Data points: referent finding 8/20/10

From the NYT Science Times of August 17 (John Markoff, “Step 1: Post Elusive Proof. Step 2: Watch Fireworks.”):

“The difference between the alchemists and the chemists was that the printing press was used to coordinate peer review,” [NYU professor Clay Shirky] said. “The printing press didn’t cause the scientific revolution, but it wouldn’t have been possible without it.”

Among the reader’s tasks is to find the referents for the two it‘s in

(1) … it wouldn’t have been possible without it

There are two candidates in the text for each it: the printing press and the scientific revolution.

The printing press is (locally) highly topical in the discourse, and highly topical referents are especially available for interpreting anaphoric pronouns. And the printing press is the subject of the first coordinated clause, so that’s it’s structurally parallel to the subject it in the second coordinated clause, and therefore a strong candidate for the antecedent of it. That is, two considerations favor the printing press as the referent of that subject it (the scientific revolution would then be the referent of the it that’s the object of without).

But the scientific revolution is closer to the subject it than the printing press is, so that the scientific revolution is mentioned more recently than the printing press, which might predict it should be preferred as the referent of the subject it (with the printing press as the referent of the object it).

However, neither structural parallelism nor nearness/recency is a generally reliable guide in referent-finding (and in any case, they often run counter to one another). Consider this excerpt from what I just wrote:

(2) the scientific revolution is mentioned more recently than the printing press, which might predict it should be preferred as the referent of the subject it

Again, two possible referents — for the it that’s the subject of the object clause “it should be preferred as …”: the scientific revolution (by structural parallelism) and the printing press (by nearness/recency). In (1), nearness seems to prevail; in (2), parallelism.

In fact, considerations of discourse organization and real-world plausibility prevail over these structural considerations, as I’ve maintained several times on Language Log.

In fact, I have a large collection of examples where there are two (or more) candidates for the referent of an anaphoric pronoun. Some of these are genuinely problematic, in that a cooperative reader might reasonably be unsure, even given the context, which referent was intende; but most are innocuous in context. Nevertheless, such examples (and examples of potential attachment ambiguities) are a staple for people who have become so sensitive to certain structures that they can’t help seeing potential ambiguities everywhere and condemning writing that contains them, as here:

On the website of the Boston Globe, Eric Funston found a blurb for an article on Emily Dickinson: “A book reveals the poet began to romance her father’s best friend following his death.” We knew that Emily Dickinson had her quirks, but post-mortem romance? (World Wide Words #692 6/26/10; Funston goes for the nearest candidate, but surely only someone who has taken the Nearest Rule fully to heart would do this)

Someone who has taken the Nearest Rule fully to heart and is also looking at the example out of context (and without any guidance from real-world plausibility). (The father’s best friend is in some sense the nearest available referent, but her father’s in her father’s best friend is structurally parallel to his in his death, so her father ought to be a candidate for the referent of his on those grounds.)

3 Responses to “Data points: referent finding 8/20/10”

  1. Rick Barr Says:

    I have taken your comments on cooperative readers to heart, and have used that notion to resolve potential ambiguities that before would’ve bugged me on principle. I think you’re right in that real-world considerations trump the myriad of possibilities grammar could offer in a given sentence.

    Still, some ambiguities still crop up that leave the reader, as you say, reasonably unsure as to the referent. The most recent example I could think of was a sentence from ZZ Packer’s story “Dayward,” published in the June 14 & 21, 2010, number of The New Yorker (here).

    Allow me to quote from the entire paragraph, as to not fault lack of context for the possible ambiguity (“They” in the first sentence means two runaway slaves, Lazarus and Mary Celeste): “They’d never met their Aunt Minnie, but she was nonetheless their last living kin on earth. And yet when his mother had told stories about Aunt Minnie Lazarus had only half listened to the reminiscences, which seemed to have no real beginning or end […]. It was always something about how Minnie could sew the best, or how she had a piece of mirrored silver that she wouldn’t let anyone else use. How she had loved her sister more than anything, but had stolen away away at the age of fifteen, wearing every dress their mistress owned” (p. 111).

    This last sentence I had to read twice or thrice. The pronouns in the previous sentence (“she had a piece […],” for instance) refer to Aunt Minnie, not to Lazarus’s mother. By starting with “How,” as the latter half of the previous sentence did when the referent was Aunt Minnie, we are induced to see Minnie as the “she” in the final sentence. But we could also read the whole sentence as referring to Lazarus’s mother, since there is no helpful indicator of identity in the final stretch of the paragraph. So who exactly left? (Aunt Minny, but the answer is supplied by bringing together disperse elements of the story.) A clumsy setup, in my opinion, and I would’ve strongly suggested recasting it.

  2. David Says:

    “A book reveals the poet began to romance her father’s best friend following his death.”

    Isn’t there a rule that states a possessive can’t act as a pronominal antecedent?

    • arnold zwicky Says:

      To David: There’s a fictitious rule I’ve called the Possessive Antecedent Proscription (PAP), discussed here with links to other postings, in particular a detailed handout of mine. I was hoping not to have to return to the PAP here, but I see that that is not to be.

      In any case, I didn’t entertain a PAP account, because the PAP is crap. In fact, it’s far from clear that the PAP even applies to this example, since the anaphoric pronoun is itself possessive, and many, maybe even most, attempts to formulate the PAP carefully exempt possessive anaphoric pronouns from the scope of the putative rule (the reasoning is that possessive nouns and possessive pronouns are both adjectives, and it’s ok for an adjective to be an antecedent for another adjective; I’m


      this, not espousing it.)

Leave a Reply

%d bloggers like this: