+ All Categories
Home > Documents > We are taking the language L to be a way of computing expressions, a recursive definition of a set...

We are taking the language L to be a way of computing expressions, a recursive definition of a set...

Date post: 03-Apr-2015
Category:
Upload: agate-turpin
View: 106 times
Download: 2 times
Share this document with a friend
Popular Tags:
45
We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. •(i) a set of features i) principles for assembling features into lexical Thus, UG might postulate that FL provides: •(iii) operations that apply successively to form syntactic objects of greater complexity; call them CHL, the computational system for human language Sémantique et Grammaire Générative
Transcript
Page 1: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP.

•(i) a set of features

•(ii) principles for assembling features into lexical items

Thus, UG might postulate that FL provides:

•(iii) operations that apply successively to form syntactic objects of greater complexity; call them CHL, the computational system for human language

Sémantique et Grammaire Générative

Page 2: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

Which book do you think that Mary read?

Énumération: which, book, Mary, think, that, you, do

Dérivation

Forme « phonologique » Forme « logique »

/witbukdujuinkǽtmerired/ quel x, x = livre, tu penses que marie a lu x

Exemple

Page 3: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

Irene Heim & Angelika Kratzer, Semantics in Generative Grammar

Associer des contreparties sémantiques non plus à des « règles » mais à des principes généraux tels que:

- merge

- move

Heim & Kratzer, 1998

Page 4: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

exemple

Which book do you think that Mary read?

Forme « logique »

quel x, x = livre, tu penses que marie a lu x

Page 5: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

exemple

Forme « logique »

a_lu: z. y. a_lu(y,z)

marie:D

penser:x. y. penser(y,x)

tu:D

livre:x.livre(x)

quel:?

Which x (x = book) do you think that Mary read x

Page 6: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

exemple

Forme « logique »

[z. y. a_lu(y,z)](x) ->y.a_lu(y, x)

tu:D

quel:?

Which x (x = book) do you think that Mary read x

penser:x. y. penser(y,x)

livre:x.livre(x)

Page 7: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

exemple

Forme « logique »

tu:D

quel:?

Which x (x = book) do you think that Mary read x

livre:x.livre(x)

penser:x. y. penser(y,x)

[y.a_lu(y, x)](Marie) ->a_lu(Marie, x)

Page 8: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

exemple

Forme « logique »quel:?

Which x (x = book) do you think that Mary read x

a_lu(Marie, x)

penser:[x. y. penser(y,x)](a_lu(Marie, x) ->y. penser(y, a_lu(Marie, x))

livre:x.livre(x)

Page 9: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

exemple

Forme « logique »quel:?

Which x (x = book) do you think that Mary read x

a_lu(Marie, x)livre:x.livre(x)

penser:[y. penser(y, a_lu(Marie, x))](tu) ->penser(tu, a_lu(Marie, x))

Page 10: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

après?

Forme « logique »quel:?

Which x (x = book) do you think that Mary read x

livre:x.livre(x)

penser:penser(tu, a_lu(Marie, x))

Page 11: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

proposition

quel:?

livre:x.livre(x)

penser:penser(tu, a_lu(Marie, x))

x. penser(tu, a_lu(Marie, x))

quel(x, livre(x) penser(tu,a_lu(Marie, x))

Page 12: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

proposition

quel:?

livre:x.livre(x)

penser:penser(tu, a_lu(Marie, x))

x. penser(tu, a_lu(Marie, x))

quel(x, livre(x) penser(tu,a_lu(Marie, x))

Une fonction ayant pour arguments deux propriétéset qui retourne une proposition sous forme de question

Page 13: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

problème

• D’où vient le pas d’abstraction :

penser:penser(tu, a_lu(Marie, x))

x. penser(tu, a_lu(Marie, x))

Page 14: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

SNwhich book

CP

C’

Cdo

SNyou

Vthink

that

SNMary

Vread

SNt

VP

CP

V’

V’

VP

Page 15: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

SNwhich book

CP

C’

Cdo

SNyou

Vthink

that

SNMary

Vread

SNt

VP

CP

V’

V’

VP

Page 16: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

SNwhich book

CP

C’

Cdo

SNyou

Vthink

that

SNMary

Vread

SNt

VP

CP

V’

V’

VP

t<<e, t>, t>P.?(x, book(x) & P(x))

think(you, read(mary, x))

TYPE MISMATCH

Page 17: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

SNwhich book1

CP

C’

Cdo

SNyou

Vthink

that

SNMary

Vread

SNt1

VP

CP

V’

V’

VP

t<<e, t>, t>P.?(x, book(x) & P(x))

think(you, read(mary, x))1

x. think(you, read(mary, x))

BINDER

<e, t>

Page 18: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

SNwhich book1

CP

C’

Cdo

SNyou

Vthink

that

SNMary

Vread

SNt1

VP

CP

V’

V’

VP

t<<e, t>, t>P.?(x, book(x) & P(x))

think(you, read(mary, x))

x. think(you, read(mary, x))

OU BIEN…

<e, t>

Page 19: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

SNwhich book1

CP

Cdo

SNyou

Vthink

that

SNMary

Vread

VP

CP

V’

V’

VP

tROTATE !!!!

Page 20: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

SNwhich book1

CP

Cdo

SNyou

Vthink

that

SNMary

Vread

VP

CP

V’

V’

VP

xceci est un arbre de preuve

Page 21: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

SNwhich book1

CP

Cdo

SNyou

Vthink

that

SNMary

Vread

VP

CP

V’

V’

VP

xceci est un arbre de preuve

hypothèse

déchargement de l’hypothèse

e

t

e t(e t) t

Page 22: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

règles

A B A

B

« élimination » de

[A]hypothèse

B

A B

Déchargement de l’hypothèse

« introduction » de

Page 23: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

Autre exemple

Nskieur

APgrenoblois

Nskieur grenoblois

Dun

DPun skieur grenoblois

Vaime

VPaime un skieur grenoblois

NPMarie

SMarie aime un skieur grenoblois

Page 24: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

Marie aime un skieur grenoblois

Nskieur

APgrenoblois

Nskieur grenoblois

Dun

DPun skieur grenobloisP.ex(x, ski(x)&gre(x)&P(x))

Vaime

VPaime un skieur grenoblois

NPMarie

SMarie aime un skieur grenoblois

Page 25: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

Déplacement (covert)

DPun skieur grenobloist(race)

Vaime

VPaime un skieur grenoblois

NPMarie

SMarie aime un skieur grenoblois

N AP

ND

P.ex(x, ski(x)&gre(x)&P(x))

Page 26: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

Mais…

Vaime

VPaime un skieur grenoblois

NPMarie

S aime(Marie, xm)Marie aime un skieur grenoblois

N AP

ND

P.ex(x, ski(x)&gre(x)&P(x))

DPun skieur grenobloist(race) -> variable xm

Encore mismatch!

Page 27: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

solution Heim & Kratzer

N AP

ND

P.ex(x, ski(x)&gre(x)&P(x))

Vaime

VPaime un skieur grenoblois

NPMarie

S aime(Marie, xm)Marie aime un skieur grenoblois

DPun skieur grenobloist1(race) -> variable xm

1

Heim & Kratzer: binder

xm. aime(Marie, xm)

Page 28: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

variante

N AP

ND

P.ex(x, ski(x)&gre(x)&P(x))

Vaime

VPaime un skieur grenoblois

NPMarie

S aime(Marie, xm)Marie aime un skieur grenoblois

DPun skieur grenobloist1(race) -> variable xm

S xm. aime(Marie, xm)

Page 29: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

Présentation sous forme de preuve

S xm. aime(Marie, xm)

N AP

ND

P.ex(x, ski(x)&gre(x)&P(x))

Vaime

VPaime un skieur grenoblois

NPMarie

S aime(Marie, xm)Marie aime un skieur grenoblois

DPun skieur grenobloist1(race) -> variable xm

Page 30: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

Vers un système logique

• Cf. déduction naturelle (document)

• mais quel système de déduction naturelle?

Page 31: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

Différences avec la logique classique

• En logique classique :

A, A(A B) |-- A B, mais aussi:

A, A(A B) |-- B (A peut être utilisé deux fois)• Aussi:

A, B |-- B (A utilisé 0 fois!)• Dans un calcul syntaxique, les prémisses ne

sont pas réutilisables

ex : n, n(n s) |-- ns (pas s!)

Page 32: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

Logique classique et logique intuitionnistecf. règles de la déduction naturelle

Règles d’introduction pour:

• Règles d’élimination pour:

Logique classique : rajouter règle d’élimination de la double négation

Logique intuitionniste

Page 33: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

Logique intuitionniste

• Une preuve possède une et une seule conclusion

• Les prémisses = les inputs• La conclusion = l’output• donc une preuve peut être vue comme une

fonction:

A1, …., An B• Il y a un flux d’information dans une direction

privilégiée : des inputs vers l’output

Page 34: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

calcul des séquents

• Gentzen, 1934

• (voir document)

• Logique intuitionniste :– séquents asymétriques : A1, …, An|-- B

• Logique classique :– séquents symétriques : A1, …,An|-- B1,…,Bm

(virgule à gauche : comme un , virgule à droite : comme un )

Page 35: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

représentations géométriques

• Logique intuitionniste :– Les preuves sont des arbres (plus ou moins

enrichis avec des annotations!)

• Logique classique :– Les preuves sont : ?

(des réseaux?)

Page 36: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

Le calcul de Lambek

• Une préfiguration de la logique linéaire…

• Cependant : reste un calcul intuitionniste (les preuves sont représentées par des arbres)

• Sensibilité aux ressources : y compris à l’ordre

Page 37: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

calcul des séquents

Séquent (intuitionniste)

BAAAA ni ,...,,...,, 21

antécédent conséquent

Page 38: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

pour prouver : C ,,,A/B

prouvez :

B

puis prouvez :

A C ,,

Page 39: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

Calcul de Lambek(séquents)

AA

CBACBA

,\,,,,

CABCBA

,,/,,,

BAAB/

,

ABAB\

,

CCAA

, ,, cut

Page 40: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

snsnssnsnsn)//(sn)\(nnnsn /)\(/

Page 41: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

snsnssnsnnnsn /)\(/ sn)//(sn)\(n

sn\nnnnsnsn/ssnssnsn //)\(

Page 42: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

snsnssnsnnnsn /)\(/ sn)//(sn)\(n

sn\nnnnsnsnssnsn //)\( sn/sssnsnssnsn /)\(

Page 43: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

snsnssnsnnnsn /)\(/ sn)//(sn)\(n

sn\nnnnsnsnssnsn //)\( sn/sssnsn sns)/\(snsssnsnsnsn \

Page 44: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

snsnssnsnnnsn /)\(/ sn)//(sn)\(n

sn\nnnnsnsnssnsn //)\( sn/sssnsn sns)/\(snssnsnsn s\snsssnsn

Page 45: We are taking the language L to be a way of computing expressions, a recursive definition of a set EXP. (i) a set of features (ii) principles for assembling.

snsnssnsnnnsn /)\(/ sn)//(sn)\(n

snnnsnsnssnsn n\nsn/s //)\(ssnsn sns)/\(snssnsnsn s\snsssnsn

snnnsnnn /snsnnn


Recommended