This study reports on the development of a Setswana tonal minimal pair word list, which could be implemented as research tool in the field of Bantu language linguistics and in speech pathology in South Africa. The development of the list was conducted in four phases. These are described as four separate studies. All involved Setswana L1 participants living in the urban areas of Gauteng. In Study 1, a 45-pair preliminary list was compiled from dictionaries. During Study 2, eleven L1 speakers' familiarity with each word was determined. Based on these results the list was narrowed down to 20 pairs. Study 3 entailed the validation of pictorial stimuli, which illustrate the target words. Ten different participants took part. Four pairs were not consistently familiar and were removed, resulting in the experimental list of 16 pairs. This list was validated in Study 4 and involved nine typical L1 speakers and five listeners. Word-specific analyses revealed that some words had a negative impact on the results. Six pairs were removed. A final list of 10 pairs rendered results more aligned to the expectation of typical speakers and listeners. Validation should continue to determine applicability in populations from exclusively rural or urban areas.

Purpose. The aim was to determine if the presence of a voice disorder in speakers of Setswana, an African tone language, will negatively impact the accuracy of identification by typical first language judges of words belonging to tonal minimal pairs. Method. A quasi-experimental between-group comparison and individual case studies were conducted. Five participants with different types and degrees of voice disorders and nine control participants produced 10 tonal minimal word pairs. Five judges had to identify which of a pair was produced.Result. The mean scores of the control and experimental speakers as groups differed, but the difference was not statistically significant. Control participants scored between 19.6/20 and 14.2/20 words correctly identified. Individual data revealed that four of the nine control participants attained at least one perfect score across judges and six had mean scores of 18.0/20 and higher. The highest scoring experimental participant, presenting with a mild voice disorder, attained a mean of 18.0/20. The lowest scoring participant, presenting with the most severe dysphonia, had a mean of 12.2/20 words correctly identified. Conclusion.These preliminary results appear to suggest that a severe voice disorder could compromise lexical tone variation and by implication the intelligibility of a message.