Towards Validation of South African Communicative Development Inventories: An object naming/identification task for South African English and Afrikaans

Husselmann, Nicole

Towards Validation of South African Communicative Development Inventories: An object naming/identification task for South African English and Afrikaans

Master Thesis

2022

Abstract

Background: South Africa is a country rich in diversity, with language constituting a large proportion of this diversity as the country has eleven official languages with Afrikaans, isiXhosa and South African English (SAE) being the provincial languages of the Western Cape. Research on language acquisition in South Africa is limited, but vital for the early identification, assessment and management of children who are not developing language in a typical way. A recent project in South Africa has led to the development of Communicative Development Inventories (MB-CDIs) in several of the country's official languages. MB-CDIs are parent-report questionnaires about young children's language acquisition. It is challenging to validate these MB-CDIs in the local context as there are few other language assessments available for comparative purposes. Aims and Objectives: This project aimed to develop a protocol for language assessments that can be used as part of the validation process for the MB-CDIs in South African languages. The objectives were (1) to devise a protocol for an object naming and object identification task for use with six South African languages (Afrikaans, isiXhosa, SAE, Sesotho, Setswana, and Xitsonga); (2) to describe the expressive and receptive language of children acquiring SAE based on assessment with two different tools/approaches (the object naming/identification task and parent-report MB-CDI); and (3) to describe the expressive and receptive language of children acquiring Afrikaans based on assessment with two different tools/approaches (the object naming/identification task and parent-report MB-CDI). Methodology: This study used a mixed-methods, descriptive-linguistic approach. The first objective relied on literature only and no participants were needed. I devised a protocol for a novel object naming/identification task1 which could be used to validate the MB-CDIs in six languages (Afrikaans, isiXhosa, SAE, Sesotho, Setswana, and Xitsonga). For objectives 2 and 3, focusing solely on SAE and Afrikaans, I used the protocol to devise an object naming/identification task to be used with each language, respectively. Participants (toddlercaregiver dyads) were recruited and a pilot study undertaken in which the object naming/identification task was administered to each child participant. I collected parental report data by either administering the MB-CDI in SAE and Afrikaans manually to the adult participants (parents/guardians of the children), or by providing them with a link to the online version of the MB-CDI which they were able to complete in their own time. A total of 35 dyads were included in this study: 17 child and 17 adult participants for SAE and 18 child and 18 adult participants for Afrikaans. Results: SAE: The descriptive statistics suggested variability in the overall scores on both the object naming/identification task and the MB-CDI. Correlations between the 25 items in the object naming/identification task and the same 25 items in the MB-CDI indicated a significant correlation between the two assessments overall (r=0.928; p<0.05) and a significant correlation was indicated for toddlers only (r=0.935; p<0.05) and a significant correlation was indicated for toddlers only (r=0.901; p< 0.01). Correlations between the object naming/identification task and the entire MB-CDI indicated a significant correlation (r=0.901; p< 0.01) for overall vocabulary scores. Internal reliability scores for only the 25 items included in the object naming/identification task and the 25 items in the MB-CDI also indicated a high internal consistency for the language domains (α=0.911). For the gesture domain, a reasonable but lower internal consistency (α=0.673) was found. Internal reliability scores overall indicated a high internal consistency for the language domains (α=0.947). In contrast, the internal consistency indicated for the gesture domain was not satisfactory (α=0.293). Afrikaans: The descriptive statistics indicated variability among scores obtained on the MB-CDI and the object naming/identification task. Correlations between the 25 items included in the object naming/identification task and only the same 25 items included in the MB-CDI, indicated a strong correlation (r=0.906; p< 0.01). No correlation was found for infants only (r=0.692; p< 0.05) and a significant correlation was indicated for toddlers only (r=0.901; p< 0.01). Correlations between the object naming/identification task and the entire MB-CDI indicated a strong correlation for overall vocabulary scores (r=0.909; p< 0.01). Internal reliability scores on only the 25 items included in both assessments indicated a high internal consistency for the language domains (α=0.960). Internal reliability scores between the object naming/identification task and the entire MB-CDI indicated a high internal consistency for the language domains (α=0.950). Similarly, a strong internal consistency was found for the gesture domain (α=0.914). Conclusion: The findings indicate that when correlating the responses to the 25 items included in the object naming/identification task and the MB-CDI only, as well as from the entire vocabulary and gesture sections of the MB-CDI, there is high concurrent validity between the two assessments overall for both SAE and Afrikaans. Focusing on toddlers and infants more specifically, although high concurrent validity was found overall, the findings for both SAE and Afrikaans suggest that there is high concurrent validity for both assessments for toddlers, but no correlation between the assessments for infants. High internal consistency indicated reliability for both assessments for SAE and Afrikaans. The object naming/identification task, which was developed and piloted in this study, has the potential to be used as a measure to validate full MB-CDIs in SAE and Afrikaans. Furthermore, it has the potential to be used as a valid and reliable screening tool in its own right, although further refinement is needed for the infant component of the task. The protocol developed in the first objective of this study can now be applied to other local languages, ultimately contributing to the creation of a valid and reliable set of assessments of young children's early language development in South Africa.