Double passives, real grammar, and finding fault

At Macmillan Dictionary Blog I’ve been writing about double passives, beliefs about grammar, and usage criticism. Excerpts and links follow.

In The double passive is suggested to be avoided (sometimes), I look at a construction often criticised in writing manuals, reporting on why double passives are (sometimes) problematic, and what writers can do to avoid them:

The double passive, as its name suggests, is when a phrase contains two passive constructions yoked together. There’s one in the title of this post. How acceptable it is depends principally on how legible or awkward is the result. Phrases like ‘It must be seen to be believed’ and ‘He was sentenced to be shot’ are fairly straightforward and unobjectionable. ‘The order was attempted to be carried out’ (a line cited in Burchfield’s revision of Fowler) begins to pose a problem, because it’s unnecessarily complicated.


Reflections on Real Grammar follows up on Macmillan’s recent series on that topic, which included a quiz in which over 13,000 people took part. In a Twitter chat I was asked if the results surprised me. Some did, such as the 24.7% who said they would say Whom did you see at the coffee shop? rather than Who…? in a conversation with their sister:

This seems a very high proportion. Remember, it’s a hypothetical chat with one’s sister, not a formal job application. Some answers were probably an attempt at the ‘right’ answer – the more formally ‘correct’ or ‘proper’ one – rather than a realistic and honest answer. Instead of saying what they would say, some people may have said what they thought they should say. This often happens in surveys. But it might not explain all the thousands of people saying they would use whom in a casual conversation with a family member.


Finally, in Finding fault in the right places I examine the practice of using examples of people’s language to make a point about correctness, and stress the importance of doing this appropriately:

Criticising language use is a political act. If we say, ‘This is bad English’ or ‘X here should be Y’, then it matters who we use to illustrate our point. There is the option of making up examples, but existing ones can be more meaningful, showing readers how and where someone’s grammar or style went awry in real life.

For centuries grammarians have used examples from books and other printed material to analyse or deplore certain writing practices, often stating that their intent is not to shame but to educate. . . . Edited copy is fair game: criticism goes with the professional territory. But the same high standards should not apply to casual contexts like everyday conversation.

You can also browse my full archive of articles for Macmillan Dictionary.

32 Responses to Double passives, real grammar, and finding fault

  1. Tim Martin says:

    “the 24.7% who said they would say Whom did you see at the coffee shop?”

    Before clicking on the quiz, I thought to myself “there’s no way that one quarter of people actually speak that way.” The proportion of people who have a tin ear for the language is (fortunately) not that high, and while many people are great at changing a “to who?” at the end of a sentence to a “to whom?”, very few are good at choosing the “correct”* who/whom at the start of a sentence.

    *by prescriptivist standards

    Then, I clicked on the link to the survey. I took this survey a while ago! (Probably when you shared it?) I must say, this strikes me as a terrible survey to infer actual usage data from. While this survey’s introductory text is a little bit different from many of these “test your grammar” surveys, the fact is that overall the survey looks very much like a test. I would be very surprised if people didn’t treat it as such.

    • Stan Carey says:

      Tim: Whatever people’s motivation for answering whom (or awkwardly avoiding a split infinitive, etc.), the high proportion leaning towards such superstitious hyper-formality is telling in in own right. This is why I described the post as a reflection on ‘beliefs about grammar’, not ‘facts about usage’ or anything like that.

      • Tim Martin says:

        Oh, maybe I misunderstood you. Are you saying that, even if this were a traditional, prescriptivist-type grammar test, you would find it surprising if as many as 25% of people chose the sentence with “whom” in it?

        I actually think the opposite. Among readers of a *dictionary blog,* I would expect the percentage of people who can give the right answer on such a test to be much higher! …Though my intuition isn’t based on any scientific data.

      • Stan Carey says:

        If it were a prescriptivist-type test, the question would likely be phrased differently, perhaps with ‘should’ instead of ‘do’ and almost certainly without strong hints at the informal context (sister, coffee shop). So I’m wary of extrapolating directly. If it were such a test and the question weren’t so pointedly aimed at conversational register, I would expect a much higher figure than 24.7% to answer whom.

  2. Mel says:

    When I did the Macmillan test I wasn’t sure whether it was aimed at native or non-Native speakers. There’s a lot of “You selected Answer A, but you should have selected Answer B because B is more natural” in the feedback screens – which is okay if it’s aimed at non-Native speakers as a guide to actual usage, but if it’s aimed at native speakers then I find it pretty inappropriate. It’s one thing being descriptivist and stating which versions are more common and which ones are gaining usage, but to dismiss native speakers’ responses to questions as “unnatural” and “not a good answer” just because we’re not all using the hippest grammatical structures is basically reverse-prescriptivism.

    Plus in the “different from/to” question if you select anything other than “Both correct” it assumed you selected only “different from” as correct, even though I selected “different to”. After that I wondered if the intention was to bait people into giving the older/formaller responses in order to be able to correct them. If the quiz was intended as a survey of actual usage, then so much of it was badly written and badly phrased for that purpose – calling it a “quiz”, the subtitle “time to learn […]”, asking about “correctness” all the way though, marking things as right or wrong rather than commenting on frequency/prevalence… Of course the results skew formal, people thought they were being tested not polled.

    • Stan Carey says:

      Hi Mel, I’ll be sure to pass on your feedback to the editors at Macmillan Dictionary. I don’t think it’s about using the ‘hippest’ grammar, more about asserting the legitimacy of using what comes naturally in informal situations, despite the traditional hard line taken by some prescriptivists.

      • Mel says:

        If you select “May I…” over “Can I…” it says “That’s not such a good answer. […] In this case [‘Can I…’] would be the more natural choice.”

        I’m a native speaker and I’m not hung up on what’s traditionally correct (e.g., I’d say “who” over “whom”, “bored of” over “bored with”, “I would like to” over “I should like to”). Yet, I genuinely would say “May I…” in the example given. It’s not ‘asserting the legitimacy of using what comes naturally’ if it’s specifically dismissing my natural usage as not being natural, in fact it’s trying to guide me towards changing my usage which is, erm, prescriptive. I don’t see why the answer’s explanation couldn’t be more along the lines of “Both are correct, but ‘can’ is used much more frequently than ‘may’.”

      • Tim Martin says:

        Mel, you’ve called the grammar quiz “prescriptivist” twice now, but you must understand – prescriptivists and descriptivists *both* make prescriptions. The difference is that the latter base their prescriptions on evidence. For example, on occasion I use constructions such as “me and him did X” and I don’t see any reason to stop, but I would recommend *not* doing this while speaking at a formal affair. That’s a prescription.

      • Mel says:

        Both prescriptivists and descripvitists have “evidence”, they just focus on different sources of evidence – the former preferring things such as etymology and historical tradition, the latter prevalence and current usage. (“‘Decimate’ comes from a Latin word meaning ‘to take one tenth’.” is evidence just as much as “‘Decimate’ is frequently used and understood to mean ‘to destroy almost completely’.”, different people will however give different weight to each.) The main difference is in normativism, and presciptivism’s attempts to derive the normative from the empirical, e.g. “Usage X is backed up by this given evidence whereas usage Y isn’t, so you should use X”. A truly descriptive approach would allow the evidence to speak for itself (simply “Usage Y is backed up by this given evidence whereas usage X isn’t”), leaving the individual to decide how they will act (or not) on the information. If descriptivists are going to prescribe too, then we end up trying to arbitrate between competing normative propositions, which is a pretty big step back for linguistics.

      • Tim Martin says:

        I guess I’m not entirely convinced that saying “X is more natural than Y” is anything other than a statistical fact. If “May I bring my sister to the zoo” is considered unnatural by, say, 2 out of 3 native speakers, or perhaps considered unnatural in 2 out of 3 related contexts, doesn’t that make “can” more natural than “may”? Would you prefer a more specific wording – such as “‘can’ is more frequently natural than ‘may,’ but ‘may’ also works”?

        This is exactly the kind of advice I would give to a non-native speaker of English who was wondering which word to use. “Can” is usually the more natural choice, and it is less likely to make you sound like a non-native who speaks textbook grammar. I wouldn’t say that “may” is ungrammatical, but how else should I describe the usage differences between may and can?

        As for what you say about prescriptivists, I think you’re giving them a lot of credit. I haven’t seen any prescriptivists who are consistent in their arguments. A prescriptivist may argue for using the original meanings of certain words (such as “decimate”), but at the same time they ignore *every other* word that has changed meaning since the beginning of the English language. If the argument goes “We should use the original meanings of words,” then the result is that at least half of your words have meanings that native English speakers would no longer recognize. I don’t know any prescriptivists who use “awful” and “awesome” as synonyms, or who use “girl” to mean “a child of either sex”, or who use “protest” to mean “declare.”

        Prescriptivists are similarly choosy about all of the other “rules” that they claim to conform to. It seems to me that they are more interested in finding arguments to support their pre-conceived preferences rather than forming preferences based on fact.

      • Mel says:

        Frequency is objective and specific. Naturalness is subjective and vague. The Macmillan advice is based on frequency (that ‘may I’ is used ten times more often than ‘can I’) and extrapolating naturalness from that, which is flawed. Even speakers who prefer to use “Can I…” might not consider “May I…” unnatural. Whether something is natural also covers a large number of separate concepts – “I not go can” is ungrammatical, “Colourless green ideas sleep furiously” is illogical, “Thou shalt see” is archaic, “She’s gotten married” is regional. If things are going to be put in terms of naturalness, then the specific reason that something is unnatural needs to be addressed. In this case I would drop the word “natural” from it entirely, they have no information on that anyway, and “‘Can I…’ is the more common usage, it is used ten times more frequently than ‘May I…'” conveys the same idea.

        As I said in my first comment, I have no problem with the use of “natural” in advice to non-native speakers where it can be shorthand for “what a native speaker is most likely to say”, and contrast with the “unnatural” of non-native speakers. Learners of English are also often exposed to older learning materials, and I don’t object to guiding them towards the most common modern usage and phrasing that as “natural”, giving them the knowledge and opportunity to adjust their own speech accordingly. But that is a very different situation to feedback for native speakers who already have their own intuitions about English and don’t need to be pushed towards adopting the most common usage in order for their language to sound more “natural”.

        Prescriptivism is a broad church, and I don’t think focussing on the most extreme/irrational/ridiculous voices in any movement is productive. There are plenty of prescriptivists who have slightly more substance to their views than “you should say it this way because that’s ow I like it”. And I don’t see why one would necessarily expect/require prescriptivists to come up with consistent, logical rules when language is so rarely consistent or logical. That said I’m not really trying to defend prescriptivism, normativism in language is a pointless endeavour. Anyone telling other people how they should be speaking is fighting the tide, it’s pretty irrelevant whether they’re justifying it on the basis of frequency, etymology, “naturalness” or “it’s just better!”. (The obvious exceptions being when they can engage an emotional response from speakers tied to a wider societal movement – “you shouldn’t say the n-word, it’s upsetting and offensive and you’ll be considered a bad person” or “you shouldn’t say Spital, it’s a French loanword and as a a true German you should say Krankenhaus otherwise we’ll assume you’re a traitor” or whatever – but it’s hard to get people collectively that impassioned about things like auxiliary verbs.)

        (Incidentally, “girl” meaning a child of either gender is still sometimes used, at least in bits of South West England where we traditionally call female children “maids”, although the standard usage of “girl” is much, much more common these days. Also we pronounce “can” as /kæːn/ (stressed) and /kæn/ (unstressed), and “can’t” as /kæːnʔ/ (stressed) and /kænʔ/ (unstressed), in fast speech the glottal stop often gets lost meaning the two become homophones. This might be why I use “May I…” – as a child of the 90s, it’s unlikely to be from having learnt any sort of rule, we were not taught any grammar at school. Might just be a bit behind the times though.)

  3. bevrowe says:

    I answered “whom” to the first question as I thought the quiz was about technically-correct grammar. (And I do use “whom” more often than most people). After that, I got the spirit of the thing and had every answer right.
    To say “whom” is the wrong answer is nonsense. To say both answers are correct would be acceptable.

    • Stan Carey says:

      Bev: I agree that whom is not ‘wrong’ in that context. It might seem unsuitably formal or excessively proper to some listeners, but that doesn’t amount to incorrectness.

  4. astraya says:

    Even though it comes from so august a source as Macmillan, I find the quiz to be an unhelpful mix of ‘What would you say?’ and ‘Which do you think is correct?’, which are two very different questions.

    • Stan Carey says:

      About half were ‘What do you say?’ and half ‘Which do you think is correct?’ (plus one ‘Which is correct?’ and one ‘Which do you think sounds more natural?’). Collectively they draw attention to the stubbornly popular but mistaken perception of correctness as equivalent to formal correctness, when in fact it varies according to contextual conditions.

      • bevrowe says:

        The problem with ll this is that the acceptability of an item needs to be plotted on a two-dimensional array, and the two dimensions are defined by rather lumpy continuous measures.
        On one scale we have the response of a native speaker and on the other scale we have register. If we arbitrarily number the scales 0 to 10 then “she cans go” rates 0 acceptability across all registers. “Whom …” probably ranges from about 4 in some registers to 10 in others. Split infinitives probably vary from 0 to 10!
        The essential points are that grammaticality is a continuous variable, not binary, and that registers overlap and merge into each other.

  5. David L says:

    On the double passives: my amateur judgment is that “It must be seen to be believed” is fine because “it” is the thing that needs to be seen and also the thing that needs to be believed. But “the orders were attempted to be carried out” is ungainly because while the orders are the things that would be carried out, they are not the things attempted. It’s the carrying out that is the thing to be attempted.

    Does this make “the orders were attempted to be carried out” grammatically incorrect”? I want to say yes, but I don’t have the technical means to say quite why.

    • Stan Carey says:

      Maybe, David, but I’m not convinced; I think the line is awkward but definitely grammatical. Burchfield provides an analogous example: arms were allowed to be shipped to Iran, which appeared in the New Yorker. To follow your logic: arms are the things that would be shipped, but they are not the things allowed: the shipping is. Yet the phrase works fine, I think – maybe because it doesn’t feel like it’s lacking an agent as conspicuously as the orders line.

      • David L says:

        Hmm, I have to disagree. “Arms were allowed to be shipped to Iran” sounds wrong to my ear. I would rearrange the sentence to avoid that construction. Personal reaction, of course.

      • Stan Carey says:

        That’s an interesting difference. It’s fine for me. (Whether as an editor I would be tempted to rephrase it is a separate issue.) Have you ever seen an ungrammatical construction get past the New Yorker’s copy editors?

  6. David L says:

    Oh, and I just took the quiz. I think there’s a mistake on question 5 — the one about rain having an impact on a picnic (or something, I can’t recall the exact wording). The first option, “having an impact on” is clearly correct, but the second used the phrasing “has impacted on” (my italics). I said only the first was correct but the “answer” is that both are. Not so, because of the “on.” Rain can impact a picnic; it can’t impact on a picnic.

    • Stan Carey says:

      The two options are: The heavy rain has already had an impact on food prices, and The heavy rain has already impacted on food prices. I disagree with your impression here: for me, rain can impact something, impact on something, and have an impact on something. I wouldn’t use any of these constructions, though, preferring affect and have an effect.

      • David L says:

        Hmm, again. “Impact on” sounds completely wrong to me — worse than the double passives above! FWIW, I’m originally British but have been in the US for 30 years.

      • David L says:

        PS “Impinge on” is fine, of course. Maybe there’s some crosstalk between the two expressions.

      • Stan Carey says:

        The phrasal verb impact on is recorded in most of the major dictionaries. The OED’s first citation is from William George Hardy, Father Abraham, 1935:

        For there was about them an air of eagerness and of shuddering expectation which impacted on his consciousness and fascinated even while it repelled him.

  7. astraya says:

    Re ‘Finding fault’: in my blog I sometimes discuss things my students have said or written, or I have heard or read elsewhere. I hope that it’s clear to anyone reading that I’m not laughing at or judging anyone, but rather using the example to seriously consider the language point – what is the language point, why has it arisen, and what can we learn about English, learning it or using it?. (I also occasionally discuss things that I have said or written!)

    • Stan Carey says:

      It’s certainly clear to me, David, and I’m sure the same goes for anyone else reading your blog. In my post there wasn’t room to elaborate on exceptions, but educational contexts are among them. For one thing, your tone and approach are manifestly respectful and interested, not disparaging or cynical. Also, when people are learning a language I think there’s a tacit understanding that what they say or write in the classroom can potentially be used to illustrate or discuss a point of grammar or usage, the reasonable assumption being that this will be done in an appropriate way: without belittling, etc.

  8. Ramakant Baunthiyal says:

    there is only one way to learn a grammar, read a lot and write a lot, nothing else work on me.

  9. Reblogged this on Notes from An Alien and commented:
    Today’s Re-Blog poses some interesting issues of grammar yet it is really firting with you to come visit the Macmillan Dictionary Site :-)

  10. Cheryl says:

    Hi, Stan.

    Which of the following is correct, concise and idiomatic?
    1. Words in italics are English in the original. 
    2. Words in italics are originally in English. 
    3. Italics indicate that the original word is in English.
    4. A word in italics indicates that the original of the word is in English

    Also, should Sentence #1 be re-written as “Words in italics are in English in the original”?

    • Stan Carey says:

      Hi Cheryl. I’ve seen this information presented in many different ways. None of your examples are incorrect, and you yourself can see which are more or less concise. Numbers 3 and 1 (with ‘in’ added as you suggest) work best. Other possibilities are:
      5. Italics mark words in English in the original.
      6. Italicized words are in English in the original.
      7. Words in English in the original are italicized.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: