We could start with a small budget in the $10-20K range, spent mostly on market research and promotion. Then we could use statistical measures of the success of that campaign to request the funding of a full-time administrative position and a continuation or scaling up of the advertising.
The same conditions exist in many developing countries -- the rich people have Internet access and English skills, and the poor people don't have either. Those rich people might be moved to write in their native language if we can convince them that they will be doing a great service to the non-English speakers in their country, and that we will be able to distribute the content to at least some of that audience.
Spend over 10000 US$ on advertising Wikipedia to the rich, in the hope they will be moved to write articles for the poor. This might work in India, but in Africa, that seems a total waste of money to me. I might be stating this a bit too harsh, but... First of all, where are you going to advertise? Second, I think many rich (the ones who can afford internet at home) in Africa don't give a frack about African languages, they speak French or English and they think it's ridiculous to educate children in their native language. Third, Africa people are not used to work for free if it's not for someone within their extended family.
I think providing internet to small community centers in poor areas under the condition they add something to Wikipedia - you can set up contests to see if it could work, or to simply directly pay for every article of sufficient size (give out a buck, so you'll have 10.000 articles at the end, if you'll forget about overhead costs...) is much more cost-effective. Once there are a 1000 articles in a Wikipedia it will easily advertise "itself". Sending out mails will be much more effective than when there is nothing...