http://wikis.swarthmore.edu/ling073/api.php?action=feedcontributions&feedformat=atom&user=Eschalk1LING073 - User contributions [en]2026-10-25T04:35:48ZUser contributionsMediaWiki 1.27.7http://wikis.swarthmore.edu/ling073/index.php?title=Eschalk1/Final_project&diff=7753Eschalk1/Final project2018-05-15T18:39:18Z<p>Eschalk1: </p>
<hr />
<div>I improved the Apertium transducer for Latin by adding stems and paradigms. I wrote a Python script to automatically generate stem entries from a vocab list downloaded from Haverford's The Bridge.<br />
<br />
My transducer can be found at https://github.com/cmdalton/apertium-lat. The Python script which automates stem generation can be found at https://github.com/eschalk/apertium-lat-stemgenerator<br />
[[File:Slide1.jpg]]<br />
<br />
There is much ongoing work to be done on the Latin transducer before it can be called a mature tagger. Some of these future goals include expanding the verb forms covered by the Latin transducer, restructuring the verb paradigms, adding disambiguation rules, and further developing the Python script that generates vocab.<br />
<br />
[[Category:Sp18_FinalProjects]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Eschalk1/Final_project&diff=7752Eschalk1/Final project2018-05-15T18:38:48Z<p>Eschalk1: </p>
<hr />
<div>I improved the Apertium transducer for Latin by adding stems and paradigms. I wrote a Python script to automatically generate stem entries from a vocab list downloaded from Haverford's The Bridge.<br />
<br />
My transducer can be found at https://github.com/cmdalton/apertium-lat. The Python script which automates stem generation can be found at https://github.com/eschalk/apertium-lat-stemgenerator<br />
[[File:Slide1.jpg]]<br />
<br />
There is much ongoing work to be done on the Latin transducer before it can be called a mature tagger. Some of these future goals include expanding the verb forms covered by the Latin transducer, restructuring the verb paradigms, adding disambiguation rules, and further developing the Python script that generates vocab.</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Eschalk1/Final_project&diff=7751Eschalk1/Final project2018-05-15T18:36:20Z<p>Eschalk1: </p>
<hr />
<div>I improved the Apertium transducer for Latin by adding stems and paradigms. I wrote a Python script to automatically generate stem entries from a vocab list downloaded from Haverford's The Bridge.<br />
<br />
My transducer can be found at https://github.com/cmdalton/apertium-lat. The Python script which automates stem generation can be found at https://github.com/eschalk/apertium-lat-stemgenerator<br />
[[File:Slide1.jpg]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=File:Slide1.jpg&diff=7750File:Slide1.jpg2018-05-15T18:35:51Z<p>Eschalk1: Poster presentation for Ling073 about Apertium Latin transducer updates.</p>
<hr />
<div>Poster presentation for Ling073 about Apertium Latin transducer updates.</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Eschalk1/Final_project&diff=7749Eschalk1/Final project2018-05-15T15:39:30Z<p>Eschalk1: </p>
<hr />
<div>I improved the Apertium transducer for Latin by adding stems and paradigms. I wrote a Python script to automatically generate stem entries from a vocab list downloaded from Haverford's The Bridge.<br />
<br />
My transducer can be found at https://github.com/cmdalton/apertium-lat. The Python script which automates stem generation can be found at https://github.com/eschalk/apertium-lat-stemgenerator<br />
[[File:LingPoster.pdf]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Eschalk1/Final_project&diff=7748Eschalk1/Final project2018-05-15T15:38:57Z<p>Eschalk1: </p>
<hr />
<div>I improved the Apertium transducer for Latin by adding stems and paradigms. I wrote a Python script to automatically generate stem entries from a vocab list downloaded from Haverford's The Bridge.<br />
<br />
My transducer can be found at https://github.com/cmdalton/apertium-lat. The Python script which automates stem generation can be found at https://github.com/eschalk/apertium-lat-stemgenerator<br />
[[File:lingposter.pdf]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=File:LingPoster.pdf&diff=7747File:LingPoster.pdf2018-05-15T15:38:35Z<p>Eschalk1: Poster presentation of Latin transducer improvements</p>
<hr />
<div>Poster presentation of Latin transducer improvements</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Eschalk1/Final_project&diff=7746Eschalk1/Final project2018-05-15T15:37:21Z<p>Eschalk1: </p>
<hr />
<div>I improved the Apertium transducer for Latin by adding stems and paradigms. I wrote a Python script to automatically generate stem entries from a vocab list downloaded from Haverford's The Bridge.<br />
<br />
My transducer can be found at https://github.com/cmdalton/apertium-lat. The Python script which automates stem generation can be found at https://github.com/eschalk/apertium-lat-stemgenerator</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Eschalk1/Final_project&diff=7745Eschalk1/Final project2018-05-15T15:25:39Z<p>Eschalk1: Created page with "I improved the Apertium transducer for Latin by adding stems and paradigms. I wrote a Python script to automatically generate stem entries from a vocab list downloaded from Ha..."</p>
<hr />
<div>I improved the Apertium transducer for Latin by adding stems and paradigms. I wrote a Python script to automatically generate stem entries from a vocab list downloaded from Haverford's The Bridge.<br />
<br />
My transducer can be found at https://github.com/cmdalton/apertium-lat. The Python script which automates stem generation can be found at</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Uzbek/Universal_Dependencies&diff=7691Uzbek/Universal Dependencies2018-05-03T16:02:07Z<p>Eschalk1: /* @acl */</p>
<hr />
<div>== Evaluation ==<br />
<br />
{| class="wikitable"<br />
! <br />
! uzb.annotated.ud.conllu<br />
! uzb.annotated2.ud.conllu<br />
|-<br />
! uzb.withmorph.udpipe<br />
| UAS: 82.07%, LAS: 79.68% || UAS: 42.98%, LAS: 35.09%<br />
|-<br />
! uzb.nomorph.udpipe<br />
| UAS: 72.11%, LAS: 69.72%<br />
| UAS: 43.86%, LAS: 35.09%<br />
|}<br />
<br />
=== uzb.annotated.ud.connlu ===<br />
<br />
forms: 251<br />
<br />
sentences: 30<br />
<br />
=== uzb.annotated2.ud.connlu ===<br />
<br />
forms: 114<br />
<br />
sentences: 9<br />
<br />
== Dependency relations ==<br />
<br />
=== @nsubj ===<br />
<br />
@nsubj (nominal subject) denotes a semantically agent-like and syntactically subject-like nominal. Note that it cannot be used for pleonastic subjects.<br />
<br />
In Uzbek, @nsubj will usually be assigned to a noun in the nominative case. Generally speaking, a noun in the nominative case is the subject of the sentence and the doer of the action. An example of @nsubj can be seen in the following sentence from our corpus.<br />
<br />
<code><br />
"<Azalda>"<br />
<br />
"azal" n loc @obl #1->6<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #2->6<br />
<br />
"<osmon>"<br />
<br />
"osmon" n nom @obj #3->6<br />
<br />
"<bilan>"<br />
<br />
"bilan" cnjcoo @cc #4->5<br />
<br />
"<yerni>"<br />
<br />
"yer" n acc @conj #5->3<br />
<br />
"<yaratdi>"<br />
<br />
"yarat" v tv past p3 sg @root #6->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #7->6<br />
<br />
</code><br />
<br />
''Azalda Xudo osmon bilan yerni yaratdi.''<br />
<br />
In.the.beginning God heaven and earth created.<br />
<br />
"In the beginning, God created the heavens and the earth."<br />
<br />
Note that not all nouns in the nominative case are @nsubj. For some reason, ''osmon'' "heaven" in this sentence is in the nominative case even though it is an object and a patient. Perhaps the accusative ending on ''yerni'' "earth" is able to cover both conjuncts.<br />
<br />
Sometimes @nsubj can be assigned to things that are not prototypically nouns. This occurs in the following sentence.<br />
<br />
<code><br />
"<Kech>"<br />
<br />
"kech" adv subst nom @nsubj #1->2<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @root #2->0<br />
<br />
"<,>"<br />
<br />
"," cm @cc #3->5<br />
<br />
"<erta>"<br />
<br />
"erta" adv subst nom @nsubj #4->5<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @conj #5->2<br />
<br />
"<–>"<br />
<br />
"-" guio @cc #6->7<br />
<br />
"<bu>"<br />
<br />
"bu" prn dem nom @conj #7->2<br />
<br />
"<birinchi>"<br />
<br />
"bir" num ord @nmod #8->9<br />
<br />
"<kun>"<br />
<br />
"kun" n nom @nsubj #9->7<br />
<br />
"<edi>"<br />
<br />
"e" cop past p3 sg @cop #10->7<br />
<br />
"<.>"<br />
<br />
"." sent @punct #11->2<br />
<br />
</code><br />
<br />
''Kech bo'ldi, erta bo'ldi - bu birinchi kun edi.''<br />
<br />
Evening was, morning was - this first day was.<br />
<br />
"And the evening and the morning were the first day."<br />
<br />
According to both the transducer and the CTILD dictionary, ''kech'' "evening" and ''erta'' "morning" are actually adverbs, but in this sentence, they fill the role of nominal subject.<br />
<br />
=== @obj ===<br />
<br />
@obj is usually applied to a patient-like constituent. In Uzbek, the constituent designated @obj will often be a noun in the accusative case. This is seen in the sentence below from our corpus.<br />
<br />
<code><br />
"<Ushbu>"<br />
<br />
"ushbu" prn dem nom @det #1->2<br />
<br />
"<gumbazni>"<br />
<br />
"gumbaz" n acc @obj #2->7<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @advmod #3->7<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #4->7<br />
<br />
"<osmon>"<br />
<br />
"osmon" n nom @xcomp #5->7<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #6->7<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @root #7->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #8->7<br />
</code><br />
<br />
''Ushbu gumbazni esa Xudo osmon deb atadi.''<br />
<br />
This dome whereas God heaven that called.<br />
<br />
"God called the expanse heaven."<br />
<br />
However, sometimes the @obj can be in other cases. In the sentence below, the obj ''suvlarning'' "waters" is in the genitive case.<br />
<br />
<code><br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->5<br />
<br />
"<quruqlikni>"<br />
<br />
"quruqlik" n acc @obj #2->5<br />
<br />
"<yer>"<br />
<br />
"yer" n nom @xcomp #3->5<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #4->5<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @root #5->0<br />
<br />
"<,>"<br />
<br />
"," cm @cc #6->12<br />
<br />
"<suvlarning>"<br />
<br />
"suv" n pl gen @obj #7->12<br />
<br />
"<jamini>"<br />
<br />
"jami" adj subst acc @nmod #8->7<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @advmod #9->12<br />
<br />
"<dengizlar>"<br />
<br />
"dengiz" n pl nom @xcomp #10->12<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #11->12<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @conj #12->5<br />
<br />
"<.>"<br />
<br />
"." sent @punct #13->5<br />
</code><br />
<br />
''Xudo quruqlikni yer deb atadi, suvlarning jamini esa dengizlar deb atadi.''<br />
<br />
God land earth that called, waters gathering whereas sea that called.<br />
<br />
"God called the dry land earth, and the gathering of the waters He called seas."<br />
<br />
=== @parataxis ===<br />
<br />
@parataxis is the relationship between the main predicate of a sentence and other elements "placed side by side without any explicit coordination, subordination, or argument relation with the head word." Parataxis is used in a variety of situations where the relationship between main sentence and clause is not explicit, especially in direct or reported speech (and so can often be identified w/ punctuation like ";" or ":"). @parataxis can be contrasted against @ccomp, which is used when the reported speech is clearly in a subordinate clause of the main verb. We used @parataxis to identify reported speech in our corpus.<br />
<br />
<code><br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->8<br />
<br />
"<:>"<br />
<br />
":" sent @punct #2->8<br />
<br />
"<«>"<br />
<br />
"«" lquot @punct #3->5<br />
<br />
"<Yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #4->5<br />
<br />
"<bo‘lsin>"<br />
<br />
"bo'l" v tv perm @parataxis #5->8<br />
<br />
"<»>"<br />
<br />
"»" rquot @punct #6->5<br />
<br />
"<,>"<br />
<br />
"," cm @punct #7->8<br />
<br />
"<dedi>"<br />
<br />
"de" v tv past p3 sg @root #8->0<br />
<br />
"<va>"<br />
<br />
"va" cnjcoo @cc #9->12<br />
<br />
"<yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #10->12<br />
<br />
"<paydo>"<br />
<br />
"paydo" adj @nmod #11->10<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @conj #12->8<br />
<br />
"<.>"<br />
<br />
"." sent @punct #13->8<br />
<br />
</code><br />
<br />
@Parataxis is especially helpful in these examples because the syntax of reported speech in Uzbek is even less coordinated than in English: the sentence begins with the speaker, launches into the reported speech, and we only get the speaking verb at the end of the sentence. <br />
<br />
<code><br />
<br />
"<Yana>"<br />
<br />
"yana" adv @advmod #1->17<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #2->17<br />
<br />
"<:>"<br />
<br />
":" sent @punct #3->17<br />
<br />
"<«>"<br />
<br />
"«" lquot @punct #4->13<br />
<br />
"<Suvliklar>"<br />
<br />
"suv" n loc @nmod #5->11<br />
<br />
"<o‘rtasida>"<br />
<br />
"o'rta" n px3sp loc @case #6->5<br />
<br />
"<ularni>"<br />
<br />
"ular" prn pers p3 pl acc @obj #7->9<br />
<br />
"<bir-biridan>"<br />
<br />
"bir-bir" prn ind px3sp abl @nmod #8->7<br />
<br />
"<ajratib>"<br />
<br />
"ajra" v iv past p3 sg @acl #9->11<br />
<br />
"<turadigan>"<br />
<br />
"tur" v iv past p3 sg @conj #10->9<br />
<br />
"<gumbaz>"<br />
<br />
"gumbaz" n nom @nsubj #11->13<br />
<br />
"<paydo>"<br />
<br />
"paydo" adj @nmod #12->11<br />
<br />
"<bo‘lsin>"<br />
<br />
"bo'l" v tv perm @parataxis #13->17<br />
<br />
<br />
"<»>"<br />
<br />
"»" rquot @punct #14->13<br />
<br />
"<,>"<br />
<br />
"," cm @punct #15->17<br />
<br />
"<–>"<br />
<br />
"-" guio @punct #16->17<br />
<br />
"<dedi>"<br />
<br />
"de" v tv past p3 sg @root #17->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #18->17<br />
<br />
</code><br />
<br />
We also have examples where @acl (clausal modifiers of nouns) are embedded within @parataxis. Which brings us to our next point...<br />
<br />
=== @acl ===<br />
<br />
We used @acl for clauses acting as adjectives to modify nouns. This was used often in our corpus to describe the various things created by god in the Bible.<br />
<br />
<code>"<Yer>"<br />
<br />
"yer" n nom @root #1->0<br />
<br />
"<turiga>"<br />
<br />
"tur" n px3sp dat @orphan #2->1<br />
<br />
"<ko‘ra>"<br />
<br />
"ko'ra" post @case #3->4<br />
<br />
"<o‘simliklarni>"<br />
<br />
"o'simlik" n pl acc @nmod #4->2<br />
<br />
"<:>"<br />
<br />
":" sent @punct #5->1<br />
<br />
"<urug‘>"<br />
<br />
"urug'" n nom @obj #6->7<br />
<br />
"<beradigan>"<br />
<br />
"ber" v tv past p3 sg @acl #7->8<br />
<br />
"<o‘t-o‘lanlar>"<br />
<br />
"o't" n nom @appos #8->2<br />
<br />
"<hamda>"<br />
<br />
"hamda" cnjcoo @cc #9->14<br />
<br />
"<mevasida>"<br />
<br />
"meva" n px3sp loc @amod #10->11<br />
<br />
"<urug‘i>"<br />
<br />
"urug'i" n loc @obl #11->14<br />
<br />
"<bor>"<br />
<br />
"bor" adj @amod #12->14<br />
<br />
"<mevali>"<br />
<br />
"mevali" adj @amod #13->14<br />
<br />
"<daraxtlarni>"<br />
<br />
"daraxt" n pl acc @conj #14->8<br />
<br />
"<undirdi>"<br />
<br />
"*undirdi" @amod #15->14<br />
<br />
"<.>"<br />
<br />
"." sent @punct #16->1</code><br />
<br />
In many cases, these clauses seem to convey some sort of purpose.<br />
<br />
<code>"<Yana>"<br />
<br />
"yana" adv @advmod #1->17<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #2->17<br />
<br />
"<:>"<br />
<br />
":" sent @punct #3->17<br />
<br />
"<«>"<br />
<br />
"«" lquot @punct #4->13<br />
<br />
"<Suvliklar>"<br />
<br />
"suv" n loc @nmod #5->11<br />
<br />
"<o‘rtasida>"<br />
<br />
"o'rta" n px3sp loc @case #6->5<br />
<br />
"<ularni>"<br />
<br />
"ular" prn pers p3 pl acc @obj #7->9<br />
<br />
"<bir-biridan>"<br />
<br />
"bir-bir" prn ind px3sp abl @nmod #8->7<br />
<br />
"<ajratib>"<br />
<br />
"ajra" v iv past p3 sg @acl #9->11<br />
<br />
"<turadigan>"<br />
<br />
"tur" v iv past p3 sg @conj #10->9<br />
<br />
"<gumbaz>"<br />
<br />
"gumbaz" n nom @nsubj #11->13<br />
<br />
"<paydo>"<br />
<br />
"paydo" adj @nmod #12->11<br />
<br />
"<bo‘lsin>"<br />
<br />
"bo'l" v tv perm @parataxis #13->17<br />
<br />
"<»>"<br />
<br />
"»" rquot @punct #14->13<br />
<br />
"<,>"<br />
<br />
"," cm @punct #15->17<br />
<br />
"<–>"<br />
<br />
"-" guio @punct #16->17<br />
<br />
"<dedi>"<br />
<br />
"de" v tv past p3 sg @root #17->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #18->17</code><br />
<br />
=== @cc ===<br />
<br />
@cc links a coordinating conjunction to the conjunct that comes after it. In Uzbek, one common coordinating conjunction is ''va'', which appears in the sentence below.<br />
<br />
<code><br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->8<br />
<br />
"<:>"<br />
<br />
":" sent @punct #2->8<br />
<br />
"<«>"<br />
<br />
"«" lquot @punct #3->5<br />
<br />
"<Yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #4->5<br />
<br />
"<bo‘lsin>"<br />
<br />
"bo'l" v tv perm @parataxis #5->8<br />
<br />
"<»>"<br />
<br />
"»" rquot @punct #6->5<br />
<br />
"<,>"<br />
<br />
"," cm @punct #7->8<br />
<br />
"<dedi>"<br />
<br />
"de" v tv past p3 sg @root #8->0<br />
<br />
"<va>"<br />
<br />
"va" cnjcoo @cc #9->12<br />
<br />
"<yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #10->12<br />
<br />
"<paydo>"<br />
<br />
"paydo" adj @nmod #11->10<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @conj #12->8<br />
<br />
"<.>"<br />
<br />
"." sent @punct #13->8<br />
<br />
</code><br />
<br />
''Xudo: "Yorug'lik bo'lsin", dedi va yorug'lik paydo bo'ldi.''<br />
<br />
God: "Light be", said and light appear did.<br />
<br />
"God said, 'Let there be light,' and there was light."<br />
<br />
This ''va'' isn't in the CTILD dictionary, but based on this sentence, it seems to mean something like "and." Another common @cc is ''esa'' "whereas":<br />
<br />
<code><br />
"<Xudoning>"<br />
<br />
"xudo" n gen @nmod #1->2<br />
<br />
"<Ruhi>"<br />
<br />
"ruh" n px3sp nom @root #2->0<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @cc #3->2<br />
<br />
"<suv>"<br />
<br />
"suv" n nom @nmod #4->6<br />
<br />
"<yuzida>"<br />
<br />
"yuz" num subst px3sp loc @case #5->4<br />
<br />
"<parvoz>"<br />
<br />
"parvoz" n nom @nsubj #6->2<br />
<br />
"<etar>"<br />
<br />
"etar" v tv past p3 sg @aux #7->8<br />
<br />
"<edi>"<br />
<br />
"e" cop past p3 sg @cop #8->2<br />
<br />
"<.>"<br />
<br />
"." sent @punct #9->2<br />
</code><br />
<br />
''Xudoning Ruhi esa suv yuzida parvoz etar edi.''<br />
<br />
God's Spirit whereas water face flight did was.<br />
<br />
"And the spirit of God was hovering over the surface of the waters."<br />
<br />
[[Category: sp18_UD]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Uzbek/Universal_Dependencies&diff=7690Uzbek/Universal Dependencies2018-05-03T15:30:42Z<p>Eschalk1: /* @acl */</p>
<hr />
<div>== Evaluation ==<br />
<br />
{| class="wikitable"<br />
! <br />
! uzb.annotated.ud.conllu<br />
! uzb.annotated2.ud.conllu<br />
|-<br />
! uzb.withmorph.udpipe<br />
| UAS: 82.07%, LAS: 79.68% || UAS: 42.98%, LAS: 35.09%<br />
|-<br />
! uzb.nomorph.udpipe<br />
| UAS: 72.11%, LAS: 69.72%<br />
| UAS: 43.86%, LAS: 35.09%<br />
|}<br />
<br />
=== uzb.annotated.ud.connlu ===<br />
<br />
forms: 251<br />
<br />
sentences: 30<br />
<br />
=== uzb.annotated2.ud.connlu ===<br />
<br />
forms: 114<br />
<br />
sentences: 9<br />
<br />
== Dependency relations ==<br />
<br />
=== @nsubj ===<br />
<br />
@nsubj (nominal subject) denotes a semantically agent-like and syntactically subject-like nominal. Note that it cannot be used for pleonastic subjects.<br />
<br />
In Uzbek, @nsubj will usually be assigned to a noun in the nominative case. Generally speaking, a noun in the nominative case is the subject of the sentence and the doer of the action. An example of @nsubj can be seen in the following sentence from our corpus.<br />
<br />
<code><br />
"<Azalda>"<br />
<br />
"azal" n loc @obl #1->6<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #2->6<br />
<br />
"<osmon>"<br />
<br />
"osmon" n nom @obj #3->6<br />
<br />
"<bilan>"<br />
<br />
"bilan" cnjcoo @cc #4->5<br />
<br />
"<yerni>"<br />
<br />
"yer" n acc @conj #5->3<br />
<br />
"<yaratdi>"<br />
<br />
"yarat" v tv past p3 sg @root #6->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #7->6<br />
<br />
</code><br />
<br />
''Azalda Xudo osmon bilan yerni yaratdi.''<br />
<br />
In.the.beginning God heaven and earth created.<br />
<br />
"In the beginning, God created the heavens and the earth."<br />
<br />
Note that not all nouns in the nominative case are @nsubj. For some reason, ''osmon'' "heaven" in this sentence is in the nominative case even though it is an object and a patient. Perhaps the accusative ending on ''yerni'' "earth" is able to cover both conjuncts.<br />
<br />
Sometimes @nsubj can be assigned to things that are not prototypically nouns. This occurs in the following sentence.<br />
<br />
<code><br />
"<Kech>"<br />
<br />
"kech" adv subst nom @nsubj #1->2<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @root #2->0<br />
<br />
"<,>"<br />
<br />
"," cm @cc #3->5<br />
<br />
"<erta>"<br />
<br />
"erta" adv subst nom @nsubj #4->5<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @conj #5->2<br />
<br />
"<–>"<br />
<br />
"-" guio @cc #6->7<br />
<br />
"<bu>"<br />
<br />
"bu" prn dem nom @conj #7->2<br />
<br />
"<birinchi>"<br />
<br />
"bir" num ord @nmod #8->9<br />
<br />
"<kun>"<br />
<br />
"kun" n nom @nsubj #9->7<br />
<br />
"<edi>"<br />
<br />
"e" cop past p3 sg @cop #10->7<br />
<br />
"<.>"<br />
<br />
"." sent @punct #11->2<br />
<br />
</code><br />
<br />
''Kech bo'ldi, erta bo'ldi - bu birinchi kun edi.''<br />
<br />
Evening was, morning was - this first day was.<br />
<br />
"And the evening and the morning were the first day."<br />
<br />
According to both the transducer and the CTILD dictionary, ''kech'' "evening" and ''erta'' "morning" are actually adverbs, but in this sentence, they fill the role of nominal subject.<br />
<br />
=== @obj ===<br />
<br />
@obj is usually applied to a patient-like constituent. In Uzbek, the constituent designated @obj will often be a noun in the accusative case. This is seen in the sentence below from our corpus.<br />
<br />
<code><br />
"<Ushbu>"<br />
<br />
"ushbu" prn dem nom @det #1->2<br />
<br />
"<gumbazni>"<br />
<br />
"gumbaz" n acc @obj #2->7<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @advmod #3->7<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #4->7<br />
<br />
"<osmon>"<br />
<br />
"osmon" n nom @xcomp #5->7<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #6->7<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @root #7->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #8->7<br />
</code><br />
<br />
''Ushbu gumbazni esa Xudo osmon deb atadi.''<br />
<br />
This dome whereas God heaven that called.<br />
<br />
"God called the expanse heaven."<br />
<br />
However, sometimes the @obj can be in other cases. In the sentence below, the obj ''suvlarning'' "waters" is in the genitive case.<br />
<br />
<code><br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->5<br />
<br />
"<quruqlikni>"<br />
<br />
"quruqlik" n acc @obj #2->5<br />
<br />
"<yer>"<br />
<br />
"yer" n nom @xcomp #3->5<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #4->5<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @root #5->0<br />
<br />
"<,>"<br />
<br />
"," cm @cc #6->12<br />
<br />
"<suvlarning>"<br />
<br />
"suv" n pl gen @obj #7->12<br />
<br />
"<jamini>"<br />
<br />
"jami" adj subst acc @nmod #8->7<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @advmod #9->12<br />
<br />
"<dengizlar>"<br />
<br />
"dengiz" n pl nom @xcomp #10->12<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #11->12<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @conj #12->5<br />
<br />
"<.>"<br />
<br />
"." sent @punct #13->5<br />
</code><br />
<br />
''Xudo quruqlikni yer deb atadi, suvlarning jamini esa dengizlar deb atadi.''<br />
<br />
God land earth that called, waters gathering whereas sea that called.<br />
<br />
"God called the dry land earth, and the gathering of the waters He called seas."<br />
<br />
=== @parataxis ===<br />
<br />
@parataxis is the relationship between the main predicate of a sentence and other elements "placed side by side without any explicit coordination, subordination, or argument relation with the head word." Parataxis is used in a variety of situations where the relationship between main sentence and clause is not explicit, especially in direct or reported speech (and so can often be identified w/ punctuation like ";" or ":"). @parataxis can be contrasted against @ccomp, which is used when the reported speech is clearly in a subordinate clause of the main verb. We used @parataxis to identify reported speech in our corpus.<br />
<br />
<code><br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->8<br />
<br />
"<:>"<br />
<br />
":" sent @punct #2->8<br />
<br />
"<«>"<br />
<br />
"«" lquot @punct #3->5<br />
<br />
"<Yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #4->5<br />
<br />
"<bo‘lsin>"<br />
<br />
"bo'l" v tv perm @parataxis #5->8<br />
<br />
"<»>"<br />
<br />
"»" rquot @punct #6->5<br />
<br />
"<,>"<br />
<br />
"," cm @punct #7->8<br />
<br />
"<dedi>"<br />
<br />
"de" v tv past p3 sg @root #8->0<br />
<br />
"<va>"<br />
<br />
"va" cnjcoo @cc #9->12<br />
<br />
"<yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #10->12<br />
<br />
"<paydo>"<br />
<br />
"paydo" adj @nmod #11->10<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @conj #12->8<br />
<br />
"<.>"<br />
<br />
"." sent @punct #13->8<br />
<br />
</code><br />
<br />
@Parataxis is especially helpful in these examples because the syntax of reported speech in Uzbek is even less coordinated than in English: the sentence begins with the speaker, launches into the reported speech, and we only get the speaking verb at the end of the sentence. <br />
<br />
<code><br />
<br />
"<Yana>"<br />
<br />
"yana" adv @advmod #1->17<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #2->17<br />
<br />
"<:>"<br />
<br />
":" sent @punct #3->17<br />
<br />
"<«>"<br />
<br />
"«" lquot @punct #4->13<br />
<br />
"<Suvliklar>"<br />
<br />
"suv" n loc @nmod #5->11<br />
<br />
"<o‘rtasida>"<br />
<br />
"o'rta" n px3sp loc @case #6->5<br />
<br />
"<ularni>"<br />
<br />
"ular" prn pers p3 pl acc @obj #7->9<br />
<br />
"<bir-biridan>"<br />
<br />
"bir-bir" prn ind px3sp abl @nmod #8->7<br />
<br />
"<ajratib>"<br />
<br />
"ajra" v iv past p3 sg @acl #9->11<br />
<br />
"<turadigan>"<br />
<br />
"tur" v iv past p3 sg @conj #10->9<br />
<br />
"<gumbaz>"<br />
<br />
"gumbaz" n nom @nsubj #11->13<br />
<br />
"<paydo>"<br />
<br />
"paydo" adj @nmod #12->11<br />
<br />
"<bo‘lsin>"<br />
<br />
"bo'l" v tv perm @parataxis #13->17<br />
<br />
<br />
"<»>"<br />
<br />
"»" rquot @punct #14->13<br />
<br />
"<,>"<br />
<br />
"," cm @punct #15->17<br />
<br />
"<–>"<br />
<br />
"-" guio @punct #16->17<br />
<br />
"<dedi>"<br />
<br />
"de" v tv past p3 sg @root #17->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #18->17<br />
<br />
</code><br />
<br />
We also have examples where @acl (clausal modifiers of nouns) are embedded within @parataxis. Which brings us to our next point...<br />
<br />
=== @acl ===<br />
<br />
We used @acl<br />
<br />
<code>"<Yer>"<br />
<br />
"yer" n nom @root #1->0<br />
<br />
"<turiga>"<br />
<br />
"tur" n px3sp dat @orphan #2->1<br />
<br />
"<ko‘ra>"<br />
<br />
"ko'ra" post @case #3->4<br />
<br />
"<o‘simliklarni>"<br />
<br />
"o'simlik" n pl acc @nmod #4->2<br />
<br />
"<:>"<br />
<br />
":" sent @punct #5->1<br />
<br />
"<urug‘>"<br />
<br />
"urug'" n nom @obj #6->7<br />
<br />
"<beradigan>"<br />
<br />
"ber" v tv past p3 sg @acl #7->8<br />
<br />
"<o‘t-o‘lanlar>"<br />
<br />
"o't" n nom @appos #8->2<br />
<br />
"<hamda>"<br />
<br />
"hamda" cnjcoo @cc #9->14<br />
<br />
"<mevasida>"<br />
<br />
"meva" n px3sp loc @amod #10->11<br />
<br />
"<urug‘i>"<br />
<br />
"urug'i" n loc @obl #11->14<br />
<br />
"<bor>"<br />
<br />
"bor" adj @amod #12->14<br />
<br />
"<mevali>"<br />
<br />
"mevali" adj @amod #13->14<br />
<br />
"<daraxtlarni>"<br />
<br />
"daraxt" n pl acc @conj #14->8<br />
<br />
"<undirdi>"<br />
<br />
"*undirdi" @amod #15->14<br />
<br />
"<.>"<br />
<br />
"." sent @punct #16->1</code><br />
<br />
=== @cc ===<br />
<br />
@cc links a coordinating conjunction to the conjunct that comes after it. In Uzbek, one common coordinating conjunction is ''va'', which appears in the sentence below.<br />
<br />
<code><br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->8<br />
<br />
"<:>"<br />
<br />
":" sent @punct #2->8<br />
<br />
"<«>"<br />
<br />
"«" lquot @punct #3->5<br />
<br />
"<Yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #4->5<br />
<br />
"<bo‘lsin>"<br />
<br />
"bo'l" v tv perm @parataxis #5->8<br />
<br />
"<»>"<br />
<br />
"»" rquot @punct #6->5<br />
<br />
"<,>"<br />
<br />
"," cm @punct #7->8<br />
<br />
"<dedi>"<br />
<br />
"de" v tv past p3 sg @root #8->0<br />
<br />
"<va>"<br />
<br />
"va" cnjcoo @cc #9->12<br />
<br />
"<yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #10->12<br />
<br />
"<paydo>"<br />
<br />
"paydo" adj @nmod #11->10<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @conj #12->8<br />
<br />
"<.>"<br />
<br />
"." sent @punct #13->8<br />
<br />
</code><br />
<br />
''Xudo: "Yorug'lik bo'lsin", dedi va yorug'lik paydo bo'ldi.''<br />
<br />
God: "Light be", said and light appear did.<br />
<br />
"God said, 'Let there be light,' and there was light."<br />
<br />
This ''va'' isn't in the CTILD dictionary, but based on this sentence, it seems to mean something like "and." Another common @cc is ''esa'' "whereas":<br />
<br />
<code><br />
"<Xudoning>"<br />
<br />
"xudo" n gen @nmod #1->2<br />
<br />
"<Ruhi>"<br />
<br />
"ruh" n px3sp nom @root #2->0<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @cc #3->2<br />
<br />
"<suv>"<br />
<br />
"suv" n nom @nmod #4->6<br />
<br />
"<yuzida>"<br />
<br />
"yuz" num subst px3sp loc @case #5->4<br />
<br />
"<parvoz>"<br />
<br />
"parvoz" n nom @nsubj #6->2<br />
<br />
"<etar>"<br />
<br />
"etar" v tv past p3 sg @aux #7->8<br />
<br />
"<edi>"<br />
<br />
"e" cop past p3 sg @cop #8->2<br />
<br />
"<.>"<br />
<br />
"." sent @punct #9->2<br />
</code><br />
<br />
''Xudoning Ruhi esa suv yuzida parvoz etar edi.''<br />
<br />
God's Spirit whereas water face flight did was.<br />
<br />
"And the spirit of God was hovering over the surface of the waters."<br />
<br />
[[Category: sp18_UD]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Uzbek/Universal_Dependencies&diff=7689Uzbek/Universal Dependencies2018-05-03T15:29:18Z<p>Eschalk1: /* @acl */</p>
<hr />
<div>== Evaluation ==<br />
<br />
{| class="wikitable"<br />
! <br />
! uzb.annotated.ud.conllu<br />
! uzb.annotated2.ud.conllu<br />
|-<br />
! uzb.withmorph.udpipe<br />
| UAS: 82.07%, LAS: 79.68% || UAS: 42.98%, LAS: 35.09%<br />
|-<br />
! uzb.nomorph.udpipe<br />
| UAS: 72.11%, LAS: 69.72%<br />
| UAS: 43.86%, LAS: 35.09%<br />
|}<br />
<br />
=== uzb.annotated.ud.connlu ===<br />
<br />
forms: 251<br />
<br />
sentences: 30<br />
<br />
=== uzb.annotated2.ud.connlu ===<br />
<br />
forms: 114<br />
<br />
sentences: 9<br />
<br />
== Dependency relations ==<br />
<br />
=== @nsubj ===<br />
<br />
@nsubj (nominal subject) denotes a semantically agent-like and syntactically subject-like nominal. Note that it cannot be used for pleonastic subjects.<br />
<br />
In Uzbek, @nsubj will usually be assigned to a noun in the nominative case. Generally speaking, a noun in the nominative case is the subject of the sentence and the doer of the action. An example of @nsubj can be seen in the following sentence from our corpus.<br />
<br />
<code><br />
"<Azalda>"<br />
<br />
"azal" n loc @obl #1->6<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #2->6<br />
<br />
"<osmon>"<br />
<br />
"osmon" n nom @obj #3->6<br />
<br />
"<bilan>"<br />
<br />
"bilan" cnjcoo @cc #4->5<br />
<br />
"<yerni>"<br />
<br />
"yer" n acc @conj #5->3<br />
<br />
"<yaratdi>"<br />
<br />
"yarat" v tv past p3 sg @root #6->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #7->6<br />
<br />
</code><br />
<br />
''Azalda Xudo osmon bilan yerni yaratdi.''<br />
<br />
In.the.beginning God heaven and earth created.<br />
<br />
"In the beginning, God created the heavens and the earth."<br />
<br />
Note that not all nouns in the nominative case are @nsubj. For some reason, ''osmon'' "heaven" in this sentence is in the nominative case even though it is an object and a patient. Perhaps the accusative ending on ''yerni'' "earth" is able to cover both conjuncts.<br />
<br />
Sometimes @nsubj can be assigned to things that are not prototypically nouns. This occurs in the following sentence.<br />
<br />
<code><br />
"<Kech>"<br />
<br />
"kech" adv subst nom @nsubj #1->2<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @root #2->0<br />
<br />
"<,>"<br />
<br />
"," cm @cc #3->5<br />
<br />
"<erta>"<br />
<br />
"erta" adv subst nom @nsubj #4->5<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @conj #5->2<br />
<br />
"<–>"<br />
<br />
"-" guio @cc #6->7<br />
<br />
"<bu>"<br />
<br />
"bu" prn dem nom @conj #7->2<br />
<br />
"<birinchi>"<br />
<br />
"bir" num ord @nmod #8->9<br />
<br />
"<kun>"<br />
<br />
"kun" n nom @nsubj #9->7<br />
<br />
"<edi>"<br />
<br />
"e" cop past p3 sg @cop #10->7<br />
<br />
"<.>"<br />
<br />
"." sent @punct #11->2<br />
<br />
</code><br />
<br />
''Kech bo'ldi, erta bo'ldi - bu birinchi kun edi.''<br />
<br />
Evening was, morning was - this first day was.<br />
<br />
"And the evening and the morning were the first day."<br />
<br />
According to both the transducer and the CTILD dictionary, ''kech'' "evening" and ''erta'' "morning" are actually adverbs, but in this sentence, they fill the role of nominal subject.<br />
<br />
=== @obj ===<br />
<br />
@obj is usually applied to a patient-like constituent. In Uzbek, the constituent designated @obj will often be a noun in the accusative case. This is seen in the sentence below from our corpus.<br />
<br />
<code><br />
"<Ushbu>"<br />
<br />
"ushbu" prn dem nom @det #1->2<br />
<br />
"<gumbazni>"<br />
<br />
"gumbaz" n acc @obj #2->7<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @advmod #3->7<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #4->7<br />
<br />
"<osmon>"<br />
<br />
"osmon" n nom @xcomp #5->7<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #6->7<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @root #7->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #8->7<br />
</code><br />
<br />
''Ushbu gumbazni esa Xudo osmon deb atadi.''<br />
<br />
This dome whereas God heaven that called.<br />
<br />
"God called the expanse heaven."<br />
<br />
However, sometimes the @obj can be in other cases. In the sentence below, the obj ''suvlarning'' "waters" is in the genitive case.<br />
<br />
<code><br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->5<br />
<br />
"<quruqlikni>"<br />
<br />
"quruqlik" n acc @obj #2->5<br />
<br />
"<yer>"<br />
<br />
"yer" n nom @xcomp #3->5<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #4->5<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @root #5->0<br />
<br />
"<,>"<br />
<br />
"," cm @cc #6->12<br />
<br />
"<suvlarning>"<br />
<br />
"suv" n pl gen @obj #7->12<br />
<br />
"<jamini>"<br />
<br />
"jami" adj subst acc @nmod #8->7<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @advmod #9->12<br />
<br />
"<dengizlar>"<br />
<br />
"dengiz" n pl nom @xcomp #10->12<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #11->12<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @conj #12->5<br />
<br />
"<.>"<br />
<br />
"." sent @punct #13->5<br />
</code><br />
<br />
''Xudo quruqlikni yer deb atadi, suvlarning jamini esa dengizlar deb atadi.''<br />
<br />
God land earth that called, waters gathering whereas sea that called.<br />
<br />
"God called the dry land earth, and the gathering of the waters He called seas."<br />
<br />
=== @parataxis ===<br />
<br />
@parataxis is the relationship between the main predicate of a sentence and other elements "placed side by side without any explicit coordination, subordination, or argument relation with the head word." Parataxis is used in a variety of situations where the relationship between main sentence and clause is not explicit, especially in direct or reported speech (and so can often be identified w/ punctuation like ";" or ":"). @parataxis can be contrasted against @ccomp, which is used when the reported speech is clearly in a subordinate clause of the main verb. We used @parataxis to identify reported speech in our corpus.<br />
<br />
<code><br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->8<br />
<br />
"<:>"<br />
<br />
":" sent @punct #2->8<br />
<br />
"<«>"<br />
<br />
"«" lquot @punct #3->5<br />
<br />
"<Yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #4->5<br />
<br />
"<bo‘lsin>"<br />
<br />
"bo'l" v tv perm @parataxis #5->8<br />
<br />
"<»>"<br />
<br />
"»" rquot @punct #6->5<br />
<br />
"<,>"<br />
<br />
"," cm @punct #7->8<br />
<br />
"<dedi>"<br />
<br />
"de" v tv past p3 sg @root #8->0<br />
<br />
"<va>"<br />
<br />
"va" cnjcoo @cc #9->12<br />
<br />
"<yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #10->12<br />
<br />
"<paydo>"<br />
<br />
"paydo" adj @nmod #11->10<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @conj #12->8<br />
<br />
"<.>"<br />
<br />
"." sent @punct #13->8<br />
<br />
</code><br />
<br />
@Parataxis is especially helpful in these examples because the syntax of reported speech in Uzbek is even less coordinated than in English: the sentence begins with the speaker, launches into the reported speech, and we only get the speaking verb at the end of the sentence. <br />
<br />
<code><br />
<br />
"<Yana>"<br />
<br />
"yana" adv @advmod #1->17<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #2->17<br />
<br />
"<:>"<br />
<br />
":" sent @punct #3->17<br />
<br />
"<«>"<br />
<br />
"«" lquot @punct #4->13<br />
<br />
"<Suvliklar>"<br />
<br />
"suv" n loc @nmod #5->11<br />
<br />
"<o‘rtasida>"<br />
<br />
"o'rta" n px3sp loc @case #6->5<br />
<br />
"<ularni>"<br />
<br />
"ular" prn pers p3 pl acc @obj #7->9<br />
<br />
"<bir-biridan>"<br />
<br />
"bir-bir" prn ind px3sp abl @nmod #8->7<br />
<br />
"<ajratib>"<br />
<br />
"ajra" v iv past p3 sg @acl #9->11<br />
<br />
"<turadigan>"<br />
<br />
"tur" v iv past p3 sg @conj #10->9<br />
<br />
"<gumbaz>"<br />
<br />
"gumbaz" n nom @nsubj #11->13<br />
<br />
"<paydo>"<br />
<br />
"paydo" adj @nmod #12->11<br />
<br />
"<bo‘lsin>"<br />
<br />
"bo'l" v tv perm @parataxis #13->17<br />
<br />
<br />
"<»>"<br />
<br />
"»" rquot @punct #14->13<br />
<br />
"<,>"<br />
<br />
"," cm @punct #15->17<br />
<br />
"<–>"<br />
<br />
"-" guio @punct #16->17<br />
<br />
"<dedi>"<br />
<br />
"de" v tv past p3 sg @root #17->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #18->17<br />
<br />
</code><br />
<br />
We also have examples where @acl (clausal modifiers of nouns) are embedded within @parataxis. Which brings us to our next point...<br />
<br />
=== @acl ===<br />
<br />
We used @acl<br />
<br />
<code><br />
<br />
"<Yer>"<br />
"yer" n nom @root #1->0<br />
"<turiga>"<br />
"tur" n px3sp dat @orphan #2->1<br />
"<ko‘ra>"<br />
"ko'ra" post @case #3->4<br />
"<o‘simliklarni>"<br />
"o'simlik" n pl acc @nmod #4->2<br />
"<:>"<br />
":" sent @punct #5->1<br />
"<urug‘>"<br />
"urug'" n nom @obj #6->7<br />
"<beradigan>"<br />
"ber" v tv past p3 sg @acl #7->8<br />
"<o‘t-o‘lanlar>"<br />
"o't" n nom @appos #8->2<br />
"<hamda>"<br />
"hamda" cnjcoo @cc #9->14<br />
"<mevasida>"<br />
"meva" n px3sp loc @amod #10->11<br />
"<urug‘i>"<br />
"urug'i" n loc @obl #11->14<br />
"<bor>"<br />
"bor" adj @amod #12->14<br />
"<mevali>"<br />
"mevali" adj @amod #13->14<br />
"<daraxtlarni>"<br />
"daraxt" n pl acc @conj #14->8<br />
"<undirdi>"<br />
"*undirdi" @amod #15->14<br />
"<.>"<br />
"." sent @punct #16->1<br />
<br />
</code><br />
<br />
=== @cc ===<br />
<br />
@cc links a coordinating conjunction to the conjunct that comes after it. In Uzbek, one common coordinating conjunction is ''va'', which appears in the sentence below.<br />
<br />
<code><br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->8<br />
<br />
"<:>"<br />
<br />
":" sent @punct #2->8<br />
<br />
"<«>"<br />
<br />
"«" lquot @punct #3->5<br />
<br />
"<Yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #4->5<br />
<br />
"<bo‘lsin>"<br />
<br />
"bo'l" v tv perm @parataxis #5->8<br />
<br />
"<»>"<br />
<br />
"»" rquot @punct #6->5<br />
<br />
"<,>"<br />
<br />
"," cm @punct #7->8<br />
<br />
"<dedi>"<br />
<br />
"de" v tv past p3 sg @root #8->0<br />
<br />
"<va>"<br />
<br />
"va" cnjcoo @cc #9->12<br />
<br />
"<yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #10->12<br />
<br />
"<paydo>"<br />
<br />
"paydo" adj @nmod #11->10<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @conj #12->8<br />
<br />
"<.>"<br />
<br />
"." sent @punct #13->8<br />
<br />
</code><br />
<br />
''Xudo: "Yorug'lik bo'lsin", dedi va yorug'lik paydo bo'ldi.''<br />
<br />
God: "Light be", said and light appear did.<br />
<br />
"God said, 'Let there be light,' and there was light."<br />
<br />
This ''va'' isn't in the CTILD dictionary, but based on this sentence, it seems to mean something like "and." Another common @cc is ''esa'' "whereas":<br />
<br />
<code><br />
"<Xudoning>"<br />
<br />
"xudo" n gen @nmod #1->2<br />
<br />
"<Ruhi>"<br />
<br />
"ruh" n px3sp nom @root #2->0<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @cc #3->2<br />
<br />
"<suv>"<br />
<br />
"suv" n nom @nmod #4->6<br />
<br />
"<yuzida>"<br />
<br />
"yuz" num subst px3sp loc @case #5->4<br />
<br />
"<parvoz>"<br />
<br />
"parvoz" n nom @nsubj #6->2<br />
<br />
"<etar>"<br />
<br />
"etar" v tv past p3 sg @aux #7->8<br />
<br />
"<edi>"<br />
<br />
"e" cop past p3 sg @cop #8->2<br />
<br />
"<.>"<br />
<br />
"." sent @punct #9->2<br />
</code><br />
<br />
''Xudoning Ruhi esa suv yuzida parvoz etar edi.''<br />
<br />
God's Spirit whereas water face flight did was.<br />
<br />
"And the spirit of God was hovering over the surface of the waters."<br />
<br />
[[Category: sp18_UD]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Uzbek/Universal_Dependencies&diff=7668Uzbek/Universal Dependencies2018-05-01T16:43:16Z<p>Eschalk1: /* @parataxis */</p>
<hr />
<div>== Evaluation ==<br />
<br />
{| class="wikitable"<br />
! <br />
! uzb.annotated.ud.conllu<br />
! uzb.annotated2.ud.conllu<br />
|-<br />
! uzb.withmorph.udpipe<br />
| UAS: 82.07%, LAS: 79.68% || UAS: 42.98%, LAS: 35.09%<br />
|-<br />
! uzb.nomorph.udpipe<br />
| UAS: 72.11%, LAS: 69.72%<br />
| UAS: 43.86%, LAS: 35.09%<br />
|}<br />
<br />
=== uzb.annotated.ud.connlu ===<br />
<br />
forms: 251<br />
<br />
sentences: 30<br />
<br />
=== uzb.annotated2.ud.connlu ===<br />
<br />
forms: 114<br />
<br />
sentences: 9<br />
<br />
== Dependency relations ==<br />
<br />
=== @nsubj ===<br />
<br />
@nsubj (nominal subject) denotes a semantically agent-like and syntactically subject-like nominal. Note that it cannot be used for pleonastic subjects.<br />
<br />
In Uzbek, @nsubj will usually be assigned to a noun in the nominative case. Generally speaking, a noun in the nominative case is the subject of the sentence and the doer of the action. An example of @nsubj can be seen in the following sentence from our corpus.<br />
<br />
<code><br />
"<Azalda>"<br />
<br />
"azal" n loc @obl #1->6<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #2->6<br />
<br />
"<osmon>"<br />
<br />
"osmon" n nom @obj #3->6<br />
<br />
"<bilan>"<br />
<br />
"bilan" cnjcoo @cc #4->5<br />
<br />
"<yerni>"<br />
<br />
"yer" n acc @conj #5->3<br />
<br />
"<yaratdi>"<br />
<br />
"yarat" v tv past p3 sg @root #6->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #7->6<br />
<br />
</code><br />
<br />
''Azalda Xudo osmon bilan yerni yaratdi.''<br />
<br />
In.the.beginning God heaven and earth created.<br />
<br />
"In the beginning, God created the heavens and the earth."<br />
<br />
Note that not all nouns in the nominative case are @nsubj. For some reason, ''osmon'' "heaven" in this sentence is in the nominative case even though it is an object and a patient. Perhaps the accusative ending on ''yerni'' "earth" is able to cover both conjuncts.<br />
<br />
Sometimes @nsubj can be assigned to things that are not prototypically nouns. This occurs in the following sentence.<br />
<br />
<code><br />
"<Kech>"<br />
<br />
"kech" adv subst nom @nsubj #1->2<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @root #2->0<br />
<br />
"<,>"<br />
<br />
"," cm @cc #3->5<br />
<br />
"<erta>"<br />
<br />
"erta" adv subst nom @nsubj #4->5<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @conj #5->2<br />
<br />
"<–>"<br />
<br />
"-" guio @cc #6->7<br />
<br />
"<bu>"<br />
<br />
"bu" prn dem nom @conj #7->2<br />
<br />
"<birinchi>"<br />
<br />
"bir" num ord @nmod #8->9<br />
<br />
"<kun>"<br />
<br />
"kun" n nom @nsubj #9->7<br />
<br />
"<edi>"<br />
<br />
"e" cop past p3 sg @cop #10->7<br />
<br />
"<.>"<br />
<br />
"." sent @punct #11->2<br />
<br />
</code><br />
<br />
''Kech bo'ldi, erta bo'ldi - bu birinchi kun edi.''<br />
<br />
Evening was, morning was - this first day was.<br />
<br />
"And the evening and the morning were the first day."<br />
<br />
According to both the transducer and the CTILD dictionary, ''kech'' "evening" and ''erta'' "morning" are actually adverbs, but in this sentence, they fill the role of nominal subject.<br />
<br />
=== @obj ===<br />
<br />
@obj is usually applied to a patient-like constituent. In Uzbek, the constituent designated @obj will often be a noun in the accusative case. This is seen in the sentence below from our corpus.<br />
<br />
<code><br />
"<Ushbu>"<br />
<br />
"ushbu" prn dem nom @det #1->2<br />
<br />
"<gumbazni>"<br />
<br />
"gumbaz" n acc @obj #2->7<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @advmod #3->7<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #4->7<br />
<br />
"<osmon>"<br />
<br />
"osmon" n nom @xcomp #5->7<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #6->7<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @root #7->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #8->7<br />
</code><br />
<br />
''Ushbu gumbazni esa Xudo osmon deb atadi.''<br />
<br />
This dome whereas God heaven that called.<br />
<br />
"God called the expanse heaven."<br />
<br />
However, sometimes the @obj can be in other cases. In the sentence below, the obj ''suvlarning'' "waters" is in the genitive case.<br />
<br />
<code><br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->5<br />
<br />
"<quruqlikni>"<br />
<br />
"quruqlik" n acc @obj #2->5<br />
<br />
"<yer>"<br />
<br />
"yer" n nom @xcomp #3->5<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #4->5<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @root #5->0<br />
<br />
"<,>"<br />
<br />
"," cm @cc #6->12<br />
<br />
"<suvlarning>"<br />
<br />
"suv" n pl gen @obj #7->12<br />
<br />
"<jamini>"<br />
<br />
"jami" adj subst acc @nmod #8->7<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @advmod #9->12<br />
<br />
"<dengizlar>"<br />
<br />
"dengiz" n pl nom @xcomp #10->12<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #11->12<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @conj #12->5<br />
<br />
"<.>"<br />
<br />
"." sent @punct #13->5<br />
</code><br />
<br />
''Xudo quruqlikni yer deb atadi, suvlarning jamini esa dengizlar deb atadi.''<br />
<br />
God land earth that called, waters gathering whereas sea that called.<br />
<br />
"God called the dry land earth, and the gathering of the waters He called seas."<br />
<br />
=== @parataxis ===<br />
<br />
@parataxis is the relationship between the main predicate of a sentence and other elements "placed side by side without any explicit coordination, subordination, or argument relation with the head word." Parataxis is used in a variety of situations where the relationship between main sentence and clause is not explicit, especially in direct or reported speech (and so can often be identified w/ punctuation like ";" or ":"). @parataxis can be contrasted against @ccomp, which is used when the reported speech is clearly in a subordinate clause of the main verb. We used @parataxis to identify reported speech in our corpus.<br />
<br />
<code><br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->8<br />
<br />
"<:>"<br />
<br />
":" sent @punct #2->8<br />
<br />
"<«>"<br />
<br />
"«" lquot @punct #3->5<br />
<br />
"<Yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #4->5<br />
<br />
"<bo‘lsin>"<br />
<br />
"bo'l" v tv perm @parataxis #5->8<br />
<br />
"<»>"<br />
<br />
"»" rquot @punct #6->5<br />
<br />
"<,>"<br />
<br />
"," cm @punct #7->8<br />
"<dedi>"<br />
"de" v tv past p3 sg @root #8->0<br />
"<va>"<br />
"va" cnjcoo @cc #9->12<br />
"<yorug‘lik>"<br />
"yorug'lik" n nom @nsubj #10->12<br />
"<paydo>"<br />
"paydo" adj @nmod #11->10<br />
"<bo‘ldi>"<br />
"bo'l" v tv past p3 sg @conj #12->8<br />
"<.>"<br />
"." sent @punct #13->8<br />
</code><br />
<br />
@Parataxis is especially helpful in these examples because the syntax of reported speech in Uzbek is even less coordinated than in English: the sentence begins with the speaker, launches into the reported speech, and we only get the speaking verb at the end of the sentence. <br />
<br />
<code><br />
"<Yana>"<br />
"yana" adv @advmod #1->17<br />
"<Xudo>"<br />
"xudo" n nom @nsubj #2->17<br />
"<:>"<br />
":" sent @punct #3->17<br />
"<«>"<br />
"«" lquot @punct #4->13<br />
"<Suvliklar>"<br />
"suv" n loc @nmod #5->11<br />
"<o‘rtasida>"<br />
"o'rta" n px3sp loc @case #6->5<br />
"<ularni>"<br />
"ular" prn pers p3 pl acc @obj #7->9<br />
"<bir-biridan>"<br />
"bir-bir" prn ind px3sp abl @nmod #8->7<br />
"<ajratib>"<br />
"ajra" v iv past p3 sg @acl #9->11<br />
"<turadigan>"<br />
"tur" v iv past p3 sg @conj #10->9<br />
"<gumbaz>"<br />
"gumbaz" n nom @nsubj #11->13<br />
"<paydo>"<br />
"paydo" adj @nmod #12->11<br />
"<bo‘lsin>"<br />
"bo'l" v tv perm @parataxis #13->17<br />
"<»>"<br />
"»" rquot @punct #14->13<br />
"<,>"<br />
"," cm @punct #15->17<br />
"<–>"<br />
"-" guio @punct #16->17<br />
"<dedi>"<br />
"de" v tv past p3 sg @root #17->0<br />
"<.>"<br />
"." sent @punct #18->17<br />
</code><br />
<br />
We also have examples where @acl (clausal modifiers of nouns) are embedded within @parataxis. Which brings us to our next point...<br />
<br />
=== @acl ===<br />
<br />
We used @acl <br />
<br />
[[Category: sp18_UD]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Uzbek/Universal_Dependencies&diff=7667Uzbek/Universal Dependencies2018-05-01T16:42:35Z<p>Eschalk1: /* @parataxis */</p>
<hr />
<div>== Evaluation ==<br />
<br />
{| class="wikitable"<br />
! <br />
! uzb.annotated.ud.conllu<br />
! uzb.annotated2.ud.conllu<br />
|-<br />
! uzb.withmorph.udpipe<br />
| UAS: 82.07%, LAS: 79.68% || UAS: 42.98%, LAS: 35.09%<br />
|-<br />
! uzb.nomorph.udpipe<br />
| UAS: 72.11%, LAS: 69.72%<br />
| UAS: 43.86%, LAS: 35.09%<br />
|}<br />
<br />
=== uzb.annotated.ud.connlu ===<br />
<br />
forms: 251<br />
<br />
sentences: 30<br />
<br />
=== uzb.annotated2.ud.connlu ===<br />
<br />
forms: 114<br />
<br />
sentences: 9<br />
<br />
== Dependency relations ==<br />
<br />
=== @nsubj ===<br />
<br />
@nsubj (nominal subject) denotes a semantically agent-like and syntactically subject-like nominal. Note that it cannot be used for pleonastic subjects.<br />
<br />
In Uzbek, @nsubj will usually be assigned to a noun in the nominative case. Generally speaking, a noun in the nominative case is the subject of the sentence and the doer of the action. An example of @nsubj can be seen in the following sentence from our corpus.<br />
<br />
<code><br />
"<Azalda>"<br />
<br />
"azal" n loc @obl #1->6<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #2->6<br />
<br />
"<osmon>"<br />
<br />
"osmon" n nom @obj #3->6<br />
<br />
"<bilan>"<br />
<br />
"bilan" cnjcoo @cc #4->5<br />
<br />
"<yerni>"<br />
<br />
"yer" n acc @conj #5->3<br />
<br />
"<yaratdi>"<br />
<br />
"yarat" v tv past p3 sg @root #6->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #7->6<br />
<br />
</code><br />
<br />
''Azalda Xudo osmon bilan yerni yaratdi.''<br />
<br />
In.the.beginning God heaven and earth created.<br />
<br />
"In the beginning, God created the heavens and the earth."<br />
<br />
Note that not all nouns in the nominative case are @nsubj. For some reason, ''osmon'' "heaven" in this sentence is in the nominative case even though it is an object and a patient. Perhaps the accusative ending on ''yerni'' "earth" is able to cover both conjuncts.<br />
<br />
Sometimes @nsubj can be assigned to things that are not prototypically nouns. This occurs in the following sentence.<br />
<br />
<code><br />
"<Kech>"<br />
<br />
"kech" adv subst nom @nsubj #1->2<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @root #2->0<br />
<br />
"<,>"<br />
<br />
"," cm @cc #3->5<br />
<br />
"<erta>"<br />
<br />
"erta" adv subst nom @nsubj #4->5<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @conj #5->2<br />
<br />
"<–>"<br />
<br />
"-" guio @cc #6->7<br />
<br />
"<bu>"<br />
<br />
"bu" prn dem nom @conj #7->2<br />
<br />
"<birinchi>"<br />
<br />
"bir" num ord @nmod #8->9<br />
<br />
"<kun>"<br />
<br />
"kun" n nom @nsubj #9->7<br />
<br />
"<edi>"<br />
<br />
"e" cop past p3 sg @cop #10->7<br />
<br />
"<.>"<br />
<br />
"." sent @punct #11->2<br />
<br />
</code><br />
<br />
''Kech bo'ldi, erta bo'ldi - bu birinchi kun edi.''<br />
<br />
Evening was, morning was - this first day was.<br />
<br />
"And the evening and the morning were the first day."<br />
<br />
According to both the transducer and the CTILD dictionary, ''kech'' "evening" and ''erta'' "morning" are actually adverbs, but in this sentence, they fill the role of nominal subject.<br />
<br />
=== @obj ===<br />
<br />
@obj is usually applied to a patient-like constituent. In Uzbek, the constituent designated @obj will often be a noun in the accusative case. This is seen in the sentence below from our corpus.<br />
<br />
<code><br />
"<Ushbu>"<br />
<br />
"ushbu" prn dem nom @det #1->2<br />
<br />
"<gumbazni>"<br />
<br />
"gumbaz" n acc @obj #2->7<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @advmod #3->7<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #4->7<br />
<br />
"<osmon>"<br />
<br />
"osmon" n nom @xcomp #5->7<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #6->7<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @root #7->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #8->7<br />
</code><br />
<br />
''Ushbu gumbazni esa Xudo osmon deb atadi.''<br />
<br />
This dome whereas God heaven that called.<br />
<br />
"God called the expanse heaven."<br />
<br />
However, sometimes the @obj can be in other cases. In the sentence below, the obj ''suvlarning'' "waters" is in the genitive case.<br />
<br />
<code><br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->5<br />
<br />
"<quruqlikni>"<br />
<br />
"quruqlik" n acc @obj #2->5<br />
<br />
"<yer>"<br />
<br />
"yer" n nom @xcomp #3->5<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #4->5<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @root #5->0<br />
<br />
"<,>"<br />
<br />
"," cm @cc #6->12<br />
<br />
"<suvlarning>"<br />
<br />
"suv" n pl gen @obj #7->12<br />
<br />
"<jamini>"<br />
<br />
"jami" adj subst acc @nmod #8->7<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @advmod #9->12<br />
<br />
"<dengizlar>"<br />
<br />
"dengiz" n pl nom @xcomp #10->12<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #11->12<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @conj #12->5<br />
<br />
"<.>"<br />
<br />
"." sent @punct #13->5<br />
</code><br />
<br />
''Xudo quruqlikni yer deb atadi, suvlarning jamini esa dengizlar deb atadi.''<br />
<br />
God land earth that called, waters gathering whereas sea that called.<br />
<br />
"God called the dry land earth, and the gathering of the waters He called seas."<br />
<br />
=== @parataxis ===<br />
<br />
@parataxis is the relationship between the main predicate of a sentence and other elements "placed side by side without any explicit coordination, subordination, or argument relation with the head word." Parataxis is used in a variety of situations where the relationship between main sentence and clause is not explicit, especially in direct or reported speech (and so can often be identified w/ punctuation like ";" or ":"). @parataxis can be contrasted against @ccomp, which is used when the reported speech is clearly in a subordinate clause of the main verb. We used @parataxis to identify reported speech in our corpus.<br />
<br />
<code><br />
"<Xudo>"<br />
"xudo" n nom @nsubj #1->8<br />
"<:>"<br />
":" sent @punct #2->8<br />
"<«>"<br />
"«" lquot @punct #3->5<br />
"<Yorug‘lik>"<br />
"yorug'lik" n nom @nsubj #4->5<br />
"<bo‘lsin>"<br />
"bo'l" v tv perm @parataxis #5->8<br />
"<»>"<br />
"»" rquot @punct #6->5<br />
"<,>"<br />
"," cm @punct #7->8<br />
"<dedi>"<br />
"de" v tv past p3 sg @root #8->0<br />
"<va>"<br />
"va" cnjcoo @cc #9->12<br />
"<yorug‘lik>"<br />
"yorug'lik" n nom @nsubj #10->12<br />
"<paydo>"<br />
"paydo" adj @nmod #11->10<br />
"<bo‘ldi>"<br />
"bo'l" v tv past p3 sg @conj #12->8<br />
"<.>"<br />
"." sent @punct #13->8<br />
</code><br />
<br />
@Parataxis is especially helpful in these examples because the syntax of reported speech in Uzbek is even less coordinated than in English: the sentence begins with the speaker, launches into the reported speech, and we only get the speaking verb at the end of the sentence. <br />
<br />
<code><br />
"<Yana>"<br />
"yana" adv @advmod #1->17<br />
"<Xudo>"<br />
"xudo" n nom @nsubj #2->17<br />
"<:>"<br />
":" sent @punct #3->17<br />
"<«>"<br />
"«" lquot @punct #4->13<br />
"<Suvliklar>"<br />
"suv" n loc @nmod #5->11<br />
"<o‘rtasida>"<br />
"o'rta" n px3sp loc @case #6->5<br />
"<ularni>"<br />
"ular" prn pers p3 pl acc @obj #7->9<br />
"<bir-biridan>"<br />
"bir-bir" prn ind px3sp abl @nmod #8->7<br />
"<ajratib>"<br />
"ajra" v iv past p3 sg @acl #9->11<br />
"<turadigan>"<br />
"tur" v iv past p3 sg @conj #10->9<br />
"<gumbaz>"<br />
"gumbaz" n nom @nsubj #11->13<br />
"<paydo>"<br />
"paydo" adj @nmod #12->11<br />
"<bo‘lsin>"<br />
"bo'l" v tv perm @parataxis #13->17<br />
"<»>"<br />
"»" rquot @punct #14->13<br />
"<,>"<br />
"," cm @punct #15->17<br />
"<–>"<br />
"-" guio @punct #16->17<br />
"<dedi>"<br />
"de" v tv past p3 sg @root #17->0<br />
"<.>"<br />
"." sent @punct #18->17<br />
</code><br />
<br />
We also have examples where @acl (clausal modifiers of nouns) are embedded within @parataxis. Which brings us to our next point...<br />
<br />
=== @acl ===<br />
<br />
We used @acl <br />
<br />
[[Category: sp18_UD]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Uzbek/Universal_Dependencies&diff=7666Uzbek/Universal Dependencies2018-05-01T16:41:02Z<p>Eschalk1: /* @parataxis */</p>
<hr />
<div>== Evaluation ==<br />
<br />
{| class="wikitable"<br />
! <br />
! uzb.annotated.ud.conllu<br />
! uzb.annotated2.ud.conllu<br />
|-<br />
! uzb.withmorph.udpipe<br />
| UAS: 82.07%, LAS: 79.68% || UAS: 42.98%, LAS: 35.09%<br />
|-<br />
! uzb.nomorph.udpipe<br />
| UAS: 72.11%, LAS: 69.72%<br />
| UAS: 43.86%, LAS: 35.09%<br />
|}<br />
<br />
=== uzb.annotated.ud.connlu ===<br />
<br />
forms: 251<br />
<br />
sentences: 30<br />
<br />
=== uzb.annotated2.ud.connlu ===<br />
<br />
forms: 114<br />
<br />
sentences: 9<br />
<br />
== Dependency relations ==<br />
<br />
=== @nsubj ===<br />
<br />
@nsubj (nominal subject) denotes a semantically agent-like and syntactically subject-like nominal. Note that it cannot be used for pleonastic subjects.<br />
<br />
In Uzbek, @nsubj will usually be assigned to a noun in the nominative case. Generally speaking, a noun in the nominative case is the subject of the sentence and the doer of the action. An example of @nsubj can be seen in the following sentence from our corpus.<br />
<br />
<code><br />
"<Azalda>"<br />
<br />
"azal" n loc @obl #1->6<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #2->6<br />
<br />
"<osmon>"<br />
<br />
"osmon" n nom @obj #3->6<br />
<br />
"<bilan>"<br />
<br />
"bilan" cnjcoo @cc #4->5<br />
<br />
"<yerni>"<br />
<br />
"yer" n acc @conj #5->3<br />
<br />
"<yaratdi>"<br />
<br />
"yarat" v tv past p3 sg @root #6->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #7->6<br />
<br />
</code><br />
<br />
''Azalda Xudo osmon bilan yerni yaratdi.''<br />
<br />
In.the.beginning God heaven and earth created.<br />
<br />
"In the beginning, God created the heavens and the earth."<br />
<br />
Note that not all nouns in the nominative case are @nsubj. For some reason, ''osmon'' "heaven" in this sentence is in the nominative case even though it is an object and a patient. Perhaps the accusative ending on ''yerni'' "earth" is able to cover both conjuncts.<br />
<br />
Sometimes @nsubj can be assigned to things that are not prototypically nouns. This occurs in the following sentence.<br />
<br />
<code><br />
"<Kech>"<br />
<br />
"kech" adv subst nom @nsubj #1->2<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @root #2->0<br />
<br />
"<,>"<br />
<br />
"," cm @cc #3->5<br />
<br />
"<erta>"<br />
<br />
"erta" adv subst nom @nsubj #4->5<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @conj #5->2<br />
<br />
"<–>"<br />
<br />
"-" guio @cc #6->7<br />
<br />
"<bu>"<br />
<br />
"bu" prn dem nom @conj #7->2<br />
<br />
"<birinchi>"<br />
<br />
"bir" num ord @nmod #8->9<br />
<br />
"<kun>"<br />
<br />
"kun" n nom @nsubj #9->7<br />
<br />
"<edi>"<br />
<br />
"e" cop past p3 sg @cop #10->7<br />
<br />
"<.>"<br />
<br />
"." sent @punct #11->2<br />
<br />
</code><br />
<br />
''Kech bo'ldi, erta bo'ldi - bu birinchi kun edi.''<br />
<br />
Evening was, morning was - this first day was.<br />
<br />
"And the evening and the morning were the first day."<br />
<br />
According to both the transducer and the CTILD dictionary, ''kech'' "evening" and ''erta'' "morning" are actually adverbs, but in this sentence, they fill the role of nominal subject.<br />
<br />
=== @obj ===<br />
<br />
@obj is usually applied to a patient-like constituent. In Uzbek, the constituent designated @obj will often be a noun in the accusative case. This is seen in the sentence below from our corpus.<br />
<br />
<code><br />
"<Ushbu>"<br />
<br />
"ushbu" prn dem nom @det #1->2<br />
<br />
"<gumbazni>"<br />
<br />
"gumbaz" n acc @obj #2->7<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @advmod #3->7<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #4->7<br />
<br />
"<osmon>"<br />
<br />
"osmon" n nom @xcomp #5->7<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #6->7<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @root #7->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #8->7<br />
</code><br />
<br />
''Ushbu gumbazni esa Xudo osmon deb atadi.''<br />
<br />
This dome whereas God heaven that called.<br />
<br />
"God called the expanse heaven."<br />
<br />
However, sometimes the @obj can be in other cases. In the sentence below, the obj ''suvlarning'' "waters" is in the genitive case.<br />
<br />
<code><br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->5<br />
<br />
"<quruqlikni>"<br />
<br />
"quruqlik" n acc @obj #2->5<br />
<br />
"<yer>"<br />
<br />
"yer" n nom @xcomp #3->5<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #4->5<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @root #5->0<br />
<br />
"<,>"<br />
<br />
"," cm @cc #6->12<br />
<br />
"<suvlarning>"<br />
<br />
"suv" n pl gen @obj #7->12<br />
<br />
"<jamini>"<br />
<br />
"jami" adj subst acc @nmod #8->7<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @advmod #9->12<br />
<br />
"<dengizlar>"<br />
<br />
"dengiz" n pl nom @xcomp #10->12<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #11->12<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @conj #12->5<br />
<br />
"<.>"<br />
<br />
"." sent @punct #13->5<br />
</code><br />
<br />
''Xudo quruqlikni yer deb atadi, suvlarning jamini esa dengizlar deb atadi.''<br />
<br />
God land earth that called, waters gathering whereas sea that called.<br />
<br />
"God called the dry land earth, and the gathering of the waters He called seas."<br />
<br />
=== @parataxis ===<br />
<br />
@parataxis is the relationship between the main predicate of a sentence and other elements "placed side by side without any explicit coordination, subordination, or argument relation with the head word." Parataxis is used in a variety of situations where the relationship between main sentence and clause is not explicit, especially in direct or reported speech (and so can often be identified w/ punctuation like ";" or ":"). @parataxis can be contrasted against @ccomp, which is used when the reported speech is clearly in a subordinate clause of the main verb. We used @parataxis to identify reported speech in our corpus.<br />
<br />
<code><br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->8<br />
<br />
"<:>"<br />
<br />
":" sent @punct #2->8<br />
"<«>"<br />
<br />
"«" lquot @punct #3->5<br />
"<Yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #4->5<br />
"<bo‘lsin>"<br />
<br />
"bo'l" v tv perm @parataxis #5->8<br />
"<»>"<br />
<br />
"»" rquot @punct #6->5<br />
"<,>"<br />
<br />
"," cm @punct #7->8<br />
"<dedi>"<br />
<br />
"de" v tv past p3 sg @root #8->0<br />
"<va>"<br />
<br />
"va" cnjcoo @cc #9->12<br />
"<yorug‘lik>"<br />
<br />
"yorug'lik" n nom @nsubj #10->12<br />
"<paydo>"<br />
<br />
"paydo" adj @nmod #11->10<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @conj #12->8<br />
"<.>"<br />
"." sent @punct #13->8<br />
<br />
</code><br />
<br />
@Parataxis is especially helpful in these examples because the syntax of reported speech in Uzbek is even less coordinated than in English: the sentence begins with the speaker, launches into the reported speech, and we only get the speaking verb at the end of the sentence. <br />
<br />
<code><br />
<br />
"<Yana>"<br />
<br />
"yana" adv @advmod #1->17<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #2->17<br />
"<:>"<br />
<br />
":" sent @punct #3->17<br />
"<«>"<br />
<br />
"«" lquot @punct #4->13<br />
"<Suvliklar>"<br />
<br />
"suv" n loc @nmod #5->11<br />
"<o‘rtasida>"<br />
<br />
"o'rta" n px3sp loc @case #6->5<br />
"<ularni>"<br />
<br />
"ular" prn pers p3 pl acc @obj #7->9<br />
"<bir-biridan>"<br />
<br />
"bir-bir" prn ind px3sp abl @nmod #8->7<br />
"<ajratib>"<br />
<br />
"ajra" v iv past p3 sg @acl #9->11<br />
"<turadigan>"<br />
<br />
"tur" v iv past p3 sg @conj #10->9<br />
"<gumbaz>"<br />
<br />
"gumbaz" n nom @nsubj #11->13<br />
"<paydo>"<br />
<br />
"paydo" adj @nmod #12->11<br />
"<bo‘lsin>"<br />
<br />
"bo'l" v tv perm @parataxis #13->17<br />
"<»>"<br />
<br />
"»" rquot @punct #14->13<br />
"<,>"<br />
<br />
"," cm @punct #15->17<br />
"<–>"<br />
<br />
"-" guio @punct #16->17<br />
"<dedi>"<br />
<br />
"de" v tv past p3 sg @root #17->0<br />
"<.>"<br />
"." sent @punct #18->17<br />
<br />
</code><br />
<br />
We also have examples where @acl (clausal modifiers of nouns) are embedded within @parataxis. Which brings us to our next point...<br />
<br />
=== @acl ===<br />
<br />
We used @acl <br />
<br />
[[Category: sp18_UD]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Uzbek/Universal_Dependencies&diff=7663Uzbek/Universal Dependencies2018-05-01T15:27:43Z<p>Eschalk1: /* Dependency relations */</p>
<hr />
<div>== Evaluation ==<br />
<br />
{| class="wikitable"<br />
! <br />
! uzb.annotated.ud.conllu<br />
! uzb.annotated2.ud.conllu<br />
|-<br />
! uzb.withmorph.udpipe<br />
| UAS: 82.07%, LAS: 79.68% || UAS: 42.98%, LAS: 35.09%<br />
|-<br />
! uzb.nomorph.udpipe<br />
| UAS: 72.11%, LAS: 69.72%<br />
| UAS: 43.86%, LAS: 35.09%<br />
|}<br />
<br />
=== uzb.annotated.ud.connlu ===<br />
<br />
forms: 251<br />
<br />
sentences: 30<br />
<br />
=== uzb.annotated2.ud.connlu ===<br />
<br />
forms: 114<br />
<br />
sentences: 9<br />
<br />
== Dependency relations ==<br />
<br />
=== @nsubj ===<br />
<br />
@nsubj (nominal subject) denotes a semantically agent-like and syntactically subject-like nominal. Note that it cannot be used for pleonastic subjects.<br />
<br />
In Uzbek, @nsubj will usually be assigned to a noun in the nominative case. Generally speaking, a noun in the nominative case is the subject of the sentence and the doer of the action. An example of @nsubj can be seen in the following sentence from our corpus.<br />
<br />
<code><br />
"<Azalda>"<br />
<br />
"azal" n loc @obl #1->6<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #2->6<br />
<br />
"<osmon>"<br />
<br />
"osmon" n nom @obj #3->6<br />
<br />
"<bilan>"<br />
<br />
"bilan" cnjcoo @cc #4->5<br />
<br />
"<yerni>"<br />
<br />
"yer" n acc @conj #5->3<br />
<br />
"<yaratdi>"<br />
<br />
"yarat" v tv past p3 sg @root #6->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #7->6<br />
<br />
</code><br />
<br />
''Azalda Xudo osmon bilan yerni yaratdi.''<br />
<br />
In.the.beginning God heaven and earth created.<br />
<br />
"In the beginning, God created the heavens and the earth."<br />
<br />
Note that not all nouns in the nominative case are @nsubj. For some reason, ''osmon'' "heaven" in this sentence is in the nominative case even though it is an object and a patient. Perhaps the accusative ending on ''yerni'' "earth" is able to cover both conjuncts.<br />
<br />
Sometimes @nsubj can be assigned to things that are not prototypically nouns. This occurs in the following sentence.<br />
<br />
<code><br />
"<Kech>"<br />
<br />
"kech" adv subst nom @nsubj #1->2<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @root #2->0<br />
<br />
"<,>"<br />
<br />
"," cm @cc #3->5<br />
<br />
"<erta>"<br />
<br />
"erta" adv subst nom @nsubj #4->5<br />
<br />
"<bo‘ldi>"<br />
<br />
"bo'l" v tv past p3 sg @conj #5->2<br />
<br />
"<–>"<br />
<br />
"-" guio @cc #6->7<br />
<br />
"<bu>"<br />
<br />
"bu" prn dem nom @conj #7->2<br />
<br />
"<birinchi>"<br />
<br />
"bir" num ord @nmod #8->9<br />
<br />
"<kun>"<br />
<br />
"kun" n nom @nsubj #9->7<br />
<br />
"<edi>"<br />
<br />
"e" cop past p3 sg @cop #10->7<br />
<br />
"<.>"<br />
<br />
"." sent @punct #11->2<br />
<br />
</code><br />
<br />
''Kech bo'ldi, erta bo'ldi - bu birinchi kun edi.''<br />
<br />
Evening was, morning was - this first day was.<br />
<br />
"And the evening and the morning were the first day."<br />
<br />
According to both the transducer and the CTILD dictionary, ''kech'' "evening" and ''erta'' "morning" are actually adverbs, but in this sentence, they fill the role of nominal subject.<br />
<br />
=== @obj ===<br />
<br />
@obj is usually applied to a patient-like constituent. In Uzbek, the constituent designated @obj will often be a noun in the accusative case. This is seen in the sentence below from our corpus.<br />
<br />
<code><br />
"<Ushbu>"<br />
<br />
"ushbu" prn dem nom @det #1->2<br />
<br />
"<gumbazni>"<br />
<br />
"gumbaz" n acc @obj #2->7<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @advmod #3->7<br />
<br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #4->7<br />
<br />
"<osmon>"<br />
<br />
"osmon" n nom @xcomp #5->7<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #6->7<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @root #7->0<br />
<br />
"<.>"<br />
<br />
"." sent @punct #8->7<br />
</code><br />
<br />
''Ushbu gumbazni esa Xudo osmon deb atadi.''<br />
<br />
This dome whereas God heaven that called.<br />
<br />
"God called the expanse heaven."<br />
<br />
However, sometimes the @obj can be in other cases. In the sentence below, the obj ''suvlarning'' "waters" is in the genitive case.<br />
<br />
<code><br />
"<Xudo>"<br />
<br />
"xudo" n nom @nsubj #1->5<br />
<br />
"<quruqlikni>"<br />
<br />
"quruqlik" n acc @obj #2->5<br />
<br />
"<yer>"<br />
<br />
"yer" n nom @xcomp #3->5<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #4->5<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @root #5->0<br />
<br />
"<,>"<br />
<br />
"," cm @cc #6->12<br />
<br />
"<suvlarning>"<br />
<br />
"suv" n pl gen @obj #7->12<br />
<br />
"<jamini>"<br />
<br />
"jami" adj subst acc @nmod #8->7<br />
<br />
"<esa>"<br />
<br />
"esa" cnjcoo @advmod #9->12<br />
<br />
"<dengizlar>"<br />
<br />
"dengiz" n pl nom @xcomp #10->12<br />
<br />
"<deb>"<br />
<br />
"deb" cnjsub @advmod #11->12<br />
<br />
"<atadi>"<br />
<br />
"ata" v tv past p3 sg @conj #12->5<br />
<br />
"<.>"<br />
<br />
"." sent @punct #13->5<br />
</code><br />
<br />
''Xudo quruqlikni yer deb atadi, suvlarning jamini esa dengizlar deb atadi.''<br />
<br />
God land earth that called, waters gathering whereas sea that called.<br />
<br />
"God called the dry land earth, and the gathering of the waters He called seas."<br />
<br />
=== @parataxis ===<br />
<br />
@parataxis is the relationship between the main predicate of a sentence and other elements "placed side by side without any explicit coordination, subordination, or argument relation with the head word." Parataxis is used in a variety of situations where the relationship between main sentence and clause is not explicit, especially in direct or reported speech (and so can often be identified w/ punctuation like ";" or ":"). @parataxis can be contrasted against @ccomp, which is used when the reported speech is clearly in a subordinate clause of the main verb. We used @parataxis to identify reported speech in our corpus.<br />
<br />
<code><br />
<br />
"<Xudo>"<br />
"xudo" n nom @nsubj #1->8<br />
"<:>"<br />
":" sent @punct #2->8<br />
"<«>"<br />
"«" lquot @punct #3->5<br />
"<Yorug‘lik>"<br />
"yorug'lik" n nom @nsubj #4->5<br />
"<bo‘lsin>"<br />
"bo'l" v tv perm @parataxis #5->8<br />
"<»>"<br />
"»" rquot @punct #6->5<br />
"<,>"<br />
"," cm @punct #7->8<br />
"<dedi>"<br />
"de" v tv past p3 sg @root #8->0<br />
"<va>"<br />
"va" cnjcoo @cc #9->12<br />
"<yorug‘lik>"<br />
"yorug'lik" n nom @nsubj #10->12<br />
"<paydo>"<br />
"paydo" adj @nmod #11->10<br />
"<bo‘ldi>"<br />
"bo'l" v tv past p3 sg @conj #12->8<br />
"<.>"<br />
"." sent @punct #13->8<br />
<br />
</code><br />
<br />
@Parataxis is especially helpful in these examples because the syntax of reported speech in Uzbek is even less coordinated than in English: the sentence begins with the speaker, launches into the reported speech, and we only get the speaking verb at the end of the sentence. <br />
<br />
<code><br />
<br />
"<Yana>"<br />
"yana" adv @advmod #1->17<br />
"<Xudo>"<br />
"xudo" n nom @nsubj #2->17<br />
"<:>"<br />
":" sent @punct #3->17<br />
"<«>"<br />
"«" lquot @punct #4->13<br />
"<Suvliklar>"<br />
"suv" n loc @nmod #5->11<br />
"<o‘rtasida>"<br />
"o'rta" n px3sp loc @case #6->5<br />
"<ularni>"<br />
"ular" prn pers p3 pl acc @obj #7->9<br />
"<bir-biridan>"<br />
"bir-bir" prn ind px3sp abl @nmod #8->7<br />
"<ajratib>"<br />
"ajra" v iv past p3 sg @acl #9->11<br />
"<turadigan>"<br />
"tur" v iv past p3 sg @conj #10->9<br />
"<gumbaz>"<br />
"gumbaz" n nom @nsubj #11->13<br />
"<paydo>"<br />
"paydo" adj @nmod #12->11<br />
"<bo‘lsin>"<br />
"bo'l" v tv perm @parataxis #13->17<br />
"<»>"<br />
"»" rquot @punct #14->13<br />
"<,>"<br />
"," cm @punct #15->17<br />
"<–>"<br />
"-" guio @punct #16->17<br />
"<dedi>"<br />
"de" v tv past p3 sg @root #17->0<br />
"<.>"<br />
"." sent @punct #18->17<br />
<br />
</code><br />
<br />
We also have examples where @acl (clausal modifiers of nouns) are embedded within @parataxis. Which brings us to our next point...<br />
<br />
=== @acl ===<br />
<br />
We used @acl <br />
<br />
[[Category: sp18_UD]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese&diff=7636Latin and Mandarin Chinese2018-04-25T22:20:41Z<p>Eschalk1: /* zho-lat */</p>
<hr />
<div>==Resources for machine translation between [https://wikis.swarthmore.edu/ling073/Latin Latin] and [https://wikis.swarthmore.edu/ling073/Mandarin_Chinese Mandarin Chinese]==<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Lexical_selection Lexical Selection Documentation]<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Structural_transfer Structural transfer]<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho Machine Translator]<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho-corpus Parallel Corpus]<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Contrastive_Grammar Contrastive Grammar]<br />
<br />
== lat → zho evaluation ==<br />
<br />
Sentences WER: 586.79%<br />
<br />
Sentences PER: 586.79%<br />
<br />
Tests WER: 609.9%<br />
<br />
Tests PER: 609.9%<br />
<br />
== zho → lat evaluation ==<br />
<br />
Sentences WER: 97.42%<br />
<br />
Sentences PER: 93.55%<br />
<br />
Tests WER: 95.45%<br />
<br />
Tests PER: 89.39%<br />
<br />
==Proportion of stems translated correctly==<br />
<br />
lat.sentences.txt: 314:173<br />
<br />
lat.tests.txt: 66:17<br />
<br />
zho.sentences.txt: 53:0<br />
<br />
zho.tests.txt: 10:0<br />
<br />
== Lexical Selection ==<br />
<br />
One case of one-to-many mapping from Chinese to Latin has to do with the preposition 在. This is a general locative preposition in Chinese, and it can be used to mean anything from 'in' to 'above' to 'outside' to 'to in front of' to 'west of.' Usually, the object of 在 is either a place word (such as 'China' or 'the library') or a word that means 'in', 'above', etc. which is then modified by a noun, resulting in a phrase sort of equivalent to "在 the inside of the box" or "在 the top of the table." In contrast, Latin has many different prepositions which specify different locational relationships to the object of the preposition, much like English.<br />
<br />
Another case has to do with pronouns; each Latin pronoun has many different forms for different cases. In contrast, in Chinese each pronoun has at most two different forms: one singular and one plural.<br />
<br />
It's a bit more difficult to find one-to-many mappings from Latin to Chinese due to the nature of the beast. However, note that Chinese is a classifier language, meaning that you need a classifier or measure word in order to couple a noun with a numeral or determiner. Different nouns can have different measure words depending on the shape of the object they represent, among other factors. This is different from Latin, in which you don't have to worry about all of these different measure words when pairing a noun with a numeral; you just have to make sure the numeral agrees with the noun in terms of gender, number, and case. In that sense, this might be considered a one-to-many mapping (and sometimes literally a "one" to many mapping).<br />
<br />
Also note that Chinese has a number of sentence-ending particles that can be used to "soften" or otherwise modulate a sentence. For example, the particle 呢 often marks a rhetorical question. In Latin, a rhetorical question would, I think, have the same form as a genuine question, making this another possible one-to-many mapping from Latin to Chinese.<br />
<br />
== Additions ==<br />
<br />
=== zho ===<br />
<br />
Added 2 new lexical selection rules<br />
<br />
Added 2 new structural transfer rules<br />
<br />
===lat===<br />
<br />
Added 6 new paradigms<br />
<br />
Added 5 new disambiguation rules<br />
<br />
== Final Evaluation ==<br />
<br />
=== zho ===<br />
<br />
'''Precision and recall'''<br />
<br />
Totals: 135 forms, 356 tp, 0 fp, 0 tn, 0 fn<br />
<br />
Precision: 100.00000%<br />
<br />
Recall: 100.00000%<br />
<br />
'''Coverage over large corpus:''' 100.00%<br />
<br />
'''# of words in large corpus:''' 232,490<br />
<br />
'''# of stems in transducer:''' ~8574<br />
<br />
=== lat-zho ===<br />
<br />
'''longer corpus'''<br />
<br />
WER: 103.17%<br />
<br />
PER: 87.53%<br />
<br />
Number of words in reference: 441<br />
<br />
Number of words in test: 516<br />
<br />
Number of unknown words (marked with a star) in test: 139<br />
<br />
Percentage of unknown words: 26.94 %<br />
<br />
Trimmed coverage: 100.00%<br />
<br />
'''large corpus'''<br />
<br />
Number of tokenised words in the corpus: 4997<br />
<br />
Coverage: 40.70%<br />
<br />
Top unknown words in the corpus:<br />
<br />
95 e<br />
<br />
65 f<br />
<br />
64 The<br />
<br />
58 c<br />
<br />
50 A<br />
<br />
45 d<br />
<br />
42 C<br />
<br />
41 h<br />
<br />
37 of<br />
<br />
36 a<br />
<br />
36 b<br />
<br />
31 g<br />
<br />
29 M<br />
<br />
22 B<br />
<br />
21 AMC<br />
<br />
21 ISBN<br />
<br />
21 T<br />
<br />
20 the<br />
<br />
19 Dead<br />
<br />
18 s<br />
<br />
=== lat ===<br />
<br />
'''Precision and recall'''<br />
<br />
Totals: 100 forms, 217 tp, 21 fp, 0 tn, 31 fn<br />
Precision: 91.17647%<br />
Recall: 87.50000%<br />
<br />
'''Coverage over large corpus''': 51.04%<br />
<br />
'''# of words in large corpus''': 7,279,816<br />
<br />
'''# of stems in transducer''': 1,034<br />
<br />
=== zho-lat ===<br />
<br />
WER: 120.48 %<br />
<br />
PER: 117.82 %<br />
<br />
Proportion of stems translated correctly: 8.2%<br />
<br />
Longer corpus trimmed coverage:<br />
<br />
Number of tokenised words in the corpus: 404<br />
<br />
Coverage: 100.00%<br />
<br />
Large corpus trimmed coverage:<br />
<br />
Number of tokenised words in the corpus: 54696<br />
<br />
Coverage: 100.00%<br />
<br />
[[Category:Sp18_TranslationPairs]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese&diff=7635Latin and Mandarin Chinese2018-04-25T21:57:16Z<p>Eschalk1: /* zho-lat */</p>
<hr />
<div>==Resources for machine translation between [https://wikis.swarthmore.edu/ling073/Latin Latin] and [https://wikis.swarthmore.edu/ling073/Mandarin_Chinese Mandarin Chinese]==<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Lexical_selection Lexical Selection Documentation]<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Structural_transfer Structural transfer]<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho Machine Translator]<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho-corpus Parallel Corpus]<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Contrastive_Grammar Contrastive Grammar]<br />
<br />
== lat → zho evaluation ==<br />
<br />
Sentences WER: 586.79%<br />
<br />
Sentences PER: 586.79%<br />
<br />
Tests WER: 609.9%<br />
<br />
Tests PER: 609.9%<br />
<br />
== zho → lat evaluation ==<br />
<br />
Sentences WER: 97.42%<br />
<br />
Sentences PER: 93.55%<br />
<br />
Tests WER: 95.45%<br />
<br />
Tests PER: 89.39%<br />
<br />
==Proportion of stems translated correctly==<br />
<br />
lat.sentences.txt: 314:173<br />
<br />
lat.tests.txt: 66:17<br />
<br />
zho.sentences.txt: 53:0<br />
<br />
zho.tests.txt: 10:0<br />
<br />
== Lexical Selection ==<br />
<br />
One case of one-to-many mapping from Chinese to Latin has to do with the preposition 在. This is a general locative preposition in Chinese, and it can be used to mean anything from 'in' to 'above' to 'outside' to 'to in front of' to 'west of.' Usually, the object of 在 is either a place word (such as 'China' or 'the library') or a word that means 'in', 'above', etc. which is then modified by a noun, resulting in a phrase sort of equivalent to "在 the inside of the box" or "在 the top of the table." In contrast, Latin has many different prepositions which specify different locational relationships to the object of the preposition, much like English.<br />
<br />
Another case has to do with pronouns; each Latin pronoun has many different forms for different cases. In contrast, in Chinese each pronoun has at most two different forms: one singular and one plural.<br />
<br />
It's a bit more difficult to find one-to-many mappings from Latin to Chinese due to the nature of the beast. However, note that Chinese is a classifier language, meaning that you need a classifier or measure word in order to couple a noun with a numeral or determiner. Different nouns can have different measure words depending on the shape of the object they represent, among other factors. This is different from Latin, in which you don't have to worry about all of these different measure words when pairing a noun with a numeral; you just have to make sure the numeral agrees with the noun in terms of gender, number, and case. In that sense, this might be considered a one-to-many mapping (and sometimes literally a "one" to many mapping).<br />
<br />
Also note that Chinese has a number of sentence-ending particles that can be used to "soften" or otherwise modulate a sentence. For example, the particle 呢 often marks a rhetorical question. In Latin, a rhetorical question would, I think, have the same form as a genuine question, making this another possible one-to-many mapping from Latin to Chinese.<br />
<br />
== Additions ==<br />
<br />
=== zho ===<br />
<br />
Added 2 new lexical selection rules<br />
<br />
Added 2 new structural transfer rules<br />
<br />
===lat===<br />
<br />
Added 6 new paradigms<br />
<br />
Added 5 new disambiguation rules<br />
<br />
== Final Evaluation ==<br />
<br />
=== zho ===<br />
<br />
'''Precision and recall'''<br />
<br />
Totals: 135 forms, 356 tp, 0 fp, 0 tn, 0 fn<br />
<br />
Precision: 100.00000%<br />
<br />
Recall: 100.00000%<br />
<br />
'''Coverage over large corpus:''' 100.00%<br />
<br />
'''# of words in large corpus:''' 232,490<br />
<br />
'''# of stems in transducer:''' ~8574<br />
<br />
=== lat-zho ===<br />
<br />
'''longer corpus'''<br />
<br />
WER: 103.17%<br />
<br />
PER: 87.53%<br />
<br />
Number of words in reference: 441<br />
<br />
Number of words in test: 516<br />
<br />
Number of unknown words (marked with a star) in test: 139<br />
<br />
Percentage of unknown words: 26.94 %<br />
<br />
Trimmed coverage: 100.00%<br />
<br />
'''large corpus'''<br />
<br />
Number of tokenised words in the corpus: 4997<br />
<br />
Coverage: 40.70%<br />
<br />
Top unknown words in the corpus:<br />
<br />
95 e<br />
<br />
65 f<br />
<br />
64 The<br />
<br />
58 c<br />
<br />
50 A<br />
<br />
45 d<br />
<br />
42 C<br />
<br />
41 h<br />
<br />
37 of<br />
<br />
36 a<br />
<br />
36 b<br />
<br />
31 g<br />
<br />
29 M<br />
<br />
22 B<br />
<br />
21 AMC<br />
<br />
21 ISBN<br />
<br />
21 T<br />
<br />
20 the<br />
<br />
19 Dead<br />
<br />
18 s<br />
<br />
=== lat ===<br />
<br />
'''Precision and recall'''<br />
<br />
Totals: 100 forms, 217 tp, 21 fp, 0 tn, 31 fn<br />
Precision: 91.17647%<br />
Recall: 87.50000%<br />
<br />
'''Coverage over large corpus''': 51.04%<br />
<br />
'''# of words in large corpus''': 7,279,816<br />
<br />
'''# of stems in transducer''': 1,034<br />
<br />
=== zho-lat ===<br />
<br />
'''WER and PER''' : <br />
<br />
Edit distance: 453<br />
Word Error Rate (WER): 120.48 %<br />
Number of position-independent correct words: 31<br />
Position-independent word error rate (PER): 117.82 %<br />
<br />
[[Category:Sp18_TranslationPairs]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese&diff=7634Latin and Mandarin Chinese2018-04-25T21:49:07Z<p>Eschalk1: </p>
<hr />
<div>==Resources for machine translation between [https://wikis.swarthmore.edu/ling073/Latin Latin] and [https://wikis.swarthmore.edu/ling073/Mandarin_Chinese Mandarin Chinese]==<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Lexical_selection Lexical Selection Documentation]<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Structural_transfer Structural transfer]<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho Machine Translator]<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho-corpus Parallel Corpus]<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Contrastive_Grammar Contrastive Grammar]<br />
<br />
== lat → zho evaluation ==<br />
<br />
Sentences WER: 586.79%<br />
<br />
Sentences PER: 586.79%<br />
<br />
Tests WER: 609.9%<br />
<br />
Tests PER: 609.9%<br />
<br />
== zho → lat evaluation ==<br />
<br />
Sentences WER: 97.42%<br />
<br />
Sentences PER: 93.55%<br />
<br />
Tests WER: 95.45%<br />
<br />
Tests PER: 89.39%<br />
<br />
==Proportion of stems translated correctly==<br />
<br />
lat.sentences.txt: 314:173<br />
<br />
lat.tests.txt: 66:17<br />
<br />
zho.sentences.txt: 53:0<br />
<br />
zho.tests.txt: 10:0<br />
<br />
== Lexical Selection ==<br />
<br />
One case of one-to-many mapping from Chinese to Latin has to do with the preposition 在. This is a general locative preposition in Chinese, and it can be used to mean anything from 'in' to 'above' to 'outside' to 'to in front of' to 'west of.' Usually, the object of 在 is either a place word (such as 'China' or 'the library') or a word that means 'in', 'above', etc. which is then modified by a noun, resulting in a phrase sort of equivalent to "在 the inside of the box" or "在 the top of the table." In contrast, Latin has many different prepositions which specify different locational relationships to the object of the preposition, much like English.<br />
<br />
Another case has to do with pronouns; each Latin pronoun has many different forms for different cases. In contrast, in Chinese each pronoun has at most two different forms: one singular and one plural.<br />
<br />
It's a bit more difficult to find one-to-many mappings from Latin to Chinese due to the nature of the beast. However, note that Chinese is a classifier language, meaning that you need a classifier or measure word in order to couple a noun with a numeral or determiner. Different nouns can have different measure words depending on the shape of the object they represent, among other factors. This is different from Latin, in which you don't have to worry about all of these different measure words when pairing a noun with a numeral; you just have to make sure the numeral agrees with the noun in terms of gender, number, and case. In that sense, this might be considered a one-to-many mapping (and sometimes literally a "one" to many mapping).<br />
<br />
Also note that Chinese has a number of sentence-ending particles that can be used to "soften" or otherwise modulate a sentence. For example, the particle 呢 often marks a rhetorical question. In Latin, a rhetorical question would, I think, have the same form as a genuine question, making this another possible one-to-many mapping from Latin to Chinese.<br />
<br />
== Additions ==<br />
<br />
=== zho ===<br />
<br />
Added 2 new lexical selection rules<br />
<br />
Added 2 new structural transfer rules<br />
<br />
===lat===<br />
<br />
Added 6 new paradigms<br />
<br />
Added 5 new disambiguation rules<br />
<br />
== Final Evaluation ==<br />
<br />
=== zho ===<br />
<br />
'''Precision and recall'''<br />
<br />
Totals: 135 forms, 356 tp, 0 fp, 0 tn, 0 fn<br />
<br />
Precision: 100.00000%<br />
<br />
Recall: 100.00000%<br />
<br />
'''Coverage over large corpus:''' 100.00%<br />
<br />
'''# of words in large corpus:''' 232,490<br />
<br />
'''# of stems in transducer:''' ~8574<br />
<br />
=== lat-zho ===<br />
<br />
'''longer corpus'''<br />
<br />
WER: 103.17%<br />
<br />
PER: 87.53%<br />
<br />
Number of words in reference: 441<br />
<br />
Number of words in test: 516<br />
<br />
Number of unknown words (marked with a star) in test: 139<br />
<br />
Percentage of unknown words: 26.94 %<br />
<br />
Trimmed coverage: 100.00%<br />
<br />
'''large corpus'''<br />
<br />
Number of tokenised words in the corpus: 4997<br />
<br />
Coverage: 40.70%<br />
<br />
Top unknown words in the corpus:<br />
<br />
95 e<br />
<br />
65 f<br />
<br />
64 The<br />
<br />
58 c<br />
<br />
50 A<br />
<br />
45 d<br />
<br />
42 C<br />
<br />
41 h<br />
<br />
37 of<br />
<br />
36 a<br />
<br />
36 b<br />
<br />
31 g<br />
<br />
29 M<br />
<br />
22 B<br />
<br />
21 AMC<br />
<br />
21 ISBN<br />
<br />
21 T<br />
<br />
20 the<br />
<br />
19 Dead<br />
<br />
18 s<br />
<br />
=== lat ===<br />
<br />
'''Precision and recall'''<br />
<br />
Totals: 100 forms, 217 tp, 21 fp, 0 tn, 31 fn<br />
Precision: 91.17647%<br />
Recall: 87.50000%<br />
<br />
'''Coverage over large corpus''': 51.04%<br />
<br />
'''# of words in large corpus''': 7,279,816<br />
<br />
'''# of stems in transducer''': 1,034<br />
<br />
=== zho-lat ===<br />
<br />
<br />
<br />
[[Category:Sp18_TranslationPairs]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese/Structural_transfer&diff=7513Latin and Mandarin Chinese/Structural transfer2018-04-10T16:22:19Z<p>Eschalk1: </p>
<hr />
<div>This is the page for the structural transfer of Latin and Mandarin Chinese. The main page for this language pair can be found [https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese here].<br />
<br />
==Pre-evaluation==<br />
<br />
[[Category:Sp17_StructuralTransfer]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese/Structural_transfer&diff=7512Latin and Mandarin Chinese/Structural transfer2018-04-10T16:20:59Z<p>Eschalk1: </p>
<hr />
<div>[[File:Example.jpg]]wgehrretj<br />
<br />
[[Category:Sp17_StructuralTransfer]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese&diff=7511Latin and Mandarin Chinese2018-04-10T16:20:12Z<p>Eschalk1: </p>
<hr />
<div>==Resources for machine translation between [https://wikis.swarthmore.edu/ling073/Latin Latin] and [https://wikis.swarthmore.edu/ling073/Mandarin_Chinese Mandarin Chinese]==<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Lexical_selection Lexical Selection Documentation]<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Structural_transfer Structural transfer]<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho Machine Translator]<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho-corpus Parallel Corpus]<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Contrastive_Grammar Contrastive Grammar]<br />
<br />
== lat → zho evaluation ==<br />
<br />
Sentences WER: 586.79%<br />
<br />
Sentences PER: 586.79%<br />
<br />
Tests WER: 609.9%<br />
<br />
Tests PER: 609.9%<br />
<br />
== zho → lat evaluation ==<br />
<br />
Sentences WER: 97.42%<br />
<br />
Sentences PER: 93.55%<br />
<br />
Tests WER: 95.45%<br />
<br />
Tests PER: 89.39%<br />
<br />
==Proportion of stems translated correctly==<br />
<br />
lat.sentences.txt: 314:173<br />
<br />
lat.tests.txt: 66:17<br />
<br />
zho.sentences.txt: 53:0<br />
<br />
zho.tests.txt: 10:0<br />
<br />
== Lexical Selection ==<br />
<br />
One case of one-to-many mapping from Chinese to Latin has to do with the preposition 在. This is a general locative preposition in Chinese, and it can be used to mean anything from 'in' to 'above' to 'outside' to 'to in front of' to 'west of.' Usually, the object of 在 is either a place word (such as 'China' or 'the library') or a word that means 'in', 'above', etc. which is then modified by a noun, resulting in a phrase sort of equivalent to "在 the inside of the box" or "在 the top of the table." In contrast, Latin has many different prepositions which specify different locational relationships to the object of the preposition, much like English.<br />
<br />
Another case has to do with pronouns; each Latin pronoun has many different forms for different cases. In contrast, in Chinese each pronoun has at most two different forms: one singular and one plural.<br />
<br />
It's a bit more difficult to find one-to-many mappings from Latin to Chinese due to the nature of the beast. However, note that Chinese is a classifier language, meaning that you need a classifier or measure word in order to couple a noun with a numeral or determiner. Different nouns can have different measure words depending on the shape of the object they represent, among other factors. This is different from Latin, in which you don't have to worry about all of these different measure words when pairing a noun with a numeral; you just have to make sure the numeral agrees with the noun in terms of gender, number, and case. In that sense, this might be considered a one-to-many mapping (and sometimes literally a "one" to many mapping).<br />
<br />
Also note that Chinese has a number of sentence-ending particles that can be used to "soften" or otherwise modulate a sentence. For example, the particle 呢 often marks a rhetorical question. In Latin, a rhetorical question would, I think, have the same form as a genuine question, making this another possible one-to-many mapping from Latin to Chinese.<br />
<br />
[[Category:Sp18_TranslationPairs]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese&diff=7510Latin and Mandarin Chinese2018-04-10T16:19:14Z<p>Eschalk1: </p>
<hr />
<div>==Resources for machine translation between [https://wikis.swarthmore.edu/ling073/Latin Latin] and [https://wikis.swarthmore.edu/ling073/Mandarin_Chinese Mandarin Chinese]==<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Lexical_selection Lexical Selection Documentation]<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Structural_transfer]<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho Machine Translator]<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho-corpus Parallel Corpus]<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Contrastive_Grammar Contrastive Grammar]<br />
<br />
== lat → zho evaluation ==<br />
<br />
Sentences WER: 586.79%<br />
<br />
Sentences PER: 586.79%<br />
<br />
Tests WER: 609.9%<br />
<br />
Tests PER: 609.9%<br />
<br />
== zho → lat evaluation ==<br />
<br />
Sentences WER: 97.42%<br />
<br />
Sentences PER: 93.55%<br />
<br />
Tests WER: 95.45%<br />
<br />
Tests PER: 89.39%<br />
<br />
==Proportion of stems translated correctly==<br />
<br />
lat.sentences.txt: 314:173<br />
<br />
lat.tests.txt: 66:17<br />
<br />
zho.sentences.txt: 53:0<br />
<br />
zho.tests.txt: 10:0<br />
<br />
== Lexical Selection ==<br />
<br />
One case of one-to-many mapping from Chinese to Latin has to do with the preposition 在. This is a general locative preposition in Chinese, and it can be used to mean anything from 'in' to 'above' to 'outside' to 'to in front of' to 'west of.' Usually, the object of 在 is either a place word (such as 'China' or 'the library') or a word that means 'in', 'above', etc. which is then modified by a noun, resulting in a phrase sort of equivalent to "在 the inside of the box" or "在 the top of the table." In contrast, Latin has many different prepositions which specify different locational relationships to the object of the preposition, much like English.<br />
<br />
Another case has to do with pronouns; each Latin pronoun has many different forms for different cases. In contrast, in Chinese each pronoun has at most two different forms: one singular and one plural.<br />
<br />
It's a bit more difficult to find one-to-many mappings from Latin to Chinese due to the nature of the beast. However, note that Chinese is a classifier language, meaning that you need a classifier or measure word in order to couple a noun with a numeral or determiner. Different nouns can have different measure words depending on the shape of the object they represent, among other factors. This is different from Latin, in which you don't have to worry about all of these different measure words when pairing a noun with a numeral; you just have to make sure the numeral agrees with the noun in terms of gender, number, and case. In that sense, this might be considered a one-to-many mapping (and sometimes literally a "one" to many mapping).<br />
<br />
Also note that Chinese has a number of sentence-ending particles that can be used to "soften" or otherwise modulate a sentence. For example, the particle 呢 often marks a rhetorical question. In Latin, a rhetorical question would, I think, have the same form as a genuine question, making this another possible one-to-many mapping from Latin to Chinese.<br />
<br />
[[Category:Sp18_TranslationPairs]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese/Structural_transfer&diff=7509Latin and Mandarin Chinese/Structural transfer2018-04-10T16:19:00Z<p>Eschalk1: Created page with "wgehrretj"</p>
<hr />
<div>wgehrretj</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese/Lexical_selection&diff=7493Latin and Mandarin Chinese/Lexical selection2018-04-08T21:16:09Z<p>Eschalk1: </p>
<hr />
<div>== lat → zho ==<br />
<br />
Our examples for one-to-many mappings from Latin to Chinese from last assignment were (1) the many different classifiers that can go with different nouns and (2) the sentence-ending particle 呢, often marking a rhetorical question (which, in Latin, would probably not be formally different from a genuine question). However, we later realized that these aren't exactly lexical selection problems. Instead, we decided to address the words ''tempestas'' and ''soror''.<br />
<br />
=== tempestas ===<br />
<br />
The Latin word ''tempestas'' can mean "weather" or "storm". In contrast, Chinese has 天气 "weather" and 风暴 "storm". One way we can distinguish between these two senses is by looking at the adjectives that are associated with ''tempestas''. For example, if the ''tempestas'' is ''turbulenta'' "turbulent," it's probably talking about a "storm."<br />
<br />
tempestas ferox "fierce storm" → 风暴<br />
<br />
tempestas atrox "terrible storm" → 风暴<br />
<br />
tempestas foeda "ugly storm" → 风暴<br />
<br />
tempestas turbida "turbid storm" → 风暴<br />
<br />
tempestas perfrigida "freezing storm" → 风暴<br />
<br />
tempestas turbulenta "turbulent storm" → 风暴<br />
<br />
tempestas magna "large storm" → 风暴<br />
<br />
We found our adjectives associated with "tempestas" by doing a usage search on http://www.perseus.tufts.edu/hopper/.<br />
<br />
Since "The ship was wrecked because of the weather" is a bit more understandable as to what is going on than "If the storm tomorrow is nice, I may go for a walk," we will have 天气 as the default translation.<br />
<br />
'''Example Sentences'''<br />
<br />
Tempestas pulchra est → 天气很好。<br />
<br />
"The weather is beautiful."<br />
<br />
Tempestas turbulenta est. → 风暴很汹涌。<br />
<br />
"The storm is turbulent."<br />
<br />
=== soror ===<br />
<br />
Latin has just one word for "sister", ''soror''. Chinese has 姐姐 "older sister" and 妹妹 "younger sister". This is pretty hard to select for in most cases, but at least we can tell if the Latin happens to pair ''soror'' with the adjective "big" or "small" (the Latin way of saying "older" and "younger") or some other clear indicator of age.<br />
<br />
soror major "older sister" → 姐姐<br />
<br />
soror minor "younger sister" → 妹妹<br />
<br />
soror infans "infant sister" → 妹妹<br />
<br />
'''Example Sentences'''<br />
<br />
Puella parva soror minor sua est. → 小女孩是他的妹妹。<br />
<br />
"The little girl is his younger sister."<br />
<br />
Nosco sororem majorem eius. → 我认识她的姐姐。<br />
<br />
"I know her older sister."<br />
<br />
== zho → lat ==<br />
<br />
Our examples for one-to-many mappings from Chinese to Latin were (1) the preposition 在, which can express many different locative relationships for which Latin has distinct prepositions and (2) the fact that Latin pronouns have many different forms depending on case etc., while Chinese only has one form for each pronoun. The former would be a good option if it were not for the fact that Latin often uses the locative case rather than prepositions anyway, and the latter is actually more of a grammatical issue than a lexical selection problem, so we decided to work on 开 and 头.<br />
<br />
=== 开 ===<br />
<br />
开 ''kai'' is a verb with a lot of meanings. It can mean, among other things, "to open", "to drive", "to bloom", "to hold/run/convene (e.g. a meeting)", and "to start/establish (e.g. an organization)" (see [http://www.aclweb.org/anthology/Y95-1013 Wang 1995]). Unsurprisingly, Latin has separate words for many of these concepts. For this assignment, we will focus on selecting between 开 in the sense of "to open" and 开 in the sense of "to hold/run", since 开 in the sense of "to drive" seems mostly to be used with motorized vehicles, which Latin oddly doesn't have a lot of words for. The Latin word for "to open" is ''aperiō'' (whence ''Apertium'' "open-source software for machine translation"). One Latin word for "to hold/run/convene" is ''convenio''. We can select between these words by looking at the object of the verb:<br />
<br />
开门 "to open a door" → aperiō<br />
<br />
开盒子 "to open a box" → aperiō<br />
<br />
开窗口 "to open a window" → aperiō<br />
<br />
开抽屉 "to open a drawer" → aperiō<br />
<br />
开会 "to hold a meeting" → convenio<br />
<br />
开课 "to hold a class" → convenio<br />
<br />
'''Example Sentences'''<br />
<br />
老师让学生开门。→ Magister petivit discipulum aperire ianuam.<br />
<br />
"The teacher asked the student to open the door."<br />
<br />
我们每天开会。→ Convenimus concilium quotidie.<br />
<br />
"We hold a meeting every day."<br />
<br />
=== 头 ===<br />
<br />
头 ''tou'' "head" can also refer to hair in some situations (see [http://studentsrepo.um.edu.my/5407/1/LV_YETONG_(TGC110022).pdf Lv 2014]), which is not the case in Latin. In Latin, "head" is ''caput'' and "hair" is ''capillum''. One way to select between these two senses is to look at the verbs and adjectives that go with ''tou''. Some verbs and adjectives would probably only be seen with ''tou'' in the sense of "hair."<br />
<br />
梳头 "to comb hair" → capillum<br />
<br />
剃头 "to shave hair" → capillum<br />
<br />
洗头 "to wash hair" → capillum<br />
<br />
平头 → "flat hair/crewcut" → capillum<br />
<br />
潮头 → "fashionable hair" → capillum<br />
<br />
You might think "long hair," "curly hair," etc. might be good choices, but I think Chinese uses a different word for hair in these phrases. Since the primary meaning of 头 is "head," we will have ''caput'' as the default translation.<br />
<br />
'''Example Sentences'''<br />
<br />
他的妈妈不愿意给他梳头。→ Mater eius non vult pectere capilium eius.<br />
<br />
"His mother isn't willing to comb his hair."<br />
<br />
点头就表示 “是”。→ Adnuere caput tuum significat 'ita'.<br />
<br />
"Nodding your head means 'yes.' "<br />
<br />
[[Category:Sp18_LexicalSelection]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese&diff=7351Latin and Mandarin Chinese2018-04-03T01:29:49Z<p>Eschalk1: </p>
<hr />
<div>==Resources for machine translation between [https://wikis.swarthmore.edu/ling073/Latin Latin] and [https://wikis.swarthmore.edu/ling073/Mandarin_Chinese Mandarin Chinese]==<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho Machine Translator]<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho-corpus Parallel Corpus]<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Contrastive_Grammar Contrastive Grammar]<br />
<br />
== lat → zho evaluation ==<br />
<br />
Sentences WER: 586.79%<br />
<br />
Sentences PER: 586.79%<br />
<br />
Tests WER: 609.9%<br />
<br />
Tests PER: 609.9%<br />
<br />
== zho → lat evaluation ==<br />
<br />
Sentences WER: 97.42%<br />
<br />
Sentences PER: 93.55%<br />
<br />
Tests WER: 95.45%<br />
<br />
Tests PER: 89.39%<br />
<br />
==Proportion of stems translated correctly==<br />
<br />
lat.sentences.txt: 314:173<br />
<br />
lat.tests.txt: 66:17<br />
<br />
zho.sentences.txt: 53:0<br />
<br />
zho.tests.txt: 10:0<br />
<br />
== Lexical Selection ==<br />
<br />
One case of one-to-many mapping from Chinese to Latin has to do with the preposition 在. This is a general locative preposition in Chinese, and it can be used to mean anything from 'in' to 'above' to 'outside' to 'to in front of' to 'west of.' Usually, the object of 在 is either a place word (such as 'China' or 'the library') or a word that means 'in', 'above', etc. which is then modified by a noun, resulting in a phrase sort of equivalent to "在 the inside of the box" or "在 the top of the table." In contrast, Latin has many different prepositions which specify different locational relationships to the object of the preposition, much like English.<br />
<br />
Another case has to do with pronouns; each Latin pronoun has many different forms for different cases. In contrast, in Chinese each pronoun has at most two different forms: one singular and one plural.<br />
<br />
It's a bit more difficult to find one-to-many mappings from Latin to Chinese due to the nature of the beast. However, note that Chinese is a classifier language, meaning that you need a classifier or measure word in order to couple a noun with a numeral or determiner. Different nouns can have different measure words depending on the shape of the object they represent, among other factors. This is different from Latin, in which you don't have to worry about all of these different measure words when pairing a noun with a numeral; you just have to make sure the numeral agrees with the noun in terms of gender, number, and case. In that sense, this might be considered a one-to-many mapping (and sometimes literally a "one" to many mapping).<br />
<br />
Also note that Chinese has a number of sentence-ending particles that can be used to "soften" or otherwise modulate a sentence. For example, the particle 呢 often marks a rhetorical question. In Latin, a rhetorical question would, I think, have the same form as a genuine question, making this another possible one-to-many mapping from Latin to Chinese.<br />
<br />
[[Category:Sp18_TranslationPairs]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese&diff=7350Latin and Mandarin Chinese2018-04-03T01:22:33Z<p>Eschalk1: /* zho → lat evaluation */</p>
<hr />
<div>==Resources for machine translation between [https://wikis.swarthmore.edu/ling073/Latin Latin] and [https://wikis.swarthmore.edu/ling073/Mandarin_Chinese Mandarin Chinese]==<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho Machine Translator]<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho-corpus Parallel Corpus]<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Contrastive_Grammar Contrastive Grammar]<br />
<br />
== lat → zho evaluation ==<br />
<br />
Sentences WER: 586.79%<br />
<br />
Sentences PER: 586.79%<br />
<br />
Tests WER: 609.9%<br />
<br />
Tests PER: 609.9%<br />
<br />
== zho → lat evaluation ==<br />
<br />
Sentences WER: 97.42%<br />
<br />
Sentences PER: 93.55%<br />
<br />
Tests WER: 95.45%<br />
<br />
Tests PER: 89.39%<br />
<br />
== Lexical Selection ==<br />
<br />
One case of one-to-many mapping from Chinese to Latin has to do with the preposition 在. This is a general locative preposition in Chinese, and it can be used to mean anything from 'in' to 'above' to 'outside' to 'to in front of' to 'west of.' Usually, the object of 在 is either a place word (such as 'China' or 'the library') or a word that means 'in', 'above', etc. which is then modified by a noun, resulting in a phrase sort of equivalent to "在 the inside of the box" or "在 the top of the table." In contrast, Latin has many different prepositions which specify different locational relationships to the object of the preposition, much like English.<br />
<br />
Another case has to do with pronouns; each Latin pronoun has many different forms for different cases. In contrast, in Chinese each pronoun has at most two different forms: one singular and one plural.<br />
<br />
It's a bit more difficult to find one-to-many mappings from Latin to Chinese due to the nature of the beast. However, note that Chinese is a classifier language, meaning that you need a classifier or measure word in order to couple a noun with a numeral or determiner. Different nouns can have different measure words depending on the shape of the object they represent, among other factors. This is different from Latin, in which you don't have to worry about all of these different measure words when pairing a noun with a numeral; you just have to make sure the numeral agrees with the noun in terms of gender, number, and case. In that sense, this might be considered a one-to-many mapping (and sometimes literally a "one" to many mapping).<br />
<br />
Also note that Chinese has a number of sentence-ending particles that can be used to "soften" or otherwise modulate a sentence. For example, the particle 呢 often marks a rhetorical question. In Latin, a rhetorical question would, I think, have the same form as a genuine question, making this another possible one-to-many mapping from Latin to Chinese.<br />
<br />
[[Category:Sp18_TranslationPairs]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese&diff=7349Latin and Mandarin Chinese2018-04-03T01:20:11Z<p>Eschalk1: /* lat → zho evaluation */</p>
<hr />
<div>==Resources for machine translation between [https://wikis.swarthmore.edu/ling073/Latin Latin] and [https://wikis.swarthmore.edu/ling073/Mandarin_Chinese Mandarin Chinese]==<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho Machine Translator]<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho-corpus Parallel Corpus]<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Contrastive_Grammar Contrastive Grammar]<br />
<br />
== lat → zho evaluation ==<br />
<br />
Sentences WER: 586.79%<br />
<br />
Sentences PER: 586.79%<br />
<br />
Tests WER: 609.9%<br />
<br />
Tests PER: 609.9%<br />
<br />
== zho → lat evaluation ==<br />
<br />
== Lexical Selection ==<br />
<br />
One case of one-to-many mapping from Chinese to Latin has to do with the preposition 在. This is a general locative preposition in Chinese, and it can be used to mean anything from 'in' to 'above' to 'outside' to 'to in front of' to 'west of.' Usually, the object of 在 is either a place word (such as 'China' or 'the library') or a word that means 'in', 'above', etc. which is then modified by a noun, resulting in a phrase sort of equivalent to "在 the inside of the box" or "在 the top of the table." In contrast, Latin has many different prepositions which specify different locational relationships to the object of the preposition, much like English.<br />
<br />
Another case has to do with pronouns; each Latin pronoun has many different forms for different cases. In contrast, in Chinese each pronoun has at most two different forms: one singular and one plural.<br />
<br />
It's a bit more difficult to find one-to-many mappings from Latin to Chinese due to the nature of the beast. However, note that Chinese is a classifier language, meaning that you need a classifier or measure word in order to couple a noun with a numeral or determiner. Different nouns can have different measure words depending on the shape of the object they represent, among other factors. This is different from Latin, in which you don't have to worry about all of these different measure words when pairing a noun with a numeral; you just have to make sure the numeral agrees with the noun in terms of gender, number, and case. In that sense, this might be considered a one-to-many mapping (and sometimes literally a "one" to many mapping).<br />
<br />
Also note that Chinese has a number of sentence-ending particles that can be used to "soften" or otherwise modulate a sentence. For example, the particle 呢 often marks a rhetorical question. In Latin, a rhetorical question would, I think, have the same form as a genuine question, making this another possible one-to-many mapping from Latin to Chinese.<br />
<br />
[[Category:Sp18_TranslationPairs]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese&diff=7348Latin and Mandarin Chinese2018-04-03T01:01:31Z<p>Eschalk1: </p>
<hr />
<div>==Resources for machine translation between [https://wikis.swarthmore.edu/ling073/Latin Latin] and [https://wikis.swarthmore.edu/ling073/Mandarin_Chinese Mandarin Chinese]==<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho Machine Translator]<br />
<br />
[https://github.swarthmore.edu/cdalton2/ling073-lat-zho-corpus Parallel Corpus]<br />
<br />
[https://wikis.swarthmore.edu/ling073/Latin_and_Mandarin_Chinese/Contrastive_Grammar Contrastive Grammar]<br />
<br />
== lat → zho evaluation ==<br />
<br />
== zho → lat evaluation ==<br />
<br />
== Lexical Selection ==<br />
<br />
One case of one-to-many mapping from Chinese to Latin has to do with the preposition 在. This is a general locative preposition in Chinese, and it can be used to mean anything from 'in' to 'above' to 'outside' to 'to in front of' to 'west of.' Usually, the object of 在 is either a place word (such as 'China' or 'the library') or a word that means 'in', 'above', etc. which is then modified by a noun, resulting in a phrase sort of equivalent to "在 the inside of the box" or "在 the top of the table." In contrast, Latin has many different prepositions which specify different locational relationships to the object of the preposition, much like English.<br />
<br />
Another case has to do with pronouns; each Latin pronoun has many different forms for different cases. In contrast, in Chinese each pronoun has at most two different forms: one singular and one plural.<br />
<br />
It's a bit more difficult to find one-to-many mappings from Latin to Chinese due to the nature of the beast. However, note that Chinese is a classifier language, meaning that you need a classifier or measure word in order to couple a noun with a numeral or determiner. Different nouns can have different measure words depending on the shape of the object they represent, among other factors. This is different from Latin, in which you don't have to worry about all of these different measure words when pairing a noun with a numeral; you just have to make sure the numeral agrees with the noun in terms of gender, number, and case. In that sense, this might be considered a one-to-many mapping (and sometimes literally a "one" to many mapping).<br />
<br />
Also note that Chinese has a number of sentence-ending particles that can be used to "soften" or otherwise modulate a sentence. For example, the particle 呢 often marks a rhetorical question. In Latin, a rhetorical question would, I think, have the same form as a genuine question, making this another possible one-to-many mapping from Latin to Chinese.<br />
<br />
[[Category:Sp18_TranslationPairs]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese/Contrastive_Grammar&diff=7266Latin and Mandarin Chinese/Contrastive Grammar2018-03-27T16:27:30Z<p>Eschalk1: /* lat-zho-tests */</p>
<hr />
<div>==Differences between Latin and Mandarin Chinese==<br />
{{comment|you need the regular transfer tests too!}}<br />
<br />
===Use of the verb 'to be'===<br />
<br />
Latin and Mandarin Chinese both have a verb 'to be' but use it in different contexts. Latin uses this verb in more situations than Chinese; for instance, forms of the Latin verb ''esse'' can be used to link a noun to an adjective, a predicate noun, and even helps form certain definite verbs from its present tense forms and a participle (acting as a copula). Chinese, for the most part, uses its form of this verb, 是, only to link a noun to a predicate noun; otherwise a verb is not necessary to link nouns to adjectives. <br />
<br />
Jamus et Maria in horto sunt. <br />
小明和小红在花园里面。<br />
James and Mary are in the garden.<br />
<br />
Jamus puer parvus est.<br />
小明是一个小男孩。<br />
James is a small boy.<br />
<br />
Tempestas pulchra est hodie, admodum calda est.<br />
今天天气好，很暖。<br />
The weather is nice today, and it is very warm.<br />
<br />
===Word order===<br />
<br />
Latin generally follows a subject-object-verb word order, with exceptions for most poetry, and is a highly inflected language (hence the flexible word order). Chinese has a subject-verb-object word order and is highly uninflected (hence the inflexible word order).<br />
<br />
Maria matrem suam post fenestram adspicit, mater sua ridet.<br />
小红看着在屋子里面的妈妈。妈妈正在笑<br />
Mary sees her mother behind the window (zho: inside the house), and her mother laughs.<br />
<br />
Tum etiam manus parvus videt et cum appropinquat etiam caput suum videt!<br />
她看到了一只手， 走近了一点儿，看到了他！<br />
Then she saw a little hand and when she came closer even his own head!<br />
<br />
===Possession===<br />
<br />
Both Latin and Chinese have a couple of different ways of expressing relationships of possession. Latin forms possessive relationships between nouns with a dative form of the possessing noun (or in other cases, a genitive, but there were no examples of this in our corpus). The same sentences were expressed in Chinese with the verb 有 'to have'; the option of using the dative case is obviously not available in Chinese, as there is no dative case.<br />
<br />
Jamo canis parvus est, etiam canis nunc in horto est.<br />
小明有一只狗，那只狗现在在花园里面。<br />
James has a small dog, the dog is now in the garden.<br />
<br />
Mariaene non canis est, eae catta est.<br />
不，她没有狗，她有只猫。<br />
Mary does not have a dog, she has a cat.<br />
<br />
{{comment|in description also mention use of be verb in Latin versus have verb in Mandarin}}<br />
<br />
===Adverbs===<br />
<br />
While both Latin and Chinese have adverbs that describe the performance of a verb, Chinese can also modify a verb through a complement: the use of a particle, 得, with a following adjective. Alternatively, a verb can be preceded by an adjective with the particle 地. Latin can form an adverb from an adjective using inflection, which it does with most regular adjectives.<br />
<br />
Nunc id bene audire potest.<br />
她现在听得很清楚了。<br />
Now she is able to hear it clearly.<br />
<br />
Auscultat diligenter.<br />
她仔细地听。<br />
She listens carefully.<br />
<br />
===Case===<br />
<br />
Latin inflects all nouns for case (nominative, genitive, accusative, dative, ablative, and sometimes vocative); Chinese expresses the same concepts with word order or prepositions. For instance, Latin uses the accusative case to express the direct object of a verb. Chinese would do the same thing with word order, placing the direct object after the verb.<br />
<br />
Petet Jamum: quo vadebat?<br />
想找小明。他在哪儿呢？<br />
She looks for James; where was he hiding (zho: where is he?)?<br />
<br />
Maria matrem suam post fenestram adspicit, mater sua ridet.<br />
小红看着在屋子里面的妈妈。妈妈正在笑。<br />
Mary sees her mother behind the window, her mother laughs.<br />
<br />
Tum etiam manus parvus videt et cum appropinquat etiam caput suum videt!<br />
她看到了一只手， 走近了一点儿，看到了他！<br />
Then she saw a little hand and when she came closer even his own head!<br />
<br />
==lat-zho-tests==<br />
<br />
* {{transferTest|lat|zho|Jamus et Maria in horto sunt.|小明和小红在花园里面。}}<br />
*: {{transferMorphTest|lat|zho|Jam{{tag|n}}{{tag|nom}}{{tag|m}}{{tag|sg}}et{{tag|conj}}Mari{{tag|n}}{{tag|nom}}{{tag|f}}{{tag|sg}}in{{tag|prep}}hort{{tag|n}}{{tag|abl}}{{tag|m}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|pl}}{{tag|pres}}{{tag|act}}{{tag|ind}}|小明和小红在花园里面。}}<br />
<br />
* {{transferTest|lat|zho|Jamus puer parvus est.|小明是一个小男孩}}<br />
*: {{transferMorphTest|lat|zho|Jam{{tag|n}}{{tag|nom}}{{tag|m}}{{tag|sg}}puer{{tag|n}}{{tag|nom}}{{tag|m}}{{tag|sg}}parv{{tag|adj}}{{tag|nom}}{{tag|m}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|小明是一个小男孩。}}<br />
<br />
* {{transferTest|lat|zho|Tempestas pulchra est hodie, admodum calda est.|今天天气好，很暖}}<br />
*: {{transferMorphTest|lat|zho|tempest{{tag|n}}{{tag|nom}}{{tag|f}}{{tag|sg}}pulchr{{tag|adj}}{{tag|nom}}{{tag|f}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}hodie{{tag|adv}}admodum{{tag|adv}}cald{{tag|adj}}{{tag|nom}}{{tag|f}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|今天天气好，很暖。}}<br />
<br />
* {{transferTest|lat|zho|Maria matrem suam post fenestram adspicit, mater sua ridet.|小红看着在屋子里面的妈妈。妈妈正在笑}}<br />
*: {{transferMorphTest|lat|zho|Mari{{tag|n}}{{tag|nom}}{{tag|f}}{{tag|sg}}matr{{tag|n}}{{tag|acc}}{{tag|f}}{{tag|sg}}su{{tag|prn}}{{tag|poss}}{{tag|refl}}{{tag|acc}}{{tag|f}}{{tag|sg}}post{{tag|prep}}fenestr{{tag|n}}{{tag|acc}}{{tag|f}}{{tag|sg}}adspic{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}matr{{tag|n}}{{tag|nom}}{{tag|f}}{{tag|sg}}su{{tag|prn}}{{tag|poss}}{{tag|refl}}{{tag|nom}}{{tag|f}}{{tag|sg}}rid{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|小红看着在屋子里面的妈妈。妈妈正在笑}}<br />
<br />
* {{transferTest|lat|zho|Tum etiam manus parvus videt et cum appropinquat etiam caput suum videt!|她看到了一只手， 走近了一点儿，看到了他!}}<br />
*: {{transferMorphTest|lat|zho|tum{{tag|adv}}etiam{{tag|adv}}man{{tag|n}}{{tag|acc}}{{tag|m}}{{tag|sg}}parv{{tag|n}}{{tag|acc}}{{tag|m}}{{tag|sg}}vid{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}et{{tag|conj}}cum{{tag|conj}}appropinqu{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|subj}}etiam{{tag|adv}}capit{{tag|n}}{{tag|acc}}{{tag|m}}{{tag|sg}}su{{tag|prn}}{{tag|poss}}{{tag|refl}}{{tag|acc}}{{tag|m}}{{tag|sg}}vid{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|她看到了一只手， 走近了一点儿，看到了他！}}<br />
<br />
* {{transferTest|lat|zho|Jamo canis parvus est, etiam canis nunc in horto est.|小明有一只狗，那只狗现在在花园里面}}<br />
*: {{transferMorphTest|lat|zho|Jam{{tag|n}}{{tag|dat}}{{tag|m}}{{tag|sg}}can{{tag|n}}{{tag|nom}}{{tag|m}}{{tag|sg}}parv{{tag|adj}}{{tag|nom}}{{tag|m}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}etiam{{tag|adv}}can{{tag|n}}{{tag|nom}}{{tag|m}}{{tag|sg}}nunc{{tag|adv}}in{{tag|prep}}hort{{tag|n}}{{tag|abl}}{{tag|m}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|小明有一只狗，那只狗现在在花园里面}}<br />
<br />
* {{transferTest|lat|zho|Nunc id bene audire potest.|她现在听得很清楚了。}}<br />
*: {{transferMorphTest|lat|zho|Nunc{{tag|adv}}id{{tag|prn}}{{tag|acc}}{{tag|nt}}{{tag|sg}}bene{{tag|adv}}audire{{tag|vblex}}{{tag|pres}}{{tag|act}}{{tag|inf}}possum{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|她现在听得很清楚了。}}<br />
<br />
* {{transferTest|lat|zho|Auscultat diligenter.|她仔细地听。}}<br />
*: {{transferMorphTest|lat|zho|auscult{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}diligen{{tag|adv}}|她仔细地听。}}<br />
<br />
* {{transferTest|lat|zho|Petet Jamum: quo vadebat?|他在哪儿呢？}}<br />
*: {{transferMorphTest|lat|zho|pet{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}Jam{{tag|n}}{{tag|acc}}{{tag|m}}{{tag|sg}}quo{{tag|prn}}{{tag|interr}}{{tag|abl}}{{tag|m}}{{tag|sg}}vad{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|impf}}{{tag|act}}{{tag|ind}}|他在哪儿呢？}}<br />
<br />
* {{transferTest|lat|zho|Mariaene non canis est, eae catta est.|不，她没有狗，她有只猫。}}<br />
*: {{transferMorphTest|lat|zho|Mari{{tag|n}}{{tag|dat}}{{tag|f}}{{tag|sg}}{{tag|neg}}non{{tag|neg}}can{{tag|n}}{{tag|nom}}{{tag|m}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}eae{{tag|prn}}{{tag|dat}}{{tag|fem}}{{tag|sg}}catt{{tag|n}}{{tag|nom}}{{tag|fem}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|不，她没有狗，她有只猫。}}<br />
<br />
==zho-lat-tests==<br />
<br />
* {{transferTest|zho|lat|小明和小红在花园里面。|Jamus et Maria in horto sunt.}}<br />
*: {{transferMorphTest|zho|lat|小明{{tag|n}}和{{tag|cnjcoo}}小红{{tag|n}}在{{tag|pr}}花园{{tag|n}}里面{{tag|n}}|Jamus et Maria in horto sunt.}}<br />
<br />
* {{transferTest|zho|lat|小明是一个小男孩|Jamus puer parvus est.}}<br />
*: {{transferMorphTest|zho|lat|小明{{tag|n}}是{{tag|vbser}}一{{tag|num}}个{{tag|clf}}小{{tag|adj}}男孩{{tag|n}}|Jamus puer parvus est.}}<br />
<br />
* {{transferTest|zho|lat|今天天气好，很暖|Tempestas pulchra est hodie, admodum calda est.}}<br />
*: {{transferMorphTest|zho|lat|今天{{tag|tw}}天气{{tag|n}}好{{tag|adj}}，很{{tag|adv}}暖{{tag|adj}}|Tempestas pulchra est hodie, admodum calda est.}}<br />
<br />
* {{transferTest|zho|lat|小红看着在屋子里面的妈妈。妈妈正在笑|Maria matrem suam post fenestram adspicit, mater sua ridet.}}<br />
*: {{transferMorphTest|zho|lat|小红{{tag|n}}看着{{tag|vblex}}{{tag|pii}}在{{tag|pr}}屋子{{tag|n}}里面{{tag|n}}的{{tag|part}}妈妈{{tag|n}}。妈妈{{tag|n}}正在{{tag|adv}}笑{{tag|vblex}}|Maria matrem suam post fenestram adspicit, mater sua ridet.}}<br />
<br />
* {{transferTest|zho|lat|她看到了一只手， 走近了一点儿，看到了他!|Tum etiam manus parvus videt et cum appropinquat etiam caput suum videt!}}<br />
*: {{transferMorphTest|zho|lat|她{{tag|prn}}{{tag|f}}看到了{{tag|vblex}}{{tag|prf}}一{{tag|num}}只{{tag|clf}}手{{tag|n}}， 走近了{{tag|vblex}}{{tag|prf}}一点儿{{tag|inc}}，看到了{{tag|vblex}}{{tag|prf}}他{{tag|prn}}{{tag|m}}!|Tum etiam manus parvus videt et cum appropinquat etiam caput suum videt!}}<br />
<br />
* {{transferTest|zho|lat|小明有一只狗，那只狗现在在花园里面|Jamo canis parvus est, etiam canis nunc in horto est.}}<br />
*: {{transferMorphTest|zho|lat| 小明{{tag|n}}有{{tag|vhaver}}一{{tag|num}}只{{tag|clf}}狗{{tag|n}}，那{{tag|det}}{{tag|dst}}只{{tag|clf}}狗{{tag|n}}现在{{tag|tw}}在{{tag|pr}}花园{{tag|n}}里面{{tag|n}}|Jamo canis parvus est, etiam canis nunc in horto est.}}<br />
<br />
* {{transferTest|zho|lat|不，她没有狗，她有只猫| Mariaene non canis est, eae catta est.}}<br />
*: {{transferMorphTest|zho|lat|不{{tag|neg}}，她{{tag|prn}}{{tag|f}}没{{tag|neg}}有{{tag|vhaver}}狗{{tag|n}}，她{{tag|prn}}{{tag|f}}有{{tag|vhaver}}只{{tag|clf}}猫{{tag|n}}| Mariaene non canis est, eae catta est.}}<br />
<br />
* {{transferTest|zho|lat|她现在听得很清楚了|Nunc id bene audire potest.}}<br />
*: {{transferMorphTest|zho|lat|她{{tag|prn}}{{tag|f}}现在{{tag|tw}}听{{tag|vblex}}得{{tag|part}}很{{tag|adv}}清楚了{{tag|adj}}{{tag|cos}}|Nunc id bene audire potest.}}<br />
<br />
* {{transferTest|zho|lat|她仔细地听|Auscultat diligenter.}}<br />
*: {{transferMorphTest|zho|lat|她{{tag|prn}}{{tag|f}}仔细{{tag|adj}}地{{tag|part}}听{{tag|vblex}}|Auscultat diligenter.}}<br />
<br />
* {{transferTest|zho|lat|想找小明。他在哪儿呢？|Petet Jamum: quo vadebat?}}<br />
*: {{transferMorphTest|zho|lat|想{{tag|vaux}}找{{tag|vblex}}小明{{tag|n}}。他{{tag|prn}}{{tag|m}}在{{tag|pr}}哪儿{{tag|prn}}{{tag|itg}}呢{{tag|ij}}？|Petet Jamum: quo vadebat?}}<br />
<br />
{{comment|format this as a list:}}<br />
'''Tags not found in the Apertium tagset:'''<br />
* clf: classifier<br />
<br />
* tw: time word<br />
<br />
* part: particle (I know we shouldn't say particle, but I don't know what actual linguists usually analyze them as)<br />
<br />
* prf: perfect<br />
<br />
* inc: incremental ("a little closer")<br />
<br />
* cos: change of state<br />
<br />
<br />
{{comment|you forgot to include the relevant category, which made this page a little harder to find}}<br />
<br />
[[Category:Sp18_ContrastiveGrammars]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese/Contrastive_Grammar&diff=7209Latin and Mandarin Chinese/Contrastive Grammar2018-03-27T02:03:36Z<p>Eschalk1: /* lat-zho-tests */</p>
<hr />
<div>==Differences between Latin and Mandarin Chinese==<br />
<br />
===Use of the verb 'to be'===<br />
<br />
Latin and Mandarin Chinese both have a verb 'to be' but use it in different contexts. Latin uses this verb in more situations than Chinese; for instance, forms of the Latin verb ''esse'' can be used to link a noun to an adjective, a predicate noun, and even helps form certain definite verbs from its present tense forms and a participle (acting as a copula). Chinese uses its form of this verb, XXX, only to link a noun to a predicate noun (with some exceptions); otherwise a verb is not necessary to link nouns to adjectives. <br />
<br />
Example sentences:<br />
<br />
Jamus et Maria in horto sunt. <br />
小明和小红在花园里面。<br />
James and Mary are in the garden.<br />
<br />
Jamus puer parvus est.<br />
小明是一个小男孩。<br />
James is a small boy.<br />
<br />
Tempestas pulchra est hodie, admodum calda est.<br />
今天天气好，很暖。<br />
The weather is nice today, and it is very warm.<br />
<br />
===Word order===<br />
<br />
Latin generally follows a subject-object-verb word order, with exceptions for most poetry, and is a highly inflected language (hence the flexible word order). Chinese has a subject-verb-object word order and is highly uninflected (hence the inflexible word order).<br />
<br />
Maria matrem suam post fenestram adspicit, mater sua ridet.<br />
小红看着在屋子里面的妈妈。妈妈正在笑<br />
Mary sees her mother behind the window, and her mother laughs.<br />
<br />
Tum etiam manus parvus videt et cum appropinquat etiam caput suum videt!<br />
她看到了一只手， 走近了一点儿，看到了他！<br />
Then she saw a little hand and when she came closer even his own head!<br />
<br />
===Possession===<br />
<br />
Latin forms possessive relationships between nouns with a dative form of the possessing noun (or in other cases, a enitive); Chinese forms possessive relationships between nouns with a particle, XXX, that connects the possession to the possessing noun.<br />
<br />
Jamo canis parvus est, etiam canis nunc in horto est.<br />
小明有一只狗，那只狗现在在花园里面。<br />
James has a small dog, the dog is now in the garden.<br />
<br />
Mariaene non canis est, eae catta est.<br />
可是，猫儿在屋子里面睡觉。<br />
Mary does not have a dog, she has a cat.<br />
<br />
===Adverbs===<br />
<br />
While both Latin and Chinese have adverbs that describe the performance of a verb, Chinese can also form its adverbs through the complement: the use of a particle, XXXX, with a following adjective. Latin can form an adverb from an adjective using inflection, which it does with most regular adjectives.<br />
<br />
Nunc id bene audire potest.<br />
她现在听得很清楚了。<br />
Now she is able to hear it clearly.<br />
<br />
Auscultat diligenter.<br />
她仔细地听。<br />
She listens carefully.<br />
<br />
===Case===<br />
<br />
Latin inflects all nouns for case (nominative, genitive, accusative, dative, ablative, and sometimes vocative); Chinese expresses the same concepts with word order or prepositions. For instance, Latin uses the accusative case to express the direct object of a verb. Chinese would do the same thing with word order, placing the direct object after the verb.<br />
<br />
Petet Jamum: quo vadebat?<br />
他在哪儿呢？<br />
She looks for James; where was he hiding?<br />
<br />
Maria matrem suam post fenestram adspicit, mater sua ridet.<br />
小红看着在屋子里面的妈妈。妈妈正在笑。<br />
Mary sees her mother behind the window, her mother laughs.<br />
<br />
Tum etiam manus parvus videt et cum appropinquat etiam caput suum videt!<br />
她看到了一只手， 走近了一点儿，看到了他！<br />
Then she saw a little hand and when she came closer even his own head!<br />
<br />
==lat-zho-tests==<br />
<br />
{{transferMorphTest|lat|zho|Jam{{tag|n}}{{tag|nom}}{{tag|masc}}{{tag|sg}}et{{tag|conj}}Mari{{tag|n}}{{tag|nom}}{{tag|fem}}{{tag|sg}}in{{tag|prep}}hort{{tag|n}}{{tag|abl}}{{tag|masc}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|pl}}{{tag|pres}}{{tag|act}}{{tag|ind}}|小红看着在屋子里面的妈妈。妈妈正在笑。}}<br />
<br />
{{transferMorphTest|lat|zho|Jam{{tag|n}}{{tag|nom}}{{tag|masc}}{{tag|sg}}puer{{tag|n}}{{tag|nom}}{{tag|masc}}{{tag|sg}}parv{{tag|adj}}{{tag|nom}}{{tag|masc}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|小明是一个小男孩。}}<br />
<br />
{{transferMorphTest|lat|zho|tempest{{tag|n}}{{tag|nom}}{{tag|fem}}{{tag|sg}}pulchr{{tag|adj}}{{tag|nom}}{{tag|fem}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}hodie{{tag|adv}}admodum{{tag|adv}}cald{{tag|adj}}{{tag|nom}}{{tag|fem}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|今天天气好，很暖。}}<br />
<br />
{{transferMorphTest|lat|zho|Mari{{tag|n}}{{tag|nom}}{{tag|fem}}{{tag|sg}}matr{{tag|n}}{{tag|acc}}{{tag|fem}}{{tag|sg}}su{{tag|prn}}{{tag|poss}}{{tag|refl}}{{tag|acc}}{{tag|fem}}{{tag|sg}}post{{tag|prep}}fenestr{{tag|n}}{{tag|acc}}{{tag|fem}}{{tag|sg}}adspic{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}matr{{tag|n}}{{tag|nom}}{{tag|fem}}{{tag|sg}}su{{tag|prn}}{{tag|poss}}{{tag|refl}}{{tag|nom}}{{tag|fem}}{{tag|sg}}rid{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|小红看着在屋子里面的妈妈。妈妈正在笑}}<br />
<br />
{{transferMorphTest|lat|zho|tum{{tag|adv}}etiam{{tag|adv}}man{{tag|n}}{{tag|acc}}{{tag|masc}}{{tag|sg}}parv{{tag|n}}{{tag|acc}}{{tag|masc}}{{tag|sg}}vid{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}et{{tag|conj}}cum{{tag|conj}}appropinqu{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|subj}}etiam{{tag|adv}}capit{{tag|n}}{{tag|acc}}{{tag|masc}}{{tag|sg}}su{{tag|prn}}{{tag|poss}}{{tag|refl}}{{tag|acc}}{{tag|masc}}{{tag|sg}}vid{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|她看到了一只手， 走近了一点儿，看到了他！}}<br />
<br />
{{transferMorphTest|lat|zho|Nunc{{tag|adv}}id{{tag|prn}}{{tag|acc}}{{tag|neut}}{{tag|sg}}bene{{tag|adv}}audire{{tag|vblex}}{{tag|pres}}{{tag|act}}{{tag|inf}}possum{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|她现在听得很清楚了。}}<br />
<br />
{{transferMorphTest|lat|zho|auscult{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}diligen{{tag|adv}}|她仔细地听。}}<br />
<br />
{{transferMorphTest|lat|zho|pet{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}Jam{{tag|n}}{{tag|acc}}{{tag|masc}}{{tag|sg}}quo{{tag|prn}}{{tag|interr}}{{tag|abl}}{{tag|masc}}{{tag|sg}}vad{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|impf}}{{tag|act}}{{tag|ind}}|他在哪儿呢？}}<br />
<br />
==zho-lat-tests==</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese/Contrastive_Grammar&diff=7208Latin and Mandarin Chinese/Contrastive Grammar2018-03-27T02:01:12Z<p>Eschalk1: </p>
<hr />
<div>==Differences between Latin and Mandarin Chinese==<br />
<br />
===Use of the verb 'to be'===<br />
<br />
Latin and Mandarin Chinese both have a verb 'to be' but use it in different contexts. Latin uses this verb in more situations than Chinese; for instance, forms of the Latin verb ''esse'' can be used to link a noun to an adjective, a predicate noun, and even helps form certain definite verbs from its present tense forms and a participle (acting as a copula). Chinese uses its form of this verb, XXX, only to link a noun to a predicate noun (with some exceptions); otherwise a verb is not necessary to link nouns to adjectives. <br />
<br />
Example sentences:<br />
<br />
Jamus et Maria in horto sunt. <br />
小明和小红在花园里面。<br />
James and Mary are in the garden.<br />
<br />
Jamus puer parvus est.<br />
小明是一个小男孩。<br />
James is a small boy.<br />
<br />
Tempestas pulchra est hodie, admodum calda est.<br />
今天天气好，很暖。<br />
The weather is nice today, and it is very warm.<br />
<br />
===Word order===<br />
<br />
Latin generally follows a subject-object-verb word order, with exceptions for most poetry, and is a highly inflected language (hence the flexible word order). Chinese has a subject-verb-object word order and is highly uninflected (hence the inflexible word order).<br />
<br />
Maria matrem suam post fenestram adspicit, mater sua ridet.<br />
小红看着在屋子里面的妈妈。妈妈正在笑<br />
Mary sees her mother behind the window, and her mother laughs.<br />
<br />
Tum etiam manus parvus videt et cum appropinquat etiam caput suum videt!<br />
她看到了一只手， 走近了一点儿，看到了他！<br />
Then she saw a little hand and when she came closer even his own head!<br />
<br />
===Possession===<br />
<br />
Latin forms possessive relationships between nouns with a dative form of the possessing noun (or in other cases, a enitive); Chinese forms possessive relationships between nouns with a particle, XXX, that connects the possession to the possessing noun.<br />
<br />
Jamo canis parvus est, etiam canis nunc in horto est.<br />
小明有一只狗，那只狗现在在花园里面。<br />
James has a small dog, the dog is now in the garden.<br />
<br />
Mariaene non canis est, eae catta est.<br />
可是，猫儿在屋子里面睡觉。<br />
Mary does not have a dog, she has a cat.<br />
<br />
===Adverbs===<br />
<br />
While both Latin and Chinese have adverbs that describe the performance of a verb, Chinese can also form its adverbs through the complement: the use of a particle, XXXX, with a following adjective. Latin can form an adverb from an adjective using inflection, which it does with most regular adjectives.<br />
<br />
Nunc id bene audire potest.<br />
她现在听得很清楚了。<br />
Now she is able to hear it clearly.<br />
<br />
Auscultat diligenter.<br />
她仔细地听。<br />
She listens carefully.<br />
<br />
===Case===<br />
<br />
Latin inflects all nouns for case (nominative, genitive, accusative, dative, ablative, and sometimes vocative); Chinese expresses the same concepts with word order or prepositions. For instance, Latin uses the accusative case to express the direct object of a verb. Chinese would do the same thing with word order, placing the direct object after the verb.<br />
<br />
Petet Jamum: quo vadebat?<br />
他在哪儿呢？<br />
She looks for James; where was he hiding?<br />
<br />
Maria matrem suam post fenestram adspicit, mater sua ridet.<br />
小红看着在屋子里面的妈妈。妈妈正在笑。<br />
Mary sees her mother behind the window, her mother laughs.<br />
<br />
Tum etiam manus parvus videt et cum appropinquat etiam caput suum videt!<br />
她看到了一只手， 走近了一点儿，看到了他！<br />
Then she saw a little hand and when she came closer even his own head!<br />
<br />
==lat-zho-tests==<br />
<br />
{{transferMorphTest|lat|zho|Jam{{tag|n}}{{tag|nom}}{{tag|masc}}{{tag|sg}}et{{tag|conj}}Mari{{tag|n}}{{tag|nom}}{{tag|fem}}{{tag|sg}}in{{tag|prep}}hort{{tag|n}}{{tag|abl}}{{tag|masc}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|pl}}{{tag|pres}}{{tag|act}}{{tag|ind}}|小红看着在屋子里面的妈妈。妈妈正在笑。}}<br />
<br />
{{transferMorphTest|lat|zho|Jam{{tag|n}}{{tag|nom}}{{tag|masc}}{{tag|sg}}puer{{tag|n}}{{tag|nom}}{{tag|masc}}{{tag|sg}}parv{{tag|adj}}{{tag|nom}}{{tag|masc}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|小明是一个小男孩。}}<br />
<br />
{{transferMorphTest|lat|zho|tempest{{tag|n}}{{tag|nom}}{{tag|fem}}{{tag|sg}}pulchr{{tag|adj}}{{tag|nom}}{{tag|fem}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}hodie{{tag|adv}}admodum{{tag|adv}}cald{{tag|adj}}{{tag|nom}}{{tag|fem}}{{tag|sg}}esse{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|今天天气好，很暖。}}<br />
<br />
{{transferMorphTest|lat|zho|Mari{{tag|n}}{{tag|nom}}{{tag|fem}}{{tag|sg}}matr{{tag|n}}{{tag|acc}}{{tag|fem}}{{tag|sg}}su{{tag|prn}}{{tag|poss}}{{tag|refl}}{{tag|acc}}{{tag|fem}}{{tag|sg}}post{{tag|prep}}fenestr{{tag|n}}{{tag|acc}}{{tag|fem}}{{tag|sg}}adspic{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}matr{{tag|n}}{{tag|nom}}{{tag|fem}}{{tag|sg}}su{{tag|prn}}{{tag|poss}}{{tag|refl}}{{tag|nom}}{{tag|fem}}{{tag|sg}}rid{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|小红看着在屋子里面的妈妈。妈妈正在笑}}<br />
<br />
{{transferMorphTest|lat|zho|tum{{tag|adv}}etiam{{tag|adv}}man{{tag|n}}{{tag|acc}}{{tag|masc}}{{tag|sg}}parv{{tag|n}}{{tag|acc}}{{tag|masc}}{{tag|sg}}vid{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}et{{tag|conj}}cum{{tag|conj}}appropinqu{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|subj}}etiam{{tag|adv}}capit{{{{tag|n}}{{tag|acc}}{{tag|masc}}{{tag|sg}}su{{tag|prn}}{{tag|poss}}{{tag|refl}}{{tag|acc}}{{tag|masc}}{{tag|sg}}vid{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|她看到了一只手， 走近了一点儿，看到了他！}}<br />
<br />
{{transferMorphTest|lat|zho|Nunc{{tag|adv}}id{{tag|prn}}{{tag|acc}}{{tag|neut}}{{tag|sg}}bene{{tag|adv}}audire{{tag|vblex}}{{tag|pres}}{{tag|act}}{{tag|inf}}possum{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}|她现在听得很清楚了。}}<br />
<br />
{{transferMorphTest|lat|zho|auscult{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}diligen{{tag|adv}}|她仔细地听。}}<br />
<br />
{{transferMorphTest|lat|zho|pet{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|pres}}{{tag|act}}{{tag|ind}}Jam{{tag|n}}{{tag|acc}}{{tag|masc}}{{tag|sg}}quo{{tag|prn}}{{tag|interr}}{{tag|abl}}{{tag|masc}}{{tag|sg}}vad{{tag|vblex}}{{tag|p3}}{{tag|sg}}{{tag|impf}}{{tag|act}}{{tag|ind}}|他在哪儿呢？}}<br />
<br />
==zho-lat-tests==</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese/Contrastive_Grammar&diff=7207Latin and Mandarin Chinese/Contrastive Grammar2018-03-26T22:54:00Z<p>Eschalk1: </p>
<hr />
<div>==Differences between Latin and Mandarin Chinese==<br />
<br />
===Use of the verb 'to be'===<br />
<br />
Latin and Mandarin Chinese both have a verb 'to be' but use it in different contexts. Latin uses this verb in more situations than Chinese; for instance, forms of the Latin verb ''esse'' can be used to link a noun to an adjective, a predicate noun, and even helps form certain definite verbs from its present tense forms and a participle (acting as a copula). Chinese uses its form of this verb, XXX, only to link a noun to a predicate noun (with some exceptions); otherwise a verb is not necessary to link nouns to adjectives. <br />
<br />
Example sentences:<br />
<br />
Jamus et Maria in horto sunt. <br />
小明和小红在花园里面。<br />
James and Mary are in the garden.<br />
<br />
Jamus puer parvus est.<br />
小明是一个小男孩。<br />
James is a small boy.<br />
<br />
Tempestas pulchra est hodie, admodum calda est.<br />
今天天气好，很暖。<br />
The weather is nice today, and it is very warm.<br />
<br />
===Word order===<br />
<br />
Latin generally follows a subject-object-verb word order, with exceptions for most poetry, and is a highly inflected language (hence the flexible word order). Chinese has a subject-verb-object word order and is highly uninflected (hence the inflexible word order).<br />
<br />
Maria matrem suam post fenestram adspicit, mater sua ridet.<br />
小红看着在屋子里面的妈妈。妈妈正在笑<br />
Mary sees her mother behind the window, and her mother laughs.<br />
<br />
Tum etiam manus parvus videt et cum appropinquat etiam caput suum videt!<br />
她看到了一只手， 走近了一点儿，看到了他！<br />
Then she saw a little hand and when she came closer even his own head!<br />
<br />
===Possession===<br />
<br />
Latin forms possessive relationships between nouns with a genitive form of the possessing noun; Chinese forms possessive relationships between nouns with a particle, XXX, that connects the possession to the possessing noun.<br />
<br />
<br />
<br />
===Adverbs===<br />
<br />
While both Latin and Chinese have adverbs that describe the performance of a verb, Chinese can also form its adverbs through the complement: the use of a particle, XXXX, with a following adjective. Latin can form an adverb from an adjective using inflection, which it does with most regular adjectives.<br />
<br />
Nunc id bene audire potest.<br />
她现在听得很清楚了。<br />
Now she is able to hear it clearly.<br />
<br />
Auscultat diligenter.<br />
她仔细地听。<br />
She listens carefully.<br />
<br />
===Case===<br />
<br />
Latin inflects all nouns for case (nominative, genitive, accusative, dative, ablative, and sometimes vocative); Chinese expresses the same concepts with word order or prepositions. For instance, Latin uses the accusative case to express the direct object of a verb. Chinese would do the same thing with word order, placing the direct object after the verb.<br />
<br />
Petet Jamum: quo vadebat?<br />
他在哪儿呢？<br />
She looks for James; where was he hiding?<br />
<br />
Maria matrem suam post fenestram adspicit, mater sua ridet.<br />
小红看着在屋子里面的妈妈。妈妈正在笑。<br />
Mary sees her mother behind the window, her mother laughs.<br />
<br />
Tum etiam manus parvus videt et cum appropinquat etiam caput suum videt!<br />
她看到了一只手， 走近了一点儿，看到了他！<br />
Then she saw a little hand and when she came closer even his own head!<br />
<br />
==lat-zho-tests==<br />
<br />
==zho-lat-tests==</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese/Contrastive_Grammar&diff=7206Latin and Mandarin Chinese/Contrastive Grammar2018-03-26T22:46:50Z<p>Eschalk1: </p>
<hr />
<div>==Differences between Latin and Mandarin Chinese==<br />
<br />
===Use of the verb 'to be'===<br />
<br />
Latin and Mandarin Chinese both have a verb 'to be' but use it in different contexts. Latin uses this verb in more situations than Chinese; for instance, forms of the Latin verb ''esse'' can be used to link a noun to an adjective, a predicate noun, and even helps form certain definite verbs from its present tense forms and a participle (acting as a copula). Chinese uses its form of this verb, XXX, only to link a noun to a predicate noun (with some exceptions); otherwise a verb is not necessary to link nouns to adjectives. <br />
<br />
Example sentences:<br />
<br />
Jamus et Maria in horto sunt. <br />
小明和小红在花园里面。<br />
James and Mary are in the garden.<br />
<br />
Jamus puer parvus est.<br />
小明是一个小男孩。<br />
James is a small boy.<br />
<br />
Tempestas pulchra est hodie, admodum calda est.<br />
今天天气好，很暖。<br />
The weather is nice today, and it is very warm.<br />
<br />
===Word order===<br />
<br />
Latin generally follows a subject-object-verb word order, with exceptions for most poetry, and is a highly inflected language (hence the flexible word order). Chinese has a subject-verb-object word order and is highly uninflected (hence the inflexible word order).<br />
<br />
Maria matrem suam post fenestram adspicit, mater sua ridet.<br />
小红看着在屋子里面的妈妈。妈妈正在笑<br />
Mary sees her mother behind the window, and her mother laughs.<br />
<br />
Tum etiam manus parvus videt et cum appropinquat etiam caput suum videt!<br />
她看到了一只手， 走近了一点儿，看到了他！<br />
Then she saw a little hand and when she came closer even his own head!<br />
<br />
===Possession===<br />
<br />
Latin forms possessive relationships between nouns with a genitive form of the possessing noun; Chinese forms possessive relationships between nouns with a particle, XXX, that connects the possession to the possessing noun.<br />
<br />
<br />
<br />
===Adverbs===<br />
<br />
While both Latin and Chinese have adverbs that describe the performance of a verb, Chinese can also form its adverbs through the complement: the use of a particle, XXXX, with a following adjective. Latin can form an adverb from an adjective using inflection, which it does with most regular adjectives.<br />
<br />
Nunc id bene audire potest.<br />
她现在听得很清楚了。<br />
Now she is able to hear it clearly.<br />
<br />
Auscultat diligenter.<br />
她仔细地听。<br />
She listens carefully.<br />
<br />
==lat-zho-tests==<br />
<br />
==zho-lat-tests==</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Latin_and_Mandarin_Chinese&diff=7201Latin and Mandarin Chinese2018-03-26T21:46:19Z<p>Eschalk1: </p>
<hr />
<div>==Resources for machine translation between [https://wikis.swarthmore.edu/ling073/Latin Latin] and [https://wikis.swarthmore.edu/ling073/Mandarin_Chinese Mandarin Chinese]==<br />
<br />
https://github.swarthmore.edu/cdalton2/ling073-lat-zho-corpus<br />
<br />
<br />
<br />
[[Category:Sp18_TranslationPairs]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Neo-Aramaic/Disambiguation&diff=7176Neo-Aramaic/Disambiguation2018-03-23T04:08:16Z<p>Eschalk1: </p>
<hr />
<div>This is our page for the construction of our Neo-Aramaic disambiguator. It can be found [https://github.swarthmore.edu/cdalton2/ling073-aii/blob/master/apertium-aii.aii.rlx here].<br />
<br />
== Initial Evaluation of Ambiguity ==<br />
<br />
Initially, our transducer did not pick up on any ambiguity. This is probably because we haven't implemented that much of our language's lexicon yet. We decided to add forms to our transducer with the potential for ambiguous analyses, particularly Neo-Aramaic's independent and attributive pronouns:<br />
<br />
ܐܗܵܐ<prn><sg><m><near><nom><ind> ↔ ܐܗܵܐ aha 'this (independent pronoun)'<br />
ܐܗܵܐ<prn><sg><m><near><nom><atr> ↔ ܐܗܵܐ aha 'this (attributive pronoun)'<br />
<br />
We had difficulty finding evidence for these forms in our Sureth dictionary we've come to rely on so much, but found an older dictionary by [https://books.google.com/books/about/A_Dictionary_of_the_Dialects_of_Vernacul.html?id=6YJIxCtHEKcC AJ Mclean] that did include them. <br />
<br />
The difference between the use of the two forms is illustrated below. <br />
<br />
ܐܵܬܹܐ ܐܗܵܐ 'this kills'<br />
ܐܵܬܹܐ ܢܘܼܝܢ ܐܗܵܐ 'this fish kills'<br />
<br />
We implemented this ambiguity across the near, medium, and far forms of these pronouns and for their masculine singular, feminine singular, and plural forms.<br />
<br />
Since we could not find these forms in our existing corpus, in order to test our new analyses and disambiguator, we wrote our own sentences and places them in a new text file in our larger corpus. The text we used, plus translations, is show below:<br />
<br />
ܦܵܬܚܹܢ ܠܹܗ ܐܵܗܵܐ ܒܸܬܵܐ 'I open this house'<br />
ܐܵܗܵܐ ܡܵܕܡܹܚ ܠܹܗ ܟܵܢܬܘܼܢܬܵܐ 'This causes the cat to sleep'<br />
ܐܵܒܒܵܐ ܣܵܒܘܼܢܵܐ ܡܘܼܠܸܕ ܠܹܗ ܒܵܒܵܐ 'This grandfather begat the father'<br />
ܟܵܢܬܘܼܢܬܵܐ ܩܵܛܠ ܠܹܗ ܐܵܝܝܵܐ 'The cat kills this'<br />
ܐܵܝܝܸܐ ܦܬܹܚܠܵܐ ܠܹܗ ܒܝܼܬܵܐ 'That opened the egg'<br />
ܡܵܠܟܵܐ ܦܵܬܚ ܠܹܗ ܐܵܝܝܵܐ 'The king opens this'<br />
<br />
After running tests on this small corpus, our transducer was generating 25 forms with 31 analyses. Our constraints were very basic but got the job done; we wanted to remove independent pronouns when they were followed by a noun phrase (a noun or an adjective), and use attributive pronouns when they were followed by a noun phrase (noun or adjective). These are very basic but make sense with the grammar and syntax of Assyrian Neo-Aramaic, as seen in Geoffrey Khan's grammar.<br />
<br />
To implement this rule in our rlx file, our lists were<br />
<br />
LIST Independent = (prn sg m near ind) (prn pl near ind) (prn sg m med ind) (prn sg f med ind) (prn sg m far ind) (prn sg f far ind) (prn pl far ind) ;<br />
LIST Attributive = (prn sg m near atr) (prn pl near atr) (prn sg m med atr) (prn sg f med atr) (prn sg m far atr) (prn sg f far atr) (prn pl far atr) ;<br />
LIST Noun = (n sg nom) (n pl nom) ;<br />
LIST Adjective = adj ; <br />
<br />
Our constraint rules were <br />
<br />
REMOVE Independent IF (1 Noun OR Adjective) ;<br />
REMOVE Attributive IF (NOT 1 Noun OR Adjective) ;<br />
<br />
==Final evaluation of ambiguity==<br />
<br />
Across our small corpus, we now have no ambiguity.<br />
<br />
[[Category:Sp18_Disambiguation]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Neo-Aramaic/Disambiguation&diff=7175Neo-Aramaic/Disambiguation2018-03-23T03:30:14Z<p>Eschalk1: </p>
<hr />
<div>This is our page for the construction of our Neo-Aramaic disambiguator.<br />
<br />
== Initial Evaluation of Ambiguity ==<br />
<br />
Initially, our transducer did not pick up on any ambiguity. This is probably because we haven't implemented that much of our language's lexicon yet. We decided to add forms to our transducer with the potential for ambiguous analyses, particularly Neo-Aramaic's independent and attributive pronouns:<br />
<br />
ܐܗܵܐ<prn><sg><m><near><nom><ind> ↔ ܐܗܵܐ aha 'this (independent pronoun)'<br />
ܐܗܵܐ<prn><sg><m><near><nom><atr> ↔ ܐܗܵܐ aha 'this (attributive pronoun)'<br />
<br />
The difference between the use of the two forms is illustrated below. <br />
<br />
ܐܵܬܹܐ ܐܗܵܐ 'this kills'<br />
ܐܵܬܹܐ ܢܘܼܝܢ ܐܗܵܐ 'this fish kills'<br />
<br />
ܦܵܬܚܹܢ ܠܹܗ ܐܵܗܵܐ ܒܸܬܵܐ 'I open this house'<br />
ܐܵܗܵܐ ܡܵܕܡܹܚ ܠܹܗ ܟܵܢܬܘܼܢܬܵܐ 'This causes the cat to sleep'<br />
ܐܵܒܒܵܐ ܣܵܒܘܼܢܵܐ ܡܘܼܠܸܕ ܠܹܗ ܒܵܒܵܐ 'This grandfather begat the father'<br />
ܟܵܢܬܘܼܢܬܵܐ ܩܵܛܠ ܠܹܗ ܐܵܝܝܵܐ 'The cat kills this'<br />
ܐܵܝܝܸܐ ܦܬܹܚܠܵܐ ܠܹܗ ܒܝܼܬܵܐ 'That opened the egg'<br />
ܡܵܠܟܵܐ ܦܵܬܚ ܠܹܗ ܐܵܝܝܵܐ 'The king opens this'<br />
<br />
We implemented this ambiguity across the near, medium, and far forms of these pronouns and for their masculine singular, feminine singular, and plural forms.<br />
[[Category:Sp18_Disambiguation]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Neo-Aramaic/Disambiguation&diff=7174Neo-Aramaic/Disambiguation2018-03-23T03:14:02Z<p>Eschalk1: </p>
<hr />
<div>This is our page for the construction of our Neo-Aramaic disambiguator.<br />
<br />
== Initial Evaluation of Ambiguity ==<br />
<br />
Initially, our transducer did not pick up on any ambiguity. This is probably because we haven't implemented that much of our language's lexicon yet. We decided to add forms to our transducer with the potential for ambiguous analyses, particularly Neo-Aramaic's independent and attributive pronouns:<br />
<br />
ܐܗܵܐ<prn><sg><m><near><nom><ind> ↔ ܐܗܵܐ aha 'this (independent pronoun)'<br />
ܐܗܵܐ<prn><sg><m><near><nom><atr> ↔ ܐܗܵܐ aha 'this (attributive pronoun)'<br />
<br />
The difference between the use of the two forms is illustrated below. <br />
<br />
ܐܵܬܹܐ ܐܗܵܐ 'this kills'<br />
ܐܵܬܹܐ ܢܘܼܝܢ ܐܗܵܐ 'this fish kills'<br />
<br />
We implemented this ambiguity across the near, medium, and far forms of these pronouns and for their masculine singular, feminine singular, and plural forms.<br />
[[Category:Sp18_Disambiguation]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Neo-Aramaic/Disambiguation&diff=7134Neo-Aramaic/Disambiguation2018-03-22T15:57:27Z<p>Eschalk1: </p>
<hr />
<div>This is our page for the construction of our Neo-Aramaic disambiguator.<br />
<br />
== Initial Evaluation of Ambiguity ==<br />
<br />
Initially, our transducer did not pick up on any ambiguity. This is probably because we haven't implemented that much of our language's lexicon yet. We decided to add forms to our transducer with the potential for ambiguous analyses, particularly Neo-Aramaic's independent and attributive pronouns:<br />
<br />
<br />
ܐܗܵܐ<prn><sg><m><near><nom><ind> ↔ ܐܗܵܐ aha 'this (independent pronoun)'<br />
ܐܗܵܐ<prn><sg><m><near><nom><atr> ↔ ܐܗܵܐ aha 'this (attributive pronoun)'<br />
<br />
The difference between the use of the two forms is illustrated below. <br />
<br />
ܐܵܬܹܐ ܐܗܵܐ 'this kills'<br />
ܐܵܬܹܐ ܢܘܼܝܢ ܐܗܵܐ 'this fish kills'<br />
<br />
[[Category:Sp18_Disambiguation]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Neo-Aramaic/Transducer&diff=7061Neo-Aramaic/Transducer2018-03-20T15:10:24Z<p>Eschalk1: </p>
<hr />
<div>The code for the transducer can be found in [https://github.swarthmore.edu/cdalton2/ling073-aii this] Github repository.<br />
<br />
==Getting set up: Part One==<br />
<br />
Getting set up for this stage of the project was an ongoing process for us; while it was easy to bootstrap our transducer and start working with Apertium, we found that grammar tags, our <code>Alphabet</code>, and our <code>tests/</code> all needed continuous updating. This was because our grammar rules used tags and characters not found in our defined tags or alphabet, and these changes had to trickle down at some point to <code>tests/</code>, too. We continue to update these to reflect the growing number of tags we use to complete our grammar coverage. We now have tags for verb features, parts of speech, an archiphoneme, person morphology, gender morphology, number morphology, case morphology, TAM morphology, pronoun morphology, and punctuation. <br />
<br />
Our <code>Alphabet</code> includes Syriac characters, Latin characters, a null morpheme (useful for analyzing vowel diacritics), and punctuation.<br />
<br />
Alphabet<br />
ܐ ܒ ܓ ܕ ܗ ܘ ܙ ܚ ܛ ܝ ܟ ܠ ܡ ܢ ܣ ܥ ܦ ܨ ܩ ܪ ܫ ܬ<br />
ܸ ܲ ܵ ܸ ܹ ܼ<br />
◌:0<br />
. ؛ ! - — ، ؟ ' " « » ” “ ( ] [ ) ܀ \ /<br />
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z<br />
a b c d e f g h i j k l m n o p q r s t u v w x y z<br />
<br />
This section also includes our archiphoneme, which is used to realize the vowel sound <code>A</code> as the alaf when it appears at the beginning or end of a word. This is helpful when working with stems where the alaf often disappears after prefixes and suffixes are added to the stem.<br />
<br />
%{A%}:0<br />
! %{A%}:ܐ<br />
<br />
==The hard stuff: Part One==<br />
<br />
When we moved on to the hardest part of putting together a morphological transducer, our biggest challenge was working with the many prefixes and infixes found in Neo-Aramaic. We focused in the beginning on easy forms, like plural nouns, that used mostly suffixes in their morphology. When we moved on to our prefix morphology, we used the solution suggested on the course wiki, where our twoc file only processes forms with a plus and minus version of any given feature: <br />
<br />
Rules<br />
"Remove paths without matching suffix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
_ :* Fy:0 ;<br />
where Fx in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fy in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
"Remove paths without matching prefix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
Fy:0 :* _ ;<br />
where Fy in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fx in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
<br />
This guarantees that if a verb picks up a <code>-hab</code> or <code>-fut</code> tag, it will only analyze these forms as having a habitual or future prefix if they pick up a later <code>+hab</code> or <code>+fut</code> tag.<br />
<br />
== Evaluation ==<br />
<br />
We have 2556 tokenised words in our corpus, with a coverage of 9.66%. We have about 37 stems in our transducer (definitely a place we should be concentrating work right now). <br />
<br />
Top unknown words in the corpus:<br />
<br />
66 ܠܹܗ<br />
<br />
62 ܕ<br />
<br />
54 ܘ<br />
<br />
52 ܐ<br />
<br />
50 ܚܕ<br />
<br />
49 ܡ<br />
<br />
44 ܒ<br />
<br />
40 ܪ<br />
<br />
39 ܡܘܼܠܸܕ<br />
<br />
37 ܠ<br />
<br />
35 ܢ<br />
<br />
31 ܡܢ<br />
<br />
28 ܟܠ<br />
<br />
28 ܝܢ<br />
<br />
28 ܝ<br />
<br />
27 ܗܩܘܬܐ<br />
<br />
25 ܠܐ<br />
<br />
22 ܐܝܬܠܗ<br />
<br />
20 ܗ<br />
<br />
19 ܵܐ<br />
<br />
<br />
== Notes ==<br />
<br />
The first unknown word we're going to analyze is ܡܘܼܠܸܕ, which is the ninth most common unknown word and appears 39 times in the corpus. This one is pretty easy because we were able to find it in the Sureth dictionary, and it translates as "to breed, to beget, to conceive." It's not surprising that this one appears so often in our corpus, since the bible is so concerned with lineage. <br />
<br />
ܡܘܼܠܸܕ:ܡܘܼܠܸܕ PresentVerbInfl; ! molidh "to begat, to conceive"<br />
<br />
Adding this form to our transducer bumped our coverage up to 12.03%!<br />
<br />
However, besides this word, we started noticing that a LOT of these uncovered words are single letters; they don't seem to be actual words per se. But we couldn't figure out why so many letters would be sprinkled in our corpus, except as an error. It turns out these are a copy-paste error, as seen in this sentence taking from a pdf piece of our corpus:<br />
<br />
[[File:noticescreenshot.png]]<br />
<br />
This sentence came over to our aii-basic-corpus.txt file looking like this:<br />
<br />
ܡܬܪܓܡܹܢܐ ܡܗܼܝܹܪ ܸܒܠܫܢ ܼ ܕܪܘܫ ܹܡ ܿ ܵ ܐ ܐ ܵ ܵ o ܐ<br />
<br />
We knew we would need to find a more powerful way to get text out of our pdf sources, and after some research decided to find an [https://en.wikipedia.org/wiki/Optical_character_recognition OCR] with Syriac support. Unfortunately, our first thought, Adobe Acrobat Pro, had a tiny selection, and even ABBYY FineReader didn't include Syriac in its list of 192 recognized languages. Fortunately, our friend George Kiraz (a local Syriac scholar) came through and recommended Tesseract, an open-source OCR developed by Google!!! Our .txt file outputted by Tesseract looks a lot more like the original:<br />
<br />
ܡܬܪܓܡܠܝܐ ܡܗܝܠܐ ܒܠܦܫܐ ܕܪܢܕܫܠܪܩ<br />
<br />
However, after looking more carefully at our corpus, we decided to remove most of our pdf source because it was too messy for apertium to work with; it was full of problematic characters like Latin script and even a few instances of � that were causing errors when we tried to later calculate ambiguity. We replaced these lines with a translation of [https://unicode.org/udhr/d/udhr_aii.html the universal declaration of human rights]. <br />
<br />
Adding the declaration significantly changed the list of unknown words and removed a lot of those single letters we were worried about; the next word we decided to investigate was ܟܠ, which was our twelfth most common unknown word and appeared 28 times in our corpus. This was also easy to find in the Sureth dictionary and is a simple adjective that means "any, all." It's used often in the Declaration of Human Rights since each statement in this document begins with ܟܠ ܚܕ, which translates as "any person."<br />
<br />
ܟܠ:ܟܠ AdjectiveInfl ; ! had 'any, all'<br />
<br />
Since our coverage went down after adding the Universal Declaration of Human Rights, we went from 9.67% to 10.75% by adding this word!<br />
<br />
We next added ܚܕ, which was our fifth most common unknown word with 50 appearances in the corpus. The Sureth dictionary showed that this word acts as a sort of indeterminate noun, meaning "a, an, one." As previously stated, it often appears in our corpus in the Declaration of Human Rights in conjunction with ܟܠ to mean "anyone."<br />
<br />
ܚܕ:ܚܕ RegNounInfl ; ! 'had- "a, an, one"<br />
<br />
<br />
All together, adding these words to our transducer brought our coverage up to 12.72%.<br />
<br />
[[Category:Sp18_Transducers]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Neo-Aramaic/Transducer&diff=7049Neo-Aramaic/Transducer2018-03-18T17:46:32Z<p>Eschalk1: /* Notes */</p>
<hr />
<div>The code for the transducer can be found in [https://github.swarthmore.edu/cdalton2/ling073-aii this] Github repository.<br />
<br />
==Getting set up: Part One==<br />
<br />
Getting set up for this stage of the project was an ongoing process for us; while it was easy to bootstrap our transducer and start working with Apertium, we found that grammar tags, our <code>Alphabet</code>, and our <code>tests/</code> all needed continuous updating. This was because our grammar rules used tags and characters not found in our defined tags or alphabet, and these changes had to trickle down at some point to <code>tests/</code>, too. We continue to update these to reflect the growing number of tags we use to complete our grammar coverage. We now have tags for verb features, parts of speech, an archiphoneme, person morphology, gender morphology, number morphology, case morphology, TAM morphology, pronoun morphology, and punctuation. <br />
<br />
Our <code>Alphabet</code> includes Syriac characters, Latin characters, a null morpheme (useful for analyzing vowel diacritics), and punctuation.<br />
<br />
Alphabet<br />
ܐ ܒ ܓ ܕ ܗ ܘ ܙ ܚ ܛ ܝ ܟ ܠ ܡ ܢ ܣ ܥ ܦ ܨ ܩ ܪ ܫ ܬ<br />
ܸ ܲ ܵ ܸ ܹ ܼ<br />
◌:0<br />
. ؛ ! - — ، ؟ ' " « » ” “ ( ] [ ) ܀ \ /<br />
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z<br />
a b c d e f g h i j k l m n o p q r s t u v w x y z<br />
<br />
This section also includes our archiphoneme, which is used to realize the vowel sound <code>A</code> as the alaf when it appears at the beginning or end of a word. This is helpful when working with stems where the alaf often disappears after prefixes and suffixes are added to the stem.<br />
<br />
%{A%}:0<br />
! %{A%}:ܐ<br />
<br />
==The hard stuff: Part One==<br />
<br />
When we moved on to the hardest part of putting together a morphological transducer, our biggest challenge was working with the many prefixes and infixes found in Neo-Aramaic. We focused in the beginning on easy forms, like plural nouns, that used mostly suffixes in their morphology. When we moved on to our prefix morphology, we used the solution suggested on the course wiki, where our twoc file only processes forms with a plus and minus version of any given feature: <br />
<br />
Rules<br />
"Remove paths without matching suffix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
_ :* Fy:0 ;<br />
where Fx in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fy in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
"Remove paths without matching prefix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
Fy:0 :* _ ;<br />
where Fy in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fx in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
<br />
This guarantees that if a verb picks up a <code>-hab</code> or <code>-fut</code> tag, it will only analyze these forms as having a habitual or future prefix if they pick up a later <code>+hab</code> or <code>+fut</code> tag.<br />
<br />
== Evaluation ==<br />
<br />
We have 2556 tokenised words in our corpus, with a coverage of 9.66%. We have about 37 stems in our transducer (definitely a place we should be concentrating work right now). <br />
<br />
Top unknown words in the corpus:<br />
<br />
114 ܐ<br />
<br />
86 ܕ<br />
<br />
66 ܠܹܗ<br />
<br />
62 ܡ<br />
<br />
54 ܒ<br />
<br />
46 ܢ<br />
<br />
44 ܘ<br />
<br />
44 ܠ<br />
<br />
39 ܡܘܼܠܸܕ<br />
<br />
36 ܝ<br />
<br />
31 ܪ<br />
<br />
30 ܡܢ<br />
<br />
23 ܵܐ<br />
<br />
21 ܗ<br />
<br />
20 ܹܐ<br />
<br />
18 ܠܐ<br />
<br />
17 ܫ<br />
<br />
14 ܵ<br />
<br />
13 ܚ<br />
<br />
12 ܡܕܝܢܬܐ<br />
<br />
== Notes ==<br />
<br />
The first unknown word we're going to analyze is ܡܘܼܠܸܕ, which is the ninth most common unknown word and appears 39 times in the corpus. This one is pretty easy because we were able to find it in the Sureth dictionary, and it translates as "to breed, to beget, to conceive." It's not surprising that this one appears so often in our corpus, since the bible is so concerned with lineage. <br />
<br />
ܡܘܼܠܸܕ:ܡܘܼܠܸܕ PresentVerbInfl; ! molidh "to begat, to conceive"<br />
<br />
Adding this form to our transducer bumped our coverage up to 12.03%!<br />
<br />
However, besides this word, we started noticing that a LOT of these uncovered words are single letters; they don't seem to be actual words per se. But we couldn't figure out why so many letters would be sprinkled in our corpus, except as an error. It turns out these are a copy-paste error, as seen in this sentence taking from a pdf piece of our corpus:<br />
<br />
[[File:noticescreenshot.png]]<br />
<br />
This sentence came over to our aii-basic-corpus.txt file looking like this:<br />
<br />
ܡܬܪܓܡܹܢܐ ܡܗܼܝܹܪ ܸܒܠܫܢ ܼ ܕܪܘܫ ܹܡ ܿ ܵ ܐ ܐ ܵ ܵ o ܐ<br />
<br />
We knew we would need to find a more powerful way to get text out of our pdf sources, and after some research decided to find an [https://en.wikipedia.org/wiki/Optical_character_recognition OCR] with Syriac support. Unfortunately, our first thought, Adobe Acrobat Pro, had a tiny selection, and even ABBYY FineReader didn't include Syriac in its list of 192 recognized languages. Fortunately, our friend George Kiraz (a local Syriac scholar) came through and recommended Tesseract, an open-source OCR developed by Google!!! Our .txt file outputted by Tesseract looks a lot more like the original:<br />
<br />
ܡܬܪܓܡܠܝܐ ܡܗܝܠܐ ܒܠܦܫܐ ܕܪܢܕܫܠܪܩ<br />
<br />
However, after looking more carefully at our corpus, we decided to remove most of our pdf source because it was too messy for apertium to work with; it was full of problematic characters like Latin script and even a few instances of � that were causing errors when we tried to later calculate ambiguity. We replaced these lines with a translation of [https://unicode.org/udhr/d/udhr_aii.html the universal declaration of human rights]. <br />
<br />
Adding the declaration significantly changed the list of unknown words and removed a lot of those single letters we were worried about; the next word we decided to investigate was ܟܠ, which was our twelfth most common unknown word and appeared 28 times in our corpus. This was also easy to find in the Sureth dictionary and is a simple adjective that means "any, all." It's used often in the Declaration of Human Rights since each statement in this document begins with ܟܠ ܚܕ, which translates as "any person."<br />
<br />
ܟܠ:ܟܠ AdjectiveInfl ; ! had 'any, all'<br />
<br />
Since our coverage went down after adding the Universal Declaration of Human Rights, we went from 9.67% to 10.75% by adding this word!<br />
<br />
53 of 83 tests pass now. Yay!<br />
<br />
[[Category:Sp18_Transducers]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Neo-Aramaic/Transducer&diff=7048Neo-Aramaic/Transducer2018-03-18T17:02:06Z<p>Eschalk1: /* Notes */</p>
<hr />
<div>The code for the transducer can be found in [https://github.swarthmore.edu/cdalton2/ling073-aii this] Github repository.<br />
<br />
==Getting set up: Part One==<br />
<br />
Getting set up for this stage of the project was an ongoing process for us; while it was easy to bootstrap our transducer and start working with Apertium, we found that grammar tags, our <code>Alphabet</code>, and our <code>tests/</code> all needed continuous updating. This was because our grammar rules used tags and characters not found in our defined tags or alphabet, and these changes had to trickle down at some point to <code>tests/</code>, too. We continue to update these to reflect the growing number of tags we use to complete our grammar coverage. We now have tags for verb features, parts of speech, an archiphoneme, person morphology, gender morphology, number morphology, case morphology, TAM morphology, pronoun morphology, and punctuation. <br />
<br />
Our <code>Alphabet</code> includes Syriac characters, Latin characters, a null morpheme (useful for analyzing vowel diacritics), and punctuation.<br />
<br />
Alphabet<br />
ܐ ܒ ܓ ܕ ܗ ܘ ܙ ܚ ܛ ܝ ܟ ܠ ܡ ܢ ܣ ܥ ܦ ܨ ܩ ܪ ܫ ܬ<br />
ܸ ܲ ܵ ܸ ܹ ܼ<br />
◌:0<br />
. ؛ ! - — ، ؟ ' " « » ” “ ( ] [ ) ܀ \ /<br />
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z<br />
a b c d e f g h i j k l m n o p q r s t u v w x y z<br />
<br />
This section also includes our archiphoneme, which is used to realize the vowel sound <code>A</code> as the alaf when it appears at the beginning or end of a word. This is helpful when working with stems where the alaf often disappears after prefixes and suffixes are added to the stem.<br />
<br />
%{A%}:0<br />
! %{A%}:ܐ<br />
<br />
==The hard stuff: Part One==<br />
<br />
When we moved on to the hardest part of putting together a morphological transducer, our biggest challenge was working with the many prefixes and infixes found in Neo-Aramaic. We focused in the beginning on easy forms, like plural nouns, that used mostly suffixes in their morphology. When we moved on to our prefix morphology, we used the solution suggested on the course wiki, where our twoc file only processes forms with a plus and minus version of any given feature: <br />
<br />
Rules<br />
"Remove paths without matching suffix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
_ :* Fy:0 ;<br />
where Fx in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fy in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
"Remove paths without matching prefix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
Fy:0 :* _ ;<br />
where Fy in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fx in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
<br />
This guarantees that if a verb picks up a <code>-hab</code> or <code>-fut</code> tag, it will only analyze these forms as having a habitual or future prefix if they pick up a later <code>+hab</code> or <code>+fut</code> tag.<br />
<br />
== Evaluation ==<br />
<br />
We have 2556 tokenised words in our corpus, with a coverage of 9.66%. We have about 37 stems in our transducer (definitely a place we should be concentrating work right now). <br />
<br />
Top unknown words in the corpus:<br />
<br />
114 ܐ<br />
<br />
86 ܕ<br />
<br />
66 ܠܹܗ<br />
<br />
62 ܡ<br />
<br />
54 ܒ<br />
<br />
46 ܢ<br />
<br />
44 ܘ<br />
<br />
44 ܠ<br />
<br />
39 ܡܘܼܠܸܕ<br />
<br />
36 ܝ<br />
<br />
31 ܪ<br />
<br />
30 ܡܢ<br />
<br />
23 ܵܐ<br />
<br />
21 ܗ<br />
<br />
20 ܹܐ<br />
<br />
18 ܠܐ<br />
<br />
17 ܫ<br />
<br />
14 ܵ<br />
<br />
13 ܚ<br />
<br />
12 ܡܕܝܢܬܐ<br />
<br />
== Notes ==<br />
<br />
The first unknown word we're going to analyze is ܡܘܼܠܸܕ, which is the ninth most common unknown word and appears 39 times in the corpus. This one is pretty easy because we were able to find it in the Sureth dictionary, and it translates as "to breed, to beget, to conceive." It's not surprising that this one appears so often in our corpus, since the bible is so concerned with lineage. <br />
<br />
ܡܘܼܠܸܕ:ܡܘܼܠܸܕ PresentVerbInfl; ! molidh "to begat, to conceive"<br />
<br />
Adding this form to our transducer bumped our coverage up to 12.03%!<br />
<br />
However, besides this word, we started noticing that a LOT of these uncovered words are single letters; they don't seem to be actual words per se. But we couldn't figure out why so many letters would be sprinkled in our corpus, except as an error. It turns out these are a copy-paste error, as seen in this sentence taking from a pdf piece of our corpus:<br />
<br />
[[File:noticescreenshot.png]]<br />
<br />
This sentence came over to our aii-basic-corpus.txt file looking like this:<br />
<br />
ܡܬܪܓܡܹܢܐ ܡܗܼܝܹܪ ܸܒܠܫܢ ܼ ܕܪܘܫ ܹܡ ܿ ܵ ܐ ܐ ܵ ܵ o ܐ<br />
<br />
We knew we would need to find a more powerful way to get text out of our pdf sources, and after some research decided to find an [https://en.wikipedia.org/wiki/Optical_character_recognition OCR] with Syriac support. Unfortunately, our first thought, Adobe Acrobat Pro, had a tiny selection, and even ABBYY FineReader didn't include Syriac in its list of 192 recognized languages. Fortunately, our friend George Kiraz (a local Syriac scholar) came through and recommended Tesseract, an open-source OCR developed by Google!!! Our .txt file outputted by Tesseract looks a lot more like the original:<br />
<br />
ܡܬܪܓܡܠܝܐ ܡܗܝܠܐ ܒܠܦܫܐ ܕܪܢܕܫܠܪܩ<br />
<br />
However, after looking more carefully at our corpus, we decided to remove most of our pdf source because it was too messy for apertium to work with; it was full of problematic characters like Latin script and even a few instances of � that were causing errors when we tried to later calculate ambiguity. We replaced these lines with a translation of [https://unicode.org/udhr/d/udhr_aii.html the universal declaration of human rights]. <br />
<br />
53 of 83 tests pass now. Yay!<br />
<br />
[[Category:Sp18_Transducers]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Neo-Aramaic/Transducer&diff=7047Neo-Aramaic/Transducer2018-03-18T15:43:36Z<p>Eschalk1: </p>
<hr />
<div>The code for the transducer can be found in [https://github.swarthmore.edu/cdalton2/ling073-aii this] Github repository.<br />
<br />
==Getting set up: Part One==<br />
<br />
Getting set up for this stage of the project was an ongoing process for us; while it was easy to bootstrap our transducer and start working with Apertium, we found that grammar tags, our <code>Alphabet</code>, and our <code>tests/</code> all needed continuous updating. This was because our grammar rules used tags and characters not found in our defined tags or alphabet, and these changes had to trickle down at some point to <code>tests/</code>, too. We continue to update these to reflect the growing number of tags we use to complete our grammar coverage. We now have tags for verb features, parts of speech, an archiphoneme, person morphology, gender morphology, number morphology, case morphology, TAM morphology, pronoun morphology, and punctuation. <br />
<br />
Our <code>Alphabet</code> includes Syriac characters, Latin characters, a null morpheme (useful for analyzing vowel diacritics), and punctuation.<br />
<br />
Alphabet<br />
ܐ ܒ ܓ ܕ ܗ ܘ ܙ ܚ ܛ ܝ ܟ ܠ ܡ ܢ ܣ ܥ ܦ ܨ ܩ ܪ ܫ ܬ<br />
ܸ ܲ ܵ ܸ ܹ ܼ<br />
◌:0<br />
. ؛ ! - — ، ؟ ' " « » ” “ ( ] [ ) ܀ \ /<br />
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z<br />
a b c d e f g h i j k l m n o p q r s t u v w x y z<br />
<br />
This section also includes our archiphoneme, which is used to realize the vowel sound <code>A</code> as the alaf when it appears at the beginning or end of a word. This is helpful when working with stems where the alaf often disappears after prefixes and suffixes are added to the stem.<br />
<br />
%{A%}:0<br />
! %{A%}:ܐ<br />
<br />
==The hard stuff: Part One==<br />
<br />
When we moved on to the hardest part of putting together a morphological transducer, our biggest challenge was working with the many prefixes and infixes found in Neo-Aramaic. We focused in the beginning on easy forms, like plural nouns, that used mostly suffixes in their morphology. When we moved on to our prefix morphology, we used the solution suggested on the course wiki, where our twoc file only processes forms with a plus and minus version of any given feature: <br />
<br />
Rules<br />
"Remove paths without matching suffix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
_ :* Fy:0 ;<br />
where Fx in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fy in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
"Remove paths without matching prefix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
Fy:0 :* _ ;<br />
where Fy in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fx in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
<br />
This guarantees that if a verb picks up a <code>-hab</code> or <code>-fut</code> tag, it will only analyze these forms as having a habitual or future prefix if they pick up a later <code>+hab</code> or <code>+fut</code> tag.<br />
<br />
== Evaluation ==<br />
<br />
We have 2556 tokenised words in our corpus, with a coverage of 9.66%. We have about 37 stems in our transducer (definitely a place we should be concentrating work right now). <br />
<br />
Top unknown words in the corpus:<br />
<br />
114 ܐ<br />
<br />
86 ܕ<br />
<br />
66 ܠܹܗ<br />
<br />
62 ܡ<br />
<br />
54 ܒ<br />
<br />
46 ܢ<br />
<br />
44 ܘ<br />
<br />
44 ܠ<br />
<br />
39 ܡܘܼܠܸܕ<br />
<br />
36 ܝ<br />
<br />
31 ܪ<br />
<br />
30 ܡܢ<br />
<br />
23 ܵܐ<br />
<br />
21 ܗ<br />
<br />
20 ܹܐ<br />
<br />
18 ܠܐ<br />
<br />
17 ܫ<br />
<br />
14 ܵ<br />
<br />
13 ܚ<br />
<br />
12 ܡܕܝܢܬܐ<br />
<br />
== Notes ==<br />
<br />
The first unknown word we're going to analyze is ܡܘܼܠܸܕ, which is the ninth most common unknown word and appears 39 times in the corpus. This one is pretty easy because we were able to find it in the Sureth dictionary, and it translates as "to breed, to beget, to conceive." It's not surprising that this one appears so often in our corpus, since the bible is so concerned with lineage. <br />
<br />
ܡܘܼܠܸܕ:ܡܘܼܠܸܕ PresentVerbInfl; ! molidh "to begat, to conceive"<br />
<br />
Adding this form to our transducer bumped our coverage up to 12.03%!<br />
<br />
However, besides this word, we started noticing that a LOT of these uncovered words are single letters; they don't seem to be actual words per se. But we couldn't figure out why so many letters would be sprinkled in our corpus, except as an error. It turns out these are a copy-paste error, as seen in this sentence taking from a pdf piece of our corpus:<br />
<br />
[[File:noticescreenshot.png]]<br />
<br />
This sentence came over to our aii-basic-corpus.txt file looking like this:<br />
<br />
ܡܬܪܓܡܹܢܐ ܡܗܼܝܹܪ ܸܒܠܫܢ ܼ ܕܪܘܫ ܹܡ ܿ ܵ ܐ ܐ ܵ ܵ o ܐ<br />
<br />
We knew we would need to find a more powerful way to get text out of our pdf sources, and after some research decided to find an [https://en.wikipedia.org/wiki/Optical_character_recognition OCR] with Syriac support. Unfortunately, our first thought, Adobe Acrobat Pro, had a tiny selection, and even ABBYY FineReader didn't include Syriac in its list of 192 recognized languages. Fortunately, our friend George Kiraz (a local Syriac scholar) came through and recommended Tesseract, an open-source OCR developed by Google!!! Our .txt file outputted by Tesseract looks a lot more like the original:<br />
<br />
ܡܬܪܓܡܠܝܐ ܡܗܝܠܐ ܒܠܦܫܐ ܕܪܢܕܫܠܪܩ<br />
<br />
53 of 83 tests pass now. Yay!<br />
<br />
[[Category:Sp18_Transducers]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Neo-Aramaic/Transducer&diff=7046Neo-Aramaic/Transducer2018-03-17T17:38:46Z<p>Eschalk1: /* Notes */</p>
<hr />
<div>The code for the transducer can be found in [https://github.swarthmore.edu/cdalton2/ling073-aii this] Github repository.<br />
<br />
==Getting set up: Part One==<br />
<br />
Getting set up for this stage of the project was an ongoing process for us; while it was easy to bootstrap our transducer and start working with Apertium, we found that grammar tags, our <code>Alphabet</code>, and our <code>tests/</code> all needed continuous updating. This was because our grammar rules used tags and characters not found in our defined tags or alphabet, and these changes had to trickle down at some point to <code>tests/</code>, too. We continue to update these to reflect the growing number of tags we use to complete our grammar coverage. We now have tags for verb features, parts of speech, an archiphoneme, person morphology, gender morphology, number morphology, case morphology, TAM morphology, pronoun morphology, and punctuation. <br />
<br />
Our <code>Alphabet</code> includes Syriac characters, Latin characters, a null morpheme (useful for analyzing vowel diacritics), and punctuation.<br />
<br />
Alphabet<br />
ܐ ܒ ܓ ܕ ܗ ܘ ܙ ܚ ܛ ܝ ܟ ܠ ܡ ܢ ܣ ܥ ܦ ܨ ܩ ܪ ܫ ܬ<br />
ܸ ܲ ܵ ܸ ܹ ܼ<br />
◌:0<br />
. ؛ ! - — ، ؟ ' " « » ” “ ( ] [ ) ܀ \ /<br />
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z<br />
a b c d e f g h i j k l m n o p q r s t u v w x y z<br />
<br />
This section also includes our archiphoneme, which is used to realize the vowel sound <code>A</code> as the alaf when it appears at the beginning or end of a word. This is helpful when working with stems where the alaf often disappears after prefixes and suffixes are added to the stem.<br />
<br />
%{A%}:0<br />
! %{A%}:ܐ<br />
<br />
==The hard stuff: Part One==<br />
<br />
When we moved on to the hardest part of putting together a morphological transducer, our biggest challenge was working with the many prefixes and infixes found in Neo-Aramaic. We focused in the beginning on easy forms, like plural nouns, that used mostly suffixes in their morphology. When we moved on to our prefix morphology, we used the solution suggested on the course wiki, where our twoc file only processes forms with a plus and minus version of any given feature: <br />
<br />
Rules<br />
"Remove paths without matching suffix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
_ :* Fy:0 ;<br />
where Fx in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fy in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
"Remove paths without matching prefix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
Fy:0 :* _ ;<br />
where Fy in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fx in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
<br />
This guarantees that if a verb picks up a <code>-hab</code> or <code>-fut</code> tag, it will only analyze these forms as having a habitual or future prefix if they pick up a later <code>+hab</code> or <code>+fut</code> tag.<br />
<br />
== Evaluation ==<br />
<br />
We have 2556 tokenised words in our corpus, with a coverage of 9.66%. We have about 37 stems in our transducer (definitely a place we should be concentrating work right now). <br />
<br />
Top unknown words in the corpus:<br />
<br />
114 ܐ<br />
<br />
86 ܕ<br />
<br />
66 ܠܹܗ<br />
<br />
62 ܡ<br />
<br />
54 ܒ<br />
<br />
46 ܢ<br />
<br />
44 ܘ<br />
<br />
44 ܠ<br />
<br />
39 ܡܘܼܠܸܕ<br />
<br />
36 ܝ<br />
<br />
31 ܪ<br />
<br />
30 ܡܢ<br />
<br />
23 ܵܐ<br />
<br />
21 ܗ<br />
<br />
20 ܹܐ<br />
<br />
18 ܠܐ<br />
<br />
17 ܫ<br />
<br />
14 ܵ<br />
<br />
13 ܚ<br />
<br />
12 ܡܕܝܢܬܐ<br />
<br />
== Notes ==<br />
<br />
The first unknown word we're going to analyze is ܡܘܼܠܸܕ, which is the ninth most common unknown word and appears 39 times in the corpus. This one is pretty easy because we were able to find it in the Sureth dictionary, and it translates as "to breed, to beget, to conceive." It's not surprising that this one appears so often in our corpus, since the bible is so concerned with lineage. <br />
<br />
ܡܘܼܠܸܕ:ܡܘܼܠܸܕ PresentVerbInfl; ! molidh "to begat, to conceive"<br />
<br />
Adding this form to our transducer bumped our coverage up to 12.03%!<br />
<br />
However, besides this word, we started noticing that a LOT of these uncovered words are single letters; they don't seem to be actual words per se. But we couldn't figure out why so many letters would be sprinkled in our corpus, except as an error. It turns out these are a copy-paste error, as seen in this sentence taking from a pdf piece of our corpus:<br />
<br />
[[File:noticescreenshot.png]]<br />
<br />
This sentence came over to our aii-basic-corpus.txt file looking like this:<br />
<br />
ܡܬܪܓܡܹܢܐ ܡܗܼܝܹܪ ܸܒܠܫܢ ܼ ܕܪܘܫ ܹܡ ܿ ܵ ܐ ܐ ܵ ܵ o ܐ<br />
<br />
53 of 83 tests pass now. Yay!<br />
<br />
[[Category:Sp18_Transducers]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Neo-Aramaic/Transducer&diff=7045Neo-Aramaic/Transducer2018-03-17T17:38:19Z<p>Eschalk1: /* Notes */</p>
<hr />
<div>The code for the transducer can be found in [https://github.swarthmore.edu/cdalton2/ling073-aii this] Github repository.<br />
<br />
==Getting set up: Part One==<br />
<br />
Getting set up for this stage of the project was an ongoing process for us; while it was easy to bootstrap our transducer and start working with Apertium, we found that grammar tags, our <code>Alphabet</code>, and our <code>tests/</code> all needed continuous updating. This was because our grammar rules used tags and characters not found in our defined tags or alphabet, and these changes had to trickle down at some point to <code>tests/</code>, too. We continue to update these to reflect the growing number of tags we use to complete our grammar coverage. We now have tags for verb features, parts of speech, an archiphoneme, person morphology, gender morphology, number morphology, case morphology, TAM morphology, pronoun morphology, and punctuation. <br />
<br />
Our <code>Alphabet</code> includes Syriac characters, Latin characters, a null morpheme (useful for analyzing vowel diacritics), and punctuation.<br />
<br />
Alphabet<br />
ܐ ܒ ܓ ܕ ܗ ܘ ܙ ܚ ܛ ܝ ܟ ܠ ܡ ܢ ܣ ܥ ܦ ܨ ܩ ܪ ܫ ܬ<br />
ܸ ܲ ܵ ܸ ܹ ܼ<br />
◌:0<br />
. ؛ ! - — ، ؟ ' " « » ” “ ( ] [ ) ܀ \ /<br />
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z<br />
a b c d e f g h i j k l m n o p q r s t u v w x y z<br />
<br />
This section also includes our archiphoneme, which is used to realize the vowel sound <code>A</code> as the alaf when it appears at the beginning or end of a word. This is helpful when working with stems where the alaf often disappears after prefixes and suffixes are added to the stem.<br />
<br />
%{A%}:0<br />
! %{A%}:ܐ<br />
<br />
==The hard stuff: Part One==<br />
<br />
When we moved on to the hardest part of putting together a morphological transducer, our biggest challenge was working with the many prefixes and infixes found in Neo-Aramaic. We focused in the beginning on easy forms, like plural nouns, that used mostly suffixes in their morphology. When we moved on to our prefix morphology, we used the solution suggested on the course wiki, where our twoc file only processes forms with a plus and minus version of any given feature: <br />
<br />
Rules<br />
"Remove paths without matching suffix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
_ :* Fy:0 ;<br />
where Fx in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fy in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
"Remove paths without matching prefix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
Fy:0 :* _ ;<br />
where Fy in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fx in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
<br />
This guarantees that if a verb picks up a <code>-hab</code> or <code>-fut</code> tag, it will only analyze these forms as having a habitual or future prefix if they pick up a later <code>+hab</code> or <code>+fut</code> tag.<br />
<br />
== Evaluation ==<br />
<br />
We have 2556 tokenised words in our corpus, with a coverage of 9.66%. We have about 37 stems in our transducer (definitely a place we should be concentrating work right now). <br />
<br />
Top unknown words in the corpus:<br />
<br />
114 ܐ<br />
<br />
86 ܕ<br />
<br />
66 ܠܹܗ<br />
<br />
62 ܡ<br />
<br />
54 ܒ<br />
<br />
46 ܢ<br />
<br />
44 ܘ<br />
<br />
44 ܠ<br />
<br />
39 ܡܘܼܠܸܕ<br />
<br />
36 ܝ<br />
<br />
31 ܪ<br />
<br />
30 ܡܢ<br />
<br />
23 ܵܐ<br />
<br />
21 ܗ<br />
<br />
20 ܹܐ<br />
<br />
18 ܠܐ<br />
<br />
17 ܫ<br />
<br />
14 ܵ<br />
<br />
13 ܚ<br />
<br />
12 ܡܕܝܢܬܐ<br />
<br />
== Notes ==<br />
<br />
The first unknown word we're going to analyze is ܡܘܼܠܸܕ, which is the ninth most common unknown word and appears 39 times in the corpus. This one is pretty easy because we were able to find it in the Sureth dictionary, and it translates as "to breed, to beget, to conceive." It's not surprising that this one appears so often in our corpus, since the bible is so concerned with lineage. <br />
<br />
ܡܘܼܠܸܕ:ܡܘܼܠܸܕ PresentVerbInfl; ! molidh "to begat, to conceive"<br />
<br />
Adding this form to our transducer bumped our coverage up to 12.03%!<br />
<br />
However, besides this word, we started noticing that a LOT of these uncovered words are single letters; they don't seem to be actual words per se. But we couldn't figure out why so many letters would be sprinkled in our corpus, except as an error. It turns out there are a copy-paste error, as seen in this sentence taking from a pdf piece of our corpus:<br />
<br />
[[File:noticescreenshot.png]]<br />
<br />
This sentence came over to our aii-basic-corpus.txt file looking like this:<br />
<br />
ܡܬܪܓܡܹܢܐ ܡܗܼܝܹܪ ܸܒܠܫܢ ܼ ܕܪܘܫ ܹܡ ܿ ܵ ܐ ܐ ܵ ܵ o ܐ<br />
<br />
53 of 83 tests pass now. Yay!<br />
<br />
[[Category:Sp18_Transducers]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=File:Noticescreenshot.png&diff=7044File:Noticescreenshot.png2018-03-17T17:37:23Z<p>Eschalk1: </p>
<hr />
<div></div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Neo-Aramaic/Transducer&diff=7043Neo-Aramaic/Transducer2018-03-17T17:36:34Z<p>Eschalk1: /* Notes */</p>
<hr />
<div>The code for the transducer can be found in [https://github.swarthmore.edu/cdalton2/ling073-aii this] Github repository.<br />
<br />
==Getting set up: Part One==<br />
<br />
Getting set up for this stage of the project was an ongoing process for us; while it was easy to bootstrap our transducer and start working with Apertium, we found that grammar tags, our <code>Alphabet</code>, and our <code>tests/</code> all needed continuous updating. This was because our grammar rules used tags and characters not found in our defined tags or alphabet, and these changes had to trickle down at some point to <code>tests/</code>, too. We continue to update these to reflect the growing number of tags we use to complete our grammar coverage. We now have tags for verb features, parts of speech, an archiphoneme, person morphology, gender morphology, number morphology, case morphology, TAM morphology, pronoun morphology, and punctuation. <br />
<br />
Our <code>Alphabet</code> includes Syriac characters, Latin characters, a null morpheme (useful for analyzing vowel diacritics), and punctuation.<br />
<br />
Alphabet<br />
ܐ ܒ ܓ ܕ ܗ ܘ ܙ ܚ ܛ ܝ ܟ ܠ ܡ ܢ ܣ ܥ ܦ ܨ ܩ ܪ ܫ ܬ<br />
ܸ ܲ ܵ ܸ ܹ ܼ<br />
◌:0<br />
. ؛ ! - — ، ؟ ' " « » ” “ ( ] [ ) ܀ \ /<br />
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z<br />
a b c d e f g h i j k l m n o p q r s t u v w x y z<br />
<br />
This section also includes our archiphoneme, which is used to realize the vowel sound <code>A</code> as the alaf when it appears at the beginning or end of a word. This is helpful when working with stems where the alaf often disappears after prefixes and suffixes are added to the stem.<br />
<br />
%{A%}:0<br />
! %{A%}:ܐ<br />
<br />
==The hard stuff: Part One==<br />
<br />
When we moved on to the hardest part of putting together a morphological transducer, our biggest challenge was working with the many prefixes and infixes found in Neo-Aramaic. We focused in the beginning on easy forms, like plural nouns, that used mostly suffixes in their morphology. When we moved on to our prefix morphology, we used the solution suggested on the course wiki, where our twoc file only processes forms with a plus and minus version of any given feature: <br />
<br />
Rules<br />
"Remove paths without matching suffix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
_ :* Fy:0 ;<br />
where Fx in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fy in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
"Remove paths without matching prefix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
Fy:0 :* _ ;<br />
where Fy in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fx in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
<br />
This guarantees that if a verb picks up a <code>-hab</code> or <code>-fut</code> tag, it will only analyze these forms as having a habitual or future prefix if they pick up a later <code>+hab</code> or <code>+fut</code> tag.<br />
<br />
== Evaluation ==<br />
<br />
We have 2556 tokenised words in our corpus, with a coverage of 9.66%. We have about 37 stems in our transducer (definitely a place we should be concentrating work right now). <br />
<br />
Top unknown words in the corpus:<br />
<br />
114 ܐ<br />
<br />
86 ܕ<br />
<br />
66 ܠܹܗ<br />
<br />
62 ܡ<br />
<br />
54 ܒ<br />
<br />
46 ܢ<br />
<br />
44 ܘ<br />
<br />
44 ܠ<br />
<br />
39 ܡܘܼܠܸܕ<br />
<br />
36 ܝ<br />
<br />
31 ܪ<br />
<br />
30 ܡܢ<br />
<br />
23 ܵܐ<br />
<br />
21 ܗ<br />
<br />
20 ܹܐ<br />
<br />
18 ܠܐ<br />
<br />
17 ܫ<br />
<br />
14 ܵ<br />
<br />
13 ܚ<br />
<br />
12 ܡܕܝܢܬܐ<br />
<br />
== Notes ==<br />
<br />
The first unknown word we're going to analyze is ܡܘܼܠܸܕ, which is the ninth most common unknown word and appears 39 times in the corpus. This one is pretty easy because we were able to find it in the Sureth dictionary, and it translates as "to breed, to beget, to conceive." It's not surprising that this one appears so often in our corpus, since the bible is so concerned with lineage. <br />
<br />
ܡܘܼܠܸܕ:ܡܘܼܠܸܕ PresentVerbInfl; ! molidh "to begat, to conceive"<br />
<br />
Adding this form to our transducer bumped our coverage up to 12.03%!<br />
<br />
However, besides this word, we started noticing that a LOT of these uncovered words are single letters; they don't seem to be actual words per se. But we couldn't figure out why so many letters would be sprinkled in our corpus, except as an error. It turns out there are a copy-paste error, as seen in this sentence taking from a pdf piece of our corpus:<br />
<br />
<br />
<br />
53 of 83 tests pass now. Yay!<br />
<br />
[[Category:Sp18_Transducers]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Neo-Aramaic/Transducer&diff=7042Neo-Aramaic/Transducer2018-03-17T16:58:20Z<p>Eschalk1: </p>
<hr />
<div>The code for the transducer can be found in [https://github.swarthmore.edu/cdalton2/ling073-aii this] Github repository.<br />
<br />
==Getting set up: Part One==<br />
<br />
Getting set up for this stage of the project was an ongoing process for us; while it was easy to bootstrap our transducer and start working with Apertium, we found that grammar tags, our <code>Alphabet</code>, and our <code>tests/</code> all needed continuous updating. This was because our grammar rules used tags and characters not found in our defined tags or alphabet, and these changes had to trickle down at some point to <code>tests/</code>, too. We continue to update these to reflect the growing number of tags we use to complete our grammar coverage. We now have tags for verb features, parts of speech, an archiphoneme, person morphology, gender morphology, number morphology, case morphology, TAM morphology, pronoun morphology, and punctuation. <br />
<br />
Our <code>Alphabet</code> includes Syriac characters, Latin characters, a null morpheme (useful for analyzing vowel diacritics), and punctuation.<br />
<br />
Alphabet<br />
ܐ ܒ ܓ ܕ ܗ ܘ ܙ ܚ ܛ ܝ ܟ ܠ ܡ ܢ ܣ ܥ ܦ ܨ ܩ ܪ ܫ ܬ<br />
ܸ ܲ ܵ ܸ ܹ ܼ<br />
◌:0<br />
. ؛ ! - — ، ؟ ' " « » ” “ ( ] [ ) ܀ \ /<br />
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z<br />
a b c d e f g h i j k l m n o p q r s t u v w x y z<br />
<br />
This section also includes our archiphoneme, which is used to realize the vowel sound <code>A</code> as the alaf when it appears at the beginning or end of a word. This is helpful when working with stems where the alaf often disappears after prefixes and suffixes are added to the stem.<br />
<br />
%{A%}:0<br />
! %{A%}:ܐ<br />
<br />
==The hard stuff: Part One==<br />
<br />
When we moved on to the hardest part of putting together a morphological transducer, our biggest challenge was working with the many prefixes and infixes found in Neo-Aramaic. We focused in the beginning on easy forms, like plural nouns, that used mostly suffixes in their morphology. When we moved on to our prefix morphology, we used the solution suggested on the course wiki, where our twoc file only processes forms with a plus and minus version of any given feature: <br />
<br />
Rules<br />
"Remove paths without matching suffix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
_ :* Fy:0 ;<br />
where Fx in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fy in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
"Remove paths without matching prefix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
Fy:0 :* _ ;<br />
where Fy in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fx in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
<br />
This guarantees that if a verb picks up a <code>-hab</code> or <code>-fut</code> tag, it will only analyze these forms as having a habitual or future prefix if they pick up a later <code>+hab</code> or <code>+fut</code> tag.<br />
<br />
== Evaluation ==<br />
<br />
We have 2556 tokenised words in our corpus, with a coverage of 9.66%. We have about 37 stems in our transducer (definitely a place we should be concentrating work right now). <br />
<br />
Top unknown words in the corpus:<br />
<br />
114 ܐ<br />
<br />
86 ܕ<br />
<br />
66 ܠܹܗ<br />
<br />
62 ܡ<br />
<br />
54 ܒ<br />
<br />
46 ܢ<br />
<br />
44 ܘ<br />
<br />
44 ܠ<br />
<br />
39 ܡܘܼܠܸܕ<br />
<br />
36 ܝ<br />
<br />
31 ܪ<br />
<br />
30 ܡܢ<br />
<br />
23 ܵܐ<br />
<br />
21 ܗ<br />
<br />
20 ܹܐ<br />
<br />
18 ܠܐ<br />
<br />
17 ܫ<br />
<br />
14 ܵ<br />
<br />
13 ܚ<br />
<br />
12 ܡܕܝܢܬܐ<br />
<br />
== Notes ==<br />
<br />
The first unknown word we're going to analyze is ܡܘܼܠܸܕ, which is the ninth most common unknown word and appears 39 times in the corpus. This one is pretty easy because we were able to find it in the Sureth dictionary, and it translates as "to breed, to beget, to conceive." It's not surprising that this one appears so often in our corpus, since the bible is so concerned with lineage. <br />
<br />
ܡܘܼܠܸܕ:ܡܘܼܠܸܕ PresentVerbInfl; ! molidh "to begat, to conceive"<br />
<br />
Adding this form to our transducer bumped our coverage up to 12.03%!<br />
<br />
53 of 83 tests pass now. Yay!<br />
<br />
[[Category:Sp18_Transducers]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Neo-Aramaic/Transducer&diff=7041Neo-Aramaic/Transducer2018-03-17T03:39:48Z<p>Eschalk1: /* Evaluation */</p>
<hr />
<div>The code for the transducer can be found in [https://github.swarthmore.edu/cdalton2/ling073-aii this] Github repository.<br />
<br />
==Getting set up: Part One==<br />
<br />
Getting set up for this stage of the project was an ongoing process for us; while it was easy to bootstrap our transducer and start working with Apertium, we found that grammar tags, our <code>Alphabet</code>, and our <code>tests/</code> all needed continuous updating. This was because our grammar rules used tags and characters not found in our defined tags or alphabet, and these changes had to trickle down at some point to <code>tests/</code>, too. We continue to update these to reflect the growing number of tags we use to complete our grammar coverage. We now have tags for verb features, parts of speech, an archiphoneme, person morphology, gender morphology, number morphology, case morphology, TAM morphology, pronoun morphology, and punctuation. <br />
<br />
Our <code>Alphabet</code> includes Syriac characters, Latin characters, a null morpheme (useful for analyzing vowel diacritics), and punctuation.<br />
<br />
Alphabet<br />
ܐ ܒ ܓ ܕ ܗ ܘ ܙ ܚ ܛ ܝ ܟ ܠ ܡ ܢ ܣ ܥ ܦ ܨ ܩ ܪ ܫ ܬ<br />
ܸ ܲ ܵ ܸ ܹ ܼ<br />
◌:0<br />
. ؛ ! - — ، ؟ ' " « » ” “ ( ] [ ) ܀ \ /<br />
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z<br />
a b c d e f g h i j k l m n o p q r s t u v w x y z<br />
<br />
This section also includes our archiphoneme, which is used to realize the vowel sound <code>A</code> as the alaf when it appears at the beginning or end of a word. This is helpful when working with stems where the alaf often disappears after prefixes and suffixes are added to the stem.<br />
<br />
%{A%}:0<br />
! %{A%}:ܐ<br />
<br />
==The hard stuff: Part One==<br />
<br />
When we moved on to the hardest part of putting together a morphological transducer, our biggest challenge was working with the many prefixes and infixes found in Neo-Aramaic. We focused in the beginning on easy forms, like plural nouns, that used mostly suffixes in their morphology. When we moved on to our prefix morphology, we used the solution suggested on the course wiki, where our twoc file only processes forms with a plus and minus version of any given feature: <br />
<br />
Rules<br />
"Remove paths without matching suffix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
_ :* Fy:0 ;<br />
where Fx in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fy in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
"Remove paths without matching prefix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
Fy:0 :* _ ;<br />
where Fy in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fx in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
<br />
This guarantees that if a verb picks up a <code>-hab</code> or <code>-fut</code> tag, it will only analyze these forms as having a habitual or future prefix if they pick up a later <code>+hab</code> or <code>+fut</code> tag.<br />
<br />
== Evaluation ==<br />
<br />
We have 2556 tokenised words in our corpus, with a coverage of 9.66%. We have about 37 stems in our transducer (definitely a place we should be concentrating work right now). <br />
<br />
Top unknown words in the corpus:<br />
<br />
114 ܐ<br />
<br />
86 ܕ<br />
<br />
66 ܠܹܗ<br />
<br />
62 ܡ<br />
<br />
54 ܒ<br />
<br />
46 ܢ<br />
<br />
44 ܘ<br />
<br />
44 ܠ<br />
<br />
39 ܡܘܼܠܸܕ<br />
<br />
36 ܝ<br />
<br />
31 ܪ<br />
<br />
30 ܡܢ<br />
<br />
23 ܵܐ<br />
<br />
21 ܗ<br />
<br />
20 ܹܐ<br />
<br />
18 ܠܐ<br />
<br />
17 ܫ<br />
<br />
14 ܵ<br />
<br />
13 ܚ<br />
<br />
12 ܡܕܝܢܬܐ<br />
<br />
== Notes ==<br />
<br />
53 of 83 tests pass now. Yay!<br />
<br />
[[Category:Sp18_Transducers]]</div>Eschalk1http://wikis.swarthmore.edu/ling073/index.php?title=Neo-Aramaic/Transducer&diff=7040Neo-Aramaic/Transducer2018-03-17T03:29:30Z<p>Eschalk1: /* The hard stuff: Part One */</p>
<hr />
<div>The code for the transducer can be found in [https://github.swarthmore.edu/cdalton2/ling073-aii this] Github repository.<br />
<br />
==Getting set up: Part One==<br />
<br />
Getting set up for this stage of the project was an ongoing process for us; while it was easy to bootstrap our transducer and start working with Apertium, we found that grammar tags, our <code>Alphabet</code>, and our <code>tests/</code> all needed continuous updating. This was because our grammar rules used tags and characters not found in our defined tags or alphabet, and these changes had to trickle down at some point to <code>tests/</code>, too. We continue to update these to reflect the growing number of tags we use to complete our grammar coverage. We now have tags for verb features, parts of speech, an archiphoneme, person morphology, gender morphology, number morphology, case morphology, TAM morphology, pronoun morphology, and punctuation. <br />
<br />
Our <code>Alphabet</code> includes Syriac characters, Latin characters, a null morpheme (useful for analyzing vowel diacritics), and punctuation.<br />
<br />
Alphabet<br />
ܐ ܒ ܓ ܕ ܗ ܘ ܙ ܚ ܛ ܝ ܟ ܠ ܡ ܢ ܣ ܥ ܦ ܨ ܩ ܪ ܫ ܬ<br />
ܸ ܲ ܵ ܸ ܹ ܼ<br />
◌:0<br />
. ؛ ! - — ، ؟ ' " « » ” “ ( ] [ ) ܀ \ /<br />
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z<br />
a b c d e f g h i j k l m n o p q r s t u v w x y z<br />
<br />
This section also includes our archiphoneme, which is used to realize the vowel sound <code>A</code> as the alaf when it appears at the beginning or end of a word. This is helpful when working with stems where the alaf often disappears after prefixes and suffixes are added to the stem.<br />
<br />
%{A%}:0<br />
! %{A%}:ܐ<br />
<br />
==The hard stuff: Part One==<br />
<br />
When we moved on to the hardest part of putting together a morphological transducer, our biggest challenge was working with the many prefixes and infixes found in Neo-Aramaic. We focused in the beginning on easy forms, like plural nouns, that used mostly suffixes in their morphology. When we moved on to our prefix morphology, we used the solution suggested on the course wiki, where our twoc file only processes forms with a plus and minus version of any given feature: <br />
<br />
Rules<br />
"Remove paths without matching suffix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
_ :* Fy:0 ;<br />
where Fx in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fy in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
"Remove paths without matching prefix feature"<br />
Fx:0 /<= _ ;<br />
except<br />
Fy:0 :* _ ;<br />
where Fy in ( %[%-hab%] %[%-fut%] %[%-obl%] )<br />
Fx in ( %[%+hab%] %[%+fut%] %[%+obl%] )<br />
matched ;<br />
<br />
This guarantees that if a verb picks up a <code>-hab</code> or <code>-fut</code> tag, it will only analyze these forms as having a habitual or future prefix if they pick up a later <code>+hab</code> or <code>+fut</code> tag.<br />
<br />
== Evaluation ==<br />
<br />
Number of tokenised words in the corpus: 2556<br />
<br />
Coverage: 9.66%<br />
<br />
Top unknown words in the corpus:<br />
<br />
114 ܐ<br />
<br />
86 ܕ<br />
<br />
66 ܠܹܗ<br />
<br />
62 ܡ<br />
<br />
54 ܒ<br />
<br />
46 ܢ<br />
<br />
44 ܘ<br />
<br />
44 ܠ<br />
<br />
39 ܡܘܼܠܸܕ<br />
<br />
36 ܝ<br />
<br />
31 ܪ<br />
<br />
30 ܡܢ<br />
<br />
23 ܵܐ<br />
<br />
21 ܗ<br />
<br />
20 ܹܐ<br />
<br />
18 ܠܐ<br />
<br />
17 ܫ<br />
<br />
14 ܵ<br />
<br />
13 ܚ<br />
<br />
12 ܡܕܝܢܬܐ<br />
<br />
== Notes ==<br />
<br />
53 of 83 tests pass now. Yay!<br />
<br />
[[Category:Sp18_Transducers]]</div>Eschalk1