Modelling English diphthongs with dynamic articulatory targets

The nature of English diphthongs has been much disputed. Bynow, the most influential account argues that diphthongs arephoneme entities rather than vowel combinations. However,mixed results have been reported regarding whether the rate offormant transition is the most reliable attribute in the perceptionand production of diphthongs. Here, we used computationalmodelling to explore the underlying forms of diphthongs. Wetested the assumption that diphthongs have dynamicarticulatory targets by training an articulatory synthesiser witha three-dimensional (3D) vocal tract model to learn Englishwords. An automatic phoneme recogniser was constructed toguide the learning of the diphthongs. Listening experiments bynative listeners indicated that the model succeeded in learninghighly intelligible diphthongs, providing support for thedynamic target assumption. The modelling approach paves anew way for validating hypotheses of speech perception andproduction.