November 23, 2006

Let's meet at mine


The sharp-eyed Éamonn McManus noticed a gap in the list of independent possessive constructions in my recent "Overpossessive" posting: I illustrated the anaphoric zero, predicative, and double genitive constructions with both pronominal and non-pronominal possessives (mine, Sandy's), but gave only a non-pronominal illustration for the locative construction: Let's meet at Sandy's.  This was not an oversight -- I find Let's meet at mine unacceptable in a context where there's no antecedent for the missing head, and most other English speakers make the same judgment -- but now McManus provides some attestations of pronominal locative possessives (from the U.K. and Ireland), which suggests that some speakers are beginning to simplify their grammars by eliminating an odd constraint on one specific construction.

Which reminded me of Baker's Paradox:  although learners generalize ("project") from the language they hear, producing many utterances that are not directly modeled for them, in some cases they resist obvious generalizations and seem to conclude that things they haven't heard just aren't grammatical; they learn lexical exceptions and very specific constraints. 


The paradox gets its name from C. L. Baker, author of the 1979 Linguistic Inquiry paper "Syntactic theory and the projection problem", in which the issue (for lexical exceptions) was clearly presented.  (Baker was the first student to write a Ph.D. dissertation under my direction, back in the  Pleistocene Epoch, so it pleases me to refer to his work here.)  More recently, Peter Culicover's book Syntactic Nuts (1999) examined a series of puzzling constructions in English, using (apparently) arbitrary differences in syntactic behavior between lexical items to conclude that learning must be, among other things, "conservative".  And now, for her dissertation, Stanford student Liz Coppock is looking at the cases from this literature, plus some others, so Baker's Paradox has been very much on my mind.

The question is how people learn things like the following:

You can give $100 to the library, give the library $100, or donate $100 to the library, but not *donate the library $100.

You can be the likely winner, be likely to win, or be the probable winner, but not *be probable to win.

You can be happy, be a happy person, or be glad, but not *be a glad person.

(Searching on the net will get you small numbers of examples like the asterisked ones above.  While most people are conservative learners, a few are more adventurous.)

On to syntactic constructions, a world in which construction-specific (though systematic) constraints are rife.  A couple of well-known examples from English:

In main-clause wh-interrogatives, prepositions can be stranded or (in a rather formal style) fronted, but in wh-interrogative complement clauses, fronted prepositions are much less acceptable:

Which city did they fly from?  [main, stranded]
From which city did they fly?  [main, fronted]
I wonder which city they flew from.  [embedded, stranded]
?? I wonder from which city they flew. [embedded, fronted]

In two serial-verb-like constructions -- which I'll call GoV and TryAndV -- for most speakers the verbs must obey the Inflection Condition of Pullum 1990 ("Constraints on intransitive quasi-serial verb constructions in modern colloquial English", in OSU WPL 39.218-39), which requires that they be in a form identical to their base form (either the base form itself, or the non-3sg present); in other similar constructions, in particular GoAndV, there is no constraint:

I'll go and see what I can do. [GoAndV, base]
  I'll go see what I can do. [GoV, base]
  I'll try and see what I can do. [TryAndV, base]
I always go and see what I can do. [GoAndV, 1sg pres]
  I always go see what I can do. [GoV, 1sg pres]
  I always try and see what I can do. [TryAndV, 1sg pres]
He always goes and sees what he can do. [GoAndV, 3sg pres]
  *He always goes see(s) what he can do. [GoV, 3sg pres]
  *He always tries and see(s) what he can do. [TryAndV, 3sg pres]
I went and saw what I could do. [GoAndV, 1sg past]
  *I went see/saw what I could do. [GoV, 1sg past]
  *I tried and see/saw what I could do. [TryAndV, 1sg past]

In all such cases, the puzzle is why so few people generalize, why so few eliminate the wrinkles in the grammar.  Locative possessives provide yet another instance of the puzzle: the other three types of independent possessives are unconstrained, but the locative construction maintains its constraint against personal pronouns.

Until recently, that is.  In McManus's words:

Up until recently I would have assumed that nobody would ever say Let's meet at mine/yours/hers/ours/theirs but in fact it seems to be current usage in England and spreading to Ireland. I'm pretty sure nobody ever used that construction when I was growing up in Dublin, but my slightly-younger brother now uses it all the time. Google finds a few hits for Let's meet at mine, including [here], where it's the title of the page and the product it's selling. There are very few hits, though (none at all for ours, theirs, hers and only one possibly non-native one for yours), for what should be a very common phrase in chat forums so it may be that it's not yet all that widespread. But I'd bet on it spreading further because it's handy, immediately comprehensible, and logical.

Getting rid of the let's pulls in a modest number of examples, most of which seem to be from British sources, for example:

Looking forward to seeing the others I agreed to meet at theirs at 8pm.  (link)

(this from a blog full of British English features).  Searches varying other parts of the search string will no doubt yield many more examples.  The independent possessive pronouns are on the march!

[Addendum: two others have written to confirm that this usage is widespread in speech in the U.K. and also in Australia and New Zealand.  Both were under the impression that it spread fairly recently.]

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at November 23, 2006 02:47 PM