Hi Bokeh,
Becasue of your reply, it made me take a second look at the original question. I also did some research and found this paper written on the subject of Semantic Composition of Subjunctive Conditionals by Michela Ippolito of MIT/Tübingen University. I am not completely sure if I understood all thwt he wrote, but I do agree whole-heartedly with his view from what I understood. It's obvious that there are several subjunctive moods and conditionals discussed in great legnth which was exactly the reason causing the confusions on this thread. I find it absoulutely useful so I've extracted a small portion which I beleive was related the posted question.
http://www.google.com/search?hl=en&q=type+of+subjunctive
michela@alum.mit.edu
3. The Semantic Analysis of Subjunctive Conditionals
In answering the question of what the correct semantic analysis of subjunctive
conditionals is we will raise and answer the following questions too: (1) What is the
difference between indicative an subjunctive conditionals? (2) What is the role of the
past morphology in the composition of the meaning of a subjunctive conditional? (3)
What is the contribution of the second layer of past to the meaning of subjunctive
conditionals? As Iatridou observes, the past morphology in subjunctive conditionals is
not interpreted temporally, as the event of playing baseball in example (2) is supposed
to take place in the future (tomorrow). What follows in this paper is inspired by her
work and by the intuition behind it, i.e. that the temporal morphology we see in modal
constructions actively contributes to the construction of the modal meaning. However,
I depart from her idea that tense morphology has a “core meaning” that can apply to
different kinds of entities (i.e. her idea that if it applies to times, it is interpreted
temporally; if it applies to worlds, then it is interpreted modally). My claim is that
tense (aspectual) morphology has a single, definite interpretation: the temporal
(aspectual) one. The way tense morphology contributes to the composition of modal
meaning is by being interpreted in different positions in the structure of a modal
sentence, i.e. either in the restriction or in the nuclear scope of the modal operator.
Recall that I am arguing that accessibility relations are of type <s<i<st>> (where i is
the type for times and s the type for worlds): the notion of accessible world is relative
not only to a world but also to a time so that a world will be accessible if it satisfies
certain conditions with respect to an evaluation world and an evaluation time. The
past that we see in subjunctive conditionals such as If Charlie played baseball
tomorrow, we would lose the game is the morphological realization of a perfect
operator interpreted in the modal domain. I will develop an analysis of the meaning of
subjunctive conditionals and show how it solves the puzzle of the presupposition
projection for subjunctive conditionals discussed in Heim 1992; finally, I will answer
the three questions I raised above.
3.1 Felicity Conditions for Conditionals
Recall what the puzzle was. The antecedent of a subjunctive conditional can be
inconsistent with the common ground, and consequently, the set of worlds the modal
operator quantifies over cannot be restricted to the worlds in the context set (the
epistemically accessible worlds) (see (19) below). Furthermore, this set cannot be the
empty context (W) either because, if it were, we would expect conditionals with
antecedents with presuppositions to be infelicitous since the modal base does not have
the right entailments. However, this is incorrect: subjunctive conditionals whose
antecedents have presuppositions are felicitous, which means that the antecedent’s
presuppositions can be entailed by the modal base (cf. (20)). In fact, they must (cf.
(21)).
(19) Jack is dead. If he were alive, he would come to the ceremony.
11
(20) Jack smokes. If he quit smoking tomorrow, which he won’t, he would run
the marathon.
(21) Jack quit smoking last year. If he quit smoking tomorrow, he would run
the marathon.
Heim 1992 concluded that the only way to reconcile these two requirements of
subjunctive conditionals is to stipulate that the modal base is neither the set of
epistemically accessible worlds (the main context) nor the totally empty modal base
W, but the (largest) set of worlds obtained by suspending all the speaker’s
assumptions except the presuppositions of the antecedent, which then remain entailed.
However, I showed above that this stipulation does not work for all subjunctive
conditionals: in particular, it does not account for the difference between one-past
subjunctive conditionals and mismatched two-pasts subjunctive conditionals, as
shown below.
(22) Jack died last year.
a. #If he came to the ceremony tomorrow, he would be proud of Sally.
b. If he had come to the graduation tomorrow, he would have been proud
of Sally.
We are back where we were: how is the set of worlds to which modal operators apply
selected? Clearly, the felicity conditions for indicative, one-past and two-pasts
subjunctive conditionals are all different. But what is the difference and how is the
difference determined?
It seems correct to hold that for a sentence to be felicitously uttered in the
context c, c must entail the presuppositions of . In the common ground theory of
presuppositions developed by Stalnaker (1973, 1974, 1975), the common ground is
the set of all the propositions known or assumed to be true by all the participants in
the conversation, and the context set is the set of worlds where all the propositions in
the common ground are true. Assertions are meant to update the common ground. If
the assertion is made and accepted, the common ground expands and the context set
shrinks. Thus, if a sentence presupposes p, then asserting requires that the
common ground entail p, i.e. it requires that the speaker assume that it is true in the
common ground that p, modulo accommodation.10 It is explicit in Heim’s context
change semantics (and implicit in Stalnaker’s idea of a derived context) that a clause
(that is to say, the structural description of a clause at the level of Logical Form) is not
always evaluated with respect to the context of utterance: the context with respect to
which a structure is evaluated depends on the level of embedding of the clause, the
most unembedded clause being interpreted with respect to the main (utterance)
context. We can then reformulate the principle above: what is responsible for the
felicity of a sentence is not whether its presuppositions are entailed by the utterance
context but whether they are entailed by the evaluation context (which may be
identical to the utterance time in some cases). Call this principle PREP.
10 Stalnaker (1972, 1973, 1974, 1978, 1988, 1998). Kartunnen (1974), Lewis (1979), Heim (1982,
1983, 1992), Thomason (1990) and von Fintel (2000) also contributed important work in the tradition
of the common ground theory of presuppositions.
3.2 What Looks Like Past is Perfect
I propose that the past morphology we see in subjunctive conditionals in English is
the morphological realization of a perfect operator. The English perfect, and
especially the present perfect, has raised a lot of interest in the linguistic literature
because of the properties that distinguish it from both the present and the simple past
tense. McCoard (1978) offers a survey of possible theories of the perfect: the current
relevance theory, the indefinite past theory, the embedded past theory and, finally, the
theory that he argues to be the best, the Extended Now theory. Very briefly, according
11 The claim that the presuppositions of the antecedent of a conditional have to be entailed by the
context is a standard claim of a dynamic approach to meaning (Heim 1992). However, we will see later
that the issue is more intricate and I will have more to say on this topic later on in the paper.