Some Nobel Laureates Are More Coherent Than Others: Measuring Literary Quality in a Corpus of High Prestige Contemporary Literature

Publikation: KonferencebidragKonferenceabstrakt til konferenceForskningpeer review


The definition of literary quality is a highly complex problem that different schools of thought have approached through very different lenses. Studies that link literary quality to stylometric properties of the texts often look at global features, such as average sentence length,1 but recent studies show that a text’s development can be a better predictor of readers’ appreciation.2 Hypothesising that a balance between coherence and novelty improves the reading experience,3 examined the entropy of features linearly across literary texts, while4 looked at the fractal dynamics of their sentiment arcs.
Recently,5 used sentiment arcs’ coherence to tell Nobel laureates’ works from control groups, based on the idea that the Nobel Prize is the ultimate recognition of literary quality. However, Nobel laureates’ production can vary, and some laureates’ text can be more appreciated than others.
In our work, we use GoodReads’ scores to find out whether metrics that correlate with quality in broader corpora keep their predictive power within the Nobel canon itself. Defining the coherence of sentiment arcs through three complementary mea- sures - fractality,4 approximate entropy3 and arc compressibility6 - we find that novels within certain intervals of these measures tend to elicit higher ratings, suggesting that they might capture a balance between predictability and novelty which remains powerful even among highly competitive writers.
StatusUdgivet - 2023
Begivenhed7th Conference Digital Humanities in the Nordic and Baltic Countries: Sustainability: Environment, Community, Data - Online, Oslo, Norge
Varighed: 8 mar. 202310 mar. 2023
Konferencens nummer: 7


Konference7th Conference Digital Humanities in the Nordic and Baltic Countries