40 students aged 9-11 predicted their spelling performance using a 3-point scale (yes, I can; maybe I can; no, I can't spell it) before spelling 18 medium-difficulty words. 1/2 hour later students looked at each word and judged its relative correctness on the 3-point scale they used to make predictions (yes, it is; maybe it is; no, it is not correct). In the second phase of the study, 3 experimental groups "looked" at their spellings under different conditions: silent pronunciation, out loud pronunciation by student, or teacher pronunciation. A control group simply predicted and wrote spellings again. Results showed that students' predictions were significantly correlated with their actual spelling accuracy. The number of "maybe" words decreased in the "looking" phase, when evaluations of these words became more definite and accurate. Looking at a word after spelling it, compared to just writing it again, led to more accurate self-evaluations. Results are discussed in terms of the kinds of spelling inform...