Sociolinguistic Studies in a Data Base of Spoken Swedish (Sprakdata) is a project to describe the language from a semantic-communicative point of view. The purposes are to (1) investigate the possibility of grouping informants based on results of a semantic-communicative analysis of recorded spoken language and (2) examine possible interrelationships between these groups and social data. The language corpus consists of about 100 recordings of free conversation, comprising about 500,000 words. Detailed linguistic criteria concerning how a given message is communicated have been established for ten semantic-communicative categories. The categories emphasize modality and causal relations, attitudes toward the message receiver, and the form of the message. The corpus has been transcribed and coded for machine reading so that the linguistic analysis can be done interactively at display terminals. (MSE)