9...

The paper is concerned with the type of contexts that surround words and how these contexts affect the learning of unknown words. The paper informs that it is traditionally accepted that vocabulary learning is most effective using context-based as opposed to definition-based and other word acquisition methods. Two types of contexts are referred to: pedagogical contexts, which are created with the intent of illuminating unknown words, and natural contexts, which are less directed. Natural contexts are said, by the authors, to constitute most of the reading material used in vocabulary development. The authors describe a hypothesis they generated regarding the latitude of effectiveness of natural contexts, speculating that on one end of a “continuum” the contexts would be effectively unhelpful in word acquisition, and on the other end, the contexts would be “directive.” They report that the hypothesis was tested and the data supported the notion of the continuum. Strategies are presented to supplement vocabulary learning, as it is noted that these are likely to benefit less skilled readers, those “most in need of vocabulary development.” The place of teachers is discussed, with suggestions made on how to aid students’ vocabulary development, and, in concluding, the paper makes recommendations about how vocabulary development in general can be more effective—by the use of pedagogical contexts and by the supplementary methods discussed.
If the authors’ experiment is to be taken seriously, and it ought to be, the general idea of context categories is particularly relevant to the CVA project as it raises the issue of how, by design, SNePS exhibits a “sense” of context differentiation. It also raises the issue of how to reconcile the authors’ suggestion that pedagogical contexts be the primary source for vocabulary development, and the idea of the intuitively natural contexts of STEM texts. In the event that the authors’ leanings in this regard, however well-informed, are impractical, the development of a curriculum based on the computational theory underlying the SNePS approach offers an alternative. This is to say that whatever the shortcomings of the contexts available in STEM texts, with the implications concerning reader comprehension, SNePS may well provide a way to counter any resulting ill-effects. As for the goal of creating AI systems that operate independent of human assistance, there is little need to be concerned about the types of contexts available, as these systems would most likely not be involved in tasks where they are privileged to choose preferred contexts.

a "word" is a graph with a root letter and an arithmetic series of probabilities with in abc space... context is constrained by how relevant any output is to its counterpart in tension... so... whose shit is better?