Constructional semantics as a limit to grammatical alternation: The two genitives of English

Abstract

It has often been claimed that the distribution of the s-genitive and the of-genitive is determined by considerations of information structure, more specifically by linear precedence preferences related to animacy, givenness, or syntactic weight. This paper shows that such claims are untenable on empirical as well as theoretical grounds. First, corpus analyses simply do not bear out the predictions made by these claims. Second, such claims assume that the two genitives are semantically equivalent. I show that this assumption is wrong and offer a systematic account of the s-genitive and the of-genitive as distinct semantic-role constructions, arguing that the former encodes a possessor-possessee relation and the latter a part-whole relation unless the head noun itself inherently specifies a different relation. Only in the case of such an inherently specified relation does the possibility arise that information structure may play a role. I then show that in such cases animacy and (to a lesser degree) length have an influence, but that givenness has an optional stylistic influence at best.

The genitive alternation in Chinese and German ESL learners

“…For example, Jason's research interests is more acceptable than research interests of Jason because research interests is not specific and, therefore, less specific than Jason; however, when a definite article is added to research interests, the of-genitive variant the research interests of Jason is just as acceptable as the s-genitive variant. Rosenbach (2002) and Stefanowitsch (2003) demonstrate that the choice of construction is also impacted not just by morphosyntactic features that have broad semantic correlates, but furthermore constrained by the different meanings and functions that are associated with the two genitive variants. Stefanowitsch (2003) presents a detailed account of the different 'semantic relations' that are encodable by the two genitives; Table 1 provides an overview.…”

“…Rosenbach (2002) and Stefanowitsch (2003) demonstrate that the choice of construction is also impacted not just by morphosyntactic features that have broad semantic correlates, but furthermore constrained by the different meanings and functions that are associated with the two genitive variants. Stefanowitsch (2003) presents a detailed account of the different 'semantic relations' that are encodable by the two genitives; Table 1 provides an overview. Adopting the theoretical perspective of construction grammar, Stefanowitsch (2003) argues that the two variants are in fact semantic-role constructions that only partially overlap in terms of their semantic potential: "the s-genitive assigns the roles POSSESSEE and POSSESSOR to its head and modifier respectively, and the of-genitive assigns roles that I have called, for want of a better term, ENTITY and INTRINSIC ENTITY" (Stefanowitsch 2003: 20).…”

“…Stefanowitsch (2003) presents a detailed account of the different 'semantic relations' that are encodable by the two genitives; Table 1 provides an overview. Adopting the theoretical perspective of construction grammar, Stefanowitsch (2003) argues that the two variants are in fact semantic-role constructions that only partially overlap in terms of their semantic potential: "the s-genitive assigns the roles POSSESSEE and POSSESSOR to its head and modifier respectively, and the of-genitive assigns roles that I have called, for want of a better term, ENTITY and INTRINSIC ENTITY" (Stefanowitsch 2003: 20).…”

Extending collostructional analysis

“…As previous studies have shown that givenness is not a factor anyway (Altenberg 1980;cf. Gries 2002;Stefanowitsch 1998Stefanowitsch , 2003, we will focus on animacy here. The predictions of linear-precedence accounts of s-genitive and of-construction concerning the two constructions' distinctive collexemes are straightfor- ward: the s-genitive should attract animate modifiers and inanimate heads, and the associations should be reversed for the of-construction.…”

Go-and-V, come-and-V, go-V and come-V

“…Wulff (2006: 120-121), citing Shopen (1971), observes that, "the linkage between V 1 and V 2 is stronger in the constructions lacking a conjunction in comparison to those including and. Moreover the auxiliary-like function of go as noted by Stefanowitsch (2001) appears to be even stronger for go-V than for go-and-V, which is also reflected in its closer position to V 2 . " It follows from these claims that V 2 in the go/come-V construction will be more subjectively construed than V 2 in the go/come-and-V construction.…”