Overview

This document shows a simple way for a phone manufacturer to

• Allow over 100 million new customers to communicate quickly by text messaging in their native language.
• Reduce the memory requirement of phones.
• Reduce the costs and risks associated with phone production, inventory and distribution.
• Make products simpler and easier to use.
• Catch the eye of the press through innovation.

Predictive text entry on mobile phones

Virtually every phone manufacturer offers some form of predictive text-entry system on their mobile phones. The most widely used system is T9, produced by AOL. Others include iTap (Motorola), and eZiText (Zi Corporation). All these methods require the phone manufacturer to install a large dictionary of words for every input language in their phones. Dictionary-based systems frequently fail. When this happens, users must fall back on using a laborious and slow secondary system, known as Multi-Tap.

Product differentiation in a market driven by replacement sales

The cell phone market is increasingly driven by replacement sales. In 2001, over 50% of sales were replacements. That figure is projected to rise to 75% by the end of 2002. Manufacturers must confront the fact that replacement customers are more discerning in their purchasing and need to see clear reasons to select a new phone. They will use experience to make decisions based on comparison of available features. As a result, phone manufacturers have to work harder to differentiate their product. Thus there has been a rapid proliferation of new (non-functional) features such as games, downloadable ring tones, changeable face plates, etc. In the midst of this activity and the flurry of increasingly exotic features, a fundamental user need has been ignored: giving people the option of rapidly and easily sending text messages in their preferred language.

Eatoni Ergonomics

Eatoni Ergonomics has developed LetterWise, an easy-to-use predictive text entry system currently being introduced into GSM, CDMA, and DECT phones as well as other devices that use a reduced keypad for text entry. The principle objectives driving the development of LetterWise were to provide high-quality predictive text entry and simplicity in user interface, using minimal memory. Unlike other systems, LetterWise does not use a dictionary and never requires the use of Multi-Tap. LetterWise is available today in over forty languages.

LetterWise makes it possible for a manufacturer to stand out in a new and important way: by providing rapid text entry to people who speak a language other than the dominant one in a region. The aggregate numbers involved are enormous. There is the potential for manufacturers to increase sales by millions of units and to increase brand loyalty, and for carriers to reduce churn and increase ARPU as new and repeat customers significantly increase their use of text messaging.

In this document we measure the scale of this opportunity, explain why it exists, and show why Eatoni's LetterWise solution is uniquely positioned to take advantage of it.

Linguistic research

Eatoni has done extensive linguistic research during the development of software for text entry in 120 languages. Our linguistic database contains detailed information on the numbers of speakers of 7,000 languages in 260 countries, on cell phone penetration by country, on country population, on predictive text entry language availability by software supplier, by phone manufacturer sales and by country.

By combining information from linguistic research sites such as Ethnologue, and from sources such as the CIA World Factbook, government census figures, analyst research reports, numerous manufacturer phone manuals, corporate web sites, news reports and press releases, Eatoni is uniquely positioned to provide detailed analysis of global linguistic demand and to produce manufacturer- and operator-specific reports.

Unsatisfied linguistic demand

There are three gaping holes in support for predictive text entry in today's market:

1. Ignored foreign linguistic minority: Imagine yourself as a Norwegian speaker living in New York (there are 612,000 Norwegian speakers in the U.S.). You have a number of Norwegian friends in your neighborhood, and you have many friends and relatives back in Norway. You would like to send text messages in Norwegian, but it is not included in the languages provided by your U.S.-bought mobile phone. Furthermore, you are unable to find any phone for sale in New York that offers Norwegian as a language for predictive text entry. You friends have a similar complaint and you all use the laborious Multi-Tap method to send messages in Norwegian to each other and to family in Norway.
2. Ignored national minority language: Or, imagine you are Welsh, and that although you speak fluent English, you are also one of the 500,000 speakers of Welsh. All your friends speak Welsh and you all use Multi-Tap to send each other messages in Welsh. You would happily use predictive text entry in Welsh and would also like to be able to use Welsh words in your English text messages. But, T9 does not allow you to freely mix words from other languages in a single message, and of course Welsh is not available in your phone in the first place.
3. Ignored country: Finally, suppose you are Icelandic, and live in Iceland, the country with the world's highest cell phone penetration. Predictive text entry with T9 or eZiText in Icelandic is not available. Why not? Because there are only 250,000 people in Iceland. Including Icelandic in a phone destined for the Northern European market at the expense of a language for the far more populous other countries of the region makes no sense. Phone manufacturers are sending a clear message to their customers in Iceland: You do not matter. One can only imagine the reception of a phone on the Icelandic market that offered high quality predictive text entry for SMS in Icelandic.

These people have something in common: their preferred language is either not available in phones in the region where they live, and, in many cases, has never been available in any phone anywhere. Their joint predicament is an opportunity for phone manufacturers.

How many of these people are there? What languages do they speak? Where do they live? Eatoni's research answers these questions. Eatoni's LetterWise is the key to delivering a differentiated and attractive product to these people.

Aggregate linguistic impact

LetterWise makes it possible to offer predictive text entry using just 3K bytes per language. It is probable that phones will appear on the market offering 50, 100, or even more languages. The Panasonic KX-TCD755 DECT phone, which uses LetterWise, has 24 languages installed.

We will now examine the aggregate linguistic market for such phones. A couple of examples have already been mentioned: 612,000 Norwegian speakers in the U.S. and half a million Welsh speakers in the U.K. These are relatively small numbers. The interesting question is: What does the picture look like when we consider all the languages in all the countries in Europe, Asia or the Americas?

Imagine the following experiment. A cell phone manufacturing company sends their market research team to question every person in every country in Europe. Interviewees are asked to list the languages they speak. For each language a person speaks that is not available on any phone that can be purchased locally, the interviewer adds one to the total for that language. At the end of this process, the researchers pool their results across all countries and languages.

Assume that before the market research begins, the phone manufacturer has divided Europe into four regions (North, South, East, West) and distributed phones with five T9 languages as follows:

Region	Languages
North	Danish, Dutch, Finnish, French, German
South	English, French, Greek, Italian, Spanish
East	Bulgarian, Hungarian, Romanian, Russian, Slovak
West	Dutch, English, French, German, Portuguese

Given this distribution, what would the market study show? Eatoni's research indicates that:

• The sum of the interviewer's counts across all unsupported languages is 210 million.
• Therefore, if people in Europe speak an average of two languages each, there is a market opportunity to address the needs of over 100 million people.
• A total of 74 different languages would be found to be unsupported, counting only those languages with at least 100,000 speakers.

The conclusion of the study is clear: a very large number of people are being denied a basic need by the company's current linguistic policy. How can a manufacturer turn these people into customers?

Dictionary-based systems cannot address this opportunity. Actually, they create it.

Although interesting as individual stories, markets such as the 200,000 Turkish speakers in The Netherlands have until now been too small for manufacturers to address. The main reason for this is that T9, the dictionary-based predictive text entry software available from AOL, uses approximately 60K of memory for each language dictionary, and over 100K for some languages. To provide even five languages in a phone, manufacturers have been forced to dedicate over 300K of memory per phone. As memory is both limited and expensive, including T9 and one or more languages has a very significant cost, both in terms of money paid for memory and in terms of lost opportunity. Opportunities are lost because the existing memory could have been used to provide other features (additional games, larger address book, storage of more SMS messages, longer voice memos, screen savers, etc.).

Thus the choice of which T9 languages to include in a new phone has been a serious problem for manufacturers. Omitting languages means ignoring the linguistic needs of millions of potential customers, while including languages rapidly consumes memory and thus greatly limits other functionality. As phones have begun to incorporate more memory intensive applications, such as WAP browsers, the competition for scarce phone memory has intensified. With the white hot pressure to reduce phone costs in order to survive as a manufacturer, the answer is not to install larger quantities of expensive memory.

Memory intensive dictionary-based predictive text solutions limit the ability of manufacturers to appeal to a wider market. Manufacturers must resort to dividing their markets geographically, offering different small subsets of languages in each. Manufacturers deliver embarrassing and awkward solutions that offer phone menus in over thirty languages, but predictive text software in just a subset of these. As a result, customers have to puzzle over setting their "phone language" versus their "Tegic language". The situation is made worse by the need for users to resort to Multi-Tap to enter names into their address book because the dictionary-based methods are useless for generalized entry of names.

In these ways, the use of memory-intensive dictionary-based predictive text solutions forces the manufacturer's hand. With a product such as T9 or eZiText, even if the needed language databases were available, there is simply no way to make phones appeal to a wider linguistic audience without dedicating megabytes of memory to store these databases. Manufacturers who use these products are forced to draw gross linguistic and geographical divisions that necessarily allow the needs of hundreds of millions of people to fall through the cracks.

Using Eatoni's LetterWise to tap the linguistic opportunity

Fortunately, there is a quick solution to this dilemma

• Put predictive text in 80 languages into every phone.
• Do not divide up the market linguistically and geographically: address it all at once, with every phone.

LetterWise makes this simple solution possible. LetterWise is currently available in 120 languages. With its extremely modest per-language memory requirement, a manufacturer can today produce a phone with 50 languages, using just half the memory that T9 or eZiText would use for only 5 languages. Only with LetterWise can a manufacturer simultaneously address the broad linguistic needs described above. This a the key difference that LetterWise provides. Minority linguistic communities can be addressed without sacrificing support for larger ones. As we've shown, these minority communities frequently number in the millions and sometimes comprise entire countries.

Why is predictive text entry so important?

The availability of predictive text entry drives increased SMS usage. For this reason, predictive text is important to ordinary phone users and also to the phone carriers whose revenue is increasingly buoyed by the rise of SMS and other data traffic. Ease of messaging implies increased messaging. Ease of phone use increases brand loyalty. In addition, the ability to rapidly enter product names, song titles, search strings, email addresses etc., with minimal difficulty will also be a important component complementing applications provided through Java, WAP and other interfaces. In short, predictive text entry on reduced keyboards will become part of the expected framework in which people will operate their phones.

LetterWise offers reduced production overhead

By installing the same 80 languages into every phone, a manufacturer can avoid costs in software configuration management. All phones can be shipped to all regions without the need for special builds to install language databases. All phones use the same amount of memory for language databases, so there is no complex juggling of applications depending on language memory. Phones do not need to be linguistically pre-configured for a particular region, a practice which creates inefficiencies in inventory and limits flexibility.

What other advantages does LetterWise offer?

There is only one predictive text entry method that makes it possible for a manufacturer to take advantage of the linguistic market opportunity described above: Eatoni's LetterWise.

There are a number of other important ways in which LetterWise distinguishes itself from dictionary-based methods such as T9 and eZiText. These include the ability to enter non-words, simpler user interface, and offering consistent text entry in all areas of phone functionality.