By being both an Attic and a conservative Koine system, the Alexandrian System respects the centrality of classical Athens while also acknowledging the need for broader applicability – the very concerns which drove the formation of Koine in the first place.

Introduction to the Alexandrian System

This article presents a system for the pronunciation of Ancient Greek, called the “Alexandrian Koine System,” which is designed for readers focusing primarily on works written before 150 BC. 

The Alexandrian Koine System is a system of both Attic and Koine pronunciation. Koine Greek came into existence when “the Macedonian kings, in search of a ‘civilization’ to underpin their growing military and political power,…established the study of classical Greek literature, much of it literary Attic, as a central plank of their educational system and adopted contemporary Great Attic as their own official language of state” (Horrocks 2014: 80).  The dialect of ancient Greek spread by Alexander the Great, which became the Greek of the Hellenistic kingdoms, more or less was Attic Greek (see Horrocks (2014: 82 ff.) for the differences). Moreover, in Ptolemaic Egypt, the royal court at Alexandria maintained a highly conservative pronunciation – more conservative even than Athens (Horrocks 2014: 166). An Alexandrian Koine pronunciation system, therefore, is appropriate for Classical Athens as well.

Key Features

1. An ultra-conservative and a majority variant, in which ῃ = the diphthong [eːi] ¹ and [iː], respectively.
2. ει = [iː] except pre-vocally, where it is [eː]. υι is a diphthong [yi] (approx.), as it was in pre-4th-century BC Attica, but may optionally be monophthonized into [yː].
3. The ι element of the long diphthongs ᾳ and ῳ are pronounced.
4. Aspirated φ, θ, and χ, rather than the later fricative sounds of Modern Greek.
5. Velarization and labialization of ν at the end of words when followed by initial velars and labials (e.g. ἐμ πόλει for ἐν πόλει)
6. Phonemic vowel length is preserved, and pitch accent is encouraged.
7. Follows the sounds of Reconstructed Attic (as presented by Allen and Threatte) as much as possible.
8. No need to worry about “spurious” v.s. “non-spurious” diphthongs, a distinction irrelevant after 350 B.C (Threatte 180: 299).

The preference for an Alexandrian pronunciation rather than an early 4th century BC Athenian system is partially an acknowledgement of the role of the Alexandrians in textual transmission. According to Marrou, “it was only in the generation following Aristotle and Alexander the Great that education assumed its classical and definitive form; thereafter it underwent no substantial change” (1956: 137). It was in this period, in Alexandria especially, that canonization (e.g. the “Ten Attic Orators”) took place. And thereafter, it was in Alexandria that old books – including official copies “borrowed” from Athens and elsewhere- were copied and even stolen for the library (Barnes 2004: 65). According to Graham Shipley, “the library [of Alexandria] became the main source of reliable texts of authors such as Homer, who were increasingly acknowledged to be classics” (2000: 240).

But the preference for an Alexandrian pronunciation is also a recognition of the need for a system to be broadly applicable. Attic pronunciation underwent many changes during the classical period from 500-350 B.C. (especially if one follows Theodorsson’s (1974) reconstructions), such that one using Attic pronunciation(s) accurate to the time of authorship would require a slightly different pronunciation for Aeschylus, Xenophon, and Aristotle. This problem is only compounded if one also wishes to read non-Attic texts, such as Homer, Sappho, and Herodotus, both because their pronunciations differed substantially from 5th century B.C. Attic and, more importantly, because our MSS and texts likely tell us more about how the Alexandrians would have pronounced their words than the authors themselves (see Horrocks (2014: 62-63 ff.), “we may be quite sure that the markings of initial aspiration [in Herodotus] is purely editorial, not only because eastern Ionic was psilotic but because there is no trace of the aspirate in compounds”). Only the most narrowly focused of classics scholars could plausibly use a historical pronunciation accurate to the time of authorship of his focus; anyone who reads more broadly needs an alternative solution.

I propose the Alexandrian Koine System as the alternative. First, the Alexandrian System is an Attic system – especially in its conservative variety. Secondly, the Alexandrian system is a broadly applicable historical pronunciation system, which achieves its breath of applicability by reconstructing the pronunciation of the readers of classical Greek texts who were alive when the texts first became the cornerstone of western literature. Effectively, the preference for an Attic pronunciation accurate to Aristotle and Demosthenes rather than Sophocles or Xenophon allows one to use the same historically valid pronunciation system to read all texts from Homer to Polybius – including the Septuagint. By being both an Attic and a conservative Koine system, the Alexandrian System respects the centrality of classical Athens while also acknowledging the need for broader applicability – the very concerns which drove the formation of Koine in the first place.

Effectively, the preference for an Attic pronunciation accurate to Aristotle and Demosthenes rather than Sophocles or Xenophon allows one to use the same historically valid pronunciation system to read all texts from Homer to Polybius – including the Septuagint.

This system won’t be appropriate for everybody. For example, those focused on the New Testament, the Apostolic Fathers, or Roman period texts, should consider my Caesarian System. Further, teachers might reasonably prefer to continue with an Erasmian system for the sake of instructional ease. But for those whose primary interests lie with Greek texts from Homer to about 150 BC, and who have a special interest in Attic and Hellenistic literature, I submit that my Alexandrian System might be most appropriate.

The System¹

Grapheme	IPA Representation (w/ links to audio)	“As in…” (Approx.)	Notes
α	[a], [aː]	father
β	[b]	boy
γ	[g]	gold	Never soft, as in general.
δ	[d]	dente (Italian)	Pronounced as a laminal denti-alveolar (“dental”)
ε	[e]	pet	Avoid pronouncing like the diphthong [ei].
ζ	[zː]	wisdom (lengthened)	Geminated consonant
η	[eː]	pet (lengthened)	Long vowel; pronounced the same as ει when it comes before a vowel. ε, but long. Avoid pronouncing like the diphthong [eːi].
θ	[th]	top (But pronounced as a dental rather than an alveolar)	Dental; aspirated, but otherwise the same as τ
ι	[i], [iː]	[i] = bit; [iː] = bead	ῑ is pronounced the same as ῃ and non-prevocalic ει.
κ	[k]	tack	Unaspirated, but otherwise the same as χ
λ	[l]	like
μ	[m]	mead
ν	[n]	net
ξ	[ks]	box	Resist the urge to pronounce initial ξ as /z/ in words such as ξένος
ο	[o]	forense (Italian)	English speakers often erroneously pronounce this as [a].
π	[p]	top	Unaspirated, but otherwise the same as φ
ρ	[r] , [r̥] (when aspirated)	[r] as rojo (Spanish)	r̥ does not occur in the phonology of western European languages, but “it should be emphasized that the voiceless pronunciation of ρ in certain environments is a purely allophonic matter, and no confusion can therefore be caused if ρ is always pronounced with its voiced value, as e.g. in modern Greek” (Allen 1987: 43). If the speaker cannot produce [r], he might reasonably substitute [ɾ] (=caro (Spanish), better (American English).
σ	[s], [z] (before β, γ, δ, μ)	[s] = sing; [z] = cosmos	English speakers often pronounce s as [z] in intervocalic and final positions (e.g. Basil), which should be avoided for σ (Allen 1987: 46).
τ	[t]	tale (Italian)	Pronounced as a laminal denti-alveolar (“dental”)
υ	[y], [yː]	tu (French)	Pronounced like an English ῑ, but with lip rounded like an English u.
φ	[ph]	pot	Aspirated, but otherwise the same as π
χ	[kh]	cat	Aspirated, but otherwise the same as κ
ψ	[ps]	lapse	ψ = π + σ
ω	[oː]	Cambodia (General American English)	Avoid pronouncing as a diphthong [öʊ].
–
αι	[ai] (diphthong)	high
αυ	[au] (diphthong)	how
ᾱυ	[aːu] (diphthong)	how	“[D]iphthongs in Greek cannot strictly be distinguished as ‘short’ or ‘long’; for accentual purposes they all have the same value of 2 morae (time-units), as for a long simple vowel” (Allen 85). As a consequence, while the so-called “long diphthongs” will have the same time-value as regular long vowels and diphthongs, the relative proportion of the elements will shift to roughly a 2:1 ratio. “May be pronounced as αυ…with little ambiguity” (Allen 1987: 88).
ευ	[eu] (diphthong)	europeo (Spanish)	No English analogue. This should not be pronounced like the English neuter.
ηυ	[eːu] (diphthong)	europeo (Spanish)	“[D]iphthongs in Greek cannot strictly be distinguished as ‘short’ or ‘long’; for accentual purposes they all have the same value of 2 morae (time-units), as for a long simple vowel” (Allen 1987: 85). As a consequence, while the so-called “long diphthongs” will have the same time-value as regular long vowels and diphthongs, the relative proportion of the elements will shift to roughly a 2:1 ratio. “May be pronounced as…ευ with little ambiguity” (Allen 1987: 88). No English analogue. This should not be pronounced like the English neuter.
οι	[oi] (diphthong)	coin
υι	[yi] (diphthong; approx.) / [yː](Atticizing var.)	huit (French)	There isn’t sufficient evidence to prove that υι was pronounced as anything other than a diphthong by the majority in Ptolemaic Egypt (Teodorsson 1977: 227). In Attica, however, υι had completely monophthongized into [yː] by the 4th century B.C. (Threatte 1980: 338). This pronunciation was, according to Herodian, seen as especially Attic: “υἱὸς καὶ φυιός, ταῦτα δὲ οἱ Ἀττικοὶ ἄνευ τοῦ ι γράφουσιν” (ibid., 376). Modern speakers with a strong penchant for Atticism may nonetheless monophthonize, for “small groups of speakers [in Ptolemaic Egypt] may have had a monophthongal pronunciation, above all perhaps in the 3rd century (Teodorsson 1977: 227).
ωυ	[οːu]	owe	“[D]iphthongs in Greek cannot strictly be distinguished as ‘short’ or ‘long’; for accentual purposes they all have the same value of 2 morae (time-units), as for a long simple vowel” (Allen 1987: 85). As a consequence, while the so-called “long diphthongs” will have the same time-value as regular long vowels and diphthongs, the relative proportion of the elements will shift to roughly a 2:1 ratio.
–
ᾳ	[αːi]	high	“[D]iphthongs in Greek cannot strictly be distinguished as ‘short’ or ‘long’; for accentual purposes they all have the same value of 2 morae (time-units), as for a long simple vowel” (Allen 1987: 85). As a consequence, while the so-called “long diphthongs” will have the same time-value as regular long vowels and diphthongs, the relative proportion of the elements will shift to roughly a 2:1 ratio.
ῳ	[oːi]	coin	“[D]iphthongs in Greek cannot strictly be distinguished as ‘short’ or ‘long’; for accentual purposes they all have the same value of 2 morae (time-units), as for a long simple vowel” (Allen 1987: 85). As a consequence, while the so-called “long diphthongs” will have the same time-value as regular long vowels and diphthongs, the relative proportion of the elements will shift to roughly a 2:1 ratio.
–
ει	[iː] (before consonants or word-boundary), [eː] (before vowels)	[iː] = beat; [eː] = pet (lengthened)	ει before consonants or word-boundaries is pronounced the same as ῑ or ῃ. ει before vowels is pronounced the same as η. ει is a digraph, not the diphthong [eːi]. According to Allen (1987: 70), “The development of ει to [iː] is revealed by occasional confusion between ει and ι from the late 4 c. B.C., becoming common in the 3 c.”
ῃ	[iː] / [eːi] (conservative var.)	[iː] = beat; [eːi] = hay	ῃ is pronounced the same as ῑ and non-prevocalic ει. Modern speakers with a strong penchant for conservative Attic may also pronounce ῃ as [eːi], for which there is some evidence, and which corresponds with the pre-4th century B.C. value of Attic ῃ (Teodorsson 1977: 222). Teodorsson (1974: 286-99), whose analysis probably gives too much weight to progressive variants, holds that ῃ = [i] by 350 B.C in Attica. Allen (1987:86) more conservatively argues that it became [iː] around the beginning of the 3rd century B.C.
ου	[uː]	ooze
–
γγ	[ŋg]	angle
γκ	[ŋk]	anchor
γξ	[ŋks]	sphinx
νχ, γχ	[ŋkh]	anchor (with aspiration)

Other features

1. νγ, νκ, νξ, νχ = γγ, γκ, γξ, γχ should apply across word-boundaries (e.g. τὸγ γραμματέα for τὸν γραμματέα) (see Allen (1987: 34) & Threatte (1980: 616-636).)
2. νβ, νπ, νφ = μβ, μπ, μφ should apply across word-boundaries (e.g. ἐμ πόλει for ἐν πόλει) (see Allen (1987:33-4) & Threatte (1980:616-636).)
3. Pitch accentuation should be attempted to the best of the speaker’s ability, but can be emphasized (less desirably) with stress. If the speaker cannot produce an accurate pitch accent at all, a stress accent can be used, provided that proper vowel lengths are maintained. (English speakers erroneously tend to lengthen stressed syllables.)
4. Vowel quantity should be preserved as diligently as possible.
5. The rough breathing mark (῾) = [h] was probably only gently pronounced in pre-450 B.C. Attic, but confounding variables (e.g. adoption of the Ionic writing system) complication the evaluation of the progression of its loss (Threatte 1980: 494). Nonetheless, it was “progressively lost during the period of the Koine” (Horrocks 2014: 171). I recommend pronouncing it at the beginning of clauses, but either pronouncing it more lightly or omitting it altogether mid-clause.

Bibliography

Allen, W.S. (1987) Vox Graeca, Cambridge.
Barnes, Robert (2004) “Cloistered Bookworms in the Chicken-Coop of the Muses: The Ancient Library of Alexandria,” in R. Macleod (ed.) (2004) The Library of Alexandria: Center of Learning in the Ancient World, London.
Gignac, Francis T. (1976) A Grammar of the Greek Papyri of the Roman and Byzantine Periods: Vol. 1 Phonology, Milan.
Horrocks, Geoffrey (2014) Greek: A History of the Language and its Speakers, Oxford.
Marrou, H.I. (1956) A History of Education in Antiquity, New York.
Ruijgh, C.J. (1978) Review of S.-T. Teodorsson (1974) in Mnemosyne 31, 79-89.
Shipley, Graham (2000) The Greek World: After Alexander, London.
Teodorsson, S.-T. (1974) The Phonemic System of the Attic Dialect 400-340 BC, Göteborg.
Teodorsson, S.-T. (1977) The Phonology of Ptolemaic Koine, Göteborg.
Threatte, Leslie (1980) The Grammar of Attic Inscriptions I: Phonology, Berlin.

Notes

1. Brackets ([ ]) and slashes (/ /) contain information expressed using the International Phonetic Alphabet. Some audio samples – but not most – are of words containing the represented sounds, rather than the sounds themselves. Audio samples only capture appropriate quality, not quantity or aspiration.