Which of the following sentences is grammatical, and why does it contrast with the others?
A'Who do you wonder whether Mary met?' — extraction from an embedded interrogative is allowed in English
B'Who did you believe the claim that Mary met?' — extraction from a complex NP is permitted through complement clauses
C'Who did she say that Mary met?' — long-distance extraction through a complement clause is grammatical
D'Who did she leave after meeting?' — extraction from an adjunct is allowed when the adjunct is temporal
English allows long-distance extraction only through complement clauses — embedded clauses that are direct arguments of verbs like 'say,' 'think,' 'believe.' Option C is grammatical because 'that Mary met' is a complement clause of 'say.' Options A, B, and D involve island environments: 'wonder whether' creates a wh-island (the embedded clause is itself interrogative), 'the claim that' creates a complex NP island, and 'after meeting' is an adjunct island. All three block extraction, producing ungrammaticality under standard subjacency theory.
Question 2 Multiple Choice
Consider the contrast: 'Who do you think left?' (grammatical) vs. 'Who do you think that left?' (degraded in English). Which principle most directly explains the degraded status of the second sentence?
ASubjacency: moving 'who' crosses two bounding nodes in the second sentence but only one in the first
BThe that-trace effect: a subject trace adjacent to an overt complementizer 'that' violates the Empty Category Principle
CWh-islands: 'that' creates an interrogative island blocking subject extraction
DComplement clauses block subject extraction regardless of whether 'that' is present
The that-trace effect is the classic name for this phenomenon. When 'who' is extracted from the subject position of an embedded clause, leaving a trace immediately following 'that' (an overt complementizer), the result is degraded in English. This is attributed to the Empty Category Principle (ECP): the trace left by extraction of a subject must be properly governed, and an overt 'that' blocks the required government relationship. In 'Who do you think __ left?' the complementizer is absent, allowing proper government of the trace; in 'Who do you think that __ left?' the overt 'that' intervenes, creating the degradation.
Question 3 True / False
In English, a wh-element can be extracted from an arbitrarily deeply embedded complement clause — spanning any number of clause boundaries — without becoming ungrammatical.
TTrue
FFalse
Answer: True
This is the defining property of long-distance (unbounded) extraction through complement clauses. 'Who did she claim that he believed that Mary had seen?' spans three clause boundaries and remains grammatical because each movement step stops at the intermediate Spec,CP of each complement clause (successive-cyclic movement), crossing only one bounding node per step. The dependency is 'unbounded' in the sense that there is no fixed maximum depth, as long as each intermediate step respects subjacency.
Question 4 True / False
Island constraints on long-distance extraction are universal: most human languages block extraction from complex NP islands, wh-islands, and adjunct islands in the same way.
TTrue
FFalse
Answer: False
Crosslinguistic variation shows that island constraints are not uniformly universal. Many Scandinavian languages allow extraction from wh-islands that English disallows. Malagasy and other Austronesian languages impose much stronger restrictions on extraction. Some languages permit extraction from adjunct islands under certain conditions. This variation drives ongoing theoretical debate: do all languages share the same underlying constraints with surface differences driven by feature specifications ('strong' vs. 'weak' features), or are the constraints themselves parameterized? The crosslinguistic data suggests neither pure universality nor free variation.
Question 5 Short Answer
Why does 'Who did you say that Mary met?' succeed as long-distance extraction while 'Who did you read the claim that Mary met?' fails? What structural difference accounts for the contrast?
Think about your answer, then reveal below.
Model answer: In 'Who did you say that Mary met?', the gap is inside a complement clause — 'that Mary met' is a direct argument (complement) of 'say,' a transparent extraction domain. In 'Who did you read the claim that Mary met?', the gap is inside a noun complement clause embedded within the complex NP 'the claim that Mary met.' Complex NPs are islands: no element may be extracted from a relative clause or noun complement clause, because the NP constitutes a bounding node that cannot be crossed. The difference is not distance but the structural type of the embedded clause: complement clauses are transparent; complex NPs are barriers.
The complex NP island is one of the most robust extraction constraints across languages. The theoretical explanation varies by framework: in subjacency theory, crossing the NP boundary violates the constraint against crossing more than one bounding node per movement step. In more recent theories, the nominal structure's feature properties block movement through it. Practically, the test is whether the embedded clause is an argument of a verb (extractable) or embedded inside a nominal phrase (island). 'The claim that...' is the latter — the complement clause belongs to the noun 'claim,' not to the matrix verb 'read.'