The bestersell effect: Nuances in positional encoding of morphemes in visual word recognition

Previous studies have confirmed stem morphemes (e.g., book) are identified in any position (e.g., in both bookmark and textbook) but prefixes and suffixes (e.g., re- in replay and -er in player) cannot be recognized when moved from their typical word-initial or word-final locations. However, English words with multiple affixes (e.g., unresolved, mindfulness) suggest there must be further nuance to the positional constraints imposed on affixes in the reading system to facilitate cases where affixes occur in atypical locations but still convey meaning. We used two lexical decision experiments (N = 90 native English-speaking participants each) to investigate the positional encoding of mid-embedded suffixes. In Experiment 1, transposed tri-morphemic nonwords ending in a chain of two suffixes (e.g., spitenessful [derived from spitefulness]), and transposed nonwords with string-initial suffixes (e.g., fulyouthness [derived from youthfulness]) were compared against orthographic controls (e.g., spitementdom/domyouthment). In Experiment 2, transposed tri-morphemic nonwords ending in a stem (e.g., bestersell [derived from bestseller]) and transposed nonwords with string-initial suffixes (e.g., erwalksleep [derived from sleepwalker]) were compared against orthographic controls (e.g., bestalsell/enwalksleep). Across both experiments, the results revealed a significantly larger morpheme transposition effect relative to controls for the mid-embedded compared with the string-initial suffix conditions. Items like bestersell activated the corresponding lexical representation of "bestseller" and made it more difficult to reject the target nonword, revealing that suffixes are not as strictly positionally encoded as previously assumed. These findings challenge existing predictions of positional requirements for affixes and provide evidence calling for more nuanced theoretical models of morphological processing.