This paper attempts to quantify the emotional tendency of 18 emotionally deviated words by using the point-wise mutual information computational method based on information retrieval( PMI-IR) in Computational Linguistics. In addition,word2 vec is used for adjustment of the output of PMI-IR. The results show that emotional deviation has the characteristic of strong timeliness,and it can be easily affected. The main reasons for generation and spread of the emotionally deviated words include hot even...