World English Corpus  

A unique modern database of over 200 million words revealing fresh information on how words are used and natural examples of English as it is written and spoken now! This is the most up-to-date corpus created from British and American written and spoken text collected from a range of media – books, magazines, newspapers, e-mails, television and radio – written and spoken text from learners of English, and ELT written and spoken materials.

Professor Michael Hoey on corpora
Adam Kilgarriff on lexicography and language technology

Frequently Asked Questions

How big is the corpus?

The corpus contains a total of around 220 million words of written and spoken text.

What are the major components of the corpus?

The World English Corpus is made up of the following:

• a British English component

• an American English component

• World English component

• a corpus of learners’ text

• a corpus of ELT materials

What types of texts are included in the different corpora?

Academic discourse, print and broadcast journalism, fiction, recorded conversations (including telephone calls), recorded business meetings, general non-fiction, answerphone messages, emails, legal texts, academic seminars, cultural studies texts, radio documentaries, broadcast interviews, ELT course books, text written by learners of English, including essays and examination scripts.

What is the ratio of the written and spoken texts in the corpus?

The ratio is about 9:1 (written:spoken).

How was the corpus used in creating the dictionary?

The corpus is our primary source of information about the way words behave. It forms the basis of our description of word meanings and of the way words combine with each other (syntactically and collocationally). It also provides information about frequency – of words, meanings, grammatical patterns, and collocations. And finally, it is the main source of the example sentences shown in the dictionary.

What types of computer program were used to get information from the corpus?

Like most dictionary publishers, we used ‘concordancing’ software to investigate word behaviour and word patterning. In addition, the MED team used new, state-of-the-art ‘lexical profiling’ software, which gives us the most detailed and most reliable information about collocations that has ever been available.

For more on how information was used from the corpora, see How the MED was Written.