Module 4 Lexical Semantics Foundations and Issues - PowerPoint PPT Presentation

1 / 12
About This Presentation
Title:

Module 4 Lexical Semantics Foundations and Issues

Description:

Linguistics Visiting Prof. USC 96. HRL Labs (used to be Hughes Research Labs) ... two disciplines: linguistics and computational linguistics: cases where notation ... – PowerPoint PPT presentation

Number of Views:42
Avg rating:3.0/5.0
Slides: 13
Provided by: robert693
Category:

less

Transcript and Presenter's Notes

Title: Module 4 Lexical Semantics Foundations and Issues


1
Module 4 - Lexical Semantics Foundations and
Issues
  • Who am I?
  • Robert Belvin
  • Linguistics Visiting Prof.
  • USC 96
  • HRL Labs (used to be Hughes Research Labs)
  • Office hours just before or after class--make
    appt. via email rsbelvin_at_hrl.com
  • Scope of Linguistics Component of Class
  • linguistics background and areas of research
  • overview areas of potential importance to NLU

2
Module 4 - Linguistic Issues
  • Prof. Robert S. Belvin (Linguistics Dept.)
  • October 4, 6, 11, 25 November 1, 3, 29
  • Topics
  • Linguistic study of lexical semantics with focus
    on areas of potential benefit for automated
    systems
  • Lexical semantics vs. lexicography vs. formal
    semantics
  • Different views of the language faculty and how
    that can impact how we do semantic representation
    and mapping
  • Some classics Fillmore (Cases),
    Gruber(Thematic Relations)
  • Lexical semantic decomposition
  • some semantic primitives, cross-categorial
    semantic features, languages which have
    transparent decomposition/morphologically complex
    forms, lexicalization patterns (in English and
    cross-linguistically)

3
Module 4 - Linguistic Issues
  • Topics (continued)
  • Verb Classes and Alternations
  • valence-changing operations (in English and
    cross-linguistically)
  • the "unaccusative" hypothesis
  • Some Formalisms and Notation
  • Lexical Conceptual Structure
  • Parsons Event Semantics
  • Image-schemas also some Cognitive
    Semantics/metaphor - guest lect. (Dr. Tim
    Clausner).
  • Mapping
  • how do you get semantic representation out of
    syntactic structure?
  • subcategorization and selection
  • thematic role hierarchy and universal theta
    alignment hypothesis
  • semantic contribution of syntactic structure

4
What is Lexical Semantics?
  • the semantics of lexical items (words)
  • isnt semantics in linguistics always about
    words?
  • lexical vs. propositional is proper contrast
  • propositional John killed Bill P is all you
    get,
  • or FOL representation kill (j,b) (just tells you
    2 args)
  • lexical asks question how does kill mean what
    it does?

5
What is Lexical Semantics?
  • the semantics of lexical items (words)
  • lexical semantics is often contrasted with formal
    semantics
  • but this is misleading, though it is generally
    true that formal semantics, with a few
    exceptions, does not ask questions like
  • how does kill mean what it does?
  • in formal semantics, kill is usually taken as a
    unit
  • The simple FOL meaning of Joe killed a rat
    is
  • ?x (rat(x) ? kill(Joe,x))

6
Lexical versus Formal
  • Somewhat artificial distinction
  • Many highly formalized Lexical Semantic studies
    (e.g. Pustejovsky, Wierzbicka)
  • Pustejovskys representation of the template for
    an Agentive change-of-state predicate (like bake
    or kill)
  • ?y?x?e1?e2?P?R ? AGENTIVER(e1,x,y)?FORMALP(
    e2,y)?e1lt?e1
  • Formal semantics tradition looking within words
    (e.g. Terence Parsons)
  • ?ekilling(e) ? Agent(x,e) ? Patient(y,e) ?
    Cul(e)
  • (this is the meaning of kill )

7
Lexical versus Formal
  • Somewhat artificial distinction
  • There are many mechanisms within the field of
    formal semantics that one could use for lexical
    semantic characterizations
  • could do lexical properties as meaning postulates
    (entailments), etc.
  • e.g. kill-event die-event
  • kill-event cause-event
  • etc.
  • lambda-abstract templates (like Pustejovsky)
  • many others

8
Both formal and lexical traditions have benefits
  • Formal Semantics has mathematical (set theory)
    and formal logic traditions for operating on
    representations
  • Well-known logical proof methods Modus Ponens,
    Modus Tollens, etc. good for reasoners
  • P ? Q and P from this infer Q
  • P ? Q and Q from this infer P etc.
  • DeMorgan laws
  • handy for converting and to/from or
  • (P ? Q) ? (P ? Q)
  • (P ? Q) ? (P ? Q) etc.
  • A LOT of these double negation, contraposition,
    etc.. BUT non-compliant NL exs!
  • Negating a universal or existential
    quantification
  • Negate the proposition AND toggle quantifier
    (between ? and ?)
  • ?x P(x) ? ?x P(x) similarly with ?x ?x
    P(x) ? ?x P(x)
  • But in fact these arent excluded from use in
    Lexical Semantics

9
Both formal and lexical traditions have benefits
  • Lexical Semantic Structures more closely resemble
    syntactic structures
  • EVENTCAUSE (THING Joe , EVENT DIE (THING
    rat ))

EVENT
CAUSE
EVENT
THING
DIE
Joe
THING
rat
10
Lexical vs. Formal Semantics
  • Difference is really more a question of focus
  • formal semantics tradition concerned with
    problems which are as much philosophical as
    semantic, such as reference and quantification
    much is based on set theory. Notion of set is
    central premise. Difficulties arise with real
    world (fuzziness, prototypes, etc.)
  • lexical semantics tradition has always been
    multilingual, borrowing a lot from descriptive
    linguistics and cognitive psychology. Some
    accept set as axiomatic, others dont.

11
Lexical vs. Formal Semantics
  • in-class mapping exercise

12
Lexical vs. Formal Semantics
  • BUT In spite of usefulness of formal semantics
    for developing systems for encoding lexical
    semantic knowledge, there is potential for
    advances to be made in understanding of
    I-language which are not likely with formal
    semantics.
  • different goals of two disciplines linguistics
    and computational linguistics cases where
    notation and formalism have allowed for genuine
    progress to be made
  • often-recited distinction between building birds
    versus planes also applies are we building a
    brain or a question-answering machine?
  • answer will affect how you proceed and priorities
Write a Comment
User Comments (0)
About PowerShow.com