The burgeoning literature on fake news reveals that the emotional context of the message is a major factor that drives its diffusion on social media. However, studies have largely missed a major aspect of the diffusion process, which is the morphing of the textual contents themselves during this process. Our study first visually illustrates through hazard functions that, while falsehoods morphs aggressively at the initial stages, correction messages morph more aggressively in the long run. In addition, we leverage on cosine distance and econometric modeling to empirically investigate how sentiment affects the morphing of fake news and their correction messages. We find that positive sentiments, emotionally charged messages and correction messages positively affect the morphing of messages during the diffusion process. Our results also show that, as time goes by, the impacts of sentiments on morphing change.