Date of Completion


Embargo Period



impersonal structures, Brazilian Portuguese, language acquisition, null Subjects

Major Advisor

Diane Lillo-Martin

Associate Advisor

Zeljko Boskovic

Associate Advisor

William Snyder

Field of Study



Doctor of Philosophy

Open Access

Open Access


This dissertation focuses on the acquisition of null impersonal structures in Brazilian Portuguese (BP), a partial null-subject language (PNSL). PNSLs allow definite null subjects under more restricted conditions than consistent null-subject languages (CNSLs). However, PNSLs are characterized by having impersonal constructions with generic null subjects which are absent in CNSLs.

Using the Truth Value Judgment Task, I investigated whether BP-speaking children know that they are acquiring a PNSL, in which null subjects should be understood as generic. The results show that children as young as 4-years-old correctly reject the definite reading of null subjects in impersonal structures and correctly accept their generic reading. The data suggest that BP-speaking children have early knowledge that they are acquiring a PNSL.

Using the Felicity Judgement Task, I tested children’s knowledge of restrictions imposed on the well-formedness of impersonal structures in BP. Particularly, impersonal structures in BP are completely well-formed only when they have an overt marker of genericity (e.g., a deontic modal or an impersonal clitic se). 7-year-olds and 6-year-olds exhibited sensitivity to the adult grammar. 5- and 4-year-olds exhibited worse performance than the other groups, over-accepting the null subject in structures without an overt marker of genericity. It seems that at the age of 5-years-old children did not completely acquire the impersonal pronoun se. This can be explained by the fact that impersonal se is not frequent in the input.

Building upon Holmberg’s (2010a) Null Subject Parameters, I propose a learning model to address the problem of learnability that PNSLs and other (non-)null-subject languages impose. There is no evidence of missetting of the Null Subject Parameters by children acquiring any (non-)null-subject language, supporting the hypothesis that parameters are set in the earliest observable stages (Wexler 1995).