Skip to main content


Cohesion is what makes us understand a text as “a unified whole rather than a collection of unrelated sentences” (Halliday and Hasan, 1974: 1). It uses five key resources: reference, substitution, ellipsis, conjunction and lexical cohesion.

In the case of reference, personal, possessive and demonstrative pronouns (as well as other devices) are the ‘phoric’ resources that denote people, places, things and actions in a way that can only be understood by reference to other textual elements. For instance, we cannot understand who ‘she’ is without referring back to ‘Lydia’ in the previous sentence:

As usual, Lydia went swimming alone.
She liked it best that way

In other words, it is presumed that we can identify who ‘she’ is (presuming reference) because she has been introduced by name earlier (introducing reference). Usually the introduction precedes the phoric item  (anaphora) but it may also follow, holding the listener or reader in suspense for a time (cataphora), and reference may also be retrieved from the immediate, situational context (exophora), or from the relevant cultural context (homophora) – for instance, people who live in Denmark are supposed to know who the Queen is so that she will not usually be introduced by name – (’the Queen’, instead of Margrethe II, Queen of Denmark).

In the case of Substitution, semantically empty items, whether nominal (e.g. one), verbal (e.g. do) or clausal (e.g. so, as in ‘he said so’), refer to earlier textual items.  We cannot, for example, understand what ‘one’ refers to without going back to ‘perch’ in the first of the two clauses:

She didn’t see any perch today.
Once she was so close to one that she pricked her hand on its fins

Ellipsis is defined as ‘substitution by zero’. In the following sentence we do not know who did the ‘waiting’ without reference to ‘she’ (which in turn refers to Lydia) in the previous sentence

She waded out until the water reached above her waist,
then waited with raised arms

Conjunction relates what is stated in one clause to what is stated in the next in terms of temporal and  logical relationships such as time, cause, condition, etc., as realized by conjunctive items such as ‘then’, ‘meanwhile’, ‘because’, ‘if’,  etc., e.g. ‘then’ in the example above.

Lexical cohesion is created by lexical relations such as repetition, collocation, synonymy, antonymy,  meronymy (whole-part relations) and hyponymy (‘kind of’ relations), as with the relation of synonymy between ‘the emerald-green deep’ and ‘the water’ in the first, or the meronymical relation between ‘body’, ‘stomach’ and ‘back’ in the second example below

She swam out over the emerald-green deep.
She enjoyed the feeling of the water bearing her

First she lay on her stomach to let the sun shine on her back.
Her whole body was very tanned already

In Halliday and Hasan’s pioneering book (1976), cohesion was, for the most part, treated as a kind of glue that makes sentences cohere into a text, although they also acknowledged the semantic function of cohesion, since cohesion makes textual elements dependent on each other for their interpretation. Their work established a distinction between cohesion and coherence in which cohesion is “understood as a textual quality, attained through the use of grammatical and lexical elements that enable readers to perceive semantic relationships within and between sentences”, while coherence “refers to the overall consistency of a discourse – its purpose, voice, content, style, and so on” (Enos, 2010: 390).

Martin’s discourse semantics (1992) further developed the theory, focussing much more strongly on the semantic role of cohesion. His approach to reference focused on tracking the various references to one and the same participant throughout the text. This then allowed such participants to be coupled to the relevant processes and circumstances, and, where necessary, put into chronological order creating ‘activity sequences’. An abbreviated activity sequence extracted from the story of Lydia’s swim would be:

Lydia   x     went swimming
She       x     ran the towel over her body    + quickly
She       x     stretched out                                       + on a flat rock
She       x     let her thoughts wander

Martin also used lexical cohesion, which was relatively underdeveloped in Halliday and Hasan (1976), to reconstruct the classifications inherent in texts, by first tracking, and then bringing together the hyponyms and co-hyponyms in a text. In this way he showed how cohesion realizes the ideational structure of the text as a whole, the underlying discourse, rather than only ‘organizing’ already formed ideational and interpersonal meanings. Van Leeuwen made ‘activity sequences’ the core of his theory of discourse (Van Leeuwen, 2008).

By right, cohesion should be an important part of the study of multimodality, as it can show how the different modes in multimodal texts cohere into integrated multimodal texts.  So far it has mostly been applied to the study of film and video. Van Leeuwen (2005) has mapped out which conjunctive relations are possible between the shots (or parts of complex shots) in film and video, and which between the verbal and visual part of the film. Tseng (2013) studied filmic reference. Clearly when, in a film, we see a shot of a hand doing something, we need to identify whose hand it is for a full understanding of the shot. But here reference is realized by various filmic ‘matching’ techniques (and of course also by text-image relations) rather than by pronouns, as Tseng discusses in detail.  Multimodal analysis has, until recently, mostly focused on writing grammars for individual modes, rather than on studying how modes are integrated into a whole. Cohesion is one promising avenue for studying such integration, not just for time-based, but also for space-based texts, where correspondences of colour and graphic shape can create cohesion between elements, in a kind of visual rhyme. Much work still remains to be done.

Citing this entry

van Leeuwen, Theo. 2015.  “Cohesion.” In Key Terms in Multimodality: Definitions, Issues, Discussions, edited by Nina Nørgaard. Retrieved


Enos, T.J., ed. (2010) Encyclopedia of Rhetoric and Composition. London: Routledge

Halliday, M.A.K. and Hasan, R. (1976) Cohesion in English. London, Longman

Martin, J.R. (1992) English Text: System and Structure. Amsterdam, Benjamins

Tseng, C. (2013) Cohesion in film: tracking film elements. London, Palgrave Macmillan

Van Leeuwen, T. (2005) Introducing Social Semiotics. London, Routledge

Van Leeuwen, T. (2008) Discourse and Practice: New Tools for Critical Discourse Analysis. New York, Oxford University Press