March 21, 2004

Don't say "lexeme" or we'll break your legs

David Elworthy, a natural language processing engineer who keeps his journal of Massachusetts life here, writes to me with the astonishing story of how linguist James Pustejovsky and others set up a company in 1997 to build technology based on ideas developed in Pustejovsky's book The Generative Lexicon, and called the company Lexeme, and got threatened out of it by an already existing corporation. You might like to try and guess which corporation.

Lexeme is a fairly familiar technical term in linguistics (especially morphology). It stands for the abstract entity that correlates with separate dictionary entries. Notice that in most dictionaries you can't separately look up take, taken, takes, taking, and took; they are treated as just inflectional forms of a single entity take. That is the lexeme. The forms that are not predictable are listed at the beginning of the entry for the lexeme. The Cambridge Grammar of the English Language uses the distinction between word forms and lexemes throughout, and distinguishes the two notationally: word forms like taken in plain italics and lexemes like take in bold italics. It is hard to see how we could have written the book without the distinction in question.

So which company used its lawyers to threaten Pustejovsky's fledgling company for trying to register this anodyne and unambiguous lexical item as a name? One might have thought it would be Lexis-Nexis, who at least have business in the area of handling text. But no: the legal challenge came from Lexus, a division of Toyota making luxury cars. And they assert a right to all names that begin with the letters l e x, apparently.

Rather than waste money on a legal fight when it wasn't even born yet, Lexeme caved, and changed its name to LingoMotors. The company once held the URL, but that seems no longer to exist (maybe the Lexus company still wasn't satisfied with the name change and hired some ugly guys to come round and threaten them completely out of business). But what a piece of insanity. Why on earth should a luxury car company have the right to prevent a small natural language processing and search technology company from adopting a technical term of linguistics as its company name? Aren't the owners of the English language ever going to rise up against greedy corporations like Lexus and Microsoft and Star bucks who lay claim to whole regions of the phonetosphere as if their financial power gave them arbitrary dominion over any set of possible words they take a fancy to, when no use of their trademark is being made and no possible confusion threatens?

Posted by Geoffrey K. Pullum at March 21, 2004 08:39 PM