Typing as a window into chunking in language: top-down effects from multiword units

Abstract Top-down effects of larger-grained linguistic chunks on their smaller-grained constituent parts have been established in both reading and speaking. However, typing as a domain of language production has been less thoroughly investigated in this regard. In the current paper, we present a copy task in which participants were shown a stimulus and asked to type it. Their keystrokes were recorded, allowing insight into both typing fluency (in interkey intervals, or IKIs) and latency to typing onset (in response times, or RTs). Critically, stimuli varied in both lexical status (words vs. non-words) and collocational status (frequently co-occurring vs. novel word pairs). As expected, non-words were reacted to and typed more slowly than words. At the group level, collocated word pairs were initiated and typed slightly faster than non-collocated pairs, but this effect was not statistically significant. However, evidence emerged for considerable individual differences in the trade-off between RTs and IKIs, suggesting that typers differ in the stage at which they benefit from top-down facilitation when typing collocated word pairs. This complements previous research on top-down effects and is consistent with the view that the mental processing blocks supporting written language production and comprehension may align—though the extent and timing of such alignment appear to vary across individuals.