On the Amplification of Linguistic Bias through Unintentional Self-reinforcement Learning by Generative Language Models -- A Perspective