Early-acquired words are processed faster than late-acquired words. This is a well-accepted effect within the word recognition literature. Different explanations have been proposed, either localizing the effect of age of acquisition (AoA) in a particular substage of word processing or seeing it as the result of the way in which information is stored and accessed in the brain in general. The cumulative-frequency hypothesis is an example of the latter type of explanation: it states that the total number of times a system has come across a particular stimulus will determine the speed with which the stimulus can be recognized. The present multi-task investigation provides a critical test of the different explanations. Results show that in a variety of word processing tasks the effects of frequency and AoA are highly correlated, and that the impact of AoA is consistently higher than would be expected on the basis of the cumulative-frequency hypothesis. The findings are interpreted as evidence for recent demonstrations of a loss of plasticity in neural networks due to training and/or for mathematical models that describe the growth of the lexico-semantic network as the attachment of new nodes to existing nodes.