Consider an utterance by me of (1), with “she” used to refer to some salient woman:
(1) She thinks that I was drinking a martini earlier.
According to many people, such an utterance of (1) imposes no special demands on how the referent of “she” needs to think of me in order for what is said in the utterance to be true: she might think of me as (in a way he would naturally express by using) “that man,” “the man in the corner smoking a pipe,” “Guy,” and so forth.
By contrast, consider an utterance by me of (2), in this case addressing the same woman as “you”:
(2) You think that I was drinking a martini earlier.
According to some people, (2) differs from (1) in that, although it allows an understanding on which, like (1), it will be true however the referent of “you” thinks of me, it also has a different reading. On the second reading, some people hold that “I” is forced (in effect) to take narrow scope, so that in uttering (2) I would say something true only if the referent of “you” thought of me in a particular way, namely as (in a way he would naturally express by uttering) “you”. On this view, we can disambiguate (2) into wide scope (3) (with (4) indicating one amongst other ways in which the referent of “you” might express the reported psychological state) and narrow scope (5) (with (6) indicating the most natural way in which the referent of “you” would express the reported state of mind).
(3) You think that I<wide> was drinking a martini earlier.
(4) He was drinking a martini earlier.
(5) You think that I<narrow> was drinking a martini earlier.
(6) You were drinking a martini earlier.
I have three questions about (2).
Q1. Is it true that (2) has a narrow scope reading, or is it rather, for example, that in some cases in which (2) is natural, it will be natural to expect the referent of “you” to express their thought via (6) rather than (4)?
Q2. Is it true that (1) has no such narrow scope reading?
Q3. Assuming that something like the narrow scope reading is available in English, though not marked in articulation, are there any human languages that mark the distinction between wide and narrow readings? (Perhaps, for example, there are human languages in which analogues of (3) and (5) differ in structure; or perhaps there are such languages in which different (or differently pronounced) expressions would be used to translate “I” in order to readings (3) and (5) respectively.
I’m grateful to unpublished work by M. G. F. Martin for drawing my attention to this issue, which has also been discussed, as he notes, by Mark Richard, and John Perry and Mark Crimmins.