Argument prominence and the nature of superiority violations
This paper studies the mapping of argument structure into higher parts of the clause and examines the relation of argument structure to <i>multiple wh-structures</i>and Superiority phenomena. Superiority effects are commonly assumed to arise when <i>wh </i>-movement triggered by the feature-checking requirements of a [+Q] C 0 violates economy restrictions on movement (Shortest Move). The paper, however, points out certain serious diffi culties for a purely structural approach to Superiority patterns, and suggests an alternative analysis of the data. First, it is shown that the non-occurrence of Superiority effects in cases of multiple <i>wh</i>fronting in Bangla seems to contradict the fact that pair-list answers to multiple<i>wh </i>-questions are expected/required to the same degree as they are in languages with clear <i>wh </i>-movement such as English. The obvious question that is raised is how one should reconcile the lack of Superiority effects in a language with the assumption that genuine <i>wh </i>-movement nevertheless occurs in the language? The paper shows, re-examining the generalizations about English, that Superiority effects are not the result of a purely structural fi lter such as Shortest Move but are rather controlled by a variety of factors: animacy distinctions among <i>wh </i>-phrases, thematic relations of the <i>wh </i>-phrases, stressing and prosodic weight of the <i>wh </i>-phrases and referential familiarity of the expected answer to a <i>wh </i>-question. Finally, the “Superiority” effects observable in multiple <i>wh </i>-sluices in Bangla are attributed to the tendency to copy the argument prominence relation in the non-sluiced clause.
