The Age of Acquisition (AoA) effect is such that words acquired early in life are processed more quickly than later-acquired words. One explanation for the AoA effects is the arbitrary mapping hypothesis (Ellis & Lambon-Ralph, 2000), stating that the AoA effects are more likely to occur in items that have an arbitrary, rather than a systematic, nature between input and output. Previous behavioural findings have shown that the AoA effects are larger in pictorial than word items. However, no behavioural studies have attempted to directly assess the AoA effects in relation to the connections between representations. In the first two experiments, 48 participants completed a word-picture verification task (Experiments 1A and 2A), together with a spoken (Experiment 1B) or written (Experiment 2B) picture naming task. In the third and fourth experiments, 48 participants complete a picture-word verification task (Experiments 3A and 4A), together with a spoken (Experiment 3B) or written (Experiment 4B) word naming task. For each pair of experiments the subtraction of the naming latencies from the verification tasks for each item per participant was calculated (Experiments 1-4C; e.g. Santiago, Mackay, Palma & Rho, 2000). Results showed that early-acquired items were responded to more quickly than late-acquired ones for all experiments, except for Experiment 3B (spoken word naming) where the AoA effect was shown for only low-frequency words. In addition, the subtraction results for pictorial stimuli demonstrated strong AoA effects. This strengthens the case for the AM hypothesis, also suggesting the AoA effect resides in the connections between representations.