Success of engineered systems and devices is contingent upon their approval by expert committees. For example, advisory panels to the Food and Drug Administration make crucial recommendations regarding the approval and diffusion of medical devices in the United States. Systems and devices are subject to opinions and institutional bias that might drive decision outcomes in an unfavorable direction. Computational linguistics algorithms, such as Latent Semantic Analysis, provide a quantitative metric for the dynamics of the multi-stakeholder decision making underlying approval meetings. Insights from the analysis can lead to recommendations for how such committees might be structured to improve system performance with minimal cost and rework.