While listeners tend to downgrade speakers' accent and comprehensibility when they perceive them to be from a different language community--a process known as reverse linguistic stereotyping (RLS)--research has generally relied solely on quantitative data such as Likert scale ratings. The current study sought to extend the analysis further by investigating the reasons which informed raters' decisions. A sample of 210 Japanese university students (six groups of n = 35) were asked to listen to recorded speeches by native Japanese speakers. In a matched-guise design, the groups were either shown photos of a Japanese, Caucasian, or Southeast Asian male, and asked to provide ratings of accentedness, comprehensibility, and intelligibility. They were then asked to report what factors influenced their comprehensibility ratings. In line with previous research, listeners rated non-Japanese guises as significantly more accented than the Japanese one, though differences in intelligibility were non-significant. A key finding was that while comprehensibility ratings were statistically comparable, the rationales given were qualitatively different. Groups who viewed the Caucasian or Southeast Asian photographs cited pronunciation issues significantly more than the Japanese group which reported grammatical and lexical factors as the reasons behind their downgraded ratings.