GOOD READERS DON'T GUESS‐TAKING ANOTHER LOOK AT THE ISSUE OF WHETHER CHILDREN READ WORDS BETTER IN CONTEXT OR IN ISOLATION

The notion that we read words more easily in story context than in isolation owes much to the influence of a study by Goodman (1965), who found a 60‐80 per cent improvement in reading accuracy when children read words in context rather than in isolation. This study has been widely cited in the reading literature, even though its validity has at times been questioned. In the present .study, Goodman's findings were re‐examined by conducting three experiments. The first involved asking 60 children, all eight years of age, to read stories which varied in difficulty and contextual completeness, and within which target words occurred in varying sentence position. The second experiment, using easy stories, sought to find out whether the results for the first experiment would also be similar for ten good and ten poor readers, also about 8 years of age. The third experiment, involving a further 8 good and 8 poor readers, at the same age level, extended the search for context effects by using difficult stories as well. The results of this series of experiments suggest that Goodman was wrong about the effects of context on word recognition. What seems to separate good readers from poor is the ability to decode words independently of context.     

Surface cues reduce the latency to name rotated images of objects.

Jolicoeur (1985, Memory & Cognition 13 289-303) found a linear increase in the latency to name line drawings of objects rotated (0 degrees to 120 degrees) from the upright (0 degrees) in the initial trial block. This effect was much shallower in later blocks. He proposed that the initial effect may indicate that mental rotation is the default process for recognising rotated objects, and that the decrease in this effect, seen with practice, may reflect the increased use of learned orientation-invariant features. Initially, we were interested in whether object-colour associations that may be learned during the initial block, could account for the reduced latency to name rotated objects, seen in later blocks. In experiment 1 we used full-cue colour images of objects that depicted colour and other surface cues. Surprisingly, given that Jolicoeur's findings were replicated several times with line drawings, we found that even the initial linear trend in naming latency was shallow. We replicated this result in follow-up experiments. In contrast, when we used less-realistic depictions of the same objects that had fewer visual cues (ie line drawings, coloured drawings, greyscale images), the results were comparable to those of Jolicoeur. Also, the initial linear trends were steeper for these depictions than for full-cue colour images. The results suggest that, when multiple surface cues are available in the image, mental rotation may not be the default recognition process.     

Children's use of verbal collaterals in DRL performance

This study examines the effects of a verbal collateral (counting) on a series of three differential reinforcement of low rates of response (DRL) tasks with progressively longer interresponse time requirements (DRL-2, DRL-6 and DRL-12). Forty-eight 10-year-old children, divided according to sex and conceptual tempo (fast-inaccurate vs. slow-accurate) participated in DRL training, half of them being instructed to count aloud between responses. Counting was found to be related to faster learning of all the tasks and to greater efficiency in obtaining reinforcement in most cases. It also eliminated differences due to sex and conceptual tempo except in DRL-12. However, it had the disadvantage of causing inexact temporal discrimination. The results were replicated in a group of 8-year-old subjects performing the same tasks.