Digital Assistant for Financial Transactions

In a country of 11 official languages, only 2 languages, namely, English and Afrikaans, are dominantly used in the financial domain. Many banks and other institutions offering Fintech services for improving financial literacy are not fully accessible to the South African population as they do not support all of the indigenous languages of South Africa, such as isiZulu. In our work, we have addressed this gap by developing a mobile application that assists users to manage their money. The application supports isiZulu, the most widely spoken indigenous language in South Africa, and English. The mobile application adapts a trust inducing interface proposed by Wang and Emurian and incorporates a Natural Language Generator (NLG) that will generate summaries to inform users about their finances. We invited isiZulu and English speakers to evaluate our system, and 67% of the participants found the interface trustworthy, 83% revealed that they think the application will positively impact their financial literacy, 78% revealed that they found that the NLG produced understandable summaries and 83% revealed that the NLG system was accurately verbalising numbers. From the evaluation, we concluded that the interface is trust inducing and that the application will positively impact users’ financial literacy. We could also conclude that the NLG system was accurately verbalising numbers and producing understandable informative summaries. We hope our work will inspire existing and future Fintech products to incorporate indigenous languages.