UI Design Newsletter – April, 2006

If it feels right... It is right... right?

A few years ago, a research study – purportedly from Cambridge University – grabbed quite a bit of attention. The study said that the order of letters within words was really not so important to reading. Just to reinforce the point, the email that circulated the Internet included a dramatic demonstration, something like the following:

...In fact, sentences in whcih lettres aer transpsoed (or jubmled up), as in the setnence you are now raeding, aer no more difficult to raed tahn setnencse in whcih teh lettres aer in teh rihgt oerdr...

Readers were convinced. They could read that sentence. It really didn't feel hard. Human brains are amazing, aren't they? Viral marketing took over.

Everybody got the email messages. Psycholinguists and linguists (researchers particularly interested in how language works – I'm one) got the email twice a day.

What's Usability got to do with it?

Have you ever had that same experience in the interface design world?

Some sort of a "prevailing wisdom" about what citizens/ consumers/ staff need percolates up. A reasonable-sounding (folk usability) explanation is generated to explain why. Often there is an interesting user anecdote that led to the explanation in the first place.

The usability team – sensing a mismatch – tries to push back with explanations and similar examples. But the momentum of folk explanations continues to build...

It happens a lot, actually.

One way to deal with a prevailing "received wisdom" is to clearly show that it's not accurate. Typically, this means drawing on existing evidence and (perhaps) extending current theory to show why the naive theory isn't quite right.

If yuo can raed this yuor brian wroks. But that’s all you can conclude.

When the jumbled words email came out, psycholinguists did that. They tried to fight back against the folk psychology intuition with detailed scientific explanations, invoking well-understood constructs like top-down processing, phonetic neighborhood activation and race-based models of lexical access.

Their alternative explanations were also based in experimental science. They presented logically argued steps leading to an obvious conclusion: the Cambridge study didn't make sense. To no avail. Converts to the Cambridge-order-doesn't-matter-school-of-reading felt they also had "data." They could read the sentence, after all.

Eventually most of the psycholinguists/ linguists/ cognitive scientists (including me) gave up on trying to fix it. When the email arrived (again), they just cringed (again), hit delete (again) and secretly hoped it wouldn't come up (again) at the next family get-together.

The case of the jumbled words has a happy ending, though. Rayner, White, Johnson, Liversedge (2006) stepped up to provide evidence that seems to work to debunk the jumbled words legend for lay people. They report both current eye-tracking studies and previous work (Rayner & Kaiser, 1975) that captures the relative difficulty (or cognitive cost) of reading sentences with words that have

• no jumbled letters
• jumbled letters
• letter substitutions (replace one letter with another, inappropriate letter)

Their work shows that:

• Some presentations of jumbled letters are easier to read than others (e.g., it's easier if the words are jumbled at the end of the sentence or the letters are jumbled at the end of the word).
• Sentences with letter substitutions are harder to read than sentences with jumbled words.
• Reading words with jumbled letters always costs more (in cognitive processing terms) than reading normal text.

On making explanations usable.

The last statement – if you make something more complicated to do, it gets harder to do it – seems so obvious. So why was it so hard to convince people it's true?

Psycholinguists' attempts to debunk the urban legend didn't work because they provided explanations by cognitive scientists framed for cognitive scientists. In fact, the early counter-explanations are based on solid research. But to draw the right conclusions, listeners had to:

1. suspend their disbelief.
2. follow an theoretical explanation that drew on research studies designed to address similar but not identical questions.
3. speak cognitive science.

In contrast, the Cambridge study email presented a simple premise with one simple example to back it up. It presented one example in isolation:

If yuo can raed this yuor brian wroks.

Work hard to understand the alternative or accept the explanation that seems obvious because you've experienced it. Which would you choose?

Wow! That was easy.

Perhaps psycholinguists would have enjoyed more traction if, instead of offering a mini-lecture on lexical access, they offered the following one counter example with a relative baseline to compare against:

No, really... Which is easier?

a. if oyu nac eadr shti rouy narbi swork.

or

b. If yuo can raed this yuor brian wroks.

or

c. If you can read this, your brain works.

Sure, this explanation is less explanatory. But the goal was debunking the jumbled letters myth, not to enlist a new psycholinguist.

Sometimes convincing does not require educating.

Usability professionals fall into the same trap.

Because we are enthusiastic, we try to educate colleagues indiscriminately. We are effusive about affordances and primary noun analyses. We explain why the method used in user-centered design is the right one; how the logic of the interaction works; the gestalt of the visual hierarchy. We present detailed findings of usability tests when, perhaps a single "WOW!" moment from the usability testing highlights tape would do the trick.

Though none of the lessons are about language processing, there are several lessons to be learned from the "Cambridge study"...

1. Having an evidence-driven theory may not be enough.
2. Explanations do not need to be long.
3. Explanations do need to be usable.

References

Rayner, K., White, S., Johnson, R., Liversedge, S. (2006). Raeding Wrods with jumbled Lettres; There is a cost. Psychological Science 17(3), 192-193.

Rayner, K., & Kaiser, J.S. (1975). Reading mutilated text. Journal of Educational Psychology, 67, 301–306.

There are two valuable lessons here. One is the power of convincing demonstrations. There is no amount of theory that will be as useful as giving your executive sponsor a brain cramp trying to use a proposed design. Then demo the alternative that feels like a warm oil massage.

The second lesson is how very powerfully users can adapt – they can read the jumbled words, and they can use the jumbled interface – BUT AT A COST. The argument that "The users could do it" is pretty weak. If they CAN'T do it that is amazing considering the adaptability of people. But even if they CAN do it, how long does it take and what is the effort involved? The business consequences of a jumbled interface are significant, even if the users CAN sort it out.