Goodall Colloquium

April 24
Wieboldt 408
University of California, San Diego
Two ways to escape from an island

Island violations can be ameliorated in a number of ways. Two possible ways are explored here using formal sentence acceptability experiments, with the goal of understanding why and to what extent the amelioration occurs, and ultimately, what this can tell us about the nature of extraction. The first case involves the amelioration that comes from making the extracted element more specific, the phenomenon known as D-linking. Our experimental results show that this effect occurs in all kinds of extraction, not just in potential island violations, suggesting that the effect may result from a general processing advantage associated with D-linked fillers. The second case involves subextraction from subjects. It has been widely claimed that these subject island violations improve if the subject is first moved to a left-peripheral position and the subextraction takes place from there. Our experimental results suggest that once the full range of facts is taken into account, this type of movement actually makes subextraction worse, not better. Taken together, the D-linking and subextraction cases point to a simpler view of extraction than is sometimes assumed: the grammar perhaps does not attend to the specificity of the extracted element for the purposes of extraction itself, nor does it allow complex derivations involving subextraction from phrases that have themselves been extracted.