Yang Wang: On regular copying languages

Yang Wang: On regular copying languages

🎙 Yang Wang 👥 3K 📅 July 23, 2026 ⏱ 43 min 👁 38 📄 original study 🧭 2026-08-16
Available in: English (current) Français

Keywords

reduplicationregular languagescopyingfinite-state buffer machinesformal language theory

Summary

Yang Wang presents her joint work with Tim Hunter on regular copying languages, proposing a new formal model called finite-state buffer machines (FSBMs) to characterize reduplication in natural language. The talk begins by motivating reduplication as a common and important morphological phenomenon, citing typological data from WALS showing that over 300 languages exhibit productive reduplication. She contrasts reduplication with string reversal, noting that while copying is common, reversal is rare and confined to language games. The central puzzle is that reduplication requires context-sensitive power, yet reversal, which is context-free, is virtually unattested. The proposed FSBMs extend finite automata with a queue buffer and modes for buffering and matching, allowing recognition of the copy language WW while excluding reversal and crossing dependencies. The talk covers formal definitions, examples, closure properties (closed under union, concatenation, Kleene star, homomorphism, and intersection with regular languages; not closed under inverse homomorphism, intersection, or complement), and a pumping lemma. The paper also discusses typological implications and determinism. The talk concludes with a Q&A session and references to the published paper in the Journal of Language Modelling.

180 words

Critical Evaluation

Value of the Information & Strength of the Argument

The talk provides a novel formal characterization of reduplication, addressing a long-standing challenge in computational linguistics. The argumentation is solid: the speaker clearly motivates the problem, presents a precise formal model, and demonstrates its properties with examples and closure results. The value lies in offering a framework that captures copying while excluding unattested patterns like reversal, aligning with linguistic typology. The presentation is logically structured, though some technical details are condensed due to time constraints.

Scientific Rigor, Source Quality, Title Accuracy

The talk is scientifically rigorous, based on a published paper in the Journal of Language Modelling. The speaker cites relevant literature (e.g., Johnson, Kaplan, Kay, Gasser and Pulum) and provides a formal proof of a pumping lemma. The title accurately reflects the content. The presentation includes a clear definition of FSBMs and discusses closure properties, demonstrating methodological soundness. The sources are appropriate and the claims are well-supported.

157 words

Title / Content Match

The title accurately reflects the content, which focuses on regular copying languages and their formal characterization.

Quality & Reliability

8/10

The talk presents original research with formal definitions, proofs, and published paper. The speaker is an expert in linguistics and formal language theory. The content is rigorous and well-structured, though the presentation is concise and assumes background knowledge.

Key Moments

Cited Sources

Concurring Sources

Contribution & Novelties

The talk introduces a novel formal model, finite-state buffer machines, that characterizes reduplication as a regular copying language, filling a gap in formal language theory. This provides a precise upper bound for morphological and phonological patterns involving copying, while excluding unattested patterns like reversal. The work offers a new perspective on the expressive power needed for natural language morphology.

Pour aller plus loin :

97 words

Radar Profile

The radar profile shows high scores in quality of information, technical level, and reliability, with slightly lower but still strong scores in quantity of information. This indicates a technically dense and reliable presentation, though it may be challenging for a general audience.

Reliability 8/10