Case Study: Interrogative Wh-in-Situ in Coptic
In Coptic Egyptian, the final developmental stage of the Ancient Egyptian language, most constituent questions involve wh-in-situ. Both wh arguments and wh-adjuncts can occur in situ, as shown in (1). (The canonical word order is SVO. For details of the textual sources of the examples, see Reintges 2004, 597–600.)
(1)
Wh-in-situ has a broad syntactic distribution and occurs in matrix and embedded clauses. Main and embedded wh-in-situ differ, however, with respect to the scope of the wh-phrase. In main clauses, the wh-phrase takes matrix scope and the clause is interpreted as a direct question, as seen in (2a). If the wh-phrase occurs in situ in a finite embedded clause, it must take embedded scope and only the indirect question reading is available, as seen in (2b).
(2)
It is also possible, though not common, for the interrogative phrase to surface at the left edge of the clause, in a position that, following Green and Reintges (2004) and Reintges and Green (2004), we analyze as the specifier of a Focus head located below C and above T. This displacement, we assume, is the result of wh-movement. Constituent questions with overt wh-movement are exemplified in (3a–c).
(3)
Because Coptic is no longer spoken, it is impossible to establish with certainty that constituent questions of types (1)–(3) are sensitive to islands, and so involve wh-movement. The best that can be said is that there are no attested examples of questions of either type in which the interrogative phrase originates inside an island and takes scope outside the island.
What interests us here is that wh-in-situ questions are flagged by relative tenses, so called because they are derived from ‘‘basic’’ (absolute) tenses by the addition of the relative markers e- and e nt- to the left edge of the verbal cluster. The relative prefix e nt- is selected in the context of the perfect marker a, while the allomorph e- is chosen in all other contexts. Since the morphological alternation of the relative marker is dependent on the following tense inflection (a null morpheme in the present tense), it provides prima facie evidence for the correlation between this special inflection and the tense/finiteness system. In general, relative tenses occur in the classic constructions formed by wh-movement, such as relative clauses and declarative focus sentences. In relative constructions (which in Coptic are externally headed), there is no subordinating complementizer besides the relative tense marker, as shown in (4a–b).
(4)
In the question-answer pair in (5), the focus-in-situ e-p-topos n-Apa MEna ‘to the shrine of Apa Mena’ in the answer corresponds to the wh in-situ locative argument t c n ‘where’ in the question.
(5)
As might be expected by this point, we claim that relative tense is a form of wh-agreement: it spells out the Agree relation holding between a probe of some type and a goal that bears the active feature [wh]. Wh-agreement takes the form of a relativizing complementizer that is merged in the left peripheral Focus head. To determine whether this wh-agreement is of flavor (i) or flavor (ii), we must ask how relative tense patterns in long distance constituent questions. Reintges’s (2002, 2003, 2004) work has uncovered no examples in which the wh-in-situ occurs within an embedded clause containing a finite verb, but takes scope over the matrix clause (see above, (2b)). However, there are attested examples in which a wh-in-situ occurs as the nonverbal predicate of an embedded clause, but nonetheless takes matrix scope. In these examples, relative tense marking surfaces in the left periphery of the matrix clause.
(6)
There are also examples in which the wh-in-situ occurs within an embedded infinitive, but nonetheless takes matrix scope.5 Once again, relative tense marking occurs in the highest clause over which the wh-in-situ takes scope.
(7)
This pattern argues that relative tense is wh-agreement of flavor (ii): it surfaces in the highest clause along the path of wh-movement, flagging only the highest copy—which, of course, is unpronounced. In other words, relative tense is entirely analogous to Chamorro ‘‘operator-C agreement,’’. Crucially, the parallel holds even though constituent questions in Chamorro require overt wh movement, whereas in the relevant constituent questions in Coptic the wh-phrase surfaces in situ.
The alternation in the form of the relative marker reveals that Coptic operator-C agreement interacts with the realization of tense and aspect in ways that could indicate movement of T to C, or some equivalent (see Pesetsky and Torrego 2001). We pass over this here (but see Reintges 2003). What is important is that we take relative tense in Coptic to transparently reflect the Agree relation in overt syntax, where this relation holds between a probe that bears the active features [X] and [Q] (presum ably because it has acquired these features from the closest C above it) and a goal that is a wh-phrase. The main difference between Coptic relative tense and Chamorro ‘‘operator-C agreement’’ lies in which copy of wh-movement is pronounced: in Chamorro, only the highest copy is pronounced, whereas in Coptic wh-in-situ constructions such as (1), (2), (6), and (7a–b), only the lowest copy is pronounced.
We can take the analysis further by recognizing that when interrogative phrases in Coptic undergo overt wh-movement, which raises them to the specifier of FocusP, relative tense marking does not occur. Consider (8) above as well as the long-distance constituent question in (8), which exhibit overt wh-movement but no relative tense marking.
(8)
How is the absence of relative tense in these constructions to be explained? An answer emerges within our approach, which maintains that constituent questions in Coptic always involve wh-movement in overt syntax. Suppose further that relative tense is spelled out only in the phonology, at roughly the point when it is determined which copy of wh-movement is pronounced. Then the two pronunciation issues would be interrelated. If only the lowest copy of the interrogative phrase is pronounced, then relative tense is spelled out overtly (see (9a)). But if only the highest copy of the interrogative phrase is pronounced, then the spell out of relative tense, which would occur within the same minimal projection, is blocked (see (9b)).
(9)
The situation is reminiscent of the Doubly Filled Comp Filter in English (Chomsky and Lasnik 1977), which allows the pronunciation of either a complementizer or the wh-phrase in its specifier, but not both.
