I’m giving a talk tomorrow (Tues 21 Feb) at the University of Bath, for the AI folk in the computer sciences department. My thanks to Joanna Bryson for being adventurous and inviting an anthropologist to speak to CompSci folk!
Counting coconuts for the chief: coevolution in language and culture
Across the world, languages vary in their ways of enumeration. Some languages, but not others, have dedicated linguistic mechanisms for counting certain objects and/or large numbers. Numeral classifiers are words or affixes to nouns that are used for counting certain classes of objects, such as “animate things” or “coconuts”. Specific counting systems go a step further and count specific classes of objects by units greater than one, such as (e.g.) pairs or twenties. Examining Oceanic languages, Bender and Beller have advanced the idea that numeral classifiers and specific counting systems are object-specific, refer to culturally-salient semantic domains, and are often used to enumerate large quantities. Here we test their hypothesis that these linguistic features may have co-evolved with aspects of socioecology, specifically, norms of redistribution such as chiefly tribute that are found in socially stratified societies. We use comparative data across a sample of Austronesian ethnolinguistic groups, lexical phylogenies of these languages as a model of population history, and statistical methods from evolutionary biology to (a) reconstruct the most likely model of history of counting systems and social structure and (b) test for causal co-evolutionary processes. Using phylogenetic approaches not only allows us to control for Galton’s Problem but allows us to test these language-culture coevolutionary hypotheses in a framework that delivers estimates of the processes of cultural change. These results speak to broader issues regarding the flexibility of human numerical cognition, as well as shed light on the specific development of counting systems within the Austronesian cultural context.