Audio Narration vs. Onscreen Text

Updated 7/30/2007 based on looking at Audio Narration vs. On-Screen Text issues. I had read about the related issues in Clark and Mayer - Science of Instruction and found a good summary article at - http://www.adobe.com/products/captivate/pdfs/captivate_leveraging_multimedia.pdf

Good post by Will Thalheimer on what stops us from using more research in our designs.

It seems like a big part of the reason that we don't use research is that we often disagree on what the research really says and how it applies to us in particular situations.

There was a recent debate on a newsgroup (Brandon-Hall AT Group) around audio narration vs. on-screen text. After much debate, there was some agreement on what the research says, but still a lot of debate about the specific situation. I.e., no one could come in and say definitively that the research says to do X.

What I took away was:

1. For learners who will not be getting audio, you need to have an option where the narration can be read. This can be an alternative delivery mode.

2. The best result when using text + audio was when the text was short blurbs that reinforced key points - single words best.

3. Having the narration text on-screen has mixed results in studies and does not appear to add value except in cases such as non-native speakers. Many users find it annoying because the narrator speaks slower than they read and will turn off the audio.

4. Having long text passages on screen that are supposed to be read simultaneous with DIFFERENT audio narration is highly distracting and is detrimental to learning.

5. (Clark & Mayer) Use first and second person language and learning agents where possible.

Learners will be more engaged and learn more by using "you" and "we" in the learning. Further, establish a person who is talking to you and directing the learning. According to Clark & Mayer - the appearance and style (realistic, avatar) doesn't make much difference in learning - of course, it does need to match with the audience. These "learning agents" should speak via audio narration, use a conversational tone and have familiar human accents. On-screen text or machine-generated audio does not work as well.