Handling multiple romanizations for the same sound

If you're new to these arts, this is the place to ask "stupid" questions and get directions!
Post Reply
Akubra
cuneiform
cuneiform
Posts: 116
Joined: 24 May 2024 20:33
Location: Where "amai", "plezant", and "goesting" are ordinary words.

Handling multiple romanizations for the same sound

Post by Akubra »

Have you ever used different characters for the same sound when deciding on the spelling of your conlang? I’m using Latin characters rather than a conscript, and I’ve run into this situation. Here are some examples:
  • For /ʃ/, I use <x> in words influenced by Maya languages and <sh> in words influenced by Hindi.
  • For /t͡ʃ/, I write <ch> (Maya and Hindi) and <tj> (Aboriginal languages).
  • For /d͡ʒ/, I use <j> (Hindi) and <dj> (Aboriginal languages).
I want to keep these romanizations because they add to the aesthetic and etymological depth of the language. By looking at the spelling, you can tell which language influenced the word. Many natlangs exhibit this kind of variation.

However, this makes using a sound changer quite difficult. I have to maintain three separate word lists and apply the same rules to slightly different romanizations, which can be overwhelming.

What do you recommend? Should I aim for one sound = one (set of) character(s), or are there strategies to manage these different romanizations effectively?
:con: Rautahi, K'aach
User avatar
Creyeditor
MVP
MVP
Posts: 5257
Joined: 14 Aug 2012 19:32

Re: Handling multiple romanizations for the same sound

Post by Creyeditor »

Applying the sound changes to phonemic IPA would solve your prpblem right?
Creyeditor
"Thoughts are free."
Produce, Analyze, Manipulate
1 :deu: 2 :eng: 3 :idn: 4 :fra: 4 :esp:
:con: Ook & Omlűt & Nautli languages & Sperenjas
[<3] Papuan languages, Morphophonology, Lexical Semantics [<3]
Akubra
cuneiform
cuneiform
Posts: 116
Joined: 24 May 2024 20:33
Location: Where "amai", "plezant", and "goesting" are ordinary words.

Re: Handling multiple romanizations for the same sound

Post by Akubra »

Not exactly, that's actually what I'm already doing.

For instance, I deromanize <x> to /ʃM/ and <sh> to /ʃH/ to distinguish them. If I don't, I lose the difference between <x> and <sh>. However, this causes issues with Lexurgy because /ʃM/ and /ʃH/ are phonetically identical, and Lexurgy doesn't allow defining two symbols with the same phonetic qualities.

Perhaps I need to explore a different approach with Lexurgy, but that might be beyond the scope of this discussion.

In any case, I'll keep looking for a solution.
:con: Rautahi, K'aach
User avatar
Creyeditor
MVP
MVP
Posts: 5257
Joined: 14 Aug 2012 19:32

Re: Handling multiple romanizations for the same sound

Post by Creyeditor »

So you want to convert back to Romanization again? Then maybe you need to have several list+sound change combinations. So for Australian-derived you would start with a a set of Australian-derived Inputs and Australian-Romanization decoding and then your general sound changes and Australian-Romanization de-decoding. Would that work?
Creyeditor
"Thoughts are free."
Produce, Analyze, Manipulate
1 :deu: 2 :eng: 3 :idn: 4 :fra: 4 :esp:
:con: Ook & Omlűt & Nautli languages & Sperenjas
[<3] Papuan languages, Morphophonology, Lexical Semantics [<3]
Akubra
cuneiform
cuneiform
Posts: 116
Joined: 24 May 2024 20:33
Location: Where "amai", "plezant", and "goesting" are ordinary words.

Re: Handling multiple romanizations for the same sound

Post by Akubra »

That's correct, I want to re-romanize the lists after applying all sound changes.

Currently, I have three lists with the same sound changes because I want to treat them as one language, not three separate ones. This approach works, but I need to ensure that all three sets of sound changes remain identical. This is what concerns me: when I modify existing sound changes, I'm worried about losing track and ending up with three slightly different sets of sound changes.

So, while this method does work, maintaining three parallel lists of identical sound changes is challenging, especially when making frequent adjustments. That’s why I’m looking for a solution that is less error-prone.
:con: Rautahi, K'aach
Salmoneus
MVP
MVP
Posts: 3091
Joined: 19 Sep 2011 19:37

Re: Handling multiple romanizations for the same sound

Post by Salmoneus »

I wouldn't have thought this would be a big problem in a sound changer.

I just still use SCA, and I would think it would be quite manageable in that. I would use the rewrite rules (basically just an initial set of soundchanges to convert from romanization into IP) to convert each romanised symbol or digraph into a unique symbol. Then I would define classes containing those symbols that have the same realisation, and define all the sound changes in terms of those classes. Then rewrite the unique symbols back into romanization.
User avatar
Creyeditor
MVP
MVP
Posts: 5257
Joined: 14 Aug 2012 19:32

Re: Handling multiple romanizations for the same sound

Post by Creyeditor »

@Sal: So you would create categories that include different unique symbols that code different romanizations for the same sounds, right? And then apply the rules to these categories such that the romanization information is kept?
Creyeditor
"Thoughts are free."
Produce, Analyze, Manipulate
1 :deu: 2 :eng: 3 :idn: 4 :fra: 4 :esp:
:con: Ook & Omlűt & Nautli languages & Sperenjas
[<3] Papuan languages, Morphophonology, Lexical Semantics [<3]
Akubra
cuneiform
cuneiform
Posts: 116
Joined: 24 May 2024 20:33
Location: Where "amai", "plezant", and "goesting" are ordinary words.

Re: Handling multiple romanizations for the same sound

Post by Akubra »

In Lexurgy, you can work with symbols, but each symbol must be unique and have distinct qualities. Therefore, defining the following is not allowed:

Symbol ʃM [unvoiced postalveolar sibilant fricative consonant]
Symbol ʃH [unvoiced postalveolar sibilant fricative consonant]

Perhaps I should just work with classes instead. It would simplify things.
:con: Rautahi, K'aach
Akubra
cuneiform
cuneiform
Posts: 116
Joined: 24 May 2024 20:33
Location: Where "amai", "plezant", and "goesting" are ordinary words.

Re: Handling multiple romanizations for the same sound

Post by Akubra »

Or, maybe I can separate the deromanization/romanization process from the pure sound changes.

Here's what I'm doing right now:

Code: Select all

                  => Maya list  => deromanize / apply sound changes separately / romanize => new Maya list  =>
general word list => Hindi list => deromanize / apply sound changes separately / romanize => new Hindi list => new general list
                  => Aust. list => deromanize / apply sound changes separately / romanize => new Aust. list =>
If I separate the deromanization/romanization, it would look like this:

Code: Select all

                  => Maya list  => deromanize =>                               => romanize => new Maya list  =>
general word list => Hindi list => deromanize => apply sound changes generally => romanize => new Hindi list => new general list
                  => Aust. list => deromanize =>                               => romanize => new Aust. list =>
With this approach, I would only need to keep the word lists separate and feed them one by one to the sound changer. This seems to be much easier than maintaining three identical sound change files.
:con: Rautahi, K'aach
Salmoneus
MVP
MVP
Posts: 3091
Joined: 19 Sep 2011 19:37

Re: Handling multiple romanizations for the same sound

Post by Salmoneus »

Akubra wrote: 14 Jun 2024 14:51 In Lexurgy, you can work with symbols, but each symbol must be unique and have distinct qualities. Therefore, defining the following is not allowed:

Symbol ʃM [unvoiced postalveolar sibilant fricative consonant]
Symbol ʃH [unvoiced postalveolar sibilant fricative consonant]

Perhaps I should just work with classes instead. It would simplify things.
The advantage of the SCA is that it's entirely mechanical - you can define anything however you want.

Does Lexurgy force you to use its own quality definitions, or can you make you own? Because if so you could just define one as [Hindi unvoiced postalveolar sibilant fricative consonant] and the other as [Maya unvoiced postalveolar sibilant fricative consonant].
User avatar
Arayaz
mayan
mayan
Posts: 1665
Joined: 07 Sep 2022 00:24
Location: Just south of the pin-pen merger

Re: Handling multiple romanizations for the same sound

Post by Arayaz »

Salmoneus wrote: 15 Jun 2024 19:53
Akubra wrote: 14 Jun 2024 14:51 In Lexurgy, you can work with symbols, but each symbol must be unique and have distinct qualities. Therefore, defining the following is not allowed:

Symbol ʃM [unvoiced postalveolar sibilant fricative consonant]
Symbol ʃH [unvoiced postalveolar sibilant fricative consonant]

Perhaps I should just work with classes instead. It would simplify things.
The advantage of the SCA is that it's entirely mechanical - you can define anything however you want.

Does Lexurgy force you to use its own quality definitions, or can you make you own? Because if so you could just define one as [Hindi unvoiced postalveolar sibilant fricative consonant] and the other as [Maya unvoiced postalveolar sibilant fricative consonant].
You can make your own. That's a good idea, and what I'd do.
proud member of the myopic-trans-southerner-viossa-girl-with-two-cats-who-joined-on-september-6th-2022 gang

my garbage
areyaxi family, alushi, ǫǫ (active)
2c2ef0, arskiilz, kahóra (inactive)
xúuuatxia family, njabad (tenuous)

she/her
User avatar
lsd
greek
greek
Posts: 812
Joined: 11 Mar 2011 21:11

Re: Handling multiple romanizations for the same sound

Post by lsd »

3SDeductiveLanguage(1Sense=1SIgn=1Sound)
is presented with different romanizations :
- in French and English, depending of the audience...
- and some which describe their phonographic sides (phonetically),
- others their semantic sides (in the form of back translation)
- the most exotic their sign sides (using logographic romanization)
the native writing is purely logographic...
Akubra
cuneiform
cuneiform
Posts: 116
Joined: 24 May 2024 20:33
Location: Where "amai", "plezant", and "goesting" are ordinary words.

Re: Handling multiple romanizations for the same sound

Post by Akubra »

Salmoneus wrote: 15 Jun 2024 19:53 Does Lexurgy force you to use its own quality definitions, or can you make you own? Because if so you could just define one as [Hindi unvoiced postalveolar sibilant fricative consonant] and the other as [Maya unvoiced postalveolar sibilant fricative consonant].
Arayaz is right, you can do that. However, I think (though I'm not certain) you might need to define each symbol with Maya/Hindi/Aust even if it's not absolutely necessary. AFAIK, Lexurgy cannot compare symbols unless they contain the same number of qualities. I'll have to verify this; hopefully, I'm mistaken...

If that's the case, you'd have something like this:
Symbol ʃM [Maya unvoiced postalveolar sibilant fricative consonant] which corresponds to <x>.
Symbol ʃH [Hindi unvoiced postalveolar sibilant fricative consonant] which corresponds to <sh>.

Additionally, you'd have:
Symbol lM [Maya voiced alveolar lateral approximant consonant] which corresponds to <l>.
Symbol lH [Hindi voiced alveolar lateral approximant consonant] which also corresponds to <l>.
:con: Rautahi, K'aach
Akubra
cuneiform
cuneiform
Posts: 116
Joined: 24 May 2024 20:33
Location: Where "amai", "plezant", and "goesting" are ordinary words.

Re: Handling multiple romanizations for the same sound

Post by Akubra »

I've figured it out!

Here's how I solved the problem:

Code: Select all

Symbol x   [MayRom unvoiced postalveolar sibilant fricative]
Symbol ʃM  [MayIpa unvoiced postalveolar sibilant fricative]
Symbol sh  [HinRom unvoiced postalveolar sibilant fricative]
Symbol ʃH  [HinIpa unvoiced postalveolar sibilant fricative]
In the deromanizer, I simply write:

Code: Select all

[MayRom] => [MayIpa]
[HinRom] => [HinIpa]
This way, <x> maps to /ʃM/ and <sh> maps to /ʃH/.

There was another small issue with the character <j>, which is used for /d͡ʒ/ and is the IPA symbol for <y>. I solved it similarly:

Code: Select all

Symbol j   [GenRom voiced postalveolar affricate]
Symbol d͡ʒ  [GenIpa voiced postalveolar affricate]
Symbol y   [GenRom voiced palatal approximant]
Symbol jG  [GenIpa voiced palatal approximant]
Now, <j> maps to /d͡ʒ/ and <y> maps to /jG/.

With this issue resolved, I can start writing the sound changes. Thanks to everyone for your input!
:con: Rautahi, K'aach
Post Reply