Instructions forwarded to the class on August 30:

The file containing the revised ATN (no reference to the word "to*" as input)
and dictionary (intransitive -> intrans) is attached.

In looking at the dictionary that I have in my office I noticed that it
might be helpful if we use the same abbreviations for the same parts of 
speech, so I've listed the most common ones below.

For verbs, please use v
For nouns, please use n
For adjectives, adj
For adverbs, adv
For coordinate conjunctions, use conj
For pronouns, pron

All of the above abbreviations are probably the ones in whatever 
dictionary you pick up.  But you will need to be different from 
most of the older dictionaries in the following:

Please do not include all the possible parts of speech for 'a' and
'an'.  If you have these words in your sentences, simply list them as 'det'
for 'determiner'.  Many dictionaries do not list determiner as a part
of speech at all, so determiners are listed as definite article, indefinite
articles, or adjectives in publisher's dictionaries.

The most common determiners which occur in our sentences are 
a
an
the
this
that   (e.g., "that older text")
these  (e.g., "these first five examples")
each   (e.g., "each individual simplified expression")

Although "many" seems to occur in the same contexts, it is not a
determiner, because no noun phrase has more than one determiner, and
it is possible to say "the many unusual occurrences of ...".  Note
that it is NOT possible to say "the each unusual occurrence" or "the
these..."  or "the that ...".  So "many" should be listed as an
adjective.

This, that, these, each, and many should also be listed as pronouns (pron). 

Someone in the class has the word "our" which is a possessive pronoun
(please abbreviate "pospron").  Possessive pronouns occur exactly
where determiners can, so you may simply add another edge with
category pospron from the same initial state and to the same target
state as your det edge.  In fact, since I don't mind your storing the
possessive pronoun in the det register, you can actually simply add
pospron to the category list for the existing edge.

Publishers' dictionaries don't usually distinguish between coordinate
and subordinate conjunctions.  Coordinate conjunctions are 'and',
'but', and 'or', and for our purposes, 'either'.  Since either can
also be a pronoun, please list it as both conj and pron, if you have
the word in your sentences.  The coordinate conjunctions should all be
listed as conj.

The subordinate conjunctions (please abbreviate 'subconj') include 
after   
as
before
if
when
where (normally, "where" functions simultaneously as an adverb and relative
 	pronoun, as in  "the place where we ate last night", but it is used
 	only as a subordinate conjunction in our assignment, and I not only 
 	give permission but strongly recommend that 'where' be listed only as
 	a subconj)
while  
 
 
'After' and 'before' are both prep's and subconj's.  'While' is both
subconj and n.

Most publishers' dictionaries also do not mention that relative
pronouns are distinct from other pronouns.  The relative pronouns
should be abbreviated relpron.  In our assignment, I've only noticed
two:  'which' and 'that'.  But everyone must list those two words as
relpron, if they have sentences that use those words.

The sample dictionary that you originally received did not distinguish
between nouns and proper nouns.  The vast majority of the time, a
proper name is not modified ("the cute little Janice" or "the Rajeev
in the middle" would be extremely unusual, and even "cute little
Raymond" is infrequent).  So you have permission to have an edge
marked nprop which extends from the start state of your np recognizer
to the final state, like the already existing edge marked pron in that
subatn.  A couple of students have a compound proper noun "World Wide
Web" which they may list in the dictionary with underscores instead of
blanks:  World_Wide_Web.  If they do, however, they must allow for the
word "the", which happens to accompany this particular compound proper
noun.  Craig, I think you are the only person that really has a
dilemma regarding this - if you allow "the World Wide Web", you also
allow "the Yarowski" and "the Brin".  If you really want to, you may
include "the" as part of the name for the World Wide Web, but then let
me know, so my dictionary includes it as well.

The sample handout did not include a present participle.  They look like
the following:

(tagging v (transitive intrans prespart) tag)
(extracting v (transitive prespart) extract)
(voting v (intrans prespart) vote)
(vote (n v) (intrans))
(extract (n v) (transitive))
(tag (n v) (transitive intrans))

Another feature that your publishers' dictionary probably won't
provide is the 'beverb' feature.  These are all the verbs which can
take a predicate adjective or predicate complement.  For example, "she
looks pretty" is a sentence in which "looks" is a beverb, because it
has the same syntactic structure as "she is pretty".  "The song
sounded soft and melodious" shows that the verb 'sound' is also
sometimes a beverb.  At the same time, 'look' is not always a beverb.
Consider "I looked into the barrel."  That's just a normal
intransitive use of the verb 'look'.  Our assignment includes several
verbs which much be given the 'beverb' feature.  One is "seem".
Another is "look" and its variants "looks" and "looked", even though
they aren't used as beverbs in these sentences.  And of course, all
the forms of "be" should have that feature as well.


Hope the above is helpful.  Don't hesitate to ask questions by email or
come by my office.  Or in class, of course.

-LB-