Many thanks for this post. I missed a lot of the Twitter discussion when it was happening. I tried to read back through it, but Twitter really doesn’t make this easy! In any case, I wanted to raise (or maybe re-raise) the question of what we should expect by way of influences between these fields, especially as it relates to time and citation practices.

Consider the field of semantic parsing (NLP). First, all the successful papers in this area have at least one author who would pass an intro to semantics exam! Maybe they didn’t literally take the classes, but the ideas came from somewhere!

More importantly, the formalisms are mostly CCG-based, and these saw very active development in the late 1980s through to about 2000. This might seem old, but the alternative is that they would use work like that of Chris Barker, Chung-shieh Shan (@ccshan), and their colleagues on continuation-based composition and scope-taking. I believe that this would be an outstanding formalism for semantic parsing, but it’s a tall order to ask an NLPer to already have absorbed all of this. Even linguists are still digesting it!

A related point: lots of classic papers in compositionality are now getting cited by NLPers. Most are way older than 20 years. But it’s only recently that statistical NLP could approximate compositionality in ways that it finds appealing and useful.

This relates to my point about citation practices. By the time these ideas do get distilled into something that another field can consume, they often don’t get cited in the expected way. They seem like they are just part of the intellectual environment.

I’m sensitive to the arguments in your post about cherry-picking, but I still feel compelled to say that I could repeat the above with Gricean pragmatics (the core implicature pattern is now getting baked into people’s objective functions) and coreference (eventually we will all see that this isn’t reducable to text spans as Karttunen saw in 1972; see Marta Recasens’ work), and your examples from psycholinguistics and morphology/phonology are great. (In the reverse direction, at least for me personally, that Wilson and Hayes paper was a huge deal.)

In any case, I’m partly sympthetic to @yoavgo’s criticism of my field, but I feel he might have to wait a while to be sure of his assertion.

Professor of Linguistics and Director of CSLI, Stanford. Computational semantics and pragmatics, running, hacking.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store