ITR Atlas
← All interfaces

Twiddler Chording Keyboard (HandyKey, 1991)

Measured by Lyons, Plaisted & Starner · IEEE ISWC 2004 (2004)

Manual Chord keyboard 1991

Inputs

The measured or assumed values behind the calculations, each with its source.

rate = 47 wpm
Expert average across 5 subjects after ~25 hours of practice; one subject reached 67 wpm, matching a 10-year Twiddler user. A separate longitudinal study had 10 novices average >26 wpm after 400 minutes. The system date follows a HandyKey Corporation Twiddler brochure copyrighted January 1991, listed in Google Patents' non-patent citations; the throughput is measured in the 2004 ISWC study.
H = 1.0 bits/char
English-text entropy (Shannon).
N = 30
Output alphabet size for the raw-character Wolpaw ceiling (uniform prior). The chord layout changes production speed, not the size of the character set produced.
T_char = 0.2553 s/char
Gross character interval for the Wolpaw ceiling: 60 / (47 wpm × 5) = 0.2553 s.

Strictest ITR

Each scoring method is an upper bound on the channel, so the headline is the strictest (smallest) one for this entry. Use the score selector on the home page to view any single method across entries.

Strictest Shannon (text) Recomputed
Character-entropy throughput
Net of English redundancy
3.92 bits/s

  1. Characters per minute

    47 wpm × 5 chars/word = 235 chars/min

    Uncorrected word-level error was ~0.3% for experts (chording errors fell below 5% after the second session). Users mostly left errors in, so the 47 wpm already reflects them; documented, not re-applied.

  2. Bits per character

    H(English) ≈ 1.0 bit/char (Shannon)

  3. Information transfer rate

    235 char/min × 1.0 bit/char ÷ 60 s/min = 3.92 bits/s

What counts as a bit depends on the action space. The number of distinguishable actions and how likely each one is are design choices of the task, not the sensing hardware. The same modality can present a fixed set of targets, a set pruned per step by a grammar or language model, or a continuous control space. Each of these changes how many actions are live and how the probability mass is spread, and therefore the information per selection. Read the action space below before comparing headline numbers across entries.

Action space

What the user can produce at each step, and how those options are distributed.

Structure
Fixed set of targets
Size
30 distinguishable actions
Prior
Context-conditioned: likelihoods depend on prior actions
Notes
A one-handed keyboard where each character is a finger chord (button combination) rather than a key. The alphabet is the same ~30-symbol set as a keyboard, so the reference counts the English character stream at Shannon entropy, consistent with the other typing entries; the chording layout changes how fast the symbols can be produced, not how much information each carries.

Comparability The strictest bound here is the Shannon entropy of the output text, under one predictor held constant across the whole atlas (≈1 bit per character). That shared predictor makes it directly comparable to every other text entry (keyboards, spellers, silent speech and speech BCIs) regardless of prior or vocabulary size. For most text interfaces it comes out tighter than the raw-selection bounds, but not always. Where a small vocabulary makes Wolpaw tighter, that wins instead. Any Fitts, Wolpaw or log₂(N) figure shown below is another bound on the same channel. Switch the home-page score selector to compare one across entries.

Other bounds considered for the headline

Also valid upper bounds for this entry and eligible to be the headline. They just came out looser than the strictest above. Pick any of these in the home-page score selector.

Wolpaw Recomputed
Wolpaw bitrate over the raw character set
Uniform-prior, perfect-copy ceiling on the character channel
19 bits/s

  1. Bits per selection (Wolpaw formula)

    B = log2(N) + P*log2(P) + (1-P)*log2((1-P)/(N-1))
  = log2(30) + 1*log2(1) + 0*log2(0/29)
  = 4.907 bits / selection

    Term 1 is the information if every choice were correct; terms 2-3 subtract the bits lost to the error rate, assumed spread evenly over the other N-1 targets.

  2. Selections per second

    T = 0.2553 s/selection  ->  1 / 0.2553 = 3.917 selections/s

  3. Information transfer rate

    ITR = B * selections/s = 4.907 * 3.917 = 19.22 bits/s

Other score types

Bounds the atlas keeps out of the default strictest headline: as-reported figures, alternate task conditions, or raw-channel ceilings that shouldn't win the headline by default. Each still carries a score type, so the home-page selector ranks this entry on it when you choose that type. Read its derivation before comparing across entries.

Nuyujukian Recomputed
Nuyujukian achieved bitrate over the raw character set
Achieved-bitrate view of the character channel, shown for comparison
19 bits/s

  1. Achieved-bitrate credit per net-correct character

    N = 30 → log2(N − 1) = log2(29) = 4.86 bits per net-correct selection (field-standard achieved bitrate, e.g. Webgrid; Nuyujukian 2015, which introduced the metric, used log2(N)).

  2. Net-correct character rate

    Experts left most errors in, so the 47 wpm already reflects them and no separate accuracy is applied: this is a perfect-copy ceiling with net-correct = 47 wpm × 5 = 235 char/min = 3.92 correct/s.

    Same N and character interval (0.2553 s) as the entry's perfect-copy Wolpaw ceiling. With no error term the achieved and Wolpaw ceilings differ only by log2(N − 1) vs log2(N), so both land near 19 bits/s: the uniform-prior character channel, above the 3.92 bits/s Shannon headline.

  3. Achieved bitrate

    4.86 bits × 3.92 correct/s = 19.0 bits/s.

Source

Authors
Lyons, Plaisted & Starner
Publication
IEEE ISWC 2004, 2004
Paper
10.1109/ISWC.2004.19
Reference
System date: HandyKey Corporation Twiddler brochure, copyright January 1991