![Sample our Language & Literature journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5941/TOC-Subject-Sample-2021-Language-Literature.jpg"})
ABSTRACT
We investigate whether grammatical information is accessed in processing noun phrase ellipsis (NPE) and other anaphoric constructions. The first experiment used an agreement attraction paradigm to reveal that ungrammatical plural verbs following NPE with an antecedent containing a plural modifier (e.g. Derek’s key to the boxes … and Mary’s_ probably *are safe in the drawer) show similar facilitation to non-elided NPs. The second experiment used the same paradigm to examine a coordination construction without anaphoric elements, and the third examined anaphoric one. Agreement attraction was not observed in either experiment, suggesting that processing NPE is different from processing non-anaphoric coordination constructions or anaphoric one. Taken together, the results indicate that the parser is sensitive to grammatical distinctions at the ellipsis site where it prioritises and retrieves the head at the initial stage of processing and retrieves the local noun within the modifier phrase only when it is necessary in parsing NPE.
Acknowledgements
We are grateful to the anonymous reviewers for their invaluable suggestions and comments. We would like to also thank Brian Dillon, Matt Goldrick, Alex Krauska, Dave Kush, Ellen Lau, Akira Omaki, Colin Phillips, Shayne Sloggett, Patrick Sturt, Matt Wagers, Alexis Wellwood, members at Syntax, Semantics and Sentence Processing Lab at Northwestern, and the audiences at the CUNY2016, AMLaP2016, CUNY2017 and the 90th annual meeting of the Linguistics Society of America for their invaluable discussions and comments. This work has been supported in part by NSF DDRI Grant: BCS-1749580.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 In our study, the condition is called “ungrammatical”, but we do not mean that the mismatch between the antecedent site and the ellipsis site in terms of syntactic structure is ungrammatical. In the literature, it has been observed in many places that such mismatch is possible (Arregui et al., Citation2006; Frazier, Citation2008; Kim, Kobele, Runner, & Hale, Citation2011). Rather, by “ungrammatical”, we intend that the number mismatch between the “retrieved” antecedent and the verb is ungrammatical. For example, when the antecedent which has the singular noun does not match in number with the subsequent verb, this situation is very similar to the ungrammatical conditions in non-ellipsis baseline conditions, where the head noun does not match in number with the subsequent verb. Because we are calling such conditions in the non-ellipsis baseline conditions, ungrammatical conditions, we are calling the comparable conditions in the ellipsis conditions, “ungrammatical” conditions.
2 As an anonymous reviewer pointed out, an assumption of linear mixed-effects model is that the residuals should be normally distributed. Residuals were distributed symmetrically around zero, suggesting normality (Min = −3.14; Median = 0.02; Max = 3.02). Following a reviewer’s suggestion, we also carried out a cumulative logit model (also known as proportional odds model) for Experiment 1a. This revealed similar results to the linear models reported below, with a significant main effect of NPE (β = 0.60, SE = 0.09, z = 6.34, p < 0.001), Grammaticality (β = −1.00 = 0.10, t = −10.51, p < 0.001), a significant interaction between Local noun number and Grammaticality (β = 0.60, SE = 0.19, z = 3.23, p < 0.01) and a significant interaction between NPE and Grammaticality (β = 0.65, SE = 0.19, z = 3.49, p < 0.001).
3 There is an alternative account with regards to whether the whole structure is retrieved at the NPE site. In cases where the head is initially retrieved, it is possible that the parser accesses the head and calculates agreement at the verb. If the number mismatch between the head and the verb arises, the cue-based retrieval mechanism is employed. Even in this scenario, the parser is sensitive to the structural information such as the head and the modifier. Thus, the parser privileges the head noun over the local noun in the modifier, using structural information. In other words, the parser distinguishes the head and the modifier when it accesses an antecedent. At this point, it is hard to tease apart whether the whole structure or the head noun is retrieved at the initial stage of the retrieval processes.
4 Given that coordinate structures were used in these experiments, an expectation of parallelism could have led to easier retrieval of the head, weakening the potential interference from the modifier.
5 Again, residuals followed a symmetrical distribution around zero, suggesting normality (Min = −3.15; Median = −0.02; Max = 3.99). As in Experiment 1a, following a reviewer’s suggestion, we also carried out a cumulative logit model (also known as proportional odds model) of Experiment 1b. In this analysis, we found significant main effects of Local noun number (β = 0.24, SE = 0.08, z = 2.87, p < 0.01), NPE (β = −0.25, SE = 0.08, t = −3.04, p < 0.01) and Grammaticality (β = −2.53, SE = 0.10, z = −25.80, p < 0.001), a significant interaction between Local noun number and Grammaticality (β = 0.52, SE = 0.17, z = 3.12, p < 0.01) and no other interactions.
6 Residuals followed a symmetrical distribution around zero, suggesting normality (Min = −4.53; Median = 0.06; Max = 3.12). As in Experiments 1a and 2a, we also carried out a cumulative logit model (also known as proportional odds model) of Experiment 3a. This disclosed main effects of Anaphoric one (β = −0.26, SE = 0.08, t = −3.16, p < 0.01) and Grammaticality (β = −1.99, SE = 0.09, z = −21.36, p < 0.001) and an interaction between Local noun number and Grammaticality (β = 0.59, SE = 0.17, z = 3.48, p < 0.001). There was also a marginal interaction between Local noun number, Grammaticality and Anaphoric one (β = −0.65, SE = 0.34, z = −1.92, p = 0.06).
7 This was suggested by an anonymous reviewer. We are grateful to the reviewer for pointing out this possibility.
8 Note that, “ungrammatical” conditions are called “ungrammatical” conditions because the number marking of Anaphoric one and the verb are not matched. They are ungrammatical not because the number marking of one and the antecedent mismatches, but because number marking of one and the verb mismatches.