“Communicative competence is the ability to create coherent text that

is appropriate for a given situation within a social setting. Discourse

analysis is a description of the many sub-systems that promote

coherence and the social constraints that operate on those sub-

systems.” Hatch (1992:318)

Introduction

Prior to the 1960’s, the study of language was largely focused on sentence level in

order to reveal how language works and to establish rules and constraints

accordingly. In the 1960’s and 70’s, however, linguists, psychologists and sociologists

began to assume and suggest far broader views of language and communication

(Pennycook 1994). They recognized that there is much more to language than the

identification and production of correct sentences and started to explore how

language is used beyond the sentence level to create coherence and achieve what

Hymes (1971 in Cook 1989) calls communicative competence. Out of this work in

different disciplines grew, in accordance with McCarthy (1991), discourse analysis.

Terminology

Firstly, it is necessary to define what is meant in this paper by discourse, text,

coherence, and discourse analysis.

Discourse and text

Cook (1989:156) defines discourse as “stretches of language perceived to be

meaningful, unified, and purposive.” However, Cook’s definition fails to consider

context and, in accordance with Nunan (1993), context is a key feature of discourse

as it refers to the situation in which the discourse arises and is embedded. Nunan

defines discourse as “communicative events involving language in context” (Nunan

ibid.:118). A problem with this definition is that Nunan’s use of the term “language”

is rather vague in comparison to Cook’s use. As such, an improved definition of

discourse may be realized through the amalgamation of Nunan’s and Cook’s

definitions – that is, discourse is a piece of oral or written interaction that involves

meaningful, unified, and purposive language in context.

Coherence

Coherence is the way in which discourse appears to ‘hang together’ rather than

being unrelated sequences of sentences or utterances (Nunan 1993). At a more

sophisticated level, and in accordance with Cook (Op.Cit.:156), coherence is defined

as “the quality of meaning, unity and purpose perceived in discourse.”

A crucial question regarding coherence is how can it be identified in discourse?. One

answer is the use of a top-down or “contextual” approach where knowledge of the

world, of the speaker, and of social convention is considered first before attending to

detailed features. Another solution is to apply a bottom-up or “formal” approach

where the focus is initially on language rules that operate between sentences, as well

as within them, then proceeds to more general, less definable features.

Discourse Analysis
Discourse analysis involves the identification of regularities and patterns in

language in order “to show and to interpret the relationships between these

regularities and the meanings and purposes expressed through discourse.” (Nunan

1993:7). Also, as Brown and Yule (1983) point out, an important aspect of any

discourse analysis is that the language under scrutiny should be fully contextualised

so as to enable appropriate conclusions to be drawn with regard to the language

used in that context1. In summary, discourse analysis is the study of discourse in

order to describe regularities in language with reference to context of use.

Of importance with regard to discourse analysis is the fact that there are clear

distinctions between spoken and written discourse that derive from the fact that

speech is basically transitory and writing is intended to be permanent (Brown and

Yule 1983). One of the major differences discussed by Brown and Yule is in manner

of production. A speaker may employ a variety of paralinguistic resources,

continually monitor her speech and its reception by the hearer and rapidly plan

appropriate responses that fit into the overall pattern of conversation whereas a

writer, in contrast, may review and reorder what has been written without worrying

about time or interruption aided by referral to notes and a dictionary. Another

major difference Brown and Yule discuss is difference in form. Typically, written

language is more structured than spoken language, metalingual markers and

rhetorical organisers are more numerous, syntactic forms are less repetitive, and

passive constructions are less frequent. Clearly, these distinctions will impact on

1
See Brown and Yule (1983:36) for a detailed example of how context affects the message language
conveys.
how a particular piece of discourse is analysed, the interpretation of the findings

and the conclusions made.

Reference

Referential forms are those which “instead of being interpreted semantically in their

own right … make reference to something else for their interpretation” (Halliday

and Hanson 1976:31). That is, the reader has to refer elsewhere to discover the

meaning of a word or phrase. Halliday and Hanson identify two major categories of

reference: exophoric and endophoric.

An exophoric reference is one in which referral is made to something outside the

discourse but within the context of the situation (Halliday and Hasan ibid.).

However, exophoric reference will not be considered here as, in accordance with

Halliday and Hasan (1976), exophoric reference does not play a part in cohesion and

is therefore outside the focus of this paper.

Where a reference is internal to the text, it is called endophoric and such references
may be cataphoric or anaphoric (Halliday and Hasan ibid.). Cataphoric references
point the reader ‘forward’ in the discourse to identify elements subsequently
mentioned and are often used for dramatic effect by writers to encourage the reader
to continue on reading (McCarthy 1991). An anaphoric reference directs the reader
‘backwards’ in the discourse to an element previously mentioned (Nunan 1993).

Halliday and Hasan (1976) identify and detail three sub-types of cohesive reference

– personal, demonstrative and comparative. Examples of each type2 follow.

2
For a full analysis of each reference type see Halliday and Hasan (1976:31-87).
Personal reference

Personal references are expressed by means of personal and possessive pronouns

and possessive determiners.

Although Mr. Hart has denied falling asleep at the wheel, it has been
suggested that he had been surfing the Internet for hours before
leaving his home in Strubby, Lincs, before dawn.

(The Weekly Telegraph 11-17 April 2001:3)

Demonstrative reference

Demonstrative references are expressed by means of selective and non-selective

determiners and selective adverbs.

The front of the Land Rover was struck by a Newcastle to King’s

Cross express travelling at 125 mph. This in turn was deflected into
the path of a coal train.

(The Weekly Telegraph ibid.:3)

Comparative reference

Comparative references are expressed as a “general” or “particular” comparison.

General comparison relates to a special class of adjectives and adverbs that compare

identity, similarity or difference.

Substitution
Substitution is a cohesive device in which a particular set of words replaces a word,

phrase or clause that appeared in a previous sentence (Cook 1989). Three types 3 of

substitution are proposed by Halliday and Hasan (1976) - nominal, verbal, and

clausal.

Ellipsis

Ellipsis is the omission of words, clauses or phrases that can only be recovered by

referring to an element elsewhere in the text (Nunan 1993). The process is essentially

the same as substitution in that “ellipsis can be interpreted as that form of

substitution in which the item is replaced by nothing.” (Halliday and Hasan

1976:86). That is, ellipsis is substitution by zero 4.

Ellipsis, like substitution, can be used to create ties to nominals, verbals, and
clausals5.

Conjunction

Conjunction is a cohesive device that is used in discourse to mark logical

relationships (Nunan 1993). However, conjunction, as McCarthy (1991) suggests, is

different to reference, substitution and ellipsis in that it does not refer backward or

forward to an element. Rather, its role is to indirectly signal relationships between

elements that occur in succession in the discourse.

3
For a full analysis of each substitution type see Halliday and Hasan (1976:88-141).
4
For grammatical reasons, Halliday and Hasan (1976) consider substitution and ellipsis independently. In a
subsequent publication, substitution and ellipsis are combined by Halliday (1985) into a single category.
5
For a full analysis of each type of ellipsis see Halliday and Hasan (1976:142-225).
According to Halliday and Hasan (ibid.), there are four kinds of conjunction -
additive, adversative, causal, and temporal6.

Lexical cohesion

Halliday and Hasan (1976) identify two sub-categories of lexical cohesion –

collocation and reiteration. They use the term “collocation” with regard to the way
cohesion is generated through the relationship between pairs of lexis occurring
within an identical lexical environment. Examples of collocations given by Halliday
and Hasan (1976:286) include “…bee…honey, door…window, king…crown, boat…
row…”. Their use of the term “collocation” differs from the long held notion that it
refers to, as Hoey (1991) and McCarthy (1991) suggest, the way in which lexical
items co-occur with more than random probability

Repetition, in accordance with refers to the way in which items in a sentence “ …tell
the reader or listener nothing new but reinstate some element(s) from earlier
sentences so that something new can be said about them” (Hoey ibid:268). It may be
simple or complex, achieved through paraphrasing, or categorized as superordinate,
hyponymic or co-reference repetition.

6
Halliday and Hasan (1976:226-273) give full details of different types of conjunction.