EPS
← All batches·2605.16758

Language Acquisition Device in Large Language Models

topic: current_projecttop score: 100released: 2026-05-19first surfaced: 2026-05-19arXivPDFlinked_to_results2026-05-19

Authors: Masato Mita, Taiga Someya, Ryo Yoshida et al.

arXiv · PDF

Summary

arXiv:2605. 16758v1 Announce Type: new Abstract: Large Language Models (LLMs) remain substantially less data-efficient than humans.

Relevance

Read next because Language Acquisition Device in Large Language Models overlaps with clean result "LoRA persona trained on alone emits at 23.5% when a co-trained partner learns ..., vs 0% control on Qwen2.5-7B-Instruct (MODERATE confidence)", clean result "Leakage rate is a usable signal for recovering trigger-shaped phrases on Gaperon-1125-1B without knowing the hidden trigger itself (MODERATE confidence)", clean result "Language-mismatch LoRA SFT on Qwen2.5-7B leaks the trained completion language into bystander directives the model was never trained on, absent under same-language SFT (LOW confidence)". Matching terms: code, strong, token, line, position, language, model. Source: arxiv cs.CL (NLP).

Abstract

arXiv:2605.16758v1 Announce Type: new Abstract: Large Language Models (LLMs) remain substantially less data-efficient than humans. Pre-pretraining (PPT) on synthetic languages has been proposed to close this gap, with prior work emphasizing highly expressive formal languages such as $k$-Shuffle Dyck. Inspired by the Language Acquisition Device (LAD) hypothesis, which posits that innate constraints preemptively restrict the learner's hypothesis space to natural-language-like structure, we propose LAD-inspired PPT: pre-pretraining on MP-STRUCT, a formal language whose strings encode hierarchical composition, feature-based dependencies, and long-distance displacement via MERGE, AGREE, and MOVE. A brief 500-step PPT with MP-STRUCT matches strong formal-language baselines in token efficiency while additionally imparting a human-like resistance to structurally implausible languages (e.g., REVERSE). Analyzing simplified variants, we find that MP-STRUCT CORE outperforms $k$-Shuffle Dyck despite not being definable in C-RASP (a formal bound on transformer expressivity), challenging the prior hypothesis that effective PPT languages must be both hierarchically expressive and circuit-theoretically learnable. We show that functional landmarks, which reduce dependency resolution ambiguity, are a key driver, suggesting that effective PPT design depends not only on expressivity but also on the accessibility of dependency resolution.