How Culturally Aware are Vision-Language Models?