The stigma of ungrammaticality

On the Stanford linguistics newsletter site (the Sesquipedalian) yesterday, Arto Anttila asked who was the first person to use the asterisk to mean ungrammaticality:

This question was brought up in the Foundations of Linguistic Theory class on Friday… I have been able to confirm that * was used to mean ungrammaticality as early as in 1963 by R. B. Lees and E. S. Klima in their article ‘Rules for English Pronominalization’, Language 39(1), 17-28. The relevant sentence is on p. 18:

(8) *I see himself.

The example is followed by a long footnote where Lees and Klima patiently explain what * means. They cite no precedents. These facts together strongly suggest that one of them is the originator of the notation. But we may never know which. Lees passed away in 1996 and Klima in 2008.

… Thanks to Martin Kay for asking the question and to Paul Postal for suggesting the answer.

Beth Levin checked Lees’s The Grammar of English Nominalizations (1960) and found that it has asterisks of ungrammaticality in it, starting on p. 7, where an asterisked example is given without comment.

Certainly, by the time I entered grad school at MIT (1962), asterisks were widely deployed as marks of ungrammaticality.

Given the central role that ungrammatical (as well as grammatical) examples played in Chomsky’s early writing on syntax, I expected to see asterisks in Syntactic Structures (1957) and in the giant work from which Syntactic Structures was extracted, The Logical Structure of Linguistic Theory (dittoed ms. of 1955, reprinted in 1975 with revisions and a new introduction). But no: examples were characterized in the text as grammatical or ungrammatical (beginning on p. 15 of Syntactic Structures, the third page of the actual text), but without any metalinguistic mark, or stigma, associated with them.

So it seems that the stigmata appeared in generative syntax late in the 1950s, possibly at the hand of Bob Lees.

The OED — at least OED2 (1989), which has the most recent available entries for asterisk, n. and v. — is not at all helpful (the entry for the noun star is even less helpful). The relevant subentry for the noun:

The figure of a star (*) used in writing and printing.

a. as a reference to a note at the foot or margin of the page,

b. to indicate the omission of words or letters,

c. to distinguish words and phrases as conjectural, obscure, or bearing some other specified character,

d. as a dividing mark, or for similar typographical purposes.

The syntactic stigma falls under subentry c, but the gloss there is far too broad to be helpful. Not only is the use of the asterisk in syntax missing, but so is the much older use to mark reconstructed forms in historical-comparative linguistics.

Once the asterisk was available for marking ungrammaticality, the way was open for other marks to be deployed for related purposes: the question mark (or numbers of question marks) for examples of dubious or uncertain grammaticality, the check mark (or prepended ok) for grammatical examples, the percent sign for examples with different grammaticality in different varieties, the pound sign (octothorp, double-cross, etc.) for examples that are pragmatically anomalous in a given context. Each of these has its own history.


7 Responses to “The stigma of ungrammaticality”

  1. arnold zwicky Says:

    Pieter Seuren, who is among other things, a historian of linguistics (Western linguistics: An historical introduction (1998)), writes with his own recollections:

    As regards the earliest use of the asterisk to mark ungrammaticality, I clearly remember seeing the asterisk appearing in the “new” linguistic literature during the late 1950s, just as your [posting] says. But I have no idea which publication was first. I simply registered the fact that, apparently, this had become established usage, and I followed it.

  2. Bruno Estigarribia Says:

    Perhaps somebody with a handy copy of Bloomfield (1933) should check this?
    “In formal linguistics, ungrammatical sentences are labelled using an asterisk (*):

    *The subject of this sentence is.

    This convention goes back at least as far as Leonard Bloomfield’s Language (1933).”

    • arnold zwicky Says:

      Luckily, I have an extra copy of Bloomfield’s Language at home — luckily, because my original copy is in my library, which I’ve been unable to get to for many months (see postings on my disintegrated hip saga). And there, in a endnote on section 10.1 (p. 516), we get:

      The asterisk before a form (as, *cran) indicates that the writer has not heard the form or found it attested by other observers or in written documents. It appears, accordingly, before forms whose existence the writer is denying (as, *ran John), and before theoretically constructed forms (such as *cran, the theoretically posited independent word corresponding to the compound-member cran- in cranberry). Among the latter the most important are ancient speech-forms not attested in our written records, but reconstructed by the linguist.

      Here Bloomfield connects what are in later literature treated as very different uses of the asterisk. It remains for someone to find instances of the asterisk of ungrammaticality before its rise to common practice in the syntactic literature of the 1950s.

  3. Bruno Estigarribia Says:

    Hmmm, I cannot find the old entries, while sillily (?) hoping that they refer to a very hip saga of yours that somehow disintegrated, and not the other possible reading, Arnold 🙁
    Any tips on how to search for this (authors in the 40’s, perhaps? Morris, Pike, etc.)? It is an interesting question for the historiography of Linguistics. Also, it seems to me that it might fit nicely with a sociological analysis of the rise of generativism.

    • arnold zwicky Says:

      On the hip saga, search my site for “hip” as keyword, and you’ll get links to some of the bulletins (though many more went out on Facebook; stuff got/gets posted on my blog only when I had/have a linguistic hook for the reports). I’m still regaining abilities; every day is a little better than the one before, and I can’t tell you how wonderful it is not to be living with excruciating pain, almost unable to walk at all.

      But yes, the historiographic question is very intriguing. Given the tantalizing bit in Bloomfield, I’d look first to those most directly and explicitly influenced by Bloomfield, *especially* Charles Hockett. (He and I eventually established an amiable and mutually respectful correspondence, late in his life, and I wish now that I’d thought to ask him about this point, but of course you never think about these things at the time.)

      As I have to keep noting, I am very much not a historian of ideas, and I don’t have the background or resources to investigate many of the questions I can frame. (Like historical reconstruction in linguistics, the history of ideas is a dauntingly difficult field of study, requiring absurdly broad sheer knowledge and techniques for investigating and interpreting the data, and I just don’t have these abilities, though I do caution people that if they think they can just walk into this stuff with some bright ideas and a few pieces of data, they’re fools.)

  4. arnold zwicky Says:

    1/4/13, e-mail from Arto Anttila:

    If it goes as far back as Bloomfield, we may have no hope of ever getting to the bottom of this.

    Your message reminded me of the 1996 CLS meeting where Arthur Merin, a German linguist, made the following connection between the asterisk of ungrammaticality and the asterisk of reconstruction: “In syntax, an asterisk means that native speakers ought to complain. In historical linguistics, it means no one ever will.”

  5. arnold zwicky Says:

    And now from Fritz Newmeyer in e-mail:

    Fred Householder takes credit for the asterisk of ungrammaticality, or at least its spread:

    Householder, F. W. (1973). On arguments from asterisks. Foundations of Language, 10, 365-376.

    See page 366. He sort of implies that he devised it at the 1958 institute and that’s where Lees got it from.

Leave a Reply

%d bloggers like this: