The psychological reality of the morphological difference between compounds (teacup) and single words (crescent) is controversial. Decomposition accounts vary regarding time-course (early vs. late) and presence of constraints such as novelty and length. We investigated decomposition in compounds using magnetoencephalography (MEG) in a visual lexical decision paradigm comparing compounds (CW) and single words (SW).
Monolingual English speakers responded to 60 disyllabic noun-noun CW, 60 pairwise-matched SW, and 120 nonwords including 12 word-nonword foils (boxshep) while a 160-channel whole-head MEG system (Kanazawa Institute of Technology, Japan) recorded brain signals.
Response time (RT) reflects both early and late processes, whereas the MEG component around 350ms post-onset (M350) reflects lexical activation. While the CW and SW were pairwise-matched for overall frequency (mean log-frequency 0.455) and letter-length (mean 7.8 letters), the first and second CW constituents had higher log-frequency (mean log-frequency 1.96, both for first and for second constituents) and shorter letter-length (first constituent mean letter-length 3.82, second constituent mean letter-length 4.0). Consequently, early decomposition predicts earlier M350 and faster RT, reflecting constituent over whole-word properties. Late decomposition predicts faster RT but not earlier M350. Given lexicalized CW with short constituents, lexicalization/length constraints predict neither RT nor M350 differences. Whole-word-only accounts likewise predict no differences.
Faster M350 latency (341ms vs. 375ms) and RT (682ms vs. 740ms) for CW vs. SW supported early decomposition. Comparison of the lowest-frequency CW vs. SW items showed faster M350 for CW (340ms vs. 387ms), but not faster RT (761ms vs. 755ms), predicted by reduced post-access facilitation from constituents. Among the highest-frequency items, however, both M350 (344ms vs. 364ms) and RT (635ms vs. 663ms) were faster. M350 patterned with overall frequency only within SW, and constituent frequency within CW.