Competence Collapse in Code-Mixed Generation: Spectral Evidence and Mechanistic Recovery via Cross-Lingual Activation Steering

Tanushree Ravindra Pratap Yadav


Abstract
As Large Language Models (LLMs) approachhuman-level reasoning in English, their performance in low-resource, code-mixed languagesremains surprisingly brittle. We identify Competence Collapse, a distinct pathology wheremodels capable of complex reasoning in English exhibit severe utility degradation whenprompted in Hinglish (Hindi-English). Wequantify this as a Service Gap, observing astatistically significant decline in instructionalquality (∆D ≈ −11.3%, p < 0.001) across9 diverse architectures. Spectral analysis suggests that this stems from a representationaldivergence between the model’s High-UtilityDirection and its Generation Subspace. Tobridge this gap, we propose Cross-LingualActivation Steering (CLAS), an inferencetime intervention that injects a "CompetenceGap Vector" into the residual stream. Evaluated across 6 open-weight models (using alightweight calibration set, N = 50), CLASrecovered utility by ∆D = +2.22 (d = 0.60)while preserving code-mixed fidelity (CMI ≈0.4) and reinforcing safety protocols.
Anthology ID:
2026.loreslm-1.3
Volume:
Proceedings of the Second Workshop on Language Models for Low-Resource Languages (LoResLM 2026)
Month:
March
Year:
2026
Address:
Rabat, Morocco
Editors:
Hansi Hettiarachchi, Tharindu Ranasinghe, Alistair Plum, Paul Rayson, Ruslan Mitkov, Mohamed Gaber, Damith Premasiri, Fiona Anting Tan, Lasitha Uyangodage
Venue:
LoResLM
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
29–40
Language:
URL:
https://aclanthology.org/2026.loreslm-1.3/
DOI:
Bibkey:
Cite (ACL):
Tanushree Ravindra Pratap Yadav. 2026. Competence Collapse in Code-Mixed Generation: Spectral Evidence and Mechanistic Recovery via Cross-Lingual Activation Steering. In Proceedings of the Second Workshop on Language Models for Low-Resource Languages (LoResLM 2026), pages 29–40, Rabat, Morocco. Association for Computational Linguistics.
Cite (Informal):
Competence Collapse in Code-Mixed Generation: Spectral Evidence and Mechanistic Recovery via Cross-Lingual Activation Steering (Yadav, LoResLM 2026)
Copy Citation:
PDF:
https://aclanthology.org/2026.loreslm-1.3.pdf