| This thesis is concerned with the need to allow a world formation rule to apply to the internal portion of an existing word. In order to resolve apparent paradoxes with respect to the ordering of reduplication and affixation in Tagalog (Carrier, 1979), we propose that WFRs employing reduplication can subcategorize for an affixed form but copy material that is internal to an affix. We call the material that is copied in reduplication the 'focus'. Unlike previous analyses which allow an internal constituent of a word to be copied in reduplication (Carrier, 1979; Aronoff, 1988), the focus of the reduplication is distinct from the subcategorized element of the WFR. By specifying that the internal element subject to copying is not the element that is subcategorized by a WFR, we distinguish focus from morphological head.;The use of focus creates forms in violation of an assumption that the linear order of morphemes reflects their order of attachment (locality). We suggest that locality may constrain affixation but does not govern process morphology. Locality, therefore, presents a problem for focus only if we define reduplication as a case of affixation. Analysis of data from Sanskrit (Steriade, 1988), Woleaian and Nasioi suggests that this definition is not descriptively adequate. We reject affixation of empty templates and adopt the process-oriented theory of parameters to explain reduplication (Steriade, 1988). The use of focus, therefore, does not violate locality.;Most instances of process morphology identify phonological targets; reduplicative foci generally correspond to morphological constituents. The morphological nature of reduplicative foci presents a curious anamoly. Specification of focus in a variety of Austronesian languages is examined to determine if it is different from other targets of process morphology. Investigation reveals that the use of morphological information to specify focus is not qualitatively different, and the objections to foci in reduplication are without basis. We conclude that the use of foci to resolve disparities between linear and structural order in reduplication is justified. |
