3 ms·
I fully agree with you. I started studying linguistics since I come from a math background and Chomskian promise of mathematical linguistics was something I fou
by ribit 6y ago
I fully agree with you. I started studying linguistics since I come from a math background and Chomskian promise of mathematical linguistics was something I found very enticing. I quickly became disillusioned. Most of the Chomsky-inspired research boils down to fitting carefully selected data into an artificial formal machine and manipulating the said machine into accepting more and more data, until the machinery ends up being turing complete therefore rendering any result trivial. Don't get me wrong, there are a lot of very smart people working in that particular field and they have produced some very interesting results and observations, but IMO the real academic value of these models if very limited. It's all about solving intellectual puzzles in a theoretical void. But does it advance our understanding of human brain, learnability or language change? After being a linguist for over 15 years, I'd say "not really".
And yes, as you say, Chomsian followers are a highly guarded, closed camp. There is a strong tradition of giving maximal support to acolyte who wants to work in this area while casting out the "heretics".
- foldr 6y agoAs another (ex but not disillusioned) linguist, can I ask what’s the alternative? Say I’m puzzled by the acquisition of Antecedent Contained Deletion structures. I know how to tell a broadly Chomskian story about it. What other options are on the table? I hear lots of broad brush criticism of generative linguistics, but in almost all cases it turns out that the proposed alternative approaches are just working on different problems, and aren’t really alternatives at all.
- ribit 6y agoI don't think that you will find my answer satisfactory, but does there have to be an alternative? I think that many syntactic mysteries are mysterious only because syntacticians assume that they work in particular ways. Once you start looking at these problems from the perspective of pragmatics, competing interpretations and conventionalisation, many of these mysteries disappear. The point to make is that interpretation makes use of multiple sources of information, of which syntax is just one. You don't need an elegant, tight fitting, constrained syntactic structure to use language. Which incidentally is why we can understand things even they are not grammatical. I firmly believe that the issue of "grammaticality" simply boils down to convention. Not to mention that many constructed examples used to motivate formal grammar research (be it syntax or semantics) are borderline acceptable themselves to many native speakers.
- foldr 6y agoOk. How would you apply this approach to ACD? It is not clear to me how to make the mystery “disappear” in this case.
- shp0ngle 6y agoI studied NLP in Prague, and we tend to use "dependency grammars" more than "phrase structure grammars" of Chomsky https://en.wikipedia.org/wiki/Dependency_grammar https://en.wikipedia.org/wiki/Dependency_grammar I didn't work in the field at all after finishing the studies, so I cannot really tell you any actual implications of the different grammar approach anymore.
- foldr 6y agoDependency grammars are just another grammar formalism. They’re an alternative to Chomsky’s specific technical proposals but not to his overall approach.
- tralarpa 6y ago> Most of the Chomsky-inspired research boils down to fitting carefully selected data into an artificial formal machine and manipulating the said machine into accepting more and more data, until the machinery ends up being turing complete therefore rendering any result trivial. Ah, that's interesting. Because that describes very well what I felt when I read the early Chomskian works. Very cool theories and models with very little data to support them. At some point, I was wonderering whether the authors of those theories had any practical experience with languages apart from their own (English) mother tongue. And then the models became more and more complex, seemingly in an attempt to make them somehow work with real data, and at the end they became so complex that any data could be matched and no scientifically verifiable predictions could be made.