This research is intended to develop a technique for computerised content analysis. The goal is not to use computers as an aid in the management of information generated through manual coding, but rather to develop algorithms which permit computer coding of texts without having each text read and coded by human researchers. Manual coding is not, however, entirely eliminated. Manual coding of a subset of text(s) is employed to provide the ‘seed’ information needed to generate dictionaries of words related to specific issues. One of the objectives of this research is to determine the amount of human coding which is required in order to obtain reasonably accurate computer coding of virgin texts. For this reason, the existing stock of content-coded party manifestos is an invaluable resource.