Linguistics as a field is different from how it was 20, 30, 40 years ago. There was a time when linguistics was dominated by a couple of sub-fields that relied overwhelmingly on similar analytical tools, working at roughly the same level of analysis. Things are different now. There are more people working in more areas, using a more tools to ask questions at multiple levels of analysis.
The broadening of the field is clearly reflected in hiring trends for young faculty, and in the interests of students who are applying to join linguistics programs. The scientific demographics of the field has clearly evolved.
But curricula have not kept pace with the changes in the field. Almost all leading linguistics graduate programs in the US have a curriculum that closely follows the model that has dominated for the past 40 years, giving privileged status to some sub-fields and treating others as second class citizens (‘boutique’ or ‘Cinderella’ sub-fields). This is hurting a generation of young researchers, and I think it’s also missing a big opportunity for the field.
This post is a reaction to similar conversations that I have had with many people over the past couple of years.
Typical linguistics graduate programs in the US follow a model in which beginning students are expected to follow a prescribed set of courses. These consume most of a student’s attention in the first year. These courses are generally dominated by courses in ‘theoretical linguistics’, in phonology, syntax, and semantics. Courses in fields such as psycholinguistics and computational linguistics are more commonly offered as electives to be taken later in the student’s program, around the same time that students are also selecting research topics for their PhD qualifying projects (often called ‘generals’). This curriculum follows a model established at MIT, probably in the 1960s, and that has been copied across the continent.
The fixed curriculum made perfect sense 30-40 years ago, when the field was small, when the required courses covered the sub-fields that the faculty specialized in and that almost all of the students wanted to focus on in their research. But as the research in the field and the composition of departments has evolved, curricula have not kept pace.
Exploding the curriculum
My own department at the University of Maryland is an outlier. (I’m sure there are others — use the comments section to point to these.) In the early 2000s the department decided to throw out the traditional linguistics curriculum, and to have no required courses. Students must take courses, of course, but there’s no specific course that they are required to take. We offer a lot of different courses at the foundational graduate level, and students must take at least 6 of those, including two 2-course sequences, to encourage depth. But the choice and timing of these courses is entirely up to the student and their advisor(s), who are free to tailor the curriculum to fit the student’s individual needs.
At the time when we exploded the curriculum, we were also an outlier department in other ways, e.g., barely half of our faculty had a PhD in linguistics. This meant that we had more need to liberalize our curriculum than other programs had at the time. For example, at the time when we dropped all required courses, a major part of departmental funding was coming from computational linguistics, and faculty in that area were supporting their students through grants at a far higher rate than in other areas. It just didn’t seem right to recruit computational students, whose funding would be coming mainly from computational grants, and to then tell them that they should set aside their interests for the first year while they focused on other areas that were more important.
This approach has worked out really well for us. It allows students to take courses in the areas that they are most interested in as soon as they arrive. It has allowed students to take ownership of their own course of study. It certainly hasn’t led to our students being narrowly trained. On the contrary, they have gone far beyond the minimum that is required of them, within and beyond our department. Students are satisfied because they’re never taking something that they’re forced to take. Faculty are satisfied because they never have students in their classrooms under duress. The department has benefited in many other ways. It would be hard to argue that the department has suffered: applications, recruitment, job placement, funding, and pretty much any other measure that you can think of has looked better since the change. (Of course, correlation is not causation, and the curriculum wasn’t the only aspect of our program that changed.)
Cinderella subjects
We expected that our curriculum change would raise some eyebrows when we first implemented it. It did. But we have been surprised at what has and has not happened over the intervening years. In terms of hiring and recruitment, we’re less of an outlier now than we were a dozen years ago, e.g., most leading linguistics programs now have one or more psycholinguists on their roster, and many also have computational linguists. They have diversified their faculty in other ways too. But our curriculum is still an oddball. In almost all other leading programs, the traditional curriculum is largely intact. And that’s the problem.
Many young faculty have been hired with the goal of diversifying departments’ research portfolio, broadening the tools and methods available, and possibly increasing external funding. New students have been attracted by the new opportunities. But these faculty feel like second class citizens when it comes to the curriculum. It’s hard for beginning students to take their courses. That makes it harder for the students to choose their area for their initial research projects and qualifying papers. And so they face a disadvantage in training students. In contrast, students in psychology or computer science programs are more likely to be diving into research in their primary area right from the start of their graduate careers, making them more competitive by the time that they graduate. And since research in these fields depends so much on faculty-student collaborations, it also holds back the faculty research. The feeling of being second class citizens is made worse when some colleagues refer to sub-fields like phonology, syntax, and semantics as “core” areas of linguistics. (I think the current fashionable term for that is “micro-aggression”.)
The situation that I’m describing is not fabricated or isolated. I’ve heard the same story from faculty and students in many different institutions. [Edit: and since writing this post, I have heard from more young faculty who have shared their version of this story.]
I have heard various arguments for why the traditional curriculum should be preserved. I’ll list them here, with comments. You won’t be surprised that I’m not particularly convinced by them.
1. Students need training in ‘core’ areas before applying this knowledge in other domains.
Yes, it’s helpful to know some linguistics in order to do good work in computational linguistics or psycholinguistics. But you don’t need a whole year of graduate level course work in syntax, phonology, etc. For better or for worse, the state of the art in computational linguistics and psycholinguistics generally does not depend on the very latest results in theoretical linguistics. And beginning linguistics graduate students know more than they used to. Advanced linguistic expertise is definitely an asset, but it’s not a pre-requisite.
2. Required courses are needed, because students need breadth.
Yes, students benefit from breadth. But there are many different ways to be broad as a linguist. Our graduate students at Maryland have no required courses, yet their training is extremely broad. Different individuals are broad in different ways. So no need to worry.
3. If students haven’t taken courses in X, Y, or Z, they won’t be able to get an academic job. You have to know these things to teach an introductory course.
Not true. Any smart linguist can deal with an introductory course. And few are hired for that ability. If you really want to ensure that students are employable in academia, load them up on courses in computational linguistics and second language acquisition. That’s where the jobs are.
There is a variant on this concern that perhaps is true. Most hiring is still done by faculty who specialize in traditional sub-fields. Many of them are worried about the invasion of hot-headed youngsters whose work looks quite different from theirs. If you’re looking to get hired, but you give the impression that you don’t understand or appreciate what the people doing the hiring care about, then you’re at a disadvantage.
Another variant on this (thanks to Brian Dillon for raising it) is that many teaching jobs require individuals who are versatile enough to teach advanced undergraduate courses in multiple areas, especially the ones that dominate the traditional graduate curriculum. I’d agree that broad teaching skills are valuable, and that we need to do more to prepare students for this. But I don’t see how it follows from this that everybody needs exactly the same breadth. High demand undergraduate courses include second language acquisition, language and gender, historical linguistics, language and computers, language and advertising, and so on. The argument favors breadth, not uniformity.
4. How can you call somebody a linguist if they don’t know X?
Quaint.
The field has grown up now. It covers a broad space, and it’s just not possible to keep up with everything. We should expect that different experts in our field should have different kinds of expertise. That’s a mark of a mature field.
5. It’s important for beginning students to go through a fixed set of courses, so that they build a cohort.
We agree with the goal, but in our experience this can be achieved without forcing all students into the same mould.
6. We need to make certain courses required, so that the courses reach minimum enrollments.
This reflects a harsh reality for many graduate programs. They’re expected to offer graduate courses, and those courses have to meet enrollment targets. But that is hard to do in the face of declining student support and declining student numbers. A compulsory curriculum helps to create some degree of certainty. In our PhD program we cannot meet enrollment targets based on beginning PhD students from our own programs. But enrollments haven’t been a serious problem for us. This is in part because students take the foundational courses on different schedules, but also because we draw in many students from other departments (just as our own students often take courses in other departments), and because we draw in a small number of our best undergraduates. My own foundational course in psycholinguistics, which extends over a full year, serves beginning linguistics graduate students, but they are often in the minority in the course.
Why this matters to linguistics
But why complain about this? After all, I’m delighted with our non-curriculum, and it has helped us to attract excellent students and faculty. So why care about the curriculum elsewhere? In part, it’s a concern about the impact on individual students and individual faculty careers. When we hire faculty and recruit students we need to set them up to succeed. But beyond that, I think that the fossilization of the linguistics curriculum is holding back the field, certainly in psycholinguistics, and possibly in computational linguistics. Why so?
For a long time, researchers with linguistics training have been marginal constituencies in psychological and computational research on language. But that may be changing. In psychology departments, language is not a growth area nowadays. Other areas are hotter, and more lucrative, e.g., social cognitive neuroscience. There are a few psychology departments that have big language groups, e.g., Illinois, Connecticut, Edinburgh, but it’s rare for that to be a priority these days. When I go to psycholinguistics conferences like CUNY, I encounter vastly more more linguists than I did 15-20 years ago. It’s conceivable that in the next few years a lot of psycholinguistic research will shift from psychology departments to linguistics departments. But fossilized curricula put the brakes on that. I’m less knowledgeable about the state of computer science departments, but there’s also a general sense that linguistic expertise is on the ascendancy again, after many years in the wilderness. As brute force big data methods reach their limits in NLP, and as tools like Siri and Google Translate raise consumer expectations for the quality of language tools, there’s more and more need for a deeper understanding of how human language works. This creates an opportunity for linguistics programs to play a big role (… and one that can bring funding with it). But, again, fossilized curricula create a barrier to training computational linguists in linguistics programs.
If you have more arguments why the traditional curriculum is better for the field, I’m curious to hear them.
Yeah. I’m debating in my mind, not having really had to do this but having seen a few examples of curricula. There’s no question that Maryland has got the best curriculum I’ve seen in cognitive science. How other institutions should implement a quality curriculum, I’m less sure.
One thing would be to make the difference between core courses and (personalized) required courses. De facto, even if we didn’t have core courses, we all I think pretty much listened to our advisors on what courses we should take. So not having core courses is not the same as anarchism. It’s more that top down guidance is then free to do what makes sense for each student.
More than that, I think de facto pretty much everybody at Maryland takes syntax and psycholinguistics. You can correct me with real stats. I don’t know if everyone was top down encouraged to do so, but I do know that amongst our year we were generally in agreement that we would’ve been foolishly missing a great opportunity if we didn’t take at least the first semester’s worth of both.
At that point I start debating whether there is some value to having courses that a large swathe of the department takes, de facto or de jure. Most departments would probably debate core courses in terms of whether they deliver X, Y, and Z pieces of information, which might be necessary for understanding something else, and that’s a bit the way you’ve got it posed here. But there’s even more important high level stuff sitting in the courses we all took. The important takeaway from syntax for experimental people was that linguistics isn’t crazy, and the theory is motivated, non-trivial, and interesting. Equally, the important message from your psycholinguistics class for theory people was that psychologists aren’t stupid, we’re all grappling with same elephant. I think that getting that right should be a sine qua non for a curriculum in language science, required courses or no.
So I don’t think Maryland has got a buffet the way this post might make it sound. There’s still strong and quality guidance from the faculty, and there are still core things that the curriculum wants to be delivering. One could take that either way. It’s positive evidence that you can avoid throwing out the coherent baby with the bathwater, but it’s also evidence that you probably shouldn’t throw out core courses if the only alternative is turning the education into a free-for-all.
I do think there’s something particularly functional about Maryland that makes it possible to be coherent without core courses, which is that, at least within Linguistics, and to a very nice degree outside, everyone has a fairly good sense of what everyone else is teaching, and why that might be important to their own students. And everyone seems to have a pretty clear idea of what a well rounded education means. And the students talk to each other. About science. In that case you can probably trust that students are going to wind up in the right places without central control. I’m not so sure that would work equally well in every place.
Well, I wouldn’t say that we have the ‘best’ solution, but the ‘choose your own adventure’ approach leaves freedom for people to create some good individualized solutions. And it avoids creating a caste system among sub-fields. You raise a good point that the label on a course doesn’t necessarily reflect the main moral of that course.
I know that when I became interested in psycholinguistics, there was not much support within the Linguistics Department at my alma mater. Fortunately, there was a psycholinguist in the Psychology Department who was open to experimental work, and doing that work in an indigenous N American language. And trying to find a job as a psycholinguist in a linguistics department was TOUGH 12 years ago, which is why I am in Educational Psychology now (where research on second language acquisition and pedagogy, reading, and individual differences were/are valued). More opportunities outside of the “core” areas now, but still blinkered by tradition…
I agree with you Colin, in principle, but the variant of point 3 that you make is absolutely on the money, and part of our job is to professionalize students to equip them for the actual academic job market they face, at least in linguistics departments–ideally, of course, psychology and CS depts would start hiring such grads, but–correct me if I’m wrong–the hiring of newly minted PhDs in psycholing and computational ling from linguistics depts tends to be in linguistics depts, not in psych and CS, sadly. This is not a hypothetical: a recent PhD with excellent work in comp.ling. was on a campus interview and was introduced to one of the senior faculty in the hiring dept who is famous for certain things in sem/prag, and the candidate not only didn’t know his work, but even asked him “So what do you work on?”. This is a level of ignorance that isn’t unique to such “Cinderella” (new term to me) fields, but it certainly made sure that this particular candidate didn’t get the job.
(I can also attest to your point about CS needing more savvy ling: went to a talk by Slav Petrov from Google Research recently whose upshot was “we need syntax”. That doesn’t mean “we need two semesters of Minimalist syntax”, of course, but it means more syntax than a typical CS or psych grad has ever seen.)
@Jason. Yes, students need to be prepared for future success. In multiple career paths. That includes professional development, diverse skill sets, broad communication abilities, collaborative skills, and much more. And it also includes celebrating rather than stigmatizing careers outside of academia. But I don’t see any of this as motivation for the standard caste system in the linguistics curriculum. It’s a motivation for more extensive rethinking of how we prepare students.
On the question of where graduates in psycholinguistics and computational linguistics are landing jobs, they wind up all over the place. Among our computational linguistics grads, Chris Dyer is on the CS faculty at CMU; Mona Diab is in CS at George Washington U. In that field what matters is the kind of work you do, not what it says on your degree certificate. Those two have serious NLP chops, in addition to linguistics. Among psycholinguistics grads from here and elsewhere, there are a few who are in psych departments: Laura Wagner (Ohio State), Jennifer Arnold (UNC), Nina Kazanina (Bristol), Julien Musolino (Rutgers), Whit Tabor (UConn), Florian Jaeger (Rochester). Many will end up in linguistics departments, but that’s partly due to where the jobs are currently, different expectations for the CVs of new hires in ling. and psych., and where the linguistics grads prefer to land (they like being in ling. departments).
Anyway, I’m glad that you agree in principle. And it’s good to follow principles.
This argument seems circular: Why should a psychology or computer science department want to hire students who are trained extensively in theoretical linguistics, with psychology or computation being an afterthought?
Being a good computational psycholinguist requires a different kind of breadth than is found in any of the three traditional disciplinary curricula in linguistics, psychology, or computer science. The appropriate ‘core’ set of courses encompasses elements of all three. Students who pursue this training are quite versatile. For example, Sharon Goldwater has a PhD in linguistics and works in informatics; Micha Elsner has a PhD in computer science and works in linguistics; Lisa Pearl has a PhD in linguistics and works in cognitive science; I have a PhD in cognitive science and work in linguistics.
I’m always amused when linguists fail to recognize why theoretical syntax and semantics have ‘lost steam’… The reason is blatantly clear and very depressing: the fact is that 99% of theoretical syntax and semantics is useless (not only for comp ling, but for any practical use whatsoever). The field is stuck in an early 80’s view of language that produces virtually zero facts about how languages actually work. To put it differently, Siri won’t be benefiting from current syntactic and semantic theories any time soon, given that the latter are a joke. Computational linguists have naturally stopped caring about theoretical syntax for decades, and so have psycholinguists (except a small faithful handful, of course). But there is hope, I agree. Once theoretical linguistics departments start vanishing, maybe the field will recover and finally become a respectable science, with palpable, observable, clear results. Some of which may even help develop new technology to actually benefit people, like any other respectable science.
Frank, I very much disagree. Those fields have a great deal to offer, keep delivering new findings, and create eminently testable claims. I have zero interest in overthrowing those fields, and have argued strongly in defense of their methods elsewhere. Incidentally, state of the art NLP is now looking more towards linguistics again. The pendulum is swinging. The aim of this post is not to question the scientific credibility of certain subfields. It is to question the way that individuals and programs hurt some of their colleagues and students by holding onto privileges in the curriculum.
NLP has begun to become more linguistically informed (using tree structure rather than Markovian shallow methods, first-order logic semantic representations, etc.), but you are leaving out the fact that it is not being informed by modern syntax and semantic theory in any way shape or form. You won’t see any ‘little v’ or ‘force’ or any other such diacritics on current treebanks. NLP is being informed by good-old (descriptive) linguistic information. Nothing else on the curriculum is of any use. Job offers for theoretical syntax and theoretical syntax are vanishing, and good riddance to them.
Interesting discussion. I would like to raise a point about the difference between the US and Europe (or at least Norway) which I think is relevant. It is good and important that many undergraduate programs are now delivering basic foundational syntax, phonology and general linguistics from a broadly modern perspective. This frees up graduate programs to build breadth and strength in the way that Colin is advocating and the way that Maryland is doing. In such a context, I agree that it is antiquated and not particularly useful to train ALL linguistics phds through 2 terms of Advanced Minimalism. The situation is often different in countries where there are NO or few undergraduate linguistics programmes and where phd candidates are admitted with Masters degrees in some `language’ as taught by very traditional methods and with different educational goals. In such countries, it unfortunately makes sense for everyone to get Phonology I, Syntax I and Semantics I in addition to the psycholinguistics, language acquisition etc. that the department offers. Maybe this just means that such countries are going to end up lagging behind. I am currently fighting to get a new undergraduate programme on the books in Tromsø but I am afraid it will be an uphill battle.
Thanks Gillian. I agree that any program needs to be responsive to the circumstances of the students that enter the program, and it could make a difference if your cohort is starting from already having published research articles vs. knowing very little of relevance. Your situation raises two issues that we should probably distinguish: what do students need by the time that they’re done, and what need is there to put some subfields ahead of all others. My main worry is about the second of these, the ordering question. It creates an implicit hierarchy that has negative effects. In fact, more negative effects than I realized. This post has elicited a surprising number of private emails from faculty in different institutions who have described their frustration with the way that their program’s curriculum impacts them and their students’ research development. In our PhD program, both Howard Lasnik (syntax) and I (psycholinguistics) *claim* that our grad courses are set up in a way that allows students to take them who enter with practically no background. And we have students from other fields who take the courses as evidence that it can be done. In both cases, I think it’s a little optimistic, and students who have almost no background in the field(s) find the courses to have a steep learning curve. But in both cases I do think it’s true that you can do a surprising amount with minimal linguistic background.
Hear hear! This really strikes a chord with me. I am a “Cinderalla” (psycho)linguist, who had to blaze my own way through graduate school. It took me eight (yes eight) years from entering grad school to finish my Ph.D., in part because we had to spend three years taking courses in “core” linguistic areas, with very little chance to actually do research. And this was after I had an undergraduate degree in linguistics from the same institution. They treated the first few years of graduate school as a warm up period, and the assumption was that people would only go on to do “normal” linguistics (syntax, phonology, etc), and that you couldn’t possibly do that kind of research until after several years of coursework.
I had an epiphany during my first year that I really wanted to pursue psycholinguistics, but in my department there was really no way to do so, so I had to find my way to the psychology department, where we luckily had a pre-eminent psycholinguist who was willing to let me hang out in his lab. But this was only *after* my third year of graduate study, when most of my coursework was done. I when I started doing psycholinguistics, I essentially had to start a new degree from scratch, which explains the long post-MA period.
I did a post-doc in a psychology department, where most of the graduate students in both the linguistic programs and the graduate students in psychology doing psycholinguistics really did take a lot of shared courses (both in statistics and experimental design and linguistic theory), but importantly, because of the lab-based structure, students were doing publishable-quality research for their first year projects. MA projects from these sorts of programs (including other psycholinguists from psychology departments in other parts of the country) often end up in top-tier journals (JML, JEP, Cognition, etc), and importantly, they don’t seem to lack any linguistic sophistication to me. I am still floored, because my graduate school experience was quite different — It was expected that students wouldn’t have the background to conduct “real” linguistics research (i.e., syntax or semantics or whatnot) until several years in the program. But psychology students (and students from Colin’s lab) routinely graduate with 3, 4, 5, or even 6 first-authored publications.
I landed a fantastic job in a linguistics department, where I am very happy — even though I am a “Cinderalla” linguist, they value and appreciate what I do, and have been extremely supportive in getting me space and resources, and I have wonderful, bright, enthusiastic students.
But right now I’m struggling a bit to figure out a way to mentor my incoming students (who have strong backgrounds in linguistics, but maybe not experimental design) to start doing publishable-quality research from the get-go, while they take 3-4 courses per semester for the first years of the program. In our department, luckily, a course in psycholinguistics is required for all graduate students and most take it in their first year, but it is quite introductory (so it is appropriate not just for psycholinguistics, but also syntacticians, pragmaticists, etc).
Still, this doesn’t leave much time for research in the first year, and over the years. This is especially problematic for people working with me doing ERPs, where it can take a 6 months to write a good stimulus set — and this happens only after you’ve had time to come up with an idea, read the relevant literature, formulate a specific research question, an experiment design, etc. And then collecting the data can be another 6 months, especially if you’re looking for linguistically atypical people (in our case, bilinguals and second language learners). So doing a publishable-quality 2nd year qualifying paper becomes quite tricky, but I’m trying find ways to make this happen for my students. And then, over the years, experimentalists will have to take many additional courses over and above the normal ling curriculum (statistics, Matlab, etc), making the course burden exceptionally high.
I know students who come out of my lab are going to be competing for jobs with psychology students who start research right away, and who have multiple quality publications at graduation. A real challenge will be finding a way to structure linguistics graduate programs so they are still that — LINGUISTICS graduate programs — but also embrace the fact that linguistics is increasingly a big tent field, with strong ties to other field. I hope change comes to our curricula, though I know it will be slow. I’m doing what I can in the meantime to make sure my students both have the linguistic and (important for ERPs) technical backgrounds to thrive and compete. I’m really glad Colin posted this and I am glad it’s creating a buzz across the field.
We’ve had something similar in place in the Arizona graduate program for many years, though only ‘semi-exploded’, not fully exploded. Our students have to take Syntax I and Phonology I, and typically do that in the fall of their first semester. Otherwise, they get to do something like what you describe, Colin: they have to take 5 other courses in any of 5 subfields of their choosing, and also ‘Major’ in one of them by taking a couple more, to encourage depth, as you say. We find it works well in supporting the development of a diverse, engaged, and excited group of researchers, but also turns out students with a good basic understanding of sound patterns and syntactic structure, which we feel still does seem to be important for understanding Language with a capital L.
In addition to the usual suspects, including an NLP/computational/human-language-technology stream, a psycholinguistics (adult or developmental) stream, and a joint PhD between Linguistics and Anthropology, we also offer the only concentration within a PhD program on language documentation and revitalization that I know of. And our students in recent years have created a strong cross-current of interaction with our Speech, Language and Hearing Disorders department, so we have a strong group specializing in that kind of work as well. And we’ve also got the Second Langauge Acquisition and Theory PhD program here as well, which feeds us a lot of students who are working on SLA. Anyway, moral of the story is: Big Tent linguistics programs like ours have a lot to offer.
BUT: MANY linguistics departments and programs are too small to be able to do that kind of eclectic thing. They do better by focussing in a specific subareas, and recruiting students in those subareas, and would not find it easy to create a program and keep offerings going that covered the whole range of possible topics in the language sciences. Those programs are still turning out super duper, qualified academics who do great work and succeed in the job market. Fox and Hedgehog, etc. etc. etc., place for everyone, let a thousand flowers bloom, etc.
Oh — and re writing a ‘good stimulus set’ in one’s first year — in my experience, you just can’t do that without having a strong background in both syntax and phonology and (usually) morphology. Just can’t control for many relevant factors unless you have some idea of what might be going on under the hood. Of course, that’s what lab meetings are for, so others can help you workshop that kind of thing; maybe that’s enough of a resource….
Thanks Heidi, good to hear about the Arizona experience, which seems to work well. Re: smaller programs, I agree that if a program’s research spans X and Y but not Z, then it makes perfect sense for the research training to emphasize X and Y. The specific concern is what happens once a program broadens its research expertise, and how the curriculum should respond to that. This is exactly the situation in linguistics in recent years, as so many programs have broadened their research portfolio.
If you look at the papers by current students and recent PhDs that are accepted under anonymous review to linguistics conferences, you see a number of good departments well represented, including yours and mine (your linguistic alma mater), with different curricular models and different emphases. I think that is sufficient to tell us that there is more than one good way to educate linguists for the current intellectual and professional world. (As is the fact that departments with these different curricular models hire each other’s PhDs.) Will the future be very different, and what will be the best way to educate students then? We can all place bets on the answers, but none of us has a reliable crystal ball, which is why it is healthy for the field to have some diversity of models right now. Thank you for the post.
Thanks David. I wholeheartedly agree that trying out different models is a Good Thing. And I think that there has been much to celebrate in that regard in recent years, in terms of how departments have chosen to develop their faculty expertise in specific directions. Different places have different profiles, and we shouldn’t want them all to conform.
The focus of my post is on a different issue. It’s about the alignment of research expertise with the research training curriculum, and about how misalignment creates disadvantages and the sense that some colleagues and students are second class citizens. Personally, I haven’t experienced this problem directly, but I have heard from so many who have expressed this (and more in the past couple of days). There may be multiple solutions to the problem, and our program’s approach is just one. It would be great to hear about other possible solutions, or to hear arguments that the problem is imaginary.
At the risk of making you blush, I’ll mention one example from my own past. When I was a graduate student, I was part of a government-funded training program that you (co-)directed. Funding agency rules treated foreign students like myself as second class citizens, but you, as PI, made a special effort to erase any differences between US and foreign students. That made a huge difference to me, for which I’m very grateful. More recently I’ve faced similar issues with programs that I’ve been running, and I have been strongly influenced by my own (good) experience as a student. I have also repeatedly highlighted to the funding agency the problem that their policy creates. I couldn’t claim any causal role, but I’m happy that the agency has recently modified its rules, so that foreign students are less marginalized than they were before.
Thanks for this very useful post, Colin! I especially appreciate the point, which I hadn’t considered, that having large numbers of required courses in ‘core’ areas tends to disadvantage profs and students in traditionally less central areas.
I very much agree with what you are saying here, Colin. I don’t think the current 60’s style graduate program in linguistics is even adequate for training students in a theoretical field like mine. Here is a post on my own blog expressing a similar sentiment.
http://blogs.umass.edu/kratzer/2015/06/18/on-teaching-graduate-students/
“When I visited the University of Maryland during NASSLLI last year, I got to know some of the graduate students there. I saw that even first year semantics students were already working on their own original research projects. It’s not that they weren’t taking any classes. They were, but they were at the same time contributing members of research groups. This is also the way PhD students are trained in the Berlin School of Mind and Brain or at ILLC in Amsterdam. The PhD students in all three of those programs are enthusiastic and self-confident scholars who are deeply immersed in their research, mentored by specialists in their field, as well as by a Graduate Program Director. Maryland has a post-baccalaureate program providing a bridge between College and Graduate School. The Berlin and Amsterdam programs have associated MA programs that offer introductory and more specialized graduate-level classes, as well as ‘methods’ classes. In addition, there are workshops and mini-courses targeting PhD students. I think programs like these might be the future of graduate education in the Cognitive Sciences, including linguistics. The time of the ’60s-style PhD program in linguistics that was so successful in the US may be over. Training in linguistics is bound to become more like training in the sciences. “
There’s another distinct (but related) sense in which UMD’s curriculum is unusually flexible I think: as well as flexibility with respect to the balance between the “traditional core subfields” (syntax, phonology, etc.) and “non-traditional/Cinderella subfields” (psycholing, comp ling, etc.), there’s flexibility with respect to the balance *within* the traditional subfields. For example, I ended up lining up most closely with some of the traditional subfields (syntax/semantics), and yet never took a course in phonetics, phonology or morphology in my life; instead, my sidelines were dabbling in bits of computational linguistics and psycholinguistics. So as well as endorsing the idea that you can, say, be a useful psycholinguist having taken syntax courses but not necessarily phonology courses, the curriculum also endorses the idea that you can be a useful syntactician by having psycholing and/or comp ling rather than phonology as the additional strings to your bow. This issue — whether knowledge of all traditional subfields should be required/expected of those whose primary subfield is a traditional one — also seems to be a somewhat controversial one, and perhaps slightly less debated than the balance of traditional and Cinderella subjects.
Colin, what are your Ph.D. placement rates over the past five years? This is not a rhetorical question/point; it’s against the background of the kind of argument that Brian Dillon raised.
Our program’s PhD placement has been quite good in recent years. An up-to-date list is on our alumni page. The vast majority of recent PhD graduates are currently in academic positions: tenure track, research scientist, or postdoc. We can point to a number of cases where students’ breadth training was directly linked to their post-graduation employment. I would not claim that the curricular changes are exclusively responsible for this, but I would say that they were a key enabler for other things that happened later.
I can’t believe that absolutely no one is stepping up to take on the role of stick-in-the-mud. Allow me, then!
These à la carte models of grad curricula are fine if the student already has an informed view of what they want to do within linguistics (an involved and proactive advisor can help you try to find the answer, but it’s the kind of thing that the advisor can’t really “do” for you).
Now, obviously, most if not all students applying to a grad program in linguistics think they know what they want to do. But we all know many stories of people who came in thinking they want to do one thing, and ended up doing quite another. And I’m not just talking about people who switched subfields entirely (there are plenty of “defectors” from syntax to phonology and vice versa). I’m also talking about people who made subtle shifts within their subfield, shifts that would not have been possible without the requisite exposure to a broad base of (da-da-dum) “core” courses.
Let me take myself as an example. I came into grad school thinking I would be working on what I now like to affectionately call Big Syntax: wh-dependencies, islandhood; very long-distance/macro stuff. It turns out I ended up being a morphosyntactician, the kind of syntactician that is obsessed with little morphemes and gets ‘accused’ by other syntacticians of actually being a morphologist. (Heavens to Betsy.) Would this have happened if I had been allowed to navigate my way around any curricular exposure to phonology and morphology? Hard to say, but I’m betting no.
One thing you could say is who knows if I and/or the world would be better off if I had stayed in Big Syntax. That’s hard to adjudicate, but I can tell you I’m professionally very happy with this outcome. The more general point is that an advanced education, I think, involves a gradual whittling-down of what is initially an overly-wide focus, until you hit on what you really want to do (think of how you answered the question of “what are you interested in doing” in high-school, then in college, then in grad school). And as much as we like to think of ourselves as rational agents, that process often requires that a bit of noise be injected from the outside. Something like an intellectual “eat your vegetables.”
Of course, no one knows if, for person X, the relevant “vegetable” will be theoretical phonology or theoretical syntax or psycholinguistics or computational semantics etc. etc. Any of these is just as likely to spur an unexpected scientific interest as the others. So the traditional theoretical subfields don’t have any privileged status with respect to the point I’m making. In a perfect world, a person would be required to take all of these different courses before achieving PhD candidacy in linguistics. And since we don’t live in a perfect world, I don’t think that establishing an implicit caste system among the subfields is the answer.
I’m just saying that the idea that nothing is lost by relaxing these traditional requirements, or that what is lost can be recouped via “cohort effects” and interaction with your immediate intellectual community, strikes me as a tad optimistic.
Fortunately or unfortunately, I happen to be in the same department Colin was talking about (hi, Colin!), and not in a position to enact unilateral curricular changes (thank god), so I will have a chance to see first hand if/how I am wrong. I’m willing to have my mind changed.
This doesn’t sound like a stick-in-the-mud position, Omer. I just got back from a very interesting couple of days in Ann Arbor, hanging out with Jon Brennan, who made a similar observation. Flexibility creates choices, but it is incumbent on students, mentors, and programs/communities to encourage good choices. And what counts as a good choice is going to be different for different people in different programs. Your observations are entirely consistent with what this post advocates. I wouldn’t encourage any PhD student to focus their training so narrowly that they only learn about one thing.
Another reason for greater curricular flexibility may be related to how adults learn. Some research suggests that we learn better when we know why we’re learning what we’re learning. There’s some “need to know” learning in what Colin describes here.