Several studies at the beginning of the century demonstrated that extensive practice enables subjects to adequately perform two language tasks presented simultaneously. This result was recently confirmed by Hirst et al. These authors concluded that practice eventually results in the ability to attend to both tasks simultaneously, since evidence shows that both primary and secondary tasks are processed semantically. Two arguments are presented showing that this conclusion is questionable. First, the processing of category and semantic relationships between words and sentences can be automatic. Second, the conclusion that true divided attention is possible after training can only be made if evaluated against several criteria of automatic performance. Hirst et al. reported data pertaining to only one of these criteria. Therefore this study is inconclusive and fails to elucidate the nature of the changes in processing that occur.