Pretext is a text-crawler software that tries to find meaning in a large database of texts. It systematically browses this database to index found links between words. After linking these words, short sentences are created whereupon links between these short sentences are generated. A linguistic algorithm then arranges new, longer sentences using the similarities it encountered between related sentences in the past 50 minutes. Thus Pretext tries to find new meaning in the database as source material for a new text. The resulting machine texts balance between decisiveness and accuracy. The output can sometimes appear to be sentences of somebody who is misunderstood and is forced to explain himself in a language that is not his own. The machine’s primary language being binary after all, the software struggles with the chaotic swirls and possible misconceptions of (human) language and linguistics. After each exhibition the “Pretexts” are bundled into a thick book.