Some people in the softer sciences, including much of psychology, are very wary of “reductionism,” the practice of understanding via an analysis of the interaction of the parts of complex systems.
Yet there are scientists, especially statistical mathematicians, who love to tear apart complex systems as a way of classifying or codifying them.
So it’s not entirely surprising that a group of mathematicians has applied statistical analysis to mythic literature. In an article titled “Universal Properties of Mythological Networks,” a team from Coventry University has analysed three classical myths to see how much their historicity can be established statistically.
The latest study is similar to the work of Stanford’s Franco Moretti (reviewed on this blog last year). In Moretti’s work, a literature specialist attempted to apply statistical models to novels and plays. Moretti first studied Hamlet, while the present study looks at The Iliad, Beowulf and the Irish epic Táin.
Moretti conceded at the time that he is not an expert in statistical analysis, nor are the present mathematicians experts in literature. However, the mathematicians are on safe enough ground to the extent that they are not interested in the qualitative merits of the epics they’re studying. Instead, they apply entirely statistical comparisons between each epic and the others and, more important, between each epic and the kinds of character relationships that are typical in real social networks.
The central interest of the statistical study is to see if it’s possible to identify fictional stories by comparing their social networks to typical real networks. As well, the degree to which each epic corresponds to real networks indicates the likelihood that at least parts of an epic story reflect real history. In other words, the more that the social interactions in an epic look like real social networks, the more likely it is that the epic’s story is based on actual events and real people.
The statistical analysis carried out by the Coventry team goes far beyond the simple “character relationships chart” generated by Moretti.
Here we statistically compare networks underlying mythological narratives from three different cultures to each other as well as to real, imaginary and random networks. In this way we quantitatively explore universality in mythology and attempt to place mythological narratives in the sprectrum from the real to the imaginary.
The methods used include clustering analysis (how much a network is closed or “cliqued”), hierarchical analysis (how smaller groups aggregate in power relationships into larger groups), giant component analysis (how much the network depends on the most involved individuals), assortive analysis (how much individuals of similar power rank interact), how vulnerable networks are to breakdown when under “attack” (by removing a high, low, or random individual) — and so on and so forth.
This kind of multiple factor investigation is quite complicated, and the details are not all necessary to follow and appreciate the analysis. (If you do want to read the entire study, formulae and all, you can find it here.)
The results of this statistical analysis compare well with traditional estimates of the historicity of each epic. The Iliad and Beowulf are believed to contain at least some real characters and events, while Táin is thought to be mostly or entirely fictional. These were the same conclusions reached by the mathematicians. Specifically, the statistical analysis suggests that The Iliad operates most like real social networks. Beowulf is also realistic, but this realism is diminished to the extent that the central character has been “imposed” artificially onto existing stories. The networks in Táin look like those in the Marvel Comics superhero series, analyzed separately, suggesting that the Irish epic is likely highly fictional.
In summary, the Coventry team discovered that the “friendly” (related or allied) relationships in mythological literature are very similar to those in real social networks. Purely fictional social networks, in contrast, differ significantly from real networks on most of the key factors of the analysis.
There are several areas of interest here beyond the bare numbers. One is that the study supports the idea that many epic mythologies are probably based in real history. Rather than just making it up, the chroniclers of many culture-defining epics used real people and real events at the base of their imaginative stories. The dragons and gods may be fanciful, but the people whose stories include them are often real.
Another issue is the question of whether, or to what extent, statistical tools can effectively and appropriately be applied quantitatively to qualitative experiences like the reading of literature. In the present work, we see what appears to be a real contribution to mythology study. This suggests that at least some quantitative methodologies are applicable to the study of literature. But there is an as yet ineffable quality to the literary experience, something that has always gone beyond what’s on the page. As long as reading literature involves a stimulated human mind as part of the equation, analyzing just the concrete, on the page part of the activity must fall short of a complete explanation.
Nonetheless, sophisticated empirical analysis of traditionally non-emprical content is an interesting and evolving field, and it bears watching.