239 In summary, it is clear that some degree of low-vowel assimilation occurs,
especially in noun and verb stems. However, there is no convincing evidence which suggests that it could be categorical assimilation. Prefix raising is even less clear, once
again suggesting that low-level phonetic assimilation is the most we can confidently propose.
6.8 Summary
This chapter has given a detailed look at a number of Ikoma verbal suffixes and suffix combinations. The table in 203 below summarizes all of the root and suffix vowel
alternations which have been described in this chapter. Under each suffix heading, I list the underlying form of the suffixes, and then for each vowel I list the surface form of
the root vowel and suffix when the two are combined. Parentheses draw attention to root vowel alternations. Question marks indicate patterns for which limited data is available,
indicating that the analysis is tentative. 203
Summary of root and suffix vowel alternations
V1 Applicative Subjunctive
Inversive Causative,
etc. Applicative
-Causative Inversive-
Causative
- r -e
-o , -ok -i
- -i -o -i
c
Note the wide variety of suffix vowels which follow a. For both the applicative and subjunctive, a is followed by the underlying form of the suffix, which is [-ATR] and
[+ATR], respectively. In both cases, a does not appear to cause alternations. This is in
240 line with the noun stem co-occurrence patterns described in §4.1. There we saw that both
[+ATR] and [-ATR] mid vowels can following a, and that [+ATR] mid vowels are more common. We can also attribute the lack of suffix alternations to a different behavior of a
in prefix harmony and suffix harmony. Leitch 1997 describes the case of Mbosi Oléé Bantu C20 in which low vowels cause [rtr] harmony in prefixes but not in suffixes.
Ikoma seems to have similar behavior in this respect. This chapter has also described in detail two types of vowel raising. First, before
dominant suffixes, is categorically raised to [e]. Second, before the same dominant suffixes, is noticeably raised, but it is not categorical. The low vowel a, and in fact
all other vowels, are also raised in a number of environments, but again, it is most likely not categorical. These patterns are decisive evidence that featureal asymmetry explains
the data much more simply than any type of morphological asymmetry, such as that proposed in root-contol harmony languages. In Ikoma, superficial [+ATR] spreading to
the stem is more important than preserving the quality of the root vowel. This is different than the tendency described by Beckman 1997 3 in her work on Shona vowel harmony.
She notes that “vowel harmony is initiated by the root-initial syllable, because faithfulness to underlying contrasts in this position is paramount.” Though preservation
of contrasts in the root is a common pattern in many languages, it is also violated in many harmony languages, including Ikoma.
A final theme throughout this chapter is the asymmetry between front and back [-ATR] vowels. Harmony patterns indicate that there is a hierarchy in the “strength” of
[-ATR] vowels as triggers of [-ATR] spreading and targets of [+ATR] spreading or, [-ATR] delinking, as outlined in 204 below. All three [-ATR] vowels induce
241 dissimilation in prefixes, but the table below shows their different behaviors concerning
stem and suffix harmony in verbs. 204
[-ATR] vowel strength hierarchy Vowel Strength
Triggers prefix dissimilation
Spreads [-ATR] in verb stems
Target of [+ATR] spreading
Stronger a
81
Weaker The back vowel is the strongest [-ATR] trigger. It always spreads [-ATR] rightward
except to high vowels and is never a target of [+ATR] spreading. The front vowel is clearly weaker. The most obvious evidence of its weakness is that it is a target of leftward
[+ATR] spreading. There is no evidence of spreading [-ATR] rightward in the stem, but this is perhaps a result of other factors, such as the fact that the applicative suffix is
already [-ATR], and spreading [-ATR] to the subjunctive, which is a potential target, would result in - being in a restricted FV position.
Concerning the low vowel a, once again, other factors prevent us from getting a full picture of its character as a [-ATR] vowel. Since the applicative is [-ATR], we do not
know if it would normally spread [-ATR] within the stem. Its behavior with the inversive is also unusual, since it triggers an alternation from -o to [-u ]. Overall, we might
consider a to be weaker than since it does not spread [-ATR] to the subjunctive, but it is stronger than since it is not a target of leftward [+ATR] spreading. Of course this
fact could also be explained by the common constraint against low [+ATR] vowels. Harmony in verb stems and suffixes is complex, and the analysis which I propose
in this chapter must remain tentative for a number of reasons, some of which I have
81
Even though a does not seem to trigger alternations of the subjunctive and applicative suffix, it does cause alternations of the inversive. As I describe above, the inversive, which is -o underlyingly, alternates
to [-u ] after a, perhaps as some sort of dissimilation.
242 already touched upon. The first major complication to the overall analysis is that the
underlying form of the applicative suffix is not entirely clear. Recall that the most obvious and transparent analysis is that it is underlyingly [-ATR], since this variant
occurs after all root vowels except for e o, where we find the [+ATR] variant. However, if we posit a [-ATR] UF, it introduces several other challenges into the analysis, as
described more fully in §6.1. In particular, it requires us to account for apparent rightward [+ATR] spreading from root to suffix—a pattern which occurs nowhere else in
in the language. Aside from the applicative suffix, the second complication concerns the inversive
suffix. Because of a lack of inversive examples involving root vowels, the inversive pattern is also not entirely clear. There is a single potential example of the inversive with
this underlying root vowel, but the relationship between the basic and inversive form is questionable. Without additional examples, it is impossible to be sure about whether or
not there is [+ATR] spreading from the inversive suffix to the root vowel . As I discuss below, I set aside this issue until further evidence is found.
243
Chapter 7: Understanding Ikoma’s [ATR] markedness
patterns
Chapters 3 through 6 are primarily descriptive in nature, first describing the vowel phonemes from a number of perspectives, and then describing the vowel harmony
patterns which occur in prefixes, noun stems and verb stems. These chapters also include a significant amount of acoustic phonetic data in order to better describe the patterns and
to justify a number of analytical issues which must be dealt with during the process of description, especially concerning the nature of root and suffix vowel alternations.
In the description of vowel alternations in prefixes, stems and suffixes, we have encountered an interesting situation in Ikoma in which it seems that two opposite values
of the feature [ATR] are active in the language. Also, though both values of [ATR] are marked in some sense, the resulting processes are not like we might expect. The [-ATR]
markedness exhibited in prefix alternations is strange; instead of a more normal process of assimilation to the marked value, in Ikoma [-ATR] vowels trigger a prefix height
dissimilation. We have also observed an asymmetry in the strength of different [-ATR] vowels. For example, is more prone to spread [-ATR] and to resist [+ATR] spreading,
whereas is more often a target of [+ATR] spreading. Concerning the apparent [+ATR] spreading in roots and suffixes, it not only applies asymmetrically to front and back
vowels, but it also targets only short vowels. When evaluating the many vowel patterns described in this thesis, at first glance it
seems intuitive that there is much more evidence of regular [+ATR] markedness and dominance in Ikoma, especially since so many suffixes seem to spread [+ATR] to the
root. Nonetheless, despite the many places of clear alternation to [+ATR] vowels, there