Version classiqueVersion mobile

Studies in Semitic Vocalisation and Reading Traditions

Aaron D. Hornkohl
Geoffrey Khan

The Representation of Gutturals by Vowels in the LXX of 2 Esdras

Peter Myers

Texte intégral

1.0. Introduction

  • 1 This, as well as other phonetic transcriptions, represents the Tiberian pronunciation tradition, wh (...)

Greek transcriptions in the LXX are an important source of data for reconstructing the sounds of ancient Hebrew. Yet, given that Greek and Latin both possess a single laryngeal consonant /h/, opinions differ on the extent to which transcriptions into their scripts can provide evidence for the realisation of Hebrew gutturals, which include both laryngeals and pharyngeals. A minimalist view is that “with the exception of the quiescent Latin h in certain positions, the glottals are practically never represented by a transcription sign” (Murtonen 1981, 68). Rather than direct transcription, evidence for gutturals can instead be detected by their effect on nearby vowel changes, “in the Septuagint, a helping vowel can occasionally be found in the vicinity of original gutturals, e.g. νωε ַ נ ֹ ח [noːaħ]1 ‘Noah’. In the Hexapla, one finds helping vowels after expected gutturals, e.g. νεεμαν Image [nɛːʔɛmɔːn] ‘enduring’ (89.38)” (Yuditsky 2013, 805b).

A well discussed exception are two gutturals that are sometimes represented by consonants in the LXX. The graphemes < ח > and < ע > were originally polyphonous, each representing two phonemes. < ח > represented a voiceless velar fricative /*ḫ/ [x] (corresponding to Arabic Image) and a voiceless pharyngeal fricative /ḥ/ [ħ] (corresponding to Arabic Image). < ע > represented a voiced velar fricative /*ǵ/ [ɣ] (corresponding to Arabic Image) and a voiced pharyngeal fricative /ʿ/ [ʕ] (corresponding to ArabicImage). Blau argued that in the LXX “most proper nouns” containing < ח > are transcribed “by zero/vowel mutation or by χ” (Blau 1983, 43 [147§ ]12) and those containing < ע > “by zero/vowel mutation or by γ” (Blau 1983, 5 [109] §6). < ח > and < ע > are transcribed by <χ> and <γ> when they represent the velar fricatives /*ḫ/ and /*ǵ/. These correspondences are most consistent in Genesis, then the rest of the Pentateuch (Blau 1983, 39 [143] §9.2). They are less consistent in the rest of the LXX books, which were translated later, due to the loss of the velar fricatives /*ḫ/ and /*ǵ/ from “the spoken language.” He argued that there are no cases of < ח > and < ע > reflected by <χ> or <γ> in Ezra–Nehemiah, which therefore demonstrates that these books must have been translated last (Blau 1983, 71 [175] §15.1), and that by this time the velar fricatives /*ḫ/ and /*ǵ/ had also been lost from the “literary solemn language, as in the public reading of the Bible in synagogues” (Blau 1983, 39–40 [143–44] §2–3).

In contrast to the above authors, Krašovec describes gutturals as sometimes being directly represented in the LXX by Greek vowel graphemes. In the case of /ḥ/ he cites an example from Deut. 3.8, where Image is represented as Ἀερμών (Krašovec 2010, 24). Krašovec’s monograph on Biblical Hebrew names in Greek and Latin has not caught the attention of many Semitists. Yet, from my observation of the extant textual witnesses to 2 Esdras, the Greek translation of Ezra–Nehemiah, the phenomenon occurs far more often in this corpus than Krašovec describes for the LXX as a whole.

1If this observation is correct, then it is quite intriguing, because it would mean not only that the LXX provides more evidence for the pronunciation of gutturals than is often realised, but also that the direct representation of gutturals in 2 Esdras occurred much later than one might expect. Building on Blau’s work by comparing transcriptions attested in inscriptions, Steiner (2005) dated 2 Esdras to the mid-late second century CE. One of the characteristic features of Hebrew in the Second Temple and Tannatic periods is the weakening of guttural consonants, which is reflected by confusion of guttural graphemes in some sources (Mor 2013). Therefore, if the translator of 2 Esdras did attempt to render gutturals directly, his work may provide helpful data for nuancing our understanding of how the pronunciation of these consonants developed.

2.0. Textual Transmission

2Transcription spellings where gutturals are apparently represented by vowels are rarely attested by all, or even most, manuscripts at any given place in 2 Esdras. Neither do these spellings tend to be attested in the modern critical editions. Such spellings could be discounted as having arisen due to corruptions in transmission. The corruptions required to produce them, however, would be the random addition of <α> or <ε> that by chance alone happen to correlate with the distribution of gutturals in the underlying Hebrew word. A simpler explanation is that the transcription of gutturals as vowels creates spellings that are unusual in Greek. So, given that manuscripts are copies made by Greek speakers, who very likely had little to no knowledge of Hebrew, errors in transmission are more likely to remove these spellings than create them.

3One justification for this claim is that transcription of gutturals by vowels creates vowel hiatus in the Greek transcription, e.g., the aforementioned example cited by Krašovec, Ἀερμών, where the transcription of < ח > by <α> has resulted in the hiatus <αε>. Vowel hiatus was not comfortable for a Greek speaker and, therefore, such spellings, especially in foreign and unfamiliar words, were more liable to undergo development in transmission. Such changes were probably unintentional, but unintentional does not mean entirely random. Whatever the method by which a manuscript was copied or the mechanism by which a mistake was made, the most likely output is a spelling that more closely resembles Greek phonotactic patterns.

  • 2 The Greek manuscript sigla and notation used in this article are taken from Hanhart (1993) with min (...)

All typical developments in transmission that reduce vowel hiatus can be illustrated from transcriptions of Image ‘Reaiah’, which occurs at Ezra 2.47 and Neh. 7.50. In both places all variants can be explained as developments from ρααια. This form is attested at Neh. 7.50 by A V Gᴸ a-group (except ραδαια 370) b-group (except ραια 98– [379]) 119, but not attested in any manuscript at Ezra 2.47:2

(1) ραβαια 55, ραδαια 370 (Neh. 7.50): insertion of a consonant
(2) αραια Gᴸ (Neh. 7.50): metathesis
(3) ραεα B– [122] S (Neh. 7.50): phonetic substitution of a simpler grapheme, in this case αι: ε
(4) ραια 98– [379] (Neh. 7.50): haplography
(5) ρεηα B–55 > ρεηδ 122 (Ezra 2.47): graphic confusion of a vowel grapheme with a consonant grapheme, in this case Α: Δ

4When evaluating the manuscript readings, I suggest the harder reading is usually the reading that involves vowel hiatus. Therefore, when reconstructing the original text of transcriptions, developments that remove vowel hiatus are more likely to have occurred than developments that create vowel hiatus. Applying this principle to the extant witnesses of 2 Esdras, there are a number of places where it suggests a vowel should be reconstructed in the original text, or a minority reading with a vowel should be accepted over a majority reading. On almost all these occasions the extra vowel corresponds to a guttural in the Hebrew-Aramaic consonantal text or to a vowel in the Tiberian reading tradition. All the examples cited below are my reconstructions of the best text for the transcription in the Old Greek (G) or the Lucianic recension (Gᴸ) via application of this method. For the sake of caution, I have been deliberately ambivalent regarding other less-clearly attested spelling features, placing them in square brackets, []. A starred, *, spelling is my reconstruction of the text that best explains the extant readings, but is not itself attested in any manuscript. In all cases the reader can assess my decisions against the manuscript data by consulting the relevant place in the apparatus of Hanhart’s (1993) edition.

3.0. Gutturals in Syllable Onset

5The most straightforward examples of Hebrew gutturals represented by Greek vowels are word-medial (or construct-chain-medial) gutturals in syllable onset.

3.1. ʾAleph /ʾ/

6All such cases involving /ʾ/ are in the divine element א ֵ ל . The phoneme is usually represented by <ε>, in one lexeme by <α>, and in one, possibly two, cases by <ι>:

(6) Image ‘Bezalel’ (Ezra 10.30) βεσ [σ] ελ [ε] ηλ G βεσσελεηλ Gᴸ

7The <ε> corresponding to /ʾ/ is attested only by some later G manuscripts in the a-group. It is probably not original to G, but due to harmonisation to the Gᴸ tradition, where the guttural has been transcribed.

(7) Image‘Hananel’ (Neh. 3.1) ανανεηλ G ανενεηλ Gᴸ
(8) Image‘Tabel’ (Ezra 4.7) ταβεηλ G Gᴸ
(9) Image ‘Mehetabel’ (Neh. 6.10) μεηταβεηλ G μετεβεηλ Gᴸ
(10) Image ‘Mahalalel’ (Neh. 11.4) μαλελεηλ G Gᴸ
(11) Image ‘Mesezabeel’ (Neh. 10.22) μεσωζεβηλ G μασση ζαβιηλ Gᴸ and Image (Neh. 11.24) *μασηζα βεηλ G μασσιζαβεηλ Gᴸ

Note in the first case /ʾ/ is represented by <ι>. In the following instance of this word there is no evidence the guttural is represented: Image (Neh. 3.4) μασεζεβηλ G *μασσηζαβελ Gᴸ.

(12) Image‘Natanel’ (Ezra 10.22) ναθαναηλ G Gᴸ

8As in (11), /ʾ/ may also be represented by <ι> in the following example. The Greek transcription, however, reflects a different syllabification from the Tiberian tradition, so <ι> may simply represent a vowel:

(13) Image‘Azarel’ (Ezra 10.41) εζριηλ G Gᴸ

3.2. Ḥet /ḥ/

9The phoneme /ḥ/ is usually represented by <α>, but in two lexemes by <ε>:

(14) Image‘Esarhaddon’ (Ezra 4.2) *ασαρεαδδων G
(15) Image‘the offering’ (Neh. 13.5) μαναα G
(16) Image ‘Pilha’ (Neh. 10.25) φαλα [ε] ι G φαλλαε ι Gᴸ
(17) Image‘Pashur’ (Ezra 2.38) φασ [σ] ουρ G φαδδας Gᴸ, Image (Ezra 10.22) φασουρ G φασσουρ Gᴸ, Image (Neh. 7.41) φασεουρ G φαδασσουρ Gᴸ, Image (Neh. 10.41) φασουρ G φασσουρ Gᴸ, Image (Neh. 11.12) φασεουρ G φασσουρ Gᴸ
(18) Image‘Siha’ (Ezra 2.43) σουαα G σουδαι Gᴸ, Image (Neh. 7.46) σιαα G σουδαι Gᴸ

10The spellings in Ezra 2 reflect a Vorlage read as * אחוצ . In Neh. 7, Gᴸ has been harmonised to Ezra 2.

(19) Image‘Tel Haresa’ (Ezra 2.59) θελαρησα G θελααρησ [σ] α Gᴸ

3.3. ʿAyin /ʿ/

11The phoneme /ʿ/ is evenly represented by <α> and <ε>:

(20) Image‘Balaam’ (Neh. 13.2) βαλααμ G Gᴸ
(21) Image‘Gibeon’ (Neh. 3.7) γαβαων G Gᴸ
(22) Image‘the Gibeonite’ (Neh. 3.7) γαβαωνιτης G γαβαων [ε] ιτης Gᴸ
(23) Image ‘the Gileadite’ (Ezra 2.61) γαλααδιτου G Gᴸ
(24) Image‘and with Sarah’ (Neh. 11.29) σαραα Gᴸ
(25) Image‘with Paroh’ (Neh. 9.10) φαραω G Gᴸ
(26) Image‘Paros’ (Neh. 3.25) φορος G φορεως Gᴸ

In the following instance of this word there is no evidence the guttural is represented: Image (Neh. 7.8) φορος G φαρες Gᴸ.

(27) Image‘Eleazar’ (Ezra 7.5) ελεαζαρ G Gᴸ
(28) Image‘Simeon’ (Ezra 10.31) σεμεων G συμεων Gᴸ
(29) Image‘and Simei’ (Ezra 10.23) σαμου G σεμεει Gᴸ
(30) Image ‘Simei’ (Ezra 10.33) σεμ [ε] ει G Gᴸ

3.4. He /h/

12There are no transcriptions of words where /h/ is attested in word-medial syllable onset. When a construct chain is transcribed with the definite article, the vowel is transcribed, but never the consonant /h/, e.g.,

(31) Image ‘Pakeret of the Sebaim’ (Ezra 2.57) φαχεραθ ασεβωειμ G φακεραθ σαβωειμ Gᴸ

4.0. Gutturals in Syllable Coda

13In contrast to gutturals in syllable onset, it is harder to evaluate the data relating to gutturals in the coda. Both word-medial and word-final gutturals in the coda are often accompanied by epenthetic vowels in the Tiberian reading tradition. Therefore, the transcription of gutturals by vowels must be carefully distinguished from cases where gutturals have conditioned vowel changes.

14For example, the final vowels in the following transcriptions correlate with furtive pataḥ in the Tiberian reading tradition, and so can be interpreted as transcriptions of this phenomenon:

(32) Image‘Zanoah’ (Neh. 11.30) ζανωε Gᴸ
(33) Image ‘Neziah’ (Ezra 2.54) *νασουε G

15In some cases a vowel change is transcribed even though in the Tiberian tradition the guttural has weakened such that no furtive pataḥ is pronounced. Either the guttural is strong enough to effect the sound change in the speech of the Greek translator, or the translator is transcribing the guttural itself:

(34) Image‘Zattua’ (Ezra 2.8) ζαθθουα G Gᴸ

16There are a small number of cases where a word-final Greek vowel grapheme correlates with a guttural and the grapheme is harder to explain as merely representing an epenthetic vowel. In these cases the penultimate vowel is written as either <α> or <ε>. Therefore, the final vowel does not correspond to a furtive pataḥ, as no significant change to the quality of the vowel is necessary in order to articulate the following guttural. Therefore, the vowel grapheme probably represents the guttural itself:

(35) Image ‘Arah’ (Ezra 2.5) *ηραε G ωρεε Gᴸ, Image (Neh. 6.18) ηραε G ηιρα Gᴸ

In the following instance of this word there is no evidence that the guttural is represented: Image (Neh. 7.10) ηρα G ηιρα Gᴸ

(36) Image ‘Tamah’ (Ezra 2.53) θεμα G θεμαα Gᴸ, Image (Neh. 7.55) θημα G θεμαα Gᴸ
(37) Image ‘and in Beer Sheba’ (Neh. 11.27) βεηρσαβεε G βηρσαβεαι Gᴸ, Image (Neh. 11.30) βεηρσαβεε G βηρσαβεαι Gᴸ
(38) Image‘and Geba’ (Ezra 2.26) γαβαα G Gᴸ, Image (Neh. 7.30) γαβαα G Gᴸ, Image (Neh. 11.31) γαβαα G Gᴸ, Image (Neh. 12.29) γαβαε Gᴸ
(39) Image‘Yoyada’ (Neh. 3.6) ιο [ε] ιδα G ιωδαε Gᴸ

17In the other four places where this lexeme occurs, Neh. 12.10, 11, 22; 13.28, the transcription is spelt ιωιαδα in both traditions.

5.0. Conclusion

18To summarise the data presented above:

In syllable onset /ʾ/ in the morpheme Image is often represented by <ε> and in one lexeme by <α>, /ḥ/ is usually represented by <α>, but in two lexemes by <ε>, while representations of /ʿ/ are evenly distributed between <α> and <ε>.
In a small number of lexemes, word-final /ḥ/ and /ʿ/ are represented by <α> or <ε>.
In 2 Esdras, there are no examples of the transcription of /h/.

The lexemes Image and Image etymologically possessed the velar fricative /*ḫ/, and the lexeme Image etymologically possessed the velar fricative /*ǵ/. These lexemes were among those identified by Blau as evidence that those consonants had been lost by the time of the translation of Ezra-Nehemiah. Our data suggests that Blau’s case is even stronger than he claimed, as these lexemes not only lack χ and γ, but the gutturals are transcribed with Greek vowel letters, which are typical ways the translator transcribes the phonemes /ḥ/ and /ʿ/.

19There are no transcriptions of gutturals in word-initial position, which may reflect the weakening or loss of gutturals in this position. However, in the majority of cases when /ʾ/, /ḥ/, or /ʿ/ occur in a word-medial syllable coda after a consonant (and in example (18) after a vowel) they are transcribed with a Greek vowel grapheme. Gutturals may therefore have been stronger within the word than at the beginning, though this conclusion would be typologically unusual. These findings will be better contextualised by a similar analysis of transcription spellings in other LXX books that also takes into account the specific factors that are likely to have affected the development of this particular class of word.


6.0. References

Blau, Joshua. 1983. ‘On Polyphony in Biblical Hebrew’. In Proceedings of the Israel Academy of Sciences and Humanities 6, 105–83. Jerusalem: Israel Academy of Sciences and Humanities.

Hanhart, Robert, ed. 1993. Esdrae Liber II. Göttingen: Vandenhoeck & Ruprecht.

Krašovec, Jože. 2010. The Transformation of Biblical Proper Names. London: T & T Clark.

Mor, Uri. 2013. ‘Guttural Consonants: Pre-Masoretic’. In Encyclopedia of Hebrew Language and Linguistics, edited by Geoffrey Khan at al., 2: 161–65. Leiden: Brill.

Murtonen, A. 1981. ‘Methodological Preliminaries to a Study of Greek (and Latin) Transcriptions of Hebrew’. Abr-Nahrain 20: 60–73.

Rahlfs, A. 2012. Offizielles Verzeichnis der Rahlfs-Sigeln: Herausgegeben vom Septuaginta-Unternehmen der Akademie der Wissen-schaften zu Göttingen. Göttingen: Lagarde-Haus.

Steiner, Richard C. 2005. ‘On the Dating of Hebrew Sound Changes (*ḫ > Ḥ and *ġ > ʿ) and Greek Translations (2 Esdras and Judith)’. Journal of Biblical Literature 124 (2): 229–67.

Yuditsky, Alexey (Eliyahu). 2013. ‘Transcription into Greek and Latin Script: Pre-Masoretic Period’. In Encyclopedia of Hebrew Language and Linguistics, edited by Geoffrey Khan, et al., 3: 803–22. Leiden: Brill.


1 This, as well as other phonetic transcriptions, represents the Tiberian pronunciation tradition, which does not necessarily correspond in all details to the pronunciation tradition reflected by the Greek transcriptions.

2 The Greek manuscript sigla and notation used in this article are taken from Hanhart (1993) with minor modifications. Bibliographic information for all manuscripts is available in Rahlfs (2012). A, B, and S are the majuscule codices commonly known, respectively, as Alexandrinus, Vaticanus, and Sinaiticus. V is a tenth-century majuscule codex. 122 is a fifteenth-century minuscule based on the exemplar B. 55 is a tenth-century minuscule with a text similar to that in B. The a-group (71–74–106–107– [44–125–610]–120–121–130–134–236–314– 370–762) and b-group (46– [52]–64–98– [379]–243–248–381–728– 731– [68]) are comprised of miniscules from the tenth–sixteenth centuries that probably derive from two different textual recensions made sometime in the fifth–ninth centuries. Sigla connected by n-dash, – , indicate manuscripts whose texts have a likely genetic affiliation. Sigla in square brackets, [], indicate manuscripts whose scribe [s] likely used the preceding manuscript as their exemplar. G indicates my best text for the Old Greek and Gᴸ my best text for the Lucianic recension (fourth c.?), usually witnessed by the minuscules 19–108–93, a long correction to 728 (labelled 728ᴵ), sometimes 121, less often 44–125, and at times possibly also 248.


(PhD, University of Cambridge, 2019) is lecturer in Hebrew and Old Testament at the Ethiopian Graduate School of Theology. He completed his PhD in Hebrew phonology and Greek transcriptions under the supervision of Prof. Geoffrey Khan. He is an ordained Anglican minister and a member of SIL.


Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search