Selected Research Projects

See my CV for a complete list of projects, publications and presentations.

How do Lexical Factors Condition Speech Perception?

DPrime of Onset Consonants in Kaqchikel

Acoustic and lexical effects on speech perception in Kaqchikel (Mayan)

With Ryan Bennett (UC Santa Cruz).

Kaqchikel (Mayan) has a phonemic contrast between voiceless plosives and `glottalized' plosives (implosives and ejectives) at corresponding places of articulation. Our study investigates acoustic and perceptual similarity within the plosives of Kaqchikel, focusing on how acoustic similarity and lexical factors affect consonant identification.

To find out more, click here

* One new publication added on May, 2018

Conversational Correlation of Frequency

Expectation and Lexical Retrieval in Naturalistic and Experimental Misperception

With Andrew Nevins (University College London).

Naturalistic mishearings provide data about lexical retrieval, as they instantiate cases where instead of the intended word, a listener accesses an incorrect -- but often similar word in terms of multiple lexical properties. Here, we examine the relationship between the intended word and the the actually-perceived word for one of the main lexical properties, namely token frequency using two naturalistic corpora of conversational and sung speech and an experimental corpus. Our results suggest that listeners are sensitive to their experience and expectation of their phonetic and lexical knowledge during speech perception.

To find out more, click here

Slips Logo

Naturalistic Speech Misperception

Supervised by Andrew Nevins (University College London).

For my PhD thesis, I was investigating speech misperception in its most naturalistic form, namely slips of the ear. The erroneous patterns in speech perception shed lights on psycholinguistics, speech segmentation, models of diachrony and more. Furthermore, I cross-examined the patterns in naturalistic and laboratory data in English. The naturalistic corpus of 5000 naturally occurring misperception of conversational English yielded as part of the thesis is available at the SEAR Project (

To find out more, click here

How do Lexical Factors Condition Speech Production?

Effect of predictability on duration

Contextual predictability influences word and morpheme duration in a morphologically complex language (Kaqchikel Mayan)

With Ryan Bennett (UC Santa Cruz).

Probability is one of the many factors which influence phonetic variation. Contextual probability, which describes how predictable a linguistic unit is in some local environment, has been consistently shown to modulate the phonetic salience of words and other linguistic units in speech production (the probabilistic reduction effect). In this paper we ask whether the probabilistic reduction effect, as previously observed for majority languages like English, is also found in a language (Kaqchikel Mayan) which has relatively rich morphology.

To find out more, click here

* A new published JASA article added on August, 2018 [pdf]

Italian G:NG ratio by FL

A Functional-Load Account of Geminate Contrastiveness: a Meta-Study

With John Harris (University College London).

Phonetic studies of geminates do not typically address the question of what motivates the fine-grained linguistically-relevant variation, particularly along the primary acoustic correlate of a singleton-geminate contrast, constriction duration (Kawahara, 2015). In the present study we argue that a major factor contributing to the variation is the measure Functional Load (Hockett, 1955).

To find out more, click here

How do Lexical Factors Condition Morpho-phonological Patterns?

non productivity of L-shaped morphome decline in productivity

The rise and fall of the L-shaped morphome: diachronic and experimental studies

With Andrew Nevins (University College London) and Cilene Rodrigues (Pontifical Catholic University of Rio de Janeiro).

It has been suggested that the Romance first person singular indicative constitutes a natural class with the present subjunctive paradigm for the purposes of stem selection (Maiden 2005), thus forming a kind of ‘diagonal syncretism’, as the latter shares no morphosyntactic features with the former. The existence of such patterns has been taken to be an argument for autonomous morphology and the existence of unnatural ‘morphomes’, in the sense of Aronoff (1994). Our experimental investigations with native speakers of Portuguese, Italian, and Spanish reveal that this pattern is underlearned. We then study the underlearning of the L-morphome in terms of historical change in the salience of these patterns. Even though many of the morphomic verbs have maintained a very high token frequency (allowing them to survive as memorized), their productivity has diminished over time, and hence they go unlearned as a generalizable pattern. When the distribution of irregular alternations is overshadowed in the lexicon, a morphologically unnatural pattern may cease to maintain its status as part of the grammar.

To find out more, click here

awT violation

Phonotactics with[awt] rules: the learnability of a simple, unnatural pattern in English

With John Harris (University College London) and Nick Neasom (University College London).

In this paper, we examine the English phonotactic pattern where consonants following /aw/ are restricted to coronals. The pattern (‘awT’) is pretty regular (more so than velar softening), general (it affects a large swathe of the lexicon), and formally quite simple (more so than the -s pattern). And it is not natural. We report the results of a large scale non-word judgement experiment. We found that to the extent that speakers have any tacit inkling of the pattern at all, it is probably not encapsulated in anything like a phonologist’s rule or constraint. Rather it's based on lexical statistics such as neighbourhood density and phonotactic probability.

To find out more, click here

Phonetic Bases of Phonological Patterns

Formal Analysis of Kaqchikel Root Restrictions

Laryngeal co-occurrence restrictions as constraints on sub-segmental articulatory structure

With Ryan Bennett (UC Santa Cruz).

Laryngeal co-occurrence restrictions are prevalent in Mayan languages. Multiple ejectives are not allowed in a /CVC/ root, unless they are identical. The labial implosive /ɓ/ and glottal stop /ʔ/ are exempt from this restriction. One proposal called phonetic realism (Gallagher, 2010) proposes that it is grounded in considerations of acoustic similarity. This project investigates whether surface phonetic patterning can consistently reflect their phonological classhood as well as proposing an alternative account referring to sub-segmental articulatory structure.

To find out more, click here


Lenition as modulation reduction: consonant allophony in Ibibio

With John Harris (University College London) and Eno-Abasi Urua (University of Uyo).

We present an account of the impact of lenition on consonants that is founded on the model of speech as a modulated carrier signal. Lenition has the uniform effect of reducing the extent to which a consonant modulates the carrier. Unlike approaches based on sonority or articulatory opening, the modulated-carrier account provides a unitary treatment of all the classically recognised lenition types. We examine the effects that lenition in this language has on the amplitute and periodicity properties of consonants and propose a method for combining these under a single measure of modulation reduction.

To find out more, click here

Modelling Phonology

Positional importance by syllable structure

OCP Effects in Turkish Partial Reduplication: Locality and Feature Specificity

With Faruk Akkuş (UPenn).

This paper presents a comprehensive analysis of Turkish Particial Reduplication. We highlight that the OCP constraints are more graded than they have been previously proposed. Our findings support previous work on the formulations of OCP constraints that treat individual features as free parameters in the similarity computation. The surprising finding with regard to locality is that not only the OCP effect extends all the way from C1 to C4, but it is a function of both the proximity from LC and the position in syllable structures.

To find out more, click here

Interdisciplinary work

Distribution of RU1-1, RU1-2, and deletion of READ1 in 43 world populations.

Genetic variation and phoneme variation across languages

With Mellissa M. C. DeMille (Department of Pediatrics, Yale University School of Medicine) and Jeffrey R. Gruen (Department of Pediatrics and Department of Genetics, Yale University School of Medicine).

Languages evolve rapidly due to an interaction between sociocultural factors and underlying phonological processes that are influenced by genetic factors. DCDC2 has been strongly associated with core components of the phonological processing system in animal models and multiple independent studies of populations and languages. To characterize subtle language differences arising from genetic variants associated with phonological processes, we examined the relationship between READ1, a regulatory element in DCDC2, and phonemes in languages of 43 populations across five continents. Variation in READ1 was significantly correlated with the number of consonants. Our results suggest that subtle cognitive biases conferred by different READ1 alleles are amplified through cultural transmission that shape consonant use by populations over time.

To find out more, click here

* One new publication added on April, 2018 (PNAS:

Universal Non-Word Repetition Test (UNWR) -- Part of the FIST project

With Peter Howell (University College London) and John Harris (University College London).

This study designed and validated a non-word repetition test (UNWR), applicable across 20 languages. It was validated by comparing groups of children identified by their speech and language symptoms as having either stuttering or WFD. This non-word repetition test validates the symptom procedure. Both the symptom procedure and the non-word repetition measure are appropriate for use with children who have diverse language backgrounds.

To find out more, click here
To download the toolkit, click here

Top 100 Sound Symbolic Words

Sound Symbolism Revisited -- a Reconstruction Approach

With Pawel Mandera (Ghent) and Emmanuel Keuleers (Tilburg).

This study revisited the topic of sound symbolism with a reconstruction approach. A well-known linguistic fact about the linkage between sound and meaning is that the relationship is almost or completely arbitrary (Hinton, Nichols, and Ohala, 2006). By using machine-learning techniques, our results suggested that the connection between sound and meaning is not entirely arbitrary but consistently above random and allegedly sound-symbolic words (such as "clunk", "pop" and "slash") are more reconstructable than just any words.

To find out more, click here