TITLE: 		EXTENDING FOCUSING FRAMEWORKS TO PROCESS COMPLEX SENTENCES
		AND CORRECT THE WRITTEN ENGLISH OF PROFICIENT SIGNERS OF
		AMERICAN SIGN LANGUAGE
AUTHOR:		Linda Z. Suri
COMMENTS:	Ph.D. dissertation. (c) 1993 Linda Z. Suri

ABSTRACT:
In order for a computer system to understand written text, it must be
able to interpret pronouns. A local focusing framework can be used to
suggest a referent for a pronoun, and an inferencing mechanism that
uses semantic and world knowledge can then accept or reject the
proposed referent. The local focus is the person, property, object or
concept which a sentence is most centrally about within the discourse
context in which the sentence occurs. A local focusing framework
consists of a (set of) focus tracking algorithm(s), and a (set of)
algorithm(s) for suggesting referents for pronouns. A focusing
framework is useful because it is much easier to confirm a
co-specification using inferencing than to find a co-specification
using only inferencing.

Previous local focusing research (including Sidner's focusing research
and centering research) did not explicitly and adequately address
processing complex sentences, although complex sentences are prevalent
in written English. We identify a methodology for determining how to
extend a local focusing framework to process complex (i.e.,
multi-clausal) sentences. Furthermore, we argue that our methodology
is superior to alternative methodologies. We show how we extended our
local focusing framework, RAFT/RAPR, based on our findings using this
methodology. We argue that our methodology is also useful for
assessing and local comparing focusing frameworks and we show why
other methodologies for comparing frameworks are inadequate.

Our focusing research was partly motivated by a project to design an
effective Computer Assisted Language Learning (CALL) system for second
language learners. We explain how one should develop an error taxonomy
and indicate how that taxonomy can affect the design of an entire CALL
system. We explain how one can use a local focusing framework to
correct errors involving inappropriate pronominalization and the
inappropriate omission of noun phrases. While we discuss our CALL
design methodology and justify our claims in terms of our project to
build a CALL system for proficient signers of American Sign Language
(ASL), our findings are applicable to the design of CALL systems for
other second language learners.