Generalization of object naming after training with picture cards and with objects.

Delamater et al. (1986) compared two ways to teach object naming to children with intellectual disability. One group learned with real objects. The other group learned with picture cards.

After training, the team tested if each child could name new objects they had never seen. This check showed whether learning transferred beyond the teaching set.

Kids who practiced with real objects named more untrained items correctly. Kids who saw only picture cards rarely named the real thing when it appeared.

When some picture-trained children failed, the trainers switched to real objects. Adding several different examples during retraining fixed the problem.

Rosenthal et al. (1980) had already shown that showing all items together beats one-at-a-time presentation. The 1986 study adds the next layer: the items themselves should be real, not flat photos.

Sprague et al. (1984) taught vending machine use and found that training with several varied examples created the best generalization. The 1986 naming study mirrors this: more exemplars and real objects beat single pictures.

Mazur (1983) taught low-functioning children to communicate with picture cards and saw wide generalization. That success with pictures seems to clash with the 1986 result, but the tasks differ. E taught requesting; J taught labeling. Requesting can thrive on pictures, while naming real objects needs the real object.

If you want a child to name things in the real world, put the actual item on the table. Picture cards can save space, but they are a weak bridge to the three-dimensional world. Start with objects, then add variety. When generalization stalls, rotate several similar objects during the same lesson. This small shift can save weeks of retraining.