![]()
A study of ATR harmony patterns indicates that a serial derivation model best accounts for the surface forms of Karimojong verbs. Five affix types in verbal complexes in the language may undergo harmony, trigger it or remain neutral, resulting in three separate harmony domains: bidirectional root-controlled harmony,[-ATR] suffix-controlled harmony, and [+ATR] suffix-controlled harmony. ATR specification may also be determined by adjacency effects. The [-ATR] suffix-controlled process is proposed in part to emanate from the phonologization of tongue retraction required to pronounce the itive suffix [-Ar]. The continuation of consonant voicing produces a [+ATR] feature that will block even dominant [-ATR] processes. Ease of perception and a desire to conserve vowel height information are proposed to be responsible for an [ATR] dissimilation rule affecting stem-final mid and high back vowels. The presence of neutral vowels in active affixes is accounted for by a system of a headmarking and feature percolation (Van derHulst& Van derWeier, 1995; van derHulst& Dresher, 1998) within a hierarchical prosodic word (PW) structure that channels the ATR specification to recipient vowels. The feature specification of the dominant prosodic word node and the domains of consonant generated features explain the outcome of the intersection of harmonic domains. Surface forms are explained as derived in three morpho-phonological levels, identified by the application or absence of ATR harmony under a stratal optimality–theoretic model. The AGREE constraint family used in this analysis employs agreement with a dominant phonological head to define the extension of a harmony domain and is employed within a framework of positional faithfulness. To account for consonant-generated features impacting harmony domains, DOMAIN C is proposed. Among the strategies to preserve a parallel derivation model, sympathy theory is shown to be implausible at best, and nothing more than serial derivation disguised by formalism. Output-Output correspondence fails to correctly account for surface forms. Due to the irregularities generated by suffix-controlled [-ATR] domains, headed spans are not an option. Whether candidate chains (McCarthy, 2006) can be considered truly parallel derivation remains to be explored. Transitional phases of incorporation measured by their participation in harmony demonstrate diachronic evolution in the language. Pronominal prefixes are largely neutral, but those found in high-frequency narrative forms alternate under dominant suffix-controlled harmony processes and are indicative change in progress. Tense/Aspect/Mood (TAM) markers also show both neutral and phonologically active behaviors. Frequentive suffixes with differing behaviors provide evidence for the genesis and evolution of reduplicated derivational suffixes, a sub-process in the evolution in the agreement system, indicated by patterns of differentiation, uniformity, and loss in TAM marker paradigms. . |
Return to Colloquia Page || LSO Home || MSU Home
....