Predicting Pro-Drop in Russian and Polish
Marjorie McShane
In this talk I will describe my work to date on developing a predictive model of pro-drop in Russian and Polish, with predictive defined as sufficiently explicit to permit a non-native speaker (man or machine) to correctly use overt and null subject pronouns in all contexts. The approach mirrors that taken in A Theory of Ellipsis (Oxford University Press, 2005), which treats pro-drop only in passing for reasons I will explain.
The talk is composed of the following parts:
- salient typological features of Russian and Polish
- the landscape of empty categories in Russian and Polish
- other contributions to ellipsis in Slavic
- data analysis and classification of pro-drop in Russian and Polish
Despite significant typological similarities, Russian and Polish have quite different pro-drop profiles, with Polish showing a far stronger tendency to drop subject pronouns than Russian. A practical approach for modeling is this: for Russian, assume an overt subject pronoun then explain cases when it is dropped; for Polish, assume a dropped subject pronoun then explain cases when it is overt. By explanations, I mean something more fine-grained and practically useful than broad generalizations like‘pro-drop is used in all non-emphatic contexts’.
As concerns generative theory, I will make no claims but will offer certain observations. If Polish is a canonical pro-drop language, which it would appear to be based on traditional definitions, then all canonical pro-drop languages should be revisited and fully described, since the use of overt pronouns in Polish surpasses what would be expected by the literature on pro-drop. (I will argue, as an important aside, that the field in general needs to be more rigorous about the use and presentation of data when dealing with ellipsis: as an only partially syntactic phenomenon, ellipsis cannot be handled by the one-example-with-no-preceding-context approach, even if one’s only interest is in syntax.)As for Russian, its elided subjects might well be licensed not by a special pro-drop parameter but by the same discourse-oriented strategy that licenses the ellipsis of other categories in this highly elliptical language.
Among the goals of the talk is to demonstrate the need to put pro-drop in the context of other elliptical phenomena in a language before deciding whether there is need for a special parameter devoted to the ellipsis of namely subjects.