Multilingual BERT (mBERT), a language model pre-trained on large multilingual corpora, has impressive zero-shot cross-lingual transfer capabilities and performs surprisingly well on zero-shot POS tagging and Named Entity Recognition (NER), as well as on cross-lingual model transfer. At present, the mainstream methods to solve the cross-lingual downstream tasks are always using the last transformer layer's output of mBERT as the representation of linguistic information. In this work, we explore the complementary property of lower layers to the last transformer layer of mBERT. A feature aggregatResearch goal: How does multimodal alignment (e.g., using CLIP or BLIP-2) affect the zero-shot cross-lingual transfer performance of mBERT on XTREME-R when compared to text-only feature aggregation methods?Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 9.2/10.