Verb Phrase Ellipsis in Children's Answers to Questions

Abstract

The emergence of verb phrase ellipsis is investigated in two 2-year-old English-speaking children's speech by studying their answers to yes/no questions over the period of about a year. The investigation is framed using a generative linguistics model, the “PF-deletion model,” which assumes that there is syntactic structure in the ellipsis site that is simply not pronounced at the level of Phonological Form (PF). On this model, if a child's use of elliptical verb phrases is delayed, in principle he or she could pronounce the material that would otherwise be elided and produce full sentences instead. Two potential reasons for delay in verb phrase ellipsis are considered: (i) learning of the auxiliary system of English and (ii) the intonation requirements on the auxiliary verb. One child produced sentences with verb phrase ellipsis early in the course of acquisition, although there were gaps in her knowledge of the auxiliary system. The second child did not produce verb phrase ellipsis for several months and, instead, initially produced full sentence answers to yes/no questions a high proportion of the time. This child appeared to refrain from using VP ellipsis until the intonational requirements of its use had been mastered, a finding that is consistent with the PF-deletion model.

ACKNOWLEDGMENTS

The research reported in this paper was supported in part by NSF Grant #BCS 9905199 to Rosalind Thornton. The grant also provided doctoral support for Dr. Graciela Tesan. Many thanks go to Graciela, who collected the longitudinal data with me and was instrumental in analyzing and interpreting the data at the early stages of the project. A preliminary analysis of these data was presented at the Linguistic Society Annual Meeting in Boston in 2004. I also thank audiences at the Beijing Language and Culture University, the Second Yuelu Workshop on Language Acquisition, Hunan University, and at Macquarie University, Sydney, for their comments, and Stephen Crain, Jason Merchant, and Juan Uriagereka, for their support of this project over the years. I would also like to thank the three anonymous reviewers for their challenging and helpful reviews.

Notes

¹The searches of Nina (Suppes) and Eve (Brown) were carried out by searching for utterance final auxiliary verbs. This will have identified most instances of VP ellipsis but will have failed to capture utterances such as I did too or the more unlikely John ate beans and so did Sally. Potential examples of VP ellipsis with null subjects, verbatim repetitions of a parent, repetitions of the child's previous utterance, songs and rhymes, and imperatives were excluded from the counts. Nina's data from 2;1 to 2;6 are from files 1 to 31, and the data from 2;9 to 3;0 are from files 32 to 42. All of Eve's data from files 1 to 20 was searched.

²According to López and Winkler (2000), the functional head that licenses VP ellipsis is  (not Infl), the same head that hosts the auxiliary verb for verum focus, the emphatic auxiliary verb in affirmative sentences such as Mary DOES speak Mandarin (which might be uttered in a context where it is assumed that Mary does not speak Mandarin).

³The term SpecCP refers to the specifier position in the Complementizer Phrase (CP); this is the position to which wh-phrases move.

⁴Notice that the fragment answer Chili is not simply formed by eliding everything except chili from the string Daddy can cook chili, as in Daddy can cook chili. This derivation of the fragment answer would violate the licensing requirement that only complements to functional heads can be elided (Lobeck, 1995). Merchant, therefore, proposes that the fragment chili moves to FP (focus phrase), as shown in (8b). The projection FP, as part of the extended CP domain, is a functional projection, so the complement to the focus head, the entire sentence Daddy can cook can be targeted for deletion, leaving just the fragment chili for pronunciation.

⁵Thanks to an anonymous reviewer for suggesting the inclusion of this argument.

⁶The children's names and the names of their family members are fictitious.

⁷Nonadult answers were of the wrong syntactic category. For example, the VP answer go here in reply to Where does it go? is coded as nonadult, although it would be reasonable to claim that it is a sentence with a null subject.

⁸This child quite frequently used high inflection in sentences with a third person singular subject. That is, the third person agreement ending is expressed before the main verb, rather than being attached to the verb, as in the adult grammar. This phenomenon is documented in more detail in Thornton and Tesan (2007).

⁹VPs were counted as repeated VPs in a number of cases when the repeated VP was not identical. For example, an answer with pronominal reduction counted as a repeated VP (e.g., Q: Does the bear like cheese? A. Yes, he likes it). In addition, some permutations of semantically close verbs were permitted, such as got for have.

¹⁰Thanks to an anonymous reviewer for pointing out the relevance of Potsdam's 1997 paper.

¹¹There are a few earlier uses of don't, but analysis of Georgia's use of negation suggests that this form is an unanalyzed chunk at this point (see Thornton & Tesan, 2007).

¹²The example from the transcript at 2;5.07 is as follows:

• Georgia:  Did the rabbit ring the bell?

• Experimenter:  Rabbit did?

• Georgia:  Rabbit did.

¹³As a reviewer points out, Georgia's use of pitch accent is odd in this response. Since Daddy had already been introduced into the conversation, the subject noun phrase Daddy should receive the accent.