The experts taught specific strategies that were intended to be deliberately applied when encountering an unfamiliar word. These range from graphic organizers to the use of cloze exercises. An example is the Jenkins, Matlock, and Slocum study, which compared the effects of two approaches to vocabulary instruction: individual word meanings and deriving word meaning with varying amounts of instruction and practice. They measured effects of these two approaches both on learning individual words and learning from context. The deriving-meaning group was taught a general strategy for deriving meaning of unknown words that emphasized using context clues. They developed this strategy after talking through the process the authors themselves used. Jenkins et al. described this general strategy with the acronym SCANR (Substitute a word or expression for the unknown word; Check the context for clues that support your idea; Ask if substitution fits all context clues; Need a new idea? Revise your idea to fit the context). Lessons were taught using a script, with each step taught, assessed, and reinforced. Students received either much practice (270 words over 20 sessions) or little practice (45 words over 9 sessions). A second group received instruction in specific word meanings. Jenkins et al. found significant effects for both treatment and amount of practice on multiple-choice measures as well as measures of words in context, with the group receiving instruction in specific word meanings performing better on measures of knowledge of those words and the group given the strategy instruction performing better on measures of deriving words from text. Other studies taught more generic strategies, without specific algorithms. Buikema, teachers, developed a 5-day sequence of whole-group teaching learners to use context clues. She began with word riddles and used a riddle metaphor to help her students derive words from context. They found significant effects for the treatment on two measures of deriving words from context and on a measure of incidental learning using an Edgar Allen Poe short story followed by an unexpected vocabulary test. More recent approaches have moved away from specific lists of processes to a more general modeling of learning words from context. An example is Goerss et all’s (1994) training program, in which students were given general outlines on how to derive word meanings from context and were told to think aloud while reading texts with nonsense words inserted in a one-on-one tutoring session. The instruction involved general delicateness to unknown words and the use of both knowledge of the overall context and specific cues within the text. They taught their students that it was not necessary to derive full definitions, but to get whatever information was attainable about a word from that context. They found growth in learning from context among the students with which they worked, but they did not use a control group. It is not possible to appraise whatever students would have made similar progress with unguided practice alone.
Thus, the trend in this area of research has moved from providing clear taxonomies of context clues, to providing cognitive strategies and instructing students in how to use them flexibly, and finally, to providing more general guidelines and modeling. But, are any of these approaches effective in teaching learners to become more efficient at using context? The remainder of this article addresses that question.
There are several major flaws common to research. First is the identification of a control group. Studies compared students who were taught a general strategy for using context clues with students learning specific words or with students who received no attitude in this area. There are problems with both types of comparisons, which we outline below. Rather than continuing with this type of design, we suggest that a practice-only treatment in addition to a no treatment control is likely to prepare the most useful comparison. A second flaw is the measures used to assess learner’s knowledge of word meanings. For example, metalinguistic measures, such as verbalization of the process of deriving word meanings or using definitions, may not be useful as measures of implied knowledge, such as identification measures or even checklists, would be. These are discussed in turn.