Evolutionary dynamics of maximal syllable complexity show no universal bias rooted in jaw biomechanics

Abstract

Linguistic evolution has given rise to variation in the degree of syllable complexity that any given language permits. However, several models of language converge in predicting a universal bias towards the consonant-vowel (CV) structure (or a broader strong-weak asymmetry). This is closely tied to incremental speech processing, and in evolutionary terms has been linked to certain rhythmic orofacial displays in non-human primates. Yet, whether linguistic evolution really reflects such a universal tendency is an open question, one that we aim to address using phylogenetic comparative methods. Modeling change according to a continuous-time Markov process, we estimate transition rates for maximal onset and coda sizes (i.e. grammatical rules) across 5 language families. We demonstrate that a lineage-specific model outperforms a global model, and further explore the nature of lineage-specific long-term preferences and rates of change on multiple levels. We show that the former drive inter-family variation more than the latter. Generally, particulars turn out to impact syllable complexity more than any universal biomechanically or neurocognitively motivated biases do, implying that evolutionary continuity with our primate ancestors may have been overstated for this domain of language.