lingdb / Sound-Comparisons

Exploring phonetic diversity across language families —
http://www.soundcomparisons.com
Other
13 stars 8 forks source link

Using Colour on the Malakula Website #459

Open PaulHeggarty opened 7 years ago

PaulHeggarty commented 7 years ago

At the moment, the Malakula website looks very strange and uninformative about the linguistic diversity on the island. And we are wasting the value of colour to help with this. And making the site look unnecessarily monochrome.

I appreciate that grouping and colouring by region doesn’t make much sense, and gives a misleading impression of a geographical structure that is not there.

But, @AvivaShimelman, you seem clear that there is no real problem distinguishing between the language level and the dialect level (at least in most cases?). Indeed you use it in the tree structure. So, how about we colour by language, and keep dialects of the same language in the same colour? I know, that would need 40+ distinctive colour shades, which is actually too much for overall eyeballing, but it would still be much better also to help people find languages on the map.

What this means also is that rather than having our (linguistically arbitrary) six geographical language groups on the left, we would have 40 odd, mostly with just one to three dialects each.

It would also be possible to switch off (by default, or by a new toggle switch) all but one representative dialect of each language.

Thoughts, @AvivaShimelman and @LauraWae?

LauraWae commented 7 years ago

Hi! Good thoughts!

I like the idea of the toggle switch. This indeed can be very helpful.

About the colours, I agree, too. We should also ask ourselves why we have different colours in the left bar menue, but no such on the speech bubbles on the maps!

As we have 42 different languages, I agree also that this is a very huge amount to be using a different colour for each. My suggestion is to mix between geography and language here. Let's maintain the geography as an overall colour determiner (North = red, South = green, etc.) and then use different shades of those determiner colours in order to assign the different languages. This would mean that we make no differences between the dialects of different languages any more, which is actually what we are doing with the Romance languages, for example. This would also mean that we would have to take some (necessarily not completely satisfying?) decisions for some language which do not fit in this scheme.

This sounds quite easy and I am very interested to hear what Aviva has to contribute about it, because I am sure it is possibly not that easy ;)


As a reference, I have collected those languages from the Reference Sheet:

Venen Taute MaluaBay Najit Nese Vao Tirax Uripiv Wala Rano Atchin Alavas Larevat Neveei Novol Nasarian Nahavaq Naati Ninde Naahai Axamb Neverver Tape Naman Unua Pangkumu Nitita Fifti Tesmbol Aulua Mbwenelang Burmbar Nisvai PortSandwich Avol Uliveo Batarxopu Njav Letemboi Navwien Nasvang Natangan

PaulHeggarty commented 7 years ago

No need to create new lists, though. We already have all the info we need. I can change some lines of code and that will create new groups automatically, on the basis that level 5 = language, level 6 = dialect. The colours could be assigned automatically.

Not sure about using shades by region, though, I think that would still make people thing that languages in a region were more similar to each other than they are. I’d rather have 40+ different colours for each language, and keep the very similar colours in different parts of the map, as you would colour countries on a politcal map.

LauraWae commented 7 years ago

This should look beautiful, then. I am sure you can draw on your large colour pool from COBL in this case ;)

AvivaShimelman commented 7 years ago

Excellent. There are, of necessity, larger aggregates (8 or 9) (ask Laura, I already color my transcriptions this way). On-the-ground perception (i.e. speaker perception) and Simon's trees pretty much agree as to where it would be appropriate to make the chromatic cuts. ANd (bonus!), not unexpectedly, there's a lot of geographic bundling.

A.

On Tue, Apr 11, 2017 at 2:47 AM, Paul Heggarty notifications@github.com wrote:

At the moment, the Malakula website looks very strange and uninformative about the linguistic diversity on the island. And we are wasting the value of colour to help with this. And making the site look unnecessarily monochrome.

I appreciate that grouping and colouring by region doesn’t make much sense, and gives a misleading impression of a geographical structure that is not there.

But, @AvivaShimelman https://github.com/AvivaShimelman, you seem clear that there is no real problem distinguishing between the language level and the dialect level (at least in most cases?). Indeed you use it in the tree structure. So, how about we colour by language, and keep dialects of the same language in the same colour? I know, that would need 40+ distinctive colour shades, which is actually too much for overall eyeballing, but it would still be much better also to help people find languages on the map.

What this means also is that rather than having our (linguistically arbitrary) six geographical language groups on the left, we would have 40 odd, mostly with just one to three dialects each.

It would also be possible to switch off (by default, or by a new toggle switch) all but one representative dialect of each language.

Thoughts, @AvivaShimelman https://github.com/AvivaShimelman and @LauraWae https://github.com/LauraWae?

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/lingdb/Sound-Comparisons/issues/459, or mute the thread https://github.com/notifications/unsubscribe-auth/ARGKM256JTskb1gVZy6qrFICssSRWJnpks5ruyHlgaJpZM4M5rzU .

-- Aviva Shimelman, PhD

AvivaShimelman commented 7 years ago

Yikes! No geography/"genetic" doubling possible, I fear. "TMI" (too much information ... and too much useless information). In any case, as I mentioned in my last post, the language grouping turn out to be regional in the majority of the cases, in any case. I'll bundle the languages.

A.

On Tue, Apr 11, 2017 at 5:21 AM, LauraWae notifications@github.com wrote:

Hi! Good thoughts!

I like the idea of the toggle switch. This indeed can be very helpful.

About the colours, I agree, too. We should also ask ourselves why we have different colours in the left bar menue, but no such on the speech bubbles on the maps!

As we have 42 different languages, I agree also that this is a very huge amount to be using a different colour for each. My suggestion is to mix between geography and language here. Let's maintain the geography as an overall colour determiner (North = red, South = green, etc.) and then use different shades of those determiner colours in order to assign the different languages. This would mean that we make no differences between the dialects of different languages any more, which is actually what we are doing with the Romance languages, for example. This would also mean that we would have to take some (necessarily not completely satisfying?) decisions for some language which do not fit in this scheme.

This sounds quite easy and I am very interested to hear what Aviva has to contribute about it, because I am sure it is possibly not that easy ;)

As a reference, I have collected those languages from the Reference Sheet:

Venen Taute MaluaBay Najit Nese Vao Tirax Uripiv Wala Rano Atchin Alavas Larevat Neveei Novol Nasarian Nahavaq Naati Ninde Naahai Axamb Neverver Tape Naman Unua Pangkumu Nitita Fifti Tesmbol Aulua Mbwenelang Burmbar Nisvai PortSandwich Avol Uliveo Batarxopu Njav Letemboi Navwien Nasvang Natangan

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub https://github.com/lingdb/Sound-Comparisons/issues/459#issuecomment-293200943, or mute the thread https://github.com/notifications/unsubscribe-auth/ARGKMy0uxqtJhiqGCW9rwcBIycRlooK5ks5ru0YmgaJpZM4M5rzU .

-- Aviva Shimelman, PhD

AvivaShimelman commented 7 years ago

Right. Simon's most recent tree has my endorsement. Depending on what makes sense chromatically (i.e., how many colors/shades are actually visible and interpretable at a glance), just pick node levels. I can do this by hand in 5 minutes, too.

A.

On Tue, Apr 11, 2017 at 5:55 AM, Paul Heggarty notifications@github.com wrote:

No need to create new lists, though. We already have all the info we need. I can change some lines of code and that will create new groups automatically, on the basis that level 5 = language, level 6 = dialect. The colours could be assigned automatically.

Not sure about using shades by region, though, I think that would still make people thing that languages in a region were more similar to each other than they are. I’d rather have 40+ different colours for each language, and keep the very similar colours in different parts of the map, as you would colour countries on a politcal map.

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub https://github.com/lingdb/Sound-Comparisons/issues/459#issuecomment-293209514, or mute the thread https://github.com/notifications/unsubscribe-auth/ARGKM2jWKgasmIvXBJI8d0dOMBLeYIZAks5ru04FgaJpZM4M5rzU .

-- Aviva Shimelman, PhD

PaulHeggarty commented 7 years ago

There are, of necessity, larger aggregates (8 or 9) (ask Laura, I already color my transcriptions this way). On-the-ground perception (i.e. speaker perception) and Simon's trees pretty much agree as to where it would be appropriate to make the chromatic cuts.

OK, we could structure and colour it like that if you wish, 8-9 geographical groups.

I would need:

Best in an excel sheet.

AvivaShimelman commented 7 years ago

Wait a couple of minutes and I'll group by major colors (Eliminating white and black, I think English still gives us a perfect 8 to work with).

On Tue, Apr 11, 2017 at 2:41 PM, Paul Heggarty notifications@github.com wrote:

No need to create new lists, though. We already have all the info we need. I can change some lines of code and that will create new groups automatically, on the basis that level 5 = language, level 6 = dialect. The colours could be assigned automatically.

Not sure about using shades by region, though, I think that would still make people thing that languages in a region were more similar to each other than they are. I’d rather have 40+ different colours for each language, and keep the very similar colours in different parts of the map, as you would colour countries on a politcal map.

— You are receiving this because you were assigned. Reply to this email directly, view it on GitHub https://github.com/lingdb/Sound-Comparisons/issues/459#issuecomment-293362392, or mute the thread https://github.com/notifications/unsubscribe-auth/ARGKM15ZMSk2NZI04BPOx8qlzunBUx4Gks5ru8llgaJpZM4M5rzU .

-- Aviva Shimelman, PhD

AvivaShimelman commented 7 years ago

Doing that now. By "order within group," practically, would that correspond to adjacent shades?

A.

On Tue, Apr 11, 2017 at 5:58 PM, Paul Heggarty notifications@github.com wrote:

There are, of necessity, larger aggregates (8 or 9) (ask Laura, I already color my transcriptions this way). On-the-ground perception (i.e. speaker perception) and Simon's trees pretty much agree as to where it would be appropriate to make the chromatic cuts.

OK, we could structure and colour it like that if you wish, 8-9 geographical groups.

I would need:

  • the group names,
  • the list of all varieties you want in each group
  • with their full index numbers
  • in the order you want them in within that group.

Best in an excel sheet.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/lingdb/Sound-Comparisons/issues/459#issuecomment-293412804, or mute the thread https://github.com/notifications/unsubscribe-auth/ARGKM5lZJQLACoGpQUkK_7bQ2veI1IT3ks5ru_eTgaJpZM4M5rzU .

-- Aviva Shimelman, PhD

AvivaShimelman commented 7 years ago

Was waiting for Simon's tree. Back on it now.

On Tue, Apr 11, 2017 at 5:58 PM, Paul Heggarty notifications@github.com wrote:

There are, of necessity, larger aggregates (8 or 9) (ask Laura, I already color my transcriptions this way). On-the-ground perception (i.e. speaker perception) and Simon's trees pretty much agree as to where it would be appropriate to make the chromatic cuts.

OK, we could structure and colour it like that if you wish, 8-9 geographical groups.

I would need:

  • the group names,
  • the list of all varieties you want in each group
  • with their full index numbers
  • in the order you want them in within that group.

Best in an excel sheet.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/lingdb/Sound-Comparisons/issues/459#issuecomment-293412804, or mute the thread https://github.com/notifications/unsubscribe-auth/ARGKM5lZJQLACoGpQUkK_7bQ2veI1IT3ks5ru_eTgaJpZM4M5rzU .

-- Aviva Shimelman, PhD

AvivaShimelman commented 7 years ago

Attached is a spreadsheet with my suggestions for grouping and shading languages. Here are some notes about how to read it:

Notes for Malakula languages groupings and color codings. As it turns out, we have 10 tight families and one set of three languages that really aren't tight enough to group with any of the color sets (maybe these could all be grey?)

1.I have given SC indes numbers out through level 5. All languages with the same 8-place prefix should be colored the same (i.e., all level 6 lects with the same first 8 didgits in their indexes should be colored/shaded the same) 2.the (a)/(b) distinction is a bit beside the point – it is meant just to indicate that there are bigger jumps between these languages and others in their color group (I have operationalized this by assigning larger chromatic distance (see (3)) 3.Column F “chromatic distance” is supposed to indicate relative shading to correspond – very roughly -- to relative distance between languages. It is of course up to whoever has to program this how to operationalize that, but I think it would be either a dark-to-light (or light-to-dark) series with some kind of regular stepping (i.e., I don't mean that an 8, say, should correspond to 80% opacity ... )

Let me know what is missing/unclear.

2017 04 12 Malakula languages groupings for color coding.xlsx

PaulHeggarty commented 7 years ago

Great, thanks. For the moment there is no option to do colour-coding within groups, but now that you raise the idea, it is something good to consider adding to Sound Comparisons in due course.
The basic colouring by group will be done, although quite when I'll get the time to implement this I don’t know. But it will happen eventually!

AvivaShimelman commented 7 years ago

Sorry. The shading was what I thought you meant by "order within that group." If that's something I still need to deal with, please elaborate.