LINGUIST List 13.848

Wed Mar 27 2002

Diss: Discourse Analysis: Taboada "Collaborating..."

Editor for this issue: Karolina Owczarzak <karolinalinguistlist.org>

Directory

1. mtaboada, Discourse Analysis: Taboada "Collaborating through Talk"

New Dissertation Abstract

Institution: Universidad Complutense de Madrid
Program: English Philology
Dissertation Status: Completed
Degree Date: 2001

Author: Maite Taboada 

Dissertation Title: 
Collaborating through Talk: The Interactive Construction of
Task-Oriented Dialogue in English and in Spanish

Dissertation URL: http://www.sfu.ca/~mtaboada/thesis.html

Linguistic Field: 
Discourse Analysis, Computational Linguistics

Subject Language: 
Spanish, English 

Dissertation Director: Julia Lavid 


Dissertation Abstract:
 
This thesis constitutes a study of coherence at all levels in
conversation. Typical conversations are considered to be coherent, by
those participating, and also often by those observing them. However,
it is difficult to pinpoint exactly what makes them coherent, and how
that coherence is achieved dynamically. I study some of the resources
that speakers employ when building conversations. These resources
contribute to the overall coherence, which is constructed
interactively as speakers build on each other's contributions.

The study is cross-linguistic, based on a corpus of task-oriented
conversations between dyads of two speakers each. The corpus contains
conversations between two native speakers of English on the one hand,
and two native speakers of Spanish, on the other. The interlocutors
have conflicting agendas that cover a two to four week period, and are
told to agree on an appointment within that time. The speakers try to
find a date on which they are both free to meet. The conversations
are viewed as texts that are built interactively by the speakers, as a
collaborative effort. The analyses are concerned with the discourse
characteristics that hold the conversations together as a
text.

The main framework of study is the analysis of speech genres (Bakhtin
1986). These conversations are an instance of a genre, as a "staged,
goal-oriented, purposeful activity in which speakers engage as members
of our culture" (Martin 1984). As such, they are divided in clear
stages, the main three of which are: Opening, Task Performance and
Closing. For these stages to be distinguishable, they need to contain
different patterns of linguistic realization. The linguistic and
discourse characteristics studied include: thematic progression,
rhetorical structure analysis and cohesion. These three types of
analyses are unified under the textual metafunction of language, in
the Hallidayan tradition. Halliday and Hasan (1976) divide the textual
metafunction into structural and non-structural components. The
structural one, at the clause level, is the thematic structure of the
clause. The non-structural one is cohesion. One of the elements of
cohesion is conjunction, which I represented here not through
Halliday's and Martin's conjunctive relations, but through a similar
theory, Rhetorical Structure Theory (Mann and Thompson
1988).

This study is, then, one of text, textual metafunction and texture in
a spoken genre. It sheds light on how two speakers create a text
interactively. The second goal of this work is to find out whether and
if so, how that process is different in English and in Spanish.

The main contributions of the thesis are summarized in: (1)
corpus-based characterization of a dialogic genre; (2) compilation of
a body of analysis tools for generic analysis; (3) application of
English-based analyses to Spanish and comparisons between English and
Spanish text-building strategies; and (4) a study of the
characteristics of each stage in the dialogues, for both languages.

The thesis contains descriptions of the three types of analyses:
thematic realization and progression, rhetorical structure and
cohesion. The description covers the English and Spanish data, with
comparisons between the two. Finally, a computational model of the
structure of the dialogues is provided, divided into stages and speech
acts found in each of the stages. The computational model is based on
the generic structure of the dialogues. It formalizes the sequencing
of stages, and the sequencing of speech acts within those stages. The
model has been used in a Machine Translation system. The complete
characterization of the stages is used throughout the study, in order
to delimit the stages as units in the conversations within which the
different phenomena are explored.