The structure and evolution of story networks.

With this study, we advance the understanding about the processes through which stories are retold. A collection of story retellings can be considered as a network of stories, in which links between stories represent pre-textual (or ancestral) relationships. This study provides a mechanistic understanding of the structure and evolution of such story networks: we construct a story network for a large diachronic collection of Dutch literary retellings of Red Riding Hood, and compare this network to one derived from a corpus of paper chain letters. In the analysis, we first provide empirical evidence that the formation of these story networks is subject to age-dependent selection processes with a strong lopsidedness towards shorter time-spans between stories and their pre-texts (i.e. 'young' story versions are preferred in producing new versions). Subsequently, we systematically compare these findings with and among predictions of various formal models of network growth to determine more precisely which kinds of attractiveness are also at play or might even be preferred as explicatory models. By carefully studying the structure and evolution of the two story networks, then, we show that existing stories are differentially preferred to function as a new version's pre-text given three types of attractiveness: (i) frequency-based and (ii) model-based attractiveness which (iii) decays in time.