Collocations

The patterns of lexis are called collocations. Collocations, like the structures of grammar, consist of ‘things’ occurring one after another in a sequence. For instance just as the structure exemplified by the clause The waitress spilt the soup on the tablecloth consists of things — the elements S, P, C, A – occurring one after another in a sequence, so too the collocation exemplified by the same utterance consists of things — the items The, waitress, spilt, the, soup, on, the, tablecloth — occurring one after another in a sequence. Similarly, just as the structure exemplified by the clause I have read the first fifty pages of that book consists of the elements S, P, C, occurring one after another in a sequence, so too the collocation exemplified by the same utterance consists of the items I, have, read, the, first, fifty, pages, of, that, book, occurring one after another in a sequence.

There are, however, important differences between the collocations of lexis and the structures of grammar. One difference between them lies in the part played by sequence in the formation of the patterns. Although, as has been shown, the ‘things’ which make up both kinds of pattern occur sequentially, the position of each ‘thing’ in the sequence is unimportant for a lexical collocation, though important for a grammatical structure. The conversation

A. Read the first fifty pages of that book.

B. I have.

exemplifies different grammatical structures from those of the utterance I have read the first fifty pages of that book, but exemplifies exactly the same collocation. [...]

Another difference between the collocations of lexis and the structures of grammar lies in the generality of the things which make up the patterns. An element of structure is an abstraction (that is, a generalisation) from a large number of similar things. For instance the element subject is a generalisation from such items as Theodore, Tiddles, They, Alexander and He in the clauses.

Theodore seems fond of cats, Tiddles scratched Aunt Jemima. They like cream.

Alexander plays the flute.

He is musical.

These items share certain properties and therefore a general label can be applied to them all. An item of a collocation, on the other hand is a particular unique thing. For instance if, instead of considering Theodore, Tiddles, They, Alexander and He as items realising elements of structure, we consider them simply as items of collocations, we find that each is unique. Their shared properties are no longer relevant. It is the properties which make them unique which enable them to enter into collocations. Because each is unique, there is no way in which we can link them together and make a general statement about them. Theodore is Theodore, Tiddles is Tiddles, They is They, Alexander is Alexander and He is He. That is all we can say about them.

Just as the statements made about items of collocations are less general than the statements made about the elementsof structures, so too the statements made about collocations themselves are less general than the statements made about struc­tures. We can say of the five clauses cited in the previous paragraph that they are all alike in that they have the same structure, SPC. From a lexical point of view thefive statements have nothing in common at all. If we add to these five utterances a sixth The cat who likes music scratched Alexander, we can say that, lexically, each of the first five utterances has something in common with the sixth, in that each as a collocation which shares at least one item with the sixth. This kind of likeness between collocations is a much less general kind of likeness than the kind of likeness which exists between structures consisting of the same elements.

Because collocations and their items are less general than grammatical structures and their elements, they cannot be studied in quite the same way.

A greater amount of data is necessary to produce worthwhile results in a lexical study than in a study of grammatical structure. Because an element of structure is more general (can be realised by more items), than an individual item (which can be realised only by itself, though “realised” is perhaps the wrong term here), it can be found more frequently in a text. For instance, if a study were being made of the element predicator in a text, it would be a fair assumption that a predicator could be found in each of most of the clauses in the text. On the other hand, if a particular item, such as spill, were being studied, this item would certainly occur less frequently than the element predicator and might even occur only once or twice in the text. Thus in order to study a number of occurrences of the item comparable with thenumber of occurrences of the element, it would be necessary to study other texts in addition to the one text which would on its own yield worthwhile information about the element.

Because each element occurs more frequently than each item, a native speaker of a language has a greater experience, conscious or subconscious, of the relationships which exist between two items. For this reason a linguist studying grammatical structure can rely more onhis own intuitions than can a linguist studying collocations resort more quickly to the use of statistical techniques in order to obtain more objective verification of their observations.

The relationships between the items of a collocation are, in fact, usually expressed in terms of statistical probability. The relationship between any two items is the probability (that is, the greatness of the likelihood) that one item will occur with the other. Some pairs of items have a strong probability of co-occurrence relationship, while other pairs have a weak probability of co-occurrence relationships. There is a greater likelihood, for instance, that cat will occur with scratch than that cat will occur with music, cat and scratch havea relatively strong probability of co­occurrence relationship, while cat and music have a relatively weak probability of со-occurrence relationship.