A Survey of Localisation in African Languages, and its Prospects: A Background Document

5. Technical Context

As part of the effort to better understand the interface of language and ICT in African societies, this section will review the technical context encountered by localisation efforts. It deals with two areas:

  1. A nexus of issues relating to connectivity, infrastructure, computer hardware, and operating systems in Africa
  2. International trends in two interrelated concerns – FOSS and localisation – and their impact in Africa.

These two areas might merit treatment as separate sections but for fundamental connections in (a) the key concern of "access," and (b) their relationships (actual or potential) to ICT for development programs. In effect, all efforts active in ICT in Africa, whether internal, international, or both, have an ultimate (or at least ostensible) aim of increasing access to the technology. Access, however, is a subject with broad implications.

ICT4D programs have focused in one way or another on access, and as such are often implicated in bringing hardware to Africa, increasing connectivity, and, less consistently, in FOSS and localisation. ICT4D programs and projects then are important contributors to shaping the technical environment for localisation.

With this context, this section will begin with a brief discussion of access and the factors it involves. It will then treat infrastructure issues, hardware and operating systems, and connectivity, followed by international trends in FOSS and localisation, and their bearing on Africa. Mention of ICT4D initiatives will be made as appropriate throughout.

5.1 Access: Physical and soft

Access to ICT is a fundamental defining factor of the digital divide,40 and hence the focus of a range of activities relating to the technology in Africa. It is generally understood as involving more than mere proximity to and permission to use computers (or other devices) that are connected to the internet (or other network). Various sources have sought to elaborate levels or types of access. For instance, Telecommons (2000), in an early evaluation of the potentials for ICT for rural development discusses "'physical access' to ICT infrastructure and applications, and 'soft access,' which [they] define as software and applications which are designed to enable rural African users to utilise ICTs for their own needs and uses once the physical access has been established." In a broader context, the organisation Bridges.org went further to define twelve dimensions of what it called "real access"41 to the technology, of which "relevant content" specifically mentions language. In effect several points emerge in discussing access and localisation:

  • Foundations for of access – hard realities
    • Availability of functioning computers, power to make them run, and connections to link them are the starting point for discussion of access
    • Costs of establishing and maintaining access, which are generally beyond local means and hence often involve outside support or even initiative
    • Permissions to use the devices, whether by fees or other means, often represents a significant cost relative to potential users’ resources (a potential barrier to physical access)
  • Access and localisation – how providing access meets the user
    • Two aspects of access – software interface and interactive software on the internet – are ones in which choice of language is important,
    • We can adopt Telecommons' term of "soft access" to refer to how well these anticipate user needs
    • Localisation is the major part of the process of assuring soft access.
  • User skills – how the user meets physical and soft access
    • User profiles in terms of, among other things, language and literacy
    • Implies attention to developing user skills, including basic literacy (see above, section 4.4)

Figuratively speaking, access builds towards the user from one end, from hardware and connections to interfaces understandable by potential users (with other factors involved also). User skills, in effect, build from the other end, involving education, training and so on. There is, then, arguably a trade-off or complementarity between soft access and user skills. The more skilled or experienced users are less needy in terms of soft access, while less skilled or experienced users require more attention, including localised interfaces.42

Localisation for enhanced soft access, and the enhancing of users’ skills, therefore emerge as two complementary and essential elements to extending physical access to ICT into effective or "real" acces (which in turn imply connections to other concerns, per localisation ecology). However, it all begins with physical access.

5.2 Infrastructure

In discussing efforts to expand ICT in Africa, whether localised or not, there is often reference to other basic technical and infrastructure indicators such as number of telephone lines and level of electrification.43 In effect, the realities of poor communications infrastructure and unavailability of (reliable) power sources limit even well-funded efforts to establish access to computers and the internet. These basic factors are changing slowly, while other solutions such as alternative power sources (primarily solar) and the dramatic growth in numbers of cellphones are altering the equation in some ways.

For purposes of this study, however, these important infrastructural factors will be considered as givens in order to give attention to other technical variables that affect actual and potential localisation in Africa.

5.3 Computer hardware and operating systems

The most basic measure of ICT penetration and of the "digital divide" is the availability or not of devices in working order – primarily computers, but to an increasing degree also portable and handheld devices – that can process, store, and transfer information. In their absence of course, any discussions of connectivity, access and localisation are moot. One reflection of this fact are efforts such as the one to develop the Simputer in India44 and the recent OLPC project effort to develop and supply inexpensive laptop computers to schoolchildren in poor countries.45

By any reckoning, numbers of computers in Africa are low by comparison to other world regions, and very often those machines are of older make and operating system. There are obvious economic reasons for this state of affairs. Even the efforts by outside agencies to address their perception of the this hardware side of the digital divide in Africa by supplying new or used computers have limited effect (however well designed or funded such projects might be, they are by nature addressing limited objectives within much larger contexts).

The existence in Africa of many older computers and systems also has implications for the potential use of various kinds of software and multilingual web content. In many cases, older operating systems cannot run newer software, use Unicode fonts, or take full advantage of internet connections. In time these systems will be retired, but given the persistence of the root causes for resort to used computers, it would seem likely that Africa will continue to have a high percentage of computers in use that cannot handle the latest operating systems and software. In other words, there will still be in use computers and systems that cannot take advantage of the most recent advances in internationalisation and localisation, in effect always being a step or two behind the latest technology.

This discussion has dealt mainly with computers, though the rapid spread of mobile telephony in Africa and the increasing capacity of handhelds should be understood together as a harbinger of significant changes in the way we plan for ICT use on the continent, including localisation. The potential for multilingual SMS is already being explored. Longer-term potential for more powerful handheld devices is discussed below (section 9).

5.4 Connectivity and internet policy

Another measure of the digital divide in Africa is the level of connectivity – the presence and quality of internet connections.

Basic connectivity on the country level was a major focus for ICT in Africa in the 1990s. Since the success of projects like the Leland Initiative (funded by USAID) and the Internet Initiative for Africa (funded by UNDP),46 and the introduction of other infrastructures like those for cell phones and VSATs, attention has turned to extending connectivity from the capital cities out to other regions of African countries.

This evolution has been accompanied, and to varying degrees guided, by discussions and elaboration of policies governing the use of expanded national bandwidth. The Leland project itself involved policy prescriptions as much as it did infrastructure assistance, with the aim being to foster a sustainable organisational configuration to manage optimal use of the bandwidth. This was guided by market philosophy requiring the national level beneficiary of increased bandwidth to resell that to privately held ISPs.

More broadly, the UN Economic Commission for Africa (UNECA), through its African Information Society Initiative (AISI) encouraged countries to develop national information and communications infrastructure (NICI) plans to help them determine how connectivity was to be expanded. Altogether, Internet connectivity and policies to guide its use have been closely interrelated.

In technical terms, bandwidth in Africa continues to increase (IDRC 2005) and the number of internet connections is growing a rapid rate (USINFO 2006). This has been accomplished largely by the use of satellite uplinks in each country, however the deployment of undersea cables around the coast of the continent is also becoming a factor: In West Africa, the Sat-3/ WASC cable and in Southern and East Africa the FAST cable.47

However, despite efforts to increase connectivity to all African countries, the actual levels of connectivity between countries and within each country tends to vary significantly.

Some of the structural issues mentioned above affect the potential for expansion of access. In particular, connecting rural centres is a challenge. Some countries have implemented a phone tariff system where all parts of the country can dial-up access at the same rates, which removes one disadvantage, but still connection quality and actual accessibility may not be good.

It is in connectivity that some of the clearest connections can be seen between policy and technology (per localisation ecology and the PLETES model).

This is important to consider since the patterns of language and connectivity/access may indicate priority areas for localisation, and the issue of access links to the motivation for undertaking localisation.

5.5 Trends in localisation

As discussed above (section 2.2), localisation as an interest began probably the moment that people started to use computers to store and manipulate human language. With the advent of personal computers and their spread throughout the world, and also with the growth of the internet, various challenges of localising for specific languages – and indeed of internationalising ICT to facilitate that - have arisen and been dealt with (see below, section 6).

Today, localisation is the focus or a concern of a wide range of commercial and voluntary activities. Localisation of internet content is particularly prominent for major international languages, although apparently not as developed for African languages (see below, 7.1).

Localisation of computer software has various aspects and some complexities. First of all, one might suggest that in this area we are really speaking about four different things:

  1. Operating systems (Windows, Unix, Linux)
  2. Software application or package completely translated (such as OpenOffice, Microsoft Office, or specialised programs such as web browsers)
  3. Partial package, in which only the most frequently encountered commands are translated (this is the strategy of the Microsoft Language Interface Project [LIP])
  4. Production tools for the target language without commands or help files translated (i.e., only language settings, input methods [keyboard layout], spelling dictionaries) (this may be seen as an extension of the category "Equipping Systems" discussed above, 2.2)

Numbers 2-4 represent different levels of localisation of commonly used software. Most of the focus now with regard to African languages is on these levels. While operating systems (#1) are translated into major international languages, relatively little is being done for African languages.

Although Microsoft has increased the number of languages in which it offers its software, and has announced projects for others, there is still a lot more localisation in FOSS. The OpenOffice suite is localised or being localised into about 90 languages currently.48

Other proprietary general office software such as Corel? WordPerfect? has focused on major languages with perhaps production tools in less widely used ones.

The dynamics of evolution of increased software localisation has thusfar been driven mainly by improved internationalisation (mentioned above) and increased demand in major markets. With regard to text, progressive improvements in the ability to render text in diverse writing systems, notably the single standard for all scripts – Unicode – facilitates multilingual production and use of content (Unicode and scripts are discussed further below, 6.2). Also, as the linguistic background of users diversifies (as ICT becomes cheaper and more readily available in more places and forms), the demand for diverse language interfaces is manifested. Along with the latter, the potential pool of contributors to fill that demand for localisation (including where necessary developing terminology) expands.

Localisation has also been undertaken by various companies and projects internationally for their specific uses. On the other hand, localisation has not generally figured prominently in ICT4D projects in Africa with some exceptions.

5.6 Trends in FOSS

FOSS deserves specific attention in discussing the technical context of localisation in Africa due to its demonstrated potential for localisation in diverse languages. This in turn is due to cost advantages and the accessibility of its code.

FOSS includes a range of applications from servers to common "office"-type software (word-processors, spreadsheets, etc.) to more specialised programs like geographic information systems (GIS). But FOSS is as much a movement49 as an approach to developing and marketing software. The FOSS movement has spread in Africa and the Arabic-speaking world much as it has elsewhere, although the extent of use is less, which would reflect the smaller overall user communities. Nevertheless, there exist in many countries of the continent, FOSS user groups.

With a large, growing and diversifying base of enthusiasts internationally it presents an unusual range of possible support as well as some challenges for tapping it. (One hope of this PAL project is to facilitate communication and coordination of efforts among FOSS localisers in and interested in Africa.)

FOSS and some support from major corporate entities (perhaps ironically) such as Sun Microsystems and IBM. The initiative to develop the popular Debian Linux-based system, "Ubuntu," was funded by the South African entrepreneur, Mark Shuttleworth.

All that said, and despite the potential for localisation, FOSS in Africa, as in much of the global South, is most often conceived of as a way to reduce costs and dependence on proprietary software (notably that of Microsoft). The tie-ins of FOSS with localisation in Africa, however, so far seem to have been somewhat slow to develop (with a few notable exceptions, such as Translate.org.za in South Africa). It seems that, at least initially, various country-level FOSS associations mentioned above are focused on promoting use of software such as OpenOffice in the official languages. Regional groupings also reflect this.

So, while FOSS and localisation seem to be meeting to an ever greater degree on the international level, one question then is how to involve the FOSS communities in Africa to a greater degree in the localisation processes in their respective countries and regions.

Tools to facilitate localisation of FOSS

On a technical level, the potential for localisation of any software, especially for smaller language communities, is limited by the number of people with the requisite knowledge of programming and the languages in question. One tactic for facilitating localisation, then is to provide tools designed to make it easier for potential localisers who may have the motivation and the language skills, but not the technical background to translate software.

An example is that Dwayne Bailey of Translate.org.za and Javier Sola of Khmeros have developed an interface called "Pootle" to facilitate localisation of OpenOffice by people without high levels of technical expertise.50 The object in effect is to "lower the bar" of entry into localisation such that people with language expertise but not much technical or programming background can undertake localisation projects.

This approach has also been undertaken in other contexts such as the compiling of locale data in an online tool. It will be interesting to monitor the effect of opening aspects of localisation to people with language skills and FOSS enthusiasm, but without high computer specialisation.

< 4. Linguistic Context | Survey Document | 6. Africa & the I18n of ICT >