Whatever their linguistic and geographical affiliations, the vast majority of post-tenth-century bce inscriptions throughout the Levant show similarities to Phoenician that can only be explained by direct descent or at least strong influence from that script. Scholars accordingly posit that, along with Phoenician language, Phoenician script had a certain caché. It was prestigious and influential from Anatolia (Yakubovich 2015) to the southern Levant (Lemaire 2013) and beyond (e.g., figure 17.1). Each of the “national” scripts—including Ammonite, Aramaic, Hebrew, and Moabite—eventually (at different dates in the ninth and eighth centuries) developed particular features that warrant labeling each separately (Naveh 1987: 53–112; Vanderhooft 2014), but their parentage is all primarily if not entirely Phoenician. As it would turn out, this was to be only half—if that!—of the Phoenician alphabetic legacy. (For the letter shapes, see table 17.1.)
Figure 17.1 Ivory box inscription (KAI 29). Ur, ca. seventh century bce. British Museum 120528.
Source: Drawing by M. Richey.
Into the West
Already Herodotus understood the Greeks to have acquired their alphabet from the Phoenicians (Hist. 5.58.1–2). Minus a few dissenting voices (cf. inter alia Segert 1963; Knauf 1987), scholars generally agree with this genealogy. Aside from circumstantial historical factors—that is, the apparent dominance of Phoenicians in early Iron Age Mediterranean seafaring—the fact that certain Greek letter names (ἰῶτα, ρῶ) show the so-called Canaanite Shift *ā > o(:) is decisive for a Phoenician source (McCarter 1975: 100n87; Naveh 1987: 183; Amadasi Guzzo 1991: 296; Ruijgh 1997: 545). The fact that certain early Greek epigraphs refer to letters as ϕοινικηια, “Phoenician (letters)” (IG XII.2, 967 [Aeolia]; E 180, 152–53 [Crete]; Jeffery 1982: 819n2) may support this derivation, but of course there is some debate as to the topographical referent of this Greek term (for the Phoenicians in Greek sources, see chapter 44, this volume). There is even less agreement regarding when, where, how, and why the westward transmission of alphabetic writing began.
From Phoenician to Greek
As noted earlier, the Phoenician alphabet consisted of twenty-two graphemes generally hypothesized to represent twenty-two and only twenty-two consonantal phonemes. Hebrew and Aramaic scribes began early on to employ certain graphemes—{h}, {w}, and {y}—to represent vocalic elements, mostly in word-final position. Among Phoenician scribes, on the other hand, this practice was very rare, if operative at all, throughout the Iron Age; it did not really catch on until the fifth century bce (Amadasi Guzzo 1999: 57–58; cf. Lemaire 2008). The earliest witnesses to the Greek alphabet, on the other hand, already use six graphemes—{A}, {E}, {I}, {O}, and {Y}—to represent the vowels ă/ā, ĕ(/ē), ĭ/ī, ŏ/ō, and ŭ/ū, respectively. Greek dialects that retained the phoneme h used the grapheme {H} (< Phoenician {ḥ}) for this sound, but some of these dialects and others that had lost h employed {H} for the vowel ē; this usage is that which eventually became standard in Greek (e.g., Woodard 2010: 36). A seventh grapheme, {Ω}, was introduced to differentiate ō from ŏ in the alphabets of Ionia, Knidos, Paros, and Melos. Since it appears always at the end of the alphabetic sequence, it was probably last to be invented, probably by variant imitation of Phoenician {ˁ}, before 600 bce (Jeffery 1982: 824, 830, 1990: 38). Each of these letters has a fairly clear Phoenician source, each of which represented in Phoenician laryngeal, pharyngeal, and glide phonemes lost or never extant in most or all Greek dialects. There is, though, much discussion as to why, within this listed group, particular Phoenician graphemes were chosen to represent particular Greek vocalic phonemes. For {y} > {I} for ĭ/ī and {w} > {Y} for ŭ/ū, the rationale is transparent. The Greek graphemes represent vocalic phonemes homorganic with the Phoenician consonantal phonemes represented (Brixhe 1991: 344–54). Scholars often attribute the use of the grapheme {ˀ} as Greek {A} for ă/ā to Greek speakers hearing primarily the first vowel—rather than the glottal consonant—of the Phoenician letter name ˀalp- “ox” and thereby assuming the Phoenician letter to itself denote a. But the use of {h} and {ḥ} for ĕ and/or ē and especially of {ˁ} for o/ō are not so amenable to the same principles (Brixhe 1991: 317–23; Krebernik 2007: 130–33).
Turning to the consonants, we see most Greek adaptations of Phoenician letters are straightforward representations of consonantal phonemes equivalent or nearly so in the two languages. The simple plosives and the sonorants fall into this group. The letter {Ξ} and the three letters preceding {Ω} at the end of most Greek alphabets—{Φ}, {Χ}, and {Ψ}—had different values in different regions. “Green” alphabets (Crete, Thera, Melos, Anaphe) lacked these letters entirely. “Red” alphabets (Euboea, Boeotia, Arcadia, Laconia) used {Φ} for ph, {Χ} for ks ({Ξ} is absent in these alphabets), and {Ψ} for kh (ps is expressed as {ΦΣ}). “Blue” alphabets (the most broadly attested since it was used for Attic-Ionic dialects) all use {Φ} for ph and {X} for kh but can be split into subgroups based on the presence or absence of {Ξ} and {Ψ}. “Light blue” alphabets (Attica, some Ionic Aegean) lack both letters and spell ks and ps as {ΚΣ} and {ΦΣ}, respectively. “Dark blue” alphabets (Ionic Dodekapolis, some Ionic Aegean, Corinth, Argos) use {Ξ} for ks and {Ψ} for ps—that is, essentially as does Classical Attic Greek (system of Kirchhoff 1887: 180; see Brixhe 1991: 340–344; Wachter 1989: 34–49; Woodard 2010: 38–39).
The graphemes representing Phoenician sibilants—{z}, {s}, {ṣ}, {š}—present a special problem (Wachter 1989: 49–61; Krebernik 2007: 128–30). Most scholars agree that Greek {Ζ}, {Ξ}, {Ϻ} (archaic “san”; perhaps ts), and {Σ} are based on those four Phoenician graphemes, respectively, but the phonemes represented do not match precisely, nor do the names in each tradition. The discussion about the precise Phoenician phonetic values of the letter is itself complex (Amadasi Guzzo 1999: 24–26; for Greek phonetic values, see Wachter 1989: 49–61; Brixhe 1991: 323–27; and Woodard 2010: 32, table 3.1h) The Phoenician letter names, in turn, are essentially those hypothesized by Hamilton 2006 for proto-Semitic and daughter traditions, including the likely Phoenician phonetic developments. The following describes the differences:
Scholars have often suggested that the sibilant names were thoroughly jumbled, so that ζῆτα is from ṣadē-, ξεῖ is from šann-, σάν is from zēn-, and σίγμα is from samk- (Jeffery 1990: 25–28; Brixhe 1991: 332–33; Powell 1991: 46–48). The correspondences, though, are quite irregular, and most have difficulty explaining why this jumbling should have occurred. A simpler hypothesis has been put forward by Woodard (1997: 147–88, 2010: 31–33; Willi 2008: 415–17) and seems plausible: the letters {ṣ} and {š} were both adopted to represent the phoneme s, but one of these (in most archaic local scripts, the reflex of {ṣ}) was gradually marginalized or dropped until only the reflex of {š}, with its letter name šann-, remained. The letter name ṣadē- was thus eliminated. The other supposed jumblings turn out to be phantoms: the name ζῆτα is analogical with preceding ῆτα; the name ξεῖ follows an innovative pattern known from other letter names, e.g., ϕεῖ, χεῖ, ψεῖ; and the name σίγμα is quite a late employment of a noun from onomatopoetic σίζω “to hiss.” Therefore, the older Semitic letter names were displaced by these innovations.
Most of the Phoenician letter names were, however, faithfully transmitted to the Greek sphere. The earliest attestation of most letter names in Greek appears already in the late fifth century bce, in the Γραμματικὴ τραγῳδία “Letter Tragedy” play of Callias (see Willi 2008: 402–404 for this and other early attestations). It was in part because these names were patently related to Semitic rather than Greek lexemes that early authors were certain of the Greek alphabet’s Semitic heritage (Euseb. Praep. evang. 10.5). By and large, the referents of these Semitic words are visible in the earliest, Sinaitic forms of the letters. Thus Greek ἄλϕα is clearly from Phoenician ˀalp- “ox” (originally Egyptian hieroglyph F1), βῆτα is bēt- (< *bayt-) “house” (O1 [/O4]), and so on. Not all of the derivations are so straightforward (Hami
lton 2006; Krebernik 2007: 148–61; Willi 2008), but enough are clear to demonstrate the connection and some basic phonological principles—for example, operation of the Canaanite Shift *ā > o(:) in the source language. Additionally, the final -a of many Greek letter names only occurs in particular phonological environments: after consonant clusters in ἄλϕα, γάμμα, δέλτα, κάππα, λάβδα, σίγμα (if archaic), and after the sequence -ēt- or -ōt-, perhaps by localized analogy with the former category and then extension (cp. βῆτα, ζῆτα, ἦτα, θῆτα, ἰῶτα; Krebernik 2007: 146–47; Willi 2008: 413–14). If the names had these final −a elements in the source language, one might expect to see this element on most if not all of the letter names. Since this universality is not achieved, it is less likely that this −a represents a Semitic accusative case ending (Tropper 2000: 358) or the Aramaic definite article (Segert 1963: 48–52; Knauf 1987: 45–48).
A troublesome question in the “how” sphere involves whether transmission was actually singular or whether multiple local adaptations eventually coalesced to produce the script(s) we know as the Greek alphabet. Many scholars (e.g., Jeffery 1982: 822) regard a set of universal adaptations, such as the uniform treatment of vowel letters and sibilants and the use of Phoenician {w} as both {Ϝ} and {Y}, as too arbitrary to have proceeded from coincidence rather than a single adaptation. (For the letter shapes, see table 17.2 at the end of the chapter.)
A Long Darkness?
The oldest currently known Greek alphabetic inscriptions are datable to approximately the second quarter of the eighth century bce (ca. 775–750; see, e.g., overviews by Jeffery 1990 [1st ed., 1961] and McCarter 1975). At the time of P. Kyle McCarter’s survey in 1975, the oldest Greek inscriptions known were the Diplyon Oinochoe and the Lacco Ameno Sherd (from Pithekoussai, modern Ischia), both ca. 750/25. Some recent additions to the early corpus include the Osteria dell’Osa flask, also from Pithekoussai (Ridgway 1996; Woodard 2010: 44 dates it as early as ca. 830 bce), one roughly contemporary short graffito from the temple of Apollo Daphnephoros in Euboia (Theurillat 2007), and inscriptions from Kommos, Crete (Csapo et al. 2000; for the intriguing Fayum plaques, see Woodard 2014). Of course, these oldest inscriptions only fix a date by which some Greeks must have encountered and exploited the Phoenician prototype. The early eighth century is thus merely a terminus ante quem for the initial adaptation. Perhaps unsurprisingly, the range of opinions as to the actual date of the Greek innovation touches both conceivable ends of the spectrum. Early on, the Classicist Rhys Carpenter aimed to shatter an “old illusion of the great antiquity of the Greek alphabet” (Carpenter 1933: 8); the transmission occurred, according to Carpenter, on Rhodes and Crete in the late eighth century bce (1933: 29). Although some more recent finds have rendered Carpenter’s very low date untenable, many scholars are still disinclined to date the transmission much before ca. 775 or, at most, at the turn of the century, with a date “ca. 800 bce” as a frequently employed shorthand (Amadasi Guzzo 1991: 307; Röllig 1995: 202; Wachter 1989: 69–76; Krebernik 2007: 123).
And yet, as usual, not everyone agrees. For those proposing an earlier adaptation, the absence of Greek inscriptions securely dating to, say, ca. 850 bce or even earlier is sometimes claimed to be a fluke of the data, attributable to the deterioration of fragile papyrus supports (Ruijgh 1997: 536) or to the technology’s restriction to a tiny cadre of adepts (Mavrojannis 2007: 317). This is not impossible; a similar evidence gap is often claimed for the Old South Arabian alphabet(s), the oldest of which are at the earliest from the eighth century bce, but whose letter forms are most easily traceable from the late Bronze Sinaitic and/or Old Canaanite material described earlier (Naveh 1987: 43–44; Sass 2005: 96–132).
The most straightforward way to determine the date of the Phoenician–Greek transmission is to compare the graphemes themselves. Most agree that the Phoenician script of ca. 850–750 bce yields the most likely prototypes for the archaic Greek scripts (cf. overview in McCarter 1975: 93–102). The lower bound is more easily established: in the late ninth and early eighth centuries bce, the vertical stem soon developed by Phoenician {d} is still only nascent or even absent, as in Greek {Δ}. Pre-750 bce Phoenician {k} generally retains its oblique trident shape—that is, two smaller arms meet a longer oblique shaft at a single point. This is the shape borrowed in Greek {K}. By ca. 750 bce, the leftmost arm of Phoenician {k} has shifted left and up, so that it touches the middle arm’s upper terminus instead of its lower one. At around the same time, the head of Phoenician {m} was simplified from a W-shape—taken over by Greek {M}—to a bisected horizontal bar (Peckham 1968: 115–75; McCarter 1975: 39–63).
The upper bound, ca. 850 bce, is more difficult to argue, primarily because there are few non-Byblian Phoenician inscriptions from the early ninth century bce or before. The most problematic grapheme for an early ninth- or tenth-century borrowing is {w}, which has a cup-headed shape in all Byblian, South Canaanite, and even Phoenician scripts of the tenth and early ninth centuries, but not in early Greek. The Greek letter {Ϝ} seems instead to be based on the ninth-century hook-headed Phoenician {w} or a prototype, as in line 2 of the Nora Stele (KAI 46).
Those who favor a high date sometimes point to the angular forms of the Greek letters or to boustrophedon writing as features most explicable by derivation from earlier Semitic prototypes (Naveh 1987: 177–78; for other high dates, see Ruijgh 1997; Bernal 1990). But angularity is not confined to the pre-ninth-century bce Semitic scripts; this feature is dependent more on medium and genre than on date. Boustrophedon writing is actually a very rare feature in early Northwest Semitic inscriptions; aside from a few very debatable Sinaitic cases, there is only one inscription—a late Bronze Age bowl sherd from Lachish—that is plausibly so written (Cross 2003: 293–96). It has also been suggested that the Nora fragment (CIS I.145) was to be read as boustrophedon, but this produced no satisfactory understanding of the text (Cross 2003: 254–64; cf. Sass 1988: 91–93).
By Land or By Sea?
Another issue that complicates the chronological question involves whether the Greeks adapted their alphabet directly from the Phoenicians or through an intermediary, who had therefore adopted it themselves at an earlier date. Most models have assumed direct Phoenician–Greek interaction and then located a site at which such interaction could have occurred. Popular proposals include: (1) Al Mina, a site on the Syrian coast at which large quantities of Greek pottery were discovered and a Greek–Phoenician trading post accordingly imagined (Jeffery 1982: 830, 1990: 11–12)—but new studies have minimized or lowered the date of Greek presence at this site (summary in Woodard 2010: 40); (2) Cyprus, home to late Bronze Age syllabic scripts that, like the later Greek alphabet, included a single grapheme for the combination [k] + [s] (Woodard 1997); (3) Euboea, with posited connections to early Homeric literary production (Powell 1991; Röllig 1995: 202); and (4) various other Aegean islands, argued from simple proximity to the Levant and/or their (“Green”) alphabets lacking the supplementary graphemes (Carpenter 1933: 23).
The Phrygian alphabet (corpus in Brixhe 1984), closely related to that of the Greeks, may yet have a decisive part to play. New radiocarbon dates place a major destruction at Phrygian Gordion ca. 830–800 bce, which may provide a terminus post quem for some graffiti (G-104, -237, and -249), if Claude Brixhe is correct in dating them “immediately after the catastrophe” (Brixhe 2007: 278; cf. Brixhe 2002: 26). If these graffiti are indeed so early, one is tempted to pay greater attention to those models that posit an overland stream of transmission through Asia Minor—perhaps one among many—for the early alphabet (Bordreuil 2005: 23; Brixhe 2007: 282).
Once it passed into the Greek realm, the alphabet followed the same routes as did general colonization and trade (overview in Jeffery [1982: 823–30], incorporating the analysis of Jeffery [1990 and 1st ed., 1961]). This proliferation produced in no time a great number of differentiated epichoric (“regional”) alphabets. While some have viewed the rising distinctions as sim
ply the product of branching typological development, others have argued that script differentiation served political centralization (Luraghi 2010). By 403/2 bce, the Attic script (in turn based on the Ionic one) had achieved declared hegemony over neighboring systems, and though these latter continued for some time, they were increasingly marginalized.
Beyond Tyre and Athens
Certain and Uncertain Children
As the Phoenician-Punic alphabet developed and spread throughout the Mediterranean (Peckham 1968; Amadasi Guzzo 2011), it engendered numerous more and less productive offshoots. The Phoenician–Greek transmission may receive the most attention, but it is far from the end of the story. By the seventh century bce, the western Greek Euboean alphabet was adapted to create the Etruscan alphabet, which was in turn remodeled for Latin and other Italic scripts (Wallace 2016, with previous bibliography). Meanwhile, in Anatolia, the fifth–fourth century bce Lycian and Lydian alphabets are clearly derived from a “Red” Greek script, while the parentage of the neighboring Carian script is less well understood (Adiego 2007: 230–33).
Deeper in the realm of the uncertain lie the Iberian and Berber alphabets. In southern Spain, the local population employed as early as the seventh century bce a group of “Iberian” or “(Paleo-)Hispanic” scripts to write both Celtic Indo-European and non-Indo-European languages (corpus in Untermann 1975–1997). Because both Phoenicians and Greeks operated in Spain as early as the ninth and eighth centuries bce, a source in either group’s script or a combination of both is theoretically possible. Over the combinatory theory (Untermann 1975–1997: I.70–71), de Hoz has argued repeatedly and persuasively for a Phoenician source, adapted first in Tartessos (southwest Iberia) with minimal graphic mutation and transparent reworking of the representational system (de Hoz 2010: 485–525).
The Oxford Handbook of the Phoenician and Punic Mediterranean Page 36