Sketch Engine review

[mcthulhu]

The latest issue of Jost Zetzsche's Tool Box Journal has a review of Sketch Engine (https://www.sketchengine.co.uk/), which provides various options for analyzing corpora in over 80 languages and studying word usage. "The sizes of the corpora differ widely (from just a few million words in Maori to more than 800 billion in English), and they are available for a number of analysis purposes for any paying trial user (the annual subscription price is 100 euros for non-academic users, with the trial period ending after 30 days)."

(Even a few million words of Maori sounds like a lot to me. )

Per their FAQ, the free trial gives "access to the complete functionality (concordance, word sketch, thesaurus, keyword lists, word list, n-grams etc.) of Sketch Engine and also access to 100 ready-made corpora in more than 75 languages." You can also upload your own corpora and process them, which is what really interested Zetzsche. "A word sketch is a summary of a word's grammatical and collocational behavior (collocational refers to the analysis of how often a word co-occurs with other words or phrases)." This sounds similar to the Leipzig corpora search results.

The term extraction tool looks really good - you can drag and drop a file in various formats from your own computer onto the Web interface, and get back an analysis of term frequency, key terms, collocations, and so forth, which you can download. You can even generate a bilingual glossary from uploaded parallel texts; this might be useful for language learners, too.

I'm tempted to try the free trial.