Previous studies [2], [6] have shown that combinations whose modifiers are typically associated with the instantiated relation are interpreted reliably faster than those whose modifiers are less frequently associated with the instantiated relation. Gagné and Shoben attributed this effect to the influence exerted by prior experience of the modifier. They proposed that speakers maintain relation type distributions regarding the frequency with which modifier nouns have been used with each of 16 possible relation types. However, others [4], [5], [7] have claimed that the differences in response times observed in Gagné and Shoben’s study may have arisen due to the nature of the concepts being combined. We contrasted these views by investigating whether these differences persisted when the constituent concepts of a combination were presented individually. We presented noun pairs without a modifier-head syntax; as a result interpretation could not be influenced by previous experience of how those nouns had been used as modifiers in the past. The results revealed that differences in response time remained using either method of presentation. This finding is problematic for the CARIN theory and, as a result, we consider other factors that might influence the difficulty of linking two arbitrary concepts.