Loading presentation...

Present Remotely

Send the link below via email or IM


Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.


Halima Husic

No description

Halima Husic

on 10 July 2015

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Halima Husic

Open American National Corpus OANC (14 Mill. words)
POS annotations, Dependency

Noun senses from Wordnet
Annotations from native speakers
What do we do differently?
We assume that,
Halima Husic
Bochumer English Countability Lexicon 2.0

rice cheese
water furniture
money wood
biology weather

Countability is a feature which applies to nouns
Our database
The annotation
the main annotation part consisted in answering questions about the syntactic and semantic properties of senses of the nouns
annotations - questionnaire
Test I
Test II
Test III
to identify countability classes we restrict the database to those noun sense pairs which were annotated unanimously
possible answers
the classes can be grouped together into four main categories

coin horse
man idea
bottle plate
rainstorm car

: singular and plural, can be combined with the indefinite article

: no plural form, not used with indefinite article - rather a classifier construction
a "something" of

a little
Distinction in grammar
indefinite article:

A dog is an animal.
*A wine is a drink. / Wine is a drinkable liquid.
My horses are hungry.
The *cheeses are/cheese is in the fridge.
I haven't got many coins./ I've got a few dollars.
I haven't got much rice./ I've got a little money.
Other approaches:
Criteria such as cumulativity, distributivity and homogeneity are applied to the referents
conceptual/ lexical
ontological criteria are applied to the concepts of words
countability is not a lexical feature, it is rather derived from the contextual environment
What about...
John baked cake for dessert. Mary likes cake. Cake is healthier than ice-cream.
John baked a cake for dessert. Mary liked the cake. That cake is healthier than those bonbons.
Some theories following Keith Allen (1980) claim that all nouns are count and mass to a certain extent.
At least some nouns appear to be both: count and mass
We call such nouns
dual-life nouns
a binary division of nouns into countable and uncountable is not possible.
the assignment of a [±count] feature cannot be based on the lemma
We investigate different types of nouns, not only the prototypes like
car, dog, wine, water...
We consider the senses of nouns
Native speakers of English annotate the noun-sense-pairs in terms of answering questions about some syntactic-semantic features
long-term goal
: a token based classification of nouns
Extraction of
nouns with a frequency > 10
the first four senses from Wordnet

ca. 14.000
noun sense pairs
already annotated
While considering the noun
we extract the first four senses of car and pass them to the annotators
four canadian assistants annotated pairwise a set of noun sense pairs
The values of the IAA (Krippendorfs α) for each question range from 0.635 to 0.762
tests the grammaticality of a certain construction by inserting the annotated noun sense
Three two-part questions:
Test I.1
Test I.2
Test II.1
Test II.2
Test III.1
Test III.2
Test I.1
Test I.2
Test II.2
Test II.1
Test III.1
Test III.2
The boy ate more fruitcake than the girl --> Yes
*John bought more car than Bill --> No
Ted owns more furniture than Ross --> Yes
more fruitcake --> not number
more furniture --> number
*more car --> not applicable
John bought more cars than Bill. --> Yes
He drank more whiskeys than her. --> Yes
He drank more glasses of whiskey than her.
--> equivalent
(*) John bought more kinds of car than Bill.
--> not equivalent
A car is a vehicle. ---> Yes
#A whiskey is a liquid. ---> No
Whiskey is a drinkable liquid containing alcohol. ---> Yes
*Car is a vehicle. ---> No
For every question three possible answers are given

Test I.1 yes, no, NA
Test I.2 number, not number, NA
Test II.1 yes, no, NA
Test II.2 equivalent, not equivalent, NA
Test III.1 yes, no, NA
Test III.2 yes, no, NA
the results form a set of distinct countability classes

the number of possible answers for each question lead to a logical class space of 729 classes
from 729 possible classes, only 18 are identified in the current stage of annotation
from the overall 14,000 annotated noun sense pairs remain 10.667. These 10,667 noun sense pairs form the basis of the detection of countability classes
anyone who does not belong in the environment in which they are found (235)
the cardinal number that is the product of ten and seven (721)
a method of tending to or managing the affairs of a some group of people (especially the group's business affairs) (528)
running at a jog trot as a form of cardiopulmonary exercise (519)
furnishings that make a room or other area ready for occupancy; "they had too much furniture for the small apartment" (531)
fluid matter having no fixed shape but a fixed volume (510)
the state of being friends (or friendly) (726)
a location in the northern part of a country, region, or city (523)
the countries of Asia (37)
a rosy color (especially in the cheeks) taken as a sign of good health (190)
food remaining from a previous meal; "he had leftovers for dinner last night" (371)
(law) the unlawful act of capturing and carrying away a person against their will and holding them in false imprisonment (729)
(usually plural) the components needed for making or doing something; "the recipe listed all the makings for a chocolate cake" (73)
dishware made of high quality porcelain (513)
the state of relying on or being controlled by someone or something else (514)
the point or degree to which something extends; "the extent of the damage"; "to a certain extent she was right" (199)
the outside boundary or surface of something (28)
violent pangs of suffering; "death throes" (353)
Full transcript