Markov Chain Monte Carlo Procedure Dating Of Homer

Eric Lewin Altschuler, Andreea S. Calude, Andrew Meade, and Mark Pagel have just published “Linguistic evidence supports date for Homeric epics.” Their paper offers a “likelihood-based Markov chain Monte Carlo procedure” estimate for the date of composition of Homer’s epics. Here’s the abstract,

The Homeric epics are among the greatest masterpieces of literature, but when they were produced is not known with certainty. Here we apply evolutionary-linguistic phylogenetic statistical methods to differences in Homeric, Modern Greek and ancient Hittite vocabulary items to estimate a date of approximately 710–760 BCE for these great works. Our analysis compared a common set of vocabulary items among the three pairs of languages, recording for each item whether the words in the two languages were cognate – derived from a shared ancestral word – or not. We then used a likelihood-based Markov chain Monte Carlo procedure to estimate the most probable times in years separating these languages given the percentage of words they shared, combined with knowledge of the rates at which different words change. Our date for the epics is in close agreement with historians’ and classicists’ beliefs derived from historical and archaeological sources.

This sentence stands out from the University of Reading’s news release announcing the paper, “The research dated the Homerian epics with a 95% certainty within a date range of 376 BCE and 1157 BCE, with a mean estimate of 762 BCE.” Yep, that is helpful. The late date in the 95% range is historically impossible and the early is, well, extremely unlikely. The news release doesn’t vouchsafe to us the certainty percent for the 710–760 BCE range.
The paper itself is behind a pay wall so it will be a little while before I can even attempt to evaluate it but here is a teaser chart presumably from the paper.

The dating of Homer

I’m not kneejerk opposed to such studies. In fact, I often find them abnormally interesting. I once proposed something like this to help sort out the relationships between the various Northwest Semitic languages. But I hope that the authors or someone has used the same methodology to determine the dates of Plato’s Republic and Origen’s Contra Celsum, not because I don’t know the dates of these works but because I do. Maybe someone has done such a study; maybe the authors of “Linguistic evidence supports date for Homeric epics” have. If such studies exist, I’d just like to see them.
Via Rogue Classicism

One thought on “Markov Chain Monte Carlo Procedure Dating Of Homer”

  1. The paper appears to now be available if you scroll down to Linguistic evidence supports date for Homeric epics.
    After reading … well, it’s remarkably thin, both in its use of comparative data and in the specificity of its conclusions. And they don’t seem to have considered the fact that Homer-as-we-have-it is undoubtedly different from Homer-as-it-was-composed, which would confound their calculations. Still, with a 95% interval of nearly thirteen centuries it is hard to say that the weakness of their data made their conclusion less meaningful.

Comments are closed.