Peter J. Kennedy

Development of a Digram-Based Typing Key Layout for Single-Finger/Stylus Input

Hand-held devices often include a keyboard on which a user types with one finger or a stylus. When a user types with one finger, conventional key layouts are not optimal for typing throughput. We used a predictive human performance model based on the frequency of English-language digrams and interkey distances to evaluate alternative key layouts. We also analyzed English digrams with a path-analysis program to determine the strongest links among English letters. The path analysis combined with the predictive human performance model provided a basis for developing and evaluating typing-key layouts for single-finger input. A nonstandard typing-key layout (in a roughly 5 times 5 key matrix) based on digram analysis should, after the development of skilled use, be about 27% better than the QWERTY layout. A nonstandard alphabetic typing-key layout (in a roughly 5 times 5 key matrix) should be about 13% better than the QWERTY layout for single-finger entry.

EVALUATION OF TYPING KEY LAYOUTS FOR STYLUS INPUT

We studied initial user performance with and preference for alternative typing-key layouts for typing with a stylus. Previous analyses have indicated that, after asymptotic practice with this kind of typing, users should type fastest with certain nonstandard layouts. Although asymptotic performance is important in selecting a layout, it also is important to evaluate initial performance with and preference for layouts. Twelve participants used paper models of six different layouts to type four sentences. The layouts were the QWERTY, Dvorak, standard alphabetic, square alphabetic, and two digraph-based layouts. Performance and preference favored the QWERTY over the other layouts. The square alphabetic layout took a distant second place. QWERTY should be a products default layout even if users can type with only a single finger or a stylus. If the QWERTY layout is not feasible, designers should consider using the square alphabetic layout for limited typing or a digraph-based layout for extensive typing.

Handheld Electronic Devices

From PDAs to cell phones to MP3 players, handheld electronic devices are ubiquitous. Human factors engineers and designers have a need to remain informed about advances in research on user interface design for this class of devices. This review provides human factors research summaries and research-based guidelines for the design of handheld devices. The major topics include anthropometry (fitting the device to the hand), input (types of device control and methods for data entry), output (display design), interaction design (one-handed use, scrolling, menu design, image manipulation, and using the mobile Web), and data sharing (among users, devices, and networks). Thus, this review covers the key aspects of the design of handheld devices, from the design of the physical form of the device through its hardware and software, including its behavior in networks.

CRT Typeface Design and Evaluation

Reading text from a video display screen is a relatively new modality of human communication. This new technology presents advantages and limitations never considered in the design of typefaces for printed communication. It is well known that people read faster from printed text than from a Cathode Ray Tube (CRT). This paper discusses some of the current work to improve CRT readability. Specifically, we have developed software to be used by a typography designer to design anti-aliased (or “grayscale”) CRT typefaces. A new anti-aliased typeface was designed and tested for readability. It was found that readability was significantly improved with this new typeface. This paper describes some of our recent work in this area. The typeface design software and examples of text rendered in a new typeface will be demonstrated on an analog color display.

Automated Data Collection

Human factors organizations frequently are requested to provide recommendations about design choices to a product design team. These recommendations should be based on operator performance data. However, in many cases the required information may not be present in the current literature and time and personnel resources may be limited. This paper describes how this problem has been solved by the Human Factors group in IBM at Boca Raton. We are using the IBM Personal Computer as a process controller for studies in our lab. This low cost tool facilitates the automation of studies which require the recording of data for subsequent analysis. The Personal Computer has also been used to analyze the data.

Development and Testing of the Operator Training Package for a Small Computer System

The training package for the System/23 Datamaster was developed in accordance with human factors principles and an interactive testing strategy. The viability of the training was assessed by observations, interviews, and post-training task assignments, using surrogate operators. Revisions were incorporated into the training materials and testing was repeated. The result of this interactive testing process was a highly successful training package.

A Method of Analyzing Personal Computer Use in an Application Environment

This paper describes a program for the IBM Personal Computer (PC) which runs in the background to record and time-stamp keystrokes as operators use real PC applications. The value of the program as a human factors research tool is discussed. In addition to this, an example of a study using the program to estimate the frequency of acquisition of the Backspace and Enter keys in PC applications and a comparison of two IBM keyboards are described.