ity, then a linear projection is probably reasonable: children learn a certain number of new vocabulary words a year, and there is no strong reason to believe, for example, that during some years they can only learn fewer and during other years can learn more. But the depth-of-processing dimension is limited by children’s cognitive capacities—their theory of mind level, their capacities to take others’ perspectives, to coordinate multiple perspectives, and to distinguish belief from knowledge. We would need to call on the knowledge built up from basic research in cognitive development in order to establish reasonable expectations about children’s capacities for deeper processing of texts. Elaborations of that work might well be helpful in deciding at what age the majority of children would be most susceptible to being taught about multiple perspectives in text, or about the use of textual features to raise doubts or questions, or other subtleties of processing.
Thus benchmarks for comprehension, while not themselves a simple matter of drawing conclusions from research findings, could be deeply informed by a set of research activities that considered both basic cognitive development and the array of standards identified by various groups with a wide array of interests and experiences. Establishing an initial set of benchmarks, and developing assessments of them, would help improve outcomes in reading comprehension simply by proffering a common understanding of what needs to be taught and learned. The initial set of benchmarks should be subject to regular review and recalibration. In fact, the comprehension-instruction agenda outlined above and the benchmarking agenda sketched here should be constantly confronting one another. As improved instruction in reading comprehension raises learners’ capacities, the benchmarks can be ratcheted up to ensure that the proper balance between high standards and opportunities to succeed is maintained.