Sometimes when I see an example sentence in a linguistics textbook which is supposed to be incorrect or unparseable, I get annoyed because the sentence seems just fine to me. Garden paths aren't the same to me anymore, either, because after studying them I can parse new ones without much problem. For example, most linguists know the famous 'the horse ran past the barn fell'; because of that, 'the boat floated sank' and 'the car turned went straight' are easier to parse. How do linguists maintain a sense of what non-linguists can and can't do? There are always surveys, of course, but a linguist can't just write every sentence on a survey and ask for opinions on every one.

2 more examples. It has been claimed that starting a sentence with multiple 'that' is not possible; but I am just fine with saying

That that she did that was a bad idea was not a good thing to hear.

Perhaps not the best wording, but I would give it a '?', not a '*'. Triple embeddings are fine sometimes, too, if there's an adverb. For example,

The computer the guy your mom saw yesterday bought broke

is fine with me. Am I abnormal, or am I just a linguist?:)

  • 6
    Anyone can learn, given enough experience. One way to keep in touch with intuitions is to teach Intro Ling and watch what surprises people.
    – jlawler
    Commented Feb 16, 2012 at 20:56
  • 3
    Two classics to add to those mentioned by @Alex B.: Bar-Hillel (1971) "Out of the Pragmatic Wastebasket"; Botha (1973) "The justification of linguistic hypotheses". Three ways to sidestep the issue: (i) do not let important theoretical claims rest on "delicate" judgments; (ii) use a corpus for your examples; (iii) don't study your native language.
    – user483
    Commented Feb 16, 2012 at 22:10
  • 3
    You're presuming that linguists tend to work on their own language - there are many of us who work with languages other than our own, and in fact the chance to use our intuition in our work actually grows over time!
    – LaurenG
    Commented Feb 17, 2012 at 15:24
  • 1
    (Oops, hit 'enter' before I was done) This is only useful in that the linguist can say "Structure X doesn't appear in all this data so it's probably not used", but does makes it pretty hard to know if it could be accepted in certain circumstances. But, it does avoid the problem of the speakers you work with reluctantly saying "yes, maybe you could use that" when in reality they wouldn't. It's not only linguists that can begin to be convinced that something might work, when you think about it hard enough :) Commented Feb 18, 2012 at 9:38
  • 1
    Increasingly people are including detailed contexts for their examples, to show that there is a specific context where the sentence is grammatical and/or felicitous. This is especially true in semantics, where judgements can rest on very delicate issues in scope, binding, and quantification. Syntax benefits from providing detailed contexts too, even for English examples, because it helps others understand the discourse frame in which utterances should be analyzed and consequently helps tease out discourse-dependent phenomena from pure syntax.
    – James C.
    Commented Feb 18, 2012 at 23:26

3 Answers 3


This question suffers from a presupposition failure. In general, linguists don't lose their "normal intuitions", if you assume for example that only experienced linguists are publishing in Linguistic Inquiry — Jon Sprouse and Diogo Almeida have tested a bunch of examples from LI and from Adger's intro syntax book on naive informants using sophisticated methods and have found that the judgments confirm those reported in print. You can get into this literature at this page.

What your discussion points to is something that is often pointed out to students as a cautionary note is that even though you might be able to extract the intended meaning of a given ungrammatical string, it doesn't mean that the string HAS that meaning. We can often understand what a string with an island violation in it is supposed to mean, but that doesn't make it any less ungrammatical. It's facts like these, too, that are often taken to claim that natural language is not identical to the language of thought (whatever that might be).


Contrary to some of the other answerers, I do think "intuition blurring" is a real thing. It's a bit like what happens if you say the word "zebra" a few dozen times in a row and suddenly it stops sounding like an English word and starts sounding like a meaningless noise. If you read dozens of complicated, multiply embedded sentences to yourself, they'll all start sounding equally good or equally funny after a while.


  • This isn't a permanent loss of ability. It's more like temporary sensory overload. Going out for a walk or having a conversation that isn't about linguistics usually fixes it in my experience.
  • Like Knitter says, it's a bit different from what you're experiencing, because you're looking at processing effects. You can definitely teach yourself to process some "hard" sentences faster than a normal person. But researchers who do language processing tend to work with controlled experiments rather than intuitive data, so it's not a huge issue for them.
  • If you need grammaticality data and you really feel like your intuition is hopelessly fuzzy, you can ask your friends. Hang out in a linguistics department long enough and someone will come running out of their office and ask "Hey, you speak English, right? Can you say X?"

Your question is a bit confused because it conflates ungrammaticality with things like garden paths and center embeddings. Both of these (garden paths and center embeddings) are indeed grammatical but are often at first perceived to be ungrammatical because of processing difficulty. In this way, the linguist is in fact more equipped to determine the relevant sentences' grammaticality than your average Joe.

It's also the case that there is a normal amount of variation in intuition, sometimes because of dialect and sometimes just differing acquisition. In other words even two people who grow up in relatively similar circumstances are probably walking around with slightly different grammars of English. (Clearly, in yours, multiple that is fine: in mine, it's not.) Some grammaticality judgments are more rigid than others, of course; everybody would say that a determiner following its noun in English is ungrammatical, but certain ECP violations will sound better or worse depending on the exact sentence and the listener. (And in fact adding words like "yesterday" to a triple embedding does help its parse-ability.) It does not necessarily mean that you are out of touch if people differ in their grammaticality judgments, although you might be an outlier. That is why it's important to check with other real people, even as haphazard as that can be. (Lack of evidence for a construction, for example on the Internet or in a corpus, does not necessarily mean it is ungrammatical, as is the case for triple center embeddings.) Rarely will an argument ride entirely on one judgment, though.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.