Grammar tools (nouns, verbs, adjectives)

Started by Wllìm, August 06, 2014, 09:02:59 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Tìtstewan

This thread could help you. :)
nga is easy, only the genitive is different: nga - ngeyä
Same ending appears on po - peyä and oeng - oengeyä and fko - fkeyä

At oeng, with case ending the a appears: oeng - oengal
Plus the plural of ayoeng could be also awnga

Fko, poe and poan appears only in singular, that mean it was never seen a word like *pxefko, *ayfoe or *mefoan
(I personally see no reason why it shouldn't exist, but this would be a question for Pawl...)


-| Na'vi Vocab + Audio | Na'viteri as one HTML file | FAQ | Useful Links for Beginners |-
-| Kem si fu kem rä'ä si, ke lu tìfmi. |-

Wllìm

Irayo, that is useful, I'll look into it when I start making the pronoun tool :)

The adjective tool is ready now. :D

Tìtstewan


-| Na'vi Vocab + Audio | Na'viteri as one HTML file | FAQ | Useful Links for Beginners |-
-| Kem si fu kem rä'ä si, ke lu tìfmi. |-

archaic

Pasha, an Avatar story, my most recent fanfic, Avatar related, now complete.

The Dragon Affair my last fanfic, non Avatar related.

Plumps

A suggestion for the verb section, under mode ...

Concerning ‹us› and ‹awn›: it's noted that a "participle is an adjective made from a verb" but it doesn't say that they can only be used attributively. Is it possible to insert the -a- function for these two infixes as you did in the adjective section?

Also, if you have ‹us› and ‹awn›, not only are aspect, tense, intent and mode not possible but the infixes of transitivity as well. ;)

`Eylan Ayfalulukanä

By 'infixes of transitivity', I assume you mean <äp> and <eyk>?

This is turning into quite the useful little tool. Irayo nìtxan, ma Willim!

Yawey ngahu!
pamrel si ro [email protected]

Tirea Aean

Quote from: Plumps on August 09, 2014, 01:08:23 PM
Also, if you have ‹us› and ‹awn›, not only are aspect, tense, intent and mode not possible but the infixes of transitivity as well. ;)

How then would you say stuff like "the bathing animal" and such?

Wllìm

#27
Quote from: Plumps on August 09, 2014, 01:08:23 PM
Concerning ‹us› and ‹awn›: it's noted that a "participle is an adjective made from a verb" but it doesn't say that they can only be used attributively. Is it possible to insert the -a- function for these two infixes as you did in the adjective section?
Good idea, I'll try if that is possible :)

For the question whether <äp>/<eyk> can occur in a participle, I'll open a separate thread to avoid a mess here ;)
Edit: done, see here.

Wllìm

#28
Quote from: Plumps on August 09, 2014, 01:08:23 PM
Concerning ‹us› and ‹awn›: it's noted that a "participle is an adjective made from a verb" but it doesn't say that they can only be used attributively. Is it possible to insert the -a- function for these two infixes as you did in the adjective section?

Done :) The tool now shows these -a- forms when you ask for a participle. For example for taron:


Note: participles can only be used attributively. That is, you cannot say "<noun> lu tusaron"; only "<noun> a-tusaron" or "tusaron-a <noun>".



Quote from: Plumps on August 09, 2014, 01:08:23 PM
Also, if you have ‹us› and ‹awn›, not only are aspect, tense, intent and mode not possible but the infixes of transitivity as well. ;)

Pending the discussion in the other thread, I made it so that the tool shows a warning with those forms:


Warning: it is not known whether this form exists. See this discussion.



Edit: based on this, removed the <äp> + <awn> form (and <äpeyk> + <awn>) from the tool.

Tìtstewan

Quote from: Wllìm on August 09, 2014, 04:48:13 AM
By the way: can I report omissions in NaviData.sql somewhere?
Fìtsenge
I posted that stuff for you, before it get lost. :)




As for <eyk> and <us>, Plumps found this:
Quote from: Plumps on August 16, 2014, 04:45:18 AM
Part of an answer is already in a comment by Paul on naviteri.org concerning the infix use in zeyko (< zo) and if you write him again, I would make him aware of that lest he should have forgotten about that.

Quote from: Paul, 17 July, 2010Tsalsungay tsalì'u alu zeykuso lu eyawr. Slä zene fko pivlltxe san zäpeyko sìk. (*Zeykäpo lu keyawr.) Ulte kawkrr ke tsun fko pivlltxe san *zeykeyko!

[...]

The question of how the "pre-first position" infixes relate to each other and to the rest of the infix inventory is important, so I'll save it for a future post.
I'm going to remind Pawl on this. :)

-| Na'vi Vocab + Audio | Na'viteri as one HTML file | FAQ | Useful Links for Beginners |-
-| Kem si fu kem rä'ä si, ke lu tìfmi. |-

Tirea Aean

Quote from: Tirea Aean on September 21, 2014, 01:09:51 PM
Found an error on here http://wimiso.nl/navi/verbs/:

Result:

kät-eyk-ilv-äng-eng

käteng is a head-initial compound verb (not head-final compound like the result above would imply). it's made of:

vin. kä + adj. teng

therefore all infixes should go in the kä portion.

It was a Database error.

Quote from: http://naviteri.org/2010/09/getting-to-know-you-part-3/
käteng [k••äteng] vi. 'spend time with, hang out with.' No implication of dating or romance—simply passing time with friends.

Quote from: Eana Eltu
6864    käteng    ˈk·æ.tɛŋ    kät<1><2><3>eng    vin.

EDIT: But as you see, the dot is in the correct place. Eana Eltu still put <1><2><3> in the wrong place. This may be beyond our control. (except those with shell access to the perl scripts)

Beware of all compound verbs. If your project actually makes queries to the database directly, you will need a way to hard code correct them upon getting query results. For vrrtepcli, I have actually made a serverside hax.sh file that does sed operations on the database dumps for the certain compound verbs to which EE has incorrectly given infix location data. <1><2><3> may be in the incorrect place, even if the raised dot is in the correct place in the IPA.

An alternative way to get more correct results from compound verbs is to actually query for the IPA and put the infixes where the raised dot characters are, NOT where the <1><2><3> are. EE apparently assigns those using some weird algorithm having to do with vowels, and not taking into account head-initial or head-final type compound verbs.

Wllìm

(Long post, only read if you're interested in technical stuff ;))

I'll explain how my project gets the infix positions. Basically, I downloaded the Eana Eltu database dump (in SQL format), and created a little script to convert this to the JSON format needed for my application. (You can find this file here). The infix data comes from the infixes column; the script just detects "<1><2>" and "<3>" and replaces them by dots.

Unfortunately, I had to manually correct many problems where either no infix data was given in this column, and I had to remove forms like zeyko manually (to avoid people making zeyk-eyk-o in the application ;) ). I thought that the infixes column contained manually-entered data, so I did not expect that there were other mistakes in it. Sigh...

All manual changes make synchronizing the JSON with newer versions of the SQL dumps quite impossible, except for doing it manually. This is clearly sub-optimal, but in the SQL file there is just not enough information to do for example removal of forms like zeyko automatically :(

The point is that I don't even really know what Eana Eltu is or does; I guess it is a database that contains definitions for words in various languages, but it is hard for newcomers like me to find information about it ;) I didn't even know that there was any database dump of Na'vi words until Tìtstewan pointed that out to me. It seems that there are aspects of Eana Eltu that can be improved, e.g. the database layout. So my question is: is Eana Eltu still maintained? If not, it might be an idea to either modify it, or start from scratch... :-\ (I have some ideas for that :))

For now, I fixed käteng manually in the JSON file; I guess there are more mistakes. I'll go over the entire file sometime this week to try to fix them.

Tirea Aean

#32
For reference, here is my server-side script for vrrtepcli that corrects these incorrect entries in EE [only pasting relevent portions]:


as you see, the trouble words so far are:

emrey, yemstokx, 'awnìm, newomum, pllngay, and käteng.

You also see that wayä aylì'u has been corrected from having i to having ì.

vrrtepcli process:
Download the .sql;
recreate the database with the sql file;
dump the tables to .txt files;
dump certain columns to .txt files;
apply the hax.sh to the .txt tables;

vrrtepcli uses the raw .txt dumps (tsv format) of the actual database tables.

--

Eana Eltu is actually the very MySQL database from which the Learn Na'vi Dictionary PDF gets generated. Eana Eltu is also considered to be the Perl scripts which generate the MySQL database of the same name. Those Perl scripts and the process of using those to generate the PDF files were written/created by user Tuiq, to help out Taronyu, who create the first Na'vi Dictionary PDFs manually from LaTeX documents or something.

the Eana Eltu database, as generated by that .sql file has these tables of interest:

metaWords -- the table whose columns/rows look like this:

`id`,`navi`,`ipa`,`infixes`,`partOfSpeech`


and

localizedWords -- the table whose columns/rows look like this:

`id`,`languageCode`,`localized`,`partOfSpeech`


respectively. As you see, each Na'vi word in metaWords has a unique database ID number. And then its definitions are contained by that same ID in the localizedWords table in all the supported languages. (the Russian ru language has question marks because this was copy pasted from tirea.learnnavi.org/dictionarydata where I serve in UTF-8, but for some reason the Cyrillic characters don't show up)

There are also these tables in Eana Eltu:

metaInfixes -- list of all Na'vi affixes, whose columns/rows look like this:

`id`, `navi`, `ipa`, `shorthand`, `position`


and localizedInfixes -- definitions of all affixes, whose rows/columns look like this:

`id`, `languageCode`, `meaning`, `habitat`


and so on.

Basically, for this project, while converting to JSON, you will need to scan for these certain hardcoded words and fix them right before you assign their infix positions... Until someone fixes the Perl code for EE so that it generates correct data in the infixes column of the metaWords table.

Wllìm

I reorganized my server a bit, and in the process modified and cleaned up quite a bit of code for the grammar tools. Everything should still work as before; if there are any problems please let me know.

Also, if you have any ideas for future grammar tools, please share them here ;)

Vawmataw

Quote from: Wllìm on October 28, 2014, 04:33:46 PM
I reorganized my server a bit, and in the process modified and cleaned up quite a bit of code for the grammar tools. Everything should still work as before; if there are any problems please let me know.

Also, if you have any ideas for future grammar tools, please share them here ;)
One thing: Zeyko is a verb.
Fmawn Ta 'Rrta - News IN NA'VI ONLY (Discord)
Traducteur francophone de Kelutral.org, dict-navi et Reykunyu

Wllìm

Quote from: Vawmataw on October 28, 2014, 04:35:15 PM
One thing: Zeyko is verb.

Yes, but technically the root form is zo. So if you want to get zeyko, you can enter zo and click Causative.

The reason that you can't enter zeyko: consider what would happen if you then click Causative. You would get zeykeyko ;)

Vawmataw

#36
QuoteThe reason that you can't enter zeyko: consider what would happen if you then click Causative. You would get zeykeyko.
There are the causative of zo and the causative of zeyko. zeyko: make work (not to be confused with 'heal'); zeykeyko: make heal.
Fmawn Ta 'Rrta - News IN NA'VI ONLY (Discord)
Traducteur francophone de Kelutral.org, dict-navi et Reykunyu

Tìtstewan

Quote from: Vawmataw on October 28, 2014, 04:40:09 PM
QuoteThe reason that you can't enter zeyko: consider what would happen if you then click Causative. You would get zeykeyko.
There are the causative of zo and the causative of zeyko. zeyko: make fix; zeykeyko: make heal.
No! :o
The root verb of zeyko is zo, and zeyko itself is not a verb root in which you can add <eyk> again!
Pawl explained that in this comment on Na'viteri:
http://naviteri.org/2010/07/vocabulary-update/comment-page-1/#comment-95
Quote[...] Kemaweyan asked a good question, so let me explain this in English so it's available to more people. He noted that zeyko appears to be a garden-variety derivative of zo, with the causative infix -eyk- added: zeyko = z-eyk-o. If that's true, then why give it a special place in the vocabulary? This led Kemaweyan to speculate it might actually be a separate lexical item not derived via -eyk-, which could then lead to a form like *zeykeyko.

K.'s first impression was correct: zeyko is just the causative form of zo. I called it out only to emphasize its usefulness: it's the standard way to say "fix" or "heal." But he's right: It didn't have to be included.

-| Na'vi Vocab + Audio | Na'viteri as one HTML file | FAQ | Useful Links for Beginners |-
-| Kem si fu kem rä'ä si, ke lu tìfmi. |-

Vawmataw

#38
That's strange... Every verb in Na'vi has a causative, except zeyko. Does vezeyko have a causative?
So it's impossible in Na'vi that someone makes fix something.
Fmawn Ta 'Rrta - News IN NA'VI ONLY (Discord)
Traducteur francophone de Kelutral.org, dict-navi et Reykunyu

Tìtstewan

#39
Quote from: Vawmataw on October 28, 2014, 04:51:53 PM
Who is right? Wllìm or me?
Wllìm is correct as he did mean that you just use root zo and the button for causative. *zeykeyko is not possible as "zeyko" is not a root verb like zo.

Quote from: Vawmataw on October 28, 2014, 04:53:43 PM
That's strange... Every verb in Na'vi has a causative, except zeyko. Does vezeyko have a causative?
No, because vezeyko is created by the same way like zeyko.

-| Na'vi Vocab + Audio | Na'viteri as one HTML file | FAQ | Useful Links for Beginners |-
-| Kem si fu kem rä'ä si, ke lu tìfmi. |-