Speech Sound Category Learning
Being able to understand and comprehend a language involves mapping a continuous, multidimensional acoustic signal to discrete abstract speech categories. While knowledge of a native language system is acquired early in life, the complexity of this mapping poses a difficult learning problem, particularly for second language learners who struggle to acquire the speech sounds of a non-native language. The apparent rigidness of the system does not allow for much plasticity in representations of the native language. The goal of my research is to advance our understanding of the speech perception system by focusing on the situations where it is robust to change, where it shows plasticity, and the systems that enable this. I study effective mechanisms for updating one's speech category system to accommodate for the speech sounds of a second language and what neural circuits are active during these different learning paradigms. My approach involves combining evidence from neural work with computational modelling of human behaviour.
Continuous Processesing of Speech in Non-Native English Speakers
While listening to speech humans are constantly predicting upcoming information as several levels on the linguistic hierarchy. It has been shown that native English speakers make predictions about upcoming phonemes using both a local context model which account for purely sublexical information, and - separately - a global context model which uses information and word and sentence levels. We are currently investigating how these context models may differ in non-native English speakers, using continuous MEG recordings.
The Language Familiarity Effect In Infancy
Human listeners are better at telling apart speakers of their native language than speakers of other languages, a phenomenon known as the language familiarity effect. While most accounts of this effect in adults require abstract phonological knowledge or comprehension of the speech itself, the effect has also been observed in infants as young as 4.5 months of age who are unlikely to have such sophisticated knowledge. Using algorithms from unsupervised machine learning and automatic speech recognition we built a model to demonstrate how children may show this effect without requiring any sophisticated linguistic knowledge.
Communicating Language Science to K-12 Students
Joint work with Kathleen Oppenheimer, Lauren Salig, Erika Exton, London Dixon and Alex Krauska
When the Covid-19 pandemic forced elementary and middle schools to move to online education, many forms of typical scientific outreach such as science fairs and school visits were no longer possible. In the last two years, we have pivoted to virtual outreach activities - making innovative, fun long-form demonstrations that can be presented to classrooms over zoom. We use green screens and a high degree of student interaction and distribute surveys to engage effectiveness. Our research show students enjoy the activities and learn basic linguistic concepts and principles of the scientific method
More Information: @TheLanguageScientists Instagram, Website (coming soon!)