Text this: Understanding Social Biases in Large Language Models