This study investigated the effect of language complexity and group size on knowledge construction in two online debates. Knowledge construction was assessed using Gunawardena et al.âs Interaction Analysis Model (1997). Language complexity was determined by dividing the number of unique words by total words. It refers to the lexical variation. The results showed that knowledge construction and group size are significantly and negatively correlated. Also, the study revealed that knowledge construction and language complexity are significantly and positively correlated. Furthermore, the study demonstrated that language complexity is a significant predictor of knowledge construction in online debates. Some actions should be undertaken to increase language complexity in order to foster knowledge construction in online debates.