THEORY: Cross-Referencing Terms of Sub-Ordinate Clauses

Imagine two languages, named "Aristotelian" and "Goclenian" for the aid of my memory,
in which clauses may appear subordinate within clauses, to arbitrary depth.

In either language, a verb in a subordinate clause can sometimes have markings
indicating some of its "terms" -- that is to say, some of its "core arguments"
-- are the same as some of the "terms" ("core arguments") of, either its
immediate parent clause, or of the main "root-ancestor" clause containing it.

In "Aristotelian", the available markings are limited to these three:
1) The subordinate clause's subject may be the same as its immediate parent's object.
2) The subordinate clause's object may be the same as its immediate parent's object.
3) The subordinate clause's subject may be the same as the "root ancestor"'s subject.

In "Goclenian", the available markings are limited to these three:
4) The subordinate clause's object may be the same as its immediate parent's subject.
5) The subordinate clause's subject may be the same as its immediate parent's subject.
6) The subordinate clause's object may be the same as the "root ancestor"'s object.

----------------------------------------------------------------------------------------------------------------------------------------

In an "Aristotelian" complex sentence with a subordinate clause subordinated to a
subordinate clause, schematically
[S1 V1 O1 SC[S2 V2 O2 SC[S3 V3 O3]]],
there is no way for Clause 3 (SC[S3 V3 O3]) to refer to S2, the subject of Clause 2;
nor to O1, the object of Clause 1; unless either S1 or O2 refers to one of
them.

On the other hand in a "Goclenian" complex-complex sentence with the same schema,
there would be no way for the third clause to refer to O2 or S1 unless either
S2 or O1 referred also to one of them.

But suppose five nominals, N1 N2 N3 N4 N5, and five verbs, V1 V2 V3 V4 V5.
I will write "N3-subj" to show N3 appears as the subject of the appropriate
verb, "N2-obj" to show N2 appears as the object of the appropriate verb,
"Mark1-V4" to show verb V4 appears with "Aristotelian" marking #1 (subordinate
clause's subject same as immediate parent's object), "Mark4-V5" to show verb V5
appears with "Goclenian" marking #4 (subordinate clause's object same as
immediate parent's subject), etc.

The "Aristotelian" sentence
[N1-subj V1 N2-obj SC2[N2-subj V2 N3-obj SC3[N3-subj V3 N4-obj SC4[N4-subj V4 N5-obj
SC5[N1-subj V5 N5-obj]]]]]
is equivalent to the "Aristotelian" sentence
[N1-subj V1 N2-obj SC2[Mark1-V2 N3-obj SC3[Mark1-V3 N4-obj SC4[Mark1-V4 N5-obj
SC5[Mark2-Mark3-V5]]]]]

The "Goclenian" sentence
[N2-subj V1 N1-obj SC2[N3-subj V2 N2-obj SC3[N4-subj V3 N3-obj SC4[N5-subj V4 N4-obj
SC5[N5-subj V5 N1-obj]]]]]
is equivalent to the "Goclenian" sentence
[N2-subj V1 N1-obj SC2[N3-subj Mark4-V2 SC3[N4-subj Mark4-V3 SC4[N5-subj Mark4-V4
SC5[N5-subj Mark5-Mark6-V5]]]]]

---------------------------------------------------------------------------------------------------------------------------------------

Questions:

Many languages contain texts like "The House that Jack Built" and "the loaf of
bread under the seat on the train that my mother bought for two zlotys" (if I
remembered that right, and I'm pretty sure I didn't) with deeply-imbedded
subordinate clauses.

Languages also have means for different clauses of the same utterance to share terms
without completely restating them.

My imaginary languages "Aristotelian" and "Goclenian" introduced above are based
on the hypothesis that, if a subordinate clause needs to refer to a term of a
containing clause, it will most likely be either the immediately containing
clause, or the main clause.

Is that probably so?

There is no way to test that without compiling a corpus of utterances in which
subordination is at least four-deep; that is, the subordinate clause that is
doing the referring is immediately contained within a subordinate clause, which
itself is not immediately contained within the main clause.

How big a corpus of that kind exists?

(The reason why it must be at least four deep; Suppose it was only three deep, SC3
contained within SC2 contained within MC1. Then every term in SC3 would either
be new (introduced by SC3), or come from SC2 (SC3's immediate parent), or come
from MC1 (the main clause).
Only by having the referring clause be at least SC4 can we have a level that can
get skipped -- or fail to get skipped.)

Are there natural languages in which, if a subordinate-within-subordinate clause
shares a term with a containing clause, it is marked differently depending on
whether the shared term is shared with its immediately containing clause or the
main clause?

----------------------------------------------------------------------------------------------------------------------------------------

Something like this whole thing could also be done with consecutive co-ordinated
clauses. For instance, it might be that a verb could be marked to show that any
one of its terms was the same as any one of the terms of the previous verb, or
of the first verb of the chain, but not both at the same time for the same
verb. Or instead, perhaps every verb could share terms with up to two earlier
verbs; always with the previous verb, and then have to choose between the first
verb in the chain, or the verb just before the previous verb.

----------------------------------------------------------------------------------------------------------------------------------------

But a question arises in cases where a clause has more than one subordinate
clause. Suppose a subordinate clause has a term which is markable as shared
with another clause. Do natlangs exist which mark it differentially, depending
on whether it came from another subordinate clause, or from the main clause?

--------

Thanks to anyone who answers anything, even with a guess.

Tom H.C. in MI






		
---------------------------------
 Sell on Yahoo! Auctions  - No fees. Bid on great items.

Reply