Semantic training facilitates the acquisition of novel morphemes: MEG evidence

Viktória Balla, Yury Shtyrov, Miika Leminen, Alina Leminen

Tutkimustuotos: KonferenssimateriaalitPosterivertaisarvioitu


Learning to recognize morphemic boundaries is essential for fluent language use. In languages with rich morphology, the question of morphological learning is particularly relevant. Some neurocognitive models propose that morphologically complex words are parsed and their constituents are stored as separate units in lexical memory. Neural mechanisms underlying the acquisition of such novel morphological units, however, remain largely unexplored.
Here, we trained 19 native Finnish-speaking participants with novel derivational suffixes through a word-picture association task. Following this short training session, we used magnetoencephalography to record the participants’ brain responses to trained and untrained suffixes combined with new real and pseudoword stems in a 22-minute passive listening task. Existing Finnish suffixes served as controls. We compared the responses measured early and late (first/last 5 minutes) during passive exposure to investigate the online build-up of novel suffix representations.
In the left frontal and temporal cortices, source activation for real suffixes was significantly higher than for both trained and untrained novel suffixes in the 60-80 ms, 120-140 ms and 220-260 ms time-windows (around ERF peaks) following the suffix onset, suggesting the activation of pre-existing long-term memory traces. However, we also found increased source amplitudes for the trained as opposed to the untrained suffixes already early in exposure. Although such a brief exposure may not be sufficient for the integration of novel morphological units into lexical memory, our findings suggest that a short semantic training of novel affixes facilitates morphological decomposition and speeds up suffix memory trace formation in the left fronto-temporal language networks.
TilaJulkaistu - syysk. 2019
OKM-julkaisutyyppiEi sovellu
TapahtumaThe 25th AMLaP: Architectures and Mechanisms of Language Processing Conference - Higher School of Economics, Moscow, Venäjä
Kesto: 6 syysk. 20198 syysk. 2019
Konferenssinumero: 25


KonferenssiThe 25th AMLaP: Architectures and Mechanisms of Language Processing Conference
LyhennettäAMLaP 2019


  • 3112 Neurotieteet
  • 6121 Kielitieteet

Siteeraa tätä