Unacceptable grammars? an eye-tracking study of English negative concord

Abstract

This paper uses eye-tracking while reading to examine Standard English speakers’ processing of sentences with two syntactic negations: a negative auxiliary and either a negative subject (e.g., Nothing didn’t fall from the shelf) or a negative object (e.g., She didn’t answer nothing in that interview). Sentences were read in Double Negation (DN; the ‘she answered something’ reading of she didn’t answer nothing) and Negative Concord (NC; the ‘she answered nothing’ reading of she didn’t answer nothing) biasing contexts. Despite the social stigma associated with NC, and linguistic assumptions that Standard English has a DN grammar, in which each syntactic negation necessarily contributes a semantic negation, our results show that Standard English speakers generate both NC and DN interpretations, and that their interpretation is affected by the syntactic structure of the negative sentence. Participants spent more time reading the critical sentence and rereading the context sentence when negative object sentences were paired with DN-biasing contexts and when negative subject sentences were paired with NC-biasing contexts. This suggests that, despite not producing NC, they find NC interpretations of negative object sentences easier to generate than DN interpretations. The results illustrate the utility of online measures when investigating socially stigmatized construction types.

Asymmetries in the Acceptability and Felicity of English Negative Dependencies: Where Negative Concord and Negative Polarity (Do Not) Overlap

“…Regarding the asymmetry between NC acceptability and felicity, we note that this finding both supports and complements previous work which compared NC with DN, its truth conditional opposite (Blanchette, 2017; Blanchette et al, 2018; Blanchette and Lukyanenko, 2019). In these studies, preceding context was employed to elicit an NC or a DN reading for sentences a subset of which were parallel to the critical sentences presented here.…”

“…While vernacular English varieties are known for instantiating NC (Wolfram and Fasold, 1974; Nevalainen, 2006), a series of recent experimental studies show that SE speakers also have reliable intuitions about this construction type. The studies show that SE speakers have a clear knowledge of the syntactic distribution of NC (Blanchette, 2017), an understanding of its meaning and prosodic properties in relation to DN (Blanchette et al, 2018), and an apparent proclivity toward building NC structures during online processing (Blanchette and Lukyanenko, 2019). These studies all involve comparison of sentences with a negative noun phrase in direct object position following a negative marker as in (11) (and (6/7) above), and sentences with a negative noun phrase in canonical subject position preceding a negative marker, as in (12) 6 ,7 :…”

“…Both strict and non-strict NC languages have sentences like (11), in which a negative noun phrase is preceded by and acts in concord with a negated auxiliary, but only strict NC languages have sentences like (12), in which the negative noun phrase both precedes and acts in concord with the negated auxiliary. On the basis of their findings, Blanchette and Lukyanenko (2019, p. 24) therefore suggest that SE may be categorized as “non-strict 8 .” They further note a similarity between speakers’ subtle intuitions about NC in SE, and more obvious intuitions about parallel NPI constructions. To illustrate, consider the following contrast:…”

“…However, along with theories such as Zeijlstra (2004, et seq. ), it is difficult to extend to grammatical systems that generate both NC and DN (e.g., Déprez, 2011; Puskás, 2012; Déprez et al, 2015, among others), including English (e.g., Blanchette and Lukyanenko, 2019). For example, if English negative phrases are NPI-like, then they should not be able to occur in DN constructions.…”

Processing Sentences With Multiple Negations: Grammatical Structures That Are Perceived as Unacceptable

“…In fact, some theoretical proposals (e.g., Zeijlstra, 2004; Tubau, 2008; Blanchette, 2013, 2015) have hypothesized that the underlying structure of Standard English is that of negative concord. In this vein, Blanchette and Lukyanenko (2019) demonstrate that, in the absence of the necessary licensing conditions, native speakers of English can actually interpret double negation dependencies as negative concord. The participants in our experiments were not explicitly tested for having grammars that allowed negative concord dependencies in order to avoid calling attention to the manipulation.…”