A Spellchecker for Dyslexia
Luz Rello, Miguel Ballesteros and Jeffrey P. Bigham
The 17th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2015)
Lisbon, Portugal, October 26-28, 2015
Abstract
Poor spelling is one of the main challenges that people with dyslexia face through all their lives. Spellcheckers are there fore a crucial tool for people with dyslexia, but current spellcheckers do not detect real-word errors, which are a common error type for people with dyslexia. Real-word errors are spelling mistakes that result in an existing word, for instance, form instead of from. Nearly 20% of the errors that people with dyslexia make are real-word errors. In this paper we present a method to detect and correct real-word errors using Google n-grams along with a probabilistic language model and a statistical dependency parser. To test the accuracy of our system we carried out an evaluation by using real examples written by people with dyslexia. To test its competitiveness we compared our system against widely used spellcheckers. To evaluate its usefulness we carried out an experiment with 34 people, 17 of them with dyslexia. Our results show that people with dyslexia corrected sentences significantly more accurately and in less time when they use the output of the system.
START
Conference Manager (V2.61.0 - Rev. 3862)