It feels not long ago that I was worried about my job applications, but in reality, I have done this long enough that some of my first MSc supervisees are professors of their own now. Every few months, a lab member goes on the academic job market, and we have a conversation about how to best approach it.
So I decided to collect those thoughts on what to do if you plan to become a professor (or, more generally, a professional researcher) in NLP. It’s entirely based on my own experience: I’ve worked two tenure track jobs, one more research-focused in a CS dept and one with more teaching in a Business School. I fact-checked with some people, but it’s probably still not always generalizable.
If it helps you make up your mind, though, go ahead :)
Where to go
The first question is where to go. Part of that is your active decision: Do you enjoy teaching or research more? Part of that is out of your hands: There might not be a position in the city, region, or even country you would like to go to. So you will have to be flexible, but it pays to have a robust set of criteria. Also, keep an open mind. There are plenty of options: traditional university jobs, business schools, think tanks, industry research labs, or government research facilities. There might be even more. They all come with pros and cons, but it is a good idea to include all applicable ones in your search. Much of what follows is outside your control, but choosing to apply (and where) is most definitely within it.
Finding a position
Give it your best go, but do not get discouraged. You do not know what a committee is looking for. They might look for people with a specific research profile, a specific demographic, or a specific skill. While this uncertainty is grating, it can also be consoling. You might not have gotten rejected because you were not good enough in the things you have control over – you might just not be the person they want. Nothing you can do about that. After being placed in the first position, I once got rejected because the hiring department decided to build a bridge to another department that required a specific subject area. One of the other candidates happened to have that specialization. I did not. They hired him. It wasn’t part of the initial call, and it wasn’t something they had planned. It happens.
Some (though luckily not all) positions you apply for are already spoken for but have to run a public call nonetheless. You will do the whole song and dance, but you never actually had a chance to get it. Unfortunately, being rejected from a competition that was never open holds very little information. You will not know beforehand which kind of competition it is.
Expect to apply for 20 positions or more for any match (depending on the field). Try to get practice interviews. Your first interview should not be for your dream job.
In general, you will be judged by these four aspects, in descending order of importance:
Research productivity
Funding ability
Teaching
Service
1. Research productivity
Different places have very different ways of measuring research productivity, from your Google Scholar index, to a list of preferred venues, to a list of your three most impactful papers. It is impossible to optimize for all. You will not know beforehand which venues a university values (they might not even know themselves), so it’s good to go for a mixed strategy. The easiest metric to track is your h-index. It is a measure of both the quality and quantity of your research. Aim for the top venues in the field, but throw in some “easy” venues with higher acceptance rates to bulk out the CV. Often, those supposedly second-tier venue publications garner more citations than top venues. Having a lot of papers and a lot of citations is a helpful signal to employers. However, even candidates with high citation count who have only second-tier publications will likely get sorted out early if there is any expert in the committee.
Do not be shy to self-cite! I have heard the argument in a tenure case that “If even he doesn’t cite himself, how important is the work?” Ideally, you are developing a stream of research, so put in citations to your own prior work (if appropriate, of course). Before submitting a camera-ready, go over your prior work and check: would it make sense to include it here? It’s easy to overlook a paper that could add a valid point, even if it’s your own. Being able to cite your own prior work also indicates that you have a consistent research profile. Also, seeing a reference in a paper makes others more likely to re-cite it.
Research is a crowded business, and it is hard to keep track. People might not cite you for any number of reasons, but typically, it is simply a lack of awareness. Be active on social media and make it easy for people to find you. Maintain an updated website and tell people what you are researching (and why).
2. Funding ability
Universities care about your research. They also care about money. If you have a proven track record of getting grants, you will be much more desirable as a candidate than someone with the same CV but no grants. Grants also signal that your research is interesting to outside stakeholders, something universities value. People typically only list their successful grants, the ones they got but declined, and maybe the current ones.
Grant writing is very different from academic writing: it is story-telling, not research. Do not expect to be good at it from the start. It takes practice, and the odds are low, but persist, and you will be rewarded. As always, grant systems are auto-correlated. If you got a grant in the past, you are more likely to get one in the future (they even ask you to list past grants to assess you). So aim for small grants first, even if they are primarily symbolic. They will pad your resume and increase your chance for future grants.
Do write grants. Lots of them. You will not be naturally good at it; it is a weird genre. Accept that most will get rejected and that you might not know why. If you get a success rate of 30% over your lifetime, you will be outstanding. Do go back and read old grants, and think about whether you would have given yourself money. Read other people’s grants and learn to parse them into components.
Realize that you are writing for a lay audience. The “so what” of the grant is equally or more important than the “what.” Identify the need for your work, and explain why we will be better off afterward (the “unrejectionable”): It is the funders’ choice whether they want to let this problem persist…
Grants give you freedom and independence. Freedom to explore the areas you want and independence from university politics. Grants will often allow you to reduce your teaching load to have more time for research. They will also give you a bargaining chip in salary negotiations and other decisions.
3. Teaching
If you apply for an academic position, you will be expected to teach. There are sometimes incentives for this (teaching awards as positive, student evaluations as a negative incentive), but it will be a given. You should have a teaching statement and some prepared material, but do not sweat this too much: early on, you are not expected to have a ton of teaching already. For better or worse, teaching quality rarely enters into hiring decisions (we all had that one prof). Fight hard to get a class on a topic you know well, and fight even harder to keep it. Preparing a new class is a bottomless time sink, especially in areas you do not know well. If you can use the materials of a more experienced colleague, do so. Resist the urge to make your own or to tinker too much. If you have to make your own class from scratch, make sure you will teach it for the foreseeable future. Plan ahead. Make it count. You do not have to have the perfect material in your first year. It takes about three times to get a class fine-tuned and make it your own. Don’t expect to know everything after one year. More than with anything else, perfection is the enemy of the good in teaching prep. You can spend any amount of time preparing.
Don’t.
Realize that students are not out to expose you: They want to learn from you. You do not have to know every obscure detail. If you do not know some obscure detail, say so, make a note, and look it up for next time. It shows you are engaged, and it is an excellent chance for you to learn.
Prepare an outline of the class, and think about what you want students to learn from it. Gear everything towards that learning outcome. Refer back to that outline often. Do not be afraid to change it if you see it makes more sense in a different order. You will not know ahead of time. Use plenty of examples, and wrap up each class with take-home points. Worry more about the learning outcome than about grade curves or evaluations. Teaching can be gratifying and rewarding if you learn what works for you. It is worth investing in that. It can be a great complement to research, or it can be a drag. It is up to you how to approach it.
Know that evaluations are not objective. They measure how much students liked you as much as how well they learned. If you are female, more junior, or not white, your evaluations will likely be worse than that of white, male, more senior colleagues. Use this to discount the numerical outcomes.
Do pay attention to valid concerns in the written part, but be prepared for contradictions. I have been told in one of my first classes that I was simultaneously too fast, too slow, too technical, not technical enough, too strict, and the nicest teacher all year. Discard the emotions and look for constructive feedback.
Structure your interaction with students and be transparent but firm about it. It is usual for them to have many questions, and it is easy to send an email. But not every question needs an email. There are more of them than you, and you have limited time. Even if every student just has a single question, that still means dozens of emails for you. Collect email questions and answer them in batch in class. Many will also resolve themselves (we all ask questions and realize the answer once we say it aloud). Point to the syllabus, write everything out as clearly as possible and refuse to answer things that are on there (and tell them you do that).
Only meet during dedicated office hours and after appointments (I tried an open-door policy and did not get anything done). It’s also good to set a time limit for meetings: if you have an online system, use it. Otherwise, set up a Google Form with name, time, and topic, and link to it on your website. Ask students to come prepared to those meetings. It will help them focus their thoughts, and it will make it easier to answer them.
4. Service
Research is a community effort. Show that you are a good citizen. If a university wants to hire you potentially for life, they like to see that you are involved. Get engaged in organizing workshops, panels, or initiatives (either write a proposal, or ask an existing committee to let you join). Choose the ones that align with your profile. Become a regular reviewer for a good journal, or eventually an editor.
Take some time to understand how your workplace functions: structurally, procedurally, politically. Those things are not the most natural fit for researchers and can be quite dry and boring, but knowing and understanding them can make your life easier and your work more effective.
Structure is the number and types of department, research centers, labs, hubs, groups, etc. It’s the positions and roles that make up the governance of a place. Knowing who’s responsible for what can help you waste less time by asking the right person.
Procedure is everything from grant management to hiring. What are the steps, which forms do you have to complete, who needs to sign off? This will make your work run more smoothly and save you unpleasant surprises.
Politics is everything that connects the former two, but on a personal level. Who works with whom, who wants what position, what are their goals and motivations. If a professor is known to hate AI, don’t waste time trying to convince them of your new hire or grant proposal. If two people don’t speak to each other, be careful not to get in the middle.
Agree to help out with one committee in your university, but choose which one. Say no to all others. Especially junior female professors are often expected to do a lot of service, and it can easily take over too much time that should go to research and building your profile.
Publicity
Ideally, good work would bubble to the top. That’s not what happens, though. All things being equal, research from big schools or well-known co-authors get cited more. Research that was promoted on social media and is easy to find gets cited more. Research that made it into the news gets cited more. As a researcher, it is your job to publish and then publicize your findings. Getting it into a venue is not enough.
You need a website that is easy to find and up to date. You need to post things on social media: talks you give, workshops you organize, papers you published. Make it easy for people to find your work and understand it. If two people talk about your area, they need to mention your name. It is not their job to be aware of you. It is your job to make them aware. You will have to say the same thing many times before people know it. Have a clear message and profile, and reinforce it.
Network widely, go to conferences, give invited talks. Give plenty of invited talks. “Invited” does not mean they cold-called you. If you visit a city with a relevant research group, let them know and offer to give a talk. They will usually be happy to slot you in.
When you go up for tenure, schools will send out requests for reference letters from people you haven’t worked with. Cultivate a network of those people early! You will be able to list some reference givers, but they might choose some more. So make sure your name is known.
Making it into the press is usually a matter of supplying your press office with a release and hoping they get it out or promoting it on Twitter. Newspapers rarely trawl conference proceedings. Journalists are similar to researchers in that they are looking for interesting new findings. However, they are under much more time and space constraints, which often does not leave room for subtlety. When giving an interview, prepare: What single sentence do you want them to take away from the interview? Repeat that one. Make your points clear and simple: people have no time to read up on the background knowledge. Use simple examples and images. Take notes of all the things you want to cover in the interview and make sure to tick them off as you go. Ask for written questions ahead of time. After the interview, ask for a draft or at least the quotes of you before publication. Sharing this should be standard, but sometimes it is not. Those proofs are your only chance to correct the record if you have been misunderstood (it happens, even with preparation). Remember that you can mark certain things as “off the record.”
Staying creative and managing projects
Whatever route you take, you will have to generate new research ideas. When you are in industry, you should never be irreplaceable. When you are in academia, you need to be one of a kind. Either way, you should be original and creative. It does get easier over time, but it takes work. Find out what inspires you, and seek that out. For me, it is interacting with people on visits or conferences and then distracting myself with something creative but unrelated to push it forward. Trust your instinct: if you need a day of slacking and playing video games, it might just be your brain recharging and making new connections.
You will have to find out how many parallel projects you can pursue to always have an iron in the fire without slowing progress on everything. It should be more than one, but the exact number depends on your multi-tasking affinity and the projects’ complexity. For some, that is 2. For others, 15. Try to push yourself to find that limit, but realize when you reach it. If you feel like you are standing still, don’t be afraid to shelf a project.
Those projects ideally fit together to advance your research profile, but do not be afraid to try something new out every once in a while. Nothing is sadder than a researcher clinging to outdated ideas.
How much you can choose and experiment also depends on your career stage. Earlier on, it is probably reasonable to explore more and to push many lines. As you find your niche, focus on that. At the same time, getting tenure allows you to be more experimental. Use the security to explore more far-fetched and ambitious projects, but don’t risk your collaborators’ careers.
But do not get stuck in minor variations of the same theme either.
Running a lab
Depending on your choices, you might have the chance to build up a lab. This is sometimes even expected. If you have a choice, consider the options. Working with a lab allows you to explore more research areas, but it will replace a large chunk of your time with managerial duties, admin, and bureaucracy. That chunk usually comes out of your active research time. Working alone means less collaboration on a day-to-day basis, but also no managerial responsibilities. Either way, fight hard to keep a day clear of all meetings and admin to do your own research. Mark it in your calendar as blocked, and do not make exceptions. Ever. If you make an exception for a good reason, you’ll soon make one for a bad reason.
If you do build a lab, realize that you will spend more time each day with these people than with your partner. Don’t hire assholes. Work on the group atmosphere you want: you will set the tone and the norms. Treat people with respect and trust them that they want to do good work. Do not be afraid to give up responsibility and control. Do not make them do something you would not do. Be clear about what you expect from them, and keep their desks free of unnecessary bureaucracy and admin. Listen to their ideas and concerns, and take them seriously. Say “Yes, and?” and see where it takes them.
It is your job to help them succeed and solve their problems, not the other way around.
Be flexible and don’t be afraid to change things. The meeting day does not work for most of the group? Find a different day. The format of the internal talks feels boring? Try shortening it, restrict the topics, make it an interpretative dance. You are not forced to stick to something if it does not work. Occasionally, it can be good to try something new even if nobody feels the need. You can always change it back.
Final thoughts
Make sure to take at least one day a week completely off. That also means no email. Keep some hobbies, especially sports. You can get away with a lot of neglect in your 20s, but it will catch up, and you don’t want to lose time fixing a preventable injury. Our jobs require a lot of sitting at a computer: counteract that with whatever sports you enjoy.
Go on vacation, and set your OOO response for one more day than you are away. You can use it to catch up on email.
Don’t feel like you have to be perfect. Make a plan, be prepared, but know that things will blindside you, that things will not work out. And that’s ok. So will you be. I have failed many times, and usually it didn’t matter. Our jobs are not heart surgery.
Lastly, realize that you are given one of the potentially best and most self-directed jobs there are, but that it’s still a job. Don’t let it distract you from the importance of friends, family, and your health.
(Some of the better advice in here is based on tips from Dan Jurafsky, Ed Hovy, Lyle Ungar, and Nel Dutt. Thanks!)
“Things of great importance have to be taken lightly.”
– Hagakure
“Realize that your choices are half chance. So are everybody else’s.”
– Baz Luhrman, “Everybody’s free (to wear sunscreen)”