Modeling morphology with Linear Discriminative Learning: considerations and design choices

Abstract

This study addresses a series of methodological questions that arise when modeling inflectional morphology with Linear Discriminative Learning. Taking the semi-productive German noun system as example, we illustrate how decisions made about the representation of form and meaning influence model performance. We clarify that for modeling frequency effects in learning, it is essential to make use of incremental learning rather than the endstate of learning. We also discuss how the model can be set up to approximate the learning of inflected words in context. In addition, we illustrate how in this approach the wug task can be modeled in considerable detail. In general, the model provides an excellent memory for known words, but appropriately shows more limited performance for unseen data, in line with the semi-productivity of German noun inflection and generalization performance of native German speakers.

arXiv comment38 pages, 5 figures, 10 tables; acknowledgements added

References in corpus (5)

• Efficient Estimation of Word Representations in Vector Space
• Distributed Representations of Words and Phrases and their Compositionality
• Syntactic Structure from Deep Learning
• The Greedy and Recursive Search for Morphological Productivity
• Inflecting when there's no majority: Limitations of encoder-decoder neural networks as cognitive models for German plurals