Summary A case study of secondary palatalization in Isthmus Mixe

64 Ci. Overall, this does not affect the F2 mean of C j ‚. On the average then, no vowel quality shifts into the formants of another vowel quality, even when fronted by the effects of palatalization. Each vowel quality demonstrates two allophones, one which occurs adjacent to plain consonants and a second, more fronted version, which occurs adjacent to palatalized consonants. ← F2 F1 ↓ 24 00 2100 1800 1500 1200 900 200 j i i j ‚ 300 ‚ j u u j e 400 e j o o 500 j a a 600 Figure 4.19. Isthmus Mixe vowels adjacent to palatalized and plain consonants. The symbol j preceding the vowel indicates vowel means adjacent to palatalized consonants. A plain vowel indicates vowel means adjacent to plain consonants.

4.7 Summary

Secondary palatalization of every consonant in the Isthmus Mixe inventory is attested, including the laryngeal consonants h and Ð. Occasional reference in the literature is made to palatalized h, such as in foreign words in Lithuanian Kenstowicz 1971:8, and to both palatalized h and Ð, such as in Copainalá Zoque Sagey 1986:108–109, see section 5.4, and in Coatlán Mixe Hoogshagen 1984:4, Van Haitsma [Dieterman] and Van Haitsma 1976:5–11, where the entire consonant inventory is said to be modified by morpheme-induced palatalization, which implies that the laryngeal consonants h and Ð are also palatalized. However, Wichmann 1995:xix does not recognize palatalization of the glottal stop in the Oaxacan Mixe languages. The lack of references to palatalized h and Ð in the literature other than descriptions of Mixe-Zoque languages suggests that palatalization of these laryngeal consonants is of infrequent occurrence in the world’s languages. Because the palatalization of these consonants is morphemically induced in the Oaxacan Mixe languages, any description of the grammar of these languages necessitates an adequate treatment of the palatalized laryngeal consonants further discussed in sections 5.4 and 5.5. In the past, when the linear model was used in the Mixe languages to describe secondary palatalization, the third person marker was represented by the phoneme j word initially Wichmann 1995 or by metathesis of j with the consonant Crawford 1963. A most unusual set of consonant clusters occurs, either as jC Wichmann or Cj Crawford in that nearly every consonant in the inventory occurs in these clusters as C. 81 In either case, the phonetic reality of simultaneity is obscured and the link to the 81 Wichmann 1995 omits the glottal stop as being palatalized, as well as the palatal consonant; Crawford 1963 omits only the palatal consonant. 65 morpheme is not overtly expressed. 82 In addition to the evidence that secondary palatalization modifies every consonant in Isthmus Mixe except j, it has been shown that every vowel quality is also affected, in that the fronted allophone occurs adjacent to palatalized consonants. See section 5.3 for further discussion regarding the linear model representation. To date, there has been no explanation of secondary palatalization in the Oaxacan Mixe languages that encompasses all of the aspects of this phenomenon in one unified description. It will be shown in section 5.5.1 that a single autosegmental feature may be used to describe all occurrences of morpheme-induced secondary palatalization and its phonetic effects on all of the consonants and vowels, thus, explicating the relationship between the morphemes represented by secondary palatalization and the phonetic manifestations.

5. Theoretical Implications

5.1 Introduction

Although the previous sections have been mainly descriptive in character, it is essential to consider the theoretical implications of the data. According to Chomsky and Halle 1968, generative phonological theory has as its foundation the phonetic structure of speech. Their phonological representation is a linear arrangement of phonetic matrices to represent each sound segment. There is still some controversy over minor phonetic details, but in general, the universal inventory of contrastive elements is described according to the vocal tract articulators. Linear phonological representations characterize each sound as a feature matrix fully specified with distinctive features, as provided by Universal Grammar. Phonological theory textbooks, such as Durand 1990 and Kenstowicz 1994, describe this linear arrangement in detail and also present a hierarchical display of features in a nonlinear approach, known as feature geometry, which has largely superseded the linear model. 83 A nonlinear representation has, at minimum, three tiers that are essential to the representation. The most basic tier is the timing tier, which may be characterized by Xs which suggest that each slot equals one abstract unit of time, or by consonants and vowels that suggest a tier divided into two types of timing elements. The second tier, located above the timing tier, organizes the units of timing into syllables that are formed per language specific constraints. A third tier is below the timing tier, representing the featural makeup of the individual segments.

5.2 Feature geometry representations

The Isthmus Mixe data shown in the previous chapters are more adequately represented using feature geometry models, rather than linear models, since feature geometry provides a better balance between phonetic and phonological perspectives than the linear models are able to do. In applying feature geometry models to the Isthmus Mixe data, nasal place assimilation and voicing spread will be explained in section 5.2.1. Although a relatively simple model adequately explains the nasal process, this application of feature geometry to the nasal-plosive consonant cluster is complicated by the addition of a secondary palatalization morpheme that expresses a verbal suffix clause-type marker section 5.5.1. Both consonants in the nasal- plosive consonant cluster are mutated by this palatalization, and the vowels which precede and follow this 82 Crawford 1963:39, 79–81 discusses the third person marker morpheme as the j of the Cj cluster and the importance of morphophonemics in his description of Totontepec Mixe. However, his description was written before the autosegmental theory was developed. 83 It is assumed that the reader is familiar with generative phonology and the classic works of Chomsky and Halle 1968, Kenstowicz and Kisseberth 1979, the more recent work of Kenstowicz 1994, and other notable works too numerous to mention here.