Section: New Results

Test of the psychological validity of AI algorithms.

In this section, we focus on the utilisation of machine learning algorithms of speech and language processing to derive testable quantitative predictions in humans (adults or infants).

• In [24], we compare the performance of humans (English and French listeners) versus an unsupervised speech model in a perception experiment (ABX discrimination task). Although the ABX task has been used for acoustic model evaluation in previous research, the results have not, until now, been compared directly with human behaviour in an experiment. We show that a standard, well-performing model (DPGMM) has better accuracy at predicting human responses than the acoustic baseline. The model also shows a native language effect, better resembling native listeners of the language on which it was trained. However, the native language effect shown by the models is different than the one shown by the human listeners, and, notably , the models do not show the same overall patterns of vowel confusions.

• Word learning relies on the ability to master the sound con-trasts that are phonemic (i.e., signal meaning difference) in a given language. Though the timeline of phoneme development has been studied extensively over the past few decades, the mechanism of this development is poorly understood. In [20], we take inspiration from computational modeling work in language grounding where phonetic and visual information is learned jointly. In this study, we varied the taxonomic distance of pairs of objects and tested how adult learners judged the phonemic status of the sound contrast associated with each of these pairs. We found that judgments were sensitive to gradients in the taxonomic structure, suggesting that learners use probabilistic information at the semantic level to optimize the accuracy of their judgements at the phonological level. The findings provide evidence for an interaction between phonological learning and meaning generalization in human learning.