Part 1:  Back in my posting on “Words to eliminate”, I looked at a site that proposed to get you to improve your writing by eliminating 15 words from it. (Yes, a silly idea.) One of these was that:

[Mashable advice] It’s superfluous most of the time. Open any document you’ve got drafted on your desktop, and find a sentence with “that” in it. Read it out loud. Now read it again without “that.” If the sentence works without it, delete it.

The idea is fraught with problems, most turning on the fact that there are several distinct lexical items that, with a large number of uses, and with distinct syntax, discourse functions, and sociolinguistic statuses for each use.

Part 2: On one of these items, the complementizer that, and its use to mark the object complement of a verb, as in

They know (that) pigs can’t fly.

(where the that variant and the ∅ variant are both fine).

But then I started an e-mail to a friend:

 I do wish people would credit sources.

(with the ∅ variant; the that variant is also possible) and thought to link to previous context with a though — but then the ∅ variant struck me as very awkward indeed:

?? I do wish, though, people would credit sources.

though the that variant is fine:

 I do wish, though, that people would credit sources.

What’s crucial is that material intervenes between the complement-taking verb and the complement. It turns out that this intervention effect is well-known in the variation literature.

Part 1: lots of that. There are at least three lexical items that: complementizer that (above and in subject complements, as in That pigs can’t fly distresses me, and in complements to nouns, as in the idea that pigs can’t fly), relativizer that (an idea (that) I can’t accept) and demonstrative that (that idea; That distresses me). Plus some further types: for example, so (that) in my previous posting; in order that; and a degree modifier, as in It’s not that important.

There are, in fact, so many different uses that I could easily give a whole graduate seminar on them.

Omissibility differs from use to use in very complex ways.

Demonstrative that is is not normally omissible, and when it is, omission doesn’t preserve meaning (something the Mashable advice doesn’t take into account): That money is tainted and Money is tainted are not semantically equivalent, so you shouldn’t omit that in this case.

Part 2: intervention and its discontents. I took the intervention example above to my colleague Tom Wasow, who has made something of a career studying the omissibility of relativizer that, but knows a lot about complementizer that as well. As luck would have it, Tom is teaching a seminar this quarter (Seminar in Sociolinguistics: Sociogrammar) with a session on optional complementizers, and Tom was not only able to provide me with some bibliography on the subject but could e-mail me pdf files of four relevant articles, the most important of which is:

[T&S] “No momentary fancy! The zero ‘complementizer’ in English dialects” by Sali Tagliamonte and Jennifer Smith, English Language and Linguistics 9.2. 289-309 (2005)

(Ah, the kindness of colleagues!)

T&S explore a large set of factors (both linguistic and extra-linguistic) that might contribute to omissibility and use Varbrul software to assess their contributions to the use of the  variant. The article, drawing on the existing literature and adding a substantial amount of new data from working-class Engish dialects, is rich and thought-provoking.

T&S report a significant contribution from speech vs. writing (more ∅ in speech), from the specific complement-taking verb, and from the subject of this verb. These factors combine to make think with 1st-person subject virtually categorical for ∅ in speech; I think then looks more like an adverbial than like a complement-taking verb.

One of the significant factors favoring ∅, in this study and in others, is, yes, material intervening between the verb and its complement, as in the dubious example above. The effect is statistical, not absolute, but in the example above, it’s striking; you really need the that to easily recognize what follows this material as the verb’s complement.

One Response to “Intervention”

  1. arnold zwicky Says:

    From Mike Pope on Facebook:

    In our technical-writing work, we’ve been exhorted to include “that” at all places where it would otherwise be elided. This is for a pragmatic reason: to reduce ambiguity in text that will be preprocessed for localization by running it through machine translation. As an editor, I’ve certainly had my share of arguments with writers who felt that the inclusion of explicit “that” created awkward prose, and they’re not wrong. But it was editorial policy that was even policed by grammar-checking tools.

    Well, this makes real sense in the context. Special circumstances make for special rules.

Leave a Reply

%d bloggers like this: