Adam Ledyard and Sanae Sato
Lab4 - Miyako


Addressing Lab3 remarks:

-----

>>Why are you adding -0 instead of just using a non-inflecting
lexical rule?  (Either one works, I'm just curious about the motivation.)

The tense can be marked or not marked.  It's actually supposed to not be
obligatory which we changed in this iteration of the grammar

------
        
>> It should be possible to do the separate word case, too:

1) Define a lexical rule for the non-finite (combining) form of
the main verbs.  This will be in a position class that attaches
directly to the stem, and doesn't have any other position classes
that take it as input.

2) Define the aspect auxiliaries, and have them select non-finite
complements.

3) Let the other affixes take the auxiliaries as their input.

Let me know what questions you have about doing this.

------------> 	We didn't have time to address this at the moment, but it has been added
		to the to-do list        

>> What does class2 mean?

---->	Class 1 verb stems all end in 'i' whereas class 2 verb stems end in anything
	else and will need an thematic vowel depending on the suffix and if the stem
	has no vowel in coda position of the stem

	Example:  
	tub => to fly
	tuv-tar => past tense requires the athematic stem and the stop is fricativized
	tub-a-baa => the optative mood suffix requires an 'a' thematic vowel
	tub-i => Imperative requires an 'i' thematic vowel

---------------

        Mood is attached to the very end and can express the optative,
        intentional, negative sentential, prohibitive, imperative and
        realis.  The prohibitive and imperative have slightly
        different syntax in that the subject is not optional, it is
        actually prohibited.  Realis is overtly expressed and the
        irrealis is not.  This works since only one mood suffix is
        permitted

>> Is "negative sentential" really sentential negation, i.e., not
a mood but a negative morpheme that occupies the same position
class as the mood markers?

---> that is a typo, it should be negative intention...

	Negation is separate from mood since the mood affix and the negative affix
	can occur in the verb:

		#Example 43 grammatical use of NEG > TENSE(PAST) > -baa MOOD
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: tam
		ba=ga tigami=u kakattabaa
		ba=ga tigami=u kaka-t-ta-baa
		1SG=NOM letter=ACC write-NEG-PAST-OPTA
		I did not want to write a letter.

        
>> "Cannot handle" meaning it overgenerates?  Why?  If you
just put the negative marker in the same position class as the
other moods, you should avoid parsing such examples.

---> see above. negative intention is actually separate from the negation affix.

	#Example 44 grammatical use of -djaan MOOD suffix
	Source: author
	Vetted: f
	Judgment: g
	Phenomena: tam
	ba=ga tigami=u kakadjaan
	ba=ga tigami=u kaka-djaan
	1SG=NOM letter=ACC write-NEGINTENT
	I won't write a letter.

	-di is intent and -djaan is negative intent

-----------

        The nominal case is optional in all sentences and is seen in
        the grammar and there seemed to be no restrictions.  The
        accusative case might be optional but there is no mention of
        this in the grammar so we are going to play it safe and say
        that is prohibited.

>> I think you mean "nominative marked argument" when you
say "nominal case" and "accusative marked argument" when
you sayd "accusative case".  I would be a little surprised if
you don't ever get object drop.  Keep an eye out for such examples
as you are working with your materials.

---->	We haven't seen any object drop yet, so we weren't comfortable saying that is
	possible until we find an example.

----------

        Sentential negation is a suffix added to the verb and is seen in the TAM
        section.  It can be -n or -t depending on if the next sound is 't' which
        isn't restricted and is not handled as of yet.

>> This is a morphophonological alternation that you can abstract
away from.  Just call it -n in the morpheme segemented line. (And
then target that line with your grammar rather than the surface one.)
Alternatively, when it comes to tdl editing, you can write rules
that change the -n to a -t if the following affix starts with a t,
though this will get a little fiddly.

However, 'kakabaa'/'kakadi'/'kakadjaan' is natural form as input so we
left as it is.  Also, we do not have a syntactical way to distinguish
-djaan (negative intentional) from  negation of intentional. We were
not sure if the sentential negation of intentional form is really
incorrect form in Ryukyuan, however we assumed that it is an incorrect
form (for now) because it has -djaan (negative intentional) form
separately. Similary, other suffixes/converbs which have negation
feature already embedded in semantics such as -rna (prohibition) and
-da (converve).

>> I need a little more info here.  What is the negative intentional?
What are some examples?

---> this is dealt with above.

------------------

We implemented TENSE as obligatory. Therefore LKB will not parse
sentences which has no TENSE suffix. In fact, it is easy fix these by
adding 0 morpheme if we re-define our testsuite.txt.

>> Or you can add a lex rule to that position class that fills in whatever
TENSE information is appropriate in the absence of an overt morpheme
but has no affix associated.  Let me know if it's not clear how to do this.

---> we switched it to not being obligatory since the irrealis examples are not
marked for tense

----------

There is another phenomena which we need to be addressed in the
future. With similar reason, ungrammatical example (Example 45) "ba=ga
tigami=u kaka-n-di" will be parsed if we add -0 (TENSE suffix) as
"tigami=u kaka-0-n-di" which should not be parsed because there is
negative intentional form -djaan in Irabu Ryukyuan.

        #Example 45 ungrammatical use of -di MOOD suffix with NEG (should use -djaan)
        Source: author
        Vetted: f
        Judgment: u
        Phenomena: tam
        ba=ga tigami=u kakandi
        ba=ga tigami=u kaka-n-di
        1SG=NOM letter=ACC write-NEG-INTENT
        I will not  write a letter.

>> Do NEG and INTENT in fact fill the same slot? In that case,
putting them in the same position class should rule this out.

	---> NEG and INTENT do fill separate slots since tense can be inserted between 
		(see above)

----------

Changes and additions to the choices file with their representation in the test suite

Auxilaries:

We added auxilaries for later development since aspect can be expressed using an
auxilary or as a suffix:

	< has-aux=no
	> has-aux=yes
	> aux-comp-order=after
	> aux-comp=v
	> multiple-aux=no

Some examples of this are:

ba=a tigami=u [kaki-i=du u-tar].
1SG=TOP letter=ACC write-CVB=FOC PROG-PAST
‘I was writing a letter’

Perfect
ba=a tigami=u kaki-i njaa-n-Ø.
1SG=TOP letter=ACC write-CVB PRF-NEG-NPST
‘I have written a letter.’ 

New cases:

We needed a new case for NP coordination since it is an adposition and a feature
is required so the ASsoCiative case was added.  Also, some example sentences have
the INStrumental particle so that case was also added.

	>   case3_name=asc
	>   case4_name=ins


NP Coordination:

The particle =tu is used for coordinating NPs and replaces the case marker.

	>   cs1_np=on
	>   cs1_pat=omni
	>   cs1_mark=word
	>   cs1_orth=tu
	>   cs1_order=after

and this combines with the new adposition:

	>   adp5_orth=tu
	>   adp5_order=after
	>     adp5_feat1_name=case
	>     adp5_feat1_value=asc

It goes after the whole NP and is illustrated with NPs with an adjective within
the example sentence:

		#Example 56 both NPs are marked with ASC label
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: crd
		pztu=tu midum=tu iditar
		pztu=tu midum=tu iditar
		man=ASC woman=ASC exit-PAST
		A man and a woman exited

		#Example 57 both NPs are marked with ASC label
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: crd
		ujakiiujaki=nu pztu=tu kibannkiban=nu pztu=tu niv0
		ujakiiujaki=nu pztu=tu kibannkiban=nu pztu=tu niv0-0
		rich=GEN man=ASC poor=GEN man=ASC sleep-NPST
		The rich man and the poor man sleep.

		#Example 58 Case marking is not allowed to coincide with crd 
		Source: author
		Vetted: f
		Judgment: u
		Phenomena: crd
		pztu=tu=nu midum=tu iditar
		pztu=tu=nu midum=tu iditar
		man=ASC=NOM woman=ASC exit-PAST
		A man and a woman exited





Interrogatives:

The particle -ru can be used as a question marker and is at the end of a sentence:

	> q-part=on
	> q-part-order=after
	> q-part-orth=ru

The focus marker -ru can also be used to signify that it is an interrogative and then
makes the question marker optional.  This means that all three of these questions are
grammatical:

		#Example 53 -ru Focus particle on subject, makes it an interrogative 
		Source: a
		Vetted: f
		Judgment: g
		Phenomena: wo
		vva=ga=ru uri=u aztar
		vva=ga=ru uri=u az-tar
		2SG=NOM=FOC that=ACC say-PAST
		Did you say that?

		#Example 54 -ru Question particle on verb, makes it an interrogative 
		Source: a
		Vetted: f
		Judgment: g
		Phenomena: wo
		vva=a uri=u aztar=ru
		vva=a uri=u az-tar=ru
		2SG=TOP that=ACC say-PAST=Q
		Did you say that?

		#Example 55 -ru Question particle on verb and FOC marker on subject makes it an interrogative 
		Source: a
		Vetted: f
		Judgment: g
		Phenomena: wo
		vva=ga=ru uri=u aztar=ru
		vva=ga=ru uri=u az-tar=ru
		2SG=NOM=FOC that=ACC say-PAST=Q
		Did you say that?



Lexicon:

We added some more nouns and verbs since we wanted to diversify the example sentences a bit:

	>     noun1_stem5_orth=saki
	>     noun1_stem5_pred=_saki_n_rel
	>     noun1_stem6_orth=sinsii
	>     noun1_stem6_pred=_sinsii_n_rel
	>     noun1_stem7_orth=cc
	>     noun1_stem7_pred=_cc_n_rel


	>     verb4_stem2_orth=tuv
	>     verb4_stem2_pred=_tuv_v_rel
	>     verb4_stem3_orth=niv
	>     verb4_stem3_pred=_niv_v_rel
	>     verb6_stem8_orth=num
	>     verb6_stem8_pred=_num_v_rel
	>     verb6_stem9_orth=jum
	>     verb6_stem9_pred=_jum_v_rel
	>     verb6_stem10_orth=kurus
	>     verb6_stem10_pred=_kurus_v_rel
	>     verb6_stem11_orth=kurusa
	>     verb6_stem11_pred=_kurusa_v_rel


Addition of Copula:

The copula is the only verb that requires agreement with the subject so a new feature was introduced,
animacy since a different verb is needed depending on the subject.  The copula is needed for the non-verbal
predicate so this needed to be addressed.

	
	>   feature3_name=ANIMATE
	>   feature3_type=head
	>   feature3_cat=both
	>   feature3_new=no
	>   feature3_existing=bool

Examples of the non-verbal predicates:

		# nominal predicate (NP)
		#Example 71
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: cop
		kari=a sinsii jar0
		kari=a sinsii jar-0
		3SG=TOP teacher COP-NPST
		He is a teacher

		#(35) Adj
		#Example 72
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: cop
		kari=a takaf=du ar0
		kari=a taka-f=du ar-0
		3SG=TOP tall-ADV=FOC COP-NPST
		He is tall

		# nominal locative (PP)
		#Example 73
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: cop
		pztu=nu cc=n ur0
		pztu=nu cc=n ur-0
		man=NOM moon=DAT COP-NPST
		The man is on the moon 

We are still a bit confused with this since (j)ar (ur being the animate one)is the 
inanimate copula but it is used in the NP and Adj predicates when the subject is 
animate but these are ungrammatical:

		*pžtu=nu=du ar 
		‘there is a man’

		*jama=nu=du ur 
		‘there is a mountain’


New Particles:

The topic and focus are marked with adpositions so these were added.

	> section=info-str
	>   focus-marker1_type=adp
	>   topic-marker1_type=adp

	>   adp6_orth=a
	>   adp6_order=after
	>     adp6_feat1_name=information-structure marking
	>     adp6_feat1_value=tp
	>     adp6_feat2_name=information-structure meaning
	>     adp6_feat2_value=topic

	>   adp7_orth=du
	>   adp7_order=after
	>     adp7_feat1_name=information-structure marking
	>     adp7_feat1_value=fc
	>     adp7_feat2_name=information-structure meaning
	>     adp7_feat2_value=focus

	>   adp8_orth=baa
	>   adp8_order=after
	>     adp8_feat1_name=information-structure marking
	>     adp8_feat1_value=tp
	>     adp8_feat2_name=information-structure meaning
	>     adp8_feat2_value=topic

'baa' is used to mark topic of the object and 'a' is used elsewhere.  'du' marks focus. 
Our grammar can't handle multiple adposition attached to one constiuent so some of these
are not parsing.  You will also notice that mutiple TOP markers can be within a sentence
and the 'du' FOC marker can mark many types of constiuents (verb, noun, adverb)

		#Example 75 Focus particle
		Source: a
		Vetted: s
		Judgment: g
		Phenomena: info
		pztu=nu=du ff0
		pztu=nu=du ff-0
		man=NOM=FOC come-NPST
		A man comes over.

		#Example 76 Focus particle must come after case marker
		Source: a
		Vetted: s
		Judgment: u
		Phenomena: info
		pztu=du=nu ff0
		pztu=du=nu ff-0
		man=FOC=NOM come-NPST
		A man comes over.

		#Example 77 There can be more than one TOP tagged element
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: info
		ba=a mm=mu=baa faan0
		ba=a mm=mu=baa faa-n-0
		1SG=TOP potato=ACC=TOP eat-NEG-NPST
		I don't eat potatoes.

		#Example 78 TOP and FOC can be in the sentence together
		Source: a
		Vetted: s
		Judgment: g
		Phenomena: info
		ba=a zzu=u=du turtar
		ba=a zzu=u=du tur-tar
		1SG=TOP fish=ACC=FOC take-PAST
		I caught a fish.

		#Example 79 FOC can't be used with REAL suffix
		Source: a
		Vetted: s
		Judgment: u
		Phenomena: info
		ba=a zzu=u=du turtam
		ba=a zzu=u=du tur-ta-m
		1SG=TOP fish=ACC=FOC take-PAST-REAL
		I caught a fish.

We also added the INStrumental adposition:

	>   adp9_orth=sii
	>   adp9_order=after
	>     adp9_feat1_name=case
	>     adp9_feat1_value=ins



New suffix:

We added a new suffix to the morphology for coordination of clauses and subordination.
It attaches to the stem without any other suffixes being allowed and we only added
two of them.  There are quite a few more but for testing purposes, we started small.

	>   verb-pc12_name=converb
	>   verb-pc12_order=suffix
	>   verb-pc12_inputs=verb1, verb2
	>     verb-pc12_lrt2_name=medial
	>       verb-pc12_lrt2_lri1_inflecting=yes
	>       verb-pc12_lrt2_lri1_orth=i
	>     verb-pc12_lrt3_name=conditional
	>       verb-pc12_lrt3_lri1_inflecting=yes
	>       verb-pc12_lrt3_lri1_orth=tigaa

The medial suffix is for string clauses together and is needed for the other form
of ASPECT that uses the auxilary:

		#Example 59
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: crd
		kuzmi=u burii ai=sii=du fau-tar
		kuzmi=u bur-i-i, ai=sii=du fau-tar
		small.claw=ACC break-THM-MED, that.way=INS=FOC eat-PST
		(I) broke the small claws and ate it in that way.



The conditional is used for subordination:

		#Example 60
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: emb-d
		saki=u numtigaa, ba=a sugu=du niv0
		saki=u num-tigaa, ba=a sugu=du niv-0
		Sake=ACC drink-CVB.COND 1SG=TOP right.away=FOC sleep-NPST
		When (I) drink Sake, I sleep easily.

	


Additions to the test suite that aren't dealt with in the choices-generated grammar:

Embedded clauses:

Embedded clauses are marked with either a converb as in example 69 or as a VP to the 
left of the modified item as in example 70 which seems akin to the way it is dealt with
in Japanese.

		#Example 69
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: emb-d
		ba=a saki=u numtigaa, sugu=du niv0
		ba=a [saki=u num-tigaa], sugu=du niv-0
		1SG=TOP [Sake=ACC drink-CVB.COND] right.away=FOC sleep-NPST
		[When I drink Sake], I sleep easily.

		#Example 70
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: emb-d
		ba=ga kuuttar pztu=u=baa kurustar
		ba=ga [kuu-t-tar] pztu=u=baa kurus-tar
		1SG=NOM [come-NEG-PST] man=ACC=TOP kill-PST
		I killed the man [that didn't come.]


clausal coordination/subordiation:

The matrix questionaire cannot handle clausal coordination that is handled with a suffix on the verb
which is how it is seen in Miyako:

		#Example 59
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: crd
		kuzmi=u burii ai=sii=du fau-tar
		kuzmi=u bur-i-i, ai=sii=du fau-tar
		small.claw=ACC break-THM-MED, that.way=INS=FOC eat-PST
		(I) broke the small claws and ate it in that way.

		#Example 60
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: emb-d
		saki=u numtigaa, ba=a sugu=du niv0
		saki=u num-tigaa, ba=a sugu=du niv-0
		Sake=ACC drink-CVB.COND 1SG=TOP right.away=FOC sleep-NPST
		When (I) drink Sake, I sleep easily.


A converb suffix is added to the verb and a complete phrase can be attached to the right, either coordination
as in 59 or subordination as in 60.

Adverbs:

Adverbs can either be created with an inherently labeled adverb (example 60 above) or an adjectival stem with
a suffix, -f:

		#Example 66 Adverbs take a stem and add -f to adverbialize it.
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: adv
		ba=ga vva=u japa-f kurusadi
		ba=ga vva=u japa-f kurusa-di 
		1SG=NOM 2SG=ACC soft-ADV kill-INTENT
		I kill you softly

Semantically this sentence might not make sense but there are very little adverbs in the document
so we had to improvize.

Adjectives:

Adjectives are realized by taking a stem (either noun or adjective) and reduplicating it.  At this
time we aren't dealing with reduplication and have just put the whole word in as is:

		#Example 64 Demonstrative must go before adjective in the NP
		Source: author
		Vetted: f
		Judgment: g
		Phenomena: adj
		kunu ujakiiujaki=nu pztu=nu niv0
		kunu ujakiiujaki=nu pztu=nu niv-0
		PROX rich=GEN man=NOM sleep-NPST
		This rich man sleeps

The reduplicated stem then takes the GEN particle and is attached to the left of the head noun but to the right
of the demonstrative.  Adjectives and the genative need to be added for us to be able to handle this.

Coverage and performance:

We are getting 62 percent coverage but this is an understandable drop since we added quite a few phenomena
that could not be handled with the choices file.

Overgeneration:

We are overgenerating in a few areas, intransitive unmarked case, the zero marked suffixes, and ambiguity in
spellings, verb-class specific suffixes.

The zero marked suffix for intentional, imperative and nonpast cause problems if we make tense not obligatory.
We did this for the irrealis verbs which do not require tense so we will need to restrict these somehow 
by adding constraints or dealing with the zero suffix differently.

Pro-drop is conflicting with some of the parses as well:

		#Example 12 1SG Pronoun
		Source: author
		Vetted: f
		Judgment: u
		Phenomena: c
		ba ga iditar
		ba=ga idi-tar
		1SG=GEN exit-PAST
		I exited.

		#Example 13 1SG Pronoun
		Source: author
		Vetted: f
		Judgment: u
		Phenomena: c
		ba n iditar
		ba=n idi-tar
		1SG=DAT exit-PAST
		I exited.

These are being treated as the subject being dropped which needs to be addressed.

There is ambiguity in spelling which is legitimate where this:

	#Example 31 ungrammatical use of -u ASPECT suffix w/ -tar PAST suffix
	Source: author
	Vetted: f
	Judgment: u
	Phenomena: tam
	ba=ga tigami=u kakitaru
	ba=ga tigami=u kaki-tar-u
	1SG=NOM letter=ACC write-PAST-PROG
	I was writing a letter.

is being parsed as an imperative:

	ba=ga tigami=u kakitaru
	ba=ga tigami=u kaki-ta-ru
	1SG=NOM letter=ACC write-PAST-IMP
	I was writing a letter.

which is also ungrammatical but that will go away when we deal with the imperative and 
restrict it to a structure with no nominal.  

The animacy requirement for the copula isn't being caught as well:


		#Example 74 Animacy is important in the copula
		Source: author
		Vetted: f
		Judgment: u
		Phenomena: cop
		pztu=nu cc=n jar0
		pztu=nu cc=n jar-0
		man=NOM moon=DAT COP-NPST
		The man is on the moon


Errors:

The errors from last time are still present but that will never change since we are switching
the suffixes around and the grammar is not recognizing it.  The rest of the errors are from the
adjectives and adverbs which could not be entered into the matrix.
