Text Typology and Statistics Explorations in Italian press subgenres

According to Biber’s definition, text types are represented by groupings of texts which are similar in their linguistic form, while genre categories are assigned on the basis of use. It is important to stress that texts from a single genre might be classified into different text types. In the present research, an experiment will be carried out to single out different text typologies among Italian press subgenres only on the basis of morphosyntactic features. The approach will be corpus-based and the aim merely exploratory. A virtually random sample will be extracted from two Italian tagged corpora , the sample divided into 22 files, a file for each subgenre, and 41 morphosyntactic features will be counted per article within each subgenre. The raw frequencies will be normalised. The dataset built from the normalised frequencies (quantitative variables) will be submitted to Descriptive Statistics and Factor Analysis. Descriptive statistics give information about the distribution of the variables. It is a preliminary step in any exploration and gives a better understanding of the set of data. Factor analysis studies the correlations among a large number of interrelated quantitative variables by grouping these variables into a small number of factors, which help to understand the structure of correlations or the underlying construct. The aim of this research is to investigate to which extent statistical techniques can help in classifying texts in a steady and reliable way.