What a Meeting Transcript Shows About How You Speak
By Noam Shemla · Data & Method
What can you learn about your speaking from a meeting transcript? Five things are easy to read off it: how much of the meeting you spoke, which filler words you lean on, how often you soften your claims, whether you ask questions or only make statements, and how your important points begin. Compare a few meetings, pick one pattern to change, and check the next transcript to see if it moved.
Most video meeting tools can now produce a transcript automatically. Most people never read their own. That is a shame, because a transcript is one of the few honest records of how you actually speak at work, as opposed to how you think you speak. You do not need special software to learn from it: a transcript, a quiet ten minutes and a highlighter are enough.
Here are five things to look for, roughly in order of how easy they are to see.
1. How much of the meeting you spoke
Scroll through and get a rough sense of your share of the talking. Do you speak for a third of a five-person meeting, or almost not at all? Neither is automatically wrong, but it is worth knowing, because talk time carries weight. In a study of leaderless groups, MacLaren and colleagues found that the amount a person spoke kept a direct effect on whether they emerged as the group's leader, even after accounting for intelligence, personality and gender.1
That is not an argument for talking more for its own sake; it is an argument for noticing when your share is so small that your contribution is not registering. If you were silent in a meeting where you had something to say, the transcript will show it.
2. Your filler words
Search the transcript for "um", "uh", "so", "like", "you know", "basically" and "kind of". Most people have one or two favourites, and they cluster in predictable places: at the start of a turn, and just before a point the speaker is unsure of.
Fillers are not all bad. Herbert Clark and Jean Fox Tree argued that "uh" and "um" work as words in their own right, signalling a short or a longer delay to come.2 The problem is concentration, not existence: a dozen fillers in front of your main recommendation make it sound less sure. Note where yours cluster, not just how many there are. What is filler rate? explains how fillers are counted and why a rate per hundred words is fairer than a raw count.
Be aware that automatic transcripts are inconsistent about fillers: some tools drop them, some keep them. If your transcript has none at all, the tool probably removed them.
3. How often you soften your claims
Look for hedges: "I think", "maybe", "sort of", "I'm not sure but", "this might be a silly question". Highlight every one in a single meeting. A hedge on something genuinely uncertain is honest. A run of them on things you know well is a habit, and it costs credibility. What is hedging rate? sets out how hedges can be counted, and the speech pattern that undercuts you covers the research on what they do to a listener.
4. Questions against statements
Count how many of your turns are questions and how many are statements. Some people only ever state; others only ever ask. Questions, especially follow-up questions that build on what someone just said, do social work: in experiments by Karen Huang and colleagues, people who asked more of them were liked more by the person they were talking to.3 If your transcript shows a meeting full of statements and no questions, try one deliberate follow-up question next time.
5. How your important points begin
Find the two or three moments in the meeting where you said something that mattered: a proposal, a concern, a decision. Read the first ten words of each. Did they lead with the point, or did they wander towards it ("So, I was just thinking, and this might not be relevant, but...")? Rewrite each opening as you wish you had said it. That exercise alone, done for a few meetings, changes how people start their important sentences.
Do it for three meetings, then change one thing
One transcript is a snapshot; three show a pattern. Read three, pick the single pattern that bothers you most, and work on only that for two weeks. Then read the next transcript and see whether it moved. This is the same approach any good feedback loop uses: measure, change one variable, measure again.
What a transcript cannot tell you
A transcript records words, not voice. It cannot show your pace, your pauses, your tone or whether you sounded confident while saying "I think". For that you need the audio, or a tool that measures it. And automatic transcription makes mistakes, especially with accents, technical terms and people talking over each other, so treat any single line with caution. What we actually measure in a talk covers what can and cannot be measured from speech.
References
- MacLaren, N. G., Yammarino, F. J., Dionne, S. D., Sayama, H., Mumford, M. D., Connelly, S., Martin, R. W., Mulhearn, T. J., Todd, E. M., Kulkarni, A., Cao, Y., & Ruark, G. A. (2020). Testing the babble hypothesis: speaking time predicts leader emergence in small groups. The Leadership Quarterly, 31(5), 101409. https://doi.org/10.1016/j.leaqua.2020.101409
- Clark, H. H., & Fox Tree, J. E. (2002). Using uh and um in spontaneous speaking. Cognition, 84(1), 73–111. https://doi.org/10.1016/S0010-0277(02)00017-3
- Huang, K., Yeomans, M., Brooks, A. W., Minson, J., & Gino, F. (2017). It doesn’t hurt to ask: question-asking increases liking. Journal of Personality and Social Psychology, 113(3), 430–452. https://doi.org/10.1037/pspi0000097