Weak Crossover and the Absence of Traces

We provide a new definition of the linear prominence constraints between pronouns and operators (wh-words and quantifiers) which correctly rules out examples that violate weak crossover. Previous analyses of weak crossover relied on the presence of a trace in the extraction site of a wh-question; in contrast, our analysis enables a traceless account of examples previously cited in support of traces. Since no other incontrovertible evidence for traces has been put forward, our account allows a return to the traceless and therefore more constrained account of long-distance binding in LFG proposed by Kaplan and Zaenen (1989). 1 Long-distance dependencies in LFG Traces were introduced in transformational grammar (Chomsky, 1973) in order to give a phrasal account for a wide range of long-distance dependency phenomena: a long-distance dependency between a displaced element in a wh-question or relative clause was assumed to arise by movement of the displaced element, which leaves behind a ‘trace’ in its original position. The original LFG treatment of long distance dependencies (Kaplan and Bresnan, 1982) was based on an adaptation of the transformational/phrase-structure scheme. The relation between the displaced wh-phrase and its within-clause function was defined in terms of a relation between the wh-word and an empty c-structure constituent, a trace, within the clause. In example (1), the relation between the NP node dominating who and the NP node dominating its trace, represented by , ensures that the wh-phrase who is both the FOCUS and the OBJ of the sentence: