==Evaluation==

==Dependency Relations==
===advcl===
A number of the sentences in the corpora I used for this assignment had subordinating clauses, many of which had きゃー (while) as the subordinating conjunction. ぞぅが (but) is another common one. That subordinating conjunction is dependent on the root of the main sentence.
*Here is a sentence with this relation:
<pre>
"<むるうきゃー>"
"きゃー" cnjsub
"むる" v acc @advcl #1->7
"<っみゃ>"
"っみゃ" ij @discourse #2->7
"<やらびぬ>"
"やらび" n gen @nsubj #3->7
"<っみゃ>"
"っみゃ" ij @discourse #4->7
"<じてんしゃから>"
"じてんしゃ" n abl @obl #5->7
"<っみゃ>"
"っみゃ" ij @discourse #6->7
"<ってぃ>"
"ってぃ" v cvb_abs @root #7->0
"<っみゃ>"
"っみゃ" ij @discourse #8->7
"<。>"
"。" sent @punct #9->7
</pre>
*Here is the relevant bit of another example:
<pre>
"<あい>"
"あい" adv @advmod #12->13
"<むるうきゃー>"
"きゃー" cnjsub
"むる" v acc @advcl #13->16
"<やらびぬ>"
"やらび" n gen @nsubj #14->16
"<じてんしゃから>"
"じてんしゃ" n abl @obl #15->16
"<ってぃ>"
"ってぃ" v cvb_abs @root #16->0
</pre>

===aux===
There are a handful of words which I am treating as auxiliary verbs. The two main examples are the continuous and the resultative.
*An example with the continuous:
<pre>
"<やみ>"
"やみ" v cvb_abs #4
"<うい>"
"う" vaux @aux #5->4
</pre>
*An example with the resultative:
<pre>
"<ふぃー>"
"ふぃ" v cvb_abs @root #6->0
"<あいば>"
"ば" vaux
"あ" vaux @aux #7->6
</pre>

===discourse===
There are a large number of words tagged as discourse markers and interjections in this corpus. These words are dependent on the root of the sentence.
*Here is one example:
<pre>
"<うらあ>"
"うら" prn dem p3 sg nn top1 @dobj #1->3
"<ひてぃったま>"
"ったま" cl abs
"ひてぃ" num @nmod #1->2
"<とぅいー>"
"とぅい" v cvb_abs
"<ふぁうっちゃーんな>"
"ふぁう" v cvb_sim top1
"<ふぅーきゃー>"
"きゃー" cnjsub
"ふぅ" v cvb_abs @root #5->0
"<っみゃ>"
"っみゃ" ij @discourse #6->5
"<。>"
"。" sent @punct #7->5
</pre>
*There is another example here:
<pre>
"<なうがら>"
"なうがら" ij @discourse #1->6
"<うぬ>"
"うぬ" det dem @det #2->3
"<おじさんな>"
"おじさん" n top1 @nsubj #3->4
"<っさんそぅが>"
"そぅが" cnjsub
"っさ" v neg
"<ぬっさ>"
"ぬす" n top1 @nsubj #5->6
"<っさんそぅが>"
"そぅが" cnjsub
"っさ" v neg @root #6->0
</pre>

===dobj===
Direct objects are often marked with the accusative marker う in Miyako. However, in this corpus, there were a number of cases where I tagged a word as a direct object despite the use of a different particle, since that seemed to be how it was being used.
*じてんしゃうくぎー is an example with the accusative marker.
<pre>
"<じてんしゃう>"
"じてんしゃ" n acc @dobj #14->15
"<くぎー>"
"くず" v cvb_abs #15
</pre>
*This same combination of words also appears without the accusative marker.
<pre>
"<じてんしゃ>"
"じてんしゃ" n abs @dobj #6->7
"<くぎー>"
"くず" v cvb_abs
</pre>

===nmod===
This is the connection between a noun and another noun that is modifying it. That often happens with the genitive marker, but not always. Furthermore, I suspect that the genitive and nominative glosses were at times confused when the pear story was being glossed, as they can both be either が or ぬ, and there are instances where the gloss simply does not make sense. There are also multiple instances with number words.
*An example with the genitive marker:
<pre>
"<びきやらびぬ>"
"びきやらび" n gen @nmod #2->3
"<みちゃーいや>"
"ちゃーい" cl top1
"みー" num @nsubj #3->4
</pre>
*Here is an example with a number word:
<pre>
"<うらあ>"
"うら" prn dem p3 sg nn top1 @dobj #1->3
"<ひてぃったま>"
"ったま" cl abs
"ひてぃ" num @nmod #2->1
</pre>

[[Category:Sp17_UD]]
[[Category:Miyako]]

Miyako/Universal Dependencies

2017-05-13T05:01:27Z

2017-05-08T02:45:12Z

Doldham1: /* twol */

== Resources for machine translation between [[Okinawan|Okinawan]] and [[Miyako|Miyako]] ==

* [[Okinawan and Miyako/Contrastive Grammar|Contrastive Grammar]]
* [[Okinawan and Miyako/Lexical selection|Lexical Selection]]
* [[Okinawan and Miyako/Structural transfer|Structural Transfer]]
* [https://github.swarthmore.edu/aconant1/ling073-ryu-mvi-corpus Parallel corpus repo]
* [https://github.swarthmore.edu/aconant1/ling073-ryu-mvi Language pair repo]

[[Category:Sp17_TranslationPairs]]

==Lexical Selection ==
*In Miyako, はい can mean field, needle, or south, and is also a causative auxiliary verb. If it is followed by string, it is probably needle.
*Hand and arm are the same word in Okinawan (てぃい), but different words in Miyako. If てぃい is followed by an instrumental, we are assuming it is hand.

==Evaluation==
===Evaluation as of lexical selection===
====ryu → mvi Evaluation ====
=====Evaluation of tests=====
*WER: 100 %
*PER: 100 %
*Coverage: 89%

=====Evaluation of sentences=====
*WER: 90.32 %
*PER: 90.32 %
*Coverage: 62.1%

====mvi → ryu Evaluation ====
=====Evaluation of tests=====
*WER: 88.89 %
*PER: 77.78 %
*Coverage: 100 %

=====Evaluation of sentences=====
*WER: 100%
*PER: 100%
*Coverage: 80.65%

===Final Evaluation===
====ryu Evaluation====
*Precision:
*Recall:
*Coverage of large:
*Number of words in large
*Number of stems in the transducer:

====mvi Evaluation====
*Precision:
*Recall:
*Coverage of large:
*Number of words in large
*Number of stems in the transducer:

====ryu → mvi Evaluation ====
=====Evaluation of longer=====
*WER:
*PER:
*Coverage:
*Trimmed coverage:
*Number of tokens:

=====Evaluation of large=====
*Trimmed coverage:
*Number of tokens:

====mvi → ryu Evaluation ====
=====Evaluation of longer=====
*WER:
*PER:
*Coverage:
*Trimmed coverage:
*Number of tokens:

=====Evaluation of large=====
*Trimmed coverage:
*Number of tokens:

====Expansion of Miyako transducer and ryu → mvi====
=====Morphology=====
*I added more numbers. 6 is not included because I am confused as to what it is.
*I expanded classifiers, adding days, portion, group, and people
*I changed how the focus marker worked, such that it was a lexicon rather than being hard-coded. This also means that the verbs which take focus markers can now get them.
*: Old output: <code>^ぼーしなてぃどぅ/*ぼーしなてぃどぅ$</code>
*: New output: <code>^ぼーしなてぃどぅ/ぼーし<n><abs>+な<mod><quot><foc>$</code>
*: Old output: <code>^そぅだてぃどぅ/*そぅだてぃどぅ$</code>
*: New output: <code>^そぅだてぃどぅ/そぅだてぃ<v><cvb_abs><foc>$</code>
*I implemented the resultative and causal.
*: Old output: <code>^あいば/*あいば$</code>
*: New output: <code>^あいば/あい<vaux><vaux>$</code>
*I hard-coded the forms I have for do.
*I added an additional lexicon certain verb forms go through to attach but to the end of them.

=====twol=====
*I added a rule such that the accusative marker goes to う after something ending in a or u.
*I added a rule such that the accusative marker changes to the appropriate thing, such as going to ぬ after something ending in ん.
*: Old output: <code>^みんぬ/みん<n><gen>$</code>
*: New output: <code>^みんぬ/みん<n><acc>/みん<n><gen>$</code>
*I added a rule such that {っ} goes to じ after a word ending in ず.
*I added a rule such that ず goes to っ before the topic and accusative markers.
*I added a rule such that す goes to っ before the accusative marker.
*I added a rule such that {っ} goes to そ after a word ending in す and before an accusative marker.

====Expansion of Okinawan transducer and mvi → ryu====