Null subjects: comments on Valian (1990)

In this paper, I do not claim that any particular parameter-setting approach is correct, or even provide a characterization of subjectless sentences in children's speech. The only point of this paper is to show that Valian's argument that single-value solutions for setting the null subject parameter have insoluble problems is incorrect. Valian is correct that being able to analyze and interpret triggering data is a prerequisite for setting parameters, but a single-value solution of the sort described in this paper (and implicitly assumed in parameter-setting acquisition theories) is sufficient to do so; there is no need to invoke the dual-value solution that Valian argues is necessary. Furthermore, I argue that the single-value solution should be preferred on the grounds that (i) the mechanism I propose maintains many of the niceties of idealized parameter-setting acquisition theories whereas Valian's approach explicitly gives up on these attractive features of standard parameter-setting models, and (ii) it follows from dual-value theories but not single-value theories (depending on the nature of parameters and how many parameters there are) that parsing and speech production involve over-whelmingly difficult computations for children.