Main Menu

Na'vi vowels

Started by omängum fra'uti, May 18, 2010, 02:19:51 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

omängum fra'uti

In an attempt to double check my own pronunciations, I charted a bunch of Frommer's vowels based off the formants.  I'm not sure how many people here even know what that means, let alone find it useful (I know at least a few understand the acoustic aspects of vowels, anyway) but I figured I'd post it here.  If anyone has use for the raw data points let me know and I can send you the data.  This is based off four sources.

1. NY. Times dialog (Oel hu Txewì...)
2. Four questions
3. Golden Rule pronunciation
4. 'Upxare ta Karyu Pawl

I tried to select a wide variety of clean vowels for this.  (Meaning vowels that settled onto a sound for a big enough period to get a good sample, and not diphthongs.)  I didn't look into the ì outlier, but the u outlier was one I picked up from speech that was trailing off and softly spoken by the time the u was spoken.

In case it's not obvious, the Y axis is f1, the X axis is f2, and they are both reversed so the chart falls into an arrangement similar to the IPA vowel chart.

The second chart is adjusted with f2-f1, which is supposed to be a better indication of frontness than f2 alone.  However I didn't bother marking the groups there.
Ftxey lu nga tokx ftxey lu nga tirea? Lu oe tìkeftxo.
Listen to my Na'vi Lessons podcast!

kewnya txamew'itan

What do f1 and f2 mean? I'm trying to work out what this data means and, it's a bit tricky without knowing that.  ;)
Internet Acronyms Nìna'vi

hamletä tìralpuseng lena'vi sngolä'eiyi. tìkangkem si awngahu ro
http://bit.ly/53GnAB
The translation of Hamlet into Na'vi has started! Join with us at http://bit.ly/53GnAB

txo nga new oehu pivlltxe nìna'vi, nga oer 'eylan si mì fayspuk (http://bit.ly/bp9fwf)
If you want to speak na'vi to me, friend me on facebook (http://bit.ly/bp9fwf)

numena'viyä hapxì amezamkivohinve
learnnavi's

omängum fra'uti

http://en.wikipedia.org/wiki/Formants

Basically, f1 roughly corresponds to the openness of the vowel, and f2 roughly corresponds to the frontness.  (Actually, f2 - f1 is supposed to be a better measure of the frontness, but I was going for frequency rather than tongue articulation.)
Ftxey lu nga tokx ftxey lu nga tirea? Lu oe tìkeftxo.
Listen to my Na'vi Lessons podcast!

kewnya txamew'itan

What does the data look like using f2-f1? This might make it a bit easier to interpret from a phonetic point of view.
Internet Acronyms Nìna'vi

hamletä tìralpuseng lena'vi sngolä'eiyi. tìkangkem si awngahu ro
http://bit.ly/53GnAB
The translation of Hamlet into Na'vi has started! Join with us at http://bit.ly/53GnAB

txo nga new oehu pivlltxe nìna'vi, nga oer 'eylan si mì fayspuk (http://bit.ly/bp9fwf)
If you want to speak na'vi to me, friend me on facebook (http://bit.ly/bp9fwf)

numena'viyä hapxì amezamkivohinve
learnnavi's

omängum fra'uti

Pretty much the same, but the bottom is skewed to the right so that, for example, the "a" set is straight below the middle of the "u" set.  What are you trying to get out of it?
Ftxey lu nga tokx ftxey lu nga tirea? Lu oe tìkeftxo.
Listen to my Na'vi Lessons podcast!

kewnya txamew'itan

That's very interesting.

If f2-f1 is closer to frontness then a being under u implies that Frommer's been pronouncing it [ ɑ ] whic we thought it was for a while until he told us it wasn't.

The only reason I ask is that, if, as you say frontness is better represented by f2-f1 then surely it would be easier to match to a vowel trapezium if you used f2-f1.
Internet Acronyms Nìna'vi

hamletä tìralpuseng lena'vi sngolä'eiyi. tìkangkem si awngahu ro
http://bit.ly/53GnAB
The translation of Hamlet into Na'vi has started! Join with us at http://bit.ly/53GnAB

txo nga new oehu pivlltxe nìna'vi, nga oer 'eylan si mì fayspuk (http://bit.ly/bp9fwf)
If you want to speak na'vi to me, friend me on facebook (http://bit.ly/bp9fwf)

numena'viyä hapxì amezamkivohinve
learnnavi's

omängum fra'uti

#6
Well there's a third value that enters into it and messes that all up.  Roundness can also influence both numbers.  Rounded vowels will tend to push the pitch while unrounded vowels will push the pitch up.

It turns out that IPA isn't as firm as far as sounds as it is made out to be.  There is quite a bit of wiggle room, and sounds that would be transcribed with the same IPA symbol from different countries can end up with variances in the formants.  For example, I found a chart that compares the sounds of Danish to English and they are quite different.  Go on, try and find a chart for formants of vowels.  You'll find a few, and they will all say different things.

Also, due to variances in the shapes of peoples throats and mouths, generally a chart is taken averaged over a population, rather than compared between individuals.  For example, when I charted my own pronunciation, I tended towards the low end frequency compared to Frommer.  Whereas his typical speech didn't go below 300hz formants, I regularly got down to 150-250hz for my u and i.

So while the f2-f1 chart lines up closer with the IPA, using it as a basis for the associated IPA symbols is a bad idea.

If it were as easy as just lining up charts, computer speech recognition would be a LOT easier.
Ftxey lu nga tokx ftxey lu nga tirea? Lu oe tìkeftxo.
Listen to my Na'vi Lessons podcast!

kewnya txamew'itan

Fair point.

Interesting with the Danish too, irayo.
Internet Acronyms Nìna'vi

hamletä tìralpuseng lena'vi sngolä'eiyi. tìkangkem si awngahu ro
http://bit.ly/53GnAB
The translation of Hamlet into Na'vi has started! Join with us at http://bit.ly/53GnAB

txo nga new oehu pivlltxe nìna'vi, nga oer 'eylan si mì fayspuk (http://bit.ly/bp9fwf)
If you want to speak na'vi to me, friend me on facebook (http://bit.ly/bp9fwf)

numena'viyä hapxì amezamkivohinve
learnnavi's

Ftiafpi

#8
For anyone who doesn't understand this (like me, until a minute ago) read this: http://en.wikipedia.org/wiki/Formant
and this: http://en.wikipedia.org/wiki/Vowel

Anyway, great work, I'm sure this will be useful but I wonder how much we can go off of this since Frommer himself has admitted that some pronunciations of his are off. However, I think you have given us something that we could call more or less canon since the statistical probability of error is small for each vowel "area" on the graph given the sample size (except for "ì", I don't think we have enough samples of that one).

Anyway, from what I can see, it looks like the Na'vi "e" follows the Danish IPA and not the English as far as F2 but follows the English on F1, interesting. Hard to say with the Na'vi "i" but it seems to be following the English there. I'll dive into this later I guess, but good work.

omängum fra'uti

#9
Quote from: kemeoauniaea on May 18, 2010, 03:12:46 AM
That's very interesting.

If f2-f1 is closer to frontness then a being under u implies that Frommer's been pronouncing it [ ɑ ] whic we thought it was for a while until he told us it wasn't.
Actually I'm going to go better than implies and say that it is certainly NOT ɑ.  Though, the example given on the community generated spelled pronunciation guide implies it would be.  (Father is ˈfɑðər.)  I checked my pronunciation of "father" against Frommer's Na'vi "a" and it was nowhere close, even considering the slightly difference between my and his pronunciations.  It was, in fact, to the right of it, exactly where you'd expect /ɑ/ to land in relation to /a/.

In fact, I'd go so far as to say that /a/ doesn't exist alone in American English.  It only exists as part of diphthongs, such as "mine" /maɪn/, /kaʊ/, etc.  It is the first part of the diphthong repeated in the phrase "How now brown cow", or as part of triphthongs like in fired or hour.

Anyway, I added labels to make the chart more clear, and a second chart that shows f2-f1.

Quote from: Ftiafpi on May 18, 2010, 09:43:59 AM
Anyway, from what I can see, it looks like the Na'vi "e" follows the Danish IPA and not the English as far as F2 but follows the English on F1, interesting. Hard to say with the Na'vi "i" but it seems to be following the English there. I'll dive into this later I guess, but good work.
The English vs Danish is actually f2-f1 vs f1, not the straight f2 vs f1 I charted.  Na'vi is pretty close to the English one, unsurprisingly.
Ftxey lu nga tokx ftxey lu nga tirea? Lu oe tìkeftxo.
Listen to my Na'vi Lessons podcast!

kewnya txamew'itan

I have to agree with you about a not being in most accents of English, hence wikipedia keeps changing its pronunciation guide's for English for the letter (it used to just say the a in father, then it said that was GA and now it's dropped English from the main IPA page and given a detailed breakdown on the vowel's own page).

It seems then that, for Frommer's mouth at least, even f2-f1 isn't an accurate description of frontness  :D is 2*f2-f1 any closer?

Hmmm... I guess I'm probably just trying to hard to this onto the IPA chart. Irayo though.
Internet Acronyms Nìna'vi

hamletä tìralpuseng lena'vi sngolä'eiyi. tìkangkem si awngahu ro
http://bit.ly/53GnAB
The translation of Hamlet into Na'vi has started! Join with us at http://bit.ly/53GnAB

txo nga new oehu pivlltxe nìna'vi, nga oer 'eylan si mì fayspuk (http://bit.ly/bp9fwf)
If you want to speak na'vi to me, friend me on facebook (http://bit.ly/bp9fwf)

numena'viyä hapxì amezamkivohinve
learnnavi's

omängum fra'uti

Yeah there is no perfect mapping between this and the chart, it's just approximate.  I actually disagree with some of Wikipedia's English examples.  The best source I found of pronunciation examples is actually dictionary.com.  When you look at a word, it gives the spelled pronunciation, but you can switch that to IPA, then click the ? and it gives example words for all sounds you find in English, by IPA.
Ftxey lu nga tokx ftxey lu nga tirea? Lu oe tìkeftxo.
Listen to my Na'vi Lessons podcast!