透過您的圖書館登入
IP:3.138.114.38
  • 學位論文

以語言學特徵擷取技術探討影響聽眾滿意度之表達能力-以四草生態解說為例

Apply Linguistic Feature Extraction Technologies to Discover the Influence of Expressional Ability Factors on Audience Satisfaction – for the Ssu Ts'ao Ecological Narration

指導教授 : 王献章

摘要


語言是知識傳承最主要的工具,而解說在現今的社會中也因此被大家重視,表達能力更是影響其優劣的重要因素。本研究嘗試透過分析解說語料找出具有顯著影響聽眾滿意度的解說特徵因素。 本研究將解說分為專有名詞、形容詞、比喻詞、講話速度、聲調、手勢、贅詞、問答八個部分,分別計算其語言學上之特徵,並透過問卷瞭解聽眾之滿意度,分析解說特徵與聽眾滿意度之間的關係,找出解說時對滿意度有顯著影響之特徵。 本研究收集四草生態解說為測試語料,使用多變量變異數分析得知聽眾對解說者之專業能力的滿意度分數有顯著差異。透過複迴歸分析結果顯示,講話速度與贅詞兩項解說特徵對聽眾所認知的專業能力滿意度有顯著的相關性。實驗結果顯示解說者必須加強其講話速度的控制與贅詞的減少,以提升其專業程度,增加聽眾的滿意度。 因個人口語解說特徵具有較多不確定性,目前本論文在擷取語言特徵時大多採用人工校對之方法,未來之研究方向可嘗試增加特徵項目個數及特徵擷取之自動化來努力。

並列摘要


Language is the main method to pass down knowledge between people. Narration is one of the most powerful tools and the expression ability is an important factor to affect narration quality. Current evaluation criteria for narration expression ability are mostly not automatically. This paper tries to automatically find out principles that affect the audience satisfaction by analyzing the relationship between narrator spontaneous features and satisfaction scores. The narration is decomposed into eight spontaneous features and then we discover the importance for the overall narration quality. The eight features are domain terms, adjectives, analogy terms, oral speed, tone, gesture, stopping words, and interactive dialogues. We also collect satisfaction scores of audience by questionnaires. Using MANOVA analysis, we discover significant difference in the satisfaction about professional knowledge of the narrators. Further analysis conducted by using multiple regression analysis shows that two spontaneous features, i.e., oral speed and stopping words, are related to the satisfaction of the narrator’s expression ability. The experimental result suggests that the narrator should enhance their oral speed and reduce the usage of stopping words to make them more professional. Since there are many uncertainties in the narrations, current feature extraction of some of the spontaneous features are done manually. An interested future work will be to make the feature extraction more convenient and automatically.

參考文獻


[7] 林欣慧,「解說成效對休閒效益體驗之影響研究-以登山健行為例」,國立臺灣師範大學運動休閒與管理研究所碩士論文,2002。
[21] Montoyo A. and Palomar M., “Word Sense Disambiguation with Specification Marks in Unrestricted Texts,” Proceedings of 11th International Workshop on Database and Expert Systems Applications, IEEE, pp.103-107, 2000.
[23] Freeman T., Interpreting our heritage(3rd ed.). The University of North Carolina Press, 1967.
[26] Yarowsky D., “Unsupervised Word Sense Disambiguation rivaling Supervised Method,” Proceedings of the 33rd Annual Meeting of the Association for Computational Linguistics, pp.189-196, 1995.
[1] 李如珍,「高雄都會公園解說志工解說能力養成因素及解說現況之研究」,高雄師範大學環境教育研究所碩士論文,2004。

延伸閱讀