Stacks Xchange: Field Notes– Why do languages have the sounds they do?

This presentation by Muhammad Rehan, UCLA PhD candidate, investigates an age-old question in phonological typology: why are particular combinations of sounds more common than others and, more fundamentally, why do languages have the sounds they do? I argue that part of the answer lies in perceptual similarity, the relative discriminability of particular phonetic contrasts. Focusing on young infants, who possess the best discrimination abilities among humans, we use data from 151 experiments across 79 studies to perform a meta-analysis of infants' relative discriminability of phonetic contrasts. We then predict the frequency of these contrasts in 3,000+ languages, after controlling for genetic and spatial autocorrelation, using a Bayesian Binomial regression model with weakly informative priors. 

The second part of this talk discusses an innovative Continuous-Time Markov Chain that models contextual sound change on a phylogenetic tree using a rate matrix that is 1) derived from a log-linear function of 22 features relating to the physiological dimensions of different sounds and 2) boosts changes in particular contexts while hindering them in others. We use this model to infer rates of segmental changes and then regress these rates on the relative ease infants have in discriminating these sounds (measured through Hedges' g) to provide a causal link between infant perception and phonological typology. The main claim of this talk is that universal perceptual similarity of contrasts predicts their typological frequency through a possible causal link of segmental stability: sounds that are harder to perceive are less stable and thus more likely to change diachronically, explaining why they are less frequent in the languages spoken around the world today.

Muhammad Rehan is a 3rd year PhD student in the Program in Indo-European Studies at UCLA. Trained as a historical linguist and a phonologist, he is broadly interested in the evolution and spatial distributions of phonological features. A part of his research applies Bayesian statistical methods to model the spatial and spatio-temporal dynamics of phonological features, while another strand of research aims to develop integrative phylogenetic methods (including sequence alignment methods) that incorporate phonological characters in phylogenetic inference.