Lexical exceptionality in paradigm-specific learning: modeling stem-final obstruent alternations in Korean verbs and adjectives

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Science & Research — Social Sciences & Behavioral Studies, Artificial Intelligence & Machine Learning, Research Methodology & Innovation · Depth: Expert, quick

Summary

Stella Eunsoo Hong's study investigates how lexically-specific phonological patterns are acquired in Korean verbs and adjectives, focusing on stem-final obstruent alternations. Specifically, it examines /p/-, /t/-, and /s/-final stems that undergo intervocalic lenition before vowel-initial suffixes, where irregular classes exhibit distinct alternation patterns. The research employs a lexically scaled MaxEnt model, building on work by Linzen et al. (2013) and Hughto et al. (2019), to learn these paradigm-specific alternations. Simulations were run under two scenarios: one where repair strategies occur at equal frequencies, and another where one strategy significantly outnumbers others. Results indicate that the model favors a "least-cost" solution, treating statistically dominant morpheme classes as the general pattern. The study discusses the model's sensitivity to lexical statistics and its implications for language acquisition.

Key takeaway

For research scientists modeling language acquisition or phonological processes, understanding how lexical statistics influence learning is crucial. Your models should account for the "least-cost" solution bias, where statistically dominant morpheme classes are generalized as the primary pattern. This insight suggests that frequency data is a powerful predictor for how learners resolve ambiguity when multiple repair strategies are available. Consider incorporating lexical frequency weighting into your computational models to better reflect natural language acquisition.

Key insights

Lexical statistics guide phonological pattern acquisition, favoring least-cost solutions for dominant morpheme classes.

Principles

• Lexical exceptionality interacts with phonological processes.
• Learners identify roots triggering lenition and repair strategies.
• Models favor statistically dominant patterns as general.

Method

A lexically scaled MaxEnt model learns paradigm-specific alternations by simulating acquisition under varying repair strategy frequencies.

In practice

• Model acquisition of complex phonological rules.
• Analyze how frequency impacts linguistic generalization.

Topics

• Korean Phonology
• Lexical Exceptionality
• MaxEnt Models
• Language Acquisition
• Obstruent Alternations
• Morphological Learning

Best for: NLP Engineer, AI Scientist, Research Scientist

Related on AIssential

• Modelling the Morphology of Verbal Paradigms: A Case Study in the Tokenization of Turkish and Hebrew — Tokenization strategy critically impacts Transformer models' ability to represent complex morphology, especially in non-concatenative languages.
• A Scalable Tool for Measuring Manner and Result Verbs in Developmental Language Research — A RoBERTa classifier, trained on LLM-generated annotations, accurately distinguishes manner and result verbs at scale.
• A Theory of When and How Learners Construct Tiers: Implications for Opaque and Transparent Vowels — Learners construct phonological tiers based on specific input conditions, impacting vowel transparency and opacity.
• Semantic Chunking and the Entropy of Natural Language — A new statistical model explains natural language redundancy through hierarchical semantic chunking and predicts variable entropy rates.
• Probing Semantic Alignment, Lexical Invariance, and Syntactic Influence in LLM Metaphor Processing — LLMs struggle with deep metaphor comprehension, often misinterpreting concepts and relying on lexical associations over context or full syntactic understanding.
• A computational study of how Sanskrit-specific morphophonological phenomena behave under noisy spoken input — The study computationally analyzes Sanskrit morphophonology's resilience to noisy speech.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.