Ecclesiastes tells us “For everything there is a season, and a time for every matter under heaven” (3:1 ESV). We celebrated 23 years in Wycliffe Bible Translators this year, and we are feeling it’s time to make some changes. Much of what follows could easily be seen as the next logical steps for us, given where we have been, and the needs facing the global Bible translation movement. But you may disagree, and if you do, we would be honored to hear that, and work through it with you as well as possible.
In short, we’re looking at working from Texas for our next assignment, rather than from Cameroon, and we’re looking at shifting my work towards computer tool development to serve more broadly the global need for writing systems. These two points are connected (the team I would work on tools with is based in Texas), but I’ll address them here one at a time here.
Africa
Over 20 years ago, we were in a missions conference with a church that told us they were looking for a missionary going to Africa, since they didn’t have anyone in their mission portfolio currently stationed there. They had missionaries in other parts of the world. They supported Christian workers who worked from the US, for various reasons. So the idea of a young couple going out to actually work in Africa was appealing.
We have been proud to work in Africa, to rub shoulders with Africans on a daily basis. We talk about incarnational ministry; leading a Bible study in Romans was one of the most fulfilling parts of this last term —but it was just life, not my job. Job or not, it was a privilege to build both good Bible study skills and solid theological material into the leaders of our local church.
So it is more than a bit sad to think that we should work from the US for the next couple years, and we understand if you agree. Anyway, we want to think through what this does and doesn’t mean, and get your thoughts. For instance, working from the states would mean being more locally available for James (as he looks for a post-university job) and Joel (in his second year at LeTourneau University) and Anna (doing her senior year in the same school where she did her junior year, then going to college!). But it would not mean cutting ourselves off from Africa; one of my mandates for the next couple years is travelling to test A−Z+T in languages where it hasn’t been used before.
In addition to the need to step away from the daily grind of cross-cultural work, I’m getting more and more requests to respond to needs outside of Africa. For instance, A−Z+T mostly deals with roman scripts, like the consonants and vowels we use in English. Frankly, this isn’t because I don’t care about non-roman scripts, I’ve just never run across someone working in Africa asking for help with them. But while working up a proposal for ETEN, I found that non-roman script support is one of their criteria for tools, as tools they support should be useful everywhere in the world. So I will likely work on that soon, even though it doesn’t necessarily add to the work in Africa much, because it will make the tool more useful globally.
Linguistics Consultancy
For some time, I have described my job as helping people to develop writing systems in their languages, including training and mentoring others to do this. Also for some time, I have seen the immensity of the task (e.g., one linguist for 60+ language writing systems) as requiring smarter working with new tools. So my work as always including trying out new tools that I thought might help us, corresponding with their developers and submitting bug reports.
As a linguist corresponding with software developers, though, I have often felt like I’m straddling two worlds. The people building the tools aren’t the ones actually using them. This isn’t necessarily a problem for established tools with established user bases, where developers can get good, broad feedback. But I have been asking for new tools, to do new things in new ways, which means I spent a lot of time explaining myself, and I had a hard time getting my ideas prioritized by (admittedly short-staffed) development teams.
Then COVID came, and while I still had the requirements of my conscience and calling, suddenly I couldn’t leave my house to work. So I tried and eventually succeeded building the tool I needed myself, which today we know as A−Z+T.
Much of my software development felt like poking around in the dark, working from exigence and internet searching more than experience or training. But from my first testing I could see that the tool was already helpful, so continuing to invest made sense. Many of you have seen the tool demoed since, and I think most have agreed. If you haven’t seen it yet and want to, please just ask —we can help you find how many vowels you have in your language (hint: English has more than five)!
When we came back on furlough this last year, and I thought through what was next for A−Z+T, I realized that there were possibilities for applying the databases we create in new ways. One of these is the development of a language specific Automatic Speech Recognition tool — like your phone speech to text, but it would work in a minority language, and use the correct writing system.
But now I have an opportunity to work on a team with a couple other software developers, with the support (and accountability!) of people with more training and experience in this work. this would provide value in both directions: I would have the support and accountability of people who understand computer science, and they would have the support and accountability of a doctoral linguist with extensive field experience.
Since drafting the above, I had a long meeting with the man who will likely be my next supervisor. We talked about Audio Project Manager, the tool headlining the team, and how it would interact with A−Z+T, and how they would work together to integrate value from other development groups.
It seems likely (though this remains to be tested) that data resulting from a normal run of A−Z+T (where a community says “Help us figure out our consonants and vowels, please”) could be used to train an Automated Speech Recognition (ASR) tool for that language. This would provide, for each language that does this work in A−Z+T:
- An ability to speak to the computer where implemented, rather than type
- A path to transcribing Oral Bible Translation (OBT) audio files for import into text based tools. Audio Project Manager currently allows for manual transcription, but this can take a long time, especially compared to us providing an automated draft to correct.
- A path to collecting language texts much more efficiently. Anyone who has collected natural speech texts in the field knows that a three minute story can easily take a couple hours to transcribe manually. This is one reason the Basic Oral Language Development (BOLD) method was developed —to help people collect stories with their translations, without depending on transcription to any writing system. If we could add to the BOLD method the ability to subsequently transcribe efficiently (both for the language and the translations), then we would have an efficient way to collect lots of texts.
- With a large body of natural languages texts, we would have the possibility of training an AI engine that could produce natural language text, and potentially provide initial drafts of scripture in that language very quickly. These drafts would of course need to be checked and corrected, but this would be a major advantage to translators!
Conclusion
The more I think through this, the more I think that the move toward tool development in the US is the right thing for my family and the global Bible Translation movement, but also a logical progression for my work, moving from working in one language, to planning for 60, to thinking of all of central Africa, now to organizing for the next (or rather last?) 1,000 writing systems —and the Bible translations they will support. I have largely found agreement on this with Wycliffe and SIL administrative hierarchies, potential future colleagues, and our ministry partners (as we’ve been able to talk to them so far). But this conversation remains in process; what do you think?
This work by Kent is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International
0 Comments