While the world swings wildly from celebrating the achievements of AI to worrying about it taking over, it is mind-boggling that some of us are still unable to use our real names to fill in online forms. The reason? Diacritics – the marks we see above or sometimes next to letters – or, more specifically, the shift to online forms read by automated machines that can’t understand them.
When’s the last time you filled in an online form? There’s a lot around, from a subscription to a newsletter, to heavier business and accreditation forms. We’re now used to doing this online and having the results machine read and processed. But these systems are still too old fashioned to understand some vital information and the standards need to be raised.
They might seem to be a small thing, well, they are physically small, but diacritics are an important part of names and identities. If you’re not familiar, these are the marks you see above and sometimes next to letters of the alphabet that denote tone, and sometimes meaning. I have one on my surname – Molnár, but I can’t write my name correctly in most online forms. Not with the acute accent over the letter a.
Accents are a reflection of a different alphabet, one that is not accepted in UK online forms, even though the UK would like to consider itself to be an international and savvy country, Brexit or otherwise.
Diacritics appear in a number of languages including French, Polish, Portuguese, Arabic, Swedish, Czech, Italian, and German. That’s a lot of the world that could be better included in online activity. But you don’t even have to look outside of the UK to find languages that feature diacritics, as Welsh, Irish Gaelic, Scottish Gaelic, and Cornish use them widely.
This wouldn’t be a problem if forms were still written on paper. When this was still happening, it’s more likely they were being read by a human and not a machine. By using machine reading we’re homogenising language and alphabets, wiping out characteristics that are still important.
I worry about this when it comes to things like accreditation and certificates. If I earn a diploma or a certificate here in the UK, it’s likely I’ll have those markers dropped, but will it still be valid if I go to work in other countries where accented alphabets are more commonly used? If I earn UK citizenship, my surname will be stripped of accents, effectively changing my name. This is due to ‘IT considerations’, which I personally don’t find an acceptable reason in a culturally diverse country like the UK.
It might be more efficient to use machine reading and this is being used more and more often for things like forms but also CVs and qualifications. What if your name doesn’t match exactly? Because we have cut out the common sense of a human, we are running into problems.
There are solutions that can be explored to make the use of diacritics easier – hotkeys are the usual user method of trying to correct the problem, but when some characters are not accepted in online forms, this doesn’t help. Unicode is the most used current standard and, natively, it supports diacritics. This however falls short on the human factor as it requires active adoption and testing by the developers to ensure it does what it’s intended to do.
Another current approach is the use of T5 transformers – text to text formats for natural language processing. When it is natural to use diacritics in your language, this is an important addition and has been used successfully in 13 languages with a 94% accuracy rate. So this is not an impossible challenge, it’s one that is overdue and should be easily applied in a multicultural world.
While the world loves to experiment with the latest cutting edge AI experiments online, and witness the latest moves in the space race, we’re failing at what might seem like little things but should be solved problems by now. It’s one thing going to the moon, but less impressive when you can’t write your own name. When clever minds put their thoughts together, there’s hope that we can change these simple things and remove the barriers that help us celebrate our cultural – and text-based – differences.
