The Greek prototypes of the city names Sidon and Tyre: evidence for phonemically distinct initials in Proto-Semitic or for the history of Hebrew vocalism?

• Author: Woodhouse, Robert

1. The once virtually standard affirmative answer to the first part of the question posed above was based on a belief that the initial of Gk. [TEXT NOT REPRODUCIBLE IN ASCII] "Tyre" represented a direct reflex of Proto-Semitic (PS) *t while the initial of Gk. [TEXT NOT REPRODUCIBLE IN ASCII] "Sidon" was different because it reflected PS *s, cf., e.g., Koehler/Baumgartner (1958 and 1983 s.v. sor II), (1) Friedrich/Rollig (1970: 9), Hadas-Lebel (1981: 43f.), Segert (1993: 89). (2) This has been replaced in Maria Amadasi Guzzo's third edition of Friedrich/Rollig (1999: 11f. n. 4) with what is probably now the generally accepted negative answer based on Steiner (1982: 66f.). An important feature of the argument, first noted by Wild (Steiner 1982: n. 117), is that Ugaritic does not distinguish between the initials of the two names any more than Phoenician or Hebrew do, a fact, incidentally, that appears to have been known to Segert well before 1993 (cf. Segert 1984: 198f., s.vv. sdynm et srm). (3)

2. Unfortunately, Steiner's (and the new Friedrich/Rollig/Amadasi Guzzo) account based on the idea of differing Greek attempts to represent an affricate absent from the Greek of the Classical period (cf. Steiner 1982: 62f.) is in danger of anachronism since the date of adoption of both names is likely to have preceded the period of Classical Greek (fifth to fourth centuries B.C.E.).

   Steiner (1982: 66) himself points out that the vocalism of [TEXT NOT REPRODUCIBLE IN ASCII] suggests that this form of the name is "considerably older" than its first attestation in Herodotus (fifth century B.C.E.) and cites the more contemporary Greek form [TEXT NOT REPRODUCIBLE IN ASCII] (Philistus, fifth to fourth centuries B.C.E.). Steiner's position is supported by the prominence of the name in the form [TEXT NOT REPRODUCIBLE IN ASCII] in Greek mythology as the seat of a king Agenor, father of the Cadmus who traditionally gave the Greeks the alphabet.

   "Sidon" and the adjective "Sidonian" are attested in Homer's Odyssey and Iliad, respectively, works which certainly precede Classical Greek linguistically, cf. their frequent metrical reliance on the digamma (=[TEXT NOT REPRODUCIBLE IN ASCII]), which is not recorded in the manuscripts and was generally absent from the literary dialects of Classical Greek. These epic poems are in fact traditionally thought by many to have been composed c. 800 and c. 900 B.C.E., respectively. Both are also thought to "belong to a tradition reaching back into Mycenaean times (c. 1400-1100 B.C.E.)" (Pollard 1975: 402), though it must be said that neither Tyre nor Sidon appears to figure in the Mycenaean records.

   If we regard these facts as a warrant to search for a date of adoption of "Sidon" into the Greek tradition somewhere in the period of the eleventh to the ninth centuries B.C.E., (4) then according to Bartonek (1961: 50-67, 148-50, 185-89) this takes us broadly to a time when all the Greek dialects passed through a stage of possessing at least one affricate of the *ts type deriving from pre-Greek *tw, (5) and some other sources, and developing word-initially into *s- > [sigma]- (Rix 1976: 90, 93). (6) Thanks to its derivation from *tw, this affricate had the further property of entering into a morphophonological alternation with *tu > [TEXT NOT REPRODUCIBLE IN ASCII], cf., e.g., Homeric Greek (nom.) [TEXT NOT REPRODUCIBLE IN ASCII] "thou" beside (acc.) [TEXT NOT REPRODUCIBLE IN ASCII] (*twe) "thee."

   Now if, as seems likely, [TEXT NOT REPRODUCIBLE IN ASCII] "Tyre" was adopted during the same period, we have an obvious explanation for the different initials of the two city names in Greek, viz. that the Semitic affricate was adapted in Greek ears to the above-mentioned morphophonological alternation such that Semitic *[TEXT NOT REPRODUCIBLE IN ASCII] of "Sidon" was heard as Greek *[TEXT NOT REPRODUCIBLE IN ASCII] > [sigma]i-, in [TEXT NOT REPRODUCIBLE IN ASCII], while Semitic *[TEXT NOT REPRODUCIBLE IN ASCII] (or *[TEXT NOT REPRODUCIBLE IN ASCII]; see 5 below) of "Tyre" was heard as Greek *[TEXT NOT REPRODUCIBLE IN ASCII] > [TEXT NOT REPRODUCIBLE IN ASCII]- in [TEXT NOT REPRODUCIBLE IN ASCII].

3. The next question is whether the date of adoption of [TEXT NOT REPRODUCIBLE IN ASCII] can reasonably be shown to have fallen within this period.

   The critical phonetic feature to be used in answering this question, as Steiner has indicated (2 above), is the Northwest Semitic change in the vowel of the first syllable, as evidenced in the fifth/fourth-century Greek form [TEXT NOT REPRODUCIBLE IN ASCII] and in the Massoretic pointing of "Tyre," viz. sor/s[o.sup.w]r.

   This change involves both lowering and lengthening. The shortness of the first vowel of [TEXT NOT REPRODUCIBLE IN ASCII] is guaranteed by the short vowel of the final syllable following the acute accent of the first. It agrees with (1) the short first vowel of Akkadian sur-ru "Tyre," (2) the short vowel of Arabic zirr "flint," which corresponds to Steiner's (1982: 67 n. 117) preferred etymology of the name, and (3) the regular development into the shape represented by the Massoretic forms, as seems to be generally acknowledged (cf., e.g., Bergstrasser 1918: 117; Bauer/Leander 1922: 196, cf. also 193; Goetze 1939: 433; Birkeland 1940: 49; Harris 1941: 145; Rabin 1970: 311; Blau 1976: 35; Kutscher 1982: 25; Meyer 1982: 106; Malone 1990: 470 n. 30; Jouon/Muraoka 1993: 42f.; Friedrich/Rollig 1999: 37, 44). (7)

   The length of the first vowel of [TEXT NOT REPRODUCIBLE IN ASCII] is guaranteed by the spelling with [omega] (omega) since there is little likelihood that [omega] was preferred to o (omicron) on a point of vowel quality (Allen 1987: 63, 75). The idea that this length reflects erstwhile length in the more developed Canaanite, i.e., Hebrew, form of the name sor is supported by the view of Goetze (1939: 433) that an earlier long vowel was shortened in analogous unbound forms such as 'am "people," leb "heart" and kol "entire" by analogy with the short vowel in the corresponding suffixbase forms with retained PS vocalism 'amm-, libb- and kull-. Thus "Tyre," being a proper noun, never bore a pronominal suffix (Andersen/Forbes 1986: 158) and thus never had an opportunity for analogical shortening. Although this view conflicts with Blau's (1976: 31) belief that lengthening was prevented in syllables closed by geminates, Philippi's Law provides evidence that geminates did not always behave as if they caused closure of a syllable (see 8 below). Indeed in Blau's example, *qallu > qal "light, swift," the short vowel can likewise be due to analogy with an unstressed stem, viz. that of the feminine *qallatu > [TEXT NOT REPRODUCIBLE IN ASCII] [.sup.h]. (8)

   It is questionable whether evidence for a long first syllable in the Semitic prototype of [TEXT NOT REPRODUCIBLE IN ASCII] can be seen in the proportion of plene writings of sor/s[o.sup.w]r in the Massoretic text. Andersen/Forbes (1986: 95) indicate that vowel letters were "almost never" used to write short vowels (other than u), but their treatment of short [TEXT NOT REPRODUCIBLE IN ASCII] at this point (98-100) deals only with short qames ([TEXT NOT REPRODUCIBLE IN ASCII]) embellished with waw. Short holem in verbs like [TEXT NOT REPRODUCIBLE IN ASCII] is eventually given theoretical recognition (126f.) and the proportion of plene to defective writings for verbs of this class is given as 32 to 134 (199), i.e., about one in four. The proportion is suitably reversed for the figures pertaining to the similarly spelled nouns in the absolute state with their expected long vowel (cf. Goetze 1939: 433; Blau 1976: 31), viz.: 2307 plene against 591 defective (198), with once again the ratio of "expected" writings to "non-expected" being close to four to one. (9)

   Now, a count of sor/s[o.sup.w]r spellings in the Stuttgart Bible (Elliger/Rudolph 1987), omitting variant readings at Hos 9.13 and Zech 9.2, yields thirty-one defective (10) vs. nine plene, (11) i.e., just under a quarter of all uncontroversial occurrences of the name in the Old Testament. (12) This relatively low frequency of plene spellings of a vowel expected on other grounds to be long is no doubt to be explained by the conservatism that frequently attends the spelling of proper nouns, cf. the persistence throughout the Massoretic text of h to write the final -o of par'[o.sup.h] "Pharaoh," slom[o.sup.h] "Solomon," etc. (Andersen/Forbes 1986: 39; Barr 1989: 165).

   "Sidon" confirms that Canaanite vocalic length can be reflected as Greek length if it is accepted that the constant plene writing of the first syllable and common plene writing of the second syllable in the Massoretic text indicate length. The length of the stem vowels of the Greek form [TEXT NOT REPRODUCIBLE IN ASCII] "Sidon," is assured by their filling the second spondee of Odyssey 15.425: [TEXT NOT REPRODUCIBLE IN ASCII] "on the one hand from Sidon." The length of the first vowel of the gentilic shows up in its occurrence in the first and second dectyls, respectively, of Iliad 6.290 and 291: [TEXT NOT REPRODUCIBLE IN ASCII] "(of the women)/the Sidonian ones, whom godlike Alexander (= Paris) himself/led from the Sidonian [territory]." The short vowel of the second syllable of the Greek gentilic may mirror the constant defective writing in Hebrew s[i.sup.y]don[i.sup.y] or may have been shortened for metrical reasons. (13)

4. Steiner's imprecision over the dating of the lowering and the lengthening of the first vowel in the prototype of [TEXT NOT REPRODUCIBLE IN ASCII] (2 above) may point either to a crux of Hebrew diachronic phonology or to a latent suspicion that the best date might cast doubt upon Steiner's own reliance on Classical Greek phonology to explain Herodotus' initial T-. With respect to Hebrew diachronic phonology the best date given, e.g., by Bergstrasser (1918: 163) and Birkeland (1940: 49), (14) is that the lowering of PS *i, *u to *e, *o in closed accented syllables and in...