In a character decision task, phonetic compound targets (composed of a sema
ntic radical and a phonetic component) followed primes that shared (a) the
target's radical and were semantically related (R+S+), (b) the target's rad
ical and were not semantically related (R+S-), (c) no radical but were sema
ntically related (R-S+), and (d) no radical and were not semantically relat
ed (R-S-). Target radicals also varied as to the number of compounds in whi
ch they appeared (i.e., combinability). When targets followed primes immedi
ately (Experiment 1; SOA 243 ms), target latencies following R+S- primes we
re slowed relative to R-S- controls but those following R+S+ and R-S+ prime
s were facilitated equivalently. Increases in combinability significantly r
educed decision latencies. When 10 items separated primes and targets (Expe
riment 2), facilitation was evident only after R+S+ primes. Results indicat
e that one type of component, the semantic radical, is processed in the cou
rse of Chinese character recognition and that orthographic similarity due t
o repetition of a radical is not an adequate account. (C) 1999 Academic Pre
ss.