Lexical Semantics: An Introduction

Loading...

Lexical Semantics: Introduction Timothy Baldwin

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

In the Beginning There was a Question ...

1

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

In the Beginning There was a Question ...

What is Lexical Semantics?

1

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

What is Lexical Semantics? • What is a lexical item/unit? ? a word? ? what about phrases? ? what about sub-word structure and morphology?

(Fillmore, 2003)

2

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

What is Lexical Semantics? • What is semantics? ? meaning? ? can we ignore syntax, then? ? can we ignore pragmatics (domain effects, ...)?

(Fillmore, 2003)

2

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

What is Lexical Semantics? • Working definition: the study of what individual lexical items mean, why they mean what they do, how we can represent all of this, and where the combined interpretation for an utterance comes from

2

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Example of Lexical Semantics in Action (1) • Predict the morphosyntax [countable/non-countable]) of:

(esp.

countability

? coagulopathy: group of conditions of the blood clotting (coagulation) system in which bleeding is prolonged and excessive; a bleeding disorder ? muntjac: small Asian deer with small antlers and a cry like a bark

Baldwin and Bond, 2003

3

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Example of Lexical Semantics in Action (1) • Predict the morphosyntax [countable/non-countable]) of:

(esp.

countability

? coagulopathy: group of conditions of the blood clotting (coagulation) system in which bleeding is prolonged and excessive; a bleeding disorder ? muntjac: small Asian deer with small antlers and a cry like a bark • More tomorrow from me ... Baldwin and Bond, 2003

3

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Example of Lexical Semantics in Action (2) • Interpret the following compound nominalisations: ? risk recognition ? doctor involvement • Cf.: ? police failure ≡ (the) police SUBJ fail ? player selection ≡ [SB] selects (the) player OBJ • More tomorrow from me ... Lapata, 2002; Nicholson and Baldwin, 2005

4

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Example of Lexical Semantics in Action (3) • Given the following interpreted Japanese relative clause constructions: ? the [magnate-bought] shop ≡ (the) shop the magnate bought ? the [small-child-bought] shop ≡ (the) shop at which the small child shopped

Baldwin, 1998

5

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• Interpret the following: ? the [bought-book] author ? the [bought-book] person ? the [bought-book] reason • Lexical semantics vs. syntax vs. pragmatics • The role of construction semantics (more from Collin Baker ...)

Baldwin, 1998

6

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Lexical Semantics and Brethren • Lexical semantics overlaps crucially with fields such as: ? ? ? ? ? ? ?

lexicography phraseology philosophy corpus linguistics syntax pragmatics child language acquisition ..

Baldwin, 1998

7

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• ... and computationally ? ? ? ?

natural language understanding computational lexicography computational language learning (lexical acquisition) knowledge representation ..

Baldwin, 1998

8

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Computational Lexical Semantics: Recognised Sub-tasks • Word sense discrimination/disambiguation • Semantic role labelling • Multiword expression compositionality/decomposability • Ontology learning and population • Semantic language modelling Baldwin, 1998

9

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Structure of Course 1. Introduction to Lexical Semantics (L1) 2. Frame Semantics, Constructions, and the FrameNet Lexical Database (L2,4) 3. Learning Lexical Semantic Representations (L3) 4. Lexical Semantics and Ontologies (L5,7) 5. Approximating Semantic Similarity with Distributional Similarity (L6) 6. Semantic Distance and Lexicographic Resources (L8) 10

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Lecture Format • 8 × 1.5 hour lectures • (Hopefully) modular course structure • Recommended readings for each lecture listed on the web page • Slides posted on the web page

11

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Caveats • Language orientation (English-centricity) • Coverage of topics (instructor-centricity) • Selection of readings (accessibility-centricity) • Widely-varying accents, spelling and examples (dialecteccentricity)

12

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

INTRODUCTION TO LEXICAL SEMANTICS

13

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Starting at the Beginning ... • Lexical semantics is concerned with the identification and representation of the semantics of lexical items • If we are to identify the semantics of lexical items, we have to be prepared for the eventuality of a given word having (shock, horror) multiple interpretations = polysemy (cf. monosemy)

14

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Polysemy • Polysemy = the condition of a single lexical item having multiple meanings • Polysemy vs. homonymy (cf. bank vs. bass) • Polysemy vs. indeterminacy (cf. father vs. uncle) • Regular/logical vs. irregular polysemy (cf. ash) • Regular vs. complementary polysemy (cf. door vs. farm) Cruse, 1986; Hirst, 1987; Ravin and Leacock, 2000; Pustejovsky, 1995

15

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Distinguishing Polysemes • The polysemy of a word can be tested by a variety of means, including: ? antagonism: can the word be used in a sentence with multiple competing interpretations? (cf. Kim can’t bear children) ? zeugma: can the word be used in a context where multiple competing interpretations are simultaneously evoked? (cf. Kim and her visa expired )

Cruse, 1986; Hirst, 1987; Ravin and Leacock, 2000; Pustejovsky, 1995

16

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

? independent truth conditions: can the word be used in a given sentence with different truth conditions according for different interpretations? (cf. Kim is wearing a light jacket) ? definitional distinctness: it is impossible to come up with a unified definition which encompasses the different sub-usages of the word .. • Note the importance of attested usages in gauging polysemy Cruse, 1986; Hirst, 1987; Ravin and Leacock, 2000; Pustejovsky, 1995

17

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Approaches to Lexical Semantic Categorisation • Attributional semantic categorisation • Semantic clustering • Relational semantic categorisation

18

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Attributional Semantic Categorisation

19

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Attributional Semantic Categorisation • For each lexical item, come up with a semantic description of each of its distinct usages, in isolation of the categorisation of other lexical items, e.g.: enrichment (n) the act of making fuller or more meaningful or rewarding • Methodologies: ? definitional semantics ? decompositional semantics 20

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Definitional Semantics • Standard lexicographic approach to lexical semantics: semantics = the study of language meaning tailor = a person whose occupation is making and altering garments • Definitions are conventionally made up of a genus (what class the lexical item belongs to) and differentiae (what attributes distinguish it from other members of that class) Jackson, 2002; Wilks et al., 1996

21

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• “Decoding” vs. “encoding” dictionaries • Pros: ? familiarity (look-up and annotation) • Cons: ? subjectivity in sense granularity (splitters vs. lumpers) and definition specificity ? circularity in definitions ? consistency, reproducibility, ... ? often focus on diachronic rather than synchronic semantics Jackson, 2002; Wilks et al., 1996

22

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

The Corpus Revolution in Definitional Semantics • Moves towards corpus-based lexicography in an attempt to reduce subjectivity in sense granularity and definition specificity (e.g. COBUILD) = move from type- to discrimination/annotation

Jackson, 2002; Sinclair, 1987

token-based

sense

23

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Decompositional Semantics • Define words by way of a constrained representation language, in an attempt to avoid circularity and enforce consistency of annotation, e.g. LCS: (:DEF_WORD "give" :LCS (cause (* thing 1) (go poss (* thing 2) ((* to 5) poss (thing 2) (at poss (thing 2) (thing 6)))) (give+ingly 26)))

Jackendoff, 1990; Dorr, 1997; Wierzbicka, 1996; Pustejovsky, 1995

24

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

e.g. definition of love in Natural Semantic Metalanguage (Goddard, 2000): X loves Y = X often thinks about Y X thinks good things about Y X wants to do good things for Y X wants good things to happen to Y when X thinks about Y, X often wants to be with Y when X thinks about Y, X often feels something good

• Also: Generative Lexicon (to be covered by Paul Buitelaar ...)

Jackendoff, 1990; Dorr, 1997; Wierzbicka, 1996; Pustejovsky, 1995

25

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• Pros: ? systematic representation/in-built definition of wellformedness • Cons: ? obscurity of representation ? disagreement about primitives/semantic language ? subtle semantic distinctions can be impossible to make due to restrictions in the representation language

Jackendoff, 1990; Dorr, 1997; Wierzbicka, 1996; Pustejovsky, 1995

26

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Decomposability and Multiword Expressions • The decomposability of a MWE = degree to which the semantics of an MWE can be ascribed to those of its parts kick the bucket → die0 spill the beans → reveal0(secret0)

Nunberg et al., 1994; Sag et al., 2002; Riehemann, 2001

27

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Decomposability and Syntactic Flexibility • Consider: *the bucket was kicked by Kim Strings were pulled to get Sandy the job. The FBI kept closer tabs on Kim than they kept on Sandy. ... the considerable advantage that was taken of the situation

• The syntactic flexibility of an idiom can generally be explained in terms of its decomposability

Nunberg et al., 1994; Sag et al., 2002; Riehemann, 2001

28

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Attributional Semantics and the Syntax-Semantics Interface • Countability: ? A syntactico-semantic property of the noun phrase ? Bounded, indivisible individuals prototypically countable: a dog, two dogs ? Unbounded, divisible substances prototypically uncountable: gold

Allan, 1980; Bond, 2005; Wierzbicka, 1988

29

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

vs.

Allan, 1980; Bond, 2005; Wierzbicka, 1988

30

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Countability Classes • countable: book, button, person (one book, two books) • uncountable: equipment, gold, wood (*one equipment, much equipment, *two equipments) • plural only: clothes, manners, outskirts (*one clothes, clothes horse) • bipartite: glasses, scissors, trousers (*one scissors, scissor kick, pair of scissors) Allan, 1980; Bond, 2005; Wierzbicka, 1988

31

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Countability and the Syntax-Semantics Interface • Semantic properties of a given noun are strong predictors of both its countability (lexical semantics) and surface manifestation (syntax): ? (simple) enumerable ↔ countable ? usable as bare singular NP ↔ uncountable • I.e., syntax offers a powerful semantic validation tool

Allan, 1980; Bond, 2005; Wierzbicka, 1988

32

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Differences in Conceptualisation • Knowing the referent is not enough to determine countability, e.g. scales

Allan, 1980; Bond, 2005; Wierzbicka, 1988

33

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Differences in Conceptualisation • Knowing the referent is not enough to determine countability, e.g. scales 1. Thought of as being made of two arms: (British) a pair of scales 2. Thought of as a set of numbers: (Australian) a set of scales 3. Thought of as discrete whole objects: (American) one scale/two scales • Also Lego – countable or uncountable? Allan, 1980; Bond, 2005; Wierzbicka, 1988

33

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Differences in Realisation • Looking at corpus data to determine countability leads to its own challenges, e.g. enrichment

Allan, 1980; Bond, 2005; Wierzbicka, 1988

34

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Differences in Realisation • Looking at corpus data to determine countability leads to its own challenges, e.g. enrichment Education itself provides enrichment to ... ... would bestow great enrichment upon ... Job enrichment is part of ... It was a developmental enrichment. ... an enrichment of life. ... received many enrichments ... Allan, 1980; Bond, 2005; Wierzbicka, 1988

34

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Basic vs. Derived Uses • Countability categorisation is confused by the existence of highly-productive conversion rules, e.g.: ? the Universal Grinder: countable noun with individuated semantics → uncountable noun with “piecemeal” semantics (e.g. the floor was littered with computer ) ? the Universal Packager: uncountable noun with substance semantics → countable noun with portion of substance semantics (e.g. two beers) Copestake and Briscoe, 1995; Jackendoff, 1991; Pustejovsky, 1995

35

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• Rather than consider all nouns as both countable and uncountable, we generally identify the “basic” uses of a given noun and derive alternate uses through the use of lexical rules (but consider chicken vs. dog vs. worm) • Cf. regular/logical polysemy

Copestake and Briscoe, 1995; Jackendoff, 1991; Pustejovsky, 1995

36

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Lexical Semantics, Sense and Context • There is growing awareness in lexical semantics of: ? context-sensitivity ? sense specificity ? basic vs. derived word usages (and the fuzziness of the boundary) ? difficulties in making categorical judgements for a word • Case in point: aspectual (aka. Vendler) verb classes • Lots more from Paul Buitelaar ... Vendler, 1957; Dowty, 1979

37

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Semantic Clustering

38

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Semantic Clustering • Capture semantically-related word groupings implicitly in the form of clusters • Obviates the need for the determination of definitions (attributional semantics in reverse)

Wilks et al., 1996

39

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• Example semantic clustering-based resource: Thesaurus (based on 1913 Roget’s):

Moby

Great Leap Forward, advance, advancement, amelioration, amendment, ascent, augmentation, bettering, betterment, boost, complement, enhancement, enrichment, eugenics, euthenics, fecundation, fertilization, fortification, furtherance, headway, impregnation, improvement, insemination, lift, melioration, mend, mending, pickup, preferment, progress, progression, promotion, recovery, restoration, revival, rise, upbeat, uplift, upping, upswing, uptrend, upward mobility, vitaminization

• Polysemy = occurrence in more than one class Wilks et al., 1996

40

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• Pros: ? implicit definition ? possibilities for combining word classes into a single class • Cons: ? no explicit indication of what a word means in a given class ? (generally) flat structure not amenable to the representation of cross-cluster correspondences ? multi-dimension classification Wilks et al., 1996

41

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Semantic Clustering and the Syntax-Semantics Interface • As typified most famously by the work of Levin, 1993, many semantic distinctions correlate closely with (observable) syntactic phenomena : it is possible to carry out semantic clustering based on (readily reproducible) syntactic tests

Levin, 1993

42

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Diathesis Alternations • A diathesis alternation is a regular variation in the argument structure of a verb, and can be either: ? morphosyntactic – truth-functionally equivalent; main difference in information structure Kim saw Sandy ↔ Sandy was seen by Kim ? morphosemantic – variation in grammatical form correlates with variation in lexical semantics Kim opened the door ↔ The door opened Levin, 1993

43

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Example Diathesis Alternations • Causative/inchoative alternation: Kim broke the window ↔ The window broke • Middle construction alternation: Kim cut the bread ↔ The bread cut easily

Levin, 1993

44

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• Conative alternation: Kim hit the door ↔ Kim hit at the door • Body-part possessor ascension alternation: Kim cut Sandy’s arm ↔ Kim cut Sandy on the arm

Levin, 1993

45

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Diathesis Alternations and Verb Classes • A verb’s (in)compatibility with different alternations is a strong predictor of its lexical semantics: Causative Middle Conative Body-part

break YES YES NO NO

cut NO YES YES YES

hit NO NO YES YES

touch NO NO NO YES

break = {break, chip, crack, crash, crush, ... } cut = {chip, clip, cut, hack, hew, saw, ... } hit = {bang, bash, batter, beat, bump, ... } touch = {caress, graze, kiss, lick, nudge, ... } Levin, 1993

46

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• Corollary: we can predict the syntax of novel words we are given the semantic class for (cf. countability examples earlier) • The principal weakness of syntax-based verb classification is that there are often subtle divergences in semantics between synonyms (cf. eat vs. devour vs. gobble)

Levin, 1993

47

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Lexical Semantics and the Lexicon-Semantics Interface • Firth, 1957 famously made the observation: you shall know a word by the company it keeps which is commonly known as the distributional hypothesis • Blunter instrument but with greater flexibility than a strict interpretation of the syntax-semantics interface • More from James Curran ... Firth, 1957

48

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

A Case in Point Acyclovir is a specifically anti-viral drug ... Acyclovir has been developed and marketed by ... Acyclovir given intravenously, ... Coagulopathy is a well recognised complication ... ... could stimulate a coagulopathy ... ... is also probably responsible for a coagulopathy ... ... a patient with a coagulopathy.

Firth, 1957

49

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Lexical Semantics and Context (1) • Lexicosyntactic context is commonly used by corpus linguists to analyse lexical semantics, through a combination of: ? ? ? ?

concordancing analysis of common verb–argument collocations analysis of passives and other constructions analysis of co-occurrence with certain adverbs/auxiliaries ..

Stubbs, 2001

50

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Relational Semantic Categorisation

51

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Relational Semantic Categorisation • Capture correspondences between lexical items by way of a finite set of pre-defined semantic relations • Methodologies: ? lexical relations ? constructional relations

52

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Lexical Relations • (Possibly directed) semantic relations which pertain to pairs of lexical items, e.g. bush

orange

tea

tree

bush orange tea tree

Cruse, 1986; Miller et al., 1990; Fellbaum, 1998; Widdows, 2005

53

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Synonymy • Propositional synonymy: X is a propositional synonym of Y if (i) X and Y are syntactically identical, and (ii) substitution of Y for X in a declarative sentence doesn’t change its truth conditions (e.g. violin and fiddle) • Why propositional synonymy is over-restrictive: ? syntactic identity (cf. eat and devour ) ? collocations (cf. cemetery and graveyard ) ? gradability (cf. sofa/settee vs. boundary/frontier ) Cruse, 1986; Miller et al., 1990; Fellbaum, 1998; Widdows, 2005

54

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• More pragmatic view of synonymy via syntax: synonyms are inter-substitutability in some/most rather than all contexts • Synonymy via semantics: synonyms share “common traits” or attributional overlap, walking the fine line between “necessary resemblances” and “permissible differences” (Cruse, 1986), cf: grain vs. granule green vs. purple alsation vs. spaniel Cruse, 1986; Miller et al., 1990; Fellbaum, 1998; Widdows, 2005

55

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• Permissible differentiation via clarification: Here is a grain, or granule, of the substance. #The cover is green, {or,that is to say} purple. and contrast: Here is a grain or, more exactly, granule #He likes alsations, or more exactly, spaniels

Cruse, 1986; Miller et al., 1990; Fellbaum, 1998; Widdows, 2005

56

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• Properties of synonymy: ? ? ? ?

symmetric applies only to lexical items of the same word class applied at the sense or lexical item-level? ≈ converse of polysemy

Cruse, 1986; Miller et al., 1990; Fellbaum, 1998; Widdows, 2005

57

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Hypernymy and Hyponymy • Hyponymy: X is a hyponym of Y iff f (X) entails f (Y ) but f (Y ) does not entail f (X): Kim has a pet dog → Kim has a pet animal Kim has a pet animal 6→ Kim has a pet dog N.B. complications with universal quantifiers and negation: Kim likes all animals → Kim likes all dogs Kim likes all dogs 6→ Kim likes all animals Cruse, 1986; Miller et al., 1990; Fellbaum, 1998; Widdows, 2005

58

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• Hypernymy: Y is a hypernym of X iff X is a hyponym of Y • Properties of hypernymy/hyponymy: ? asymmetric ? applies only to lexical items of the same word class ? applies at the sense level

Cruse, 1986; Miller et al., 1990; Fellbaum, 1998; Widdows, 2005

59

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Basic-level Categories • Basic-level categories have maximal “cue validity”, i.e. are easy to draw/visualise (cf. furniture vs. chair vs. throne) • In attribute terms, basic-level categories: ? maximise the number of attributes shared by members of the category ? minimise the number of attributes shared with other categories Rosch, 1975

60

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Antonymy • Complementarity: X and Y are complementaries if X and Y define mutually-exclusive sets which encompass all of a conceptual domain, cf.: ?The door is neither open nor shut I am feeling neither good nor bad • Antonymys: ? are fully gradable ? when intensified move in opposite directions along Cruse, 1986; Miller et al., 1990

61

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

their scale of domain (cf. heavy vs. light) ? do not bisect their domain of operation • Similarity with synonymy, in terms of attributional overlap • Antonymy is generally considered to operate at the lexical item-level (cf. rise/fall vs. ascend/descend ) • Morphological influences lengthen/shorten)

Cruse, 1986; Miller et al., 1990

(cf.

long/short

vs.

62

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• Other properties of antonymy: ? symmetric ? applies only to lexical items of the same word class (esp. adjectives and verbs)

Cruse, 1986; Miller et al., 1990

63

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Other Lexical Relations • There are many, many more lexical relations advocated by various theories (esp. MTT), including: ? ? ? ? ? ?

meronymy/holonymy (part-whole) troponymy/hypernymy (cf. walk vs. lollop) entailment (cf. snore vs. sleep) Sing(ular)/Plu(ral) (cf. bee vs. swarm) Oper(ator) (cf. question vs. ask ) Magn(ifier) (cf. wound vs. badly) ..

Cruse, 1986; Fellbaum, 1998; Mel’ˇcuk and Polgu`ere, 1987; Mel’˘cuk, 1995; Mel’ˇcuk, 1998

64

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

WordNet • WordNet is an open-source electronic lexical database of English, developed at Princeton University http://wordnet.princeton.edu/perl/webwn2.1 • Made up of four separate semantic nets, for each of nouns, verbs, adjectives and adverbs • Lexical items are categorised into ∼115K (and counting) glossed synsets (= synonym sets) Miller et al., 1990; Fellbaum, 1998

65

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

1. enrichment -- (act of making fuller or more meaningful or rewarding) 2. enrichment -- (a gift that significantly increases the recipient’s wealth)

• Lexical relations at either the synset level or sense (= combination of lexical item and synset) level • Strongly lexicalist: ? synsets only where words exist ? (near) absence of frame semantics

Miller et al., 1990; Fellbaum, 1998

66

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• Other quirks/properties: ? 25 unique beginners in noun semantic net ? taxonomic vs. functional hyponymy (cf. chicken vs. bird/food ) ? few proper nouns and no separate classification for proper nouns

Miller et al., 1990; Fellbaum, 1998

67

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Psycholinguistic Foundations of WordNet • Strong foundation on inheritance) based on

hypo/hypernymy

(lexical

? response times to sentences such as: a canary {can sing/fly,has skin} a bird {can sing/fly,has skin} an animal {can sing/fly,has skin}

? analysis of anaphora: I gave Kim a novel but the {book,?product,...} bored her Kim got a new car. It has shiny {wheels,?wheel nuts,...}

? selectional restrictions 68

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Lexical Similarity • A lexical similarity function maps two senses/lexical items onto a real number value (conventionally ∈ [0, 1]), representing their relative semantic overlap • Models of lexical similarity: ? attributional ? cluster-based ? relational (all relations equal?)

G¨ardenfors, 2000; Widdows, 2005

69

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

• Properties of lexical similarity: ? symmetric ? can potentially apply to lexical items of differing word classes ? applies at the lexical item or sense level • More from Graeme Hirst ...

G¨ardenfors, 2000; Widdows, 2005

70

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Constructional Relations • Semantic relations which pertain to particular construction types, whereby we seek to interpret instances of a given construction by way of a finite set of relation types. • Largely focused on (productive) multiword expressions for my purposes: ? compound nouns ? compound verbs in Japanese ? relative verb constructions in Japanese 71

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Compound Noun Semantic Relations • Compound nominals semantically

are

largely

unrestricted

diesel truck/oil/tanker, phone book, cloud bus, apple juice seat • Compound nominalisations tend to occur with subject or object interpretation: machine performance, museum construction, student education BUT ALSO soccer competition Isabelle, 1984; Downing, 1977; Levi, 1978; Lauer, 1995b; Lapata, 2002; Nicholson and Baldwin, 2005

72

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Semantic Theories • Every linguist commonalities

has

her

own

theory,

but

with

• Import of syntax, semantics, discourse and knowledge representation in different theories • Claims that finite enumeration of semantic relations are psychologically untenable (Downing, 1977)

Isabelle, 1984; Downing, 1977; Levi, 1978

73

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Example Theory 1: Levi, 1978 • 4 roles for nominalisations: ? act, product, agent, patient truck driver = agent student discontinuation = act • 9 recoverably deletable predicates for compound nominals: ? in, for, from, about ? cause, make, have, use, be power station = make steel box = use baby crocodile = be Levi, 1978

(prepositional) (relative clauses)

74

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Example Theory 2: Lauer, 1995a • Interpret compound nominals according to 7 prepositions: ? ? ? ? ? ? ? ?

of : state law = law of state for: baby chair = chair for baby in: morning prayer = prayer in morning at: airport food = food at airport on: Sunday television = television on Sunday from: reactor waste = waste from reactor with: gun man = man with gun about: war story = story about war

Lauer, 1995a

75

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Example Theory 3: Copestake, 2003 • Cateogrise compounds as first category that “fits”: 1. 2. 3. 4. 5. 6. 7. 8.

listed compounds: home secretary hypernmic compounds: tuna fish, oak tree deverbal compounds: satellite observation relational compounds: jazz fan made-of compounds: steel sword, polystyrene box prepositional compounds: airshow accident non-deverbal verb compounds: oil town non-paraphrasable compounds: listeria society

Copestake, 2003

76

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Constructional Semantics • How do lexical items interact with each other and context? • What is the linkage between lexical items and language as a communicative tool/discourse semantics? • Can we extend the research on lexical item-level semantics to larger units (i.e. constructions)? • All this and much, much more from Collin Baker ... 77

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Summing Up • What is lexical semantics? • Different approaches to lexical semantic categorisation: ? attributional semantics ? semantic clustering ? relational semantics • Recurring theme: the syntax-semantics interface

78

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

To Come • Frame Semantics, Constructions, and the FrameNet Lexical Database (L2,4) • Learning Lexical Semantic Representations (L3) • Lexical Semantics and Ontologies (L5,7) • Approximating Semantic Similarity with Distributional Similarity (L6) • Semantic Distance and Lexicographic Resources (L8) 79

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

(A Random Selection of) Online Resources • On-line bibliography: dingo.sbs.arizona.edu/~hharley/522/522Spring1999/LexSemBiblio. html • Levin verb classes: www-personal.umich.edu/~jlawler/levin.verbs • WordNet: wordnet.princeton.edu • VerbNet: verbs.colorado.edu/~kipper/verbnet.html • FrameNet: framenet.icsi.berkeley.edu • LCS Database: www.umiacs.umd.edu/~bonnie/LCS_Database_Documentation. html 80

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

References Allan, K. (1980). Nouns and countability. Language, 56(3):541–67. Baldwin, T. (1998). The Analysis of Japanese Relative Clauses. Master’s thesis, Tokyo Institute of Technology. Baldwin, T. and Bond, F. (2003). Learning the countability of English nouns from corpus data. In Proc. of the 41st Annual Meeting of the ACL, pages 463–70, Sapporo, Japan. Bond, F. (2005). Translating the Untranslatable: A Solution to the Problem of Generating English Determiners. CSLI Publications, Stanford, USA. Copestake, A. (2003). Compounds revisited. In Proc. of the 2nd International Workshop on Generative Approaches to the Lexicon, Geneva, Switzerland. Copestake, A. and Briscoe, T. (1995). Semi-productive polysemy and sense extension. Journal of Semantics, pages 15–67. Cruse, D. A. (1986). Lexical Semantics. Cambridge University Press, Cambridge, UK. Dorr, B. J. (1997). Large-scale dictionary construction for foreign language tutoring and interlingual machine translation. Machine Translation, 12(4):271–322. Downing, P. (1977). On the creation and use of English compound nouns. Language, 53(4):810–42. Dowty, D. (1979). The Semantics of Verbs and Times in Generative Semantics and in Montague’s PTQ. Springer, Dordrecht, Netherlands. Fellbaum, C., editor (1998). WordNet: An Electronic Lexical Database. MIT Press, Cambridge, USA. Fillmore, C. J. (2003). Topics in lexical semantics. In Form and Meaning in Language, chapter 6. CSLI Publications, Stanford, USA. 81

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Firth, J. R. (1957). A synopsis of linguistic theory, 1930-1955. Studies in Linguistic Analysis, Philological Society, Oxford. G¨ardenfors, P. (2000). Conceptual Spaces: The Geometry of Thought. MIT Press, Cambridge, USA. Goddard, C. (2000). Polysemy: A problem of definition. In (Ravin and Leacock, 2000), chapter 7, pages 129–151. Hirst, G. (1987). Semantic Interpretation and the Resolution of Ambiguity. Cambridge University Press, Cambridge, UK. Isabelle, P. (1984). Another look at nominal compounds. In Proc. of the 10th International Conference on Computational Linguistics (COLING ’84), Stanford, USA. Jackendoff, R. (1990). Semantic Structures. MIT Press, Cambridge, USA. Jackendoff, R. (1991). Parts and boundaries. In Levin, B. and Pinker, S., editors, Lexical and Conceptual Semantics, pages 1–45. Blackwell Publishers, Cambridge MA and Oxford UK. Jackson, H. (2002). Lexicography: An Introduction. Routledge, London, UK. Lapata, M. (2002). The disambiguation of nominalizations. Computational Linguistics, 28(3):357–88. Lauer, M. (1995a). Corpus statistics meet the noun compound: Some empirical results. In Proc. of the 33rd Annual Meeting of the ACL. Lauer, M. (1995b). Designing Statistical Language Learners: Experiments on Noun Compounds. PhD thesis, Macquarie University. Levi, J. N. (1978). The Syntax and Semantics of Complex Nominals. Academic Press, New York, USA. Levin, B. (1993). English Verb Classes and Alterations. University of Chicago Press, Chicago, USA. Mel’˘cuk, I. (1995). Phrasemes in language and phraseology in linguistics. In Everaert, M., van der Linden, E.-J., Schenk, A., and Schreuder, R., editors, Idioms: Structural and Psychological Perspectives, chapter 8. Lawrence Erlbaum Associates. 82

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Mel’ˇcuk, I. (1998). Collocations and lexical functions. In Phraseology: Theory, Analysis, and Applications, pages 23–54. Oxford: Clarendon Press. Mel’ˇcuk, I. and Polgu`ere, A. (1987). A formal lexicon in meaning-text theory (or how to do lexica with words). Computational Linguistics, 13(3–4):261–275. Miller, G. A., Beckwith, R., Fellbaum, C., Gross, D., and Miller, K. J. (1990). Introduction to WordNet: an on-line lexical database. International Journal of Lexicography, 3(4):235–44. Nicholson, J. and Baldwin, T. (2005). Statistical interpretation of compound nominalisations. In Proc. of the Australasian Language Technology Workshop 2005, pages 152–9, Sydney, Australia. Nunberg, G., Sag, I. A., and Wasow, T. (1994). Idioms. Language, 70:491–538. Pustejovsky, J. (1995). The Generative Lexicon. MIT Press, Cambridge, USA. Ravin, Y. and Leacock, C., editors (2000). Polysemy: Theoretical and Computational Approaches. Oxford University Press, Oxford, UK. Riehemann, S. (2001). A Constructional Approach to Idioms and Word Formation. PhD thesis, Stanford, USA. Rosch, E. (1975). Cognitive representations of semantic categories. Journal of Experimental Psychology, 104:192–233. Sag, I. A., Baldwin, T., Bond, F., Copestake, A., and Flickinger, D. (2002). Multiword expressions: A pain in the neck for NLP. In Proc. of the 3rd International Conference on Intelligent Text Processing and Computational Linguistics (CICLing-2002), pages 1–15, Mexico City, Mexico. Sinclair, J., editor (1987). Looking Up: An Account of the COBUILD Project in Lexical Computing. Collins ELT, London, UK. Stubbs, M. (2001). Words and Phrases: Corpus Studies of Lexical Semantics. Blackwell, Oxford, UK and Malden, USA. Vendler, Z. (1957). Verbs and times. The Philosophical Review, 66:143–60. Widdows, D. (2005). Geometry and Meaning. CSLI Publications, Stanford, USA. 83

ACL/HCSNet Advanced Programme in NLP

Lexical Semantics: An Introduction

Wierzbicka, A. (1988). The Semantics of Grammar. John Benjamin. Wierzbicka, A. (1996). Semantics: Primes and Universals. Oxford University Press, Oxford, UK. Wilks, Y., Slator, B. M., and Guthrie, L. M. (1996). Electric Words: Dictionaries, Computers and Meanings. MIT Press, Cambridge, USA.

84

Loading...

Lexical Semantics: An Introduction

Lexical Semantics: Introduction Timothy Baldwin ACL/HCSNet Advanced Programme in NLP Lexical Semantics: An Introduction In the Beginning There was...

445KB Sizes 8 Downloads 16 Views

Recommend Documents

Theories of Lexical Semantics
Great Clarendon Street, Oxford ox2 6dp. Oxford University Press is a department of the University of Oxford. It furthers

Introduction to Semantics
Linguistics. Formation rules (IV). 8. A wff which does not contain any free variables is a proposition. 9. Only the form

An Introduction to English Semantics and Pragmatics - WordPress.com
Jan 25, 2017 - titles in the series include. An Introduction to English Syntax. Jim Miller. An Introduction to English P

Verbs semantics and lexical selection - ACM Digital Library
S. K. Dwivedi , Parul Rastogi, Critical analysis of WSD algorithms, Proceedings of the International Conference on Advan

THEORIES OF LEXICAL SEMANTICS Dirk Geeraerts - Research Unit
Feb 15, 2009 - designandum is chosen as the motif expressed in the name. In parasémie, the source domain is the same on

Semantics
(1) A Texan tried to assassinate the President. (2) A Texan tried to murder the President. (3) A Texan tried to kill the

Makalah Semantics | Semantics | Proposition - Scribd
Makalah Semantics - Download as Word Doc (.doc / .docx), PDF File (.pdf), Text File (.txt) or read online. Makalah seman

LEXICAL STRUCTURE, LEXICAL CONCEPTS AND - Dialnet
dominio emocionológico. Granada: Método Ediciones. Ruiz de Mendoza Ibáñez, F. 1997a. “On “relational frames”:

Lexical Grammar
like positive/negative, singular/plural, active/passive, and so on. If we look at them carefully, these terms are of cou

FRAME SEMANTICS
clearly demonstrating the above thesis is the difference in meaning between the ... After this the relationship of frame