J. Svartvik and R. Quirk (eds): A Corpus o f English Conversation. Lund:

more than twenty years old and access to it, according to the introduction to the present volume, is ‘keenly sought by... grammarians from all over the world'. The Survey corpus comprises texts, each 5000 words in length, sampling written and spoken material with a variety of origins. Among the spoken texts are 34 recordings of conversations made without the knowledge (at the time of the recording) of the speakers, and these 34 5000-word samples are transcribed in the present volume. The London material is filed (on paper slips in filing cabinets) according to a multitude of classifications, but no attempt to reproduce this classificatory system is made here. In a collaborative project at the University of Lund, all of the spoken London material has been reproduced on a computer; the material presented here is the Lund version of the 34 texts in question. This differs from the London version in that it embodies a more simplified prosodic notation and omits represen tation of phonation type and other paralinguistic features