SlideShare ist ein Scribd-Unternehmen logo
1 von 32
Downloaden Sie, um offline zu lesen
ECAP 6 — Sixth European Congress of Analytic Philosophy
                  Krakow August, 26, 2008



“Unifying Logical Form & The linguistic Level
                   of LF” 


                  Mouhamadou El Hady BA
             Institut Jean Nicod ENS-EHESS Paris
Overview of the Talk:


 The Generative Framework

 The Linguistic level of LF

 Logicians Logical Form

 Unification via generalized quantification

 Interest & problems




    Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   2
Introduction:

    Logic: an artificial & well defined language + rules of inference

               −   Looking for sound derivations
               −   regardless of actual implementation into the human mind




    Generative Linguistics: define an abstract grammar but above all an empirical
    science looking for the laws actually used by the speaker


    Claim: these two research programs converge via their use of Generalized
    quantification
     − logical laws of thought



              Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   3
The Generative Framework:



What's a Grammar?

   "The grammar of a language purports to be a description of the intrinsic competence
  of an ideal speaker-listener. If the grammar is, moreover, perfectly explicit – i.e. if it
  doesn't simply accept the understanding if intelligent reader but gives an explicit
  analysis of the activity that the speaker-listener display- we can, with some
  redundancy, call it a generative grammar. "

Noam Chomsky (1965) : Aspects of the theory of syntax


             Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008          4
The Generative Framework:


What's a Grammar?

     −   (p) Colorless green ideas sleep furiously

     −   (q) Walk pixel hands color john foolishly


    Distinction Phonology/Syntax/Semantics: distinct and somewhat independent
    modules of the language faculty manage these aspects of natural languages


    That's why we recognize (p) as a sentence, understand it and judge it absurd when
    we can't even say of (q) that it really is a sentence.


                 Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   5
The linguistic Level of LF



 “The basic elements we consider are sentences; the grammar generates mental
 representations of their form and meaning. I will call these representations,
 respectively, ''phonetic forms'' and LF (which I will read, ''logical form,'' though with
 a cautionary note)”
Chomsky[1980], Rules & Representations, p. 143


 ''representations of their form and meaning''→ LF a semantic level?



             Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008         6
The Linguistic Level of LF


LF as a semantic level?

 “LF is thus the interface between grammar and the conceptual-intentional properties
 of language, just as the level of Phonetic Form (PF) is an interface between grammar
 and the audio-perceptual properties of utterances. LF is not to be equated with the
 level of semantic structure anymore than PF is to be treated as a level specifying the
 sound waves of any given utterance. It expresses only aspects of semantic structure
 that are [...] contributed by grammar.”

CT James Huang [1994], “Logical Form”



             Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008       7
The Linguistic Level of LF


                                                 
                                                  SS: Surface Structures

                                                 
                                                  PF: Phonetic Forms

                                                 
                                                  LF: Logical Form




     We are only interested in the SS/ LF branch


        Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   8
The Linguistic Level of LF

  Defining Move α

  Minimalism:

      •   Only one rule: Movement (Move α) with constraints

      •   E.g.: Move NPs to the head of the sentence leaving traces at their former
          place

    “The transformational mapping to S-structure can be reduced to (possibly
    multiple application of) a single general rule “Move α,” where α is an arbitrary
    phrase category”

   Chomsky [1980], Rules & Representations, p. 145


             Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008       9
The Linguistic Level of LF



    Why should we add a covert level of LF?

       To understand the interpretation of Wh-phrases

       To uncover the reference of pronouns

       To explain syntactically why some ambiguities arise:

             Ex: Let's consider this sentence with two NPs :

               (d) Every philosopher loves some linguist




              Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   10
The Linguistic Level of LF



              (d) Every philosopher loves some linguist

      Application of Move α : Quantifier Raising

  –    Move the NPs at the head of the sentence

  –    Leave traces at their former place

  –    Traces are kind of variables like in FOL's predicate language




              Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   11
The Linguistic Level of LF



 Two possibles LF formula:
         (d1) [Every philosopher]1[Some linguist]2 [t1 loves t2]

         (d2) [Some linguist]2 [Every philosopher]1[t1 loves t2]

 Hence the ambiguity: possibly many linguists in (d1) but same beloved linguist in (d2).




              Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008     12
The Linguistic Level of LF

 Traces:
         (d1) [Every philosopher]1[Some linguist]2 [t1 loves t2]

         (d2) [Some linguist]2 [Every philosopher]]1[t1 loves t2]


    The traces t1 & t2 variables binded by the NPs
    Variables, Quantified Noun Phrases, Rings a bell? That's logic, isn't it?


    But lets do some real logic and we will come back to see if we can unify logic and
linguistics


               Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   13
Logician's Logical Form
Logic: Consensus 'till Montague: natural languages can't handle the task of
formalizing our reasonings

“Languages are not made so as to match logic's ruler. Even the logical element in
language seems hidden behind pictures that are not always accurate.”

Frege [1906] Letter to Husserl



“The syntax of ordinary language, as is well known, is not quite adequate for this
purpose. It does not in all cases prevent the construction of nonsensical pseudo-
propositions”

Wittgenstein [1929]: Some remarks on Logical Form


           Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   14
Logician's Logical Form
Russell On denoting:

Breaking up Denoting Phrases:

      “The phrase per se has no meaning, because in any proposition in which it
 occurs the proposition, fully expressed, does not contain the phrase, which has been
 broken up”

     “Consider the next the proposition 'all men are mortal'. This proposition is really
 hypothetical and states that if anything is a man, it is mortal.”

  Russell (1905)

          ∀ x [M(x) →D(x)]

                    → Grammatical structure is irrelevant

              Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008     15
Logician's Logical Form




Solution:

Translate NL utterances into an artificial language displaying their “real” logical
structure and using correct laws of entailment




            Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   16
Logician's Logical Form


Frege on quantification:

 Numbers and quantifiers (∀, ∃) are second order predicates expressing relations
 between concepts rather than properties of objects

     For example:

         (f) The shirt is red

         (g) Every shirt is red




            Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   17
Logician's Logical Form

Frege on quantification:

Universe: {shirts}
    (f): ( ) is red is attributed to an object (a particular shirt)
    (g): a bit more complex
Analyse of (g) according to Frege: two concepts: ( ) is a shirt and ( ) is red, (g) states
a relation between these concepts namely that whatever satisfy ( ) is a shirt satisfy ( )
is red.
    So quantifiers are second order concepts expressing something about first order
concepts



            Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008         18
Logician's Logical Form


Russell On denoting:

Variables first :

“I take the notion of the variable as fundamental...”

           → Quantifiers are just variable binding operators

  Do not really state that Quantifiers are not second order predicates but limits the
  explorations in this direction




              Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   19
Logician's Logical Form


Frege, Russell, Quine....


Consensus: Natural languages are improper for good reasoning. A logician has to use
 an artificial language in order to display the real structure and make correct
 inferences.


                                  Enters Montague!
                            Respected philosopher & logician



               Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   20
Unification via generalized Quantifiers


Montague:

 English as a formal Language, PTQ, UG
1- English, like any other NL, is just an interpreted formal language
  “There is in my opinion no important theoretical difference between natural
  language and artificial languages of logicians; indeed, I consider it possible to
  comprehend the syntax and semantics of both kinds of language within a single
  natural and mathematically precise theory.”
      Universal Grammar
2- NPs are Generalized quantifiers



              Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   21
Unification via generalized Quantifiers


Generalized quantifiers:

Every king of France is bald

∀x [K(x) → B(x)]

What about Most Kings of France are Bald?

We can't analyse it in predicate logic.
Even if we introduce “Q”, a quantifier, meaning “most”

           Qx [K(x) → B(x)]




              Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   22
Unification via generalized Quantifiers




Generalized quantifiers:

         Qx [K(x) → B(x)]

 This impossibility shows that Quantifiers are not just variable binding operators but
 true second order concepts letting us define and manipulate sets into the universe




             Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008       23
Unification via generalized Quantifiers
Generalized quantifiers:

Back to Frege: with quantifiers, we compare sets

 “Every King of France is bald” means that the set of kings of France is a subset of
 the set of Bald persons.



 Likewise, “Most Kings of France are bald” means that the set of bald kings of
 France is larger than the set of hairy Kings of France



How could we formalize this?



             Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   24
Unification via generalized Quantifiers


Generalized quantifiers:

How could we formalize this?

          [Qx K(x)] [B(x) ]

Reading: “For most x such that x is King of France, x is bald ”

Works for every and some

          [∀x K(x)] [B(x) ]

          [∃x K(x)] [B(x) ]




             Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   25
Unification via generalized Quantifiers


Generalized quantifiers:

It even works for proper names

            [Johnx K(x)] [B(x)]

Reading: “For John x such that John is King of France, John is bald ”

Nota:

        [Johnx K(x)] is not a name but a quantifier denoting the intersection of all the
 sets of which John King of France is an element




               Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008    26
Unification via generalized Quantifiers


Generalized quantifiers:

          [∀x K(x)] [B(x) ]

          [∃x K(x)] [B(x) ]

          [Qx K(x)] [B(x) ]

          [Johnx K(x)] [B(x)]

Note that we are back to the grammatical NP/VP structure.

Moreover, in generative linguistics also QNPs are Generalized Quantifiers cf. for
 example Fox[2002]



             Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   27
Unification via generalized Quantifiers



Fox (2002): 

«QNPs denote second order predicates. They convey information about basic (first
order) predicates like 'tall'; they tell us something about the set of individuals that a
given (first order) predicate is true of. So in the sentences in (1) the relevant
predicate is tall. And, given the meaning of the specific QNPs, the sentences convey
the information that the predicate is true of at least one girl, (1)a, many girls, (1)b,
every girl, (1)c, or no girl, (1)d. »




            Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008        28
Unification via generalized Quantifiers



Quantifier Raising just reveal the true logical structure of Natural Language
sentences

ergo

        Linguist's LF = Logicians Logical Form




            Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   29
Interest & Problems


    Interest:

        * Naturalising logic

        * Flesh out the LOT hypothesis

        * Clarify the debates over contextualism [Wordnet, conceptual
    graphs → Free Enrichment]

        * Translate our reasoning mechanisms into a well known
    mathematical theory: set theory




      Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008   30
Interest & Problems



    Problems:

        * How to go from parallelism to identification

        * Even if LF is like a GQ Logic, it doesn't necessarily mean that rules of
    Grammar are logical rules of inference

        * Is using grammar identical to reasoning? Looks awfully whorfian




          Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008    31
Many Thanks to you,




to Pierre Stanislas Grialou                               & to Santiago Echeverry

             Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008     32

Weitere ähnliche Inhalte

Was ist angesagt?

Principles and parameters of grammar report
Principles and parameters of grammar reportPrinciples and parameters of grammar report
Principles and parameters of grammar reportAubrey Expressionista
 
Bosch1991a bermuda
Bosch1991a bermudaBosch1991a bermuda
Bosch1991a bermudagorin2008
 
7. ku gr.sem 2013: Syntax
7. ku gr.sem 2013: Syntax7. ku gr.sem 2013: Syntax
7. ku gr.sem 2013: SyntaxTikaram Poudel
 
Lecture 2: From Semantics To Semantic-Oriented Applications
Lecture 2: From Semantics To Semantic-Oriented ApplicationsLecture 2: From Semantics To Semantic-Oriented Applications
Lecture 2: From Semantics To Semantic-Oriented ApplicationsMarina Santini
 
NLP_KASHK:Context-Free Grammar for English
NLP_KASHK:Context-Free Grammar for EnglishNLP_KASHK:Context-Free Grammar for English
NLP_KASHK:Context-Free Grammar for EnglishHemantha Kulathilake
 
Prosodic Morphology
Prosodic Morphology Prosodic Morphology
Prosodic Morphology Maroua Harrif
 
Black max models-and_metaphors_studies_in_language and philosophy
Black max models-and_metaphors_studies_in_language and philosophyBlack max models-and_metaphors_studies_in_language and philosophy
Black max models-and_metaphors_studies_in_language and philosophymarce c.
 
Talmy lexicalizationpatterns
Talmy lexicalizationpatternsTalmy lexicalizationpatterns
Talmy lexicalizationpatternsBrendaWongUdye
 
International Journal of Humanities and Social Science Invention (IJHSSI)
International Journal of Humanities and Social Science Invention (IJHSSI)International Journal of Humanities and Social Science Invention (IJHSSI)
International Journal of Humanities and Social Science Invention (IJHSSI)inventionjournals
 
Idealization in cognitive and generative linguistics
Idealization in cognitive and generative linguisticsIdealization in cognitive and generative linguistics
Idealization in cognitive and generative linguisticsBarbara Konat
 
Principles of parameters
Principles of parametersPrinciples of parameters
Principles of parametersVelnar
 
Debugging Chomsky's Hierarchy
Debugging Chomsky's HierarchyDebugging Chomsky's Hierarchy
Debugging Chomsky's HierarchyHussein Ghaly
 
MELT 104 - Construction Grammar
MELT 104 - Construction GrammarMELT 104 - Construction Grammar
MELT 104 - Construction GrammarGlynn Palecpec
 

Was ist angesagt? (20)

Principles and parameters of grammar report
Principles and parameters of grammar reportPrinciples and parameters of grammar report
Principles and parameters of grammar report
 
Bosch1991a bermuda
Bosch1991a bermudaBosch1991a bermuda
Bosch1991a bermuda
 
7. ku gr.sem 2013: Syntax
7. ku gr.sem 2013: Syntax7. ku gr.sem 2013: Syntax
7. ku gr.sem 2013: Syntax
 
Lecture 2: From Semantics To Semantic-Oriented Applications
Lecture 2: From Semantics To Semantic-Oriented ApplicationsLecture 2: From Semantics To Semantic-Oriented Applications
Lecture 2: From Semantics To Semantic-Oriented Applications
 
NLP_KASHK:Context-Free Grammar for English
NLP_KASHK:Context-Free Grammar for EnglishNLP_KASHK:Context-Free Grammar for English
NLP_KASHK:Context-Free Grammar for English
 
Prosodic Morphology
Prosodic Morphology Prosodic Morphology
Prosodic Morphology
 
Black max models-and_metaphors_studies_in_language and philosophy
Black max models-and_metaphors_studies_in_language and philosophyBlack max models-and_metaphors_studies_in_language and philosophy
Black max models-and_metaphors_studies_in_language and philosophy
 
Talmy lexicalizationpatterns
Talmy lexicalizationpatternsTalmy lexicalizationpatterns
Talmy lexicalizationpatterns
 
International Journal of Humanities and Social Science Invention (IJHSSI)
International Journal of Humanities and Social Science Invention (IJHSSI)International Journal of Humanities and Social Science Invention (IJHSSI)
International Journal of Humanities and Social Science Invention (IJHSSI)
 
Textual equivalence
Textual equivalenceTextual equivalence
Textual equivalence
 
Idealization in cognitive and generative linguistics
Idealization in cognitive and generative linguisticsIdealization in cognitive and generative linguistics
Idealization in cognitive and generative linguistics
 
Principles of parameters
Principles of parametersPrinciples of parameters
Principles of parameters
 
Debugging Chomsky's Hierarchy
Debugging Chomsky's HierarchyDebugging Chomsky's Hierarchy
Debugging Chomsky's Hierarchy
 
Textual equivalence
Textual equivalenceTextual equivalence
Textual equivalence
 
cohesion
cohesioncohesion
cohesion
 
Paris manfred
Paris manfredParis manfred
Paris manfred
 
MELT 104 - Construction Grammar
MELT 104 - Construction GrammarMELT 104 - Construction Grammar
MELT 104 - Construction Grammar
 
Lfg and gpsg
Lfg and gpsgLfg and gpsg
Lfg and gpsg
 
Phrase structure grammar
Phrase structure grammarPhrase structure grammar
Phrase structure grammar
 
CONTEXT RETENTION
CONTEXT RETENTIONCONTEXT RETENTION
CONTEXT RETENTION
 

Andere mochten auch

Andere mochten auch (20)

Phonetics: The Sounds of Language
Phonetics: The Sounds of LanguagePhonetics: The Sounds of Language
Phonetics: The Sounds of Language
 
Grammar
GrammarGrammar
Grammar
 
Serco corporate brochure
Serco corporate brochureSerco corporate brochure
Serco corporate brochure
 
Periodontics gum lift
Periodontics gum liftPeriodontics gum lift
Periodontics gum lift
 
League Of Nations Revision
League Of Nations RevisionLeague Of Nations Revision
League Of Nations Revision
 
Nepal
NepalNepal
Nepal
 
Mathura of my Dreams by Vasundhara Agarwal
Mathura of my Dreams by Vasundhara AgarwalMathura of my Dreams by Vasundhara Agarwal
Mathura of my Dreams by Vasundhara Agarwal
 
Where We End
Where We EndWhere We End
Where We End
 
Idea 1
Idea 1Idea 1
Idea 1
 
Sound of Our Action
Sound of Our ActionSound of Our Action
Sound of Our Action
 
Nou paradigma
Nou paradigmaNou paradigma
Nou paradigma
 
ประวัติ
ประวัติประวัติ
ประวัติ
 
Value Flow ScoreCards - For better strategies, coverage & processes (2008)
Value Flow ScoreCards - For better strategies, coverage & processes (2008)Value Flow ScoreCards - For better strategies, coverage & processes (2008)
Value Flow ScoreCards - For better strategies, coverage & processes (2008)
 
Asaco pla salut 2013
Asaco pla salut 2013Asaco pla salut 2013
Asaco pla salut 2013
 
Thoughts and Actions
Thoughts and ActionsThoughts and Actions
Thoughts and Actions
 
Cerita kanak
Cerita kanakCerita kanak
Cerita kanak
 
Isn power point copy
Isn power point   copyIsn power point   copy
Isn power point copy
 
3d adshock
3d adshock3d adshock
3d adshock
 
Presentation1 power point april 2013
Presentation1   power point april 2013Presentation1   power point april 2013
Presentation1 power point april 2013
 
Embarazo precoz
Embarazo precozEmbarazo precoz
Embarazo precoz
 

Ähnlich wie Unifying Logical Form and The Linguistic Level of LF

Bosch1991a bermuda
Bosch1991a bermudaBosch1991a bermuda
Bosch1991a bermudagorin2008
 
Syntactic Structures (2nd Edition).pdf
Syntactic Structures (2nd Edition).pdfSyntactic Structures (2nd Edition).pdf
Syntactic Structures (2nd Edition).pdfMeryana5
 
An Outline Of Type-Theoretical Approaches To Lexical Semantics
An Outline Of Type-Theoretical Approaches To Lexical SemanticsAn Outline Of Type-Theoretical Approaches To Lexical Semantics
An Outline Of Type-Theoretical Approaches To Lexical SemanticsTye Rausch
 
The Linguistic Components of Contrastive Analysis
The Linguistic Components ofContrastive AnalysisThe Linguistic Components ofContrastive Analysis
The Linguistic Components of Contrastive Analysiszahraa Aamir
 
Numerical Cognition, linguistic relativity and the ontology of numbers
Numerical Cognition, linguistic relativity and the ontology of numbersNumerical Cognition, linguistic relativity and the ontology of numbers
Numerical Cognition, linguistic relativity and the ontology of numbersHady Ba
 
AN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECH
AN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECHAN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECH
AN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECHgerogepatton
 
AN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECH
AN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECHAN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECH
AN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECHijaia
 
Philosophical assumptions in cognitive and generative linguistics
Philosophical assumptions in cognitive and generative linguisticsPhilosophical assumptions in cognitive and generative linguistics
Philosophical assumptions in cognitive and generative linguisticsBarbara Konat
 
Microliguistic contrastive analysis.
Microliguistic contrastive analysis.Microliguistic contrastive analysis.
Microliguistic contrastive analysis.zahraa Aamir
 
The LSA breaks downanalyzes what constitutes a good and bad a.docx
The LSA breaks downanalyzes what constitutes a good and bad a.docxThe LSA breaks downanalyzes what constitutes a good and bad a.docx
The LSA breaks downanalyzes what constitutes a good and bad a.docxarnoldmeredith47041
 
contributions of lexicography and corpus linguistics to a theory of language ...
contributions of lexicography and corpus linguistics to a theory of language ...contributions of lexicography and corpus linguistics to a theory of language ...
contributions of lexicography and corpus linguistics to a theory of language ...ayfa
 
History of linguistics overview
History of linguistics overviewHistory of linguistics overview
History of linguistics overviewJordán Masías
 
05 linguistic theory meets lexicography
05 linguistic theory meets lexicography05 linguistic theory meets lexicography
05 linguistic theory meets lexicographyDuygu Aşıklar
 
Syntactic Analysis Based on Morphological characteristic Features of the Roma...
Syntactic Analysis Based on Morphological characteristic Features of the Roma...Syntactic Analysis Based on Morphological characteristic Features of the Roma...
Syntactic Analysis Based on Morphological characteristic Features of the Roma...kevig
 
SYNTACTIC ANALYSIS BASED ON MORPHOLOGICAL CHARACTERISTIC FEATURES OF THE ROMA...
SYNTACTIC ANALYSIS BASED ON MORPHOLOGICAL CHARACTERISTIC FEATURES OF THE ROMA...SYNTACTIC ANALYSIS BASED ON MORPHOLOGICAL CHARACTERISTIC FEATURES OF THE ROMA...
SYNTACTIC ANALYSIS BASED ON MORPHOLOGICAL CHARACTERISTIC FEATURES OF THE ROMA...kevig
 
Some Issues of Contention.
Some Issues of Contention.Some Issues of Contention.
Some Issues of Contention.zahraa Aamir
 
Ontology as a formal one. The language of ontology as the ontology itself: th...
Ontology as a formal one. The language of ontology as the ontology itself: th...Ontology as a formal one. The language of ontology as the ontology itself: th...
Ontology as a formal one. The language of ontology as the ontology itself: th...Vasil Penchev
 

Ähnlich wie Unifying Logical Form and The Linguistic Level of LF (20)

Bosch1991a bermuda
Bosch1991a bermudaBosch1991a bermuda
Bosch1991a bermuda
 
Transformational grammar
Transformational grammarTransformational grammar
Transformational grammar
 
Structural semantics2
Structural semantics2Structural semantics2
Structural semantics2
 
Fillmore case grammar
Fillmore case grammarFillmore case grammar
Fillmore case grammar
 
Syntactic Structures (2nd Edition).pdf
Syntactic Structures (2nd Edition).pdfSyntactic Structures (2nd Edition).pdf
Syntactic Structures (2nd Edition).pdf
 
An Outline Of Type-Theoretical Approaches To Lexical Semantics
An Outline Of Type-Theoretical Approaches To Lexical SemanticsAn Outline Of Type-Theoretical Approaches To Lexical Semantics
An Outline Of Type-Theoretical Approaches To Lexical Semantics
 
The Linguistic Components of Contrastive Analysis
The Linguistic Components ofContrastive AnalysisThe Linguistic Components ofContrastive Analysis
The Linguistic Components of Contrastive Analysis
 
Numerical Cognition, linguistic relativity and the ontology of numbers
Numerical Cognition, linguistic relativity and the ontology of numbersNumerical Cognition, linguistic relativity and the ontology of numbers
Numerical Cognition, linguistic relativity and the ontology of numbers
 
AN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECH
AN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECHAN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECH
AN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECH
 
AN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECH
AN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECHAN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECH
AN ONTOLOGICAL ANALYSIS AND NATURAL LANGUAGE PROCESSING OF FIGURES OF SPEECH
 
Philosophical assumptions in cognitive and generative linguistics
Philosophical assumptions in cognitive and generative linguisticsPhilosophical assumptions in cognitive and generative linguistics
Philosophical assumptions in cognitive and generative linguistics
 
Microliguistic contrastive analysis.
Microliguistic contrastive analysis.Microliguistic contrastive analysis.
Microliguistic contrastive analysis.
 
The LSA breaks downanalyzes what constitutes a good and bad a.docx
The LSA breaks downanalyzes what constitutes a good and bad a.docxThe LSA breaks downanalyzes what constitutes a good and bad a.docx
The LSA breaks downanalyzes what constitutes a good and bad a.docx
 
contributions of lexicography and corpus linguistics to a theory of language ...
contributions of lexicography and corpus linguistics to a theory of language ...contributions of lexicography and corpus linguistics to a theory of language ...
contributions of lexicography and corpus linguistics to a theory of language ...
 
History of linguistics overview
History of linguistics overviewHistory of linguistics overview
History of linguistics overview
 
05 linguistic theory meets lexicography
05 linguistic theory meets lexicography05 linguistic theory meets lexicography
05 linguistic theory meets lexicography
 
Syntactic Analysis Based on Morphological characteristic Features of the Roma...
Syntactic Analysis Based on Morphological characteristic Features of the Roma...Syntactic Analysis Based on Morphological characteristic Features of the Roma...
Syntactic Analysis Based on Morphological characteristic Features of the Roma...
 
SYNTACTIC ANALYSIS BASED ON MORPHOLOGICAL CHARACTERISTIC FEATURES OF THE ROMA...
SYNTACTIC ANALYSIS BASED ON MORPHOLOGICAL CHARACTERISTIC FEATURES OF THE ROMA...SYNTACTIC ANALYSIS BASED ON MORPHOLOGICAL CHARACTERISTIC FEATURES OF THE ROMA...
SYNTACTIC ANALYSIS BASED ON MORPHOLOGICAL CHARACTERISTIC FEATURES OF THE ROMA...
 
Some Issues of Contention.
Some Issues of Contention.Some Issues of Contention.
Some Issues of Contention.
 
Ontology as a formal one. The language of ontology as the ontology itself: th...
Ontology as a formal one. The language of ontology as the ontology itself: th...Ontology as a formal one. The language of ontology as the ontology itself: th...
Ontology as a formal one. The language of ontology as the ontology itself: th...
 

Unifying Logical Form and The Linguistic Level of LF

  • 1. ECAP 6 — Sixth European Congress of Analytic Philosophy Krakow August, 26, 2008 “Unifying Logical Form & The linguistic Level of LF”  Mouhamadou El Hady BA Institut Jean Nicod ENS-EHESS Paris
  • 2. Overview of the Talk:  The Generative Framework  The Linguistic level of LF  Logicians Logical Form  Unification via generalized quantification  Interest & problems Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 2
  • 3. Introduction:  Logic: an artificial & well defined language + rules of inference − Looking for sound derivations − regardless of actual implementation into the human mind  Generative Linguistics: define an abstract grammar but above all an empirical science looking for the laws actually used by the speaker  Claim: these two research programs converge via their use of Generalized quantification − logical laws of thought Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 3
  • 4. The Generative Framework: What's a Grammar? "The grammar of a language purports to be a description of the intrinsic competence of an ideal speaker-listener. If the grammar is, moreover, perfectly explicit – i.e. if it doesn't simply accept the understanding if intelligent reader but gives an explicit analysis of the activity that the speaker-listener display- we can, with some redundancy, call it a generative grammar. " Noam Chomsky (1965) : Aspects of the theory of syntax Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 4
  • 5. The Generative Framework: What's a Grammar? − (p) Colorless green ideas sleep furiously − (q) Walk pixel hands color john foolishly  Distinction Phonology/Syntax/Semantics: distinct and somewhat independent modules of the language faculty manage these aspects of natural languages  That's why we recognize (p) as a sentence, understand it and judge it absurd when we can't even say of (q) that it really is a sentence. Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 5
  • 6. The linguistic Level of LF “The basic elements we consider are sentences; the grammar generates mental representations of their form and meaning. I will call these representations, respectively, ''phonetic forms'' and LF (which I will read, ''logical form,'' though with a cautionary note)” Chomsky[1980], Rules & Representations, p. 143 ''representations of their form and meaning''→ LF a semantic level? Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 6
  • 7. The Linguistic Level of LF LF as a semantic level? “LF is thus the interface between grammar and the conceptual-intentional properties of language, just as the level of Phonetic Form (PF) is an interface between grammar and the audio-perceptual properties of utterances. LF is not to be equated with the level of semantic structure anymore than PF is to be treated as a level specifying the sound waves of any given utterance. It expresses only aspects of semantic structure that are [...] contributed by grammar.” CT James Huang [1994], “Logical Form” Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 7
  • 8. The Linguistic Level of LF  SS: Surface Structures  PF: Phonetic Forms  LF: Logical Form We are only interested in the SS/ LF branch Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 8
  • 9. The Linguistic Level of LF Defining Move α Minimalism: • Only one rule: Movement (Move α) with constraints • E.g.: Move NPs to the head of the sentence leaving traces at their former place “The transformational mapping to S-structure can be reduced to (possibly multiple application of) a single general rule “Move α,” where α is an arbitrary phrase category” Chomsky [1980], Rules & Representations, p. 145 Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 9
  • 10. The Linguistic Level of LF  Why should we add a covert level of LF? To understand the interpretation of Wh-phrases To uncover the reference of pronouns To explain syntactically why some ambiguities arise: Ex: Let's consider this sentence with two NPs : (d) Every philosopher loves some linguist Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 10
  • 11. The Linguistic Level of LF (d) Every philosopher loves some linguist Application of Move α : Quantifier Raising – Move the NPs at the head of the sentence – Leave traces at their former place – Traces are kind of variables like in FOL's predicate language Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 11
  • 12. The Linguistic Level of LF Two possibles LF formula: (d1) [Every philosopher]1[Some linguist]2 [t1 loves t2] (d2) [Some linguist]2 [Every philosopher]1[t1 loves t2] Hence the ambiguity: possibly many linguists in (d1) but same beloved linguist in (d2). Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 12
  • 13. The Linguistic Level of LF Traces: (d1) [Every philosopher]1[Some linguist]2 [t1 loves t2] (d2) [Some linguist]2 [Every philosopher]]1[t1 loves t2] The traces t1 & t2 variables binded by the NPs Variables, Quantified Noun Phrases, Rings a bell? That's logic, isn't it? But lets do some real logic and we will come back to see if we can unify logic and linguistics Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 13
  • 14. Logician's Logical Form Logic: Consensus 'till Montague: natural languages can't handle the task of formalizing our reasonings “Languages are not made so as to match logic's ruler. Even the logical element in language seems hidden behind pictures that are not always accurate.” Frege [1906] Letter to Husserl “The syntax of ordinary language, as is well known, is not quite adequate for this purpose. It does not in all cases prevent the construction of nonsensical pseudo- propositions” Wittgenstein [1929]: Some remarks on Logical Form Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 14
  • 15. Logician's Logical Form Russell On denoting: Breaking up Denoting Phrases: “The phrase per se has no meaning, because in any proposition in which it occurs the proposition, fully expressed, does not contain the phrase, which has been broken up” “Consider the next the proposition 'all men are mortal'. This proposition is really hypothetical and states that if anything is a man, it is mortal.” Russell (1905) ∀ x [M(x) →D(x)] → Grammatical structure is irrelevant Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 15
  • 16. Logician's Logical Form Solution: Translate NL utterances into an artificial language displaying their “real” logical structure and using correct laws of entailment Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 16
  • 17. Logician's Logical Form Frege on quantification: Numbers and quantifiers (∀, ∃) are second order predicates expressing relations between concepts rather than properties of objects For example: (f) The shirt is red (g) Every shirt is red Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 17
  • 18. Logician's Logical Form Frege on quantification: Universe: {shirts} (f): ( ) is red is attributed to an object (a particular shirt) (g): a bit more complex Analyse of (g) according to Frege: two concepts: ( ) is a shirt and ( ) is red, (g) states a relation between these concepts namely that whatever satisfy ( ) is a shirt satisfy ( ) is red. So quantifiers are second order concepts expressing something about first order concepts Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 18
  • 19. Logician's Logical Form Russell On denoting: Variables first : “I take the notion of the variable as fundamental...” → Quantifiers are just variable binding operators Do not really state that Quantifiers are not second order predicates but limits the explorations in this direction Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 19
  • 20. Logician's Logical Form Frege, Russell, Quine.... Consensus: Natural languages are improper for good reasoning. A logician has to use an artificial language in order to display the real structure and make correct inferences. Enters Montague! Respected philosopher & logician Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 20
  • 21. Unification via generalized Quantifiers Montague: English as a formal Language, PTQ, UG 1- English, like any other NL, is just an interpreted formal language “There is in my opinion no important theoretical difference between natural language and artificial languages of logicians; indeed, I consider it possible to comprehend the syntax and semantics of both kinds of language within a single natural and mathematically precise theory.” Universal Grammar 2- NPs are Generalized quantifiers Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 21
  • 22. Unification via generalized Quantifiers Generalized quantifiers: Every king of France is bald ∀x [K(x) → B(x)] What about Most Kings of France are Bald? We can't analyse it in predicate logic. Even if we introduce “Q”, a quantifier, meaning “most” Qx [K(x) → B(x)] Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 22
  • 23. Unification via generalized Quantifiers Generalized quantifiers: Qx [K(x) → B(x)] This impossibility shows that Quantifiers are not just variable binding operators but true second order concepts letting us define and manipulate sets into the universe Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 23
  • 24. Unification via generalized Quantifiers Generalized quantifiers: Back to Frege: with quantifiers, we compare sets “Every King of France is bald” means that the set of kings of France is a subset of the set of Bald persons. Likewise, “Most Kings of France are bald” means that the set of bald kings of France is larger than the set of hairy Kings of France How could we formalize this? Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 24
  • 25. Unification via generalized Quantifiers Generalized quantifiers: How could we formalize this? [Qx K(x)] [B(x) ] Reading: “For most x such that x is King of France, x is bald ” Works for every and some [∀x K(x)] [B(x) ] [∃x K(x)] [B(x) ] Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 25
  • 26. Unification via generalized Quantifiers Generalized quantifiers: It even works for proper names [Johnx K(x)] [B(x)] Reading: “For John x such that John is King of France, John is bald ” Nota: [Johnx K(x)] is not a name but a quantifier denoting the intersection of all the sets of which John King of France is an element Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 26
  • 27. Unification via generalized Quantifiers Generalized quantifiers: [∀x K(x)] [B(x) ] [∃x K(x)] [B(x) ] [Qx K(x)] [B(x) ] [Johnx K(x)] [B(x)] Note that we are back to the grammatical NP/VP structure. Moreover, in generative linguistics also QNPs are Generalized Quantifiers cf. for example Fox[2002] Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 27
  • 28. Unification via generalized Quantifiers Fox (2002):  «QNPs denote second order predicates. They convey information about basic (first order) predicates like 'tall'; they tell us something about the set of individuals that a given (first order) predicate is true of. So in the sentences in (1) the relevant predicate is tall. And, given the meaning of the specific QNPs, the sentences convey the information that the predicate is true of at least one girl, (1)a, many girls, (1)b, every girl, (1)c, or no girl, (1)d. » Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 28
  • 29. Unification via generalized Quantifiers Quantifier Raising just reveal the true logical structure of Natural Language sentences ergo Linguist's LF = Logicians Logical Form Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 29
  • 30. Interest & Problems  Interest: * Naturalising logic * Flesh out the LOT hypothesis * Clarify the debates over contextualism [Wordnet, conceptual graphs → Free Enrichment] * Translate our reasoning mechanisms into a well known mathematical theory: set theory Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 30
  • 31. Interest & Problems  Problems: * How to go from parallelism to identification * Even if LF is like a GQ Logic, it doesn't necessarily mean that rules of Grammar are logical rules of inference * Is using grammar identical to reasoning? Looks awfully whorfian Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 31
  • 32. Many Thanks to you, to Pierre Stanislas Grialou & to Santiago Echeverry Hady Ba, "Unifying Logical Form & FL", ecap 6, Krakow, August 2008 32