Tracking early spelling development in English for bilingual children: A Latent Transition Analysis

Tracking early spelling development in English for bilingual children: A Latent Transition Analysis

First Author: Malikka Habib -- Nanyang Technological University
Additional authors/chairs: 
Beth O'Brien
Keywords: Latent profile analysis, Growth Modeling, Spelling errors, Alphabetic writing, Development
Abstract / Summary: 

Purpose:
Children acquire spelling skills over an extended period of time, and this developmental process has been described through phase models (Ehri, 1989). Accordingly, children accrue specific sets of spelling feature knowledge, which has been tracked by the types of spelling errors they make (Bear, Invernizzi, Templeton, & Johnston, 2000, 2016). Despite support for such models, their generalization to different types of learners is not established. This study aims to examine the developmental progression of spelling in English by bilingual children using a latent transition analytic approach. We also examined whether their additional language may impact the growth pattern of English spelling skills.
Method:
Spelling patterns of 1401 bilingual children in Singapore were collected 3 times over 2 consecutive years of kindergarten, from age 57.33 mos (SD=15.10) to age M= 69.77 mos (SD= 14.22). Performance on the first 10 items of the WRAT-IV spelling test was scored on 9 developmental features outlined by Bear et al. (2000), and 5 latent groups of spellers were classified using latent profile analyses (LPA) for each wave of data. These data were then entered into a latent transition analysis (LTA mixture model, Mplus v.8, 2017), to track children’s movement through different spelling classes. Children’s additional language (either Mandarin, Malay or Tamil) was then entered as a covariate to the LTA model.
Results/Conclusion:
After the first step of establishing best fitting, 5-class LPA models at each time point, latent transition analysis showed a wide range of growth patterns. The largest sets of individuals started from an early phase of random letter writing, and then transitioned to either an alphabetic phase or an indiscriminant use of a wider range of strategies. In addition, when entered as a covariate, the type of children’s additional language showed an interaction with the transition patterns.