<![CDATA[Detecting deception in institutional interrogations remains a significant challenge for law enforcement and investigative agencies. This study investigates the relationship between cognitive load and deceptive speech in anti-corruption interrogations through a modified Activation–Decision–Construction Model (ADCM) that incorporates iterative cognitive processing. The modified ADCM posits that truth-telling and deception engage qualitatively distinct cognitive pathways: Direct Truth Path (rapid veridical information access), Strategic Lie Path (deliberate false-narrative construction), and Iterative Lie-Construction Path (multiple cycles of cognitive revision and backtracking). We hypothesized that these pathways would be distinguishable through acoustic-linguistic markers reflecting differential cognitive load. The study analyzed 102 question-answer segments from 15 speakers across 12 authentic anti-corruption interrogations in Indonesian. For each segment, we measured normalized fundamental frequency (F0), pause frequency and duration, reaction time (RT), and assigned ADCM-phase classifications via temporal coding. Segments were independently verified as Likely Truthful, Likely Deceptive, or Ambiguous based on corroborating evidence obtained during investigations. Results demonstrate that Likely Deceptive segments exhibited significantly elevated F0 (mean z-score +0.52 vs. –0.38 for truthful; d = 0.90), increased pause frequency (3.8 vs. 1.5 pauses; t = 8.34, p < .001), prolonged total pause duration (2.1 vs. 0.6 seconds; t = 9.12, p < .001), and extended RT (2.8 vs. 1.2 seconds; t = 7.89, p < .001). The Iterative Lie-Construction Path predominated in deceptive segments (47%), while the Direct Truth Path dominated truthful segments (76%). These findings provide strong empirical support for the modified ADCM and demonstrate that cognitive load in deceptive speech is reliably indexed through convergent acoustic-linguistic markers. The results have implications for evidence-based interrogation practice and deception detection in institutional settings]]>
Copyrights © 2025
