Verbal prefixes Vowel alternations in prefixes

150 118 Low-vowel prefixes K B K B K B K B K B K B K B K B 9 9 . , ́ 2 : ́ 0000 6 8 : 8 ́ , -- F 3333 = ́ - 6 2 1 : - 9 , - . 8 ́ 6 2 - Note that low-vowel prefixes are sometimes raised before i u. The amount of raising is gradient, and depends significantly on the speaker, the word, and the specific class prefix. For example, in the class 2 prefix [a a-], the augment is never raised, whereas the class prefix vowel is sometimes pronounced higher. Oddly, there is less tendency for raising with the class 9 and 10 prefixes. Note that gradient raising of a occurs not only with prefixes, but it is sometimes perceptible in noun and verb stems as well. These issues are discussed in more detail in §6.7 below, where I give some sample formant measurements and tentatively conclude that the raising is gradient and phonetic, not something which needs to be accounted for in the phonology.

5.2 Verbal prefixes

Vowel dissimilation in verbal prefixes is in many ways identical to the process observed in nominal prefixes. As with nominal prefixes, all verbal prefixes with underlying [e o] alternate to the high vowels [i u] before [-ATR] stem vowels. In nouns, only the noun 151 class prefix alternates, and the augment is always [e o a]. Similarly, in verbs, only the prefix nearest the root alternates, and all other leftward prefixes remain [e o a]. The remainder of this section shows examples of verbal prefix alternations, including a description of vowel dissimilation in noun class agreement and TAM prefixes. I also include a number of examples of verbs with multiple prefixes. First, however, we begin with simple infinitive verbs which have the infinitive prefix ko-. Because the prefix has an underlying mid vowel, it alternates to ku- before the [-ATR] vowels [ a]. In 119 below, I show the infinitive prefix with each root vowel. Two examples are given for each vowel. Note that the consonant of the prefix alternates as well, from underlying k to its voiced counterpart as a result of Dahl’s Law, which is discussed in §2.3.4 above. The examples in a show [+ATR] stems, and those in b show [-ATR] stems. In a, all prefixes have the mid-vowel [o], and in b all prefixes have the high-vowel [u]. 119 Infinitive prefix ko- with each root vowel N N N N A A A A 7 8 8 . 8 2 : . 8 8 9 - : 8 8 . -7 8 - 2 8 8 F 8 9 152 All other verbal prefixes, such as subject agreement, object agreement and TAM prefixes, behave in the same way as the infinitive prefix and the nominal prefixes described in the previous section. In all cases, non-low prefixes which are directly adjacent to the verb stem alternate between mid- and high-vowels. Mid-vowel prefixes occur before [i e o u] roots, and high-vowel prefixes occur before [ a ]. Low-vowel prefixes do not alternate, and there are no underlying high-vowel verb prefixes see more below. I now discuss prefixes of inflected verbs, which are more morphologically complex than infinitive verbs. For ease of reference, in 120 below I repeat the verb inflection template from 18 above. 120 Verb inflection template FM – Subject – Neg – TA – Object – Root – Extensions – Final – Locative stem See §2.2.2 for a more thorough introduction to verbal inflection. In 121 below I give examples of three subject prefixes which represent the three possible prefix vowels e o a. Their alternations are shown using two different verb roots, one with a [+ATR] vowel tuka and one with a [-ATR] vowel sa a. These six forms are all marked with the perfective aspect, which is represented by the suffix [-i i]. Since this aspect form uses only a suffix and initial focus marker FM, there is no prefix between the verb root and the subject prefix, allowing us to observe subject prefixes directly adjacent to verb stems. 153 121 Subject agreement prefixes on perfective verbs -; . H -; . H -; . H -; . H 8 8 8 8 2 : 2 : 2 : 2 : ? D ? D ? D ? D DDDD VVVV ---- 8 8 8 8 XXXX VVVV ---- XXXX D 8 D D D MI ? MK MI ? 2 : MK ? D ? D ? D ? D DDDD VVVV 8 8 8 8 XXXX VVVV XXXX D 8 D D D MI ? MK MI ? 2 : MK + D8 + D8 + D8 + D8 DDDD VVVV1111 8 8 8 8 XXXX VVVV1111 XXXX D 8 8 D D 8 D MI + MK MI + 2 : MK The 3PL subject prefix is a-, and since it has a low vowel it does not alternate. The 1PL subject prefix is to-. The vowel remains mid before a [+ATR] root such as [tuk], but it is realized as [tu-] before a [-ATR] root such as [sa ]. The Class 7 prefix ke- operates just like it’s nominal counterpart. It is [ke-] before [+ATR] roots, and it alternates to [ki-] before [-ATR] roots. In order to see how multiple prefixes interact, in 122 below I give examples of the same two verb roots and the same three subject prefixes used in 121 above. This time we see them in the imperfective aspect, which is formed with the prefix ako- between the subject prefix and the stem. 154 122 Multiple prefixes with the imperfective ako- prefix 64 -; . H -; . H -; . H -; . H 8 8 8 8 2 : 2 : 2 : 2 : ? D ? D ? D ? D DDDD VVVV ---- 8 8 8 8 XXXX VVVV ---- XXXX D 8 8 D D 8 D MI ? JI M MK MI ? JI M 2 : MK ? D ? D ? D ? D DDDD VVVV 8 8 8 8 XXXX VVVV XXXX D 8 8 D D 8 D MI ? JI M MK MI ? JI M 2 : MK + D8 + D8 + D8 + D8 DDDD VVVV1111 8 8 8 8 XXXX VVVV1111 XXXX D 8 8 8 D D 8 8 D MI + JI M MK MI + JI M 2 : MK We see that the IMPF prefix always surfaces as [ o-] before [tuk], but [ u-] before [sa ], which is what we expect since it is a mid-vowel prefix directly adjacent to the stem. 65 The subject prefixes, however, do not alternate since they are not adjacent to the stem. With both verb roots above, the 1PL subject prefix is always [to-], and the Class 7 prefix is always [ e-]. Alternation patterns described in previous sections show that only the [-ATR] vowels [ a ] trigger the change to high vowels, so we actually would not expect the subject prefixes in 122 above to alternate, since in all cases they precede the [+ATR] vowels [o u]. However, additional data shows that even [a] in a prefix does not trigger alternations in a preceding prefix. For example, see the narrative tense verbs in 123 64 Recall from §2.2.2 that the imperfective, narrative and incipient prefixes each have the shape VCV-, and the initial V of the prefix can be analyzed as an “empty mora.” It always contributes length to the preceding vowel usually the vowel of the subject prefix, and it never influences the quality of the preceding vowel. That is, it does not induce glide formation or any type of coalescence. As the examples in 122 illustrate, the combination of effects from this “empty mora” along with prefix vowel dissimilation and Dahl’s Law, results in surface forms of the imperfective prefix which vary significantly from its underlying form. 65 Recall that Dahl’s Law causes k in prefixes to alternate to [ ] before a voiceless consonant. 155 below, once again showing the same three subject prefixes with the same two verb roots used above. 123 Multiple prefixes with the narrative [aka-] prefix -; . H -; . H -; . H -; . H 8 8 8 8 2 : 2 : 2 : 2 : ? D ? D ? D ? D DDDD VVVV 8 8 8 8 XXXX VVVV XXXX D 8 8 D D 8 D ? AHN MK ? AHN 2 : MK ? D ? D ? D ? D DDDD VVVV 8 8 8 8 XXXX VVVV XXXX D 8 8 D D 8 X ? AHN MK ? AHN 2 : MK + D8 + D8 + D8 + D8 DDDD VVVV 8 8 8 8 XXXX VVVV XXXX D8 8 8 D D8 8 D ? AHN MK ? AHN 2 : MK The second vowel of the narrative prefix remains [a] before both verb stems, and it does not cause the 1PL and Class 7 subject prefixes to alternate from mid to high. We see, then, that not even [a] triggers harmony in prefixes which are not adjacent to the verb stem. This is parallel to the pattern which we saw in nominal prefixes in which only the noun class prefix alternates, but not the augment, since the augment is not adjacent to the stem. In order to give a more complete account of verbal prefix harmony, in 124 below I show subject agreement prefixes for all personal pronouns and noun classes. I list the noun class or personal pronoun, the possible surface forms of the prefix, and examples using the perfective aspect. Two examples are given for each class, one with the [+ATR] verb [ eha] ‘pay,’ and one with the [-ATR] verb [k a] ‘do.’ The perfective verbs are all formed with the focus marker prefix n-, followed by the subject prefix, 156 then the verb root, then the perfective suffix -i i. These examples can all be glossed as ‘I have paid,’ ‘I have done,’ ‘you have paid,’ ‘you have done,’ etc. 124 Subject agreement prefix alternations with perfective verbs -; -; -; -; TTTT 2 : 2 : 2 : 2 : 8888 D 8 D 8 8 ? D 8 ? D 8 ? - - 8 7 D 8 D 1 1 8 D 8 1 1 8 + 8 D 8 1 1 8 D - - 8 D 8 3 D 8 D 8 8 1 1 8 D - - 8 8 D 8 1 1 8 8 D 8 3 D 1 1 8 Just as we have seen elsewhere, all of the non-low prefixes alternate based on the root vowel. All non-low prefixes are mid [e o] before the stem [ eha], but they surface as the high vowels [i u] with the stem [k a]. Additional examples with multiple prefixes from various classes are in 125 below. All four examples show object prefixes as well as subject and TAM prefixes. 157 Examples a and b both use the 3S object prefix mo-, which has a high vowel in a and a mid vowel in b. Both c and d use the 2S object prefix ko-, which has a mid vowel in c and a high vowel in d. 125 Additional examples with multiple prefixes 7777 VVVV FFFF XXXX D F D AG 7 C H ? MK -7 -7 -7 -7 VVVV 8888 XXXX D 8 D AHN 7 C 2: 0 2 .7 .7 .7 .7 VVVV XXXX D 8 8 D JI M 7 C 2 H ? MK J 9 2 : 7 7 7 7 VVVV 8888 XXXX D 8 8 8 D JI M 7 C H ? MK J 9 : We see, then, that object prefixes are nearly always identical in form and behavior to subject prefixes. The description in this section, especially concerning verbs with multiple prefixes, leads to an important point in the analysis of Ikoma prefixes. In both nouns and verbs, all prefixes not adjacent to the stem are always [e o a], and this is highly suggestive that these are the underlying prefix vowels. Given this pattern, it is very difficult to make a case for underlying high prefix vowels. As already mentioned in §5.1, once we are convinced that the mid vowel prefixes are underlying, it is only a short jump to the conclusion that the [-ATR] vowels are the ones which trigger the alternation. For this 158 reason, despite the fact that [-ATR] does not actually spread to the prefixes, I consider Ikoma prefix harmony to be a functional equivalent to other types of [-ATR] dominance, such as that found in Bantu C and Komo. See §5.5 for more discussion of this point.

5.3 Acoustic analysis of prefixes