Mobile Text Entry

Definition

Mobile text entry is the problem of getting written language into a device that a person is carrying rather than sitting at. It is one of the defining constraints of mobile and wearable computing, because the QWERTY keyboard — the solution the desktop inherited — requires a stable surface, two hands and visual attention, and a mobile device reliably has none of the three.

The Twiddler

This territory is best defined by the Twiddler, created by Handykey Corporation. The Twiddler is a one-handed chording keyboard: it is gripped in the palm, with the thumb operating a small block of modifier and cursor keys and the fingers pressing a twelve-key grid, and characters are produced by pressing combinations — chords — rather than individual keys.

Chording is the crucial idea. A keyboard with one key per character needs at least as many keys as the alphabet; a keyboard that recognises combinations needs only enough keys that the combinations outnumber the alphabet. Twelve keys yield far more chords than a language needs characters, so the device can be small enough to hold in one hand and still address the full character set, without the modal cycling that made phone keypads slow.

The Twiddler became standard equipment in the wearable computing community around the MIT Media Lab in the 1990s precisely because it satisfies the constraints that matter for a wearable: one hand, no surface, and — once the chords are learned — no need to look at it at all. Thad Starner and colleagues studied its performance systematically, finding that novices reached roughly 26 words per minute after around 400 minutes of practice, with expert users substantially faster, which put it well ahead of multi-tap on a phone keypad.

The Trade-off

The Twiddler illustrates the central trade-off in this domain with unusual clarity. Chording is fast, eyes-free and compact, and it is difficult to learn — the chords must be memorised before the device is usable at all, and the learning curve is measured in hours of deliberate practice. Every mass-market text entry method has chosen the opposite side of this trade: immediately usable by someone who has never seen it, and permanently mediocre thereafter.

The lineage of that choice runs through multi-tap keypads, T9 predictive text, handwriting recognition on stylus devices, the on-screen touch keyboard, swipe-path input, and dictation. Each is discoverable in seconds. None reward practice with expert-level speed, because none have an expert mode to reach.

The problem returns with every new form factor that lacks a surface. Smartwatches, head-mounted displays, and any interface controlled by voice in a public place all reopen the same question, and the same options remain: constrain the vocabulary, predict aggressively, transfer the work to speech, or ask the user to learn something. Chording persists in niches where the last of these is acceptable — stenography, accessibility devices, and the enthusiast keyboard community — which suggests the approach was never wrong, only badly matched to a market that will not train.

Related Reading

• Wearable Computing
• Designing Interactions
• Frictional Interface
• Mobile technologies