3.4 Ablaut 135application of Default Accentuation, so there is no accentual trace of the
deleted suffixal /EC/. The output e is the regular contraction of /aae/ at a stem-
suffix boundary (§3.2.3.3).
With 3MaPl -aen, we get the problematic aksae-n (108.d). The accent
cannot shift farther left, and we see as rather than e at the morpheme boundary.
If we analyse sksae-n as due to Presuffixal α-Shortening followed by
VV-Contraction (37.b), we have the derivation in (109.a). Here stem-final χ
could be identified with deletable /A/. If VV-Contraction (37.d) by itself is
posited, we get the derivation in (109.b).
(109) Alternative Derivations of aksae-n 'they-Ma ate'
a. with Presuffixal α-Shortening (preferred)aksa-asn underlying
sksas-sen Default Accentuation, Presuffixal α-Shortening
sksae-n VV-Contraction (37.c)b. without Presuffixal α-Shortening (dispreferred)aksa-aen underlying
aksa-aen Default Accentuation
aks-aen VV-Contraction (37.d)In (109.a), Presuffixal α-Shortening results in a sequence of two /as/'s,
which then contract to a single se. (It is difficult to decide whether to
hyphenate the surface form as sksae-n or aks-aen). In (109.b), the stem-final α
abruptly disappears before the suffixal short V. Since this is not the way
VV-Contraction works in other stem-suffix combinations, the derivation
(109.b) is questionable, and I prefer (109.a).
Presuffixal α-Shortening is blocked (or undone) when the relevant stem-
final V is "protected" by an ablaut lengthening formative χ-pcl or χ-f. A
Resit example is 3FePl aksd-nset 'they-Fe have already eaten', cf. PerfP
sksae-nast. A LoImpfP example is t-imtalli-naet 'they-Fe are confused', where
stem-final i is not shortened. The apparently exceptional failure to shorten the
V visibly can be accounted for in either of two ways. In one version,
Presuffixal α-Shortening is simply blocked by this ablaut formative. In the
second version, Presuffixal α-Shortening does apply at an early stage, but the
shortened V is then (re-)lengthened by applying the relevant χ formative. The
latter solution is preferable on several grounds, but chiefly because the
relengthening of /ae/ is appropriately a, never e.
The same pattern of shortening (i.e. provided that an χ formative does not
protect the full V) is observed with the only C-initial Participial suffix, PI -nen
(contrast MaSg -aen and FeSg -Eet). Participles are used in subject relatives
(§8.5). Participles for -vksi> 'eat' are in (110).