Angelo MERCADO Phonology and Poetics: The Problem of Greek amp^hip^horeús and amp^horeús

The existence of Greek amp^horeús beside Homeric amp^hip^horeús 'two-handled pitcher or jar' has been explained by the phonological phenomenon known as haplology, a special type of dissimilation consisting in the loss of a whole syllable before or after a phonetically similar syllable. While rare because of the conditions that must prompt it, haplology occurs most commonly in compound constructions, as indeed in amp^hi-p^horeús/am-p^horeús, similarly *apó-poina/á-poina, he:mi-médimnon/he:-médimnon and a number of others in Greek (Schwyzer; Lejeune 1972).

However, with the decipherment of Linear B and the discovery of the Mycenaean Greek dialect, the assumption that early Greek amp^hip^horeús (as in Homer) became amp^horeús (as in Herodotus, Aristophanes+) by the process just described came into question: the Mycenaean tablets attest both the long and short forms of the word, a-pi-po-re-we and a-po-re-we. It then became unclear whether Hom. amp^hip^horeús indeed preserves the long Mycenaean form and later gives rise to amp^horeús, attested elsewhere in inscriptional and literary Greek, or whether amp^hip^horeús is already a back-formation based on amp^horeús (Morpurgo Davies 1985). The most recent discussion of the problem, by I. Hajnal (1998), is likewise non-committal. Nevertheless, although the matter remains unresolved, a more careful look at the Mycenaean and especially the Homeric data may yet shed a bit more light on the situation.

In the extant tablets, Myc. a-pi-po-re-we /amp^hip^hore:wes/ (nom. pl. m.) occurs once (KN), and the reduced form a-po-re-we /amp^hore:we/ (nom. du. m.) twice (MY and PY) (Docs.²; DMic.). There is no evidence that the dichotomy reflects dialectal distinctions within Mycenaean, on a geographic or any other basis. Moreover, since the short form is attested twice and corresponds to 1st-millennium Gk. amp^horeús, error must also be excluded. Therefore, a-po-re-we had begun to coexist with older a-pi-po-re-we already in Mycenaean times.

Now the full form amp^hip^horeús occurs nine times in Homer (Il. 2x; Od. 7x). The simple form amp^horeús, being for all intents and purposes metrically impossible in any case form, is not attested in the epics. A detailed analysis of the Homeric material shows the essentially formulaic behavior of amp^hip^horeús, with non-formulaic attestations having arisen by analogy. The Homeric data themselves, then, strongly suggest that the long form was indeed a survival, and not merely a metrically convenient back-formation.