Does literature evolve one funeral at a time?

The cultural evolution of literary fiction is rarely studied, but rich literary data can help address some of the general problems of cultural change. In this article, we use a massive dataset of Anglophone fiction (over 23 000 books) and the tools of natural language processing to understand whether the long-term change of topics in books is driven by the individual change of authors or by the cohort turnover in author populations. To answer this question, we borrow a method from evolutionary ecology: decomposition analysis based on the Price equation. To prove the suitability of this method, we first apply it to simulated data and show that it does allow distinguishing between these two processes. Afterwards, we decompose the temporal trajectories of topics and measure the relative effects of the arrival of newcomer authors (entrances), the retirement of authors (exits) and the change of topic preferences during authors' lifetimes (individual change). We find that cohort turnover is a stronger force than individual change. Within the cohort effects, the effect of entrances is almost twice as strong as the effect of exits. Using simulated data, we discover that this difference stems from the unequal lengths of authors' careers.