Floating tone

Abstract: This paper provides robust empirical evidence for floating tones in Ga, a Kwa language of Ghana. As will be shown, floating tones are crucial to an analysis of verbal tense/aspect/mood distinctions. I begin by describing two tonal processes, the HL rule and Plateauing. While these are regular processes of the language, both are blocked in the perfective. I show that the blockage is the result of a floating low tone that marks the perfective, and that the floating tone marker explains other anomalous tonal effects in the perfective. I then give an analysis of floating tone prefixes that mark certain tenses/aspects/moods by associating to the subject prefix, thus overwriting the lexical tone of the subject prefix. Finally, I give examples of suffixed floating tones that mark tense/aspect/mood by associating to verb stems, causing the underlying stem tones to delink. In these tenses/aspects/moods, we find evidence for an underlying L vs. toneless contrast, constituting another phenomenon where, as with floating tones, there is a mismatch between the number of tones and tone-bearing units. Thus, a major prediction of Autosegmental Phonology (Goldsmith 1976, Clements and Ford 1979) is borne out in Ga.

Abstract: Complex phenomena of grammatical tone, well-described for many African languages, are increasingly attested also in the Tibeto-Burman family. This paper describes the tone assignment rule and two cases of tonal expression of grammatical categories in the Tibeto-Burman language Anal. The typologically unusual rule involves tone spreading, tonal polarity on a non-edge constituent and additional spreading, resulting in constant tonal patterns across grammatical suffixes. In two different cases the combination of the tonal pattern assigned by this rule with peculiar morpho-tonological processes results in a marking of a grammatical category (future and 1sg-person) by grammatical tone, by vowel-length, or only by the overall tonal pattern of the verbal form. Both cases are related to the omission of an explicit marking of the category, although the outcome cannot be explained only by the concept of a floating tone.

Abstract: This dissertation presents a phonological description and acoustic analysis of the word prosody of Ixpantepec Nieves Mixtec, which involves both a complex tone system and a default stress system. The analysis of Nieves Mixtec word prosody is complicated by a close association between morphological structure and prosodic structure, and by the interactions between word prosody and phonation type, which has both contrastive and non-contrastive roles in the phonology. I contextualize these systems within the phonology of Nieves Mixtec as a whole, within the literature on other Mixtec varieties, and within the literature on cross-linguistic prosodic typology.The literature on prosodic typology indicates that stress is necessarily defined abstractly, as structured prominence realized differently in each language. Descriptions of stress in other Mixtec varieties widely report default stress on the initial syllable of the canonical bimoraic root, though some descriptions suggest final stress or mobile stress. I first present phonological evidence—from distributional restrictions, phonological processes, and loanword adaptation—that Nieves Mixtec word prosody does involve a stress system, based on trochaic feet aligned to the root. I then present an acoustic study comparing stressed syllables to unstressed syllables, for ten potential acoustic correlates of stress. The results indicate that the acoustic correlates of stress in Nieves Mixtec include segmental duration, intensity and periodicity.Building on analyses of other Mixtec tone systems, I show that the distribution of tone and the tone processes in Nieves Mixtec support an analysis in which morae may bear H, M or L tone, where M tone is underlyingly unspecified, and each morpheme may sponsor a final +H or +L floating tone. Bimoraic roots thus host up to two linked tones and one floating tone, while monomoraic clitics host just one linked tone and one floating tone, and tonal morphemes are limited to a single floating tone. I then present three studies describing the acoustic realization of tone and comparing the realization of tone in different prosodic types. The findings of these studies include a strong directional asymmetry in tonal coarticulation, increased duration at the word or phrase boundary, phonation differences among the tone categories, and F0 differences between the glottalization categories.

Abstract: This paper focuses on the association of floating tones. One type we call "phonological association", where the tone-bearing unit (TBU) the floating tone associates to is determined by the phonological grammar, e.g. to a default position like a stressed syllable or domain edge. We contrast this with another type we call "targeted association", whereby the floating tone targets a numerical position , e.g. a floating H associates to the fourth TBU of the stem. Targeted association is idiosyncratic to the sponsoring morpheme, and does not necessarily target a default position. Such data have been taken as evidence for counting in grammar. Instead, we propose a novel representation we call "phantom structure", without counting. Phantom structure formalizes the observation that certain morphemes require structure to be present in order for their floating tone to be realized, but cannot provide this structure themselves. Under the targeted association type, the floating tone is pre-associated to a specific position in the phantom structure. We show that a simple grammar with two partially overlapping correspondence relations - substantive input and output structure (SubO-CORR) and phantom and output structure (PhO-CORR) - derives the association of the floating tone to the equivalent numerical position in the output as in the phantom plane input.