Consistency in <hom> tagging

[Andhrabharati]

1. In general, the <hom> number (if not with a letter) is followed by a dot and the <hom> letter is without a dot. But the odd-mans are--

<hom>1.</hom> 4706 <hom>1</hom> 70 <hom>2.</hom> 4362 <hom>2</hom> 59 <hom>3.</hom> 882 <hom>3</hom> 16 <hom>4.</hom> 226 <hom>4</hom> 3 <hom>5.</hom> 126 <hom>5</hom> 12 <hom>6.</hom> 16 <hom>6</hom> 1

<hom>a</hom> 5179 <hom>a.</hom> 62 <hom>b</hom> 4963 <hom>b.</hom> 55 <hom>c</hom> 144 <hom>c.</hom> 6

2. And the <hom> tag placements wrt the Skt. word are--

"<hom>[0-9](.*?)</hom> <s>" 9579 "<hom>[0-9](.*?)</hom><s>" 1 "</s> <hom>[0-9]" 227

"</s> <hom>[a-z]" 10317 "<hom>[a-z](.*?)> <s>" 1788 "<hom>[a-z](.*?)><s>" 4

Should these not be made consistent throughout? [The <hom> number to precede the <s> word (with a space) & the <hom> letter to follow the <s> word (with a space).]

--------------- [Side-note: There are 11 "<s>" and 1 "</s>" occurrences.]

[Andhrabharati]

One interesting observation:

The <hom>letter tags 'a', 'b' etc. are introduced additionally in the text, which are NOT in the print, for some reason.

As a minimum, the 'a' and 'b' counts should be equal, i.e. every 'a' should have associated 'b' somewhere in the text; otherwise having an isolated 'a' has no meaning. (Having lesser numbers for the next letters is understandable.)

But the counts are different-

Resolving this, to achieve no count differences would be a good exercise indeed. -------------------------- And, the same condition (equal numbers) applies to <hom>1 (total counts: and <hom>2 as well, the counts being as below-

[funderburkjim]

some progress

Some of the inconsistencies pointed out above have been addressed.

(158) <hom>n</hom> -> <hom>n.</hom>  (n a digit, 1 to 6)
(123) <hom>n\.</hom> -> <hom>n</hom>  (n a lower-case letter , [a-c]
(   1) </hom><s> -> </hom> <s>
( 31) in the 'headline' (line after metaline), numeric hom precedes headword
   <s>X</s> <hom>n.</hom> ¦   -> <hom>n.</hom> <s>X</s>  ¦ 
( 15) numeric hom to precede Sanskrit after = 
   = <s>X</s> <hom>n.</hom>   -> = <hom>n.</hom> <s>X</s>
NOTE:  There remain 47 cases of form '<s>X</s> <hom>n.</hom>', but these
   require no changes since the form is always
   '<s>X</s> <hom>n.</hom> <s>Y</s>'

number-letter homs

There were several (about 140) cases of either <h>[0-9][a-z] or `[0-9][a-z]' After changes there remain 66 of each.

[funderburkjim]

homonyms in the list display

The dictionary identifies for about 5700 entries (according to mw.txt digitization). Sometimes the homonym variants appear 'next to' or 'near' each other in the dictionary ordering. But often they appear some distance apart, and it is sometimes useful to navigate among the homonyms in the list display to see the dictionary context of the variants. The 'arrows' in the list display provide this functionality. For instance, clicking on the 'yellow' arrow will change the left-hand list pane to be centered at that second homonym

[funderburkjim]

why 'letter' homonyms ?

In MW, a given headword can appear at different dictionary locations, yet these different locations are not identified by the author as homonym variants.

Letter homonyms were 'invented' by Peter Scharf to permit the list display navigation to these otherwise unmarked entries. Since MW's printed homonym codes are numeric, there is no confusion between MW's printed homonyms and the synthetic letter homonyms.

The list display homonym navigation feature uses only the metaline <h>X values. The <hom>X</hom> code is unused by the navigation, but is used to style the homonym value X (red color).

Here is an example:

[funderburkjim]

letter homonym in metaline

The <h>[a-z] codes were originally assigned by a Java program developed by Pawan Goyal, working with Peter. And I have adapted this to the current format of mw.txt.

This display feature makes use of the <h> element in the metaline of entries. 10645 matches for "<h>[a-z]" in buffer: temp_mw_1.txt.

I don't recall whether Pawan's original work introduced the few (100+) 'letter-number' h-values, or whether that was done by me at some time.

letter homonym in entry body

As I recall, it was my choice to put the <hom>X</hom> (X a letter) into entry text. And I must have for some reason chosen to put the letter homs AFTER the headword, such as <s>X</s><hom>Y</hom> ¦

My current view is that I should have

• used a different markup. Maybe, <hom n="a"/> instead of <hom>a</hom>.
  • This would be preferable because the 'a' is NOT part of the text, but <hom>a</hom> implies, by usual xml markup conventions, that the 'a' IS part of the text being marked.
  • By contrast, the empty tag <hom n="a"/> is interpreted as being purely markup.
  • The display programs could be adapted to <hom n="a"/> to display the 'a' as it is currently shown.
• inserted the hom markup BEFORE the headword,
  • This would be consistent with the text's convention for numerical homonyms
  • Displaying the 'a' AFTER the headword might visually emphasize the difference from numeric homonyms, but this distinction is already present via the letter-digit distinction.

I have not made the changes just suggested in the 'I should have...' items.

Request comment by others as to whether this should be done.

Even if the suggested change is made, there are still cases where it is difficult to know how to add markup. For example, in 'or' groups, where the second word has a homonym designation.

[funderburkjim]

I am sure there is further work to be done regarding homonyms in mw, but will put this aside for now as the situation is still unclear to me in several aspects.

[gasyoun]

By contrast, the empty tag is interpreted as being purely markup.

Makes sense.

[Andhrabharati]

I have "handled" the matter in my current full review, and thus this issue can be closed now.