Stability Soyaltepec Mazatec in Autosegmental Phonology

242 designated structure, i.e., whether or not tones are specifically linked in the underlying representation, and not just in the fact that they are contour tones. The processes described above demonstrate both tonal mobility which is a characteristic of African tone systems and the greater affinity of some tones to specific TBUs which is more similar to Asian tone systems. The tonal behavior was demonstrated using RTT structures; however, no specific advantages to using these structures emerged. Tonal mobility could just as easily be demonstrated without the expanded geometry of RTT. The only distinction that occurres using the RTT geometry is that OCP motivated merger is demonstrated on the tonal melody tier in 6a and on the tonal register tier in 2d, 3d, 6b and 7d.

5.2.2 Stability

Yip 2002: 65 defines stability as the survival of a tone after the loss of the original host segment. Arguments for stability are either diachronic or synchronic. For the purposes of this study I address an example of synchronic stability, the case of the diminutive morpheme ʔi˩tʃ︢i˦˥ ‘small.’ When the morpheme in question is said in isolation, it is a two syllable word which begins with a low-toned syllable followed by a syllable with a Mid-high to High rising contour. The most common usage of the morpheme; however, is as a suffix to the noun which it describes. In this case, it consists of a single syllable with segments and tone equivalent to the second syllable of the stand alone morpheme: - tʃ︢i with a Mid-high to High rising tone. The first syllable of the morpheme is subject to elision. 118 However, when the morpheme is affixed as in 8, a Low tone is added to the final syllable of the stem of the noun 8c. This demonstrates that even though the segments of the first syllable have been deleted, the tone remains. In 8 the stages of the derivation of ti˧˩tʃ︢i˦˥ ‘small boy’ are explained beginning in 8a with the underlying representation of the individual morphemes, a noun with a level Mid tone and the disyllabic diminutive in question. This is followed by the elision required for suffix formation in 8b. The Low tone which remains 118 A counter argument could be made in favor of the insertion of an epenthetic syllable necessary to support the tone in isolation rather than elision. There is not enough evidence at this time for a definitive argument. 243 after the segments are deleted then docks to the left 8c in violation of the UAC. Finally in 8d the OCP is satisfied by the merger of adjacent features and the surface form emerges. 8 Tonal stability of the L in ʔi˩tʃ︢i˦˥ despite segmental elision a. UR h l l h L L H H ti ʔi tʃ︢i ‘boy’ ‘small’ b. Suffix formation - elision h l l h L L H H ti t ʃ︢i c. Linkages form contra the UAC? h l l h L L H H ti t ʃ︢i d. OCP merger yielding surface representation h l h L H ti t ʃ︢i ‘small boy’ In 8a the underlying representations, the noun has one TBU and one TRN a level Mid tone while the diminutive has two TBUs and three TRNs a Low, a Mid-high and a High. When the suffix is formed, the 244 segmental material of the first syllable is deleted 8b; however, the tones remains unchanged leaving three TRNs and one TBU. In both 8a and 8b an association line is indicated between the Mid-high tone and the final syllable of ʔi˩tʃ︢i˦˥. It might be argued that the process of elision creates the link or it could be present in the UR. This link must be in place before the linkages are formed in order to force the Low tone to associate to the left of the suffix in the newly formed word 8c. In 8c the combination of the two morphemes shows that the L tone TRN which was sponsored by the first syllable of the diminutive morpheme docks to the stem of the noun which is indicated by the dashed association line from the TRN to the TBU. This association is not expected based on the standard application of the UAC. 119 Even the existence of a pre-linked TBU does not block application of the UAC which would link the Low tone to the right. In this case, the tone docks preferentially to the stem in a leftward direction; therefore, the UAC must not be strictly followed in Soyaltepec Mazatec. The final syllable of the stem is the stressed syllable. It is my hypothesis that when surrounded by two linked tones word medially, a floating tone attaches preferentially to the stressed syllable rather than to the suffix which is not stressed. 120 This pressure to link to the stressed syllable unexpectedly forms a contour on the first syllable of the newly formed word. Finally, the OCP is violated two times by the linkages in 8c: there are adjacent L melodies on the first syllable and there are adjacent H melodies on the second syllable. The merger of these tonal melodies is demonstrated in 8e. The final result is a two syllable word with a falling M 2 -L contour on the first syllable and a rising M 1 -H contour on the final syllable. The presence of a low tone preceding the suffix demonstrates tonal stability in the presence of segmental deletion. It also gives an additional argument for the necessity of linking at least some tones in the underlying representation. Specifically, the M 1 must be linked to the suffix. If the M 1 is not linked in the UR and if there was a strict adherence to the left to right directionality of the UAC, then we would expect 119 It might be hypothesized based just on this example that two tones per TBU is superior to three as would happen if the tones associated according to the UAC leaving any unassociated tones to form a contour on the final syllable, therefore forcing the association of two tones on each syllable. The problem with this supposition is that the realities of the language do not require or encourage such a restriction, the L will dock to the left even if a rising tone already exists on the initial morpheme as will be shown in §5.2.4 and §5.3.2. 120 Alternatively, the leftward linkage could be lexically specified; however, this is intellectually unsatisfying because it seems arbitrary. 245 that the Low tone would dock to the suffix rather than docking to the preceding syllable of the stem. Therefore, linkages exist in the UR and the UAC is not strictly obeyed.

5.2.3 One-to-Many