, with the imāla, i.e., the raising of a to ɛ, attested in vocalised Judaeo-Arabic texts from the Genizah (Wagner 2010, 63), being a possible culprit.

8.0. Study Bible, T-S AS 59.215

93T-S AS 59.215 is a paper bifolium containing Prov. 27.27–28.21 and 30.7–24 from a small-format codex. It is fully vocalised, with cantillation and (phonetic) gaʿya, and it contains a number of paratextual Masoretic features, including a seder sign and spaces marking the parashiyyot. The script is square, the paper was ruled, and there are line fillers consisting of the first letter, or letters, of the following word. The vocalisation and cantillation are in the same ink as the consonantal text, however, suggesting that one hand produced the whole work. It can be classified as a good-quality small-format Bible, intended for private study or as a ‘lap Bible’.

94The margin contains a decorated seder marker (at Prov. 28.16) as well as four qere readings (some of which are in different ink and perhaps in different hands). They are marked with the masoretic circle in the body of the text and by ק in the margin. They fall at Prov. 28.10 ( ,) ותוחשב 30.10 ,) וינדא ( 30.15 עבראו ( ), and 30.18 הע (  for העבראו ). The instances of qere at Prov. 28.10 and 30.15 are not noted in Codex Leningrad or the Aleppo Codex, and עבראו at 30.15 reflects a consonantal difference from the text of both of these codices (which have עברא ). There is perhaps sufficient variation in the script and ink to suggest that these two instances of qere might be the work of a subsequent corrector of the fragment, who used the device to correct the text, rather than to record ketiv and qere in the standard sense.

95The parashiyyot petuḥot at Prov. 28.16, 30.9, 30.14, 30.17, and 30.20 are all marked in accordance with the MT, leaving a large space and starting the following verse on a new line. Only at Prov. 28.4 does it appear that no extra space was left at the end of the verse (the manuscript is damaged at this point, but not so much that we cannot be reasonably sure), where both the Aleppo Codex and Codex Leningrad have a petuḥa section.

96Despite the apparent quality of the work, the copyist erred in omitting a whole verse, Prov. 30.11, probably through haplography on the initial word רוד . A further error in the divine name in Prov. 30.10 was fixed by the scribe in the course of writing: a Tetragrammaton was deleted with supralinear dots and the correct form, ינדא (with the qere וינדא given in the margin), written immediately after it. Yet another error missed out a quiescent ʾalef, but was again immediately corrected by deletion and rewriting: ‘and they will eat it’ (Prov. 30.17).

97Dagesh and rafe (including rafe on final mater lectionis ה -) are used throughout, though with some variance from the MT. We find b for an expected ḇ in:

(MT ‘and the needy’ (Prov. 30.14); (MT-Leningrad ‘children of’ (Prov. 30.17); (MT-Leningrad ‘in the middle of the sea’ (Prov. 30.19)

98And ḇ for b in:

(MT ׀ ‘I will be full’ (Prov. 30.9)

99Dagesh forte is frequently omitted, e.g.,

(MT ‘he who covers up’ (Prov. 28.13); ‘his teeth’ (Prov. 30.14); ‘they’ (Prov. 30.15); (MT ‘to obey’ (Prov. 30.17); (MT ‘he will pick it out’ (Prov. 30.17)

100In contrast, dagesh lene is more consistently applied. Not only is mappiq omitted in Prov. 30.23 (MT ‘her mistress’, but the final he is given rafe, as if a mater lectionis.

8.1. Consonantal Text of T-S AS 59.215

101The consonantal text of the fragment differs from the MT when it comes to the representation of the o vowel, without showing a strong tendency towards plene or defective forms overall:

(MT ‘when rejoicing’ (Prov. 28.12); (MT ‘three’ (Prov. 30.15); (MT ‘ravens of’ (Prov. 30.17); (MT ‘three’ (Prov. 30.21)

102The u vowel is written plene in the fragment:

(MT ‘will be searched for’ (Prov. 28.12); ׃ (MT ׃ ‘they will have mercy’ (Prov. 28.13); (MT ‘oppressed’ (Prov. 28.17); (MT ‘not washed’ (Prov. 30.12)

103The text has plene e in contradiction to the MT in (MT ‘vanities’ (Prov. 28.19) and a defective i in (MT ‘to obey’ (Prov. 30.17). Furthermore, yod has been added above the line a number of times, probably by the original hand, where it is present in the MT:

(MT ׃ ‘rich’ (Prov. 28.6); (MT ‘in his eyes’ (Prov. 30.12); including (MT ‘poverty’ (Prov. 30.8), where the quiescent ʾalef is replaced by yod

104Further corrections are evident, e.g., (MT ‘oppressor’ (Prov. 28.16). The sheer number shows the care taken to produce an accurate consonantal text, but one uncorrected difference remains at ‘and in the place of’ (Prov. 28.12), where the MT reads ‘when they rise’. The fragment’s reading is possibly under the influence of the earlier phrase at .) םוקמבו םילדג ( 25.6 .Prov

8.2. Ḥaṭef and Shewa in T-S AS 59.215

105Ink transfer and staining leave some of the vocalisation signs in the fragment ambiguous or unreadable. There is evidence, however, of some systematic editing of the vocalisation, correcting the more phonetic elements towards a standard Tiberian rendering. For instance, pataḥ has been erased (faint traces remain) and replaced with shewa in:

(originally ‘he covers’ (Prov. 28.13); (originally ‘to the blood of’ (Prov. 28.17)

106Other variations in vocalisation from the standard MT, mostly involving shewa and ḥaṭef, remain uncorrected, however. In Prov. 30.14  ‘his teeth’, an apparent ḥaṭef pataḥ gaʿya stands in place of the MT’s shewa gaʿya. This is a graphic difference only, as the two are pronounced identically, and reflects the more phonetically transparent approach attributed to the school of Aharon ben Moshe ben Asher, which tends to extend the use of ḥaṭef pataḥ across the full consonantal range of Hebrew (Yeivin 1968, 24–25). There are ink spots, bleeding of ink, and mirrored text down this side of the page, however, so whether this is a correction from an original or or was always so, is unclear.

107Pataḥ is found in place of ḥaṭef pataḥ in ‘the poor’ (Prov. 30.14), with omission of dagesh too. Pataḥ is similarly preferred to vocalic shewa in ‘let no one hold him back’ (Prov. 28.17), which, unlike and noted above, was not subsequently corrected to shewa.

8.3. Differences in Vowel Quality in T-S AS 59.215

108T-S AS 59.215 exhibits a small number of variations in vowel quality. Segol substitutes for pataḥ under ʿayin in:

‘in the transgression of’ (Prov. 28.1); ‘unjust gain’ (Prov. 28.16)

109The use of two segols in segolates with a ḥet or ʿayin as the third radical is characteristic of the Palestinian vocalisation system (Yahalom 1997, 25). Although the spelling of the divine name [ʔaḏoːˈnɔːy]) ‘the LORD’ (Prov. 30.9) has a shift of a > ɛ unconditioned by gutturals. There is a shift of ē > ā under a he in:

‘a growling lion’ (Prov. 28.13)

110Two other changes in vowel quality from the MT can be noted:

‘and that despises’ (Prov. 30.17); ɛ>e, marked with rafe in MT-Leningrad, ‘in the middle of the sea’ (Prov. 30.19)

111The interchange of ɛ and e is typical of the Sefardi reading tradition (Henshke 2013b). Similarly, ū>ō in a closed stressed syllable is also characteristic of the Palestinian vocalisation tradition (Yahalom 1997, 17–18; Heijmans 2013a, §3c). However, the attendant Palestinian shift of ō>ū in an open syllable is not attested in (Prov. 28.13).

8.4. T-S AS 59.215 in Conclusion

112This is the best quality Bible manuscript of the fragments assembled here. In most respects it is similar to the MT; it is fully vocalised and cantillated, and it employs almost the full range of paratextual features, such as the marking of qere and the graphic representation of the parashiyyot. It also appears to have undergone some later correction towards the MT, in both the consonantal and vocalic layer; it may have passed through the hands of more than one owner, as a valued object.

113Where it does deviate from the MT, it does so in similar fashion to the other texts assembled here, though to a lesser extent. There is a tendency for pataḥ to replace ḥaṭef and shewa, but, unlike most of the other texts, also for ḥaṭef to replace shewa. None of these reflect phonetic changes from the standard Tiberian pronunciation. The variations in vowel quality that do occur, though comparatively isolated, might reflect the influence of Palestinian pronunciation. The prominent and decorated seder marker at Prov. 28.16 suggests that the triennial pericope was particularly meaningful to the intended user of this text. Probably this was a Bible manuscript produced for a member of the Palestinian congregation of Fusṭāṭ and perhaps therefore we might expect artefacts of the Palestinian pronunciation to turn up in its vocalisation.

9.0. Conclusions

114The fragments under inspection here are a mixed bunch, which is deliberate, as they have been chosen to display something of the range of Common Bibles found in the Cairo Genizah, with a necessary emphasis on those with Tiberian vocalisation. Despite sharing commonalities of form, in function they might have been quite different: to prepare for liturgical reading, for writing practice, as a study Bible, or as a lap Bible. The principal feature that they all share, however, is a lack of masora proper, and, for the purposes of classification, this provides a clear point of demarcation between ‘Common’ and ‘Masoretic’ Bibles.

115The Genizah is undoubtedly an important source of Masoretic Bibles too. Leaves from dozens of Great Bibles are scattered through the different Genizah collections, reflecting the vitality of the Jewish communities who were able to produce such weighty and expensive codices. Recent research has revealed the presence in the Cairo Genizah of leaves from Masoretic Bibles produced by some of the greatest scribes of their day, such as Samuel ben Jacob, who also copied the Leningrad Codex (Phillips 2016). The documentary evidence they left behind reveals how much the Egyptian Jewish elite were prepared to pay for such prestige works of biblical art and how specific they were in the details of their commissions, regarding the consonantal text, the vocalisation, the masora, and other features (Outhwaite 2018, 330–33). The ownership, commissioning, and production of Bibles was at the heart of Jewish culture in the Genizah world, with a highly skilled scribal community and knowledgeable consumers.

116From the same world come the thousands of leaves of Common Bibles that saturate the Genizah Collection. The appetite for Bible ownership extended beyond the topmost level of society, but the capacity for commissioning expert scribes to fill this need did not; hence the widescale production of Common Bibles, penned by less-expert hands. Many words have been used to describe them—‘popular’, ‘vulgar’, ‘sub-standard’—but ‘common’ suits them best, if only because they are indeed the most common form of the Bible among all the manuscript fragments found in the Genizah Collection. As a medium that, in many cases, may never have left the ownership of its original producer, they can be quite unregulated, perhaps the closest we might come to witnessing the reading of the Bible by the Jewish community of the Middle Ages. It is this unofficial nature, this potential lack of mediation, that makes them such a valuable group of manuscripts.

117Of course, no two Common Bibles are alike. Those that most closely mimic the MT are liable to give us the least evidence of the realia of the reading tradition of the medieval congregations. They do, however, point at the expertise available in the community generally, which was able to produce small-format, relatively cheap Bibles to this quality. The Proverbs fragment, T-S AS 59.215, is a case in point. It was carefully produced, and it has even undergone later correction towards the MT to remove some of its idiosyncrasies. Even so, it has preserved a number of examples that allow us a glimpse into how the creator of this manuscript pronounced their Hebrew, in this case that the shewa was pronounced as an a vowel, in line with Tiberian phonology. Entirely at the other end of the scale are the more rustic manuscripts, of which T-S AS 44.53 is a good example. Here, the scribe has reimplemented the Tiberian graphemic system according to rules of their own devising, and has used unorthodox spellings such as בש (for אוש ) and םשור (for םשאר ). As a consequence, we can see exactly which elements of the reading tradition were of more importance to the owner of the manuscript, and the phonetics of much of it are revealed.

118Given these two extremes, and the wealth of material that sits between them, it is hard to generalise about the Common Bible in the Cairo Genizah, especially given the potential geographical and temporal spread of the manuscripts that found their way into the collection. However, drawing on the analysis above, we can point to a number of features that can be found across different manuscripts, and draw some broad conclusions concerning the reading traditions associated with them.

9.1. The Consonantal Text

119We ought to begin with the question of the consonantal text. I have already said that I do not believe these manuscripts are particularly useful for textual criticism in the traditional sense of establishing the reading of the consonantal text of the Hebrew Bible. Goshen-Gottstein shares this view, though he states it more baldly: “There is perhaps one chance in a thousand that any ‘deviation’ might turn out to be something else than either the outcome of non-TBT [=non-Tiberian Bible Text] reading habits (‘Sefardi’, ‘Yemenite’, etc.) or simple bowdlerization because of lack of care or ignorance” (Goshen-Gottstein 1963, 40). What facts can we derive from the analysis of the five Common Bibles here that back up his and my assertions? There are only a few consonantal differences that reflect different readings (other than some obvious errors of omission). Two minor differences occur in the text that departs most frequently from the MT, and gives the impression of not having been copied from a Vorlage at all, T-S AS 44.35, and can probably be ignored as errors. Two more occur in ‘better’ texts, at 1 Kgs 3.27 in the hafṭara lection T-S AS 53.90 and at Prov. 28.12 in the study Bible T-S AS 59.215. In both cases, we can find similar readings elsewhere (1 Kgs 3.26 and Prov. 25.6) that probably influenced these, and they too should therefore be regarded as simple errors. The former, however, is a reading shared by a number of other manuscripts, so, if an error, it is one frequently made, and this in itself is worth noting. The latter is in a high-quality manuscript, though not one without errors (it has missed out, for example, the whole of Prov. 30.11), and its careful presentation of the parashiyyot and qere readings deserves attention (and, indeed, should dispel calumnies of ‘ignorance’). In particular, it preserves several instances of qere that are not found in the MT (Aleppo and Leningrad), one of which also has a minor consonantal difference, עבראו at Prov. 30.15 instead of עברא . Its orthography is interesting too, as it frequently does not match the MT’s, in both plene and defective forms, yet it shows signs of careful correction. As a copy of the Bible, even without a masora to safeguard it, it appears to conform to some kind of textual tradition, just one not identical to the mainstream MT.

120Across all the fragments, there is not a prevalence of matres lectionis, as perhaps might be expected, or a plethora of respellings. There are exceptions: T-S AS 44.35 has more the character of a text produced by dictation, or from memory, than by copying. But the other fragments have a mix of plene and defective forms that show a general adherence to the norms of MT spelling. We do not know how these texts were produced, and it is likely that they have a variety of different origins: copied from codices preserved as public property in the synagogues (which numbered in the dozens, according to the booklists), copied from other Common Bibles begged or borrowed off friends or family, produced by dictation or, perhaps even, by prodigious feats of recollection. It is fair, then, to slightly moderate the earlier assertions, and suggest that while most will not provide useful evidence for textual criticism (beyond assisting us with further knowledge of the kinds of errors that Bible copyists are capable of), that is not to say that none of them will. The careful text and paratext of T-S AS 59.215 give all appearance of having been copied from, or at least collated with, a reliable Vorlage, and should therefore be given due consideration for their textual value.20

121Beyond the variable value of the consonantal text, the phonetic value of the manuscripts is unquestionable, as I hope I have displayed above. Far from the mixed results of the consonantal survey, the analysis of these Bibles’ vocalisation clusters around a number of interesting features, speaking to the vitality of the Tiberian pronunciation tradition in the post-Masoretic period, and the conservatism of the Genizah world’s Bible reading.

9.2. Lack of the Ḥaṭef Vowel

122The commonest feature, found in four out of the five Common Bibles analysed here (only T-S AS 59.215, the finest of the bunch differs), is a reluctance to use the ḥaṭef sign. Three of the fragments have no occurrences of it (T-S AS 44.35; T-S AS 53.90; T-S AS 5.144), and even the closest manuscript to the MT, T-S AS 53.215, occasionally replaces ḥaṭef with pataḥ. Of morphophonological significance in the Tiberian system, the sign is phonetically superfluous if the intended user of the text is familiar enough with the pronunciation tradition. One of the hallmarks of Aharon ben Moshe ben Asher’s approach to marking vowels was a preference for ḥaṭef signs, providing greater clarity to the presence and timbre of the vocalic shewa (Yeivin 1968, 24–25). It was a sign intended to remove doubt and ambiguity. In our fragments we find a similar dislike of ambiguity, of simple shewa in particular, but the solution is different. These fragments tend to employ a full vowel, usually pataḥ, in place of the missing ḥaṭef. We can only speculate whether the full vowel is deliberate, a consequence of their education, or a lapse; it may well be different for the various fragments, as might be guessed from their varying qualities.

123In some cases, there is a different approach: ḥaṭef is replaced with simple shewa, e.g., as a vocalic shewa under gutturals, against standard Tiberian practice. T-S AS 68.100 does this on a number of occasions. There is no resulting ambiguity, however, because a vowelless guttural, where the MT would have a simple (silent) shewa, in that fragment is unmarked. Thus, the simple shewa sign is always vocalic in that fragment, and ḥaṭef is not required to avoid ambiguity.

9.3. Pataḥ for Vocalic Shewa

124A related feature to the avoidance of ḥaṭef, common to four out of the five fragments, is the use of pataḥ for shewa in some or all cases when it is to be pronounced vocalically. This accords with the standard Tiberian pronunciation of shewa as [a] when unconditioned by a following guttural or yod (Khan 2013b), a feature preserved also in the Yemeni reading tradition of Biblical Hebrew (Ya‘aqov 2013). This contrasts with the Sefardi pronunciation tradition, such as the reading traditions of Tunisia, Aleppo, and Baghdad, where unconditioned shewa is pronounced as an [e] of varying lengths (Henshke 2013b).21 This realisation of shewa as a front vowel ultimately derives from the Palestinian pronunciation tradition (Khan 2013b), where e vowels commonly occur in place of Tiberian shewa (Heijmans 2013a).22 All the fragments discussed here retain the original Tiberian realisation of shewa, even T-S AS 53.219, which appears closer to the Palestinian reading tradition in other ways. Israel Yeivin’s analysis of Tiberian vocalised piyyuṭ manuscripts from the Cairo Genizah reveals that they similarly often use pataḥ in place of shewa, e.g., ב ַ נ ִ י ‘son’ (Cambridge University Library Add.3367.8). He reaches the same conclusion, that the [a] pronunciation of unconditioned shewa is pervasive (Yeivin 1990, 176–77). The reading traditions evidenced in all these sources point to the retention of the Tiberian pronunciation of shewa, and show very little evidence for the influence of the Sefardi-Palestinian reading tradition on the vocalisation.

125The use of a pataḥ sign for the shewa vowel avoids the ambiguity inherent in the shewa sign itself. In some of the texts pataḥ only substitutes in particular circumstances, such as when shewa occurs under the first of two identical consonants following a long vowel. This was a problematic linguistic circumstance discussed widely in the medieval literature, e.g., by Aharon ben Moshe ben Asher in his Diqduqe haṭ-Ṭeʿamim and by Abū al-Faraj Hārūn in his treatise, Hidāyat al-Qāriʾ (Heijmans 2018, 98–100). It must have been a pronunciation considered prone to error. Such cases put a great reliance on either the reader’s expert knowledge of the Tiberian system or the presence of additional linguistic signage, such as the secondary gaʿya accent. Neither of these is a given with the fragments here: gaʿya rarely occurs in any but the highest-quality Common Bibles, since it is an advanced component of the cantillation system, and, indeed, its occurrence varies greatly even in Masoretic Bibles. Nor were the users of these fragments necessarily likely to have been masters of the Masoretic arts. Uncertainty is avoided, therefore, through the use of pataḥ. The outlier in its approach is again the Lamentations fragment, T-S AS 44.35, which is not content just to replace shewa in difficult circumstances, but, showing no fidelity to the accepted system, replaces vocalic shewa universally with a pataḥ sign: ‘your ear for my relief’ (Lam. 3.56).

9.4. Other Vowel Signs for Vocalic Shewa

126It is less frequently that we find a vowel other than pataḥ substituting for shewa. Two out of the five fragments show examples, where MT shewa occurs before a guttural or yod. In each of the cases, the alternation is phonetically in accord with Tiberian pronunciation, replicating the vowel under the guttural or, when before yod, giving ḥireq. Sporadic occurrences of such vocalisations can be found across the Common Bible corpus of the Genizah—a few other examples:

‘my chosen ones’ (Isa. 65.22, T-S AS 48.187); ‘you saw’ (Deut. 4.15, T-S AS 49.125); ‘the Israelite’ (Lev. 24.10, T-S AS 53.45); ‘a swelling’ (Lev. 13.28, T-S AS 57.167); ‘in his hand’ (Isa. 53.10, T-S AS 65.47); ‘and be’ (Lev. 11.44, T-S AS 48.141)

127Similar vocalisations can be found among liturgical poetry manuscripts from the Genizah, with ḥireq in place of shewa before a guttural (itself with ḥireq) or before yod relatively common (Yeivin 1990, 161, 166, 168–69, 177), e.g., ‘into the hand of’ (Cambridge University Library Add.3367.8). Other vowels though are less frequently found, and Yeivin believes that the levelling of these conditioned shewas to that of the uncoloured vocalic shewa [a] is ongoing in this period (i.e., the tenth–twelfth centuries CE, the assumed period of the piyyuṭ manuscripts’ production). This levelling is complete in Sefardi-Palestinian pronunciations (e.g., [viḥiːˈrɔːy] > [veḥiːˈraːy]), such as the reading traditions of North Africa (Henshke 2013a, §6). The evidence of these Common Bible fragments strongly differs and shows that the traditional Tiberian pronunciation of shewa was followed in most linguistic circumstances. We do not see the same levelling to the uncoloured pronunciation of shewa at all. That this is also different from the evidence of roughly contemporary piyyuṭ manuscripts is not a contradiction. The biblical reading traditions generally display a more conservative pronunciation than those of the non-biblical—the Mishna, prayers, and liturgical poetry.

9.5. Variation in Vowel Quality

128Among the five Common Bible fragments there are few divergences from standard Tiberian vowel quality. Most occur in the Lamentations manuscript, T-S AS 44.35, the most transparently phonetic in its vocalisation. It shows evidence of an almost complete shift of ɛ > e, suggesting the influence of Palestinian pronunciation, as happened in the Sefardi pronunciation, with its neutralisation of segol/ṣere and pataḥ/qameṣ (Henshke 2013b). The concomitant shift of ɔ > a also occasionally occurs in T-S AS 44.35, but is not consistent and may in fact be a morphological variant (loss of pausal forms).

129Slightly more frequent in the fragments is variation between a and ɛ. T-S AS 59.215 shows segolate nouns such as עצב with two segols [ˈbɛːṣɛʕ], suggestive of Palestinian pronunciation. Whereas the hafṭara lectionary T-S AS 53.90 and the writing exercise T-S AS 5.144, both at the more basic end of the Common Bible scale, show a shift of ɛ > a around the guttural consonants. A few other interchanges (i > e, a > ɛ, e > a, u > o) occur in such small numbers, that they can probably only be considered the sort of isolated occurrences that are liable to occur in any manuscript. What is probably most significant therefore is the extent to which, T-S AS 44.35 aside (and even that not wholly), the fragments tend to reproduce the original quality of Tiberian vowels without much variation. Certainly, there is not the evidence of a wholesale neutralisation of e/ɛ and ɔ/a as in the Sefardi pronunciation.

9.6. Variation in Consonantal Quality

130The only fragment to show more than minor variation in the pronunciation of the consonants is T-S AS 59.215, which has ב [b] for the MT’s ] v [three times and [v] for [b] once. This is surprising given the otherwise polished nature of this fragment, yet it does show considerable difference from the MT in regard to orthography as well. Perhaps this shows a lack of distinction in pronunciation between the plosive and fricative allophones, such as is found in some Yemeni pronunciations (Ya‘aqov 2013), or perhaps a free variation, such as is found in the pronunciation of Baghdad’s Jews (Shatil 2013, §2). However, sufficient regard is shown to maintaining the distinction graphically across most of the fragments presented here, suggesting, on balance, that this is not likely. Even T-S AS 44.35, which uses neither dagesh nor rafe signs, still maintains the fricative pronunciation of bet, [v], as shown by its spelling of אוש as בש at Lam. 2.14 as well as its singular use of shewa as an apparent marker of the fricative pronunciation, e.g., [haˈṭoːv] ‘the best’ (Lam. 4.1). In the other fragments the rafe sign occurs to mark fricatives (with the exception of the writing exercise, T-S AS 5.144).

9.7. The Absence of Dagesh

131The use of rafe or, in T-S AS 44.35’s case, shewa, shows that the plosive versus fricative pronunciation of the bgdkpt consonants is still operative. Dagesh, however, occurs quite infrequently in these fragments; this contrasts with its greatly increased presence in the Extended Tiberian system. On the contrary, the Common Bible fragments given here largely manage without dagesh even to distinguish the allophones of the bgdkpt. The further lack of dagesh forte to mark the gemination of consonants is striking. The two most complete fragments in their vocalisation, T-S AS 68.100 and T-S AS 59.215, both mark dagesh forte, but the other three fragments do not. This suggests that consonantal length may not have been a significant feature in their pronunciation of the text, at least in the informal environment in which these texts figured, or that it was sufficiently familiar not to require explicit marking.23 In formal reading of the Bible, the pronunciation traditions of Tunisia (Henshke 2013a, §4), Aleppo (Henshke 2013b), and Baghdad (Shatil 2013, §3) all retain the gemination of most consonants, as does the Yemeni pronunciation (Ya‘aqov 2013). While the absence of dagesh in these Common Bibles should not be taken alone as evidence for the absence of gemination, we can perhaps conclude that gemination was of less importance to the reading tradition in the eyes of these fragments’ users than the correct pronunciation of the plosive and fricative allophones of the bgdkpt consonants, since these are marked far more frequently (mainly through the presence or absence of rafe) than the geminated consonants.

9.8. Overall Conclusion

132This analysis has looked at only five fragments from the Taylor-Schechter Additional Series. This is but a drop in the Common Bible ocean, and generalisation should therefore be avoided. Indeed, there a number of significant differences between the fragments—from their presentation of the biblical text and its paratext to their quite varied approaches towards the marking of vowels. The degree of casualness and of competence can be wholly different between any two Common Bible manuscripts. But, overall, and allowing for both lapses in competence and more casual approaches to reproducing the text, we should note the clear knowledge of the Tiberian reading tradition displayed by all the fragments. A couple do suggest some influence of the Sefardi-Palestinian pronunciation in aspects only of their reading; others show occasional laxity, but nevertheless aspire to the prestige Tiberian pronunciation. There is a tendency to drop signs that are of less immediate importance to the readers, either because they have no effect on phonetic quality, since their role is performed by other signs, or because they facilitate aspects of the reading tradition that may not have been significant or even discernible to these average users: all ḥaṭef vowels, shewa when it denotes ø vowel, dagesh, some or all cantillation signs and particularly gaʿya. It was ḥaṭef (for Ben Asher), gaʿya (for the difference between Ben Naftali and Ben Asher), and dagesh (for the innovation of Extended Tiberian) that so occupied the Masoretic experts, but it is evident from these texts that the level of phonological detail these signs provided was irrelevant to most users. To this we could also add the parashiyyot, which were such a marker of quality in Maimonides’ eyes, but which are rarely represented in Common Bibles. It does not necessarily imply ignorance or lack of competence in the reading tradition, only a lack of interest or necessity. Although some of the peculiarities of the vocalisation tradition (e.g., that a simple shewa under a guttural should always be ø) might have been less than perfectly understood.

133One feature is phonetically in accordance with the Tiberian reading tradition again and again in the fragments: the pronunciation of shewa—its quality under normal circumstances, before guttural consonants, before yod, and on the first of two identical letters. No matter what other phonetic changes they display and despite the idiosyncrasies of marking the vocalisation, they all assiduously maintain this aspect of the Tiberian reading tradition. This testifies to the conservatism of the biblical reading tradition and the continued and pervasive influence of the prestige reading tradition, the Tiberian reading tradition, in the Genizah world of the High Middle Ages.