Katie Sickles
LING 567
Lab 4

Customization Choices:
	
	For the Word Order choice I seleceted SVO. In the research I've done of Icelandic,
I haven't found any information that explicitly states that it is an SVO language.
However, all the examples I've seen have been in SVO, and the information has stated
that the verb always comes second. This information was not in a linguistic source,
so I'm not sure if that means that Icelandic is a V2 language or not. So for the sake
of simplicity I chose SVO since that is the data that I have seen, and I have constucted
my test suite in SVO.
	The Word Order page also had a selection for determiners as free words. This was a
complex choice since Icelandic does have free determiners, but only in certain cases.
When the determiner is a free word, it appears before the noun (as well as the 
adjective). However, when the determiner appears as a affix (which is the most common 
case), it is suffixed to the noun. Since the customization only asked to specify if 
determiners can appear as free words, I chose the "yes" option with det-noun as the 
order. I will add the suffix case to the grammar later.
	For the Sentential Negation section, I chose the independent adverb option. This
adverb appears directly following the verb. This is why I modifier option and selected
that it modifies the verb adn appears to the right of the verb. In the corupus that I
have looked at for Icelandic, however, shows this adverb appearing in many places. It
could be that it is used for negations other that sentential negation, or is more free
than indicated in the Icelandic grammar. Since my test suite only contains the negative
adverb as a verb modifier, this choice should predict the correct grammatical judgement
of the test suite sentences.
	In choosing the words for the lexicon portion of the customization, I made some
choices that were different than what the standard dicitonary translations defines as 
words. For the words "cat" and "dog", the dictionary translates them as "köttur" and
"hundur". I chose to write them in as "kött" and "hund" I think that the nomative for
form for these types of words is adding the "-ur" suffix since the accustive is just
of the form "kött". I might end up changing this, but it seems like it would be easier
to add a nominative suffix then to try to strip a word for accusative form.

Sentence Parsing:

	The first sentence I tried parsing was
	
	köttur elskar hund
	cat.nom.masc love.3sg dog.acc.masc
	"a cat loves a dog"
	
As would be predicted, this was sentence was unable to parse for a few reasons. The
first being that the ascii "ö" was not in the lexicon as the website for the 
customization didn't allow it. So the first change I made was to add the correct ascii 
characters into the lexicon.
	The next problem was that there was no verb argreement, so "elska" was in the
lexicon, but "elskar" wasn't. There was the same problem with "köttur". There is the
nomanitive suffix "-ur" in the lexicon, but there are no rules to attach it to nouns.
Since adding the rules and agreement wasn't in the assignment, I have not yet dealt
with this. So in order, to parse a sentence I tried

	kött elska hund
	
	This sentence is ungrammatical, but gets a correct parse as if there were no
case markers or agreement. This is expected since there is nothing in the grammar yet
to account for this.

Test Suite:

	At the time of submitting this version, I am unable to run the test suite due to
the stability of my linux machine. However, I have attempted to individually parse 
the sentences in the suite so that I can have parsing results. None of the sentences
currently parse (before and after lexicon expanding). This is due to the fact that 
there is not agreement or case. The only verb form I have (base form), is the form
used for 1sr person singular, and there are no pronouns in the lexicon, so this form
can't be used grammatically.
Note: I've included the tsdb in the tarball, but it has not been tested due to the
	problems stated above.

Important Phenomena:

	For Icelandic the most important phenomenon will be agreement. This is because
practically every word must have some type of agreement. This is the reason that no
sentences from the test suite were able to parse.
	Icelandic has agreement in gender, number, person and even case in some cases.
The verbs have to agree with their subjects in person in number. For example:

	köttur elskar hund
	cat.nom.masc love.3sg dog.acc.masc
	"a cat loves a dog"
	
	*köttur elska hund
	cat.nom.masc love.1sg dog.acc.masc
	"a cat loves a dog"
	
The first sentence is grammatical since "cat" is 3rd person singular, and the verb
is the correct 3rd person singular form. The second sentence is ungrammatical since
the verb is in the 1st person singular form.
	In addition to verbs, the determiners have to agree with their nouns in gender,
number and case. There are different determiner forms for all these agreement
combinations. For example:

	köttur-inn elskar hund-inn
	cat.nom.masc-the.nom.masc.sg love.3sg dog.acc.masc-the.acc.masc.sg
	"the cat loves the dog"
	
	*köttur-in elskar hund-na
	cat.nom.masc-the.nom.fem.sg love.3sg dog.acc.masc.pl
	"the cat loves the dog"
	
Agreement seems to be included in almost every other Icelandic phenomena. This is the
reason that it would be the most important phenomena to work on. If agreement fails,
then no sentence would be able to be parsed.