Three partial null‐subject languages: a comparison of Brazilian Portuguese, Finnish and Marathi*

Abstract

Abstract. The paper discusses properties characteristic of partial null‐subject languages, that is languages which allow null subjects but under more restricted conditions than consistent null‐subject languages. Three such languages are compared: Brazilian Portuguese, Finnish, and Marathi. It is demonstrated that they have indefinite null subjects, in particular, a null counterpart of English genericone, but allow definite 3rd person null subjects only when controlled from a higher clause, while consistent null‐subject languages do not allow null ‘one’, but do allow definite 3rd person null subjects without a controlling antecedent. A theory is proposed to explain this difference between the two types, with elements fromHolmberg (2005),Frascarelli (2007), andRoberts (2007). The structural difference is that consistent null‐subject languages have an unvalued D‐feature in T which is absent from partial null‐subject languages. The relation between a null subject embedded in a finite clause and its controller is discussed in some detail. There is some degree of variation between the three languages, yet the relation in all three of them is different from obligatory control and from non‐obligatory control, as familiar from non‐finite clauses, and also different from the antecedence relation found in consistent null‐subject languages.