"The original Survey Corpus predated modern computing. It was recorded on reel-to-reel tapes, transcribed on paper, filed in filing cabinets, and indexed on paper cards. Transcriptions were annotated with a detailed prosodic and paralinguistic annotation developed by Crystal and Quirk (1964) Sets of paper cards were manually annotated for grammatical structures and filed, so, for example, all noun phrases could be found in the noun phrase filing cabinet in the Survey. Naturally, corpus searches required a visit to the Survey.
"This corpus is now known more widely as the London-Lund Corpus (LLC), as it was the responsibility of co-workers in Lund, Sweden, to computerise the corpus" (Wikipedia article on Survey of English Usage, accessed 06-07-2010).