LINGUIST List 14.2171

Sun Aug 17 2003

Review: Cognitive Sci/Text: Louwerse & van Peer (2002)

Editor for this issue: Naomi Ogasawara <naomilinguistlist.org>

Directory

1. Dominic Forest, Thematics: Interdisciplinary Studies

Louwerse, Max and Willie van Peer (2002) Thematics: Interdisciplinary
Studies, John Benjamins Publishing Company, Converging Evidence in
Language and Communication Research 3.

Announced at http://linguistlist.org/issues/13/13-2420.html


Dominic Forest, Universit� du Qu�bec � Montr�al

SYNOPSIS

The aim of the book ''Thematics: Interdisciplinary Studies'' is to
define, from an interdisciplinary point of view, the concept of
theme. The book is composed of 22 essays (from various disciplines
such as literature, linguistics, psychology and cognitive sciences)
grouped into two main parts (each part is preceded by a useful, clear
and concise introduction).

The first part, entitled ''Structure and Processing'', focuses on
language and linguistic aspects of thematic structures and
processes. This first part is divided into three sections.

The section ''Theoretical approaches'' discusses the concept of theme.
Graesser, Pomeroy and Craig (chapter 1) give an overview of
psychological and computational researches on theme comprehension. The
goal of their paper is to ''investigate psychological theories that
explain how humans comprehend text'' (p. 19). This goal can be
achieved in part by a better comprehension of how the reader
identifies and understands themes in a text. Zwaan, Radvansky and
Written (chapter 2) present a detailed examination of the situation
models within the framework of the Event-Indexing Model, focusing on
their applications to language comprehension. The authors argue that
situation models are an interesting strating point to explain how the
readers extract themes from stories, but those models seem not to be
necessary and sufficient for the abstraction of theme-motifs. Finally,
Van Oostendorp, Otero and Campanario (chapter 3) present an
interesting explanation of how the situation-models update during text
processing.

The section ''Experimental and corpus linguistic approaches'' concerns
the relation between textual structures and theme emergence. The
problem is to identify the potential relation between specific
linguistic expressions and the emergence of themes in the reader's
mind. From this point of view, Shen (chapter 4) discusses the relation
between the concept of point structure theory of stories (where the
point is defined as the raison d'�tre of the story) and the affective
state of the reader. Similarly, Emmott (chapter 5) discusses the
importance of linguistic style in relation to thematic meaning in the
context of cohesive chains of referring expressions. Gernsbacher and
Robertson (chapter 6) defend the idea that the definite article
''the'' constitutes a structural marker of thematic cohesion, a cue to
map thematic information. In the last chapter of this section, Kim
(chapter 7) demonstrates the interaction between Korean case markers
and the thematic structure of the discourse.

The section ''Computational approaches'' is dedicated to computer
models that, in various ways, process textual and linguistic elements
to retrieve themes in a text. Kintsch's (chapter 8) presents how the
knowledge used to understand a text can be presented in an algebraic
form. He presents a specific model of text comprehension (the
construction-integration (CI) theory). An important part of this paper
is dedicated to the presentation and discussion of the Latent Semantic
Analysis (LSA) and to the analysis of themes in terms of hierarchical
macrostructures. In chapter 9, Le presents an original point of view
on theme analysis based on the idea that units of text larger than the
sentence are composed of concepts and are therefore relevant to theme
analysis. Thus, she presents the grounds of thematic analysis and
organization based on discursive structures at the paragraph level. In
the last chapter of this section, Louwerse (chapter 10) presents and
evaluates a connectionist model of theme retrieval. The computer
program presented by Louwerse, called CoCon (Coherence Connectionist
model), is based on the concept of coherence in text. For Louwerse,
the referential, locational, causal and temporal coherence can help
defining what text is about, facilitating the retrieval of themes
texts. The evaluation of the model presented was done by comparing the
results obtained by the computer model and readers' summaries of the
same text. The results obtained showed the success of this approach
in performing relatively accurate summaries of texts and theme
retrieval.

The second part, entitled ''Content and Context'', discusses the
content and context of thematic issues. This second section is also
divided into three sections: ''Theoretical Approaches'',
''Interpretive Approaches'' and ''Computational Approaches''.

In the first section, ''Theoretical Approaches'', Sollors (chapter 11)
raises a few questions concerning the role of thematic approaches in
contemporary literary criticism. In the first part of his paper, the
author presents an overview of current perspective on theme analysis.
This clear and useful overview leads the author to discuss
possibilities regarding the problem of identifying themes in texts. In
chapter 12, Petterson gives an overview of recent studies in
thematics. This overview is developed through seven trends identified
by the author. These seven trends are: ''Theme as unifying element'',
''focus on motif'', ''focus on communicative and interpretive
aspects'', ''theme as an interrelation between text and world'',
''humanist thematic'', ''computer content analysis'' and ''empirical
thematics''. Keeping these seven trends in mind, the author concludes
with a reflection on the future of literary thematics. In the next
chapter, Van Peer (chapter 13) gives some relevant answers to the
question of why there are themes in literature. He argues that
literary themes are to be considered differently (and also operate
differently) from themes in non-literary texts. This distinction lies
in the foregrounded nature of literary themes. Also, literary themes
are characterized by specific features (their ''meaning maximizing''
potential and their emotional coloring, for example) and their
tendency toward intercultural and interperiodical
representation. Finally, he underlines the importance of an
interdisciplinary approach to the study of themes. In the last part of
this section, Roque (chapter 14) presents an explanation of thematics
in visual arts based on Panofsky's theory of visual meaning. His
objective is to demonstrate the importance of the artist's motives in
order to understand artisitic works.

The second section, ''Interpretative approaches'', is dedicated to
detailed analysis of specific themes from textual and cultural
perspectives. For instance, in her paper, Giora (chapter 15) explains
how some taboo themes in some cultures can only be exposed by means of
irony. Hjort (chapter 16) presents some important aspects of themes of
nation and argues that thematizations of nation usually imply elements
of self-deception, mythmaking and motivated irrationality. Daemmrich
(chapter 17) focuses on the theme of war in contemporary German
literature. He attempts to identify some of the textual elements by
which thematization of war is realized and the meaning of such a
theme. Finally, in the last chapter of this section, Wolf (chapter
18) discusses the variance between the literary modes of presentation
(often characterized by its subjectivity and irrationality) and the
non-literary modes of presentation (often characterized by its
objectivity) of a theme. Tosupport his ideas, the author uses the
theme of ''money'' as example.

As Louwerse and van Peer point out in their introduction to part 2,
one important point about themes is that they ''allow the grouping of
meanings into manageable chunks'' and therefore text processing
technologies based on thematic analysis can be seen as a form of
''knowledge management''. From this point of view, ''Themes render the
multitude of information meaningful by streamlining individual pieces
of information into a meaningful whole which can then be processed
more efficiently and linked to ongoing cultural concerns''
(p. 215). The last section of this book, ''Computational Approaches'',
is dedicated to this topic of text processing techniques and their
applications to thematic analysis.

In chapter 19, Hogenraad illustrates the potential of a text-mining
approach to identify themes (defined as groups or clusters of words)
in literary texts. Hogenraad also presents the results he obtained by
using the software PROTAN on Flaubert's ''L'�ducation
sentimentale''. Researches using computer programs based on
text-mining methods for theme identification in textual corpus are
upon the most promising one. Similar recent work using text-mining
techniques applied to thematic analysis has also been done by Forest
and Meunier (2000), Forest (2002) and Forest, Meunier and Piron
(2001). Martindale and West (chapter 20) also defend the thesis that
themes in texts can be identified by clustering techniques. They
describe a quantitative hermeneutical method based on the Regressive
Imagery Dictionary (used to categorize documents) and using the COUNT
computer program. In his chapter, Fortier (chapter 21) presents how an
analysis of lexical frequencies of texts (identifying prototypes or
rarity effects in word occurrences) can lead to the identification of
textual themes. Throughout the chapter, examples from Andr� Gide's
''L'Immoraliste'' are presented. In the last chapter of the book
(chapter 22), Meister focuses on the link between psychological
processes of theme identification and computational theme discovery
and analysis. Meister's aim is to find out whether or not the
emergence of themes in the mind of a reader can be identified by a
computational analysis of text.

DISCUSSION

''Thematics: Interdisciplinary Studies'', edited by Max Louwerse and
Willie van Peer, is obviously an essential book for academics from
various disciplines concerned by thematics. The book is well balanced
between theoretical and practical aspects. It also demonstrates
clearly the importance of an interdisciplinary point of view to the
study and analysis of thematics. Furthermore, it presents a good
overview of the ''classical'' perspectives as well as a more recent
computational perspective to the problem of thematics. It allows the
reader to understand the importance and the complexity of the
subject. Researchers concerned by the computational approach to
thematic analysis will find in the section ''Computational
Approaches'' very interesting methods of analysis endorsed by relevant
experiments.

On the other hand, some chapters of the book are very technical and
deal with very specific aspects of thematic analysis. Therefore, in
our point of view, these chapters can only be relevant to academics
that are already familiar with thematic analysis. To this respect, we
would not recommend this book as an introduction to this matter.

We also regret that this book does not mention Rastier's work on
thematic analysis. We think it would have been relevant to include in
this book a chapter presenting Rastier's work on thematic analysis
(Rastier, 2001; Rastier and Martin, 1995).

Nevertheless, we would definitely recommend this book to any
researcher concerned by thematics.

REFERENCES

Forest, D. (2002). Lecture et analyse de textes philosophiques
assistees par ordinateur : application d'une approche classificatoire
mathematique a l'analyse thematique du Discours de la methode et des
Meditations metaphysiques de Descartes. Memoire de maitrise, Montreal,
Universite du Quebec a Montreal.

Forest, D., Meunier, J.-G. et Piron, S. (2001). From mathematical
classification to thematic analysis of philosophical texts. Actes du
colloque ACH/ALLC 2001, 13-16 juin 2001, New York University, New
York, U.S.A.

Forest, D. et Meunier, J.-G. (2000). La classification mathematique
des textes : un outil d'assistance a la lecture et a l'analyse de
textes philosophiques. In Rajman, M. & Chappelier, J.-C. (eds.). Actes
des 5es Journees internationales d'Analyse statistique des Donnees
Textuelles, 9-11 mars 2000, EPFL, Lausanne, Suisse. Volume 1, pages
325 � 329.

Rastier, F. (2001). Arts et sciences du texte. Paris : Presses
Universitaires de France.

Rastier, F. et Martin, E. (1995). L'analyse thematique des donnees
textuelles : l'exemple des sentiments. Paris : Didier Erudition.

ABOUT THE REVIEWER

After completing a Master degree in philosophy on the application of
computer technologies to thematical analysis of philosophical texts,
Dominic Forest is now a doctorate candidate in the cognitive computer
science program at the Universit� du Qu�bec � Montr�al. Since
1996, he is also a researcher at the Laboratoire d'ANalyse Cognitive
de l'Information (LANCI). His research interests mainly concern
computer-assisted reading and analysis of texts, analysis of textual
data, text mining as well as the impacts and applications of computer
technologies to knowledge management.