LAB 4
Writeup
Gracheva, Suskic                                                                     



Part 1: documentation fo the new or revised choices we made in the customization system illustrated with
examples from our testsuite.


						MATRIX YES-NO QUESTIONS


//INTONATION USE FOR YES-NO QUESTIONS

Russian can have intonation as the sole marker of the syntactic function for the yes-no questions.  
Yes/No questions thus can retain the normal declarative order and are identified only phonetically, 
i.e. by intonation.  This is the intonation "IK-3" ("intonational contour") as described by Cubberley 
(adapted from Bryzgunova 1960s).  It is referred to as the "peculiarly Russian contour in shape", 
with the central syllable marked by a sharp rise in the speaker's top level, followed by an equally 
sharp fall to the bottom level.  Again, since Russian is a free word order language and we cannot 
really mark intonation syntactically, so we only have grammatical examples for this part.  

We selected the option to use the Separate question particle "li" and sentence final; as the position. 
However, this did not work since in Russian, the particle is placed usually after the first stressed word 
in the question (the verb or a predicate word) indicating the focus of the question. The end result is that 
our grammatical questions with "li” in the second place fail to parse in lkb and itsdb, while our ungrammatical 
questions with “li” at the end of the sentence are recognized as grammatical in itsdb (see further below for 
examples)


#160   (subj verb obj)
Source: a, e
Vetted: t
Judgment: g
Phenomena: Matrix yes-no questions
Ivan kupil        sobaku?
Ivan kupi-l       sobak-u
Ivan buy-3SG.PST  dog-ACC.SG
Did Ivan buy a dog?

#161   (obj verb subj: this word order places the emphasis on the "dog")
Source: a, e
Vetted: t
Judgment: g
Phenomena: Matrix yes-no questions
Sobaku      kupil       Ivan?
sobak-u     kupi-l      Ivan
Dog-ACC.SG  buy-3SG.PST Ivan
Did Ivan buy a dog? 

#162    (verb  subj  obj):  that sounds like a reconfirmation, i.e. describing a situation in which we know that
#	Ivan was going to buy a dog, and now we are asking to reconfirm that "Did Ivan buy a dog?")
Source: a, e
Vetted: t
Judgment: g
Phenomena: Matrix yes-no questions
Kupil        Ivan sobaku?
Kupi-l       Ivan sobak-u
buy-3SG.PST  Ivan dog-ACC
Did Ivan buy a dog? 



//PARTICLE "LI" USE FOR YES-NO QUESTIONS

An interrogative particle "li" can be also used to mark the interrogative: its position is 
usually second in the clause, following the word central to the question (usually the verb), 
which leads to the inversion of S-V.  Intonation remains IK-3, centered on the last word.  
The use of particle "li" is optional.  Also, based on the data from two native speakers, "li" 
can also occupy a position other than the third (although this is a much less common form) and 
also it can follow the pos other than the verb.  The further away from the second position the
particle "li" is, the more ungrammatical the sentence becomes.  

#163   (subj	li	verb 	obj)
Source: a, e
Vetted: t
Judgment: g
Phenomena: Matrix yes-no questions
Ivan li 	     kupil        sobaku?
Ivan li 	     kupi-l       sobak-u
Ivan INTERROGATIVE   buy-3SG.PST  dog-ACC.SG
Did Ivan buy a dog?

#164   (obj	li	verb	subj): that sounds like "was it the dog that Ivan bought?")
Source: a, e
Vetted: t
Judgment: g
Phenomena: Matrix yes-no questions
Sobaku      li			kupil       Ivan?
sobak-u     li			kupi-l      Ivan
Dog-ACC.SG  INTERROGATIVE	buy-3SG.PST Ivan
Did Ivan buy a dog? 

#165    (verb li subj obj: that sounds like "Ivan was going to buy a dog, and now we are asking to reconfirm 
#	that "Did Ivan buy a dog? as in "did Ivan end up buying a dog?"
Source: a, e
Vetted: t
Judgment: g
Phenomena: Matrix yes-no questions
Kupil        li			Ivan sobaku?
Kupi-l       li			Ivan sobak-u
buy-3SG.PST  INTERROGATIVE	Ivan dog-ACC
Did Ivan buy a dog? 

#166   (*obj	verb	subj	li)
Source: a, e
Vetted: t
Judgment: u
Phenomena: Matrix yes-no questions
Li		Ivan  	kupil        sobaku?
Li		Ivan    kupi-l       sobak-u
Interrogative	Ivan 	buy-3SG.PST  dog-ACC.SG
intended: did Ivan buy a dog?

#167   (*obj	verb 	subj	li)
Source: a, e
Vetted: t
Judgment: u
Phenomena: Matrix yes-no questions
Sobaku      kupil       Ivan	li?
Sobak-u     kupi-l      Ivan	li
Dog-ACC.SG  buy-3SG.PST Ivan	INTERROGATIVE
intended: did Ivan buy a dog? 

#168    (*verb	subj	li 	obj)
Source: a, e
Vetted: t
Judgment: u
Phenomena: Matrix yes-no questions
Kupil        Ivan 	li		sobaku?
Kupi-l       Ivan 	li		sobak-u
buy-3SG.PST  Ivan 	INTERROGATIVE	dog-ACC
ungrammatical: did Ivan buy a dog? 

When we run the unanalyzed and wellformed examples in tsdb++, we get several problems with "li":
sentence #164 (Kupil li Ivan sobaku?) is analyzed as ungrammatical because the particle "li"
is not placed in one of the positions licensed by Grammar Customization, although it is a
grammatical sentence. The same applies to sentence #163 (Sobaku li kupil Ivan?) and sentence 162
(Ivan li kupil sobaku?)




					//COORDINATION

Russian equivalent of English "and" is "i", which can join words, phrases, or clauses.  In Russian it 
appears that they can be joined in different ways: 

1). either only one (last) coordinand is marked (A, B and C), 
2). or all coordinands are marked (and A, and B, and C), with the latter giving a reading similar to 
   (both ... and...), 
3). or none of the coordinands are marked (A, B, C).  

To cater for this we defined 3 coordination strategies in Matrix Customization: 

We chose monosyndeton (gives us polysydenton), omnisyndenton and asyndenton. In the first two cases the 
conjunction is marked with the word “I”. For the first option we chose nouns, NPs and VPs as the coordinants, 
while for the latter two we chose nopuns and NPs only. We were getting multiple parses when the latter two 
options also coordinated VPs and sentences. Itsdb was also not able to parse that many edges and was failing.

This is what we have in the choices file for coordination:
 
section=coordination
  cs1_n=on
  cs1_np=on
  cs1_vp=on
  cs1_pat=mono
  cs1_mark=word
  cs1_orth=i
  cs1_order=before
  cs3_n=on
  cs3_np=on
  cs3_pat=omni
  cs3_mark=word
  cs3_orth=i
  cs3_order=before
  cs4_n=on
  cs4_np=on
  cs4_pat=a

In the lexicon file we had to define 3 different entries for "i":

In our testsuite we have examples both for the sentences with only two coordinands (examples #168-173)
and with 3 coordinands (#174-189). We cover N, NP, V, VP and S coordination in these testsuite examples. 
All of these use the same coordination strategies as described above.    

The ungrammatical examples used in our testsuite place "i" in the positions which are not licensed as
defined by our coordination strategies (below are the examples for Ns).  

#187    (*N i N i N)
Source: a, e
Vetted: t
Judgment: u
Phenomena: coordination
Maljchik,  i    devochka,   i     sobaka    letjat       samoljotom. 
Maljchik   i    devochk-a   i     sobak-a   let-jat      samoljot-om.  
Boy.NOM    and  girl-NOM    and   dog-NOM   fly-3PL.PRS  plane-INST.SG
intended: The boy, the girl and the dog are flying by airplane.

#188    (*i noun i noun noun)
Source: a, e
Vetted: t
Judgment: u
Phenomena: coordination
Maljchik,  i    devochka,   sobaka    letjat       samoljotom. 
Maljchik   i    devochk-a   sobak-a   let-jat      samoljot-om.  
Boy.NOM    and  girl-NOM    dog-NOM   fly-3PL.PRS  plane-INST.SG
intended: The boy, the girl and the dog are flying by airplane.

#189    (*i noun i noun noun)
Source: a, e
Vetted: t
Judgment: u
Phenomena: coordination
I    maljchik,  i    devochka,   sobaka    letjat       samoljotom. 
I    maljchik   i    devochk-a   sobak-a   let-jat      samoljot-om.  
And  boy.NOM    and  girl-NOM    dog-NOM   fly-3PL.PRS  plane-INST.SG
intended: The boy, the girl and the dog are flying by airplane.

 





Part 2: descriptions of properties of Russian illustrated in our test suite but not covered by our starter
grammar and/or customization system.


					//ATTRIBUTIVE ADJECTIVES

In Russian attributive adjectives appear usually before the nouns (while predicative adjectives 
usually follow the noun), with the exception being when they describe nouns in menus or lists or 
when the noun is generic (as in example "Ivan - student userdnyj" ('Ivan is a hardworking student').  
They agree with the nouns they modify in gender, number, and case features. We have covered this 
agreement for adjectives in the previous testsuite for Lab 3 (#107-116), in which we had the 
adjectives, appearing before the nouns, and agreeing with them in Gender, Number, and Case features.  
This time we are adding the examples that demonstrate that unlike English, other circumstantial 
information can come between attributive adjectives and the nouns they modify.  It is also not 
mentioned in the sources, but, based on the native speakers' data, it seems that the attributive 
adjectives usually follow the determiners (if determiners are present) and not the other way around.  

#190 comment: n-sg-nom verb-3sgsubj adj-sg-acc prep adj-sg-inst n-sg-inst n-sg-acc
Source: a, e
Vetted: t
Judgment: g
Phenomena: adjectives
Devochka           chitaet        krasnuju          s     boljshim             perepljotom            knigu.
Devochk-a          chita-et       krasnu-ju         s     boljsh-im            perepljot-om   	      knig-u
Girl-FEM.3SG.NOM   read-3SG.PRS   red-FEM.SG.ACC    with  large-INST.SG.MASC   binding-INST.SG.MASC   book-FEM.SG.ACC
The girl is reading a red book with a big binding. 

The other testsuite examples we have cover these constructions:

*n-sg-nom verb-3sgsubj prep adj-sg-inst n-sg-inst adj-sg-acc n-sg-acc
det adj noun
*adj det noun

						//ADVERBS
//ADVERB FORMATION

The most common formation of adverbs in Russian is by adding the -o to the stem of the adjective 
(Cubberley refers to it as not the suffixation, but instead to as just making the short neuter form 
of the adjective, which is formed by adding -o to the stem), i.e. from the adjective "horosh-ij" 
(good) we can form an adverb "horosh-o" (well).  The other examples include bystr-yj/bystr-o 
(quick/quickly),"ploh-oj/ploh-o" (bad.ADJ/bad.ADV), skuchn-yj/skuchn-o (boring.ADJ/boring.ADV)

The adverbs that do not have a short form (these are the adjectives with the suffix -(e)sk-, 
form the adverbs by adding the suffix -i to the stem, i.e. druzhesk-ij/druzhesk-i, vorovsk-oj/vorovsk-i.  

The adverbs of time and quantity were often formed long time ago, from the nouns, with their 
etymology now rather obscure, as in vchera, ochenj, zdesj.  

Lastly, some adverbs have been formed from the case forms of temporal or quantity nouns, i.e. 
let-o/let-om (summer/in summer), utr-o/utr-om (morning/in the morning).

Comparative adverbs are usually homophonous with the comparative adjectives and their pos can be 
distinguished by the element they modify: i.e. if it modifies the noun, then it is a comparative 
adjective, and if it modifies another adjective or a verb, then it is a comparative adverb.  



//ADVERB POSITION
Adverbs can attach to V, VP, and S.  They are usually positioned next to the verb they qualify (Wade 534).  
The word order, however, is slightly more complicated phenomenon. Per Wade, the adverbs of time usually 
precede the verb (i.e. adverb "vsegda" (always)), although 4 native speakers say that though this might be a 
more common way of positioning the adverbs, they can also position the adverbs after the verb they modify 
and it will still be grammatical.  The manner adverb "quickly", on the other hand, is more commonly positioned 
to the right of the verb, per data gathered from 4 native speakers.  The adverbs conveying essential new 
information usually appear in the final position (Wade 535).  However, based on the data from 4 native
speakers, both positions (sentence-initial preceding the verb and sentence-final following the verb) 
are equally acceptable for these "new information" adverbs.  This might indicate a difference in the 
positioning of adverbs depending on what group they belong to (manner vs. time vs. introducing new information). 
Therefore, for now we will treat them as following the free word order.  

Per Wade, if there is more than one adverb or adverbial phrase in the sentence, their order of appearance is 
usually in the following order:
1. adverbs/adverbial phrases of time
2. adverbs/adverbial phrases of place
3. other types of adverb/adverbial phrase (manner, cause, etc.)


We cover these possibilities in our testsuite: 
-for adverbs attaching to V we look at these possibilities: (noun adv verb), (noun verb adverb), (adv noun verb)
-for adverbs attaching to S: (adverb S), (S adverb)

We don't really have ungrammatical examples for the adverbs since they can attach to the verbs in any order.

Some of the examples in our testsuite are:

#199    (adv VP)
Source: a, e
Vetted: t
Judgment: g
Phenomena: adverbs
Ivan      bystro   letit        samoljotom.  
Ivan      bystro   let-it       samoljot-om
Ivan.NOM  quickly  fly-3SG.PRS  plane-INST.SG
Ivan is flying quickly by the plane.    

#200    (VP adv)
Source: a, e
Vetted: t
Judgment: g
Phenomena: adverbs
Ivan      samoljotom        letit        bystro.  
Ivan      samoljotom        let-it       bystro   
Ivan.NOM  plane-INST.SG     fly-3SG.PRS  quickly
Ivan is flying quickly by the plane.   

#201    (adv noun VP)
Source: a, e
Vetted: t
Judgment: g
Phenomena: adverbs
Ivan      bystro       samoljotom        letit.  
Ivan      bystro       samoljot-om       let-it  
Ivan.NOM  quickly      plane-INST.SG     fly-3SG.PRS
Ivan is flying quickly by the plane.   
 

				//EMBEDDED CLAUSES (declarative, interrogative)


Russian uses complementizers for embedded declarative clauses and particle "li" for the interrogative
clauses.

//EMBEDDED DECLARATIVE CLAUSES: 

WORD ORDER: 
The complementizer occupies the first position in the embedded clause, the rest of the clause also seems 
to follow the free word order (there wasn't anything written on this in our sources, but 3 native speakers 
judged these sentences below as grammatical, noting that although every word order carries a different 
emphasis/intonation, all sentences are grammatical).  

COMPLEMENTIZER: 
Complementizer "chto" ("that") is used for declarative clauses of saying, thinking, etc.  It cannot be 
omitted as in English in such contexts.  Other conjuctions can be used for other types of declarative 
clauses (i.e. "chtoby" ("in order to") for conjunctions of purpose; "tak chto" ("so") for conjunctions 
of result; "esli" ("if") for conditional conjunctions; "hotja/hotj" ("although") and "pustj" (even if) 
for concessive conjunctions, "kak" ("as, like") for comparative conjunctions, "do togo kak" ("until"), 
"posle togo kak" ("after"), "prezhde chem" ("before"), "poka" ("while"), "kogda" ("when") for temporal 
conjuctions, etc.)  

#204    (declarative clause: comp noun verb)
Source: a, e
Vetted: t
Judgment: g
Phenomena: embedded clauses
Ivan      dumaet          chto    sobaka      spit.   
Ivan      duma-et   	  chto    sobak-a     sp-it.        
Ivan.NOM  think-3SG.PRS   that    dog-SG.NOM  sleep-3SG.PRS  
Ivan thinks that the dog is sleeping.

The ungrammatical example below shows that the complementizer in embedded declarative clause is obligatory.

#205    (declarative clause: *noun verb)
Source: a, e
Vetted: t
Judgment: u
Phenomena: embedded clauses
Ivan      dumaet          sobaka      spit.   
Ivan      duma-et   	  sobak-a     sp-it.        
Ivan.NOM  think-3SG.PRS   dog-SG.NOM  sleep-3SG.PRS  
intended: Ivan thinks (that) the dog is sleeping.  

We have the following examples of different word order in the embedded clauses in our testsuite:
(declarative clause: comp verb noun)
(declarative clause: comp noun-nom tr-verb noun-acc)
(declarative clause: comp noun-acc tr-verb noun-nom)(declarative clause: comp tr-verb noun-nom noun-acc)
(declarative clause: comp tr-verb noun-acc noun-nom)


//EMBEDDED INTERROGATIVE CLAUSES: INTERROGATIVE PARTICLE "LI" AND WORD ORDER
Embedded yes-no questions use the c (also referred to as particle in other sources), 
which is placed second in the clause, usually after the verb (Cubberley 237).  Data gathered from 
3 native speakers shows that the interrogative particle "li" can occupy the third position in the 
sentence as well, but as with the matrix questions, the further it moves from the second position, 
the more questionable its grammaticality becomes, therefore we are treating any position besides the 
second position in the embedded sentences as ungrammatical for the interrogative particle "li".  


#212    (interrogative clause: tr-verb li noun-nom noun-acc)
Source: a, e
Vetted: t
Judgment: g
Phenomena: embedded clauses
Ivan      sprosil           kupila		li		devochka	knigu.   
Ivan      sprosi-l   	    kupi-l-a		li		devochk-a	knig-u    
Ivan.NOM  ask-3SG.MASC.PST  buy-3SG.FEM.PST	INTERROGATIVE	girl-FEM.3SG	book-3SG.ACC	 	 
Ivan asked whether the girl bought a book.

#213    (interrogative clause: noun-acc   li   tr-verb   noun-nom)
Source: a, e
Vetted: t
Judgment: g
Phenomena: embedded clauses
Ivan      sprosil           knigu		li		kupila		devochka.   
Ivan      sprosi-l   	    knig-u		li		kupi-l-a	devochk-a   
Ivan.NOM  ask-3SG.MASC.PST  book-3SG.ACC	INTERROGATIVE	buy-3SG.FEM.PST	girl-FEM.3SG	 	 
Ivan asked whether the girl bought a book.


We have several un grammatical examples by placing "li" in the non-licensed position.  
(interrogative clause: noun-nom   li  tr-verb   noun-acc)
(interrogative clause: *tr-verb chto li noun-nom noun-acc)
(interrogative clause: ??tr-verb noun-nom li noun-acc )
(interrogative clause: *tr-verb noun-nom noun-acc li)




						//NON-VERBAL PREDICATES

The "simplest form of the copula" in Russian language is the verb "bytj" ("be") and it can serve 
different functions.  It is not used in Present Tense, when it serves a locational function 
("na stole kniga" (the book is on the table).  The only case when "bytj" is used in the Present 
Tense is when it the existential sense, as in "na stole estj kniga" ("there is a book on the table"). 
The copula "bytj" is ised in the Future Tense and the Past Tense obligatorily.  There are also two 
additional copular verbs that are equivalent to "bytj", i.e. 1). "byvatj" (means "to be" but with an 
additional frequentative meaning), as in "Ivan byvaet v Moskve"; or 2). "javljatjsja" (used in formal 
written style), as in "Ivan javljaetsja doktorom".  We are not looking at these two copular verbs, as 
they are less common.


#220    (noun noun)
Source: a, e
Vetted: t
Judgment: g
Phenomena: non-verbal predicates
Ivan		doktor. 
Ivan		doktor		 
Ivan.NOM  	doctor.NOM.SG.MASC	 	 
Ivan is a doctor.    

#221    (*noun be.PRS noun)
Source: a, e
Vetted: t
Judgment: u
Phenomena: non-verbal predicates
Ivan		estj		doktor. 
Ivan		estj		doktor		 
Ivan.NOM  	be.3SG.PRS	doctor.NOM.SG.MASC	 	 
intended: Ivan is a doctor. 

#222    (noun be.PST noun)
Source: a, e
Vetted: t
Judgment: g
Phenomena: non-verbal predicates
Ivan		byl		doktorom. 
Ivan		byl		doktor-om		 
Ivan.NOM  	be.3SG.MASC.PST	doctor-SG.MASC.INST	 
Ivan was a doctor.  

#223    (noun be.FUT noun)
Source: a, e
Vetted: t
Judgment: g
Phenomena: non-verbal predicates
Ivan		budet		doktorom. 
Ivan		budet		doktor-om		 
Ivan.NOM  	be.3SG.FUT	doctor-SG.MASC.INST	 
Ivan will be a doctor.  
  
We have similar testsuite examples for the Adjectives:
(noun adj)
(*noun be.PRS adj)
(noun be.PST adj)
(noun be.FUT adj)


						INFORMATION STRUCTURE

The word order is one of the most important markers of the information structure in Russian language.  
Thus, we are repeating the examples from our word section, with a more detailed explanation. 

The emphatic stress can be used on the words as well in order to make this word more "informative" 
than the others.  The word order nevertheless is preserved, retaining its usual functions 
(Timberlake 455).

We have testsuite examples for SV, SVO:

The most neutral word order in Russian is SV and SVO (i.e. when the subject precedes the verb).  

#228   SV
Source: a, e
Vetted: t
Judgment: g
Phenomena: information structure
Ivan spit.
Ivan sp-it
Ivan sleep-3SG.PRS
Ivan is sleeping.

#229 SVO 
Source: a, e
Vetted: t
Judgment: g
Phenomena: information structure
Ivan kupil        sobaku.
Ivan kupi-l       sobak-u
Ivan buy-3SG.PST  dog-ACC.SG
Ivan bought a dog.



We have testsuite example for OVS:
The order OVS, per Timberlake, can usually serve two functions.  It: 
1). establishes a relationship between a known entity (object) and an abstract condition (subject).
2). introduces a new event, which is an interaction between the known object and a new individual.  

Timberlake bases his findings on corpus consisting of 359 examples.  There seem to be other possibilities, 
though, for using the OVS order.  The example we are using in the testsuite seems to belong to neither 
of these categories, but it demonstrates the OVS order (and it is also judged as grammatical by 4 native 
speakers, although the reason for that might be an addition of emphatic stress, which in combination with 
this word order makes this sentence grammatical). 

#230  OVS
Source: a, e
Vetted: t
Judgment: g
Phenomena: information structure
Knigu       kupil       Ivan.
Knig-u      kupi-l      Ivan
Book-ACC.SG buy-3SG.PST Ivan
Ivan bought a book.


We have testsuite examples for: SOV, OSV
The orders SOV and OSV are used to establish the entities as bases (topics), with the predicate 
(as focus) stating the relation among them. SOV order links the current predicate to surrounding 
text through the effect on the object.  OSV is used to contrast this particular entity (which is 
the object in this case) to the others (Timberlake 452).  

#231 SOV, the object is a weak basis
Source: a, e
Vetted: t
Judgment: g
Phenomena: information structure
Ivan sobaku      kupil.
Ivan sobak-u     kupi-l
Ivan dog-ACC.3SG buy-3SG.PST
Ivan bought a dog.

In our testsuite example with OSV, the object is a strong basis:

#232 OSV, the object is a strong basis
Source: a, e
Vetted: t
Judgment: g
Phenomena: information structure
Sobaku     Ivan kupil.
Sobak-u    Ivan kupi-l
Dog-ACC.SG Ivan buy-3SG.PST
Ivan bought a dog.

We also have the testsuite examples for VS(X), VSO, VOS: 
-VS(O)X allows the speaker to insert a strong focus and answer an implicit question (Timberlake 453).  
-VOS introduces a new scene involving the object and then an unknown subject (as a strong focus). 






Part 3:  Documentation of the coverage of our grammar over the testsuite.

SUMMARY OF RESULTS FROM TSDB++

For coverage we have 46.1%, which is up from 29% coverage that we had for Lab 3.  

We have 5 wellformed/unanalyzed sentences (three of which were due to the position of "li" particle, discussed
in the question formation section). For wellformed and unanalyzed examples we also get:
#43 (Ivan kupil sobake knigu) (i-category: c).  We think that this sentence didn't parse because it has Dative case which we haven't
dealt with yet.  
#56 (Sobaka jeta spit) (i-category: det). We are not yet clear on why it's not parsing.  

For illformed but parsed results we get:
#44 ("Ivan kupil sobaka knigu")   
We think the reason for #44 not-parsing might be the fact that we haven't analyzed the Dative case.
#166 (Sobaku kupil Ivan li?") parses because the interrogative particle "li" is placed in the sentence-final position, 
which is licensed by Grammar Customization system while in fact it is ungrammatical in Russian.  






Part 4: documentation of changes we made to our grammar to improve its performance.  

Lexicon
We made a change in the lexicon to remove one noun-slot which was duplicating a rule, causing inamimate fem 
inanimate nouns to generate two parses. We firs set up a noun inflection slot for feminine nouns and chose both 
animate and inamimate feminine noun types as input. We then created a separate noun slot to inflect (with the 
same morphemes) only inanimate feminine nouns. We corrected this mistake in Lab 4.

Also in Lab 4, we took a closer look at our test suite to see why some grammatical sentences were not parsing. 
About six sentences were marked as ungrammatical, but were being passed in lkb. We noticed that the error was 
in the test suite, where the example in transliteration did not match the example in the morpheme boundary line.
Easy fix and our coverage improved (currently at over 40%).

