Colligation, lexis, pattern, and text

Editors’ introduction

One purpose in analysing a large corpus of texts, exemplified by Scott in this volume, is to focus on what the texts are about: to see the corpus as reflecting the concerns of the society which produces the texts. Another, complementary, purpose is to focus on how the texts are worded: to see the corpus as simulating the kind of constant exposure to ‘ways of saying things’ in the society which we all get and which shapes the way we ourselves use the language. This is the perspective that Hunston takes in this paper.

As she points out, repetition across texts is the foundation of a native speaker’s sense of what sounds ‘natural’ in a language. This idea has become most familiar in the study of collocation: the tendency of particular words to occur together. There is no objective reason why, for example, ‘the driver managed to regain control’ sounds more natural than ‘the driver managed to recover control’ — in isolation the meaning of ‘recover’ would make it appropriate. But when we read ‘regain control’, some kind of memory of all the other times we have read, heard or said the same combination of words (but not ‘recover control’) makes us accept it as natural — and, of course, reading the phrase adds one more example to our mental concordance for each of the two words (the image of a ‘mental concordance’ comes from Michael Hoey).

However, Hunston argues that collocation is merely one aspect of a wider phenomenon. This is the fact that words or phrases typically behave in certain ways in texts, appear in certain contexts but not others. These contexts include other words (as in collocation), but they also include features such as the grammatical company that a word keeps — the grammar patterns into which it fits — and the place in a sequence that a word prefers — for example, at the start of a sentence rather than the end, or as the subject of a clause rather than as the object. Following Hoey, who himself takes the term from Firth, she calls this kind of grammatical patterning ‘colligation’. Another key concept in this broader view is that of semantic prosody: the fact that certain words and phrases have become associated, through repeated use, with negative or positive contexts. For example, as Stubbs (1996) has noted, the verb ‘cause’ has a strong tendency to be followed by negative effects (‘cause cancer/dismay/riots’) rather than positive ones (say, ‘cause delight’). This is a kind of collocation, but it involves relations between whole groups of words of a particular type rather than between individual lexical items.

Hunston illustrates these patternings at work in a particular text, a book review. She uses a corpus to identify the main colligational patterns and semantic prosodies associated with a number of key words and phrases from the text; and she shows that the reader’s perception of the coherence of the text depends on sensitivity to these patterns — and that that sensitivity depends upon the reader having met the same patterns in innumerable other texts. The patterns sometimes operate within a clause; but in many cases they extend over relatively long stretches and thus play an important role in the organisation of the text.

What is particularly significant about Hunston’s approach is that she demonstrates that the meaning even of individual words can only be fully understood by looking at them from the perspective of text. We note in our introduction to Fries’s paper in this volume that text should be seen as the organism in which words and structures function. Like Fries and all the other writers in this collection in their different ways, Hunston is helping to re-establish the primacy of text rather than words and structures in isolation as the essential focus of linguistic enquiry.