Summary and implications for [ATR] markedness

173 that the perceptual motivation has some role to play. It is possible that perceptual and theoretical motivations such as the OT analysis presented above have conspired together to cause the height-alternating prefixes. 5.5 Summary and implications for [ATR] markedness This chapter has described vowel harmony in both nominal and verbal prefixes in §5.1 and §5.2, respectively, focusing especially on the prefixes which alternate from mid to high. The acoustic analysis in §5.3 supports the claims made concerning both alternating and non-alternating prefixes. In §5.4 I suggest possible theoretical and perceptual explanations for these patterns. An important issue addressed in several of these sections is the two interrelated questions concerning the underlying height of the prefix vowels mid or high?, as well as the feature which triggers the height alternation [-ATR], [+ATR] or a height feature?. I have argued that the alternating prefixes must have mid vowels underlyingly, which strongly suggests that [-ATR] is the trigger for prefix dissimilation. To summarize the argument, despite the fact that there is no [-ATR] spreading to prefixes, there are two main pieces of evidence in favor of [-ATR] as the dominant value the distribution restrictions against [-ATR] mid vowels in prefixes, as well as the evidence pointing toward the mid-vowel prefixes, not the high-vowel prefixes, as the underlying forms. As mentioned above, the main indication of mid-vowel underlying forms is the occurrence of only mid vowels and not high vowels in prefixes which are not adjacent to the stem, seen not only in the nominal augment but especially in verbal prefixes as well. This evidence of [-ATR] markedness is discussed further in Chapter 7 when looking at Ikoma’s vowel harmony as a whole and its implications for our 174 theoretical and typlogical understanding of vowel harmony. I make one final point here to set the stage for that discussion. In a presentation by Casali and Leitch 2002, they make a connection between a language’s dominant [ATR] value and the default value of prefixes which are outside of the harmonic domain. In particular, they focus on the Bantu C languages in which [-ATR] is dominant. There is a correlation in Bantu C for affixes to surface as [+ATR] mid vowels in cases in which [-ATR] spreading does not apply. Therefore, [+ATR] is default, and affixes only surface as [-ATR] mid vowels when the active value spreads. This resonates well with the patterns found in Ikoma in which [+ATR] is the default value in prefixes. This is in line with the conclusion that [-ATR] is the active value in Ikoma prefix dissimilation, just as it is the active value in Bantu C. Compare this to neighboring Zanaki, in which [+ATR] is clearly the dominant, marked value. In Zanaki, prefixes are underlyingly [-ATR], and they surface as such unless they precede the [+ATR] vowels [i u]. The underlying prefix forms are, for example, [ m -] for Class 1 and [ k -] for Class 7. Just as [+ATR] is default in a [-ATR] dominant system, in Zanaki we see that [-ATR] is default in a [+ATR] system. Therefore, just as we would not expect primarily underlyingly [+ATR] prefixes in a language with [+ATR] dominance, we would also not expect [-ATR] prefixes to be the standard in a language with [-ATR] dominance. Therefore, if we concede that [-ATR] is dominant in Ikoma at the very least in the process of prefix dissimilation, then it is in fact expected that Ikoma would have underlyingly [+ATR] mid-vowel prefixes, which is indeed the case. 175 As a final point on this subject, see also the following quote from Beckman 1997:1 from her article on positional faithfulness and positional neutralization in Shona vowel harmony. In languages which exhibit positional neutralisation of vowel contrasts, one or more vowels generally, the most marked members of the vowel inventory may occur distinctively in only a small subset of the structural positions available in the language. Outside of these positions, the marked vowels may surface only if they harmonise with a similar vowel in the privileged position. Once again, we see that if [ ] are the most marked members of the vowel inventory, these are the vowels which we would expect to have a more restricted distribution to only certain positions. Since they do not occur at all in prefixes, this is perhaps an indication that their status as being marked plays a major role in the nature of prefix alternations. For these reasons, I argue that [-ATR] is the marked and dominant value in prefix dissimilation. This is an important point to understand, because there is also evidence elsewhere in Ikoma in stems and suffixes that [+ATR] is the marked and dominant value. Though this is an unusual and complex markedness situation, we must allow for this complexity in order to adequately account for the facts. See Chapter 7 for more discussion of this markedness quandary. 176

Chapter 6: Vowel harmony in verb stems and suffixes

Vowel harmony in verbal suffixes is well-attested in Bantu languages. In fact, in many Bantu languages it is the only place in which vowel harmony occurs. The typically cited cases of vowel harmony involve the applicative and stative front-vowel extensions and the inversive and inversive-stative back-vowel extensions see Hyman 1999. Ikoma has harmony in these suffixes, and it also has a number of suffixes that induce harmony in verb roots. As described above in the introduction to verbal morphology §2.2, there are both derivational and inflectional suffixes in Bantu languages. Derivational suffixes follow the verb root, and they are considered to be part of the verb stem. In Bantu literature, derivational suffixes are usually called “extensions” since they extend the stem. These include the applicative, stative, causative, passive, inversive, and reciprocal extensions. Multiple derivational suffixes can be added to a single verb root, sometimes causing a stem which consists of several morphemes. Concerning inflectional suffixes, the basic imperfective affirmative suffix is the final vowel FV [-a]. Ikoma has two other inflectional suffixes as well the perfective [- i i] and the subjunctive [-e]. There are two main types of harmony involving suffixes, and the harmony type depends not on whether or not the suffix is derivational or inflectional, but instead on the underlying vowel of the suffix and its position in the word. For that reason, in this chapter I group suffixes according to their harmony patterns, not based on their status as either derivational or inflectional. See §2.2.2 for a basic review of inflectional verb morphology and §2.2.3 for a review of derivational verb morphology.