Mind Your Moras: Orthography-Aware Error Analysis of Neural Japanese Morphological Generation

• URL: http://arxiv.org/abs/2605.20043v1
• Date: Tue, 19 May 2026 16:04:33 GMT
• Title: Mind Your Moras: Orthography-Aware Error Analysis of Neural Japanese Morphological Generation
• Abstract summary: We present an orthography-aware error analysis of Japanese past-tense morphological inflection.<n>We evaluate two character-level sequence-to-sequence architectures on past-tense formation using datasets formatted according to the SIGMORPHON 2020 and 2023 shared task conventions.<n>Despite high aggregate accuracy, models exhibit systematic, linguistically interpretable errors that cluster around specific orthographic properties of hiragana.
• Score: 6.1031222024562695
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: We present an orthography-aware error analysis of Japanese past-tense morphological inflection, treating hiragana not merely as a transcriptional medium, but as a representational system encoding morphophonological distinctions that may influence model generalization. We evaluate two character-level sequence-to-sequence architectures on past-tense formation using datasets formatted according to the SIGMORPHON 2020 and 2023 shared task conventions. Despite high aggregate accuracy, models exhibit systematic, linguistically interpretable errors that cluster around specific orthographic properties of hiragana. We introduce a concise error taxonomy capturing seven primary failure modes and provide both quantitative and qualitative analyses. Gemination-related errors dominate residual failures, accounting for 75-80% of errors, particularly in verbs whose stems end in the vowel e and require gemination before the past-tense suffix. Error patterns remain highly consistent across architectures and random seeds, suggesting a robust interaction between orthographic representation, morphological structure, and data frequency effects in shaping model generalization. These results underscore the necessity of orthography-aware evaluation for understanding neural generalization in morphologically complex languages.

Related papers

• PACUTE: Phonology-, Affix-, and Character-level Understanding of Tokens for Filipino [1.6481883307499954]
  PACUTE is a diagnostic benchmark of 4,600 tasks designed to evaluate morphological understanding in Filipino.<n>We find that open-weight models perform near chance on morpheme decomposition regardless of scale.<n>These results identify productive morphological composition, rather than character access alone, as the persistent bottleneck for Filipino word-structure understanding.
  arXiv  Detail & Related papers  (2026-06-13T06:12:56Z)
• When Irregularity Helps: A Subclass Analysis of Inductive Bias in Neural Morphology [6.1031222024562695]
  We show that a very small, structurally specific irregular subtype accounts for a disproportionate share of model errors.<n>These findings suggest that error concentration is driven by the interaction between extreme low-frequency morphological patterns and specific morphophonological processes.
  arXiv  Detail & Related papers  (2026-05-19T23:18:47Z)
• Disaggregation Reveals Hidden Training Dynamics: The Case of Agreement Attraction [6.531546527140474]
  Language models generally produce grammatical text, but they are more likely to make errors in certain contexts.<n>We show that it is possible to better understand the intermediate stages of grammatical learning in language models.<n>We argue that taking this approach to analyzing language model behavior more generally can serve as a powerful tool for understanding the intermediate learning phases, overall training dynamics, and the specific generalizations learned by language models.
  arXiv  Detail & Related papers  (2025-10-28T19:59:26Z)
• Boosting Semi-Supervised Scene Text Recognition via Viewing and Summarizing [71.29488677105127]
  Existing scene text recognition (STR) methods struggle to recognize challenging texts, especially for artistic and severely distorted characters. We propose a contrastive learning-based STR framework by leveraging synthetic and real unlabeled data without any human cost. Our method achieves SOTA performance (94.7% and 70.9% average accuracy on common benchmarks and Union14M-Benchmark.
  arXiv  Detail & Related papers  (2024-11-23T15:24:47Z)
• Morphological Inflection with Phonological Features [7.245355976804435]
  This work explores effects on performance obtained through various ways in which morphological models get access to subcharacter phonological features. We elicit phonemic data from standard graphemic data using language-specific grammars for languages with shallow grapheme-to-phoneme mapping.
  arXiv  Detail & Related papers  (2023-06-21T21:34:39Z)
• Improving Pre-trained Language Models with Syntactic Dependency Prediction Task for Chinese Semantic Error Recognition [52.55136323341319]
  Existing Chinese text error detection mainly focuses on spelling and simple grammatical errors. Chinese semantic errors are understudied and more complex that humans cannot easily recognize.
  arXiv  Detail & Related papers  (2022-04-15T13:55:32Z)
• Modeling Target-Side Morphology in Neural Machine Translation: A Comparison of Strategies [72.56158036639707]
  Morphologically rich languages pose difficulties to machine translation. A large amount of differently inflected word surface forms entails a larger vocabulary. Some inflected forms of infrequent terms typically do not appear in the training corpus. Linguistic agreement requires the system to correctly match the grammatical categories between inflected word forms in the output sentence.
  arXiv  Detail & Related papers  (2022-03-25T10:13:20Z)
• Comparative Error Analysis in Neural and Finite-state Models for Unsupervised Character-level Transduction [34.1177259741046]
  We compare the two model classes side by side and find that they tend to make different types of errors even when achieving comparable performance. We investigate how combining finite-state and sequence-to-sequence models at decoding time affects the output quantitatively and qualitatively.
  arXiv  Detail & Related papers  (2021-06-24T00:09:24Z)
• Minimal Supervision for Morphological Inflection [8.532288965425805]
  We bootstrapping labeled data from a seed as little as em five labeled paradigms, accompanied by a large bulk of unlabeled text. Our approach exploits different kinds of regularities in morphological systems in a two-phased setup. We experiment with the Paradigm Cell Filling Problem over eight typologically different languages, and find that, in languages with relatively simple morphology, orthographic regularities on their own allow inflection models to achieve respectable accuracy.
  arXiv  Detail & Related papers  (2021-04-17T11:07:36Z)
• Compositional Generalization via Semantic Tagging [81.24269148865555]
  We propose a new decoding framework that preserves the expressivity and generality of sequence-to-sequence models. We show that the proposed approach consistently improves compositional generalization across model architectures, domains, and semantic formalisms.
  arXiv  Detail & Related papers  (2020-10-22T15:55:15Z)
• On Long-Tailed Phenomena in Neural Machine Translation [50.65273145888896]
  State-of-the-art Neural Machine Translation (NMT) models struggle with generating low-frequency tokens. We propose a new loss function, the Anti-Focal loss, to better adapt model training to the structural dependencies of conditional text generation. We show the efficacy of the proposed technique on a number of Machine Translation (MT) datasets, demonstrating that it leads to significant gains over cross-entropy.
  arXiv  Detail & Related papers  (2020-10-10T07:00:57Z)
• Mechanisms for Handling Nested Dependencies in Neural-Network Language Models and Humans [75.15855405318855]
  We studied whether a modern artificial neural network trained with "deep learning" methods mimics a central aspect of human sentence processing. Although the network was solely trained to predict the next word in a large corpus, analysis showed the emergence of specialized units that successfully handled local and long-distance syntactic agreement. We tested the model's predictions in a behavioral experiment where humans detected violations in number agreement in sentences with systematic variations in the singular/plural status of multiple nouns.
  arXiv  Detail & Related papers  (2020-06-19T12:00:05Z)
• Evaluating Neural Morphological Taggers for Sanskrit [23.87775187927048]
  We evaluate the efficacy of four standard sequence labelling models on Sanskrit. We find that although some neural models perform better than others, one of the common causes for error for all of these models is syncretism.
  arXiv  Detail & Related papers  (2020-05-21T20:36:32Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.