Evidence is presented that the Sentential Subject Constraint presented by Ross in "Constraints on Variables in Syntax" is a global constraint which makes reference to surface structure and to earlier stages of a derivation. (SC)

Kuno's direct discourse analysis is examined and rejected, and the Prague School concepts of theme and rheme are shown to be relevant to Kuno's data and additional data. It is further shown that an incompatible application of two or more transformations produces sentences that tend to be bad. (SC)

This article hypothesizes that causative constructions among the languages of the world share some fundamental characteristics, even though they have different ordering restrictions and varying degrees of fusion. (Author/NCR)