La Charité, Darlene.2009-04-172009-04-1719931993Source: Dissertation Abstracts International, Volume: 68-06, Section: A, page: 2430.http://hdl.handle.net/10393/10920http://dx.doi.org/10.20381/ruor-17073The conclusion of this thesis is that affricates are represented phonologically as stops. There is no phonological notion of affricatehood, nor is there any way of phonologically defining an obstruent class containing all possible affricates and fricatives, to the exclusion of stops. Three types of evidence support this hypothesis. The first consists of the patterns of distribution of affricates in the segment inventories of over two hundred languages. In almost half the inventories considered, affricates and stops are in complementary distribution with respect to place of articulation. This is unaccounted for by any theory which assumes that affricates comprise a distinct obstruent class. In nearly all of those inventories where both an affricate and a stop occur at a single place of articulation, overlap occurs in the dental/alveolar region. However, these cases do not require the assumption of a distinct affricate class because the affricate can always be distinguished from the homorganic stop by an independently required feature. Typically, this feature is [+strident]. Dental affricates [ts], [dz] and alveolar affricates [ts], [dz] are strident stops. It is further argued that the feature [strident] should be interpreted as sibilant, which defines the following set of segments: [s, z, ts, dz, ts, dz, s, z, c, jˇ]. The second and third types of evidence not only justify the proposed affricate representation but, in so doing, validate the idea that the distribution of segments in inventories provides a valuable source of information about the featural makeup of those segments. The second body of evidence consists of analyses of rules that have been used to argue that affricates pattern with fricatives. These rules and constraints (from Modern and Classical Yucatec Maya, English, Chumash, and Basque) demonstrate that only the sibilant fricatives and affricates pattern together. Notably absent are rules that involve a broader range of affricates and fricatives. The rules and constraints that show affricates patterning with fricatives are profitably reanalyzed as targeting segments that are [+strident]. The third type of evidence supporting the view that affricates are stops focuses on analyses of several morphological processes in southern Bantu languages, specifically Shona and Setswana, that derive affricates from a variety of different segments, including both stops and fricatives. These processes have often been considered suppletive. However, the assumptions that affricates are stops, and that the alveolar affricates are strident stops, provide the tools for an elegant and insightful analysis of the consonant mutations involved. Despite apparent lack of relationship between the input and output of consonant mutation, the two can be shown to be featurally identical, or very similar. Individual feature differences between the input and output of consonant mutation are explained by independently attested feature co-occurrence restrictions of the language. These co-occurrence restrictions, which may be violated in causative, passive, and diminutive formation, trigger repair strategies whose goal is to produce feature combinations that reflect those occurring in the underlying representation.297 p.Language, Linguistics.Thesis