A study compared two groups of second grade children (one receiving code-emphasis reading instruction and the other receiving eclectic reading instruction) on their ability to generate target words deleted from sentence frames, given minimal graphemic information. Target words were either a high-frequency word or low-frequency synonym while sentence frames provided three levels of contextual information. Results suggested a small advantage for code-emphasis subjects. This was primarily noticeable in responses to low-frequency target words, particularly when contextual constraints were weak. The findings indicate that both contextual constraints and word frequency were significant factors in ease of word identification. (TJ)