The role of chunking and analogy in early vocabulary acquisition and processing

Cabiddu, Francesco 2023. The role of chunking and analogy in early vocabulary acquisition and processing. PhD Thesis, Cardiff University. Item availability restricted.

Preview	PDF - Accepted Post-Print Version Download (9MB) | Preview
	PDF (Cardiff University Electronic Publication Form) - Supplemental Material Restricted to Repository staff only Download (3MB)

Abstract

Chunking and analogy, learning through associations and similarities respectively, are crucial cognitive processes in a usage-based theory of language development. Assessing their roles in child naturalistic word learning has posed significant challenges. In this thesis, I offer methodological solutions to examine the developmental plausibility of these processes. Chapter 2 discusses limitations in studies of early word segmentation from naturalistic speech, affecting conclusions about the processes' developmental plausibility. I present a new chunking-based model, CLASSIC Utterance Boundary (CLASSIC-UB), to study how English infants discover words from continuous naturalistic speech. Its plausibility is assessed through new metrics focusing on child production vocabularies from large-scale conversational corpora. I show the advantages of using large word production samples and how this can improve the refinement of early word segmentation and learning theories. In Chapter 3, conclusions about CLASSIC-UB’s plausibility are supported by extending this approach cross-linguistically, using Italian as a case study. Across Chapters 2 and 3, CLASSIC-UB more accurately captures child productions than other chunking and non-chunking accounts, supporting its plausibility in early word segmentation and learning. In Chapter 4, I identify methodological challenges in assessing the independent effects of chunking and analogy in child word processing. I focus on how children use sentence context to resolve ambiguous word meanings (word sense disambiguation). I present ChiSense-12, a new open-access sense-tagged corpus of child-directed speech, and describe its use in creating experimental stimuli to disentangle variables (verb-object associations and verb-event structures) that are informative about the independent role of chunking and analogy. Using this corpus, I showed - for the first time - that 4-year-old children exploit both bottom-up verb-object associations and top-down verb-event structures to resolve lexical ambiguities. Overall, this thesis makes a significant contribution to usage-based theories of language development and improves our understanding of how children acquire language in real-life contexts.

Item Type:	Thesis (PhD)
Date Type:	Completion
Status:	Unpublished
Schools:	Psychology
Date of First Compliant Deposit:	2 January 2024
Last Modified:	03 Jan 2024 14:39
URI:	https://orca.cardiff.ac.uk/id/eprint/165140

Actions (repository staff only)

Edit Item