February 09, 2007

"Based on"

During the question period after last night's "Abusage and usage" panel at the Philadelphia Free Library, David Cuff expressed concern about sentences like "People shop for a product based on price". According to him, that sentence is confusing, because readers will mistakenly take it to mean "people shop for a product that is based on price", rather than "when people shop for a product, they compare prices". He suggested that in this case we should substitute "People shop for a product on the basis of price".

That example may be unfair, since it's not the example that he gave, but one that I just found on the internet. So here are two longer examples from an email that Prof. Cuff sent me earlier today, with his comments.

  • A story in Time reported " …the Association last year adopted a policy allowing pharmacists to refuse to fill prescriptions based on ethical, moral or religious grounds". I think we all would suspect such prescriptions.
  • Sen. Bill Frist, after viewing a video of Terri Schiavo, presumed to disagree with the attending doctor’s diagnosis. Paul Krugman wrote, " …I think the American Medical Association disapproves of politicians who second-guess medical diagnoses based on video images…". The reader can be forgiven for thinking the original diagnosis was based on video images.

It's easier to see the ambiguity in his examples than in mine, though I don't think that I would have noticed a problem in the normal course of reading.

But let's step back and look at this in a wider perspective. The general case is a structure of the form Verb Object X, where X might be a participial phrase (like "based on price", "covered with snow" or "owing to her owner's ill health"), a prepositional phrase (like "on the basis of price", "for the Iraq war" or "with a brass key"), or various other things.

(The issues are the same when the verb is replaced by an adjective or noun that takes a complement, as in "the choice of a product based on price", or when the intervening noun comes from a prepositional phrase, e.g. "EMI will decide whether to forge ahead with the strategy based on the size of the offers.")

There's a general problem here for readers -- and therefore for writers. The phase labelled X can have many different functions, and there's no general structural way to distinguish among them.

Thus a gerund-participial clause might be a predicative adjunct modifying the subject:

He answered the phone crying bitterly.

Alternatively, it might be an adverbial adjunct modifying the verb:

We played the game according to the rules.

Or it might be a modifier of the object:

They elected a man lacking integrity.

The same set of options exist for phrases heading by past participles:

He answered the phone overcome with remorse.
We played the game based on the instructions we were given.
They elected a man spoiled by privilege.

And similarly for prepositional phrases:

He answered the phone in a foul mood.
We played the game by the rules.
They elected a man from a wealthy family.

Now, these structural alternatives can certainly be confusing. Sometimes, as in the examples I've just given, the choice of words determines one plausible interpretation to the effective exclusion of the rest. But in other cases, the sentence is amusing, if not confusing -- consider Groucho Marx's line

I shot an elephant in my pajamas. What he was doing in my pajamas, I'll never know.

For better or for worse, there's no systematic cure for this kind of ambiguity -- the English language is just like that. And it's been like that for many centuries -- we could find examples from Shakespeare for all of the alternative analytic categories listed above.

OK, back to "based on".

Prof. Cuff's suggestion, as I understand it, goes like this. When X in the frame Verb Object X has the particular form "based on ...", it's usually a modifier of the object (or, perhaps, this was true at some point in the past). But when X has the form "on the basis of ...", it's usually a modifier of the verb. So we should eliminate a source of ambiguity and confusion by establishing a stylistic rule, turning those statistical tendencies into categorical preferences.

(It's possible that Prof. Cuff thinks that English grammar actually requires these choices -- but that would be wrong, and so I'm going to attribute to him a position that's both consistent with the facts and logically coherent.)

The first difficulty with this argument is that it's hard to make people obey such stylistic rules. In this case, it would be impossible, in my opinion, even if all the nation's linguists were to agree that there's really a problem here that needs to be solved, and that Prof. Cuff's proposed solution would be an improvement. As a result, the whole discussion is essentially an academic one -- though at the end of the post, I'll discuss what we can do to address Prof. Cuff's valid concern for clear writing.

A second difficulty is that "based on..." adverbials occur in many structures where something other than a noun phrase precedes. A small sample from today's New York Times:

Based on the contents of his e-mail account, Jeremy was charged with an extra count of possession of child pornography.
Soens contends the lawsuits should be dismissed based on an Iowa law that shields school officials from liability five years after a student leaves the school.
Deadlines vary based on individual plans submitted by the mining companies.
I don't think it's appropriate, based on the content of the film, to be screening it at this time.

Should all adverbial uses of "based on" be banned, or just those that follow a noun and therefore create the potential ambiguity that worries Prof. Cuff? I can't claim to have exact figures, but based on reading through the 50 examples of "based on" in today's New York Times, I'll guess that less than 20% of the uses of adverbial "based on..." are post-nominal. You could substitute "given" or "on the basis of" or other near-synonms for the other 80% -- but why?

A third difficulty is the shakiness of the claim that when post-nominal "based on..." is used as an adverbial adjunct rather than as a nominal modifier, this inevitably (or even usually) causes confusion. The three examples that I found in today's New York Times all seem perfectly clear to me, in context:

The Pentagon informed Pelosi's staff this week that she would get a plane, based on availability, and that nonstop service could not be guaranteed. (link)
One of the unidentified people said EMI would decide whether to forge ahead with the strategy based on the size of the offers. (link)
Revenues from sales of super-premium vodka jumped the most -- more than 43 percent -- a figure sure to catch the attention of distillers because vodka, based on sales, is the most popular spirit in the country. (link)

A fourth difficulty -- but one that points the way to the solution -- is that even the nominal-modifier use of "based on..." is potentially ambiguous, as in this sentence from today's New York Times:

But “The Other Here,” performed at Japan Society on Wednesday, never manages to attain the magic of “Shunkin,” an older work by the company based on the writings of another Japanese writer, Jun’ichiro Tanizaki. (link)

Here "based on" needs to skip over the immediately preceding "company" so as to reach its target, the "older work".

Though this sentence is no jewel, it doesn't seem perniciously ambiguous to me. I wouldn't have noticed it if I weren't scanning every instance of "based on" in today's online edition -- but it's surely no better than the earlier sentence about EMI's deciding whether to forge ahead based on the size of the offers.

The fact is, the ambiguous affinities of English words and phrases are always ready to make trouble for writers and readers. There's no magic way to avoid this. You can't eliminate the potential for structural ambiguity by inventing stylistic rules for individual English words and phrases. You need to look at your writing from the reader's point of view, and fix any constructions that are likely to be confusing. There are often good reasons to create and to follow stylistic rules, but in the end, the only way to ensure clear writing is to write clearly.

[Ben Yagoda put it more succinctly, in a response to Prof. Cuff's email (which included a much longer list of allegedly problematic examples):

Based on my reading of the examples, I conclude that the use of "based on" to mean "on the basis of" is harmless, in and of itself, and has a certain degree of economy and forcefulness. However, like many locutions, it can be used ineptly by poor writers and lead to ambiguity or confusion in certain sentence orderings.


Posted by Mark Liberman at February 9, 2007 09:15 PM