How do syntacticians explain object pronouns in the subject position ("Me and him" or "Lui et moi")?

Question

Me and a lot of other native English speakers sometimes use object pronouns as the subject of sentences if there's an "and" in the subject. This has been mentioned on Stack Exchange before but I don't think anyone discussed the syntactic theories of it:

• "He and I", "Him and me"

• "I and someone", "me and someone" or "I and someone we"

• Universals and emphatic pronouns

In school, my fellow English speakers and I learned how to force ourselves to say "he and I". This has a certain logic, that the subject of the sentence always use subject pronouns. So why is "him and me" possible as a subject?

In English this is considered an error or a low-prestige variant, but my question is not about errors or the social status of language. My question is about how a syntax can allow object pronouns in the subject position. In the comments, the user @jlliagre said that French also has the exact same object-pronoun syntax that I would like to have explained:

In French, the only used forms are Lui et moi savons[...] or Moi et lui savons[...]. Using subject pronouns is impossible and never done in French when a conjunction is used. In particular, while 'You and I are going on vacation' is taught to be the "best" English form, neither Toi et je partons en vacances nor Tu et je partons en vacances are remotely possible in French.

If it's easier to describe the phenomenon in terms of high-prestige French than low-prestige English, then that would answer my question nicely.

I'm sure different syntax theories address this in different ways. I took a class in a Chomskyan X-bar theory, but I'm not as smart as I was back then. If you can describe it in those terms, it will make a little more sense to me. I think that the way we would have done it in class is that the underlying form of the pronouns are all object case like "us", and that the pronouns can only take nominative case if they are moved up the nominative case position in the tree. I would guess that there is something that blocks heavier multi-part constituents like "him and me" from moving that far up the tree. I don't remember the theories of blocking.

Answer 1

This raises a more basic question: How does generative linguistics account for apparent errors?

1. Are the rules sometimes optional or multivalent?
2. Do the rules have exceptions we haven't identified?
3. Do people fail to acquire their language?
4. Do people fail to execute their linguistic knowledge?

Different accounts can be given for an utterance like "me and him" according to which of these hypotheses we select.

1. The first hypothesis is probably the least attractive to anyone trying to create a good linguistic generator. Functions are not supposed to yield multiple outputs for a single input. Nevertheless, we do have the concept of free variation, at least in phonetics. Why not also say that me and I are both acceptable surface representations of the morpheme?

2. The second hypothesis seems to underlie your question as asked. Is there some way we can formulate the structure and derivation rules such that me is expected after all? This seems like quite a stretch if we want to somehow make an object a subject (but then, I'm probably not as smart as you were in your X-bar–studying days either). But the commentators have pointed to very interesting phenomena around ordering and mixing cases: [he and I], ?[he and me], [me and him], *[me and he], *[him and I], ?[him and me], *[I and him], *[I and he]. These suggest something interesting, perhaps even across branches of linguistics (e.g. a phonological reason).

3. The hypothesis that people fail to acquire their language fully is easy to cite as a cause of language change. The creation of the modal future with will, the breaking of a napron to an apron, etc. show that we can see them as cases of the reanalysis of a surface representation such that it yields a different deep representation. More contemporary ones are still debated, such as whether than is a relative conjunction ("faster than I am") or a preposition ("faster than me"). According to this hypothesis, "me and him" is used by those who fail to acquire the same distribution of case distinctions.

4. Similarly, generative linguistics also knows of "production errors", where the language has been acquired cognitively but the linguistic apparatus is subject to misfiring. I doubt that this is the correct hypothesis here because unlike with your typical production error, I'm sure we all know people who we could press to find something wrong with "me and him" and who wouldn't know what it was, or who might only know "some pedant told me it was wrong, so I guess it's wrong", which doesn't count as having acquired the rule.

How we account for the use of "me and him" depends on which hypothesis we start with. Personally, I suspect it's a mix of the second and third: something about the context pushes toward the "me" realization (I don't know what, and it's the interesting part of your question); and if it happens often enough then people begin to acquire a different version of the language in which either subject case is not needed there or "me and him" is the realization of the subject case. (A good account of the context would yield a test to disambiguate those two descriptions.)

Answer 2

Almost by definition, different syntactic theories should not explain why many native English speakers find "him and me" more natural, because syntactic theories are about the principles underlying a person's ability to produce and understand utterances, and not about their attitude towards utterances. Sociolinguistics is about attitude, caused by social facts. There could be a well-defined sub-area of sociolinguistics, analogous to sociophonetics (i.e. sociosyntax) which uses the devices of a syntactic theory to account for attitudes (I don't know if there yet is such a sub-area). Syntactic theory simply accepts that some people's grammars generate one pattern, and some people's grammars generate a different pattern, and it gives you the tools to talk about each pattern.

Variation is "the big problem" in the study of grammar. There is a tendency to treat all variation as essentially the same, but that is clearly mistaken – there is individual variation, and social variation. Social variation is where a person says something about English which turns out to be true of American English but not Indian English. Individual variation is where a single individual allows more than one output: individual variation is the interesting kind of variation from the perspective of grammatical theory. If I tell you that English and Norwegian differ in some way, you probably wouldn't find that to be a kind of "variation", yet because various different languages are called "English", we have a hard time accepting that people in Newcastle speak a different though related language from people in Seattle.

Individuals have the ability to adjust their linguistic behavior especially when repeatedly exposed to other dialects, therefore I might actually say, sometime, that "I and a bunch of others are going to the pub", which is clearly forced linguistic retraining due to the school system. My revulsion for "I" in a conjunction is strong, greater than that for "he, she, they" ("he and you should get together" is okay-ish, though "him and you should get together" is better). I guess there must be others who are more accepting of the normative dialect.

If we set aside the problem of how to dispose of individual variation, we can certainly ask "what structural facts determine this pattern of pronouns". We can also change the pattern and ask "What differentiates pattern A from pattern B?". Here is a table of pertinent combinations in subject position (no judgment attached).

1  He is leaving soon.
2  I am leaving soon.
3  Me is/am leaving soon
4  Him is leaving soon
5  Him and me are leaving soon.
6  Him and I are leaving soon.
7  Me and him are leaving soon.
8  Me and he are leaving soon.
9  He and I are leaving soon.
10 He and me are leaving soon.
11 I and he are leaving soon.
12 I and him are leaving soon.

Some of these are complete gibberish, some are perfectly natural, and some are questionable. You decide.

3,4 are utterly ungrammatical; 1,2 are wonderful. 5 is the best conjunction, followed by 7 but I know the rule that I was told which forces that order. 11 and 12 are the worst of the conjunctions. 6,8,9,10 are hard to sort but I'd say that 6 is better than 9; 8 is the worst of those 4.

I left out from the paradigm of inversions structures like "Are {he and I; me and him; him and me...} supposed to go soon". Also considering how terrible it is to put "I" or "he" in object position (conjoined or otherwise), and considering the fact that the natural 1-word (pronoun) answer to a question "Who Verbed X?" is "me, him, her, them" and never "I, he, she, they", I conclude that "I, he, she, they" are only used in subject position. You get the "I am leaving" / "Me and him are leaving" difference by appeal to the syntactic notion of C-command. In some (less-flat) theories of syntax, the tendency to use the unmarked pronoun form increases when the pronoun is structurally further from the following verb.

So this is a question that is eminently addressable by syntactic theory, as long as you only ask for the structural correlates of a pattern. Non-syntactic theory may explain the motivation for the arrangement of preferences, for example

the reason why "me and him" is dispreferred is that I was told as a child that it makes people think that he is mean.