3.5 Syntactically controlled phonological processes 155presence of a pronominal-subject prefix (separating the accented
demonstrative from the accented verb stem). For example, 3MaSg /w-d
i-baeddaed-aen/ 'he who stands' has no clashing (i.e. adjacent) accented
syllables, so it appears straightforwardly as w-α i-baeddaed-aen after accent
rules. On the other hand, underlying 3FeSg /t-d t-baeddaed-aet/ becomes /t-d
0-baeddaid-aet/after Prefixal t-Deletion (§7.4.1), and so at this stage has
adjacent accents. This clash would be resolved by Rightward Accent Shift,
producing /t-a baeddaed-aet/ and surface t-ά baeddaed-aet.
While clash avoidance might have been a key factor in the historical origin
of Rightward Accent Shift, it is a dubious synchronic analysis. To begin with,
when a clitic is hosted by the demonstrative at the beginning of a definite
relative, the clitic (whether accented or not) has no effect on Rightward Accent
Shift on the following verb although the clitic alters the metrical structure of
the phrase. In addition, accent within an accentual phrase works right to left,
and once a primary accent is established the general pattern is alternating-
syllable secondary accents going to the left, so any lexical or grammatical
accents to the left of the primary accent are simply overridden. In the normal
course of events, /t-d baeddaed-aet/ should therefore appear as #t-a bseddaed-ast
(with no overt accent on the demonstrative). This is not the actual output, but it
would be perfectly pronounceable and would obviate accent clashes. I
conclude that Rightward Accent Shift is now a morphophonological rule with
no clear synchronic phonological motivation.
In (133), Rightward Accent Shift occurs (along with χ-pcl Erasure) in a
prepositional relative (cf. LoImpfP -nass 'lie down'). The initial demonstrative
t-d is followed by the cliticized preposition, which has no inherent accent, but
gets a secondary accent by the usual right to left secondary accentuation
process. Rightward Accent Shift operates on the verb in the same way it would
without the intervening cliticized preposition. Therefore a clash-avoidance
analysis, by which t-d induces rightward shift of an immediately following
accent, cannot work synchronically.
(133) t-e-sela-t-t [t-a-\fsel naess-aer]
Fe-Sg-mat-Fe-FeSg [Fe-Dem.Sg-\on Iie.down.LoImpfP-lSgS]
'the mat on which I am lying.' [K-d]
As t-ά 0-basddaed-aet 'she who stands' shows, a relative with 3FeSg
subject is subject to Rightward Accent Shift just like other subject categories
as long as there is a syllable for the shifted accent to appear on. The 0- here
represents 3FeSg subject prefix t-, which is audible on FeSg participles (and
3FeSg subject inflected verbs) whose stems begin with a V, but is deleted
before a C-initial stem.