近年來由於中國的崛起，全世界有越來越多人把中文當作第二外語學習，在臺灣由於社會形式改變，外籍配偶從2002年的23萬人，至今已超過了44萬人次。依內政部規定，這些新移民必須通過考試，並且達到一定的識字量，才可取得中華民國國籍，這些現象顯示了漢語識字學習需求的重要。然而，把華語作為第二外語與把英文作為第二外語的學習者相比，華語在認識一個新的漢字上需要形音義三方面結合，而英文的單字只要音跟義結合即可，可以說學習漢字的成本比學習英文要多費些工夫，且目前市面上漢字識字教材的編撰，並未考慮使用者的學習效率。漢字總共分成六大部分，也就是俗稱的六書，其中在常用字方面，形聲字約佔70%到80%，對於初學者來說掌握形聲字轉音規則，對於學習漢字識字來說，事半功倍，為此我們提出以聲符部件為主的漢字識字教學，再匹配適合的發音規則，用以幫助漢字的學習。本系統類似於以字帶字識字教學，與傳統的教學法不同地方在於，以字帶字識字教學在教導一個部件時，會連帶教導有相同聲符部件且有相似發音的延伸字，在此種教學法裡面，相同聲符會與不同的部首去做結合，產生發音相似的複合字，此種方法的優點在於教導少量部件與少量部首，卻可以認識大量的字，達到聞一知十的學習效率。本文與中央大學中文系合作，從中研院漢字構形資料庫裡面，分析出1453筆常用部件，配合形音義及字詞教材，建立以聲符部件為主之漢字識字線上學習系統，並且經由發音強度、筆畫數及其延伸字出現頻率作為教學順序。最後在模擬中，當學習前400個部件與其延伸字時，對於現今南一版國小課本中的課文與小學生撰寫的小短文或時事新聞，即可達到60%以上的識字率，並在學習前800個部件就可達到9成左右的識字率，遠高於使用傳統式小學課本學習了三年的學習曲線，成效非常驚人。Owing to the rise of China in recent years, an increasing number of people learn Chinese as second language in the world. In Taiwan, the number of foreign spouses from south-easten countries and mainland China has been growing from 230 thousand to 440 thousand in the last decade. In accordance with Ministry of the Interior Affairs’ law, these new immigrants need to pass Chinese tests in order to get R.O.C nationality. This phenomenon shows that learning Chinese is getting more important.For native speakers who learn a language from speaking to reading, learning Chinese as a second language is more difficult than learning English as a second language. Learning a new Chinese character requires the connection among graph, pronunciation and meaning, while learning a new English word requires only the connection between pronunciation and meaning. This means that the cost of learning Chinese is higher than that of learning English. Although there are six categories of Chinese characters, about 70%~80% of Chinese characters are picto-phonetic compounds that are composed of a phonetic component (PC) and semantic component. Therefore, one can make a guess at a character’s pronunciation and meaning from its phonetic and semantic component for a new character. For this reason, we propose an order of phonetic components based on pronunciation strength, frequency and number of strokes for efficient learning with proper pronunciation rules and graph recognition. We adopt stem-deriving instructional method (以字帶字) which extends each phonetic component with different radical component to derive new picto-phonetic compounds of similar pronunciation. Via simulation, the top 400 phonetic components and their picto-phonetic extensions are enough for the recognition of 60% of general articles; and top 800 phonetic components can help 90 percents of general articles.
