Introduction to the Study of Language 2

Semantics 3

Evelien Keizer, SS17

Semantics sessions

Semantics 1: Word meaning

Semantics 2: Meaning relations between words
Semantics 3: Sentence meaning and meaning in
context

ISL 2 2
Summary Semantics 2

Paradigmatic vs. syntagmatic relations

Paradigmatic relations
synonymy, antonymy, hyponymy

Syntagmatic relations
collocation

Complications in meaning representation

idioms
metaphorical extension
homonymy/polysemy

ISL 2 3
Todays programme

Words in context
lexical ambiguity
mismatches of meaning
semantic roles
words in corpora
Sentence meaning
what is sentence meaning?
semantic relations between sentences

ISL 2 4
Lexical ambiguity

"How is bread made?"

"I know that!" Alice cried eagerly.
"You take some flour"
"Where do you pick the flower?" The White Queen asked. "In
a garden or in the hedges?"
"Well it isnt picked at all," Alice explained; "its ground"
"How many acres of ground?" said the White Queen.
Lewis Carroll, Through the looking glass

ISL 2 5
Lexical ambiguity

"How is bread made?"

"I know that!" Alice cried eagerly.
"You take some flour"
"Where do you pick the flower?" The White Queen asked. "In
a garden or in the hedges?"
"Well it isnt picked at all," Alice explained; "its ground"
"How many acres of ground?" said the White Queen.
Lewis Carroll, Through the looking glass

homophony, homonymy

ISL 2 6
Lexical ambiguity

(1) There's something wrong with the cat.

a. It hasn't even touched its food.
b. I think you'll have to call the mechanic.
(2) She cannot bear children.
a. ... when they are noisy.
b. ... because she is sterile.

Homonymy/polysemy: disambiguation through context

pragmatics

ISL 2 7
Lexical ambiguity

(3) Mary had a little lamb,

a. Its fleece was white as snow
Everywhere that Mary went,
The lamb was sure to go
http://www.kididdles.com/lyrics/m003.html

ISL 2 8
Lexical ambiguity

(3) Mary had a little lamb,

a. Its fleece was white as snow
Everywhere that Mary went,
The lamb was sure to go
http://www.kididdles.com/lyrics/m003.html

b. roast potatoes and vegetables.

ISL 2 9
Mismatches of meaning

*The table is reading the newspaper.

The ___ is reading the newspaper.
*The dog is reading the newspaper.

ISL 2 10
Mismatches of meaning

*The table is reading the newspaper.

= asterisk: ill-formed
here: semantic anomaly

The ___ is reading the newspaper.

N [+ human]
*The dog is reading the newspaper.
N [ human]
Selectional restrictions

ISL 2 11
Mismatches of meaning

Colourless green ideas sleep furiously.

Noam Chomsky, Syntactic Structures, 1957

ISL 2 12
Mismatches of meaning

Colourless green ideas sleep furiously.

Noam Chomsky, Syntactic Structures, 1957

Violation of selectional restrictions:

meaning clashes
semantically ill-formed/anomalous (but syntactically ok)

ISL 2 13
Semantic roles

Dr. Jones carried the handouts from his office

to the lecture hall.

ISL 2 14
Semantic roles

Dr. Jones carried the handouts from his office

who what where from

to the lecture hall.

where to

ISL 2 15
Semantic roles

Semantic role (thematic role, Theta role):

role that a participant plays in an event
Event Predication
( Clause)
Participant Argument
(e.g. Noun Phrase)

ISL 2 16
Semantic roles

Dr. Jones carried the handouts from his office

who what where from
Agent Theme
to the lecture hall.
where to

carry Agent Theme (Source) (Goal)

Predicate Arguments

Predication
ISL 2 17
Semantic roles

1. William stole the bicycle.

Agent Theme
2. Peter opened the safe with a key.
Agent Theme Instrument
3. Peter was in the living room.
Theme Location
4. William saw the bicycle.
Experiencer Theme

ISL 2 18
Semantic roles

Agent entity performing an action

Experiencer entity experiencing / undergoing the effect of
an event or situation
Theme entity undergoing action or movement,
entity described
Instrument entity used to carry out an action
Source starting point for a movemen
Goal end point for a movement
Location place at which entity/action is located

ISL 2 19
Semantic roles

William stole the bicycle.

steal Agent Theme

William gave the bicycle to Harry.

give Agent Theme Goal

ISL 2 20
Semantic roles

Meaning of the verb determines:

1. Number of arguments (participants)
2. Semantic roles of the arguments

Three place verbs/predicates: give, donate, hand, put

Two place verbs/predicates: steal, carry, open, see
(What about: I see you to the door?)
One-place verbs/predicates: ??

ISL 2 21
Semantic roles

Meaning of the verb determines:

1. Number of arguments (participants)
2. Semantic roles of the arguments

Three place verbs/predicates: give, donate, hand, put

Two place verbs/predicates: steal, carry, open, see
(What about: I see you to the door?)
One-place verbs/predicates: laugh, sleep, sit, fall
(What about live?)
Zero-place verbs/predicates: ??

ISL 2 22
Semantic roles

Meaning of the verb determines:

1. Number of arguments (participants)
2. Semantic roles of the arguments

Three place verbs/predicates: give, donate, hand, put

Two place verbs/predicates: steal, carry, open, see
(What about: I see you to the door?)
One-place verbs/predicates: laugh, sleep, sit, fall
(What about live?)
Zero-place verbs/predicates: rain, snow, hail

ISL 2 23
Semantic roles

(1) a. Peter opened the safe with a key.

b. The key opened the safe.
c. The safe opened.

(2) a. Harry broke the window with a stone.

b. The stone broke the window.
c. The window broke.

ISL 2 24
Semantic roles

(1) a. Peter opened the safe with a key.

b. The key opened the safe.
c. The safe opened.

ISL 2 25
Semantic roles

(1) a. Peter opened the safe with a key.

b. The key opened the safe.
c. The safe opened.

Peter opened the safe with a key.

Agent Theme Instrument
The key opened the safe.
Instrument Theme
The safe opened.
Theme

ISL 2 26
Semantic roles

Janet lost the ball.

Agent Theme

The ball was lost by Janet.

Theme Agent

Same predication, different form (clause)

From next week: relation with syntax

ISL 2 27
Semantic roles

Problem 1: How many roles are there?

First argument:
Agent, Theme, Experiencer, Instrument, Location, Force ...
Second argument:
Theme, Patient, Location, Instrument ...
Third argument:
Goal, Location, Instrument, Recipient, Beneficiary ...

ISL 2 28
Semantic roles

Problem 2: One role or several roles?

The chimney smoked.
Agent/Source
The bath filled.
Theme/Goal
This machine has saved many lives.
Agent/Instrument
The wind closed the door.
Agent/Force
I like Vienna.
Theme/Location
ISL 2 29
Semantic roles

Solution 1: semantic features

Different "Aktionsarts" (Modes of Action):
[control]
[dynamic]
[telic]
[experience]
(e.g. S.C. Dik, The Theory of Functional Grammar, 1997)

ISL 2 30
Semantic roles

Features: Aktionsart: Example:

[+con] [+dyn] Action John (Ag) kicked the ball (Th)

[+con] [+dyn] [+tel] John (Ag) walked to the station (Go)

[-con] [+dyn] Process The wind (Fo) closed the door (Th)
[-con] [+dyn] [+tel] The ball (Th) fell into the water (Go)

[+con] [-dyn] Position John (Pos) stood on the platform (Loc)

[+con] [-dyn] [+exp] John (Exp) did not believe the story (Th)

[-con] [-dyn] State The money (Th?) is in the safe (Loc).

[-con] [-dyn] [+exp] John (Exp) did not know the story (Th)

ISL 2 31
Semantic roles

Solution 2: Prototypes
Semantic roles as abstract functions: proto-roles
Proto-agent:
volitional (control)
causing a state of change in another participant
etc. experience (perception)

Proto-patient:
undergoes change of state
causally affected by other participant
direction/endpoint
etc.
(D. Dowty, Thematic protoroles and argument selection, 1991)
ISL 2 32
Semantic roles

John built this house

John: prototypical Agent
this house: prototypical Patient/Theme
John resembles his father
John: non-prototypical Agent
his father: non-prototypical Patient/Theme
This machine has saved many lives
this machine: non-prototypical Agent

ISL 2 33
Words in text corpora

Corpus:
a large, balanced collection of natural texts (written and/or
spoken).
typically based on a specific type of text, from a particular
period
aimed at being representative of the type under
examination.

ISL 2 34
Words in text corpora

Corpus linguistics:
linguistic description based on the extensive accumulation of
actually occurring language data and its analysis by
computer
empirical discipline, using statistical methods

ISL 2 35
Text in corpora

Some well-known English corpora:

COBUILD corpus (Collins Birmingham University International Language
Database)
BNC British National Corpus (BYU: http://corpus.byu.edu/bnc/)
COCA Corpus of Contemporary American English (BYU:
http://www.americancorpus.org/)
ICE corpus (International Corpus of English)
HELSINKI Corpus of English Texts (Old Middle - Early Modern English)
VOICE (Vienna Oxford International Corpus of English),
http://www.univie.ac.at/voice/

ISL 2 36
Words in text corpora

Corpus analysis software:

programmes for building ones own corpus (e.g.
concordancing, tagging)
e.g. WordSmith, AntConc

Sophisticated search software:

e.g. ICE-CUP (fuzzy trees)

ISL 2 37
Words in text corpora

Insights into semantics:

frequency of a word
frequency of different meanings for a given word
systematic collocations with other words
systematic association with particular registers / genres /
dialects / social, ethnic, age groups
etc.

ISL 2 38
Annotation in text corpora

Most corpora are tagged:

word classes: noun, verb, adjective, preposition
noun: proper name, pronouns, count/mass nouns
Occasionally corpora are parsed (ICE-GB):
syntactic categories: NP, clause, active / passive etc.
Most (bigger) corpora allow you contain different types of
text:
spoken - written
public - private
academic, business, newspaper, literature etc
interviews, telephone conversations, lectures etc.

ISL 2 39
Annotation in text corpora

Some corpora are specialized:

variety (British, American, Australian, Scottish, etc. ELF)
period: OE, ME, ModE, PDE
Some corpora also provide information about the speakers:
gender
age
education
geographical background

ISL 2 40
Words in text corpora

How many meanings of right can you identify?

Any other meanings (not in this sample)?
(Tribble, Chris and Glyn Jones. 1997. Concordances in the classroom.
Houston: Aethelstan)
ISL 2 41
Words in text corpora

Longman-Lancaster-corpus (ca. 6 million words, BE/AE, 50%

academic/50% fiction) (Biber, Conrad, Reppen. 1998. Corpus
Linguistics. CUP) ISL 2 42
Words in text corpora

Biber, Conrad, Reppen. 1998. Corpus Linguistics.CUP

ISL 2 43
Words in text corpora

(Kennedy, Graeme. 1998. An introduction to corpus linguistics.

London:Longman)

ISL 2 44
Sentence meaning

Lexicon:
large but finite; (core) meanings can be listed
Sentences
infinite number; meanings cannot be listed; meanings
are compositional
When know the meaning of individual words and the
(semantic and syntactic) rules for combining them, we can
make in infinite number of sentences (infinite number of
meanings).

ISL 2 45
Sentence meaning

Word meaning vs. sentence meaning

(1) large red balloon red balloon (absolute, non-gradably)
large balloon (relative, gradable)
(2) sky blue
blue sky
(3) Peter hit John
John hit Peter
(4) I had made the copies.
I had the copies made.

ISL 2 46
Sentence meaning

Sentence meaning (semantics)

vs. Speaker meaning / intention (pragmatics)
or:
What is coded in language (semantics)
vs. what is implied / intended (pragmatics)

Were closing.
Could you pass me the bread?
Do you have the time?
I cant hear you.
ISL 2 47
Ambiguous sentences

Ambiguity: lexical and/or structural

The Rabbi married my sister. (lexical)

1. ..
2. ..
Old men and women may enter. (structural)
1. Old [men and women] may enter.
2. [Old men] and women may enter.

ISL 2 48
Representing sentence meaning

Proposition:
What is stated/asserted in a sentence.
Truth value:
Is the propositional content true or false?

Max loves Anne. True | False

The table is reading the newspaper. True | False ???
The King of Austria is very rich. True | False ???

ISL 2 49
Representing sentence meaning

Proposition or not?

A: B:
John was late Thats not true. (He wasnt.)
Was John late? *Thats not true.
Dont be late. *Thats not true.
John promised he would come. Thats not true. (He didnt.)
I promise I wont be late. ?

ISL 2 50
Semantic relations between sentences

Paraphrase ( synonymy)

Entailment ( hyponymy)

Contradiction ( antonymy)

Presupposition

ISL 2 51
2.2 Meaning relations

(1a) Mary sold a car to Tom.

(1b) Tom bought a car from Mary.

(2a) The police chased the burglar.

(2b) The burglar was chased by the police.

Paraphrase
A and B express the same proposition (same truth value)

ISL 2 52
Meaning relations

(1a) Mr. Sellers killed his wife.

(1b) Mrs. Sellers is dead.

(2a) Jane ate oatmeal for breakfast this morning.

(2b) Jane ate breakfast this morning.

Entailment
If A is true, B must also be true.

ISL 2 53
Meaning relations

(1a) Chris is Janes brother.

(1b) Jane is an only child.

(2a) Charles is a bachelor.

(2b) Charles is married.

Contradiction
If A is true, B must be false (opposite truth values)

ISL 2 54
Meaning relations

(1) I regret leaving London.

I left London
(2) She cried before she finished her thesis.
She finished her thesis
(3) I have stopped smoking
I used to smoke

Presupposition
If A is true, then B is implied to be true

ISL 2 55
Meaning relations

Entailment vs. Presupposition:

Entailment can only occur in declarative sentences
(statement); presupposition can be found in different kinds
of sentences
Do you regret leaving London? (presupposition remains)
Did Mr. Sellers kill his wife? (entailment disappears)
Presuppositions remain under negation
I do not regret leaving London (presupposition remains)
Mr. Sellers did not kill his wife (entailment disappears)
Presuppositions can be cancelled
She died before she finished her thesis. (presupposition cancelled)

ISL 2 56
Semantics: important terms

conceptual vs. associative meaning, denotation vs. sense,

semantic features
(+ applications: sense relations, meaning mismatch in
sentences), prototype theory
paradigmatic vs. syntagmatic relations, synonymy, antonymy
(+ types), hyponymy, collocation, idioms, metaphor &
metonymy, polysemy vs. homonymy
ambiguity (lexical, structural), semantic roles, corpora,
proposition, truth value, sentence meaning vs. speaker
meaning, paraphrase, entailment, contradiction,
presupposition

ISL 2 57
Transcription exercise (3)

1. Using the symbols you have learned, make a broad transcription of

the following passage, including weak forms and marking rhythmically
stressed syllables:

Since we moved here a year ago I have been very frustrated by my

inability to communicate fluently. I have much, some would say too
much, to say on any subject. I have always been known as someone
who is willing, even eager, to express an opinion on almost any topic,
and suddenly have found myself with this curious new disability that
prevents me from doing so.

ISL 2 58
Transcription exercise (3): key

| sns wi mu:vd hr jer | a v bi:n very frstretd

ba ma nblti t kmju:nket flntli | a hv m |
sm wd se tu: m | t se n eni vn sbkt | a v
:lwez bi:n nn z smwn hu z wl | i:vn i: | t
e hr pnjnz n lmst eni tpk | nd sdnli v
fand maself w s kjrs nju: dsblti | t
privents mi frm du: s |

ISL 2 59
Transcription exercise (4)

a. Transcribe the following utterance as it might be pronounced in slow

careful speech.

There isnt a place for that kind of sixth sense.

b. Rewrite it so that your transcription shows the three different kinds of

connected speech processes present. (not including weak forms).

ISL 2 60
Transcription exercise (4): key

a. Transcribe the following utterance as it might be pronounced in slow

careful speech.

There isnt a place for that kind of sixth sense.

e znt ples f t kand v sks sens

b. Rewrite it so that your transcription shows the three different kinds of

connected speech processes present. (not including weak forms).

er znt ples f k kand sks sens

linking r assimilation elision (2x: /v/ and //)

ISL 2 61
Readings for next week

Before: Plag et al. 4.1, Yule ch. 7

After: Plag 4.2 & 4.3

ISL 2 62