H567/03 Section B Option 1 - Wood et al. (1976) and improving learning

The tutorial question and scaffolding

Problem solving requires combining component skills to achieve a goal. Recognising an appropriate solution can be easier than producing all the actions needed to make it. A child may recognise a correct construction but be unable to select and coordinate its pieces independently.

The researchers asked how a skilled adult could assist with those difficult elements while allowing the child to perform achievable ones. They described scaffolding as controlling parts initially beyond the learner's competence so the learner can concentrate on manageable actions. Eventually the aim is independent competence, rather than merely an adult-assisted finished product.

Their investigation sought systematic descriptions of a tutorial interaction, including responses to different forms of assistance. They explicitly cautioned against treating it as a hypothesis test of the whole tutoring process. Comprehension preceding production was an important interpretation: meaningful feedback requires some recognition of the intended result, although exploration can also help the learner discover the goal.

Sample, observation and the pyramid task

Thirty children attended individual sessions lasting 20 minutes to an hour. There were ten children aged three, ten aged four and ten aged five, with five girls and five boys in each group. Their parents answered volunteer advertisements and lived within five miles of Cambridge, Massachusetts. Families were predominantly middle or lower-middle class, and parents accompanied the children.

This was a controlled observation in an arranged research setting at Harvard. Age groups were naturally occurring; children were not assigned an age. All worked with one tutor, Gail Ross, the third author, on a specified task. There was no randomly assigned no-tutor group. Age differences in these observations therefore do not isolate maturation or establish the effect of tutoring alone.

The purpose-designed wooden toy had 21 blocks forming a pyramid about nine inches high with a nine-inch-square base. Its six levels comprised a single top block and five four-piece layers. Each layer involved two interlocking pairs. Peg-and-hole connections could fit in misleading arrangements, while correct orientation formed larger circular connections needed to stack layers. Simply attaching pieces was not always a correct construction.

The task was chosen to be attractive, challenging, varied and manageable without excessive strength or manipulation demands. Its repeated structure allowed an earlier solution to help with later layers. These design choices also limit transfer: the results concern this particular task and an encouraging adult interaction, not every school subject or learner.

Tutoring procedure and behavioural coding

The child sat at a small table with the blocks in a jumble and had about five minutes of free play. The tutor then demonstrated a small correct pair, or used a pair the child had already made, and invited similar constructions. She aimed to let the child do as much as possible, normally offering verbal instruction before more direct intervention.

The next response depended on what the child did. If the child ignored the task, the tutor offered constrained material or a further example. If a child selecting pieces overlooked a feature, she drew attention verbally to the discrepancy. When she had supplied the construction material herself, she corrected resulting errors herself. Otherwise she allowed the child to continue, intervening when construction stopped or difficulty arose. Her appreciative approach encouraged participation, although shifting from imaginative play to the prescribed pyramid sometimes reduced enthusiasm.

Coding distinguished manipulation of separate pieces from existing assemblies, assisted from self-selected constructions, correct from mismatched outcomes, rejection of errors, and disassembly followed by reassembly. Assisted meant that the tutor presented or specifically indicated material; it was not simply any action in the tutor's presence.

Every intervention was classified as direct assistance, a verbal error prompt or a general verbal direction to continue the task. Subsequent child behaviour was coded. Two independent scorers agreed on 94% of 594 videotaped events. This provides measured evidence of coding consistency for that event pool, rather than proof of perfect validity or agreement for every possible judgement.

Findings and the six scaffolding functions

Overall activity was similar: median acts were 39, 41 and 32 for the three-, four- and five-year-olds respectively, with no significant overall age-group difference reported. The composition of activity differed. No three-year-old made a correct four-block assembly, whereas every four- and five-year-old did so at least once. Older children selected matching pieces more readily. Younger children could nevertheless recognise appropriate assemblies, shown by their greater likelihood of restoring correct than incorrect constructions they had taken apart.