Textual Relations: A Case Study in Bengali From the Viewpoint of RST

.........................................................................................................................

आभ्यंतर (Aabhyantar)      SCONLI-12  विशेषांक         ISSN : 2348-7771

.........................................................................................................................

2. Textual Relations: A Case Study in Bengali From the Viewpoint of RST

Abahan Datta, Soumya, Shankar Ghosh : School of Language and Linguistics, Jadavpur University

Abstract

This paper wants to investigate the relational network in which information is situated. Brief news articles will be taken as example. Through this way, investigation of the networked relations can be explained. Network of the relations, then, will be represented with the help of Rhetoric Structure Theory to explain how meaning is construed in a discourse which is larger than a sentence.

1. Introduction and Research Question

“Rhetoric may be defined as the faculty of observing in any given case the available means of persuasion.” (Aristotle) So it can be said, rhetoric is a kind of text used to circulate some opinions. And if we define Rhetoric Structure Theory (RST): it is a descriptive theory of a major aspect of the organization of natural text. It is a linguistically useful method for describing natural texts, characterizing their structure primarily in the terms of relation that holds between two parts of the text.

This paper seeks to investigate the relational network in which information is situated. Newspaper articles will be looking at to investigate the way relations are networked. Network of the relations, then, will be represented with the help of Rhetoric Structure Theory to explain how meaning is construed in a discourse which is larger than a sentence.

2. Literature Review

Exploring relational network of a discourse gets its momentum in Discourse Representation Theory or DRT (1981). In formal linguistics, DRT is a framework for exploring meaning under a formal semantic approach. It concentrates primarily on various types of anaphoric relations which are crucial in establishing coherence and cohesion across the sentence boundaries. Though this approach has its root in the traditional logic, it is often claimed that it is in some way different from the Montagovian approach.

Rhetoric Structure Theory (hereafter, RST) has also its impact on the analysis and the representation of discourse relations. It concentrates on how a coherence text is well formed from the viewpoint of its relational structure and how this structure can be parsed to have a sense of the internal constituencies. This, in turn, will enhance our understanding of the way meaning is constructed both syntactically and semantically at the level of discourse.

Several studies use RST as a descriptive framework for investigating linguistic issues. Some of the studies are the following (Mann & Thompson, 1987):

a.       RST provides a general way to describe the relations among the clauses of a text. They may or may not be grammatically or lexically signalled. Thus it is an useful framework for relating the meanings of conjunctions, the grammar of clause combining and non-signalled parataxis.

b.       Descriptive RST has been used as an analytical tool for a wide range of text types. It can be used to explain various kinds of expository texts. Thus it can be a tool to characterise the features of news broadcast.

c.       Descriptive RST lays a foundation for the studies of contrastive rhetoric.

d.       RST is proven to be useful to analyse narrative discourse. It is valuable to describe grammatical as well as narrative properties produces by different kinds of speakers.

e.       RST provides a framework for investigating relational propositions, which are unstated but inferred propositions that arise from the text structure in the process of interpreting texts. Since the coherence of a text depends in the part of these relational propositions, this theory has been useful in the study of text coherence.

RST had been originated in the 1980s. Since then it enjoys attention, especially from computational linguistics. Because RST was used to plan coherent text and to parse the structure of text. It was also used in approaches like discourse analysis, theoretical linguistics and psycholinguistics.

2.1 Theoretical Framework

As a descriptive framework for text, RST provides a combination of features that is useful in several kinds of discourse studies. It can identify the hierarchic structure that is present in a text. It describes the relation of text parts in functional terms. RST can also provide a comprehensive analysis rather than some selected description. Empirical evidences show that it is insensitive to the text size. It can be applied to a wide variety of sizes of texts.

Theoretical ingredients of RST are (a) relations, (b) schemas, (c) schema applications, and (d) structures. Relation is defined by the particular relationship between two constituents of a text. Often these constituents are discrete, hence non-overlapping (see figures). These constituents are of two types namely nucleus (represented with vertical lines) and satellite (represented with curve lines with arrows).Furthermore, relation consists of following four fields: (1) constraints of the nucleus, (2) constraints of the satellite, (3) constraints of the combination of nucleus and satellite, and (4) the effect. Each field specifies particular judgement that the text analyst must make in building the RST structure. According to the analysis, these are only plausibility. In case of effect, analyst judges whether it is plausible that the writer desires the specified condition. Schemas define pattern in which a particular span of a text can be analysed in respect of another span. In other words, it defines the structural constituency arrangements of text. They are abstract patterns consisting of a small number of constituent text spans, a specification of relations between them and a specification to certain spans related to the whole collection. RST recognises five kinds of schemas. Those are represented by the five diagrams. The notion of the structure of an entire text is defined in terms of composition of schema applications. Three conventions determine the possible applications of a schema: (a) unordered spans (b) optional relations (c) repeated relations. Unordered spans are the schemas that do not constrain the order of nucleus or satellites in the text spans in which the schema is applied. Optional relations are for multi-relational schemas, all relation are optional, but at least one of the relations must hold. Repeated relations are a relation that is part of a schema can be applied any number of times in that application of that schema.

       Fig 1            Fig 2                 Fig 3                        Fig 4                              Fig 5

This is to mention that all the figures follow the simple pattern represented by a single relation with nucleus and satellite. The schema names for these are the same as the corresponding relation name.

Now, the very first step of analysing a text is dividing it into units. Unit size is arbitrary, but the division of the text into unit should be based on some theory-neutral classification. That is, for interesting results, the units should have independent functional integrity. In the analyses, units are essentially clauses. These units are then classified either as nucleus or as satellite depending on the nature of their respective participations in a given relation.

A structural analysis of a text is a set of applications such that the following constraints hold: (a) completeness, (b) connectedness, (c) uniqueness and (d) adjacency. Completeness is the set contains one schema application that contains a set of text spans that constitute the whole text. Connectedness is except for the entire text as a text span, each text span in the analysis is either a minimal unit or a constituent of another schema application of the analysis. Uniqueness is each schema application consists of a different set of text spans, and within a multi-relational schema each relation applies to a different set of text spans. Adjacency is the text spans of each schema application constitute one text span.

Before analysing the structural theory through empirical evidence we have to present the defined relations by name and representative sample of definitions. The following shows the defined relations grouped according to a specific kind of resemblance.

Organization of relational definitions:

Circumstance                                                                      Antithesis and Concession

Solutionhood                                                                                   Antithesis

Elaboration                                                                                     Concession

Background                                                                                    Condition and otherwise

Enablement and Motivation                                                             Condition

      Enablement                                                                               Otherwise

Motivation                                                                    Interpretation and Evaluation

Evidence and Justify                                                                       Interpretation

      Evidence                                                                                  Evaluation

      Justify                                                                          Restatement and Summary

Relations of Cause                                                                          Restatement

      Volitional Cause                                                                       Summary

      Non-Volitional Cause                                                    Other Relations

      Volitional Result                                                                       Sequence

      Non-Volitional Result                                                               Contrast

      Purpose

In this chart, each relation definition is accompanied by the analysis of a natural example of its occurrence.

Here follows an example from Mann & Thompson (1987):

If we consider that Evidence and Justify form a subgroup, then both involves the reader’s attitude towards the nucleus. Evidence satellite is intended to increase the reader’s belief in the nuclear material. As an example of Evidence relation, following units are taken:

1.       The program as published for calendar year 1980 really works.

2.       In only a few minutes, I entered all the figures from my tax return and got a result which agreed with my hand calculation to the penny.

                     Fig 6

The RST diagram in the figure shows unit 2-3 in an Evidence relation with unit 1. They are provided to increase the readers belief in the claim expressed in unit 1.

Justify relation a sentence can be taken:

1.       The next music day is scheduled for July 21 (Saturday), noon-midnight.

2.       I’ll post more details later,

3.       but this is a good time to reserve the place on your calendar.

In this text, unit 2-3 are in Justify relation with unit 1. They tell readers why the writer believes he has the right to say unit 1 without giving more details, in particular without giving the location. The figure will look like following:

                   Fig7

3. Data Analysis

First we should discuss that what kind of text we are taking for analysis. The following task will be performed over a text called ‘news brief’. What is that? A broadsheet newspaper is generally designed in eight columns. It contains some brief news in the leftmost or the rightmost column. They are typically restricted into single column and consist of 5 to 10 sentences. The space allotted to the briefs is termed as Doric column for having symbolic resemblance with an archaic form of architectural order developed in Greece and Rome. If a Doric column contains three or four brief news, then they are ordered according to their significance. This may vary from one newspaper to another depending upon their editorial policies. And the most important point for this paper is that the briefs are structured.

The framework that we have presented in the previous section will be discussed in this section in the light of RST analysis of a Bangla text. The text mentioned in example (1) is taken form a Bangla newspaper Anandabazar Patrika, dated 18 November 2017.

1.

kɔmpiuʈar

niontrito

inʈarlɔkiŋ

bæbostha

calu

hocche.

khɔrogpure.

computer

controlled

interlocking

system

start

be-CONT-3

kharagpur-LOC

tai

bohu

ekspres

o

lokal

ʈren

batil

kɔray

bhug-te

that’s why

many

express

and

local

train

cancel

do-CAUS-NF

suffer-INF

holo

jatri-der

balicɔk

theke

tãra

jate

ʈren

dhorte

be-PST-3

passengers

balichok

from

they-HON

so that

train

catch-INF

paren,

sei

jonno

barti

baser

aʃʃaʃ

deoa

hoechilo.

can-PRS-3 HON

that

reason

extra

bus-POSS

assurance

give

be-PRF-PST-3

ʃukrobar

ʃɔkale

ɔboʃʃo

bas

mele

ni.

phɔle

durbhoge

Friday

morning

but

bus

get-PRF-PRS

NEG-PRF

result

woe

pɔren

jatrira.

fall-PRS-3 HON

passengers

In Kharagpur computer controlled interlocking system is on the edge of starting. Passengers have suffered a lot as many express and local trains have been cancelled due to this reason. Even though authority has assured that they will provide more buses in order to ensure that passengers can catch the train from Balichok station but they didn’t get any bus on Friday morning. As a result of which passengers fall into a deep trouble.

In order to analyze this text we will first determine the elementary discourse units (EDUs) which are the minimum building blocks of the discourse tree. Having done that we will establish the inter discourse relations that these EDUs posses. Hence we will get a discourse sub-tree of the mentioned text as represented below: 

Figure 8

The entire RST structure as presented in Figure 8 seems to indicate that kɔmpiuʈar niontrito… hocc^he is somewhat more central for the whole stretch of text as it includes the nucleus of both purpose and antithesis relation on a rather high level of the RST hierarchy.

4. Discussion

In this point of the discussion, it becomes important for the paper to explore the significance of RST in construing meaning in the level of a connected written discourse. A close look on the RST structures, presented in Figure 8, we will see that a complex knitting structure among the EDUs make information to travel from micro level to a macro level. By doing that it depicts a plausible account of what the writer wanted to achieve with each part of the text. Therefore, within the broader theoretical scope meaning making and information building can be visualized as an integral domain that consist of (i) basic discourse units, i.e. EDUs and (ii) discourse relations that these EDUs possess.

5. Bibliography

·        Maite Taboada and William C. Mann. 2006. Rhetoric Structure Theory: Looking Back and Moving Ahead

·        William C. Mann and Sandra A. Thompson. 1987. Rhetoric Structure Theory: A Theory of Text Organization

·        Lynn Carlson and Daniel Marcu. 2001. Discourse Tagging Reference Manual