With stenotypes, courtroom transcriptionists press multiple keys simultaneously, allowing them to transfer information to their computer much faster. Using this technology, professionals type at 225wpm,1 more than three times faster than professional QWERTY typists.2 With new open source software, it costs only $50 to try to switch over. My investigation showed that stenotype can probably improve our typing speed, but it would take months to years to learn. Investigation of my typing speed showed that I rarely exceed 40-60wpm, and that I rarely reach my ceiling of 100wpm, thus switching to stenotype would be unproductive. For those who type more slowly than average, a viable alternative to QWERTY and stenotype is voice recognition software, which merits further investigation.
Many readers of this essay probably spend half of their waking hours on a computer. This requires us to spend a lot of time funnelling our thoughts through our input devices. Last week, I found a type of keyboard that promises to make this easier and more productive: the courtroom typewriter, or stenotype. As the stenotype is much less widely discussed than alternative keyboard layouts like DVORAK and Colemak, I have focussed this investigation exclusively on stenotype.
How does stenotype work?
A stenotype is a chorded keyboard, meaning it permits you to press combinations of keys together – chords – rather than just one key at a time. Chorded keyboards transfer information faster than ordinary keyboards as the number of unique chords increases exponentially with the number of keys. A ten-keyed stenotype can play 1023 chords. A standard twenty-two keyed stenotype can play 4.2 million chords. The 200,000 words3 in the English language allocated to these chords using a phonetic coding system,4 so that chord-word pairs can be learned easily. Some long words require multiple chords and some multi-word phrases are encoded by single chords so that on average, each chord encodes around 0.8 words.5 You can see chords being typed into a classical stenotype here.6
- An n-rollover keyboard
- Software that translates chords into words
These used to be bought together for thousands of dollars. Now, it is possible to instead use 1) a $50 X4 Sidewinder keyboard, with 2) The open source stenotype software Plover.
What improvement in typing speed can we expect?
To work as a professional stenotypist, you must type at 180-225wpm7 compared to 50-70wpm to work as a QWERTY typist.8 The average QWERTY user can type at 40wpm, according to data gathered in 1997,9 although the figure may have increased since then. You can measure your typing speed here,10 although you will not be able to sustain this 1-minute typing speed over a longer period. The world record sustained typing speed for stenotype is 360wpm,11 compared to 174wpm for QWERTY.12
Stenotypists are more extensively trained than QWERTY typists, and are selected by a difficult training program. One steno school advertises a retention rate of 75% for three years,13 while pessimists describe dropout rates of 85%.14,15 Although it is possible to reach sufficient speed to pass one’s stenography tests within 18 months,16 this is uncommon. Two of the programs are 31.517 and 48 months long.18 Anecdotally, the amount of time taken to pass one’s tests can blow out to 5-7 years.19
Amateur attempts at learning to type with Plover are scattered across online forums, and range from 40wpm after 6 months20 to 60wpm after 12 months21 to 210wpm after one year,22 and none of these users specify their previous QWERTY typing speed.
Learning a chorded keyboard takes tens of hours. In a study of a one-handed chorded keyboard, typists naïve to chorded keyboards would initially type at 4.6wpm. With 95 minutes of practice, this improved to 13.5wpm, 20% of their original QWERTY typing speed.23 With a two-handed keyboard, subjects naïve to both stenotype and QWERTY keyboards were split into a group that learnt QWERTY and a group that learnt a chorded keyboard. After 20 hours, the chorded-keyboard typists were entering 32 words per minute (in Hebrew), and after 60 hours they reached 55wpm, and were almost twice as fast as the QWERTY group.24
In conclusion, the switching costs here are very high. The question for the reader is what speed they ought to expect to be typing after about six months. The quality of this data is poor, but my guess is that the reader would reach 40-100wpm.
Are we limited by our typing speed?
In conversation our speech averages 164wpm.25 However, we rarely if ever need to type this fast. It feels like my typing speed is in greatest demand when I have a burst of writing activity. So I downloaded Revealer Keylogger26 and set it to monitor my keystrokes for five days. Most of the time, I would only type a few thousand words per day, and this was mostly a couple of words at a time into a google search, form, or similar. Eventually, I captured an 88 minute burst of activity, in which I typed 1705 words, at an average 19% of my maximum possible speed of 100wpm (the 100wpm speed was recorded in a 1 minute trial).
Even at the hump from 9:00 to 9:10, I typed less at than 60wpm.
So my typing speed seems not to limit my computer use. If I typed at 40-60wpm, then typing speed would start to become a limiting factor.
Trends in input devices
Mark Kislingbury, world record stenotypist reports that 5% of courtroom transcription is now done by voice rather than by stenotype, and this is rising.27 If machine learning continues to improve, then voice recognition ought to replace keyboards in many contexts.
There is also a widespread trend toward more mobile computing, in smartphones, laptops, tablets and Google Glass, which is being reflected in input design.
A third potential perspective argues that we can make efficiency gains by mapping input devices more naturally across the human body. It is already possible to configure 24 chorded keys across a glove. After tens of hours of practise, users typed into chorded glove devices at around a third of regular typing speed.28,29 It may become possible to develop adaptive, personalised chording keyboards, for portable use to improve further upon this.30
Switching to a keyboard might make QWERTY typists faster, but this is uncertain, and the cost of switching is large. Moreover, my computer use is not limited by typing speed anyway, and it is unclear whether typing speed is a limiting factor for the average typist. Thus switching to a chorded keyboard would probably not make sense for most readers. It would, however, be a useful experiment for any worse than average QWERTY typist who writes a lot. Useful next steps in research of efficient computer inputs would be to:
- Investigate voice recognition software (e.g. Dragon Naturally Speaking)
- Ask amateur stenotype users how much work it takes to get to an acceptable stenotype speed, and whether they lose typing speed with QWERTY.
- Try switching to stenotype, and then report one’s findings. This job would suit a slow-typing novel-writer or programmer.
- Find better keylogging software. This should output .xls or .csv files and should make timestamps at less than ten second intervals.
- Compare the usefulness trajectory of brain-computer interfaces compared to ordinary interfaces.
15. Downey, G. J. (2008). Closed captioning: subtitling, stenography, and the digital convergence of text with television. JHU Press.books.google.com/books?isbn=0801893437
23. Anderson, A. M., Mirka, G. A., Joines, S. M., & Kaber, D. B. (2009). Analysis of alternative keyboards using learning curves. Human Factors: The Journal of the Human Factors and Ergonomics Society, 51(1), 35-45. http://people.engr.ncsu.edu/dbkaber/papers/Anderson_etal_HF_09.pdf
24. Gopher, D., & Raij, D. (1988). Typing with a two-hand chord keyboard: will the QWERTY become obsolete?. Systems, Man and Cybernetics, IEEE Transactions on, 18(4), 601-609. http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=17378&url=http%3A%2F%2Fieeexplore.ieee.org%2Fiel1%2F21%2F631%2F00017378.pdf%3Farnumber%3D17378
25. Yuan, J., Liberman, M., & Cieri, C. (2006, September). Towards an integrated understanding of speaking rate in conversation. In INTERSPEECH. http://languagelog.ldc.upenn.edu/myl/ldc/llog/icslp06_final.pdf
26. http://www.logixoft.com/en-gb/index. Note that this software records my keystrokes 10-30second batches, making my typing speeds appear more lumpy than they actually are.
28. Lee, S., Hong, S. H., & Jeon, J. W. (2003, November). Designing a universal keyboard using chording gloves. In ACM SIGCAPH Computers and the Physically Handicapped (No. 73-74, pp. 142-147). ACM. http://bauhaus.ece.curtin.edu.au/~iain/PhD%20BU/A_Phd%20docs/To%20read/A_printed%20and%20refed/ACM%20articles/p142-lee.pdf
29. Rosenberg, R., & Slater, M. (1999). The chording glove: a glove-based text input device. Systems, Man, and Cybernetics, Part C: Applications and Reviews, IEEE Transactions on, 29(2), 186-191.
30. Pham, T., Kim, K., McKay, B., & Nguyen, X. H. An Adaptive, Personalised Chording Keyboard. http://sc.snu.ac.kr/PAPERS/KBD_HCI2009.pdf