LAB 6
Writeup
Gracheva, Suskic



							MATRIX YES-NO QUESTIONS


//INTONATION QUESTIONS

Intonation-only questions are covered in our grammar.  They parse and have a correct INDEX.SF value.
For example, if we parse sentence #160 from testsuite.txt, the INDEX.SF value is the prop-or-ques, which is a correct value 
for this feature, because in Russian the matrix yes-no questions are formed by intonation.  




//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.  


We followed the instructions for constraining the position of the clitics in the interrogative sentences, as described in Parts 1
and 2.  Part 1 allowed us to now parse correctly interrogative matrix sentences with particle "li".  Part 2 is still in progress.
We need to find out what causes the inability of grammar to parse any sentences.  





Part 1


We added a feature that checks the position in the sentence.  

canonical-synsem :+
 [ L-PERIPH bool ].




We then added the constraint on all binary rules: i.e. that it copies up the L-PERIPH value from its left-hand 
daughter and constraints the right-hand DTR to be compatible with [L-PERIPH -].

basic-binary-phrase :+
 [ SYNSEM.L-PERIPH #periph,
   ARGS < [ SYNSEM.L-PERIPH #periph ], [ SYNSEM.L-PERIPH - ] > ].




We then  made a type for the unary phrases that also has the constraint of copying up the L-PERIPH.  

same-periph-unary-phrase := unary-phrase &
 [ SYNSEM.L-PERIPH #periph,
   ARGS < [ SYNSEM.L-PERIPH #periph ] > ].





We then added the same-periph-unary-phrase as a subtype for the following phrases:

bare-np-phrase := basic-bare-np-phrase & same-periph-unary-phrase &
  [ C-CONT.RELS <! [ PRED "exist_q_rel" ] !>,
    SYNSEM.LIGHT - ].

basic-head-opt-comp-phrase :+
 [ SYNSEM.L-PERIPH #periph,
   ARGS < [ SYNSEM.L-PERIPH #periph ] > ].

basic-head-opt-subj-phrase :+
 [ SYNSEM.L-PERIPH #periph,
   ARGS < [ SYNSEM.L-PERIPH #periph ] > ].



We then created a lexical type for the "li" clitic. We could also assign values to the MOD feature, thus specifying which 
constituents the "li" can attach to.  This would exclude the negative adverbs, prepositions, or wh-words and it would
prevent us from generating ungrammatical sentences like "in   li   store  did   he   buy   a   book" .  However, this 
would not yet solve our problem in which "li"attaches sometimes to the second word in a larger constituent and thus takes the
third position in the sentences.  If there was a way to group neg adverbs and prepositions and create a subtype for them 
that would disallow licensing appearance of the "li" after them, that would be very useful for Russian.  


question-clitic-lex := no-hcons-lex-item  &
 [ SYNSEM.LOCAL [ CAT [ VAL [ SPR < >, COMPS < >, SUBJ < >, SPEC < >],
                                        HEAD adv &
                                                [ MOD < [ LIGHT +,
                                                          L-PERIPH +,
                              LOCAL intersective-mod ] > ]],
                             CONT.RELS <! !> ]].



We also had to make a change in the basic-head-mod-phrase-simple identifying the L-PERIPH values:

basic-head-mod-phrase-simple :+ [ SYNSEM.LOCAL.CAT.MC #mc,
    NON-HEAD-DTR.SYNSEM.LOCAL.CAT [ MC #mc,
                        HEAD.MOD < [ L-PERIPH #periph ] > ],
    HEAD-DTR.SYNSEM.L-PERIPH #periph ].



Now we get only one parse for sentence:  

 #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?

Coverage: in LKB sentence #231 is now parsed as grammatical, but in itsdb it is a "wellformed" non parsed 
sentence.  





Part 2
NOTE: THIS IS A PART MOST OF WHICH WE HAD TO TAKE OUT FOR SUBMITTING THIS ASSIGNMENT BECAUSE THEY CAUSED THE 
RUNAWAY ERROR.  WE ARE WORKING ON FIXING THAT, SO THAT OUR "LI"-CLAUSES HAVE CORRECT SEMANTICS.  

We added a new non-local feature YNQ, which is accumulated from both daughters.  


non-local :+
 [ YNQ 0-1-dlist ].




We specified that all words except interrogative "li" should have an empty YNQ value (since we also had a 
redundancy error when loading the grammar, we had to have insert the SYNSEM.  

;;; All words except question clitic should have an empty YNQ value

non-ynq-word := lex-item &
 [ SYNSEM.NON-LOCAL.YNQ 0-dlist ].

basic-zero-arg :+ [ SYNSEM.NON-LOCAL.YNQ 0-dlist ].
basic-one-arg :+ [ SYNSEM.NON-LOCAL.YNQ 0-dlist ].
basic-two-arg :+ [ SYNSEM.NON-LOCAL.YNQ 0-dlist ].
basic-three-arg :+ [ SYNSEM.NON-LOCAL.YNQ 0-dlist ].
intersective-mod-lex :+ [ SYNSEM.NON-LOCAL.YNQ 0-dlist ].



We constrained binary phrases to gather up YNQ from both daughters:

basic-binary-phrase :+
 [ SYNSEM [ L-PERIPH #periph,
           NON-LOCAL.YNQ [ LIST #list, LAST #last ] ],
   ARGS < [ SYNSEM [ L-PERIPH #periph,
                    NON-LOCAL.YNQ [ LIST #list, LAST #middle ] ] ],
          [ SYNSEM [ L-PERIPH -,
                 NON-LOCAL.YNQ [ LIST #middle, LAST #last ] ] ] > ].




And then we constrained some unary phrases to copy up YNQ:

same-ynq-unary-phrase := unary-phrase &
 [ SYNSEM.NON-LOCAL.YNQ #ynq,
   ARGS < [ SYNSEM.NON-LOCAL.YNQ #ynq ] > ].




Then we added same-ynq-unary-phrase as a supertype to bare-np, opt-subj and opt-comp:

bare-np-phrase := basic-bare-np-phrase & same-periph-unary-phrase & same-ynq-unary-phrase &
  [ C-CONT.RELS <! [ PRED "exist_q_rel" ] !>,
    SYNSEM.LIGHT - ].

basic-head-opt-subj-phrase :+ same-periph-unary-phrase & 
 [ SYNSEM.L-PERIPH #periph,
   ARGS < [ SYNSEM.L-PERIPH #periph ] > ].



!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

and we wrote this in russian.tdl:  (adding the supertype after the ":+" seemed kind of strange, hope that's ok!)

basic-head-opt-comp-phrase :+ same-periph-unary-phrase & 
 [ SYNSEM.L-PERIPH #periph,
   ARGS < [ SYNSEM.L-PERIPH #periph ] > ].

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!


Then we constrained the question clitic "li" to have a non-empty YNQ value:


question-clitic-lex := no-hcons-lex-item  &
 [ SYNSEM [ LOCAL [ CAT [ VAL [ SPR < >, COMPS < >, SUBJ < >, SPEC < >],
                                        HEAD adv &
                                                [ MOD < [ LIGHT +,
                                                          L-PERIPH +,
                              LOCAL intersective-mod ] > ]],
                             CONT.RELS <! !> ],
                 NON-LOCAL.YNQ <! *top* ! > ] ].




Then we constrained the root symbol to require an empty YNQ value:

root := phrase &
  [ SYNSEM [ LOCAL [ COORD -,
                   CAT [ VAL [ SUBJ < >,
                               COMPS < > ],
                         MC +,
                         HEAD +vc &
                              [ FORM finite ] ] ],
         NON-LOCAL.YNQ 0-dlist ] ].




We created non-branching int-cl and decl-cl types (and associated rule instances) in russian.tdl:


int-cl := head-only & interrogative-clause &
  [ SYNSEM [ LOCAL.CAT [ VAL #val,
                         MC bool ],
             NON-LOCAL.YNQ <! !> ],
    HEAD-DTR.SYNSEM [ LOCAL.CAT [ MC na,
                                            VAL #val ],
                                NON-LOCAL.YNQ <! *top* !> ]].

decl-cl := head-only & declarative-clause & same-ynq-unary-phrase &
  [ SYNSEM.LOCAL.CAT [ VAL #val,
                       MC bool ],
    HEAD-DTR.SYNSEM [ LOCAL.CAT [ MC na,
                                  VAL #val ],
                      NON-LOCAL.YNQ 0-dlist ]]




In rules.tdl we created these instances:

int-cl := int-cl.
decl-cl := decl-cl.  




We constrained other headed phrases to produce MC na mothers and
take MC na head daughters.

mc-na-headed-phrase := headed-phrase &
  [ SYNSEM.LOCAL.CAT.MC na,
    HEAD-DTR.SYNSEM.LOCAL.CAT.MC na ].


??????????????????????????????????????????????????

However, after writing this:

        binary-headed-phrase :+ mc-na-headed-phrase.

we don't get any parses any longer and I am not sure what it's the problem.  When I parse "Ivan spit" ("Ivan 
sleeps"), I don't get a parse, but there is no unification error at the highest level "S" licensed by Head-Subj
Rule.  

??????????????????????????????????????????????????




We also added these constraints:



bare-np-phrase := basic-bare-np-phrase & same-periph-unary-phrase & mc-na-headed-phrase &
  [ C-CONT.RELS <! [ PRED "exist_q_rel" ] !> ].

basic-head-opt-comp-phrase :+ same-periph-unary-phrase & mc-na-headed-phrase &
 [ SYNSEM.L-PERIPH #periph,
   ARGS < [ SYNSEM.L-PERIPH #periph ] > ].

basic-head-opt-subj-phrase :+ same-periph-unary-phrase & mc-na-headed-phrase & 
 [ SYNSEM.L-PERIPH #periph,
   ARGS < [ SYNSEM.L-PERIPH #periph ] > ].



We then added mc-na-headed-phrase to supertypes for bare-np, opt-subj, and opt-comp. 

bare-np-phrase := basic-bare-np-phrase & same-periph-unary-phrase & same-ynq-unary-phrase & mc-na-headed-phrase &
  [ C-CONT.RELS <! [ PRED "exist_q_rel" ] !>,
    SYNSEM.LIGHT - ].

basic-head-opt-comp-phrase := same-ynq-unary-phrase & mc-na-headed-phrase &
 [ SYNSEM.L-PERIPH #periph,
   ARGS < [ SYNSEM.L-PERIPH #periph ] > ].

head-opt-subj-phrase := basic-head-opt-subj-phrase & same-ynq-unary-phrase & mc-na-headed-phrase.





Then we constrained the complementizers and clause-embedding verbs to take [MC -] complements.


decl-embed-clausal-verb-lex := embed-clausal-verb-lex &
   [ SYNSEM.LOCAL [ CAT.VAL.COMPS < [ LOCAL [ CAT.MC -, 
                               CONT.HOOK.INDEX.SF prop ] ] > ] ].

interrogative-embed-clausal-verb-lex := embed-clausal-verb-lex &
   [ SYNSEM.LOCAL [ CAT.VAL.COMPS < [ LOCAL [ CAT.MC -, 
                               CONT.HOOK.INDEX.SF ques ] ] > ] ].

complementizer-lex-item := raise-sem-lex-item & basic-one-arg &
  [ SYNSEM.LOCAL.CAT [ HEAD comp &
                            [ MOD < > ],
                       VAL [ SPR < >,
                             SUBJ < >,
                             COMPS < #comp > ] ],
    ARG-ST < #comp &
             [ LOCAL.CAT [ MC -,
                           HEAD verb,
                           VAL [ SUBJ < >,
                                 COMPS < > ] ] ] > ].




RESULTS: if we comment out this line

 binary-headed-phrase :+ mc-na-headed-phrase.

we still get parses, but more than after completion of Part 2, i.e. now we get two parses for sentence:  

#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?

In first parse "li" attaches to preceding N "Ivan" and in the second parse "li" attaches to the 
following V "kupil". 

And if we add this line in

 binary-headed-phrase :+ mc-na-headed-phrase.

Then we don't have any parses, even for sentence like "Ivan spit".  However, examining highest S licensed
by Head-Subject Rule does not give the unification errors, so it's very unclear what is going on, we are 
looking into that.   


							EMBEDDED CLAUSES


1.  We created clause-embedding verbs in russian.tdl and two subtypes for declarative and interrogative clauses:

embedded-clausal-verb-lex := verb-lex & clausal-second-arg-trans-lex-item &
   [ SYNSEM.LOCAL [ CAT.VAL.COMPS < [ LOCAL [ CAT.HEAD comp,
                                     CONT.HOOK.INDEX.SF prop-or-ques ] ] > ] ].

declarative-embedded-clausal-verb-lex := embedded-clausal-verb-lex &
   [ SYNSEM.LOCAL [ CAT.VAL.COMPS < [ LOCAL.CONT.HOOK.INDEX.SF prop ] > ] ].

interrogative-embedded-clausal-verb-lex := embedded-clausal-verb-lex &
    [ SYNSEM.LOCAL [ CAT.VAL.COMPS < [ LOCAL.CONT.HOOK.INDEX.SF ques ] > ] ].




2.  We also had to make the following change (before we had the [HEAD +nvjrpdmo] constraint on the head-comp-phrase, which 
was no longer the correct choice.  

head-comp-phrase := basic-head-1st-comp-phrase & head-initial-head-nexus.

comp-head-phrase := basic-head-1st-comp-phrase & head-final-head-nexus &
  [ SYNSEM.LOCAL.CAT [ HEAD +nvjrpdmo & 
                               [ AUX - ] ] ].



3.  We created three more verbs in the lexicon.tdl, since we didn't have any clause-embedding verbs before:

duma := declarative-1-conj-imp-clausal-verb-lex &
  [ STEM < "duma" >,
    SYNSEM.LKEYS.KEYREL.PRED "_think_v_rel" ].

sprashiva := interrogative-1-conj-imp-clausal-verb-lex &
  [ STEM < "sprashiva" >,
    SYNSEM.LKEYS.KEYREL.PRED "_ask_v_rel" ].

spros := interrogative-2-conj-perf-clausal-verb-lex &
  [ STEM < "spros" >,
    SYNSEM.LKEYS.KEYREL.PRED "_ask_v_rel" ].



4.  And we changed the conjugation for for the lexemes they were inheriting from, in the russian.tdl:

declarative-1-conj-imp-clausal-verb-lex := past-regular-rule-dtr & infinitive-rule-dtr & declarative-embedded-clausal-verb-lex &
  [ INFLECTED -,
    SYNSEM.LOCAL.CONT.HOOK.INDEX.E.ASPECT imperfective ].


interrogative-1-conj-imp-clausal-verb-lex := past-regular-rule-dtr & infinitive-rule-dtr & interrogative-embedded-clausal-verb-lex &
  [ INFLECTED -,
    SYNSEM.LOCAL.CONT.HOOK.INDEX.E.ASPECT imperfective ].

interrogative-2-conj-perf-clausal-verb-lex := past-regular-rule-dtr & suffix-add-i-rule-dtr & infinitive-rule-dtr & present-1-conj-2-rule-dtr & interrogative-embedded-clausal-verb-lex &
  [ INFLECTED -,
    SYNSEM.LOCAL.CONT.HOOK.INDEX.E.ASPECT perfective ].




5.  We added several sentences in our testsuite, from which #232, 233, 236 now parse correctly because we cover the embedded
declarative sentences.

#232   decl-emb-v COMP decl-emb-clause
Source: a, e
Vetted: t
Judgment: g
Phenomena: embedded questions
Ja       dumal         chto   Ivan     kupil        sobaku.
Ja       duma-l        chto   Ivan     kupi-l       sobak-u
1.SG     think-1SG.PST that   Ivan.NOM buy-3SG.PST  dog-ACC    
I thought that Ivan bought a dog.

#233   *decl-emb-v decl-emb-clause
Source: a, e
Vetted: t
Judgment: u
Phenomena: embedded questions
Ja       dumal         Ivan     kupil        sobaku.
Ja       duma-l        Ivan     kupi-l       sobak-u
1.SG     think-1SG.PST Ivan.NOM buy-3SG.PST  dog-ACC    
intended: I thought that Ivan bought a dog.

#236   decl-emb-v COMP decl-emb-clause
Source: a, e
Vetted: t
Judgment: g
Phenomena: embedded questions
Ja       dumal         chto   Ivan     spit.
Ja       duma-l        chto   Ivan     sp-it
1.SG     think-1SG.PST that   Ivan.NOM sleep-3SG.PRS    
I thought that Ivan was sleeping.




6.  The interrogative embedded clauses still remain a problem because we haven't solved the "li"-questions in matrix yet.  
Sentences 234 and 235 do not yet parse correctly.  Also, the sentence #234 does not parse now because 
embedded-clausal-verb-lex requires HEAD comp and we are not treating interrogative particle "li" as a comp anymore 
(it is now a question-clitic-lex, which we have done in order to have coverage over the matrix "li"-interrogative clauses).  
We have to figure out then how to place the question-clitic-lex instead of the complementizer-lex into interrogative
embedded clauses.  

#234   interrogative-embedded clause: v  li  subj obj
Source: a, e
Vetted: t
Judgment: g
Phenomena: embedded questions
Ja       sprosil           kupil        li   Ivan        sobaku.
Ja       sprosi-l          kupi-l       li   Ivan        sobak-u
1.SG     ask-PERF.1SG.PST  buy-3SG.PST  Q    Ivan.NOM    dog-ACC    
I asked whether Ivan bought a dog.

#235   interrogative-embedded clause: *v subj obj
Source: a, e
Vetted: t
Judgment: u
Phenomena: embedded questions
Ja       sprosil            kupil         Ivan        sobaku.
Ja       sprosi-l           kupi-l        Ivan        sobak-u
1.SG     ask-PERF.1SG.PST   buy-3SG.PST   Ivan.NOM    dog-ACC    
intended: I asked whether Ivan bought a dog.







=============================================================

THE FEATURE MC

Russian matrix and embedded clauses have different syntactic properties (i.e. presence of particle "li" is required in 
interrogative embedded clauses and presence of complementizer "chto" is required in declarative embedded clauses).  
So, we had to set the MC feature in root condition, the embedding verb, and the constructions/lexical rules/etc (inside CAT).  

1. The root condition in the roots.tdl was already constrained to have [MC+]:


root := phrase &
  [ SYNSEM.LOCAL [ COORD -,
                   CAT [ VAL [ SUBJ < >,
                               COMPS < > ],
                         MC +,
                         HEAD +vc &
                              [ FORM finite ] ] ] ].


2. We added [MC-] constraint on the embedding verbs into the russian.tdl:

embed-clausal-verb-lex := main-verb-lex & clausal-second-arg-trans-lex-item &
    [ ARG-ST < [ LOCAL.CAT.HEAD noun ],
             #comps &
             [ LOCAL.CAT [ VAL [ SPR < >,
                                   COMPS < > ],
                   HEAD comp ] ] >,
      SYNSEM.LOCAL [ CAT.VAL.COMPS < #comps >,
                     CONT.HOOK.INDEX.SF prop-or-ques ] & 
             [ CAT.MC - ] ].

Actually that seemed to cover what we need because we already get the correct MRSs for the verb "duma" ("think").
For example, in the attached image for sentence 236 the ARG2 position of _think_v_rel (h8) is linked to the LBL of 
_sleep_v_rel (h14) and the SF value of e15 is PROP.  


#236   decl-emb-v COMP decl-emb-clause
Source: a, e
Vetted: t
Judgment: g
Phenomena: embedded questions
Ja       dumal         chto   Ivan     spit.
Ja       duma-l        chto   Ivan     sp-it
1.SG     think-1SG.PST that   Ivan.NOM sleep-3SG.PRS    
I thought that Ivan was sleeping.



(SEE IMAGE: "CheckYourMRSs" for more info)


And we also get the [MC luk] when we parse sentences like #232, which was already discussed on the GoPost and we don't need 
to do anything here.


                                                                     
     

								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.




Adjective Predicates

As mentioned above, in Russian, for Present tense, there is no copula while for past and future it is a form of verb "be".


Example #1: Copulaless AP

For this example, we created a lexical rule for making a predicative adjective from a atttributive one:

attributive-adj-to-predicative-adj-lex-rule := cat-change-only-lex-rule &
 [ DTR hard-adjective-1-lex-rule & 
           [ SYNSEM [ LKEYS.KEYREL.ARG0 #ev, 
                      LOCAL.CAT.HEAD.MOD < [ LOCAL.CONT.HOOK.INDEX #ind ] > ]],
   	     SYNSEM.LOCAL [ CAT [ HEAD adj & [ PRD +, MOD < > ],
                            VAL [ SUBJ < [ LOCAL [ CONT.HOOK.INDEX #ind, CAT [ HEAD noun, VAL.SPR < > ]]] >,
				  SPR < >,
				  SPEC < >,
				  COMPS < > ]],
                            CONT.HOOK [ INDEX #ev, XARG #ind ]]].

The above rule constrins the mother of the rule as HEAD adj, PRD + (a new boolean feature we added to HEAD to differentiate 
attributive and predicative adjectives) and MOD < >. Furthermore, the rule constrains the SUBJ of the mother to HEAD noun and 
SPR < > (the last not allowing the attributive adjective to form with nouns and thus reducing overgeneration.

We also identified the INDEX of the mother with its own ARG0 and the index of the daughter's MOD

In order to inflect the attributive adjective, we added a supertype lex rule which would then ensure that the attributive 
adjectives as well as the predicative adjectives went through the same inflection and respective constraints on the SUBJ.
                            CONT.HOOK [ INDEX #ev, XARG #ind ]]].

Finally, we added a root condition which would allow sentences such as the first example below:

root-ap := phrase &
  [ SYNSEM.LOCAL [ COORD -,
                   CAT [ VAL [ SUBJ < >,
                               COMPS < > ],
                         HEAD +vjp & [ PRD +, MOD < > ] ] ] ].
  
IGT Example:

Kniga			boljshaja. 
Knig-a			boljsh-aja		 
Book-FEM.SG.NOM  	large-FEM.SG.NOM 	 
The book is large.





Example #2: AP in past tense

For this example, we added a copula-lex-item:

copula-verb-lex := verb-lex & trans-first-arg-raising-lex-item-2 &
  [ SYNSEM.LOCAL [ CAT.VAL [ SUBJ < #subj >,
                             COMPS < #comps >,
                             SPR < >,
                             SPEC < > ],
                   CONT.HOOK.XARG #xarg ],
    ARG-ST < #subj &
             [ LOCAL [ CONT.HOOK.INDEX #xarg,
                       CAT [ VAL [ SPR < >,
                                   COMPS < > ],
                             HEAD noun ] ] ],
             #comps &
             [ LOCAL.CAT [ VAL [ COMPS < > ],
                           HEAD +jp ] ] > ].

And:

by-copula-verb-lex := copula-verb-lex & past-regular-rule-dtr.

We also created a past-regular-rule-dtr so we could inflect this copula in the same way as the regular verbs.

Finally we created a lexical item:

by := by-copula-verb-lex &
  [ STEM < "by" >,
    SYNSEM.LKEYS.KEYREL.PRED "_be_v_id_rel" ].

IGT example:

Kniga			byla		boljshaja. 
Knig-a			byl-a		boljsh-aja		 
Book-FEM.SG.NOM  	be-3SG.FEM.PST	large-FEM.SG.NOM 	 
The book was large.




Example #3

Same as example #2 where we created a different copula item (as well as a lex rule daughter supertype to allow for inflection:

bud-copula-verb-lex := copula-verb-lex & future-be-lex-rule-dtr.

We then added an instance in lexicon:

bud-copula := bud-copula-verb-lex &
  [ STEM < "bud" >,
    SYNSEM.LKEYS.KEYREL.PRED "_be_v_id_rel" ].

IGT Example

Kniga			budet		boljshaja. 
Knig-a			bud-et		boljsh-aja		 
Book-FEM.SG.NOM  	be-3SG.FUT	large-FEM.SG.NOM 




Prepositional predicates

For this phenomena, we created an adposition lexical type:

location-adposition-lex := basic-int-mod-adposition-lex &
	[ INFLECTED +,
	  SYNSEM.LOCAL.CAT [ HEAD [ PRD + ],
			     VAL [ COMPS.FIRST.LOCAL [ CAT.HEAD noun & [ CASE pre ]],
				 SUBJ.FIRST.LOCAL.CAT [ HEAD noun & [ CASE nom ],
						        VAL [ SPR < >,
							      COMPS < >,
							      SUBJ < > ]]]]].

This type is restricted to HEAD.PRD + and the first item on its COMPS list to HEAD noun and CASE pre. On the VAL feature, we 
restricted the first item on the SUBJ list to HEAD noun and CASE nom and the rest of VAL is empty.

We were not able to parse the below example and had trouble forming the PP via the HEAD-COMPS rule. The unification error seemed
to indicate a problem with INFLECTED feature, but we were not able to replicate succesfully. We did get the edge limit exhausted 
error consistently.

IGT Example:

Kniga			na	stole. 
Knig-a			na	stol-e  
book.FEM.SG.NOM  	on	table-SG.MASC.PREP 	 	 
The book is on the table.  






Noun Predicates

We did not have time to implement the noun predicates to cover the below examples:

Ivan		doktor. 
Ivan		doktor		 
Ivan.NOM  	doctor.NOM.SG.MASC	 	 
Ivan is a doctor.    


Ivan		byl		doktorom. 
Ivan		byl		doktor-om		 
Ivan.NOM  	be.3SG.MASC.PST	doctor-SG.MASC.INST	 
Ivan was a doctor.  


Ivan		budet		doktorom. 
Ivan		bud-et		doktor-om		 
Ivan.NOM  	be-3SG.FUT	doctor-SG.MASC.INST	 
Ivan will be a doctor.  






						CHANGES FROM LAB 5:


1.  We had trouble parsing the sentence:

Jeti                devochki           chitajut         boljshimi 		knigi.
Jet-i               devochk-i          chita-ju-t       boljsh-imi 		knig-i 
These-FEM.3PL.NOM   girl-FEM.3PL.NOM   read-3PL-PRS     big-FEM.3PL.INST        book-FEM.3PL.ACC 
These girls are reading large books.

And were instructed that there is a missing relation in the semantics. We fixed that by changing the demontrative lex type to say:


demonstrative-determiner-lex := determiner-lex-supertype &
[ SYNSEM [ LOCAL [ CAT.VAL.SPEC.FIRST.LOCAL.CONT.HOOK [ INDEX #ind, LTOP #ltop ],
                   CONT.RELS. <! [], #altkeyrel !> ],
                   LKEYS.ALTKEYREL #altkeyrel & [ ARG1 #ind, LBL #ltop ]]].

instead of:

demonstrative-determiner-lex := determiner-lex-supertype &
[ SYNSEM [ LOCAL [ CAT.VAL.SPEC.FIRST.LOCAL.CONT.HOOK [ INDEX #ind, LTOP #ltop ],
                   CONT.RELS.LIST.REST #altkeyrel ],
                   LKEYS.ALTKEYREL #altkeyrel & [ ARG1 #ind, LBL #ltop ]]].

And that worked. However, after working on the lab 6 grammar, the sentence no longer parses. It looks like the root condition for
AP is licensing te boljshimi knigi and the head-comp phrase is unable to combine the jeti devochki chitajut V with the bolhshimi
knigi AP. when we take out the adjective it seems to work. We dug around in the unification results and did not get a unification
error when dragging the ADJ node to the second argument of the HEAD-COMPS rule.


2.  We added several grammatical and ungrammatical examples for embedded clauses in our testsuite.

3.  We also created three new verbs in the lexicon.tdl for think.Imperfective, ask.Perfective and ask.Imperfective because
we didn't have any clause-embedding verbs before.  We had to update all rules in russian.tdl to get the right conjugations for
these verbs (as indicated in "Embedded clauses" section).  





					TESTSUITE COMPARISON (BASELINE COMPARISON)

Grammar Coverage Lab 5:
Analyze   |    Coverage 62.6%
Analyze   |    Overgeneration: 5.3%

After adjectives work in Lab 6 we had:
Analyze   |    Coverage 64.3%
Analyze   |    Overgeneration: 3.9%

After embedded clauses work in Lab 6 we had: 
Analyze   |    Coverage 66.9%
Analyze   |    Overgeneration: 2.6%







