The interweaving of three different sorts of software based on different algorithms (co-word analysis and downward hierarchical classification) and applied on a file (in the field of risk assessment through the introduction of transgenic plants) extracted from the CAB (Commonwealth Agricultural Bureau) data base, has enable us to provide three types of results: Leximappe provides a synthetic image from clusters of key-words. The main themes were identified. Alceste improves a corpus' characterization and allows a logical reading of it, thanks to the creation of categories, along with their mutual dependencies, the peculiar meaning of each and their division in time. Moreover, Alceste allows us to perceive the contexts of the contents previously identified under Leximappe. Sampler allows us to go into the details of the terms association in graphical form and detail the specific orientations of the corpus, especially with the inscription of weak signals. Finally, this software, applied from the categories drawn from Alceste, offers for each category a meaningful graphic representation. We can argue that the different ways of measuring and presenting results are complementary since they highlight different aspects of risk assessment carried by different actors, as it is underlined in social science studies of public controversy. Moreover we can follow these actors through the categories and clusters (socioeconomic, scientific and risk assessment linked to regulation and policy) which are more and more differenciated in time. This methodology allows the study of emerging processes in the social construction of issues within controversies.

No abstract