An Approach to Measure Pronunciation Similarity in Second Language Learning Using Radial Basis Function Kernel
Christos Koniaris · 2014
This paper shows a method to diagnose potential mispronunciations in second language learning by studying the characteristics of the speech produced by a group of native speakers and the speech produced by various non-native groups of speakers from diverse language backgrounds. The method compares the native auditory perception and the non-native spectral representation on the phoneme level using similarity measures that are based on the radial basis function kernel. A list of ordered problematic phonemes is found for each non-native group of speakers and the results are analyzed based on a relevant linguistic survey found in the literature. The experimental results indicate an agreement with linguistic findings of up to 80.8% for vowels and 80.3% for consonants.