Should you get a PhD in linguistics? Possibly.

Rachael Tatman, a graduate student in linguistics at the University of Washington, has a new blog post up titled Should you get a PhD in linguistics. Her answer is a resounding NO. As someone who just graduated from a linguistics PhD program and doesn’t fully share her bleak conclusions, this seemed like a good topic to inaugurate this blog. The fact that the PhD is very recent in my mind means that I have some pretty strong opinions about the matter and that I am enthusiastic to share them, but also that my perspective is fairly limited, so I’d love to hear other opinions in the comments.

According to Tatman’s post, there are two main reasons you shouldn’t get a PhD in linguistics: the job market in linguistics is terrible, and grad school is grueling. I fully agree with the second point: getting a PhD is, in fact, really hard. Most graduate students have monthly crises where they decide to quit the program, then change their mind, then decide to quit again. And many end up actually quitting. Hal Daumé found in an informal Twitter poll last month that 78% of his followers who finished a PhD had had at least one “emotional crisis when they wondered if they could cut it”. Their crises were probably not brought on by job market anxiety: I imagine that like Hal most of the people who took the poll got their PhDs in computer science, a field that is facing such a spectacular shortage of professors that there are now one-year Master’s programs to train people with PhDs in other fields to teach in computer science departments. No: getting a PhD is actually, truly very hard.

Where I disagree with Tatman is the state of the job market. There are a lot of depressing stories about the academic job market in the humanities. Hordes of Yale English PhDs are reportedly shuttling between four different community colleges to scrape together $20,000 a year as adjuncts. I have no reason to doubt these stories. But I’m not convinced that linguistics is in the same boat. I don’t know the stats (if you do, leave a comment!), but my impression is that linguistics departments aren’t facing the same decline in undergraduate enrollments as many of the “core” humanities departments. If anything, some linguistics departments are expanding, which means that there is a reasonable number of job openings (though those positions are still very competitive, of course).

More important, though, is the fact that more and more linguists have computational and data analysis skills that make them eminently employable in the private sector. These jobs are really, really not the preserve of nerds who learned to be software engineers in their parent’s basement as teenagers. There is a huge demand right now for smart people with PhDs and quantitative research skills. Given how easy it was for people in my school who didn’t have any formal training in computer science to find a job as junior software engineers or data scientists, I’d say most of us are probably underestimating our chances of landing one of those jobs (perhaps after a two-month “boot camp”). If you’re less excited about the private sector, there’s a growing number of quantitative positions in governments and NGOs. And there are also plenty of non-quantitative jobs that linguistics PhDs are qualified to do, but I’m less familiar with those career paths and haven’t personally considered them.

The common perception that private sector (“industry”) jobs are looked down on by professors is almost certainly overblown, and even if that attitude still exists here and there, it’s on the decline. It reflects very well on a department if its graduates are gainfully employed at well-known tech companies like Nuance or Google or startups like Duolingo. There is increasing pressure in academia to collaborate with industry; academics need industry connections. And even if non-academic jobs are looked down on by some out-of-touch member of the old guard or another, are you sure you really care? Your starting salary as a data scientist is probably going to be higher than what they make as full professors.

Between academia and the private sector, recent graduates from top linguistics programs don’t seem to have a very hard time finding full-time employment. For some examples, see the placement records for Stanford, NYU and UMD. (Unfortunately some linguistics programs don’t have their placement records in an easily accessible form on their websites. We should be more transparent about this – prospective PhD students need to be able to make an informed decision!)

The big “but” here is that linguistics is made up of several subdisciplines, some of which are closer to the humanities and others closer to the behavioral sciences and computer science. Not all subfields of linguistics have the same number of faculty openings. And those subfields where competition for academic positions is the fiercest (such as syntax and morphology, according to this analysis) tend to be the ones where many students don’t receive sufficient quantitative training to get one of those cushy tech jobs, or simply don’t have any interest in doing quantitative research in the private sector. I’m not sure if the situation in those fields is as dire as it looks from the figure I linked to; I do know several talented and accomplished theoretical linguists who haven’t been able to find a permanent position three or four years after their PhD, but so far most of them have been employed in reasonably paid full-time temporary positions in good departments.

At any rate, I’m not really sure what the best way is to address this issue. Perhaps students who write theoretical dissertations should make sure they get a reasonable amount of quantitative research training. And perhaps more graduate students are currently specializing in certain subfields than we need. Of course, most of us would agree that the number of humanities PhDs that “we need”, by which we typically just mean “the number of available faculty positions”, is too low. But that’s part of a much larger debate about whether the humanities are undervalued in our society. I think they are, but as a prospective PhD student you’re not going to resolve that debate single-handedly: Tatman is absolutely right that the decision of whether or not to get a PhD and what research opportunities to pursue as part of that PhD should be informed not only by idealistic intellectual considerations but also by the realities of the job market.

So my answer to the question “should I get a PhD in linguistics” is “it depends”. Are you absolutely determined to be a professor? (If you are, don’t be.) Do the non-academic career options in your subfield appeal to you? Can you receive solid quantitative training in your program? Would you be at all interested in a non-academic quantitative job? What is the placement record of your program? How worried are you about the opportunity cost (all of that money you could save if you worked in consulting instead of getting a PhD)? Are you prepared for the emotional difficulties involved in going to grad school (and, as my professors tell me, in being on the tenure track)? And – it goes without saying – are you very, very passionate about this intellectual journey?

9 thoughts on “Should you get a PhD in linguistics? Possibly.”

  1. I came across the following much more pessimistic comments on Facebook from someone who prefers to stay anonymous, and thought I’d post them here:

    I would still lean on the ‘no’ side. Most of the people who I know and who left academia don’t do anything related to their PhDs, including myself. In fact, leaving academia for a linguist usually involves a full blown career change. Depending on your age and ability to be flexible, as well as ability to take things with humor, and ability to not earn for a while, this may be a very hard, humiliating and frustrating experience. And then when you do find a job, your manager is likely to be 5 years younger than you, because unlike you, they spent the past five years out there in the work force, climbing the career ladder. You can of course take your chances with the academic job market, but it’s naive to think that there are good employment opportunities for linguistics outside of it. Anyways, this has been my experience, and I am 2 years out of grad school.

    Also, where he is wrong is that most people fail to see the value of a ling PhD. Usually when people look for Data Scientists, they look for PhDs from the math and physics departments. Most people associate ling with English lit. And even if you do get that job, you are very likely to be underestimated by your boss with an undergraduate degree in engineering.

    I know that there are linguists who got really good non academic jobs. But there are also people who get great tenure track jobs. In reality most people don’t though.

    Finally, I don’t know anyone in Linguistics who seriously collaborated with anyone in the industry, and to get a full time job at Google is exceptionally difficult for anyone (so don’t count on it). Nuance does hire linguists, but it is one company. One. Out of many others who don’t.

    If you’re doing a PhD in ling because you want to be a data scientist, shouldn’t you be doing a PhD in statistics or computer science instead? Why do a PhD that will require that you go through a career change later in life?

    The no children plays a big role. Once you have a family to support the monetary considerations become more and more important. Spousal career was another constraint for me. A spouse in academia with inflexible location constraints can make things harder.

    [About non-academic jobs for linguists:] Let me ask you this: how many job interviews did they go to? How many ‘nos’ did they hear? How long did it take them to get their first job? Were they happy at their first job? Was it a compromise? In what respects? How old are they? Do they have a family to support? It is not simple or straightforward to go through a career change. It is easiest if you are young, single and male.

    I think there are many many Ling graduates, Not just from NYU or Stanford. In a way, your sample will be biased if you consider only schools that are in large tech hubs, because for these people the tech jobs are more accessible. A lot of good programs are in small towns, or in places where there is little to no technology. I think to have a representative sample we should talk about students from other schools too.

  2. I wholeheartedly agree with your post (haven’t read the comment yet), but one small quibble: I do think the jobs situation, while certainly less bleak than e.g. English lit, is still somewhat more bleak than suggested by looking at e.g. NYU and Maryland grads. Those are great schools with great reputations, and do great in job placement (not to suggest that people from there who get jobs haven’t legitimately earned them; everyone I know from those places who’s gotten tenure-track jobs recently is excellent and totally deserved that job). It may be harder for people from schools without such a strong reputation, even if those people are just as great. It’s hard to make this not sound self-serving (but at least I have a job too so hopefully it doesn’t sound like I’m talking about myself), but I have anecdotally heard lots of people having the impression that their job search (or that of someone they know) has been hurt by the lower-tier status of their program. Of course, who knows how big that effect really is–it’s only natural when getting passed over for a job to think “But I’m so great, those bastards must only have passed me over because I’m not from one of those fancy schmancy schools”–but at least there seems to be some empirical evidence too (http://advances.sciencemag.org/content/1/1/e1400005). Although yeah, all that being said, I still would not call the job market bleak, especially compared to humanities; it’s rough, sure, but everything is.

    1. Thanks, that’s a very important point. I agree that there’s a bias throughout academia for fancy schools and research groups. You get on future search committee members’ radars just by working with a famous person, and I’m sure that letters from big names carry more weight than letters from people who are possibly just as brilliant but are less effective at self-marketing or work at a less famous department. I really think that’s something that the field needs to work on, but again, as a prospective graduate student you’re not going to single-handedly change people’s biases, and there’s not reason not to take those biases into account deciding whether and where to get your PhD. To be clear, I’m definitely not advocating choosing an advisor or a department just (or even mostly) because they’re famous – the most important factor is where you’d be able to do good research and where you’d be personally happy.

  3. FYI, if linguistics grads are okay with having an appointment in CSD/SPLH, there are job openings (you could do the same research). This is because there aren’t enough post-graduates to fill these positions. The last 2 hires in SPLH/CLDP at KU came out of psychology and genetics programs, and the SPLH department carried out another search this past year. CAPCSD is where all the jobs are listed. My adviser has told me that NIDCD is very interested in neurolinguistics research (read: grants!) related to language acquisition, so it is an opportunity to obtain a tenure-track position. Steve has said that many descriptions seem like they want clinical experience, but based on the last 2 hires (neither of whom are clinicians – one studies speech production and the other molecular genetics), the job description criteria may be flexible.

  4. I don’t regret the PhD at all. I had a great time and learned many things, some of which are even theoretically useful skills for non-academic jobs. But, in 7+ months of job hunting, I’ve found the non-academic job market not-so-friendly. for a couple of reasons. One part of it is probably just not knowing how to present skills and experience for a non-academic audience. Whether they’re willing to consider academic research relevant experience also seems to be an open question. Another part of this is networking and things like summer internships that people who start out intending to find industry jobs seem to do as standard part of their degree.

    As far as response from people in my grad program to my not pursuing an academic job, my advisor has been very supportive through the whole thing, but they don’t have a lot of practical advice or relevant connections for doing this sort of thing.

    Then again, I met someone who got the same type of job I’ve been applying for with less technical background after about a week of job hunting, so luck and timing may be important too.

    1. Thanks for this comment. Glad to hear that your advisor was supportive, but I agree that many linguistics professors aren’t particularly informed about the non-academic job market. I hope that will change as more linguistics PhDs get non-academic jobs: their advisors and their peers who later become faculty will have a wider non-academic professional network and will be able to advise graduate students on the best way to make themselves marketable in multiple career paths. If the student is considering a career in tech, for example, I don’t see a reason why their advisor shouldn’t encourage them to take summer internships in the private sector, attend a data science summer program, develop open source projects and post them on github, collaborate with research groups that do more applied research, even write cute blog posts that show off their data analysis skills.

  5. A few thoughts, if I may. First, I think you should disclose that “this post was written by someone who very recently landed a (very well deserved!) tenure-track job at a great institution” 🙂 (Okay, you can leave out the “very well deserved” part, but I thought I’d editorialize.) I say this because it’s hard to avoid that outcome coloring your impressions of the situation, even if you sincerely try (which I believe you have).

    Second, there seems to be an implicit assumption towards the end of the post that quantitative methods & skills are desirable in and of themselves, perhaps even independently of their utility on the non-academic job market (cf. “Can you receive solid quantitative training in your program?”). For some of us that is simply not true. I happen to think that the obsession with quantitative methods is something of a scourge on language science nowadays – part of the more general cargo-cult of p-values and science-y-ness. But even if you disagree with me on that (and I’m fairly certain that you do), my point is merely that some of us do think this. And for those of us who do, “Can you receive solid quantitative training in your program?” is kind of like “Can you receive some training in public accounting while doing your linguistics PhD?” – it is simply not what we came to linguistics to do. And you’re probably right to suspect that for people whose interests trend in these directions, the non-academic prospects for the PhD-holder aren’t as good.

    1. Thanks for these comments. In case that wasn’t clear, my goal in writing this post was to balance (to some extent) Rachael Tatman’s categorically negative post. It was definitely not my intention to argue that everyone should get a computational linguistics PhD. I think we agree that there’s value in both quantitative and qualitative approaches to linguistics (and that there’s plenty of bad quantitative work). My point was not that no one should write theoretical syntax or sociolinguistic ethnography or “grammar of X” dissertations, but that it makes sense to keep your non-academic job prospects in mind when you choose what subfield of linguistics to focus on or which department to go to for your PhD, and keep them in the back of your head as you choose what internships to do or what connections to try to develop as you’re working on your PhD. Just as it makes sense to take these considerations into account in the (unlikely) event that you’re equally excited by the idea of a PhD in English and a PhD in Computer Science. Those non-academic job prospects don’t necessarily have to be in the tech sector, of course; that just happens to be the industry I’m most familiar with.

  6. I did some googling of the blog poster’s background. He received a Bachelor’s degree in Maths (and Linguistics) and his research is on computational linguistics. I think we all know that computational linguists already have an easier time finding employment in either academic or non-academic settings. So it is no wonder he has a rosier view of things. Linguistics is not Computational Linguistics; and most linguistics PhDs do not have a Maths BSc degree (the poster is making a point on the importance of acquiring quantitative skills).

Leave a Reply

Your email address will not be published. Required fields are marked *