Re: CHAT: Esperanto and prepositions taking the nominative [was Re: CHAT: Back on the list; Anti-conlanging bigots]
| From: | Thomas R. Wier <trwier@...> |
|---|
| Date: | Thursday, December 6, 2001, 22:29 |
|---|
Quoting Christian Thalmann <cinga@...>:
> --- In conlang@y..., "Thomas R. Wier" <trwier@M...> wrote:
>
> > It is arguable, however, that this allative sense is really layered
> > over the underlying case assignment of null-nominative. Note that
> > we also see such layering in a natural language, German, where the
> > dative is retained even when the passive is used:
> >
> > (3) Der(NOM) Mann hat mir(DAT) geholfen.
> > The man helped me.
> > (4) Mir(DAT) wird geholfen.
> > I am helped.
>
> The German passive typically moves the accusative object of a
> transitive verb into the subject slot:
[...]
> In a sentence with both accusative and dative objects, the dative
> remains unchanged:
[...]
> The problem arises when there isn't an accusative object to transform
> into a subject. In your sentence, there is no subject present, though
> it seems you're dropping the generic stand-in subject "es" that could
> be used in this situation:
>
> (e) Es (NOM) wird dem Mann (DAT) geholfen.
> The man is helped. [lit. *"It is helped to the man."]
>
> The "es" doesn't really have any other duty than to provide a subject
> to the passive sentence, it doesn't correspond to any former accusative
> object (which the verb "helfen" couldn't support anyway).
> Correct me if I'm wrong, you seem to be more of a linguist than I.
No, you're right about this. This is going to take a LONG time
to explain, so please bear with me. In an X-bar theoretical analysis,
here is how it would work:
Every word has a subcategorization frame that is stored in the lexicon
along with the meaning and various other kinds of information. This
subcategorization frame tells you whether that particular word can take
NP subjects or NP objects or complementizer phrases like those in English
headed by "that", "to", etc. For example, the English word "help" and
"paint" would have subcat frames kinda like this:
<help, V[case: ACC, -PASS, -AUX], [T1; ___ TP(TFORM: to; T2)]>
part of case semantic compliment sem.
speech role role
("theta-
role")
<paint, V[case: ACC, -PASS, -AUX], [T1; ___ DP(T2)]>
Now, when you're constructing the sentence, the subcategorization frame
tells you what a licit deep-structure sentence will look like. So, for
this frame, we might construct a deep-structure tree for "He helped her
to paint the house":
TP
\
T'
/ \
T VP
DP[T1] V'
he \
V TP[-tense][T2]
helped \
T'
/ \
T VP
to DP[T1] V'
she V DP[T2]
paint the house
Now look back at the subcat frame. The subcat frame accurately
predicts that "she" is the agent of the painting. So, why is
"she" underlyingly *after* "to" in the syntax, and not before?
That's because the deep-structure that we have posited is not
licit in itself; there are other structural principles that also
need to be obeyed before the sentence can become grammatical at
the surface. One of these is, in X-bar theory, that all specifier
positions (i.e., the left-branches) of a TP phrase must be filled
with something. Therefore, to compensate for this, the underlying
"she" gets moved up to that specifier position at surface structure,
where it gets case from "help", which assigns ACC case as marked
above in the subcat frame:
TP
\
DP[T1] T'
He / \
T VP
(t) V'
/ \
V TP[-tense][T2]
helped / \
DP[T1] T'
her / \
T VP
to (t) V'
V DP[T2]
paint the house
[(t) = "trace", place where the word is posited to move from]
Now, other verbs may take very different subcat frames. "Seem",
for example, does not take an external theta role (T1, i.e., does
not take any subject):
<seem, V[case: N/A; -PASS; -AUX], [ ___ TP(TFORM: to; T2)]>
no T1
here
She seemed to paint the house
Deep Structure:
TP
\
T'
/ \
T VP
V'
\
V TP[-tense][T2]
seemed \
T'
/ \
T VP
to DP[T1] V'
she V DP[T2]
paint the house
This deep-structure does not, however, immediately obey our
metarules, so we need to invoke a movement rule: "she" moves
out from the embedded phrase, and up to fill the specifier of
both the embedded and the matrix TP phrase. We see a surface
structure like the following:
TP
DP[T1] \
She T'
/ \
T VP
(t) V'
\
V TP[-tense][T2]
seemed \
(t) T'
/ \
T VP
to (t) V'
V DP[T2]
paint the house
"She" gets case from the T node, which always assigns nominative
in English. Now, here is where we get back to your problem:
German and English both share the rule mentioned above that SpecTP
must be filled by something, even if it's meaningless. That's why
we get expletive "it" in English and expletive "es" in German like
below:
Deep structure of "It has rained."
TP
T'
T VP
V'
V VP
has DP V'
it V
rained
Deep structure of "Es hat geregnet."
TP
T'
T VP
V'
V VP
hat DP V'
es V
geregnet
In both of these structures, the expletive pronoun (it and es
respectively) must surface at the topmost SpecTP:
Surface structure of "It has rained."
TP
DP T'
T VP
it (t) V'
V VP
has (t) V'
V
rained
Surface structure of "Es hat geregnet."
TP
DP T'
T VP
es (t) V'
V VP
hat (t) V'
V
geregnet
German works like English in this respect (or English like
German, depending on your point of view), because also like
English, it doesn't *require* an expletive pronoun. Something
else may move up to fill that place, too. We may posit a
subcat frame for the German verb <helfen> like the following:
<helfen, V[case: DAT; -PASS; -AUX], [T1; ___ DP[T2]]>
From this, we could come up with a deep structure of an active and
passive sentence:
Active: Er hilft mir. (He helps me)
TP
T'
T VP
DP V'
er V DP[T2]
hilft mir
Passives, in both German and English, are lexicalized
transformations from the active: they loose their external
subject (here, "er"), and don't take core cases. Helfen,
however, does not assign a core case, but an oblique one:
Passive:
TP
T'
T VP
V'
V VP
wird V'
DP V
mir geholfen
("Geholfen" switches the position of the V and DP here.)
At the surface, this DP "mir" needs to move up to fill SpecTP:
TP
DP T'
T VP
Mir (t) V'
V VP
wird (t) V'
(t) V
geholfen
So, in other words, unlike the core cases you mentioned,
the dative has a semantic connotation that accusative does
not. So it stays in that case, overriding the nominative
it would normally get from the T. In X-bar theory, this
would also override the number checking that would normally
happen, so that you get "wird" (third person singular) instead
of the expected "werde" (first person singular). So, at long
last, what you pointed out about "es" filling in the position
is actually integrally related to case-marking (in X-bar Theory,
anyways), because the desire to mark case seems to be the
impetus to move in the first place.
(I have tried to keep my opinions about what's going on distinct
from what X-Bar Theory says. There are many aspects of X-bar
Theory that really bug me, so if there is a problem with the
analysis above, it might be because there is a problem with
X-Bar Theory, not me. :) I hope I have not simply succeeded in
muddying the waters further. I'm not fully convinced that the
above argument is a satisfactory explanation of the data.)
=====================================================================
Thomas Wier <trwier@...> <http://home.uchicago.edu/~trwier>
"...koruphàs hetéras hetére:isi prosápto:n /
Dept. of Linguistics mú:tho:n mè: teléein atrapòn mían..."
University of Chicago "To join together diverse peaks of thought /
1010 E. 59th Street and not complete one road that has no turn"
Chicago, IL 60637 Empedocles, _On Nature_, on speculative thinkers