Welcome back everyone. This is part four in our series on presenting your work. This is the final screencast in the series. We're going to be talking about giving talks, especially the talk that you might give at an NLP workshop or conference. I should say at the start that I'm going to give a lot of advice about how to give effective talks, but the only advice that I feel really confident in is that you should cultivate your own style as a speaker, and you should do that by giving talks and then really and truly reflecting on how it went, looking for ways you could do better, and then heading off in that direction.

The precise direction that you go is highly individual. This will depend on your goals and your preferences and the style you're cultivating. But I can promise that if you do this act of self-reflection, you will become a better speaker. It can be painful, but it is very productive. This also means that the advice that you do here about giving effective talks should always be filtered through this highly personal thing that you're trying to do.

That said, let's dive in. I'm going to offer a bunch of advice. The first piece of advice I think is pretty straightforward. For the kind of academic talks that we're talking about, it's a good default to assume that the talk structure should mirror the paper structure, but the talk needs to be much simpler.

You absolutely cannot convey all the detail that you packed into a paper. It's just the wrong mode for that. So what you can really hope for is that you find a way of talking about the ideas that conveys their essence, teaches the core lessons, and entices people to actually invest in all of the details that are in your paper.

To do that, just as with the paper, I think you should open with context setting. What problem are you solving? Why is it important? And what's been tried before? And why does it fall short? And if you do this context setting right, the audience will be really well set up to understand the essence of the talk.

That's the middle part. You should do concrete details like what are the data, what are the metrics, things like that. What's the task structure? But really what you want to do is make sure that you convey the essence of your idea. What is your approach? If it's a model or a metric or a task structure or something else, this is where you really shine a light on those ideas.

And that should feel like the essence of the talk, the ideas behind your work. Having done that, you can kind of add some quantitative results to support what you did. You can talk about ablation studies and do error analyses and so forth. It can kind of branch out from there, assuming that you've done the job of really conveying the approach.

And then finally, at the end, you'll want to quickly summarize. But also think ahead to the future. This is your opportunity to convey to people how they might build on the ideas, what the important open questions are, and why the work was ultimately significant. Jeff Pullum has what he calls golden rules for giving talks.

He says there are five of them, but there are actually six. I like all these rules. I especially like five and six, but I thought I would offer all of them to you. The first one is don't ever begin with an apology. I think this is reasonable. I think the concern here is that if you start by apologizing for how bad the talk is going to be because you didn't prepare or whatever, people might believe you.

Their impression of you might be that, in fact, you're about to give a very bad talk, and that could cause them to check out. And it could certainly shape their opinions of the talk they do here. So start strong instead of starting with an apology. Don't ever underestimate the audience's intelligence.

I think the spirit of this is that we've already baked in that we're going to give a simple version of the ideas we're trying to teach a core lesson. And with that in mind, you want to approach the audience as peers, people who might surprise you with new ideas, but certainly people who are equipped to understand what you're trying to do.

And if you're in that mode, I think everyone will feel good and feel connected around what you're trying to do. Respect the time limits. This is nice and concrete, and this is absolutely crucial. It is tragic when people use up all their time doing context setting, and they never get to describe their idea.

The way to respect the time limits is to do a lot of practice. I think there is no shortcut here. You should practice. You should watch yourself practicing, and you should find a way to fit the ideas in in a way that does what you want to do. If you don't practice, you are at serious risk of going way past the time limits.

Don't survey the whole damn field. This aligns well with the advice I gave before. What you want to do is provide context for your ideas, but it should be just the context we need in order to understand the essence of what you've done. And if you cast the net too widely, people will get lost, and you might simply run out of time.

And this one I really like. Remember that you're an advocate, not the defendant. The spirit of this is like you're a lawyer, and you're defending the ideas. You're invested in doing a good job, but you yourself are not on trial in any sense. And so this is like the right level of investment.

It's not so personal, but you are, of course, trying to do the best thing that you can by the ideas that you're presenting. And then finally, expect questions that will floor you. On the assumption that you're talking to peers, very intelligent people, you've prepared for the discussion section. We'll talk about that a bit later.

But even having done that, surprising things can happen. That's part of the reason that we're giving talks for the exploratory aspect, to learn new things. So you should expect questions that are surprising and hard to resolve, maybe questions that even terrify you. That should be considered part of the exploratory educational thing that you are doing.

On top of all of that, I really like to return to Patrick Blackburn's fundamental insight about talks. I mentioned this before in the context of writing. Blackburn intended it to be about talks. It applies everywhere. He asks, where do good talks come from? And he says, honesty. A good talk should never stray far from simple, honest communication.

I like to think of this in the context of education. The goal of the talk is not to broadcast ideas or to advertise, but rather to educate. And if you approach it from the perspective of genuinely trying to teach your ideas to your audience, I think you'll be in the right place.

You'll connect with them. And they'll take those further steps of checking out the paper and more deeply understanding what you've done. People joke about PowerPoint. And they worry about the kind of distorting effects that it can have on the way we present our ideas. I think it's reasonable to be a bit worried.

But obviously, I'm not too worried because this entire course is based in screencasts where I'm using slides, PowerPoint-style things. But we do need to be thoughtful about how we use this device. As you read about advice for talks, you see kind of two camps emerge, the minimalist and the comparative.

The minimalist view is that your slide should be as spare as possible. The audience should spend most of the time listening to you and looking at you. Individual slides don't stay up for long. And they don't get used in more than one way. They're kind of just punctuation to the talk track.

You could compare that with the comparative view. Your slide should be as full as possible without sacrificing clarity. Your talk should make it easy for people to spend time studying your slides. Individual slides stay up for a long time and get used to make multiple comparisons and establish numerous connections.

I'll talk a little bit in a second about how I think these approaches compare to each other, how this relates to style and to your goals. The one thing I will say here now is that whatever your camp is, this slide is making a colossal mistake. The mistake it makes is that I put all the information up all at once.

And I know that that means that for the most of the time that I was talking, a lot of you were reading around on the slide. And we got disconnected. What I should have done is show you the information in the order I presented it so that you were with me as I was giving my talk track.

Whether the slide is simple or complex, I feel like that has to be good advice, because this is a lot about joint attention. But whether you're a minimalist or a comparativist is really going to be a personal matter. The minimalist view seems right to me for telling a story.

It's often the best mode when time is of the essence and the audience is mainly there to learn at a high level about what your paper contains. Whereas the comparative view seems right to me for teaching. It's the closest slides come to a full, well-organized chalk board. Good advice for a chalk board is that you never erase things until you absolutely have to, because you might need them later.

Same thing with a slide. If you can get a lot of information up there when it's still perspicuous, then you'll have lots of chances to do comparisons, answer questions, and fundamentally to educate. Find the style that works for you, though. As long as you think long and hard about what it will be like to listen to your talk and you make adjustments, you'll shine.

You will become a better speaker in whatever mode you feel is a good way for you. Guiding audience attention is crucial, though. That's the joke that I made before about my over-full slide with no overlays. Use overlays to fill a slide while still keeping the audience with you. This is how you can be a comparativist while not losing the attention of the audience.

Color used systematically to create distinctions is wonderful, as long as you're sure that the colors really do create contrast that will survive color blindness from your audience members and also really poor projectors, which can be quite a problem. Size is great to draw attention to things. Boxes, arrows, and other devices help people navigate plots, especially model diagrams, long prose statements, and so forth.

It's incredibly helpful to put boxes around the part of a results table or the part of a model diagram that you are talking about in the order that you talk about it so that people know exactly where to look as you make your crucial points. These devices are absolutely wonderful.

In the minimalist mode, I guess you would say, overlays, color, size, boxes, arrows. I honestly find this really hard to track a lot of the time. I understand that it's kind of dynamic feeling, but you can tell in the way I'm talking that I am more of a comparativist than a minimalist when it comes to slides.

Let's wrap up with some more mundane things. Turn off notifications that might appear on your screen. If you get up in front of thousands of people at an ACL conference to give an important talk, it's your moment to shine. You do not want personal messages flashing by as notifications in the upper corner of your screen.

Shut those down. Make sure your computer is out of power save mode so that the screen doesn't shut down while you're talking. The most tragic thing here is that you step away from the computer to make a point. You're doing that long enough that the screen shuts down, and then the projector won't come back, and you lose many minutes of your allotted time just trying to get the projector to cooperate.

So sad. Just make sure your screen doesn't do that, and you'll be better off. Shut down running applications that might get in your way. They could eat up memory and cause problems, or they could just start to ask that you update them or some other needy thing right in the middle of the talk, and it could distract everyone.

And you might not notice the little bouncing icon that everyone is now studying instead of listening to you. Make sure your desktop is clear of files and notes that you wouldn't want the world to see. Again, imagine you're up in front of thousands of people. They are definitely going to study whatever they can see of your desktop as they sit idly waiting for your talk to start or whatever else is happening.

Make sure that's information that you want everyone to be looking at. And if you're using PowerPoint or Keynote or Google Slides or some specialized software that might depend on having net access, make a PDF backup so that it's there just in case. And really the ultimate in survival here, projectors can fail, so you should be prepared to give the talk without any slides.

The audience will be on your side if you have to do this. It's a real opportunity to shine. Be ready to do it as a talk track alone. And then finally, the discussion period. This is often the most anxiety-producing part of giving any talk. It's very important. It should be the chance for the audience to gain a deeper understanding of your ideas.

And when that is the spirit of the audience, this is wonderful. You really feel like you're engaged in a new kind of exploration and education. Sometimes, though, other things do happen. Hostile questioners, confused questioners, people who won't give up the floor, all of that other stuff that can be so anxiety-producing.

You have to live with that. When people ask questions, my standard advice is to pause for a second before answering each one, if you can possibly help it. That will make sure that they're done talking, and it will also make you appear deliberative. And I think that's always good in terms of honoring the contributions people make.

I grant that you've thought about the ideas forever. You might know what answer you want to give from word one from this person, but at least keep up the pretense that you are actively listening. You might be surprised, and also you want to honor the contributions of these people.

Avoid saying, "I have no idea," and leaving it at that. You can probably get away with doing that once if it's a really blazing insight. But ideally, you say something like, "I have no idea, but let's think in this direction," or, "Let's meditate on this possible extension," or whatever it is to create a discourse again, 'cause that's, after all, the goal of these things.

Most questions won't make total sense to you. You've thought about the ideas a lot. Your questioner may be entirely new to them, and they might be struggling as part of their question to figure out exactly how to come to grips with the ideas. So you have to work with that and find a way for this to feel productive.

You'll be a hit if you can warp every question, whether confused or not, hostile or not, whatever it is, into one that makes sense and leaves everyone with the impression that the questioner raised an important issue. Then everyone will feel validated. You'll feel like you've connected with people, and they'll feel like they connected with you.

Notice that I have called this the discussion period. It's often called the Q&A period. I think that's a misnomer. It's really not about question and answer. It's about discussion. And if you ease into that mode, I think you'll overall feel more satisfied by these experiences, recognizing, though, that sometimes they kind of do go off the rails.

(upbeat music) you