Question: What was the word for “two” used by people living in the Pontic-Caspian steppes (modern Ukraine and Southern Russia) 5,000 years ago?

Answer: *dwóH₁

This is how historical linguists reconstruct “two” in the Proto-Indo-European (PIE) language (see Indo-European vocabulary on Wikipedia). And now evidence accumulates that PIE speakers belonged to what archaeologists call the Yamnaya culture. Sure, there is still a bit of a controversy lingering about some aspects of this reconstruction, but recent aDNA evidence, in my opinion, has quite decisively put a wooden stake into the heart of  the alternative Anatolian hypothesis. Of course, nothing in science is 100% certain (if you want certainty, your best recourse is Divine Revelation). But to me, 99% certainty is good enough.

At a very fundamental level, historical linguists can reconstruct the PIE vocabulary with a high degree of certainty because language change is an example of cultural evolution. Or, as Darwin could have said, it’s ”descent with modification.” Many aspects of language change slowly and in a remarkably law-like manner. For example, I never studied German and don’t intend to do so formally. But I’d like to learn the language as I will be spending a lot of time in Austria in the next five years. So I listen to announcements on the bus and eavesdrop on other passengers. When walking in the streets of Vienna I play the game “figure out the the meaning of this word” (on a street sign or advertisement). The game goes like this: if the first letter in a German word is “z”, try substituting it with “t”; if “t”, with “d”, “d” with “th” and so on. Replace a “b” in the middle with “v” (see here). In many cases you will recover a word that is remarkably like its English equivalent. For example, the mysterious “Diebe” after substitutions becomes “thieve”–thieves!

Back to the PIE. It is absolutely remarkable that we can reconstruct how a word sounded 5,000 years ago—and remember, the Yamnaya people had no writing!

Historical linguistics is clearly the best developed case-study of phylogenetic reconstruction in cultural evolution. But why stop with language—what about religion? My good friend and colleague David Sloan Wilson has long argued that we should use the methods of evolutionary science in the study of religion (read his great book, The Darwin’s Cathedral). Religions evolve. Early Christianity evolved from Judaism. It then split into different branches: Monophysites, Arians, Chalcedonians, …, with Chalcedonians splitting later into the Eastern Orthodox and Western Catholic branches. Indic religion gave rise to Hinduism and Buddhism, with the latter splitting into Hinayana and Mahayana branches.

Of course, it’s not only descent with modification; different languages and religions also borrow elements from each other. The tree model of linguistic evolution needs to be supplemented by reticulations connecting different branches. And so do trees of religions (reticulations denoted with broken lines):

These ideas have been much on my mind during the past year. With the publication of our Nature article on moralizing gods the Seshat project has broken new ground—we are now testing evolutionary theories of religion. Some critics charged that we are trying to do the impossible. Texts and records become sparse as you go back in time. And once you go to the time before writing, they maintain, you cannot say anything about religion.

But that’s clearly wrong. If historical linguists can reconstruct the sounds of languages that disappeared well before writing was invented, why shouldn’t we able to do the same with religion? In fact, it is already being done.

I just finished reading a remarkable book by one of Seshat contributors, Patrick Kirch, co-authored with Roger Green. In Hawaiki, Ancestral Polynesia, published in 2001, Kirch and Green develop the phylogenetic model and apply it to cultural evolution in the Pacific Ocean before the Europeans arrived there (note: the Polynesians never developed writing!). They use a “triangulation method” in which historical linguistics, archaeology, comparative ethnology, and biological anthropology are integrated for the purpose of historical reconstruction. And I would add that in the eighteen years since they published the book, we have acquired an additional powerful source of information: ancient DNA.

An integrated approach is key. For example, historical linguistics is not great at timing when different languages split (it is now clear that glottochronology is much more difficult than initially thought). But archaeology fills this gap by telling us when people arrived at different islands. And so on. Where one avenue of reconstruction fails, another comes to rescue.

Kirch and Green use the phylogenetic model to yield extraordinary insights into the world of Ancestral Polynesians: which islands they inhabited before colonizing most of the Pacific, what they ate and how they prepared their food, how their material culture and socio-political organization evolved, and how their rituals and beliefs about gods and ancestors changed with time.

From their reconstruction it is clear that such fundamental concepts as mana and tapu were well-established in Ancestral Polynesia, but they also have undergone additional evolution in different branches occupying different archipelagos in the Pacific. The deification and ritual supplication of ancestors was also virtually universal.

Of particular interest is their reconstruction of what they call “an elaboration of the pantheon” which particularly affected the Eastern Polynesian societies. First to be added to the single Proto-Polynesian god, *Taangaloa (* indicates reconstruction) was a god of war, *Tu(q)u. Later four more named gods were added to the pantheon. These innovations were accompanied by an elaboration of the ritual. As an example, Kirch and Green suggest that an important innovation during the Proto Central-Eastern Polynesian phase was *tiki (“carved human image”).

Central-Eastern Polynesians (CEP) are of great interest to us, because one of the regions that we code in the Seshat World Sample-30 is Hawaii, which belongs to this branch. Although Kirch and Green don’t directly address the moralizing aspects of the Polynesian religion, Patrick Kirch has been very helpful in answering questions about moralizing supernatural punishment (MSP) that members of the Seshat project posed to him.

According to our informal and very tentative reconstruction (which still needs more expert advice!), there are clear MSP elements in Hawaiian religion. In particular, the kapu system (tapu/tabu in other Polynesian languages), which denotes what is sacred or forbidden (for more information, see this article on Wikipedia), included injunctions against deceit, theft, and murder. But these moralizing elements were of secondary concern compared to ritual infractions. Furthermore, offenses against kapu were primarily policed by human agents (chiefs and their retinues), rather than by supernatural agents (spirits and gods).

Our survey of Hawaii’s “sister cultures” (Maori, Rarotonga, Tahiti, Tuamotu, and the Marquesas) suggests that even these, relatively weak, MSP elements were largely absent in other CEP branches, with a possible exception of the Marquesas, where theft and murder also could be subject to supernatural punishments (but with the same limitations as in Hawaii). The idea of punishment/reward in the afterlife appears to be universally absent in all Polynesian cultures.

This survey raises a number of intriguing questions. Are MSP elements, which we see in Hawaii, a relatively recent innovation? Note that Hawaii was settled by colonists from the Marquesas. It is perhaps significant that MSP elements in the CEP are found only in these two cultures.

All of this is quite speculative—remember that I often use my blog as a platform for airing new ideas and soliciting comments and critique. What we need is the application of the phylogenetic model to this question by specialists on Polynesian culture (of which I am, most assuredly, not one). Careful reconstruction using the triangulation method of Kirch and Green could be complemented by more quantitative Bayesian phylogenetic models that have been developed by such cultural evolutionists as Ruth Mace and Russell Gray. In fact, Gray’s group recently (in 2015) published a Bayesian analysis of moralizing religion in Austronesia (a broader linguistic grouping that includes the Polynesians). I am in contact with the first author of the article, Joseph Watts, about the details of their data and analyses.

In conclusion, Polynesia (and, more broadly, Austronesia) is a great “polygon” that has served us well in developing a variety of approaches for reconstructing cultural evolution of prehistoric societies. But it’s not the only one. Take Indo-Europeans. Jean Haudry in 1993 compared oath formulas from a number of Indo-European languages (Old Norse, Russian, Sanskrit, and Persian) and found that they share the image of the perjurer struck by his own weapon. Was this MSP element present in the PIE culture? There is a great potential for employing the phylogenetic model to reconstruct not only past languages, but also elements of past religions. And this potential has hardly been tapped.