The Poets’ Corner Ps. Q. Vol XVI, No 4 Contents ██████████ █████ ███████—J. Fraser Bennett

The Native Speaker Fallacy

Linguists are people, and must thus be expected to believe some foolish thingsif perhaps fewer than the general populace. However, some commonly held opinions are simply too ludicrous to be allowable; it is one such absurdity that I will here disprovethe Native Speaker Fallacy.

For years, linguists have appealed to the judgements of native speakers of various languages as to the grammaticality, semantic import, and/or phonological contrastiveness of various natural or constructed utterances. While this method of supporting one’s arguments is in itself suspect, it is nonetheless only an unfortunate result of an unspoken but deeply held conviction which rests on two key principles:

  1. There are native speakers of any given language, and
  2. these native speakers can be identified with certainty.
Both of these principles have usually been considered self-evident; however, both can be shown to be essentially ad hoc, and therefore untenable, in their usual formulations.

Disproof of the Native Speaker Fallacy has eluded linguists in the past for two major reasons. Primarily, the requisite method of demonstrating its falsity has remained unimagined. Additionally, no one has bothered to attempt to imagine it. Until now.

The vehicle for disproving the Native Speaker Fallacy is a fairly simple experiment involving grammaticality judgements of five sentences taken from the text of an early campaign speech by Richard Nixon. The sentences, presented in Table I, include two which are clearly grammatical, two which are clearly ungrammatical, and one which is most definitely strange:


  1. The leaves that are on the tree are green.
  2. A giant squid is eating John’s sister in the backyard.
  3. Dog the in dragged big yellow stupid because in in at.
  4. When yesterday the yield sign, tomorrow that the stop light came again and went.
  5. Ikan akan sedang dimakan oleh lakilaki.

Subjects were simply asked to discriminate verbally between the choices “grammatical” and “ungrammatical.” Results of this discrimination are presented in Table II.


Sentence “Yes” “No” Unsure “Huh?”

1 56% 31% 10% 3%
2 50% 25% 22% 3%
3 28% 41% 28% 3%
4 20% 33% 44% 3%
5 11% 58% 28% 3%

Σ = 12       μ > .478333       ξ < 5       ε = 96.8%

The subjects for this investigation were 36 self-professed “native speakers” of English. None had ever left the United States; nor had any studied any foreign languages at all. Thirty-five were senior undergraduate students of Political Science at a small Eastern liberal arts college, and one was a tobacco farmer. All were between the ages of 10 and 87. Two students who had taken introductory linguistics classes were excluded from the sample.


The data in TABLE II show that, generally, “native speakers” are able to correctly identify grammatically correct and incorrect sentences. However, the percentages are not nearly so clear-cut as many linguists might have hoped. Fully 44% of the subjects failed, for example, to correctly identify sentence 1 as grammatical. Likewise, 67% were unable to recognize the fact that sentence 4 is clearly ungrammatical. Even more disturbing is the failure of a whopping 97% to realize that sentence 5 is not even English at all, but simply some Indonesian that the future Chief Executive had been practicing and inadvertently slipped into his speech.

The obvious conclusion from these facts is that not all of the subjects in this test can honestly be called “native speakers” of English. Performance errors could conceivably account for some of the mistakes, but certainly not so many as are clearly present here. These facts must force us to reconsider our conception of “native speaker” status for English, and by implication, for any language.

The simple fact that a person has never spoken any language except hypothetical language X does not, apparently, guarantee that he/she is a native speaker of language X. Conceivably, one could be trained (with years of practice) to imitate a native speaker of some language, without actually being one. There is another, equally plausible explanation: a person might simply not be a native speaker of any language at all. Together, these two possibilities provide ample explanatory justification for the results obtained in this investigation.

Keith Slater Michigan State University

The author gratefully acknowledges the support of the National Science Foundation, without which this research could have not been undertaken. In deference to the Foundation, the original title “Linguistics ruined by NSF” was discarded.

The Poets’ Corner
[redacted] [redacted] [redacted]—J. Fraser Bennett
Ps. Q. Vol XVI, No 4 Contents