How many words can we read at once? More intervenor effects in masked priming

It is argued that the existence of masked translation priming from L1 to L2 with a 50 ms prime implies that processing of the prime must continue well after it has been replaced by the target, since it is estimated that the meaning of a word is not established until at least 120 ms after stimulus onset. This fact implies that the lexical processor must be equipped to handle two words simultaneously. However when a masked word intervenes between the prime and the target, three words must be processed simultaneously. Under these conditions, form priming is eliminated altogether, and identity priming is reduced, suggesting that the capacity of the lexical processor does not extend to three words. Four experiments are reported showing that this disruption of priming only occurs when the intervenor triggers lexical processing. It is argued that the differential effect of the intervenor on identity and form priming can be explained on the assumption that priming takes place at the level of form, and at the level of meaning. As support for this interpretation, it is shown that an identity prime is capable of generating a congruence effect in a semantic categorization experiment despite the presence of a masked intervenor.