E-mail this message to a friend

Dear all, 
 
Consider sentences such as 
 
(1) John thought that Harry will leave, 
 
(2) John forgot that Mary will come, 
 
(3) Mary believed that Harry will be late, 
 
or, generally, sentences whose matrix clause has a (stative) verb in simple 
past tense, followed by an embedded clause whose verb is in simple future 
tense _and_ interpreted deictically. (If English grammar as a matter of 
fact forbids sentences of this general form, which I'm not in a position to 
decide, I would be grateful to learn this, and furthermore to know if in 
some other language the relevant pattern of tense interactions can however 
occur.) 
 
It appears to me that these types of sentences have, logically, two 
readings. I refer to them as the non-uniform reading and the uniform reading. 
 
Take (1). As thinking is something that extends over time, (1) states 
something about an extended interval of past time. According to (1), during 
such an extended period of past time, ''John thinks that Harry leaves after 
t_0'' was true, if t_0 is the time of utterance of (1). 
 
On the other hand, what it takes to witness a simple future statement like 
''Harry will leave'' is to present a future time at which ''Harry leaves'' 
is true. 
 
This leads to two readings: 
 
(non-uniform) For all relevant t_1 < t_0 there is t_2 > t_0 such that at 
t_2 Harry leaves (according to what John thought at t_1) 
 
(uniform) There is t_2 > t_0 such that for all relevant t_1 < t_0, we have: 
at t_2 Harry leaves (according to what John thought at t_1) 
 
The difference between the two readings is that in the former it suffices 
that t_2 be just any function of t_1, while in the latter t_2 is the same 
for all t_1. Now I would like to ask: 
 
(a) Are both readings really possible in English? If both are, probably the 
former would be the default reading. Is this so? If both are not possible, 
what is the general linguistic rationale that excludes the relevant reading? 
 
(b) If the present examples are ungrammatical, are there examples of 
interactions between temporal expressions (not necessarily tenses) having a 
natural uniform reading (other than the evident cases where explicit 
denoting expressions such as ''tomorrow at 11.15 AM'' are used)? That is, a 
reading where the interpretation of the grammatically subordinate temporal 
expression would be *logically* independent of the grammatically precedent 
temporal expression, in the sense that the expression in the embedded 
clause would have to receive the same interpretation (t_2) for all 
interpretations (t_1) of the expression in the matrix clause. 
 
Many thanks in advance. 
 
Kind regards, 
 Tero Tulenheimo