I am a fourth-year Linguistics PhD student at Linguistics and English Language within the School of Philosophy, Psychology and Language Sciences at the University of Edinburgh.

In my PhD dissertation, I use simulations and experimental methods to investigate the claim that sound change is driven by phonetic biases. Phonetic biases are physiological and psychoacoustic constraints on speech. One example is vowel undershoot: vowels sometimes fail to reach their phonetic targets due to limitations on the speed of the articulators. Phonetic biases are often paralleled by phonological patterns. For instance, many languages exhibit vowel reduction, a phonologised version of undershoot. It is often claimed that such parallels exist because phonological patterns arise through the repeated application of phonetic biases. My dissertation argues that this cannot be the case, as the influence of individual phonetic biases becomes greatly attenuated when other aspects of sound systems (such as category competition) are also taken into account. It also demonstrates that while biases do not cause sound change, they affect the probabilities of different types of sound change. Specifically, (i) the strength of phonetic biases and (ii) lexical factors related to phonetic biases can be used to predict cross-linguistic patterns in sound systems.

This investigation is unique both in its scope and methodology. Although phonetic biases are often implicated in sound change, their exact role is rarely investigated. The main reason for this lies in the elusiveness of the concept of biases and the difficulty of tracking their influence in complex systems. My dissertation overcomes these difficulties by (i) using computer simulations to generate predictions related to phonetic biases and (ii) matching these predictions to experimental data and cross-linguistic observations.

(1) The emergence of intrusive-r in English
This project investigates the claim that intrusive-r emerged through analogical extension. I look at frequency distributions within 18th century English (based on a corpus compiled for this project) and find that they support an analogical approach. Moreover, I show that a computational model of analogy can be used simulate the emergence of intrusive-r in an empirically accurate way.

(2) Echo-pairs in Hungarian
This project searches for phonological regularities in a corpus of Hungarian echo-pairs. Echo-pairs (cf. English itsy-bitsy) are notorious for their irregularity. I use statistical techniques like multidimensional scaling and methods from NLP to make sense of the data, and find meaningful patterns of phonological variation.

(3) L-deletion in Hungarian
This project is a collaboration which investigates l-deletion in Hungarian through a large production experiment. l-deletion is a phonetically gradient and variable phenomenon, which makes the interpretation of the results difficult. The project overcomes these methodological hurdles by using discriminant analysis and mixed effects modelling, and presents a detailed analysis of the social and phonological factors influencing l-deletion.