Training in word recognition based on a child's dominant sensory modality (visual, auditory, or kinesthetic) was compared with training based on nondominant modality. Twenty-five first-grade children with no prior instruction in reading were given visual, auditory, and kinesthetic modalities tests and a pretest of recognition of the words used in the experimental training. In addition, all subjects experienced three 1-week experimental training periods, one for each modality, with each period including tests of immediate and delayed recall of words taught. Due to the lack of any available suitable group learning-modalities test, the author developed one which proved to be unreliable. However, further analysis indicated that the experimental training test procedure was actually itself a reliable learning-modalities test. Author-made tests, tables, and references are given. (Author/DE)