Unix for Poets

-1-

Unix™ for Poets

Kenneth Ward Church AT&T Research kwc@research.att.com Text is available like never before. Data collection efforts such as the Association for Computational Linguistics’ Data Collection Initiative (ACL/DCI), the Consortium for Lexical Research (CLR), the European Corpus Initiative (ECI), ICAME, the British National Corpus (BNC), the Linguistic Data Consortium (LDC), Electronic Dictionary Research (EDR) and many others have done a wonderful job in acquiring and distributing dictionaries and corpora.1 In addition, there are vast quantities of so-called Information Super Highway Roadkill: email, bboards, faxes. We now has access to billions and billions of words, and even more pixels. What can we do with it all? Now that data collection efforts have done such a wonderful service to the community, many researchers have more data than they know what to do with. Electronic bboards are beginning to fill up with requests for word frequency counts, ngram statistics, and so on. Many researchers believe that they don’t have sufficient computing resources to do these things for themselves. Over the years, I’ve spent a fair bit of time designing and coding a set of fancy corpus tools for very large corpora (eg, billions of words), but for a mere million words or so, it really isn’t worth the effort. You can almost certainly do it yourself, even on a modest PC. People used to do these kinds of calculations on a PDP-11, which is much more modest in almost every respect than whatever computing resources you are currently using. I wouldn’t bring out the big guns (fancy machines, fancy algorithms, data collection committees, bigtime favors) unless you have a lot of text (e.g., hundreds of million words or more), or you are trying to count really long ngrams (e.g., 50-grams). This chapter will describe a set of simple Unix-based tools that should be more than adequate for counting trigrams on a corpus the size of the Brown Corpus. I’d recommend that you do it...