Q: I’m aware of the current usage rules for “which” and “that,” but reading older literature suggests that this was not always the case. In fact, there doesn’t even seem to be a consensus now. Can you enlighten me?
A: We’ve written previous posts, most recently in 2008, about the modern American use of “that” and “which.” Our views largely reflect those in popular usage guides like The Chicago Manual of Style, Garner’s Modern American Usage, The New York Times Manual of Style and Usage, and Pat’s Woe Is I.
You should be aware, however, that other language authorities have legitimate differences of opinion here. And we believe there are times when writers are justified in putting euphony ahead of convention, as long as there’s no possibility of a misunderstanding.
We’ll get to the opposing arguments later, but first let’s discuss the established convention for the use of “that” and “which” in educated American speech and writing.
As we explain in our blog post, when you have a clause (a group of words with a subject and a verb) and you could start the clause with either “that” or “which,” here’s how to choose between the two:
If the information is essential and defines what came before it, use “that.” If the information is not essential and merely adds to what came before, use “which” and set the clause off with commas.
Information that’s defining is called “restrictive” and is introduced with “that.” Information that’s nondefining (it’s like a parenthetical aside) is “nonrestrictive” and is introduced with “which.”
We’ll use two examples from Woe Is I (note the underlined clauses):
Restrictive: “The dog that won best in show was Buster’s bulldog.”
Nonrestrictive: “Buster’s bulldog, which had one white ear, won best in show.”
Again, reasonable people have legitimate differences of opinion here. Scholarly works on grammar do not recognize any “rule” that would limit “which” to nonrestrictive clauses. Neither do standard dictionaries or Merriam-Webster’s Dictionary of English Usage.
But we believe, as we said up above, that most educated Americans are with us in using “that” for restrictive clauses and “which” for nonrestrictive ones.
So much for the modern American use of “that” and “which.” The history of their usage as relative pronouns is a much longer story.
In the beginning, “that” was our only relative pronoun. It was used to introduce both kinds of clauses – restrictive (defining) as well as nonrestrictive (nondefining). Both uses were recorded in writing as far back as the 800s.
“Which” was around, but it was used for other purposes – largely in questions, as an adjective, or in the sense of “what” or “who.” It wasn’t used as a relative pronoun, according to Merriam-Webster’s Dictionary of English Usage, until the 14th century. And “which,” like “that,” was used to introduce both kinds of clauses.
Merriam-Webster’s explains that “by the early 17th century, which and that were being used pretty much interchangeably.”
But in the later 17th century, literary writers simply stopped using “that” as a relative pronoun and used “which” exclusively. Perhaps “that” was considered less scholarly or erudite.
At any rate, when “that” reappeared in the early 18th century it was used mainly in restrictive (defining) clauses, though “which” wasn’t limited to one or the other.
By the early 20th century, however, usage commentators had begun to look askance at the all-purpose “which.”
Their thinking, according to Merriam-Webster’s, was, “If that was being confined to introducing restrictive clauses, might it not be useful (as well as symmetrical) to confine which to nonrestrictive clauses?”
One of the most influential cheerleaders for this idea was Henry Fowler, the author of A Dictionary of Modern English Usage (1926). He flatly stated, “which is appropriate to non-defining and that to defining clauses.”
Fowler argued for a “restoration of that to the place from which, in print, it tends to be ousted. … If writers would agree to regard that as the defining relative pronoun, & which as the non-defining, there would be much gain both in lucidity & in ease.”
Fowler also indicated that British writers were more likely to need this advice than Americans. And he said it was a “false inference” to regard “that” as a colloquial usage and “which” as a literary one.
The mistaken belief that “which” is more literary than “that” is also mentioned by Otto Jespersen in his Essentials of English Grammar.
Jespersen says that ever since “which, whom, and who came into use as relative pronouns,” they’ve been “gaining ground at the expense of that, chiefly in the last few centuries and in the more pretentious kinds of literature.”
“One of the reasons for this preference,” he says, “was probably that these pronouns reminded classical scholars of the corresponding Latin pronouns.”
Despite Fowler’s advice in his influential usage guide, it’s not clear whether much has changed in the last century.
In American usage, writers tend to use “that” in restrictive (or defining) clauses and “which” in nonrestrictive (or nondefining) clauses. But British writers often use “which” freely for both kinds of clauses, as they did when Fowler wrote his usage guide.
As the Oxford English Dictionary says, “which” is used to introduce both (1) “a clause defining or restricting the antecedent and thus completing the sense,” and (2) “an additional statement about the antecedent, the sense of the principal clause being complete without the relative clause.”
If writers use “which” both ways, how can we tell what they mean? The OED says that in modern printing, the use of “which” in restrictive clauses is usually distinguished by the lack of a preceding comma, as it is in speech by the absence of a pause.
Fowler agreed that when writers use “which” for both kinds of clauses, “it is important to have another means of distinguishing. A comma preceding which shows that the which-clause is non-defining, & the absence of such a comma shows that it is defining.” But he added: “That right interpretation should depend on a mere comma is a pity.”
So as things stand, writers who persist in using “which” in all cases at least have a way to make their meaning clearer. We’ll invent a couple of examples, both of which seem clear enough:
Restrictive: “Sue threw away the clothes which were outdated.” (She dumped only the outdated clothes.)
Nonrestrictive: “Sue threw away the clothes, which were outdated.” (She dumped all the clothes.)
We also mentioned euphony above. If a sentence already contains a nearby “that,” adding one to introduce a restrictive clause can be clunky. A writer who cares how a sentence sounds would be justified in substituting “which,” as in “It’s clear that the shirt which is stained should be bleached.”
There are other kinds of euphony too. One of our readers sent us an example from G. K. Chesterton:
“We call wine ‘white wine’ which is as yellow as a Blue-coat boy’s legs. We call grapes ‘white grapes’ which are manifestly pale green.” In such a passage, “that” simply wouldn’t work. (Of course, a writer today would use a simpler and more natural style.)
On the whole, we believe that the conventional distinction between “that” and “which” can be an aid to clarity. If a “which” is ambiguous, why not use “that”? The presence or absence of a comma isn’t always enough.
But perhaps the best argument of all is “that” seems more natural and idiomatic in restrictive clauses, especially in speech.
[Note: This post was updated on Dec. 27, 2022.]