Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A hierarchical Bayesian model for syntactic priming

Weijie Xu, Richard Futrell

Abstract

The effect of syntactic priming exhibits three well-documented empirical properties: the lexical boost, the inverse frequency effect, and the asymmetrical decay. We aim to show how these three empirical phenomena can be reconciled in a general learning framework, the hierarchical Bayesian model (HBM). The model represents syntactic knowledge in a hierarchical structure of syntactic statistics, where a lower level represents the verb-specific biases of syntactic decisions, and a higher level represents the abstract bias as an aggregation of verb-specific biases. This knowledge is updated in response to experience by Bayesian inference. In simulations, we show that the HBM captures the above-mentioned properties of syntactic priming. The results indicate that some properties of priming which are usually explained by a residual activation account can also be explained by an implicit learning account. We also discuss the model's implications for the lexical basis of syntactic priming.

A hierarchical Bayesian model for syntactic priming

Abstract

The effect of syntactic priming exhibits three well-documented empirical properties: the lexical boost, the inverse frequency effect, and the asymmetrical decay. We aim to show how these three empirical phenomena can be reconciled in a general learning framework, the hierarchical Bayesian model (HBM). The model represents syntactic knowledge in a hierarchical structure of syntactic statistics, where a lower level represents the verb-specific biases of syntactic decisions, and a higher level represents the abstract bias as an aggregation of verb-specific biases. This knowledge is updated in response to experience by Bayesian inference. In simulations, we show that the HBM captures the above-mentioned properties of syntactic priming. The results indicate that some properties of priming which are usually explained by a residual activation account can also be explained by an implicit learning account. We also discuss the model's implications for the lexical basis of syntactic priming.
Paper Structure (22 sections, 6 equations, 3 figures, 1 table)

This paper contains 22 sections, 6 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Empirical Background
  3. Related Work
  4. The Current Study
  5. Modeling Framework
  6. Representing Syntactic Knowledge
  7. Updating Syntactic Knowledge
  8. Beta-Binomial Process.
  9. Hierarchical Bayesian Model.
  10. Model Evaluation
  11. Simulation 1: Lexical Boost and Inverse Frequency
  12. Estimating the prior.
  13. Inferring the posterior.
  14. Result.
  15. Discussion.
  16. ...and 7 more sections

Figures (3)

  • Figure 1: Hierarchical representation of syntactic statistics.
  • Figure 2: Results of Simulation 1. Panel A: Model-estimated average prior and posterior probability of DO for the target verbs in pickering1998representation; Same refers to the condition with verb overlap between the prime and the target; Different refers to the condition without verb overlap. Panel B: Model-predicted priming effect size, calculated as the difference of log-odds between the posterior and the prior for DO.
  • Figure 3: Simulation 2 model-predicted priming effect size as a function of the number of additional batches of post-priming data. Effect size calculated as in Simulation 1.