Search papers, labs, and topics across Lattice.
Department of English Language and Linguistics, Heinrich Heine University D眉sseldorf
3
0
3
Character-level transformers encode complex morphological patterns as item-specific abstractions, but they struggle with generalization like humans do.
A freely available, offline-first keyboard suite now enables digital communication in the endangered Idu Mishmi language, preserving its unique orthography.
Despite capturing irregular morphological patterns during training, transformers fail to generalize these patterns to novel forms in a human-like way, highlighting a key difference between statistical learning and true morphological abstraction.
