Compares effects of teaching words in either semantic random groupings. Finds that the treatment had very strong effects, but student performance was not affected by whether the words were grouped in categories. Suggests that, given rich and elaborate instruction, semantic grouping is not necessary. (RS)