ABSTRACT
One of the unseen revolutions currently taking place as a result of the development of large linguistic corpora and databases concerns the way linguists have to think about data. Until very recently, it was desirable when representing the spoken language to compress as much information as possible into the simplest possible symbol. The virtues of this approach include readability and economy in printing, as can be seen in conventional phonemic and prosodic transcriptions, but it is not in principle a matter of linguistic theory. There is no reason in the 1990s to continue to solve the problems of paper-and-pencil technology.
