Re: THEORY: CP-V2 vs. IP-V2

I'm glad you were able to get something out of my long-winded
explanation.  Here are some comments on your comments...

On Wed, 19 May 1999, John Cowan wrote:

> Matt Pearson wrote:
>=20
> > To account for the fact that V2 only occurs in main clauses, certain
> > linguists have suggested that the V2 position is the same as the
> > position occupied by the complementizer "that" (this position is
> > called C).
>=20
> So in this syntax model, the complementizer belongs to the
> embedded clause?  That seems to go against the historical evidence,
> at least for English,
> which is that that "that" is the same "that" that is used
> as a demonstrative, and that "I know that.  John read the book
> yesterday" was the original form, with the "that" referring to
> the whole of the next sentence.
Personally, I'm highly skeptical of this account, since it would=20
seem to suggest that Old English (or its Germanic ancestor; not sure how
far back we're going here) didn't have embedded clauses, which seems
very unlikely.  A more plausible story, I think - although I confess I
don't know the facts - is that the "that" originally started out as a
determiner restricting the embedded clause, suggesting the following=20
as a historical parallel:

=09We know [ that [ boy ] ]
=09We know [ that [ John is sick ] ]

=2E..where the determiner "that" can take either a noun ("boy") or a
clause ("John is sick") as its restriction.  IMHO, this analysis
makes much more sense than the biclausal analysis from a cross-
linguistic point of view:  There are tons of languages out there
(Turkish, Quechua, and Warlpiri leap to mind) where embedded clauses
behave morphologically like noun phrases, and even languages
(Malagasy, some Mayan and Salishan languages) where embedded clauses
take determiners like "the" or "that".  Here's an example from Malagasy:

=09Mahafinaritra ahy ny  novidin'i John ny  akanjo
=09pleases       me  the bought    John the dress
 =09
This is literally "The [John bought the dress] pleases me" - i.e.
"It pleases me that John bought the dress".

Anyhow, even if the biclausal story on the origin of English "that"
is true, it doesn't contradict the claim that synchronically,=20
complementizers are part of the embedded clause rather than the
main clause.  And I think there's pretty good evidence for saying
that "that" is part of the embedded clause, e.g.:

(1) The complementizer "that" can show up in places where the
determiner/pronominal "that" is not allowed, e.g. after the
predicate in an impersonal "it" construction.  Compare:

=09 [That John is a liar] is clear
=09 [That] is clear

But:=09 It is clear [that John is a liar]
=09*It is clear [that]

If "that" were part of the main clause, we would need a special
stipulation to explain why "It is clear that" is ungrammatical.

(2) "That" moves along with the embedded clause when the latter
becomes a derived subject.  Compare:

=09It seems to have been proven [that John is a liar]
=09[That John is a liar] seems to have been proven

You can't say "John is a liar seems to have been proven that".
If "that" were part of the main clause, we would have to explain
why it is possible to front it with the embedded clause, whereas
it is not possible to front other pronouns with a following
embedded clause.

(3) When two embedded clauses are coordinated, the "that" can be
repeated:

=09We know [that John is sick] and [that Bill is injured]

> > Languages which have the German/Dutch/Swedish pattern are called "CP-V2=
"
> > languages, reflecting the fact that the verb-second position is C, whil=
e
> > languages which have the Icelandic/Yiddish pattern are called "IP-V2",
> > reflecting the fact that in these languages the verb only moves up to I=
=2E
>=20
> What about infinitives?  In German we have:
>=20
>         Es ist schwer, ein J=FCde zu sein.
>=20
> but in Yiddish (proverbially)
>=20
>         Es ist shver tsu zayn a Yid.
>=20
> Looks like the same basic pattern, but does the same theory account
> for it?
A Yiddish-speaking friend of mine (who is not a native speaker, but
used to be quite fluent) informs me that it's possible to say EITHER "Es
is shver tsu zayn a Yid" ("is", not "ist") OR "Es is shver a Yid tsu
zagn".  So you can get either the English order or the German order.
According to her intuitions, this only holds if the direct object is
a full noun phrase.  If you have a pronominal direct object, it must
precede the infinitive:

=09Ikh vil a beygel essn=09"I want a bagel to eat"
=09Ikh vil essn a beygel=09"I want to eat a bagel"

=09Ikh vil im essn=09=09"I want it to eat"
=09*Ikh vil essn im=09"I want to eat it"

This is reminiscent of the word order possibilities with verb-particle
constructions in English:

=09He took the garbage out
=09He took out the garbage

=09He took it out
=09*He took out it

It's possible that the non-finite verb in Yiddish is occupying the
same position as particles like "out" in English.

Anyhow, that's just me thinking out loud.  The answer to your
question is no, the contrast between Yiddish and German cannot
be attributed to the IP-V2/CP-V2 parameter, since it involves the
position of non-finite verbs rather than finite verbs.  Where
non-finite verbs go seems to cross-cut the IP-V2/CP-V2 dichotomy,
leading to a finer-grained typology of verb-second languages:

=09=09=09IP-V2=09=09CP-V2

=09  Obj - Inf=09Kashmiri(?)=09German
=09=09=09=09=09Dutch

=09  Inf - Obj=09Icelandic=09Swedish
=09=09=09=09=09Norwegian
=09=09=09=09=09Danish

=09  both=09=09Yiddish=09=09??

Old English was another language like Yiddish, where you could=20
apparently get the infinitive either preceding or following the=20
direct object (if the latter was a noun phrase), but I don't=20
recall whether Old English was supposed to be IP-V2 or CP-V2 -
probably IP-V2, since having V2 in embedded clauses seems
like a reasonable first step towards losing V2 altogether
(as English did sometime during the Middle English period).

Matt.