This paper describes a process of laryngeal reduction in San Martín Peras Mixtec (SMPM; ISO: jmx), an Otomanguean language spoken in Oaxaca and by diasporic communities throughout Mexico and the US. In this process, roots containing a laryngealized vowel often appear in a highly reduced form in fast speech. Laryngeal reduction is gradient, dependent on speech rate, and lacks a phonologically-defined conditioning environment, giving it the characteristics of a phonetic process. However, it is at least sometimes correlated with a phonological process of mora deletion, as evidenced by the fact that some highly reduced laryngealized roots—but no unreduced laryngealized roots—undergo a phonological tone sandhi alternation that applies only to mono-moraic rising tones. The phonological process of mora deletion is argued to be conditioned by the same phonetic factors that drive laryngeal reduction, constituting an instance of a phonological process triggered by purportedly phonetic factors.

Why do certain morphemes spread their tones, while other morphemes do not? We address this fundamental question in Kalabari (Kalaḅarị-Ịjọ), where certain clitics trigger a process of ‘low tone spread’ targeting following high tones (e.g. à ‘I’ in /à páḅụrụ tẹ↓ẹ/ → [à pàḅụrụ tẹ↓ẹ] ‘I have stammered’). We provide a comprehensive description of this process, establishing that its only triggers are a small class of prosodically-deficient pronominal clitics, all of which are low-toned, monosyllabic, and onsetless. We claim that these properties together prevent it from being parsed as a separate phonological word, and instead, the low tone of these clitics must tonally incorporate into a neighboring prosodic domain. We argue that the domain for low tone spread is the phonological phrase and show independent evidence for this exact constituent from grammatical tone. Finally, low tone spread is unbounded and targets a contiguous string of high tones within the relevant domain. We attribute its unboundedness to a consequence of tonal incorporation: this creates new LHH sequences which are independently marked in the language and consequently repaired by low tone spread. In total, our study demonstrates that tone spreading can profitably be decomposed into several sub-operations triggered by multiple interacting factors (here, word-minimality, prosodic constituency, and *LHH tonotactics).

In this paper, I provide a new type of evidence for sub-tonal features (Yip 1980, 2002; Bao 1999) from the Eastern Sudanic language Gaahmg (Stirtz 2011): Gestalt contour formation where specific morphological categories change the tone of a base word to a falling contour, but with different absolute tone values (High-Mid, Mid-Low, and High-Low) depending on the input tone. I show that the three different Gestalt contours in Gaahmg can be captured succinctly via feature affixation using Register Tier Theory (Snider 1990, 1998, 1999), and that this analysis receives independent support by other general patterns in the morphophonology of the language. Thus, following McPherson (2017) and Meyase (2021), the paper undermines the major objection against tonal feature geometries that they lack broad support in language-specific tonal grammars (Hyman 2010; Clements, Michaud & Patin 2011). By developing a virtually complete tone-affixation analysis of Gaahmg's tonal morphology in Autosegmental Colored Containment Theory (Trommer 2015; Zaleska 2020; Paschen 2021), the paper also provides evidence for the viability of this formalism in the context of the Generalized Nonlinear Affixation (GNA) program to reduce all productive non-concatenative morphology to affixation of partial phonological representations (Bermúdez-Otero 2012).