From the perspective of the psychology memory and knowledge, a hallmark of a great crossword clue is one that is obvious once you see the answer, but difficult or impossible if you see just the clue. This highlights how our knowledge is both associative and content-addressable. What this means is that we associate some concepts, words, phrases, etc. with other concepts, words, phrases, but not always in both directions. A trick constructors use is to look for the second or third meaning of a clue, to send the solver down a garden path.
This week, I saw some great clues that worked like this, at least for me. For example, in Sunday March 16 NYT puzzle called ITS BETTER THIS WAY (by Jeremy Newton), there was: (spoilers below)
73D They're 18 to 21 _ _ _ _
This was a real stumper, because it leads you to think of college-age young adults. Of course, nothing with five letters works to describe these folks. I needed to get 3/4 of the crossing words to figure it out (a miserable failure), to discover that the number of letters in the answer and the number of numbers between 18 and 21 should have been a good hint to the answer: RSTU.
Another great clue was:
31D Scratch, say. _ _ _ _
This takes some thinking. Here, scratch has at least three meanings that are verbs-- of dealing with an itch, pulling an athlete out of competition, or damaging something, or scrawling something, and so on. Words like PULL seem too remote, ITCH might have worked because people often say 'I need to itch my back', but this is improper and so it should have had a ? at the end. I think I've see scratch clue the answer MAR (a verb, but it could be a noun). This might lead you to MARK, which also isn't right, but is also both a verb and a noun. Only after revisiting it with F - - W, as I able to rethink of scratch as a noun rather than a verb and get FLAW. So, there are two things that make this clue great; I got lead into thinking of the verb rather than the noun, which makes FLAW difficult to access. Also, it is absolutely true that a scratch is a flaw; but I think the connection is much stronger from FLAW to SCRATCH than from SCRATCH to FLAW.
This is somewhat backed up by Doug Nelson's free association norms. Here is the distribution of first associates to SCRATCH:
SCRATCH FSG BSG MSG OSG QSS TSS QFR TFR QMC TMC QUC TUC
ITCH 0.840.3590.0000.0004 4 13 5 9 0.50 1.67 1 1
SCRIBBLE 0.080.0000.0000.0000 16 13 1 9 1.20 1.67 0 1
DENT 0.080.0340.0000.0000 18 13 2 9 1.20 1.67 0 1
CLAW 0.060.0140.0010.0073 15 13 1 9 2.21 1.67 1 1
WOOL 0.050.0000.0240.0104 16 13 10 9 2.25 1.67 1 1
SANDPAPER 0.050.0000.0140.0000 11 13 1 9 1.10 1.67 1 1
FINGERNAIL 0.030.0000.0000.0014 15 13 2 9 1.47 1.67 1 1
ROUGH 0.030.0000.0020.0007 17 13 41 9 1.85 1.67 1 1
SCAR 0.020.0000.0010.0069 26 13 10 9 2.82 1.67 1 1
BRISTLE 0.020.0000.0010.0000 12 13 3 9 0.92 1.67 1 1
WHISKERS 0.020.0000.0000.0070 8 13 3 9 1.75 1.67 1 1
CUT 0.020.0480.0000.0065 21 13 192 9 1.35 1.67 1 1
PAW 0.020.0000.0030.0068 13 13 3 9 2.62 1.67 1 1
RASH 0.020.0000.1860.0789 18 13 1 9 1.56 1.67 1 1
RUB 0.020.0000.0000.0000 23 13 6 9 1.40 1.67 1 1
SCRAP 0.010.0000.0000.0017 20 13 8 9 1.44 1.67 1 1
BANDAID 0.010.0000.0080.0209 14 13 0 9 3.75 1.67 1 1
BELLY 0.010.0000.0000.0000 12 13 23 9 0.50 1.67 1 1
IRRITATE 0.010.0000.0410.0172 12 13 1 9 2.33 1.67 1 1
PICK 0.010.0000.0000.0000 13 13 55 9 0.50 1.67 1 1
SCRUB 0.010.0000.0000.0007 19 13 9 9 2.06 1.67 1 1
BRUISE 0.010.0000.0010.0110 20 13 3 9 2.95 1.67 1 1
DANDRUFF 0.010.0000.0110.0047 13 13 0 9 1.62 1.67 1 1
FLEA 0.010.0000.0320.0153 11 13 2 9 1.82 1.67 1 1
SNIFF 0.010.1860.0000.0000 12 13 2 9 1.40 1.67 0 1
ICE 0.010.0000.0000.0000 18 13 45 9 1.29 1.67 1 1
Of course, FLAW never appears, with high associates like ITCH and SCRIBBLE. However, DENT is also fairly high, which is at least somewhat similar to FLAW.
This is the the essence of a great clue, and it has a lot to do with how we tend to associate words together.
Nice work, Jeremy!
No comments:
Post a Comment