MW CAyA example

[funderburkjim]

In the recent batch of Correction Form corrections, one in particular suggested further research.

The correction was to CAyA in MW.

Before correction, here is the Basic Display :

(H1B) छाया a [p= 406,1] [L=75839]   f. σκιά, shade, shadow, a shady place (" a covered place, house " Naigh.  iii, 4) RV.  i, 73, 8 ; ii, 33, 6 ; vi, 16, 38 AV.  VS.  v, xv AitBr.  vii, 12 ṠBr.  &c
(H2) छाया b [p= 406,1] [L=75859]    » °य.

And After correction:

H1B) छाया a [p= 406,1] [L=75839]    f. σκιά, shade, shadow, a shady place (" a covered place, house " Naigh.  iii, 4) RV.  i, 73, 8 ; ii, 33, 6 ; vi, 16, 38 AV.  VS.  v, xv AitBr.  vii, 12 ṠBr.  &c
>>> and all of these <<<
(H1B) छाया [p= 406,1] [L=75840] f. the shadow of a gnomon Sūryas. 
(H1B) छाया [p= 406,1] [L=75841] f. shelter, protection, Hit.  iii, 8 ,1/2
(H1B) छाया [p= 406,1] [L=75842] f. a reflected image, reflection RV.  v, 44, 6 ; x, 121, 2 VS.  ii, 8 AV.  v, 21, 8 PraṡnUp.  Mn.  &c
(H1B) छाया [p= 406,1] [L=75843] f. shading or blending of colours, play of light or colours, lustre, light, colour, colour of the face, complexion, features Suṡr.  VarBṛS.  lxviii, 89 ff. Ragh.  iv, 5 Megh.  (ifc. f(आ).) &c
(H1B) छाया [p= 406,1] [L=75844] f. gracefulness, beauty, 77 & 101 VP.  iv, 4, 31 Kathās.  iic
(H1B) छाया [p= 406,1] [L=75845] f. a series, multitude (पङ्क्ति) Pañcat.  i, 16, 8
(H1B) छाया [p= 406,1] [L=75846] f. a Sanskrit gloss on a Prakrit text
(H1B) छाया [p= 406,1] [L=75847] f. a copy (of a MS.)
(H1B) छाया [p= 406,1] [L=75848] f. a little (ifc.) Veṇis.  vi, 13/14, 1
(H1B) छाया [p= 406,1] [L=75849] f. nightmare, Buddh.  L. 
(H1B) छाया [p= 406,1] [L=75850] f. a bribe L. 
(H1B) छाया [p= 406,1] [L=75851] f. " Shadow ", (like संज्ञा) wife of the sun and mother of the planet Saturn Hariv.  545 ff. VP.  iii, 2 BhP.  vi, viii MatsyaP.  Kathās.  cv
(H1B) छाया [p= 406,1] [L=75852] f. (N. of a शक्ति) Hcat.  i, 5, 197
(H1B) छाया [p= 406,1] [L=75853] f. the sun L. 
(H1B) छाया [p= 406,1] [L=75854] f. a metre of 4 x 19 syllables
(H1B) छाया [p= 406,1] [L=75855] f. a kind of rhetorical figure Sarasv.  ii, 5
(H1B) छाया [p= 406,1] [L=75856] f. (in music) N. of a राग
(H1B) छाया [p= 406,1] [L=75857] f. N. of कात्यायनी (or दुर्गा W. ) L. 
(H2) छाया b [p= 406,1] [L=75859]

The problem was that all of the records from L=75840 to 75859 were marked as CAya, n.:

H1B) छाय [p= 406,1] [L=75840]   n. the shadow of a gnomon Sūryas. 
(H1B) छाय [p= 406,1] [L=75841]  n. shelter, protection, Hit.  iii, 8 ,1/2
 ETC

[funderburkjim]

I am sure that there are other such cases lurking within MW.

It is likely that a program can winnow the MW records to a manageable set of cases.

The algorithm might be something like:

• search for an Hx that is a nominal, and that has HxB sub-records.
• Assume W is the headword (key1)
• check that the Hx record is m or n or mfn for its normalized lex.
• check that there is an HxB record which is f, keyword spelled Y
• check that there is a subsequent HxB (or HxA) record Z which is
  • not f for gender
  • with spelling W

Then record Z is a candidate that might be wrong.

Hard to estimate how many false positives there might be.

[gasyoun]

Hard to estimate how many false positives there might be.

But the algo makes sense.