Other Areas of Investigation: Negation and Late Argument Drop

Abstract

From the investigation of infinitives and null subjects in Wh-questions and from the results presented in Chapter 8, it emerges that the truncation approach as well as the ATOM approach must assume two kinds of null subjects in order to explain the fullness of the data. (2000) considers the child null subject occurring in the specifier of the root as the primary source of the null subject phenomenon in child language but admits the occurrence of PRO null subjects in infinitival structures because these are always licensed by UG. The first assumption explains the results of the previous chapter and the occurrence of null subjects in finite and non-finite environments, the latter explains the higher rate of null subjects in infinitival structures. The ATOM approach has always suggested that in addition to the PRO null subject licensed by [-TNS] infinitives, null subjects occurring in finite structures may be topic-drop (see Bromberg and Wexler 1995). This conjecture was motivated precisely by the observation that null subjects do practically not occur in finite Wh-questions in English. It was the correlation of the developmental curve of infinitives and null subjects in finite structures as shown for Danish by (1997), 1998 that made a topic-drop approach to finite null subjects unlikely. However, only a topicdrop approach can take care of the data distribution in Wh-questions in languages like Danish, German and French. So in the ATOM approach, it is assumed that topic-drop still plays a role — even if (in this approach) the bulk of finite null subjects in Danish is explained by their ‘disguised’ non-finite nature.¹ However, if we admit topic-drop as a possibility for an initial finite null subject (especially in topic-drop languages like German), there remains the rarity of infinitives in Wh-questions which needs another, unrelated explanation.