Computational analysis of morphology remains unsolved

Establish a complete and effective solution to the computational analysis of morphology.

Background

The paper introduces labeled morphological segmentation (LMS) and the Chipmunk semi-CRF model to jointly model segmentation and morphotactics, demonstrating improvements across multiple languages and tasks (segmentation, stemming, and morphological tag classification). Despite these advances, the authors note in the 2024 Retrospective that the broader goal of computational morphology remains unresolved.

This underscores that while specific components (e.g., supervised segmentation with morphotactics via semi-CRFs) show measurable progress, a comprehensive, generally effective solution to computational morphological analysis across settings has not yet been achieved.

References

All in all, 9 years later, the first author at least still finds the computational analysis of morphology a challenging and unsolved problem.

— Labeled Morphological Segmentation with Semi-Markov Models  (2404.08997 - Cotterell et al., 2024) in Retrospective, final sentence

A further problem is that it is far from clear how the cognitive system would know when to set up the individual transformation matrices.

— Compound interpretation is based on analogy  (2610.01688 - Shen et al., 1 Oct 2026) in General Discussion, paragraph beginning “Fourth, the CAOSS model is conceptualized...”