Incidental word learning from print: The role of semantic diversity

Incidental word learning from print: The role of semantic diversity

First Author: Matthew HC Mak -- University of Oxford
Additional authors/chairs: 
Yaling Hsiao; Kate Nation
Keywords: Word Learning, Implicit Learning, Role of context, Computational modelling, Lexical Quality
Abstract / Summary: 


In three experiments, we investigated how semantic diversity (SemD) of the context in which a novel word was experienced influences how well it is learnt.

Method & Results

In Experiment 1, 40 undergraduates read passages containing pseudowords (e.g., ossanic); half of them were experienced in different-themed passages (high SemD) while the other half were experienced in same-themed passages (low SemD). Results from pseudoword inferiority, speeded recognition, and semantic judgement suggest that in the initial stage of word learning, low-SemD contexts facilitated word learning.

Experiment 2 adopted a similar design, but all pseudowords were first experienced in low-SemD passages. Afterwards, half of them were experienced in a different semantic context (high SemD) while the other half remained in the same context (low SemD). In contrast to Experiment 1, words in the high SemD condition showed better learning.

Experiment 3 is a computer simulation study. We hypothesized that low and high SemD influence the standing of a target word in a lexical network, thereby affecting spreading activation. Using the reading materials from the two behavioural experiments, we built lexical networks for each pseudoword and simulated spreading activation. Results from two sets of simulation provided further credence to the behavioural data.


(1) Context of low SemD (same-themed context) helps anchor new words into pre-existing memory, facilitating the learning of new words in the early stage. (2) In the longer run, high SemD is required to further refine mental representation of new words.