Crowdsourced speech and automatic alignment: New frontiers for laboratory phonology @ LabPhon 20

 

This workshop explores the shift from controlled laboratory recordings to crowdsourced and automatically aligned speech data. Advances in speech technology and annotation tools now enable large-scale phonetic research but raise questions about data reliability, interpretability, and ethics. Alignment errors and variable recording conditions are especially common in spontaneous and heterogeneous data, where they challenge traditional analytical assumptions. Bringing together perspectives from phonetics, phonology, speech technology, and the social sciences, the session examines how these new data practices reshape laboratory phonology and invites discussion on developing transparent, linguistically informed, and socially responsible approaches to large-scale speech analysis.

 

Program
9:00 - 9:05Welcome
9:05 - 10:05Invited Talk (Jane Stuart-Smith, University of Glasgow)
  
Oral Session 1 
10:05 - 10:25Ka Ki So, Chenzi Xu, Grace Wenling Cao and Peggy Mok. Using Montreal Forced Aligner for Coding Phonetic Variations in Spontaneous Speech
10:25 - 10:45Philip Hoole and Miriam Oschkinat. Using time-warping and amplitude-oriented non-discrete boundaries for automated alignment of repetitive speech data
10:45 - 11:05Chun Man Manson Wong, Yitian Hong and Pik Ki Peggy Mok. How well can MFA detect sound change in large-scale sociophonetic research?
  
11:05 - 11:25Coffee Break
  
Oral Session 2 
11:25 - 11:45Robert Fromont. The Ethics of Automatic Speech Processing in the Cloud.
11:45 - 12:05Miriam Oschkinat, Melanie Weirich and Stefanie Jannedy. Regional perceptual variation in a large-scale crowdsourced database: The German /eː/–/ɛː/ merger
  
12:05 - 12:15General Discussion

 

 

 

Topics of interest include, but are not limited to:

  • Data variability, recording conditions, and alignment challenges

  • Integrating phonetic and phonological knowledge into automatic methods

  • Cross-dialectal, cross-linguistic, and socio-demographic variation

  • Statistical and computational methods for large, heterogeneous corpora

  • Ethical issues and interdisciplinary collaboration

 

Organizing committee 

  • Martine Adda-Decker, LPP/CNRS, France

  • Ioana Chitoran, Paris Cité University, France

  • Johanna Cronenberg, LPP/CNRS, France

  • Adèle Jatteau, University of Lille, France

  • Lori Lamel, Vocapia Research, France

  • Mélanie Lancien, University of Lorraine, France

  • Mark Liberman, University of Pennsylvania, USA

  • Anisia Popescu, LISN/CNRS, France

  • Laura Spinu, City University of New York, USA

  • Paola Tubaro, CREST/CNRS, France

  • Ioana Vasilescu, LISN/CNRS, France

  • Yaru Wu, University of Caen Normandy, France (Coordinator)

 

Submission instructions

The formatting should adhere to the LabPhon abstract formatting requirements:

  • Written in English

  • Maximum of one page of text; references, examples, and/or figures may be included on a second page

  • Submitted as a PDF file

  • Times New Roman font, size 12, single spacing, 1-inch margins

  • Filename format: Paper_title.pdf (e.g., Looking_back_and_looking_forward.pdf)

  • Do not include author names or affiliations in the filename or in the abstract itself

 

Link for submission: https://easychair.org/conferences/?conf=csaa2026

 

Important dates

  • Submission deadline (extended): Sunday March 22, 2026, 11:59PM, Anywhere on Earth (AoE)

  • Notification of acceptance: April 8, 2026.

  • Date/Time of the workshop: 9:00–12:15 (local time), June 25, 2026
  • Location: TBA (but the same place as the conference venue, in Montréal, Québec, Canada)