Tactile Feedback to Improve Text Messaging Speed

According to surveys, 85% of Americans own cell phones. In addition, Americans also send on average 568 text messages per year. This being said, I’d guess that many people reading this blog send/receive exponentially more text messages than that! With these things in mind, cell phones are arguably one of the most pervasive mobile computing and communication devices. In addition, according to the CTIA, Americans text more than they talk. With this in mind, cell phones have definitely proven their convenience through portability, but there are some significant shortfalls that have yet to be ironed out. Looking at the various methods for text entry in cell phones, there are two main varieties – predictive (such as Tegic’s T9), and non-predictive entry. Along with these two methods, there are keyboards ranging from condensed dial pads with multiple letters per button to full QWERTY pads (a standard keyboard). As a side note, condensed keyboards with multiple letters per button use a technique called multitap in order to make all letters available.

At the ACM Conference on Human Factors in Computing System (CHI) this past Spring, a group of researchers from Scotland proposed a novel idea to increase the rate at which we text message particularly on condensed cell phone keyboards. As rationale, they cited the overall increasing trend of predictive text entry, text messaging in general, and the trend towards smaller devices. Their idea hinges on the notion that with predictive text messaging, if users were given feedback during text entry, this would improve text entry speed and prevent syntax mistakes in text messages.

However, mobile devices are often used in “noisy” or disabling environments such as loud subways, sunny beaches, or quiet workplaces. This rules out auditory and visual feedback. Given that users are inherently holding the cell phone, the researchers reasoned that tactile feedback would be most appropriate. This research breaks the total time taken to enter a given phrase into three categories:

1) Th – Time taken to settle hands on the keyboard (.4 seconds)

2) Tk – Time taken to press a key (.28 seconds)

3) Tm – Time taken to mentally respond to system action (1.38 seconds)

This research targets 3). In short, the proposed feedback is a 75ms long vibration using the phones’ built-in vibrator. When the predictive text mechanism thinks the user should look at the screen, it triggers the tactile feedback. Given a suggestion of n letters with m letters typed, and Tk’ the user’s average typing speed over the past 10 characters, if (n – m)Tk’ > Tm, then the system fires the vibration. Basically, if the potential time saved is greater than the mental processing time (3), then the mechanism tries to save the user some time. In another scenario, if the predictive text mechanism cannot find any words in its dictionary that match the input text, the system fires a longer, 150ms long vibration.

The experimental setup consisted of twenty participants with varied experience using predictive text entry, age, gender, etc. In order to quantify these variables, this experiment looked at the data with a per-character time span. Each participant entered six phrases into a Nokia E65 cell phone using predictive T9 technology.

After conducting the experiment, it was apparent that using tactile feedback in the given scenarios improved overall text entry speed by about 5% and is statistically significant.

No Tactile Feedback     Tactile Feedback

600 ms/key                          519 ms/key
20wpm                                  23wpm

Future research in this project aims to target highly ambiguous words with unique feedback, in order to more clearly alert the user. Overall, this would help to “fine-tune” the algorithm so that it doesn’t just treat every ambiguous word the same.


Dunlop, D, Taylor, F. “Tactile Feedback for Predictive Text Entry,” in Proc. CHI ’09, Boston, Massachusetts, USA, 2257-2260, 2009.


~ by andrewsporter on November 30, 2009.

5 Responses to “Tactile Feedback to Improve Text Messaging Speed”

  1. Is this system capable of learning new words such as slang and jargon? Many of us often use words that a cell phone shouldn’t recognize in its dictionary, so if you were typing those words, would the tactile feedback system alert you every time or would it eventually ‘learn’ the slang and jargon a person uses frequently and add it to the dictionary? I could see where it would be irritating to have your phone buzzing at you every time you typed someone’s name or a shorthand, etc. If I am correct, today’s phones already have a little bit of a learning ability if you use an uncommon word often enough. Do they plan to improve this feature or not? And when you mentioned targeting highly ambiguous words with unique feedback, what would the feedback be? One last question: would there be any move toward making predictive text more intuitive? Right now many phones’ predictive text will predict a word that isn’t very useful and you have to tap through a list of equally stupid choices before getting to the common word you were looking for. Thoughts?

  2. Good question,

    This research is actually independent of your cell phone prediction dictionary. You are correct that phones will sometimes “learn” your slang or jargon. I also know that with some phones you can simply add an entry to the weighted dictionary. At this point, the weight of your slang is really up to its usage, which is a process underneath this research.

    Basically, the way that this works is: Given a word suggestion (or the absence of one), looking at the # of words typed in comparison to the suggested word length, if the function returns true, then trigger the vibration.

    Elizabeth – I completely agree with your frustrations with text messaging suggestions. Predictive text isn’t the best, but it’s certainly getting better. While this is still evolving, I think it’s important to note that the physical keys and entry methods (i.e. touch screens, extremely small buttons, poor placement) are equally inhibiting.


  3. I’m a big fan of multi-modal interfaces, especially of the tactile variety. I can’t speak for everyone but I know that I would respond well if my phone started buzzing once I spelled a word wrong or forgot a space, almost like mini “shock collar”. In fact, I am pretty sure this method of text input would not only decrease the time it would take me to text but also improve my texting spelling and grammar. While this study is interesting, I feel that they could have done much more testing, including testing over a longer period of time to allow users to learn the system. I believe that the results from a longer study would show an even more dramatic drop in texting time.

    In response to the above comments, I believe that most phones store your contact names in the “dictionary” and can suggest them to you as with any other word.

  4. Hey I was wondering how the research was conducted to generate the times for the various actions prior to and during the texting process on a mobile device as revealed in your article. How many people were used in this experimental survey?, who or what were the controls of the experiment?, are the results (in seconds)?, averages of the sample? (However many people the experiment was conducted on). Because I know it takes me a little longer to do all those things.lol

  5. Was any of the research done on phones that have a touch-screen keyboard, such as the iphone? Are text speeds generally faster with touch-screen keyboards? And is there any data behind the increases in texting speed with phones (e.g., the iphone, again) that offer visual — not auditory or tactile — predictive suggestions that allow the user to decline the suggested word if desired?

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: