This research aims to analyze language patterns in hate speech found in comments on YouTube about Rohingya refugees in Thailand and Syrian refugees in Europe. Data were collected from 4,113 comments in Thai and 13,960 comments in English and appeared on a video-sharing website specifically from news clips about Rohingya refugees in Thailand during the year 2015- 2019 and news on Syrian refugees in Europe during 2013-2019. After applying critical discourse analysis theory on the data, three types of hate speech were found: 1) name calling; 2) verb phrases; and 3) modifiers. The hate speech addresses race, religion, gender, body shape, taste, potential, ability, and individual or group identity. There were five types of name-calling, namely regarding race, religion, threats, animals/evil, and being unwanted; three types of verbs, namely danger/threats, behaviors/actions, and eviction/expulsion; and two types of modifiers, namely degrading quality and degree intensity. The three sets of vocabulary equally form hate speech in Thai and in English. In both languages, refugees are portrayed as villains, devalued as unwanted, dangerous, and offensive. Modifiers are used to magnify the degree of intensity driven by the underlying emotional implications.