Three types of negated constituent movement during the history of English -NegV1, Negative Inversion and Negative Movement -are linked to the role of NegP in the syntax of English up to the early Modern period, building on proposals made by Haegeman (1995), van Kemenade (2000) and Zeijlstra (2004). NegP is analysed as involving the presence of a Neg Operator, null in languages with a head negator. A high NegP licensed NegV1 in Old and early Middle English, the optionality of a Neg Operator in NegP triggered Negative Movement in Late Middle English until NegP was lost, and the loss of NegP in Early Modern English permitted Negative Inversion. The absence of NegP is proposed for the earliest attested stages of English as well as of other Indo-European languages, as a way of accounting for a stage where NegV1 and Negative Concord were absent. 1 We thank the editor and an anonymous reviewer for their many helpful suggestions aiming to clarify and improve the article. Thanks also go to Ans van Kemenade and Susan Pintzuk for their critique of some the ideas presented here. All remaining errors of fact and interpretation are our own.

Optional OV order in Later Middle English (LME) has given rise to conflicting theoretical accounts. Earlier analyses postulating movement to AgrOP or alternative base orders are found to be inadequate to deal with the occurrence of OV in non-literary LME; in a large database of 15th century private familial correspondence, residual OV order is found to have been productive only with negated objects. Multiple subject constructions with there expletives showed the same restriction. These phenomena are accounted for by postulating overt Neg Movement (Haegeman 1995) as a permitted option in LME. In this framework, it is argued that LME showed a mixed typology having both Neg movement and a null Neg operator. LME had three ways of satisfying the NEG Criterion (Haegeman 1995): Merge not in Spec NegP, coindex [OP]i … [XP(Neg)]i, and Move XP(Neg) to Spec NegP. Modern English has only the first two. The distribution in this period of negative concord with not is shown to support our analysis.

No abstract

Using a large database of familial correspondence, it is shown that NP positions in 15th-century English were essentially those of Present-Day English, contrary to claims that the syntactic structure of Late Middle English still had a position for preposed object NPs, and that 15th-century English possessed multiple subject constructions with expletive there. The only form of OV order to remain productive in familial correspondence of this period occurred in the configuration: finite verb–negated NP–lexical verb. Neither ordinary NPs nor other quantified NPs were productive in this configuration. It is also shown that negated subject NPs, but not ordinary or other quantified subject types, were commonly found in multiple subject constructions. All these phenomena can be conveniently accounted for in terms of movement to a Neg projection (Haegeman, 1995). A theoretical account of noncanonical subject and object positions in terms of Neg movement thus receives strong empirical support from the distribution of argument NPs in 15th century English.

The Uiqiversit)! of ReadingIt has been proposed that grammatical specific language impairment (SLI) is characterized as a deficit affecting only feature-related aspects of grammar. The research reported here indicates a wider irnpairnient involving aspects of grammar not determined by feature checking. in particular to the structure of the verb phrase (VP) with resultative s~condary predication. The results of two video elicitation tasks showed children with SLI to have a significant deficit on such VP constructions compared with chronological age matches and younger vocabulary matches. These findings are accounted for by the deficit in dependency relations proposed by van der Lely (1996) to the effect that nonbinary syntactic dependencies are vulnerable in grammatical SLI. Additionally, it is shown that research using mean length of utterance matching of children with SLI may obscure syntactic problems revealed by matching with younger children on another language trait.