D-Linking/Specificity and a Subjacency-Type Account
Returning to the main subject, I now propose a minimalist account of the selective extractability of wh-phrases out of infinitive/subjunctive indirect questions based on the claim that specific DPs in the sense of Enc¸ (1991) move out of VP to the vP-edge position while nonspecific DPs do not, as advocated by, among others, Mahajan (1990), Diesing (1992), Runner (1994), and Torrego (1998). Essentially following Diesing (1992), I argue that specific (what Diesing calls ‘‘presuppositional’’) DPs are required to move out of VP to the vP-edge position before a derivation reaches LF, in accordance with the Mapping Hypothesis. One of Diesing’s arguments supporting this contention is based on the antecedent-contained deletion (ACD) construction (see, e.g., Bouton 1970; Sag 1976; May 1985; Larson and May 1990; Fiengo and May 1994; Pesetsky 2000).
Mary read every book that John will
The meaning of the elided VP complement of will cannot be under stood as the matrix VP read every book that John will, since this would lead to an infinite regress. If the DP every book that John will moves out of the VP covertly, then it leaves behind a VP of the form read t, which provides the appropriate antecedent for the elided VP, as shown in (2).
(2)
Hence, as argued by Pesetsky (2000), the ACD construction can be used as a diagnostic for phrasal movement: unless phrasal movement takes place, there is no way of licensing the ACD construction. Diesing observes that while specific DPs allow the ACD construction, as shown in (3), nonspecific DPs do not, as shown in (4).
(3)
(4)
The contrast between (3) and (4) supports the Mapping Hypothesis, which requires specific DPs, but not nonspecific DPs, to move out of VP covertly.
As argued by Cinque (1990), among others, D-linked wh-phrases also count as specific, since they are understood as presupposing the existence of a known set of specific elements. The D-linked wh-phrase which of the new books presupposes the existence of a known set of specific new books, and the question is interpreted as asking which of such new books a person has bought. Further evidence for this contention comes from ACD facts. As observed by Pesetsky (2000), exactly like specific DPs, D-linked wh-phrases in situ also license the ACD construction, which suggests that D-linked wh-phrases are themselves specific and thus required to move out of VP covertly in ac cordance with the Mapping Hypothesis.
(5)
(6)
During its derivation, we construct the embedded vP phase (7).
(7)
According to Diesing’s analysis, the D-linked wh-phrase which of the new books, being specific, is required to move out of the VP covertly in accor dance with the Mapping Hypothesis. Recall that we are assuming that a phase Ph1 is interpreted/evaluated at the next phase Ph2. Hence, this semantic/interpretive condition on specific DPs requires the D-linked/ specific wh-phrase which of the new books to move out of the VP into the vP-edge position covertly until the derivation reaches the next phase level, namely, the embedded CP phase (8).
(8)
Note in passing that in (8), the other wh-phrase when moves first to the vP-edge position and then to the Spec of the embedded C.
As the derivation proceeds, we construct the matrix vP phase; the relevant structure is represented in (9).
(9)
In (8), the embedded AspP undergoes covert restructuring. Recall that the D-linked/specific wh-phrase which of the new books resides in the embedded vP-edge (vP2-edge) position. From there, it raises to the matrix vP-edge (vP1-edge) position, as shown in (9).
Although this movement is not triggered by any formal feature at the matrix vP-phase level, its evaluation with respect to the economy condition takes places at the next phase, namely, the matrix CP phase. At the matrix CP-phase level, the Q-feature and EPP feature of the matrix C trigger movement of which of the new books to its Spec, as shown in (10).
(10)
Recall that the covertly moved embedded AspP is not pronounced in the head position of its chain (in the matrix clause); instead, it is pronounced in its trace position (in the embedded clause). Recall also that according to Pesetsky’s (2000) principle of pronunciation regarding wh-movement, the first instance of phrasal movement of a wh-phrase to the Spec of an interrogative C is overt in that the wh-phrase is pronounced in the head position of its chain and unpronounced in the other positions. In this derivation, the wh-phrases which of the new books and when are pronounced in Spec,C and unpronounced in the other positions. This derivation con verges; (6) is acceptable.
(11)
During its derivation, we construct the matrix vP (vP1) phase; the relevant structure is shown in (12).
(12)
In (12), the embedded AspP undergoes covert restructuring. Note that there is no way of raising the non-D-linked/nonspecific wh-phrase what to the embedded vP-edge (vP2-edge) position. Recall that under the proposed analysis of successive-cyclic movement, v may not be assigned an EPP feature or a P-feature; there are no formal features that trigger movement of what to the embedded vP-edge (vP2-edge) position. Furthermore, since what is non-D-linked/nonspecific, the Mapping Hypothesis does not require what to move out of VP to the vP-edge position. It should also be noted that what cannot undergo non-feature-driven movement to the embedded vP-edge (vP2-edge) position because of the economy condition. This is because what does not undergo feature-driven movement at the next phase (the matrix vP (vP1) phase), where movement in the embedded vP (vP2) phase is evaluated, since under the proposed analysis of successive-cyclic movement, v may not be assigned an EPP feature or a P-feature. The wh-phrase what stays within the embedded vP (vP2) phase and thus is not accessible to operations at the matrix CP because of the PIC. There is no way of satisfying the uninter pretable EPP feature and Q-feature of the matrix C. This derivation crashes; (11) is therefore unacceptable. wh-phrases cannot be extracted out of infinitive/subjunctive indirect questions, either. This fact can be explained on a par with the inability of non-D-linked wh phrases like who and what to be extracted out of infinitive/subjunctive indirect questions. This is because the notion of specificity is only relevant to DP denotations; thus, adjunct wh-phrases like why, how, and when, which are not DPs, do not count as specific.
