Raising, hyperraising, and thematic structure
In this talk I present a number of puzzling asymmetries in the distribution of A-movement (particularly across clauses) and revisit some old ones, arguing that many of the licensing conditions on A-movement boil down to a restriction on moving between theta positions (such as the classic Theta Criterion).
In previous work, it’s been noted that there is a connection between evidentiality/perception and hyperraising (i.e. A movement out of a finite clause) wherein predicates of direct perception block hyperraising while predicates of indirect evidence allow it (Lee & Yip 2023). I show that such a contrast holds even in ordinary nonfinite raising contexts in English, as in (1).
(1a) *This wine tastes to be sour. (Direct) (1b) The alcohol by volume for this wine tastes to be about 12%. (Indirect)
Rather than positing additional syntactic machinery to account for this raising asymmetry, I attribute the contrast between (1a) and (1b) to argument structure/thematic structural differences. The basic idea I defend is that sources of direct evidence correspond to thematic roles while indirect evidence does not, constraining A-movement to a landing site that has a direct evidence/perceptual source theta-role to assign.
Beyond this, I show that as long the theta criterion is obeyed, even languages like English allow raising out of finite clauses, going against longstanding consensus regarding the typology of hyperraising.
Finally, I revisit a number of well-motivated analyses from previous work on A-movement that do away with the theta criterion and attempt to reconcile them with the present study. We see that virtually all cases of theta-criterion violating derivations proposes for different phenomena (e.g. verbal reflexives, complex event ntransitives, control, Tamil -koL- raising [Sundaresan 2013]) merely associate an argument with multiple theta roles within a single (super)event. This leads me to revisit the theta criterion and consider refining it to be sensitive to eventhood.