Taking as her point of departure the recurrent use in the languages of
the world of "past tense" morphology to express counterfactuality,
Iatridou (2000) proposes that a unified account of temporal and modal
uses of such morphology can be constructed in terms of a relation of
set-theoretic exclusion between a topic time or world and a speaker
time or world. On such an account, the relevant morphology would
realize a feature specification [+Excl(usion)] whose meaning in the
temporal domain is that topic time excludes speaker time and whose
meaning in the modal domain is that the topic world(s) exclude the
speaker world(s).
In this talk, I argue that while Iatridou's attempt to provide a
unified account of temporal and modal uses of "exclusion morphology"
along these lines is unsuccessful, such an account is in fact
attainable. The major obstacle, not confronted by Iatridou, is the
overlapping distribution of the representatives of [+Excl] and [-Excl]
in the modal domain, namely subjunctive and indicative conditionals:
schematically speaking, subjunctive conditionals occur in environments
a and b and indicative conditionals in environments b and c, thus
threatening to derail the project of accounting for the distinction
between the two in terms of a single binary feature. I propose to
deal with this problem by specifying subjunctive and indicative
morphology, respectively, only for the distinctive environments a and
c and allowing both to appear in environment b by default. This
proposal is implemented at the semantic end by a definition of [+Excl]
that is satisfied in the environment a that allows only subjunctives,
unsatisfied in the environment c that allows only indicatives, and
undefined in the environment b that allows both. It is implemented at
the morphological end, within the framework of Distributed Morphology,
in terms of a version of Halle's (1997) Subset Principle. The first
half of the talk concludes with a brief consideration of the temporal
domain, in which I sharpen Iatridou's claim that the temporal meaning
of [+Excl] involves the relation between topic time and speaker time
by showing that the relation of topic time to eventuality time, which
Iatridou does not consider in detail, is expressed by a separate
feature.
As a result of the overlapping distributional pattern referred to,
subjunctive conditionals are in the general case ambiguous between
counterfactual (environment a) and noncounterfactual (environment b)
readings. The second half of the talk examines the factors that lead
to disambiguation in favor of one or the other interpretation, a
question that Iatridou treats only in an aprioristic fashion. It then
uses the results of this investigation to clarify the interpretation
of subjunctive conditionals relating to the future, among them a group
that Iatridou labels "future less vivid" and claims to be
counterfactual. I conclude that, in fact, future less vivids,
properly defined, typically lack a counterfactual reading, and show
that the presence or absence of that reading follows from a lexical
property of the predicates involved.
References
Halle, Morris. 1997. Distributed morphology: Impoverishment and fission.
In MIT working papers in linguistics 30, 425-449. Department of
Linguistics and Philosophy, MIT, Cambridge, Mass.
Iatridou, Sabine. 2000. The grammatical ingredients of
counterfactuality. Linguistic Inquiry 31:231-270.
