Algebraic Classification of Reduplicative Processes

Matthew Hayden


Abstract
This paper offers an updated perspective on the computational complexity of reduplication. Since one-way deterministic transducers cannot model reduplication in a straightforward way, the phenomenon has long been considered the outlier of morphology from a complexity perspective. Drawing on algebraic methods, I show that the vast majority of reduplicative processes belong to a few remarkably simple classes of subregular functions. A detailed study of the RedTyp database (Dolatian and Heinz, 2019) reveals that 100% of the surveyed reduplicative processes correspond to string-to-string functions in the class DA, while over98% are locally testable (LJ1) and over 87% are locally trivial (L1). These results indicate a new upper bound on the complexity of reduplication that is comparable to that of morphological processes in general.
Anthology ID:
2026.scil-main.42
Volume:
Proceedings of the Society for Computation in Linguistics 2026
Month:
July
Year:
2026
Address:
San Diego, CA
Editors:
Rob Voigt, Alex Warstadt, Naomi Feldman, Tal Linzen
Venues:
SCiL | WS
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
447–459
Language:
URL:
https://preview.aclanthology.org/ingest-acl-workshops/2026.scil-main.42/
DOI:
Bibkey:
Cite (ACL):
Matthew Hayden. 2026. Algebraic Classification of Reduplicative Processes. In Proceedings of the Society for Computation in Linguistics 2026, pages 447–459, San Diego, CA. Association for Computational Linguistics.
Cite (Informal):
Algebraic Classification of Reduplicative Processes (Hayden, SCiL 2026)
Copy Citation:
PDF:
https://preview.aclanthology.org/ingest-acl-workshops/2026.scil-main.42.pdf