Jingyi Chen

Jingyi Chen

Juggling equations and caffeine with equal enthusiasm.

Posts by Collection

portfolio

publications

ALIGNMENT_CONTRAST_IN_HUIYANG_HAKKA_FALLING_TONES

Published in The 96th Annual Meeting of the Linguistic Society of America, 2022

The aim of this paper is to examine the role of tonal alignment in the perception and production of two falling tones in Huiyang Hakka. Whether tonal alignment can be contrastive in contour tones within a language is a longstanding debate in tonal typology. Previous research indicates that tonal alignment can differ systematically in production; however, the question of whether tonal alignment can be (phonologically) contrastive in contour tones is still unresolved. In this study, we explored whether tonal alignment is contrastive in both production and perception of falling tones. The results of a production experiment show that the two falling tones in Huiyang Hakka have a significant difference in tonal alignment; however, tone duration also differs consis tently between the two falling tones. To find out more about the perceptual relevance of these correlates, I conducted a perception experiment to test which phonetic param eter(s) is/are the major correlate that native speakers use to encode the tonal contrast between their falling tones. The results indicate that tonal alignment is perceptually contrastive in Huiyang Hakka falling tones, providing experimental evidence of both production and perceptual relevance of alignment contrasts.

Recommended citation: Chen, Jingyi. (2022). ALIGNMENT CONTRAST IN HUIYANG HAKKA FALLING TONES. https://www.researchgate.net/publication/371938607_ALIGNMENT_CONTRAST_IN_HUIYANG_HAKKA_FALLING_TONES

Exploring_How_Generative_Adversarial_Networks_Learn_Phonological_Representations

Published in Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), 2023

This paper explores how Generative Adversarial Networks (GANs) learn representations of phonological phenomena. We analyze how GANs encode contrastive and non-contrastive nasality in French and English vowels by applying the ciwGAN architecture (Begus, 2021). Begus claims that ciwGAN encodes linguistically meaningful representations with categorical variables in its latent space and manipulating the latent variables shows an almost one to one corresponding control of the phonological features in ciwGAN’s generated outputs. However, our results show an interactive effect of latent variables on the features in the generated outputs, which suggests the learned representations in neural networks are different from the phonological representations proposed by linguists. On the other hand, ciwGAN is able to distinguish contrastive and noncontrastive features in English and French by encoding them differently. Comparing the performance of GANs learning from different languages results in a better understanding of what language specific features contribute to developing language specific phonological representations. We also discuss the role of training data frequencies in phonological feature learning.

Recommended citation: Chen, Jingyi & Elsner, Micha. (2023). Exploring How Generative Adversarial Networks Learn Phonological Representations. 3115-3129. 10.18653/v1/2023.acl-long.175. https://aclanthology.org/2023.acl-long.175.pdf

talks

teaching

LING2000 Introduction to Linguistics

Undergraduate course, Ohio State University, Linguistics Department, 2020

From AU2020 to AU2022, I worked as instructor of undergraduate course Introduction to Linguistics