After almost a month, I'm finally following up on the results of the single-question surveys that I asked Language Log readers to participate in. Each survey received an overwhelming 1500+ responses, and I didn't realize that I needed a "pro" (= "paid") account on SurveyMonkey in order to view more than the first 100. I owe special thanks to Mohammad Mehdi Etedali, to whom I transfered the surveys and who kindly sent me the overall percentages.
As it happens, the results did not change all that much after the first 100 responses: here are the survey questions and answers, with the percentage of all 1500+ respondents who selected each answer indicated to the left and the percentage of the first 100 respondents who selected each answer indicated in parentheses to the right.
Survey 1. Which of the following three conclusions can you draw from the following statement?
The final exam for that class was deceptively easy.
56.8% The final exam was easy. (60%)
36.0% The final exam was hard. (32%)
7.2% The final exam was neither easy nor hard. (8%)
Survey 2. Which of the following three conclusions can you draw from the following statement?
The final exam for that class was deceptively hard.
11.8% The final exam was easy. (12%)
84.0% The final exam was hard. (84%)
4.2% The final exam was neither easy nor hard. (4%)
Several of you wrote to me to point out that, at least for you, none of the answers provided was really an accurate reflection of your personal intuitions. This led some of you to simply not answer the question, and others to struggle to find what they considered to be the best (or "least bad") answer. I freely admit that there are several methodological problems with these surveys, but I think that the results reveal two interesting things:
- People's intuitions about what deceptively ADJ means are surprisingly variable. (This was pointed out to me by John Hawthorn, and is what originally led me to post the surveys.)
- There appear to be significant differences between people's intuitions about what deceptively ADJ means and what deceptively not-ADJ means, for at least some ADJ/not-ADJ pairs.
With regard to the variability, Simon Spero (whose email subject line is the title of my post here) writes:
The Houghton/Mifflin American Heritage Dictionary notes that for the sentence The pool is deceptively shallow, judgement of their usage panel was 50% for the logical reading, 32% for the opposite reading, with 18% finding it impossible to judge.
(Language Log's own Geoff Nunberg was chair of the usage panel of the AHD at the time, as he discusses toward the end of this Fresh Air piece from October 2003.)
Lauren Gawne also wrote to point me to this post she wrote on the topic (which was the first I'd heard of the excellent Superlinguo blog "for those who like and use language" — check it out). Lauren teases apart three meanings of deceptively there that may account for the observed variability in people's intuitions. I should note that even though my surveys only allowed for one interpretation per respondent, I believe that individuals have some access to the different meanings of deceptively and that these can be pragmatically manipulated. (More fodder for those interested in the methodological problems with my survey.)
As interesting as I think the variability is on its own, I'm even more interested in the stark difference between deceptively easy (56.8% = "easy") and deceptively hard (84% = "hard"). I think that overnegation (discussed a lot here on Language Log) is playing a role here. In response to this conjecture, John Lawler writes:
As for overnegation, note that "hard" but not "easy" triggers NPI [= negative polarity items; this is a common test for negative elements--EB]:
It's hard to believe anybody ever loved this place.
*It's easy to believe anybody ever loved this place.
And while "deceptive" doesn't, it still sets up an alternation between a "perceived" and a "real" mental space, just the way negatives do. (See http://www.umich.edu/~jlawler/negsinspace.pdf for more details.) So I think overnegation of one sort or another is a good candidate.
I'm ill-equipped to delve into this hypothesis deeply enough, both in terms of my available time and in terms of my expertise (I'm a phonologist, after all, not a lexical semanticist). But the idea would be that deceptively easy involves only the alternation between a "perceived" and a "real" mental space set up by deceptively, leading to the ambiguity between the "easy" and "not-easy = hard" interpretations; by contrast, deceptively hard involves both that alternation and the negation inherent to hard, thus… wait, I just lost it. Damn.
And Lauren Gawne writes again:
I've often wondered with these if the pairings change the relationship — for example whether "deceptively hard/easy" are more likely to be analysed with different properties than "deceptively small/big."
Me, too. Sadly, my time and expertise limitations prevent me from pursuing any of this any further. Anyone who desires to do so is free to report back in the comments.