Electronic Thesis and Dissertation Repository


Doctor of Philosophy




Archibald, Lisa M. D.

2nd Supervisor

Joanisse, Marc F.

Joint Supervisor


In this thesis, I investigated the process of statistical word segmentation using a combination of behavioural, clinical, and neuroimaging approaches. Prior research has largely focused on the outcome of statistical learning approaches, with little research attention paid to the process of learning. In this body of research, I sought to address this issue. In Chapter 2, I examined how domain-specific and domain-general working memory interference effects on a statistical word segmentation task. I found that when completing a concurrent visuospatial or verbal working memory task, statistical language learning was impaired. Thus, this study provided some evidence that domain-general working memory may support statistical language learning. In Chapter 3, I further investigated how cognitive processes, including language and working memory, are involved in statistical learning across domains. In this study, school-aged children with and without a developmental language disorder (DLD) completed a statistical language learning task and a visual statistical learning task. I found that those with DLD did not differ from typically developing children on either statistical learning task, and that performance across groups was meager for the statistical language learning task, and not above chance levels for the visual statistical learning task. Further, performance on the statistical learning tasks was not associated with other cognitive processes. This raised the possibility that an alternative measurement approach may be better suited to examine statistical learning. I addressed this issue in Chapter 4, where I measured event related potentials (ERPs) using electroencephalography (EEG) during exposure to a structured, unsegmented language. I found that statistical learning performance was related to neural responses to the structured linguistic input, and that ERPs were modulated as a function of language exposure, revealing the dynamic nature of statistical learning. Chapter 5 discusses the relevant findings of this thesis in relation to the current state of affairs in statistical learning research, and presents recommendations for future research in examining the process of statistical learning.