The hidden history of tone: Evidence from the perception of f0 perturbations

Roughly half the world's languages use pitch to differentiate word meaning. How such tonal contrasts arise-a process known as tonogenesis— remains poorly understood. Tonogenesis provides a unique window into how stable linguistic structures emerge from individual perceptual dispositions. Prominent among its proposed triggers are contrastive laryngeal features of consonants, which perturb the pitch of the following vowel.

Over time, this perturbation is amplified, and the consonantal contrast is lost, whereupon the pitch difference becomes the primary cue for lexical distinction. Critical to this exaptation hypothesis is prerequisite that listeners can perceive the relevant pitch perturbations. However, the foundational evidence for how listeners map these perturbations onto perceived pitch rests on a single study of just ten English listeners. With a large-scale, multilab team science project, we investigate how listeners across many languages match the perceived pitch of syllables varying in perturbation extent and duration. We characterise this perceptual mapping and its covariation with listeners' linguistic, musical, and demographic backgrounds to delineate the conditions for tonogenesis. In this talk we will present preliminary findings from our pilot study.

Chenzi Xu is an Assistant Professor in Linguistics and Multilingual Studies at Nanyang Technological University. Her research interests are in phonetics, psychoacoustics, neurolinguistics, and sociophonetics, with a particular focus on tone languages. Her work often uses computational and experimental approaches, bridging phonetic research with speech and language technologies. Her recent work investigates the production and perception of tonal patterns across languages, with a perspective to understanding systematic phonetic variation and i

ts implications for linguistic theory. For more information, please visit her personal website.
She holds a D.Phil. (Ph.D.) in Linguistics from the University of Oxford. Prior to joining NTU, she
was a Leverhulme Trust Early Career Fellow at the University of Oxford, and also worked as a postdoctoral research associate in forensic phonetics at the University of York, focusing on person-specific automatic speaker recognition.