First-of-a-kind study measures children’s utterances in Southern Bantu languages

A study published by the Child Language Development node of the South African Centre for Digital Language Resources (SADiLaR) is the first of its kind to measure children’s utterances in isiXhosa, Sesotho, Setswana and Xitsonga.

The research team, led by node manager and principal investigator Prof Heather Brookes, published their results in the Journal of Child Language, a peer-reviewed academic journal covering all aspects of the scientific study of language behaviour in children.

“Our study is the first study to try and measure children’s utterances in the four Southern Bantu languages of isiXhosa, Sesotho, Setswana and Xitsonga,” says Brookes. “We looked at whether using the three longest sentences a child said could indicate their language development and we found that it could.”

Prof Brookes is based at Stellenbosch University’s Department of General Linguistics, where SADiLaR’s Child Language Development Node is hosted. The main function of this node is to promote research on child language development in all South African languages and the digitisation of child language development data so that it is freely available on the SADiLaR platform for all scientists working on language, cognition, child health and development, language learning and language disorders.

Measuring early language development

The ability to assess young children’s early language proficiency accurately is crucial as any delays could lead to learning disabilities, anxiety and even behavioural problems, which could impact their success at school and life in general. While several international tools are available, a different set of tools is needed locally to accommodate the South African languages and their cultural contexts.

For their study, Brookes and eight of the Node’s team members (who are affiliated with the universities of Stellenbosch, Cape Town and Sefako Makgatho Health Sciences) analysed the utterances of 448 toddlers (aged 16 to 32 months) to measure their early language development.

“One way of measuring children’s language development is to measure how long their utterances or sentences are,” says Brookes. “As children develop, they start off with gestures, then one-word utterances (e.g., in Sesotho, metsi [water]), then two-word utterances (e.g., batla metsi [want water]), then three-word utterances (e.g., ke batla metsi [I want water]), and so on. Measuring their length of utterances can tell us how well they are doing, and speech-language therapists can measure how well a child is progressing,” she explains.

According to Brookes, this can be done by recording a child’s speech, but it can be difficult to get children to talk and their speech takes a long time to transcribe and analyse. Another way is to ask parents what the longest sentences are that their child has said in the past week.

Morphemes instead of words

“To measure utterances, in some languages, we can count the words a child says in one utterance and get the average number of words,” she continues. “However, in other languages this is more difficult because one word can contain many different parts that do different things. For example, in English we can say ‘I love you’, which is three words. However, in isiZulu we say ‘Ngiyakuthanda’, which is one word, but all of that information – Ngi (I), ya (tense marker), ku (you), Thanda (love) – shows the child is actually putting four different parts of speech together. This means we have to measure children’s language using what we call morphemes [the smallest unit of language that contains meaning] instead of words.”

The research team compared each child’s sentence lengths with the words and grammar they know, and found that measuring their three longest sentences did indeed reflect their progress. “We think that speech-language therapists might find this a useful way of gauging how well a child is putting sentences together. Our next steps will be to work out the average sentence length per age on larger samples,” Brookes adds. “Our sample was 100 per language for this study.”

Scarcity of appropriate child language assessment tools in African languages

To date, there are few language assessment tools developed and standardised as accurate diagnostic measures or indicators of either typical or atypical language development in South Africa and most other African countries.

“The scarcity of such tools places a serious burden on clinicians,” Brookes comments. “Our results have shown the measurement of utterance length in morphemes to be a better indicator of young children’s linguistic development in morphologically complex languages like isiXhosa, Sesotho, Setswana and Xitsonga, than a words-only approach which could underestimate their ability. The findings of our study are a small, but important, step towards addressing the need for linguistically appropriate child language assessment tools in African languages.”

Read the journal article:

Brookes H et al. (2024) Mean Length of Utterance: A study of early language development in four Southern Bantu languages. Journal of Child Language. 

ISSUED BY FUNDICONNECT