Vowel raising in Korean has been primarily treated as a phonological, categorical change. This study aims to show how the Korean connective suffix {-go} is realized in various environments, and propose a principle of vowel raising based on both acoustic and perceptual data. To that end, we used a corpus of spoken Korean to analyze the types of syntactic constructions, the realization of prosodic boundaries (IP and PP), and the types of boundary tone associated with {-go}. It was found that the vowel tends to be raised most frequently in utterance-final position, while in utterance-medial position the vowel was raised more when the syntactic and prosodic distance between {-go} and the following constituent was smaller. The results for boundary tone also showed a correlation between vowel raising and the discourse function of the boundary tone. In conclusion, we propose that vowel raising is not simply an optional phenomenon, but rather a type of phonetic reduction related to the comprehension of the following constituent.