Accessing linguistic competence: Evidence from children's and adults' acceptability judgments

Date of Completion

January 2000


Language, Linguistics




In order to learn about grammar, linguists primarily rely on acceptability judgments from native speakers of the language under investigation. Our hope is that these judgments allow us to tap into people's competence, or their knowledge of the language, and allow us to investigate the grammar. However, there has been some criticism raised regarding the use of judgments and what they tell us about competence and performance. ^ First, many researchers have argued that grammaticality judgments are not appropriate for studying children's competence since they are not able to perform metalinguistic tasks. However, McDaniel and her colleagues have argued that children, as young as 2;11, are capable of providing consistent and reliable judgments if they are trained. ^ In this study, I provide additional evidence that children are able to give reliable grammaticality judgments, and show that a combination of production and judgment data may reveal more about the child's grammar than production data alone. ^ In particular, I investigate children's non-adult negative questions with doubled auxiliary verbs, as in (1). (1) What did the smurf didn't buy? My studies show that children produced 2Aux questions sentences, yet judged them to be ungrammatical. I argue that these children do in fact have the adult grammar, contrary to recent proposals, and that their production of 2Aux questions is a performance error related to knowledge about constituent negation. ^ A second concern has been raised with respect to the study of the adult grammar. Linguists have noticed anecdotally that certain types of island violations become increasingly acceptable after repeated exposure. In order to determine whether this so-called “syntactic satiation” is a general performance phenomenon or constrained by syntax, Stromswold (1986) and Snyder (1994, 2000) have investigated it experimentally. In this study, I replicate Snyder (1994) and test additional types of island violations. I also examine whether subject-related, such as handedness or linguistic training, and task-related factors, such as general reading ability, response time and presentation method, are associated with satiation. The evidence from the studies suggests that syntactic satiation is constrained by syntax and that it is a reflection of competence. ^