Paradigm Completion for Derivational Morphology

2017-08-30

Conference on Empirical Methods in Natural Language Processing

AI Summary

This paper addresses the long-overlooked problem of complex derivational word formation in NLP by introducing— for the first time—the novel task of *derivational paradigm completion*, establishing its theoretical correspondence with inflectional paradigm completion. Methodologically, it adapts state-of-the-art neural sequence-to-sequence models—originally designed for inflection—to derivational modeling, incorporating morphological constraints to enhance structural well-formedness. Experiments demonstrate that the proposed model outperforms non-neural baselines by 16.4%, providing the first systematic empirical validation that data-driven approaches can effectively learn derivational regularities. The core contributions are threefold: (1) formal definition of the derivational paradigm completion task; (2) empirical confirmation of the feasibility of neural modeling for derivational morphology; and (3) identification of semantic and historical factors as key performance constraints, thereby laying both theoretical and empirical foundations for future research on derivational morphology.

Abstract

The generation of complex derived word forms has been an overlooked problem in NLP; we fill this gap by applying neural sequence-to-sequence models to the task. We overview the theoretical motivation for a paradigmatic treatment of derivational morphology, and introduce the task of derivational paradigm completion as a parallel to inflectional paradigm completion. State-of-the-art neural models adapted from the inflection task are able to learn the range of derivation patterns, and outperform a non-neural baseline by 16.4%. However, due to semantic, historical, and lexical considerations involved in derivational morphology, future work will be needed to achieve performance parity with inflection-generating systems.

Problem

• Generating complex derived word forms
• Applying neural sequence-to-sequence models
• Improving performance in derivational morphology

Innovation

• Neural sequence-to-sequence models
• Derivational paradigm completion
• Outperform non-neural baseline