In this paper, we present a method for the automatic extraction of terms from machine-readable text corpora. The method is based on a pattern grammar endowed with regular expressions and feature-structure unification capacity. The text corpus we have used consisted of a software manual by Hewlett-Packard extending to around 90000 wordforms, containing a term index against which the results of the method were evaluated. The method extracted 124 out of 214 manually coded terms, featuring a 58% recall.