CALS-CPST = Colloque d’Albi Langages et -Multidisciplinary Centre de semmiolingutique textual. Electronic version: Paris: Text, 2006: http://www.revue-texto.net/Parutions/Livres-E/Albi-2006/pincemin.pdf other edition of the same text on the Internet [http://www.revue-texto.net/Parutions/Livres-E/Albi-2006/pincemin.pdf] National audience Concordances are a way of presenting text extracts, all containing the same word or language pattern. It is a secular textual analysis methodology, the very example of “benedtin work” before the possible use of computers. Matches are IT tools that produce the desired matches from a digital corpus. The subject of interest is to understand this remarkable continuity of match and to identify and analyse the discreet but decisive transformations brought about by their automated calculation. This pathway also raises the question of a more systematic and relevant use of the possibilities offered by the computer: what general application can be made of the matching method? Typically, in textual analysis software, a concordance calculation is defined by the determination of three parameters: the data of a pivot, i.e. the language word or reason whose occurrences are intended to be studied in context; the size of the context to be viewed; and a sorting criterion (possibly multiple) setting out the order in which the contexts are presented. Our generalisation is specific: rather than multiplying settings and settings, we propose to focus the adjustments on what makes the strength of the calculated matches. We therefore consider that two essential characteristics correspond to ‘good’: contexts presented on a line and superimposed in such a way as to create visual vertical alignment effects, on the one hand; and the sorting of context lines, on one or more elements, on the other hand. As a result, we clearly disregard other records of occurrences sometimes referred to as matches. Drawing very freely on designations of types of index in information science, we distinguish the KWOC (keyword out of context), a list of certificates; the KWAC (keyword and context), as we see it, with its visual grouping device by superposition, vertical alignment and sorting; KWIC (keyword in context), a mapping of contexts with a more flexible choice of their size; the KWUT (keyword up to text), where occurrences are identified along the text. These four methods of recording occurrences are complementary and gain in being grown and used for their specific characteristics. The traditional ‘paper’ match and the outputs of a match differ more than in their production method; better, everyone makes the most of the specificities of their construction process: summary work and interpretative guide for the former, regularity, versatility and dynamics for the latter. However, by different channels, manual matches and calculated matches serve the same basic hermeneutical principle: highlighting parallelism and contrasts in the contexts of the item being studied. With this in mind, for matching (KWAC), the adjustment of the size of the contexts becomes secondary (or even harmful, as it potentially distorts the match; rather, it is a feature of KWIC). On the other hand, the way in which the pivot is defined is being refined by the introduction of a decomposition into zones, which makes it possible to multiply and relax the devices for visual alignment and alignment.