How graduate school is like (American) football

This is an email that I wrote to a graduate student as they finished up an exceptionally heroic dissertation-writing experience. All identifying information has been changed: Robin is a gender-neutral name. Dalmatian was a Romance language spoken in Croatia. (It became extinct when the last known speaker, Antonio Udina, died in an accidental explosion in 1898.) Slivovitz is a delicious plum brandy that I recommend you drink with roasted lamb. Ilse Lehiste was a major phonetician and historical linguist of the 20th century, and the author of the classic Suprasegmentals. (I applied to Ohio State planning to study with her, not realizing that she had just retired—see my post on how not to apply to grad school.) David Sternberg is the author of the excellent How to complete and survive a doctoral dissertation. “T&K” is Sarah Thomason and Terrence Kaufman, the authors of Language contact, creolization, and genetic linguistics—when my undergraduate advisor made me read it, I lost all faith in distant reconstruction (in historical linguistics, at any rate) and ended up doing lexical semantics in a medical school. Banitsa is a delicious Bulgarian pastry.

Hi, Robin,

I’ve read through the most recent chapters that you sent.  Lots of progress–that’s great.

I’m going to start with what sounds like bad news, but actually isn’t: you should rethink this–a lot.  You are chasing some very specific hypotheses, and betting a lot on them.  What I recommend that you do instead is take a step back and look at the more general contributions that your dissertation can make, focus on those throughout the first chapters, and leave the theoretical implications for a later chapter.

What follows is based pretty heavily on my own approach to science in general, and other people could certainly have their own takes on the scientific enterprise, so if you decide to take my advice in this matter of the overall direction of your dissertation, you should discuss it with your advisor at the minimum, and preferably with the rest of your committee, too.  (Feel free to forward this email to the rest of your committee–I thought about copying everyone, but decided to let you stew on this yourself and then make your own decision about whether or not to do that.)

kennedy football daylight images
Picture source: http://www.azquotes.com/quote/156221

One of my fundamental ways of thinking about how to figure out what to publish in general, and how to focus a dissertation in particular, is David Sternberg’s notion of “daylight.”  “Daylight” is actually an idea from American football.  The goal there is to move a ball in a particular direction, while a bunch of big bruisers try to stop you from doing so.  [“Big bruiser” and other obscure American English expressions explained at the bottom of the post, as always.] The most basic advice for how to accomplish this is to look for an open spot between the big bruisers–one through which you can see “daylight”–and then run through it.  (Other approaches would be to try to knock the big bruisers down and run over them–if one wants to finish one’s dissertation in this lifetime, or score a goal in American football, it’s definitely easier to run through the open spot.  I think so, anyway.  ([Note to blog readers: no, I don’t play football–the real tough people play judo.]) 

In the context of a dissertation (and, to some extent, scientific publishing in general), “daylight” is whatever you can contribute to science that hasn’t been contributed to science before.  There are a lot of forms that “daylight” can take:

1) No one has ever written about X before.

2) Someone has written about X before, and/but you looked at it again, and found different facts.

3) Some theory makes a prediction about X, but no one has ever actually tested it.

4) Even better: some theory makes a prediction about X, but no one has ever actually tested it, and you have data that suggests that the prediction is wrong.

5) There are contradictions in the literature.

There are others, but these seem like the most salient ones, at least for a linguist.

So, what do you want to see in a linguistics dissertation, in particular?  Personally, I want to see two things.  In particular, I’d like to see the two main things in (American) linguistics:

1) Description

2) Theoretical implications

football run to daylight 58106302-19ef-4fab-8b03-e665e1731df5_red_rtd_itunes_3000
Picture source: http://www.azquotes.com/quote/156221

Back to the generalities: in general, you want (or I want, at any rate), in science in general, to work on questions such that whatever the answer turns out to be, it’s interesting.  For example, (3) and (4) on my partial list of kinds of “daylight” are both versions of the same method: let’s test theory X.  Theory X works in the case of your question?  Great, if no one has ever published that before.  Theory X doesn’t work in the case of your question?  Even better.  Either way: you win.  You win because you’re not invested in any particular outcome–this is the time to be that most dispassionate of dispassionate scientists, only interested in knowing what the truth is.

 

vince lombardi run-to-daylight-9781476767178_lg
Picture source: http://www.simonandschuster.com/books/Run-to-Daylight!/Vince-Lombardi/9781476767178

If you buy all of that: let’s think about your dissertation in terms of where the daylight is.  Let’s also think about it as linguistics, per se–that is, in terms of description and theory.  When I write the following, I’m writing it on the basis of my knowledge of what the literature already covers, and you certainly know the relevant literature better than I do, so if you disagree, I’m going with your judgement, not mine.  I went through the current chapters and highlighted every place where you mention that no one has ever done something before, and you have:

1) Dalmatian is phonetically understudied.  This is the first acoustic phonetics study of Dalmatian.

2) “To my knowledge, there has never been any study performed to compare the acoustic variations in different dialect groups of Dalmatian speakers.”

3) From p. 16: “My data provides examples of phonological processes that have not been previously mentioned or evaluated.”

4) Does the acoustic observation of sound variation in Dalmatian show consistency in all Dalmatian dialects?  No one else has looked.

5) No one has ever studied code-switching in Dalmatian speakers before–you have.

Now the disagreements (see 5 on my partial list of kinds of daylight):

1) There is disagreement on the number of vowels and consonants in Dalmatian.  You have an answer.  (I didn’t say “the” answer–you don’t have to have “the” answer–just an answer that is well-supported by data and by argument.)

2) Lehiste gives an excellent account of Dalmatian contrasts, but doesn’t capture the variation between dialects.  You do.

3) P. 29: “…there is a disagreement among scholars, whether or not this sound belongs to the vowel inventory of Dalmatian.”

Predictions:

1) Slivovitz says that U and o are allophones–you show that it’s more complicated.

2) Banitsa predicts palatalization–your data shows otherwise.  (Your effort to elicit the palatalized consonants was not “unsuccessful,” as you described it–it was a *successful* effort to find out whether or not they’re there.  Turns out they aren’t, for your speakers.)

3) From p. 20: “According to Mongo (2011), words typically have a single primary stress, which falls on the final syllable….My data shows that stress can also appear on the first syllable, as well as the penultimate syllable.”

4) Slivovitz claims all three insertion processes for all three dialects; you show that it doesn’t happen in Central.

5) P. 25: “The results of my examination demonstrate whether or not Thomason and Kaufman’s assertion applies to Dalmatian.”

Would any single one of these get you a PhD?  Unlikely.  Should all of them get you a PhD?  With some examination of the theoretical implications: yes.  I just gave you THIRTEEN contributions that your dissertation makes, and we haven’t even gotten to the stuff that you’ve mainly been focussing on, which is the whole language contact thing.

So: you’ve got the descriptive part of this covered.  Now you need some theory.  The theories: (a) language contact, (b) the internal, community-identity model (T&K) versus the structure model.  You know what you think Croatian contact might throw into the mix.  You know what you think Slovenian contact might throw into the mix.  You know what you think education, urban versus rural, and age might throw into the mix.  You didn’t just make those predictions up–they’re based on what we know about how language contact usually works, as well as what education does, urbanity does, age does.  Now: is what you would expect to be true based on all of that, true?  Do the predictions all hold?  If so: that’s nice.  Not very interesting, but nice–validation that you got the facts right.  Do the predictions not all hold?  That’s GREAT.  In science, we LOVE data that doesn’t fit theories!  That’s, like, the best thing in the world.  The key: don’t have an investment in the outcome.  Think about all of this such that whatever the answer is, you know what its significance is.  You know that it fits the theories?  Great–you know the theories, and you have a bunch of new data, and that would fit the minimally accepted requirements for giving Robin a PhD, so let’s do so.  You know that it doesn’t fit the theories?  GREAT–Robin realizes that she’s found a problem with the theory, and that’s how science advances, so let’s give her a PhD.

If you buy all of this: you need to do some rewriting.  You don’t need to change the content, really–you just need to move some of it.  Start with facts–just facts. Description.  Tell your reader what the previous literature says, and what’s missing from it.  Tell the reader what your data says.  Point out where the previous literature was right, and where you’ve contributed the new knowledge that it was wrong, and where you’ve filled the empty holes. Then theory, later. Later.  Take all of your speculations, predictions, explanations, etc. about language contact, age, education, etc., out of those earlier chapters.  Put them in a later chapter or chapters.  That’s where you’re going to test the theories–theories about language contact, about the effects of education, age, etc.  When you get there, you’re already going to have half of your PhD in hand, because you’ll already have contributed 13 things (see above) to our incomplete descriptions of Dalmatian.  Now you’re going to be a good linguist and talk about theory, and then you’re going to tell your committee that it’s time to award you your PhD, and you’re going to go on with your life.

On my paper copy, I wrote tons of comments about specific points in the dissertation–things that I don’t agree with, areas where I don’t think that you present the data convincingly, but mostly, place after place after place where you get caught up in the specifics of your predictions about the effects of language contact, etc. and should pull that material out and move it later.  I’m not talking about rewriting five chapters–I’m talking about moving material out of four chapters and putting it later.  You can absolutely do this between now and when your dissertation is due.  If you want the paper copy, come by my office.

Let me know if you have questions about this.  If you agree with the approach that I’m suggesting, I suggest that you forward this email, or some condensed version of it, to your advisor and to the rest of the committee.  If they agree with the approach, too, then you’re on very safe ground.

Postscript: Robin finished their dissertation, and defended it to wide acclaim.  English notes follow–back to French next time.

  •  bruiser: Definition from Merriam-Webster: a large, strong man. (If you’ve been reading this blog for a while, you know that the combination of large plus a physical object is not typical–see this post on large and big as things that seem like synonyms, but aren’t, exactly.)  How it appeared in the post: The goal there is to move a ball in a particular direction, while a bunch of big bruisers try to stop you from doing so.
  •  tons of: a large quantity of, a lot of. How it appeared in the post: On my paper copy, I wrote tons of comments about specific points in the dissertation–things that I don’t agree with, areas where I don’t think that you present the data convincingly, but mostly, place after place after place where you get caught up in the specifics of your predictions about the effects of language contact, etc. and should pull that material out and move it later. 
  •  judgement: the British spelling of judgment. The American spelling always rubbed me wrong, somehow—sue me.
  •  investment: this has many meanings.  I meant it in this sense, from The Free Dictionary: “a devoting, using, or giving of time, talent, emotional energy, etc., as to achieve something.”  How it appeared in the post: The key: don’t have an investment in the outcome.  Think about all of this such that whatever the answer is, you know what its significance is.  That is: the advice is to be able to not care (to not devote emotional energy) to what the specific answer happens to be.
  • allophone: a technical term in linguistics. We’ll save it for another time.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s