不同策略及訪談順序下的測謊言語指標的適用性

儘管有許多測謊言語指標的相關研究，但大多研究並沒有排除每個人的說話風格。而少數分析同一人說謊與說實話的研究，皆為意見表達，而非能推論到司法實務應用的模擬犯罪實驗。因此本研究讓參與者經歷兩個模擬犯罪事件以找出可以用來測謊的言語指標。我們也進一步探討了不同策略如何影響測謊言語指標的可用性。另外，我們額外分析了訪談順序對測謊指標可用性的影響。本研究分析了53位台大學生的謊言和實話中十個由認知負荷理論、接近性理論及現實監測理論推得的詞類出現之頻率差異。發現有五個詞類可作為測謊指標：填充贅詞、第一人稱單數代名詞、第三人稱單數代名詞、空間詞、時間詞。本研究發現參與者是否使用非言語策略、仿真策略及訪談順序會影響可用指標。而是否使用言語策略則不會影響測謊指標。這可能是因為使用不同策略，對認知資源有不同影響。本研究提供一個可類推到高張情境使用的測謊言語指標，且分析了在參與者使用不同策略及在不同時間點說謊對指標的可用性的影響，司法實務工作者可在適合情境下使用這些指標測謊。文末討論了本文限制與未來相關研究可以如何改進。

關鍵字

言語指標；測謊；模擬犯罪；策略

並列摘要

Despite the many studies about verbal cues to deception in recent years, very few of them successfully addressed the severe confounding variable, communicational style. Even if they did, these studies used paradigms that participants merely falsely and truthfully express their opinions rather than experiences. Therefore, the current study investigated the differences between lies and truths from the same participants by the frequencies of ten theory-based word categories. To ensure our results can be applied in high-stake contexts, we made our participants experience two mock crimes and then being investigated. In light to present studies, we investigated the effect of lying strategies on the frequencies of word categories. We also explored how the lie position affect the verbal cues to deception. 53 participants were included in this study. 5 categories demonstrated significant differences between truths and lies: Filler words, first and three personal pronouns, space words and time words. Further, the effectiveness of some word categories varied regarding to whether participants applying noverbal, approach the truth strategies and the lie position. However, results remained the same regardless of using verbal strategies or not. This may due to different demands of cognitive resources when using different strategies and lying in different accounts. The results provide a reliable approach that may enhance individuals’ lie-detecting ability in high-stake scenarios. Most importantly, they may apply suitable cues depend on different contexts. Some limitations and suggestions to future studies were also discussed.