Handwriting recognition technology, children, and the writing process

Abstract

The research reported here investigates the match between child, technology and task in the scenario of children using handwriting recognition software and pen technology to construct writing. Children are an interesting emerging user group who have different requirements from technology to adults. Handwriting recognition technology is one of a range of novel input technologies that has not been widely investigated, and writing is a task that is known to be difficult for children, and for
which children use computers in schools.
The research was mostly carried out with children aged between six and ten who were in state education. The focus was an investigation of the usability of handwriting recognition technology for use with children, specifically as a replacement for the
QWERTY keyboard during the writing process. Specific aims were to determine whether the technology could be used in this way, to identify the usability problems that might arise and to suggest some guidelines for developers who might be making
pen-based products for children. The research was also aiming to contribute to knowledge on design and evaluation with children, to add to the literature on the acceptance of errors in recognition-based interfaces for children, to explore the methods that were used for evaluating recognition-based interfaces for text entry and to identify possible future directions for the use of digital text and digital ink to support writing
The thesis is that handwriting recognition can be used by children for text input. The thesis document reports a series of empirical studies that identify that the children were able to use the technology, that the rates for recognition were better than expected in most cases, and that the children liked using the pen and tablet. The main usability problems for the child, technology, and task are documented, and a set of design guidelines, that describe some methods by which the usability problems can be overcome, is included. A list of requirements for a recognition-based interface is presented; many of these have been implemented in CobWeb, a prototype-writing environment.
The way that children dealt with the errors at the interface is explored, and a tolerance figure for the number of acceptable errors is established. A new taxonomy of errors within the recognition interface is produced, and design solutions are presented for the different types of error. Options for the design of appropriate training for the handwriting recognition interface are explored and some of the difficulties that children have with the interface are examined by looking at the mental models that the children have of the technology and the interface.
The work concludes with a discussion of the potential for digital ink for writing and the identification of some areas that might be further developed; these include extensions to the prototype, further work on error handling and work on the design of
evaluation studies for handwriting recognition.