LINGUIST List 14.944

Mon Mar 31 2003

Diss: Computational Ling: Mohanty "ECM..."

Editor for this issue: Anita Yahui Huang <anitalinguistlist.org>

Directory

1. rajatkm, Computational Ling: Mohanty "ECM Construction in Oriya..."

Institution: Central Institute of English and Foreign Languages
Program: Doctorate of Philosophy in English
Dissertation Status: Completed
Degree Date: 2002

Author: Rajat Kumar Mohanty 

Dissertation Title: 

ECM Construction in Oriya and English and Tree Adjoining Grammar


Linguistic Field: Computational Linguistics 

Subject Language: Oriya (code: ORY)

Dissertation Director 1: Dr M Hariprasad
Dissertation Director 2: Prof. P Madhavan

Dissertation Abstract:
 
The aim of this thesis is to analyse ECM constructions in Oriya and
account for them using Tree Adjoining Grammar formalism. It has been
shown in generative grammar and other linguistic theories that in an
ECM construction, the matrix verb selects an infinitival clause as its
complement and assigns accusative Case to the embedded subject. In
other words, there is no complementizer possible between the matrix
verb and the embedded subject, which can block the ECM verb in
assigning Case to the embedded subject. But in Oriya, in an ECM
construction, there occurs a complementizer between the matrix verb
and the embedded subject. Yet the embedded subject is marked
accusative Case and the sentence is grammatical. We solved this
problem by looking at basic clause structures, Case system, and
complementizers in Oriya.

Based on a description of Case markers and basic clauses in Oriya, we
have arrived at six tree families in Oriya Tree Adjoining Grammar,
viz., 
(i) [Tnx0nx1V]: Declarative Transitive Tree, 
(ii) [Tnx0V]: Declarative Intransitive Tree,
(iii) [Tnx0nx1nx2V]: Declarative Ditransitive Tree, and
(iv) [Tnx0s1V]: Sentential Complement Tree 
(v) [TXnx0s1V]: ECM Tree,
(vi) [Tnx0Ax1]: Declarative Adjective Small Clause Tree.

We have also identified six extracted trees: 
(i) [(alpha)Wnx0Wnx1nx0nx1V]: Transitive Tree with Object wh-NP
 Extraction and Subject NP extraction,
(ii) [(alpha)Wnx1nx0nx1V]: Transitive Tree with Object NP Extraction,
(iii) [(alpha)Wnx0nx0nx1V]: Transitive Tree with Subject wh-NP
 Extraction,
(iv) [(alpha)Wnx1Wnx0nx0nx1V]: Transitive Tree with Subject wh-NP and
 Object NP Extraction,
(v) [(alpha)Wnx0Wnx1nx0nx1V]: Extracted Transitive Tree,
(vi) [(alpha)Wnx1Wnx0nx0nx1V]: Extracted Transitive Tree.

It has been claimed in Bal (1990) that there are two complementizers
in Oriya, viz., a clause-initial [JE] and a clause-final [BOLI]. Since
Oriya is a head-final language, [JE] as a complementizer is doubtful.
Getting more evidence, we argued that [JE] is not a complementizer but
an emphatic and further, following Kayne (1983), we showed that there
also exists a null complementizer [fi] in Oriya Grammar. The null
complementizer [fi] and [BOLI] are in complementary distribution.
[BOLI] is actually derived from the verb [BOL] 'to sing' or 'to say'.
So we assumed that the complementizer [BOLI] has partially retained
its verbal nature allowing the ECM verb to assign Case to the embedded
subject. In other words, [BOLI], unlike complementizers in many other
languages, is transparent. We used this property of [BOLI] and added
a feature called <transfer-case> to the feature matrix of [BOLI] in
OTAG. The <transfer-case> has the value [acc/nil]. When [BOLI]
selects a finite clause, the value of <transfer-case> becomes [nil]
and so it does not transfer the case from the matrix verb to the
embedded subject. When [BOLI] selects an infinitival clause as its
complement, the value of <transfer-case> is [acc], which matches with
the value of the <case> feature of the embedded subject and the value
of the <assign-case> feature of the matrix verb.