Semantics and Pragmatics of NLP Lexical Semantics: Polysemy Alex - PowerPoint PPT Presentation

Data Different Kinds of Polysemy How to Capture Lexical Generlisations Semantics and Pragmatics of NLP Lexical Semantics: Polysemy Alex Lascarides School of Informatics University of Edinburgh university-logo Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations Outline Why A Dictionary Won’t Do: Polysemy! 1 Different Kinds of Polysemy 2 How to Capture Lexical Generlisations 3 university-logo Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations Basic Lessons The meanings of words sometimes predict aspects of their syntactic behaviour (regular polysemy). So lexical knowledge is the interface between world knowledge and linguistic knowledge/processing Discourse/ pragmatic processing interacts with lexical semantics Lexical semantic information can be modelled using a typed feature structure formalism, extended to handle defaults university-logo Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations Introducing bake (1) a. Kim is baking b. The potatoes are baking c. Kim is baking a cake d. Kim is baking a cake for Sandy e. Kim is baking Sandy a cake f. The clay baked g. The clay baked hard h. Sandy was baking, sitting in the hot sun university-logo Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations Sense enumeration: a strawman (2) a. � bake , TakesNP , bake ′ 1 � b. � bake , TakesNP , bake ′ 2 � � bake , TakesNP .NP , bake ′ 3 � c. d. � bake , TakesNP .NP .PP , bake ′ 4 � .NP , bake ′ e. � bake , TakesNP .NP 5 � Senses connected by meaning postulates , such as: ∀ x , y [ bake ′ 3 ( x , y ) → bake ′ 1 ( x ) ∧ bake ′ 2 ( y )] ∀ x , y [ bake ′ 4 ( x , y , z ) → bake ′ 3 ( x , y )] university-logo Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations Problems Doesn’t capture generalisations (cf cook , paint ) No distinction between homonyms (accidental polysemy) and related senses Meaning postulates are unrestricted Cannot list all potential usages. university-logo Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations Terminology homonymy mogul :mound of snow vs. mogul :chinese emperor unpredictable polysemy bank :tilt of plane vs. bank :mound regular polysemy bank :building used by bank :institution nonce not recorded in the lexicon, but interpretable via generative devices; The ham sandwich is getting impatient . institutionalised recorded in the lexicon as derived by a regular process; teacher lexicalised idiosyncratically augment or override regularly derived information; in hospital . established covers institutionalised and lexicalised university-logo Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations More Terminology constructional polysemy a single sense assigned to a lexical entry is contextually specialised (3) a. That book is 500 pages long b. That book introduces syntax. c. That book is 500 pages long and introduces syntax. sense extension separate lexical entries are generated (4) a. That chicken is healthy. b. That chicken is tasty. c.??That chicken is healthy and tasty. Distinction between constructional polysemy and sense university-logo extension is not always clear cut. Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations Back to the lexical entry for transitive bake   transitive-verb ORTH : bake       CAT : v       phrase        SYN:CAT : n  SUBJ :        SEM:INDEX : x    SYN :         �  phrase  �       SYN:CAT : n  SUBCAT :        SEM:INDEX :   y          INDEX : e      _bake3_rel          � � EVENT : e  SEM :       LISZT :     ARG1 : x            ARG2 : y   HPSGish syntax and MRS — neutral wrt various approaches to university-logo lexical semantics Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations bread (NP)   phrase ORTH : bread   � �  CAT : n  SYN :   SUBCAT : ��         INDEX : w     �� ��    SEM : _bread_rel     LISZT :   INST : w    LISZTs are appended when signs are combined    _bake3_rel  � � �� EVENT : e _bread_rel      LISZT :  ,    ARG1 : x  INST : y     ARG2 : y BAKE 3 ( e , x , y ) ∧ BREAD ( y ) university-logo Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations Abbreviated description   transitive-verb ORTH : bake       CAT : v   SUBJ : NP x      SYN :      SUBCAT : � NP y �             INDEX : e      _bake3_rel          � � EVENT : e  SEM :       LISZT :     ARG1 : x            ARG2 : y   university-logo Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations Type for simple transitive verbs   transitive-verb ORTH : string       CAT : v   SUBJ : NP x      SYN :      SUBCAT : � NP y �             INDEX : e      rel         � �  EVENT : e SEM :       LISZT :     ARG1 : x             ARG2 : y   Generalisation: agents realised as subjects and patients as university-logo objects. Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations Type hierarchies ⊤ ✟ ❍❍ ✟ ✟ ❍ sign rel ❛❛❛❛❛❛ ✟ ❍❍ ✟ ✟ ❍ phrase word ✟ ❍❍ ❛ ✟ ✟ ❍ verb noun noun_rel verb_rel ✟ ❍❍ ✟ ✟ ❍ phys_rel art_rel ✟ ❍❍ ✟ ✟ ✟ ❍ ✟ ✟ phys_art_rel transitive-verb tverb_rel _bake3_rel bake3 Information on higher types is inherited by lower types (and lexical entries, such as bake3) university-logo Multiple inheritance is possible Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations Constructional polysemy: Logical Metonymy (5) a. Sandy enjoyed the film/beer/hamburger. b. Sandy enjoyed the book. c. Sandy enjoyed reading the book d. The goat really enjoyed your book. Don’t want to enumerate a million senses of enjoy ! It’s not a purely pragmatic phenomenon: ??John enjoyed the doorstop . ??John enjoyed the tunnel. Generalisation: When NP is an artifact, enjoy NP means enjoy V-ing NP , where V is its purpose. But this can be overridden in sufficiently rich discourse contexts. university-logo Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations Capturing the Generalisation (Simplified): enjoy the book book: enjoy: inherited info; begin , finish etc. coercing 2 3 np book 2 3 *2 3 + SEM : book ( y ) SEM : n [ Q ( y )] " CONST : pages 6 7 CAT SUBCAT : 6 # 7 6 7 4 5 6 7 6 7 QUALIA : TELIC : read QUALIA TELIC : act-on-pred P 4 5 6 7 4 5 AGENTIVE : write SEM : [ e ][ enjoy ( e , x , e ′ ) ∧ act-on-pred / ( e ′ , x , y ) ∧ P n ] enjoy the book:   coercing � np  �    SEM : n book ( y )  CAT SUBCAT :       QUALIA TELIC : P read     SEM : [ e ][ enjoy ( e , x , e ′ ) ∧ / P read ( e ′ , x , y ) ∧ n book ( y )] university-logo Alex Lascarides SPNLP: Lexical Polysemy

Data Different Kinds of Polysemy How to Capture Lexical Generlisations Constructional Polysemy: Sense Broadening (6) a. There are clouds in the sky — it’s going to rain. b. There was a cloud of flies round the cow. c. The flies were pestering the horse — it swished its tail at the buzzing cloud.  lex-count-noun  ORTH : cloud   CAT : noun-cat     SEM : obj-noun-formula       phys_obj /natural_obj        nomform        QUALIA : FORM : RELATIVE : indiv       ABS. : amorph         CONSTITUENCY : phys_cum /water-vapour university-logo Alex Lascarides SPNLP: Lexical Polysemy