The acquisition of allophonic rules: statistical learning with linguistic constraints. Peperkamp, S., Calvez, R. L., Nadal, J., & Dupoux, E. Cognition, 101(3):B31-41, 2006. doi abstract bibtex Phonological rules relate surface phonetic word forms to abstract underlying forms that are stored in the lexicon. Infants must thus acquire these rules in order to infer the abstract representation of words. We implement a statistical learning algorithm for the acquisition of one type of rule, namely allophony, which introduces context-sensitive phonetic variants of phonemes. This algorithm is based on the observation that different realizations of a single phoneme typically do not appear in the same contexts (ideally, they have complementary distributions). In particular, it measures the discrepancies in context probabilities for each pair of phonetic segments. In Experiment 1, we test the algorithm's performances on a pseudo-language and show that it is robust to statistical noise due to sampling and coding errors, and to non-systematic rule application. In Experiment 2, we show that a natural corpus of semiphonetically transcribed child-directed speech in French presents a very large number of near-complementary distributions that do not correspond to existing allophonic rules. These spurious allophonic rules can be eliminated by a linguistically motivated filtering mechanism based on a phonetic representation of segments. We discuss the role of a priori linguistic knowledge in the statistical learning of phonology.
@Article{Peperkamp2006,
author = {Sharon Peperkamp and Rozenn Le Calvez and Jean-Pierre Nadal and Emmanuel Dupoux},
journal = {Cognition},
title = {The acquisition of allophonic rules: statistical learning with linguistic constraints.},
year = {2006},
number = {3},
pages = {B31-41},
volume = {101},
abstract = {Phonological rules relate surface phonetic word forms to abstract
underlying forms that are stored in the lexicon. Infants must thus
acquire these rules in order to infer the abstract representation
of words. We implement a statistical learning algorithm for the acquisition
of one type of rule, namely allophony, which introduces context-sensitive
phonetic variants of phonemes. This algorithm is based on the observation
that different realizations of a single phoneme typically do not
appear in the same contexts (ideally, they have complementary distributions).
In particular, it measures the discrepancies in context probabilities
for each pair of phonetic segments. In Experiment 1, we test the
algorithm's performances on a pseudo-language and show that it is
robust to statistical noise due to sampling and coding errors, and
to non-systematic rule application. In Experiment 2, we show that
a natural corpus of semiphonetically transcribed child-directed speech
in French presents a very large number of near-complementary distributions
that do not correspond to existing allophonic rules. These spurious
allophonic rules can be eliminated by a linguistically motivated
filtering mechanism based on a phonetic representation of segments.
We discuss the role of a priori linguistic knowledge in the statistical
learning of phonology.},
doi = {10.1016/j.cognition.2005.10.006},
keywords = {Humans, Linguistics, Models, Phonetics, Statistical, Verbal Learning, 16364279},
}
Downloads: 0
{"_id":"v3Dt5rmAhTvwvX8hu","bibbaseid":"peperkamp-calvez-nadal-dupoux-theacquisitionofallophonicrulesstatisticallearningwithlinguisticconstraints-2006","author_short":["Peperkamp, S.","Calvez, R. L.","Nadal, J.","Dupoux, E."],"bibdata":{"bibtype":"article","type":"article","author":[{"firstnames":["Sharon"],"propositions":[],"lastnames":["Peperkamp"],"suffixes":[]},{"firstnames":["Rozenn","Le"],"propositions":[],"lastnames":["Calvez"],"suffixes":[]},{"firstnames":["Jean-Pierre"],"propositions":[],"lastnames":["Nadal"],"suffixes":[]},{"firstnames":["Emmanuel"],"propositions":[],"lastnames":["Dupoux"],"suffixes":[]}],"journal":"Cognition","title":"The acquisition of allophonic rules: statistical learning with linguistic constraints.","year":"2006","number":"3","pages":"B31-41","volume":"101","abstract":"Phonological rules relate surface phonetic word forms to abstract underlying forms that are stored in the lexicon. Infants must thus acquire these rules in order to infer the abstract representation of words. We implement a statistical learning algorithm for the acquisition of one type of rule, namely allophony, which introduces context-sensitive phonetic variants of phonemes. This algorithm is based on the observation that different realizations of a single phoneme typically do not appear in the same contexts (ideally, they have complementary distributions). In particular, it measures the discrepancies in context probabilities for each pair of phonetic segments. In Experiment 1, we test the algorithm's performances on a pseudo-language and show that it is robust to statistical noise due to sampling and coding errors, and to non-systematic rule application. In Experiment 2, we show that a natural corpus of semiphonetically transcribed child-directed speech in French presents a very large number of near-complementary distributions that do not correspond to existing allophonic rules. These spurious allophonic rules can be eliminated by a linguistically motivated filtering mechanism based on a phonetic representation of segments. We discuss the role of a priori linguistic knowledge in the statistical learning of phonology.","doi":"10.1016/j.cognition.2005.10.006","keywords":"Humans, Linguistics, Models, Phonetics, Statistical, Verbal Learning, 16364279","bibtex":"@Article{Peperkamp2006,\n author = {Sharon Peperkamp and Rozenn Le Calvez and Jean-Pierre Nadal and Emmanuel Dupoux},\n journal = {Cognition},\n title = {The acquisition of allophonic rules: statistical learning with linguistic constraints.},\n year = {2006},\n number = {3},\n pages = {B31-41},\n volume = {101},\n abstract = {Phonological rules relate surface phonetic word forms to abstract\n\tunderlying forms that are stored in the lexicon. Infants must thus\n\tacquire these rules in order to infer the abstract representation\n\tof words. We implement a statistical learning algorithm for the acquisition\n\tof one type of rule, namely allophony, which introduces context-sensitive\n\tphonetic variants of phonemes. This algorithm is based on the observation\n\tthat different realizations of a single phoneme typically do not\n\tappear in the same contexts (ideally, they have complementary distributions).\n\tIn particular, it measures the discrepancies in context probabilities\n\tfor each pair of phonetic segments. In Experiment 1, we test the\n\talgorithm's performances on a pseudo-language and show that it is\n\trobust to statistical noise due to sampling and coding errors, and\n\tto non-systematic rule application. In Experiment 2, we show that\n\ta natural corpus of semiphonetically transcribed child-directed speech\n\tin French presents a very large number of near-complementary distributions\n\tthat do not correspond to existing allophonic rules. These spurious\n\tallophonic rules can be eliminated by a linguistically motivated\n\tfiltering mechanism based on a phonetic representation of segments.\n\tWe discuss the role of a priori linguistic knowledge in the statistical\n\tlearning of phonology.},\n doi = {10.1016/j.cognition.2005.10.006},\n keywords = {Humans, Linguistics, Models, Phonetics, Statistical, Verbal Learning, 16364279},\n}\n\n","author_short":["Peperkamp, S.","Calvez, R. L.","Nadal, J.","Dupoux, E."],"key":"Peperkamp2006","id":"Peperkamp2006","bibbaseid":"peperkamp-calvez-nadal-dupoux-theacquisitionofallophonicrulesstatisticallearningwithlinguisticconstraints-2006","role":"author","urls":{},"keyword":["Humans","Linguistics","Models","Phonetics","Statistical","Verbal Learning","16364279"],"metadata":{"authorlinks":{}}},"bibtype":"article","biburl":"https://endress.org/publications/ansgar.bib","dataSources":["xPGxHAeh3vZpx4yyE","TXa55dQbNoWnaGmMq"],"keywords":["humans","linguistics","models","phonetics","statistical","verbal learning","16364279"],"search_terms":["acquisition","allophonic","rules","statistical","learning","linguistic","constraints","peperkamp","calvez","nadal","dupoux"],"title":"The acquisition of allophonic rules: statistical learning with linguistic constraints.","year":2006}