SCA2 questions

Discussions worth keeping around later.
CaesarVincens
Lebom
Lebom
Posts: 204
Joined: Thu Feb 25, 2010 7:26 pm

Re: SCA2 questions

Post by CaesarVincens »

I found a bug with the "…" wildcard, I think.

Try using this rule:

Code: Select all

a/à/_/à…_

on these words

Code: Select all

akakakakakaka
akkakkakkakkakkakka
I would expect that only the first "a" would become "à" and the rest would be blocked, but it seems the "…" wildcard matches only one letter in the exception area if it comes before the underline. It works as expected in the condition area matching any number of exceptions it also works after the underline in the exception area.

If it's working as designed, it would be good to note.

User avatar
vec
Avisaru
Avisaru
Posts: 639
Joined: Tue Sep 16, 2003 10:42 am
Location: Reykjavík, Iceland
Contact:

Re: SCA2 questions

Post by vec »

Just wanted to pop in and complain that the SCA2 doesn't fully support Unicode combining characters and tends to treat them as separates rather than as a part of the previous glyph. Is this something that can be fixed? I have a large number of vowel qualities, each marking nasality and stress and I'd rather not add in dozens of rewrite rules. I figure the combining characters spectrum in Unicode is quite limited so it ought to be easy enough to just say: if 0300–036F, 1AB0–1AFF, 1DC0–1DFF, 20D0–20FF or FE20–FE2F, treat as part of previous glyph, right?
vec

User avatar
jal
Sumerul
Sumerul
Posts: 2633
Joined: Tue Feb 06, 2007 12:03 am
Location: Netherlands
Contact:

Re: SCA2 questions

Post by jal »

vec wrote:Just wanted to pop in and complain that the SCA2 doesn't fully support Unicode combining characters and tends to treat them as separates rather than as a part of the previous glyph. Is this something that can be fixed? I have a large number of vowel qualities, each marking nasality and stress and I'd rather not add in dozens of rewrite rules. I figure the combining characters spectrum in Unicode is quite limited so it ought to be easy enough to just say: if 0300–036F, 1AB0–1AFF, 1DC0–1DFF, 20D0–20FF or FE20–FE2F, treat as part of previous glyph, right?
Yeah, but it's more complicated than that, not to mention that in general, JavaScript has a Unicode problem.


JAL

zompist
Boardlord
Boardlord
Posts: 3368
Joined: Thu Sep 12, 2002 8:26 pm
Location: In the den
Contact:

Re: SCA2 questions

Post by zompist »

This is what Unicode was supposed to prevent. :( I'd have to rewrite things quite a bit, as I do a lot of walking strings character by character.

(On the other hand, it might not be too hard to automatically generate a bunch of rewrite rules, so I'll look at that...)

User avatar
Salmoneus
Sanno
Sanno
Posts: 3197
Joined: Thu Jan 15, 2004 5:00 pm
Location: One of the dark places of the world

Re: SCA2 questions

Post by Salmoneus »

Ok, what in the hell?

I've been having some problems getting it to do stuff at the end of words. So i played around and found I couldn't even just add things at the end of words - it ignored me completely.

So then I just used your sample portuguese thing. And added a new final change:
/k/_#

What does this do? Well, for every word it does nothing at all. Except that it produces the words "distritok k k k k k k" and "segundok k"

Hmm. Now, since starting this post I've worked it out. So you can ignore the 'what in the hell' bit, I've worked that out. It's about spaces. You say in the help file that it now counts spaces as word-endings... but it seems it ONLY counts spaces as word-endings. For some sound changes. So each word needs one and only one blank space on the same line.

But for other sound changes, it works fine. So 'r//_#' works fine. In fact, in the test set-up, everything works, except for /*/_#. In my file, I'm also seemingly struggling with a convert-one-category-to-another-when-final rule... but it may be that there's another reason for that (I was trying to use that 'insert final symbol' as a test to see what was going on and in what order, and when it didn't work I assumed that that was the same problem, but it may be that it's not and I've just made a mistake somewhere).

But still. Inserting final characters does not work as expected.


EDIT: ok, finally worked out what I was doing wrong. So it looks like it's just the adding thing that's weird.

[Although I think some erratic things can happen if you have word-final rules AND accidentally add ' ' to one of your categories... but I can't really tell, my rules are too complicated to test easily]
Blog: [url]http://vacuouswastrel.wordpress.com/[/url]

But the river tripped on her by and by, lapping
as though her heart was brook: Why, why, why! Weh, O weh
I'se so silly to be flowing but I no canna stay!

User avatar
KathTheDragon
Smeric
Smeric
Posts: 2139
Joined: Thu Apr 25, 2013 4:48 am
Location: Brittania

Re: SCA2 questions

Post by KathTheDragon »

Salmoneus, I've found the problem. In the Portugese stuff, click after 'districtus' and 'secundus'. You'll then see why there are extra k's.

User avatar
Salmoneus
Sanno
Sanno
Posts: 3197
Joined: Thu Jan 15, 2004 5:00 pm
Location: One of the dark places of the world

Re: SCA2 questions

Post by Salmoneus »

KathAveara wrote:Salmoneus, I've found the problem. In the Portugese stuff, click after 'districtus' and 'secundus'. You'll then see why there are extra k's.
I found that problem already!
It's because there are extra spaces and the machine treats spaces as word endings.

But the real problem is, for this one rule (and maybe others I haven't tested), it doesn't recognise word endings UNLESS there's a space there!
Blog: [url]http://vacuouswastrel.wordpress.com/[/url]

But the river tripped on her by and by, lapping
as though her heart was brook: Why, why, why! Weh, O weh
I'se so silly to be flowing but I no canna stay!

User avatar
KathTheDragon
Smeric
Smeric
Posts: 2139
Joined: Thu Apr 25, 2013 4:48 am
Location: Brittania

Re: SCA2 questions

Post by KathTheDragon »

Oh, I see what you mean

zompist
Boardlord
Boardlord
Posts: 3368
Joined: Thu Sep 12, 2002 8:26 pm
Location: In the den
Contact:

Re: SCA2 questions

Post by zompist »

OK, should be fixed.

1. Insertions should work at the end of the word.
2. Blanks at the end of vocabulary items are removed.

User avatar
vec
Avisaru
Avisaru
Posts: 639
Joined: Tue Sep 16, 2003 10:42 am
Location: Reykjavík, Iceland
Contact:

Re: SCA2 questions

Post by vec »

zompist wrote:OK, should be fixed.

1. Insertions should work at the end of the word.
2. Blanks at the end of vocabulary items are removed.
Nice!

Does it replace spaces at the end of sound changes as well? If there's a space after a rule, it is treated as invalid currently.
vec

User avatar
R.Rusanov
Avisaru
Avisaru
Posts: 393
Joined: Sat Jan 05, 2013 1:59 pm
Location: Novo-je Orĭlovo

Re: SCA2 questions

Post by R.Rusanov »

This isn't really a SCA2 question but it might be useful to some of y'all.

It's the sound changes from classical latin, including orthography, to vulgar latin (with stress marked), with west Romance vowel outcomes (which you can however easily change). For lack of a unicode character I represented the stressed versions of ɛ and ɔ as έ and ώ - thanks Greek!
More: show
### Orthography
qu/kw/_
x/ks/_
v/w/_
c/k/_
ph/f/_
h//_
g/ŋ/_n
n/ŋ/_[kg]
ae/e/_
oe/ē/_
u/w/a_
i/j/_V
i/j/V_
u/w/_V
u/w/V_
S/L/_ns
n//_s
L/S²/_
S//_²²
/$/S_²
S//$_
S/L/_$
$//_
### Stress
/*/_V(C)(C)#
*/./V…_
/*/_V(C)(C)(C).
*/./V…_S(C).
.//.…_
/*/_V(C)(C)(C)./*_
.//_
### Vowels (West Romance)
e/ɛ/_
o/ɔ/_
i/e/_
u/o/_
L/S/_
Y/W/*_
*//_
### Other Changes
V//VC_C(R)W
V//W(R)C_CV
Y//VC_CV
S//V[wj]_C#/_m
w//V_V#
b/w/V_V
[eɛi]/j/V_
[eɛi]/j/_V
[oɔu]/w/V_
[oɔu]/w/_V
m/n/W_#
[dtm]//_#/n_
The categories I used were:
More: show
S=aeiou
L=āēīōū
V=aeiouāēīōūɛɔáéíóúέώ
C=wrtplkjgfdszbnmŋ
Y=aeiouɛɔ
W=áéíóúέώ
A sample input and output:
More: show
tóto [tōtum]
altárjo [altārium]
rɔgárɛ [rogāre]
entɛrrɔgárɛ [interrogāre]
dώrso [dorsum]
formóso [fōrmōsum]
róstro [rōstrum]
bέllo [bellum]
bέlla [bellam]
róstros [rōstrōs]
kɔmentjárɛ [cominitiāre]
awkέllo [avicellum]
láwro [laurum]
láwra [lauram]
láwros [laurōs]
láwras [laurās]
nénno [ninnum]
rɛspirárɛ [respīrāre]
brúno [brūnum]
kátto [cattum]
kaldárja [calidāriam]
mutárɛ [mūtāre]
formátko [fōrmāticum]
infántɛ [īnfāntem]
són [sum]
έs [est]
kiwtátɛ [cīvitātem]
kwέso [quaesum]
pagésɛ [pāgēnsem]
djórno [diurnum]
mandukárɛ [mandūcāre]
éllo [illum]
amáj [amāvi]
amáw [amāvit]
Hope this helps y'all out! If you notice any bugs or improper representations of a given sound change, let me know and I'll add it in, provided it's a regular one, of course :^)
Slava, čĭstŭ, hrabrostĭ!

User avatar
GreenBowTie
Lebom
Lebom
Posts: 179
Joined: Wed Oct 09, 2002 3:17 am
Location: the darkest depths of the bone-chilling night

Re: SCA2 questions

Post by GreenBowTie »

!!!
thank you!

User avatar
GreenBowTie
Lebom
Lebom
Posts: 179
Joined: Wed Oct 09, 2002 3:17 am
Location: the darkest depths of the bone-chilling night

Re: SCA2 questions

Post by GreenBowTie »

is there a de-SCA somewhere? a Sound Change Reverser? it would be nice to be able to insert the rules and the results and get back a list of possible origins

User avatar
pharazon
Lebom
Lebom
Posts: 192
Joined: Thu Sep 04, 2003 1:51 am
Location: Ann Arbor
Contact:

Re: SCA2 questions

Post by pharazon »

GreenBowTie wrote:is there a de-SCA somewhere? a Sound Change Reverser? it would be nice to be able to insert the rules and the results and get back a list of possible origins
I haven't tried it, but there's this.

User avatar
GreenBowTie
Lebom
Lebom
Posts: 179
Joined: Wed Oct 09, 2002 3:17 am
Location: the darkest depths of the bone-chilling night

Re: SCA2 questions

Post by GreenBowTie »

wow, nice

User avatar
jal
Sumerul
Sumerul
Posts: 2633
Joined: Tue Feb 06, 2007 12:03 am
Location: Netherlands
Contact:

Re: SCA2 questions

Post by jal »

The main problem, of course, being that there can be, and often is, a n:1 relationship between the ancestral language's phonemes and the decendent's ones.


JAL

User avatar
jal
Sumerul
Sumerul
Posts: 2633
Joined: Tue Feb 06, 2007 12:03 am
Location: Netherlands
Contact:

Re: SCA2 questions

Post by jal »

I've just been trying the SCA2 in combination with gen, and my first remark is that I'd really love to see them combined. I have some complex phonotactics going on, and I'd love to have gen just output that, instead of me having to copy/paste it to the SCA2 etc.

That said, it seems that the SCA2 doesn't handle optional characters directly after an underscore (i.e. it handles them as compulsory). If I have a rule "a/e/C_(C)Ci", this works for "paldi" (becoming "peldi") but not for "padi" (staying "padi"), though the rule "a/e/C_C(C)i" works fine for both cases. Is this a known problem? EDIT: it seems that in general, (C)C doesn't work, but C(C) does.

Secondly, what does the setting "Rewrite on output" do? If I check it, none of the rewrite rules are applied, if I uncheck it, the output is changed as expected.


JAL

User avatar
jal
Sumerul
Sumerul
Posts: 2633
Joined: Tue Feb 06, 2007 12:03 am
Location: Netherlands
Contact:

Re: SCA2 questions

Post by jal »

I also have a feature request for gen. I know it's not exactly SCA2, but I didn't want to start a new thread for it. It would be really nice if gen could show the syllable bounderies, e.g. with a dot. I have a number of SCA rules that work on syllable bounderies, and I can't use the gen output directly now.

Also, while I'm making requests, a different dropoff value for the phonemes and the syllable types would be welcome (perhaps just the %, so we could play around a bit more).


JAL

zompist
Boardlord
Boardlord
Posts: 3368
Joined: Thu Sep 12, 2002 8:26 pm
Location: In the den
Contact:

Re: SCA2 questions

Post by zompist »

All right, I made several changes to gen:

1. Added a checkbox that adds a dot between syllables.
2. Added an IPA button.
3. Added a "Molasses" drop off option.
4. Added a control for more even syllable type choice.

The program did not and does not use the Dropoff control to choose syllables. It had a fairly fast dropoff, however, and I rewrote that and also added the slower dropoff.

I though the UI would get a bit too messy if I added percentage boxes.

I will look at the SCA2 problem later. Enough Javascript wrangling for tonight. :)

User avatar
jal
Sumerul
Sumerul
Posts: 2633
Joined: Tue Feb 06, 2007 12:03 am
Location: Netherlands
Contact:

Re: SCA2 questions

Post by jal »

zompist wrote:All right, I made several changes to gen
Thanks, great! I'll take a look at it alter today.
I though the UI would get a bit too messy if I added percentage boxes.
Well, it's already a bit messy, but I think I can live with that :)
I will look at the SCA2 problem later. Enough Javascript wrangling for tonight. :)
I don't envy you :).


JAL

User avatar
dhok
Avisaru
Avisaru
Posts: 859
Joined: Wed Oct 24, 2007 7:39 pm
Location: The Eastern Establishment

Re: SCA2 questions

Post by dhok »

I'm delurking after a fairly long hiatus to report a bug. In the changes from Proto-Macro-Quazian to Proto-Quazioid, there's a law that long vowels shift (it's fairly GCS-ey), except if they're word-final in a polysyllabic word. Thus, *θɨ:ma gives sāma, with *ɨ: shifting to ā, but *pɨyčē is supposed to give īcē, with the final *ē staying put, because it's the final segment of a polysyllabic word. However, SCA2 is giving a reflex of īcī. The relevant categories and rule are:

Code: Select all

V=aeioɨāēīōȳ (ȳ represents /ɨ:/, there's a rewrite rule for it)
Ā=āēȳī
Ē=ēīāî
-------
Ā/Ē/_/V…_# (that's definitely an ellipsis character, not three periods)
pɨyčē lost its initial /p/ and changed its affricate, so that by the time this change is supposed to operate, the form in question has become ɨycē. The log of changes that applied by word indicates that the rule above did indeed change the final ē to ī, even though there's an exception saying it can't do that. I've run the standard checks on the word- there are no weird spaces or anything like that, and all the characters in question are what they claim to be. What's going on?

(EDIT: Checked another thing- the rule fails to work regardless of the presence of the word boundary in the exception, so I'm pretty sure it's the ellipsis that's screwy.)

(EDIT2: the rule above succeeds in catching *nimasī and preventing it from shifting the final *ī, but, again, not with *pɨyčē.)

(EDIT3: The rule doesn't actually catch *nimasī- a later rule about the breaking of /ī/ does. So edit 2 can be ignored. Also, I've figured out a way to work around the issue by undoing the change in monosyllables, so it's not at all pressing, but it's still an error, so I thought I'd log it.)

User avatar
Qwynegold
Smeric
Smeric
Posts: 1606
Joined: Thu May 24, 2007 11:34 pm
Location: Stockholm

Re: SCA2 questions

Post by Qwynegold »

Hi Zompist! Do you still work on SCA2? I've encountered a problem. I'm working on a language with voiceless vowels, and there's a sound change that deletes them. However, words can't be vowelless, so this sound change should be ignored in monosyllabic words. I won't post my sound changes here because they're unnecessarily complicated for this question, but I've been able to determine that you can't have two #s in the environment. I.e. the sound change won't work if the environment is e.g. #C_#. Is this something you can fix easily?
Image
My most recent quiz:
Eurovision Song Contest 2018

User avatar
KathTheDragon
Smeric
Smeric
Posts: 2139
Joined: Thu Apr 25, 2013 4:48 am
Location: Brittania

Re: SCA2 questions

Post by KathTheDragon »

I've just done some very quick testing, and I can't seem to replicate your issue. I tried an environment #C_# in both the environment and the exception fields, and in both cases it worked as expected.

User avatar
Qwynegold
Smeric
Smeric
Posts: 1606
Joined: Thu May 24, 2007 11:34 pm
Location: Stockholm

Re: SCA2 questions

Post by Qwynegold »

That's odd, now it works for me too when I test with something super simple. There must be some problem with my real sound changes. I'll investigate...
Image
My most recent quiz:
Eurovision Song Contest 2018

User avatar
Qwynegold
Smeric
Smeric
Posts: 1606
Joined: Thu May 24, 2007 11:34 pm
Location: Stockholm

Re: SCA2 questions

Post by Qwynegold »

Okay, never mind, I solved it. >_< It was because I was marking the voiceless vowels with a combining diacritic, and didn't remember to include the vowel itself in the environment. Thanks Kath!
Image
My most recent quiz:
Eurovision Song Contest 2018

Post Reply