This study investigated decoding assessment from an onset-rime perspective,
and consistent with an interactive model of cognitive constructs underlyin
g silent reading comprehension. Participants were 128 first and second grad
ers in a public elementary school. Two kinds of decoding items were examine
d: one-syllable words and nonwords, each comprised of a high- to moderate-u
tility onset and a high-utility rime. Data were analyzed mainly with stepwi
se multiple regression and conditional probability analyses. The principal
finding was that the construct validity of decoding items varied depending
on whether they were words or nonwords. Tests of knowledge of onsets and ri
mes accounted for 14% more variance in real-word test than nonword rest sco
res. The superior construct validity of words over nonwords as decoding ite
ms seemed to occur because decoding nonwords requires an additional ability
that decoding real words does not.