Mark Pickavance talks about the layout of keyboards and how their origins have dictated common aspects across the globe
Although never trained to do so, technically I can touch type, mostly through a disturbing level of repetition. I’m so familiar with the QWERTY layout I only get really confused if I’m presented to letters organised alphabetically, as they are occasionally in some applications and devices.
But how did we end up with the seemingly random nature of QWERTY, are their better alternatives, and how do other language speakers cope with their variants?
I have some answers, but not all of them, I suspect.
Origins
As most people are aware, the QWERTY layout of the modern era has its origins in the typewriter. Specifically, the Sholes and Glidden machine is generally accepted to be the first commercially successful typewriter when it became widely available in 1874.
The concept of an ad hoc printing machine was first devised in 1866 by inventor/printer Christopher Latham Sholes, engineer Carlos S Glidden and their then associate Samuel W Soule.
Eventually, Soule left the endeavour before the typewriter was finished, and his place was filled by James Densmore, a man who saw the business possibilities of this invention and was prepared to provide financial backing to propel the project forward.
James Densmore is critical to this story, because according to common legend, it was he who was the grandfather of QWERTY. Early prototypes suffered with an problem: as the speed of typing increased, the likelihood of adjacent keys being struck in quick succession causing them to collide and jam.
It was he who came up with the idea of grouping commonly used letters with less common ones to lessen the possibility of a jam, and through a nefarious limitation of the hardware that coalesced into QWERTY.
What we also know both Glidden and Sholes tweaked the layout as the device progressed into the very limited productions runs that the inventors initially attempted.
These early machines were diabolically unreliable, and as they often shipped great distances across the US to their owners, getting parts and repairs proved something of a logistical nightmare.
There was also a cultural problem regarding typed letters to overcome. Early recipients often became incensed by them, as they assumed that they’d had a typed letter because either their eyesight or education wasn’t good enough to read the handwriting of another, which they took as an insult.
With remarkably low market penetration, low reliability and a high unit cost, this is where the typewriter and QWERTY might have ended if it wasn’t for Americans’ unrelenting love for firearms.
In 1873, Sholes and Glidden approached the gun maker E Remington and Sons, which was looking to diversify, and demonstrated their machine at Remington’s New York offices.
At the end of that meeting, Remington agreed to make 1,000 machines, for which they got $10,000 (approximately $200,000 today) and royalty rights, but the originators got to serve as the exclusive sales agents.
What was critical about this agreement was that being experts in mass-producing complicated machines with a high level of reliability, Remington effectively re-engineered the Sholes and Glidden typewriter to make it both reliable, better at taking abuse and cheaper to manufacture.
These early machines still had some problems, not least the cost of $125 in an era where many skilled people earned less than $40 a month. Yet they managed to sell 400 of them, including one to the famous author Mark Twain (aka Samuel Clemens).
The breakthrough came in 1976 when the machine, along with some rubbish concocted by Alexander Graham Bell’s called the ‘telephone’, was exhibited at the Centennial Exhibition. That increased sales by tenfold of the Remington No.1 model, and the company considered it worth further design time. Two years later, the No.2 model appeared.
This was a typewriter that we’d recognise today, in that it generated both upper and lower case letters, a critical feature missing from the original design.
The number of these machines in circulation rapidly increased, volume production reduced the cost to £80, and competing machines also started to appear.
All of them used the QWERTY layout, and they began to appear outside the USA in Europe and across the globe.
They were embraced initially by writers, journalists, lawyers and businesses, where clarity of communication was paramount.
Little did Sholes, Glidden and all the other contributors to the typewriter know that their concept would live on past their device into a totally different one that would change the world almost a century after their invention.
The Dvorak Diversion
Once typewriters got into wider circulation, lots of people commented on the inefficiency of the layout, among them US academic Dr August Dvorak.
In 1936, he took a long-term research project about keyboard layout and distilled it into a patented design, the Dvorak Simplified keyboard (DSK).
What’s interesting about his approach is that all the vowels are in a single line on the left side of the ‘home row’ (where the hands naturally rest), while the most used consonants are grouped on the right.
The problem was that even in the 1930s, there were lots of people with QWERTY experience, so Dvorak needed to convince people that conversion was possible and beneficial.
In conjunction with the US Navy in 1944, he devised a sturdy in which 14 typists were converted to the new layout, and they averaged 52 hours to detrain from QWERTY and retrain on Dvorak. According to his findings, they became 74% quicker on Dvorak and significantly more accurate.
That’s because on the new layout, the fingers travel less, since there are fewer combinations of keys where the fingers have to travel large distances across the keyboard. As the distance of travel increases, so does the likelihood of hitting the wrong key, in theory.
The US Navy were convinced,and ordered 2,000 typewriters with the Dvorak layout on the basis of it.
However, that might have been premature, because later research revealed that Dvorak’s enthusiasm for his own layout might have got the better of his empirical approach.
In the early 1950s, many organisations inspired by the experience of the US Navy started to seriously consider transition to Dvorak, and the US General Services Administration, an independent agency of the United States government, commissioned a study to determine if the claims were true.
Regrettably, under strict controls the majority of the speed increases and the ease of transition seemed largely illusory.
Research done since has determined that it is possible to get typists to switch to almost any layout in a similar time frame, and having the majority of used keys on the home line is probably only worth a speed increase of 4% at best.
What really undermined Dvorak’s research was that it later became known that the navy study on which the whole technology was promoted was actually the third study he’d performed. And because the results didn’t support his hypothesis, he’d junked the collected data.
Supports of DSK will claim that Earl Strong, the man dispatched to conduct the tests disliked Dvorak, both the man and his keyboard, even before the first test was conducted, so these tests are likely tainted too.
Modern ergonomics researchers now consider that Dvorak’s ideas were good, but he didn’t really consider how complicated a physical action typing is and that the speed anyone can achieve is limited by many factors, not just the specific layout.
You can still get Dvorak keyboards today, and Windows even includes support for the layout. There are people who swear by them, and others who say that they can switch easily from QWERTY to Dvorak without breaking sweat, but the real problem that this layout has is that most people aren’t familiar with it.
And this precise issue, the devil-you-know syndrome, has also been a factor in maintaining other keyboard layouts that even more unsuitable than QWERTY.
C’est La Vie
While QWERTY has its detractors in the English speaking world, give a thought to those on the other side of the channel, who have suffered not typically in silence for a whole century with a layout that is totally reviled.
What’s mildly fascinating about AZERTY, a derivative of QWERTY, is that no one will actually take responsibility for devising it. At some point in the last decade of the 19th century, it appeared on typewriters sold in France and Belgium, and it continues to this day for most French speakers.
Yet even in the first decade or use, many declared it not fit for purpose. It was hated so much that in 1907, after much ergonomics research, Albert Navarre presented the ZHJAY layout as an alternative. But already some had learned to cope with AZERTY, and ZHJAY gained little traction, unfortunately.
Even to the casual, non-French speaking observer, AZERTY is something of a disaster. Oft-used characters are difficult to generate, while things that are almost never used get an un-shifted key to their self.
It’s also bizarrely biased to the left hand, and the centreline contains a collection of easily assessable letters that few French words use.
The most striking of these is the ù (u-grave), which gets the key that on QWERTY is the right of the ‘L’ on the centreline. Sadly, in modern French, ù is only used in one word, où, meaning ‘where’.
You would therefore have my sympathy if you get that letter while playing Scrabble in France. But you won’t, because the ‘diacritical’ marks as they’re known are ignored in that game, I’m led to believe.
What’s really odd is that where ù gets a whole key to itself, other accented characters require all sorts of finger acrobatics, even though they’re used in many common words.
Especially challenging are capitalised versions of accented letters, some of which are so difficult to generate, people can’t be bothered to use them any longer.
That’s a real problem according to the ministry Academie Francaise, which is the guardian of all things cultural in France, including the language and how it is represented. According to guidelines it’s issued, capital letters deserve the correct diacritical including acute, circumflex, accent circonflexe, diaeresis, grave and cedilla, or they’re technically not pronounced correctly.
When you also factor in the ligature letters æ and oe, neither of which you can get out of a AZERTY, it’s no surprise there is now a very serious debate about dumping AZERTY before it damages French any more than it is already considered to have done.
Much research has been done in an attempt to make a new layout that addresses the concerns of both the cultural bastions, regular users and global French speakers.
From these, a number of likely candidates have surfaced, including one very well considered design called BEPO (or BÉPO).The problem that this and other layouts face is the same one that sent ZHJAY packing a hundred years ago: once people have learned to cope with a layout, they’re never keen to relearn that skill. It would take a monumental effort to switch now, and few believe that the government in France has the appetite for rubbing its populous the wrong way by demanding AZERTY is dumped.
Compared with some countries, though, France has had it easy.
Outside The Latin Box
With the spread of the typewriter, a range of variations for the majority of European languages were developed. The common feature that most of them exhibit is that they use a ‘Latin script’ (i.e. they use letter forms that are based on Latin representations of letters). These are used in many languages, even if many of them can’t actually agree on the sound they represent. It’s been calculated that about 70% of the world uses a writing system built around these symbols, and its application is wholly phonetic and is the basis of the International Phonetic Alphabet.
The keyboards that use Latin script are broken down into those that are based on QWERTY (like AZERTY, and those that use the letter forms but in alternative organisations, like Dvorak and Neo.
But beyond these are keyboard layouts for languages that either have their own letter forms or that aren’t even phonetically represented. There are a wide range of these, including the Brahmic scripts (Bengali, Thai, Tibetan, etc.), Arabic, Armenian, Cyrillic (Russian, Bulgarian, Serbian, Ukrainian), Georgian, Hebrew, Inuktitut, Cherokee and Tifinagh.
If you’re wondering, as frankly I did before I wrote this, how hard it is to get a keyboard in Cherokee, then you’d be surprised to know that this is a supported layout and script for Windows, iPhone and Android.
Considering there are less than 25,000 native speakers of Cherokee in the US, that’s quite impressive. However, some rare languages have no written form and therefore no available keyboard representation.
But what really made this Westerner’s mind boggle was when I started to look at how Asian language keyboards work, given that their representation is often not phonetic but pictographic, or an odd combination of the two.
Special Input Methods
In English, we limit ourselves to 26 commonly used characters, and a handful of other symbols like hyphens and exclamation marks. Imagine if you can use a language where there are at least 2,500 commonly used symbols, all of which you need to get out of a keyboard that has no more keys than a QWERTY one!
That’s the sort of challenge faced by Chinese and Japanese computer users, as their characters aren’t created through a direct key=symbol translation.
Essentially, all input starts with sending the keyboard into a particular mode often by entering a pronunciation cue and then a number if there are variants.
On a computer versus a typewriter, heuristics are used to zero in on the correct symbol in a method that’s much like the guessing game that auto-complete plays on your phone.
Since the advent of graphical interfaces, typing in Japanese usually involves a small input that then generates an on-screen menu with the most likely characters you’ll need and then the selection of the desired symbol.
In the Japanese JIS standard, often the word is inputted phonetically or kana, then converted to Kanji by the input method that you’ve predefined. However, there are other input methods that are really complicated, and they use special key layouts, including the infamous Oyayubi Shifuto (thumb shift) layout where the space bar is split into two or four special modifying keys.
Of the Asian languages, Chinese is the most complicated to key, because there are multiple ways to generate a character using either pronunciation or structure. In the structural mode, shapes can be inputted that are combined to define the symbol, mimicking the way that they’re constructed when they’re actually drawn.
As that comment suggests, many Chinese users have invested in drawing accessories for their computers, because sketching a symbol is much faster than generating it from the keyboard, unless you’re an expert at using one.
The third major Asian language, Korean, is entirely separate from both Japanese and Chinese, in that their keyboard is phonetic, and the symbols are in the Korean alphabet Hangul.
Where this differs from Western languages is in how it is written, where consonants and vowels are grouped together into syllabic grids. This process is automated by the computer as the words are presented, as there is generally only one way to correctly group the combinations.
There are two keyboard layouts that the Koreans generally use, Dubeolsik and Sebeolsik. The Dubeolsik layout is the most common, and looks rather like a QWERTY keyboard with extra shifted symbols on it. Korean speakers would recognise that the symbols are organised so that the consonants are all on the left and the vowels on the right.
That the Korean keyboard sits at an intersection of European and Asian thinking is often presented as a contributing factor in that region’s success in the tech industry.
What makes Asian languages additionally complicated for Westerners to follows it that Japanese, for example, can be written as English, as in left-to-right, top to bottom. But it can also be written in Talegaki mode, where the characters start top right, go down and then the next column moves from the right to the left.
These variations and the frankly horrific input schemes provide some explanation as to why few Westerners ever master these languages to speak, and even fewer to type.
The Future Of Keyboards
The first computer I used didn’t have a keyboard; it used punch cards. But since that point, I’ve been welded to this input device for approaching 40 years. Better layouts, more ergonomic key placement and the best mechanical switches can’t ignore the fact that this peripheral is seen as a hindrance in communicating with a computer.
To that end, we’ve spent many years developing increasingly sophisticated alternatives, like voice and handwriting recognition. Yet, years after these things have become effective enough to use, we’re still going back to the keyboard. Why?
As a long-time keyboard user, I’m inclined to believe that there’s something about the keyboard that mirrors part of the way that the human brain works. And a by-product of this modelling is that once you’ve mastered a layout to a certain level, you can disassociate your conscious mind from the physical actions, allowing you to concentrate on the contents of what you’re typing rather than the action. In the same way, we don’t generally spell out the words we’re speaking or the phonetics in our head before we say something.
This is much more difficult for people to do when writing, as we don’t tend to talk in the compact style that we often type. The formality of written English isn’t something you’d generally hear in conversation, unless it was between Sherlock Holmes and Doctor Watson, perhaps.
I do know some journalists that use voice recognition, and they’ve adapted to the idea of creating fully formed sentences that they can then speak. I tend to write in a series of waves, often where I revise a sentence part way through or even change the focus entirely as I construct it. This doesn’t lend itself well to speaking, though it works well enough with a keyboard.
I’m not saying for a moment that I couldn’t convert to keyboard-less output, but given the amount I produce in the time I’ve got, it would be an expensive experiment and therefore one that I’d rather not be involved in performing.
And that is the rub, really. The QWERTY keyboard, for all its infamous points of failure, is the curse we almost all know, and because of that, it’s very difficult to change.
This is not a unique scenario, because years after very efficient automatic transmissions have been available, people are still buying cars with more pedals than feet to operate them and the opportunity to select the least efficient gear while driving.
Keyboards have been around for over 140 years, and they look unlikely to be seen off by either Cortana or Siri in the near future. However, to languages like Japanese and Chinese, their advantages may be much greater, and therefore the take-up in those regions might well be higher.
That we’re using QWERTY now, a layout designed to solve an engineering flaw in the construction of a device no longer in general use, is proof that the law of unintended consequences is alive and well within all complex systems.