It's 2024, and most software still can't handle non-English names

If there’s one experience that binds together many bearers of non-English names such as myself, it is the dreaded Starbucks interaction – that moment in the transaction where, when asked for my name to complete the order, I’ll hesitate before mumbling, shamefaced: “Err, John.” In my defence, I like to think I’m making life easier for those on both sides of the counter: the barista doesn’t have to grapple with spelling a Celtic name they’re probably unfamiliar with, and I get my americano that bit quicker. But still, it feels cowardly and inauthentic. It is perhaps a consequence of the English language’s dominance in many parts of the world , that non-Anglo names – and, by extension, identities – are too often regarded as troublesome deviations from the default cultural setting. When it comes to digital technology – something that’s now about as ubiquitous as a daily coffee – the othering of non-Anglo names is more frequent, more annoying and more consequential. Last month, a campaign group in the UK called “I am not a typo”, erected a large yellow billboard in central London that highlighted how most devices’ dictionaries had no problem with an Emma, Nigel or William but regarded Haniya, Ruairidh and Zarah as errors. According to I am not a typo, 41 per cent of different names given to babies in England and Wales are seen as “incorrect” when viewed by Microsoft’s UK English dictionary. In an open letter to several tech giants calling on them to update their products, the group rightly says that our names “are the most important words in our lives – part of our identity”. “But,” it adds, “we have noticed that while all names are created equal, the technology that shapes our world does not treat them like that. Some are more equal than others.” But is your iPhone or tablet failing to recognise your name anything more than an irritating slight? Some observers, such as Rashmi Dyal-Chand, a law professor at Northeastern University in Boston, make a strong case that this is more serious than merely being a nuisance. In a 2021 article called Autocorrecting for Whiteness, Prof Dyal-Chand says: “Autocorrect’s changes to names such as these are not just trivial product glitches.” “In a world rife with the multiplying effects of algorithmic bias in increasingly essential domains of decision-making,” she states, “autocorrect produces social and cultural harms that disproportionately affect communities of colour and those who do not have Anglo identities.” The campaigners go on to say that these unrecognised names are “disproportionately African and Asian in origin. Some are Eastern European. Some are Scottish, Welsh or Irish. This doesn’t reflect a diverse, inclusive society ”. In a bid to encourage tech designers and manufacturers to rethink their dictionaries and devices, I am not a typo helpfully offers them a spreadsheet of first names recorded by the UK’s Office of National Statistics – a list that is refreshed every year. Among this harm Prof Dyal-Chand lists “basic economic losses” in which the “efficiency, convenience and the enhancement of free and spontaneous communication” offered by modern devices is stymied by the user having to repeatedly correct their name. We pay the same money for these devices and services, but receive a lesser product. “Tort and contract claims for unfair and deceptive trade practices, breach of contract and breach of warranty, as well as discrimination claims, would be perfectly viable if a certain class of consumers was sold cars with nonfunctional cruise control or other mechanical deficiencies,” Prof Dyal-Chad says, adding that the autocorrecting of ‘wrong’ names should be considered alongside “examples of racist algorithms [that] cover a disturbingly broad range of social and legal functions, from racist sentencing guidelines, to credit scoring, to autofills that provide race-based answers to questions, to racially targeted advertisements.” This is not the first time that the marginalisation of certain names has been linked to other forms of prejudice and disadvantage. In April last year, Mayor of London Sadiq Khan lent his support to a campaign calling for mandatory ethnicity pay gap reporting. The campaign by the People Like Us non-profit organisation made the point that the autocorrection of non-Anglo identities is an example of “subconscious bias in favour of British-heritage names”. Such bias, the group argues, also finds expression in people from ethnic minorities often being paid less than their white counterparts. Or, as Mr Khan put it: “If your name is autocorrected, chances are your pay packet might be too.” The cultural dominance of English may be one contributing factor. The fact that Silicon Valley remains disproportionately white could be another. According to 2021 data from Pew Research in the US, America's tech sector is almost 70 per cent white, and the US Equal Employment Opportunity Commission says the vast majority of tech executives are white males. Given this, it should not be surprising that names and identities found outside this cultural group are passed over. Interestingly, devaluing names and identities in the 21st century can also be an in-group phenomenon. In February, Ireland’s High Court heard that the country’s flag carrier – Aer Lingus – and the Bank of Ireland both used computer systems that were unable to recognise sínte fada, which are accents on some customers' Irish-language names. Post-colonial neurosis, cheap software or simple carelessness, whichever explanation you care to believe in this case, the core issue is the same: some names are thought to be more equal than others. As for me, a quick experiment on my phone tells me that although “Declan” avoids the dreaded wavy red underline, the full-fat Gaelic version of my name – “Deaglán” – still causes some problems. I’ll live with it for now, but next time I go to Starbucks, I think I won’t adopt a monosyllabic John or Joe or Paul alter ego. I’ll just be me. That’s not too much to ask, is it?