Issue No. 04 - October-December (2004 vol. 3)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/MPRV.2004.11
Cell Phones and Innovative Text Input
Cellular phone technologies have developed dramatically. As a communication tool, the cell phone is very convenient in terms of its continually shrinking size. Also, with more and more people signing up for the service, accessing voicemail, email, and Web browsing via cell phones will certainly gain popularity. However, my personal experience tells me that using cell phones for full-scale operation seems far from a sure thing. Every time I try to use my phone to access email or browse the Web, I'm not motivated to continue because of the keypad's slow typing speed—which I believe holds the key to making cell phones fully useful.
Thad Starner's article "Keyboards Redux: Fast Mobile Text Entry" (Wearable Computing, July–September 2004) discusses but doesn't quite satisfy my concerns. I agree with his statement, "Two barriers exist for these markets: appropriate display and text-entry speed." Text-entry evaluations are usually evaluated using two measures, speed and accuracy. Neither multitap nor T9 technologies can provide reasonable input speed.
As Starner points out, "If mobile phone service providers want to enable mobile email, faster typing methods must be devised." Though T9, which works with a built-in dictionary and requires a single key press, was developed to be more efficient and less time-consuming than mere character-number-matching multitap input, the significant problem is that users must learn how to enter data into their cell phones. The more time it takes to learn data entry, the more users will stay away from these technologies. Moreover, T9 is very difficult to use in terms of getting the right feedback while typing and looking at the screen simultaneously.
Currently, the email and Web browsing functions I expect to use in the context of anytime and anywhere (cell phone) service exist in little more than name. Low performance times and high error rates cause frustration and unnecessary stress, often leading users to totally give up. To cope with this challenge, we need to develop an innovative text-input method. Due to the phone's size restriction, no text-input method will match the speed we can perform on a regular computer keyboard. As a result, there's only one way to deal with this problem—voice recognition. We can simply leave the phone keypad for its original dedicated usage—number input only.
Nowadays, people don't have time to press a key or several keys many times to enter a few characters. Furthermore, users aren't supposed to spend a lot of time training to use a phone for email. That's why they prefer a simple, easy, and quick method that will let them use their phones at will.
Information Technology Manager
University of British Columbia
Thad Starner responds:
Thanks for your letter! I really enjoy getting feedback on the Wearable Computing department.
You assert that "there's only one way to deal with this problem—voice recognition. We can simply leave the phone keypad for its original dedicated usage—number input only."
Voice recognition is a great technology. In fact, I spent a year in BBN's (Bolt, Beranet, and Newman's) Speech Systems Laboratory learning how the algorithms work and what sort of interfaces are possible. The progress in the last 10 years has been quite impressive.
However, speech is often not an appropriate interface. For example, the Tokyo subway system used to be filled with people speaking on their mobile phones. This behavior was considered so disruptive that there are now signs posted on the subway that discourage mobile phone use. Today, many commuters type email and instant messages on their phones instead. A similar phenomenon can be seen among students in classrooms in Europe.
Certainly the role of voice interfaces is going to expand on the mobile phone. However, as the mobile phone becomes more like a personal computer, key entry will continue to be a dominant interface on the phone. Typing has distinct benefits in being silent and precise. For example, most of my life as a professor is spent in meetings or in face-to-face discussions during the day. However, these are also the times I want to look up information on the Web, take notes, or send instant messages to others in my team to help with the task at hand. Murmuring notes or messages to my phone while someone else is talking is distracting to others. In addition, I can't concentrate on what other people are saying if I'm talking to my phone. Similarly, I can't listen to a voice message on my phone while maintaining a discussion. However, I can easily type a short note while maintaining a conversation as long as the note is on the same topic as the conversation.
Another option would be to have the phone transcribe entire conversations so you could later revisit a conversation when free to access additional information. Besides being quite a difficult technical challenge, such transcripts are often not very useful. Transcripts often lack the subtleties of the conversation and are relatively verbose and awkward. A simple written sentence can summarize and encapsulate the idea better. Reporters know this; they often write notes while conducting interviews instead of simply relying on a voice recorder.
If you'd like to know more, see "The Role of Speech Input in Wearable Computing" (Wearable Computing, July–September 2002, pp. 89–93), which discusses speech interface limitations and points to some references on the subject.