The impact of test language and multilingualism on digit span performance in South Africa

<b>Objective:</b> Digit span tests are neurocognitive tests that measure attention and working memory. They depend on language-based auditory input and verbal responses. This study examined the effects of language of digit span administration in a linguistically diverse South African sample. <b>Method:</b> Data were drawn from four South African normative studies (<i>N</i> = 512; ages 8-79) that included a digit span test as part of a broader neuropsychological battery. Data were stratified by age, education level and quality, and language. Forward and backward span scores were analyzed in relation to these variables. <b>Results:</b> For the forward span, amongst the educationally advantaged and disadvantaged children and the advantaged adults, the participants who heard and said the numbers in Afrikaans when it was their first language, scored significantly worse than those who heard and said the numbers in English when English <i>was</i> or <i>was not</i> their first language (<i>t</i>(23) <i>=</i> -4.48<i>, p =</i> <.001 for disadvantaged children; <i>t</i>(117) <i>=</i> -2.2<i>, p =</i> .03 for advantaged children; <i>t</i>(104) <i>=</i> -5.48<i>, p</i> < .001 for advantaged adults). Significant differences were not found for the educationally disadvantaged adults, although there was a trend toward lower scores when the numbers were said in Afrikaans. For the backward span, no significant differences were found when the numbers <i>were</i> or <i>were not</i> heard and said in the participants' first language. Findings suggest that digit span may be sensitive to auditory-linguistic interference in people who speak more than one language, complicated by language proficiency and complexity of number names. <b>Conclusions:</b> In multilingual settings, digit span test performance may be affected by both first language and language of administration. These findings support the need for neuropsychological norms that also stratify for examinees' language and the language of administration. Clinical neuropsychologists should consider factors such as language proficiency and complexity of number names when assessing working memory in linguistically diverse populations.