Arnold Zwicky's Blog

In a Law & Order: Special Victims Unit episode:

… the man we believe who attacked you

Something has gone wrong here; the question is what.

The phrase was spoken smoothly, without pauses or intonational falls. Nevertheless, it was possible that the actor was aiming for a phrase in which “we believe” is parenthetical:

… the man, we believe, who attacked you

(and this might even have been in the script). Similar sentences can be found on the web, sometimes punctuated with commas, sometimes with parentheses (dashes would be another possibility):

She also got another nasty bite in school last week, but is [?not] telling us the name of the person (we believe) who bit her. (link)

The person, we believe, who was most extensive engaged in curing [bacon] … (link)

It wouldn’t be surprising to see some web examples of this sort without the parenthetical punctuation, since informal writing often omits commas in appositives and the like (notably, in non-restrictive relative clauses); after all, the writer knows what the sentence means and how it’s supposed to sound and so doesn’t see the need for punctuation.

But not all the web examples lend themselves easily to this analysis:

Unfortunately sometimes the person we think who is right for us, sometimes isn’t. (link)

All you have to do is respond to the forum members questions and the person we think who has been the most helpful will win. (link)

There is even a variant with that:

The person that I think who made a real influence on my life is my father. (link)

The fact is that a configuration can arise in many different ways, and an analysis that’s appropriate for some people might not be appropriate for others. It’s possible, for example, that some English speakers have a (non-standard) complement construction marked by a relative pronoun (who in the examples above, but which is also possible).

Still another possibility is that some of the occurrences of the configuration here are inadvertent errors, of one sort or another. I can see at least two ways in which things like “we believe who” could arise: as a syntagmatic error or as an inadvertent blend.

One type of syntagmatic error is reversal. In the case at hand, the idea would be that a speaker or writer is planning to produce two things,

(1) a relative pronoun (who) introducing a relative clause; and

(2) a subject NP plus a complement-taking verb with that NP as subject (we believe),

with the complement clause (beginning as in (2)) serving as body of the relative clause in (1). The expected order of (1) and (2) is then (1) before (2), but it’s possible to reverse the sequence inadvertently. And such a reversal might be encouraged by the fact that (2) is much more substantial in content than (1).

Another route to we believe who is via a competition between two formulations of the relative clause:

(3) … the man we believe attacked you; and
(4) … the man who attacked you.

Such competitions sometimes result in a production that combines the (non-shared) pieces of both productions; that’s another route to we believe who.

Of course, an inadvertent error could serve as a model for other speakers and writers, who take the resulting configuration to be just one of the allowable variants in the language.

In general, it’s hard to say what’s going on in a single isolated occurence. There’s a sense in which single occurrences are inscrutable, and can be interpreted only in the context of a wider body of evidence. In the case at hand, I don’t have that evidence, but the phenomena would be worth further investigation.

This entry was posted on October 14, 2009 at 3:08 pm and is filed under Errors. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.