An Impact Study on Large-Scale Aviation English Tests Based on Data Mining

Abstract

According to the results of the study of aviation incidents and accidents that have occurred over the years, the poor use of English in aviation, i.e., the poor listening and speaking skills of pilots and air traffic controllers is one of the main causes of aviation accidents. Therefore, the International Civil Aviation Organization (ICAO) has adopted strengthened language proficiency requirements for flight crew and air traffic controllers operating along international air routes. In order to evaluate the language proficiency of civil aviation pilots in a more effective and professional manner, Pilots' English Proficiency Examination of CAAC (PEPEC) was developed by CAAC in accordance with the international standards set by ICAO. As high-stakes tests, PEPEC affects not only pilots and ATC, but also the civil aviation industry, and even the society as a whole. The current study aims to investigate the impact of PEPEC on the English learning and working of pilots. In reference to the aviation tests impact model, a muti-phase and multi-method research design based on data mining has been developed to investigate the impacts of PEPEC on pilots to explore their perception and evaluation of these tests, as well as the impact of the test on their learning, working or designing practice. Various research methods have been employed, such as questionnaire surveys, interviews and document analysis. The results reveal that PEPEC has exerted a strong promoting impact on English learning and on the test takers’ English proficiency in working and their working competence. The findings of this research provides implications for the further improvement of the candidates’ English proficiency in that even if they have achieved Level 4 in the tests and their English proficiency in working is improved, many pilots still have inadequate ability in communication in international operations and the new testing system based on artificial intelligence and machine learning should be developed.