Cross-situational learning in expressive vocabulary intervention for Cantonese late talkers

Abstract

This study presents an expressive vocabulary intervention for Cantonese-speaking late talkers that incorporated cross-situational statistical learning principles and high treatment dosage. Target words were coded for phonological neighbourhood density to compare the treatment results for dense and sparse words. Three late talkers aged 25-39 months received 7-9 weeks of twice weekly input-based treatment in which no explicit verbal production was required. The number of target words spontaneously produced relative to the control words were measured by pre-session word probes, in-treatment productions, and parent reports of word use at home. Effect sizes for each data collection method were calculated. In addition, post-treatment growth in expressive lexicon sizes was measured using a validated parental checklist. Two out of three participants obtained considerable effect sizes of target words learned across all types of data, which were comparable to a precedent study (Alt et al., 2014). Also, when data in all conditions were included, treatment dosage contributed unique variance to the target word production scores for all participants. However, pre-post vocabulary gains were limited. Additionally, dense words did not have an advantage over sparse words. The results support the use of a high-dosage, cross-situational statistical learning paradigm in vocabulary interventions for late talkers.