Halliday and Hasan (1976): Cohesion analysis

The purpose of Halliday and Hasan’s (1976) analytical model is to investigate the resources that English has for creating texture. They claim that texture is provided by instances of cohesive relation between two items in the text, and they call the occurrence of a pair of cohesively related items a “tie”. With the help of the concept of tie, texts are analyzed in terms of their cohesive proper-ties, giving a systematic account of their patterns of texture. Such analyses of the quantity and quality of cohesive ties between sentences may offer insights into the difference between spoken and written texts, the peculiarities of various genres, and personal and cultural differences in both speech and writing. The five types of cohesive ties are reference, substitution, ellipsis, conjunction (all four being cohesive relations found in the grammar), and lexical cohesion.

Reference: Reference items, as explained by Halliday and Hasan (1976), are linguistic elements which make reference to something else for their interpreta-tion, therefore are not interpreted semantically in their own right. In English there are three main groups of reference: personals (personal pronouns, pos-sessive pronouns, pospos-sessive determiners), demonstratives (e.g., this, that, these, those, here, there, the, then), and comparatives (adjectives and adverbs expressing identity, similarity, or difference, such as the same, similar, different, etc.).

Within the category of reference, Halliday and Hasan distinguish two major types: (1) exophoric or situational reference (referring to a thing as identified in the context of situation) and (2) endophoric or textual reference (refer-ring to a thing as identified in the surrounding text). Endophoric reference can also be of two kinds; it may refer back to the preceding text, or refer to the following text. The first one is called anaphoric, while the latter is called cataphoric reference.

Substitution and ellipsis: Substitution and ellipsis (both grammatical cat-egories) cover very similar processes within the text; what is more, they can be regarded as the same type of cohesive relation that takes two different forms.

Substitution is the replacement of one item by another, and ellipsis is the omission of an item. Therefore ellipsis can also be interpreted as zero substitution, i.e., substitution by nothing. Because in English the substitute may function either as a verb, a noun, or a clause, three main types of substitution can be distinguished:

verbal (do), nominal (one, ones, same), and clausal (so, not).

Conjunction: The fourth type of cohesive relation that can be found in the

81 Discourse analysis and language teaching

grammar is conjunction. What distinguishes this category from the first three types of relations is that conjunctions are not cohesive in themselves, but their cohesive force comes from their specific meanings. Hence in describing con-junction as a cohesive device, Halliday and Hasan concentrate on its function of relating to each other linguistic elements that occur in succession but are not related by other, structural means. Conjunctions can be divided into five main classes, depending on the quality of connection they make:

(1) additive (e.g., in addition, besides, in other words, that is, alternatively),

(2) adversative (e.g., however, but, instead, nevertheless, as a matter of fact), (3) causal (e.g., so, then, hence, because, it follows, to this end, as a result), (4) temporal (e.g., then, next, finally, meanwhile, in conclusion, up to now), (5) continuatives (e.g., now, of course, well, anyway, surely, after all).

Lexical cohesion: As opposed to the various types of grammatical cohesive relations (reference, substitution, ellipsis, and conjunction), lexical cohesion is the cohesive effect achieved by the selection of vocabulary. Its two main classes are reiteration and collocation. Reiteration is defined by Halliday and Hasan (1976) as “a form of lexical cohesion which involves the repetition of a lexical item, at one end of the scale; the use of a general word to refer back to a lexical item, at the other end of the scale; and a number of things in between - the use of a synonym, near-synonym, or superordinate” (p.278).

Collocation, on the other hand, refers to cohesion that is achieved through the association of lexical items that regularly co-occur. This is the most prob-lematic component of Halliday and Hasan’s analytical model, since it is based on intuition and thus is a very fuzzy category. They claim, however, that, in general, any two lexical items tending to occur in similar contexts will generate a cohesive force if they occur in adjacent sentences. Therefore the category of collocations may include synonyms and near-synonyms (climb - ascent), superordinates (elm - tree), antonyms (like - hate), converses (order - obey), part-whole relations (car - brake), co-hyponyms (chair - table), and “proximity”

relations (laugh - joke, garden - dig) which Halliday and Hasan explain as the cohesive effect deriving from the occurrence in proximity with each other.

A major advantage of Halliday and Hasan’s (1976) study is that they not only describe their cohesion model extensively, but also work out systematically the methodology of analyzing texts with the help of this analytical tool: they review the principles of analysis, provide a coding scheme for the various types of cohe-sion, and offer a detailed sample analysis of seven short passages of text.

82 Krisztina Károly 2.3.2. Hasan (1984): Cohesive harmony

Hasan (1984) is concerned with placing the concepts of coherence and cohesive harmony in relation to each other. She describes coherence as

“unity” or “hanging together,” by which she means that any text is coherent

“to the extent that its parts hang together” (p.181). She also emphasizes that coherence in a text stands in some relation to the extralinguistic universe, and thus she claims that meaning cannot be studied without the complete context. In Hasan’s interpretation, due to the fact that coherent texts “hang together,” the patterns of language manifest the existence of semantic bonds between one element and some other element in the textual environment.

In her analysis, she started out from the cohesive mechanisms identified by Halliday and Hasan (1976), but realized that they describe merely “sur-face” phenomena, the “lexico-grammatical categories” (Hasan, 1984, p.186) and as such, have only little or nothing to do with “deep” phenomena, i.e., semantic relations. The way they are related is that cohesive devices (e.g., reference, substitution and ellipsis, lexical cohesion) realize parts of indi-vidual messages (the category of cohesive conjunctive is different because they are overt indications of logical relations between two messages), and the integration between them is a product of the cohesion between specific parts of individual messages.

Hasan, in her initial analyses of coherence, encountered numerous prob-lems. First, her initial hypothesis regarding coherence was that (1) the larger the number of cohesive ties in a language piece, the greater the coherence and (2) the greater the continuity of ties relating to each other, the greater the coherence. Her analysis, however, invalidated these hypotheses, showing that coherence does not correlate with variation in either the number of cohesive ties or their proportion in chains.

The second major problem Hasan faced concerned the analysis of lexical cohesion in terms of the 1976 model. The difficulty arose from the fact that (1) due to the lack of a precise definition of the categories of lexical cohesion, their boundaries were unclear and thus produced ambiguity during the process of identifying the ties and (2) the existing categories of lexical cohesion failed to take into account certain semantic bonds, and (3) the separation of lexical and grammatical cohesive chains did violence to certain aspects of the text’s semantic organization. The analysis being predominantly quantitative, these problems caused major inconsistencies concerning the findings. Therefore the theory had to be reconsidered, which resulted in the revision of lexical cohesive categories and the definition of the concept of cohesive harmony.

83 Discourse analysis and language teaching

Hasan (1984) claims that lexical cohesion seems to belong to two main types:

(1) lexical cohesion mediated through “general” lexical relations, which are based on supratextual semantic bonds, with a “language-wide validity” (p.201) and (2) lexical cohesion mediated through “instantial” lexical relations, which are, by contrast, text-bound. Table 1 shows Hasan’s revised system of lexical cohesive categories, which allows for a more precise definition of categories and thus helps produce a more consistent analysis.

A. General

travel, leave (including co-hyponyms: leave, arrive) hand, finger (including co-meronyms: finger, thumb) B. Instantial

1. equivalence 2. naming 3. semblance

the sailor was their daddy; you be the patient, I’ll be the doctor the dog was called Toto; they named the dog Fluffy

the deck was like a pool; all my pleasures are like yesterdays Table 1: Categories of lexical cohesion revised (Hasan, 1984)

Hasan stresses that the principles of chain formation should permit the chain’s valid relation to the system and − at the same time − permit the chain’s relation to the text as a process. Thus, she distinguishes two main types of chain, which are both present in normal texts: the “identity chain” (IC), the members of which are held together by the semantic bond of co-referentiality and, as such, are text-bound, and the “similarity chain” (SC), which is not text-bound, and the semantic bond between its members is either coclassification or coexten-sion). Using these concepts, the tokens (cohesive elements) of each text may be represented in a chart and the ICs and SCs can be identified.

Tokens which are not parts of chains are called “peripheral tokens” (PT) and tokens subsumed in chains are referred to as “relevant tokens” (RT). While PTs are not crucial concerning textual meaning, RTs are relevant because they are cohesively related to each other, and they are also related to the topical develop-ment of the text. Based on this, Hasan goes further and claims that individual chains interact with each other, and as a result, a significant difference to the unity of the text is created. Therefore, RTs which participate in interaction must be distinguished, and she calls them “central tokens” (CT).

Hasan’s (1984) analysis shows that the relationship between cohesive har-mony and coherence can be summed up as follows:

84 Krisztina Károly

any text will be seen as coherent, in which the central tokens (CT) form (1) at least 50 percent of the total tokens (TT). This percentage may be

treated as a measure of their cohesive harmony;

ranking by cohesive harmony will match the ranking of the texts on (2) the cline of coherence by reference to informal reader/listener reaction;

cohesive harmony is the lexico-grammatical reflex of the semantic fact of coherence;

if two texts display no significant difference in their cohesive harmony, (3) variation in their coherence will correlate with interactive gaps: all else being equal, the larger the number of such gaps, the less coherent the text would be;

whereas the percentage of PT over RT or TT is not significant for (4) coherence [...], the ratio of CT to PT is associated with coherence: the higher the ratio of CT to PT, the more coherent the text will be, all else being assumed equal. (p.218)

In conclusion, Hasan’s (1984) analysis clearly shows that merely a clause by clause analysis of a text will not reveal the nature of coherence in it.

Instead, what is needed is mapping the functional relations of lexical to-kens which are united by their internal similarities. The echoing of these functional relations, she claims, “becomes a powerful device, but this happens only because the members of the chains echo one another in any event. Herein lies the rationale for insisting that the minimal condition for chain interaction is the ‘echo’ of a functional relation” (p.219). This means that we can only talk about chain interaction if at least two members of a chain stand in the same grammatical relation to at least two members of some other chain.