Early access effects in English compound and pseudo-compound words

Although there is substantial experimental evidence that the morphemic constituents of compound words (e.g., snowball) are activated during compound word access, it is unclear exactly how the presence of constituents impacts word access. A series of experiments using a masked repetition primed lexical decision task investigates the role played by the morphology of compounds in word access. Semantically transparent compound words show consistent advantages relative to their frequency- and length-matched non-compound controls, but opaque compound words do not. For both kinds of compounds, the effect of repetition priming is the same for the compounds and their controls at short prime durations (50 and 100 ms). However, at long prime durations (300 ms), the compounds show more priming than their controls. In short, the compound advantage appears to be independent of the facilitation provided by short duration primes, but affected by long duration primes, and it appears to depend on the semantic transparency of the compound. Pseudo-compound words, such as carpet, provide an interesting comparison to compounds, because the language system cannot, a priori, determine whether they are compounds or not. Pseudo-compounds appear to be more difficult to process than their controls, and at long prime durations they show less priming than their controls. These results suggest that the compound advantage in processing arises relatively late in processing and is sensitive to the match between the semantics/morphology of a constructed compound interpretation and the required whole word.