The easiest way to do that would to be to create your words as graphics--each the same size. Place the names of those images in a stim file and draw them randomly on each trial. This would be done by listing them as sequential stimuli, e.g., $word, $word, $word, in the case that you have three coloured word images. Make sure that in the location (LOC) column of each that you use a clear value of "0", e.g., ".75,.5,0" if you want the words to appear three quarters of the way down the screen and centered horizontally. The final "0" value is the clear value and tells DirectRT to preserve whatever is on the screen when it displays the picture. Since the word pictures are all the same size, each will overwrite the other but the remainder of the screen will appear constant--i.e., your origional picture will remain onscreen the whole time.

Finally, to have the words change automatically, use a "!" in the min-max value for the style you are applying to the trial. For example, to have the words switch every 500ms, use a min-max value of "0-500!"

How you proceed from there will depend on whether or not you want them to see all the colours or if you want the trial to end if and when they respond.

Hope that helps,
Blair