The processing of letter-position information in randomly arranged con
sonant strings was investigated using a masked prime variant of the al
phabetic decision (letter/nonletter classification) task. In Experimen
t 1, primes were uppercase consonant trigrams (e.g., FMH) and targets
were two uppercase Xs accompanied by the target letter or a nonletter
(e.g., XMX, X%X). Response times were systematically faster when targe
t letters were present in the prime string than when target letters we
re not present in the prime string. These constituent letter-priming e
ffects were significantly stronger when the target letter appeared in
the same position in the prime and target stimuli. This contrast betwe
en position-specific and position-independent priming was accentuated
when subjects responded only when all the characters in the target str
ing were letters (multiple alphabetic decision) in Experiments 2 and 3
. in Experiment 4, when prime exposure duration was varied, it was fou
nd that position-specific priming develops earlier than position-indep
endent priming. Finally, Experiment 5 ruled out a perceptual-matching
interpretation of these results. An interpretation is offered in terms
of position-specific and position-independent letter-detector units i
n an interactive-activation framework.