2

I'm looking for some good works of corpus linguistics using the Brown Corpus. Does anyone know of an important early work, and/or an important recent work that uses it?

3

The Brown corpus, which is a representative 1 M. word sample of 1960s American English, has been used extensively in research on short-term diachronic change. A number of corpora have, over the last one or two decades, been compiled that extend the Brown corpus. There is a 1990s American English corpus (Frown), and a 1960s and a 1990s British English corpus (LOB and F-LOB). Together, these have been called the Brown family of corpora.

Research based on these corpora includes:

More recently, 1930s corpora of British and American English have been compiled to add to the historical depth of the Brown family, and work has begun on additional corpora.

2

The key early works based on the Brown corpus are, of course, Kucera and Francis' 1967 work "Computational Analysis of Present Day American English" and the American Heritage Dictionary. It's greatest long-term impact was in inspiring the creation of other corpora like LOB (Lancaster-Oslo-Bergen) which then led to some innovative work on grammars.

You will find very little use of the Brown corpus in modern corpus linguistics research. There are simply too many far better resources. A modern respectable corpus of written language will start at 100 million tokens (as opposed to Brown's 1 million) and try to go much higher. It is sometimes used for teaching purposes and I still see it used for some basic frequency work in non-specialist works - this is probably in part because of its easy accessibility.

2
  • 1
    Yes, its often beneficial to have larger corpora. But: "You will find very little use of the Brown corpus in modern corpus linguistics research." - No. "Change in Contemporary English" appeared in 2009, has been cited 239 times up till now, and is based on the Brown family. There are many research questions for which larger corpora are not available. – robert Aug 29 '15 at 9:30
  • 1
    You're right @robert, I wasn't thinking about this broadly enough. What I had in mind was use of the Brown corpus on its own for synchronic research. That would indeed be only suitable for very rare uses. I also perhaps overemphasized the small size. For many uses (e.g. where detailed manual annotation is required), a million or fewer tokens have to be enough or are even better. However, your examples notwithstanding, I'd argue that the Brown corpus (or LOB) is little use in modern corpus linguistics. – Dominik Lukes Aug 30 '15 at 9:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.