The Osservatorio of Pavia publishes the report of a lexical analysis of the US Presidential debates (pdf), that represents the words used on a very clear graph of Simple-Complex and Rational-Emotional axes.
The frequency of the words used, and their grouping into clusters of meaning let the authors of the report show the main themes covered, and ethical and moral values expressed in an objective and scientific manner. The report is based on an algorithmic lexical analysis called Alceste 4.0, which can be applied to any text.
Is this a reproducible approach to semantic classification of non-structured human discourse? What could the linking of this analysis to Google’s APIs bring us to? Could a thorough system of memetic classification emerge?