Recombinative generalization of within-syllable units in prereading children.

Three kindergarten kids played a computer game. They picked printed words that shared letters, like cat, cap, and can. The game gave praise for right picks and a buzz for wrong ones.

After only one set of these overlapping CVC words, the team tested the kids on brand-new words built from the same letters. No one taught these new words directly.

All three kids could now pick the new words they had never seen. This is called recombinative generalization. The skill showed up fast and without extra teaching.

The result means the children had learned the building blocks of words, not just the trained items.

Lillie et al. (2019) got the same kind of generative leap, but with adults learning braille. Their learners built braille dots from print cues and then read untrained braille signs. Both studies show that equivalence training lets people create new correct responses after tiny amounts of instruction.

DeRoma et al. (2004) also saw reading generalization, yet they used direct instruction and repeated reading. Their child improved oral fluency across new passages. The 2000 paper adds that even simpler computer drills can spark generalization at the single-word level before fluency work begins.

Stricker et al. (2024) delivered online relational training to whole classes and lifted standardized reading scores. The 2000 finding helps explain why: once kids can recombine letters, later comprehension and fluency programs have a stronger base to build on.

You can plant the seeds of reading in minutes. Run a quick computerized match-to-sample game with overlapping CVC words. Watch for untrained words the child picks correctly. If generalization appears, you have proof the learner is acquiring letter-sound units, not just memorizing. Move on to longer words or passages with confidence that the equivalence engine is already running.