Advice from numbers

« previous post | next post »

This morning, Arnold Zwicky took a look at the general question of whether language mavens' advice to "Avoid Potential Ambiguity" is actually helpful in avoiding ambiguity. He focused on the particular case of sentence-adverbial hopefully, and part of his argument was that if you're fluent in English,

you have to know that lots of people use hopefully as a sentence adverbial; it's all over the place. (I haven't run the numbers, but I'm sure that these days sentence-adverbial hopefully vastly dwarfs nominal-modifying hopefully in both colloquial and more elevated English.)

Well, here at Language Log, we aim to leave no number unrun. So I went to Mark Davies' lovely "Corpus of American English" search page at BYU, and checked a sample of 100 instances of hopefully from each of the five genres that he offers: spoken, fiction, magazines, newspapers, and academic. I assigned each example to one of the two categories "speaker-oriented" (i.e. sentence adverb meaning "it is to be hoped") and "subject-oriented" (i.e. verbal adverb of manner, meaning "in a hopeful manner").

The results, expressed as percentage of subject-oriented examples… the envelope please…

Spoken Fiction Magazines Newspapers Academic
0% 62% 10% 8% 3%

Across all the genres, the average was 17%. Leaving fiction out, the average was 5%. (The numbers would no doubt vary in different samples — but these should be reasonable ball-park figures.)

55% of the "subject-oriented" instances of hopefully in fiction were in quotatives, e.g. "'Shall I read to you?' he suggested hopefully." 21% were manners of gazing, e.g.

"That's true enough," Bonny said and looked at Phyllis hopefully.
I sipped coffee, gazing hopefully at my watch.
She pulls the hem of her skirt up to mid thigh. She looks hopefully over to Edward.
Kay looked for support from his father, and Ector looked hopefully at Myrddin, for his interpretation of events.
Walker found himself staring into them hopefully, searching for the answers he could find nowhere else.
Doug looked up at him hopefully.

The others were manners of moving, gesturing, and so on.

Since the "spoken" part of the BYU corpus involved pretty elevated discourse (transcripts of Larry King, PBS Newhour, 20/20, etc.), I checked the first 100 hopefully's in the Switchboard corpus — same result, 0% subject-oriented hopefully.

As a further check, I looked at 100 random examples from each of the 5 genres in the British National Corpus (except the Academic genre, where there were only 79 instances altogether):

Spoken Fiction Newpaper Academic Misc
0% 63% 4% 2% 7%

Again, the manner-adverbial hopefully's are mostly in quotatives and gazes.

So we can quantify Arnold's surmise. In spoken English, even in fairly formal settings, hopefully is not ambiguous, because it's essentially never used as a manner adverb. In written English non-fiction, the manner-adverbial use is well below 10%, and probably below 5% in most genres. In fiction, the manner-adverbial usage is common, but largely limited to a few stereotyped cases — hopeful quotatives, hopeful looks and hopeful gestures account for the great majority of examples.

Since the speed and ease of linguistic perception generally tracks usage frequencies quite closely, we can turn these numbers into advice for writers. If you're writing non-fiction, don't use hopefully to mean "in a hopeful manner". If you're writing fiction, feel free to use hopefully as a manner adverb, but only to modify quotatives, verbs of looking, and verbs of motion with animate subjects. And don't use it in dialogue.

If you want, go ahead and placate the crazies by avoiding the sentence-adverbial use of hopefully that means "it is to be hoped". But don't take their advice and actually use the manner-adverbial sense, at least not without thinking very carefully about what you're doing.

Like most general prescriptions, even valid ones, this rule can certainly sometimes be broken in effective writing. But if you were to listen to most of the usage mavens on this one, you'd be wrong almost all the time.

[I've assigned this post to the category "Prescriptivist Poppycock". But I hope that's not right, because I'm the one giving the prescriptive advice here.]


  1. Meesher said,

    June 2, 2008 @ 10:29 pm

    Is "ball-part figures" some sort of eggcorn? T and K aren't very close on a qwerty keyboard — could it stem from more frequent use of "part" than "park" (see "particular" in paragraph 1)? Am I completely off base (wink wink) in assuming that "ballpark figures" came first? Is an unreleased /k/ in "ballpark" a factor?

  2. baylink said,

    June 2, 2008 @ 10:41 pm

    Spider Robinson, who agitates for 'hopeably', would be pleased…

  3. Cheryl Thornett said,

    June 3, 2008 @ 2:31 am

    David Crystal classes 'hopefully' as a disjunct, along with other sentence adverbials such as 'thankfully' and 'fortunately'. He defines disjunts as making a 'an observation about the truth of a clause… or a value judgement about it'.

    Has anyone encountered warnings against using other sentence adverbials?

  4. Greg Dyke said,

    June 3, 2008 @ 2:35 am

    I'm having trouble thinking of example sentences containing final hopefully that are not manner adverbial.


    1) Hopefully, he went home.
    2) He hopefully went home
    3) He went home, hopefully
    4) He went home hopefully

    All four of the sentences make unambiguous sense to me, with 1) being the most likely sentence-adverbail and 4 being the only manner-adverbial.

  5. Mark Liberman said,

    June 3, 2008 @ 4:42 am

    According to MWCDEU,

    The onslaught against hopefully in the popular press began in 1965 and reached a peak around 1975, which is the year the issue seems to have crossed the Atlantic. Viewers with alarm there would repeat all the things American viewers with alarm had said, and add the charge of "Americanism" to them.

    The locus classicus of most of the charges leveled against sentence-modifying hopefully is Follett 1966. Follett died in January 1963, and it is likely that his analysis was one of the earliest to have been written down. He seems to have been the originator of the theory that this use of hopefully was un-English and that it came from "hack translators" of German who used it to translate the German hoffentlich. But he doesn't produce any hack translations to back up his assertion: all his examples seem to be from American newspapers. Besides the irrelevant German objection he also complains that this hopefully lacks point of view and adds a social objection demeaning people who use vogue words. His discussion lacks only a complaint about the loss of the original sense of the word, but others were around to supply it. (The complaint appears as recently as Garner 1998.)

    MWCDEU concludes:

    There has been a considerable abatement in the fuss since [1975] and many commentators now accept the usage, but it seems safe to predict that there will be some who continue to revile it well into the new century. You can use it if you need it, or avoid it if you do not like it. There was never anything wrong with it.

    I've made a (semi-serious) suggestion in the other direction — manner-adverbial hopefully has become so rare, except in quotatives and as a modifier of verbs of seeing and moving in fiction, that you are likely to be misunderstood (or thought to be strange) if you use it in speech or in non-fiction writing.

  6. Mark Liberman said,

    June 3, 2008 @ 5:03 am

    @Greg Dyke: I'm having trouble thinking of example sentences containing final hopefully that are not manner adverbial.

    I think you're right that expressions of this general type are much more likely to be used at the beginning of a clause than at the end. But as well as thinking, you can also look.

    First, for the general pattern — sentence-initial "apparently" occurs 2,618 times in the BYU Amercican English corpus, while sentence-final "apparently" occurs just 416 times. Sentence-initial "admittedly" has 350 hits, while sentence-final "admittedly" has just 12.

    Second, for the particular facts with "hopefully". Searching for sentence-final hopefully in the BYU corpus yields mostly examples of the "he said hopefully" type, most of which are from fiction; but there are plenty like this:

    What will that do to the Democratic Party? It'll ruin it hopefully.
    …and I think this has taken us into another step, hopefully.
    …and his reflexes are reflexive enough not to do something stupid, hopefully.
    Can Brocco do that? Hopefully.
    And we're ready to expand the restaurant — worldwide, hopefully.
    …by the federal government, hopefully.

  7. Bob Ladd said,

    June 3, 2008 @ 6:59 am

    There's a good article on the development of hopefully as prescriptivist whipping-boy by Stan Whitley in American Speech in 1983. It's very accessible in the sense of being non-technical and well-written, but accessible in the sense of being easy to get hold of only if you are near a well-stocked university library or are connected to JSTOR . If you can't read it on JSTOR, the full reference is Whitley, M.S. (1983), "Hopefully, a shibboleth in the English adverb system", American Speech 58: 126-149. He does quite a nice job of showing how the stigma attached to this usage was manufactured more or less out of nothing, and makes a tentative case that the usage may actually have been around for a long time before Follett and others decided to criticize it.

  8. Ewan said,

    June 3, 2008 @ 8:35 am

    OK – so realistically there needs to be a pause if you are going to use the subject-oriented hopefully sentence-finally, and the intonation changes slightly if you use it after the subject – and of course, that's not always punctuated (particularly not in the "he hopefully" case) so it's worth looking at every context. But if frequency-tracking is really so good, simple rareness — not impossibility — is a good enough reason to prescribe against a particular using, if that's the business we're in. Judging by the numbers on "apparently," it's probably safe to update the advice to, "If you're writing fiction, feel free to use hopefully as a manner adverb . . . but not sentence-initially."

  9. Rubrick said,

    June 3, 2008 @ 1:19 pm

    Sentence position does seem key in differentiating the two usages. Pinker's eloquently absurd putative-manner-adverbial example in The Language Instinct, "Hopefully, Michael Jordan stepped up to the free-throw line", becomes reasonable if changed to "Michael Jordan stepped hopefully up to the free-throw line."

  10. Steve said,

    June 3, 2008 @ 1:38 pm

    To answer Cheryl Thornett's question, I used to ask people who objected to 'hopefully' as a sentence adverbial – or disjunctive adverb (as I was taught to call it) – whether they also objected to 'thankfully' as in 'Thankfully, it didn't rain' and at least one replied, yes, that was a barbaric usage and it definitely implied to him that the weather was expressing its gratitude. So there is at least one person in the world who would warn against 'thankfully'. But I've never met anyone who objected to 'fortunately' or 'admittedly' or any of the many other adverbs that are used this way, and obviously (another one!) one wouldn't take advice on any topic at all from somebody who could think you meant that the weather can be grateful.

    Most people who object to sentence adverbial 'hopefully' will tell you it should be replaced by 'It is to be hoped that' – because it clearly doesn't mean the same as 'I hope'. Personally, (I can't seem to stop using them now) I think that anyone who thinks that a 6-word passive infinitive with an impersonal subject is superior to a simple adverb, has just demonstrated why one should ignore any opinions they might have about correct grammar or elegant style.

    Actually, (see how many of them there are?) I suspect that what most people object to is that it is overused by politicians and sportsmen in interviews to say that 'Hopefully, the economy will recover', or 'Hopefully, we'll win the match'. Nobody has ever said it wasn't a cliche, but that doesn't make it incorrect.

  11. Breffni said,

    June 3, 2008 @ 2:48 pm

    Mark’s suggested proscription of content-oriented hopefully – to take a half-serious suggestion completely seriously – would tend to rule out perfectly innocuous constructions like “look forward hopefully to”, “wait hopefully for”, etc., of which there are a good number of instances on the Web (few of them from fiction, as far as I can see – “I look forward hopefully to the next novel”, “We all congratulate Dianne, and look forward hopefully to the national awards in October”, “The world was waiting hopefully for the sensible American people to rectify the ill-advised actions of a rogue neoconservative administration”). Relatively rare they may be, but I doubt that they cause any serious processing problems, or that their users “are likely to be misunderstood (or thought to be strange)”.

    And I think it's a problem of principle. I have my doubts about the idea that statistical tendencies, even strong ones, can be transmuted into proscriptions, or even guidelines (Advice from Numbers, as the heading has it). Evidence can expose the absurdity of proscriptions based on spurious a priori arguments, but that isn’t the same as using statistical evidence to generate counter-proscriptions against constructions which are unquestionably grammatical. That looks to me like trying to derive “ought” from “is”.

  12. Andy Hollandbeck said,

    June 3, 2008 @ 3:12 pm

    I wonder whether Follett's analysis was tinged with the common "hatred" of anything German during WWII. Or perhaps Follett used anti-Germanism simply as a way to get people to agree with him without actually having to prove his ideas, in the same way that politicians, tricksters, and idiots (but I repeat myself) can quickly turn American opinion against anything by equating it with terrorism.

    Ask Rachael Ray about that one…

  13. Mark Liberman said,

    June 3, 2008 @ 5:52 pm

    In fact, I'm not serious about proscribing manner-adverbial hopefully.

    At least not in general.

    It works well in quotatives; it works well with "gaze" verbs (including Breffni's example "…look forward hopefully to…", which is a sort of fixed expression anyhow); it generally works well with motion verbs or verbs of waiting that have animate subjects.

    But if you use the hopefully-of-manner in a phrase like "he'll hopefully try a few threes in overtime", you're almost certain to be misunderstood, or at least to make your readers waste time trying to figure out what you meant.

    In other words, it's only the manner adverbial, not the sentence adverbial, that is ever likely to be perniciously ambiguous.

  14. Honzik said,

    June 3, 2008 @ 6:48 pm

    Great work on this. Strangely, it made me laugh out loud. But then I write for a living.

  15. Ran Ari-Gur said,

    June 3, 2008 @ 8:01 pm

    @Steve: Judging by your metacomments, I think you might be misunderstanding the nature of (or putative reasoning behind) the proscription of "hopefully". "Admittedly", "apparently", "actually", "fortunately", "obviously", and so on all are all fine by this reasoning, because they assert the sentence's claim itself to be admitted, apparent, actual, fortunate, obvious, or what-have-you. Sentence-adverbial "hopefully" and "thankfully" differ in that they're speaker-oriented — it is the speaker is hopeful or thankful, not the sentence's claim or its subject's referent. (Obviously, that doesn't make the proscription right; just because there's a general pattern that applies to most sentence adverbials in -ly, that doesn't mean that every sentence adverbial must absolutely conform to that pattern. But I have to admit that this sort of logical/deductive reasoning about grammaticality is much more appealing than a lot of kinds of prescriptivist arguments.)

  16. Mark Liberman said,

    June 4, 2008 @ 5:44 am

    Ran Ari-Gur, responding to Steve: Judging by your metacomments, I think you might be misunderstanding the nature of (or putative reasoning behind) the proscription of "hopefully".

    But it's important to keep in mind how diverse the arguments against the sentence-adverb version of hopefully have been. I think that Ran has a different argument in mind from the one that Steve was ridiculing. Ran's version of the argument is more sensible, but Steve's version is also out there. And both Ran's version and Steve's version are unconnected to the issue that Arnold and I were addressing.

    Among the charges that have been lodged against hopefully are innovation, modishness, ambiguity, Americanism, being a "hack translator's" calque from German, ungrammaticality, and perhaps some others that I've forgotten. Arnold (and I) have been discussing only the argument from ambiguity. (And of course Arnold's original point has nothing special to do with hopefully, which is just a convenient example of a more general form of argument.)

    The potential ambiguity of hopefully happens to be syntactic as well as semantic, but that's not crucial to the general issue of whether "Avoid Potential Ambiguity" — as interpreted by some language mavens — is a sound piece of advice. Arnold argued that it isn't, because potential (as opposed to effective) ambiguity is so pervasive as to be unavoidable. I merely observed that a serious concern to avoid effective ambiguity would actually lead to the opposite of the usual prescription about hopefully.

    Ran and Steve were dealing with (two different versions of) a different argument: that the sentence-adverbial usage of hopefully is actually ungrammatical, not just stylistically inadvisable. (The argument from ungrammaticality is inconsistent with the argument from ambiguity, of course, though some people seem to accept both at once.)

    One form of the argument from ungrammaticality is the idea that Steve was implicitly ridiculing, namely that that adverbs must (as the name implies) modify verbs (well, adjectives or adverbs too), and therefore all attempts to use adverbs to modify sentences are solecisms. Another form of this argument relies on belief in a more general principle, namely that grammatical relations such as modification must always connect specific words, so that a sentence adverb is necessarily "dangling" and therefore unlicensed.

    I know that this idea seems so obviously contrary to fact that it's hard to believe that anyone can actually think that it's true. But you will nevertheless find passages like "Adjectives and adjectival phrases attach themselves to the nearest noun. Adverbs and adverbial phrases attach themselves to the nearest verb, adjective, or adverb. Dangling modifiers occur when a word or phrase either has no word in the sentence to modify or seems to modify the wrong word."

    There's a more sophisticated form of the argument from ungrammaticality, which is the one that Ran has charitably attributed to Steve's imaginary opponents. This argument says that sentence adverbs of the form [BASE+ly S] are always compositionally related to something like [it is BASE that S]. This permits [Admittedly S] = [it is admitted that S], but not [Hopefully S] = [?it is hopeful that S].

    This argument is not a convincing one, in my opinion, because derived words don't always have a transparent relationship to a fixed semantic structure involving their component morphemes. As Ran points out, thankfully as well as hopefully follow a different pattern, and similarly, [Frankly S] means that the speaker is being frank, not that the sentence (or its subject) is. [Happily S] and [Unhappily S] are similar.

    A strictly comparable form of argument would be to observe that there is a productive pattern whereby noun compounds of the [X oil] mean "oil made from X" — olive oil, peanut oil, flaxseed oil, soybean oil, cottonseed oil, etc. — and therefore hair oil and sewing-machine oil are ungrammatical and should be banned.

  17. Steve said,

    June 4, 2008 @ 11:25 am

    Ran is quite right to point out that I was conflating two different kinds of sentence adverbial, and I'm grateful to Mark for teasing out my meaning more clearly. Cheryl – to whose comment I was directly responding – had already conflated them to some extent by bracketing 'thankfully' and 'fortunately' together, and as Mark recognised, I was more intent on ridicule than accuracy. Still, the fact that I used 'thankfully' rather than 'fortunately' in my attempt to wrongfoot critics of 'hopefully' rather implies that I was aware of the distinction, but my further comments certainly obscure that fact.

    Still, I'm not sure that the distinction is quite as cut-and-dried as Ran suggests. After all, one of Follet's objections to 'hopefully' was that it 'lacks point of view', and I was trying to suggest something similar when I said that 'hopefully' does not mean quite the same as 'I hope'. It seems to me that an adverbial makes a statement less personal and more universal – 'Hopefully, we'll win' strikes me as much less personal than 'I hope we'll win'. (I suspect that's why 'hopefully' is so popular with politicians and sportsmen.) If my intuition is right, then, presumably (and which sort is that one?) it's because the use of a sentence adverbial tends to shift the reference away from the speaker and towards the sentence or subject, at least to a degree.

    I am also quite aware that none of this has anything to do with ambiguity – my apologies for going so far off-topic.

    (Oh, and re-reading my earlier comment, I realise it could be seen to imply that I think 'It is to be hoped that' is incorrect grammar as well as inelegant style, and that I think an infinitive can have a subject – neither of which is true.)

  18. Arnold Zwicky said,

    June 4, 2008 @ 12:51 pm

    To Steve, who asserts that "'hopefully' is so popular with politicians and sportsmen". Something similar was asserted by a commenter on LanguageHat's blog entry on "hopefully": (where it was claimed that sentence-adverbial "hopefully" originated in the speech of politicians). The historical claim is certainly false, as you can see by looking at the cites in MWDEU and the OED.

    I am suspicious of the claim about current usage; Steve's impressions are not evidence, and I know that people's impressions about who uses particular variants are often seriously mistaken (they are based on the experiences of individuals, and they are subject to well-known distortions of selective attention and confirmation bias). This is true even when people are very confident that their impressions are correct.

    But there's a testable claim here, though testing it could take some real work. We'd need samples of the speech of politicians and sportsmen (in public discourse) — however "sportsman" is defined here — and a comparable sample of the speech of others (again in public discourse). These samples would have to come from a number of different speakers — there is sure to be individual variation here — and would have to be large enough to allow for the statistics to be interpretable. Then we mark every instance of sentence-adverbial "hopefully" (possibly just the sentence-initial occurrences of "hopefully", which are easier to pick out as sentence adverbials) and calculate their frequency of occurrence (per N words or per N sentence-like units, however these are defined) in the different samples. Then compare.

    My guess is that some people are just noticing occurrences of sentence-adverbial "hopefully" in the speech of politicians and sportsmen (perhaps because they *expect* these speakers to use the variant) and disregarding occurrences elsewhere. Comparable things happen a lot, for instance in the very strong (and largely mistaken) beliefs that many people have about what sort of speakers use quotative "like".

  19. Arnold Zwicky said,

    June 4, 2008 @ 1:28 pm

    The comments have moved off into discussions of the semantics of sentence-adverbial "hopefully" vs. other variants. In my original discussion, I accepted for the sake of argument the position of many handbooks that things like "I hope that" and "It is to be hoped that" are equivalent to sentence-initial sentence-adverbial (SISA) "hopefully". This is of course not so. As a general matter (as I point out repeatedly) variants are almost never equivalent in their semantics and discourse function, and that's clearly the case here.

    SISA "hopefully" has a lot of virtues. Brevity, for one. Second, coming first, it serves as a framing adverbial for the clause that follows — like such other adverbials as "in fact", "regrettably", "indeed", "frankly", etc. Third, as one aspect of this function, it's subordinate, while "I hope that" and "It is to be hoped that" have the clause that provides the main asserted content as a subordinate clause, against the intended information structure. And fourth, it allows the speaker to include others in the set of hopeful people. (Of course, there are contexts in which you might not want to this, and then other alternatives are available, for instance parenthetical "I hope" within the clause: "Kim, I hope, is not your only friend.")

  20. Steve said,

    June 4, 2008 @ 2:08 pm

    Thanks for the link to the LanguageHat discussion, but I don't actually agree with the assertion made there that sentence adverbial 'hopefully' originated with politicians. I was making the different point that the reason people objected to it so strongly was that it was overused, and that people often justify their objection to a cliché by asserting that it is incorrect. The reference to politicians and sportsmen in my first comment was intended as an example of its popularity, not of its origin, and in my second as an illustration of why its tendency to sound more objective and less personal might be useful to many classes of speaker.

    Hopefully, that clears up the misunderstanding.

  21. Les VanderLism said,

    June 4, 2008 @ 7:03 pm

    Mark Liberman wrote, If you use the hopefully-of-manner in a phrase like "he'll hopefully try a few threes in overtime", you're almost certain to be misunderstood.

    Not if you use it twice: "hopefully, he'll hopefully try a few threes in overtime". (You're using some kind of sports reference I don't fully understand, but you gert the general idea.)

  22. Breffni said,

    June 5, 2008 @ 12:03 am

    it's only the manner adverbial, not the sentence adverbial, that is ever likely to be perniciously ambiguous.

    I recognise that that’s the real take-home lesson, but (sticking dourly to the idea of converting observed usage tendencies into usage advice) as a piece of advice – as distinct from a proscription – it would leave to the writer nearly all the work of sorting out good instances of manner-hopefully from bad. On the other hand, the fuller account is unwieldy, since it refers to a disparate collection of verb categories that aren’t themselves at most people’s fingertips as concepts (quotatives, verbs of waiting and looking with animate subjects and so on). And of course, if construed as a normative rule rather than a statistical tendency, it will still rule out good sentences (like all such rules, as you point out. Here’s one by Steven Pinker: “Any explanation of how the mind works that alludes hopefully to some single master force or mind-bestowing elixir like ‘culture’, ‘learning’ or ‘self-organization’ begins to sound hollow…” – How the Mind Works, p.19 of Norton 1999 edition).

    And there’s a risk that observed corpus-based patterns will miss the real point – it seems to me, for example, that the fiction / non-fiction split that comes out of your preliminary analysis is explained (and therefore superseded) by your account of the kind of verbs that can most comfortably be modified by hopefully.

    So my point – and maybe it's tangential to your original post – is that I’m not convinced that corpus-based accounts convert easily to advice, or at least not to readily usable advice.

  23. Melissa Bollbach said,

    June 9, 2008 @ 12:06 am

    There are other "authoritative" sources besides Strunk & White that disparage sentence-adverbial hopefully…I had no problem with it until the usage was disparaged in a Babysitters Club book by Claudia's older sister Janine, who was a genius, and in retrospect also obnoxious and somewhat misguided. After reading that I felt like I knew a smart person's secret and I avoided sentence-adverbial hopefully myself.

RSS feed for comments on this post