International Mother Language Day
has been observed annually on 21 February since 2000. Gerhard Taljaard, competence area manager of the CSIR’s human language technologies (HLT) group, sees this commemorative day as a timely affirmation of the need to promote linguistic and cultural diversity and multilingualism in South Africa.
Taljaard took on his leadership role at the CSIR Meraka Institute in November 2011. He holds qualifications from three South African universities (B. Eng (Electronic), BSc (Hons) in Computer Science and an MBA), which are matched by 16 years of experience in industry, both locally and abroad. Highlights of his career to date have been leading a South African software team to develop various speech technologies and then marketing and selling South African-designed hardware and software in multicultural contexts in Europe, the Middle-East and North and South America.
Taljaard strongly believes in the intrinsic power of a mother language to unlock the potential of individuals, “Children who learn in their mother language have fewer hurdles to overcome when learning the basics.” He supports the view of well-known South African, Dr Neville Alexander – a linguistic revolutionary – that “unless we have spaces where people use languages they know best, productivity and efficiency will continue to suffer”. Using technology to make this imperative of linguistic diversity a reality, is the vision of Taljaard and his talented group of linguists, computer scientists and engineers.
The HLT research group has made strides in this regard. Through the Lwazi project funded by the Department of Arts and Culture, the HLT team produced the technology building blocks to model South Africa and Africa’s under-resourced languages. Taljaard explains, “Resources have been developed over many years for widely-spoken languages such as Spanish and English. The CSIR has successfully developed a set of tools to collect and model speech corpora, and text-to-speech and automatic speech recognition engines to cater for South Africa’s 11 official languages.”
Servicing people in their own languages
The HLT group has had numerous successes in developing applications for Lwazi. “The first of these was a speech-driven telephony system to facilitate access to service delivery for South Africans,” Taljaard explains. “We also piloted an automated health helpline at the Botswana-Baylor Children's Clinical Centre of Excellence for caregivers of HIV-positive infants and children.
“At present, we are rolling out pilots across nine provinces to test an application for the National School Nutrition Programme (NSNP). This allows learners to give feedback in local languages via a mobile device and free of charge, on the quality, quantity and frequency of meals at their schools, which are funded by the Department of Basic Education via the NSNP. The aim is for the Department of Basic Education to monitor and assess the practical implementation and workings of the NSNP.”
Another recent success was the input by the HLT group to Google’s voice search service. Google has pioneered a technology that allows people to access the web using the spoken word via a cell phone, in the same way that one would type in a query. It was aimed at enabling a mobile voice search feature in South African English, Afrikaans and isiZulu, given the growing need for tools to overcome language barriers online. “In conjunction with North-West University, the HLT team assisted Google with the speech data collection effort and pronunciation dictionary development for these three languages,” Taljaard reveals. “Persons with disabilities are also able to use the Google voice search for ease of access to information.”
Subsequent to the Google voice search project, the team developed a mobile phone data collection tool that forms part of the language technology development toolset that is being made available for Africa. More than 1.5 million speech samples have already been collected in six provinces for South Africa’s 11 official languages – thereby contributing to the language technology research and development for these languages.
The CSIR is a partner in the European Union Framework 7 Programme’s VOICES (VOice-based Community-cEntric mobile Services for social development) project. Through VOICES, consortium researchers plan to assist rural farmers in Mali to benefit by access to content in their own language through mobile devices. “Our CSIR researchers have been able to offer their expertise in producing language tools for under-resourced languages and in mobile platforms to this multi-partner project,” Taljaard confirms.
Taljaard is proud of his team’s cutting-edge technologies and toolsets, and has defined research and development themes for the group, “Our focus is language technologies for mobile platforms; call centre optimisation for local languages; speech technologies for children and persons with disabilities; language learning; safety and security applications. We will also be doing research on service delivery optimisation, monitoring and evaluation, and information access.”
Localisation of technology and content is important. “For example, if we localise tablets for children in their own language, we help them develop language and digital literacy skills to equip them for a digital world,” Taljaard contends.
Taljaard is determined to ensure that the “good work of the HLT group continues to provide world-class solutions to challenges in South Africa and Africa. We must ensure that our solutions are both innovative and relevant.”
*NOTE: Taljaard also oversees the activities of the CSIR’s knowledge technologies research groups: the joint CSIR/University of KwaZulu-Natal’s Centre for Artificial Intelligence Research led by director Prof Tommie Meyer and the Enterprise Knowledge Engineering and Management research group led by Prof Paula Kotzé.
News contributed by Biffy Van Rooyen, CSIR Strategic Communication and Stakeholder Relations