A Syntactic Representation of the Sequence-of-Tense Phenomena
Having shown how the sequence-of-tense phenomena can be accommodated under the neo-Reichenbachian approach, I now consider how they should be represented syntactically. Essentially following Hornstein (1990) and Thompson (1996), I assume that the three times (i.e., the Event, Reference, and Speech times) are syntactically represented as follows: the Event time is represented by V; the Reference time by Aspect (Asp), which is the head of the Aspect Phrase (AspP) that resides between TP and vP; and the Speech time by a [+tense] T, as shown in (1).
(1)
Hornstein claims that tense points can interact only under government, defined as follows:
(2)
On the assumption that T (INFL in Hornstein’s analysis) is the head of CP (S0 in his analysis), the embedding V governs the embedded CP and its head, namely, T., the embedding E point (E1), which is provided by the embedding verb heard, can interact with the embedded S point (S2), which is provided by the embedded T, as represented by (3).
(3) John heard that Mary was pregnant
(4)
This allows the sequence-of-tense rule to link the embedded S point (S2) with the embedding E point (E1), which yields the derived tense structure in.
Hornstein’s analysis, however, cannot account for the sequence-of tense phenomena with infinitive/subjunctive clauses.
(5) John hopes to be in the correct room
(6)
Recall that the S point is lacking in the temporal structure of an infinitive complement clause, and thus the sequence-of-tense rule links the embed ding E point (E1), which is provided by the embedding verb hope, with the embedded R point (R2), which is provided by the embedded Asp. In (6), however, the embedding V does not govern the embedded Asp, since CP, which is assumed to be the maximal projection of T under Horn stein’s analysis, dominates the embedded Asp but not the embedding V. Hence, Hornstein’s analysis would wrongly predict that the sequence-of tense rule cannot apply to an infinitive/subjunctive complement clause.
Note also that Hornstein’s analysis crucially relies on the notion of government, which is no longer available in the Minimalist Program (MP).
I depart from Hornstein’s analysis, arguing that tense points can inter act only under a local structural relation, that is, a specifier-head or head complement relation. I assume the widespread view that the head of CP is C. I also claim that T, which hosts an S point, covertly raises to C. Let us first look at how this analysis accounts for the sequence-of-tense phenomena with finite clauses.
(7)
As shown schematically in (7), the embedded [þtensed] T, which pro vides the embedded S point, covertly raises to the embedded C.4 The matrix V (more precisely, the trace/copy left by V-raising to v), which provides the matrix E point, enters into a head-complement relation with the embedded CP, to whose head the embedded T is raised. Hence, the two tense points, the matrix E and embedded S points, can interact; the sequence-of-tense rule can link these two points.
Let us now turn to the sequence-of-tense phenomena found within finitive/subjunctive complement clauses. On the basis of insights by Burzio (1986) and Boeckx (2001), I propose covert restructuring, whereby infinitive/subjunctive complements undergo covert clause-union, becoming part of the higher clauses covertly. More specifically, I argue that the embedded AspP, whose head Asp provides the embedded R point, covertly raises to the Spec of the matrix v to which the matrix V, which provides the matrix E point, raises, as shown schematically in (8).
(8)
In (8), the matrix subject (Subj1) originates in the Spec of v and then AspP raises to the outer Spec of v. The embedded AspP enters into a specifier-head relation with the v-V complex. The matrix E and embedded R points can interact under the specifier-head relation; the sequence-of tense rule can link these two points. Hence, the covert restructuring analysis can account for the sequence-of-tense phenomena found with infinitive/subjunctive clauses.
