DOCUMENTATION OF CUSTOMIZATION CHOICES AND UNHANDLED PROPERTIES
===============================================================

TENSE

Yiddish has three tenses, past, present and future. Present tense is
indicated through the use of a simple VP (i.e. no AUX) with the main
verb being inflected to agree with the subject for number and person
-- the agreement paradigm is presented under the agreement
section. Both past and future tenses are indicated through the use of
AUX + verb constructions. In the past tense, the auxiliary is the
finite form of one of the verbs hobn (to have) and (to be) -- which
again agree with the subject -- which combines with the past particle
of the matrix verb. The past participle is formed by adding the suffix
-e or -n to the verb, and depending on the verb, a prefix ge-. The
auxiliary used is selected for by the verb with hobn being more
frequent, and zayn tending to be selected by verbs which indicate
whole body movement, body state, or motion. The past tense
construction is illustrated below:

Source: {a:217}
Vetted: t
Judgment: g
Phenomena: {tam}
Ikh hob geshribn.
Ikh hob ge-shrayb-n.
1SG.NOM have;1SG PP-write-PP.
I have written.

Source: {a:217}
Vetted: t
Judgment: g
Phenomena: {tam}
Ikh bin geblibn.
Ikh bin ge-blayb-n.
1SG.NOM be;1SG PP-remain-PP.
I remained.

These show incorrect selection of hobn and zayn as auxiliaries and
incorrect agreement on the auxiliaries:

# Wrong auxiliary used
Source: author based on {a:217}
Vetted: t
Judgment: u
Phenomena: {tam}
Ikh bin geshribn.
Ikh bin ge-shrayb-n.
1SG.NOM have;1SG PP-write-PP.
I have written.

# Wrong auxiliary used
Source: author based on {a:217}
Vetted: t
Judgment: u
Phenomena: {tam}
Ikh hob geblibn.
Ikh hob ge-blayb-n.
1SG.NOM be;1SG PP-remain-PP.
I remained.

# Wrong form of the auxiliary
Source: author based on {a:217}
Vetted: t
Judgment: u
Phenomena: {agreement}
Ikh hot geshribn.
Ikh ho ge-shrayb-n.
1SG.NOM have;3SG PP-write-PP.
I have written.

# Wrong form of the auxiliary
Source: author based on {a:217}
Vetted: t
Judgment: u
Phenomena: {agreement}
Ikh iz geblibn.
Ikh iz ge-blayb-n.
1SG.NOM be;3SG PP-remain-PP.
I remained.

Additionally, there are also some verbs which take hobn as an
auxiliary in the present tense. In these cases both verbs agree with
the subject:

Source: {a:217}
Vetted: t
Judgment: g
Phenomena: {tam}
Ikh hob lib.
Ikh hob lib.
1SG.NOM have;1SG like;1SG.
I liked.

These verbs also take gehat, the past participle of hobn in the past
tense:
 
Source: {a:217}
Vetted: t
Judgment: g
Phenomena: {tam}
Ikh hob lib gehat.
Ikh hob lib gehat.
1SG.NOM have;1SG like;1SG have;PP.
I liked.

The future tense is indicated through the finite form of the auxiliary
veln (will) + the infinitive form of the verb. This is illustrated
below:

Source: {a:218}
Vetted: t
Judgement: g
Phenomena: {tam}
Ikh vel shraybn.
Ikh vel shrayb-n.
1SG.NOM shall;1SG write-INF.
I shall write.

# Wrong conjugation of auxiliary
Source: author based on {a:218}
Vetted: t
Judgement: u
Phenomena: {tam}
Ikh velst shraybn.
Ikh vel-st shrayb-n.
1SG.NOM shall- write-INF.
I shall write.

We implemented tense thusly: 

Under the Tense, Aspect and Mood section of the questionnaire, the
default hierarchy of past, present and future was kept. Then under the
Syntax part of this section we expanded the form hierarchy by creating
the sub-type past-part, which inherits from finite.

For the tense constructions, three separate auxiliary types were
defined each for hobn, zayn, and veln, respectively. All of these were
specified to not contribute any predicate, since their function is
more grammatical, indicating tense. They also make the auxiliary's
form finite, and in the case of hobn and zayn, they specified the
tense of the auxiliary to be past, while zeln makes the auxiliary's
tense future. All were constrained such that the subject is an NP with
the case selected by the verbal complement. Hobn and zayn constrain
the complement's form to be past-part as per the past tense
construction, while veln constrains its to be nonfinite, as per the
future tense construction. We then inserted all the various inflected
forms of the auxiliaries into the relevant spelling fields. Since no
person/number constraints can be applied to these, these will
overgenerate (as discussed in the final section).

As the first two auxiliaries are both used in the same AUX +
past-participle construction to indicate the past tense, these two
auxiliary types were constrained in exactly the same way, however we
chose to keep them separate so that later on they can be selected for
by specific verbs (this limitation of the customization is also
discussed in the final section).

In order to derive the infinitive and past-participle forms of the
verb, we created new inflecting lex rules. We already had an
obligatory suffix position class (which now had to be excluded from
applying to auxiliaries) set up for finite present tense verbs, which
agree with person and number (the forms are: zero, -st, -t, -en,
-n). Since both the past participle and the infinitive form are also
indicated through a suffix attaching to the root (which both happen to
be either -e or -en), and these forms are all mutually exclusive, the
same position class was used for all three, with separate lex rules
set up for infinitive and past particple suffixes that have the same
suffixes but apply the relevant form constraint to the verb. It's not
hopelessly ambiguous though, as the past participle often takes the
prefix ge-, which was added through a prefix position class with an
inflecting lex rule and was also constrained so that the verb has past
participle form. 

Our grammar then appears to leave an ambiguous gap, where the verb has
a -n or -en suffix that is ambiguous between infinitive and past
participle interpretation when prefix ge- does not occur. This does
not occur all that often in Yiddish however, as the past participle
often -- but not always -- involves vowel mutation of the root (which
our morphological analyzer 'undoes').

We currently do not handle periphrastic verbs which take the AUX hobn
in the present tense (Ikh hob lib.) -- nor the past tense construction
of these verbs either. This could easily be modeled as an auxiliary
type, however, once again, the matrix does not support specific verbs
selecting specific auxiliaries, so we have for now left this
unimplemented.


ASPECT

Yiddish has a number of aspects, including (but not limited to)
perfective, inchoative, iterative, habitual, and singulative. We
focused on the perfective, inchoative and iterative aspects for this
lab.

The perfective aspect indicates that an action has been completed. It
is indicated by the use of either a form appearing as either a
particle appearing after the verb, or as a prefix attaching to the
verb. The particle/prefix is selected for by the verb, with the
selection varying across different dialects as well as whether it is a
particle or a prefix. Our reference grammar also makes things more
interesting by pointing out that phonologically the prefix is really a
separate word. 

Source: a:217
Vetted: f
Judgment: g
Phenomena: {tam}
Ikh shrayb on.
Ikh shrayb on.
1SG.NOM write;1SG upon.
I write to completion.

Source: {a:221}
Vetted: t
Judgment: g
Phenomena: {tam}
Der dales dergeyt di yorn.
Der dales der-gey-t di yorn.
NOM.M.SG poverty PFV-go-3SG the year.
The poverty arrived this year.

It also seems to be that in the past tense construction, it is always
the prefix form that is used and not the particle.

Source: {a:217}
Vetted: t
Judgment: g
Phenomena: {tam}
Ikh hob ongeshribn.
Ikh hob on-ge-shrayb-n.
1SG.NOM have;1SG.PST PRF-PP-write-PP.
I completed writing.

Since the Grammar Matrix does not support verbal particles, we only
implemented the prefix form of the perfective. We did this by firstly
creating a perfective aspect sub type (along with the other two
aspects described below), then by creating a position class that can
attach to any verb and an inflecting lex rules which handle the
following prefixes: on-, der-, far-, oys, op-, constraining the aspect
to be perfective.

The inchoative aspect expresses commencement of action without an
ending. This is formed using the finite form of the auxiliary nemen
(to take): nemen + verb or nemen zikh + action.

Source: {a:221}
Vetted: t
Judgment: g
Phenomena: {tam}
Er nemt arbetn.
Er nem-t arbet-n.
3SG.M.NOM begin-3SG work-INF.
He begins working.

Source: {a:221}
Vetted: t
Judgment: g
Phenomena: {tam}
Er nemt zikh arbetn.
Er nem-t zikh arbet-n.
3SG.NOM.M begin-3SG REF work-INF.
He begins working.

To implement this, we added nemen as an auxiliary which requires
its verbal complement to have infinitive form and also gives itself
finite form, as well as inchoative aspect. We have not implemented the
the nemen + zikh variant as the Matrix cannot handle complex
auxiliaries.

The habitual aspect expresses an activity that reoccurs regularly. It is
formed with the particle ton like this:  ton + infinitive.

Source: {a:221}
Vetted: t
Judgment: g
Phenomena: {tam}
Er tut shraybn.
Er tu-t shrayb-n.
3SG.NOM.M do-3SG write-INF.
He regularly writes.

Even though ton is really more of a particle, we implemented it as an
auxiliary for now, in a similar way to nemen above, so that we could
parse these sentences. We will return to this and analyze it correctly
once we start hand editing the tdl.


AGREEMENT

Yiddish shows agreement between the person and number of the subject
and the finite verb in the present tense. This is marked through
suffixation of the verb. This is illustrated below:

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement}
Ikh red.
Ikh red.
1SG.NOM talk;1SG.
I talk.

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement}	
Zey redn.
Zey red-n.
3PL.NOM red-3PL.
They talk.

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement}
Du redst.
Du red-st.
2SG.NOM red-2SG.
You talk.

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement}
Er redt.
Er red-t.
3SG.NOM.M red-3SG.
He talks.

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement}
Er redt.
Er red-t.
3SG.NOM.M red-3SG.
He talks.

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement}
Mir reden.
Mir red-en.
1PL.NOM.M red-1PL.
We talk.

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement}
Mir reden.
Mir red-en.
1PL.NOM.M red-1PL.
We talk.

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement}
Ir redt.
Ir redt.
2PL.NOM.M red-2PL.
You all talk.

The following examples show ungrammatical permutations:

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement}
Ikh redn.
Ikh red-n.
1SG.NOM talk-1PL.
I talk.

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement}
Ir redn.
Ir red-n.
2PL.NOM talk-1PL
I talk.

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement}
Er redn.
Er red-n.
3SG.NOM.M red-3PL.
He talks.

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement}
Mir red-st.
Mir red-st.
1PL.NOM red-2SG.
We talk.

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement}
Mir redt.
Mir red-t.
1PL.NOM talk-1PL.
We talk.

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement}
Zey red.
Zey red.
3PL.NOM talk-1SG.
They talk.

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement}
Es redn.
Zey redn.
3PL.NOM talk-3SG.
They talk.

We implemented this as inflecting lexical rules which constrain the
person and number as is required by the different suffixes
corresponding to the different slots in the paradigm. They also
constrain the form of the verb to be finite and the tense to be
present.

Yiddish also shows agreement between both determiners, demonstratives
and nouns. They agree for number, case and gender for third person. 
The paradigm is shown below:

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement,determiners}
Der man zingt.
Der man zing-t.
DEF.3SG.M.NOM man sing-3SG.
The man sings.

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement,determiners}
Di froy zingt.
Di froy zing-t.
DEF.3SG.NOM.F woman sing-3SG.
The woman sings.

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement,determiners}
Dos kind zingt.
Dos kind zing-t.
DEF.3PL.NOM.N child sing-3SG.
The child sings.

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement,determiners}
Di mener zingen.
Di man-er zing-en.
DEF.NOM.PL man-3PL sing-3PL.
The men sing.

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement,determiners}
Di mener zingen.
Di man-er zing-en.
DEF.NOM.PL man-3PL sing-3PL.
The men sing.

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement,determiners}
Di mener zen dem man.
Di man-er ze-n dem man.
DEF.NOM.3PL man-3PL see-3PL DEF.ACC.3P man.
The men see the man.

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement,determiners}
Di man zingt.
Di man zing-t.
DEF.3SG.NOM.F man sing-3SG.
The man sings.

The following examples show ungrammatical combinations:

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement,determiners}
Di froy zingst.
Di froy zing-st.
DEF.3PL.NOM.F woman sing-3SG.
The woman sings.

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement,determiners}
Dem froy zingst.
Dem froy zing-st.
DEF.3P.ACC woman sing-2SG.
The woman sings.

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement,determiners}
Dos man zingt.
Dos man zing-t.
DEF.3PL.NOM.N man sing-3PL.
The man sings.

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement,determiners}
Der kind zingt.
Der kind zing-t.
DEF.3SG.NOM child sing-3SG.
The child sings.

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement,determiners}
Mener di zingen.
Man-er di zing-en.
Men-PL DEF.3PL.NOM sing-3PL.
The men sing.

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement,determiners}
Mener zingen.
Man-er zing-en.
Men-PL sing-3PL.
Men sing.

Source: author
Vetted: f
Judgment: u
Phenomena: {agreement,determiners}
Dos maner zingen.
Dos man-er zing-en.
DEF.NEUT.NOM.SG man-PL sing-3PL.
The men sing.

We implemented agreement through constraints on the determiner slots,
and simply enumerated each form as appropriate since determiners are a
closed class.


### Agreement Examples: Demonstrative - Noun

[See Demonstratives]



NEGATION

Sentential negation in Yiddish is indicated by the use of the single
negative particle nit. The unmarked position is after the finite
verb. This is illustrated below:

# Unmarked sentential negation on intrans verb 
Source: author
Vetted: f
Judgment: g
Phenomena: {negation}
Der man zingt nit.
Der man zing-t nit.
DEF.NOM.M man sing-1SG NEG.
The man isn't reading the book.

# Unmarked sentential negation on transitive verb 
Source: author
Vetted: f
Judgment: g
Phenomena: {negation}
Der man leyent nit dos bukh.
Der man leyen-t nit dos bukh.
DEF.NOM.M man read-1SG NEG DEF.ACC.N book.
The man isn't reading the book.

There is also a dialectical variant, nisht, illustrated bellow, which
we currently do not handle, but plan to.

# nisht is a dialectical variant of nit
Source: author
Vetted: f
Judgment: g
Phenomena: {negation}
Der man leyent nisht dos bukh.
Der man leyen-t nisht dos bukh.
DEF.NOM.M man read-1SG NEG DEF.ACC.N book.
The man isn't reading the book.

It is possible to use nit in front of constituents other than the
verb. This still functions as sentential negation but serves to place
the focus on the particular constituent. Our grammar currently does
not cover nit occurring in a marked position. Examples of placing nit
in front of subject, direct object and adverb are below:

# alternative focus of sentential negation 
Source: author
Vetted: f
Judgment: g
Phenomena: {negation}
Nit der man leyent dos bukh.
Nit der man leyen-t dos bukh.
NEG DEF.NOM.M man read-1SG DEF.ACC.N book.
The man isn't reading the book.
# (=Someone else is reading the book.) 

# alternative focus of sentential negation
Source: author
Vetted: f
Judgment: g
Phenomena: {negation}
Nit haynt der man leyent dos bukh.
Nit haynt der man leyen-t dos bukh.
NEG today DEF.NOM.M man read-1SG DEF.ACC.N book.
The man isn't reading the book today.
# (=The man is reading the book another day.) 

When the sentence involves some combination of auxiliary/modal +
infinitive/past participle, nit occurs directly after the auxiliary in
spoken or informal language, and directly before the infinitive in
formal written language. For the fragment of Yiddish we are dealing
with, it is likely that these situations will always
coincide. Placement of nit with respect to auxiliary is shown below.

# placement of nit with auxiliaries 
Source: author
Vetted: f
Judgment: g
Phenomena: {negation}
Der man vet nit geleyent dos bukh.
Der man vet nit PTCP-leyen-INF dos bukh.
DEF.NOM.M man will;1SG NEG read-1SG DEF.ACC.N book.
The man will not read the book.

Clearly ungrammatical locations for nit then, are directly preceding
the verb, and following the final NP in a clause:

# Ungrammatical location of nit
Source: author
Vetted: f
Judgment: u
Phenomena: {negation}
Der man nit leyent dos bukh.
Der man nit leyen-t dos bukh.
DEF.NOM.M man NEG read-1SG DEF.ACC.N book.
The man isn't reading the book.

# Ungrammatical location of nit
Source: author
Vetted: f
Judgment: u
Phenomena: {negation}
Der man leyent dos bukh nit.
Der man leyen-t dos bukh nit.
DEF.NOM.M man read-1SG DEF.ACC.N book NEG.
The man isn't reading the book.

Negation was implemented using the sentential negation section of the
Grammar Matrix. In particular, we chose the simple exponence, as there
is only one negative marker required. For the syntax of the negative
particle, we opted for it to be an adverb which modifies verbs,
occurring after the verb. Aside from the aforementioned movement of
nit through topicalization, this models sentential negation in Yiddish
well.


ARGUMENT OPTIONALITY

Yiddish has optional object dropping which is lexically licensed by
the verb. The presence or absence of the object is never marked on the
verb. An example of object dropping being permitted by the verb esn
(to eat) is presented below, followed by an example showing that the
verb trogn (to carry) does not allow object dropping.

Source: author
Vetted: f
Judgment: g
Phenomena: {argument optionality}
Ikh es dem epl.
Ikh es dem epl.
1SG.NOM es;1SG DEF.ACC.M apple.
I eat the apple.

Source: author
Vetted: f
Judgment: g
Phenomena: {argument optionality}
Ikh es.
Ikh es.
1SG.NOM es;1SG.
I eat.

Source: author
Vetted: f
Judgment: u
Phenomena: {argument optionality}
Ikh trog dem epl.
Ikh trog dem epl.
1SG.NOM trog;1SG DEF.ACC.M apple..
I carry the apple.

Source: author
Vetted: f
Judgment: u
Phenomena: {argument optionality}
Ikh trog.
1SG.NOM carry;1SG.
I carry.

We handled this by indicating on the Argument Optionality page that
object dropping is lexically licensed and that both dropped and overt
objects are not marked on the verb. We then created a new verb type
trict-trans which inherits from trans-verb but also constrains its
object NP to be OPT - and then added appropriate lexical entries such
as trogn (to carry) and darfn (to need).


DEMONSTRATIVES/DEFINITENESS

Yiddish has both promximate and distant demonstratives. When stressed,
the definite articles ('der', 'di', 'dos', & 'dem') function as
proximate demonstratives. Since this is a prosodic feature of the
language, we could not model this using the matrix.

Yiddish's distant demonstratives (equivalents of 'that'/'those' in
English) manifest themselves as inflections of the root 'yen-'
('yener', 'yenem', 'yene', & 'jenc') and their distribution
corresponds closely with that of the Yiddish determiners. We chose to
implement them as determiners in the same fashion as definite
articles.

We preempted the definiteness requirement of Lab 3 last week for Lab
2. Yiddish's definite articles ('der', 'di', 'dos', & 'dem') are
marked for gender (in the third person), case, number and person (see
the Lab 2 write up for the exact distribution).  It also has
indefinite articles ('a', 'an') which work the same way as their
English counterparts. Indefinite articles can only precede singular
nouns. We imp- lemented (in)definiteness by creating a 'indef-det'
type that imposes a 'number' feature constraint. Similarly all the
other determiner types impose 'case', 'gender', and 'number'
constraints.

### Demonstratives Examples

Source: author
Vetted: f
Judgement: g
Phenomena: {agreement.demonstratives}
Yener man zingt.
Yener man zing-t.
DEM.3SG.M.NOM man sing-3SG.
That man sings.

Source: author
Vetted: f
Judgement: g
Phenomena: {agreement.demonstratives}
Yene froy zingt.
Yene froy zing-t.
DEM.3SG.NOM.F woman sing-3SG.
That woman sings.

Source: author
Vetted: f
Judgement: g
Phenomena: {agreement,demonstratives}
Yenc kind zingt.
Yenc kind zing-t.
DEM.3PL.N.NOM child sing-3SG.
That child sings.

Source: author
Vetted: f
Judgement: g
Phenomena: {agreement,demonstratives}
Yene mener zingen.
Yene man-er zing-en.
DEM.NOM.M.3PL man-3PL sing-3PL.
Those men sing.

Source: author
Vetted: f
Judgment: g
Phenomena: {agreement,determiners}
Di mener zen yenem man.
Di man-er ze-n yenem man.
DEF.NOM.3PL man-3PL see-3PL DEM.ACC.3P man.
The men see that man.

Source: author
Vetted: f
Judgment: g
Phenomena: {determiner,case,agreement}
Di mame git dem taten yenem epl.
Di mame git dem tate-n yenem epl.
DEF.NOM.F mother give;1SG DEF.DAT.M father-DAT DEM.ACC.M apple.
The mother gives the father that apple.

Source: author
Vetted: f
Judgment: u
Phenomena: {determiner,case,agreement}
yener kind zet dem taten.
yener kind ze-t dem tate-n.
DEF.NOM.M child ze-3SG DEM.DAT.F tate-DAT.
That child sees that father.

Source: a:248
Vetted: t
Judgment: g
Phenomena: {determiner,word order,agreemet}
Di mame git yenem kind yenc bukh.
Di mame git yenem kind yenc bukh.
DEF.NOM.F mother give;1SG DEM.DAT.N child DEM.ACC.N book.
The mother gives that child that book.

Source: author
Vetted: f
Judgment: g
Phenomena: {determiner,case,agreement}
Di mame git yener mamen yenem epl.
Di mame git yener mame-n yenem epl.
DEF.NOM.F mother give;1SG DEM.DAT.F mother-DAT DEM.ACC.M apple.
The mother gives that mother that apple.

Source: author
Vetted: f
Judgment: u
Phenomena: {determiner,case,agreement}
yene kind zet yener mamen.
yene kind ze-t yener mame-n.
DEM.NOM.F child ze-3SG DEM.DAT.F mame-DAT.
That child sees that mother.

# Case agreement on determiners.
Source: author based on a:224
Vetted: f
Judgment: g
Phenomena: {determiner,case,agreement}
Yener lerer shraybt yene zatsn.
Yener lerer shrayb-t yene zats-n.
DEM.NOM.M teacher write-1SG  DEM.PL.ACC.F sentence-PL.
That teacher writes those sentences.


@desc: Incorrect because wrong gender demonstratives 
Source: author
Vetted: f
Judgement: u
Phenomena: {agreement,demonstratives}
Yene man zingt.
Yene man zing-t.
DEM.3SG.NOM.F man sing-3SG.
That man sings.

@desc: Incorrect conjugation of `zingn`.
Source: author
Vetted: f
Judgement: u
Phenomena: {agreement,demonstratives}
Yene froy zingst.
Yene froy zing-st.
DEM.3PL.NOM.F woman sing-2SG.
That woman sings.

@desc: Incorrect demonstrative used. 
Source: author
Vetted: f
Judgement: u
Phenomena: {agreement,demonstratives}
Yenem froy zingt.
Yenem froy zing-5.
DEM.N.DAT.3PL woman sing-3SG.
That woman sings.

Source: author
Vetted: f
Judgement: u
Phenomena: {agreement,demonstratives}
Yenc man zingt.
Yenc man zing-t.
DEM.3PL.NOM.N man sing-3PL.
That man sings.

Source: author
Vetted: f
Judgement: u
Phenomena: {agreement,demonstratives}
Yener kind zingt.
Yener kind zing-t.
DEM.3SG.F.DAT.NOM child sing-3SG.
That child sings.

Source: author
Vetted: f
Judgement: u
Phenomena: {agreement,demonstratives}
Yenc maner zingen.
Yenc man-er zing-en.
DEM.NEUT.NOM.3SG man-PL sing-3PL.
Those men sing.


ON NOT DOING ARMCHAIR LINGUISTICS

Lastly, we would like to acknowledge that a lot of our examples are
constructed. Our reference grammar is pretty minimal, just providing
an account of each phenomenon and often doesn't go beyond single
examples of constructions -- which can often contain prepositions or
other phenomena we will not be covering for this course, making many
of the examples problematic. We would much rather be doing
evidence-based linguistics however, and to that end, we have just
found a wonderful resource: The Corpus of Modern Yiddish
(www.corpustechnologies.com:8080/YNC/search/) which contains about 400
documents in Yiddish transcribed into the YIVO Romanization we are
using. It is fully glossed, searchable and has a concordance AND fully
glossed. The search interface allows for concordance based viewing of
search results, so we are hopeful that this will help us to go through
our examples and convert some of them into attested examples
illustrating the phenomenon.


COVERAGE OVER TEST SUITE
================================== 

BUG FIXES AND REVISIONS FROM LAB2

To begin with, we corrected a number of existing problems from
lab2. These fixes included a few lexical entries that we had missed as
well as a couple of transcription mistakes in the test suite and also
changing the output of our "morphological analyzer" to convert vowel
change in the root into the base form. 

We also found a rather significant bug, which was that common nouns
were not being marked for 3rd person. Once this addition was made, no
improvements in coverage were made, but this did have the effect of
removing all ambiguity from the parsing items in the test suite. This
occurred because some items were receiving two parses, where
syncretism in the verb agreement pattern meant that number was
underspecified between two values and the NP was completely
underspecified due to the bug. For instance the following was
receiving two parses due to the fact that the -(e)n suffix on verbs
can indicate both 1st and 3rd person, and the noun was not
constraining this.

A second bug we discovered was that the number lex rules for nouns was
being applied indiscriminately to all nouns, including pronouns. Since
there is no affix for marking nouns for singular number, and the
number position class is obligatory, this meant the singular lex rule
was being applied incorrectly to pronouns. We fixed this by creating a
new noun lex type: common-noun, which fem-noun, masc-noun and
neut-noun all inherit from. The number positionn class was then
changed to take as input common-noun rather than all nouns. Once we
did this, we were then able to parse the item below, which was failing
due to the singular lex rule being obligatorily applied to an already
plural lex.

Zey redn.
Zey red-n.
3PL.NOM red-3PL.
They talk.

This is the kind of error it would be nice to catch in the validation
of the questionnaire, however would be difficult without going down
the path of compiling the grammar.

Once we made all the above additions and fixes, our coverage went up
from 34% to 61%.


COVERAGE NOTES FROM LAB3

Total items:            149
Positive items:          98
Negative items:          51
Parsing positive items:  47
Parsing negative items:  11
Coverage:                48%
Overgeneration:          22% 

All sentences that contain the wrong form of the AUX in the past
and future constructions parse when they shouldn't. This overgeneration
occurs because we have not constrained them as discussed below.

The sentence 'Ikh trog dem epl.' gets two parses because dem is
ambiguous between a definite determiner and a possessive determiner.
We will have to work out how to constrain this if possible.

The sentence 'Ikh es.' gives us two parses when it should just give
one. The spurious sentence does not have a VP but instead a V
immediately under the S, getting a full parse with the rule
basic-head-opt, instead of subj-head. We are not really sure why this
happens!


LIMITATIONS OF THE GRAMMAR MATRIX
==================================

1. Aspectual information marked with particles  

Yiddish can mark aspect through both verbal particles and also through
inflection, as illustrated by the following examples. (see above for
full gloss)

Ikh shrayb on.

Ikh hob ongeshribn.

They both show the perfective aspect in action, the first in the
present tense and the second in the past tense. The first examples
uses the particle 'on' to indicate the perfective, while the second
uses the suffix on- attaching to the past participle. (We have 
no data to suggest that the prefix version of on can be used in 
the present tense.)

Currently the Matrix can handle aspect when it is marked via
affixation, however not when it is marked through particles, so this
has been left unimplemented.


2. Verbs selecting auxiliaries

As already mentioned, Yiddish uses AUX + verb constructions to
indicate tense. The auxiliary can either be the finite form of hobn
(to have) or zayn (to be), with the specific form being selected by
each verb, as illustrated below. This auxiliary selection is currently
not possible through the matrix.

Ikh hob geshribn.
Ikh bin geblibn.


3. Irregular Auxiliary Verbs

The two main auxiliary verbs in Yiddish (hobn and zayn) have irregular
inflections for person and number. This means that full paradigm of
their forms cannot be captured through inflectional rules and each
form must be enumerated manually. While the forms of the auxiliary can
be enumerated under each auxiliary lex type, no constraints can
currently be placed upon these forms, which we need in this case to
constrain person and number. If we were dealing with verbs, it
wouldn't be too bad manually creating separate types for each form,
however auxiliaries impose many more constraints, and duplicating
these for each entry would be both tedious and clunky -- far better to
save this for hand editing the TDL.


4. Negation Focus

As already mentioned, the negative particle in Yiddish can move from
its unmarked position of to the right of the verb to before most other
constituents, serving to place the focus on that
constituent. Currently the Matrix is unable to support this kind of
flexibility in the location of negative particles. If indeed this
movement only changes the information structure and not the logical
form, adding support for this flexibility will be difficult, as it is
likely that this kind of movement in other languages could result in
actual constituent negation, thus changing the truth conditions of the
utterance (which it seemingly does not in Yiddish).

5. Alternate negative marker forms

We were unable to add the dialectical variant of nit: nisht, as The
customization system only contains one slot for the spelling of the
negative marker.
