Join us for mini-talks by ILST trainees.
Annika Heuser – “Information theoretic voicing hypothesis generation”
Phonemic voicing contrasts differ cross-linguistically, both in terms of the nature of the contrasts and their phonetic realizations. Children must learn which perceptual cues are helpful for distinguishing voiced vs. voiceless phonemes as they are realized in various contexts. Using Standard American English (SAE) as a case study, we generated hypotheses of which cues are most informative for acquisition of the voicing contrast. More specifically, we classified SAE obstruents as voiced vs voiceless using decision trees trained and tested on TIMIT. We validated our results against the findings of previous perceptual studies and we gleaned more specific hypotheses to help design future experiments on children’s acquisition of the voicing contrast.
Daoxin Li – “Modelling the distributional learning of verb argument structure”
Linguists agree there are systematic mappings between the syntax and semantics of a verb, and it is evident that children know these mapping rules from a young age. This knowledge is unlikely to be entirely universal or innate given the considerable variabilities across languages and idiosyncrasies within. In this work, we present a computational model that automatically learns productive rules between syntax and semantics based on the Tolerance/Sufficiency Principle, demonstrating that the systematic mappings are learnable without assuming any prior associations.