Processing-based measures of implicit statistical learning of artificial grammars

Abstract

Implicit statistical learning, whereby predictable relationships between stimuli are detected without conscious awareness, is important for language acquisition. Although a defining feature of implicit learning is that we are unaware that learning has occurred, implicit statistical learning is often assessed using measures that require explicit reflection (e.g., judgements about the grammaticality of a sequence of stimuli). However, implicit statistical learning can also be assessed without requiring conscious reflection, using ‘processing-based’ tasks, that instead measure other processes that are facilitated by implicit statistical learning such as reaction times or serial recall. Processing-based measures would be particularly valuable for measuring implicit statistical learning in populations such as children (who may be less adept at explicitly reflecting on implicitly learned knowledge) or individuals with dyslexia (which has been associated with a specific deficit in implicit statistical learning, rather than in explicitly reflecting on implicitly learned information). In this thesis, I developed and tested novel processing-based measures of implicit statistical learning that do not require conscious reflection and combined them with traditional reflection-based measures of learning to investigate the nature of the knowledge acquired by way of implicit statistical learning. I found evidence of implicit statistical learning across a number of experiments, which also suggest that the complexity of the grammar being learned may affect the extent to which implicit and explicit processes are recruited during the tasks. I then applied this novel serial visual recall paradigm to provide evidence that implicit statistical learning abilities are consistent across children aged 8 to 15 years and that serial visual recall may capture differences between children and adults that are not reflected in traditional measures of learning. Finally, I applied these paradigms across a number of experiments to assess differences in implicit statistical learning between individuals with and without dyslexia or other reading difficulties, and found no evidence that dyslexia is associated with a deficit in implicit statistical learning. Overall, these experiments suggest that processing-based measures are a valuable tool for measuring implicit statistical learning across a number of different populations and highlight the importance of using a combination of both processing- and reflection-based tasks to gain a more detailed insight into the nature of the knowledge acquired through implicit statistical learning.