In the field of education, especially in primary school, we can favor the development of creative thinking and written expression of the first grades pupils, using teaching (didactic) practices of selected fairy tales, with creative dynamics. Through fairy tales we encouraged children to produce original ideas, free their imagination and express themselves through creative writing. The selected texts written by the pupils as well as the criteria for creativity evaluation digitized were submitted to judges-educators and subjected linguistic processing. In this paper we present a methodology, correlating judges-teacher evaluation with the linguistic processing statistics, comparing fairy tales as for their creative potential and, finally, comparing creativity of different aged pupils based on the same fairy tales. The creativity criteria used are seven: the four of the [Vaezi & Rezaei, S, 2019; Mozaffari, 2013] rubric plus the three classic ones [Crossley et al., 2016; Τyrou, 2017; Runco & Acar, 2012]. Results of our statistics comprise a k-means clustering of tales, using all seven creativity indices (rubric), designating their relevant didactic practices. An important result is a very similar clustering of tales is produced based on either the seven creativity criteria mentioned or the percentage of the part of speech (POS) used by the pupils. The above results confirm our initial hypothesis that some aspects of creative writing can be fairly successfully evaluated using linguistic features, [McNamara et al., 2015]. Moreover, our research results can be further exploited in the development of automatic assesment systems and automated content analysis of fairy tales, written by primary school pupils.
Keywords: Creative writing assessment, fairy tales, linguistic features.