Pourquoi (et comment) l'historien doit-il compter les mots ?
Alain Guerreau · Histoire & Mesure · 1989
Alain Guerreau. Why (and how) must historian count words ? The proceedings of the international conference held in Nice in June 1985 on quantitative and computer methods for the study of texts display a great diversity and reveal that the historian must study frequencies through their variability, and through semantic structures. Lexical statistic make use of various instruments to study the variability of frequencies : first the « rank-frequency » law (Zipf law), generalised by B. Mandelbrot and which may be considered as a Pareto law. This means that the conceptual system is itself a fractal object and this fact may reactivate this field of research. Studies on variability of frequencies show that there is no adjstment to the hypergeometrical model and that text length and the speaker's idiosyncrases are to be taken into account ; since speaking is the social act par excellence. For study semantic structures, one must use semantic networks but some computers tools must be devised ; making the variation of the values of coefficients possible, they will make the relationship between text and production of meaning easier to understand. It is most informative for the historian to confront the numerical constants of a text with its semancic structures : this approach of the text ought to open new vistas to historical research.