Learning analytics (LA) involves the measurement, collection, analysis, and reporting of data about learners and their contexts, aiming to understand and optimize both the learning process and the environments in which it occurs. Among many themes that the LA community considers, natural language processing (NLP) algorithms have been widely adopted to extract information from textual data generated in learning environments (e.g., student essays and short answers, online discussion and chat). NLP can shed light on the learning process and student outcomes in different contexts. Based on the importance of NLP for education, this paper conducted a systematic literature review of the application of NLP to understand how the LA community has been applying the methods from this field. Our methodology includes automatic and manual methods to extract information about authors, relevant papers, and specific data related to educational applications and algorithms used in the field. This review selected 156 papers that reveal essential aspects of the topic; e.g., (i) the majority of the works focused on the analysis of online discussions and essay assessment; (ii) in general, the authors did not apply the developed models in real settings; (iii) recent papers selected have begun to evaluate deep learning models (e.g., BERT) more frequently; and (iv) the datasets used in the experimentation are usually small and contain English text. The results of this study and its practical implications are further discussed.