Ghent Word Test.

This test from the Ghent University Center for Reading Research has been making the rounds (I found it via Erin McKean’s Facebook feed), and it’s fun and educational, so I’m passing it on for the general delectation. My results: “You said yes to 93% of the existing words. You said yes to 0% of the nonwords. This gives you a corrected score of 93% – 0% = 93%. You are at the top level!” A warning: they say “Do not say yes to words you do not know, because yes-responses to nonwords are penalized heavily!,” and this so traumatized me I didn’t say yes to what turned out to be perfectly acceptable words like expletively and gauntleted; if I had realized that normal derivational forms were OK, my score would have been higher. (On the other hand, there’s no way I would have said yes to verticillastrate; I still have trouble believing it’s a word.) The last time I checked the MetaFilter thread about it, the highest reported score was 96%. Have fun!

Comments

  1. 88%, after I was dinged 3% for identifying ‘plinker’ as a real word. Plink is a word, in my vocabulary anyway, so I figured a plinker would be, for example, someone in the pub late on a Saturday night playing an untuned piano badly.

  2. 94% – 7% = 88%. I guessed that overlip and backload are words, and I’m pretty sure I’ve heard “backloading” used by real people, but the program said that they aren’t. I Googled for them afterwards and both are in the Wiktionary.

  3. Give them feedback about it; other people have reported similar problems.

  4. John Cowan says:

    87% – 0%, since I followed the instructions and chose “yes” only conservatively. The words I said “no” to were: vinometer, morro, epidiascope, currishly, and crap, a misplaced click and I lost the results page, never to be retrieved. Bad web design.

  5. 99% the second time around. They give you new words. Sort of: I didn’t know ‘tensity’ … twice.

    But I say those of us coming across JAINSTEAK and not incapacitated by laughter deserve extra credit.

  6. David Eddyshaw says:

    Not sure if this is in fact part of the fiendish plan, but the nonwords are of two fairly distinct sorts: words that could perfectly well be English but are (apparently) not, like “saffing”, and words that violate English structure rules and couldn’t possibly be English, so that no native speaker would be fooled by them.

  7. “epidiascope”: one was actually used in my undergraduate lectures. High tech before IT.

  8. It told me “harshing” isn’t a word, which is really harshing my mellow.

  9. OK, now I’m shocked.

  10. I accepted non-word ‘enticting’, probably because I was trying to go too fast and enticed by it.

  11. and was enticed by it.

    See! I’m doing it again.

  12. 96%, with no non-words guessed. Missed copperas, cetin, and bistort. I don’t feel bad about those failures.

  13. 93% less 3% for one non-word. Of real words I missed, “j” was the only one I was upset about, and the non-word was “airbelt” which I would totally use if I ever came across one.

  14. Giacomo Ponzetto says:

    @Glossy: Economists commonly talk of backloading but I believe the recommended spelling is back-loading, so perhaps it’s not considered a word? I’m afraid economists don’t commonly talk to linguists.

    (86% – 0%: I also went paranoid or I’d have accepted addressor, whose meaning is obvious to Romance speakers. All the other ones were down to my genuine ignorance. Yet, as a second-language speaker I’m pretty happy.)

  15. Yet, as a second-language speaker I’m pretty happy.

    As well you should be! That’s considerably better than most native speakers could do.

  16. I legitimately missed osseous (which I should have gotten), epiphysis (which was completely unfamiliar), and extravert (an alternate spelling I wasn’t familiar with, but which had be thinking for a long time). However, I was also marked down for missing “amt,” which I recognized as an abbreviation for “amount” (which is the definition they refer to), but not something I would consider a word in English. Of course, “Amt” is an unremarkable word in German, so I spent a while thinking whether I had ever come across it in English, other than as an abbreviation, and I decided not.

    Finally, most absurdly, I was informed that “stuntless” was not a word.

  17. Jonathan D says:

    For some reason I am pretty terrible at pushing the key I want to push on these sorts of things. Apart from that, I’m not happy with ‘hoi’ and ‘diem’ by themselves as English words.

  18. Trond Engen says:

    11 real words I didn’t know, 2 fake words I thought I knew (hoaxy and wakh). 84% – 7% = 78%. I think I may have been uncharacteristically skeptical to transparent compounds and derivations, having read the thread, but I don’t know if it cost me more in lost hits than it saved me in avoided goofs.

  19. I was hoping that the site gave a definition, or even a hint, as to what “counts” as a real word: do proper names count? Acronyms? Common clipped forms or abbreviations? Slang? It doesn’t say.

    I went conservative the first time and got a 77% (with no false-positives). I went bold the second time and got a 90% (after a 3% penalty for a false positive). I didn’t think to check which “not a word” I got wrong. I didn’t have any repeats, FWIW.

  20. 80% conservatively, 90% throwing caution to the wind.

    I think that the discussion in this thread already highlights a flaw in the study’s design — the assumption that a vocabulary is just a big bag of words, making word-or-not decisions completely black-and-white, rather than a jumble of words and morphemes and combinatory rules, making the issue much more complex, especially in a passive context. (I’d probably never invent the word “bracksmaid”, one of my items, myself, but if I ran across it in a poem it would seem perfectly cromulent to me and I’d assign a meaning to it from context without blinking an eye.)

    Maybe the survey should have three choices: “I’m sure this is an English word”, “I’m sure this is not an English word”, and “I’m not sure but it seems plausible to me.”

  21. First go: 88% minus 3% = 85%, UK-based British monoglot English speaker. The apparent non-word I said yes to and was penaltised for was feal, but in fact it does have a definition as a verb (connected with conceal) and adjective (connected with fealty), see http://www.merriam-webster.com/dictionary/feal , so I protested on the link button and claim 88%…

    The real words I said were non-words were intarsia, agnail, jewelweed, lowery, billbug, granulose, hoppergrass, muskellunge .

    Lowery I dithered about (3.2 secs) and guessed what it would mean if it was real (dark, threatening, lowering sky) but decided it was a fake; semarly with granulose (4.9 secs) , which I guessed was trying to trick me into thinking it was a non-version of granulous or granular but it is its own thing. A hoppergrass is a southrong US version of grasshopper, it turns out. Intarsia, agnail and billbug I didn’t know at all and just guessed (wrong). Jewelweed and muskellunge sounded like made-up sf or fantasy terms so I conjubly rejected them.

    There were a few words I didn’t know but guessed right, ie humidness (as opposed to humidity), metaphase, anionic. Intarsia, anionic… they were 50/50 words and one I guessed right, one I gesstrong.

    I like the way the age selextractor in the profilorm goes up to 125. And starts at 1, too. Apologies for the cutesy neologising… it is infaexsaeous.

  22. John Cowan says:

    Okay, ran it again, this time boldly. I got 99% – 7% = 92%, a little better, but the one rejected word skiver was a typo on my part so I’ll claim 93%. The non-words I accepted were mischant and besee, which both seem plausible, the first (which is also a surname) a little more than the second.

  23. Looking again, the arithmetic doesn’t always add up – I saw my final total was 85%, and saw there had been a 3% deduction for the so-called wrong word I had picked, so I assumed, without really looking closely, that I got 88% before the deduction. But actually, the first result reported was 89%:

    “This gives you a corrected score of 89% – 3% = 85%”

  24. I got 91% and chose “ensheet” as a word, which is apparently not a word, despite the fact that I’ve used it prior to this test!

  25. It reckons ‘chawl’ is not a word but I’ve been to a chawl – it is a type of Indian building. It also denies ‘toolwork’ is a word.

    @Giacomo Ponzetto, Glossy: As an economist, I do use the terms ‘backload’ and ‘frontload’, both unhyphenated.

  26. rewrassent is too a word! I declare it so! It means… it means…, um, brightly colored, like a wrasse?

  27. Trond Engen says:

    I got 86% on the bolder second round, and this time I had a couple of mishits. E.g. I had to think about the number of l’s in ‘traveling’ and hit the wrong button when deciding. But maybe it’s not so much about getting bolder as about getting better at tuning the level of familiarity.

    The percentages don’t add up because the number of words in each category isn’t a divisor of 100.

  28. 97% – 10% = 87%. Three wrong.
    Phonify is a thing, though. I’ve read about phonifying. Ah, well.
    This is better than that Groningen “plug in the cognate” test, which left me completely idea-less.
    I’ll lie to it someday and tell it I’m a Francophone. The Dutch (which I sometimes understand conversationally) is wildly opaque to me, so far. Maybe another day.

  29. Frans Koppenol says:

    94 % first time; did not try again. 0% non-words.
    Good maybe for a non-native speaker, but not so good for an English teacher.
    -
    Cautiousness made me reject “grisaille” which looked like a plausible loan from the French, but I had absolutely no idea what it referred to. Then “inarch” and “puttyroot”, no one uninterested in gardening can be expected to recognize those…
    -Enarch, exarch, autarch, heresiarch… but inarch???
    -
    However I feel ashamed for not recognizing interisland, I missed a hyphen after inter-, and knew for sure there is not some country populated by Interis.
    -
    And indeed, there are many non-words instantly recognizable as impossible: “contaneriian” and “attrechiiteness” have consecutive ies, very strange!!
    If you take away the superfluous ies, they will still be non-words, but more plausible.
    -
    I suppose the test was devised by Belgians…

  30. 71-0=71 and 96-10=86. Three words I did not recognise were north American plants. I also rejected coffer, thinking it was an attempt to fool me into a fake American spelling of cougher. As with many tests, the testee is testing the tester as much as vice versa.

  31. Christophe Strobbe says:

    Ah, a test developed by my former university – though a different faculty.

    @Frans Koppenol: More about the Belgians who developed the test can be found at http://crr.ugent.be/: the “Center for Reading Research is a research group connected to the Department of Experimental Psychology”, etc, etc.

    I took the test and I didn’t know the words hoecake, tommyrot, stogie and conflagrant (although I’m familiar with conflagration). Hence the score 94%-0% = 94%.

    I’m looking forward to similar tests for English and French ;-)

    The test reminded me of Jean Aitchison’s book “Words in the Mind: An Introduction to the Mental Lexicon”, which I started reading a long time ago. Part of the book discusses the size of people’s vocabulary, especially the difficulty of establishing this. If you test people with a bigger sample of words, you end up with a higher estimate (at least, that is how I vaguely remember it).

  32. Christophe Strobbe says:

    Ahem, I meant similar tests for German and French.

  33. Took some twenty tries, but I made it 90% without accepting nonwords. More averagely (that’s a real word?!) I was early eighties with two mistakes.

    Helps that a lot of their vocabulary was chemistryrelated.

  34. Narmitaj,

    “This gives you a corrected score of 89% – 3% = 85%”

    For small values of 89 and big of 3 it works.

    I didn’t count how many words it asks, but looking at the errors, 2×3%=7%, so 3% has to be 3.25-3.75%.

  35. John Cowan says:

    Christophe: I liked it best with “English and French”.

    Sili: Roundof error.

  36. 96% — should have been more, but they counted ‘flong’ as a non-word, when it is the papier-mâché substance used to mould stereotypes

  37. badgerchild says:

    100 percent on the “actual words” and I missed two of the “non-words”, both of which I was certain were actual words, and Google verified the meanings I knew they had. The two words I “missed” were “taggant” and “sackhat”. The first has at least two different meanings in the industry in which I work, and the second is a particular type of knitting pattern that produces a hat with corners, as if you put a bag on your head.

Speak Your Mind

*