Pronouns 'n' stuff

« previous post | next post »

The comments on Geoff Pullum's recent "grammar gravy train" posting have wandered into the confused territory where the grammatical terms pronoun, possessive (or genitive), and determiner live. (The first two have a long history, going back to the grammatical traditions for Latin and Greek. The third is much more recent; OED2 takes it back only to Bloomfield's Language in 1933.) We've been over this territory on Language Log several times, from several different angles. But here's one more attempt at clearing things up.

First, pronouns. Simplifying things somewhat, the grammatical tradition says two things about them. One, they constitute a part of speech (one of a small number of these, including Noun as a separate part of speech from Pronoun). Two, a pronoun "stands for", or more precisely, "takes the place of", a noun (that is, the function of pronouns is to avoid repeating nouns); people, among them "Back of beyond" in a comment on Geoff's posting, sometimes say that this claim follows directly from the etymology of the word pronoun (though I have cautioned many times that Labels Are Not Definitions; see here).

Instead of part of speech, I prefer the term syntactic category, or when the meaning is clear from context, simply category; the terminology allows me to talk about phrasal categories as well as lexical categories. But there's no substantive issue here; it's simply a matter of terminological taste.

There is, however, a set of questions about the lexical categories of English. The lexical categories of English in the Cambridge Grammar of the English Language are not quite the same as those of traditional grammar; CGEL has a nontraditional category Determinative (with members traditionally classed as a subtype of Adjective) and doesn't treat Noun and Pronoun as separate categories. I say "doesn't (exactly)" treat them as separate categories, because CGEL does treat pronouns as a subtype of the Noun category, and indeed distinguishes subtypes of pronouns, among them the (definite) "personal pronouns" that were the focus of the comments on Geoff's posting on pronouns.

As I say here every so often, the world of lexical categories is complex. In particular, no matter which ones you take as in some sense "basic", you'll have to recognize many subtypes, and supertypes as well (for instance a supertype A embracing both Adjective and Adverb). It's not even clear to me that the choice of one set of basic categories rather than another can be justified on empirical grounds — so there might not be a substantive issue here, either.

As for pronouns taking the place of nouns, here I think traditional grammar conceptualizes things very badly. Again, I've noted here every so often that the way to think of anaphors is not as pointing to linguistic expressions (whether words or larger expressions), but as pointing to referents. The latest revision of the OED (September 2009) glosses pronoun this way:

A word that can function as a noun phrase when used by itself and that refers either to the participants in the discourse (e.g. Iyou) or to someone or something mentioned elsewhere in the discourse (e.g. sheitthis).

This definition doesn't cover all types of pronouns; it doesn't work well for indefinite pronouns (someone, everybody, etc.) or interrogative pronouns (who, what, etc.), for instance, but for definite personal pronouns it does about as well as you can in a short space. It's couched in terms of referents, not linguistic expressions, and there's no talk of replacing repeated nouns. And it classifies pronouns as words (with a characteristic semantics/pragmatics) that can constitute a NP on their own.

On to possessives (or genitives). Geoff notes that he doesn't distinguish these two terms, though he has a personal preference for genitive to refer to inflected NPs, as in Kim'stheir picture, while I prefer possessive. (Some writers prefer to reserve possessive for the syntactic construction with of, as in a picture of Kim / them.) This is a matter of terminological taste, not substance.

What, then, is the category of possessive expressions? Many writers answer Adjective (and use that answer to argue, invalidly, that possessives can't be antecedents for pronouns; there's extended discussion of the Possessive Antecedent Proscription here, with criticism of the Adjective answer, and there are links to Language Log discussions here). Others give a more sophisticated answer, Determiner (or in "Back of beyond"'s variant, Article), but I've observed a number of times here that this is not quite right, because though possessive-marked NPs can serve in a Determiner function, as far as syntactic category goes, they're NPs.

CGEL is careful to distinguish category from function — not at all a new distinction with this work, but one that's somewhat tricky in the case of non-canonical Determiners like possessives and the nominal determiner a lot (of). English does have a lexical category of words whose primary function is as determiners (the articles a and the, the demonstratives this and that, some quantity modifiers, like some and every); this is the category CGEL labels Determinative.

Putting all this together: the my of my book is the possessive form of the 1st-person personal pronoun; pronouns are, among other things, words (with characteristic semantics/pragmatics) that can constitute a NP on their own; and possessive NPs can serve in the Determiner function (with the syntax associated with that function).


  1. Karen said,

    November 3, 2009 @ 3:10 pm

    Unfortunately, pronouns can replace many things, including complete clauses. Perhaps we should call them "prowords"?

    Anyway, I tell my students that people are lumpers or splitters. If you're a lumper you don't mind if sometimes (often) in your small quantity of categories are tons of "exceptions"; if you're a splitter you want every category to be uniform, even if this sometimes means you have a category of one word, not to mention dozens or more categories.

    "It's a kind of adjective" works for lumpers…

  2. Lazar said,

    November 3, 2009 @ 5:17 pm

    For English, I prefer "possessive" over "genitive", because the s-construction is a clitic, rather than a case ending – I've had the "king of Sparta's wife" example stuck in my head for years. Likewise I prefer "subjective" and "objective" over "nominative" and "accusative". The Latin case names just seem too Latinate when applied to English.

  3. Karen said,

    November 3, 2009 @ 7:22 pm

    "Accusative" is particularly ill-suited, since we use the same form for dative as well.

  4. Mark F. said,

    November 3, 2009 @ 9:25 pm

    Actually, I don't think all this terminological variability is helpful on the pedagogical front. I realize every field has it, but when I was learning grammar it seemed especially common to hear that there was some alternate name competing for mind share with the one we were actually being taught. It didn't increase my motivation to keep the terms straight. Actually, the verb forms were the worst. The possessive/genitive option wasn't a distraction because I only heard of the latter when I started taking Latin.

  5. Adrian Morgan said,

    November 3, 2009 @ 9:32 pm

    I sometimes visualise the language as an extremely complicated Venn diagram, with as many circles as words, and each circle containing the set of contexts in which that word may be grammatically used. Even this is something of an idealisation.

    Syntactic categories are a way of simplifying this picture by identifying the major patterns with some loss of accuracy in the details. To expect the language to boil down to an objectively defined group of syntactic categories without exceptions would be insane. Grammar is handled by the human brain, and the human brain does not do simple; the human brain does not do objective. There's no reason why language should be any different, and plenty of evidence that it isn't.

  6. Lazar said,

    November 3, 2009 @ 11:42 pm

    @Mark F: One nice case is the fact that the subjunctive is called the conjunctive in German.

  7. Robert T McQuaid said,

    November 4, 2009 @ 12:20 am

    While on the subject of determiners, I would love to see one of the language log experts opine on the following construction:

    John and Mary's home

    As used, it is intended to mean (John and Mary)'s home, not just John and (Mary)'s home. Some other languages have a particle that can attach to the end of any noun phrase with this meaning, but not English. I have heard the construction in the speech of educated speakers, and sometimes even in print by professional journalists. A curious extension is with pronouns:

    … in John and I's home

    Oddly, this doesn't work without the uninflected I. "In John and my home" or "in John and me's home" seem even more peculiar.

  8. Jerry Friedman said,

    November 4, 2009 @ 1:26 am

    It's always a pleasure to read an article on grammar that separates terminological from substantive issues. (Maybe I'm looking in the wrong places.)

    However, I must say that the OED definition of "pronoun" seems to work only for people who already know what a pronoun is. If a reader didn't know, it lacks something important.

    A word that can function as a noun phrase when used by itself and that refers either to the participants in the discourse (e.g. I, you) or to someone or something mentioned elsewhere in the discourse (e.g. she, it, this).

    I see nothing there that eliminates one-word names or those nouns that don't require determiners. For instance:

    "Justice is the most important thing, John."

    "Justice is important, Mary, but other things are more important."

    "John", "Mary", and "justice" can all function as noun phrases by themselves, and "John" and "Mary" refer to the participants in the discourse, while "justice" refers to something mentioned elsewhere in the discourse. There needs to be something about how pronouns 'depend on something else to establish their references.

    And can pointing count as mentioning elsewhere in a discourse?

    "I'm talking about the guy who's sort of tall, with… he looks like him."

    "Oh, that must be Joe."

    "Right. Well, Joe said…"

  9. montgomery said,

    November 4, 2009 @ 8:21 am

    @Robert T. McQuaid
    … in John and I's home

    Oddly, this doesn't work without the uninflected I. "In John and my home" or "in John and me's home" seem even more peculiar.

    Weirdly, what I find myself saying is "In John and my's home" which is totally wrong, I know. I simply don't say the more grammatical "John's and my home," maybe because I'm already past John before I realize it should have been possessive. I always stumble over this construction, to the point that I usually end up rephrasing the sentence halfway through. Any other native speakers with a similar problem?

  10. Zwicky Arnold said,

    November 4, 2009 @ 8:27 am

    To Robert T McQuaid and montgomery: see my 10/12/08 posting "Coordinate possessives", here.

  11. Aaron Davies said,

    November 4, 2009 @ 8:57 am

    @montgomery, robert t. mcquaid: i don't think anything along those lines is produceable in my idiolect–(my) english simply lacks the ability to render that concept with those words. i'd probably say something like "in our home—john and me—…" or possibly "in our home—john's and mine—…".

  12. D. Sky Onosson said,

    November 4, 2009 @ 11:44 am


    I would probably also find myself saying "John and my's home" (that is, if I knew someone named John, and shared a house with him!)

    I'm not sure that I would say "it's totally wrong", though. I say it, you say it, it makes sense. Where is the "wrong"?

  13. D. Sky Onosson said,

    November 4, 2009 @ 11:45 am

    P.S. I would also be tempted to phrase it as "me and John's home". That one sounds utterly natural to me.

  14. D. Sky Onosson said,

    November 4, 2009 @ 12:04 pm

    I should also add that I live in an area where it is common to hear phrases such as "Your guys's car". Guys's sounds the same as guises.

  15. dr pepper said,

    November 5, 2009 @ 10:27 pm

    I would say "John's and my home".

  16. Franz Bebop said,

    November 7, 2009 @ 11:21 am

    Just the humble opinion of one native speaker:

    "John and I's house" makes a kind of sense to me, but my opinion is that this is only a result of the overcorrected use of "… and I" for all phrases that include "me" and someone else. I would call this usage "affected." I would avoid this usage.

    "Me and John's house" is something I was taught not to say, but my instincts as a speaker say this is grammatical. I might label this as "slang."

    I would not say "John's and my home." I'm not sure I can label this as "incorrect" but I just wouldn't say it or write it.

    If you take the phrase "John and Mary's house" and you want to replace Mary with some kind of pronoun, I'm not sure there is a good solution. When writing carefully, I would avoid this problem altogether. If the house is "me and John's house" then it's also just "my house" and "our house" and "the house I share with John."

    It looks like Mary can't be pronouned.

    Hey, look, I verbed "pronoun" !

RSS feed for comments on this post