To print: Click here or Select File and then Print from your browser's menu

This story was printed from silicon.com, located at http://www.silicon.com/

Story URL: http://comment.silicon.com/0,39024711,10004314,00.htm

Through the fog... Predictive texting
Forget next word, think next sentence - and a boost for future handhelds

By Quocirca

Published: Thursday 22 May 2003

If you send text messages you'll either be a multi-tap or predictive text fan. But where will the next predictive texting breakthroughs mean? Quocirca analyst Bob Tarzey has been looking at the market...

Text messaging, or short message service (SMS) as it is more formally known, could be about to go through another revolution. This could have implications beyond the mobile phone and its youthful champions.

When texting, as it is normally referred to in the UK, was first introduced it was cumbersome. To type a simple message such as ‘see you later’, involved about 23 key strokes (four taps on the ‘7’ key for ‘s’, two on the ‘3’ key for ‘e’ and so on). This did not deter fans who soon developed a whole new abbreviate language for texting. ‘See you later’ became ‘CU L8R’; still 16 key strokes but quicker, nevertheless. The method of entry for such messages is known, for obvious reasons, as multi-tap. The software for doing this was proprietary to the manufactures of handsets.

In the mid-1990s new applications started to appear on handsets that tried to predict the word being typed based on a single tap of a key for each letter. Using these predictive applications reduced the number of key strokes further.

For example, on my Nokia handset, to enter ‘see you later’ involves pressing the ‘7’ key once. An ‘s’ is displayed because out of the four possible letters (‘p’, ‘q’, ‘r’ and ‘s’) ‘s’ is the most widely used. Pressing the ‘3’ key once changes the display to ‘Re’, the most common two letter sequence of the possible combinations (‘p’, ‘q’, ‘r’ or ‘s’ followed by either ‘d’, ‘e’ or ‘f’). If I press the ‘3’ key again the display changes to ‘See’, which is the word I want, correctly predicted.

To complete the phrase ‘see you later’ with the predictive texting on my Nokia handset takes 13 key strokes. The software predicts all three words correctly as they are the most common words with the possible combinations. If a word is wrongly predicted and there is an alternative, then pressing the ‘*’ key allows you to view alternatives.

For example, if you want to type ‘cat’, pressing the ‘2’ key twice and the ‘8’ key once give you the word ‘act’. Press the ‘*’ key once and you will get the word ‘cat’; four key strokes. With multi-tap this would have taken five.

A problem with predictive texting is that it only predicts words that are know to its dictionary. The dictionary on the Nokia handset predicts many proper nouns, such as ‘David’ or ‘London’. Less popular ones like ‘Ealing’ or ‘Clive’ it does not. To enter these is cumbersome: You have to select a menu item which allows you to enter new words. These are then added to your, now personalised, dictionary and will be predicted in the future.

Ironically, this allows ardent fans on the new language that has evolved from the days of multi-tap to teach the predictive software their favourite words such as ‘CU’, although it will not learn and predict ‘L8R’.

The software that enables predictive texting is available from three main suppliers – all based in North America (incidentally the market whose consumers have been some of the slowest to embrace text messaging). Motorola’s Lexicus division has a product called I-Tap, AOL owns a company called Tegic whose product is T9 (and installed on my Nokia hand set) and Zi Corporation has a product called eZiText. All these vendors have one principle route to market – agreements with handset manufacturers who embed their software.

The principle user of I-Tap is not surprisingly Motorola. Licensees for Tegic’s T9 include Ericsson, Nokia, Panasonic, Siemens and Toshiba. Zi Corporation’s licensees for eZiText include Alcatel, Alphacell, Oregan, Pace and Sony Ericsson. The capabilities of their products vary – and no doubt the number of taps to enter ‘See you later’ – but broadly speaking they do the same thing: predicting the current word you are entering. So what’s new?

The next leap in technology to make the lives of texters even easier is next word prediction. This technology is being touted by a new player called AirTx. Its tx4u product is yet to arrive in our pockets but when it does the number of taps to send a text message should decrease even further. This is because it takes personalisation to a new level. tx4u will not only learn new words but will learn about the way you commonly use word combinations.

If the first message I enter on the phone is ‘See you later’ then next time I enter and ‘s’ it will ask if I want to use the word ‘see’ again. If I accept this it will ask if I want to use ‘you’ again and so on. The most commonly used word combinations that I type will be offered up as defaults.

It does this by developing a personal language model to anticipate what I want to say as I begin my message and then predicts the end of my sentence based on my previous use of language. In some circumstances, tx4u will even predict an entire sentence. AirTx believes its pending patent for next word texting is unchallenged.

Does this mean we will all have to learn how to text again? AirTx says no. Its technology can work on top of other predictive text solutions as well as its own. T9, eZiText or simple multi-tap could all act as a front end for tx4u, the user interface thus remaining familiar. The question then is if we, the consumers, will be prepared to pay the additional cost that handset manufacturers will pass on to make next word prediction generally available.

So what difference does predictive texting – current word or next – word make to those of us who are not text junkies? The answer may not lie with handsets but with other fiddly handheld devices - PDAs, palm tops and the like. The ability to enter text quickly and easily into these devices is the key to them being adopted as serious tools for running mobile business applications. To enter anything but the shortest text strings is just too much effort for many people, who prefer the familiarity of a large keyboard – that's not exactly handheld.

If we can be persuaded to adopt predictive texting as an input mechanism then take up of these tools for serious business use may well increase. Tegic and Zi are both targeting this market with their predictive technology. AirTx is likely to do the same when tx4u hits the streets.

Will next word predictive texting have any more success in killing off the gobbledegook - incomprehensible to most adults - that teenagers have come to treat as a private language? Probably not – CU L8R!

**Quocirca is a leading, user-facing analyst house known for its focus on the 'big picture'. For a full summary of its activities see www.quocirca.com, or reach the company's founding directors by emailing quocirca@silicon.com.

Also in this series: Through the fog... Business continuity and disaster recovery Through the Through the fog... Wireless email at work dilemmas Through the fog... Storage as a service Through the fog... Buying an application server Through the fog... Corporate content management Through the fog... Automated speech recognition Through the fog... Public Key Infrastructure Through the fog... Vendor-channel relationships Through the fog... What future photo messaging?

For Quocirca's 'What's the fuss about...?' series for silicon.com, see this page

And for their earlier 'Surviving the Recession' series, see this page.