This is a bit of a second take on my original insight on double referents. I’m also going to change gears slightly by introducing two related-but-not-immediately-intuitive concepts, so I hope my dear readers will bear with me as you patiently wade through my confusion of jargon and pseudo-logic.

So I’m gonna start by explaining how these constructions work syntactically (at least, so far as my middling understanding can take it), and follow it up by why this type of error is so exciting. As a sort of refresher and motivation to work through this post, I was looking at the construction and deconstruction of the error generated by non-native speakers of English in phrases like (1) and (2).

(1) Your questions are always hard to answer (*it)!

(2) John is easy to please (*him).

In both of the above sentences, the inclusion of the final pronoun makes the sentence ungrammatical, hence its starred status. Current models for syntax have a hard time modelling these kinds of constructions, but the current-standing model is called Tough Movement, because such sentences generally contain the word tough in ’em.

Logically speaking, the preceding noun is the object of the ending verb; in (1), your questions is the direct object of the verb answer, and likewise in (2), something is the direct object of do, such that they can be rephrased as follows:

(1ʹ) It’s always hard to answer your questions!

(2ʹ) It is easy to please John.

If we assume these new sentences to be the conceptually original structures, it’s easy to claim that the object (your questions; John) move to the front, and replace it. Note that these “tough” constructions aren’t quite the same as noun phrases like “the book that I read [*it]” or “the game that I bought [*it]”, although the movement is very similar. Unfortunately, I have not yet noticed non-native speakers making such errors, but it could be argued that such noun modifiers are quite advanced for language learners.

1. Tough Movement

How/why these noun phrases (NPs) move to the front is a little beyond the scope of this blog (or indeed, even within the scope of this author’s ken), but whatever the reason, the end result is that the object NP moves to the subject position, and is not replaced by any pronouns.

In syntax, we think of the source of the movement as an unspoken placeholder (phonologically null) trace, so we mark this movement by a “t”. In order to avoid ambiguity, we also use subscript letters to index equivalent phrases, and square brackets to indicate constituency. For example:

(3) a. I killed [a man [with a knife]]
=> the man I killed had a knife, but the method/weapon I used to kill the aforementioned man is not explicitly specified;

(3) b. I [[killed [a man]] with a knife]
=> I used a knife and killed a (probably unarmed) man.

In (3-a), [a man [with a knife]] indicates that with a knife modifies a man, thereby indicating that the knife is possessed by the guy who is not the speaker. Similarly, in (3-b), with a knife modifies kill, indicating that the speaker was indeed in possession of the knife instead of the poor victim.
So, our sample sentences now look something like:

(1ʹʹ) [Your questions]i are always hard to answer ti

(2ʹʹ) [John]j is easy to please tj

As I have hopefully successfully indicated, the original position of your question leave behind a trace, while your question gets upgraded to the head of the sentence. The trace “t” obviously never gets pronounced (or, as we say, “phonologically realized”), but conceptually, it’s still there, because we understand that the new head is still suffering the verb that follows it.

2. Expletive Pronouns

But what’s even more exciting is why non-native speakers would insert an it (or him) at the end of such clauses. Certainly, certain languages like Turkish oblige an agreement pronoun to make it grammatical (so that utterances like “the book that I bought it” would be acceptable in Turkish).

English belongs to a group of languages called “non PRO-drop languages” which means that we are not allowed to drop subjects or objects just because they’re unspecified. Instead, we oblige an expletive pronoun (or a dummy pronoun) to serve as a syntactic replacement. Consider the following:

(4) a. It’s raining today.
(4) b. *The sky is raining today
(4) c. *Is raining today.

(5) a. [After five years of trying,] I finally made it!
(5) b. *I finally made my goal.
(5) c. *I finally made.

In (4-a), the expletive pronoun is in the subject position — there can’t be an explicit subject, but English requires one anyway, so we throw in the generic it to serve as a placeholder for the subject instead. Conversely in (5-a), the it is serving as a placeholder for the object of made, because made cannot be an intransitive verb.

In (4-b) and (5-b), replacing the dummy pronoun with an explicit NP doesn’t exactly work. The sky, the clouds and even the heavens can’t syntactically rain, even though rain clearly comes from one or all of the above. (5-b) is technically a grammatical sentence, but it means something totally different from (5-a); making a goal is merely identifying what the goal is; making it means you’ve already achieved the goal, so it’s been ruled out as an alternative candidate.

Finally, in (4-c) and (5-c), we see what happens when (probably speakers of PRO-drop languages, like Chinese or Japanese) make such utterances. (4-c) is unequivocally ungrammatical, but it’s definitely a kind of utterance that one would have heard with a certain subset of non-native speakers. Similarly in (5-c), native speakers listening to this utterance will lean in a bit at the end, waiting for the final bit of info that should come after made, since made is a transitive verb.

So returning to our original sentences, why are errors like “Your questions are always hard to answer it” and “John is easy to please him” so interesting? For one thing, these errors are obviously generated by non-native speakers; they are never taught as being grammatical in the classroom or by their native-speaker peers. Additionally, in the paragraph above, I mentioned how PRO-drop languages would also run the risk of dropping pronouns in non PRO-drop languages, such as when they’re learning English. However, these same speakers, once they acquire that sense of obligation for dummy pronouns, (admittedly erroneously) insert an expletive pronoun as the object for the complement verb.

Non-native speakers who generate such utterances as “your questions are hard to answer it” clearly demonstrate a knowledge of expletive pronouns, even though their native languages would not require such a thing. It implies that they’ve managed to learn how to un/consciously insert pronouns as placeholders in order to maintain the grammaticality of English sentences. Even though our main sentences in question contain errors, they demonstrate the fact that the speaker is not just blindly inserting a pronoun at the end, as indicated by (2) — the use of him clearly indicates that the speaker knows that him is referring to John, and more importantly, that it’s the object of please, since him (object) was chosen over he (subject).

My conclusions from my original post still stands: it’s a relatively sophisticated error. Speakers who generate this kind of error clearly have some knowledge of the fact that unlike their native language, they cannot drop pronouns in English; and also are somewhat aware of the rules behind NP fronting. Even though NP object fronting is the only situation where the transitive verb can be stranded, non-native speakers demonstrate that they know that transitive require objects, and that’s why they (wrongly) insert an additional pronoun in order to make it sound grammatical to them.