How to give a scientific talk

Posted on Wed 18 December 2019 in public speaking, tips, meta, teaching

I have listened to so many talks in so many different fields and topics, that I feel confident enough to provide advice on how to give a proper talk. I have listened to talks ranging from public speaking contests to PhD level seminars. I draw mainly from science talks, but this pretty much applies to any field, really.

0. Prepare your talk

If you do not feel like preparing for your talk, then please just refrain from giving one and save the people from the hassle of listening to a terrible talk. Everybody needs preparation for a talk - their experience may make the preparation quicker, but it is still needed. No preparation, no talk.

1. Define your audience

Who is your talk for? This is the single most important question you need to ask yourself. Who are you speaking for? What background can you expect them to have? Always remember that you are giving a talk for people, so they should be your first focus. Not your work, your presentation, your show - the audience come first! Two presentations on the very same topic can turn out pretty different, if they are aimed at different audiences. If you do not plan for the right audience, you are likely to give a talk that is just wrong (and useless to the world).

The most common mistake is to make the talk too hard and high level with respect to the audience it will be delivered to. This is most usually a lack of confidence, which results in the perceived need of making something difficult to understand so that

  1. people can believe the speaker is very smart;
  2. make people believe that the subject is worth.

The latter is based on the widespread implicit assumption that anything that is hard to understand is worth, which is just bullshit. Beware of anything you do not understand. It is also easy to just copy and paste some formulas or graphs and show those off, without giving away any real understanding.

So, when your talk is ready, revise the material and make sure it can be understood (to a good extent!) by a person having the background you expect from the audience. Most importantly, avoid all jargon unless it is really strictly required. And even then, question your choice of using it. It is fine to use some jargon at advanced talks, just bear in mind that if a listener has to pause and ponder about the meaning some word, you have lost them for the time being. Thus, all jargon used should be sufficiently ingrained in people's head to come without thought.

2. Define what you want to take across

What content are you trying to deliver (if any)? List around 3 points and make sure they are thoroughly covered, and that your audience clearly understands those basic pieces of information. A good measure of success is whether 90% of the audience went home with the ability of re-telling those 3 points with clarity and sufficient detail. If not, then the talk was a failure.

Always make a point of cutting out non-essential stuff. It is true that you might be telling a year worth of work in just 30 minutes, and you may be tempted to detail all the things that make it look like you have done a lot of work, but again ask yourself: is this useful to the audience's understanding? 95% of the times the answer is a sharp no.

3. Bring the material together and tell a story

The fact that you are communicating some objective piece of science does not mean that your talk should be dull. Of course, you should not aim for entertainment for its own sake. But still, strive to link all your material together and build a narrative with it, make clear to the audience how each piece is linked to the others. Aim to be a storyteller.

Also, a good question to cover, from the point of view of the audience, is "why should I care?". Do not just show your work, make it meaningful to the audience.

4. Do not give the problem for granted

All scientific results stem from a question, from an originating problem. Always make sure the audience has the setup clear, and has understood what triggered your piece of work. Never just provide answers, always provide context for them.

5. Realize you are a showman

Remember that, as Galileo put it, "Good teaching is one fourth preparation and three fourths theater." It could not be truer. The moment you start talking, you are on stage, and your thoughts should be much more those of an actor than of a scientist. Your body language, your talking, your engagement with the audience... everything should be much more those of a theater performance than those of a scientist.

6. Stay focused during the talk

One major (and frequent!) mistake is for the speaker not to be focused on the talk during the talk. Any action that results in the audience losing focus is to be avoided.

Some examples:

  • Pausing to look at the watch. You do need to keep track of time, but do so in an unobtrusive manner that does not draw the audience's attention as well!
  • Fiddling with the pointer (e.g. mistakenly go backwards with the slides instead of forward; realizing the pointer does not work well).
  • Digressing on side content, or feeling like you have to elaborate on some minor point. Stick to your main points instead and save the details for questions time. (That is, unless the details carry important insights, but then they should most likely among your main points.)

7. Try the setup in advance

It often happens that the setup fails during the talk, or that you realize something is missing. This also hinders your resolution of staying focused during the talk, and distracts your audience. Make sure the projector works, the pointer works, everything is correctly linked to your laptop, and so on. Go through every detail: is there chalk, if you need it?, and so on. All the atmosphere you have built up to some point is lost if you have to stop for some technical reason.