As regular users of this website may already know, KoineGreek.com is planning to begin recording and uploading audio recordings of the Gospel of Matthew as it is attested in Codex Vaticanus in January of the new year. In preparation for this task, as I was thinking about the genealogical sections at the beginning of Matthew, I was immediately reminded of a question I heard while teaching at a Biblical Language Center Greek workshop many years ago:
If the Greek diphthong represented by αι had already shifted to a monophthong of the [ε] or [e̞] quality by the Koine period (no longer historical *[ai]), does that mean that Hebrew proper names (PNs) like Ἰεσσαί 'Jesse' would have been pronounced as [iεˈsε] (note αι as [ε]) when Koine Greek speakers read the New Testament?
I don't actually remember what answer was given at the time (or if I even heard the conclusion of the conversation), but setting out to record a new Gospel of Matthew recording in restored Koine Greek pronunciation—especially one that may eventually become the audio narration for another LUMO project film—prompted me to give this question new consideration in light of the best evidence out there.
Naturally, I turned to some text critical scholars and friends on Twitter to help me out. I was by no means disappointed. In particular, I want to thank Pete Myers who presented some very helpful evidence on Twitter, in personal communication with me by email, and in his excellent Ph.D. thesis.
After all of the discussion, I think the issue is best treated by drawing on two kinds of evidence: (i) the epigraphical/documentary evidence from Palestine during the Koine period and (ii) the transcription of Hebrew proper names (PNs) in early manuscripts of the Septuagint (LXX) and the New Testament (NT). Each of these types of evidence will be treated in turn below.
Palestinian Koine Greek in Epigraphic/Documentary Evidence
There are many spelling interchanges in both the epigraphic and documentary material from Palestine in the Koine period that demonstrate that the original phonemes represented by αι and ε had merged by the Roman period. While comprehensive evidence and analysis will be presented in my book on the pronunciation of Palestinian Koine Greek in preparation (Kantor 2021), I list selected examples below:
αι → ε
ανετειως (for ἀναιτίως) - CIIP 3689 - Iudaea/Idumaea - 2nd c. BCE
σαβαθεου (for σαββαταίου or σαββαθαίου) - CIIP 586 - Jerusalem- 1st c. BCE/1st. c. CE
ερεσασιν (for αἱρέσεσιν) - 5/6Hev15 - Iudaea - 125 CE
προγεγραμμενες (for προγεγραμμέναις) - 5/6Hev15 - Iudaea - 125 CE
παλεαν (for παλαιὰν) - 5/6Hev15 - Iudaea - 128 CE
χερειν (for χαίρειν) - XHev/Se64 - Iudaea - 129 CE
κε (for καὶ) - CIIP 1134 - Caesarea - 1st–3rd c. CE
ε → αι
αινγαδδων (for ἐνγαδδῶν) - 5/6Hev16 - Iudaea - 127 CE - (The more common spelling is with ενγαδ- in this region/these texts. The fact that we also find interchanges with ηνγαδ- further supports regarding αι as a variant on ε rather than vice versa. A fuller discussion will be forthcoming in my book.)
βαλεται (for βάλετε) - CIIP 3860 - Iudaea/Idumaea - 3rd–5th c. CE
λαζαραι (for Λάζαρε) - CIIP 3153 - Iudaea/Idumaea - 5th c. CE
αιξοδ[ον] (for ἔξοδ[ον]) - CIIP 826 - Jerusalem- 5th/6th c. CE
While there are actually many more examples, these serve to illustrate that this merger had likely already occurred by the early Roman period. Note especially that already in the Hellenistic period we find the αι → ε interchange in CIIP 3689.
At the same time, however, it is noteworthy that we also find an interchange of αι → αει in a Semitic proper name transcribed into Greek in the Judaean Desert:
ααβαει (for Ααβαι) - Mur103 - Iudaea - 135 CE
In this example, the Semitic sequence [aj] appears to be represented in Greek transcription as αει to ensure that it was pronounced as [ai] rather than the more default value in contemporary Palestinian Koine Greek as [ε] or [e̞]. This text was almost certainly penned by a Semitic-Greek bilingual. That this αι → αει interchange served to clarify and/or indicate such a pronunciation is further supported by the following interchange found in Palestinian documentary evidence:
τροπαιεικον (for τροπαϊκόν) - 5/6Hev15 - Iudaea - 125 CE
Even though the diphthong represented by Greek αι had generally simplified to the monophthong [ε] or [e̞], those instances in which we mark iota with diaeresis (i.e., ϊ) to indicate that it was pronounced separately from the preceding vowel continued to be pronounced as α [a] + ι [i] rather than αι = [ε]—this orthography, of course, has ancient roots, but it is not typically used in epigraphy. This is almost certainly what is being represented in the variant spelling τροπαιεικον = [tropa(j)iˈkon].
In other words, in Palestinian Koine Greek, pure Greek speakers pronounced αι as [ε] and αϊ as [ai]. For Hebrew/Aramaic-Greek bilinguals, we might add that they preserved the pronunciation [ai] when reading Hebrew/Aramaic PNs transcribed into Greek.
The fact that Hebrew/Aramaic speakers would also use αι for Hebrew/Aramaic [aj] > [ai] in Greek seems to be supported by the evidence from the second column (Secunda) of Origen's Hexapla. There, αι is regularly used to represent Hebrew/Aramaic αι: e.g., ελωαι/ελωαϊ for אֱלֹהַי [ʔεloːhaːaj]. Word internally, where there might be confusion between αι = [ε] and αϊ = [ai], the variant αει can be used to clearly signify αϊ: e.g., θαειρ for תָּאִיר [tʰɔːʔiːiʀ̟].
Transcribed Hebrew PNs in Early LXX and NT Manuscripts
Let me begin this section by answering a possible objection that some readers may have when they reach the end of this section. Even though I am ultimately seeking to make a claim about how early NT manuscripts were read, the evidence I rely on in this section is all based on evidence from the Septuagint (even if the same manuscript contains the New Testament). While such a criticism is valid to a degree, we also must remember that the evidence for Hebrew/Aramaic names transcribed into Greek is far more abundant in the LXX than it is in the NT. Therefore, we have better and more comprehensive data to work with. While it is possible that the same scribe/reader would have used different pronunciation conventions for the LXX and the NT, I am operating under the assumption that this is less likely. Nevertheless, such a caveat should be kept in mind while reading the rest of this section.
In my opinion, the best work on Hebrew/Aramaic names transcribed into Greek in the LXX is found in the 2019 thesis of Pete Myers, "A Grammar of Transcriptions in 2 Esdras: Text, Philology, Phonology," (likely to be a book in the not-so-distant future). Myers bridges a gap between historical phonology and textual criticism in a way that few (or none?) before him dealing with the issue of transcribed PNs have done. It has become common practice among historical Hebrew linguists, for example, when dealing with transcriptions in Greek sources, to simply say something like, 'Because a full text-critical analysis of the material lies beyond the scope of the present work, we will proceed on the basis of the readings found in edition X'. Myers, on the other hand, integrates a comprehensive text-critical analysis of the PNs in 2 Esdras with a phonological analysis of the material. The results are excellent.
In particular, for our concerns, there are several important pieces of data he presents:
Myers suggests that the original reading of the Greek transcription of Hebrew בִלְגַּ֖י [vilgaːaj] (Neh. 10.9) in the Septuagint version of 2 Esdras is *βελγαει (Myers 2019, p. 514), reflecting the same representation/interchange of Semitic [aj] → Greek αει. If we accept the view that 2 Esdras was translated in the second century CE, this would put such a transcription right at home with the picture painted by the epigraphic/documentary material cited above.
Even in Codex Vaticanus (B) (4th c.) we find the reading βερζελλάει (Myers 2019, p. 390) for Hebrew בַרְזִלַּ֗י [varzillaːaj] (Ezra 2.61). Such a reading with -αει suggests that the final part of the name was pronounced in the Greek reading tradition of Codex Vaticanus as [ai] rather than [ε]. This evidence is especially important since the better/original reading is probably βερζελλαι with just a simple αι. This would suggest that at some point between the second century and the fourth century this proper name was modified with the convention -αει at the end. If this is the case, it indicates that possibly as late as the early fourth century this tradition of pronouncing αι in Hebrew PNs as [ai] was remembered—at least among a select few. Note the image below:
In Codex Sinaiticus (S) (4th c.), the reading σαμαει for Hebrew שַׁלְמָֽי [ʃalmɔːɔj] (Neh 7.48) also seems to reflect an attempt to preserve the [ai] pronunciation. Note, however, that the best reading in light of other manuscript evidence may be σελμει (Myers 2019, p. 493). Scribal error, then, rather than intention, may actually explain the reading here. Note the image below:
In Codex Alexandrinus (A) (5th c.), the reading γαβουαει for Hebrew בִגְוַ֖י [viʁvaːaj] (Ezra 8.14) also seems to reflect an attempt to preserve the [ai] pronunciation. Note, however, that some scribal corruption has taken place: βαγουαι → γαβουαει (Myers 2019, p. 419). Here, however, the scribal error seems to be restricted to the first three letters of the name. The presence of -αει at the end, therefore, is still relevant for the question at hand.
The fact that this convention appears to be present in one original reading (in a second-century CE translation) and in three innovative readings in three distinct early manuscripts from the fourth and fifth centuries CE strongly suggests that knowledge of the proper pronunciation of transcribed Hebrew PNs survived to quite a late date. If it survived to such a late date, it is likely that it had earlier roots and was perhaps even more common at the time of the composition of the books of the New Testament, especially those composed in Palestine and/or by authors who were from Palestine.
At the same time, we must remember that the preservation of the proper pronunciation would by no means have been a universal one. Myers is careful in his discussions of historical Greek phonology in his thesis (and in his personal correspondence with me) to make a distinction between registers of Greek. It is likely that the more educated maintained at least a memory of a more conservative pronunciation (Myers 2019, pp. 94–96), whereas those less educated (especially with respect to grammatical/scribal tradition) would have been more prone to pronounce αι as it was in contemporary Greek.
Similarly, I have shown elsewhere that Greek transcriptions of Latin in school texts in the Egyptian papyri in comparison with Greek transcriptions of Latin in non-school texts demonstrate that a more conservative pronunciation of Latin was maintained particularly in the schools (Kantor 2017, pp. 139–141).
In sum, then, it would seem that Palestinian Koine Greek speakers of the Roman period generally pronounced αι in Greek as [ε], but as [ai] in those cases that parallel αϊ in modern orthography. In Semitic proper names with [aj] transcribed into Greek, the evidence seems to suggest that Semitic-Greek bilinguals would have written αι but pronounced [ai]. The evidence of early LXX and NT manuscripts suggests that this phenomenon was preserved to a degree—perhaps only among highly educated scribes and/or those with Hebrew/Aramaic knowledge—in the manuscript tradition all the way up into the Byzantine period. On the other hand, Greek speakers without such knowledge would be prone to pronounce αι in Semitic names just like everywhere else.
Therefore, in light of my ultimately practical purposes of trying to offer prescriptive guidance in reading the New Testament in a restored Koine Greek pronunciation, my own recommendations for reading αι in Hebrew PNs like Ἰεσσαί are the following:
To emulate scribes most knowledgeable of the tradition, pronounce αι in Hebrew PNs as [ai]
To emulate scribes less knowledgeable of the tradition, pronounce αι in Hebrew PNs as [ε]
To emulate a first-century Hebrew/Aramaic-Greek bilingual, pronounce αι in Hebrew PNs as [ai]
To emulate a first-century purely Hellenistic Greek speaker, pronounce αι in Hebrew PNs as [ε]
This is, granted, a bit of an oversimplification. It is likely that there was variation both among scribes and perhaps even among Hellenistic Greeks who were not bilinguals but lived in a bilingual context. Nevertheless, sometimes oversimplifications are helpful pedagogically in a temporary capacity. Remembering these four options can at least give a simple starting point for one merely interested in having an accurate historical pronunciation for their own personal reading.
As for me, because I believe that the Gospel traditions are inextricably tied to a Hebrew/Aramaic-Greek bilingual/trilingual setting and because I have knowledge of Hebrew/Aramaic, I opt to pronounce αι as [ai] in Hebrew PNs. Therefore, I will pronounce Ιεσσαι with [ai] in the forthcoming Matthew recording.
Kantor, Benjamin. 2017. “The Second Column (Secunda) of Origen’s Hexapla in Light of Greek Pronunciation.” diss., The University of Texas.
Kantor, Benjamin. forthcoming 2021. The Apostles’ Greek: The Historical Pronunciation of New Testament Greek (Palestinian Koine).
Myers, Peter. 2019. "A Grammar of Transcriptions in 2 Esdras: Text, Philology, Phonology." Ph.D. Thesis. University of Cambridge.