La publication scientifique en langue naturelle est-elle en archéologie un discours logique? Essai de conception d’un langage cognitif d’aide à la pubblication

The project of building a cognitive framework to formalise an archaeological language, proposed here, is oriented, not to computerise any archaeological language, but to offer a tool giving a framework mainly for the formalisation and the validation of an archaeological reasoning, as well as to deliver a readable procedure, which completes the conventional natural language of the archaeological publishing. The cognitive framework is based on a decomposition of the methodological iterative procedure into three levels: 1. Acquisition, 2. Structuring, 3. Modelling, in which a cognitive grammar is defined. A cognitive grammar normally defines statements and predicates. The statements have been classified, among the more frequent archaeological statement types, which are generally, for both real and virtual objects, the results of a correlation of intrinsic and extrinsic archaeological information. The predicates are also classified following the nature of decisions they imply, either general to Human sciences, or specific to Archaeology: – identification/differentiations (generalisation of a statement at a n+1 rank), – stabilisation/destabilisations (delimiting the validity value of a statement), – exploration/renunciation (reduction of the potential ways), – paradigmatisation (hypothetical introduction of a rule at an upper level), – appropriations/disappropriations (explicit projection of the archaeologist point of view in the reasoning). The cognitive grammar is used at each of the three levels of the previously defined methodological framework. The formalisation of such a cognitive framework is materialised by a set of statement objects and predicate objects, at each three different levels. Each object may be defined as trivial (needing no more formalisation) or may be linked to another similar cognitive structure, at the origins of the decomposition of the construct into a general system of nested cognitive objects. The archaeological construct, for the scientific publishing, may be materialised by a conventional natural language, to which nested formal constructs are annexed, enabling the reader to more easily validate the logic of the reasoning. The paper is illustrated by examples of applications.