Version classiqueVersion mobile

Studies in Semitic Vocalisation and Reading Traditions

Aaron D. Hornkohl
Geoffrey Khan

Discord between the Tiberian Written and Reading Traditions: Two Case Studies

Aaron D. Hornkohl

Texte intégral

1.0. Introduction

1Like preceding Biblia Hebraica editions, the forthcoming Hebrew Bible: A Critical Edition (formerly provisionally entitled The Oxford Hebrew Bible) will have as its base text the Firkovich B19 A Leningrad Codex (= L). Defending this approach, chief editor Ronald Hendel (2016, 31–32) explains:

The copy-text will be L, our oldest complete manuscript of the Hebrew Bible. Since the accidentals of vocalization and accentuation in L are the product of medieval scribes, our critical text is open to the complaint of anachronism. This complaint is technically correct… . [B] iblical scholars already know that the consonantal text is older than the medieval vocalization system… . [However, ]… the phonology of the Tiberian vocalization system is not wholly or even mostly anachronistic… . Scholars have demonstrated that most of the phonetic features of this system accurately represent a reading tradition from the Second Temple period, and many of its features stem from the First Temple period.

  • 1 One of the best-known and oft-described examples is the shift from qal internal passive to alternat (...)
  • 2 Examples of nuanced presentations include those found of Barr (1968, 188–222; 1981, 27, 35–36; 1984 (...)

2In biblical and Hebrew language studies, one encounters seemingly incongruous views on the historical status of the Tiberian reading tradition, i.e., the specific oral realisation of the biblical text as prescribed by the Masoretic vocalisation (and accentuation). On the one hand, in the case of a small minority of certain well-known features, the vocalisation diverges from the phonetic realisations implied by the consonantal text. In most such cases the reading tradition is correctly characterised as reflecting comparatively late, secondary phonology.1 On the other hand, as Hendel notes, many authorities past and present have emphasised the antiquity of the testimony embodied in the medieval vocalisation.2

3This article plumbs the historical depths of the Tiberian reading tradition. The ‘depth’ analogy usefully comprehends two aspects of the tradition: first, its antiquity (how far back it reaches); second, its composite nature (its various constituent layers). Rarely are the two perspectives given the balanced and nuanced consideration that each deserves in exegetical, textual, literary, and even linguistic studies. All too often the Tiberian tradition’s admittedly complex textual and linguistic testimony goes undervalued and oversimplified. This frequently leads to extremes that mar studies of various types. In linguistic research, for example, the combined consonantal-vocalic text is sometimes approached uncritically, as an organic unity, its degree of linguistic heterogeneity underrepresented or entirely unacknowledged. At the other extreme are scholars who wholly discount the historical testimony of the pronunciation tradition embodied in the vocalisation, despite furnishing little to no justification for their scepticism.

4In focus here are two features in the Tiberian reading tradition whose vocalisation differs from that implied in the written tradition: (a) the qal construct infinitive and (b) the 3ms suffix that attaches to plural nouns and some prepositions. It is here argued that the Tiberian phonetic realisation, i.e., vocalisation, in the two cases both differs from that presupposed by the consonantal framework and is secondary thereto. However, far from being artificial and post-biblical, evidence is marshalled below to demonstrate that the realisations of the pronunciation tradition in both cases are organic and relatively ancient, products not of Byzantine or medieval times, but of the Second Temple Period, if not earlier.

2.0. The Qal Construct Infinitive

5The ancient Hebrew construct infinitive evolved from Biblical Hebrew (BH) to Rabbinic Hebrew (RH) and within BH itself. Developments involved phonology, morphology, and syntax. The changes are especially perceptible in certain qal forms.

2.1. Differential Treatment of Qal II-bgdkpt Construct Infinitives with Prefixed - ל

6In the Tiberian tradition, the phonetic realization of the qal II-bgdkpt construct infinitive varies depending on whether or not the form is preceded by a prefixed preposition and on the identity of the preposition. Blau (2010, 213–14) provides as clear an explanation as any:

  • 3 See also Blau (2010, 115):
    The qal infinitives construct present a complex picture, since after the (...)

The construct infinitive is frequently governed by prepositions, especially by ל . Originally this ְ ל had a fully prepositional meaning, as, e.g., ‘in order to’ (e.g., ְImageImage ‘and the Lord came down to see the town’ Gen 11 :5); later the ל became a part of the infinitive, as happened also in French and English. This is reflected both by the form and by the syntactic usage of the preposition. Formally, the ל became integrated into the infinitive. In some forms of the qal infinitive, the ל appears to be in close internal juncture: the šwa that begins the infinitive behaves as a genuine quiescent šwa, and subsequent ב ְ , ג ְ , ד ְ , כ ְ , פ ,ְ ת letters are vocalized as stops, e.g., Image ‘to fall’, as opposed to simple Image and Image ‘when falling’. In Rabbinic Hebrew the univerbalization of the infinitive with ל is even more progressed: the ל is always attached to the infinitive, even after other prepositions, and the infinitive is totally remodelled after the prefix-tense (as in Image ‘from giving’, formed after Image, in contrast to biblical Image). The special vocalization of the construct infinitive in Biblical Hebrew after ל , corresponding to the vocalization of the prefix-tense (Image ‘to remember’, matching Image) is undoubtedly in the line of Rabbinic Hebrew (and may even reflect the impact of Rabbinic Hebrew on the Masoretes). At any rate, the quiescent šwa after ל is certainly a late feature, as demonstrated by the very fact that in Image the n is not assimilated to the following consonant, because, when this assimilation operated, the šwa was not yet quiescent… . Alternatively, we could regard the vocalization of the infinitive Image as a late Mishnaic feature superimposed by the Masoretes on the biblical text, because the biblical text contained n, which had to be preserved because of the sanctity of the text.3

  • 4 On the Tiberian realisation of shewa see Khan (2013b, 546; 2013c, 775; 2020, 305–20).
  • 5 Exceptions with -ל and spirant II-bgdkpt are (...)

Recapitulating: the realisation of the second radical p is as the fricative allophone f in the bare infinitive Image [naˈfoːl] ‘to fall’ and when preceded by the prepositions - ב or - כ , e.g., Image [binˈfoːl] ‘when falling’ and Image [kinˈfoːl] ‘upon falling’. All these forms show the expected post-vocalic spirantisation of the bgdkpt consonant—this despite the fact that the preceding shewa in forms with clitic prepositions, at one time vocalic, had completely syncopated to zero in the Tiberian tradition, as reflected in the most reliable medieval codices, such as L and Aleppo (= A).4 Conversely, in the case of the infinitive with prefixed - ל , the second-radical bgdkpt consonant usually has plosive realisation, e.g., Image [linˈpoːl] ‘to fall’. The distinction illustrated here with Image [naˈfoːl] is the norm in Tiberian BH for qal II-bgdkpt construct infinitives, with very few exceptions.5

  • 6 On the two major exceptional categories to this tendency, namely n preceding a guttural and forms o (...)
  • 7 Thus Fox (2003, 205). This form is not to be confused with the presumed antecedent of the qal absol (...)
  • 8 For example, the bgdkpt consonant in the second position of (...)
  • 9 While the characteristic RH morphological similarity between prefix conjugation and construct infin (...)

Since bgdkpt fricativisation is itself a secondary development in ancient Hebrew, it might be asked whether Image [linˈpoːl] simply preserves the original plosive bgdkpt consonant that spirantised in Image [naˈfoːl], Image [binˈfoːl], and Image [kinˈfoːl]. But this explanation is problematic, because syllable-final nun normally assimilates in BH, especially in I-n forms.6 The expected form would thus be Image [lipˈpoːl], which, though absent from BH, does occur in RH. Blau’s explanation is rather that the plosive bgdkpt realisation is due to analogy to the prefix-conjugation yiqṭol form, whereby the prefix - ל of the qal infinitive construct came to be treated like the yiqṭol preformatives - י - , ת - , א , and - נ . However, whereas infinitival liqṭol descends from a form with a vowel following the first radical, perhaps lV-quṭul,7 yiqṭol represents one that never had such a vowel, i.e., yaqṭul-u/-a/-∅. Significantly, the patterning of infinitives on analogy to the yiqṭol pattern, including the infinitive’s integration of prefixed - ל , is indeed typical of RH, especially in the case of weak verbs, though important exceptions to this tendency—notably III-y verbs, on the one hand,8 and historically stative and II/III-guttural, on the other9— serve as important counterexamples. See Table 1.

  • 10 The RH forms are from Codex Kaufmann (=K) of the Mishnah.

Table 1: BH vs RH Construct Infinitives of Weak Verbs10


7Most of Blau’s account is indisputable. At least one claim, however, is open to question: namely, that in Tiberian BH the secondary plosive realisation of the middle radical in qal II-bgdkpt construct infinitives with prefixed - ל might be due to RH influence on the Masoretes and does not reflect an authentic sound shift rooted in an earlier stage of Hebrew, specifically some stage of pre-rabbinic-era BH. Before adducing evidence in favour of a more nuanced view, it is worth pointing out that any approach that takes BH as an undifferentiated whole and cites RH as the sole corpus for comparison is likely to exclude useful evidence of diversity within Tiberian BH, non-Tiberian BH, and extrabiblical material, along with information on historical development that they might provide.

  • 11 It is assumed here that the consonantal text always had an accompanying reading tradition (or tradi (...)

Even so, it is important to acknowledge the reality of the divergence between the Tiberian reading tradition and the phonological realisation that may be supposed to have accompanied the more ancient components of the consonantal text.11 Clearly, according to diverging reflexes preserved in the reading tradition, the form Image [linˈpoːl] deviates from the expected standard preserved in such forms as Image [naˈfoːl], Image [binˈpoːl], and Image [kinˈpoːl]. For though in Tiberian Hebrew the shewa of Image [linˈpoːl], Image [binˈpoːl], and Image [kinˈpoːl] was zero, its realisation in Image [naˈfoːl], the spirant allophone in the following bgdkpt consonant in Image [binˈpoːl] and Image [kinˈpoːl], and the preservation of nun in Image [linˈpoːl] are all telltale signs of its erstwhile vocal status. This implies at least some degree of phonological mismatch between the pre-Tiberian reading tradition reflected in the consonantal tradition and the comparatively more developed Tiberian reading tradition. For the former, one would expect development to Tiberian Image [linˈfoːl]; for the latter, development to Tiberian Image [lipˈpoːl]. The actual resulting Image [linˈpoːl] is either a hybrid form (as Blau seems to think) or transitional.

Against the claim that the plosive bgdkpt realisation in qal II-bgdkpt construct infinitives prefixed with - ל is necessarily due to the imposition of post-biblical phonology/morphology on the BH consonantal text, the following discussion shows that integration of - ל within the Hebrew infinitive construct was likely well underway by the Persian Period, demonstrating the historical depth of the processes that resulted organically in the plosive realisation of the second radical in II-bgdkpt liqṭol forms. Forms like Image [linˈpoːl] are, to be sure, out of step with some stage of pre-rabbinic- and pre-Tiberian-era BH as represented by the consonantal text, but are not to be explained as late post-biblical deviations under the influence of RH, much less as artificial creations of innovative medieval tradents. Rather, it is entirely plausible that this feature of the Tiberian reading tradition reveals an intermediate, perhaps vernacular, realisation linking the classical phonology and morphology expected of the BH consonantal text and RH’s more extreme phonological and morphological repatterning of construct infinitives on analogy to the prefix conjugation.

2.2. Transitional Forms in the Dead Sea Scrolls

Since Blau’s explanation might be interpreted to suggest single-step evolution between BH and RH infinitives, it is instructive to consider forms that may represent an intermediate stage, such as occasionally appear in the Hebrew of the Dead Sea Scrolls (DSS). As noted above, Tiberian BH Image [linˈpoːl] and similar (rather than RH Imagelippol and similar) suggest a vocalised first radical, i.e., one vocalism sufficient for the preservation of nun, which would otherwise presumably have assimilated. But consider the form Imagelingoaʿ in 4QSamc (4Q53) f2–5i.5 || Image [lɔˈɣaːʕaθ] in MT 2 Sam. 14.10:

(1) Image (4Q53 f2–5i.4–5)
Image (2 Sam. 14.10)
‘and the king said, “Whoever speaks to you, you should bring him to me, and he will not touch you anymore”’

  • 12 Alternatively, perhaps the nunation here results from dissimilation or prenasalisation (ng < gg) un (...)

The scroll’s scribe first wrote עוגל , presumably *liggoaʿ, and only afterwards ‘corrected’ the form by means of a supralinear nun. Of most obvious relevance for the present discussion is that the presumed pre-correction realisation *liggoaʿ is phonologically and morphologically intermediate between the respective forms expected in BH and RH, i.e., Tiberian BH [linˈgoːaʿ] < pre-Tiberian linᵊgoaʿ versus RH liggaʿ. Phonologically, the assimilation of nun is evidence that the vowel of the first radical had quiesced, as in RH and the Tiberian reading tradition. Yet, morphologically, the plene spelling with as mater waw shows that formation of the construct infinitive was not as in RH, according to analogy to yiqṭol י ִ ג ַ ע yiggaʿ which has an a, rather than o, theme-vowel (cf. RH Imageliggaʿ [M. Tohorot 5.2; 7.2–4 in K]).12

  • 13 This is an allusion to ‘when Yhwh, your God, brings you into the land to which you are coming to po (...)

8Consider also the case of לושל , presumably *liššol, ‘to clear away’, from the War Scroll:13

(2) Image
‘you are in our midst, a great and awesome God, to remove all our enemies bef [ore u] s’ (1QM 10.1–2)

  • 14 In light of the discussion above in n. 9, (...)

The expected BH form is Image (Tiberian * [linˈʃoːl] < pre-Tiberian *linǝšol). In view of the stative-type Tiberian BH imperative Image [ʃaːl] (MT Exod. 3.5; MT Josh. 5.15) and yiqṭolImage [jiʃˈʃaːl] (MT Deut. 28.40), one might expect RH-style Imageliššal in the War Scroll. Again, though, the DSS form exhibits traits characteristic of two distinct linguistic strata—the assimilation of nun typical of RH and the o-vowel typical of Tiberian BH—evidently reflecting an intermediate transitional form.14

9I-n qal infinitives in Qumran Hebrew normally behave like their BH counterparts, i.e., the nun is typically preserved or an allomorph is used, e.g., תעגל lagaʿat. Rare though the foregoing examples are, they come as evidence that the phonological process of elision of the first radical’s vowel could take place independently of the full morphological repatterning on the model of yiqṭol. One may further postulate that it was only after quiescence of the shewa of the first radical in li-qǝṭol forms had produced liqṭol, thereby resulting in phonological similarity between yiqṭol and liqṭol, that the infinitive was more fully susceptible to recasting in the mould of yiqṭol, which eventually resulted in RH-style infinitives. We will revisit this possibility below.

An attractive explanation for the aforementioned DSS forms with assimilated nun is that they represent realisations of the infinitive associated with the vernacular and/or fluent reading, in which language users pronounced no vowel following the first radical and, eventually, assimilated the nun. The inserted nun in Imagelingoaʿ in 4QSamc (4Q53) f2–5i.5 might then be attributed to a conscious attempt at careful reading appropriate for Scripture. The typologically later RH forms, in this case Imageliggaʿ, are developmentally more advanced in the direction of the vernacular, being completely under the analogical influence of yiqṭol.

  • 15 See GKC (§66f).
  • 16 While the dichotomy between biblical and non-biblical in material from the Judaean Desert is proble (...)

Syllable-final nun regularly assimilates when not word-final, but there are exceptions, even beyond II-guttural yiqṭol forms.15 It may well be that some time after quiescence of the vowel of the nun in BH Image [linˈpoːl] and DSS Imagelingoaʿ, but before wholesale RH-like repatterning on the basis of yiqṭol, language users alternated between a pronunciation preserving the syllable-final nun and one in which the nun was assimilated, perhaps reserving the realisation with nun for high-register Hebrew. Without suggesting absolute linearity between BH, the Hebrew of the DSS, and RH, the following course of development, using גנ ַ ע as an example (because its BH and RH forms also differ with respect to theme vowel), might be proposed16:


10A crucial component of this developmental scheme is that— whatever its explanation—the Tiberian form known from the authoritative medieval corpora is typologically more primitive than the QH form preserved in scrolls from the Hellenistic Period.

2.3. Further Differential Treatment of BH Construct Infinitives with Prefixed - ל

  • 17 By way of comparison, in other instances, the preposition -ל prefixed to qal I-y infinitives was ev (...)
  • 18 Blau (2010, 131).

The distinction between infinitives prefixed with - ל and those prefixed with other prepositions is not limited to qal II-bgdkpt forms. In Tiberian BH both qal I-y and II-w/y infinitives with prefixed - ל also differ from the respective forms with other prefixed prepositions. In both types, the prepositions - ב and - כ are realised with shewa, whereas - ל is vocalised with qameṣ: for I-y, consider Image [baˈlɛːðɛθ] in Image ‘and Isaac was sixty years old when (she) bore them’ (MT Gen. 25.26) versus Image [lɔːˈlɛːðɛθ] in Image ‘and she again bore, his brother’ (MT Gen. 4.2);17 for II-w/y consider Image ‘when Nathan the prophet came to him’ (MT Ps. 5122) and בכ ֵֹ או in Image ‘he came to her like coming to a prostitute’ (MT Ezek. 23.44) versus ImageImage ‘and Moses could not enter into the tent of meeting’ (MT Exod. 40.35). The phonological distinction reflects the degree to which the respective preposition was integrated into the infinitive. According to the norms of pretonic vowel development, a preposition’s originally short vowel normally shortens to shewa, as with - ב and -. כ18 The exceptional pretonic lengthening of the vowel following - ל was evidently due to the perception that it no longer served as a preposition as such—it was perhaps felt to be devoid of semantic content—but had become morphologically integral to the infinitive.

  • 19 Yeivin (1985, I, 487, 607, 641).

11The Tiberian tradition is not alone in differential treatment of - ל vis-à-vis other prefixed prepositions when it comes to construct infinitives. The Babylonian BH tradition likewise reserves differential treatment for qal II-bgdkpt, I-y, and II-w/y infinitives with prefixed -. ל19

  • 20 Exceptional forms include those with guttural root letters. Some MT qal construct infinitives are a (...)

12In the Samaritan reading tradition (as transcribed in Ben-Ḥayyim 1977), no phonological distinction marking prepositions preceding qal I-y and II-w/y infinitives, nor does the realisation of the second radical in qal II-bgdkpt infinitives depend on the presence and identity of the preceding preposition. However, the dominant qal infinitive construct pattern in strong verbs with prefixed - ל is liqṭål,20 whereas bare infinitives and those following - ב or - כ consistently bear a vowel following the first radical, e.g., עמשל lišma versus םויב עמש abyom šāma, עמשב afšāma, and עמשכ kāšāma.

  • 21 The forms are collected and discussed in Brønno (1943, 56–58); Yuditsky (2017, 131); Kantor (2017, (...)

For its part, the pronunciation tradition represented in the Second Column of Origen’s Hexapla, as preserved in Ambrosiano O 39 sup., has the forms λαμσω || Image [limˈsˤoː] ‘to find’ (MT Ps. 36.3), λαβλωμ || Image [livˈloːm] ‘to curb’ (MT Ps. 32.9), and λφνωθ (sic: probably to be read λαφνωθ) || Image [lifˈnoːθ] ‘at the turn of (cstr)’ (MT Ps. 46.6) as against the bare καρωβ* (corrected from καρωθ) || Image [qaˈʀoːv] ‘drawing near’ (MT Ps. 32.9). Unfortunately, no forms with the prepositions - ב or - כ have been preserved.21

13Thus, evidence across multiple biblical reading traditions demonstrates that qal construct infinitives with prefixed - ל were singled out phonologically among other forms of the qal construct infinitive. The simplest explanation for this affinity is that it resulted from a shared phonological heritage pre-dating the medieval or later manuscript evidence and extending back to the Second Temple period, before the traditions split.

14Yet, what of Blau’s contention that the Tiberian biblical realisation of qal II-bgdkpt construct infinitives with plosive bgdkpt allophones may be due to anachronistic reanalysis of BH on the basis of RH? Since RH is itself preserved in medieval manuscripts that reflect traditions rooted in the Second Temple period, the mere fact of demonstrating the pre-medieval character of the relevant Tiberian phonological feature does not eliminate the possibility of anachronistic superimposing of RH pronunciation on the BH infinitive.

  • 22 To illustrate by means of a phenomenon already cited, while liqṭol forms of qal II-bgdkpt construct (...)
  • 23 This is especially conspicuous when one contrasts Tiberian BH with material actually composed (as a (...)

15Two further points are in order. First, while there is no doubt that the Tiberian reading tradition and orthography exhibit non-trivial affinities with Second Temple Hebrew trends that are out of line with presumed pre-exilic phonology, it must be stressed that, overall, in respect of numerous linguistic details, the Tiberian biblical tradition presents a less advanced historical stage of Hebrew than do acknowledged post-exilic sources, e.g., the DSS, the Samaritan Pentateuch, and Rabbinic literature. The possibility of RH influence on BH or of conflation between their respective reading traditions should not be prematurely excluded, but it is clear according to the best manuscript evidence that the tradents responsible for the transmission of Tiberian BH managed with remarkable consistency to distinguish between BH and more contemporary versions of Hebrew with which they were familiar, such as RH. And this should not be thought to apply only to the consonantal tradition. High degrees of linguistic conservatism are evident in the reading tradition as well.22 Though both the Tiberian consonantal and reading traditions exhibit hallmarks of the Second Temple period in which they coalesced, since so much of their linguistic testimony points to an earlier time, neither presents a form of Hebrew that can be comfortably situated in the Second Temple period.23 The Tiberian reading tradition crystallised in the Second Temple Period. However, except where it records the pronunciation of material actually composed in the Persian Period or later, it did not originate in the Second Temple period.

16Second, it is worth discussing in the present context an observation made by Ben-Ḥayyim regarding the Samaritan tradition. Though Samaritan lišmår resembles the respective prefix tense yišmår, Ben-Ḥayyim (2000, 208) opines, on the basis of forms like לעמל alˈmāl ‘to trespass’ (SP Num. 5.6) as opposed to yiqṭol לעמת tēˈmāl, that the prefix vowel of lišmår reflects the shewa of the preposition rather than the vowel of the yiqṭol prefix. By contrast, Ben-Ḥayyim accepts the standard view that the i-vowel of the Tiberian infinitive developed via analogy to yiqṭol.

Yet, as intimated above, there seems no reason to exclude the possibility that a realisation like Tiberian Image [lišˈboːʀ] developed independently of the yiqṭol form Image [jišˈboːʀ]—via syncope of the first radical’s vowel, resolution of the preposition’s vowel to i, and maintenance of plosivisation of the following bgdkpt consonant—and that it was partially on account of the resulting similarity to yiqṭol that other construct infinitival forms, especially those of the weak verbs cited above in Table 1, were patterned after yiqṭol forms in RH. In other words, the process whereby RH weak-verb infinitives were remodelled on the basis of yiqṭol was likely organic. As such, the partial RH-like development of BH infinitives need not be considered an artificial, abrupt, top-down phenomenon orchestrated by vocalisers unduly influenced by RH according to which yiqṭol phonology was sweepingly and anachronistically applied to infinitives with prefixed - ל , but rather a natural, gradual, bottom-up process, according to which, first, li-qǝṭol simplified to liqṭol—which, in the case of qal II-bgdkpt forms, required plosive realisation of the second radical—and only subsequently, due to liqṭol’s similarity to yiqṭol, contributed to the repatterning of other qal infinitives, as in RH. Obviously, this would be mere speculation in the absence of further evidence. Thankfully, though, such evidence is available in consonantal material from the MT and other sources.

2.4. Consonantal Evidence for the Integration of - ל in the BH Infinitive Construct24

  • 24 The present section is a revised abridgement of Hornkohl (2018, 72– 79).

In the evolution of the BH construct infinitive to its RH form there is a further noteworthy morphological and syntactic development: that the preposition - ל , originally only an optional component of the BH construct infinitive, became an integral to the RH infinitive. This is most readily seen in cases in which the infinitive is preceded by a preposition other than - ל . In RH, constructions of the type Image i.e., in which the infinitive with prefixed - ל is also preceded by another preposition, whether prefixed or written separately, are not just common, but the norm. Conversely, forms preceded by prepositions and no intervening - ל are rare in RH, limited chiefly to biblical citations. This shows that for RH users, the formerly prepositional - ל had become an essential part of the infinitive. In other words, the bare infinitive is a viable option in BH, whereas the - ל is virtually inseparable from the RH infinitive.

  • 25 Conditioning factors extend beyond the purely morphological. For example, the absence of certain fo (...)

17However, the dichotomy between BH and RH as just described is potentially misleading. First, though the bare infinitive construct is especially characteristic of BH when compared to RH, it must be stressed that throughout the entire biblical corpus forms with prefixed - ל are far more common than forms without. According to the Groves-Wheeler Morphological database, in L there are 6587 infinitives construct, of which 5977 (90.7 %) follow some preposition: 4506 (68.4 % of total; 75.4 % of those with preposition) follow - ל ; only 610 (9.3 %) consist of bare infinitives. Thus RH’s extreme preference for infinitives with prefixed - ל represents no more than relative advancement in a trend in favour of the integration of - ל within the construct infinitive already well underway in BH.25

  • 26 See also MT 1ְKgs 18.29; 2ְKgs 23.10; Hab. 3.14; Ezra 10.14; 1 Chron. 5.9; 28.20; 2 Chron. 24.10; 2 (...)

18Yet even this formulation is too general. Certain strata of BH more closely resemble RH than others. RH’s regular retention of prefixed - ל following another preposition has already been mentioned. Such structures are rare in BH, where a decisive majority of the occurrences—ten of thirteen—occur in LBH (see examples 3 and 5, each contrasted with more classical parallels in examples 4 and 6, respectively).26

(3) Image
‘… is it not to reconnoitre and spy out the land that his servants have come to you? ’ (MT 1 Chron. 19.3)

Cf. the more classically formulated near-parallel without - ל after Image in

(4) Image ‘… is it not to reconnoitre the city and to spy it out and to overthrow it that David has sent his servants to you?’ (MT 2 Sam. 10.3)
(5) Image
‘in those days Hezekiah became ill to the point of death… ’ (MT 2 Chron. 32.24)

Cf. the more classically formulated near-parallel without Imagebefore the infinitive construct in

(6) Image
‘in those days Hezekiah became ill to the point of death… ’ (MT 2 Kgs 20.1 || MT Isa. 38.1)

  • 27 See also 4Q109 f1ii+3–6i.18–19 || MT Qoh. 7.5.

19The late character of such structures is further confirmed by the fact that in DSS biblical material - ל is sometimes inserted between another preposition and the infinitive when - ל is lacking in the parallel Masoretic version (examples 7–8).27

(7) Image (4Q67f1.4)
Image (Isa. 58.13)
‘and you honour it [by refraining] from going your own ways and from finding your own pleasure’
(8) Image (4Q166 2.9)
Image (Hos. 2.11)
‘and I will take away my wool and my linen from covering/to cover your nakedness’

20Similarly, in non-biblical DSS material - ל can intervene between a preposition and an infinitive (examples 9–10):

(9) Image
‘to the point of extinction, to the point of rebellion’ (4Q378 f3i.7)
(10) Image
‘and] from being party to these matters or going along w[ith them] in these things’ (4Q397 f14–21.8)

21Finally, the inscriptional and biblical distribution of construct infinitives in the function of verbal complements is instructive. Pre-exilic epigraphy and biblical material know in this function both the bare infinitive and the infinitive prefixed with - ל . A conveniently apposite illustration of mixed usage is the is the two-line sequence from the Lachish Letters in which the two alternatives appear in consecutive lines (examples 11–12).

  • 28 On the formulation of negative oaths see JM §165, especially subsections d, f and g.

(11) Image
‘And because my lord said, “You don’t know how / to read a letter!”’ (Lachish 3.8–9)
(12) Image
‘As Yhwh lives, I swear, no one has ever tried to read me a letter! ’ (Lachish 3.9–10)28

22For the situation in Masoretic BH consider Table 2. Forms with - ל are dominant throughout the Hebrew Bible, except in some poetic material (e.g., Isaiah and Job). Significantly, however, in the core LBH books and Qohelet, the bare infinitive construct as verbal complement has fallen into disuse. See Table 2 for the biblical distribution (according to L) of construct infinitives with and without - ל .

  • 29 These statistics reflect the approach of Malessa (2006, 150–66), with slight modifications, as deta (...)

Table 2: MT distribution of verbal complement infinitive construct29


23Table 3 compares the Tiberian biblical text to the Second Temple corpora of Biblical Aramaic, Ben Sira, the biblical and non-biblical DSS, and RH.

  • 30 For lists of occurrences see Hornkohl (2018, 75–76, nn. 25–28).
  • 31 While the strong BA penchant for verbal complement infinitives with prefixed -ל tallies with Second (...)

Table 3: MT, cognate, extrabiblical, and non-Masoretic distribution of infinitive construct as verbal complement with and without - ל according to corpus30 31


  • 32 See also 1QIsaa 1.14–15 || MT Isa. 1.12; 1QIsaa 1.15 || MT Isa. 1.13; 1QIsaa 7.22–23 || MT Isa. 8.4 (...)

24All material assuredly composed in the post-exilic period shows a striking preference for liqṭol over qǝṭol as verbal complement. Only the biblical DSS exhibit proportions comparable to those known from Classical Biblical Hebrew (CBH) sources, which is hardly surprising given the nature of the material. But even this similarity is somewhat deceptive. Substantiating the late replacement of bare infinitive verbal complements with forms bearing prefixed - ל , there are some eleven cases in which a Tiberian example without prefixed - ל is paralleled by a DSS case with - ל (examples 13–16) and no cases of the contrary:32

(13) Image (4Q252 1.15–16)
ְImage (Gen. 8.10) ‘and he waited another seven days and he again sent… ’
(14) Image (4Q2521.18–19)
Image (Gen. 8.12)
‘… (and) he sent forth the dove, but it did not return (to him) again’
(15) Image (4Q40 f5.6)
Image (Deut. 7.22)
‘you will not be able to finish them off’
(16) Image (1Q4 f12.2)
Image (Deut. 14.24)
‘you cannot carry it’

25It is difficult to interpret the consistency of this direction of change as casual or insignificant. From the perspective of the relevant MT material, the biblical DSS copyists regularly succeeded in reproducing classical diction, but occasionally fell under the sway of contemporary language practises that, in respect of the phenomenon under investigation, led them to use liqṭol rather than bare qǝṭol.

2.5. Summary of Case on Qal Construct Infinitive

26Against the claim that the Tiberian phonological realisation of BH qal II-bgdkpt construct infinitives is a rabbinic or later anachronism alien to older BH phonology, we have adduced phonological, morphological, and syntactic evidence to demonstrate the heretofore under-appreciated historical depth of the phonological distinction between infinitive construct forms prefixed with - ל , on the one hand, and bare infinitives and those prefixed with other prepositions, on the other. The multiplicity of traditions exhibiting similar instances of differentiation or apparent reflexes thereof (Babylonian, Samaritan, Secunda) points to a genuine Second Temple phenomenon inherited by each. Consonantal evidence from Second Temple and presumably earlier sources confirms both the diachronic character of the relevant difference between BH and RH as well as intermediate stages as witnessed in LBH and the DSS, including infinitival formations that combine BH and RH features, increased usage of infinitives prefixed with - ל following another preposition, and decreased employment of the bare infinitive as verbal complement. Of no less importance, the relative frequency of construct infinitives with - ל and the comparative rarity of bare infinitives throughout the biblical text, even in those works considered most representative of pre-exilic Hebrew, come as compelling evidence of the probable early annexation of - ל to the construct infinitive. Far from being a chronologically foreign intrusion into BH morphology, the integration of prefixed - ל within the infinitive is very much in line with morphological and syntactic trends evident in the classical stage of BH as witnessed in consonantal material. On the basis of the extant evidence, establishing a terminus a quo for syncope of the vowel of the first radical would seem to be out of the question. Even so, in the light of DSS infinitival forms that reflect syncope of the first radical’s vowel without full remodelling on the basis of yiqṭol as seen in RH, it is reasonable to hypothesise that the vocalic elision that permitted plosive realisation of the second radical in qal II-bgdkpt construct infinitives in the Tiberian tradition is not a result of reanalysis under the influence of RH, but an organic feature firmly rooted in earlier Hebrew. It is likely to have occurred first in the vernacular. Given the regularity of infinitives with prefixed - ל throughout BH (relative to the number of bare infinitives and those with other prefixed prepositions), the morphological and phonological shifts in question may well have occurred long before Second Temple Hebrew, with the expected assimilation of nun being avoided in literary registers, such as that preserved in the Hebrew Bible.

Though it is impossible to determine the full extent of the historical depth of phonological realisations like Tiberian Image [linˈpoːl] and Image [liʃˈboːʀ] there is ample evidence to show that they are phonological reflexes of a relatively early morphological development with attendant syntactic ramifications. Given that both the Samaritan and Jewish reading traditions bear witness to the phenomenon and/or to reflexes thereof, and that scholarly consensus places that schism no later than the second century BCE, such a date serves as a logical terminus ante quem for the integration of - ל into the qal infinitive construct, though some scholars would place this earlier. Clearly, the feature was sufficiently established in pre-schism scriptural reading practices as to be inherited by both the Jewish and Samaritan traditions before they diverged. In light of the dominant use of infinitives with - ל throughout the biblical text (excepting archaic poetry), supported by epigraphic evidence, it is reasonable to propose a terminus post quem as far back as the heyday of CBH, i.e., the monarchic/First Temple period, though with the disuse of bare infinitives construct as verbal complements, it is perhaps most reasonable to place the univerbalisation of liqṭol in the Persian period. This corroborates the conclusions of previous studies that emphasize the antiquity of the Tiberian reading tradition and its reliability as a linguistic witness of early Second and even First Temple Hebrew.

3.0. The 3ms Suffix for Plurals and Similar

27Once we entertain the possibility of disparity between the written and reading traditions, it opens up the possibility of alternatives to certain conventional, but dissatisfying explanations.

28One of the more counterintuitive orthographic conventions that beginning students face when learning to read Hebrew is the 3ms possessive suffix added to plural nouns and to some prepositions. Written וי -, as in ויסוס , the ending was evidently realised very early on as a diphthong along the lines of -aw, which, in turn, developed to, inter alia, Tiberian ָ וי - [-ɔːv] and Modern Israeli Hebrew [-av]. Though learners quickly equate the phonetic realisation with the spelling in question, the correspondence is decidedly anomalous from the perspective of Hebrew orthography, where, though vowel sounds are regularly left underspecified or entirely unmarked, consonants—such as the yod in וי -— are usually pronounced.

29The exceptions to this norm are the matres lectionis, namely medial and final waw and yod and final heh (and ʾalef), each of which came to be used to signal specific vowel sounds. The mater yod is associated with i- and e-class vowels. Its appearance with the a-class vowel that developed to Tiberian [ɔ] (qameṣ) calls for an explanation.

3.1. Competing Accounts: Grammatical versus Phonetic Yod

  • 33 It seems clear that יו- was indeed eventually taken as representative of the diphthong -aw (and its (...)

30Generally speaking, scholarly literature offers two competing explanations for the unexpected representation of -aw with וי -. According to one, the yod in וי - did not originally have phonetic value, but served as a mater lectionis of purely grammatical significance, introduced at some point for purposes of visually distinguishing the plural form of a possessed noun from its singular counterpart. On this view, only later, due to association with the realisation -aw in the 3ms suffix, did language users extend use of the spelling וי - to other words with a similar final diphthong or reflex thereof.33

31Andersen and Forbes (1986, 325) argue for a purely graphic genesis to the spelling וי -:

  • 34 See also Orlinsky (1942–43, 288–89); Zevit (1980, 29–30); Pardee (1988, 279–80); Knauf (1990, 20); (...)

In the old orthography it was not possible to distinguish ‘his son’ from ‘her son’, both spelled הנב , unless aided by context. The new convention wrote ונב and הנב respectively. But this created a new problem. In the old orthography ונב was the regular spelling of bānāw, ‘his sons’, but now this could be read as bānô. Other forms of plural nouns had the plene spelling of the long vowel in bānīm or of the stem-terminal diphthongs in the suffixed forms, such as םהינב , ‘their sons’. The remedy again was obvious. Spell all plural nouns (masculine or of masculine type) with י whether it was pronounced or not. Hence the artificial וינב , bānāw, ‘his sons’, in which the י is purely graphic.34

  • 35 Blau (2010, 172); Zevit (1980, 29–30).
  • 36 GKC (§91i and n. 1); Cross and Freedman (1952, 47, 54–55, 68–9); Sarfatti (1982, 65); Gogel (1998, (...)

32According to an alternative hypothesis, the yod in question was phonetic. Some seek to reconcile וי - with -aw,35 while others object that the two must reflect distinct pronunciation traditions. It is this latter possibility, that וי - and -aw reflect diverse phonetic realisations, that is examined below. If this is the case, then this is one more in a series of cases in which the dominant written form and its oral realisation in the Tiberian (and other) traditions are out of sync, i.e., represent a merger of discordant reading and written traditions.36

There is equivocal evidence for the argument that the yod in the 3ms suffix וי - began as a purely graphic morphological marker of plurality. First, a few other Hebrew suffixes appear— at least synchronically—to have a non-phonetic yod with the purely grammatical function of marking plurality, namely, the 1cpl ending in Tiberian Image [suːˈseːnuː] ‘our horses’ versus the phonetically-identical 1cpl ending in Tiberian Image [suːˈseːnuː] ‘our horse’ and the 2ms ending in Tiberian Image [suːˈsɛːχɔː] ‘your (ms) horses’ versus the homophonous Tiberian pausal Image [suːˈsɛːχɔː] ‘your (ms) horse’.

Of course, seen from a diachronic perspective, this is a non-argument. Paradigmatically, in the relevant 3ms suffix one expects a y glide, or reflex thereof in the form of a vowel produced via diphthong contraction. Thus, 1cs Imageswsy Tiberian [suːˈsaːy] ‘my horses’ and 2fs Image Tiberian [suːˈsaːyiχ] ‘your horses’ both preserve consonantal yod in a diphthong, while the e/ɛ vowels in the remaining forms are all attributable to contraction of the same ay diphthong. Leaving aside the 3ms suffix under discussion, then, a strong case can be made for an originally phonetic role for the yod in all suffixes for plurals and the relevant prepositions.

33Second, the difference between pre-exilic epigraphic orthography and biblical orthography (as represented to varying degrees in all biblical manuscript traditions) indicates that the earliest biblical compositions must have undergone a spelling revision according to which matres lectionis were frequently, but somewhat inconsistently, inserted word-medially in line with post-exilic conventions in order to facilitate reading. On the assumption that וי - is secondary to ו -, the revision in question would provide a historical scenario in which a grammatical mater yod could have been inserted. The notion of the regular insertion of a morphological mater would, however, be exceptional against the backdrop of the broader goal of phonetic transparency as well as the inconsistent use of phonetic matres.

  • 37 For inclusion of the variant 3ms endings among written-reading mismatches, see Khan (2013a, 48). Ju (...)
  • 38 For useful discussion see Gogel (1998, 159–61 and nn. 187–89) and Barkay et al. (2004, 53–54).

While the situation of orthographic revision arguably furnishes a convenient historical context in which the purely graphic change 3ms ו - -aw > וי - could take place, recognition of numerous categories involving phonetic dissonance between the Tiberian written (consonantal) and reading (vocalic) traditions demonstrates the potential reality of diverse phonetic realities behind 3ms ו - and וי -. Consistent mismatch between the written and reading components of the Tiberian Masoretic tradition is an acknowledged phenomenon in the case of a number of features, most famously the 2ms qatal verbal ending ת - and the 2ms pronominal suffix ך - versus their respective Tiberian realisations Image [-tɔː] and ָ ך - [-χɔː]. Though common in the MT, these are decidedly exceptional from the perspective of Tiberian (and other Hebrew) orthographical conventions, where final vowels are otherwise generally represented by a mater. For instance, in the vast majority of categories in which a word-final a-class vowel appears in manuscripts of the Masoretic tradition, it is accompanied by a mater heh. The lack of this heh in 2ms forms is glaringly exceptional. Variation in the realisation of the 2ms ת - and ך - endings is apparent in certain Tiberian pausal forms, e.g., Image [lɔːχ] for contextual Image [laˈχɔː]. This is found also in other traditions of Hebrew and Semitic languages more generally, where one finds both consonant-final realisations of these pronominal suffixes (Iron Age inscriptions, MT ketiv, DSS, RH, Aramaic/Syriac, Secunda) and also vowel-final realisations (Iron Age inscriptions [verbal ending only], MT qere, DSS).37 Given the reality of mismatch between the Tiberian written and reading traditions, as well as the more general variety within ancient Hebrew pronunciation traditions, one should at least consider the possibility that the majority consonantal orthography וי - actually reflects a phonetic realisation other than that which eventually developed to Tiberian [-ɔːv].38

  • 39 The dating is according to Barkay et al. (2004, 41–55). The spelling יו- is an important element in (...)
  • 40 Lemaire (1977, 599, 601); Zevit (1984, 43); Hadley (1987, 54–55); Gogel (1998, 159–60, n. 188).
  • 41 The authenticity of this inscription has been called into question; see Rollston (2003; 2006).

Of greater probative value is epigraphic evidence. Crucially, both spellings are known from sources assigned to the Iron Age. In the pre-exilic inscriptional material from the Judahite city of Lachish (early sixth century BCE) there occurs the form ושנא ‘his men’ (Lachish 3.18). Further evidence is found in the prepositional ולא ‘to him’ from the Meṣad Ḥašavyahu (Yavne Yam line 13) plea (late seventh century BCE). One might also consider the suffix of the apparently dual וחרי ‘its two months’ (Gezer 1.1 [2x], 2, 6; late tenth/early ninth century BCE). The surest occurrence of וי - in Iron Age epigraphy is in the form וינפ ‘his face’ in one of the renditions of the Priestly Blessing from Num. 6.24–26 preserved in the Ketef Ḥinnom amulets (2.9; mid-seventh century BCE).39 Other potentially relevant evidence includes the apparent polyphthong terminating הירצמו ‘and from his enemies’ (Khirbet el-Qom 3.3; 750–700 BCE—this is the consensus reading, but it is uncertain; cf. Tiberian Image [misʕ-sʕɔːˈʀɔːv] [Deut. 33.7; Jer. 46.10; Ps. 105.24]),40 and, possibly, ויחא ‘my brothers (?)’ (Moussaieff 2.7–8, though this may well represent the singular ‘my brother’).41

  • 42 See Reymond (2014, 144–47, 159) on the relative frequency of spelling variation in the DSS, though (...)

34Turning to later documentary data, the spelling ו - for expected וי - is not uncommon in Dead Sea biblical material and is even more widespread in the non-biblical scrolls. This likely indicates the persistence of various phonetic realisations, though there are alternative interpretations of the data.42

  • 43 Brønno (1943, 200–1); Yuditsky (2017, 107).
  • 44 Ben-Ḥayyim (2000, 229).

35Finally, perhaps most significant as evidence for a graphic, grammatical, non-phonetic explanation for the development of וי -as -aw is the fact that no pronunciation other than a diphthong or reflex thereof is preserved in any known Hebrew reading tradition. As already noted, the suffix came to be realised as [-ɔːv] in Tiberian Hebrew. Similarly, it is represented by -αυ (eleven times) in the Greek transcription in the Secunda of Origen’s Hexapla43 and it contracted to [-o] in Samaritan Hebrew.44

36While no extant Hebrew reading tradition evinces a pronunciation of 3ms וי - with a phonetic yod—conceivably, some sort of triphthong along the lines of -ayu or -eyu—the traditions may preserve indirect evidence indicative of such a realisation, as we shall see in what follows.

3.2. Positive Arguments for Phonetic Yod in 3ms וי -

37The remainder of this article will consider affirmative arguments for an originally phonetic yod in the 3ms suffix וי -, in which case the relevant realisation—likely something akin to -ayu or -eyu— differed from pre-Tiberian -aw. First, we return to the apparent Iron Age epigraphic evidence. It should be stressed that arguments against the pre-exilic dating of Ketef Ḥinnom’s 3ms וי - in וינפ ‘his face’ (2.9) based on the supposed lateness of the orthography must be considered circular. If the inscription is reasonably dated on other grounds to the 7th century BCE, then the spelling וי - must be accorded as much weight as spellings without yod from other Iron Age epigraphs. Further, since, as a rule, medial characters in Iron Age inscriptions serve as either consonants or vowel letters, but not grammatical matres, it is likely that וי - here has a realisation other than -aw. Of course, no certainty can be had on the exact nature of the sound in question. Crucially, though, this applies to the alternative epigraphic spelling 3ms ו -, as well; while it may coincide with pre-Tiberian -aw, it may just as well reflect a different realisation.

Potentially illuminating in respect of the phonetic reality behind 3ms וי - is the rarer alternative Image-ēhū, e.g., Tiberian Image [gibboːˈʀeːhuː] ‘his warriors’ (Nah. 2.4), Image [jɔːˈðeːhuː] ‘his hands’ (Hab. 3.10). It is commonly thought that this suffix, regularly employed only for singular III-y forms, preserves an early form of the ending that developed from *-ayhu. From והי - *-ayhu development to וי - is relatively straightforward, the presumed realisation of the latter being -ew > Tiberian [-eːv]:

*-ēv < *-ēw < *-ēū < -ēhū < *-ayhū.

This involves the routine phonetic developments of contraction of the diphthong -ay- to -e-, elision of intervocalic h, and resolution of the falling diphthong -eu- via -ew- to -ev. Cf. Tiberian Image [geːv] ‘middle, back’, Image [kisˈleːv] ‘Kislev’, Image [ʃɔːˈleːv] ‘at peace’.

Significantly, the process above accounts for the rare spelling-pronunciation combination Image-ēhū [-eːhuː], the minority spelling ו - -w as reflecting *-ēw < *-ēū (e.g., Lachish, Meṣad Hashavyahu [Yavne Yam], Gezer, and in the MT), and the dominant spelling וי -:

*-ēw < *-ēyw < *-ēyū < *-ēū < -ēhū < *-ayhū,

38In the spelling וי - the yod would, according to this reconstruction, have originally represented the glide of the diphthong *ay and subsequently, after the contraction of the diphthong, become a mater for ē. Indeed, if, for the sake of economy, it is assumed that epigraphic וי - and ו - should have represented the same realisation, from a purely phonetic perspective, it is more likely to have been - ew < -eyu than -aw. It is difficult to conceive of any single phonetic realisation underlying epigraphic וי -, epigraphic ו -, and pre-Tiberian -aw.

39But as the dominant realisation in all extant reading traditions, -aw demands an explanation. The problem is, while it is possible to get from *-ayhū to -aw, along the developmental path, there is no realisation in extant reading traditions for which the spelling וי - can reasonably be considered to be a phonetic representation.

40The simplest way to account for -aw is to posit the development

- āw < *-āū < *-āhū < *-ayhū.

  • 45 This account is based on Florentin (2016, 74). Cross and Freedman (1952, 47) note that “[o] nly in (...)

Significantly, the first step involves contraction of ay to ā, at which point a written yod became otiose, as in Image [ʔɔːn] < Image [ˈʔajin] ‘where’. The next steps are routine phonetic processes: intervocalic elision of heh, and, in Tiberian Hebrew, the shift [ɔː] < ā. Crucially, since contraction of -ay- evidently preceded elision of h, it is difficult on this view to account satisfactorily for the dominant Masoretic spelling יו -.45

41A further conjectural process may be mentioned. The evolution

*-āw < *-ayw < *-ayhū

  • 46 Blau (2010, 172); Zevit (1980, 29–30).
  • 47 In agreement with Florentin (2017, 73–74). Florentin has proposed a motivation for development of t (...)
  • 48 Barkay et al. (2004, 54). Also worthy of consideration as an explanation for the orthography יו-, b (...)

42has been proposed, ostensibly furnishing a rather straightforward account of both the וי - orthography and the Tiberian [-ɔːv] realisation.46 It must be said, though, that this developmental sequence involves the assumption of two rather arbitrary steps— elision of heh after a consonant (diphthong) and contraction of the triphthong *-ayw to *-āw. With neither standard in ancient Hebrew, it seems improbable that both would take place.47 An alternative solution involving more likely processes is preferable. Without such a solution, it is very difficult to explain both the spelling וי - and phonetic realisations reflecting -aw as results of one and the same process.48 The combined weight of the evidence arguably points to the plausibility of a phonetic realisation behind וי - different from pre-Tiberian -aw and its later reflexes.

  • 49 CD 10.9; 1QS 3.7–8; 6.17; 11.3; 1QSa 1.18, 22; 1QpHab 3.7; 5.5; 9.1; 1QHa 4.37; 4Q163 f4–7i.8; 4Q20 (...)
  • 50 1QIsaa 1.28; 11.29; 23.2; 27.23; 2Q16 f5ii–6i.1; 4Q32 f2ii+3i+4.19; 4Q56 f3ii.15; 4Q86 2.13; 4Q93 1 (...)
  • 51 Exod. 27.11; 28.28; 32.19; 37.8; 39.4, 33; Lev. 9.22; 16.21; Deut. 2.33; 7.9; 8.2; 33.9; Josh. 16.3 (...)
  • 52 See, by way of illustration, the useful examples in Golinets (forthcoming, 1–6).

One final perspective to consider is the explanatory value of the approach, especially with regard to the distribution of the various spellings in Second Temple sources and in the MT itself. In what may have more significance than is sometimes thought, biblical and non-biblical manuscripts from late antiquity show various mixtures of forms. In the non-biblical DSS, the ו - ending occurs without yod in nearly 12 percent of the relevant cases of nouns (56 of 473),49 in the biblical DSS the proportion is just over 7 percent (25 of 347).50 In the MT there are some 125 instances involving nouns,51 coming to about 5 percent of the potential 2500 cases. While various explanations could be offered for these minority spellings,52 it is here argued that the possibility that at least some reflect actual phonological variation in the realisation of the suffix should not be dismissed out of hand. It may well be that in some of these instances in the MT, many of which are marked as ketiv-qere mismatches, the disparity reflects morphological, rather than phonological, disagreement between the consonantal and pronunciation traditions. Thus, for instance, in ְ ImageImage ] ביתכ  :ְ ודי [ Image (MT Lev. 9.22), the ketiv form ודי is ambiguous—it may represent a form parallel to Tiberian dual [jɔːˈðɔːv] ‘his hands’, but may just as well reflect a singular parallel to Tiberian [jɔːˈðoː] ‘his hand’; if the latter, the qere contradicts the ketiv semantically. Further, in nearly every case it is possible that the yod of the majority spelling was simply omitted in error, so that, for example, the ketiv ודי in ImageImage ] ביתכ  :ְ ודי [(MT Lev. 16.21) does not necessarily serve as orthographic evidence of a diphthongal rather than triphthongal realisation. These considerations apply to the aforementioned DSS evidence as well.

43Be that as it may, it seems unlikely that morphological ambiguity and simple spelling inconsistency are sufficient to account for the totality of cases in which the 3ms suffix spelling ו -occurs instead of the more customary וי -. In MT Ezekiel ו - appears instead of וי - in 46 of 176 cases (26 percent of the time) and in chapter 40 alone ו - comes without yod 34 times. Job also exhibits use of the suffix nearly 10 percent of the time (12 of 122 cases). When we bear in mind other discrepancies between the Tiberian written and reading traditions, e.g., the aforementioned 2ms endings ת - and ך - and Tiberian [-tɔː] and [-xɔː], respectively, positing a similar mismatch between consonantal וי - and Tiberian [-ɔːv] helps to account for a degree of variation otherwise difficult to explain.

44If the foregoing reading of the evidence is correct, it assumes a rather curious developmental sequence leading up to the spelling and phonetic realisation of the Tiberian Masoretic tradition:

  1. a situation of mixed use of diphthongal and triphthongal phonological realisations and spelling, i.e., ו - -aw and וי - - ayu/-eyu;
  2. emergence of mismatch between the dominant diphthongal phonetic realisation and the dominant triphthongal spelling—perhaps involving the preservation of a ‘historical’ spelling paralleling the diffusion of a ‘popular’ pronunciation;53
  3. an orthographical revision strongly favouring ‘historically conservative’ וי -, reflecting either persistence of a triphthongal pronunciation or association of that spelling with the realisation -aw, leaving only a minority of the relevant forms ending in ;- ו54
  4. subsequent to the fixing of the orthographical and pronunciation traditions, the fusion of the two into the Tiberian Masoretic textual tradition, which involved the ‘correction’, by means of explicit ketiv-qere notation (from ו - to וי -) or via simple vocalisation (with [-ɔːv]), of forms one might be inclined to read otherwise, sometimes extended, significantly, to forms that do not end in the 3ms suffix in question.

4.0. Closing Remarks

45Dissonance between written and reading traditions is an acknowledged feature of the Tiberian Masoretic tradition, the authenticity and antiquity of each supported by contemporary Hebrew and cognate material. The only question regards the extent of the mismatch, i.e., its applicability and explanatory value with regard to individual features. It has been argued here that such a perspective helps to explain two vocalisation phenomena as relatively early, organic developments, rather than anachronistic post-biblicisms artificially visited upon BH. These may be added to a series of over twenty instances or categories of instances involving similar written-reading dissonance within the Tiberian tradition. The ramifications go beyond the phonetic realisations of the specific features in question, encompassing such issues as the antiquity, authenticity, and reliability of the testimony of the Tiberian reading tradition. Crucially, precisely at those points where there is the most compelling reason to suspect anachronism and artificiality on the part of the vocalisation tradition, i.e., where it deviates from a consonantal tradition generally accorded greater antiquity, the reading tradition is seen to reflect ancient and natural linguistic conventions in line with Second Temple or earlier practices.


5.0. References

Albright, William F. 1943. ‘The Gezer Calendar’. Bulletin of the American Schools of Oriental Research 92: 16–26.

Andersen, Francis I., & A. Dean Forbes. 1986. Spelling in the Hebrew Bible. Biblica et Orientalia 41. Rome: Biblical Institute Press.

Barkay, Gabriel, Marilyn J. Lundberg, Andrew G. Vaughn, and Bruce Zuckerman. 2004. ‘The Amulets from Ketef Hinnom: A New Edition and Evaluation’. Bulletin of the American Schools of Oriental Research 334: 41–71.

Barr, James. 1968. Comparative Philology and the Text of the Old Testament. Oxford: Clarendon.

——. 1981. ‘A New Look at Kethibh-Qere’. Remembering All the Way…: A Collection of Old Testament Studies Published on the Occasion of the Fortieth Anniversary of the Oudtestamenthische Werkgezschap in Nederland, edited by A. S. van der Woude, 19–37. Oudtestamentische Studiën 21. Leiden: Brill.

——. 1984. ‘Migraš in the Old Testament’. Journal for Semitic Studies 29: 15–31.

Bauer, Hans, and Pontus Leander. 1918–1922. Historische Grammatik der hebräischen Sprache des Alten Testaments. Halle: Max Niemeyer.

Ben-Ḥayyim, Zeʾev. 1977. The Literary and Oral Tradition of Hebrew and Aramaic amongst the Samaritans, vol. IV: The Words of the Pentateuch. Jerusalem: The Academy of the Hebrew Language (in Hebrew).

——. 2000. A Grammar of Samaritan Hebrew. Jerusalem: Magnes and Winona Lake, IN: Eisenbrauns.

Blau, Joshua. 2010. Phonology and Morphology of Biblical Hebrew. Linguistic Studies in Ancient West Semitic 2. Winona Lake, IN: Eisenbrauns.

Brønno, Einar. 1943. Studien über hebräische Morphologie und Vokalismus. Abhandlungen für die Kunde des Morgenlandes 28. Leipzig: Brockhaus.

Cross, Frank Moore, and David Noel Freedman. 1952. Early Hebrew Orthography: A Study of the Epigraphic Evidence. American Oriental Series 36. New Haven, CT: American Oriental Society.

Fassberg, Steven E. 2007. ‘Infinitival Forms in Aramaic’. Historical Linguistics 2005: Selected Papers from the 17th International Conference on Historical Linguistics, Madison, Wisconsin, 31 July–5 August 2005, edited by Joseph C. Salmons and Shannon Dubenion-Smith, 239–56. Amsterdam: John Benjamins.

Florentin, Moshe. 2017. ‘ʾEx naflu giborehu va-yaqumu giborav taḥtehem: Maʿamad ha-he ben tnuʿot, kinuye ha-guf shel ha-nistar ve-ha-nisteret ve-kivuṣe ha-diftongim be-ʿivrit’. Lešonénu 79: 64–82.

Fox, Joshua. 2003. Semitic Noun Patterns. Harvard Semitic Studies 52. Winona Lake, IN: Eisenbrauns.

Freedman, David Noel. 1992. ‘The Evolution of Hebrew Orthography’. In Studies in Hebrew and Aramaic Orthography, edited by David Noel Freedman, A. Dean Forbes, and Francis I. Andersen, 3–15. Winona Lake, IN: Eisenbrauns. Biblical and Judaic Studies from the University of California, San Diego 2.

Garr, W. Randall. 2007. ‘Prenasalization in Aramaic’. Studies in Semitic and Afroasiatic Linguistics Presented to Gene B. Gregg, edited by Cynthia L. Miller, 81–109. SAOC 60; Chicago, IL: University of Chicago Oriental Institute.

GKC = Gesenius’ Hebrew Grammar. Edited by Emil Kautzsch, translated by A. E. Cowley. Oxford: Oxford University Press, 1910.

Gogel, Sandra L. 1998. Grammar of Epigraphic Hebrew. Resources for Biblical Study 23. Atlanta: Scholars.

Golinets, Viktor. Forthcoming. ‘Some Considerations on Questions Philology Cannot Solve while Reconstructing the Text of the Hebrew Bible’. Proceedings of the International Colloquium Philology and Textual Criticism of the Hebrew Bible (Orbis Biblicus et Orientalis), ed. I. Himbaza.

Groves, J. Alan, and Dale Wheeler. 2005. Groves-Wheeler Westminster Theological Seminary Hebrew Morphology Database, version 4.14. Glenside, PA: J. Alan Groves Center for Advanced Biblical Research.

Hadley, Judith M. 1987. ‘The Khirbet el-Qom Inscription’. Vetus Testamentum 37 (1) 50–62.

Hendel, Ronald. 2016. Steps to a New Edition of the Hebrew Bible. Text-Critical Studies 10. Atlanta, GA: SBL Press.

Hornkohl, Aaron D. 2018. ‘Diachronic Exceptions in the Comparison of Tiberian and Qumran Hebrew: The Preservation of Early Linguistic Features in Dead Sea Scrolls Biblical Hebrew’. The Reconfiguration of Hebrew in the Hellenistic Period: Proceedings of the Seventh International Symposium on the Hebrew of the Dead Sea Scrolls and Ben Sira at Strasbourg University, June 2014, edited by Jan Joosten, Daniel Machiela, and Jean-Sébastien Rey, 61–92. Studies on the Texts of the Desert of Judah 124. Leiden: Brill.

Hughes, Jeremy. 1994. ‘Post-Biblical Features of Biblical Hebrew Vocalization’. Language, Theology, and the Bible: Essays in Honour of James Barr, edited by S.E. Balentine and J. Barton, 67–80. Oxford: Clarendon.

Hurvitz, Avi. 2014. A Concise Lexicon of Late Biblical Hebrew. Supplements to Vetus Testamentum 160. Leiden: Brill.

Joosten, Jan. 2015. ‘The Tiberian Vocalization and the Hebrew of the Second Temple Period’. Hebrew of the Late Second Temple Period: Proceedings of a Sixth International Symposium on the Hebrew of the Dead Sea Scrolls and Ben Sira, edited by Eibert Tigchelaar and Pierre Van Hecke, 25–36. Studies on the Text of the Desert of Judah 114. Leiden: Brill.

JM = Joüon, Paul and Takamitsu Muraoka. 2006. A Grammar of Biblical Hebrew. Subsidia Biblica 27. Rome: Pontifical Biblical Institute.

Kantor, Benjamin. 2017. ‘The Second Column (Secunda) of Origen’s Hexapla in Light of Greek Pronunciation’. PhD dissertation, The University of Texas.

Khan, Geoffrey. 2013a. A Short Introduction to the Tiberian Masoretic Bible and Its Reading Tradition. Gorgias Handbooks. Piscataway, NJ: Gorgias.

——. 2013b. “Shewa: Pre-Modern Hebrew.” EHLL 3, 543–54.

——. 2013c. “Tiberian Reading Tradition.” EHLL 3, 769–78.

——. 2020. The Tiberian Pronunciation Tradition of Biblical Hebrew: Including a Critical Edition and English Translation of the Sections on Consonants and Vowels in the Masoretic Treatise Hidāyat al-Qāriʾ ‘Guide for the Reader.’ 2 vols. Cambridge Semitic Languages and Cultures 1. Cambridge: University of Cambridge and Open Book Publishers.

Knauf, Ernst A. 1990. ‘War„ Biblisch-Hebräisch ‟eine Sprache? — Empirische Gesichtspunkte zur linguistischen Annäherung an die Sprache der althebräischen Literatur’. Zeitschrift für Althebraistik 3: 11–23.

Kutscher, Eduard Yehezkel. 1982. A History of the Hebrew Language. Leiden: Brill and Jerusalem: Magnes.

Lemaire, André. 1977. ‘Les inscriptions de Khirbet el-Qôm et L’Ashérah de YHWH’. Revue Biblique 84: 595–608.

Malessa, Michael. 2006. Untersuchungen zur verbalen Valenz im biblischen Hebräisch. Studia Semitica Neerlandica 49. Assen: Van Gorcum.

Mishor, Mordechai. 1993. ‘ʿAyin ha-Poʿal Rafa ba-Maqor “lifʿol”’. Ḥiqre ʿEver va-ʿArav Mugašim li-Yhošuaʿ Blau ʿal yede Ḥaverav ve-Talmidav bi-Mlot lo Šivʿim, edited by Ḥaggai Ben-Shamai, 383–95. Tel-Aviv: Tel-Aviv University.

Morag, Shelomo. 1974. ‘On the Historical Validity of the Vocalization of the Hebrew Bible’. Journal of the American Oriental Society 94: 307–15 (an abridged translation of ‘Ha-Masoret ha-Ṭavranit šel Lešon ha-Miqra: Homogeniyut ve-Heṭerogeniyut’. Peraqim—Sefer ha-Šana šel Meḵon Šoqen 2 [1974]: 105–44).

Orlinsky, Harry M. 1942–1943. ‘The Biblical Prepositions táḥaṯ, bēn, báʿaḏ, and Pronouns ʾa (or ʾanū), zōṯāh’. Hebrew University College Annual 17: 267–92.

Pardee, Dennis. 1988. Review of Andersen and Forbes 1986. Catholic Biblical Quarterly 50: 276–80.

Qimron, Elisha. 1986. The Hebrew of the Dead Sea Scrolls. Harvard Semitic Studies 29. Atlanta, GA: Scholars.

——. 2018. A Grammar of the Dead Sea Scrolls. Jerusalem: Yad Yizhak Ben-Zvi.

Reymond, Eric. 2014. Qumran Hebrew: An Overview of Orthography, Phonology, and Morphology. Society of Biblical Literature Resources for Biblical Study 76. Atlanta, GA: SBL Press.

——. 2016. ‘The Passive Qal in the Hebrew of the Second Temple Period, Especially as Found in the Wisdom of Ben Sira’. Sibyls, Scriptures, and Scrolls: John Collins at Seventy, edited by Joel S. Baden, Hindy Najman, and Eibert J. C. Tigchelaar, vol. 1, 1110–27. Supplements to the Journal for the Study of Judaism 175. Leiden: Brill.

Rollston, Christopher. 2003. ‘Non-provenanced Epigraphs I: Pillaged Antiquities, Northwest Semitic Forgeries, and Protocols for Laboratory Tests’. Maarav 10: 135–93.

——. 2006. ‘Navigating the Epigraphic Storm: A Palaeographer Reflects on Inscriptions from the Market’. Near Eastern Archaeology 68: 69–72.

Sarfatti, Gad Ben Ami. 1982. ‘Hebrew Inscriptions of the First Temple Period: A Survey and Some Linguistic Comments’. Maarav 3: 55–83.

Tov, Emanuel. 2012. Textual Criticism of the Hebrew Bible. 3rd edition revised and expanded. Minneapolis, MN: Fortress.

Yeivin, Israel. 1985. The Hebrew Language Tradition as Reflected in the Babylonian Vocalization. 2 vols. Jerusalem: Academy of the Hebrew Language (in Hebrew).

Yuditsky, Alexey (Eliyahu). 2017. A Grammar of the Hebrew of Origen’s Transcriptions. Meqorot u-Meḥqarim 16. Jerusalem: Academy of the Hebrew Language (in Hebrew).

Zevit, Ziony. 1980. Matres Lectionis in Ancient Hebrew Epigraphs. American Schools of Oriental Research Monograph Series 2. Cambridge, MA: American Schools of Oriental Research.

——. 1984. ‘The Khirbet el-Qom Inscription Mentioning a Goddess’. Bulletin of the American Schools of Oriental Research 255: 39–47.


1 One of the best-known and oft-described examples is the shift from qal internal passive to alternative forms via reanalysis as puʿʿal or hofʿal or revocalisation as nifʿal; for a recent discussion and up-to-date bibliography see Reymond (2016, 1135, nn. 5–8). Hughes (1994) collects a number of further phenomena, as do the studies listed below, n. 2. Incidentally, while the issue is not treated here, it bears mentioning that the so-called consonantal text is not, in fact, purely consonantal. While it is legitimate to suppose that the earlier portions of the Hebrew Bible were composed in a more purely consonantal orthography, it seems that they were later subjected to a revision involving the insertion of matres lectionis in line with the Second Temple orthographical conventions employed in the composition of LBH texts.

2 Examples of nuanced presentations include those found of Barr (1968, 188–222; 1981, 27, 35–36; 1984, esp. 31); Morag (1974); Khan (2013a, 46–51); Joosten (2015). See also Tov (2012, 46–47).

3 See also Blau (2010, 115):
The qal infinitives construct present a complex picture, since after the followed by bgdkpt the form has a quiescent šwa. Such forms as Image ‘in order to break’ are due to morphological reshuffling on analogy to the prefix-tense (Image‘he will break’) rather than to a genuine sound shift. The late date of this feature is indicated by forms like Image ‘that I fall’ Ps 118: 13; the n immediately preceding another consonant was not assimilated to it because at the time of the action of this shift the n was still followed by a mobile šwa. (Alternatively, one could suggest that this shift was still active, but that at the time of the vocalization of the biblical text its letters had already become hallowed and therefore the נ of לנפל could not be omitted…).

4 On the Tiberian realisation of shewa see Khan (2013b, 546; 2013c, 775; 2020, 305–20).

5 Exceptions with -ל and spirant II-bgdkpt are Image ‘to serve’ (Num. 4.23 [L]; 8.24 [L]); Image ‘to go around’ (Num. 21.4 [L]); Image ‘to harm’ (1 Sam. 22.17 [L/A]); Image ‘to pursue you’ (1 Sam. 25.29 [L/A]); ְ ImageImage ‘to uproot and demolish’ (Jer. 1.10 [L/A]; 18.7 [L/A]; 31.28 [L/A missing]); Image ‘to devastate’ (Jer. 47.4 [L/A]); Image ‘to slaughter’ (Jer. 11.19 [L/A]; 25.34 [L/A]; 51.40 [L/A]; Ps. 37.14 [L/A]); Image ‘to stray’ (Prov. 19.27 [L/A]). Exceptions with -ב or -כ and plosive II-bgdkpt are rarer: Image ‘while dwelling’ (Gen. 35.22 [L/A missing]); Image ‘upon remembering’ (Jer. 17.2 [L/A]); Image ‘by piling up’ (Ezek. 17.17 [L/A]). GKC (§45g) and Mishor (1993, 385–86) present slightly different lists.

6 On the two major exceptional categories to this tendency, namely n preceding a guttural and forms of verbs III-n, see Blau (2010, 77).

7 Thus Fox (2003, 205). This form is not to be confused with the presumed antecedent of the qal absolute infinitive, *qatāl. JM (§49a) posits underlying qṭul with initial cluster, on which assumption a secondary epenthetic vowel is responsible for the fricativisation of the following bgdkpt radical.

8 For example, the bgdkpt consonant in the second position of Image [liɣˈboːθ] ‘to collect (payment)’ may be plosive on analogy to yiqṭolImage [jiɣˈbɛː], but the ֹות- [-oːθ] ending was retained.

9 While the characteristic RH morphological similarity between prefix conjugation and construct infinitive must be considered a secondary repatterning in the case of most weak verb classes, the situation is more complicated when it comes to historically stative and II/III-guttural verbs. It is widely held that in an early stage of Hebrew, i.e., pre-Rabbinic and pre-Tiberian, the a theme-vowel of stative and II/III-guttural prefix-conjugation forms also characterised the corresponding infinitive construct, i.e., (li) qVṭal || yiqṭal. Due to the pressure of analogical leveling, (li) qVṭal infinitival forms came to have an o theme-vowel, leaving only a few Tiberian BH remnants in Image [ʃaˈxaːv] ‘lie down’ (Gen. 34.7 + 10x), Image [gaˈvaːʕ] ‘die’ (Num. 20.3; but cf. pausal Image [ligˈvoːaʕ; Num. 17.28]), and Image [ʃaˈfaːl] ‘be low’ (Prov. 16.19; Qoh. 12.4) (see Barth 1891, 106–7; Fox 2003, 216; JM, §49c). RH’s marked proclivity for liqṭal might be interpreted as a case of conservatism vis-à-vis Tiberian BH. However, it is instructive that the o theme-vowel is not at all uncommon in RH stative and II/III-guttural infinitives. Indeed, in the case of II/III-guttural verbs, the dominant orthography in RH is with mater waw, even if the corresponding prefix-conjugation form has a as theme-vowel. Given this situation, it would seem either that analogy to the RH prefix conjugation pattern led to a RH shift of liqṭol to liqṭal, which coincidentally recreated an ancient but obsolete dichotomy, or that this ancient moribund dichotomy was sporadically preserved thanks to casual identity with the results of the analogical repatterning described above. Cf. Kutscher’s (1982, 38–39) notion of ‘mirage forms’. See further n. 14.

10 The RH forms are from Codex Kaufmann (=K) of the Mishnah.

11 It is assumed here that the consonantal text always had an accompanying reading tradition (or traditions). See Barr (1981, 35) and Tov (2012, 40–41).

12 Alternatively, perhaps the nunation here results from dissimilation or prenasalisation (ng < gg) under Aramaic influence. While Ancient Hebrew exhibits sporadic examples of lC < CC and rC < CC (Blau 2010, 57–58), nC < CC is particularly characteristic of late Imperial Aramaic, including Qumran Aramaic (Garr 2007). My thanks to Steven Fassberg for alerting me to this line of argumentation.

13 This is an allusion to ‘when Yhwh, your God, brings you into the land to which you are coming to possess, he will remove Image many nations before you… ’ (MT Deut. 7.1; see also v. 22), where the verb is clearly Image Alternatively, DSS לשול is a geminate biform *laššol related to של־ Image ‘remove (from the sheaves)’ (MT Ruth 2.16), though Image in the relevant sense is a BH hapax.

14 In light of the discussion above in n. 9, Image and לשול likely represent relatively early orthographical evidence of secondary remodeling of the earlier (li) qVṭal pattern according to the dominant liqṭol alternative. That the secondary liqṭal > liqṭol shift is characteristic of both Qumran and the Tiberian reading tradition indicates the antiquity of the phenomenon. That the Hebrew of the DSS seems farther along in the process testifies to the conservative nature of the Tiberian tradition; cf. the spelling ולשכוב veliškov in 4Q51 f89–92.15 (|| MT Image [valiʃˈkaːv] 2 Sam. 11.11); see also 4Q160 f7.4; 4Q223–224 f2v.3.

15 See GKC (§66f).

16 While the dichotomy between biblical and non-biblical in material from the Judaean Desert is problematic and anachronistic, there are palpable linguistic differences between the so-called biblical scrolls and non-biblical scrolls, in that the Hebrew of the former (DSSBH) is more conservative than that of the latter (QH =Qumran Hebrew). See below, n. 23.

17 By way of comparison, in other instances, the preposition -ל prefixed to qal I-y infinitives was evidently still perceived as a true preposition not integral to the form and retaining semantic force, as in the case of so-called ‘temporal -ל’, e.g., Image ‘after Abram had lived ten years in the land’ (MT Gen. 16.3) and ImageImage ‘in the third month from the time the children of Israel had left Egypt’ (MT Exod. 19.1), in which no pretonic lengthening took place. However, the preposition -ל is regularly vocalised with qameṣ even where it retains the semantic force of ‘in order to’, e.g., ְ ImageImage ‘And the man gazed at her, keeping silent in order to know whether Yhwh had prospered his journey or not’ (MT Gen. 24.21).

18 Blau (2010, 131).

19 Yeivin (1985, I, 487, 607, 641).

20 Exceptional forms include those with guttural root letters. Some MT qal construct infinitives are analysed in the Samaritan tradition as nouns, finite verbs, or infinitives in another binyan.

21 The forms are collected and discussed in Brønno (1943, 56–58); Yuditsky (2017, 131); Kantor (2017, 339, 352). I am indebted to my friend and colleague Ben Kantor for his help in comprehending the significance of the data.

22 To illustrate by means of a phenomenon already cited, while liqṭol forms of qal II-bgdkpt construct infinitives resemble RH yiqṭol forms, the forms of other biblical infinitives consonantally amenable to RH-style vocalisation—such as statives and III-guttural forms—largely preserve BH phonology, e.g., stative BH Image [lilˈboːʃ] ‘to wear’ versus RH Imagelilbaš and III-ʿ BH Image [lizˈroːaʕ] ‘to sow’ versus RH Image lizraʿ. Note that in the case of stative BH Image [liːˈʃoːn] ‘to sleep’ (MT Qoh. 5.11) versus RH Imagelišan the full spelling of the Tiberian consonantal text also bears witness to the phonological distinction between BH and RH. The qameṣ theme vowel in the RH forms Image and Image is interpreted here as reflecting an a-vowel similar to that represented by pataḥ in standard Tiberian vocalisation; K’s vocalisation tradition does not consistently differentiate between qameṣ and pataḥ. See also n. 16, above.

23 This is especially conspicuous when one contrasts Tiberian BH with material actually composed (as against that merely copied) in the late Second Temple period, especially that which is more representative of the vernacular, e.g., some material from the Judaean Desert and from rabbinic literature. It is worth noting that alternative biblical traditions, such as those represented by the Samaritan Pentateuch and biblical DSS material, also present a form of Hebrew somewhat out of line with authentic, especially colloquial, Second Temple Hebrew usage, in that they, too, regularly preserve usages no longer typical of contemporary Hebrew. Significantly, however, in comparison to Tiberian BH, both show greater incidence of linguistic contemporisation.

24 The present section is a revised abridgement of Hornkohl (2018, 72– 79).

25 Conditioning factors extend beyond the purely morphological. For example, the absence of certain forms from RH, such as temporal clauses employing biqṭol and kiq̣tol, is at least partially conditioned by genre and by the availability of alternative syntagms, e.g., those employing the gerundive verbal nouns known as šemot peʿula. As such, the undifferentiated nature of the foregoing statistics must be acknowledged. Further study of conditioning factors remains a desideratum.

26 See also MT 1ְKgs 18.29; 2ְKgs 23.10; Hab. 3.14; Ezra 10.14; 1 Chron. 5.9; 28.20; 2 Chron. 24.10; 26.8, 16; 29.28; 31.1. Most of the relevant cases involve examples of the expression Image on the characteristic lateness of which see Hurvitz (2014, 196–98).

27 See also 4Q109 f1ii+3–6i.18–19 || MT Qoh. 7.5.

28 On the formulation of negative oaths see JM §165, especially subsections d, f and g.

29 These statistics reflect the approach of Malessa (2006, 150–66), with slight modifications, as detailed in Hornkohl (2018, 73–74, n. 24). See also JM (§124l, n. 9).

30 For lists of occurrences see Hornkohl (2018, 75–76, nn. 25–28).

31 While the strong BA penchant for verbal complement infinitives with prefixed -ל tallies with Second Temple Hebrew practice, it should be noted that infinitives with prefixed prepositions, especially -ל, are the rule throughout all historical phases of Aramaic; see Fassberg (2007).

32 See also 1QIsaa 1.14–15 || MT Isa. 1.12; 1QIsaa 1.15 || MT Isa. 1.13; 1QIsaa 7.22–23 || MT Isa. 8.4; 1QIsaa 22.13–14 || MT Isa. 28.12; 1QIsaa 24.16 || MT Isa. 30.9; 1QIsaa 39.31 || MT Isa. 47.11; 4Q111 3.6 || MT Lam. 1.14. On apparent exceptions see Hornkohl (2018, 78–79). Also worthy of consideration are MT absolute infinitives functioning as verbal complements that are paralleled in the DSS by construct infinitives with - :ל1QIsaa 36.7 || MT Isa. 42.24; 1QIsaa 47.20 || MT Isa. 57.20.

33 It seems clear that יו- was indeed eventually taken as representative of the diphthong -aw (and its reflexes), since, beyond the 3ms suffix in question, several words ending in -aw are spelled with final יו- in ancient sources, e.g., Tiberian BH Image [jaħˈdɔːv] ‘together’ (thrice in the MT ketiv, against 94 times Image frequently in the DSS); עכשיו ʿaxšav ‘now’ (consistently in K; עכשו in 4Q225 f2ii.7); Tiberian BH Image [saˈθɔːv] ‘winter/autumn, rainy season’ (MT Song 2.11 qere; cf. ketiv סתו); Tiberian BH Image [ʕɔːˈnɔːv] ‘humble’ (MT qere; ketiv ענו; non-biblical Dead Sea Scrolls); DSS ע( י )שיו ‘Esau’ (cf. Tiberian BH Image [ʕeːˈsɔːv]) and תיו ‘hook’ (cf. Tiberian BH Image [tɔːv]). In Modern Israeli Hebrew, the correlation of the spelling יו- and the pronunciation -av has led to, among other things, the convention of writing the letter names of waw and tav as וי"ו and תי"ו, respectively (orthographically differentiated from the homophonous lexemes וו vav ‘hook and תו tav ‘musical note, mark’, respectively).

34 See also Orlinsky (1942–43, 288–89); Zevit (1980, 29–30); Pardee (1988, 279–80); Knauf (1990, 20); Freedman (1992, 9–10, no. 6); Golinets (forthcoming, 1–6).

35 Blau (2010, 172); Zevit (1980, 29–30).

36 GKC (§91i and n. 1); Cross and Freedman (1952, 47, 54–55, 68–9); Sarfatti (1982, 65); Gogel (1998, 159–161, nn. 187–189); Barkay (2004, 53–54); JM (§94d and n. 7); Khan (2013a, 48).

37 For inclusion of the variant 3ms endings among written-reading mismatches, see Khan (2013a, 48). Just a few of the many other notable dissonances involve the 2/3fpl verbal ending ְן- [-nɔː]; the standard Tiberian spelling ירושלם versus the accepted realisation Image [jaʀuːʃɔːˈlaːjim], the latter of which is better matched by the minority spelling with yod to mark the triphthong -ayi- ירושלים (only five occurrences in the MT: Jer. 26.18; Est. 2.6; 1 Chron. 3.5; 2 Chron. 25.1; 32.9), which also occurs in the DSS, RH, and Second Temple epigraphic and numismatic sources; the qal internal passive, forms of which are regularly understood/vocalised as puʿʿal, hofʿal, or nifʿal when possible; and perhaps cases of apparent suppletion, e.g., forms of the verb Image ‘approach’ which are nifʿal in the qatal and participle, but pattern as qal in the yiqtol, imperative, and infinitive.

38 For useful discussion see Gogel (1998, 159–61 and nn. 187–89) and Barkay et al. (2004, 53–54).

39 The dating is according to Barkay et al. (2004, 41–55). The spelling יו- is an important element in arguments for later datation of the inscription; see, e.g., Berlejung (2008, 208–12); Golinets (forthcoming); but see below.

40 Lemaire (1977, 599, 601); Zevit (1984, 43); Hadley (1987, 54–55); Gogel (1998, 159–60, n. 188).

41 The authenticity of this inscription has been called into question; see Rollston (2003; 2006).

42 See Reymond (2014, 144–47, 159) on the relative frequency of spelling variation in the DSS, though it is to be noted that he considers the yod of יו- to be a grammatical mater and assumes that the 3ms ending on plurals and relevant prepositions was realised as  <-aw whether spelled יו- or ו-; see Qimron (1986, 33–34, 59; 2018, 270).

43 Brønno (1943, 200–1); Yuditsky (2017, 107).

44 Ben-Ḥayyim (2000, 229).

45 This account is based on Florentin (2016, 74). Cross and Freedman (1952, 47) note that “[o] nly in a dialect in which the diphthong ay was preserved, would a form -āw < *-ayhū result.” Given Samarian ינ yn ‘wine’, presumably realised [yeːn] (cf. Tiberian [ˈyaːyin]), -ēhu < *-ayhu in the north. However, with קצ qṣ ‘summer fruit’, presumably realised [qeːsˤ] (cf. Tiberian קַיִץ [ˈqaːjisˤ]), at Gezer, a mere six miles (9.6 km) north and 20.5 miles (33 km) west of Jerusalem, it seems that diphthong contraction was not limited to the dialect of the far north.

46 Blau (2010, 172); Zevit (1980, 29–30).

47 In agreement with Florentin (2017, 73–74). Florentin has proposed a motivation for development of the -āw realisation. He assumes a base form *-ayhū leading to the rarely preserved Image [-eːhuː]. He then notes that the respective singular and plural forms of III-y substantives with the 3ms suffix are phonetically identical: Image ‘his deed’ and Image ‘his deeds’ both [maːʕaˈseːhuː]. By means of the standard contraction of the diphthong ay and elision of heh *-ēyū < *-ayhū. But since *-ēyū was too similar to the 3ms suffix for singulars, -ēhū, language users intentionally opted for a discernible alternative, namely *-āw < *-āhū < *-ayhū, thereby rendering the distinction between singulars and plurals transparent. From the sizeable minority of III-y forms the -āw suffix spread to others, becoming dominant. This approach satisfies on several levels. First, it gives due weight to much of the evidence, seeking to explain both the Image [-eːhuː] and -aw spellings and realisations and linking both of them back to *-ayhū. Second, it posits motivation for what must be seen as non-standard developments in the development of -aw. However, concerned mainly with explaining the phonetic forms in the Tiberian tradition, it is unfortunate that Florentin does not discuss the potential for phonetic variety in the inscriptional sources, the DSS, or in the MT itself, nor make explicit his view of the strange relationship between orthographical יו- and phonetic [-ɔːv].

48 Barkay et al. (2004, 54). Also worthy of consideration as an explanation for the orthography יו-, but not the spelling ו- or for phonetic realisations deriving from -aw, is the entirely conjectural possibility that triphthongal *-ayyū < *-ayhū, postulating assimilation of heh to the preceding yod (cf., e.g., Image Tiberian [gamɔːˈlattuː] < gamalathu [1 Sam. 1.24]).

49 CD 10.9; 1QS 3.7–8; 6.17; 11.3; 1QSa 1.18, 22; 1QpHab 3.7; 5.5; 9.1; 1QHa 4.37; 4Q163 f4–7i.8; 4Q200 f6.3; 4Q216 5.3, 9; 4Q221 f1.2; f3.5; f5.2; 4Q228 f1i.4; 4Q255 f2.2, 6; 4Q261 f1a–b.3; 4Q262 fB.1–2; 4Q266 f2i.4; f2ii.2, 4; f5ii.2, 4; f6iii.8; 4Q270 f6iv.14, 19; 4Q299 f3c.6; 4Q365 f12biii.5; f26a–b.8; 4Q374 f2ii.8; 4Q381 f31.3; 4Q387 f2iii.1; 4Q392 f1.4–5, 9; 4Q398 f14–17ii.4, 7; 4Q403 f1i.43; 4Q404 f5.6; 4Q405 f15ii– 16.4; f20ii–22.7; f23i.13; 4Q417 f29i.7; 4Q418 f16.4; 4Q434 f1i.7; 4Q468b f1.2; 4Q472 f1.4; 4Q481d f3.2; 11Q17 7.6; 10.5.

50 1QIsaa 1.28; 11.29; 23.2; 27.23; 2Q16 f5ii–6i.1; 4Q32 f2ii+3i+4.19; 4Q56 f3ii.15; 4Q86 2.13; 4Q93 1.9; 4Q98f f1–2.1; 4Q114 1.3; 4Q128 f1.21, 29; 4Q138 f1.2, 9, 11; 4Q140 f1.26; 11Q1 fK–Li.7; 11Q5 4.15; Mur88 8.16.

51 Exod. 27.11; 28.28; 32.19; 37.8; 39.4, 33; Lev. 9.22; 16.21; Deut. 2.33; 7.9; 8.2; 33.9; Josh. 16.3; 1 Sam. 3.2; 8.3; 10.21; 23.5; 26.7, 11, 16; 30.6; 2 Sam. 1.11; 12.9, 20; 13.34; 18.17, 18; 19.19; 22.23; 24.14, 22; 1 Kgs 5.17; 6.38; 16.19; 18.42; 2 Kgs 4.34; 5.9; 11.18; 14.12; Jer. 17.10, 11; 22.4; 32.4; Ezek. 17.21; 18.21, 24; 31.5; 33.13, 16; 37.16, 19; 40.6, 9, 21 (3x), 22 (4x), 24 (2x), 25, 26 (3x), 29 (4x), 31 (2x), 33 (4x), 34 (3x), 36 (3x), 37 (3x); 43.11, 26; 44.5; 47.11; Amos 9.6; Obad. 1.11; Hab. 3.14; Ps. 10.5; 58.8; 105.18, 28 (?); 106.45; 147.19; 148.2; Job 5.18; 14.5; 20.11; 21.20; 26.14; 27.15; 31.20; 37.12; 38.41; 39.26, 30; 40.17; Prov. 16.27; 21.29; 22.25; 26.24; 30.10; Ruth 3.14; Qoh. 4.8; 5.17; Lam. 3.32, 39; Dan. 9.12; 11.10; Ezra 4.7.

52 See, by way of illustration, the useful examples in Golinets (forthcoming, 1–6).

53 Albright (1943, 22, n. 27) argues that ירחו in the Gezer Calendar ends in -êw, comparing to “the archaic uncontracted form -êhu which appears a number of times in Hebrew poetry” (see Bauer and Leander 1918– 1922, §28v). Tiberian -āw he explains (ibid.) as “an obvious conformation to the parallel Aramaic suffix which appears in Bib. Aram. as -ôhī and in Syriac as -auhī (written) and -au (pronounced).” Cross and Freedman (1952, 47, 54–55, 68–9) view Masoretic יו- as “reminiscent” (47) of North Israelite -êw and Tiberian -āw as Judahite. Pardee (1988, 179–80) objects, asking why this northern feature, and no other, should figure so prominently in the Tiberian reading tradition. Without denying an areal explanation, I am content with a more general view of variety in ancient Hebrew, incorporating such parameters as register, region, urban versus rural, diachrony, sociolect, and idiolect.

54 Presumably, some of the forms left without yod are the results of simple inconsistency in the revision or were read as suffixes for singulars. However, if yod was added to reflect a triphthongal realisation, then it is possible that it was intentionally left out of forms where, according to the reading tradition, a diphthong was intended.


(PhD, The Hebrew University of Jerusalem, 2012) is the author of Linguistic Periodization and the Language of Jeremiah (Brill, 2013), a translated adaptation of his doctoral dissertation. He holds the positions of Hebrew Language Teaching Officer and Senior Research Associate in the Faculty of Asian and Middle Eastern Studies, University of Cambridge, where he teaches modern and ancient Hebrew. His research is philological/linguistic in nature, focusing on ancient Hebrew, and encompassing diachrony and linguistic periodisation; syntax, pragmatics, and the verbal system; the Tiberian written and reading traditions and non-Tiberian Hebrew traditions; textual criticism and literary formation; and historical and contemporary exegesis.


Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search