Deceptively valuable
« previous post | next post »
A couple of weeks ago, Eric Baković posted about phrases of the form deceptively <ADJECTIVE>, and gave the results of an online survey of more than 1500 LL readers ("Watching the deceptive", 10/2/2011), who were each asked to interpret one of two phrases:
| The exam was deceptively easy. | The exam was deceptively hard | |||
| The exam was easy. | 56.8% | The exam was easy. | 11.8% | |
| The exam was hard. | 36.0% | The exam was hard. | 84.0% | |
| The exam was neither. | 7.2% | The exam was neither. | 4.2% | |
Eric suggested that this variability in judgments, and also the asymmetry between easy and hard, might be connected to the phenomenon of misnegation. And there were many other interesting observations and speculations in Eric's post and the 64 comments on it. But a simple tally of collocational frequency for the word deceptively suggests a couple of relevant factors that neither Eric nor any of the commenters noticed.
In the Google Books ngram dataset, the 100 most common adjectival values for X in deceptively X were (in descending order of frequency):
simple easy mild similar small casual complex low quiet difficult innocent benign soft gentle normal straightforward peaceful strong large high smooth bland clear sweet attractive powerful fragile modest beautiful familiar lazy placid pleasant short innocuous slow serene delicate tranquil ordinary dangerous obvious simplistic warm naive friendly complicated good frail subtle harmless sleepy hard quick thin solid neat healthy young slim transparent charming unassuming potent slight rich demure easygoing tough bright seductive tricky childlike positive neutral narrow natural modern realistic humble languid safe real brief heavy cheerful swift trivial stable clean alluring youthful slender reasonable lovely uncomplicated effective effortless optimistic boyish
Among the emergent properties of the terms on this list is the fact that nearly all of them are positively evaluated. We can see this from another angle by looking at the counts for some (approximate) antonyms:
| good | bad | innocent | guilty | ||
| deceptively ___ | 207 | 0 | 738 | 0 |
Or
| sweet | sour | quiet | loud | ||
| deceptively ___ | 240 | 0 | 815 | 0 |
And X=evil, villainous, bitter, noisy also have counts of 0.
Another aspect of this list that may be less obvious is what we might call the "milquetoast factor" or the "sandbagging dimension": words like bland and frail, in many contexts at least, might be less positively evaluated than quasi-antonyms like spicy and robust, and yet we have:
| bland | spicy | frail | robust | ||
| deceptively ___ | 308 | 0 | 146 | 0 |
Patterns like the following appear to have a similar source:
| cool | cold | warm | hot | ||
| deceptively ___ | 184 | 0 | 146 | 0 |
It's likely that these zeros are either floor effects, or artefacts of the fact that the Google n-gram collection discards less common n-grams (for n>1). Thus a general web search turns up quite a few hits for e.g. "deceptively spicy" or "deceptively cold", and so does Google Books. And we can't explain these effects in terms of the overall frequency of the adjectives in question, since cold is commoner than cool in this dataset (13.6 million vs. 4.1 million), hot is commoner than warm (9.8 million to 7.3 million), etc.
But still, deceptively does in general appear to collocate with adjectives that are metaphorically (and perhaps physiologically) mild as well as emotionally positive (or at least non-negative). This remains true in the face of some interesting exceptions, however:
| benign | safe | dangerous | deadly | poisonous | ||
| deceptively ___ | 557 | 63 | 147 | 0 | 0 |
The case that Eric used in his survey — easy vs. hard — may not be as much of an exception as it seems. Simple and easy are among deceptively's commonest collocates, and so the fact that their opposites also have non-zero counts in the n-gram corpus may just be the expected luck of the draw:
| simple | easy | difficult | hard | complex | ||
| deceptively ___ | 27,204 | 2,653 | 548 | 88 | 625 |
Of course, none of this changes the observation that deceptively <ADJECTIVE> can work out pragmatically in two quite different ways — things that are deceptively ADJECTIVE can be more ADJECTIVE than they seem, or less. Some corpus examples where things are more ADJECTIVE than you might think:
It may sound fancy, but this recipe is deceptively simple to prepare.
Hit the holiday party scene rocking one of these attention-stealing, deceptively easy looks and you'll be the girl giving everyone hair envy — guaranteed.
Getting around on Ilmatar was deceptively easy: take a bearing by inertial compass, point the impeller in the right direction, and off you go.
Here are 10 deceptively simple ideas for building your human connections that have helped employees of the companies that engage us for leadership coaching get to a whole new level of high performance.
Here are some others where things are less ADJECTIVE than they seem:
Flintknapping is a complex craft that appears deceptively simple to the inexperienced eye.
Like baking a pie, frying a chicken is a deceptively simple kitchen task with fearsome pitfalls.
Walt's arm moved in a deceptively easy movement and the big man staggered back.
Spectra that appear to be first order, but actually are not, are called deceptively simple spectra.
But if it's generally true that the oddball un-mild collocates like hard, dangerous, long, etc., tend more strongly to the "more X than it seems" interpretation than the typical collocates like easy, safe, short do — as Eric's survey results suggest — this may be connected to the fact that deceptively seems to have a preferred orientation with respect to such scales in general.
[I used Mark Davies' interface to the Google Books n-gram collection to get most of the numbers cited above.]
Spell Me Jeff said,
October 16, 2011 @ 10:21 am
To the logical side of my brain, this idiom as it is commonly used strikes me as odd. At first glance, the only example above that makes sense to my ears is this one:
"Walt's arm moved in a deceptively easy movement and the big man staggered back."
The point, I think, is not that Walt's motion proves to be NOT easy, but that its consequence proves to be more powerful that its easiness would suggest. In other words, we have to introduce a Y element for the idiom to "work."
You might likewise say something like:
"Alex is deceptively short. I've seen her drive a ball further than men a head taller."
All of which is to say that if the idiom is used logically, it might take this form:
"A is deceptively X. A is in fact more/less Y than the X-ness implies."
Of course, we all know how weak a link there is between logic and actual speech. And even from a logical perspective, my take on the matter could be very idiosyncratic.
Marc said,
October 16, 2011 @ 12:15 pm
Deceptive is deceptively deceptive.
Marc said,
October 16, 2011 @ 12:17 pm
*Deceptively
Oskar said,
October 16, 2011 @ 12:29 pm
I'm a non-native speaker of English (I didn't participate in the survey, following Eric's instructions), and I can actually remember precisely the first time I ever heard the word "deceptively". It was from an episode of Frasier, when Frasier and his dad was talking to each other. With some googling I found the exact exchange:
The the next couple of times I heard the word was in the same context, the context of art. I had an English teacher who described the poem "The Sick Rose" by William Blake as deceptively simple, meaning "it looks simple on the surface but actually hides deep complexity".
So therefore I've always interpreted "X is deceptively Y" as "X appears to be Y, but is actually non-Y". I always thought this made sense based on the word "deceptively", interpreting the sentence as "X appears to be Y, but this is a deception, it is actually non-Y".
So my instinct when reading "The test was deceptively easy" was "The test looked easy but was actually hard". I guess this puts me in the minority of speakers. I guess I'm not quite as natural with the English language as I flatter myself to be.
Kylopod said,
October 16, 2011 @ 12:34 pm
I apologize if you've covered this already, but do you take into account variations where it uses the adjective deceptive rather than the adverb deceptively? I'm sure you'll find examples like "handled with deceptive ease."
[(myl) Not something I looked at — go for it!]
teucer said,
October 16, 2011 @ 1:19 pm
The deceptively simple recipe in the first example and the deceptively simple flintknapping are actually both simpler than they "should" be, rather than contrasting. The recipe is simple and easy, but fancy; the flintknapping is simple in principle but definitely not easy.
I think these examples are all consistent with my and several other commenters' assertion that "deceptively X" means "X, in a way that is deceptive about some of its other features;" although the recipe is better explained as its other features deceiving you into not expecting the simplicity, it also makes perfect sense for it to mean that the recipe itself is so simple you wouldn't expect it to come out so fancy. The easy interpersonal action is clearly of this form – it's so easy you wouldn't expect it to work, but hey, give it a try, it's great. The only one of these that doesn't readily take such an understanding is the hairdo, for which I personally would have said "unexpectedly" instead of "deceptively."
And in fact, frying a chicken *is* simple – it's just that the difference between simplicity and ease will trip you up, so how simple the instructions are is deceptive. This is, in my opinion, the normal pattern – the chicken is simple to prepare, but this fact will mislead you about other aspects. The ease with which Walt moved his arm would lead you to expect there was less force behind it. And even the spectra (which I don't really understand but can interpret from context) are presumably simple-looking on their face, but aren't actually first-order spectra, requiring a more complex interpretation than you would expect from their straightforward appearance.
[(myl) This might be true. But if "deceptively X" can mean "seems to be X, but in fact is not X", as seems to be the case for some of these examples (e.g. "deceptive simple spectra"), some additional contortions will be needed to get things straightened out pragmatically. In any case, the main point of this post is the collocational preference of deceptively X for values of X that are (in some sense) mild and positive; I should think that your interpretation would be symmetric with respect to these dimensions. An interpretation that involves setting a one-sided bound on a scalar quantity seems to offer a bit more leverage.]
Henning Makholm said,
October 16, 2011 @ 2:12 pm
A different hypothesis: Suppose that when the "deceptively ADJECTIVE" construction was originally used as a predicative, "deceptively" logically modifies the linking verb rather than the adjective. So "It looks deceptively simple" means that it looks simple, but the looks deceive because it is not really simple. In such examples I don't think there is any real ambiguity. It doesn't mean "It looks like it is deceptively simple".
This model fails to provide a sensible meaning when the linking verb is the neutral "to be" — however, since the construction syntactically ought to fit with "to be", language users will naturally try to extrapolate a meaning to go with that context. And these extrapolations sometimes end up contradictory.
[(myl) This is a plausible idea, though you'd think, if it were true, that the pre-verbal position ("??it deceptively looks simple") would be better than it is, along the lines of "it definitely looks simple", "it always looks great", "it hardly looks different", etc. In any case, this doesn't explain why depectively X is mainly used with mild and positive values of X.]
Rubrick said,
October 16, 2011 @ 3:33 pm
The more of these I read, the odder "deceptively X" seems.
If I say "her kindness was deceptive", the implication is that she appears kind but actually has a secret mean streak. But "she was deceptively kind" implies the opposite (at least for most people): that her gruff manner conceals an inner kindness.
In other words, "deceptive" ~= "misleading", but "deceptively" ~= "suprisingly".
I of course shouldn't be surprised that an adjective and its corresponding adverb would wander apart.
[(myl) I don't think that all cases work this way. In fact, your proposed interpretation of "deceptively kind" seems to be the exception rather than the rule. Consider these Google Books examples for "that phrase":
Some pieces are marred with bloom — a deceptively kind term for what looks like mildew
… said in a deceptively kind voice (deceptive because, of course, Grandmère isn't kind) …
“It's good that you feel you can handle this,” JB said, his voice deceptively kind and soothing.
About 90 percent of all skin cancers arise from exposure to the deceptively kind, warm rays of the sun …
]
Rubrick said,
October 16, 2011 @ 3:50 pm
Catching up on some recent comments, I agree that the most often sensible interpretation of "X is deceptively simple" is "X is obviously simple, and this fact conceals surprising complexity elsewhere". The most canonical example I can think of: "The rules of Go are deceptively simple".
However, this interpretation doesn't seem to work for most adjectives other than "simple". "He was deceptively intelligent" certainly doesn't hint at a hidden stupidity.
Language is hard. Let's do math.
Aaron Toivo said,
October 16, 2011 @ 5:10 pm
This might explain why I had such a strong instinct that "deceptively easy" and "deceptively hard" both mean the same thing even though I couldn't figure out how they could. Thanks!
Dan Hemmens said,
October 16, 2011 @ 5:47 pm
This might be true. But if "deceptively X" can mean "seems to be X, but in fact is not X", as seems to be the case for some of these examples (e.g. "deceptive simple spectra"), some additional contortions will be needed to get things straightened out pragmatically.
I'm not sure "deceptively simple spectra" is a good example, because it seems to be describing a piece of technical terminology (that is, a deceptively simple spectrum is a specific type of spectrum) – I'm not a linguist, but I didn't think technical terminology was generally counted in this kind of situation. I agree with teucer's analysis of the other examples.
dw said,
October 16, 2011 @ 5:55 pm
My two cents:
"Deceptively" is currently straddling the fence that "hopefully" has just about cleared, and "apparently" vaulted a long time ago.
"The exam was apparently easy" doesn't mean "The exam was easy in an apparent manner", but something like "It appeared that the exam was easy". It doesn't imply that the exam _actually was_ easy: in fact there is a suggestion of the opposite.
I think that my internal grammar treats "deceptively" like "apparently", with a much stronger implication that the appearance is deceptive. So, for me, "The exam was deceptively easy" strongly implies that the exam was actually hard.
However, many other commenters here treat "deceptively" like a typical adverb. For these speakers, "The exam was deceptively easy" simply means something like "The exam was easy, in a deceptive manner".
No doubt there will be "deceptively" wars in the future, just as there are "hopefully" wars now.
GeorgeW said,
October 16, 2011 @ 6:49 pm
I wonder if the "milquetoast factor" has something to do with deception being anticipated more of negative behavior than positive.
'Hitler was deceptively kind' makes more sense to me than 'Mother Teresa was deceptively evil.'
Charidan said,
October 16, 2011 @ 9:12 pm
"simple spectra" is definitely a dangerous example because there is ambiguity as to whether "simple spectra" is a single term or is the technical noun "spectra" modified by the adjective "simple" which drastically alters how "deceptively" would apply to it/them.
I also believe that there is a meaning of "deceptively X" (perhaps a lesser-used one) that X is primarily easy which serves to hide some amount of not-X, and often a extreme form of not-X. Therefore, "The test was deceptively easy" would mean that for the most part the test really was easy, except for a few hidden problems that were nigh-impossible (or maybe just really hard; up for interpretation). I would also get the implication that these hard problems would be worth most of the points on the test, but that probably comes more from my knowledge of how to devise evil tests rather than the sentence itself.
Danny said,
October 16, 2011 @ 9:58 pm
Admitting that I'm a little lost in some of the nuances of these posts, I definitely feel that the 'exam' part of the question is causing the disparity. (I'm aware other commenters have touched on this, but please bear with me.)
Since the difficulty of an exam for most people equates to what result they would be able to get on it, "deceptively hard" seems to mean to the majority "hard but looked easy". In contrast, "deceptively easy" begins by meaning "easy but looked hard," but is modified in context by the fact that *looking hard* increases your chances of getting part of the test wrong, even if with hindsight you shouldn't have. Which means that the exam could be considered harder.
The obvious example to me is the type of trick question that begins, "You're a zookeeper, […complex extraneous detail about numbers of animals, visitors and feeding times…] What colour are the zookeeper's eyes?" This is a deceptively hard question, and most people are familiar with getting it wrong the first time they hear it. It is, therefore, an easy question that would nonetheless increase the difficulty of an exam (by being opaque to some people).
Substitute "the *dance move* was deceptively easy/hard" and my guess is that the results would be much more balanced (and indeed, resemble the existing result for "the exam was deceptively hard" on both sides.
But, as other people have said, the more I think about it, the less sure I am. Deceptive…
J.W. Brewer said,
October 16, 2011 @ 10:14 pm
It's certainly more milquetoast than "deceptively evil," but the collocation "deceptively gruff" is not hard to find, where the notion is that the person is nicer/softer/sweeter than first impressions would suggest. OTOH, "gruff" might in an odd way be thought itself to be a "positive" adjective insofar as it's almost a euphemism – sort of the least negative way to characterize an affect or set of behaviors that could easily be described in more pejorative terms.
Jerry Friedman said,
October 17, 2011 @ 12:01 am
Thanks for this interesting post!
The Ilmatar example is from an SF story by James Cambias, and "deceptively easy" there seems to mean "not really easy". It continues, "But occasionally Ron found himself thinking about just how hard it would be to navigate without electronic help. […] A human without navigation equipment on Ilmatar would be blind, deaf, and completely lost." (And does that happen, predictably? The page I linked to has only the beginning of the story. I must have read it, but if so I've forgotten it completely.)
The other three examples of deceptively <ADJECTIVE> meaning "more ADJECTIVE than you might think" are all cases where the deceiver is the person whose point of view we're looking from. "The recipe is deceptively simple"—you, reader, will deceive your guests. And conversely: "Flintknapping […] appears deceptively simple"—you, reader, would soon find out that it's more complex than you thought.
I wonder how good or bad, tough or effete, and more or less than you'd think correlate with whether you are the deceiver or the dupe.
Speaking of the flintknapping sentence, I'd venture to guess that appears deceptively <ADJECTIVE> almost always means "is less ADJECTIVE than you might think", and the same with seems, etc. That's true of the first two pages of GB hits on "appears deceptively", with one odd exception. I think this is something Henning Makholm was presupposing.
By the way, I was surprised that "complex" was more common than "complicated" on your list of collocates, and then more surprised to find at Google ngrams that the two words tracked each other till 1900 or so, and since then "complex" has increased to be four times as common as "complicated". Without my ever noticing.
If this were Improve Our Prose Log, I'd mention that "superficially" doesn't seem to be "skunked".
maidhc said,
October 17, 2011 @ 12:52 am
This discussion has convinced me that I should never again use the word "deceptively".
Ben Hemmens said,
October 17, 2011 @ 6:13 am
Looking at the quiz questions for more than a few seconds made me non-deceptively confused and I chickened out of answering them!
Dan H said,
October 17, 2011 @ 8:30 am
By the way, I was surprised that "complex" was more common than "complicated" on your list of collocates, and then more surprised to find at Google ngrams that the two words tracked each other till 1900 or so, and since then "complex" has increased to be four times as common as "complicated". Without my ever noticing.
Not sure what's being tracked here, but might that have something to do with "complex" – as well as being a near-synonym for "complicated" – having several other uses, particularly uses like "apartment complex" "cinema complex" or "shopping complex" which might have become significantly more common after 1900.
[(myl) You're certainly right about the nominal uses of complex — but it does seem to be true that adjectival complex has been gaining mindshare at the expense of complicated over the past half-century or so (data from the Google Books ngram dataset):
]
Boris said,
October 17, 2011 @ 9:08 am
DW: "apparently easy" does not seem acceptable in my internal grammar, at least not with "apparently" modifying "easy". If you punctuate it as "the exam was, apparently, easy" the phrase has only one meaning, that to the best of the speaker's knowledge, the exam was easy, with an implication that the speaker has no first-hand information about the exam. I can just about picture something like "the exam was, apparently, easy, but I couldn't answer a single question" which could almost mean what you say it would, but I think the actual meaning here is "The exam was easy statistically as reported by others, but I found it hard", so not really the same as deceptively easy.
beautiful familiar lazy placid pleasant short | Theresa Anderson Art said,
October 17, 2011 @ 9:20 am
[…] an excerpt from the language log at http://languagelog.ldc.upenn.edu/nll/?p=3500 […]
Tony said,
October 17, 2011 @ 12:19 pm
I wonder what proportion of people reading only the title of this post would think it was about valuable or useless data?
Though I thought I was staunchly in the "=more X than first assumed" camp, I have to admit I wouldn't have thought anything was amiss if this post was instead titled "deceptively useless". Or would I?
Dan Hemmens said,
October 17, 2011 @ 5:44 pm
You're certainly right about the nominal uses of complex — but it does seem to be true that adjectival complex has been gaining mindshare at the expense of complicated over the past half-century or so
Interesting!
I confess that I might be using the ngram viewer wrong, but something that might be worth considering is that while complex has the lead in a lot of situations, "it's complicated" is more common than "it's complex" and "really complicated" is more common than "really complex" (with a sharp upward trend in "really complicated" since perhaps 1990).
Of course I have no idea what that might *mean* but my unprofessional instinct is that the terms which give "complicated" the lead are less formal, which might explain why the result is so unintuitive. Instinctively, I'd assume that "complicated" is more commonly used than "complex", but I'm only considering which I'd be more likely to use in everyday speech, which would skew towards less formal constructions. This of course might be nonsense.
Greg said,
October 18, 2011 @ 8:57 am
It's likely that these zeros are either floor effects, or artefacts of the fact that the Google n-gram collection discards less common n-grams (for n>1). Thus a general web search turns up quite a few hits for e.g. "deceptively spicy" or "deceptively cold", and so does Google Books. And we can't explain these effects in terms of the overall frequency of the adjectives in question, since cold is commoner than cool in this dataset (13.6 million vs. 4.1 million), hot is commoner than warm (9.8 million to 7.3 million), etc.
What was your Mutual Information parameter when using Mark Davies' interface? I don't have much experience with the Google Books one, but I use COCA quite frequently, and the default MI is 3. Anyway, if it was greater than 0, that may explain why something like "cold" may not have come up as a collocate for "deceptively" even though the phrase "deceptively cold" may itself exist in the corpus.
[(myl) The numbers in question don't turn up in a search for collocates, but in a simple and direct search for the word sequences (e.g. "deceptively cold") themselves.]
Andrew (not the same one) said,
October 18, 2011 @ 9:48 am
I just heard an example of 'deceptively X' used to mean 'X, but not Y, as that would lead you to expect' used in the wild – 'deceptively sunny' with the implication 'but not warm'. I don't think there's any doubt 'deceptively X' can mean this. I would suppose that we start looking for other interpretations in cases where there is no obvious Y.
Jerry Friedman said,
October 18, 2011 @ 10:15 pm
@Dan Hemmens: Some results (knowing nothing about the Mutual Information parameter) from COCA, which covers 1990–2011:
Written corpus only:
<ADVERB> complicated: 4638
<ADVERB> complex: 6863
Spoken corpus only:
<ADVERB> complicated: 1780
<ADVERB> complex: 886
So there's definitely some correlation favoring complex in writing and complicated in speech.
Deceptively Cold Means It’s Warm? | danielmedley.com said,
October 8, 2012 @ 5:04 pm
[…] the use of “deceptively” such as deceptively cold, or deceptively large, etc. For example this site points out that when using the phrase, "The exam was deceptively easy," 56.8% of those […]