This dissertation examines evidence that a group of "Central" dialects of the Austronesian language Malagasy, spoken on the island of Madagascar, is currently developing phonological tone contrasts in place of a former consonant voicing contrast. Analysis of acoustic phonetic data from speakers of a wide range of Malagasy dialects reveals that in contrast to Non-central dialect speakers, Central speakers have nearly neutralized modal voicing contrasts in all oral obstruent series, particularly in stressed syllables, and have developed consistent pitch contrasts, in both stressed and unstressed syllables, of a magnitude and persistence comparable to those of other tone languages. Responses to identification and discrimination tasks indicate that perception of pitch is quasi-categorical for the Central group in syllables with oral obstruents, with pitch typically overriding the modal voicing cue. For the Non-central group, pitch is a secondary, non-categorical cue which is subordinate to modal voicing. Age-stratification is observed in the Central group in both production and perception, indicating that this is a recent development. Prenasalized obstruents are distinguished primarily by voicing and/or duration of nasalization for both dialect groups; however, evidence suggests incipient tonal contrasts in the Central dialects among these sounds.
Comparison of individual production and perception patterns supports the claim that perception leads production at change onset and lags production as change moves to completion. In contrast to what is seen in some other languages (e.g., Hyslop 2009), data from different consonant classes suggest that the shift to tone occurred almost simultaneously across all oral obstruent types. The data show that sound change is phonetically gradual as opposed to phonologically abrupt, and evidence of cue covariation in production and perception support Beddor (2009)'s model of a coarticulatory path to sound change. Weak support is found for Thurgood (2007)'s claim that the direct connection between voicing contrasts and tone is via the influence of laryngeal settings on phonation type. The combination of sociohistorical facts and contemporary language attitudes also suggests hypotheses concerning social factors which may be driving this change. In general, evidence of tonogenesis in progress in these dialects reveals a unique scenario of ongoing change which is ripe for further study.