As is known, the fundamental frequencies (F0) of the vowels following aspirated or lenis stops have become associated with the aspirated~lenis stop contrast while Voice Onset Time (VOT) values of them became merged in Seoul Korean. Previous studies found the effects of age, gender, lexical frequency, and vowel height. However, although lexical competition has been demonstrated to affect the trajectory of sound change regarding contrastivity, it has not been considered in this context. The present study examines the effects of lexical competition on this sound change. Through a production experiment, analyses demonstrate that the aspirated~lenis contrast is hyperarticulated in minimal pairs for both VOT and F0, but only in aspirated stops with following high vowels. Moreover, speakers advanced in the sound change produce lower F0 values for lenis stops with following non-high vowels if an aspirated competitor exists. We find that the F0 distinction in lenis stops is more hyperarticulated in speakers advanced in the sound change, but that speakers appear to hyperarticulate VOT regardless of how much of a VOT contrast they normally produce themselves. This may be related to the fact that VOT is still a robust cue to the aspirated~lenis distinction in much of the speech community.

No abstract

This paper presents our submission to the SIGMORPHON 2023 task 2 of Cognitively Plausible Morphophonological Generalization in Korean. We implemented both Linear Discriminative Learning and Transformer models and found that the Linear Discriminative Learning model trained on a combination of corpus and experimental data showed the best performance with the overall accuracy of around 83%. We found that the best model must be trained onboth corpus data and the experimental data of one particular participant. Our examination of speaker-variability and speaker-specific information did not explain why a particular participant combined well with the corpus data.We recommend Linear Discriminative Learning models as a future non-neural baseline system, owning to its training speed, accuracy, model interpretability and cognitive plausibility. In order to improve the model performance, we suggest using bigger data and/or performing data augmentation and incorporating speaker and item-specifics considerably.