Skip to Main content Skip to Navigation
Conference papers

Testing the Robustness of Online Word Segmentation: Effects of Linguistic Diversity and Phonetic Variation

Abstract : Models of the acquisition of word segmentation are typically evaluated using phonemically transcribed corpora. Accordingly, they implicitly assume that children know how to undo phonetic variation when they learn to extract words from speech. Moreover, whereas models of language acquisition should perform similarly across languages, evaluation is often limited to English samples. Using child-directed corpora of English, French and Japanese, we evaluate the performance of state-of-the-art statistical models given inputs where phonetic variation has not been reduced. To do so, we measure segmentation robustness across different levels of segmental variation, simulating systematic allophonic variation or errors in phoneme recognition. We show that these models do not resist an increase in such variations and do not generalize to typologically different languages. From the perspective of early language acquisition, the results strengthen the hypothesis according to which phonological knowledge is acquired in large part before the construction of a lexicon.
Document type :
Conference papers
Complete list of metadata
Contributor : Luc Boruta Connect in order to contact the contributor
Submitted on : Wednesday, August 29, 2012 - 12:11:09 PM
Last modification on : Wednesday, November 17, 2021 - 12:28:43 PM
Long-term archiving on: : Friday, November 30, 2012 - 2:25:08 AM


  • HAL Id : inria-00605806, version 1



Luc Boruta, Sharon Peperkamp, Benoît Crabbé, Emmanuel Dupoux. Testing the Robustness of Online Word Segmentation: Effects of Linguistic Diversity and Phonetic Variation. CMCL 2011 - Cognitive Modeling and Computational Linguistics Workshop at the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, Jun 2011, Portland, United States. pp.1-9. ⟨inria-00605806⟩



Record views


Files downloads