Do generative linguists use spoken-word corpora for data? Offhand, I don't see why at least some of them wouldn't.

I'm not suggesting that the use of spoken-word corpora vs. documented native speaker judgments of grammaticality would be mutually exclusive. Nor am I asking whether all generative grammarians would use spoken-word corpora.

But I am asking whether, for example, one or more generative grammarians or a subset of generative grammarians would make use of such things as the Santa Barbara Corpus of American English.

  • Not sure about now, but it was pretty common among Chomskyites to ignore corpora as irrelevant to their study of language. Introspection was the most common approach then or the "Hey, Sally" method.
    – Alex B.
    Mar 30, 2013 at 21:09
  • Just looked up the meaning of the "Hey Sally" method in this course description: ling.uni-potsdam.de/~vasishth/SU07/methods.pdf. Wa--y back in the late seventies, before I got my hopelessly out-of-date 1981 BA in in linguistics, we were all required to read a paper that sung the praises of the rationalist methodology in linguistics. I still don't understand why generative linguists wouldn't make use of corpora, although I obviously have to accept an expert's observation that they don't. Mar 31, 2013 at 0:34
  • Chomsky has always been interested in competence (or I-language or whatever he calls it now), not performance.
    – Alex B.
    Mar 31, 2013 at 1:05
  • True, but why couldn't a generative linguist make deductions about competence from the performance recorded in corpora? Even if examining corpora didn't replace documenting speaker judgments about whether given sentences in a language is grammatical or not, couldn't data from corpora supplement such documentation? Mar 31, 2013 at 1:07
  • Supplement - I'd say yes - but not more. Let's see what jlawler thinks about all this.
    – Alex B.
    Mar 31, 2013 at 3:15

1 Answer 1


A strict distinction between competence (the abstract grammar of a language that speakers have implicit knowledge of) and performance (the actual products of this grammar - utterance produced by speakers) prevented this. Generative grammarians argued that (1) corpora don't provide negative evidence - they don't tell us what kinds of sentences are not possible or ungrammatical, and that (2) performance is tainted by the limitations of the human mind (limited memory and attention span, tiredness etc.) (@PElliot points out in the comments that this argument is based "on misunderstanding of the difference between performance and competence)

For example, recursion (such as in The man the woman the child... saw called) is infinite in terms of competence, but in actual performance few iterations are possible - you end up losing track of how the different parts are supposed to go together.

Any attempts to filter out these confounding influences were considered futile to the extent that using corpora (collections of actual language use - performance) was considered unscientific in generative grammar. We are used to thinking about communities of scientists as the pinnacle of rational thinking but in reality, what is considered scientific or unscientific relies very much on tradition and what the consensus is at a given time. Once a certain consensus has formed, it takes enormous effort (by individual scientists and the whole community) to re-examine this consensus.

However, the last 20 years or so have seen some dialogue between generative grammarians and corpus linguists, leading to a conference on dialogue between the two fields and other publications.

EDIT: Here's an example from Carnie, Andrew. 2002. Syntax. A Generative Introduction (p. 10-11):

One obvious source [of data] is in collections of either spoken or written texts [...] called corpora [...]. While corpora are unquestionably invaluable sources of data, they can only be a partial representation of what goes on in the mind. More particularly, corpora will only contain instances of grammatical [...] sentences [...]. To really get at what we know about our languages [...], we have to know what sentences are not well-formed. [...] This kind of negative information is not available in corpora.

Carnie makes the "no negative evidence" argument here. Note that he first calls corpora "invaluable" but then proceeds to say that to "really get at what we know about" language they are not useful. In other words, he doesn't seem to consider corpora all that useful and in fact never once mentions them again in the whole textbook.

  • I'm not sure i understand how the competence/performance distinction can be used as an argument against using corpora as data. It seems to me that everything we know about competence must be inferred from performance. Grammaticality judgements are an instance of performance - you illustrate this yourself by raising the example of centre embedding constructions, which are often judged as being ungrammatical - judgements are affected by performance considerations, so if this argument is taken to its logical conclusion it undermines pretty much all of a syntactician's data.
    – P Elliott
    Aug 25, 2013 at 23:36
  • I'd say that the crucial difference between corpora and speaker judgements is that only the latter can provide us with negative evidence - i.e. evidence of what is ungrammatical. If he/she looks at corpora alone, the syntactician essentially faces the acquisition problem - an absence of negative evidence. I'm not sure the competence/performance distinction slices the pie in quite the same way.
    – P Elliott
    Aug 25, 2013 at 23:40
  • @PElliott, I totally agree with you. I was just trying to explain what I understand the common argument by generative linguists against the use of corpora used to be and still often is - I've read it in a number of text books and encountered it in conversations. I don't agree with it, though. In fact, I'm a corpus linguist, I sometimes use grammaticality judgements, and I don't work in any tradition that could be called formal or generative. (Nevertheless, I see a number of useful and valid points or concepts that are based in generative linguistics.)
    – robert
    Aug 25, 2013 at 23:46
  • The argument does seem to be based on a misunderstanding of the comp/performance distinction. As someone trained as a generative linguist, i find it disappointing that it's so prevalent! Do you know of anywhere specifically the argument is made, off the top of your head?
    – P Elliott
    Aug 25, 2013 at 23:55
  • Right, I added a quote, but it makes the 'no negative evidence' argument (which I also added), and I can't find one about the other argument without doing more research. So, would you say you have never encountered the argument that infinite recursiveness is a feature of many languages but due to the limitations of the human mind obviously does not occur in language use? Also, I'm curious, how many generative linguists do you think actually use corpora in their research at all?
    – robert
    Aug 26, 2013 at 0:26

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.