| Abstract: | 先前的研究結果顯示，統計學習與其他認知能力不同，並且與語言習得和熟練程度有關。然而，過往研究統計學習的研究，多使用接續出現在時間序列中、具有規則性的刺激材料，和拼音文字系統的構字規則較為相似。目前文獻中用來研究統計學習的實驗派典，並沒有利用和非拼音文字系統(如：中文)較為相似之刺激材料，因此統計學習和語言處理之間的關係，是否能在此類測驗中獲得支持，仍有待確定。此外，許多關於統計學習的研究使用了語言刺激材料，而過往研究已有證據指出對這些刺激材料的處理會受到人們語言經驗的影響。有鑑於此，我們需要開發一種新的統計學習測驗，採用非語言刺激、並仿照中文文字系統的特徵，以利有效地檢驗統計學習能力與中文閱讀能力之間的關係。
在本研究中，我們開發了新的”成分規則之視覺統計學習”（Componential Visual Statistical Learning，C-VSL）測驗，並以行為實驗（實驗一和二）和神經電生理學實驗（實驗三）探索此測驗的特徵。具體而言，在時間版本和空間版本的C-VSL測驗中，非語言形式的形狀分別以接續呈現以及同時呈現的方式成對出現：在學習階段，某些形狀總是出現在時序或者空間配對中的特定位置，而同一配對中的另一個形狀則非特定的形狀；在測試階段，每個在學習階段出現過的形狀會與之前沒有遇過的新形狀配對，出現在和學習階段相同或相異的位置。實驗參與者對於時間和空間版本C-VSL測驗中刺激材料之位置規則性的敏感度，則透過測試階段的熟悉度判斷和再認作業來測量。同時，實驗一和二測量參與者的基本認知能力，包括智商（IQ）和言語/非語言工作記憶，以及他們在使用視覺和聽覺刺激材料的序列統計學習測驗中的表現，以研究這些表現與C-VSL能力間的相關性。此外，實驗一和實驗二的參與者包括拼音文字系統和非拼音文字系統(中文)的母語使用者，因此我們在實驗二中利用標準詞彙判斷作業測量這些參與者對中文構字規則的敏感度，也測量所有參與者對非母語的熟練程度，以探索識字能力和成分規則之視覺統計學習能力對彼此的效果。在實驗三中，我們使用ERP研究工具，試圖找出和時序、空間之成分位置規則性敏感度相關的神經機制。
本研究的實驗結果顯示，我們新開發的C-VSL測驗能夠有效地測量實驗參與者對時序和空間刺激材料中位置規則性的敏感度；儘管這兩種敏感度彼此具有顯著的相關性，但這些能力與序列統計學習測驗或IQ測驗的分數並無顯著的相關。此外，拼音文字系統的讀者在兩種C-VSL測驗和序列統計學習測驗中的表現，與他們對中文構字規則的掌握程度有關；這與我們實驗室先前所觀察到的結果一致，支持統計學習能力與識字習得歷程之間的關係。另一方面，在以中文為母語的實驗參與者身上，沒有觀察到任何統計學習測驗的表現與其語言表現間的相關，這可能是由於這些參與者在語言測驗中的得分沒有太大變異性。至於和成分位置規則敏感度相關的神經機制，我們發現腦電波中的N400和MMN效果分別和時序及空間中的此類統計學習能力有關，這些結果也與之前的事件相關電位文獻一致。
總結來說，我們成功地開發了新的統計學習測驗(即C-VSL)，可用於測量實驗參與者對非語言刺激材料在時序和空間配對中之位置規則性的敏感度，而這樣的能力並不同於在先前文獻中常被檢視的序列統計學習能力。儘管在時間和空間C-VSL測驗中所測量到的行為表現之間存在高度相關性，但腦電波實驗結果顯示這兩種能力受到不同神經機制的支持。透過本研究所開發出的C-VSL測驗，學者可在未來研究中進一步探討統計學習能力與中文閱讀能力之間的關係。
;Previous findings have shown that statistical learning (SL) is distinct from other cognitive abilities, and is related to language acquisition and proficiency. However, the close relationship between SL and language processing is mainly supported by the findings from tests that require extraction of regularity embedded within temporal arrays, which are similar to alphabetic writing systems. Whether such relationship can be identified in tests whose stimuli are similar to non-alphabetic writing systems, such as Chinese, remains to be determined. Furthermore, a significant proportion of the research on SL had used verbal stimuli, which have been shown to be under the influence of people’s language experience. To investigate the relationship between SL and Chinese literacy optimally, there is a need to create and to validate a new SL test that employs non-verbal stimuli to resemble the characteristics of Chinese orthography.
In the present study, we developed novel componential visual SL (C-VSL) tests, and examined their characteristics in behavioral (Experiment 1 and 2) and neurophysiological (Experiment 3) experiments. Specifically, in the temporal and spatial C-VSL tests two nonverbal shapes were presented in sequential and simultaneous pairs, respectively. In the study phase, each shape always appeared in a specific position in the temporal or spatial pair, while the other shape in the same pair was not specific or unique. In the test phase, each shape encountered in the study phase, either in the same or different position as in the study phase, was paired with a novel shape that was not encountered before. Participants’ sensitivity to the stimulus positional regularity in the temporal and spatial C-VSL tests was measured by familiarity judgment and recognition in the test phase. In Experiment 1 and 2, participants’ basic cognitive abilities, including IQ and verbal/nonverbal working memory, and their performance in the conventional SL tests in both the visual and auditory modalities were measured to investigate their correlation with the C-VSL abilities. In addition, we recruited alphabetic and logographic readers for both Experiment 1 and 2, and in Experiment 2 further measured their sensitivity to Chinese orthography in a standard lexical decision task, as well as their proficiency to a non-native language, to explore the effects of literacy and C-VSL abilities on each other. In Experiment 3, the neural correlates of the sensitivity to the temporal and spatial componential regularity were identified using the ERP methodology.
The results showed that the newly developed temporal and spatial C-VSL tests are valid in measuring participants’ sensitivity to positional dependency. The behavioral accuracies in the temporal and spatial C-VSL tests correlated with each other but not with the performance in the conventional SL tests or with IQ. Consistent with the previous results from our laboratory, alphabetic readers’ performance in the tests of sequential VSL, temporal C-VSL, and spatial C-VSL correlated with their knowledge of novel Chinese orthography, supporting the relationship between SL and literacy acquisition. On the other hand, performance of none of the SL tests correlated with the language performance reliably in Taiwanese participants, possibly due to the low variability in their scores in the linguistic tests. As for the neurophysiological correlates of the performance in the temporal and spatial C-VSL tests, a significant N400 component and a sparse MMN component were identified, respectively. These results are generally in line with the previous ERP literature.
In summary, we have successfully developed SL tests that measure participants’ sensitivity to positional regularity presented in temporal and spatial pairs of nonverbal stimuli. The abilities measured in the C-VSL tests is distinct from the abilities measured in the sequential SL tests that are commonly employed in the previous literature. Despite the high correlation between the abilities measured in the temporal and spatial C-VSL tests, the ERP results suggest that these two abilities are subserved by distinct neural mechanisms. Based on the C-VSL tests developed in the present study, the relationship between SL and Chinese literacy can be explored further in future research. |
