Briefing notes on South Africa’s Information Society ICT Programmes

Human Language Technology in South Africa

In South Africa research is being undertaken on applications related to Human Language Technology (including multi-lingual systems) to facilitate a more inclusive information society. Various South African academics and institutions are engaged in research programmes that investigate how Human Language Technology(HLT) can be adapted and applied for the benefit of our developing multilingual and multicultural country.

In South Africa this goal is considered from 2 perspectives:

HLT as an enabling technology that can play a crucial role in addressing the need for information empowerment.

HLT as a support for language diversity in an affordable fashion. HLT will assist industry and government where services and documents must be made available in at least the 11 official languages. HLT will play a vital role in rectifying the historical discrimination against specific languages.

Examples of current projects with this goal include

HLT – enabled e-government service delivery

This project tests assumptions related to the provision of access to government services through the use of the extensive telephony infrastructure in South Africa. Through prototype development and a series of contrastive experiments, focus is placed on the interplay between system usabiliy, speech and language technology options and issues of culture and user exposure to technology .

The Local Language \speech Technology Initiative

This is a collaborative effort involving groups from India, the UK, Nigeria and South Africa. The aim is to develop an open source toolset for the effective development of a text-to-speech (TTS) system in any language of the world.

An affordable assistive communications device

Basic hardware components and an innovative system design are used to create an affordable hardware device for people with speech disabilities. The challenge is to minimize the cost, and to ensure flexibility of language and content.

The CSIR / University of Pretoria HLT group is involved in basic research, technology development and applications development aimed at providing South Africans with enhanced information access using Human Language Technologies.

Projects that are currently in progress include:

A general approach to the efficient development of pronunciation dictionaries in languages that use an alphabetic orthography;

The refinement and application of an open-source system for speech synthesis, to support South African languages such as isiZulu and Afrikaans;

The development of speech-recognition systems for several South African languages, including isiZulu, Sepedi and Afrikaans

Development and application of the "OpenPhone" platform, which will make it possible to deliver information in a cost-efficient manner over the telephone; the platform will support developers who are not experts in computer telephony, and encourage the development of culturally and socially appropriate interfaces.

A team of lecturers and post-graduates students at UNISA are also involved in research involving HLT. They are conducting research in the following projects

Computational morphology and its applications, and the development of machine-readable (XML) lexicons for Zulu and Xhosa.

Approaches to computational syntactic/semantic analysis.

Computational morphological analysis of Xhosa.

The Department of Arts and Culture has an initiative to develop a National Language Facility headed by Dr Mbulelo Jokweni.

The University of Stellenbosch will establish a Centre for Language and Speech Processing headed by Professor Justus Roux.

DST has recently agreed to fund a project, led by Professor Justus Roux. This project focuses on the development of a multilingual speech-based application in the domain of financial services within the government sector. A significant portion of this funding goes towards a Human Capital Development programme with specific criteria that fosters the development of HLT post graduate researchers from previously marginalized groups.

The University of Limpopo have engaged in projects in HLT. Recently they have formed a relationship with DST and CSIR who have indicated their support for these projects. An exciting new development is the coming together of different teams that are engaged in HLT. Tertiary institutions, research councils, DST and DAC have now started to collaborate with each other and bring about a unified South African effort in HLT R&D. The ICT Roadmap process is emerging as a force that is unifying a number of HLT R&D initiatives in South Africa. DST and DAC are the two government departments that have agreed to engage each other to take forward HLT.

It is envisaged that there will be a national collaborative event that brings together researchers and students in HLT and related fields on a regular basis – both to collaborate on research and to give students the opportunity to learn about advanced subjects. This event has the potential to develop into an international event that attracts a worldwide HLT fraternity.