An Intuitive Representation of Human Languages for Translation

1
An Intuitive Representation of Human Languages
for Translation

• Gábor Prószéky
• MorphoLogic
• Faculty of Information Technology,
• Pázmány University

Kalmár Workshop Szeged, October 1-2, 2003
2
Contents

• Some words on Prof. Kalmárs activity in
  computational linguistics
• Problems of human language description with
  formal tools
• A new representation with patterns
• Introduction to machine translation methods
• Application of patterns to translation

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
3
Kalmár languages

• Kalmárs paper in formal language theory An
  Intuitive Representation of Context-Free
  Languages
• Kalmárs activity in machine translation
  (conference in 1962) Representation of
  Languages with the Help of Mathematical
  Structures

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
4
Linguistic representation problems of the 60s

• Dependency structure
• Constituent structure
• X-bar theory X ? (P) X (Q)
• Related structures
• Using transformations

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
5
Structured symbols

• Linguistic categories atomic symbols
• Not enough subcategorization
• Semantic features alive, ...
• Syntactic features countable, ...
• Rule sets instead of rules
• ID/LP

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
6
Feature structures

• DAGs
• Unification problems
• Feature geometry, typed features
• LFG, GPSG, HPSG
• Parsing CF-skeleton features or feature
  structures only?

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
7
Complexity of NL grammars

• RG/FSA not enough
• CF/RTN not enough
• CS ?
• 0/ATN Turing Machine
• Transformations and metarules
• Arguments for and against

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
8
NL grammar formalisms

• Competence and performance?
• Kornai number (left-recursion, center-embedding,
  respectively construction)
• Gradually from unrestricted to regular
• (i) anbn -gtab (n is lost!)
• (ii) anbn -gte,ab,aabb,aaabbb
• Finitization by length
• No structure in FSA finite systems, however, can
  produce structural output

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
9
Syntax and semantics

• Logical representations(e.g. ?x.dog(x),
  ?x.run(x))
• World-knowledge representations(e.g. IS-A,
  PART-OF, INSTANCE-OF)
• Categorial grammar early logical representations
  of syntax (Kalmár)
• DCG interpretation representation
• Rule-to-rule hypothesis

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
10
Conflict handling

• Lexicon meets syntax who is right?
• Lexicon off-line info coming from past
  experiences
• Which is more important in a specific situation?

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
11
Open classes

• Open vs. closed classesthat is, features can or
  cannot be overridden
• Proper names, jabbers, folk etymology, loanwords,
  ...
• Grammar of closed classesminimal grammar

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
12
Finite morphology

• Finite patterns
• Finite number of entries
• Descriptions assigned to entries
• Finite open vs. infinite
  closed
• Underspecified entries for guessing

Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
13
Finite syntax

• Item and arrangement (as in morphology)
• Arrangement describes a rather free
  constituent-order
• Metawords in a meta-dictionary, e.g. (Det (Adj
  (N))) ? DAN
• Cascades without loop

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
14
The plastic box

• John is a boy.
• John is a noun.
• Go is a verb.
• Go is a verb.
• ? is a sign.
• ? is a sign.
• is a . (where is a plastic box)

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
15
Real examples

• (a) Unusual useGo is a verb.POS np ? POS v
• (b) MetaphorMy car drinks a lot.ANIMATE ?
  ANIMATE -
• (c) Unknown entryKalmár is a family name.POS
  np

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
16
Linguistic frames

• Psychology Gestalt
• Morphological complex structures treated as
  frames by humans
• Frames in AI shopping, walking, ...
• As high-level parsing relates to detailed
  on-line analysis

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
17
Translation of human languages
Kalmár Workshop 2003

• old problems (50s)
• direct (60s)
• interlingual (70s)
• transfer (80s)
• examples (90s)

Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
18
Patterns general linguistic information in
lexicalized form

• Short, fully specified patterns are lexical
  entries
• Longer, fully specified entries are multi-word
  expressions
• Partially underspecified patterns are
  collocations, phrasal verbs, idioms
• Totally underspecified patterns are linguistic
  rules
• Pattern/interpretation pairs Translation
  Description Language

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
19
The MetaMorpho principles

• No single words but contextual expressions (in
  form of patterns) only
• Pattern pairs input/interpretation structure
  pairs
• Single pass no separate transfer steps
• Target structure generation by-product of
  parsing

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
20
Jabberwocky

• Twas brillig, and the slighty tovesDid gyre
  and gimble in the wabeAll mimsy were the
  borogroves,And the mone raths outgrabe.

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
21

Kalmár Workshop 2003

• Twas , and the sDid and in the All
  were the s,And the s .

Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
22
Translation rules for Jabberwocky

• twas ? volt
• , and ? , és
• the s did ? a ok tak
• and ? és
• in the ? a ban
• all ? teljesen
• were the s ? k voltak az ok
• the s ? a ok tek

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
23

• Twas , and the sDid and in the All
  were the s,And the s .
• volt, és a oktak és tek a benteljesen
  voltak a okés a ok tek.

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
24
Translation of Jabberwocky

• Dzsebervoki Brillig volt, és a
  szlájti tóvokgájertak és gimbeltek a
  vébbenteljesen mimszik voltak a borogróvokés a
  món rátok autgrébtek.

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
25
An intuitive representation...

• X-bar based structures
• Feature-based descriptions
• Metarules (used off-line)
• Rule-to-rule principle
• Lexicon should be finite but open
• Closed classes belong to the minimal grammar
• Minimal grammar describes basically linguistic
  elements

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
26
An intuitive representation...(contd)

• Linguistic constructions can be described by
  finite patterns
• A huge finite description set is used rather
  than a limited infinite grammar
• In case of conflict, lexical information is
  either redundant or contradicting to the actual
  description
• Known constructions need no real-time analysis
  (Gestalt, frame)

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
27
An intuitive representation... (contd)

• Broken frames are analyzed real-time
• Structural (source/target) pattern pair is
  assigned to every frame to be translated
• Target structure is computed while parsing
  source structure

Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation
28
Kalmár Workshop 2003
Gábor Prószéky An Intuitive Representation of
Human Languages for Translation