Rise of canonical subjecthood
In this paper I have examined several instances in which an oblique constituent acquires canonical subject marking, i.e. nominative case and verbal agreement (in an accusative language). These instances show that an oblique constituent may acquire subject coding properties without being beforehand endowed with subject behavioural properties, if two requirements are met: (i) this oblique constituent must exhibit a considerable functional overlap with the prototypical subject in the given language and, (ii) there must be either no canonically case-marked subject in the construction at all, or the coding subject properties must be assigned to a constituent that has less functional-semantic overlap with the prototypical subject than the oblique constituent.<i> </i>Furthermore, I claim that there is often some minor semantic change concomitant with the acquisition of subject coding properties. I have also introduced the <i>control over the pre-stage </i>property (CoP) which is a weaker entailment than Dowty’s (1991) <i>volitional involvement in event or state. </i>It only denotes whether or not the experiencer had the choice to resist the experience to come about. Differently from Dowty’s (1991) approach, which presupposes that the proto-role entailments are lexical and provided primarily by the predicate, it is assumed that some of the proto-role entailments may also stem from the case frame. This becomes especially obvious with the labile predicates that allow for more than one case frame, each resulting in different sets of the proto-role entailments.