This dissertation addresses Korean Americans as speakers of English and as a unified speech community, exploring the nature and extent of sociolinguistic stratification of the English used by Korean Americans in Philadelphia. The acquisition of three linguistic features is investigated: word-medial /t/ flapping, the use of discourse markers, and the regional feature of Philadelphia short a. Statistical analyses examine these features for the effects of linguistic factors and social factors such as age, sex, occupation, age of arrival in the US, length of stay in the US, and English education. Age of arrival shows a very strong effect on flapping: immigrants who arrived in the US as children and US-born immigrants both showed a very high degree of flapping, while Korean-born adult immigrants acquired flapping to a much lesser degree. Style is also analyzed to determine whether speakers show variation along the formality continuum. In addition to production, the perceptual component of English use by the speakers is examined through a perception test. The perception test, administered to native English speakers, elicits judgments of English nativeness and ethnic identity of the Korean Americans. The results of the perception test are correlated with the production results of the linguistic features. In general, Korean Americans show varying degrees of acquisition of the three features according to sociolinguistic factors. Although the speakers exhibit stylistic variation, they have not acquired the Philadelphia dialectal feature of short a. The perception test reveals that English nativeness is accurately judged but that ethnic identification is problematic for listeners. The correlation of perception and production is positive in that an increase in the presence of the native linguistic features in the speech being judged is correlated with increased perception of the degree of English nativeness. The three features examined are not taught through formal explicit instruction to either native or non-native English speakers, which implies that speakers must engage in face-to-face interaction with native speakers in order to acquire these native speech community norms.