And:
>#1.  [This cup]i is [ _i,j (for you) to drink out of _j ].
>#2.  [This cup]i is easy [ _i,j (for you) to drink out of _j ].
>#You're assuming, I think, that "to (drink out of)" is the sharer/xcomp of
>#"is", and therefore "this (cup)" is its subject. On the other hand, I think
>#I normally assume that "(for) you" is the subject of "to (drink)". How can
>#"to" have two subjects?
>
>My analysis is that when _for_ is present, it is the sharer, and when it
>is absent, _to_ is the sharer.
## OK. So in (1),
(1)     It is for you to drink.
"for" is the sharer of "is", so "it" is the subject of "for".

>I agree, except that contra you and my thesis, but pro EWG, I think the
>subject of TO is not complement of FOR (i.e. in "for you to go", "you"
>is subj of "to" but not complement of "for"): this explains why it won't
>extract (-- because (EWG again -- and I agree with it) subjects only
>of complements of verbs extract).
## So in (1), what's the relation between "you" and "for"? If "you" is the
subject of "for" we have a problem, because "for" already has a subject -
"it". And yet there must be some relation, because "you" is obligatory
after "for" but impossible without it.

>So if you accepted my analysis, you'd say that the subject of
>the gap-containing FOR phrase is also, by definition, its extractee?
## Aha, now I understand your question!  No, I don't think it has to be
extractee by definition. For instance, we might be able to handle the
displacement entirely in terms of sharing/subject raising. So the question
is how we relate "it" to "eat" in my (1), or (more challengingly) "this
(cup)" to "of" in your 1. In fact I don't see any alternative to extraction.

## To summarise, I think my analysis of (1) is this:

(1)  It <-s- is -r-> for / -?-> you \-r-> to -r-> drink.
Plus:   it <-??- for
        it <-x- to   [x = extractee]
        it <-x- drink
The big question is what ? and ?? are.
a. If ? = s, as you suggest (and I accept in principle), then ?? can't be s
as well, so we need some kind of magic to override the usual rules for
sharers so that ?? = extractee.
b. If ? = c, then ?? could be s and we could have a special rule which
converted the s of FOR into the x of TO.

I prefer a if we can get it to work, but I'm not sure how to do it. Notice
incidentally that we can't make it conditional on being sharer of BE,
because the same construction occurs as postmodifier of a noun:
(2)     The stuff (for you) to drink is this.
or as adjunct of a verb:
(3)     I bought it (for you) to drink.

I used to think that all these examples showed that extractee and subject
had similar distributions, and I think I still think this. Conveniently
they're both pre-dependents of a verb in WG. Maybe extraction and
subject-raising are in fact one and the same thing?