Electrophysiological evidence of children's statistical learning in Chinese orthographic-semantic connections

Objectives: In the past few years, there have been many demonstrations suggesting that there are multiple cognitive mechanisms underlying statistical learning process in linguistic acquisition, such as learning orthographic regularities in artificial Chinese characters. Apart from pure orthographic regularities, Chinese characters which consist of semantic radicals possess semantic regularities embedded in their orthographic structures. Although previous studies have provided evidence that children could accommodate Chinese orthographic-semantic connections and apply them to the subsequent character encoding, the neural process and developmental change of orthographic statistical learning of Chinese semantic radicals remain unknown in children. We will address these gaps using event-related potentials.

Theoretical framework: Multiple cognitive mechanisms are engaged in Chinese children’s statistical learning of orthographic regularities in semantic radicals. However, according to the developmental invariance account, this complex learning process is not influenced by age.

Methods: Ninety-two children across first to third grade participated in a statistical learning task. Before completing the discrimination test, participants were shown a sequence of artificial characters with different semantic consistency. Each was accompanied by a picture representing its meaning.

Materials: The semantic consistency, which reflects the frequency of a semantic radical representing a specific meaning, was manipulated into high (100%), moderate (80%), and low (60%) consistency levels. At the high semantic consistency level, all pictures represented meanings in the same semantic category (e.g., the five sense organs). For the moderate semantic consistency level, pictures demonstrated two related semantic categories (e.g., animals in four pictures; a specific animal body part in one picture). At the low semantic consistency level, three pictures depicted types of food, whereas two depicted food containers (e.g., bowl and cup).

Results: Electrophysiological activities were compared among three consistency levels in four ERP components, that is, P1, N170, and P300 at the occipitotemporal area, and N400 in the frontoparietal region. The results support the multicomponent model that the statistical learning mechanisms involve two systems: an unconscious learning process attributed to neural adaptation in N170 and an explicit learning mechanism based on the attentional control in P1 and P300. Furthermore, these learning mechanisms observed in Chinese school-aged children may be related to the recognition and extraction of different representations in the early stage of statistical learning (P1 and N170), while the P300 and N400 components may reflect semantic classification in the late stage. The nonsignificant interaction between grade and semantic consistency suggested developmental invariance in orthographic statistical learning, with similar neural patterns observed from 7- to 10-year-olds.

Significance: The neural indicators associated with attentional control might contribute to monitoring the acquisition of Chinese characters when children are taught to learn a series of words containing semantic information. Copyright © 2023 AERA.