Stephen Trazskoma, California State University, Los Angeles
In the ancient grammarians we find descriptions of the outcomes of adding an enclitic to an orthotonic host. Some of these are distilled in today’s grammars and textbooks, however, an analysis of how enclitic accentuation functions—a linguistic explanation of the causes of accentual expressions—has been frustratingly elusive to formulate. This paper proposes a solution based on my own analysis of Greek recessive accentuation (Trzaskoma 2023) and that of Steriade (1988).
Both are foot-based metrical analyses of Greek prosody, although Steriade’s is syllabic while I posit a syllabic-moraic basis. Crucially, both rely upon extrametricality of some final syllables. Steriade (1988: 292) observes enclitics do not induce added tones on heavy final syllables of orthotones. In what follows, footed syllables are enclosed in ( ) and extrametrical ones in < >. Thus, we find (δαί)(μων) τις—not *(δαί)(μών) τις—and (φοῖ)(νιξ) τις—not *(φοῖ)(νίξ) τις. This is striking and seems significant. Although I did not address enclitic accentuation in my previous paper, I described a system in which only some heavy syllables are extrametrical, and Steriade’s correlation disappears: the host word in (φοῖ)(νιξ) τις has a footed ultima but not in (δαί)⟨μων⟩ τις.
A solution comes through incorporating Allen’s (1973) contonation as a complex of high pitch plus accentual fall across the rest of the syllable (circumflex accentuation) or across the whole of the following syllable (acute accentuation). The incorporation of multiple suprasegmental criteria allows us to formulate a rule that even if a final syllable is unfooted, if it hosts the lowering of the pitch in the contonation, this will invalidate it as a site for an enclitic-induced high tone.
Once this principle is in place, other rules emerge for basic enclitic accentuation. Ultimately the key is to dispense with reductive notions of clitic behavior in Greek that suppose that enclitics become part of the preceding word in some simple way during pronunciation. In my analysis, clitic boundaries create a special environment in which the host+enclitic domains (clitic boundaries below are symbolized with =) are sometimes prosodic words but, in those scenarios in which an additional acute is induced, what is actually happening is the creation of prosodic phrases (φ) with more than one prosodic word (ω). Thus, we have examples such as [[φοῖνιξ=τις]ω]φ and [[ὁδοῦ=τινος]ω]φ where ω and φ coextend, but [[δῶ]ω[ρóν=τι]ω]φ and [[λόγων=τι]ω[νῶν]ω]φ, where we have two ωs. A few simple principles (e.g., 1. the maximum window is a syllable with high tone plus two following syllables; 2. the two syllables on either side of a clitic boundary must belong to the same prosodic word; etc.) produce all of what we consider normal simple enclisis.
I outline these and conclude with thoughts on synenclisis. The proposal is difficult to reconcile with some ancient grammarians’ prescription (and the normal modern editorial practice) of accenting all but the final enclitic, but it could account for something like the “every other enclitic” accentuation described by others and found in some manuscripts, including many good ones (cf. Barrett 1964: 427).