Johns, Christopher Stephen Rowland (2007) Interpreting Agreement. Doctoral thesis, Durham University.
This dissertation is an investigation of the formal status of agreement morphology and its relationship to null arguments. Rejecting Chomsky’s (2000, 2001) claim that valuedness is the formal correlate of interpretability, it is suggested that a more elegant account of the data follows if φ-features of verbal affixes are taken to be valued in the lexicon. Interpretability is argued to be determined in the syntax, according to whether the affix is in a position to receive a Ө-role or not and as such the proposal diverges from that of Pesetsky and Torrego (2005a, 2005b), who also argue for the separation of interpretability and valuedness, but consider both to be lexical properties of a given category. Positing the existence of verbal affixes with valued interpretable φ-features is uncontroversial and this is the analysis adopted for environments such as the participial clause in Finnish, object and complementiser agreement in Modem Standard Arabic and a range of contexts in Modem Irish in which 'pronominal agreement' and full NP or free pronominal arguments are in strictly complementary distribution. It is shown that valued uninterruptible features, while not present in the lexicon according to the Chomskyan model, are necessarily created in the course of the derivation, leading to the conclusion that the computational component must contain an operation for deleting such features under identity with matching identical features of a nominal category. Since this should be no less able to apply to lexically valued uninterpretable features, there is no principled reason to suppose that such features do not exist and this is the analysis adopted for constructions such as certain Finnish adjunct clauses and SVO structures in Modem Standard Arabic, in which agreement and overt arguments co-occur. An important consequence of allowing the interpretability of φ-features to be determined in the syntax is the possibility that the same affix could be interpretable in one environment and uninterruptible in another, exhibiting different syntactic behaviour accordingly, and this is argued to be the case for Finnish possessor agreement and finite verb agreement in Modem Standard Arabic, obviating the need to posit a lexical split as other analyses (e.g. Fassi Fehri, 1993, Toivonen, 2000) have had to. Optional arguments are accounted for by recourse to the idea of a null pronominal with interpretable unvalued-features which probes an agreement head with uninterpretable valued φ-features, a natural consequence of the dissolution of the biconditional relationship between interpretability and valuedness and a direct analogue in Minimalist terms of the category pro of Chomsky (1982) and Rizzi (1986).
|Item Type:||Thesis (Doctoral)|
|Award:||Doctor of Philosophy|
|Copyright:||Copyright of this thesis is held by the author|
|Deposited On:||09 Sep 2011 09:57|