26 January: DCLRS Jan 29 4pm UCD -- Paolo Aquaviva
Dublin Computational Linguistics Research Seminar: Index of January 2010 | Dublin Computational Linguistics Research Seminar - Index of year: 2010 | Full index
Apologies for cross-posting.
The fourth in the 2009-2010 series of the Dublin Computational Linguistics
Research Seminar
will take place on Friday 29 January.
All are welcome to attend.
Time: 4pm
Location: Room B1.09, School of Computer Science and Informatics, UCD
Map to CSI entrance is at http://csiweb.ucd.ie/staff/acater/
EntranceMapUCDCSI.png
Speaker:
Paolo Aquaviva
UCD School of Languages and Literatures, UCD Dublin
Title: Nouns and the conceptualization of entities
Abstract:
Most semantic accounts treat common nouns as predicates.
Yet they differ from clear predicates. According to an influential
view, nouns encapsulate a standard of sameness that can (or
must) relativize identity in statements like "a is the same S as b",
where S must be a ?substantival? term. But a substantival
term is a noun. Can we develop this intuition into a non-circular,
explanatory account of what it means to be a noun?
This paper outlines a theory of nouns to address this question.
The theory is a linguistic one; but it rests on a specific view of how
speakers conceptualize ontology, and so it is also a hypothesis
on the relation between language, ontology, and cognition.
I argue that common nouns name kind-level entities in a speaker's
ontology. These correspond to basic concepts, and are cognitively
prior to their instances. Their domain is a taxonomy of atoms (not
a mereology). Nouns are the locus of categorization; determiner-
like elements, instead, set up and track discourse referents,
which may or may not be categorized as instances of the noun-
kind. Intermediate projections turn atomic kinds into descriptions,
whose denotation has a part structure. Only at this level does
the count-mass distinction arise.
A number of linguistic facts find a unified explanation in this
light. I will show how morphology can express part-structural
properties, or distinguish between concept- and instance-
reading. Reference to concepts also explains the semantic
filtering exemplified by argu-ment-al vs. *agree-ment-al; and
the cases where the semantic value of a noun cannot arise
from its extension.
In addition, this approach paves the way for a predictive theory
of a possible noun. As names, nouns define their referent as
one and the same across all possible worlds. Natural languages
allow nouns with no descriptive content, but not nouns for
concepts that are contradictory (?round square?) or world-contingent
(?number of planets?), although these are well-formed predicates.
Finally, not all nouns seem to have this value. Therefore,
naming a basic concept cannot be a definitional property for
common nouns; they remain a language-internal category.
Thank you to CNGL and to UCD CSI for support given to the DCLRS series.
_______________________________________________
cogsci mailing list
cogsci@scss.tcd.ie
https://lists.scss.tcd.ie/mailman/listinfo/cogsci
----- End forwarded message -----
Dublin Computational Linguistics Research Seminar - Index of January 2010 | Index of year: 2010 | Full index