Teaching children with autism spectrum disorder to ask “where” questions using a speech‐generating device

Three preschoolers with autism used an iPad that spoke for them. The team broke a favorite play routine by hiding a key piece. They then taught the kids to press "Where is [item]?" on the device to get it back.

Training used constant time delay: wait three seconds, then point to the button if the child did nothing. Sessions ran until each child asked the question without help in five different play chains.

All three kids learned the new mand in 9–14 sessions. They kept using it with new adults and new rooms. But when the toy changed, only one child asked right away; the others needed extra teaching with each new chain.

The 1992 review by M et al. looked at 36 studies and found constant time delay works for many discrete skills. Carnett’s study adds a new box to that list: asking "where" on a speech device.

Lincoln et al. (1988) compared constant time delay with system-of-least-prompts for teaching numeral names. Their kids reached mastery faster than Carnett’s, but they only had to name a card, not start a whole question. The extra time here likely comes from the harder task: forming a sentence on a screen.

Eugenia Gras et al. (2003) showed that reinforcing every small step plus using a request ladder helps kids follow directions. Carnett used the same idea—reinforce the mand and break the chain in small steps—to build language instead of compliance.

If you use SGDs, hide one piece of a loved toy and wait. Program the exact "Where is ___?" button ahead of time. Teach at least three different chains so the form transfers. Check that the missing item really matters to the child; the mand only works if the motivation is strong.