Ask Language Log: indicating or stating (that) S
« previous post | next post »
S.A. writes:
I'm working with someone who writes sentences like these:
"Past studies indicate teachers with high self-efficacy for providing nutrition instruction devote more classroom time to the subject."
"D____ and colleagues have stated teachers are better able to promote healthy eating among preschool children when they are more knowledgeable about nutrition content…"
Doesn't there need to be a "that" between indicates and teachers and between stated and teachers?
The absence of a "that" is a constant throughout her writing so I need to know if I'm correct, and if so, how I can explain it to her.
Executive summary: The complementizer that is optional in sentential complements of verbs of communication like indicate, state, suggest, etc., but it is retained much more often than not. In formal academic prose, omission is very rare (about 1%) in frames like "results indicate (that) S" or "X and Y have stated (that) S". So it's not grammatically incorrect to omit that in such cases, but it's stylistically quite odd, and may confuse readers. Systematically omitting that in examples of this type is at best very eccentric.
Details: The balanced COCA corpus has 829 instances of "results indicate". I looked at a random sample of 100, and found
- 82 cases where indicate takes a clausal complement beginning with that (e.g. "Results indicate that there were no performance differences on the first exam");
- 16 cases where indicate takes a direct object (e.g. "Results indicate positive changes such as greater self-esteem")
- 2 cases where indicate takes a clausal complement with no that ("These results indicate the coaches in this study were uniform on the importance of addressing certain ideals in conference codes of ethics"; "The results indicate your LDL reading is low, but your HDL is very high").
There are 310 instances of "have stated". Looking again at a random sample of 100, I found
- 61 cases where stated takes a clausal complement beginning with that ();
- 21 cases where stated takes a direct object ("e.g. The Ukraine and Byelorussia have stated their desire to become non-nuclear states"; "He could hardly have stated more precisely humanity's history of meddling with the Great Plains");
- 12 cases where stated is used in quotative tags ("as you have stated repeatedly"; "as I have stated personally to you");
- 6 cases where stated takes a clausal complement with no that ("Well, we have stated very clearly Iraq has to comply with all the Security Council resolutions"; "Some have stated we've sort of set up a dating service between the agencies";"…earlier council decisions have stated Iraqi law applies to the emirate";"Anton Balasingham is reported to have stated they were 'limited and inadequate,'"; "As for work overload, researchers (Cooper &; Marshall, 1976; French &; Caplan, 1972) have stated there were two kinds of work overload"; "Many elderly swimmers have stated they won't stop swimming until the day they die").
Throughout the past couple of centuries, it's certainly easy to find cases where competent writers of standard English omit the initial that in clausal complements of verbs like indicate and state:
There was a gentle courtesy and softness in her manners that seemed rather to appeal for the indulgence of others, than to indicate they needed it. [from Catherine Maria Segdwick, Clarence; or, A Tale of Our Own Times, 1830] But, that I imagine you may not have received the letter, by stating you first heard of our marriage through the papers, I should say the want of courtesy lay on your lordship's side, for having vouchsafed me no reply to it. [Mrs. Henry Wood, East Lynne, 1861]
On the other hand, it's equally clear that in formal writing, omission of the complementizer that with verbs like indicate and state is not the preferred choice. In the sample from the COCA corpus, we saw just 2% (2 of 84) omissions in the frame "results indicate (that) S", and 9% (6 of 67) omissions in the frame "have stated (that) S", for a total of 8 omissions out of 151 opportunities. This may be true because the use of that often helps to resolve the potential ambiguity between a direct object and a clausal complement. But retention of that is also to some extent a marker of formal style; and this suggests that in scientific prose, that-elision will be even less common than it is in a balanced corpus like COCA. In fact this prediction appears to be true. Searching for "results indicate" on Google Scholar, and checking the first 100 hits, I find barely 1% (1 of 91) omission of that:
- 90 cases where indicate takes a clausal complement beginning with that;
- 9 cases where indicate takes a direct object ("Results indicate significant and consistent sex differences in self-efficacy with regard to traditional vs nontraditional occupations");
- 1 case where indicate takes a clausal complement with no that ("These results indicate BAK1 is a component of BR signaling").
Similarly searching for "have stated", I again find in the first 100 examples just a bit more than 1% (1 of 77) omission of that:
- 76 cases where stated takes a clausal complement beginning with that ("Many authors have stated that there is little or no tilt in an uninjured ankle");
- 15 cases where stated takes a direct object ("we have stated it in the weakest form that we can");
- 4 cases where stated is used in quotative tags ("as Ajiferuke and Boddewyn have stated");
- 1 cases where stated introduces a direct quotation ("With regard to Dr Charlton's letter concerning validation I have stated: 'There was no attempt to develop an arbitrary definition of general practitioner workload based on consultation rates'");
- 1 case where stated takes an infinitival complement ("Penicillamine is identical with the base which we have stated in an earlier communication to occur in acid penicillin hydrolysates";
- 1 case where stated takes a wh-complement ("I have stated what I suppose to be our general theory of construction").
- 1 case where have is the main verb, and stated is a participle modifying its object ("those studies that did not have stated exclusions for depression");
- 1 case where stated takes a clausal complement with no that ("Several authors have stated there is less striate cortex in blind people.")
Overall, we see just 2 omissions out of 168 chances in the scientific and technical writing (1.2%), where the norm is even more strongly to leave that intact and explicit, compared to 8 omissions about of 151 chances in the balanced COCA corpus (5.3%).
Discussion: Perhaps S.A.'s that-omitter understands the options, and believes that omitting that is better than leaving it in. In that case, S.A. can simply point to the overwhelming general preference for retaining an explicit complementizer with verbs like indicate and state, and can argue that leaving it out is likely to confuse readers.
But maybe the individual in question is just doing what comes naturally, and is not consciously aware of the nature and even the existence of the structures where this choice arises. If that's the situation, then some explicit instruction in syntactic analysis may be required.
I'm not sure how to do this. To start with, it's likely to be embarrassing to offer such remediation to a grown-up. And also, if someone has retained this level of grammatical innocence into adulthood, there may be a certain amount of individual resistance to the concepts involved. But the logical approach would be to provide a list of relevant verbs (say, indicate, state, assert, claim, suggest, mean, believe, etc.); to point out that such verbs are often associated with a clause that explains what is indicated, stated, asserted, etc.; and then to draw attention to the choice of omitting the clause-initial that or leaving it in.
mike said,
May 30, 2013 @ 6:29 am
>"some explicit instruction in syntactic analysis may be required"
Having edited many a text in which this very issue came up, I have found that instead of offering writers some explicit instruction in syntactic analysis, it usually works better to just show them a couple of with-and-without examples. In my experience, most writers don't actually want grammar lessons. :-(
[(myl) If that method works, then the person being instructed already grasps the nature of the choice, at least to the extent of being able to generalize appropriately from a couple of examples. In some cases, though, I've found that otherwise intelligent people aren't easily able to generalize in an appropriate way, and may need a more elaborate sequence of examples, and perhaps more explicit commentary, to get them pointed in the right direction.]
Lazar said,
May 30, 2013 @ 6:44 am
The first example is a doozy, because to me at least, the direct object misinterpretation persisted all the way until devote – my instinct was to read indicate in the medical sense of indicating X for Y.
Victor Mair said,
May 30, 2013 @ 6:46 am
Where I find it really awkward is when I want to say something like "stated that that individual was the one who…." If I leave out the first "that", then the second one seems to be forced into playing the role of the first, but I really want and need it for specifying which individual. In such cases, I fell compelled to keep both of them.
Bill Walderman said,
May 30, 2013 @ 6:58 am
Over the course of my professional career, I've encountered two individuals who insisted on eliminating all "that" complementizers from documents I had written, as if use of "that" to introduce subordinate clauses were some sort of grievous offense against good writing that I should have been ashamed of. (One was in a position of authority supervisor and the other was a client, so I didn't make any effort to dissuade them.) One of them, if I remember correctly, cited a primary or secondary school teacher as the incontrovertible authority for this. Apparently, there is (or was) a zombie rule about this in circulation among some English teachers.
[(myl) Interesting. Maybe this started as an especially enthusiastic application of "omit needless words"?]
AntC said,
May 30, 2013 @ 7:01 am
S.A.'s colleague's that omission compounds abstract noun pile-up denseness.
Both examples would make excellent case studies for how not to write.
What does either even mean? Are you sure?
But maybe the individual in question is just doing what comes naturally OMG you mean she talks like that?!
fev said,
May 30, 2013 @ 7:21 am
@Bill, there's a widespread sentiment among journalists and journalism students that "that" is indeed a needless word, probably because omitting it works so well with "said." AP style itself is fairly conservative on the issue and does provide the sort of list Mark suggests.
AP sometimes has a habit of ignoring its own stylebook, as in this gem from 1996 about the Atlanta bombing investigation: "Investigators apparently believed Jewell, who had a checkered career in police and security work, including an arrest for impersonating an officer, fit a common profile for a lone bomber." Imagine the fun if the story jumps somewhere in the relative clause.
Ginger Yellow said,
May 30, 2013 @ 7:29 am
To back up fev's point, I've worked at publications with style guides that expressly prefer omission except in cases of ambiugity/confusion. It may well be the case that in academic writing omission is disfavoured, but it is favoured or at least neutral in (British) journalese. So it's not so much that there's one preference in formal writing.
Can you suggest a source of text where this preference for omission is in force? Usage with say might be different, but I'd be surprised to see that verbs like indicate and state show up with that-omission actually more common in practice — or even close to being equally common. Of course, I'm frequently surprised…]
J.B. said,
May 30, 2013 @ 7:45 am
Lately, I have also come across several online writing resources telling authors that they can always omit words like "that" to tighten up their prose because they're unnecessary. In some cases, I've seen "that" put in the same category as intensifiers like "very" which the writers-to-be are told to avoid.
Ginger Yellow said,
May 30, 2013 @ 8:05 am
MYL: I'll try to dig up the style guides from those publications, but I'm not sure if I still have copies. In the meantime, here's the Guardian's:
Do not use automatically after the word "said", but it can be useful: you tend to read a sentence such as "he said nothing by way of an explanation would be forthcoming" as "he said nothing by way of an explanation" and then realise that it does not say that at all; "he said that nothing by way of an explanation would be forthcoming" is much clearer.
A similar problem arises with verbs such as argue and warn. "He argued the case for war had not been made" and "he warned the case for war had not been made" both become much clearer if yoiu insert "that" after the verb "
Adrian said,
May 30, 2013 @ 8:25 am
Since S.A.'s colleague is not making a mistake, it might be okay to ask her if she's considered that her readers might find her style odd or confusing, but I'd think it rather arrogant to tell her off or make her change her style. One shouldn't swap one zombie rule (Omit "that") for another, opposite, one (Include "that").
[(myl) By a "zombie rule", we mean a prescriptions whose validity is denied by essentially all reasonably competent authorities, and which has never been followed by elite writers of the formal standard language, but is still taught by some badly-informed teachers, or imposed by some equally ill-informed editors. It's possible that "always omit that in sentential complements" is a zombie rule in this sense — the facts are not yet clear to me.
But there's no way to put into the "zombie rule" category the advice to retain that in scientific writing, in frames like "results indicate that S" or "X and Y have stated that S", accompanied by the observation that omitting that in such cases is also grammatical. The grounds for the advice is the simple fact that usage is roughly 80-to-one in favor of retention.]
Ginger Yellow said,
May 30, 2013 @ 8:30 am
Torygraph:
that: after verbs like "said" and "thought" the conjunction can be omitted unless "and that" is required in the second half of a sentence, in which case two "thats" are needed (The reporter said that the sub-editor was mad and that, if he cut his copy again, he would beat him).
Ginger Yellow said,
May 30, 2013 @ 8:36 am
I don't have their style guide to hand, but on a quick search for "indicated" on http://www.FT.com produces in the first 10 results: one direct object, seven clausal complements with that, and two clausal complements without.
[(myl) A longer search for 100 instances of "indicate" with non-wh clausal complements on the FT site produced 80 with that and 20 without, which is indeed a substantially higher percentage of omission than in the other sources I've looked at. (I've left out the nominal objects, which are about as common as the clausal complements with that, and wh-complements like "indicate who/how/why/whether/etc.", and various other things.
But this is still far from across-the-board or even equi-probable omission.]
Dennis Paul Himes said,
May 30, 2013 @ 9:02 am
Why is allowing a writer to maintain a personal style (so long as it's grammatical) not being considered as an option?
[(myl) It's certainly an option. So is wearing flipflops with formal dress; but if a colleague were under the mistaken impression that this is the done thing, it would be a friendly gesture to set her straight.
In the that-omission case, it may also be true that there's an argument based on not confusing your readers — but we don't have any real evidence on that point either way.]
Góðan daginn said,
May 30, 2013 @ 9:10 am
Thanks so much for this blog post — the confusing omission of "that" has always irked me.
When I read something like "The meteorologist explained weather patterns," I am expecting the sentence to end with "to the audience." I am not expecting "are changing with the rise in global temperatures."
David L said,
May 30, 2013 @ 9:16 am
@Dennis Paul Himes: I don't see it as a question of style or grammaticality, but simply clarity. My attitude as an editor is that if a sentence reads clearly and unambiguously without a 'that,' I'm happy to leave it out, but if there's a chance it could be confusing I will put one in.
Talking of which — @Ginger Yellow — I find the Torygraph example a strange one. That sentence reads fine to me without the first 'that.' I can't see any grammatical reason why it's 'needed,' as their style guide contends.
Keith said,
May 30, 2013 @ 9:28 am
Victor Mair stated Where I find it really awkward is when I want to say something like "stated that that individual was the one who…." If I leave out the first "that", then the second one seems to be forced into playing the role of the first, but I really want and need it for specifying which individual. In such cases, I fell compelled to keep both of them.
Maybe you could avoid repeating the word "that" by rewording, for example stated that it was that particular individual who or by quoting exact speech stated "that was the individual who…", etc.
K.
Ginger Yellow said,
May 30, 2013 @ 9:50 am
Talking of which — @Ginger Yellow — I find the Torygraph example a strange one. That sentence reads fine to me without the first 'that.' I can't see any grammatical reason why it's 'needed,' as their style guide contends
Style guides are not necessarily logical or driven by grammatical "necessity". It's often as much about internal consistency and the pet peeves of the curator/policymaker as anything.
C said,
May 30, 2013 @ 10:17 am
I think I am in the minority here, but I was completely comfortable reading the original author's sentences (without the "that"). I certaintly bow to the statistics that are provided above, and they are compelling, but I'm also not sure [I naturally inserted a "that" right here but then deleted it for fun :-)] it is such a huge issue to leave it out, even in formal writing. I draft legal documents and I guess I'm just wondering if there really is some introduction of ambiguity or other clear defect (beyond the statistics presented, again compellingly) with just going with the author's natural style.
Ted said,
May 30, 2013 @ 10:20 am
@David L: The usage problem in the passage quoted from the Telegraph is not that the first clause couldn't stand on its own without "that." It could, though, like myl, I would find that appropriate in an informal register but not in formal writing. The problem the editors are trying to avoid, I think, is the failure of parallelism between the two clauses. If you're introducing the second clause with "that," you have to introduce the first clause with "that" as well.
But there also seems to be a logical problem with this passage. If the reporter is threatening the sub-editor for cutting his copy, surely it is the reporter who is mad at the sub-editor, not vice versa. (I suppose the sub-editor might be mad at the sub-editor for submitting bad copy, but that's conjecture. The remainder of the passage illustrates the reporter's anger, not the sub-editor's.) Could there be a transcription error here?
Ted said,
May 30, 2013 @ 10:22 am
@C: If the author's natural style results in phrases like "self-efficacy for providing nutrition instruction," I would submit that there is no circumstance in which she should just go with it.
DEP said,
May 30, 2013 @ 10:44 am
Which "that" do you mean?
"That that, that that that modifies."
kyomi said,
May 30, 2013 @ 11:20 am
One organisation whose policy is very much to leave out "that" is the official report of the Irish Parliament, as I've noticed because this is quite a pet peeve of mine. A search using just one of the words mentioned ("indicates") threw up three examples from the debate of 29 May 2013 alone: (http://www.kildarestreet.com/search/?s=indicates):
"He indicates there has been just a 2% decrease."
"Deputy X indicated we must empower parents…"
"[I]t also indicated some schools still operate less than wholly inclusive enrolment practices"
I suspect it's a style guideline applied as a hangover from the days when the report was printed on paper and every word represented an extra cost, similarly to the world of journalism.
Although I find it annoying (unpleasant to the ear?), I think it's much less unusual to see this sentence construction in British and Irish English than in American English, and I doubt it would even be noticed by a reader on this side of the Atlantic unless it made the sentence misleading. But I agree that it might be more useful in a scientific or technical context to include all the "that"s to maximize the readability of long and involved sentences.
Haamu said,
May 30, 2013 @ 11:29 am
@Ted: The passage makes good sense if you take mad to mean "foolish," "irrational," or "crazy" rather than "angry." This is a fair assumption, since the context was a description of British journalistic practice.
Haamu said,
May 30, 2013 @ 11:30 am
Okay, I see how my second sentence could be read two ways. I'll let it stand.
Joe said,
May 30, 2013 @ 11:38 am
I don't have time to check, but my intuition tells me that omission of complementizer "that" would be pretty common in headlines. Perhaps the writer is more familiar with seeing "reports indicate" etc. in headlines than in academic writing?
Joe said,
May 30, 2013 @ 11:40 am
@Ted
re: But there also seems to be a logical problem with this passage.
I had to re-read that sentence a couple of times but I think [that] the "mad" in "The reporter said that the sub-editor was mad " is intended to mean "crazy" rather than "angry".
Joe said,
May 30, 2013 @ 11:42 am
@Ted
re: If the author's natural style results in phrases like…
I couldn't agree more!
Neal Goldfarb said,
May 30, 2013 @ 11:50 am
@myl: "In the that-omission case, it may also be true that there's an argument based on not confusing your readers — but we don't have any real evidence on that point either way."
Why do you say that there's no real evidence on this? Sentences such as the one given by fev (below) clearly show garden-path effects.
Is there any doubt about that?
[(myl) I agree that it's plausible that omitting clause-initial that sometimes causes problems for readers. But I'm reluctant as a matter of principle to assert that this is obviously so, without at least a little experimental evidence, because there's a lot of bad writing advice out there that's based on what some self-appointed authority thinks are obvious effects on readability.]
Lee B said,
May 30, 2013 @ 11:53 am
The flip-flops in formal wear analogy is striking. It's not the done thing. But frequencies aren't decisive: bow-ties and boutonnières are unthinkable for almost all men, but the 1% who regularly wear them are not thought to be making a gaffe.
[(myl) Those bow ties and boutonnières are more like whither and shall, I think…]
For writing, as in dress, I'd think the most relevant consideration is the opinion and practice of esteemed individuals. Why else would the MWDEU cite statesmen and playwrights?
[(myl) For scientific and technical writing of the kind under discussion, I think the relevant standard is not so much the writing of especially esteemed scientists, scholars, and engineers, but rather the regular standard of prose in the kind of publication venues that Google Scholar indexes — which is why I chose that particular urn to draw from.
I'm sure Mark is right that "that"-omission is "quite odd" and "very eccentric." But it only matters because it's a more-than-statistical observation. It's got normative wallop, doesn't it?
[(myl) We're talking about a particular kind of conditional probability here, where an option is created by an earlier choice, e.g. the decision to use indicate or state with a clausal complement, and where the writer is then forced to select one of a small number of outcomes (here just two). In cases like that, it seems to me that the facts (about how a certain class of people behave) have normative force pretty much by definition, at least if we believe with Horace that it is usage "quem penes arbitrium est et ius et norma loquendi".
It's true that some cases of this kind are more psychologically salient than others. And in some cases, the choice may have connections of other kinds, e.g. to readability. But even if the distribution is purely a matter of fashion or convention, it's ipso facto normative, at least in the Horatian sense, isn't it?]
Belial said,
May 30, 2013 @ 12:00 pm
I came to post an observation that I see is identical to Bill Walderman's early in the thread, including Mark's addendum that that-deletion is likely due to enthusiasm for "omit needless words." I've been fighting this battle all of my professional (legal writing) career with one colleague in particular who routinely deletes the thats that I've written. Due to strictly enforced word count limits for briefs in most courts, omitting needless words is usually the right thing to do, but avoiding ambiguity that causes even a momentary blip in the busy reader's comprehension is more important, I think.
richard said,
May 30, 2013 @ 12:11 pm
The first sentence appears to me to say "Teachers who are able to teach about nutrition are more likely to teach about nutrition." The second sentence appears to say, "Teachers who know something about nutrition are better able to promote good eating habits than teachers without such knowledge."
It's not the sentences so much as the apparent need to create those sentences that makes me sad.
Howard Oakley said,
May 30, 2013 @ 12:17 pm
Thank you, Mark, for a complete explanation.
Although an adult for far too many years, and sometimes easy to affront when it comes to such suggestions, when they are put tactfully to me, I value them in the long run. I think that SA's colleague might too.
My own style is to use 'that' whenever I can; if I am writing within a tight word limit, though, I may exercise the option to drop a 'that' or two if it helps to bring the piece within my limit.
Howard.
Jonathon Owen said,
May 30, 2013 @ 12:27 pm
I've also worked with some authors who had apparently been taught to omit that whenever possible. It makes for some jarring reading sometimes. I usually end up putting most of them back, and usually the author doesn't notice. But on the other hand, I've known some editors who delete them whenever possible, without regard to how the sentence reads afterwards.
Ginger Yellow said,
May 30, 2013 @ 12:28 pm
@Ted: The passage makes good sense if you take mad to mean "foolish," "irrational," or "crazy" rather than "angry." This is a fair assumption, since the context was a description of British journalistic practice.
The style guide entry in question seems to just pre-date this incident, but it's clearly the sort of thing it has in mind.
Ted said,
May 30, 2013 @ 1:13 pm
@Ginger Yellow: Ah, I see.
In my dialect, the copy editor's behavior was maddening, and it made the reviewer mad.
It may be worth noting, too, that neither of them has clean hands. In my dialect "nosh" is either a verb or a count noun, never a mass noun. On the other hand, "shit" is a noun, not an adjective.
Bill Walderman said,
May 30, 2013 @ 1:48 pm
I wrote earlier: "One of them, if I remember correctly, cited a primary or secondary school teacher as the incontrovertible authority for this. Apparently, there is (or was) a zombie rule about this in circulation among some English teachers."
Of course, I could be erroneously conflating this episode with one of the many instances where someone condescendingly cited their grade-school teacher as irrefutable authority for some other zombie rule, like which/that.
J.W. Brewer said,
May 30, 2013 @ 2:02 pm
Given that these are verbs which can either take a clausal complement (much more commonly with "that") or a direct object (typically w/o "that", but where the object can itself be a fairly lengthy and/or internally complex phrase), I wonder if sometimes the drafter hasn't actually completely thought through whether the rest of the sentence will end up phrased as a clausal complement or direct object by the point in time (i.e. as the verb is being typed) when (subject to subsequent proofreading/editing) the that-or-no-that decision must be made. It's pretty easy to toggle at least some of the examples given above back and forth between these options with fairly minimal wording changes. E.g. "have stated their desire etc." v. "have stated that they desire etc." are both perfectly cromulent, and even if statistically unusual "have stated they desire etc." is not so jarringly odd that I would place high odds on catching and changing it if I were proof-reading my own first draft. The semantic-nuance distinctions between these options are sufficiently subtle that I can't say I'd be strongly precommitted one way or another as I launched into the sentence and I might well change course midstream for some random not-fully conscious reason.
mollymooly said,
May 30, 2013 @ 2:36 pm
Is that-inclusion a US-UK divide analogous to the Oxford/Harvard comma? Hypothesis: US styleguides say "always do it, because it sometimes avoids ambiguity" whereas UK styleguides say, "if and when it avoids ambiguity, then do it."
@kyomi:
I'm not sure that's quite accurate. I would guess that speakers omit "that", and transcribers do not add it as part of the tidy-up. I am aware that official transcriptions do a lot of tidying up (e.g. this) but the Oireachtas style-guide may simply be silent on this particular matter, rather than having a policy of omission.
AntC said,
May 30, 2013 @ 2:39 pm
@Richard It's not the sentences so much as the apparent need to create those sentences that makes me sad.
Hear! Hear!
But look at how much mental processing it needs from the reader to work out how little those sentences are saying!
(In the second example, the referent of they is not clear. Perhaps it's the children being more knowledgeable? But I rejected that reading on the grounds [that] it's even more fatuous than the one you give.)
Ethan said,
May 30, 2013 @ 3:12 pm
@richard: The first sentence is not quite as fatuous if you unwrap the "self-efficacy" jargon. It appears to state [that] teachers who judge themselves competent to teach the subject spend more time doing so. I see less hope for redeeming the second sentence.
Matthew Stuckwisch said,
May 30, 2013 @ 3:12 pm
When "that" is omitted, I get a stronger sense that something is actually being quoted.
John said that it was raining — maybe he actually said it's pouring, or maybe he said "It is raining" and I've applied sequence of tense, etc.
John said "It was raining" — he only said "It was raining" verbatim.
So when I see "John said it was raining", I mentally add in the quotation marks. Looking at the list of words involved in the OP, some of them don't call my attention to such a distinction (like believe) but… maybe I'm the only one who reads like that?
Lazar said,
May 30, 2013 @ 3:50 pm
@Matthew Stuckwisch: No, when I see "John said it was raining", I interpret it to mean that John said, "It's raining." Imagine the syntactically similar "John said he [John] was going to the store" or "You said you were happy" or "You said I was happy" – the relevant quotes have to have been, "I'm going to the store", "I'm happy", "Lazar is happy".
Neal Goldfarb said,
May 30, 2013 @ 4:20 pm
@myl: "I'm reluctant as a matter of principle to assert that this is obviously so, without at least a little experimental evidence, because there's a lot of bad writing advice out there that's based on what some self-appointed authority thinks are obvious effects on readability."
I don't think this falls into the category of bad advice based on the pontifications of self-appointed authorities.
While I'm certainly not a psycholinguist, I'm familiar with some of the literature, and my understanding of it is that it provides good empirically-based reasons to expect that examples like the Investigators believed Jewell sentence will have garden-path effects.
The type of ambiguity at issue here (direct object vs. sentential complement) has been studied by psycholinguists, and while I haven't really looked at it, I'd be willing to bet a beer that there is in fact experimental evidence supporting the conclusion that omitting that will lead to garden-path effects in some nontrivial percentage of cases.
I'd also be willing to bet another beer that it's possible to put together a psycholinguistically-plausible explanation of why examples like the ones originally submitted by S.A. seem awkward.
Nathan Myers said,
May 30, 2013 @ 4:22 pm
Omitting "that" is by far the least of those sentences' problems. Anyone obliged to read more than six in a row might better contemplate suicide.
Rubrick said,
May 30, 2013 @ 4:54 pm
I smell a whiff of Strunkianism in "that"-aversion. While I don't think the Elements of Style specifically recommends against "that" in the situation cited here, it certainly rails against "the fact that" (it took me years to overcome the damage done by that one), and some cognitive spillover wouldn't be surprising.
David Y. said,
May 30, 2013 @ 5:32 pm
Answering an email earlier after I'd read this post, I flinched when I realized I'd written, "This suggests I should…"
Oh, gods, I did it again just above: "realized I'd written."
Am I in the 1% of academics in the minority, or is usage different when suggests/states/indicates/etc. is followed by a pronoun, etc.
[(myl) Usage is certainly somewhat different when the subject of the complement clause is a pronoun: COCA "realized I" 1480 vs. "realized that I" 929, omission 61%; "realized you" 107 vs. "realized that you" 90, omission 54%; "realized he|she" 2790, "realized that he|she" 1388, omission 67%; "realized we" 319 vs. "realized that we" 256, omission 55%; "realized they" 613 vs. "realized that they" 395, omission 61%.
The relatively high omission percentages with pronominal subjects adds plausibility to the avoiding-confusion theory. But the verb also matters: "indicated I" 18 vs. "indicated that I" 53, omission 25%; "stated I" 7 vs. "stated that I" 25, omission 22%.]
Lee B said,
May 30, 2013 @ 6:02 pm
"We're talking about a particular kind of conditional probability here […]. In cases like that, it seems to me that the facts (about how a certain class of people behave) have normative force pretty much by definition, even if the distribution is purely a matter of fashion or convention, it's ipso facto normative, at least in the Horatian sense, isn't it?"
@MYL, I agree that norms arise from usage. But if that isn't a tautology ("parameters describe distributions"), then it's the claim that what we regard as good usage is a function of what other people are doing. But don't you think it would have to be a complicated function, possibly governing some subpopulations and not others, with greater weight on the usage of those more esteemed/knowledgeable/experienced?
For that reason, I think those conditional probabilities aren't what's clinching your analysis; we don't even know whether that 1% is due to the occasional "that"-omission by your typical writer or is due to a subpopulation that consistently spurning the complementizer everyone else loves. Instead it's your personal endorsement of the conclusion that's doing all the work here, the tacit gut-agreement of an academic who writes fluently and is well placed to assess what is the "done thing" in those situations.
For comparison: how much stock would we have put in your conclusion if this blog post gave advice to a Icelandic tabloid journalist based on your analysis of some conditional probabilities in that language? Or an analysis of flickr photographs to decide whether it would be a faux pas (as opposed to unusual) for a little girl to wear a black dress to a friend's quinceañera? Undoubtedly in both cases the norms arise from people observing the usage and behavior of others; but the functions involved must be so complicated as to be totally opaque to an outsider…
(Or maybe someone will surprise me with a link to some google project is already detecting solecisms in Icelandic and faux pas in party attire…)
hanmeng said,
May 30, 2013 @ 6:08 pm
I think … it's OK to leave out that here, although I agree … it's better to leave it in. Anyway, if S.A. feels … it's necessary to instruct the that-omitter, I would suggest an appeal to authority. By citing a former teacher, perhaps?
David Morris said,
May 30, 2013 @ 6:21 pm
Two of my personal rules collide here: 'Keep it short, keep it simple" – which would favour omitting 'that' – and 'Be clear' – which would favour retaining it. I would definitely retain 'that' in these circumstances, and highly urge anyone else to do so as well, for all the reasons that Mark and the other commenters have already given. (Note the 'that' in that sentence? That 'that' was automatic – I didn't notice that I'd typed it until after I'd typed it. Oh, there's another one.)
Lazar said,
May 30, 2013 @ 7:29 pm
@David Morris: It's funny, one of the most common tics that I've noticed in my own writing – in informal online contexts – is a retention of "that" in places where, on reflection, it seems better to omit it.
Jason Eisner said,
May 31, 2013 @ 2:44 am
My guess: "that" is most strongly preferred when it helps to prevent a garden-path effect.
Consider the original examples: "Past studies indicate teachers …" and "D___ and colleagues have stated teachers …" They are arguably hard to read because one could momentarily interpret "teachers" as the direct object of the verb.
Including "that" prevents this temporary misreading. (So long as "that" is not itself incorrectly read as a direct object: "I didn't say that …")
It seems significant that many of MYL's examples where "that" is omitted are not as susceptible to misreading the subject of the clausal complement as a direct object. That subject is an unambiguously nominative pronoun, or expletive "there," or occasionally a single short word like "Iraq" or "BAK1" that is immediately followed by a verb.
I would expect optional "that" to be omitted more often in contexts like these, and also with verbs like "say" for which a direct object is a priori unlikely. Testing this hypothesis would require a bit of corpus analysis. (It is certainly possible to omit "that" in other contexts as well, as shown by MYL's first two examples of clausal complements with no "that." So I am just suggesting a statistical tendency.)
Alex Blaze said,
May 31, 2013 @ 6:18 am
I found this with a little googling, which claims that unnamed "editors" always strike out that that.
http://grammar.ccc.commnet.edu/grammar/conjunctions.htm
Who knows. The problem with S.A.'s colleague's sentences, though, aren't limited to the omitted that's.
zythophile said,
May 31, 2013 @ 7:36 am
I worked under a revise editor on the Times in London who insisted on the need for "that" after every use of "said", "told" and so on, even when there was no possible ambiguity if 'that' was missed out. Since putting all the "thats" in generally made the copy several lines too long, this would irritate me intensely, as good material would now have to be cut to make the copy fit. However, as he was higher in rank than me, I never felt able to argue with him about it.
C said,
May 31, 2013 @ 8:26 am
I still wonder, swimming against the tide here: is anyone really misreading those two original sentences? Is a competent reader not able to understand the content? I don't think either of the original sentences is close to true ambiguity. Other than that the formulation is not nearly as common as formulations using "that," I don't see any reason to force the writer to conform so strictly to the rule.
chh said,
May 31, 2013 @ 11:25 am
@C
Reading isn't just about getting to the end of the passage and being able to figure out what meaning was intended. Sentences can be easier and harder to read. This is why the proposals about garden path effects are relevant.
I doubt the verb 'stated' is very good at conditioning a robust garden path effect, since we don't do a lot of stating people in this world, but I wouldn't be surprised if you could still show slower reading times, more regressions, or some evidence of more difficult reading for sentences with stated+null complementizer in an eye tracking experiment.
@myl and Neal
I'm not familiar with this paper, but the authors find garden path effects conditioned by the absence of 'that', and also to show the kind of subcategorization effect I was thinking of.
chh said,
May 31, 2013 @ 11:26 am
http://www.bcs.rochester.edu/people/mtan/publications/1993Trueswell_JEP.pdf
fiona hanington said,
May 31, 2013 @ 1:22 pm
@richard – You made the point I was thinking of making! Missing "thats" are the least of the barriers to comprehension in those passages.
@ DEP – you're reminding me of that old "Smith where Jones had had had had had had had had had had had had had the headmaster's approval" punctuation challenge!
fiona hanington said,
May 31, 2013 @ 1:25 pm
And I'd be interested to learn more about regional differences in "that" elison. I see it a lot here (western Canada).
Joseph Bottum said,
May 31, 2013 @ 3:22 pm
Maybe it's worth noticing that, in both the 19th-century examples quoted by Prof. Liberman in his post, the sentences contained a prior "that." And the presence of a prior "that," for whatever purpose, puts a mild pressure on a written sentence to avoid a later complementizing "that."
Certainly that's the way I understand it, in my own (admittedly slight and probably mistaken) experience as a writer. Rather than being defined by a hard and simple rule, it seems a matter of pressures on the sentence.
Pressures to use a complementizer "that" in a particular sentence might include clarity of meaning, an easing of the burden on the reader, and a general sense of the eccentricity of obviously avoiding the complementizing "that."
Pressures not to use a complementizer "that" in a particular sentence might include rhythm, a desire to avoid the bloat of a sentence (especially one with parallel elements), and a general dislike of repeating words when not necessary for either clarity or rhetorical flourish.
If I had to guess, Segdwick and Wood are avoiding the complementizer "that" mostly out of a desire to avoid word repetition–and also, possibly, to avoid suggesting to the reader's ear a parallel with the prior use of the word "that."
Sid Smith said,
May 31, 2013 @ 7:35 pm
The style guide at The (London) Times, where I've been a sub editor for many years, suggests that you should favour using 'that'. (I can't quote the guide directly because I don't have access to it here at home.)
I really approve of this advice, to the extent that it's about the only occasion where the guide has influenced my own private writing style.
Ben Hemmens said,
June 1, 2013 @ 6:11 pm
To my mind, both of the examples have other weaknesses that make them hard to follow.
In the first example, there is too much stuff (including another verb) between the subject and the verb, and in the second, the writer has chosen the version that puts the verb immediately after the subject, but at the cost of an unclear antecedent of "they" which breaks up what seems to be the main piece of new information in the sentence – that teachers should know more about nutrition content.
And that's why building sentences is hard to learn: you always have to get several things right at once.
michael farris said,
June 2, 2013 @ 3:17 am
FWIW I had no problem whatsoever in understanding either example sentence and was assuming the complaint was going to be about the content which seemed a little circular.
I have no idea if my natural tendency is to keep or expell 'that' and thinking about it will just muddle things up. I suspect I would normally omit 'that' unless it seems really confusing without or I'm primarily writing for non-native readers of English. This partly might be due to some training in journalism, during which we explicitly told to prune as many words as possible.