Creating dictionaries is a labour-intensive and therefore rather expensive enterprise. But computer technology and NLP research have made it possible to do some lexicographic tasks automatically. This talk will discuss the progress made in the last 20 years in automating various parts of lexicography (data collection, corpus analysis, dictionary entry writing, etc), and will look at new developments in the pipeline. We will end by considering how far this process can go: will human lexicographers become redundant, or are there some aspects of this job which still require significant human input?

By contrast with most types of dictionary (e.g. scholarly/historical, or dictionaries for advanced learners), there is no established model for a dictionary of collocations. Existing collocations dictionaries differ wildly in which lexical items they cover and what they include in the category ‘collocation’. This talk will focus on a recent project in this area (the result of which was the Macmillan Collocations Dictionary), and will look at some of the questions we faced. How far is it possible to have a ‘scientific’ approach to selecting both headwords and collocations? Is there an ideal model for a dictionary of this type?