UC San Diego

The focus of this dissertation is twofold : First, it examines whether Korean exhibits island effects, as the status of islands in Korean and typologically related wh- in-situ languages has been unclear. Second, it explores whether Korean-English bilinguals display native-like island effects in their two languages. While most accounts of island effects claim that the input is not directly involved and that islands stem from basic properties of grammar/processing that are available to all humans, another line of accounts claims that islands actually can be learned from the input. These different approaches to the learnability issue on islands would then predict different outcomes for bilinguals. Under the accounts claiming that islands are not learned but available to all speakers regardless of one's learning environments, we might expect native-like island effects in bilinguals regardless of their learning environment. On the other hand, if input plays an important role in having island effects, as the second position argues, then bilinguals might be expected to show non-native-like and/or various types of island effects. Four acceptability experiments on island effects in Korean (involving a whether-island and an adjunct island) in native and Korean-English bilinguals (i.e. heritage speakers of Korean) revealed the presence of a whether-island but the absence of an adjunct island in Korean in both native and heritage speakers of Korean. Another five acceptability experiments on island effects in English (a whether-island, a wh-island, and three types of adjunct islands) showed that Korean-English bilinguals, grouped according to their AoA (i.e. Heritage: AoA 0-5, Early: AoA 6-10, and Late: AoA 11-14), displayed adjunct island effects but either had weaker whether/wh-island effects than native speakers or lacked these effects altogether. Their island effect sizes, measured by DD scores, indicated that islands get weaker with increasing AoA. There was also found a positive correlation between AoA and length effects in bilinguals, suggesting a difficulty with long-distance dependencies as AoA increases. These results are discussed in light of the cross-linguistic similarities and differences in islands, and the role of input in island effects