|
This thesis deals with the cross-linguistic study of the null subject phenomenon and its interaction with the contextual topics. Three types of behavior characterize the distribution of the null subject across languages.
In the so-called pro-drop languages, e.g. Polish, Spanish or Italian, null subjects are licensed by the rich personal inflection of the finite verb. In these languages, a null subject is grammatical when its antecedent is a salient contextual topic. The same is true for null subjects in Chinese, where verbal inflection is absent. While in Spanish or Italian the null subject is relatively free, in Chinese its distribution is constrained by discursive factors. Nevertheless, in these languages, null subjects are a default option. This shows that the pro-drop phenomenon probably has less to do with inflection than is traditionally assumed in the linguistic literature.
In partial pro-drop languages, such as Russian or Hebrew, null subjects depend on the near context. A pronominal subject becomes invisible only when it refers to a strong salient topic (notion introduced in Wexler 1998). In Hebrew, but not in Russian, null subjects are licensed by the personal inflection of the verb. The person (third vs. non-third) also influences the distribution of null subjects. Partial pro-drop null subjects resemble the Chinese ones, although the former are less constrained. The differences between these languages are quantitative rather than qualitative.
In English, null subjects are likewise permitted when the salient topic is strong, but topic strength is determined by the speech register. In colloquial English, informal notes or diaries, as well as in the contexts where a global contextual topic is defined (e.g. in biographies), the salient topic is strong enough to bind a null subject. Thus, if cross-linguistic distribution of null subjects is described in the terms of topic strength, English has the weakest topic among those examined.
A separate chapter is reserved for the study of French, which, depending on the speech register, can be regarded either as a pro-drop language (Zribi-Hertz 1994) or as a language not allowing null subjects of any kind.
The analyzed phenomena include the grammaticality of embedded null subjects, the intervention effects caused by the matrix subject, as well as the influence of modal elements of the functional level (IP). The gradual change of behavior from one language to the next shows that the null subject parameter cannot be described as a binary value, and that a simple variation between the established types of empty categories is insufficient for an adequate description of the cross-linguistic distribution of null subjects.
|
E-mail this message to a friend
