Spell check "ignore" and "learn" in Publisher

[DanielL]

I am not sure if this is related to this spell checking thread or something different, if different feel free to move it to a unique thread as applicable.

This is using the current 1014 beta version.  I have three instances of the same word(callsign), only difference is the capitalization and punctuation

I need the all caps to be learned as it is a control value and needs to be spelled as-is so I tell it to learn this word (right click on the one in blue).  Both versions of the all caps are now accepted versions

This leaves the lowercase situation which is a genuine spelling error and needs to be fixed, I expect to be able to right click it and fix the spelling.  However the only option available is to "UnLearn Spelling"  If its a learned spelling, why is it saying its spelled wrong still?  If capitalization is important (and thus unique), it should instead give me the option to fix the spelling.  This behavior feels inconsistant to me.

okay, so I want to unlearn this capitalized spelling issue, so I go ahead and select the option to "Unlearn Spelling", I would expect the all caps to return to being spelled wrong, but that is actually the opposite of what happens here , all versions are now accepted as being spelled correctly.

 

Continuing down the chain here, I try to use the undo feature to back out the learning.  I can undo the learning for the all caps versions, but not the lower-case version of the word.  I am now stuck with the wrong spelling being learned as lower case, and the right spelling in all caps being unlearned.  

 

This behavior is perplexing to me, seems like the spelling routines need some more work

 

[Ken Hjulstrom]

Hi DanielL,

I believe that what you're encountering is related to the same set of erroneous selections/assumptions made by the spellcheck dictionary and "learned" and "ignored" words that are apparently the cause of the inconsistent behavior that I initially reported here.  In your case, I'm wondering if the "learned" version of what you wanted to be "CALLSIGN" was actually entered into the "learned" word set as "callsign-", being consistent with what appears to me to be the spellcheck's comparison algorithm not being able to separate appended punctuation from words as they are being spellchecked.  If the "learned" word contained the trailing hyphen, that would explain why the other two instances of "callsign" weren't "learned", regardless of the different capitalization.  It's also interesting that the word shown in the "Unlearn spelling" popup menu option is in lowercase, which may imply that all words are intended to be used in a case-independent comparison for matching, which I didn't find to be the case in my original post here, nor does it seem to be the case in your example.  I also ran into a case where I couldn't "unlearn" a word (for me, it was the French word "l'honneur", possibly not entirely in lower case).  I have a software engineering background, and in my opinion, both my experiences and yours seem to be related to some sort of inconsistencies regarding case and attached punctuation between the spell checker itself and the way words are accessed and stored when added (and removed) from the "Ignore" and "Learned" lists.  Dan C's post of 17-Feb-21 seems to echo my thoughts, as he seems to imply that the underlying problem might have a larger scope than just what you and I are reporting.  I think that the following should be expected operation of the spellcheck system in Publisher:

1. By default, comparisons of document words to dictionary, "learned", and "ignored" words should be case insensitive and should ignore any contiguous punctuation.
2. There should be a way to require that "learned" or "ignored" words be specified in a case-sensitive way, so only word variants containing a specific case pattern are considered as correctly spelled or misspelled.
3. I'm not sure how important this is, but there should be sufficient leniency regarding the spellchecking of words that may contain an initial uppercase character at the beginning of a sentence, but "smart" enough to be aware of "learned" or "ignored" words that may require different capitalization at the beginning of a sentence (the word "iPhone" comes to mind).
4. There should be a way to view a list of "learned" and "ignored" words, especially if these word lists are used in a wider scope than just a single document.  But even for a single document, it's risky to not be able to view a complete list of the active "learned" and "ignored" words, since some may have been specified in error, which would damage the accuracy of the spellcheck process.

Feel free to modify the above list, as it might be used as a requirements list for future enhancements of the spellcheck facility in future versions of Publisher.

Thanks,

Ken