Ghost elements in Ende phonology
- Ghosts are phonological elements, whether consonants, vowels, features, or moras, that surface or delete in phonologically predictable contexts. This alternation with zero means that ghosts are sometimes visible and sometimes hidden. However, the quality or underlying distribution of these elements is idiosyncratic or unpredictable, which differentiates these elements from canonical epenthetic or syncopic elements. This dissertation provides an analysis for and an explanation of a critical behavioral property of ghost elements, namely the default realizational state of ghost elements in their underlying form (Zimmermann, 2019). The term default realizational state refers to the observation that some ghost elements surface by default, while others delete by default. Ghosts are lexically specified phonological elements, such as syllabic slots, melodies, or features, that may surface or not surface in order to avoid a phonologically marked form in the output. Ghost elements that are typically absent, but appear under markedness pressure, are called hero ghosts, while those that are usually present, but disappear under markedness pressure, are called martyr ghosts. Some phonological systems have ghost element patterns of both types, where one ghost element is typically absent, and another is typically present. This pattern is evident in the phonological systems of languages like Yowlumne (formerly Yawelmani; Zimmermann, 2019), a Yokutsan language of central California, and Ende, a Pahoturi River language of Papua New Guinea. In Yowlumne, an example of a martyr ghost is a ghost consonant pattern, in which some suffixal consonants are present by default but disappear if they would cause complex codas. An example of a hero ghost in Yowlumne is a ghost vowel pattern, in which certain suffixal vowels, such as the /i/ in the precative suffix, only surface to repair a complex coda, but otherwise fail to surface by default. These two ghost elements are not only distinguishable by their realizational behavior, but also by their subsegmental specification: Yowlumne ghost consonants are specified for their melodic features while Yowlumne ghost vowels are specified for their skeletal features. A very similar pattern occurs in Ende. In Ende, an example of a martyr ghost is the floating nasal pattern: lexically specified nasal segments that float through words to precede the leftmost non-initial obstruent but disappear if a such an obstruent is not available. An example of a hero ghost in Ende is infinitival theme vowels that appear in a class of infinitival verb forms and trigger reduplication, but only if the infinitival verb roots are monosyllabic. Again, the martyr ghost pattern involves ghost elements specified only for melodic features, and the hero ghost pattern involves ghost elements specified only for skeletal features. In Optimality Theory (Prince & Smolensky, 2004), ranked and violable constraints regulate the presence and absence of phonological elements in the output. If a constraint that penalizes non-realization of a phonological element is ranked higher than a constraint that penalizes realization, then the optimal output will include the element in question. This constraint ranking is necessary for a ghost element that exhibits martyr-type behavior. The flipped ranking of those two faithfulness constraints generates hero-type behavior. If ghost elements are represented uniformly as subsegments (cf. Zoll, 1996/1998), then this grammar would predict only two types of phonological systems: one in which all ghost elements are martyrs and one in which all ghost elements are heroes. This theoretical typology undergenerates the empirically observed typology of phonological patterns, as exhibited by Yowlumne and Ende. A representational distinction that splits ghost elements into two subsegmental types—those that are specified for their melodic features and those that are specified for their skeletal features—remedies this issue. This representational distinction engenders faithfulness constraints which can be ranked to indicate a language's preference to realize or not realize melodic or skeletal subsegments. This typology predicts four types of phonological systems, including Yowlumne and Ende. This work includes the first analyses for both of the Ende patterns and other languages with multiple ghost elements in an Autosegmental Phonology (Goldsmith, 1976) and an Optimality Theory framework. Besides illuminating some critical behavioral characteristics of ghost elements, these analyses also provide new data to exemplify rare linguistic phenomena. For example, chapter 3 takes a closer look at morpheme-level phonemic features like floating nasalization (also found in Máíhɨ̃̀kì (Sylak-Glassman, Farmer, & Michael, 2014)), and chapter 4 utilizes insights from the dual theory of reduplication (Inkelas, 2008) to provide another example of partial and total phonological duplication. This duplication is accounted for in an Agreement-by-Correspondence theory framework (Hansson, 2001; Rose & Walker, 2004). This work also provides the first descriptive analyses of the phonotactics, phonology, and morphology of Ende, introduces the basic typological profile of the language and the language family, and provides some basic cultural and anthropological information, including information on family structures, agriculture and subsistence, and a history of the community and the Ende language project.
|Type of resource
|electronic resource; remote; computer; online resource
|1 online resource.
|Lindsey, Kate Lynn
|Evans, Nicholas, 1956-
|Evans, Nicholas, 1956-
|Degree committee member
|Stanford University, Department of Linguistics.
|Statement of responsibility
|Kate L. Lindsey.
|Submitted to the Department of Linguistics.
|Thesis Ph.D. Stanford University 2019.
- © 2019 by Kate Lynn Lindsey
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...