Catherine Lai

Lecturer

  • Linguistics and English Language
  • Centre for Speech Technology Research
  • School of Philosophy, Psychology and Language Sciences

Contact details

Address

Street

Room 2.11, Dugald Stewart Building

City
3 Charles Street, Edinburgh
Post code
EH8 9AD

Undergraduate teaching

 2024/25:

  • Speech Processing (Hons) -  Course Organizer
  • Discourse Analysis (Hons)

Postgraduate teaching

2024/25:

  • Speech Processing (MSc) - Course Organizer
  • Discourse Analysis (MSc)
  • Researching Responsible NLP (Designing Responsible NLP CDT students only)

Office Hours by appointment. 

Open to PhD supervision enquiries?

Yes

Current PhD students supervised

As first supervisor:

  • Sarenne Wallbridge (with Peter Bell & Steve Renals)
  • Yuanchao Li (with Peter Bell)
  • Alice Ross (with Martin Corley)

As second supervisor:

  • Johannah O'Mahony (with Simon King)
  • Emelie van de Vreken  (with Korin Richmond)
  • Jie Chi (with Peter Bell)
  • Mai Dao (with Peter Bell)
  • Bonnie Liu (with Sumin Zhao, Christian Ilbury)
  • Jack Tsangou

Past PhD students supervised

  • Leimin Tian (co-supervised with Johanna Moore)
  • Pilar Oplustil Gallegos (with Simon King)
  • Nina Markl ( with Lauren Hall-Lew)

Research summary

My research focuses on how we can use the non-lexical aspects of speech (e.g. speech prosody: how we say what we say) to get at what speakers actually mean. I’m currently working mainly how prosody changes expectations of speech for spoken language understanding and speech synthesis. I’m interested in this for both theoretical and practical reasons. On the one hand, I’m interested in developing speech technologies that can better take account of contextual variation. On the other hand, I’m interested generally in understanding where non-lexical aspects of speech fit into linguistic theories. This means that I’m usually, in some way or another, working more generally on developing models of prosody in dialogue.

I’m particularly interested in how we can use ideas from both machine learning and linguistics (semantics, pragmatics and phonetics) to get a more robust understanding of the relationship between prosody, discourse structure, and information structure. Hopefully this will build some bridges between more theoretical and empirical approaches understanding of spoken communication, particularly terms of how we express and understand topic (what people are talking about) and affect (how people feel about what they’re talking about). We need these bridges to understand whether technologies actual do what people think they do and the risks of deploying them.

You can find out more about what I work on by looking at personal website.

Current research interests

Generation and perception of prosody in synthetic speech; Modelling expectations in spoken dialogue ; Topic and stance detection from spoken dialogue; Multimodal emotion recognition; Prosody, presuppositions and information structure;

Affiliated research centres