Grant by the Israeli Science Foundation: 2020-2023
Project title: Broadening the Grammatical Perspectives on Argument Ellipsis
Scientific abstract
Null arguments have been at the focus of intensive research in generative syntax for nearly four decades. Two analyses prevail most of the literature: The null pronoun (pro) analysis and the Ᾱ-variable (or topic-bound) analysis. During the last decade, an alternative analysis has been developed: Argument Ellipsis (AE). Unlike the previous two approaches, the AE approach takes an unpronounced argument to be fully syntactically articulated (at some grammatical level). Chiefly developed for East Asian languages, AE has recently been extended to other language families, e.g. Hebrew (Landau 2018). The proposed research takes off where that paper ended, exploring the full range of implications of AE in Hebrew, with extensions to Arabic and Russian. The following are the key topics of interest:
Scope. What kind of arguments can or cannot be elided? Almost all discussions of AE are limited to DPs. Do PPs and CPs undergo ellipsis? How can this possibility be distinguished from Null Complement Anaphora (NCA)? We will design experimental tests to answer these questions; our initial hypothesis is that AE is categorically unselective (hence, if DP ellipsis exists, so must PP and CP ellipsis).
Licensing. What licenses AE in one context but not in another, or in one language but not in another? The literature on VP-ellipsis, Sluicing and Stripping usually assumes that it is the presence of particular functional heads, with a particular featural make up, that triggers ellipsis (of the head's complement or some constituent in it). The empirical motivations behind this assumption, however, are mostly missing from AE environments, raising the question of whether AE licensing even lies within narrow syntax or not (note that the term "argument" in AE can be defined lexically, with no recourse to syntax).
Syntax-semantics interplay. Initial evidence suggests that certain arguments resist ellipsis solely in virtue of their meaning. Specifically, non-referential idiom chunks, and adverbial arguments (e.g., behave well). If correct, this conclusion poses an interesting challenge to the general research on ellipsis, which normally does not recognize such absolute denotational constraints (rather, if parallelism with the antecedent is respected, ellipsis should go through).
Derivation and representation. The mainstream view in studies of Indo-European languages derives ellipsis sites by PF-deletion. In contrast, the mainstream view in East-Asian languages derives AE by LF-reconstruction. We have initial evidence from Hebrew – involving extraction out of ellipsis sites – suggesting that AE too can be derived by PF deletion. This conclusion bears directly on both the licensing question and on the broader typological distribution of AE, questions that we will address at the ultimate stage of the proposed research.