Research

I study the neural mechanisms of speech motor control: how the brain monitors its own voice, corrects speech errors, and why this system fails in speech disorders such as stuttering. I combine task and resting-state fMRI, structural MRI, EEG and non-invasive brain stimulation with multivariate pattern analysis and effective-connectivity modelling. My work falls into four connected strands.

Sensorimotor control of speech

Speaking depends on continuously comparing what we intend to say with what we hear ourselves say. My doctoral work at Oxford examined this using altered auditory feedback: when the sound of a speaker’s own vowels is shifted in real time, they gradually adjust their articulation to compensate. I combined this approach with brain stimulation and MRI to study the contribution of the cerebellum and sensorimotor cortex to that adaptation.

In a preregistered study of 90 participants, we tested whether transcranial direct current stimulation (tDCS) over motor cortex or cerebellum enhances speech adaptation. The previously published effect did not replicate (Yuan et al., 2025, Cortex; data and code). A behavioural study and a functional and structural MRI study followed, identifying cortical and cerebellar mechanisms of speech adaptation, and a collaboration with Beijing Normal University ran matched behavioural and fMRI protocols in the UK and China. My thesis is available from the Oxford University Research Archive.

Stuttering and speech disorders

I use structural and resting-state fMRI to characterise how brain structure, activity and connectivity differ in people who stutter, working with a large neuroimaging dataset on developmental stuttering (~300 participants). I’m also interested in whether non-invasive brain stimulation can support speech-motor control.

Language production in bilinguals

Bilingual speakers have to select the intended language and keep the other from interfering. Using fMRI, multivoxel pattern analysis and effective-connectivity modelling, I have mapped the brain networks behind this language control, how they relate to domain-general cognitive control, and how they change with training and with language context. I have also used EEG to study how bilinguals track continuous speech in noise (preregistration).

The cerebellum in speech, language and reading

Across these strands I have a particular interest in the cerebellum. Combining fMRI with tDCS, my work examined how language control and domain-general cognitive control are organised within the cerebellum and its connections with the cerebral cortex. With collaborators, I proposed the cerebro-cerebellar mapping hypothesis, a model of the cerebellum’s contribution to reading and to developmental dyslexia (Li, Yuan et al., 2022, Neuropsychologia).