Agenda
SIL talk by Caterina Donati (LLF, CNRS)
Is there parsing evidence for Merge over Move? Two experiments on French and Hebrew
(Joint work with Francesca Foppolo, Carlo Cecchetto, Lion Oks)
Much work stemming from the classical garden-path model of parsing has focused either on how a structural heuristic in isolation (e.g., Minimal Chain Principle or Minimal Attachment) can guide the parser among different continuations, or on the comparison between structural heuristics as opposed to frequency-based explanations. Less attention has been devoted to cases where two independently motivated heuristics (such as: avoid gap vs avoid nodes) make opposite predictions about the continuation of a sentence.
Testing this experimentally becomes directly relevant for syntactic theory given the centrality that notions such as economy and optimality have assumed within the Minimalist program.
In this paper, we focus on one such case, namely on the processing of a temporary ambiguity triggered by verbs like ‘convince’ in French in sentences like (1), which are temporarily ambiguous between a transitive parse in which the DP l’enfant is modified by a relative clause, thus involving a gap (introduced by qui), and a ditransitive parse which is introduced by qu’il and is gapless.
(1) Marie a convaincu l’enfant qui/qu’il devait partir
In this configuration, the transitive parse should be preferred by Avoid Nodes, while a gapless grammatical continuation should be preferred by the Avoid Gaps.
We ran two very similar experiments exploiting this king of temporal ambiguity, one in French and one in Hebrew, where crucially the relative clauses investigated contained not a gap but a resumptive pronoun. I shall discuss the results and their potential consequences for syntactic theory.