| Abstract: | 運用英語語境線??推敲字義?僅幫助學習者輕鬆閱?，?能促進在字彙上的學習。雖然針對外語學習者之國內外相關研究在這二、三十?間逐漸受到重視，然而在臺灣，相關研究也僅聚焦於大學生（Li, 2004）或小部分的語境線?（Arden-Close, 1993; Li, 2004）。研究者亦發現，除?Yu（2006）的研究，全面探討13 種英語語境線?的研究至今仍付之闕如。因此，本研究旨在檢視臺灣高中生對於文章上下文的推?能?與字彙?之相關性為何？進而探討高中生運用英語語境線??推敲字義的能?。 本研究對象為中部地區某國?高中的 1,526 位學生，研究工具包含一千字高頻字彙測驗（Huang, 1999）、二千字高頻字彙測驗（Schmitt et al., 2001）、以及改編自Yu（2006）的語境線?測驗。在實施語境線?測驗之前，研究者使用高頻字彙測驗作為過?工具以確保研究對象字彙?充足。本研究呈現研究工具的敘述統計與信?、以及語境線?難?表。研究者使用統計軟體SPSS 17.0 for Windows ?分析所蒐集之資?。 研究結果顯示：工具的信?檢測介於.80 ~ .94，顯示本研究使用之測驗具有?好的信?。此外，在至少認?95%字彙的情況下，大部分的研究對象?能適當地使用各種語境線?來推敲字義，而其整體表現亦呈現出顯著的跨?級差?，換言之，即高?級學生表現優於低?級學生。長期以?國內外學者普遍認為，推敲字義成功?與學習者字彙?多寡有直接相關，然而本研究結果卻發現?者間缺乏線性、比?關係。是以，學習者?能具備約750 至850 的字彙?即可幫助他們成功地推敲字義；然而，當字彙?大於900 字時，對於提升推敲字義成功?效果卻相當有限，原因在於?如與世界、學科主題及特定策?等相關認知，亦會影響學習者推敲字義的表現。再者，語境線?難?表顯示出研究對象較能掌握?想（Association）、系?字組（Words connected or in series ）、因果關係（ Cause and effect ）、定義/ 敘述（Definition/description）、指示?照（Referral）與比較/對比（Comparison/contrast）等?項線??別，同義字（Synonym）與非限定用法子?（Non-restrictive clauses）則較具挑戰性。最後，根據研究結?，分別對教學者、出版業者、學習者及未?研究方向提出相關建議。
Using embedded context clues to infer meanings of unknown words facilitates reading and vocabulary learning. However, studies on context clues utilization of L2 learners to word meaning retention and text comprehension have been piled up only in the past two to three decades. With studies done in Taiwan, however, the focuses have been either on college students (Li, 2004) or on partially-selected context clue types (Arden-Close, 1993; Li, 2004). Little empirical research has been done to cover the full range of context clue types, except that done by Yu (2006). This study, therefore, aimed to examine the relationship between contextual inferencing and vocabulary size of high school students in Taiwan and investigate high school students’ ability to use embedded context clues to infer unknown word meanings in text. The participants of the current study were 1,526 students from a national senior high school in central Taiwan and the instruments included the 1,000 Vocabulary Levels Test (VLT) by Huang (1999), the 2,000 VLT by Schmitt et al. (2001), and 39-item Context Clues Tests (CCTs) adapted from Yu's (2006) 78-item CCT. Before the participants took the CCTs, they were screened by the two VLTs to ensure their vocabulary ability to be sufficient. Descriptive statistics and reliability of the instruments, and the difficulty rank order of clue types were presented. English SPSS 17.0 for Windows was employed to perform all the statistical procedures. The reliability estimates of the instruments reached .80 ~ .94, indicating that the tests were reliable instruments. Most of the participants had appropriate knowledge of the use of various context clues to infer meanings of unknown words when at least 95% of the running words were known to them. In addition, cross-grade differences were found in the participants' inferencing performance, with the senior participants performing significantly better than the junior ones. Contradictory to the long-held assumption that inferencing success is directly related to the amount of vocabulary known by a learner, the results proved that inferencing success did not linearly and proportionally correlate with vocabulary size. A vocabulary size of around 750 to 850 (out of 1,000) was sufficient for successful inferencing while reading. A vocabulary size larger than 900 words did not greatly increase the successful inferencing rate. Other factors (such as general knowledge of the world, knowledge of the subject matter, and knowledge about specific strategies) also affected the participants' inferencing performance. The difficulty rank order of the 13 clue types revealed that Association, Words connected or in series, Cause and effect, Definition/description, Referral, and Comparison/contrast did not seem to pose a challenge to the participants, while Synonym and Non-restrictive clauses were considered more difficult. Suggestions for the language instructors, textbook writers, and language learners are discussed. Future research based on different data collection approaches on diverse subjects is proposed. |
