The current problem of insufficient speaking environment for learning foreign languages in China has led to many difficulties in the process of language application. This paper proposes a language interaction system based on speech recognition technology, integrates it with AR English classroom teaching, and designs a new teaching mode. In order to verify the effect of AR technology on the enhancement of language interaction, the study is based on the collection of speech data from the experimental classroom of the preschool education program of a teacher training university, and after collecting and organizing the data, the effect of different data sets on the recognition accuracy of the language interaction system in this paper is tested by changing the size of the data set. After the data set was enlarged, the language interaction system’s accuracy increased by 4.3%. Then, the comparative analysis of teaching language between AR English language interactive classroom and traditional English classroom is carried out from three aspects of language structure, teacher tendency and question and speech, respectively, and it is found that there is a big gap between the two, such as question and speech characteristics, in the AR English language interactive classroom. The rate of active speech among students is as high as 17.8%, while the rate of active speech in traditional English classrooms is 9.5%. The experiment verified that teaching in an AR English language classroom has a significant improvement in students’ language learning interactions.