Congratulations to Linguistics PhD candidate Danielle Burgess, who successfully defended her dissertation on March 17. Committee chair is Marlyse Baptista.

Dissertation Title

Investigating the Source of Neg-First Biases in Typology and Acquisition


The tendency for negation to appear early in the sentence, dubbed the “Neg-First principle” by Horn (1989:452), has been observed in the domains of typology, language contact, and language acquisition. Based on evidence from these fields, scholars have speculated about the source and universality of Neg-First biases affecting language learning or use, which may in turn shape how linguistic systems change over time. In this dissertation, I consolidate the varied evidence and proposed explanations for the Neg-First principle and lay out the challenges and limitations of inferring the source and existence of universal cognitive constraints from typological records or acquisition evidence, due to sampling bias and uncertainty about how a purported bias should be expected to interact with other linguistic patterns and constraints. I then introduce several artificial language learning experiments to test for the presence of Neg-First biases in different tasks and populations, in order to gain a better understanding of the possible sources of Neg-First tendencies in typology and acquisition, while controlling for features of the language and the learner's input.

Through a series of three experiments, I test the presence of Neg-First biases in the early stages of second language acquisition by examining whether participants exposed to an artificial language containing both preverbal and postverbal negation demonstrate a bias to overproduce preverbal negation compared to their training input. Experiment 1, testing whether English- speaking participants demonstrate a preference consistent with a Neg-First bias, produced inconclusive results. Participants did show a small numerical tendency to produce more preverbal negation than was present in the training input, but this effect was not statistically significant. Experiment 2 addresses potential limitations of Experiment 1 by adding a dyadic interaction component, to test the hypothesis that Neg-First biases are driven by a communicative desire to reduce potential for the listener to misunderstand the intention of the sentence and removing the presence of a word bank to encourage greater regularization by making lexical retrieval more difficult. The results of Experiment 2 do indicate a bias to overproduce preverbal negation among English speakers, consistent with a Neg-First bias. However, interaction did not boost the use of preverbal negation relative to the solo production phase, except in the Majority Preverbal Negation Condition, where the bias to regularize the majority order in the training language was consistent with a bias to use preverbal negation. Finally, Experiment 3, tests for a preverbal negation preference among speakers of a postverbal negation language (Japanese), to investigate the role of previous language experience on the preference to produce preverbal negation in adult L2 acquisition. While English speakers show a preference to produce preverbal negation even in the absence of communication, Japanese speakers do not.

These results challenge the idea of a universal preference for preverbal negation in acquisition as a plausible candidate for shaping linguistic typology, though more nuanced explanations which retain the role of universal biases at play in language processing and acquisition are viable avenues for future exploration. These possibilities underscore the importance of examining the role of language experience, development, and contact when inferring universal principles propelling language variation and change.