Posted on: 27 Jul 2012
Written by: chris
Whoops! As the capital gears up in earnest for the Olympics starting later this week, It hasn’t been a great week for welcoming our Arabic-speaking brothers and sisters to the capital. Two different major London-based businesses have fallen foul of exactly the same typesetting gotcha, resulting in signs and posters proudly displaying gibberish.
Firstly the UK transport company First Capital Connect (FCC), which operates rail services in and around the London area, produced this Olympics security poster:
Next – and perhaps more embarrasingly – the Westfield Stratford shopping centre, literally across the road from the Olympic park itself, in Britain’s most multicultural borough (Newham), put up large multilingual “welcome” banners, the Arabic version of which looked like this (it’s the one in turquoise):
Unless you’re an Arabic speaker, the text above probably looks fine to you – which is probably how the problem occurred in the first place. However, it turns out the same two things are wrong with each piece of text:
- the whole message is written backwards;
- the characters aren’t joined up as they would normally be.
In both cases, the organisations in question have defended themselves, claiming they hired professional translators to produce the Arabic – and to be fair, they may well have done. So what went wrong?
Stop me if I get too geeky (which I undoubtedly will), but it’s all to do with something called Unicode, a blanket term for a series of technologies (supported to at least some extent by all modern computers) to represent world writing systems in electronic form. The most commonly-used benefit of Unicode in a multilingual context is that it enables very large fonts containing characters from all the world’s languages. In this case, it’s plain to see from the above pictures that the text in question is in Arabic characters, so it would indeed appear that the designers managed to clear the first hurdle.
However, Unicode isn’t just about large fonts – which brings us to the first point where our designers failed in this case. Arabic (in common with other Middle Eastern languages) reads from right to left (usually abbreviated to RTL)… but the two unfortunate examples above have both rendered the Arabic text in the normal Western way of reading from left to right… in other words, backwards. This probably wasn’t the fault of the designer’s computer (which no doubt supported RTL text just fine at operating-system level), but rather of the actual typesetting program they used (quite likely Adobe InDesign or Illustrator). Result: sdrawkcab tuo semoc egassem eritne eht. To give FCC’s designer some credit, they were clearly aware that Arabic is supposed to read from right to left, and so they aligned the message text against the right margin instead of the left, which is indeed what you do with Arabic text. Sadly, this didn’t actually correct the ‘backwards’ error; it just made it look to the untrained eye as if it might have done.
Secondly, and perhaps more subtly, Arabic (and also Greek, fact fans) uses a concept known as ‘contextual shaping’, which means that the actual shape of any given character (its ‘glyph’, in typographical parlance) changes depending on the other characters surrounding it. Huh? Practically speaking, this means an Arabic character will take one shape on the page if it appears on its own, another shape if it comes at the start of a word, a third shape in the middle of a word and a fourth shape at the end of a word. The closest parallel in English is our use of capital and lower-case glyphs to represent the same letters (yOU cAN rEAD tHIS tEXT bUT iT lOOKS wRONG), but evidently Arabic uses a much more developed version of the concept than that. Of course, the erroneous Arabic above hasn’t been contextually shaped, causing each individual character to take on its default ‘on its own’ glyph form. This creates an effect which non-Arabic speakers could perhaps best appreciate by imagining a piece of handwritten text in English which was supposed to be in joined-up handwriting but has actually been written out in printed capitals instead. Not actually incomprehensible, then (apart from being all backwards because of the first error, of course), but just awkward and wrong in its intended context.
Ironically, the designer’s undoing in this case was probably the decision to use a “professional” typesetting package, whereas he/she might have been better off using something more everyday like Word for Windows. As a typesetter myself, I’d normally (strongly) advise against this; but it turns out that Word is actually much better than, say, your ordinary European version of InDesign when it comes to representing Arabic, particularly when the job in question is as typographically simple as these two were. Word is far from a perfect Arabic solution, but it does actually do a reasonable job of rendering RTL and contextually-shaped text, neither of which are attempted (yet) by off-the-shelf European releases of Adobe’s software. [No, I'm not suggesting you actually do professional Arabic typesetting jobs in Word - you should really be paying a language specialist to get it right! - but I'm just saying it would have worked better in this case... ]
No doubt typesetters will one day be able to call upon a full armoury of software which silently and efficiently supports the complete range of Unicode features. Until that day, however, we’d suggest they stick to the golden rule: if you can’t proof it yourself, perhaps you shouldn’t be trying to set it.