Topics in phase theory: taking things apart and putting them back together again

Thomas McFadden

Phase theory is based on the idea that syntactic derivations involve a series of chunks of structure, called phases, that are (at least partially) opaque to each other. This is used among other things to explain locality effects. E.g., a noun phrase can move from certain non-finite embedded clauses up to the subject position of a verb like 'seem' as in (1), but the same movement is blocked if the embedded clause is finite, as in (2). Within phase theory, we can account for this by proposing that finite clauses are phases, while (at least some) non-finite clauses are not.

(1) Rory seems to be drunk.
(2) *Rory seems [that is drunk].
(3) It seems [that Rory is drunk].

One way to think about phases (essentially the mainstream one proposed by Chomsky) is that the derivation builds up structures and removes (transfers/spells out) phases (or parts of them) as they are completed. So phase impenetrability comes out of the fact that a given phase will be shipped off to the interfaces before the relevant material in the next phase even enters into the structure. A second way to think about phases is that they are constructed independent of each other, in their entirety, and then stitched together in the end to create the complete structure. Phase impenetrability on this view comes from the fact that the operations and relations that are sensitive to locality apply before the phases are stitched together with each other.

This course will explore the latter conception of derivation by phase. We will see how it compares with the standard one in its ability to handle a series of empirical phenomena -- e.g. successive-cyclic movement and other processes/configurations that cross phase boundaries and require us to build in 'escape hatches'. We will examine how the two approaches might be used to approach the question of how phases are defined and what bits of structure should (and should not) count as phasal.