seems and sentential complement

[Robert Levine]

>
> "Tibor Kiss" <afelpado at compuserve.de> writes:
>
> > Hi,
> >
> > since I was so successful recently in asking questions about English, I try
> > again. Very simple question this time: Is it correct that 'seem' requires
> > a that-complement if its complement is finite, i.e. is the following
> > distribution correct?
> >
> > (1) John seems to be certain to leave.
> > (2) It seems that John is certain to leave.
> > (3) *It seems John is certain to leave.
>
> Hi Tibor,
>
>
> All these seem amply natural to me.  I wonder why (3) seems more
> awkward than (2) to me.
>


I agree with both Shalom and Roger's judgments. I wonder if the
discrepancy between (2) and the corpus examples that Roger notes has
something to do with stress. When stress falls on the syllable
directly following `seems' the result tends to be somewhat awkward, or
worse. So maybe it's all prosody?

cheers,

Bob