MIT PhD thesis 2001AbstractThis thesis provides a novel way of looking at verb agreement in signed languages by using an interaction of several processes within the Distributed Morphology framework. At the center of the model is a phonological re-adjustment rule, called 'alignment,' which handles various forms of agreement, including orientation change, path movement, relative position of the hands, and/ or a combination of these. Further evidence is taken from cross-linguistic data fromAmerican Sign Language, German Sign Language, Australian SignLanguage, and Japanese Sign Language, as well as from interaction with several other morphemes. It is shown that the output of the alignment process is filtered by various phonetic constraints and may be replaced by an alternative form that does not otherwise violate phonetic constraints.The model outlined above leads to a new typology of signs: first there are spatial verbs, followed by plain verbs which do not have two animate arguments, followed by aligning verbs which by definition have two animate arguments. These aligning verbs contain a subset of verbs that are in theory capable of undergoing alignment without violating phonetic constraints. This subset in turn contains another subset of verbs that are listed as actually undergoing alignment in a particular language.The model rests on the assumption that the referential use of space lies outside of the grammar. Removing the referential space from the grammar removes the modality difference between spoken and signed languages with respect to 'agreement.' The remaining differences will lie in how agreement is implemented. Both spoken and signed languages make use of different processes within the morphology component to generate the agreement system (e.g. impoverishment, vocabulary insertion, and phonological re-adjustment rules), but otherwise they draw on the same set of processes made available by the grammar.