Small changes to paradigm styles

[nienna73]

What's in this PR:

• changed which elements are bold vs. italic

It looks like this now:

[codecov-commenter]

Codecov Report

Merging #864 (160d157) into main (1524a8a) will increase coverage by 0.04%. The diff coverage is n/a.

@@            Coverage Diff             @@
##             main     #864      +/-   ##
==========================================
+ Coverage   81.71%   81.76%   +0.04%     
==========================================
  Files         106      106              
  Lines        4414     4414              
  Branches      650      650              
==========================================
+ Hits         3607     3609       +2     
+ Misses        684      681       -3     
- Partials      123      124       +1     

Impacted Files	Coverage Δ
.../search_quality/management/commands/featuredump.py	32.69% <0.00%> (ø)
...ranslate/management/commands/translatewordforms.py	73.17% <0.00%> (ø)
src/CreeDictionary/CreeDictionary/views.py	81.73% <0.00%> (+1.73%)	:arrow_up:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 1524a8a...160d157. Read the comment docs.

[eddieantonio]

I will abstain from reviewing this, since I was there, but @aarppe, @andrewdotn, and @dwhieb, I'd appreciate your thoughts on these styling changes!

[eddieantonio]

Addresses #862.

[aarppe]

For separating the linguistic content from the meta-linguistic labels, bolding the latter does work well. But for distinguishing observed forms from merely generated ones, with italicizing the former, I'm not convinced that that formatting will achieve the goal. People might consider italics as lesser than regular, and think that the forms that are italicized are not observed, and not the other way around.

Honestly, the old scheme of boldfacing observed forms in contrast to unobserved forms was self-evident. Of course, that conflicted with column/row labels, if we'd want to bold-face those to distinguish them clearly.

I could imagine some options:

1. bold-face labels, bold-face and italicize observed forms, and keep unobserved forms with regular font.
2. bold-face labels, keep observed forms as regular, and present the unobserved forms with lesser grey (as long as that is legible).
3. Use a different color for the labels, though still boldfaced (I've suggested blue, as in the old itwêwina), present observed forms in bold-face and unobserved forms with regular font. See: https://github.com/UAlbertaALTLab/cree-intelligent-dictionary/issues/862#issuecomment-862876972 as well as the image below:

[eddieantonio]

For separating the linguistic content from the meta-linguistic labels, bolding the latter does work well. But for distinguishing observed forms from merely generated ones, with italicizing the former, I'm not convinced that that formatting will achieve the goal. People might consider italics as lesser that regular, and thin that the forms that are italicized are not observed, and not the other way around.

This is actually what the styling indicates. mîcisow is an unobserved form of “mîcisow”!

Honestly, the old scheme of boldfacing observed forms in contrast to unobserved forms was self-evident.

So I've had the discussion with many people: “what does the bolding mean?” and I've had to explicitly explain it to them. It is not self-evident. Plus, #405 still needs to be done that explicitly explains certain things, like the corpus frequencies.

Of course, that conflicted with column/row labels, if we'd want to bold-face those to distinguish them clearly.

I could imagine some options:

1. bold-face labels, bold-face and italicize observed forms, and keep unobserved forms with regular font.
2. bold-face labels, keep observed forms as regular, and present the unobserved forms with lesser grey (as long as that is legible).
3. Use a different color for the labels, though still boldfaced (I've suggested blue, as in the old itwêwina), present observed forms in bold-face and unobserved forms with regular font. See: #862 (comment) as well as the image below:

We'll have to iterate on the design a bit more. I will continue to resist colouring text blue if it is not a link. Of course, with #405 finished, we could make it a link!

[nienna73]

I largely agree with @eddieantonio 's comments. I think blue should be reserved for links, unless we have another very clear colour we're using for links throughout morphodict/those items actually do link to something. Also, having black and grey text side by side does weird things to my eyes and my perception of colours--I often end up thinking they're the same colour and it just looks different based on whether I'm looking directly at the text.

I think we have a lot of information to convey implicitly and getting this to a state where everyone automatically knows what's being presented is going to take multiple iterations

[aarppe]

I looked at the revision again. It does a good job in distinguishing the word-forms from the labels. But I have a really hard time distinguishing attested from unattested forms – the italicization really needs focus and is hard to discern. Weirdly, in some ways, I would that the current rendering is almost better, in that the row labels are consistently italicized, cf.

If we must still with a single color, i.e. black, I'm not really sure what the best options are. Perhaps they would be the following:

1. boldface row and column labels, as well as attested wordforms, and leave non-attested wordforms as plain regular.
2. boldface row and column labels, mark non-attested word-forms with some 50% grey, and leave the attested wordforms as plain regular.
3. Other combinations, even including borderlines.

Here are some alternative visualizations: