Within both technologically developed and developing nations, physical, social and economic structures enforce differential access to information. Whether in urban or rural communities, access to communication technology such as the Internet can be highly limited. Further, automated translation tools continue to struggle with low-resource language pairs and morphologically rich languages—those with lots of conjugations and declensions, like German and Arabic. Certain nations are therefore less integrated into global markets and cultural exchange. Particular limitations in access arise from the confluence of language barrier and Internet infrastructure. These block access to published research and general knowledge.
Language barriers aren’t going away. English maintains a hegemony over both the Internet and scientific writing, and regional languages remain firmly entrenched. Since attempts at “universal languages” like Esperanto and Interlingua have fallen flat, the next plausible method for facilitating seamless cross-language communication is a technological version of the Babel fish from The Hitchhiker’s Guide: instant text and voice translation by computer.
Machine translation (MT) uses machine learning, relying on oodles of parallel text (“bitext”). When there is no economic, legal, or national security incentive to produce massive parallel corpora (like how the EU publishes in all member languages), translation quality suffers. Speakers of languages without these incentives are therefore disadvantaged. As an example, Google and Microsoft’s new, “scarily accurate” neural-network translation models are each only available for translating between eleven languages of the thousands recognized worldwide.
Internet access is also variable, both across and within national lines. Still, with cell phones achieving 95% global penetration according to SSI, the binary question of access is outmoded. The new glaring question is bandwidth: Certain access methods limit available content. Connecting through a 56 kbit/s dialup connection doesn’t allow for streaming video; however, it represents more than 2.2 million American subscribers.
The social consequence of this diminished information access is a poverty-of-stimulus stimulation, where those without access do not have the ability to rise socially. The futurist and Google natural language expert Ray Kurzweil has predicted an exponential, accelerating-returns increase in information and productivity resulting from advances in technology—leading us to the Singularity, when man transcends biology. As the saying goes, Einstein once said that the most powerful force in the universe is compound interest (an example of an exponential process), so those who are deprived of these technological gains are kept out of the global economy and fall behind, seemingly irreparably.
Put the doom and gloom behind us, and we start looking for solutions. While we can’t guarantee an egalitarian solution to bring everyone to the same level, we can at least tighten the inequality—lowering the Gini index for translation quality. The first solution comes as a book that has been translated into virtually every language, so it provides a lengthy bitext to train on. Plus, with over twenty English editions, its impact is magnified. If you haven’t already sussed it out, this book is the Bible. Missionaries translated it into every tongue they could, and it’s one of our best sources of parallel text for languages without enough web presence for Google to crawl through them. Social media is another tool in global language barriers: it creates network effects that let information transition through polyglots and overcome geographic ties. These are the keys to holding everyone in the churn of technology as we approach the Singularity: full augmentation of our intelligence with technology, a digital neocortex. Because since before our species even walked this planet, that’s what technology has been for: breaking down the barriers of what seems possible.
This article was written by Arya McCarthy. Click here to see more of Arya’s work.