Focus projection
One striking fact about the accusative unaccusatives is that the displacement of an internal argument does not disrupt focus projection, the process by which, under neutral intonation, the focus (new, or non-D-linked material) projects from the most embedded element to the whole VP or clause, the latter resulting in maximally wide focus (see Cinque 1993, Selkirk 1995, Reinhart 1995, Zubizarreta 1998). Because the wide-focus interpretation asserts the whole sentence as new it can be felicitously uttered in an out-of-the-blue context. In contrast, scrambling disrupts focus projection (Junghanns and Zybatow 1997:300–12; Kondrashova 1996:138–48).
This distinction between constructions with and without scrambling can be demonstrated using the Russian double-object construction. As Junghanns and Zybatow (1997) show, the interpretation of focus indicates that the basic order in double-object structures such as (1) is the one in which the DAT Goal precedes the ACC Theme (see also Kondrashova 1996:142–43).
(1)
When the ACC NP scrambles over the DAT NP in (1b), it not only has the effect of removing jablokoACC “apple” from the focus, but, more importantly, focus now can no longer project to the entire VP; the DAT NP, as a result, is narrowly focused. In (1a), there is no discourse-oriented scrambling, with the result that focus can take scope over the entire sentence. The focus projection facts for (1a–b) are schematized in (2a–b), respectively:
(2)
Although the SVOO order in (1a) exhibits a well-known focal ambiguity (which in (27a) is shown to be potentially four-ways ambiguous), VO(O) without an S fails to exhibit the same ambiguity as demonstrated for accusative unaccusatives in (3–4):
(3)
(4)
In these examples, the V-initial order, which is reported as awkward in discourse-neutral speech, constitutes a non-basic, scrambled structure. The examples in (3–4) either bear a particular focus structure (narrow focus on the right edge, with all other elements interpreted as part of the presupposed segment of the clause) or are interpreted as narrative-inversion structures, familiar from the Germanic literature, with a “folksy” “story-initial” reading. Thus, the VO(O) order for Russian and Ukrainian accusative unaccusatives has two possible interpretations: (i) narrow focus on the indirect internal NPOBL argument (or PP adjunct); or (ii) narrative inversion. By analogy with the focus projection pattern in double-object constructions, we are forced to assume that the VO(O) structure (on the non-narrative-inversion reading) involves discourse-oriented scrambling. It is precisely for this reason that the expected maximally-wide focus interpretation is not available.
Note that it is not only surprising that the VO(O) structure cannot be interpreted as maximally focused, but that the OVO structures given can be. That is, what we are demonstrating here is that only the OVO structures, as in (5–6), allow for focus projection (with the accompanying focal ambiguities):
(5)
(6)
Thus, (5), for example, felicitously responds to (i) čto slučilos’ “what happened?”; (ii) čto slučilos’ s soldatom “what happened to the soldier?”; and (iii) kak ranilo soldata “how was the soldier wounded?”. The possible focus domains for
(5) are given in (6):
(7)
Focus Projection for (5)
The central claim being made here is that the surprising focus projection facts in (7) follow from EPP-motivated movement of the direct internal argument. The failure to get maximally-wide focus in the VO(O) structure, then, results from a conflict between a formal syntactic requirement of the clause (T’s EPP) and an underlying argument structure that does not trivially satisfy this requirement via Merge. It follows that the basic-order requirement for focus projection is dependent on the more general requirement of syntactic well-formedness.
The non-disruption of focus projection, which provides evidence against a scrambling analysis for these predicates, indicates that a non-discourse-oriented relationship is established with Tdef, in terms of its EPP requirement. That is, we distinguish between semantically-driven A-movement (scrambling) and EPP-driven A movement. Only the latter exhibits focus projection. This is further supported by the fact that focus projection to the clause is not disrupted by movement to Spec-TP in ordinary passives, the prototypical case of A-movement.
