Thesis
French
ID: <
2268/204408>
Abstract
In the field of linguistics, knowledge is often represented by the means of edited data (textual editions or structured databases) and diagrams. These representations can be called inscriptions of knowledge. They suggest three core questions. First of all, how to read them? The first chapter investigates the conventions at use to represent and organize information in edited data and syntactic diagrams. Secondly, how to choose what has to be inscribed? This chapter focuses on how only a part of data and analyses is represented in the inscriptions. An operation of reduction is always necessary and it accompanies the choices about what has to be represented. The choice of the formal structure used for encoding the inscription acts as a major constraint on what can and what cannot be inscribed. Thirdly, how do inscriptions allow us to discover more? This chapter is centered on the use of diagrams to unveil new knowledge about the facts and the theory. It is also shown that diagrams make it possible to investigate the relationships between formal structures themselves. A fourth chapter investigates digital inscriptions, showing how the use of the computer allows us to reevaluate the answers to all the aforementioned questions.