What does Kiki look like? Cross-modal associations between speech sounds and visual shapes in vision-and-language models