Gender asymmetries in ellipsis: an experimental comparison of markedness and frequency accounts in English
Jon Sprouse, Troy Messick, Jonathan Bobaljik
November 2020

Bobaljik & Zocca (2011) argue that ellipsis reveals the existence of (at least) two classes of gender-paired nouns, distinguished by the relationship between morphological marking and semantic specification: in the actor/actress class, the grammatically feminine form is specified for conceptual gender, while the unaffixed form is unspecified for conceptual gender, exemplifying the classic markedness asymmetry (Jakobson 1932); in the prince/princess class, despite the same morphological relation between the words, both the affixed and unaffixed form are semantically specified for conceptual gender (female and male, respectively). Bobaljik & Zocca and others (see Merchant 2014, Sudo & Spathas 2016, and Saab 2019) pursue accounts that code the difference between the classes in the linguistic representation of such nouns, incorporating (un)markedness into the explanation. By contrast, Haspelmath 2006 has suggested that differences in the relative frequency of the two forms in each pair may be the cause for the different behavior of the two classes. The frequency approach predicts that the size of the acceptability asymmetries that arise under the ellipsis test will correlate with the size of relative frequency ratio for each noun pair. In this article, we develop a formal experimental version of the Bobaljik & Zocca (2011) ellipsis test for gender asymmetries, and deploy it in two experiments: the first is a curated set of 16 noun pairs in English (by hypothesis 8 of each class), and the second is a test of 58 noun pairs forming a nearly exhaustive set of such pairs in English. We then use the freely available COCA corpus (Davies 2008) to test the prediction of the frequency approach. Our results suggest that the Haspelmath 2006 relative frequency hypothesis is not an empirically adequate competitor for the explanation of these asymmetries, as there is no evidence of a correlation between the size of the asymmetry effects in the acceptability judgment experiment and the size of the ratio of the relative frequency of the two forms in each pair for either the curated set of 16 noun pairs or the nearly exhaustive set of 58.
Format: [ pdf ]
Reference: lingbuzz/005545
(please use that when you cite this article)
Published in: Submitted
keywords: gender asymmetries, ellipsis, experimental syntax, acceptability judgments, word frequency, semantics, morphology, syntax
Downloaded:136 times


[ edit this article | back to article list ]