Thursday, October 08, 2015

Access gap in the "Connectivity Declaration"?

History sometimes moves in spirals - after years have passed and many things change, you suddenly find yourself back in a same place, but maybe on another level. I had that sense of déjà vu in Bamako in 1999, discovering that projects were still hacking mutually incompatible fonts to be able to display characters used in Malian languages on computers, much like what Prof. David Dwyer and I had to do in 1989 in order to print the first draft of the Fulfulde lexicon. The intervening decade had of course seen some remarkable advances in information technology including the unfolding of the world wide web, but there was no perceptible change in terms of ability to use African (and many other) languages on computers.

I'm getting a similar sense of already-been-there today, reading about Facebook's plan to beam the internet to remote parts of Africa and related news about the benefits of universal access to the internet including the "Connectivity Declaration."

"Access" means more than one thing

The current focus on universal internet access has me thinking back to that same period in Bamako mentioned above, when there was a great deal of enthusiasm on international development-focused email lists about how information technology and the WWW were going to transform, among other things, rural development in Africa. But while working in Mali it was clear that even if by some miracle one could get the technology to farmers, it still wouldn't speak their languages, nor even be able to properly display text in those languages, due to the font issue. (This observation was one of several leading to the Bisharat initiative, which is another story.)

From this point of view it was clear that "access" had more than one aspect, even though most discussions treated it as one issue. It was one thing to have a computer, and another to power and connect it, and another to pay for all that, and then ... with physical access secured, for a farmer or extension agent in, say, rural Mali, to actually make full (or any) use of it. The latter can be called "soft access," a term first used in contrast to "physical access" by an organization called TeleCommons (quoted here), which covers localization of content and interfaces in languages of the anticipated users. (Dwayne Bailey's reference to "last inch limitations" is another way of thinking about lack of soft access.)

Fast forward to the present, where the Connectivity Declaration mentions "access" five times and "accessible" once, but again as one undifferentiated problem to be solved and good to be achieved. And although one might interpret the intent more broadly, the focus does seem to again be technical. The Declaration certainly does not hint at the soft access issues inherent in diverse peoples taking advantage of "the tools and knowledge of the internet."

A focus on internet access as a connectivity issue is inadequate when discussing use by marginalized populations that are diverse in terms of their languages, cultures, and education levels (addressed by soft access), and in terms of other socio-economic factors (that may affect overall access). The initiative to its credit does mention "Content isn’t available in the local language" (in response to "Why aren't more people connected") as a "barrier" it wishes to address, but this is only part of the equation - there are also technical language support and interface localization needs that fit under soft access.

Current issues and potential in soft access

So we're back at more or less the same place with regard to "access," but on another level with some changes:
  • Unicode has largely resolved the font incompatibility issues mentioned above (I'd like to say completely resolved but there are still hiccups in implementation)
  • Localization of software and apps in less-resourced languages is incomplete and sometimes uneven, but it is happening
  • Input methods (hard & virtual keyboards/pads, STT) are still problematic for may extended Latin scripts and non-Latin writing systems, but the main issues now are arguably less technical than policy-based (lack of standards)
  • Human language technology / natural language processing offers the potential to 
    • bridge text ⇄ speech for speakers of less-resourced languages with new forms of soft access as well as content deliver
    • bridge languages by machine translation of content
  • Content localization and creation, which like software localization is incomplete and sometimes uneven, extends the concept to include soft access to knowledge
So we're much better situated to enable soft access today than we were 10-15 years ago, but there is still not the same focus on these issues as on the simpler (which is not to say easy) challenge of technical connectivity. Soft access means more and can be facilitated by more than was previously the case, but it will not come in the same access "box" as internet access without explicit attention. Part of the reason is that "universal internet access" is implemented on a wide scale (national, regional, continental), while soft access in the case of most less-resourced languages is particular to smaller areas and populations.

Soft access needs attention from people and organizations closer to the need - governments, NGOs, companies, and communities on the national and local levels, as well as regional and continental organizations (such as ACALAN). However I would suggest it also needs advocacy from the same people in positions of influence and power - like Mark Zuckerberg, Bill and Melinda Gates, and the other originators of the Connectivity Declaration - who are promoting universal internet access.

Whose "tools and knowledge"?

A blog post by Dr. Mark Graham of the Oxford Internet Institute in response to a tweet by Jimmy Wales (another one of the originators of the Connectivity Declaration) ...
... raises some issues about the impact of corporate-sponsored internet access that are worth considering in this context (these do not represent all of the analysis in the article). What kind of internet service or choice of content, and under what conditions (ads?) would Facebook's, or Google's Free Zone (another free concept) provide? And what might be the impact on local internet markets? Dr. Graham suggests that
In much the same way that food aid as a development strategy harmed local farmers and markets in Africa, “connectivity aid” could similarly destroy the evolution of local content, local innovation and local alternatives.
This is an interesting point, though it bears noting that approaches to food aid have evolved, and in any event, research on its impact has reached diverse conclusions. It is however a legitimate question whether "connectivity aid" would be a simple "Pareto improvement" (some gain, no one else loses) for the societies receiving it.

The impact of free internet on soft access and local(ized) content is another interesting unknown. It may cut either way, with the potential for support for languages down the long tail* if the corporations decide (perhaps in facilitating work of local initiatives?), or for an economic decision on their part to limit focus to languages at the head of the tail (i.e., the ones the most spoken). For some languages at least, this might become somewhat complicated if you take into consideration the technical and content dimensions that must be managed (note those bulleted under "Current issues and potential in soft access," above). For languages that lack standard orthography and/or have high dialectal differences without one dialect being accepted as a standard variety, a free internet operation would probably not be well placed to engage the issues necessary to have - so I imagine there could be a risk of corporate sponsors deciding to limit commitments in this area.

More on "access"

All that said, there's more to the topic of access. The physical/soft distinction is a useful but basic one. In African Languages in a Digital Age I discuss this matter in more depth (although with reference more to computers than mobile devices), including reference to one organization that disaggregated 12 dimensions of "real access."

A factor related to soft access is user skills (which may include basic literacy, computer literacy, experience with other technology), and there is some trade-off between the two, in that greater user skills require less attention to soft access, whereas lower user skills require greater attention to soft access by those providing physical access/internet access.

Soft access is not so much an issue for those of us speaking dominant languages, and who have a lot of computer experience. Another aspect of interaction with devices that normally wouldn't affect access - the quality of user experience - may get more attention in this case.

* The distribution of languages spoken in a society is always asymptotic. See earlier discussions on the long tail of languages.

No comments: