Computational Linguistics
Postdoc
caitlinr@ru.is |
|
Language and Voice Lab Reykjavík University Reykjavík, Iceland |
|
CV available by request. |
I'm a postdoc at Reykjavík University's Language and Voice Lab, where we develop computational techniques for working with speech and text in low/medium-resource language varieties. I completed my PhD in Linguistics at the University of Pennsylvania in 2021, studying the relationships between language variation, children's acquisition, mental representations, and historical change.
Over here you can usually find some demos-in-progress of a bit of whatever I'm doing lately.
About
I'm interested in finding out which acoustic information influences listeners' perceptions of difference in pronunciations, and how computational representations of the speech signal can access this information. This work, started with Mark Liberman in collaboration with Martijn Wieling and Martijn Bartelds at the University of Groningen, adapts approaches from automatic speech recognition technology to compare speech samples from different language varieties.
My dissertation research, supervised by Charles Yang and Don Ringe, proposes a quantitative model that describes when children do and don't form generalisations about the sound system of their language. Several case studies explore the model's implications, from predicting the individual time courses of different children's phonological representation learning, to offering a more concrete, explanatory, and testable account for some historical language developments.
I enjoy doing ballet and have portrayed an elephant at least twice. This animal lives in my house.
Papers and presentations
- C Richter, R Pálsson, L O'Brien, K Friðriksdóttir, B Bédi, E H Magnúsdóttir, J Guðnason. 2023. Orthography-based pronunciation scoring for better CAPT feedback. Interspeech.
- C Richter, J Guðnason. 2023. Relative dynamic time warping comparison for pronunciation errors. ICASSP.
- M Bartelds, W de Vries, F Sanal, C Richter, M Liberman, M Wieling. 2022. Neural Representations for Modeling Variation in English Speech. Phonetics. Code here.
- C Richter, B Bédi, R Pálsson, J Guðnason. 2022. Computer-assisted pronunciation training in Icelandic (CAPTinI): developing a method for quantifying mispronunciation in L2 speech. Intelligent CALL, granular systems and learner data: short papers from EUROCALL 2022
- C Richter. Allophone acquisition as informed by alternations: English alveolar flapping. 15th International Congress of the Association for the Study of Child Language, Philadelphia, 2021.
- C Richter. 2021. Consequences of lexical variability in toddlers' vocabularies. Poster presented at the 45th Penn Linguistics Conference, Philadelphia.
- M Bartelds, W de Vries, C Richter, M Liberman, M Wieling. 2020. Measuring foreign accent strength using an acoustic distance measure. Paper presented at ISSP 12 (proceedings forthcoming).
- J Mott, A Bies, S Strassel, J Kodner, C Richter, H Xu, M Marcus. 2020. Morphological Segmentation for Low Resource Languages. Paper presented at LREC 2020.
- M Bartelds, C Richter, M Liberman, M Wieling. 2020. A new acoustic-based pronunciation distance measure. Frontiers in Artificial Intelligence. Code here.
- J Kodner & C Richter. 2020. Transparent /aɪ/-Raising as a Contact Phenomenon. Invited submission to PWPL 25.2 Selected Papers from NWAV 47.
- C Richter. 2020. Menominee high back vowel split as a consequence of alternation-sensitive phoneme learning. Presented at Linguistic Society of America annual meeting, New Orleans, USA.
- C Richter. 2019. Transcription-independent measurement of pronunciation difference. Presented at the 4th Conference on Experimental Approaches to Perception and Production of Language Variation (ExAPP), 26-28 September, U. Münster, Germany.
- C Richter. 2019. Building language representations: A model of allophone acquisition. Presentation, DARPA LORELEI.
- {J Kodner & C Richter.} 2019. Emergence of partial /aɪ/-raising through child language acquisition in a mixed input setting. Presented at Linguistic Society of America annual meeting, New York, USA.
- J Kodner & C Richter. 2018. Partial /aɪ/-Raising as a Contact Phenomenon. Presented at New Ways of Analyzing Variation 47, New York, USA.
- C Richter, M Wickes, D Beser, M
Marcus. 2018. Low-resource
Post Processing of Noisy OCR Output for Historical
Corpus Digitisation. Paper presented at LREC,
May 7-12, Miyazaki, Japan.
Software (extended version) available at https://github.com/CopenhagenCityArchives/CorrectOCR/, and https://bandasavn.setur.fo/ has the original speech corpus whose transcriptions we worked on. - C Richter. 2018. Learning Allophones: What Input Is Necessary? Paper presented at BUCLD 42, November 3-5 2017, Boston, USA.
- C Richter. 2018. Computer-Aided OCR Correction for Low-Density Languages. Presentation, DARPA LORELEI.
- C Richter. 2018. An acquisitional model of phoneme learning for invariant transparency. Common Ground Seminar in Language and Communication Sciences, University of Pennsylvania.
- C Richter. 2017. Morphology Annotation Gold Standard. Presentation, DARPA LORELEI.
- C Richter, N H Feldman, H Salgado, A Jansen. 2017. Evaluating low-level speech features against human perceptual data. TACL (5); paper additionally presented at EMNLP 2017, 7-11 September, Copenhagen, Denmark.
- C Richter. 2016. Towards Annotating a Morphology Gold Standard. Presentation, DARPA LORELEI.
- C Richter, N H Feldman, H Salgado, A Jansen. 2016. A Framework for Evaluating Speech Representations. Paper presented at the Cognitive Science Society 38th Annual Conference, Philadelphia, USA.
- C Richter, N H Feldman, A Jansen. 2014. Representing speech in a model of phonetic perception. Presented at the Northeast Computational Phonology Workshop, New York, USA.
- N H Feldman, C Richter, J Falk, A Jansen. 2013. Predicting listeners' perceptual biases using low-level speech features. Presented at the Northeast Computational Phonology Workshop, Boston, USA.
- E Quirk, P Feitzinger, C Richter, M Zeitlin, V Valian. 2011. A computational analysis of grammar change and grammar similarity. Presented at the 17th International Conference on Architectures and Mechanisms of Language Processing, Paris, France.
Where I've been
-
Reykjavík University | 2021 – now
Postdoc, Language and Voice Lab.
Department of Computer Science / CADIA -
University of Pennsylvania | 2015 – 2021
PhD Linguistics.
Linguistics and CIS, funded by NSF Graduate Research Fellowship. -
University of Maryland | 2013 – 2015
Research assistant, computational linguistics.
With Naomi Feldman, focusing on models of human speech sound perception and computational representations of speech. -
University of Edinburgh | 2009 – 2013
MA Cognitive Science: informatics, philosophy, linguistics, psychology.
Dissertation on low-resource OCR post-processing, supervised by Sharon Goldwater (eventually this). Research assistant for Caroline Heycock, Alexa Morcom, and Patrick Sturt. Edinburgh Global Scholarship 2009-2013, Patterson Prize (Old English) 2012, Innovative Initiative Grant 2012. -
Vassar College | 2008 – 2009
Cognitive science.
NSF REU annotating this. - I've also spent some summers at the University of the Faroe Islands (2012, 2018) and Language Acquisition Research Center (2011).