The Syntax-PF Interface in Children’s Negative Sentences

Abstract

To test between two recent accounts of the early stages in the acquisition of negation, we conducted an elicited production study with 25 children, between 2;05 and 3;04 (mean 2;11). The experimental study produced a robust set of negative sentences, with considerable individual variation. Although 13 of the child participants mainly produced adultlike negative sentences with doesn’t, 12 children produced nontarget forms. The nonadult productions included medial negation structures, both with bare verbs (It not fit) and with inflected main verbs (It not fits, It don’t fits), as well as negative sentences with “high” inflection (It’s not fit) and doubling (It’s not fits). Although some of the findings are consistent with the account advanced by Harris & Wexler (1996), their prediction that negative sentences with do-support (It don’t/doesn’t fit) co-occur with medial negation and a bare verb (It not fit) was not borne out. The findings are amenable to the account advanced by Thornton & Tesan (2013), who contend that children’s grammars are initially restricted to adverbial negation. Children’s nontarget productions with the 3SGS morpheme positioned on the main verb or in Infl are attributed to a nonadult mapping between syntax and phonological form (PF). The steps children take to converge on the adult mapping are explained by invoking the theory proposed by Adger (2003).

Notes

¹ Once children set the pro-drop parameter to the English value, the assignment of nominative case to VP-internal subjects is discarded as an option. An empirical problem arises, however, as Déprez & Pierce (1993) observe. The problem is that the disappearance of overt VP internal subjects in children’s productions precedes the disappearance of null subjects. This leads Déprez & Pierce to argue that, although nominative case assignment to VP-internal subjects is eliminated from children’s grammar when the pro-drop parameter is reset, referential null subjects continue to be licensed by pragmatic principles. Further discussion of this issue would take us beyond the scope of the article.

² More specifically, Harris & Wexler analyze do as a semantically null modal that is base-generated in an AgrP projection (under NegP) and raises to the tense projection to check its tense feature. It is not clear how this account explains the fact that children can produce don’t without the 3SGS. If the null modal is base-generated in an AgrP projection, it could be expected to show the 3SGS agreement. We put this issue aside for the remainder of the article.

³ Assuming that not is an adverb in the adult grammar also, one could ask why the 3SGS morpheme does not appear on the main verb in adult English. This area of syntax is not well understood and is not addressed further here. The empirical observation, however, is that once children acquire do-support, utterances with inflection on the main verb soon disappear.

⁴ Schütze (2013) suggests that there is nothing that makes “superfluous” (nonemphatic) do fail to converge in the derivation of a sentence like It does fit. He proposes that there is a constraint that drives T to lower to V under adjacency. The constraint that says that if two convergent derivations based on the same numeration differ in the number of overt morphemes (aka vocabulary items) needed to spell them out, the derivation using fewer morphemes blocks the one using more morphemes. To calculate which derivation is preferable, children need to compare the two derivations, which is frequently beyond the processing capacity of children. Therefore, once children’s processing capacity matures, children will become adultlike. This proposal doesn’t render It’s not fit or It not fits any less economical than It doesn’t fit, however.

Additional information

Funding

We acknowledge that this research was funded by the Australian Research Council Centre of Excellence in Cognition and its Disorders (CE110001021) http://www.ccd.edu.au.