Convincing everyone you're smart (and how not to do it).
We need face the sordid fact that an important 9and usually unstated reason) that you want to communicate is to convince the audience of your skill, creativity and general insight -- more bluntly, you want to prove how smart you are. (Someone once said that intellectual activity of all kinds is simply one brain's way of ranking itself vis a vis other brains). Assuming this is one of your main (if hidden) reasons for speaking, you should be aware of an insidiously seductive short-cut that leads to a disasterous pitfall.

At first glance, obfuscation looks like a good strategy for making yourself look smart. The reasoning runs as follows -- "If I talk very fast, use lots of jargon and make huge inductive leaps, the audience will get lost, feel dumb and inadequate and assume that I (the speaker) must be the brightest person around." -----Wrong. The many who follow this reasoning in their written work, put out material that is tightly written, finely reasoned, extremely compact (Constipation comes to mind.) The reader endures a joyless, time-consuming task, even when the content is excellent. In lecture form this approach is dry dry dry and you find yourself sleeping in a seminar that you looked forward to hearing. A form of this short cut is to include every single glorious experiment and exactly how it was done until the audience drops from sheer tedium waiting for an idea. Obfuscation (including ponderous impressiveness) is another way of inflicting pain on an audience and no one loves pain; no one loves feeling dumb. They came to learn and you treated them like dirt.

Revert to your teaching mode. When you give a talk or write a paper, you're teaching. The ideas conveyed are your heart and soul, the product of your hands and intellect. If you can teach the intellectual message to your audience, your information becomes part of their general body of knowledge. They will remember with pleasure that they gained some insight from you; they got smarter and improved. If the material is clever they will appreciate your intellect for being able to solve a problem and for being able to convey the message clearly. If you can make your story clear even to non-specialists, you've broadened your audience and spread your message to even more people (putting a higher exponent on the information growth curve). Alternatively, if you baffle them, they'll just think you're a jerk and avoid your next paper or seminar; you've thwarted your own plan to expand your reputation and advertize your brilliance. ---- Take home lesson - TEACH 'EM; DON'T DAZZLE 'EM.

A basic philosophy to guide the public speaker
Whether or not you agree with the suggestions for seminar presentation that follow, the philosophical considerations below are intended to serve as a way of gauging the importance of a good speaking style and as an overall criterion for devices you might like to emply. Remember why you sometimes skip a seminar? Remember how you have felt when leaving a seminar that has been poorly prepared or presented? My dominant feeling is pain and rage at the amount of time wasted in discomfort -- time which could have been better or more enjoyably spent doing anything else.

The magnitude of the pain inflicted by a incompetent seminar-giver can be estimated by considering the product of the time spent (usually an hour) and the size of the audience. A painful one-hour lecture to eighty listeners inflicts aggregate pain equal to torturing one human being all day, every day for two working weeks. That is cruel and unusual. By using this as an measuring stick, you can get a fair idea of why it's important to give a good talk, especially to a large group. Think back to your outrage (as a member of the audience) at having wasted an hour of your precious time at a bad talk; multiply by the size of that audience.
An audience that comes to hear you speak has taken a considerable risk (see philosophy section above). They have given you the power to hurt and maim them without any guarantee of their own safety. They do you a great honor by showing up -- you should treat them with the respect and consideration which is their due. Think of yourself as their guide for a risky trip. The audience is naturally apprehensive that they'll be led astray. You need to reassure them with your confidence and assurance. Be solicitous of their physical and intellectual welfare; convince them of your capabilities.

Creature comforts come first.
As any capable host, you should start out by being sure that your honored guests are physically at ease. Can your guests hear, see the board? Are there enough chairs? Is the room too hot? Your audience will be able to pay better attention if you spend a moment making a few adjustments to the microphone or the seating arrangements, opening a door, turning on the board lights. Give them a chance to volunteer some suggestions -- "Is the microphone working? Can everyone hear?". Use the same philosophical attitudes regarding the "comfort of guests" to guide you in constructing and presenting your seminar.

Encourage participation
The most entertaining part of attending a seminar is having a chance to think about something new -- to try your own brain out on an unfamiliar puzzle or set of questions. You give your audience a chance at fun if you allow participation. A conversation with give-and-take is more fun than a formal lecture received passively. You can approximate a conversational atmosphere, even with a big audience, by inviting interruptions.
Try starting out by saying "If you have questions or don't understand something, interrupt me during the talk." This is a breathtaking offer to someone like me who has trouble picking up all details in a talk and can often be kept on track by a minor clarification. It's like the matador at a bullfight, turning his back on the bull and walking toward the spectators -- the audience knows you are in supreme control and worthy of their trust. The worst thing that can happen (the bull charges) is that you get some pest who starts trying to show off by dogging your every word. This isn't very common, but if it happens and threatens to take you overtime, you can always regain control and call a halt. Perhaps you could say, "That's a reasonable point but it gets us off my main topic; we can discuss it later" or even "Gosh, I'm getting behind, let's hold the rest of the questions until the end."
You can stimulate participation during the talk by interjecting invitations between sections. For example, "We've just described the basic experiment. Before we get into the results -- do you understand the basic idea? Any questions before we proceed?"

Cater to short attention spans
Someone once said that the attention span of the average human adult is about 15 minutes. This is unfortunate, since seminars usually last an hour. You can reset their clocks but building interruptions into your talk. That's one advantage of soliciting and getting questions. A question stops the formal flow of the talk. Everyone can take an intellectual breather and the tension between speaker and audience is broken for a bit by a few words of relaxing banter.

Sell the question before trying to sell the answer
A good principle of teaching is not to provide answers to questions that are not in the mind of the student. Why should a student care about the answer to a question he doesn't have? In seminars you need to teach the question first by presenting the background material that everyone accepts, outline the question inherent in this background and, only then, proceed to describe how your experiments are directed at an answer.
How to establish the background is a difficult problem. You never quite know how much your audience already knows. If you get too simple and spend too much time, they're bored. If you skimp, they can't follow your experiments. Max Delbrück was reputed to have advised seminar speakers "Assume that your audience is completely ignorant -- but infinitely intelligent". In following this advice, one reviews material that is probably familiar to most of the listeners, but one does it at a rapid pace, including only the essential points. An infinitely intelligent novice could, in principle, learn the necessary background; a partially informed audience (more commonly the case) appreciates the review and isn't bored if it moves fast.

Sell the answer before the experimental data.
Once you've established the question, it's not a bad idea to tell them where you're taking them. For example, "I just explained to you the importance of knowing the composition of the moon. Next we're going to consider a series of experiments designed to address this. Through these experiments I'm going to try and convince you that the moon is composed completely and solely of green cheese."
By doing this, you set up a concept against which the listeners can measure all that you say. Each experimental result can be evaluated for its contribution toward achieving the stated goal. This makes it much easier to frame objections or accept progress than is the case when the listeners don't know where they're headed. They can start to anticipate what experiment you might do next. They can see where they're going -- always comforting to those on a dangerous journey.
The extreme example of an undirected talk is called a "Who-done-it" in honor of detective stories that lead you through lots of detail (some relevant and some not) and provide an answer only on the final pages. In such a talk, one presents an experiment without telling it's purpose. The audience hears a lot of experiments and a lot of results before learning what has been concluded or why the experiments are being done. They must keep all this information in some sort of mental buffer and be able to recall it when conclusion time arrives. Their recall must be perfect if they're going to evaluate your evidence critically. This is tough to do.
While it may seem more honest not to prejudice the audience with the conclusion, in fact, listening to a talk is too hard for the purist approach. It's better (if you have a sharp conclusion) to tell them the answer up front, just after setting the question.

Data versus ideas
You're most likely to be talking to scientists. This is an audience that loves problems, thinking, ideas, games, hypotheses, models, paradoxes. Make sure these are the dominant features of your talk. A recitation of facts and huge piles of data are only interesting to the extent that they establish or support some engaging notion.
Audiences (at least classy ones) aren't very impressed by hearing you tell them you're a hard-working dude -- they're hard-working too. It is a very common mistake to list the "labors of Hercules" and believe you can make people gasp in awe -- mostly they just yawn at the data and hope you'll soon get to an idea, make a conclusion or eliminate a model.
When you have a lot of data, but can't make much of a conclusion, you have a serious problem. The facts can at least be arranged around an idea or a model. You can spend more time establishing the question or the difficulty of the question. Stress the explanations that are eliminated by your data and the ones the are left as possibilities. Keep tying the data to an idea, a model, a hypothesis and don't let the data take over.

Minimalism in data presentation
Frequently one does a lot of work that doesn't lead to a conclusion before coming up with the meaty experiment that tells the story. Spare them the first experiments (no matter what it costs you in anguish) and spend adequate time explaining the critical experiment fully. Every experiment you set up takes time to explain. Give the big one the attention it deserves. (If you only talk about critical experiments, they'll think you're a genius whose every experiment is ground-breaking.)
Avoid distracting side-routes. Experiments frequently suggest new ideas and pose novel questions. Avoid pursuing these. Stick to the central question you posed at the beginning of your talk. Be brutal in suppressing the clever tangents; make sure that the basic goals of the talk are completely and effectively discussed.
Above all, avoid the "Grant Proposal Coda", the hateful little trailer that follows many seminars. This is the little tail-end section, in which the speaker tells what will be done next and the cosmic conclusions that will be drawn from these un-done experiments. (Basically they are presenting their next grant application.) Interpreting results not in evidence is like saying "If my grandmother had wheels, she'd be a roller skate." The audience didn't come for science fiction or to hear you tell them directly how smart you are, they just want to have fun (Translation: to think and learn.)

Overlapping loops - the importance of redundancy
Spoken communication is inefficient because the audience has many attention lapses. They worry about dinner, the car repair, how to solve a problem they have in the lab, a sick dog, their itchy scalp etc. etc. Worst of all, they don't synchronize their lapses. Because of this, if you give a perfectly logical presentation with each step an essential and different link in the chain of evidence, many will loose track of the train of logic when they have an attention lapse.
You can accommodate this problem by having the direction of your talk follow a set of partially overlapping loops. Speak for three jumps, then recap jumps two and three and add four. Next you recap three and four and add five etc. In this way everything gets restated and you keep everyone in the game. This looping train of exposition would look terrible in printed material, but while reading you can move back up the page and reread a section missed during a lapse. Notice how often you do that when reading scientific material. You can't "re-read" what is missed in a talk, but a speaker who builds in some loops can give you a break.

Time your talk carefully
Don't ever go overtime. Remember, in the philosophy section we described the audience as a bunch of apprehensive travelers who took on the risk of a journey (attending your seminar). They contracted to risk an hour of their time (no more) to give you a chance as guide. Regardless of how well you speak, if you go overtime, you violate the contract. They gave you the honor of listening and you forced them to give even more. (They signed up for a float through the Grand Canyon and you dragged them, kicking and sceaming, all the way to the coast.) No one was ever angry because a talk was too short.

Summarize at the end
It's nice to sum up at the end. This helps those that just barely stayed with you (and those that slept through the whole thing). The former types at least get a restatement of what they heard so they can confirm and solidify the pictures they had established in their minds. For the sleepers, its a party-favor so they don't go away completely empty-handed.

The advantages (and difficulties) of chalk talks
To keep people's interest and convey information efficiently, I think it's hard to beat a chalk talk with a well-lit visible black (or white) board.
The main advantages are that the tip of your piece of chalk serves as the focus of attention and minimizes distractions.
--- Because of the time required, a diagram that you draw while speaking must have minimal detail. Thus you're presenting only the essentials.
--- All eyes and attention are directed to the line you are drawing.
--- This avoids "talking head syndrome" in which the audience must stare (hypnotized) into a face with a flapping mouth.
--- This avoids the mass of detail that is presented all at once on a detailed slide. While you describe one portion of the information on a slide, the audience is looking at the whole thing and is therefore distracted from what you're saying. They lose guidance for one part while trying to understand the rest with no guidance.
--- By drawing while speaking, you show only the diagram that is being discussed. All material visible on the board has already been described and can serve as a reminder and reference to the audience as you go ahead.
--- You can add a diagram to the board, leave it in place, and make a second diagram, moving across the board. When a second slide comes up, the first disappears and can't be referred to for those who had an attention lapse.
--- While speaking, either draw continuously or point out aspects of previously drawn material. This might be a list of key words (no long sentences) or diagrams. One point of this is that it arranges things so you don't set up an adversarial situation (you versus them). Instead, you and your audience are on the same side looking at an impersonal third object (the board). In this situation, they're more likely to formulate questions for you and think with you about the problem. In an adversarial arrangement, they're more likely to be defensive, fight against your (obvious and beautiful) conclusions or give up and think about something completely unrelated.

Disadvantages of slides (some repeated from above)
Slides take time and effort to prepare and therefore are likely to be used without revision when you give a seminar in a new way or with a new focus of interest. Thus seminars frequently contain slides with incorrect or out-of-date information.
Projectors are noisy, making it hard for those near the machine to hear you.
Slides require dimming the lights and are sleep-inducing.
It's nice to see the speaker, note the expression, animated motions, excitement. It's very hard to tell a joke in the dark because so much information is conveyed by the speaker's "body-language".
A slide contains more information than you can discuss at one moment; the extra stuff is distracting. Most people's slides contain information that will never be discussed and is therefore a distraction that has no positive effect. Whatever isn't used is hurting your presentation, because it distracts.

Problems of chalk talks
You have to know what you're doing and have the topic under complete control. This is easier when you are describing your own work because no one in the world can possibly understand that as well as you do. After all, you've been thinking about it for years, more or less full time; the audience is hearing it for the first time. If they can out-think you, you've got problems that won't be solved by use of slides.
You need to be able to think on your feet (Translation: walk and chew gum). A question may require redrawing or modifying a figure, or coming up with a new figure. This means you have to listen very carefully to questions, understand the point (usually one you have anticipated), and be ready to fill in the missing parts.
Despite the challenges, this is the high road of speaking. It keeps people's attention just to realize that you are out there alone (on the high wire without a net). It makes the talk a more intimate form of communication to have no technical devices between you and the audience. (Comparable to the difference between a live concert and hearing recorded music.)

Use of overhead projectors as an alternative A chalk talk is physically impossible when you are speaking to a very large audience, or in a small room with poor boards or poor lighting. In these cases, the board simply isn't visible and you have to find another solution. I find that overhead projectors with hand-drawn words and diagrams on the transparent sheet are the next best solution. Here's why --

-- With an overhead, you can make (in advance) the same free hand drawing you would have put on the board.. This minimizes detail and give your talk a folksy quality. (Try to spell correctly on the overhead.)
-- In principle, you can draw on the transparency while you talk. I haven't practiced this enough to feel comfortable, but I've seen it done very effectively. (Usually you have to bend over to draw, which is hard.)
-- When you put a pre-drawn transparency on the projector, you can cover the bottom part while talking about the top. This "strip-tease" approach keeps audience attention from wandering all over the page. Once you've described something, you can move the cover down, revealing the next part and leaving the top visible. This achieves the same effect as moving along the board in a chalk-talk.
-- It's easier to point out things on an overhead. You can use your finger and point on the transparency, or use the shadow of your finger to point on the screen. (Laser pointers are the pits; they're hard to spot on the screen and they shake around in a distracting way. If you're a bit nervous, the quaking laser pointer makes you look seriously terrified.

If you must use slides
By now you've considered the arguments for chalk talks; you may have found it all vaguely amusing but you have concluded that it doesn't apply to you. In your particular case, (but no one else's) slides are better. You have rationalized this decision by a complicated (and completely incorrect) train of logic and are ready to plan your slides. If this assessment of the situation is correct, I recommend that you start at the top and read the whole essay again before going on. If I've still failed to convince you and you ABSOLUTELY WILL MAKE SLIDES, read on.

Keep detail on the screen to a minimum.
Put on the slide only material you need for your argument. Anything extra is distracting and falls in the category of trying to impress the audience with what a drudge you are for collecting irrelevant data.
Five numbers on a slide constitute a discussible amount of information; fifty numbers are not. If you diagram a process, stages of mechanism, or a series of crosses, then you are morally obligated to lead the audience through every detail of the slide. If you don't, they can't possible grasp all the details and you are being deliberately confusing. Rather than confuse them, leave the slide at home or design a simple slide with only the relevant material. Either a slide is valuable and therefore deserves a complete guided tour, or its a distracting, useless ostentation. Dazzling the audience with a mass of details is usually done by speakers who intend to impress the audience with how smart they are and how dumb they can make their listeners feel. It never works and is in no way conducive to education. (If they could pool their aggregate annoyance, they would be more likely to kill the speaker than to applaud; thank God members of audiences can't pool their separate displeasures.)

Degassing your slides.
Slides frequently are made at a computer keyboard and there is a strong tendency to use lots of words in small type fonts and put lots of space between many elements of the drawing. This is terrible because usually the small fonts aren't legible. Try to fill the space on the screen. Use HUGE type fonts, broad lines, and a minimal number of elements. Basically you're squeezing the "air" out of the space and filling it with bold informative, legible symbols.

Don't use text slides. They're very popular with speakers and they are universally annoying to audiences. If the audience wanted to read your message, they would have gone to the library. It insults an audience when you put up a text slide and then read it to them. You imply that either they need help reading or you don't know what you're talking about without reading it yourself.

Mannerisms James Watson is famous for mannerisms. His contributions to science are enormous, but his contributions to public speaking are largely as a model to avoid. I once watched him being interviewed by Dick Cavett on televison. In the course of the conversation, Watson started scratching his left ear using is right hand by extending his right arm across behind his head. (Try it, it's possible.) The scene was so grotesque and continued so long than it was hard to concentrate on anything being said. He's also famous for leaning against the blackboard while lecturing and slowly sliding down the board until his elbows are in the chalk tray and his legs are extended far out toward the audience. Apart from the fact that this activity is distracting, it gets your clothes covered with chalk and you look like some kind of plasterer as you finish the talk with white smudges all over the back of your head and clothing. Another of his mannerisms is making a punctuated hissing/sucking noise by drawing little puffs of air inward across closed teeth. This makes you look like a threatened, grinning chimpanzee and has a frightening effect on the audience. I think you get the idea. Pay attention to what you're doing with your body and hands while speaking (or get your friends to watch you speak and point out physical mannerisms.)
More common distracting mannerisms are verbal, such as saying "Uhm" or "Aaah" or "Let's see now" or "You know?" or (heaven forbid} "Like........" or otherwise separating informative sentences with sound bits. Silence is better. Think quietly (or during a practice run of your talk) so your verbal utterances are all directed toward the message you want to deliver. If you need to think, do it without flapping your mouth or making sounds. There's no shame in a moment of silent prayer during a seminar.

Don't be cute when nervous
All speakers are nervous about standing before a group. A frequent response of beginners is to curry favor by being "cute". This means making giggly, silly, sometimes self-deprecatory remarks in the hopes of getting the audience to love you because of your sweet, cuddly self and not because of any message you have to offer. This technique works great for pre-school children and is socially encouraged behavior for women in some primitive societies. However it's frequently tried by science graduate students of both sexes. Examples are giggling at mistakes made or "Silly me, I forgot the critical slide" or "Shucks, I never get that right" or "I don't know the answer to that questions..... Bobby (the research advisor who's in the audience), can you explain that for little old me." FIGHT THESE MANNERISMS.
When you speak, be serious, be all business and make it clear that you are in control of the proceedings and you have an important message to convey. Why should the audience care if you don't appear to care? By asking help of someone in the audience, you loose control of the talk and things fall apart rapidly. You are being evaluated as a scientist and a teacher ---- not as a potential adoptee or as a date to the prom. It is essential that the audience take you seriously. Humor has a definite place is speaking, but you should not make yourself the object of the laughter -- this behavior makes you such an object.

The "talking head syndrome" revisited
When you face an audience and they face you, a sort of tension is set up. (Stare directly into someone's eyes for a while and see if it doesn't get tense.) If you do nothing but stand there and speak directly to the audience and they stare back while listening, this tension gets palpably hypnotic. Above, we called it the "talking head syndrome". Listeners can't stand it very long without looking away, feeling weird or having their attention lapse. You can break this tension in a variety of ways, some of which were mentioned above.
One trick is to move your eyes around the audience. Pick out several people, well-dispersed through the audience and address remarks to them for a while before switching to someone else. (Give yourself a break and pick people who are awake and trying to follow the talk.) You can even walk around a bit, but don't overdo it.
Use of the board is great for breaking this tension. You become one with the audience and join them in considering a diagram that is unfolding below your chalk. You move to the next point on the board. You turn to the audience and back to the board. Pointing to a slide is OK, but the lights are out so they can't see you at all.
Interruptions in the narrative help break the tension. Example: "So you can see the problem we're trying to address. Are there any questions before we go on to the experiments?" This breaks the mesmerizing tendency of flowing words. It's a paragraph mark in verbal flow. The concentration is relaxed for a moment and you're ready to reestablish contact for the next section.

When English isn't your native language
Trying to speak in a second language is difficult for the speaker and can be horrible for the audience. English is a plastic language and native speakers are actually pretty good at understanding imprecise pronunciations IF YOU SPEAK SLOWLY. Slower speech gives the audience a microsecond to sort out the language problems. If you speak rapidly and with some incorrect words or pronunciation. the difficulties can overwhelm the audience and they give up. There's a tendency for some non-native speakers to speak English even faster than normal English speakers. I suspect this happens if their native language has a large number of syllables per second and a lower number of ideas per syllable than English. They speak English so the rhythm sounds right to them and the resulting idea flow rate in English is very, very fast.
While speaking slowly helps the audience understand individual words, slow speech can be incredibly boring if the text has low idea content and lots of filler words. You need to compensate so as to keep the flow rate of ideas up to par. Speech that is low in idea content is OK when a native speaker is speaking rapidly; the redundancy helps the audience keep up. However if you slow your speech down to increase work understanding, don't use the inefficient filler words and empty phrases that are used by native speakers. When speaking slowly, use very information-dense sentences, so the ideas keep flowing at a reasonable rate. (Read Lincoln's Gettesburg Address some time for an example.)
These suggestions are difficult to follow and require a lot of practice. However, based on the experience of foreign students in my lab, I think they are generally correct. They are worth the effort, because being able to communicate clearly in English is important to your success here. Below is an added note of warning ----
Just because you communicate easily with English-speaking members of your English-speaking lab, don't assume that the same sort of speech will work with a group of strangers. Your closest associates get used to your linguistic imperfections. In a sense, they learn your private dialect of English and can therefore build a personal vocabulary and set of pronunciation allowances to use when listening to you. Strangers in the audience haven't heard you speak before and can't make these adjustments. For them, you need to speak a generally understandable form of English.