which vs that

Author's note (2020): This was an early post on this blog, before I learned much of what I've since learned about AmE and BrE. Thanks to those who've pointed out its problems. I've corrected it a little below. I'm happy to say the issue is treated much better in The Prodigal Tongue. I'll put a screenshot of it here (you'll probably have to enlarge it on your screen), but if you'd like the full benefit, I suggest the book!






David in Dublin emailed about the relative pronouns which and that, saying:

In American English there seems to be a strong distinction, particularly among educated speakers/writers. I'm fairly sure this distinction doesn't exist in the dialect that I speak. However, as the American standard usage is a subset of the local standard usage, it seems that the American version is used by most writers addressing both audiences (say academic publications, software documentation, ...) because American readers will assume you to be uneducated if you misuse them! I have tried to determine if I am really uneducated or if this is a real Am/Br distinction by looking up "which" and "that" in the OED. My reading of the ODE seems to support it being a distinction, but I may be deceiving myself!
You're right, David, non-American Englishes have mostly lost  American English is most particular about a distinction between which and that in restrictive relative clauses. But, as you've noticed, its persistence in American English is limited to certain types of people/discourses. The distinction is far more likely to be observed in writing, especially academic and copy-edited publications. I must admit that I have the distinction, even usually in speech, but it's something that (which!) I acquired as a doctoral student. If you have acquired the distinction, it can interfere with your ability to process writing that doesn't have the distinction--I'll explain why after explaining the distinction in a bit more detail. A relative clause (RC) is a clause (i.e. a sentence within another sentence/phrase) that's used to modify a noun. Relative clauses usually start with a relative pronoun: that, which or who/whom (and sometimes some others that aren't relevant here). There are two types of relative clause. Restrictive RCs reduce the range of things that the modified noun refers to. So, in sentence (1), I use who lives upstairs to indicate that living upstairs is the property that distinguishes this man from other men I could have talked about.
(1) The man who lives upstairs has a new piano.
In contrast, non-restrictive RCs aren't used to identify who/what the noun refers to, but to give more information about the referent of the noun, as in (2). Notice that using a non-restrictive RC is a way to fit more information about one noun's referent into a single sentence.
(2) [pointing to man] That man, who lives upstairs, has a new piano. (= That man has a new piano. He lives upstairs.)
As these examples show, who can introduce either type of clause. In speech we can tell the difference between them because restrictive and nonrestrictive RCs are spoken with different prosody (=speech melody, intonation). In writing, the non-restrictive type is correctly set off by commas. In any English dialect, that can only introduce restrictive relative clauses. In other words, non-restrictive RCs must start with which or who/whom. So all English speakers have a distinction between which and that to that extent. BrE and most other Englishes don't have a strong distinction in the restrictive RC:
(* marks grammatical impossibilities, £ marks stylistic variation common in BrE.) (3) That dress, which [*that] changed my life, is red. [non-restrictive] (4) The dress that [£ which] changed my life is red. [restrictive]
Anyone who has used Microsoft Word's grammar checker will know that if you use a restrictive which, as in (4), it highlights it. That's because it seeks out which relative clauses without commas around them in order to query whether you should have commas. Since that depends on your meaning, the grammar checker has to ask you--it can't tell from the grammar. However, if you use that, it knows that your lack of commas was purposeful. Although I learned the that/which distinction late, I have it very strongly now—probably because I've worked as a copy editor. I frequently misread relative clauses that have which where a that could be--thinking that they are non-restrictive, even though the comma rule for restrictive/non-restrictive RCs should prevent misreading. I then must go back to re-read and re-evaluate my interpretation when I reali{s/z}e that the non-restrictive interpretation doesn't make sense. This confusion almost always happens when I'm reading student writing, so it could be that I've learned to ignore punctuation since it's often fairly random. At any rate, I do appreciate it when people use the distinction in writing--and I do wonder why my BrE colleagues don't seem to have the same disambiguation problem when it comes to reading restrictive whiches. When I first started reading student work in the UK, I was also struck by some students' apparent comfort in writing the man that lives upstairs, rather than the man who lives upstairs. It seemed like I was 'correcting' that far more often here than I had in the US. However, I've not seen anyone else note it as a dialectal distinction and I've noticed it less and less, so perhaps I just had some odd students my first year here. What do the style authorities say about all this? On restrictive that/which, the 3rd edition of Fowler's Modern Usage, a British guide (though the current editor is a New Zealander), quotes the 1926 edition in saying: "Some there are who follow this [distinguishing] principle now; but it would be idle to pretend that it is the practice either of most or of the best writers." Larry Trask (an American acting as an authority to a British audience!) in Mind the Gaffe doesn't mind whether or not you distinguish that/which in restrictive RCs. For American audiences, the Modern Language Association Handbook (5th edn), the style book for many American academics, simply says, "Note that some writers prefer to use which to introduce nonrestrictive clauses and that to introduce restrictive clauses." The Chicago Manual of Style (14th edn--my AmE books are a little out of date) warns that "Although the distinction is often disregarded in contemporary writing, the careful writer and editor should bear in mind that such indifference may result in misreading or uncertainty. [...] When the commas intended to set off a nonrestrictive clause are omitted, perhaps with the intention of using which restrictively, the reader may well ponder whether the omission was inadvertent" (exactly my problem). All of the guides recommend that you use who instead of that when referring to people. A last point to make is that the American prescriptivist preference for that in restrictive RCs is undone by the other prescriptivist rule that clauses shouldn't end in a preposition. If you want your preposition at the front of the RC with the relative pronoun, then that pronoun cannot be that:
(5) the building that/which I ran into (6) the building into which I ran (7) *the building into that I ran


  1. lynneguist,
    Love your column! I really wish I had a proof reader of your caliber to ensure that my movie reviews meet the British English standards of proper grammar. If I fail my own country people in this language, I feel less disappointed. I find it incredibly difficult to proof my own work. Meanwhile, writing so much every single day and self-publishing is a challenge. Anyway, just wanted to let you know I appreciated your visit to movieEVERYday.com today. Hope you continue to enjoy.



  2. While you were leaving your comment, Scooter, I was re-posting the post for the umpteenth time after discovering yet another typo and yet another difficult-to-parse sentence. It's really hard to proof(-)read your own work and see the errors, but it's so easy to see them in other people's work.

    I wouldn't worry about making your movie (BrE prefers film) reviews fit a British English standard. You're expressing yourself, and if you're American, that's how most people would want you to express yourself. (And believe me, BrE standards of proper grammar can be pretty lax!!)

  3. Thanks to my niece, who works for UC Press, I have the new Chicago Manual of Style.

    Section 5.202 says:

    That; which. These are both relative pronouns. In polished American prose, that is used restrictively to narrow a category or identify a particular item being talked about [any building that is taller must be outside the state]; which is used nonrestrictively—not to narrow a class or identify a particular item but to add something about an item already identified [alongside the officer trotted a toy poodle, which is hardly a typical police dog]. Which should be used restrictively only when it is preceded by a preposition [the situation in which we find ourselves]. Otherwise it is almost always preceded by a comma, a parenthesis, or a dash. In British English, writers and editors seldom observe the distinction between the two words.


    You're right though, the only time I've ever seen anyone make a fuss over it was when my husband's doctoral adviser changed his whiches to thats. Or vice versa.

  4. Hmm...so Chicago has become MORE prescriptivist about which/that. Very interesting! Thanks for all that typing, Kathy!

    I do wonder whether Microsoft Word has inadvertently raised consciousness of the distinction in American writing culture.

    Americans seem to overtly care about prescriptivist rules more than some other cultures. Take, for example, the Usage Panel of the American Heritage Dictionary, which votes on which usages of words are acceptable in formal (particularly written) American English. It may be the closest thing English has to a national language academy, and it comes from the US, not the UK. Some take this to be a symptom of American "linguistic insecurity". That is, we're never sure if we're speaking English the "right" way, and look to outside authorities --the British, dictionaries, teachers-- to tell us.

  5. I think Americans in general are more fond of rules, or perhaps it's that they're less fond of gray/grey areas than the British.

    My favorite example is "Keep Your Dog on a Lead" signs, which the British habitually ignore. I'd gone to Regent's Park several times before I got up the nerve to take the lead off my dog, despite the fact every other dog was running free.

  6. There's probably something to that, Kathy, but British queu(e)ing ('waiting in line') behavio(u)r is one of the most adamantly rule-based things one can imagine.

  7. This artificial rule is gaining ground in the US, unfortunately. I'm not going to type in a huge amoung of comment, but I will refer people who are interested in the subject to the Language Log (particularly this one by Geoff Pullum http://itre.cis.upenn.edu/~myl/languagelog/archives/001464.html which has numbers!, and this rather long summary post by Arnold Zwicky http://itre.cis.upenn.edu/~myl/languagelog/archives/002146.html

  8. I'm Irish, and I could easily imagine myself saying "That dress that changed my life" ... but there would be an implication of that I already told you about in there, e.g "and that dress that changed my life is still hanging in my wardrobe all these years later".

  9. The implication of 'that I already told you about' comes from the determiner that (the first that), not the complementizer that (the one that could be a which. It's confusing to have an example with two thats in it!

  10. The that/which distinction and the preservation of the subjunctive are two examples of (standard) American English being more conservative than (standard) British English.

    Quite right about the subjunctive, quite wrong about the which/that distinction: it was invented out of whole cloth by a Brit (H.W. Fowler) and became popular in America through the efforts of writers on usage, who spread it to copy editors and editors, where it afflicts the remaining Americans who (quite properly) make no such distinction.

  11. Depends on what you mean by 'conservative' in this context--and I admit I did a bit of messing around on that. Americans do like to have rules to follow, and so in this case you'd have to take 'conservative' to mean 'prescriptive', which, ok, isn't a good way to take it!

  12. The Guardian's style manual says this about that:

    that defines, which informs:
    this is the house that Jack built, but
    this house, which Jack built, is now falling down;
    the Guardian, which I read every day, is the paper that I admire above all others, but
    the Guardian, which I admire above all others, is the paper that I read every day.

    An American Guardian reader.

  13. 12 years ago, my school English teacher in India was quite picky about not using "which" in restrictive adjective clauses. And I wouldn't be surprised if Indian teachers continue to be particular about this point — English education in India is quite conservative.

  14. Great column! Thanks a lot for clarifying that. I'm sure I'll notice it everywhere now...

  15. Hi there! I just discovered your blog, and I'm working my way through every single entry. It is fascinating -- and should be required reading for any American moving to England. I don't know if you will see this comment as it is on such an old post, but I just had to say that, as an American working as a content and copy editor on a British website, I make a point to correct every instance of which that should be a that! My English colleague thinks I'm mad :-)

  16. I do the same thing, Alexandria. You're not alone!

    (And I do see the comments on old posts!)

  17. IMO using or omitting the commas is non-negotiable, but using "which" or "that" is a stylistic or dialectal choice.

    The problem comes when people who don't really understand the difference between the two types of clause blindly follow MS Word's "corrections". You get sentences like "The article contained words which I found offensive" (fine in BrE, stylistically dubious in AmE) getting "corrected" to "The article contained words, which I found offensive" (nonsense in both dialects).

  18. Charles E. Miller, Jr. BA, MA
    At Old Dominion University in Norfolk,Virginia, I was taught that using "which vs "which" instead of "that" vs "which" was acceptable. I have always used it until I discovered the "that" vs "which" on a computer grammar check and contacted a Professor Ellis of Tidewater Community College who has a PhD in English. He agrees with the "that" vs "which." I suppose I shall change my habits.
    Charles Miller, Former Officer, Bank of America

  19. I once tried to read a book on language by linguist David Crystal, but had to give up after about 50 pages. This "expert" consistently used "which" instead of "that" in restrictive clauses and I found myself needing to circle each occurrence in pencil as I read! I lost the meaning of his writing because of this "which-hunt"!

  20. Anonymous

    This "expert" consistently used "which" instead of "that" in restrictive clauses

    I imagine that all British experts use which in both sorts of relative clauses. I don't think I'm alone in finding the American shibboleth bizarre and incomprehensible. I can believe that it used to be proscribed in British English, but I find it very hard to believe that the prescription was followed to any extent.

    I think this is one of those rues dreamed up by tidy-minded people who can't bear asymmetry. It's perfectly true that that is not standard in nonrestrictive clauses. But that's no reason to assume the corollary: that which not standard in restrictive clauses. It's like the rule that insists on the corollary of the rule that fewer is not standard with uncountable nouns. The symmetrical assertion is that less is not standard with countable nouns.

    In Britain more people pay lip-service to the fewer only + COUNTABLE prescription than to the that only + RESTRICTIVE prescription. But that only means that we apologise when we say less people. (Well, I don't apologise, but some do.) We don't apologise she we say the dress which changed my life — probably because few people alive have even heard of the prescription.

    Or was your posting just a joke?

  21. I realize that this is an old post, but like others I'm strolling through this fantastic blog after discovering it entirely by accident -- please bear with my enthusiasm.

    I'm an American, an editor, and an employee of a multinational, all of which means I work with a lot of different Englishes. But because the editorial team I work for generally edits to US standards, Garner's Modern American Usage* is my go-to style guide.

    Garner is usually an unexcitable type, starting with preface to the first edition (he's now on his third). It's a laudably emollient attempt at compromise between descriptivism and prescriptivism. I don't think it really worked, but still, I respect him for trying.

    But get him started on thats and whiches....and he'll keep going for two large, densely printed pages, including this not-so-emollient barb:

    "British writers have utterly bollixed the distinction between restrictive and nonrestrictive relative pronouns. Most commonly which encroaches on that's territory, but sometimes too a nonrestrictive which remains unpunctuated."

    And, contra John Cowan above, he argues for half a page -- including citations from seven sources going back to 1806 -- that the Fowler brothers did not "invent" the distinction. He concludes:

    "The only retrospective blame that might lie with the Fowler brothers is that they pressed their point too diffidently."

  22. Christian

    including this not-so-emollient barb:

    "British writers have utterly bollixed the distinction between restrictive and nonrestrictive relative pronouns. Most commonly which encroaches on that's territory, but sometimes too a nonrestrictive which remains unpunctuated."

    I've never encountered bollocks as a verb before — and certainly not with the coy spelling bolix

    To Mr Ganer I would point put that you annoy mess up — politely or otherwise — a dogma that/which you believe to be a nonsense.

  23. The great Comprehensive Grammar of the English Language (Quirk, Greenbaum, Leech & Svartvik) states unequivocally that the choice of restrictive pronouns is the following:

    personal..............who, that
    non-personal.......which, that

    personal..............whom, that, zero
    non-personal.......which, that, zero





  24. A Comprehensive Grammar of the English Language was published in 1985. The next blockbuster grammar was The Cambridge Grammar of the English Language (eds Huddleston & Pullum) in 2002.

    They don't treat that as formally a relative pronoun (although,of course, they acknowledge its use in relative clauses).

    They group the wh- words as

    Genitive.......... —

    Genitive.......... —

    Genitive.......... whose

    That is to say which as a pronoun is used only with a non-personal antecedent in relative clauses which are not FUSED. We can see what they mean from this extract from a table

    .................................NON PERSONAL
    INTERROGATIVE What did he want?
    FUSED RELATIVE Take what you want
    OTHER RELATIVE the car which came first

    [Which as a non-pronoun before a noun (the modern term is determiner) is treated elsewhere.]

    They divide relative clauses in two ways
    • by formal type
    • by relational type

    The formal types are illustrated thus

    i) wh-relative
    He'll be glad to take the toys which you don't want
    ii) that-relative
    He'll be glad to take the toys that you don't want
    iii) bare relative
    He'll be glad to take the toys you don't t

    The relational types are four — two of them corresponding to restrictive and non-restrictive.

    i) integrated relative
    The boys who defaced the statue were expelled.
    ii) supplementary relative
    My father, who retired last year, now lives in Florida
    iii) cleft relative
    It was Kim who wanted Pat as treasurer.
    iv) fused relative
    What you say is quite right.

    The example of integrated (= restrictive) relative i, this table ispersonal. But they go on to give these non-personalexamples:

    i) main clause
    A letter drew our attention to the problem.
    ii) wh-relative
    This is the letter which drew our attention to the problem.
    ii) that-relative
    This is the letter that drew our attention to the problem.

    i) main clause
    My neighbour gave me some advice.
    ii) wh-relative
    I accepted the advice which my neighbour gave me.
    ii) that-relative
    I accepted the advice that my neighbour gave me.
    iv) bare relative
    I accepted the advice my neighbour gave me.

    Which — and not that — also occurs in what they call complex relative phrases. Examples

    behind which
    the result of which
    prominent among which
    to refute which
    passing which
    whose essay
    which suggests

    Some of these are presented in integrated (= restrictive)
    relative clauses

    the curtain behind which Kim was hiding
    the curtain which Kim was hiding behind

    Police are looking for a Ford Escort the licence number of which ends in 7.

    problems the answers to which he already knows
    problems to which he already knows the answers
    problems which he already knows the answers to

  25. The two massive grammar that/which I've cited were based on evidence, although they don't quote much in the way of statistics. Rather different is the ever-so-slightly smaller Longman Grammar of Spoken and Written English (Biber, Johansson, leech, Corad & Finegan) published in 1999.

    In a section headed Variations within Standard English they address the dogma full on in these words

    For example, standard English uses two relative pronouns with inanimate head nouns — that and which.
    ...........I could give you figures that would shock you (FICT)
    ..........This chapter is devoted to a discussion of various
    ..........flow processes which occur in open systems.
    In most sentences either of these two forms would be grammatical, although there are a number of contextual factors that favour the use one or the other ( the existence of a standard variety has not levelled out variability of this type. In particular, the notion that the standard insists on 'uniformity' — allowing just one variant of each grammatical feature — is a serious fallacy, arising from a misleading application to language of the notion of 'standard' taken from other walks of life.

    Those tags FICT (fiction texts) and ACAD (academic texts) refer to types of text in different data banks. The other text-types are NEWS (newspaper texts) and CONV (conversation transcripts.

    That numerical reference ( is a cross reference to their Which vs that section. Here they start by demonstrating that even the association of that with restrictive relatives is not an exclusive one.

    ... With animate heads, which is rare, while that is more common, especially in conversation. A more important difference is that which commonly occurs with non-restrictive relatives — 25 per cent to 35 per cent of the time, depending on the register. In contrast, that rarely rarely occurs with non-restrictive clauses. When that does mark a non-restrictive clause, it often occurs in a series of postmodifiers and is used for stylistic effect (especially in fiction)

    Here one might say to those sliding lights, those fumbling airs, that breathe and bend over the bed itself, here you can neither touch nor destroy. (FICT)

    He gazed at the yellow, stained wall with all the spots which dead bugs, that one had crawled, had left. (FICT)

    I am talking about an organisation that probably few of you have heard of, that can and will provide to some, perhaps to some of you, a year of travel, cultural refreshment and excitement you'll remember a long time. (FICT)

    I'll post tomorrow on what they have to say about which and that in British and American English.

  26. My word, that's a reply, @David_Crosbie! Apologies for my own late response: I've been seasonally migrating 8,000 miles, during which Delta Airlines gave me a (thankfully mild) case of food poisoning. (Sad, because it was a rare instance of airline "food" that actually met non-captive-audience standards of tastiness. Back to fullly-sterilized brown-glop-on-a-tray, I guess.)

    I suspect that Garner wouldn't argue the grammar: both "that" and "which" are intelligible, particularly if you aren't conditioned to expect "which" to appear only in non-restrictive clauses. Garner is arguing only that the distinction is worth maintaining in the context of edited written English. Because his position reflects that of many American copyeditors, it's likely to continue on this side of the Atlantic for some time, notwithstanding its divergence from other written Englishes and from less-formal US English.

    Now, on bollix -- that is indeed a standard American word. Merriam-Webster Unabridged does not even mark it informal, let alone vulgar (the tag MW gives the "buttocks" sense of "ass", which is a very mild vulgarism anymore). I think I reflect the experience of most Americans in that the only context in which I've seen or heard "bollocks" is in the title of the Sex Pistols album. Indeed, MW tags it as chiefly British, usually vulgar. That unfamiliarity, combined with the change in spelling, has successfully divorced "bollix" from its origins. I know I didn't get that there was any connection between the two until I was in my 30s.

  27. Christian

    It's not just a question of what's intelligible. What the great grammars of our time state is that defining relative clauses with non-personal antecedents using which are grammatical.

    The Longman Grammar of Splen and Written English goes further — it quantifies at least some aspects of the use of that/which in such clauses. Here's an approximation of a striking diagram illustrating the use in NEWS texts

    Each ◎ represents 200 occurrences per million words in the NEWS corpus.

    restrictive that
    AmE NEWS
    BrE NEWS

    restrictive which
    AmE NEWS
    BrE NEWS

    non-restrictive which
    AmE NEWS
    BrE NEWS

    The discrepancy between BrE and AmE use of restrictive which suggest that the American shibboleth is not entirely without effect. However, the figure of about 800 instances per million words in AmE texts suggests that the shibboleth is far from universally recognised.

    The authors point to a strong association between that and AmE relative clauses in CONVERSATION texts. That is twice as common as which in AmE, although the frequencies for that and which in CONVERSATION are the same for both dialects.

    Which is also the more formal option in other genres. In ACADEMIC writing 70 per cent of texts use which more frequently than that. In the FICTION corpus, 75% of texts use that more frequently than which.

    Their conclusion is in marked contrast to the shibboleth. They take the evidence as pointing to that as the more colloquial, less formal choice in the which/that pair. And so

    The AmE preference for that over which reflects a willingness to use a form with colloquial associations more widely in written contexts than BrE.

  28. There are other patterns of use regarding which/that in restrictive relative clauses that seem to be unaffected by any dialect difference between BrE and AmE.


    The OED observes that which in restrictive realties clauses is regularly used when the antecedent is that as a demonstrative pronoun (ðæt, not ðət).

    The Longman Grammar of Spoken and Written English has a table which/that quantifies this:

    head noun [=antecedent) is a demonstrative pronoun
    with which.......... 85%
    with that................5%
    with zero.............10%

    I find this is easier to take in if you look at their examples. (I've used BLOCK CAPITALS of the antecedent rather than their somewhat confusing (BrE) square brackets.

    • of the 'dispreferred' 5%
    .............What's THIS that I'm looking at? (FICT)
    • of the 'preferred' 85%
    .............I recognised a silence like THAT which pervades a church after a service. (FICT)T

    So, contrary to the shibboleth, AmE shares the BrE preference for that which, those which, these which etc. (The inclusion of plural pronouns is important. There could be a different explanation for not choosing that that.)


    In the same table as above, the LGSWE reveals another which/that distribution

    head noun [=antecedent) is an indefinite pronoun
    with which.......... 10%
    with that..............45%
    with zero.............45%

    • of the 'dispreferred' 10%
    .............There is SOMETHING which everybody can do to alleviate the problem. (NEWS)
    .............The local authority tnds to go in dread of ANYTHING which may scandalise the electorate. (ACAD)
    • of the 'preferred' 45% with that
    .............It's just SOMETHING that we can't do, I'm afraid. (CONV)
    .............So it wouldn't want to do ANYTHING that would stuff it out. (NEWS)
    • of the equally preferred 45% with zero
    .............I'll give you ANYTHING you want, my darling. (CONV)
    .............He has SOMETHING he wants to say to you. (FICT)
    .............There could be a call from SOMEONE you haven't heard from in some time. (NEWS)

    The Cambridge Dictionary of English Grammar by Pam Peters cites all the patterns I've quoted from the LGSWE, plus couple of others form the research that went into the grammar— again, apparently, common to BrE and AmE.


    After a superlative modifying the antecedent, that is preferred over which
    e.g. It's the smallest bike that I've ever ridden


    The preference for which increases with the distance between the antecedent and the relative.

    (She doesn't give an example, nor can I find one in the LGSWE. I could make one up, but it wouldn't carry the same weight.)

  29. I've finally got round to flowing the links given by The Ridger, FCD in this post of 6 August 2006.

    The first link leads to a modified version of the table I tried to reproduce in my post of 6 October 2014.

    What I learned from both his/her links is that there is substantial opposition among American linguists to what I've been calling 'the American shibboleth'. The difference is that we Brits just dismiss it as nonsense hardly worthy of our attention, while American opponents afford it the respect of downright hatred.

    I'm less worked up because, as I see it, here in Britain the dangerous grammatical shibboleths are those which give comfort to class prejudice — not an accusation I'd level at the that/which relatives dogma.

  30. Stephen Pinker has just published an attempt at at style guide which/that avoids the dogmatism and grammatical stupidity of guides past. He calls it The Sense of Style, The Thinking Person's Guide to Writing in the 21st Century. Not surprisingly, he takes on which/that relative clauses.

    He accepts the half of the rule that forbids that introducing nonrestrictive relative clauses. Ot rather he advises against using it unless you want to come across as two hundred years old.

    The other half, the condemnation of which introducing restrictive relative clauses he dismisses as utterly incorrect. His evidence is

    1. classic quotes such as

    Render therefore unto Caesar those things which are Caesar's
    a day which will live in infamy

    2. the exercise quoted in one of The Ridger, FCD's links. In Pinker's succinct summary

    The linguist Geoffrey Pullum searched through a sample of classic novels by authors such as Dickens, Conrad Melville and Bronté, and found that on average readers will bum into a relative clause with which by the time they are 3 percent of the way into it.

    3. inconstancy on the part of the advocates. The White of Strunk and White elsewhere writes

    The premature expiration of a pig is, I soon discovered, a departure which the community marks solemnly on its calendar.

    Pinker agrees with John Cowan's posting that the rule was dreamed up by Henry Fowler.

    If writers would agree to regard that as the defining relative pronoun, & which as the non-defining, then there would be much gain in lucidity & in ease. Some there are who follow this principle now, but it would be idle to pretend that it is the practice of either most or the best writers.

    Pinker quotes with relish the riposte of a the lexicographer Bergen Evans:

    What is not the practice of most, or of the best, is not part of our common language.

  31. Just wanted to let you know which I really appreciated this article; on when to use "which" and when to use "that" ... but wasn't quite able to fully understand it on first reading.

    As my eyes were glazing over on seeing all the rules, I thought I'll just look for an example ... but I couldn't find an example quite like the phrase I want to include in a story I am writing for a monthly contest in an online s/f writing group I belong to.

    The phrase, in my story, is "the data that could establish ...such and such" (<== I think that is the correct way to write it) OR "the data which could establish ... such and such (<== I think that is probably the incorrect way.)

  32. You can say either in British English. In American, you can say either, but the 'that' is preferred by 'authorities'.

  33. This comment has been removed by the author.

  34. I'm glad to say that Bryan Garner has deleted those insulting remarks about British writers ("bollixed") in the latest edition of his guide. I complained about it and he graciously credited me (via Twitter) with some influence in making him re-consider. I'm sure Stan Carey, Geoff Pullum and others had even more effect. The Guardian have also modified their style guide somewhat.

  35. So, a query about British writers of the late 19th century: would they make the restrictive clause/ non-restrictive clause distinction? I'm thinking about a sentence like Wilde's "the little tent of blue/Which prisoners call the sky."

  36. I would read the Wilde quote as a restrictive relative (about a very restricted space, too!) but I am not sure of 19C punctuation practice, or if it was very consistent. (More research needed.) As now, the whole area seems to be more a matter of punctuation than grammar. The writer reading the sentence aloud should always clarify it. Difficult to judge non-contemporary writers, though.

  37. Well yes, but there is a little trickiness to the line, as it can--where read non-restrictively--make it sound as though we are all prisoners in calling "sky" what is really, whatever "really" means in such circumstances, a little tent of blue in a much wider universe.

  38. Thanks, Anon. Yes, possible but there are further possibilities with poetry / verse. Although I said it is largely a matter of punctuation (and I don't know how consistent 29C writers were on that), poetry might break those rules anyway. The line break often acts as a form of punctuation, perhaps a comma or a dash.

  39. Thanks, lynneguist (fantastic name) for the thorough research and information. I would love to see your book How the Americans Saved the English Language!

  40. Thanks for the kind words, L in NL. The book will probably come out in early 2018 (possibly late 2017), but with another title, as yet to be determined! (That title--my original title--is making the marketing people nervous!)

  41. I think that yes, British writers of the late 19th century would make the restrictive/nonrestrictive distinction but they would make it with commas.

  42. British writers of the late 19th century would have had no means other than commas to mark what they probably would not have termed a non-restrictive relative clause.

    [The converse is not true; it was open them to use that to mark what they probably wouldn't have termed a restrictive relative clause.]

    As for Wilde, he precluded the non-restrictive option with his choice of the word that (not the as Anonymous thought, but the grammatical result would have been the same.)

    A non-restrictive aka non-defining aka supplementary relative clause is one that can be removed while leaving the grammar and much of the sense intact. Wilde could not have written

    But I never saw a man who looked
    So wistfully at the day.

    I never saw a man who looked
    With such a wistful eye
    Upon that little tent of blue.

    because 'that tent' can't plausibly refer to 'that day'. OK, he could grammatically say 'that little tent of blue' while pointing at the sky. But a reader can't see when a writer points.

    So the only grammatical justification for that is as part of the wording

    that little tent of blue which prisoners call the sky.

    Even if Wilde shared the prejudice in favour of relative that in such clauses — which I suspect he didn't — he would no doubt have felt uneasy with

    that little tent of blue that prisoners call the sky

    This final objection wouldn't apply to

    the little tent of blue that prisoners call the sky.

    However, it would still be unsatisfactory to write:

    I never saw a man who looked
    With such a wistful eye
    Upon the little tent of blue.


    because 'that tent' can't plausibly refer to ' that day'.

    Sorry, that should read

    because ' that tent ' can't plausibly refer to ' the day '.

  44. David Crystal in Making Sense, The Glamorous Story of English Grammar identifies an interesting case of the comma rule and a non-restrictive relative.

    That stickler for rule-governed punctuation, Lynne Truss, dedicated Eats Shoots and Leaves to

    To the memory of the striking Bolshevik
    printers of St Petersburg who, in 1905,
    demanded to be paid the same rate for
    punctuation marks as for letters, and thereby
    directly precipitated the first
    Russian Revolution

    A reviewer in the New Yorker — presumably an AmE speaker — criticised her for omitting a comma before a non-restrictive relative clause. But, as David Crystal points our, observing the rule would produce

    To the memory of the striking Bolshevik
    printers of St Petersburg, who, in 1905,

    Three commas in a row? For many readers (and publishing houses) this offends against legibility and aesthetics.

  45. Lynne - have we discussed the fact that it is not only 'which' or 'who' which are used for NRRC? Surely 'when', 'where' (and possibly others) are also commonly used? Doesn't this back up a point that has been made before, that it is not so much which/that which are crucial but punctuation and meaning, and perhaps, the forgotten skill of reading aloud?

  46. Thanks to David C & L. Truss for St Petersburg printers example. Yes, I think it backs up this point: despite prescriptivists' attempts to insist on the distinction, the possible grammatical force of commas may not always be appreciated, even by well-known journalists. Style and typographical concerns may override, even in the broadsheet papers.

  47. I have taught this to journalists, both UK and overseas, and, although they get the point when text is read aloud, in practice and under pressure commas are often used quite loosely. (And the distinction often becomes muddled, then forgotten, even with professional writers.)

  48. You're right, David, non-American Englishes have mostly lost the distinction between which and that in restrictive relative clauses.

    No, John Cowan had it right: non-American Englishes have not lost the distinction; it's written American English that has enforced a distinction that did not exist.

    The problem is that 'that' is (mostly) confined to restrictive clauses. 'Which' can be used on both restrictive and non-restrictive clauses. But in the interest of 'clarity', someone decided that it would be nice to have a division of labour between the two: "Let's use 'that' for restrictive clauses and 'which' solely for non-restrictive clauses. Wouldn't that make English much tidier!" That is where the problem lies -- the decision to ban 'which' from restrictive clauses in 'good usage', not the loss of an earlier distinction.

    If you think a little further than editorial prescriptions, you will discover that this 'rule' is actually fairly arbitrary. Why? Because the logical corollary of a rule on 'which' and 'that' is that there should be a similar distinction between 'who' and 'that'. That is, there should be a distinction between:

    "The man that is driving the car..." (restrictive) and
    "The man, who is driving the car,..." (non-restrictive)

    If there is to be a division of labour, why isn't it extended to these two?

    This shows the arbitrariness of the 'rule', and the fact that the distinction between 'that' and 'which' was never lost; it was instituted by an arbiter of usage who thought that English could be 'improved', but, like most grammatical amateurs, didn't really understand the whole picture.

  49. > non-American Englishes have mostly lost the distinction between which and that in restrictive relative clauses

    I don't think so. I think they never had it. It's not there in Shakespeare or Dickens. Even more significantly, it's not there in Hawthorne or Mark Twain either.

    Rather than claiming that American English is preserving some distinction that other varieties of English have lost, shouldn't we just be debunking this "rule" as a bit of prescriptivist nonsense (originating, I think, with Fowler's English Usage and sadly perpetuated by Microsoft Word's spellchecker app)?

    1. Yes, you’re right and I should update this very old post. I treat the matter much better in The Prodigal Tongue

    2. Thanks. I shall check it out!


The book!

View by topic



AmE = American English
BrE = British English
OED = Oxford English Dictionary (online)