Saturday, March 6, 2021

Automatic translation and bilingualism

Ymddiheuraf am unrhyw drosedd y gallai'r ddelwedd hon ei hachosi.
Some of us are getting used to seeing automatic translation offered in various platforms. Google Translate has helped with communication in many and varied settings, not least the mobile app version. Facebook and Twitter will offer to translate posts. Microsoft have integrated automatic translation into their Office365 email. Simultaneous automatic translation is available to viewers of Office365 PowerPoint Online presentations. This is easier for the 'Tech Giants' with massive development budgets and a global, multilingual, market. For lesser firms, the impetus, funds and rationale may be less substantial to the point that automatic translation will just have to wait.
And why not? I will come back to this. First I must add that I do think of this post and the matters discussed within it in terms of networked learning but I will leave the reasons for that implicit for now.
In Wales, we have the incentive, backed by statute, policy and funds: the Welsh Language Act, and the Welsh Language Commissioner, are slowly but surely increasing the status and prominence of Cymraeg. For example, drivers knew they were entering predominantly Welsh-speaking areas when the order of languages on signs flipped. But recently, Welsh organisations have been expected to place Cymraeg versions of phrases first, before the English version, in signs and messages.
Whether you agree with the increasing reassertion of Cymraeg over English or not, the direction of travel is clear enough. According to this BBC article, the future is Cymraeg (and this article points out that 'Welsh' is Anglo Saxon for 'outsider', Cymru means 'friends/companions' - which are you?).
If you really want to seed adoption of Cymraeg, you have to encourage use at grass-roots, ordinary-everyday, routine, but also high-stakes situations... At 'the sharp end', things get trickier. In the last couple of years, an ambitious project began to digitise nursing documentation in the whole of GIG Cymru/NHS Wales. At the time, those running the project made the reasonable case for 'English only' for the sake of patient safety and the guarantee of understanding among practitioners using the system. This felt retrograde, from a bilingual perspective.
In the world of education, we are trying to honour the principles and direction of bilingualism by, for example, enabling students to complete assessments in Cymraeg and provide bilingual learning materials wherever possible. Since 2015, I have been involved with a project to digitise documentation and processes associated with midwives and other healthcare students' assessment of their clinical practice. For at least that long, our project partner My Knowledge Map Ltd. have been keen to integrate automatic translation with their MyProgressTM platform, having done similar work previously with the Scottish government, but project ran out of steam. For various personal reasons, I am keen to see Cymraeg re-established (more detail below). I regularly assert the need to consider bilingualism at meetings in work and pressed for this in MyProgress. Most of us are familiar with the option in Websites to flip between different languages. However, this presents them as binary opposites, but the prospect of automatic translation held singular appeal for me. Automatic translation could enable a paradigm shift from binary bilingualism towards the more fuzzy reality of everyday language learning and use. By publishing 'official' translations in two languages, they are pitched one against another. However, everyone in Cymru knows that only a small minority of purists speak the kind of Cymraeg seen on officially translated Websites. With reference to the standardised forms in student clinical portfolios, this Cymraeg is officially translated and we would not want to deviate from that. As with the All-Wales Nursing Documentation project, the problem comes when a form is completed in Cymraeg by a student and needs to be read by someone without Cymraeg. For this reason, there was hardly any take-up of the Cymraeg versions of the paper-based portfolio. The English-only version was already a very large block of A4 paper. Of course, students could make use of Cymraeg in their clinical placements where there were enough Cymraeg-speaking staff. If the student expressed an interest early enough, their placements and assessments could be completed in Cymraeg. At least, assessments could be conducted through spoken Cymraeg with the written records kept, nearly exclusively in English.
If MyProgress could at least flip between two languages it would negate the physical challenge of carrying double the documentation around.
However, to me, automatic translation at the level of an individual form response offered something more generative. Language use is not as binary as a bilingual Website, especially when trying to learn the language. I have dabbled with learning Cymraeg (see below). One of the things which makes that difficult is confidence or the lack thereof. You simply have to practice, get it wrong, make small-steps forwards, to grapple with the words, to reach for words somewhere in your memory or in a dictionary and venture out with them. There's a lot of grey area until you develop a level of confidence that means you can actually call yourself fluent, or at least try to get away with throwing in the English equivalent with a strong Welsh accent (known as 'Wenglish')! I think there are many people who inhabit that grey area, in terms of spoken and written Cymraeg. How will they become more confident?
I think that automatic translation invites people in the grey area to try out their Cymraeg whenever they want to, safe in the knowledge that people who only have English will be able to flip to that language and get the sense without any reason to castigate the student for the inconvenience of having to get someone to translate it for them. Furthermore, the system genuinely puts both languages on an equal footing which, of itself, sends a powerful signal upholding bilingual principles.
We have discussed the issue of how good the translation will be, but the kind of responses students might be expected to provide within these forms would not require pure English, so why would we insist on 'pure' Cymraeg? The primary goal is adequate sense-making. Of course, we have to explore this issue further.
Occasionally, the Welsh government calls for bids for innovation grants and, in January 2020, we were given the money to add bilingual automatic translation to MyProgress. While I did battle with bureaucracy, development at My Knowledge Maps went into overdrive, and around this time last year I was testing the nascent system. As we entered the first lock-down, I relied on my bilingual kids to help me test it and eventually I was happy to sign it off. Since then, our efforts have been somewhat diverted by gearing up to roll-out MyProgress to our nursing students. They would be using the newly developed ePAD system, which had not been factored into the Welsh Government funded grant. In spite of this, the bilingual feature is on the brink of being available for piloting between ourselves and colleagues in the School of Medicine. The nurses and medics are important groups for this project because of the sheer numbers of students increases the liklihood that some among them will wish to use Cymraeg. This greater use is hoped to provide a more robust and rounded evaluation of the innovation.
We want to know, when you embed that automatic translation functionality inside assessment documentation, what difference does that make to the uptake of Cymraeg? With a digital system, we can measure sheer uptake over time in numbers. But other questions arise, such as, how does it make learners feel about using their Cymraeg in formal settings? Most language learning targets everyday exchanges about the weather or where we hail from or what we did on the weekend. Without wanting to trivialise this type of use, can automatic translation of written Cymraeg push bilingualism into more formal, previously hard-to-reach areas? These are fascinating questions for me, and yet I struggle to find academics whose job it is to answer them. A friend recently suggested the field of applied linguistics...
Footnote: The last time I tried to learn Welsh was through an wlpan course in 2008 when my son became ill and I was trying to finish a Masters - not brilliant timing, even though all four kids were in first-language Cymraeg schools. Prior to that, the most ideal time was when I was growing up and spending long hours with babysitting grandparents in Pitman Street, Canton, (embedded streetview below) while my mum worked as a school clerk. It is one of my enduring life regrets that instead of learning Cymraeg at that age I insisted that we only spoke English. No doubt my grandparents were used to being cowed into going along with the perceptions of the inevitable terminal decline and inferiority of Cymraeg so they did not persist with me and meekly fell into line. How ignorant I was, hearing Cymraeg spoken on the phone, but never making any progress in understanding what was being said.

No comments: