LINGUIST List 4.663

Fri 03 Sep 1993

Sum: "As"-clauses

Editor for this issue: <>

Directory

1. spena, As-clauses

Summary of correspondence on 'As'-clauses
Thanks to the following who responded to my query of some
while ago on 'as'-clauses with missing subjects:

Caroline Heycock, Richard Campbell, Lynne Murphey, Kai von
Fintel, Andrew Barss, David Pesetsky, Marc Authier, Alan Munn,
Polly Jacobson, Jim Jewett, Beth Levin, Leslie Barrett, Dick DeArmond,
Max Wheeler, Dan Finer, John Lee, Stephen Spackman,Glenn Ayres, Jan Odijk.
Several people mentioned papers by Lapointe and by Stowell
bearing on the issue. Alan Munns mentioned that there has
already been some discussion of a related problem (and that he
had had email discussion with Anita Mittwoch about it).

Some of the discussants included detailed substantive points
which I append here (in no particular order), together with a
copy of the original query:

My colleague, Martin Atkinson, has pointed out to me that the
subject position indicated by ___ in (1) appears to be a case
of an expletive small pro.

(1) As ____ has been pointed out, English lacks nullsubjects.

[Interestingly, L2 learners of English often insert 'it' in
the place of the gap.]

Is this the canonical analysis within (some variety of) GB?
Cursory examination suggests that the construction is not
permitted with other types of clause, and only seems possible
in As-clauses with passive or raising predicates, and
extraposition of infinitivals from raising and tough
predicates (depending presumably on semantics; cf (5), (6)):

(2) As ____ seems obvious nowadays, ...
(3) As ____ is clear from the raw data, ...
(4) As ____ is likely to be obvious from the raw data, ...
(5) As ____ is easy to see from the data, ...
(6) *As ____ would be difficult to prove from these data, ...

Extraposition of finite clauses seems to be excluded:

(7) *As ___ seems that everyone now agrees, ...
(8) *As ___ is clear that everyone now agrees, ...
(9) *As ___ has been pointed out, that (it) is the case, ...

Are these data really representative? If so, what governs this
patterning?

- - - - -
1.

there is a paper by Steven Lapointe on this topic:
Lapointe, Steven G. 1990. "Empty Category in Reportive 'As'
Clauses".
 Proceedings of the First Meeting of the Formal Linguistics
Society of
 Midamerica, University of Wisconsin, Madison. 177-192.
That's definitely the place to look.
-
Kai von Fintel fintellinguist.umass.edu
Dept. of Linguistics finteltitan.ucs.umass.edu
South College
University of Massachusetts
Amherst MA 01003
- - - - - -
2.
 From: JACOBSON <LI700013BROWNVM.brown.edu>
 I have the feeling that analyzing examples like (1) as
involving expletive little pro's is a misanalysis:

(1) As __ has been pointed out, English lacks null
expletives.

Rather, I believe that the correct analysis is that "as" is
just a lexical CP - and hence can go anywhere where CP's can.
This analysis would account for all of your data, except for
number (6) - which seems okay to me.
 Thus note that in all of the other examples, a CP is also
possible in subject position (there have been works - e.g.,
Koster, in an LI paper in the late 70's - which claims that
the CP is in topicalized rather than subject position in my
(b) examples below; I personally don't agree with those
analyses but in fact that's orthogonal to the point - the
"as", in such analyses, could also be in topicalized position.
What's important is just that wherever "as" could occur, so
could a full CP):

(1). a. As has been pointed out,...
 b. That Bill came has been pointed out
(2) a. As seems obvious, ...
 b. That Bill came seems obvious
(3) a. As is clear from the raw data,...
 b. That this theory is correct is clear from the raw data
(4) a. As is likely to be obvious from the raw data,...
 b. That this theory is correct is likely to be obvious
from the raw data
(5) a. As is easy to see from the raw data,...
 b. That this theory is correct is likely to be easy to
see from the raw data
(6) a. As would be difficult to believe,...
 b. That this theory is correct would be difficult to
believe

(6) here is not exactly the same as your (6); I substituted in
one that seems impeccable, just as would be expected under the
account that takes "as" to be a lexical CP (I
believe,incidentally, that it is also of other categories too;
it also seems to be a lexical AP).
 Most strikingly, note that this view automatically explains
the badness of your (7)-(9):

 (7) a. *As seems that everyone now agrees
 b. *That this theory is correct seems that everyone now
agrees
 (8) a. *As is clear that everyone now agrees,...
 b. *That this theory is correct is clear that everyone
now agrees

These are of course bad, since if the CP occurs in extraposed
position it's impossible to get another CP in subject (or, if
Koster is right, in topicalized) position - hence "as" is also
bad here.
 The same analysis works for things like:

 (9) It's raining, as I had hoped.

Here I think that "as" is, in essence, a preposed CP. Note
that "hope" allows for CP complements, and so allows an "as"
here. Note that a verb like "express" allows only NP and not
CP objects:

 10. a. This theory expresses that fact.
 b. *This theory expresses that languages are
learnable.

And, just as we would expect, no "as" is possible here either:

 11. ?*Languages are learnable, just as this theory
expresses.

Incidentally, I've discovered a lot of idiosyncratic variation
with respect to (10a) and (10b) - some speakers DO like (10b)-
and I would predict that such speakers would also like (11).
I myself do find (11) a bit better than (10b), and don't know
why, but the basic pattern seems clear. Another verb like
"express" is" capture":

 12. a. This theory captures that fact.
 b. *This theory captures that languages are learnable.
And again: 13. ?*Languages are learnable, just as my
theory captures.

 Polly Jacobson
 Dept. of Cognitive and Linguistic
Sciences
 Brown University
- - - - -

3.
 From: Dan Finer <DFINERCCVM.sunysb.edu>
Organization: State University of New York at Stony Brook

<...> it seems that there a few things one could say from
a GB perspective. First, the null item isn't restricted to
subject position; it can also show up [sic] in objectposition:

 (1) As Bill has mentioned _,
 As everyone knows _,
 As Bill has been saying _ for years,
 As I argue _ in my comments below,

I don't think that an expletive is involved here; if it were,
there would be a loose thematic role in the clause. I imagine
that the thematic role is assigned to "__" in "As __ has been
pointed out", presumably via a trace in object position. The
non-
expletive hypothesis is consistent with the status of the null
element in the data in (1), and if it's not an expletive, then
we
also have an account of the ungrammatical examples you give
involving extraposition, such as

 (2) *As ___ seems that everyone now agrees,
 ...
 *As ___ is clear that everyone now agrees,
 ...
 *As ___ has been pointed out, that (it) is
 the case,

since in these cases, an expletive would be required,
i.e.,there
is no thematic role left over for the subject position. But
if
the null element isn't an expletive, it would require a
thematic
role, but if there isn't one, then etc...

The type of argument that "__" corresponds to must beclause
like
in some sense; it takes the clause it's adjoined to as an
antecedent. This may partially account for the intuition that
an
expletive pro is involved, as well as the non-native
introduction
of _it_ in these contexts. It also provides an account of the
contrast below in (3).

 (3) As _ was predicted/*arrested, Bill was
 very unhappy to be in jail.

In line with this observation, this next example can only be
paraphrased "We have read that this book is interesting," not
"We
have read this book, and it is interesting"

 (4) As we have read, this book is interesting.

It's also unlikely that it's pro, given Ross' inner islands
(83-84) (cf. (5)-(6)), picked up in Cinque's 1990 monograph,
_Types of A'-Dependencies_. The relation between the top of
the
As-clause and the null position is sensitive to weak islands,
and
Cinque takes this to indicate that movement of some
"nonreferential" item is involved. I'm not sure what kind of
null
category is moving in these examples (some type of CP?), but
the
sensitivity to weak islands that a movement/non-pro analysis
predicts certainly seems to be borne out, with "__" in either
subject or object position.
 (5) John is a happy guy, which you know _
 John is a happy guy, as you know _

 (6) John is a happy guy, which you don't know _
 *John is a happy guy, as you don't know _

 (7) As I believe _ has been proposed by
 someone, nouns are really prepositions.

 *As I don't think _ has been proposed by anyone,
 nouns are really prepositions.

 *As you wonder when _ was proposed, nouns
 are really prepositions.

 *As you wonder when Bill said __, nouns are
 really prepositions.

 As you know Bill will say __, nouns are really
 prepositions.

Your example (9) ((8) here) is an interesting case. It sounds
more or less OK, but only with a null lower subject (when the
position is filled, it's * for me, too). At first, it looked
like
a case of a parasitic gap being licensed from a c-commanding
position (horrors!), or movement across a null expletive pro
(almost as bad, given what I've said above), but after looking
at
it for awhile, I think it may be a case of relativization on
the
moved element, with extraposition of the relative clause. That
is, the sequence "that is the case" isn't a complement of
"point
out," it's rather the extraposed clause. (9) is somewhat
analogous

 (8) As ___ has been pointed out that (*it) is
 the case, ...

 (9) The proposition that is the case has been
 pointed out.

 (10) The proposition has been pointed out that
 is the case.

Correspondingly, (11), built on your (9), falls on the
grammatical
side of the fence, I think.

 (11) As it has been pointed out __ is the case,

Here, "__ is the case" is a complement to "point out", and
movement has taken place from lower subject position, as in
the
first example in (7). (12) should be out, since if the lower
clause is a complement, then movement across "that" should be
blocked (in contrast to (11)). On the other hand, if we have
the
relative extraposition structure (the clause is not a
complement),
"it" occupies the extraction site.

 (12) *As it has been pointed out that is the case,

- - - - -
4.

 From: odijkjeprl.philips.nl

(1) The `gaps' in `as' clauses are, to my knowledge, NOT
restricted to the
 subject position.
 Examples that appear correct to me (but I am not a native)
are:
 John left, as everybody knows ___ (from Cinque 1990:19)
 As Peter noticed ___, there are more possiblities.

 It appears to me then, that the construction might not
involve a little pro,
 but covert wh-movement (an empty operator).

(2) Some relevant literature is:

 G.Cinque, 1990, `Ergative Adjectives and the Lexicalist
Hypothesis',
 Natural Language and Linguistic Theory 8.1, 1-40.

 especially p. 18-21, and references cited there.

Jan Odijk
e-mail: odijkjeprl.philips.nl

- - - - -

5.

 From: Glenn Ayres - CDI San German <gayresinter.ui.clu.edu>

Just two short observations:

1) Subjects are also omitted with THAN-clauses:

 They didn't work harder than was necessary.

2) In general the verb of the clause is BE, but we do have:

 The procedure is as follows: ....

Glenn Ayres