Tuesday, 2 June 2015

Athens: Final Instalment

I would like to start off by saying, in case it wasn´t already perfectly obvious, that the posts I have been making are my own highly subjective highlights and interpretations from an extremely contentful and interesting event.  I am not even attempting to provide a proper transcript, and these are not even Minutes. They also have no official status, in the sense that the organizers have no idea what I am writing.  I thought I would start my last post in this fashion, because the final instalment will probably be even more subjective and interpretational than the previous ones.

I ended my last post with the assertion that it is hard to agree on the content and formulation of our field´s MLGs.  To illustrate this I take a toy example from  the realm of argument structure and think out loud for a bit.   Suppose I range my own commitments and things I consider consensual in a kind of hierarchical ranking going from most general to most specific. The most general level is shared I expect by all generative syntacticians, while the lowest reaches might start to get more contentious.

GG1:  The Language System discrete and symbolic, and makes crucial reference to hierarchy in its complex representations.  

GG2: A linguistic representation includes the formation of dependencies and relations.  These all seem to come with their own specific locality conditions.

MLGs (?) For Verbal Syn-Sem

1. There is a grammatically relevant notion of SUBJECT that cannot be defined purely by reference to thematic/semantic properties.

2. In  the linguistic expression of an event where both agentive and patientive participants are obligatorily represented,  the nominal constituent representing the Agent is always hierarchically superior to the nominal constituent representing the Patient in the syntactic representation (SYN-SEM generalization).

3. A monoclausal verbal structure cannot express more than one temporally non-overlapping dynamic portion (SYN-SEM generalization).

4. ARGUMENTS (thematic and notionally obligatory participants related to ta verbal expression) behave in a linguistically distinct way from ADJUNCTS.
(lots of sub-generalizations here related to the formation of dependencies into the two types).

5. Argument structure and aktionsart generalizations  are properties of  the verbal projection, not  properties of  verbal lexical items.
(Depending on who you talk to, there are different sorts of feeding relations between the lexical verb and the verbal structure it appears with).

6. In a phrase structure representing the verbal event,  argument structure projections such as CAUSE and PASSIVE appear inside (i.e. hierarchically closer to the root) than inflectional and ASPECT, and TENSE projections.  

Ok, that was just off the top of my head, and I was trying to state the MLG level in terms that would be acceptable to the maximum number of people who would consider themselves generative syntacticians.  Notice that I didn´t put in Burzio´s Generalization, or express (5) in terms of acategorial roots.  For the former, that´s because I couldn´t think of a way to express it in primes that I accept in a way that makes it both contentful and true; for the latter, I would not agree with (5) if I had to accept that extra analytic step. 

There are also a lot of other things I could write down there that I believe are correct (with a fair amount of good reason), but which I reckon that too many other people would take issue with, so they didn´t make it.  But where is the cut-off ? 

Another thing.  Groups of syntacticians that share more terms of art, will have more specific commitments in common. But are they MLGs really, or are they  just agreements about how to use the toolbox?  

Finally, some of the things one might want to write down as an MLG have been demonstrated and tested on only a small (and typologically narrow) set of the worlds languages. They are up there because they look good so far. There would be nothing on the list if  we had to confine ourselves to things that are true of every world language.   I think it is fair to concede that deep engagement with the facts and properties of currently less well understood languages can sometimes radically change the terms of the MLGs that ultimately turn out to be correct. (Dechaine was the leading voice of caution here).

It is worth emphasizing that the list above is both provisional and highly descriptive. Some of them may end up having a fan of sub-generalizations; some of them might end up being just be tendencies, or confined to certain language groups. 

In all of this, we must not lose sight of the fact that this list is not a list of Universals in the sense of Universal Grammar, since we all  think that whatever languages have in common, they have to be the abstract things that underwrite and give rise to these patterns and tendencies.  Once this is recognized, even tendencies and conditional generalizations are valuable, because they give insight into what those commonalities might be. It is an empirical issue what level of abstraction the common UG properties might exist. It might be just MERGE, plus a range of cognitive tendencies, learning biases, and 3rd factor design properties.

I found it a good exercise to try to write some of these things down. And I also found it an interesting exercise to see a whole room full of generative syntacticians trying to brainstorm a list together.   We can all agree broadly, but it is much harder to make the fine grained judgements that this kind of list requires in a consensual way.

But perhaps absolute consensus is neither possible nor desirable. The effort to transcend parochiality is good, but the list should have a more flexible and pluralistic status if it is going to have any good effects.

It certainly seems true to me that if we had such a list,  however imperfect, it could be immensely useful in guiding research questions and providing a platform for genuinely comulative advance, especially after we have made the effort to state our commitments in the maximally general way possible so as to communicate across frameworks, and ultimately across disciplines.   

What did We Accomplish?

I seriously hope that the subcommittee set up impromptu on the floor at Athens will manage to negotiate the minefield of the The List  and come up with something that at the very least can serve as a springboard for discussion and further hypothesis testing (replications and extensions). 

We also had a nice affirming experience in Athens in the sense that it was impossible to leave that event without thinking that syntacticians are serious,  smart and committed and doing a lot of good and responsible work.

Finally, we came up with a number of practical suggestions for how we can manage the outreach to schools, to the public, and to academics in other disciplines.  This was something we could all agree on.

So its all Good, Right?

The syntacticians at the Athens meeting are real live people, and so they straddle the whole spectrum of personality types  with respect to thoughts on the Road Ahead, and the reasons for the call----- Happy, Bashful, Sneezy, Sleepy, Grumpy, Dopey and Doc.  I want to concentrate for a moment on Happy and Grumpy.
Happy is the syntactician who was a little baffled by the terms of the call, and thinks that internal to syntax there is  no problem, no crisis, and no reason at all for this meeting.  Grumpy is the syntactician who sort of darkly suspects that the reason we have been so bad at communicating outside our own tribe is that we have some internal issues to resolve as well.   I speak as one who would classify herself as Grumpy in this regard.  I think, for example, my friend and colleague David Adger is Happy. (I hope David will not yell at me for this, but I think we have actually had this conversation).  This could just be a personality thing.  But if I can generalize, (and I know I am getting myself into trouble here)  I would say that Happy is a syntactician working in the US  or the UK who is comfortable using the canonical minimalist toolbox, terms and framework language.  Grumpy was usually living in non English speaking Europe, and often had fewer mainstream commitments at the implementational level.   I think Grumpy would be much happier if syntactic theorizing used a less parochial toolbox, emphasized generalizations at the MLG level more, and if it was a little bit more multilingual in its engagement with other implementational languages and of the bridging discourses to other disciplines.

I´m sure I´ve forgotten something, but it´s gone on too long already.


  1. If you have gone to the trouble of potentially getting yourself in trouble, let me do the same [sloppy reading] – on two fronts.

    Let's take (1) as a starting point (There is a grammatically relevant notion of SUBJECT that cannot be defined purely by reference to thematic/semantic properties). Let me distinguish (1) from (1'):

    (1') If there is a grammatically relevant notion of SUBJECT, it cannot be defined purely by reference to thematic/semantic properties.

    I think (1') might be a consensus among generativists; but I'm fairly sure (1) is not.

    I bring this up not to pick nits, but because I think there is another kind of discovery that generative grammar can lay claim to, which maybe isn't talked about enough: the discovery that some commonsensical principle or element of grammar doesn't actually hold or exist. For some people, "subject" falls in this group (see Ethan Poole's recent work on "subjecthood properties" for an interesting recent example). For others, "word" (in any sense other than "phonological word") falls in this group. [I was a bit surprised to read, in a previous post, that there was consensus in the group that there was definitely a syntactically-relevant unit corresponding to "word." I wonder what the arguments for this were.] Or to cite another example, that I think hasn't come up here yet, the discovery that the passive is not, contrary to the commonsensical understanding of it, an argument-structural alternation (cf. John was believed to be a liar).

    Now, none of the three items I've enumerated here are anywhere near consensus (notably, constraint-based lexicalist frameworks seem, from my outsider's view, to endorse SUBJECT and WORD as primitives, and attempt to still treat passive argument-structurally; at least HPSG does). So maybe these are not the right examples to bring up in the context of The List; but I think the general shape of this kind of discovery (e.g. "You might have thought that where we put the spaces when we write things is meaningful; turns out, it's not!") is interesting and deserves its own place in the discussion.

    The second thing I wanted to bring up has to do with the implementational language issue that you mention. I tend to agree with you, in general, that people should avoid making commitments – both technical/implementational and theoretical – that are unnecessary in making their points. [I recall one particular review process I went through that forced me to take what was a fairly framework-independent paper and add a bunch of trees to it, making it both longer and, in terms of appearances, less framework-independent.]

    But it seems that your unease goes beyond this; i.e., you have some qualms with the minimalist toolbox in particular. (Do I understand you correctly?) I ask because I have felt, and still feel, the same way towards some of the things in that toolbox, but my reaction is in some sense the opposite of striving for framework-independence: it makes me want to go looking for an argument against the specific tool I dislike. So I guess I'm curious: what is it in the toolbox that you dislike, in particular?

    1. Hi Omer,
      I take your point about the negative results. I sort of think I was trying to do something of the kind with my Subject MLG, but didn´t express it very clearly. I wanted to say that GG generally believes that there is a formal grammar internal notion of subject (regardless of whether you choose to make that a primitive as in LFG, define it by means of a particular structural position as in GB), which cannot be reduced to semantical or functional considerations. I sort of thought that the commonsensical notion was that Subject corresponded to something semantic, whereas all GGers think it is an outcome of a formal system that is related less directly to semantics. But I probably phrased it clumsily.
      The general point is well taken, I think it is true and could probably help to contribute to a number of other items of general GG consensus on that list.
      Specifically, my discontents with the minimalist toolbox would be less urgent if it weren´t so dominant. I find that the phenomena I know most about, and care most about, simply do not sit well with very basic assumptions of the standard toolbox. I do not like the way lexical items are integrated into the computation, for example. I don´t want to use acategorial roots, flavours of little v, or selectional frames for functional structure at the point of late insertion. Equally, when the grammar is described derivationally, my mind boggles at the notion of the numeration as the start point and I can get no further. I don´t care what the algorithm for labelling is---- I still think we don´t know what the number and fine grainedness of syntactic categories is and I´d rather get on with that job. I can´t assess how to implement phase and edge, and how local certain phenomena are without knowing what the actual cartography of the little v area is. The list could go on. None of these terms and architectural choices seem to have any deep justification, and I do not see a principled way of finding one. My argument is always that looking at things a different way makes the pattern come out more clearly. One gets a little bored and fed up with reading this stuff. At the same time, if one wants to do things differently, one has a lot more work to do to justify and support it. Still, I probably would have things to pick on in any framework that was dominant and provoked a disproportionate amount of interest in its own tools for their own sake. And in the end, I am of course a minimalist more than any of the other tribes.

  2. Happy here!

    Well I think you're right that I'm Happy, but wrong about why. I'm Happy because I take a very pluralistic viewpoint in general and I think generative linguistics, writ large, has a huge amount to offer the study of language and the study of cognition more generally, has made big strides intellectually over the years, and I'm confident that it has a lot more to offer, irrespective of current passing fashion for (big) data-crunching. Weirdly, what also contributes to my Happiness is that I think that whatever we have in our toolbox is lightyears away from a real theory, so I actually want there to be lots of different approaches to tackling the issues of syntax and its interfaces. And here I include non generative frameworks, sociolinguistics, pragmatics, typology and a lot more. So there are lots of opportunities for us to work with non-generativists, enriching our theories and theirs. And to do that, one has to look beyond the toolbox, and get to the core ideas to see what we can offer to those other approaches, and what we can learn from them. And I've found, doing that with, e.g., sociolinguists over the past decade, it's made a real difference. I've learned a lot, they seem to think we minimalists (or those of us who are talking) have something interesting to say. I'm even giving a plenary at NWAV this year! So another thing that makes me Happy, and makes me disagree that we have an in principle communication problem, is just, well, communicating.

    Perhaps the reason I don't mind using the canonical terms (which I don't) is that I see much of that as ephemeral, (though I can find myself completely fascinated by the intricacies of particular implementations in a geeky kind of way) - we can state the core ideas in different ways (though particular frameworks act as good methods for getting to those ideas). But I'm not actually sure Gillian's diagnosis of my Happiness as arising from a hobbit-like satisfaction with current minimalist ideas is correct. The influences that most strongly affect my latest book are Brody and Williams (hardly standard minimalists), and I end up with a system that gets rid of functional heads, makes head movement impossible, allows unary branching, makes roll-up movement structures ungenerable, ditches head-complement structures outside of the functional domain, and much else that doesn't seem to fit the characterisation of mainstream minimalist commitments at the implementational level.

    So, yes, I was a bit skeptical about the call, though having a meeting like that is going to be fun, and who can object to fun? I don't think there's a real intellectual crisis, (though there are, without doubt, more general issues in our field about research funding, early careers support, equality, etc). Internal to syntax, I think that interesting work in our field gets noticed and is influential, irrespective of how it's implemented (I count a lot of citations to First Phase Syntax on Scholar!). External to syntax, I think that, if we get off our arses and do it, there's a mass of interesting interaction to be had with those outside our field, and they are willing to listen if we talk to them.

    The final issue distinguishing Happiness from Grumpiness is, I think, more psychological. We can't control how other people do things. Some people will do techy, hacky type work, because they like it. Some people will never talk to a sociolinguist or a phonetician (!) because they think they have nothing intellectually in common with them. Some people will refuse to learn a bit of HPSG, or LFG or Construction Grammar because they think it's of no interest to them. But thinking we can control what other people do in our field is a way to eternal Grumpiness. We should just do such good work as we can ourselves, and encourage our graduate students to think about important ideas, while training them to implement them theoretically, and to think interdisciplinarily.

    1. I don´t think it´s about feeling like an outsider, or being more pluralistic. Everyone at the Athens meeting was part of a well established group, at a middle management or senior statesman level of their careers. None of the invitees could claim to be marginal or ignored---- we were there after all --- and yet some of us were Grumpy and some of us were Happy. It was not the case that the Grumpys had fewer citations than the Happys.

      I also stated a correlation, not a reason for the correlation. And I still think the correlation is related to a comfort with a certain shared language. I do think that the difference you cite, namely that you can find yourself `completely fascinated by the intricacies of different implementations´ and that you are also content with using the particular language, is a contributing factor,. You are happy with using and playing with (and of course changing and adapting) the canonical minimalist tool box. Part of it is a personality thing; part of it is a network density thing. I find that more difficult, by and large. Maybe partly because I am more singleminded than you in the particular subject matter/kind of data I have been pursuing, and the results I think I have go against the grain of some standard assumptions.

      Personality again: I am extremely bored by the techy in-talk that overreaches in the pure theory direction, and is only tangentially related to any data or MLGs. I think there´s too much of it. Most would agree there is good and bad work in every field, and GG is no different. We should apply the same standards inside and out. I also dislike tribalism, and I worry that we get pushed into an implicit tribalism by not paying attention to other views.

      I don´t think I am Grumpy because I am out of the loop, because I think I sit in a very privileged position in the field in fact. I have a job where I am allowed to do my own quirky thing, where people who care about argument structure do engage in what I have to say, but where I do not need to convince anyone to give me a job or a postdoc. When I complain about the field, its because even from my own privileged angle I can see that we are much more insular than we need to be, and that we would benefit in general as field from being more multilingual and pluralistic (I am not talking about you, I´m talking about the field as a whole). This will help my supervisees and the young people in the field who are struggling to find their niche in a rapidly changing intellectual and economic context.

      Also, as a personality thing, I do not think it is enough just to keep ones own doorstep clean. As a mid career linguist with a permanent job and lots of experience in our field, I think I should be allowed to complain, and make a pitch for higher standards whenever the situation warrants it. And I am willing to have civilized arguments with my colleagues and mentors about individual cases. Of course I believe there is a lot of great work out there, and being Happy or Grumpy about `the state of the field´ is orthogonal to whether one does good work or not. I just think the Grumpys have a little bit of a point.

      I don´t expect everyone to agree with me. I just felt I couldn´t blog this whole event without mentioning the Elephant in the Room.

  3. Well, I guess we can thrash this out over cocktails in New York ... :-). I'm puzzled about the insularity thing though. In the last year or two, I've seen a huge amount of outreach in various ways. Even in my own department, Daniel's done important stuff on Kiowa revitalisation, reported in the news, I've published a paper in PNAS that was reported in New Scientist, as well as pieces aimed at psychologists in WIRES and the piece I just wrote for Frontiers. But there's also a ton of cool stuff that Gretchen McCulloch has been involved in, George Walkden's Books in Linguistics podcast , the excellent YouTube LingVids, as well as the work we all do day to day in schools etc etc. I really think there's a lot going on. I'm not saying that we shouldn't do more, but to be honest I do think that the solution is to do it, rather than talk about doing it.

    On the correlation, is it really true? Elena's piece (I didn't see the talk, but just from the position statement) seemed on the grumpy side, but she's very comfortable with the canonical minimalist vocabulary, at least from what she publishes. Joan's work is pretty unminimalisty, but her position statement read as more happy than grumpy. So, aside from not thinking there's an explanatory mechanism to support the correlation, I'm not sure it's really a true one. But maybe I have those examples wrong. You can convince me soon!

  4. Cocktails in New York at Roots. :)