Dynamic time warping: A new method in the study of poor handwriting

Abstract

Poor handwriting is a diagnostic criterion for developmental coordination disorder. Typical of poor handwriting is its low overall quality and the high variability of the spatial characteristics of the letters, usually assessed with a subjective handwriting scale. Recently, Dynamic Time Warping (DTW), a technique originally developed for speech recognition, was introduced for pattern recognition in handwriting. The present study evaluates its application to analyze poor handwriting. Forty children attending Dutch mainstream primary schools were recruited and based on their scores on the Concise Evaluation Scale for Children's Handwriting (Dutch abbreviation: BHK), 20 good and 20 poor writers (of whom 13 were scheduled for handwriting intervention) were identified. The groups were matched for age (7-9 years), school grade (grades 2 and 3) and handedness. The children subsequently wrote sequences of the letter "a" on a graphics tablet in three conditions (normal, fast, and accurate). Classical kinematics were obtained and for each individual letter DTW was used to calculate the distance from the mean shape. The DTW data revealed much higher variability in the letter forms of the poor writers that was independent of the kinematic results of larger trajectories, faster movements, and higher pen pressure. The current results suggest that DTW is a valid and objective technique for letter-form analysis in handwriting and may hence be useful to evaluate the rehabilitation treatments of children suffering from poor handwriting. In education research it may be exploited to explore how children (should) learn to write.