The role of positive vs. negative evidence in learning a novel dialect pattern: American English speakers’ grammatical intuitions on a-prefixing in Appalachian English

Abstract

This paper focuses on the controversies surrounding the effect of linguistic input in language acquisition and investigates how learning of a dialectal pattern, a-prefixing in Appalachian English, can be modulated by positive vs. negative evidence in adult learners whose grammar does not contain this grammatical phenomenon. In an online acceptability judgment study, we measured gradient grammatical intuitions of 41 Appalachian English and 36 native speakers of American English on their perceived acceptability of carefully-matched grammatical and ungrammatical sentences with a-prefixing. We used 16 sentences (50% ungrammatical) for each of the 5 previously established constraints on a-prefixing, divided equally into a pre- and post-test session. The non-dialect speakers were further divided into 3 input groups depending on the type of input they received between the two tests: the positive-evidence group with only grammatical instances of a-prefixing, the negative-evidence group with explicit instruction concerning only ungrammaticality, and the control group with no input. Results indicated that, just like the baseline Appalachian English group, the non-dialect participants were able to distinguish grammatical sentences from ungrammatical ones, crucially in all of the 5 different constraints, prior to the treatment, although they significantly differed from the Appalachian English group in the strength of their judgments. The post-test results revealed a clear benefit of negative evidence on the learning of a novel grammatical phenomenon, albeit only concerning the responses to ungrammatical sentences, while accept-ability judgments for grammatical as well as ungrammatical sentences did not benefit from positive evidence alone. We suggest that the target-like intuitions of our non-dialect participants at the onset result from a deduction of tacitly known principles of English and linguistic universals. Concerning the differential effect of input categories on learning, we argue that exposure to only positive exemplars of the target phenomenon results in overly general hypotheses about its patterning in the language, which may push the learner to perceive new ungrammatical sentences better than their initial acceptability value. Negative exemplars, however, place a pressure on learners to notice the linguistic contexts in which the target phenomenon is not allowed to surface, thereby raising their awareness of the presence of rules.