此外，本研究欲進一步探討以非中文為母語的外籍受試者處理中文形音對應關係時的表現，以及大腦活化機制。因此，本研究招募了12位以中文為母語的本籍受試者及12位以拼音文字為母語的外籍中文學習受試者參與實驗三的核磁共振造影的研究。同時，本實驗也測量了受試者的順序性視覺統計學習能力、空間性統計學習能力中的特徵組合統計學習、隱性特徵組合統計學習、智商以及語音短期記憶。 結果發現相較於本籍受試者，外籍受試者進行形音轉換時，中文識字量與左腦梭狀回(fusiform gyrus)的活化程度有顯著相關，同時他們的視覺空間特徵組合統計學習能力與中文音旁一致性敏感度有顯著相關，此效果顯示於左腦中的梭狀回(fusiform gyrus)以及前額葉內側中回 (medial frontal gyrus)的活化。
;Recent research has shown that statistical learning (SL) ability has consistently correlated with literacy acquisition of alphabetic (e.g., Hebrew) and non-alphabetic (e.g., Chinese) languages alike. However, various forms of SL, such as spatial visual SL (SpaVSL), sequential visual SL (SeqVSL), and sequential auditory SL (ASL), have seldom been compared side-by-side, and whether these aspects contribute to literacy acquisition equally remains to be clarified.
To explore these issues, 197 alphabetic language speakers learning Chinese as a second language were recruited in our Experiment 1, with 85 among these participants were recruited in Experiment 2. Participants’ various SL abilities, as well as their intelligence quotient (IQ), verbal short-term memory (vSTM), and Chinese literacy proficiency were measured. As the majority of modern Chinese characters are phonograms, which are typically composed of a semantic and a phonetic radical on the left and right side of a character, respectively, we hypothesized that SpaVSL, particularly the sensitivity to combinatorial regularities (CVSL), would exhibit a stronger correlation with the Chinese literacy indices than SeqVSL. Our Experiment 1 results confirmed the relationship between SeqVSL and Chinese literacy measurements previously found, but showed no correlation between CVSL or ASL with Chinese literacy measurements. In Experiment 2, in addition to showing that participants who continued to learn Chinese in the past few months improved more than those who stopped to take formal lessons of Chinese, we confirmed results from Experiment 1 and previous findings for SeqVSL in predicting Chinese literacy improvement for the latter group. However, the effects of CVSL remained insignificant. A predictive effect of ASL in the participants who stopped to take formal Chinese lessons was found, contrary to previous findings. Overall, SeqVSL was a stronger predictor than CVSL in our cross-sectional and longitudinal investigations.
In Experiment 3, we recruited 12 native Chinese readers and 12 non-native Chinese readers to participate in a homophone judgment task in fMRI to investigate their sensitivity to consistency of Chinese pseudo phonograms. We also measured their SeqVSL, SpaVSL, and CVSL in addition to their IQ and vSTM. The process through which participants made homophone judgments on the pseudo phonograms is orthography-phonology conversion (OPC) in Chinese characters Due to the varying consistency information of Chinese phonetic radicals, we inferred that this process is very similar in nature to implicit learning, which can be measured via SL. Results revealed that in non-native participants, their character size was correlated with the activations at the left fusiform gyrus. For the consistency-sensitivity related effect, a significant correlation was found between SpaVSL with the left fusiform gyrus and the left medial frontal gyrus.
Our findings confirm that SL is indeed a valid predictor of second language acquisition of Chinese character recognition, and suggest that 1) different SL may be differentially correlated to various aspects of language acquisition, and 2) in addition to SeqVSL, SpaVSL may be a crucial predictor of processing efficiency of Chinese character consistency in non-native beginning learners of Chinese.