The adoption of electronics learning (e-learning) as a method of disseminating knowledge in the global educational system is growing at a rapid rate, and has created a shift in the knowledge acquisition methods from the conventional classrooms and tutors to the distributed e-learning technique that enables access to various learning resources much more conveniently and flexibly. However, notwithstanding the adaptive advantages of learner-centric contents of e-learning programmes, the distributed e-learning environment has unconsciously adopted few international languages as the languages of communication among the participants despite the various accents (mother language influence) among these participants. Adjusting to and accommodating these various accents has brought about the introduction of accents-based automatic speech recognition into the e-learning to resolve the effects of the accent differences. This paper reviews over 50 research papers to determine the development so far made in the design and implementation of accents-based automatic recognition models for the purpose of e-learning between year 2001 and 2021. The analysis of the review shows that 50% of the models reviewed adopted English language, 46.50% adopted the major Chinese and Indian languages and 3.50% adopted Swedish language as the mode of communication. It is therefore discovered that majority of the ASR models are centred on the European, American and Asian accents, while unconsciously excluding the various accents peculiarities associated with the less technologically resourced continents.