Verbal communication is not only about what is being said, the verbal content, but also how it is being said, the paraverbal dimension of verbal communication. We examine quantitatively how the polarity, the emotive positive-negative dimension, may differ across the content and paraverbal dimensions. Using earnings calls, we examine how Top Management Teams (TMTs) communicate when their organization is facing financial distress. As they may be strongly affected by the organization facing distress, we expect that negative emotions ‘leak’ via their paraverbal communication. Using a variety of state-of-the-art machine learning models, we determine the topics of conversation of the earnings calls and the polarity of the content and paraverbal dimensions. Our preliminary results seem to suggest that indeed there is a discrepancy in the emotive loading of what and how TMT members communicate. This implies without taking into account the paraverbal dimension, inaccuracies in research on verbal communication which solely uses transcripts could be widespread. Also, within industry, the accuracy of understanding managerial communication is nontrivial. A faulty interpretation due to a lack of taking into account how the TMT communicates could result in failing to spot investment opportunities, or instead committing too much to a failed cause.