Speech following sign language training in autistic children with minimal verbal language.

The team worked with autistic children who had almost no words.

They compared four ways to teach new labels: speech only, sign only, sign plus speech together, or alternating sign and speech.

Kids were randomly placed in one group and taught the same set of items.

Sessions ran until each child met a mastery rule.

Any plan that included spoken models led to more child-initiated speech than sign alone.

Children who could already imitate sounds before training added the most new spoken words.

The study found positive results, but did not report exact counts or effect sizes.

Romanowich et al. (2013) later ran a larger RCT and found medium language gains, updating this 1988 work.

That newer study also showed that receptive language level guides your choice of method, something the 1988 paper did not test.

Gabriels et al. (2001) and Plant et al. (2007) echoed the imitation cue: kids with stronger early imitation or bigger sound inventories make faster speech progress.

Matson et al. (2011) used a different path—extinguishing signed requests—to spark new vocalizations, giving you a second tool if speech stalls.

If a minimally verbal client is stuck at the sign stage, add spoken models right away.

Start by testing simple sound imitation; if it is weak, build that skill first.

You now have two ways to evoke speech: pair signs with voice, or briefly withhold reinforced signs while prompting vocals.