110 3 Phonology3.4.3 Melodies
3.4.3.1 Types of melodies
A melody is a vocalic element or sequence of such elements that can be
mapped onto unspecified vocalic positions in a stem. The concept is relevant to
the formation of MAN verb stems (perfective versus imperfective, for
example), and to nominal ablaut plural. I use the notation <...> to denote
melodies (e.g.
A string like <o a ae», extracting just the V's of a trisyllable, can be spoken
of as a surface vocalic sequence (SVS). I use «...» (double angled brackets)
for SVS's The formal study of ablaut involves examining multiple SVS's and
extrapolating (where appropriate) more basic underlying melodies from them,
along with rules for associating melodic components to particular syllables (or
vowels). One can disagree about how "deep" we should go, that is, how
abstract and morphologically inclusive the melodies should be, particularly
with reference to (allegedly) underived stems (Sg of nouns, perhaps the
Shlmpf of verbs).
The most critical decision is how to combine short {as} with full {u ο α
e i} as melodic fragments. There is undeniable evidence for treating (a u i) as
a high-V set, which I label H, and for {ae a} as an opposing low set, which I
label L.
An argument could be made that in at least some morphological contexts,
Η should be specified more exactly as rounded or unrounded, i.e. as Hu or H^1.
The purpose of this would be to specify u or i as output when the Η of a
melody is associated with a full V (only a is possible when the V is short). The
usefulness of this division between Hu and H^1 is limited, since in some cases
the u or i is really part of the basic lexical representation. To avoid making the
notation too cumbersome, I will generally use just Η but will comment on full-
V realization.
In this analysis, V-length is separated analytically from pure melodic
elements that can be described in terms of the feature [±high], perhaps
secondarily specified as [±round]. The mid-height vowels {e o} are somewhat
peripheral to the system. As conditioning elements in Short-V Harmony, {e o}
behave like L vowels. Thus, in the T-ka dialect, the output of Sg vocalic prefix
a- or e- on nouns after Prefix Reduction is is "ae- if the following syllable has
{e α ae o}, but V if the following syllable has {u a i} (§3.2.6).
There are a few morphological contexts where mid-height e or ο seems to
result from a "compromise," namely where lexical /i/ or /u/ is overlaid by an L
melodic component. See §3.4.7, below, for details.
Especially for an underived noun stem (as seen in the Sg), it is reasonable
to take the lexical representation (the "deep" form) as a spelled-out sequence
of V's and C's as in English: idam 'face', e-daemi 'light reddish-brown
animal', ae-domm 'seeds', daimbaeraku 'inability to see in dim light'. There is
no basis for separating the vocalic melody from other aspects of the stem's