Weblinks - Corpora

Sometimes you may wonder about points of grammar or usage that seem to be variable. In such cases, it is often useful to check your intuition against a corpus, to see what people actually do in real writing and speech. Some corpora can be searched online, though usually with some limitation on the results that are displayed. For example, the 100 million-word British National Corpus can be searched at: thetis.bl.uk (select 'Simple Search'). It displays a maximum of 50 results, though it also tells you how many hits it has made in total for the item you asked it to search for.

A 56 million-word subcorpus of the Bank of English can be searched online at: www.cobuild.collins.co.uk/form.html. Up to 40 results are displayed. A series of smaller corpora, including the Lancaster-Oslo/Bergen Corpus, the Brown Corpus and a number of newspaper corpora, can be searched on the Hong Kong Polytechnic University's Virtual Language Centre website: vlc.polyu.edu.hk. The number of results is unrestricted, and the site contains a large amount of other interesting material, aimed mainly at teachers and learners of English as a foreign language.

The Survey of English Usage site at University College London has details of the International Corpus of English project and a downloadable sample of the GB version of the corpus, which includes considerable analysis of each sentence: www.ucl.ac.uk/english-usage.