"It is me who likes sprouts"

[Cecil Ward]

----------------------------Original message----------------------------
[This is a modified version of a post I made on RRG-LIST.]

On the list "RRG-LIST", Emma Pavey recently brought up some interesting
facts about sentences like
        It is me who likes sprouts.
        It is him who likes sprouts.
        It is us who likes/*like sprouts.
Emma raised the point that the agreement pattern between the verb likes
and the (non-dummy) pronoun is very odd, showing number agreement but
not person agreement.

I was hoping that list members might be able to contribute some
background concerning the historical origin of this kind of sentence,
and make some comments on the basic identificatory sentence "it's me."
in English.

I am also hoping that list members would be kind enough to fill in some
of the gaps or point out some of the glaring errors in this suggestion
of my own.

I assume that sometimes languages evolve anomalous syntactic patterns
because they are trying to do two jobs in one hit, as if a single
sentence starts trying to do the work of two. Maybe such patterns even
evolve from the roll-up of sentence pairs accompanied by elision plus
various other "adjustments".

These sentences contain a "double-hit" semantic payload, in that they
perform carry out a double function in a compact package, that is, they
are identificatory "It is him." "He is the sprout-lover" - and they also
convey the secondary proposition "he likes sprouts". I am wondering if
it is reasonable to suppose that this compressed structure indeed
evolved in some way from two short sentences when a syntactic technique
became available that permitted their rolling up into one.

Another fascinating thing about these sentences is that they looks
superficially like they contain a straight relative clause, a modifier
on the non-dummy pronoun. My take on this is that this is the modern
situation is the result of historical reanalysis, rebracketing, if you
like, following which the "relative clause as modifier" interpretation
was bleached out.

Now as far as the minimal identificatory sentence is concerned, When
English is compared to German, say, what do we find? (Maybe list members
who are speakers of Germanic languages might care to chip in.)

I assume that for German we would have
        *Es ist ich/mich.
        *Es ist wir/uns.
Rather
        Ich bin es.
        Wir sind es.
And presumably,
        Ich bin es, der ...     "It's me who.."
        Wir sind es, die ...
The position of the comma, is interesting. Notice how the construction
is ripe for rebracketing. In English, maybe the demands of movement into
stressed position (rightward movement, the opposite of fronting) is at
work in English, our Germano-Celtic language, as in Celtic. Maybe this
was the driver towards the modern "It is me" vs a possible earlier "*I
am it".

Let's take the modern sentence pairs,
        It's me. The [one that likes/*like sprouts].
        It's John. The [one who/that likes sprouts].
        It's us. (=We are the ones.) The [ones who/that *likes/like
sprouts]. We can elide them together and remove the duplication
        It is us who/that *likes/like sprouts  -etc
Notice that the agreement pattern now fits.

In modern German, there is an identification between the definite
article, pronoun ("Die ist nett") and relativizer. I assume this was
true for early English (eg "that"), and presumably this played a
contributory role as well.

I assume then that we had the following situation when the dummy, always
3rd-person pronoun was on the right
        I am 3sg.  +  3sg REL likes/*like sprouts
        We are 3pl.  +  3pl REL likes/*like sprouts
We then invert to move the first pronoun into a fully stressed position,
according to the demands of its identificatory function. Meanwhile the
verb agreement on "like" retains the same pattern.

I keep returning to the question of various frameworks deal with the
results of reanalysis. How do we get past an overly literally parse of
the apparent syntactic patterns (literal reading: relative clause
modifier) in order to get to with the correct semantic load we actually
understand.

Cecil Ward