80 Similarly, the initial vowel of ‘leopard’ and ‘jackal’ is not lengthened even though it is
followed by a prenasalized consonant. Recall that throughout this thesis, length will only be written when it is phonemic or derived, but not conditioned.
3.3.3 Derived vowel length
The final source of length in Ikoma is from vowel concatenation at morpheme boundaries. When vowels meet in hiatus, a number of different processes may occur
depending on the vowels involved and their position in the word. The chart in 66, repeated from 20 above, gives examples of derived vowel length.
66 Derived vowel length at morpheme boundaries
. .
. .
7 8
- 8 - ,8
D - 8
D D
- 8
D -7
8 -
8 -
=8 -
D -
8 D
D -
8 D
In each of these examples except for mbatuki i, the subject prefix a- is adjacent to another vowel, either the TA prefix a- in the past forms or the vowel-initial stem
a ka. It is difficult to know whether the best analysis is that the first vowel is deleted
and compensatory lengthening follows, or if the vowels simply remain adjacent, resulting in a long vowel. The result, however, is clearly long.
Also, note that in the final example, mbaa éki i actually has three adjacent a vowels underlyingly, but this sequence does not surface as an extra long vowel. Instead,
it results in neutralization of the length contrast between the perfective and past forms when the verb root is vowel-initial. Clements describes a similar phenomenon in
Luganda, for which he posits a two-mora limit on surface length. He notes, “Underlying
81 sequences of three or more vowels obey the same principle they always reduce to
bimoric surface vowels” 1986 55. We see, then, that Ikoma’s neutralization of underlying sequences of two and three vowels is not unusual.
Recall from the previous section that long vowels can only occur in word-initial syllables if the syllable has an onset. For example, 67 below shows the pattern in which
the narrative verbal prefix aka- results in a long vowel in the first syllable when preceded by the 3P subject prefix a- but only in a short vowel when preceded by the
onsetless 3S subject prefix a-. 67
Word-initial vowel length only following an onset a.
3S a-aka-tuk-a
[a atuka] ‘heshe dug’
3P a-aka-tuk-a
[ aa atuka] ‘they dug’
b. 3S
a-aka-a ek-a [akaa eka]
‘heshe built’ 3P
a-aka-a ek-a [ aakaa eka]
‘they built’ c.
3S a-aka-emb-a
[akeemba] ‘heshe sang’
3P a-aka-emb-a
[ aakeemba] ‘they sang’
The sets in a, b and c each show these two prefixes with a different verb stem. The first example in each set shows the stem with the 3S prefix a-, and even though there are
two vowels underlyingly the first from the subject prefix and the second from the narrative prefix, the vowel on the surface is short. The second example in each set has
the 3P subject prefix a-, but in these words the surface vowel is long. The pattern consistently shows that vowel length across morpheme boundaries in a word-initial
syllable is preserved only when the syllable has an onset. Note also that long vowels can occur in adjacent syllables, as in [ aakaa eka] ‘they built’ and [ aakeemba] ‘they sang’.
82 Note that examples involving many other subject prefixes confirm that the lengthening is
not simply a result of a long vowel in the 3P subject prefix. Finally, lengthening also results from elision in some verbal suffixes. See
examples in 68 below. 68
Lengthening from elision a.
n-a-to- oo e -i i [nato oo ii e]
‘he has told us’
FM
-3
S
-1
P
.
OB
-tell-
PFV
b. a-ika -i i
[ iikee e] ‘they have sat’
3
P
-sit-
PFV
In a, the final of the verb stem elides resulting in a lengthened vowel [ii]. The final vowel of the suffix also alternates to e for unknown reasons. In b there are two
separate instances of derived vowel length. At the first morpheme boundary, a + i results in [ii]. Here the first vowel has elided, and the remaining vowel has undergone
compensatory lengthening.
35
The second lengthened vowel [ee] results from the elision of one of the flaps in the sequence ika + i i. This process is less straightforward, as it
seems that the two morphemes have somehow fused together, even affecting the final vowel as well, as we also saw in a. This is an example of what is often called
“imbrication” in Bantu literature.
3.4 Mekacha’s previous analysis of Nata