Vowels in both nouns and verbs Speaker B

89 sort of clusters such as palatalized or labialized consonants, with the exception of the verb ko- jeka ‘to bury.’ Otherwise, a variety of consonants were used. The infinitive form was used primarily because it is the citation form for verbs. It is also helpful because the infinitive prefix vowel alternations from o to u provide significant clues for the harmony system and are useful for prefix vowel measurements, which are discussed in Chapter 5. Also, by consistently limiting the final vowel to a, I avoid any possible influence on the root vowel made by other final vowels. For nouns, for each of the seven vowels I chose six canonical nouns with bisyllabic CVCV roots. In all of these words, both root vowels are identical, which should prevent any assimilatory effects from one vowel to another. The primary purpose of this list is to investigate vowels in the initial syllable of the root, but I also had to make decisions about noun classes and noun class prefixes, since prefix alternations are also a major point of inquiry. Out of necessity, I used nouns from a wide variety of noun classes. For recording, the list was divided into nouns and verbs and then broken down into smaller sections of approximately ten to fifteen words each. Each section was then randomized so that, for example, the nouns with CiCi stems were not all together, but were mixed in with CoCo, CaCa, etc., stems. For nouns, speakers pronounced two repetitions of the singular form followed by two repetitions of the plural form. For verbs, they pronounced three repetitions of the infinitive.

3.5.2 Vowels in both nouns and verbs Speaker B

Because of time constraints, it was not feasible to make measurements of all recorded nouns and verbs for all four speakers. Therefore, in this section, I present measurements 90 and plots from nouns and verbs from a single speaker Speaker B. In the follow section §3.5.3, I present a comparison of the vowels of all four speakers in verbs only. The table in 70 below contains average formant values for vowels in the stem- initial syllable V1 of both nouns and verbs. As mentioned in the previous section, this wordlist consisted of six nouns and five verbs per vowel, and three to four repetitions were made of each word. The measurements in 70 below are averages of a total of 221 tokens 119 from nouns and 102 from verbs. For nouns, I measured the initial root vowel of both singular repetitions and the first plural repetition, which totals three tokens per word on the wordlist. For verbs, I measured the root vowel of all three repetitions of each infinitive, once again totaling three tokens of the vowel per word. Note, however, that occasionally I was not able to measure some tokens for various reasons, so these are approximate numbers of tokens. 70 Average values of V1 Speaker B 45 year old male Noun and verb averages Noun averages Verb averages approx. 32 tokensvowel approx. 17 tokensvowel approx. 15 tokensvowel F1 F2 F1 F2 F1 F2 u 285 759 u 280 732 u 289 785 o 374 891 o 361 887 o 390 897 468 951 473 947 460 958 i 310 2271 i 304 2225 i 317 2323 e 381 2025 e 366 1965 e 396 1964 455 1873 440 1911 470 1834 a 671 1462 a 694 1474 a 640 1446 A number of comments can be made about this table. Looking at the combined averages in the first column, we see that for the back vowels, there is an approximately 90Hz difference in F1 from one vowel to the next. For the front vowels, on the other hand, the difference is only appro confined than the back are fairly similar, and c have the same seven vo The averages of below. Vowels plots tra downward, and F2 on t Maddieson 2003 16 e auditorily based vowel but has the advantage o when the formant value their place on an IPA v 71 Plot of proximately 70Hz. We see then that the front vo ck vowel space. When comparing the noun and d certainly similar enough to support the view t vowel phonemes. of nouns and verbs together are shown in the v traditionally have F1 on the vertical axis, with n the horizontal axis with values increasing from explains “This type of display closely paralle el space based on perceived ‘height’ and ‘back e of being based on verifiable measurement.” A lues are plotted in this manner, the vowels are s vowel chart. f V1 average values, with nouns and verbs toge 91 t vowel space is more and verb averages, they that both categories e vowel plot in 71 th values increasing rom right to left. llels the traditional ckness’ values used … ” As 71 below shows, e similar in position to ogether Speaker B As the plot shows, Spe figures in 70 above, t vowels, with [ ] having tendency for [u] to be a shown in the following vowels, with an F1 ave The plot in 72 individual tokens of V1 72 Plot of peaker B’s vowel space is quite symmetrical. A , the back vowels are spaced slightly more apar ing a higher F1 than [ ], and [u] having a lower e acoustically higher than [i] holds for the other ing section. The low vowel [a] is also quite far a verage of more than 200 Hz higher than the mi 72 below also represents Speaker B’s nouns an V1 are plotted. f V1 with all points, with nouns and verbs toge 92 As we saw with the part than the front er F1 than [i]. The her speakers as well, as r away from the other mid [-ATR] vowels. and verbs, but here all gether Speaker B Note that there is more back vowels. There is a than front vowels. The plot of aver for verbs only is in 74 73 Plot of re overlap in the front vowel space, with only s is also a much greater range of F2 measurement verage formant values for nouns only is in 73 74 below. f V1 average values of nouns only Speaker B 93 y slight overlap in the ents for back vowels 3 below, and the plot B 74 Plot of These plots show more well-spaced than those the noun and verb aver measurements from no all four speakers are in

3.5.3 Cross-spea