Gfrktyl

Could anyone explain why this sentence is considered ungrammatical?

You often hear quite literate people saying hideously ungrammatical things such as: "He is the kind of person who, if he had lived in the 19th century, people would not have been able to categorise him."

First, as others have pointed out in the comments, the sentence uses "who" when it should use "whom." Even native speakers fail to correctly distinguish between "who" and "whom". It is common for "who" to be used in all cases, but this is not recommended in formal speech or writing. I have made that correction throughout the rest of the answer. However, there is a larger issue in the sentence, which I believe is the intended focus in the original source.

I think the issue becomes clearer if you omit the nonessential clause "if he had lived in the 19th century".

He is the kind of person whom people would not have been able to categorise him.

One could write "People would not have been able to categorise him." as a complete sentence, or one could write "whom people would not have been able to categorise" as a relative clause describing "person". However, the example sentence combines the two, beginning as a relative clause and ending with another pronoun "him". In this sentence "whom" is already the object of "to categorise".

We can remove "him" and reintroduce the nonessential clause to get the correct sentence:

He is the kind of person whom, if he had lived in the 19th century, people would not have been able to categorise.

The nonessential clause interrupts the flow of the sentence, which can make it more likely to miss mistakes like this one both in reading and in writing. While the sentence is now correct, an even better sentence might be:

He is the kind of person whom people would not have been able to categorise, had he lived in the 19th century.

As Tashus says, this breaks the general rule that resumptive pronouns are ungrammatical in English, however, if "You often hear quite literate people saying [this]" (i.e. presumably native speakers who speak a prestige dialect), then this instance is grammatical. The rule needs to adapt to encompass modern usage (I'm reminded of people becrying split infinitives).

Wikipedia goes into more depth about the contexts in which resumptive pronouns are seen as grammatical:

... in English, "relative clauses with resumptive pronouns are officially ungrammatical [...] However, they are in fact not uncommon in speech". However, their grammaticality is influenced by linear distance from the subject, embedded depth, and extractability...

In a relative clause, resumptive pronouns are generally not seen as grammatical, however their level of grammaticality improves as they get farther from the head.

In short sentences without a subordinate clause, they are clearly ungrammatical:

He's the one who people categorised *him.

The exact rules for grammaticality aren't well understood, but broadly, the further the pronoun gets from the subject (e.g. the longer the subordinate clause is), the more acceptable it sounds (at what length is dialect dependent). The following may or may not sound acceptable:

He's the one who, had he lived then, people wouldn't've categorised (*)him.

And at the other end we have the example sentence from our "even quite literate people":

He is the kind of person who, if he had lived in the 19th century, people would not have been able to categorise him.

He is the kind of person who, if he had lived in the 19th century, people would not have been able to categorise him.

This is a rhetorical device known as anacoluthon. The grammatical structure of the sentence shifts as the sentence is spoken — or, you could view it as, two different sentence structures have been smushed together into a single sentence. (Did you notice, I did it there. And (arguably) again!)

The two sentence structures smushed into this one sentence are (with the primary subject/verb in bold):

He is the kind of person who would not have been categorizable by anyone in the 19th century.

If he had lived in the 19th century, people would not have been able to categorize him.

The actual sentence spoken ends up smushing the two grammatical sentences together, typically with a moment of grammatical confusion (indicated by a comma or dash) in the middle.

Wikipedia gives several examples of anacoluthon in English, including Milton's

Had ye been there – for what could that have done?

The two sentence structures smushed up here are

Had ye been there, you could have done something. (Well, actually, no, nothing— never mind.)

It's okay that you weren't there, for what could that have done?

The implication is that the speaker changes his mind halfway through the line.

In your original example, the speaker has not changed his mind about what he wants to say; he's just made a little tweak to how he wants to say it.

For an extreme example — in which the extreme disorder of the words does (well, is implied to) reflect the extreme disorder of ideas behind them — look at Fred Armisen's "Nicholas Fehn" character on Saturday Night Live.

https://twitter.com/sarahcpr/status/760657101404327936

You know, it's— it's the reason... I wake up— I wake up, like, anybody— I was taught... Every— well, most Americans, if you— Education. Any— any border, if Helsinki, if Oslo— I think— any publication, we would— isn't it integral, isn't it the most important— it's the substance, the very idea, that we can unite, that makes me feel, personally—