15.2004, Diss: Comp Ling/Semantics: Paul: 'An HPSG...'

[LINGUIST List]

LINGUIST List:  Vol-15-2004. Tue Jul 6 2004. ISSN: 1068-4875.

Subject: 15.2004, Diss: Comp Ling/Semantics: Paul: 'An HPSG...'

Moderators: Anthony Aristar, Wayne State U.<aristar at linguistlist.org>
            Helen Dry, Eastern Michigan U. <hdry at linguistlist.org>

Reviews (reviews at linguistlist.org):
	Sheila Collberg, U. of Arizona
	Terence Langendoen, U. of Arizona

Home Page:  http://linguistlist.org/

The LINGUIST List is funded by Eastern Michigan University, Wayne
State University, and donations from subscribers and publishers.

Editor for this issue: Takako Matsui <tako at linguistlist.org>
 ==========================================================================
To post to LINGUIST, use our convenient web form at
http://linguistlist.org/LL/posttolinguist.html.
=================================Directory=================================

1)
Date:  Mon, 5 Jul 2004 06:14:34 -0400 (EDT)
From:  soma_paul at yahoo.com
Subject:  An HPSG Account of Bangla Compound Verbs with LKB Implementation

-------------------------------- Message 1 -------------------------------

Date:  Mon, 5 Jul 2004 06:14:34 -0400 (EDT)
From:  soma_paul at yahoo.com
Subject:  An HPSG Account of Bangla Compound Verbs with LKB Implementation

Institution: University of Hyderabad
Program: Centre for Applied Linguistics and Translation Studies
Dissertation Status: Completed
Degree Date: 2004

Author: Soma Paul

Dissertation Title:An HPSG Account of Bangla Compound Verbs with LKB
Implementation

Linguistic Field: Computational Linguistics, Semantics, Syntax

Dissertation Director 1: Gautam Sengupta
Dissertation Director 2: Probal Dasgupta

Dissertation Abstract:

This thesis attempts to present a constraint-based and
semantically-grounded account of the composition of Bangla Compound
Verbs (CV) within the framework of Head-Driven Phrase Structure
Grammar (HPSG).  A Bangla CV is a two-member construction in which the
first member (V1) chooses between the usual conjunctive participial
form -e and the (rare, for this construction) infinitive form -te
while the second member (V2) bears the inflection.

1. mee-Ta  hes-e    uTh-lo
   girl-cl laugh-cp rise-3 pt
   'The girl burst into laughter'

I have identified fifteen frequently used V2s in Bangla.  A V1 does
not combine with every V2.  The combinatorial well-formedness between
a V1 and a V2 depends on the semantic compatibility between the two
verbs.  In this thesis, I propose to impose a constraint on the
semantic component of verbal lexical signs, which are arranged in a
system of multiple inheritance hierarchy.  This constraint controls
the unification of a V1 with a V2 to ensure that the grammar licenses
only well-formed CV sequences. I propose for a two-level
representation of the semantic component of verbs:

1. 	A participant level contains "grammatically relevant" information
	that defines the relations among the participants involved in the
	situation denoted by the semantic type of the verb
2. 	A supra-lexical level includes information related to temporality
	and aspect.

I implement this proposal in HPSG by introducing two attributes THEM
(representing participant level) and GRAM (representing supra-lexical
level) as the values of the semantic feature SEM | PREDS.  The
postulation of the two levels for organizing the meaning of verbs is
significant for constituting compound verb constructions.  The
semantic constraint that governs the unification of V1 with a
particular V2 is declared on the value of THEM of the verb lexeme
type, while the semantic compounding principle that builds the
temporal and aspectual features of the resultant CV is operative on
the value of GRAM.  The GRAM value of V2 is inherited by the CV of
which the V2 is a part.

The model that I propose in this thesis for the composition of the CV
constructions essentially takes into consideration the following two
premises:

1.	CVs are the lexical variants of their V1 participant
2.	CVs represent one functional-semantic unit - a predicate

I maintain that two lexical entities satisfy the constraint of lexical
variance when they are semantically related and some principle (or
lexical rule) can be formulated to regularize the formation of one
form from another.  In the course of this thesis it becomes evident
that verbs in Bangla and other Indo-Aryan languages have a kind of
lexical variant, the CVs, which are multi-word expressions.  I argue
that a predicate in Bangla and other Indo-Aryan languages can be
expressed by both one-word verb forms and multi-verb expressions.  The
significance of shifting the focus of investigation from
morphosyntactic representation of verbal expressions to their content
is the following: instead of considering the verb to be the head of
the projected phrasal structure, we can now assign the status of
'head' to a constituent that represents a predicate.  Just like a
simple verb, CVs project the sentences they head.  I lay emphasis on
the observation that morphosyntactic representation of predicates,
either a one-word form or multi-word expressions, constitutes the head
of a syntactic phrase.  As far as constituting the phrasal structure
of the CVs is concerned, I argue in favor of taking the V1 to be the
head of the phrasal structure of its CV variant even though the V2
bears the categorial information for the whole construction. Finally I
present an LKB-based implementation, mainly designed to test the
analysis for the CV constructions proposed in this thesis.

---------------------------------------------------------------------------
LINGUIST List: Vol-15-2004