以往研究认为“不很”多与褒义词连用，从而形成带有贬义色彩的搭配，通过统计发现有相反情况的存在，可以从语义韵角度做出具体的解释。语料库语言学研究语义韵有三种基本方法，分别是基于数据的方法、数据驱动的方法和两者相结合的折中方法，三种方法各有优势。利用基于数据的方法考察“不很”右侧被修饰成分，利用基于数据驱动的方法考察“不很”左侧附加性成分，得出“不很”的语义韵结构。

Previous studies believe that "not very" is often used together with positive words, thus forming a collocation with derogatory color. There is the opposite situation through statistics, and specific explanations can be made from the perspective of semantic prosody. There are three basic methods for the study of semantic prosody, namely data-based method, data-driven method and compromise of the two. The three methods have their own advantages. The data-based method is used to investigate the modified component of the "not very" right side, and the "not very" left additive component is used by the data-driven method to investigate the "不很" semantic prosody structure.