Collocations

  Concordances are a useful tool for inspecting the context particular keywords appear in. But it is not a very good basis for doing a real quantitative analysis: the number of hits you get may be very high; more importantly, the results are still unordered. Of course, the concordances can be ordered alphabetically, but to start real quantitative work you to make some decisions about which occurrences are of the same type. This requires further analysis of the concordances, and this is where collocation analysis comes in.

Collocation is the occurrence of a number of words within a short space of each other in a text. For example, you might be interested in the occurrence of a particular verb with a particular preposition. What you want to do in particular is examine to which extent this collocation of verb and preposition is different from the pattern that you would have expected for those words, and so statistical and other significance measures come in.

But before we turn to these statistical measures, we'll first describe a tool for extracting collocations from text.