English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

This paper models unsupervised learning of an identity-based pattern (orcopying) in speech called reduplication from raw continuous data with deepconvolutional neural networks. We use the ciwGAN architecture Begu\v{s} (2021a;arXiv:2006.02951) in which learning of meaningful representations in speechemerges from a requirement that the CNNs generate informative data. We proposea technique to wug-test CNNs trained on speech and, based on four generativetests, argue that the network learns to represent an identity-based pattern inits latent space. By manipulating only two categorical variables in the latentspace, we can actively turn an unreduplicated form into a reduplicated formwith no other substantial changes to the output in the majority of cases. Wealso argue that the network extends the identity-based pattern to unobserveddata. Exploration of how meaningful representations of identity-based patternsemerge in CNNs and how the latent space variables outside of the training rangecorrelate with identity-based patterns in the output has general implicationsfor neural network interpretability.

Identity-Based Patterns in Deep Convolutional Networks: Generative Adversarial Phonology and Reduplication