Two studies were conducted to test Francis Christensen's theory of generative rhetoric, which maintains that students should be taught the types of sentences that they infrequently use, especially ones with nonrestrictive phrasal modifiers (participial phrases and appositives) as a way to approach invention. The first study examined the principles of generative rhetoric as they applied to college writers and skilled adult writers. The second study assessed the effectiveness of generative rhetoric instruction in college writing classes. Findings of the first study showed that the frequency and placement of nonrestrictive modifiers seemed to be the major distinguishing syntactic features between late secondary school and early college prose and the prose of skilled writers. The second study showed that students who were taught generative rhetoric displayed greater clause length in their writing than students who were taught sentence combining. Together, these findings suggest that controlled syntactic practice is primarily responsible for the syntactic increases attributed to sentence combining at the college level, and that instruction in nonrestrictive modifiers is more important for college students than the activity of combining sentences. (Topics used in the two studies and tables of resulting data are appended.) (HOD)