English Trace Table
Luis Vicente does an excellent job of summarizing the research history of V→T(→C) movement in this handout. However, not being familiar with some of the more recent theories, I will proceed from a point of view in which affixes may still lower. Similarly, I am not sure if this particular issue has been analyzed before. (If it has, then I present this analysis merely as an independent thought exercise.)
This table consists of twelve sentences, all describing a situation in which a cat is stolen by a man. The sentences are divided into three sections: statements, interrogatives for identifying the subject or the agent (the man), and interrogatives for identifying the object or the theme (the cat). Within each section, the upper two sentences are in the active voice and the lower two sentences are in the passive voice. Within each pair of sentences, the upper sentence has a finite embedded clause and the lower sentence has a non-finite embedded clause. The main clauses and the finite embedded clauses are in the present tense, and all of the embedded clauses are in the perfective aspect.
The sentences marked in green are the sentences that are completely regular and grammatical: they are predicted by the Extended Projection Principle, expletive insertion, V→T movement, T→C movement, DP movement, wh-movement, case- and feature-checking, and the Minimal Link Condition. (The VP-internal Subject Hypothesis is also implied.) The sentences marked in yellow are grammatical sentences that are odd because they do not undergo T→C movement when they might otherwise be expected to (i.e. when forming a question). The sentences marked in red are ungrammatical sentences.
The sentences in Part I demonstrate how the rules for statements are supposed to be used, and they will act as the basis for movement that happens in Parts II and III. There is nothing unusual about these sentences.
The sentences in Part II inquire about who the agent of the sentence is. Sentence II-1 has normal wh-movement and uses expletive do to support T→C movement. Sentence II-2 does not undergo T→C movement in the main clauses, which would normally be expected in an interrogative. Sentences II-3 and II-4 are ungrammatical because they posit wh-movement of an agent that does not exist: the agent theta role was absorbed by the passive morpheme -en. (Also, such a movement would require multiple occupants of SpecTP2, which is not allowed.)
The sentences in Part III inquire about what the theme of the sentence is. Sentences III-1, III-2, and III-3 have no problem with wh-movement and use expletive do to support T→C movement. Sentence III-4, while also undergoing wh-movement without a problem, does not undergo T→C movement.
Sentences (II-2 and III-4) are odd sentences because they do not undergo T→C movement. Both of these sentences have a trace in SpecTP1 (tk and tm, respectively), whereas the other sentences have overt content in that position. In particular, that trace is for the wh-word in SpecCP1.
Since all of the sentences that undergo T→C movement have overt content in SpecTP1, rather than a trace, it is reasonable to conclude that the presence of a trace in SpecTP1 prevents T→C movement—or, conversely, that the presence of overt content in SpecTP1 promotes T→C movement.
References and Further Reading
You may respond to this page by commenting on this blog post.