Talk abstract:   

Morphemes are the smallest linguistic units that carry meaning (e.g., a complex word such as artist is composed of a stem, art-, and a suffix, -ist). Behavioural evidence suggests that complex words are processed via their constituent morphemes and that reading development benefits from the morphological structure of words and especially from the presence of stems. However, the neural underpinnings of morphological processing are still not fully understood. The present study combined MEG recordings with fast periodic visual stimulation (FPVS), a highly sensitive and behaviour-free approach, in order to investigate selective neural responses to morphological structure in both developing and skilled readers.
In an FPVS-oddball design, native English-speaking children (N=17, grade 5-6) and adults (N=28) were presented with rapid streams of base stimuli (6 Hz) interleaved periodically with oddballs (every 5 items). If present, MEG response at the oddball stimulation frequency (6/5=1.2 Hz) would index neural discrimination. Stimuli consisted of 4 types of constructs: stem+suffix (e.g., softity), stem+non-suffix (e.g., softert), non-stem+suffix (e.g., terpity), and non-stem+non-suffix (e.g., terpert). The contrast between oddball and base stimuli was manipulated to probe selective stem (e.g., stem+suffix embedded in non-stem+suffix) and suffix (e.g., stem+suffix embedded in stem+non-suffix) identification in morphologically structured pseudowords. MEG data were analysed in the frequency domain.
Preliminary sensor-level analyses revealed a neural response at the oddball frequency (1.2 Hz) and its harmonics (2.4, 3.6, 4.8 Hz), across children (p=.009) and adults (p=.05, approaching significance), in a left ventral occipito-temporal region of interest (ROI), reflecting successful suffix identification. Some evidence of stem detection also emerged in the same ROI.
In particular, the morpheme detection response was elicited in the conditions shared by the adults and children, with oddball stimuli that were entirely made up of existing morphemes (e.g., softity). In two additional conditions administered to the adult sample only, instead, there was no reliable response to oddball stimuli that only included an affix (e.g., terpity) or a stem (e.g., softert) alone.
Results provide initial neural evidence for automatic morpheme identification, even at relatively early stages of reading development, in line with all major localist accounts of morphological decomposition. Critically, these findings also suggest that morpheme identification can be modulated by the context in which the morphemes appear: a reliable detection response emerged when pseudowords were fully decomposable into existing stems and suffixes. Furthermore, cluster-based permutation analyses will complement the theory-driven ROI analysis presented here, thus revealing any additional sources of morpheme detection in the sensor space. Finally, further ongoing analyses aim at projecting the response signal at source level, where in the brain the relevant computations are likely to occur.