day 3 cg approaches to information structure rules and
play

Day 3: CG approaches to information structure Rules and derivations - PowerPoint PPT Presentation

Day 3: CG approaches to information structure Rules and derivations Functor categories can combine with their arguments by the following rules: Combinatory Categorial Grammar (CCG; Steedman 2000a,b) CCG in a nutshell (2) Forward


  1. Day 3: CG approaches to information structure Rules and derivations • Functor categories can combine with their arguments by the following rules: • Combinatory Categorial Grammar (CCG; Steedman 2000a,b) – CCG in a nutshell (2) Forward application ( > ) – Structure, intonation, and information structure X/Y Y ⇒ X – The two dimensions of information structure (3) Backward application ( < ) – Combinatory Prosody Y X \ Y ⇒ X • Other Categorial Grammar approaches: • Derivations are written as shown below, on the left side. Note the direct correspondence to the upside-down constituency tree shown on the right. – Multi-Modal Combinatory Categorial Grammar (Kruijff and Baldridge 2004) – Dependency Grammar Logic (Kruijff 2001) Marcel proved Marcel proved completeness completeness NP V NP NP ( S \ NP ) / NP NP > S \ NP VP < S S 1/33 CCG in a nutshell 3/33 CCG in a nutshell Semantics and Principle of Type Transparency • The lexical categories can be augmented with an explicit identification of their • Syntactically potent elements such as verbs are associated with a syntactic semantic interpretation and the rules of functional application are accordingly category that identifies them as functions and specifies the type and expanded with an explicit semantics. directionality of their arguments and the type of their result. (4) proved := ( S \ NP ) /NP : prove ′ • A “result leftmost” notation is used: (5) Forward application ( > ) X/Y : f Y : a ⇒ X : fa – α/β is a rightward-combining functor over a domain β into a range α – α \ β is the corresponding leftward-combining functor. • The semantic interpretation of all combinatory rules is fully determined by the – α and β may themselves be functional categories. Principle of Type Transparency : (1) proved := ( S \ NP ) /NP All syntactic categories reflect the semantic type of the associated logical form, and all syntactic combinatory rules are type-transparent versions of one of a small number of semantic operations over functions including application, composition, and type-raising. CCG in a nutshell 2/33 CCG in a nutshell 4/33

  2. Example derivation with semantics Combinators • In order to account for coordination of contiguous strings that do not constitute traditional constituents, CCG allows certain operations on functions called “combinators”, including the rule of functional composition in (7). Marcel proved completeness NP : marcel ′ (7) Forward composition ( > B ) ( S \ NP ) / NP : prove ′ NP : completeness ′ S \ NP : prove ′ completeness ′ > X/Y : f Y/Z : g ⇒ X/Z : λx.f ( gx ) < S : prove ′ completeness ′ marcel ′ • CCG includes type-raising rules, which turn arguments into functions over functions-over-such-arguments. (8) Forward type-raising ( > T ) (9) Backward type-raising ( < T ) X : a ⇒ T/ ( T \ X ) : λf.fa X : a ⇒ T \ ( T/X ) : λf.fa X ranges over argument categories (e.g., NP and PP). The rules are order-preserving, e.g., (8) can turn an NP into a rightward-looking function over leftward functions, preserving the linear order of subjects and predicates. CCG in a nutshell 5/33 CCG in a nutshell 7/33 More rule schemata Non-standard surface structures CCG includes linguistically motivated rule schemata such as the one for • Complement-taking verbs like think , VP/S, can compose with fragments like coordination of constituents of like type shown below: Marcel proved , S/NP, which accounts for right-node raising (10), and also provides the basis for an analysis of unbounded dependencies (11). (6) Coordination ( < & > ) X conj X ⇒ X (10) [I disproved] S/NP and [you think that Marcel proved] S/NP completeness. (11) the result that [you think that Marcel proved] S/NP Strings such as you think that Marcel proved are taken to be surface constituents of type S/NP. CCG in a nutshell 6/33 CCG in a nutshell 8/33

  3. Non-standard surface structures are licensed throughout Structure and intonation • Steedman assumes that the non-traditional constituents motivated for Steedman’s claims: right-node raising and similar coordinations are also possible constituents of • Surface structure and information structure coincide, the latter simply non-coordinate sentences like Marcel proved completeness . consisting in the interpretation associated with a constituent analysis of the sentence. Marcel proved completeness NP : marcel ′ ( S \ NP ) / NP : prove ′ NP : completeness ′ • Intonation coincides with surface structure (and hence information structure) > T < T S / ( S \ NP ) : λ f . f marcel ′ S \ ( S / NP ) : λ p . p completeness ′ in the sense that all intonational boundaries coincide with syntactic boundaries. > B S / NP : λ x . prove ′ x marcel ′ < S : prove ′ completeness ′ marcel ′ As a result, fragments such as Marcel proved in (12c), are not only prosodic constituents but surface syntactic constituents, complete with interpretations. Marcel proved completeness > B NP : marcel ′ ( S \ NP ) / NP : prove ′ NP : completeness ′ > T < T S / ( S \ NP ) : λ f . f marcel ′ ( S \ NP ) \ (( S \ NP ) / NP ) : λ p . p completeness ′ (12) a. Marcel proved completeness. < S \ NP : λ y . prove ′ completeness ′ y < S S S : prove ′ completeness ′ marcel ′ b. Marcel c. VP ?P completeness • The Principle of Type Transparency guarantees that all such non-standard proved completeness Marcel proved derivations yield identical interpretations. CCG in a nutshell 9/33 Structure, intonation, and information structure 11/33 Motivating non-standard surface structures Intonation and Information Structure • A sequence of one or more pitch accents followed by a boundary is referred to as an intonational phrasal tune . • According to Steedman (2000a), the non-standard surface structures are not spurious ambiguities but relevant since they subsume the intonation structures • Claim: phrasal tunes in this sense are associated with specific discourse needed to explain the possible intonation contours for sentences of English. meanings distinguishing information type and/or propositional attitude. • Intonational boundaries contribute to determining which of the possible (13) Q: I know who proved soundness. But who proved completeness ? combinatory derivations is intended. A: ( Marcel ) (proved completeness ). H* L L+H* LH% • The interpretations of the constituents that arise from these derivations are related to semantic distinctions of information structure and discourse focus. (14) Q: I know which result Marcel predicted . But which result did Marcel prove ? • Steedman’s claims: A: (Marcel proved ) ( completeness ). – Where intonational boundaries are present, they contribute to disambiguation. L+H* LH% H* LL% – Conversely, any such boundaries must be consistent with some syntactic • Evidence: Exchanging the answer tunes between the two contexts in (13) and derivation, or ill-formedness will result. (14) yields complete incoherence. CCG in a nutshell 10/33 Structure, intonation, and information structure 12/33

  4. The two dimensions of Information Structure Intonationally unmarked themes/rhemes • Theme and Rheme : • There also are intonationally unmarked themes : – The theme is to be thought of as that part of an utterance which connects (15) Q: Which result did Marcel prove? it to the rest of the discourse. A: (Marcel proved) ( completeness ). – The rheme is that part of an utterance that advances the discussion by H* LL% contributing novel information. (16) Q: What do you know about Marcel? A: (Marcel) (proved completeness ). • Focus and Background : H* LL% – The information marked by the pitch accent is called the focus , distinguishing theme focus and rheme focus, where necessary. • The same contour can also occur with an all-rheme utterance: – The term background is used for the part unmarked by pitch accent or (17) Guess what? (Marcel proved completeness !) boundary. H* LL% The two dimensions of Information Structure 13/33 The two dimensions of Information Structure 15/33 Theme and Rheme and their intonational realization Semantic characterization of theme and rheme Steedman observes the following relationship for English: • Following Jackendoff (1972), the theme is characterized semantically via functional abstraction, using the notation of λ -calculus, as in (18), • The L+H* LH% tune is associated corresponding to the theme of (14) and (15). with the theme . (18) λx.prove ′ x marcel ′ • The H* L and H* LL% tunes (among others) are associated with the rheme . • When such a function is supplied with an argument in the form of the rheme, it reduces to give a proposition, with the same predicate-argument relation as the canonical sentence. (19) prove ′ completeness ′ marcel ′ The two dimensions of Information Structure 14/33 The two dimensions of Information Structure 16/33

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend