Abstract: As text types are among important factors for translation quality assessment, this research takes an appellative text and two expressive texts with complexity as source text, high level human translation and ChatGPT translation as contrastive pairs to test ChatGPT C-E translation performance in a paragraph. The findings include that ChatGPT translation focuses the equivalence of words’ literal meaning, ignoring their deep meaning; copies the word sequence and sentence structure of the original language, failing to make the implied logic explicit; doesn't give full consideration to context; lacks personality and creativity; rarely adopts amplification and omission based on context. It is concluded that ChatGPT demonstrates limitations in translating certain texts while high level human translation has superiority.