One Pair Suffices: Unlocking Universal Zero-Shot Translation via Cross-Architecture Alignment

Current paradigms for empowering Large Language Models (LLMs) with multilingual capabilities rely heavily on massive instruction tuning. We challenge this view, proposing that the barrier is topological alignment, not data quantity. We introduce Hybrid Cross-Alignment (HCA) , fusing a frozen NLLB encoder with a Qwen decoder via a closed-loop dual-adapter architecture. HCA utilizes a Source-Side Adapter to precondition encoder features and a Query-Residual Adapter to preserve generative stability, bridged by an adaptive gated cross-modal interface. Our core finding is the phenomenon of “Source-Side Alignment Generalization.” We demonstrate that training HCA on a single language pair (German-English) unlocks state-of-the-art zero-shot transfer to dozens of unseen languages for X-to-English translation . Crucially, our “Oracle” experiments reveal that this single-pair training recovers over 96.7% of the performance achievable by training on all available pairs. This suggests that a highly generalizable, source-side projection protocol exists. Evaluated rigorously across COMET and chrF++ , our ∼ 5.25B-parameter model significantly outperforms larger baselines, surpassing TowerPlus-9B (+9.0 COMET on low-resource languages) and Aya-101 (13B). Furthermore, performance scales linearly with encoder size; upgrading from 600M to 1.3B yields immediate gains (+3.4 points on Gujarati) with minimal retraining cost.