This study explores the sound discrimination responses of Turkish speakers of English through a diagnostic listening test administered in an English as a foreign language (EFL) setting, examining their perceptions of minimal pair (MP) similarities or distinctions. The findings reveal that participants accurately marked a large majority of MPs as the same or different. However, error frequencies varied across items, with some placed in medium to very high error ranges. Errors involving vowels and consonants in these ranges were observed to be evenly distributed, with neither category outweighing the other. It was observed that salient differences among segmentals played a more prominent role in successfully distinguishing MPs in the low error range, while subtle differences, often harder to catch, contributed to higher error rates. Multiple contributing factors including participants’ first language (L1) seemed to interact in complex ways and influenced participants’ performances of sound discrimination. The study identifies several high functional load (FL) MPs in the medium to very high error ranges, highlighting their crucial role in word differentiation and meaning construction. The results also suggest the need for tailored teaching addressing specific challenges and nuanced distinctions, as well as prioritizing high FL pairs in instruction to enhance learners' communicative competence and intelligibility.