Eine kleine Nachtsyntax

Mr. Tweedy has an interesting entry today on language use. Well, 1/3 of it is interesting, anyway - the part about the use of 'I' vs. 'me' in formal and informal English.

I seize on this because it happens to be one of my biggest pet peeves about the way some people speak English (no surprise to anyone - for all we Linguists go on about being anti-prescriptivists, we ALL find ways to justify nevertheless being annoyed at how some people speak). Namely - people who overuse 'I' rank in at just barely more tolerable than those who overuse 'whom' in my registry of demons.

On the whole, I agree with Mr. Tweedy's analysis - so jump over and have a look if you have the time. The gist of it is this:

1. Frank and I (subject)- fine in both formal and informal English


2. Me and Frank (subject)- fine in both formal and informal English, though the form in (1) is preferred in formal contexts


3. Frank and me (subject)- bad in formal English - strangely worse than (2) informal English as well


4. When in object position - 'I' form is NEVER acceptable

The data is(are, whatever) correct, of course - but I'd honestly never noticed the 'Frank and me' vs. 'Me and Frank' effect - whereby the first is somehow worse than the second - in informal English before. And in fact, if you'd asked my opinion without pointing that out to me, I would've predicted just the opposite. Reason being - as Mr. Tweedy points out (though not in precisely these terms) - accusative (more properly, 'objective') case seems to be the default case in English. When someone asks you "Who wants candy?" your response is invariably 'me,' rather than 'I.' Now, as Tweedy points out, it's natural to say 'I do' (and in fact, 'me do' is completely ungrammatical) - presumably because the tensed verb(-placeholder) 'do' assigns nominative case to 'I.' Meaning that as a standalone lexical item - the first person singular pronoun is probably 'me,' and it only changes to 'I' in a specifically case-marked form; objective case is default in English.

So in a conjuntive phrase, I would rather expect the second member to be the unmarked one.

Mr. Tweedy thinks differently, and argues for an "end of phrase" effect for case marking. Basically - the second item in a phrase is more likely to be marked. So if it's "Me and Frank went to the movies," then "Frank" gets marked because it's second, but of course this marking is vacuous for proper names (indeed, any specified nouns) in English. And if it's the other way around, then "me" gets marked, and this time we can see the marking: it turns up as "I."

I think he would've been on more solid ground arguing for a "closest conjunct" effect (basically, whichever is closer to the verb gets marked), and here's why.

First of all - let me take issue with the second part of his justification. As (further) evidence for his "end of phrases get marked" thesis, he cites the case of "Jon's and Bob's car" vs. "Jon and Bob's car." Both are OK - point being, we mark the end of the phrase obligatorily, the first member optionally. Actually, I think this is a misanalysis of this particular case. In fact, in the "Jon's and Bob's car" case, it's that BOTH "Jon" and "Bob" have phrasal status. Witness things like "The king's money" vs. "The king of France's money" vs. the completely ungrammatical "The king's of France money."
Now try it with "King and queen." "The king's and queen's of France's money" simply won't do. BUT - we can say "The King's and Queen of France's money" - it's just that we understand it to be the case that the King is not the King of France (actually, no one is, but never mind ;-), but rather some other King, and "of France" modifies only "Queen." Why is this so? Well, presumably because "King" and "Queen of France" are members of separate phrases. Reality is that we're allowed only one 's marking per phrase. So if we're coordinating two items with 's on them, we're really coordinating two (single-word, in this case) phrases.

But so far I'm just nit-picking. This still works for Mr. Tweedy's explanation, right? Because in the case of "[Frank and I] went to the movies," then we can analyze "Frank and I" as a phrase and say that only the last item in it is case-marked, right?

Well, yeah, but how, then, do we explain all of this:

1. OK: He and I went to the movies.


2. BAD: Him and I went to the movies.


3. BAD: I and Frank went to the movies.


4. BAD: Me and he went to the movies.


5. OK: Me and him went to the movies.

It's easy to see that the whole thing comes tumbling down. Mr. Tweedy's theory, as stated, can get away with number (1) - although I've given reasons to doubt that marking ever shows up (even optionally) on the first item in a phrase. But let it go - he gets an explanation for (1). But if his explanation is right, and the first conjunct can optionally be case-marked, then what's wrong with (3)? Why can we optionally case-mark 'him" as "he" in (1), but not "me" (as "I") in (3)? Likewise, according to this theory, (2) - in which case marking is only on the second conjunct - should actually be better than (1), and yet it's worse. Mr. Tweedy could get around this, maybe, by claiming that "him" isn't the default form of "he" the way "me" is the default form of "I" - but he would still be unable to explain example (4). He would CERTAINLY be unable to explain how (4) is possibly worse than (5) (in which there is no nominative case marking on the second conjunct), and yet (5) is better than (4)!

So Mr. Tweedy's missing some pieces here.

Not that I have an alternate explanation myself. Conjunctive phrases are NOTORIOUSLY difficult to handle in standard Syntax. But here's my oh-so-modern Minimalist suspicion: there is some truth to the idea that pronouns in English are clitic-like. Notice that "order in the phrase" plays a role for what you can and can't get away with, but that it doesn't map onto all pronouns in the same way. That's a big clue that we're dealing with peculiarities of the lexical items themselves, and not any kind of nice syntacitc generalization about English. Rather - it seems to be a PF-level effect. "Pronounce this this way here and some other way elsewhere." And in the case of "I," it's sort of a "pronounce me last" effect. Synactically it's a full member of the conjunct phrase, but phonetically it can only be pronounced last. (Note this holds even in the hypercorrected forms - when people say crap like "I feel the magic between you and I." Even those troglodytes can't get "I feel the magic between I and you!") More evidence: note that (5) gets really bad all of a sudden if you swap places: "BAD: Him and me went to the movies." There are some dialects where you can get away with that, but in the standard dialect it's right out. Again - the ordering effect seems to be a peculiarity of the lexical items - not anything deeper than that. "Me" just likes to be first is all. Still not convinced? Try these two on for size:

1. He and him are good friends.


2. Him and he are good friends.

Neither is particularly good, but dontcha like (1) A LOT better than (2)?

I'm telling you - all it is is an ordering effect, no different from, say, the fact that in Romanian, all the object clitics go in front of the auxilliary - bucept the feminine one, which randomly goes after the main verb.

1. L-am vazut.


2. N-am vazut pe nimeni.


3. Am vazut-o

(1) and (3) are "I saw him/her" (respectively). (2) is "I didn't see anyone." It's there just to show that clitics in general go in front of the auxilliary - it's only the feminine (accusative) one that's weird and goes after the main verb.

I think something similar is at work with pronouns in English. There are no revealing syntactic principles at work here - just like there is no deep reason why the feminine clitic is the odd "man" out in Romanian. It is that way ... because it just bleedin' IS that way! The solution for English pronouns in conjunctive phrases is a simple PF-side constraint ranking problem. It's one of those times when Optimality Theory really is the way to go in Syntax.