6

In general, it seems that formal language theory questions (uncommon as they are on the site) tend not to be closed down, although I have seen users vote to close such questions (unfortunately I do not recall which questions it was).

Personally I think that in general, formal language theory questions (say, questions about reducing grammar to Chomsky normal form, or the pumping thingy) do not belong on this SE unless they are applied to natural languages in some way.

Firstly, theoretical computer science folks should be in a better position to answer such questions than linguists anyway. Since cross-posting is discouraged, it would benefit the asker to ask at CS theory, rather than here.

Secondly, linguistics is the empirical study of natural language. While formal languages are sometimes used to model natural languages, they are not themselves part of linguistics any more than Markov chains or information theory, even though the former has wide applications in computational linguistics and the latter in corpus and functional linguistics.

Do you agree that formal language theory should be off-topic on this site?

3 Answers 3

9

First, Linguistics SE also encompasses non-natural "languages" i.e. constructed languages. Second, while I agree with your characterization of linguistics as including the empirical study of natural language, it also includes the theoretical study of the empirically discerned properties of NL (which is how formal language theory got invented in the first place). The epitome of high-level theory is formalization. I do agree that many such questions are better suited elsewhere, but that is because this place is not generally populated by specialists who know the linguistic and mathematical details. Finally, as a narrow theoretician I would still maintain that there is a place for applied and methodological questions, such as how Markov models might be used in speech recognition or whether information theory has anything to say relevant to the study of meaning. So likewise, formal language questions are particularly germane to a grammarian, since theoretical linguistics has been plagued by decades of sloppy, informal thinking about what notations and conventions mean (IMO). E.g., the difference between rule and constraint, the difference between derivation and representation.

3
  • Thanks for the repsnose! I agree that the theoretical study of empirically discerned properties of NL would be on-topic - I think questions about the proof that Swiss German isn't context-free, for example, would be completely fine. The same goes for Markov models of speech recognition and information theory in semantics. Jun 28, 2017 at 17:08
  • 2
    However, what I had in mind when writing the OP was questions that were purely about formal language theory, which I feel is more akin to asking about, say, the theoretical derivation of the theorem about the PGFs of the number of states before the absorbing states of a Markov process is reached. That would belong on Mathematics rather than Linguistics. As an analogy, calculus was initially invented to solve physical problems, but I don't think a question that is purely about (say) the Stieltjes integral, without reference to physical problems, would be on topic at Physics. Jun 28, 2017 at 17:10
  • 1
    Note that once the constructed languages site proposal is started, I will be proposing that questions about conlangs without established speech communities be considered off-topic here. (Once a language has an established speech community it's on its way to becoming a normal natural language and therefore belongs here.)
    – curiousdannii Mod
    Jun 29, 2017 at 6:40
7

A subject being on-topic on one site should not be a sole reason to declare it off-topic on another site.

This is not a new phenomenon; for instance, there are quite a few overlapping topics between Politics.SE, History.SE, Law.SE, Expats.SE, and Travel.SE. Some users even end up with cross-duplicating their question to several sites.

We also have some overlap with Theoretical Computer Science.SE, Language Learning.SE (some questions about the language acquisition that were asked here even before the LL.SE was created), and even individual language sites.

I think that Linguistics.SE should be no different to the SE mainstream: automate something only after you are tired/bored doing it manually. In other words,

  • First, judge on a question-by-question basis;
  • Gather the evidence;
  • Declare something off-topic based on a solid set of questions that didn't receive a warm feedback and good answers.
1
  • If I understand you correctly, you mean that we shouldn't impose a new rules to ban formal language theory questions, right? I agree that doing so would be a bit of an overkill, but I think it's still useful to try to reach community consensus on the acceptability of such questions, since right now it seems like a grey area that remain to be clarified. Jun 28, 2017 at 17:05
-1

I personally think a good thought experiment or way to come to a decision is imagining an unbounded stack Exchange site in which everything is allowed. What would be the reasons for beginning to make certain topics prohibited?

I think in general there is simply a desire for focus on any given SE site. But I believe all of them experience these niche topics that get asked there because there isn’t a dedicated site for them. “Sociology” doesn’t have an SE site, so we have to turn to SE Philosophy and use the sociology tag.

Instead of an intrinsically normative assertion about what “linguistics” is (my vote: firstly, a word, secondly, anything to do with “language”, at all, thirdly, what is language, anyway?), it might make more sense to take a pragmatic approach. If moderators just take subjective issues with some questions, sure, ban them - but not because it’s “not linguistics”, which requires someone to have authoritative capability to define “linguistics”, and I do not think that conversation will be conclusive; but just because a majority doesn’t want those questions on the site.

There may be a slight separation between “analytical” linguistics and “humanistic”, on the site right now. I would not advocate separating them since they truly are one and the same thing (to me). But if people wanted to exclude increasingly mathematical, information-theoretic, and computational type questions from the site out of a subjective feeling that that’s just not their topic, that would be completely fine. It could lead to the creation of an SE site more strictly formal-languages approach focused. I say this only due to feeling that natural language processing is so much more a subject to be wielded by linguists than programmers who don’t know linguistic theory. So I think all of those topics should be in linguistics prior to them being on other sites, like computer science.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .