Exploring How Generative Adversarial Networks Learn Phonological Representations
It addresses the problem of understanding neural network representations in linguistics, providing insights into language-specific feature learning, but is incremental as it builds on existing ciwGAN methods.
This paper investigates how Generative Adversarial Networks (GANs) learn phonological representations, specifically analyzing contrastive and non-contrastive nasality in French and English vowels using the ciwGAN architecture, finding that GANs encode these features differently but with interactive effects that differ from linguistic models.
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 2021a). 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.