Geraldine Legendre - Blocking and anti-blocking effects at the lexicon/semantics interface: The case of French anticausatives

French has three subclasses of anticausatives alternating with transitives (e.g.transitive/intransitive break), which are defined by morphology. These include Class I (obligatory reflexive morphology with se, e.g. se briser ‘shatter’), Class II (no se, e.g. craquer ‘crack’), and Class III (optional se, e.g. (se) casser ‘break’). 
In impersonal constructions as well as aspectual contexts traditionally used as diagnostic tests for telicity interpretive (anti)-blocking effects emerge, which point to an analysis relying on competition among expression/interpretation pairs. 
To take one example, it is not the case (contra Labelle 1992) that only reflexive-marked anticausatives have a completion interpretation in sentential contexts like Le vase a mis moins de trois secondes a se briser ‘The vase took less than three seconds to shatter’ (Class I). If the lexical entry does not include a reflexive-marked option (Class II) the relevant completion reading becomes available for the non-reflexive form. If the lexical entry offers two options, reflexivemarked or not (Class III), only the reflexive-marked form has the relevant completion interpretation. 
Such effects generalize to other aspectual contexts. Elsewhere, in nonaspectual sentences with the ambiguous (referential/non-referential) subject pronoun il ‘he, there’ (Il a cassé plusieurs branches ‘He/*There broke several branches’) interpretive blocking effects are observed whereby a transitive/causative, rather than an impersonal/presentational interpretation, is the only one available --if possible for a given morphological class. 
A formal analysis of both blocking and anti-blocking effects is provided in terms of bidirectional optimization.