Comments

Sunday, February 7, 2016

A reaction to my Nijmegen provocation

Bill Idsardi sent me a link to some useful comments by Sean Roberts on David's three lectures (see here). As Bill noted, we seem to agree that the barn owl is a great paradigm of what we should all be aiming for. However, I think we agree on little else. However, this is not what I wanted to comment on.  What I found interesting is that Sean got exactly the wrong message from my comments (and that is almost certainly my fault). Let me elaborate a little bit.

First, I did not mean to imply (nor did I say) that the only questions of linguistic interest were syntactic. So, I agree with Sean that "there is more to language than syntax." But, and this is what I wanted to get across, there is at least syntax, and this is not something that neuroscientists seem happy to concede. In fact, I has been my experience that brain types have no idea what modern syntax has discovered and when they do know something they seem worried that there are many apparently different theories out there that don't agree (some also worry about the empirical basis of the discoveries). One aim I had was to assure the brain types in the audience that the standard theories make most of the same distinctions and identify effectively the same kinds of dependencies as syntactically relevant. So, once one gets by the surface orthography, many of the theories are basically saying the same thing. Thus, there is, contrary to appearances, a general consensus about some of the key syntactic features of Gs. See slides 9 and 10 (here).

Second, I argued that brain types should hope that something like Minimalism is on the right track as it will make it easier to make contact between linguistic ideas and neural ones. I mentioned bird song and phonology as an example. If they are the same kind of system (as some have argued) that we might learn something about the neural basis of phonology by studying (ahem, torturing) birds. Similarly, if large parts of FL are not linguistically specific, as Minimalism hopes, then we can study these in non-linguistic systems. So, for example, we could learn a lot about FL by studying other cognitive systems in humans and animals. What we will not be able to do is learn anything about those features of FL that are linguistically proprietary precisely because they are specific to linguistic capacity. But, if Minimalism is right, the set of things that are properly only linguistic may be small.

Third, I did not provide a parts list. I suggested that giving one will be easier if Minimalism is right for it reduces the number of moving parts to a minimum and identifies those that are purely linguistic. That should be useful for in place of looking for very complex "circuits" (e.g. passive, raising, question formation) we would be hunting for more basic ones like ones for agree or merge/combine. This, I suggested, look like the right kind of grain, unlike the GB notions that came before. Of course, I could be wrong, but…

Fourth, I did say that Chomsky is always right. And he is, ABOUT THE BIG ISSUES. I personally also think that he is often right about the details, but many have disagreed with him on these, including moi. But on the big issues he is completely correct. There is no doubt that humans have Gs that they use to produce and understand language. There is no doubt that there is something special about humans that make them linguistically capable and that involves being able to acquire Gs. These strike me as truisms and it is really a waste of time pretending that these views are controversial. They may generate controversy (I have no idea if Sean thinks them suspect) but they really aren't. They are obviously true. Of course, how they are and what this means for cognition and brains is NOT trivial or obvious. However, we have learned some things over the last 60 years about the mental mechanisms implicated. If we have learned less about how the brain does these things, then the problem lies not with what we have learned about FL from a cognitive perspective.

Fifth, where does the problem lie? Here I channel Randy Gallistel. I think that brain people will not make contact with most of cognition until they give up their neural net ways. Randy makes this point concerning dead reckoning in ants and cash behavior in birds. It applies parri passu for language and many other domains of cognition. Brain people need to start wondering about how brains represent and how they use these representations in information processing. They need to discover the analogues of standard CS notions like registers, indices, read from, writing to, stacks and all the stuff that CS types brought to cognition (and linguistics too) in the 70s and 80s. Read Mitch Marcus, Bob Berwick etc. When brain types find these mechanisms it will be much easier to relate findings in cognition (and linguistics) and brains. Why? Because neuro investigations are about how language is being used. We have no idea how brains store information (as David P noted). We can look to see how the brain does language processing given a linguistic input. But this means that we are looking at how Gs are being deployed and for this we need the technology that CS brings to the table. We have plenty of models of parsing. However, none of these make neural sense as they are all stated in classical CS idiom.

Last point: Sean took me to be dogmatic. I think I am. There are some issues that I strongly believe are simply not up for grabs and keeping an open mind concerning these is bad for your intellectual health (think flat earth or global warming denialism) At Nijmegen, I pointed out what I though these were. I outlined them, and defended them (up to a point) and, given that I had 20 minutes, I exercised my judgement and told the assembled brain worthies what I thought the uncontroversial parts were and tried to clear up some misconceptions. That's about all you can do in 20 minutes. However, what I think really irks Sean is not that I gave no arguments, but that I made a lot of claims that he didn't like (such as that GG has stuff to tell us and that these results are not controversial or really debatable). And for that I am happy. As Keynes once said: "I personally despair of results from anything except violent and ruthless truth telling--that will work in the end even if slowly." I sure hope Keynes was right. And as I know some small parts of the truth as regards language, I intend to make these known in venues like Nijmegen whenever I can as forcefully as I can. And you should too, for believe me, there is a lot of ignorance out there.

6 comments:

  1. I am sympathetic to the job description of minimalism expressed here, which is to try and isolate the smallest possible set of linguistically proprietary things that can help explain syntactic structures. It's a lofty goal, and one that is in line with Ockham's razor.

    An obvious corollary is that for those interested in the bigger picture of human linguistic prowess, the minimalist line of work can only contribute a tiny piece in the puzzle (essential though it may be). I believe Norbert sees this; this is why, for instance, animal models are mentioned as a useful approach for investigating the faculty of language in the broader sense.

    It's not just animal models of course; clearly, if minimalism sticks to its minimalist job description, language scientists would do well to embrace methods and insights from a wide range of linguistic subfields (and cognitive science neighbours) to complete the puzzle. So ultimately, my interpretation of Norberts' message is quite optimistic. The future of the language sciences is interdisciplinary.

    ReplyDelete
  2. Thanks for your excellent post and this can help us very much.Other site language is very important for us and we are learning from blogs and for word meaning english to bengali dictionary.I am eagerly waiting for your next post.

    ReplyDelete
  3. This comment has been removed by the author.

    ReplyDelete
  4. This comment has been removed by the author.

    ReplyDelete