Sunteți pe pagina 1din 29

Course in Semantics

Dr. ore Vidanovid


Introductory Notions and Terms
1
The Term
The term semantics is a new addition to the English language. It was first
used in its modern sense about one hundred years ago and appeared for
the first time in print in 1900, in the book titled Semantics: studies in the
science of meaning, by a French author, M. Breal (Michel Jules Alfred
Breal). You have to learn the whole name for your exam
Breal treated semantics in a modern fashion, claiming that it was the
science of meaning found in practical everyday language.
The Study of Meaning
Thus, semantics is the study of meaning, but meaning has many
meanings
What is the meaning of meaning?
As an illustration, what is the meaning of:
Hello
Isnt it hot in here?
You are a bright spark, arent you?
The lion is more dangerous than the unicorn.
Whats the problem with
Hello
Does it actually mean anything at all?
Is the meaning a social gesture rather than a linguistic item?
Isnt it hot in here?
Whats purpose of saying the above, is it merely asking a question? Or,
rather, saying actually I want a window open
Whats the problem with
Youre a bright spark, arent you?
This is probably sarcastic, could this be part of meaning?
The lion is more dangerous than the unicorn
Could this be true? Unicorns dont exist, but people tell and write stories
about them. So we can see that semantics cannot be only co-extensive with
the real world in the literal sense.
The study of meaning is too general!
The first definition is way too general, because there are other aspects of
meaning.
These other aspects of meaning are based on the types of knowledge
related to semantics in language.
There are at least three types or forms of knowledge related to semantics:
Types of knowledge and semantics
Semantic knowledge: knowing what words mean and how these meanings
combine in sentences to form sentence meanings.
Pragmatic knowledge: knowing how sentences are used in different
contexts and how contexts affect the interpretation of the sentence.
(Remember Isnt it hot in here?
Types of knowledge related to semantics
World knowledge: its the general knowledge about the structure of the
world that language users must have in order to maintain the conversation;
this knowledge must include what each language user must know about
the other users beliefs, goals and expectations.
Semantic knowledge
Semantic knowledge deals with what words mean and how these meanings
combine in sentences to form sentence meanings.
Sentence meanings: difficult to describe briefly, will be covered later on.
Word meanings: words can have several distinct meanings for any
syntactic class (walk, go, buy).
Semantic knowledge: examples
Bank (noun)
Bank of a river
Financial institution
Bank of lights, switches
A flight maneuver
Bank (verb)
To deposit money
To base ones hopes on something
To conduct the business of banking
Semantic knowledge: complexities
What else would semantic theory have to account for?
Some elementary logical relations such as:
Entailment: John is writing and drawing
Entails => John is writing
* An issue: does No one is writing and drawing entail No one is writing?
Semantic knowledge: complexities
Synonymy = different sentences have the same meaning:
John comes or Mary comes
John or Mary comes
Ambiguity:
The bill is large.
John was looking for the glasses
Pragmatic knowledge
Isnt it hot in here? can be interpreted in several ways:
A.I want a window open.
B.I find your lecture very boring and would like to start a new topic.
C.Ive just arrived and havent anything more intresting to say
Pragmatic knowledge: more examples
The city councillors refused the demonstrators a permit because they feared
violence.
The city councillors refused the demonstrators a permit because they feared
violence.
A Serbian example :
Vidimo se veeras u osam, kod Lepe Brene (pevaice)
Vidimo se veeras u osam, kod Lepe Brene (zgrade poznate pod tim
nadimkom)
World Knowledge
World knowledge is very often interwoven with pragmatic
knowledge, because both refer to some items that are outside the language
system.
The main difference lies in the scientific background of world
knowledge, which is something that pragmatic knowledge is not based on.
E.g. definition of salt is NaCl. Naturally, speakers/hearers neednt be aware
of world knowledge.
Semantics within Linguistics
Semantics is often viewed as a component or level of linguistics on a par
with phonetics, syntax or morphology.
Semantics is viewed as being at one end and phonetics at the other, with
grammar (morpho-syntax) somewhere in the middle.
Language is thus seen as a communication system that has a message on
one end and a set of signs on the other.
Signs and de Saussure
Saussure offered a 'dyadic' or two-part model of the sign. He defined a sign
as being composed of:
a 'signifier' (signifiant) - the form which the sign takes; and
the 'signified' (signifi) - the concept it represents.
The sign is the whole that results from the association of the signifier with
the signified. The relationship between the signifier and the signified is
referred to as 'signification.
Signifier and Signified
Sign was a psychological concept for de Saussure
More on Sign
A sign must have both a signifier and a signified. You cannot have a totally
meaningless signifier or a completely formless signified . A sign is a
recognizable combination of a signifier with a particular signified.
The same signifier (the word 'open, for example) could stand for a different
signified (and thus be a different sign) if it were on a push-button inside a lift
('push to open door'), instead of a sign on a shop-door (open for business).

The Value of a Sign
The Value of a Sign II
What Saussure refers to as the 'value' of a sign depends on its relations
with other signs within the system - a sign has no 'absolute' value
independent of this context. Saussure uses an analogy with the game of
chess, noting that the value of each piece depends on its position on the
chessboard.
The sign is more than the sum of its parts. Whilst signification - what is
signified - clearly depends on the relationship between the two parts of the
sign, the value of a sign is determined by the relationships between the sign
and other signs within the system as a whole.
More on Sign Value
As an example of the distinction between signification and value, Saussure
notes that 'The French word mouton may have the same meaning as the
English word sheep; but it does not have the same value.
There are various reasons for this, but in particular the fact that the English
word for the meat of this animal, as prepared and served for a meal, is not
sheep but mutton.
The difference in value between sheep and mouton hinges on the fact that
in English there is also another word mutton for the meat, whereas mouton
in French covers both'
Charles S. Peirce on Sign
Examples
Examples from our language:
Pozajmiti => lend / borrow (two values)
Uiti => teach / learn (two values) which can be modified and corrected on
the morphological level. How do we make up for the lack of values in this
particular case?
Semantics and Psychology
Charles Morris (Signs, language, behavior 1946), one of the founders of
behaviourism
Dog, buzzer, food. Food denoted by the buzzer if no food, then it is not
denoted but signified.
The measurement of meaning
Osgood, Suci and Tannebaum (1957) attempted to measure the meanings
of words in terms so called semantic space.
This turned out to be nothing else but positive or negative emotional
content.
Father => Is it happy or sad? Is it hard or soft? Is it slow or fast?
The final results tell little about meaning
Semantics and Communication Theory
The communication system carries information and the system can be
judged according to the efficiency with which it transmits the information.
An efficient system will have minimum redundancy (surplus of information)
and minimum noise.
Most languages have a redundancy level of about 50%!


Semantics of Naming
Vidanovic
The Scope of Semantics
Unsatisfactory Views on Semantics
Language is a system that has the signifier on one hand and the signified on the
other. What is the nature and the relationship of these two entities?
Plato (Cratylus): the signifier is a word and the signified is an object that it stands
for, refers to or denotes.
Thus, words are names or labels for things.
Naming is Attractive and Elegant
This is an attractive view.
Children surely learn many words by naming.
Ostensive definitions are particularly powerful in naming and are successfully
retained in the childs memory.
Possible Objections
This process appears to work mainly, if not only, with nouns.
Names for nouns are obviously static and solid, as opposed to adjectives, verbs,
prepositions, etc.
Nouns are, simply put, names of persons or things.
Only colour adjectival names seem possible
It may be possible to include colour labels in the general theory of naming, just in
the way it is done in colour charts.
Thus the only adjectives that might fit perfectly in the theory are colour
adjectives, as these could well be regarded as names.
Other adjectives (such as early, true or difficult) and verbs (run) are simply
impossible to identify as names.
The boy is running Which is which here?
Verbs, prepositions and pronouns are unnamable as things
In the previous sentence is it necessary to distinguish the boy and what he is
doing?
Does running involve only the feet or are the arms involved in the process too?
Is speed relevant?
What about abstract verbs such as remember, like, see?
How about prepositions (up, under) or pronouns which may denote different
things at different times?
Can the theory of naming be applied to nouns only?
Some objections:
Some nouns, e.g. unicorn, goblin, fairy relate to entities that apparently do not
exist
As they do not exist they cannot denote objects in the world (however, remember
Meinong 1904 ?).
May be the way out is to introduce two kinds of world: the real world and the
world of fairy tales?
This indicates that words probably cannot be simply names of objects of our
experience.
Further objections
Some nouns neednt refer to imaginary items, but still do not refer to physical
objects: love, hate, inspiration, nonsense, freedom, death, envy.
If nouns refer to things, and the reply is that love, hate and such are abstract
things, then the definition is faulty as nouns refer to things, not abstract things.
An aside: How do we understand each other when talking about such abstract
nouns? How do we understand each others connotations?
More objections
Even when there are physical objects that exist and which can be identified, the
meaning need not be the same as its denotation (the object that the name refers
to).
The evening star; the morning star do not have the same meaning even though
they have the same denotation!
Tito = vrhovni komandant, predsednik, ljubiica plava, Josip Broz, maral
And more
Even a restrictive definition of objects which are visible in the world around us
shows that these often denote a set of different objects.
E.g. chairs. These come in different sizes, shapes and with different backs. What
is it that makes a chair a chair, different from a stool or a settee?
The dividing lines are often very vague and sometimes they overlap.
Objections Galore
When is a hill a hill and not a mountain? When is a stream a ford and when is it a
river? In our world objects are not clearly grouped.
In contemporary logic such objects are said to belong to so-called fuzzy sets. As
such they cannot be assigned a single label (name).
This problem has bothered philosophers since Plato.
Realism vs. Nominalism
Realism: all things called by the same name have some common property there
is some kind of reality that determines what is a chair, a hill, a river, a house. (
bread, wine have the properties of paneity and vineity hm!)
Nominalism: things have nothing in common but the name. (This latter doctrine
seems to be at odds with reality: we simply do not use the word chair to refer to
objects that are completely different)
Words often reflect the interests of the people who speak a language
Words need not reflect the reality of the world around as much as the interests of
speakers.
Different cultures have different sets of words for objects that have different uses
in respective environments.
Linguistic relativity or determinism (B.L. Whorf <= E. Sapir)
Eskimos have several different words for snow; the Hopi have only one word for
flier (plane, pilot or insect); the Hopi time system
The Sapir Whorf Hypothesis
Natural Classes
Natural classes are typical products of scientific terminology, and they do exist.
If we go to the zoo well be able to see that each creature in the zoo has a
particular name and that there can be no fuzzy sets or the realism / nominalism
muddle.
No creature there can be named in two different ways, not can there be any
overlapping.
More on natural classes
This holds true of the names of animals, insects, plants and chemical substances
Bear in mind that such scientific classifications are not typical of everyday life and
that most of the things we see do not fall strictly in one class or another!
Ordinary, everyday language
This type of language differs from scientific language precisely in the fact that its
terms are not clearly defined. (Remember pitfalls of definitions, last year
linguistics course)
Ordinary language does not establish rigorous distinctions between classes.
Why couldnt a fish simply mean a creature that swims in the water and a bird
as a vertebrate that flies?
Object word vs. Dictionary word
Proposal by Bertrand Russell: object words are learned ostensibly (by pointing at
objects), while dictionary words have to be defined in terms of the object words.
Thus object words have ostensive definitions.
Still, no matter how sensible this is, it cant be a solution.
Ostensive Definitions
A non-verbal pointing definition:
What do you mean by blackboard eraser?That
thing there. (pointing)
One problem: The action done or the object
pointed to may be misinterpreted.
Another: Limited applicability, since one
cannot point to abstract objects such as love,
happiness or Gross Domestic Product.
Pitfalls of Ostensive Definitions
You have to understand exactly what is being pointed at: if I point to a chair and
say this is a chair what is it exactly I am pointing at (leg, back, wood, whole
object)?
Pointing to an object involves the identification of the object, the specifications of
the qualities that make it a chair or a stool, etc.
This requires a sophisticated understanding and a complete categorisation of
language.
More on ostensive definitions
Ludwig Wittgenstein: I must already be the master of a language to understand
an ostensive definition.
Thus a child is simply unable to handle all the complexities that we have outlined
so far.
So we come to a point at which we have to take into consideration not only
meanings of words, but of sentences as well.
In order to do that, we have to clarify the idea of a concept first!
Family Resemblances & Wittgenstein
How do we recognize that two people we know are related to one another? We
may see similar height, weight, eye color, hair, nose, mouth, patterns of speech,
social or political views, mannerisms, body structure, last names, etc. If we see
enough matches we say we've noticed a family resemblance. It is perhaps
important to note that this is not always a conscious process generally we don't
catalog various similarities until we reach a certain threshold, we just intuitively
see the resemblances. Wittgenstein suggests that the same is true of language.
We are all familiar (i.e. socially) with enough things which are games and enough
things which are not games that we can categorize new activities as either games
or not.
Language Games & Wittgenstein
Language-games Wittgenstein develops this discussion of games into the key
notion of a language-game. He introduces the term using simple examples, but
intends it to be used for the many ways in which we use language. In one
language-game, a word might stand for things to be manipulated, but in another
the same word might be used for asking questions or giving orders. "Water!", for
example, can be an exclamation, an order, a request, or an answer to a question;
but, which meaning it has depends on the language-game in which it is being
used. Thus, the word "water" has no meaning apart from its use within a
language-game.
The naming theory seems feeble
Relating words and things in a direct manner seems to be a futile attempt as there
are so many objections.
There is a more sophisticated and, probably, more plausible view.
This view relates words and things via a mediator (concepts of the mind).
This account is often called the Sign Theory (Saussure) or the Semiotic Triangle
(Ogden & Richards)
The Sign Theory of de Saussure
As we have seen the linguistic sign consists of a signifier and a signified.
More strictly put, these are a sound image and a concept, linked by a
psychological associative bond.
In this way, the noises that we make and the objects of the world we talk about
are somehow mirrored by conceptual entities.
Semiotic Triangle by Ogden and Richards
Semiotic Triangle Continued
The symbol is the linguistic element (word, sentence, etc.)
The referent is the object in the world of experience
There is no direct link between the symbol and the referent (between
language and the world); the link is via thought or concepts.
This theory avoids many problems of naming.
What is really the link between symbol and concept?
One answer is as follows: the link is psychological, so that when we think of a
name we think of a concept and vice versa.
However, the problem with this account is that we always have to have an image
when we talk about an object. However, we simply do not always visualise objects
in our minds eye as this would hinder communication.
Another explanation
We may also think about the link as some kind of permanent association stored
in the mind (brain).
This is a circular definition, says Palmer, because whenever we have a word there
will be a concept, and the concept will be the meaning of the word.
The ghost-in-the-machine argument to account for the body that has a mind
Palmer is against Mentalism
As Chomsky and Jackendoff (among others) are two of the mentalists that accept
a conceptual view of meaning, we should emphasize that Frank Palmer is opposed
to such an approach.
Chomsky has introduced introspection (intuition) as one of the key terms in
language investigation.
Palmer criticizes mentalism for several reasons:
Palmers Criticism
The ghost-in-the-machine objection: nothing is gained by moving meaning closer
to the mind (brain).
Concepts (if there were any) are inaccessible in principle to anyone but the
individual, which is complete subjectivism.
Introspection is a procedure that neglects empirical data, recordings and texts. It
is only fishing in our tank (Firth).
Palmer seems to be a Behaviourist in Disguise
As he is asking for empirical data and rejects any kind of intuitive evidence, it
looks like he might easily embrace behaviourism.
This is so as empirical data confirm or disconfirm (refute) our observations and
sightings.
They are all based on stimuli and responses (confirmation or refutation).
Palmer: meaning is elusive
Palmer insists that semantics ought to try to understand how it is that words and
sentences can mean at all, or how they can be meaningful.
Words and sentences neednt really have meaning, in the literal sense of having.
It is something they are or arent (meaningful)
Palmer uses Wittgensteins dictum: Dont look for the meaning of a word, look
for its use.
Some Dictionary Tricks and Feats
[ad. med.L. *paneitas, f. *pane-us of bread, f. pan-is bread.]
The quality or condition of being bread, _breadness'.
_1687 S. Parker Reasons Abrogat. Test (1688) 22 They could not onely separate
the Matter and Form, and Accidents of the Bread from one another, but the
Paneity or Breadishness it self from them all.
1782 Priestley Corrupt. Chr. II. vi. 42 Innocent_acknowledged that_there did
remain a certain paneity and vineity.
Sense and Reference
Reference deals with the relationship between the linguistic elements (words,
sentences) and the non-linguistic world of experience.
Sense relates to the complex system of relationships that exist between the
linguistic elements themselves (mostly words).
Sense is concerned only with intra-linguistic relations!
Can Semantics Deal only with the World?
It seems reasonable to think that semantics should deal only with the way we
relate our language to our experience (the world).
So, reference appears as the only essential element of semantics.
But we have already seen problems with object-naming and we have also seen
the power of concepts and conceptual thinking.
Therefore we must include sense relationships in our account of semantics!
Examples of Sense Relations in Language
Think about the pattern in English that includes the following pairs:
Ram / ewe; cow / bull; sow / boar
Duck / duckling; pig / piglet; father / son
Uncle / nephew; narrow / wide; male / female
Buy / sell
All of these are related words that represent sense relations in English, and are a
part of the semantic structure of the language.
Two Kinds of Semantics
Thus, using the previous distinction between Sense and Reference, we can say
that we have two kinds of semantics:
One that deals with semantic structure that is intra-linguistic (sense relations)
And the other that deals with meaning in terms of our experience outside
language.
The Language Model has a Similar Dichotomy
This dichotomy between intra-linguistic and extra-linguistic relations is inherent in
the language model.
Remember the distinction between phonetics (speech sounds) and phonology
(sound system of a language)
Or, the distinction between sentences and utterances, etc.
J.J. Katz and J.A. Fodors Structure of Semantic Theory
Katz and Fodor tried to limit semantics and keep it only within the boundaries of
sense relations:
The structure of a semantic theory, Language 39, 1963.
They discussed sentences and whether the word or the sentence should be the
basic element of semantics.
Katz and Fodor, contd.
Katz and Fodor thought that a semantic theory must account for the following
semantic properties:
Ambiguity
Anomaly
Paraphrase
Manfred Bierwisch on Semantic Properties
Bierwisch argues that a semantic theory must account for a variety of sentences
such as the following:
A. His typewriter has bad intentions.
B. My unmarried sister is married to a bachelor.
C. John was looking for the glasses.
D. The needle is too short.
E. The needle is not long enough.
F. Many of the students were unable to answer your question.
Bierwisch contd.
G. Only a few students grasped your question.
H. How long did Archibald remain in Monte Carlo?
I. Archibald remained in Monte Carlo for some time.
(A) is an example of anomalous sentence, (B) of a contradictory one, (C) of an
ambiguous one, (D-E) illustrate paraphrase
Contd
(F-G) one follows from the other, whilst in (H-I) the former implies or presupposes
the latter.
Bierwisch insisted on defining semantics in terms of truth-relations between
sentences, especially on those involving logical and analytical truth (All bachelors
are unmarried, All bachelors are miserable, 6.4)
Bierwisch: All meaning that does not belong to semantics should be called
pragmatics.
Reference Pushed to the Side, Sense Becomes Dominant in Semantics
This has happened because the relationship between words (sentences) and
reality (outside world) appears to be almost impossible to grasp and define.
What seemed essential in the beginning in our description of meaning the
relationship between language and the world is ignored or given second place.
Scholars have focused on sense because it is much easier to formulate and
describe.
Sense is Easy to Describe as it has Structure
Sense has structure and can be accurately and precisely determined.
Bear in mind, though, that sometimes it is not easy to distinguish clearly between
sense and reference because categories of our language correspond to real world
distinctions.
E.g.: ram / ewe, bull / cow
Pros and Cons for Both Approaches
In order to deal with meanings in terms of the world, should Semantics try to
include the Sum Total of human knowledge? Is it possible anyway?
Another problem is the problem of deictics, i.e. indexical experessions such as
pronouns, demonstratives, time markers, etc.
Deictics present a problem for sense relations as they cannot be paraphrased in
any successful way. Sense relations cannot handle deictics.
An impasse?
Semantics, Sense and Reference, Part II
Vidanovid
Kinds of Meaning
Sense relations appear to be concerned with factual information or with
the so-called propositions that can be either true or false.
The old Aristotelian concept of a sentence was that:
S = (M) + P
or, a sentence contains an optional modality plus a proposition
Interlude: terms: utterances, sentences and propositions
Before we go on, yet another word of caution about the terms that will be
used in the course. An utterance is created by speaking (or, even, writing) a
chunk of language. For example If I say Ontogeny recapitulates phylogeny,
this is one utterance. If another person in the same room also says [the
same+, then we would be dealing with two utterances. (Saeed, 2003, p.13,
Semantics, Blackwell).
More on the terms
Sentences, on the other hand, are abstract grammatical elements
obtained from utterances. Sentences are abstract because if a third and
fourth person in the room also say Ontogeny recapitulates phylogeny with
the same intonation, we will want to say that we have met four utterances
of the same sentence. In other words, sentences are abstracted, or
generalized, from actual language use. (Saeed, ibidem, p. 13)
And more
Thus, we can understand sentences from the point of view of the speaker,
where they are abstract elements to be made real by uttering them; or
from the the hearers point of view, where they are abstract elements
reached by filtering out certain kinds of information from utterances.
Another term used for abstraction
In trying to establish rules of valid deduction in language, logicians
discovered that certain elements of grammatical information in sentences
were irrelevant. For instance, the distinction between active and passive
sentences is, logically viewed, irrelevant.
A. Caesar invaded Gaul.
B. Gaul was invaded by Caesar.
More on propositions
Thus A. and B. are simply the same (equivalent) because whenever A is true
so is B. In the valid chain of reasoning their grammatical differences will
never be relevant or significant. In a similar vein the following
information(al) structure will also be insignificant for for a logician:
More
C. It was Gaul that Caesar invaded.
D. It was Caesar that invaded Gaul.
E. What Caesar invaded was Gaul.
F. The one who invaded Gaul was Caesar.
Why would a logician look at them in such a way? Because these sentences
share a description of the same state of affairs. Once again, if one is true all are
true, and if one is false, all are false
Contd
This aspect of meaning has to be considered and it has been called
cognitive, ideational, denotational or propositional.
Palmer claims that it is not at all clear either that the study of sense
relations is the most important kind of meaning.
He also claims that providing information or presenting facts cannot be
the prime or only function of language.
Contd
Palmer: A great deal of our meaning is not ideational at all, but is inter-
personal or social, relating ourselves to others.
Some arguments supporting the idea that language is not simply a matter
of providing factual information:
First, we do not merely make statements; we also ask questions and give
orders.
Contd
Secondly, there are a variety of what today are called speech acts: we
persuade, we warn, we insinuate, we order, we use language to influence
other people.
(Optional: the distinction between constative and performative J.L.
Austin, How to Do Things with Words, Speech Act Theory)
Contd
Thirdly, much of what we say is not a statement of fact but an evaluation.
Some semanticists have made a great play with the emotive difference
between politician and statesman, hide and conceal, liberty and freedom,
each implying approval or disapproval. The function of such words is to
influence attitudes.
Contd
This emotive or evaluative function of language is not confined to pairs of
words such as these.
Simply put, there arent many words that the dictionary marks as derog.
But there are many other words for which part of the meaning is good or
bad such as: palace, hovel, hero, villain, and, even good and bad
Contd
Fourthly, language is often deeply concerned with a variety of social
relations.
We can be rude or polite, and our decision to be one or the other often
depends on the social relationship with the person to whom we are
speaking (Shut up, Be quiet, Would you please be quiet?, Would you keep
your voice down a little please? etc.)
Contd
In his attack on the study of sense relations Palmer insists on the point that
some parts of language are wholly social and carry no information.
The best examples are, naturally, chunks of phatic communion (Good
morning, How are you? and / or all the Englishmans remarks about the
weather
Phatic Communion
The fact that human beings tend to talk when they meet, often leads to a
sense of discomfort, even hostility, when silence occurs in such a meeting.
Because talk is often a first step in establishing a relationship it is
characterised by a stock of conventional utterances which break such
silences and help to establish the participants in a mutual situation in which
awkwardness and tension gradually disappear.
Social contact is, in turn, liable to generate speech between participants
who have nothing to say. This kind of utterance was given the name phatic
communion by Malinowski, the anthropologist who studied the speech and
customs of the Trobriand Islanders. He described such talk as a means by
which 'ties of union are created by the mere exchange of words.' Typically,
in New Zealand, such phatic communion centres on comments about the
weather, on personal appearance, enquiries about health, or affirmations
about everyday things. It serves in an atmosphere-setting capacity.
Phatic communion is the initial linguistic attempt to relate to another
individual. If this relationship develops, then small talk will take over, and
that in turn may lead to serious conversation. In any real situation all three
types of talk may be intermingled according to the circumstances.
Kinds of Meaning, Critique of Sense Semantics, Contd.
Fifthly, we need not mean what we say.
By the appropriate use of intonation we can be sarcastic for example so
that Thats very clever can mean Thats NOT very clever. (I dont like
*coffee* => I like tea; Shes very *clever* => Shes very ugly)
Palmer: Semantics cant fully succeed without an investigation of the
prosodic and paralinguistic features of language.
Contd
Sixthly, the type of meaning found in presuppositions such as When did you
stop beating your wife? simply cant be subsumed under the heading of
assertion (of facts).
Neither can the other notorious example The King of France is bald which
presupposes that there is a king of France, but a mere presupposition does
NOT assert his existence.
The Word as a Semantic Unit
Although the word is one of the basic units of semantics (dictionaries),
there are some difficulties.
First, not all words have the same kind of meaning (e.g.: full words vs. form
words = content/empty, lexical/grammatical, etc.)
*Palmer claims that form words have only a grammatical meaning that they
cant be defined except in relation to the other words.
Contd
One possible objection to Palmers view is the semanticity (type of
meaning) of deictic elements such as pronouns and prepositions or
determiners.
We believe that deictic elements (that are classified as form words) have a
very clearly non-grammatical reference to the real world extended in the
space-time continuum.
Contd
Secondly, it is not clear that the word is a clearly defined unit, except as a
conventional one.
Words are only written items with spaces between them that we have
learnt to place properly.
Compare (The) White House with greenhouse. Any reasons for saying that
the former is two words and the latter two?
Contd
One of the solutions to the problem of the definition of the word is
Bloomfields minimum free form, the smallest form that may be in
isolation.
E.g.: how about the, is, by, this in isolation? Thus we have to be very
precise about what we mean by isolation
Contd
Bloomfield defined the morpheme too, an element smaller than the word,
which is an element of meaning too.
Lets take a look at the word loved that has two morphemes: love- and d.
The two morphemes have clearly distinct meanings (like someone a lot;
past)
How about took (both take and past) that cant be divided?
Contd
Still, took does have two independent bits of meaning
The best way to treat the problem is to avoid morphemes (he, he) but
introduce a new technical term, called LEXEME.
A lexeme encompasses both love and loved under the same heading in the
dictionary.
So in this way we could talk about the meaning of words (i.e. lexemes) and
about the meanings of grammatical elements, such as the past tense or
plural.
Contd
Instead of treating loved as the two morphemes love and d, we shall
analyse it in terms of the lexeme love and the grammatical category of
tense.
Still, the status of the grammatical bit is fairly difficult to define as its not
clear that the plurality is always more than one just think about wheat
and oats
Contd
Thirdly, theres a problem that arises with transparent and opaque words.
Transparent words are those whose meaning can be deduced from the
meaning of their parts (doorman, chopper).
Opaque words are those for which it is not possible to determine their
meaning from their parts (axe, porter).
Contd
If we compare the English language with, for example, German then we can
find that many words which are not transparent in English are such in
German:
Thimble, glove, linguistics => Fingerhut (finger-hat), Handschuch (hand-
shoe), Sprachwissenschaft (language-science)
The German transparency is mainly due to its being an agglutinative
language.
Contd
A note: if we decide that chopper is to be interpreted as chop and er
(instrument), then what do we do with hammer?
Thus it appears that it is next to impossible to determine the semantic
elements within a word.
Fourthly, there are many words in English that are called phonaesthetic.
In such words the initial consonantal cluster indicates the meaning of a
rather special kind.
Contd
Examples in English are slippery, slimy, sluggish, sloppy, slithery, sleazy, and
so on. It seems that the initial sl- sound gives these words their unpleasant
connotations.
When Lewis Carroll coined the nonsense word slithy in the poem
"Jabberwocky", English speakers associated it with something slippery and
unpleasant. This word took on this particular meaning by association with
the number of other words beginning with sl- that have similar
connotations.
However, there are other sl- words, like sleep, slave, slat, and slogan, that
do not have those connotations.
Contd
The sk- words refer to surfaces or superficiality: skate, skimp, skid, skim,
skin, etc.
Other examples of phonaesthetic words are those with the final -sh / /,
such as crash, smash, lash, splash, and crush; and those with a final -k, such
as crack, whack, flick, and smack.
Contd
Look at the following:
glad glass glisten glow glare glaire glaze glint glitter
gloom glory gleam glimmer glade gloss glum
glabrous gloat
Common is visibility and luminosity
Contd
Consider the following English words:
SLACK SLOUCH SLUDGE SLIME SLOSH SLASH SLOPPY SLUG SLUGGARD
SLATTERN SLUT SLANG SLY SLITHER SLOW SLOTH SLEEPY SLEET SLIP
SLIPSHOD SLOPE SLIT SLAY SLEEK SLANT SLOVENLY SLAB SLAP SLOUGH
SLUM SLUMP SLOBBER SLAVER SLUR SLOG SLATE ...
A group of words such as the above has a cumulative suggestive value
that cannot be overlooked in any consideration of our habits of speech. All
the above words are in varying degrees pejorative".(Firth 1964: 184)
Contd
Fifthly, semantic division seems to override word division.
heavy smoker / good singer
Semantically they are not heavy + smoker and good + singer (debeo, teak
pua peva sa dobrom duom). Rather, they refer to someone who
smokes heavily and who sings well.
Contd
Sixthly, we simply miss adequate binary terms: ram/ewe; stallion/mare;
but no pairs for giraffe and elephant (male giraffe, female giraffe, elephant
bull, elephant cow)
Finally, there is the problem of treating idioms. An idiom is a sequence of
words whose meaning cannot be predicted from the meanings of the
words themselves.
Contd
Familiar examples: kick the bucket, fly off the handle, spill the beans, red
herring.
The point is clear if we contrast the above with kick the table, fly off the
roof, spill the coffee, red fish.
Semantically, idioms are single units, but not single grammatical units like
words for there is no past tense like kick the bucketed*
Contd
All these objections to the definition of the word as the natural unit for
semantics can lead us to abandon that idea
But we have to be very cautious here because we cant ignore the word and
look for some modality independent meanings for semantic entities that
are totally unrelated to words.
The Finishing Touch
Thus we shall sometimes analyse semantic units that are larger than the
word and semantic units that are smaller, but never take up some
disembodied meanings
Just think about the meaning of the word bull
The full meaning is inseparable from its sense relationships with cow and
calf, regardless of the lack of description for size, weight, colour, speed,
fitness, etc
Linguistic Context
Vidanovid, Semantics, Palmer, Ch. 5
Non-linguistic context vs.. linguistic context
In chapter 3 we had something like this:
Time, gentlemen! interpreted as?
Or, Ill get it, dont bother interpreted as?
Chapter 3 dealt with the possibility that meaning might be defined in terms
of certain non-linguistic contexts.
Reminder
J.R. Firth created some categories to determine the non-linguistic context,
such as:
The relevant features of the participants: persons, personalities
The relevant objects
The effects of the verbal action
Back to Linguistic Context
The extreme linguistic view sees the meaning of the word as wholly
stateable in terms of the context.
One of the reasons is that words are determined by the distribution of
linguistic elements (Zellig Harris)
That was structuralism at its peak, in the 1950s
Contd
Structuralist thought that the task of the linguist is to:
Provide rigorous empirical methods for the classification of linguistic
elements
For example, in phonology (locus classicus) the distribution of the two l
phonemes (leaf, feel) shows that they are one and the same phoneme since
they are in complementary distribution, one occurs only in those
environments that the other does not.
Contd
Another example: -en of oxen can be proved to be the same as the plural
ending s in cats, (in terms of distribution, of course and without referring
to their meaning!)
This is exactly in the spirit of structuralism where combinations and formal
structures need not rely on semantic criteria as was thought in the heyday
of the movement.
Harris on Meaning
Harris: It may be presumed that any two morphemes having different
meanings also differ in their distribution
However, the true origin of such linguistic elements is that the difference in
meaning brought about the difference in distribution.
Simply put dog is unlikely to occur in the same linguistic contexts as apple

Distribution criteria
Some linguists have suggested that the meaning of a word or morpheme is
determined by the linguistic environment in which it occurs
Two words can be considered synonymous if and only if they are totally
interchangeable in all environments
Conditional probabilities and meaning
Martin Joos, the famous structuralist from the 1950s, suggests that the
meaning of a linguistic unit is the set of conditional probabilities and the
practical meaning of a unit is left to the analysis of sociologists.
This shows the total dissociation of structural units and reality that was the
hallmark of the whole structuralist movement.
Syntagmatic vs. Paradigmatic
Structuralists actually insisted on the so-called syntagmatic relations in
their treatment of meaning = relations within a stretch of language
This meaning could also be termed as the meaning in presented, as
opposed to the meaning of in absentia, or the paradigmatic meaning.
Paradigmatic meaning is on an axis that enables substitution or
replacement of lexical items
Examples
The dog is on the mat = syntagmatic
The cat is on the mat = syntagmatic between cat and mat, but paradigmatic
between dog and cat
Palmer criticises this approach by stressing that nothing is actually gained
by the analysis of meaning in terms of distribution when compared with the
sense relations that we studied in the previous chapters.
Distribution need NOT, in essence, have anything to do with meaning (just
think about phonemes which have no meaning at all!)
Meaning and Distribution
Sameness and difference of meaning are NOT related to sameness and
difference of distribution
E.g. antonyms will usually be found with almost identical distribution, if you
think about them right?
Wide, narrow road, hem, trouser-leg, band, interests
Interesting metaphor by Palmer
Palmer: *most importantly+, it is surely obvious that to define meaning
in terms of distribution is very largely to put the cart before the horse.
(Palmer, p. 94)
He goes on to say that words have different distribution because they
have different meanings.
Thus, structuralism is a vacuous approach, Palmer seems to suggest.
Firth and Collocation
J.R. Firth was much less extreme than the staunch structuralists in his
definition of distribution.
He was famous for the following argument: You shall know a word by the
company it keeps.
This keeping company was for Firth the concept of collocation that
contributed to meaning only to a certain extent. It was only part of a word.
He was more interested in lexical probability or the mutual expectancy of
words
What is a collocation
A COLLOCATION is an expression consisting of two or more words that
correspond to some conventional way of saying things.
The words together can mean more than their sum of parts (The Times of
India, disk drive)
Previous examples: hot dog, mother in law
Examples of collocations
noun phrases like strong tea and weapons of mass destruction
phrasal verbs like to make up, and other phrases like the rich and
powerful.
Valid or invalid?
a stiff breeze but not a stiff wind (while either a strong breeze or a
strong wind is okay).
broad daylight (but not bright daylight or narrow darkness).
Criteria for collocations
Typical criteria for collocations:
non-compositionality
non-substitutability
non-modifiability.
Collocations usually cannot be translated into other languages word by
word.
A phrase can be a collocation even if it is not consecutive (as in the example
knock . . . door).
Non-compositionality
A phrase is compositional if the meaning can predicted from the meaning
of the parts.
E.g. new companies
A phrase is non-compositional if the meaning cannot be predicted from the
meaning of the parts
E.g. hot dog
Collocations are not necessarily fully compositional in that there is
usually an element of meaning added to the combination. Eg. strong
tea.
Idioms are the most extreme examples of non-compositionality. Egg. to
hear it through the grapevine.
Non-Substitutability
We cannot substitute near-synonyms for the components of a collocation.
For example
We cant say yellow wine instead of white wine even though yellow is
as good a description of the color of white wine as white is (it is kind
of a yellowish white).
Many collocations cannot be freely modified with additional lexical material
or through grammatical transformations (Non-modifiability).
E.g. white wine, but not whiter wine
mother in law, but not mother in laws
Linguistic Subclasses of Collocations
Light verbs:
Verbs with little semantic content like make, take and do.
E.g. make lunch, take easy,
Verb particle constructions
E.g. to go down
Proper nouns
E.g. George W. Busch
Terminological expressions refer to concepts and objects in technical
domains.
E.g. Hydraulic oil filter
Example: heavy and groupings?
magnitude: dew, rainstorm, downpour, rain, rainfall, snowfall, fall, snow, shower:
frost, spindrift: clouds, mist, fog: flow, flooding, bleeding, period, traffic:
demands, reliance, workload, responsibility, emphasis, dependence: irony,
sarcasm, criticism: infestation, soiling: loss, price, cost, expenditure, taxation, fine,
penalty, damages, investment: punishment, sentence: fire, bombardment,
casualties, defeat, fighting: burden, load, weight, pressure: crop: advertising: use,
drinking:
Back to Firth
Firth was thus only concerned with selecting the relevant features of the
linguistic or non-linguistic context, not with the totality of such contexts.
The study of linguistic context is important for semantics because :
By looking at the linguistic contexts of words we can distinguish between
different meanings (examples by Nida, chair, p. 95)
Firth, contd
Nidas examples:
1. sat in a chair
2. the babys high chair
3. the chair of philosophy
4. has accepted a University chair
5. the chairman of the meeting
6. will chair the meeting
7. the electric chair
8. condemned to the chair
Nida, contd
Nida offered four different meaning of the word in context, based on
collocations.
This kind of reference to context, present in dictionaries, does NOT
establish or define differences in meaning. It may only illustrate the
differences.
Nida, contd
Although the distribution of words may appear to depend on their meaning
(not the other way round), in reality this is not true in all cases.
Note: rancid ---> bacon, butter
Addled ---> brains, eggs
This happens in spite of the fact that English has terms such as rotten, bad,
sour
*Discuss later: How about white elephant?*
Collocations in the collective words
Flock of sheep
Herd of cows
School of whales
Pride of lions
Chattering of magpies
Exaltation of larks
He, he, he
Types of collocations? An Attempt by Djordje
Unique collocations (foot the bill, shrug your shoulders)
Strong collocations (ulterior motives, rancid butter, trenchant criticism, to
be moved to tears)
Weak collocations (white wine, red hair, a black mood, a blue movie)
Medium-strength collocations (to hold a conversation, to make a mistake,
to be recovering from a major operation)
Firths formal scatter : groups of related words
Difficulties with collocations
E.g. There is no meaning distinction between herd and flock, except one is
used with cows and the other with sheep
The difficulty arises from the fact that that a word will often collocate with
a number of other words that have something in common semantically.
But: The rhododendron died is OK, whilst The rhododendron passed
away is NOT
Idioms and collocations
Idioms (semantically opaque collocations? Sufficient definition???):
bring pressure to bear (bring pressure to bear on the student's
parents for example
Semantically transparent collocations:
bring the meeting to an end
Non-collocations:
...bring a pencil to the meeting
Syntax, semantics and idioms. Integration?
Two basic approaches
Integrate semantic analysis into the parser (assign meaning
representations as constituents are completed)
Pipeline assign meaning representations to complete trees only
after theyre completed
Berkeley Restaurant Project (BeRP), testbed for speech related projects,
1300 words only
From BeRP
I want to eat someplace near campus
Somebody tell me the two meanings
Pros and Cons
If you integrate semantic analysis into the parser as its running
You can use semantic constraints to cut off parses that make no
sense
You assign meaning representations to constituents that dont take
part in the correct (most probable) parse
Non-Compositionality
Unfortunately, there are lots of examples where the meaning (loosely
defined) cant be derived from the meanings of the parts
Idioms, jokes, irony, sarcasm, metaphor, metonymy, indirect
requests, etc
English Idioms again... more formal approaches
Kick the bucket, buy the farm, bite the bullet, run the show, bury the
hatchet, etc
Lots of these constructions where the meaning of the whole is either
Totally unrelated to the meanings of the parts (kick the bucket)
Related in some opaque way (run the show)
Examples
Enron is the tip of the iceberg.
NP -> the tip of the iceberg
Not so good attested examples
the tip of Mrs. Fords iceberg
the tip of a 1000-page iceberg
the merest tip of the iceberg
How about
Thats just the icebergs tip.
More
What we seem to need is something like
NP ->
An initial NP with tip as its head followed by
a subsequent PP with of as its head and that has iceberg as the head of its
NP
And that allows modifiers like merest, Mrs. Ford, and 1000-page to modify the
relevant semantic forms
The Tip of the Iceberg
Describing this particular construction
1. A fixed phrase with a particular meaning
2. A syntactically and lexically flexible phrase with a particular meaning
3. A syntactically and lexically flexible phrase with a partially
compositional meaning
Constructional Approach
Syntax and semantics arent separable in the way that weve been
assuming
Grammars contain form-meaning pairings that vary in the degree to which
the meaning of a constituent (and what constitutes a constituent) can be
computed from the meanings of the parts.
Constructional Approach
So well allow both
VP V NP {V.sem(NP.sem)}
and
VP Kick-Verb the bucket { x Die(x)}
(ambda) = in computer programming an object extended by additional
information
Semantic Grammars
One problem with traditional grammars is that they dont necessarily
reflect the semantics in a straightforward way
You can deal with this by
Fighting with the grammar
Complex lambdas and complex terms, etc
Rewriting the grammar to reflect the semantics
And in the process give up on some syntactic niceties
BeRP example
BeRP Example
How about a rule like the following
Request I want to go to eat FoodType Time
{ some attachment }
Semantic Grammar
The term semantic grammar refers to the motivation for the grammar rules
The technology (plain CFG rules with a set of terminals) is the same as
weve been using
The good thing about them is that you get exactly the semantic rules you
need
The bad thing is that you need to develop a new grammar for each new
domain
Semantic Grammars
Typically used in conversational agents in constrained domains
Limited vocabulary
Limited grammatical complexity
Chart parsing (Earley) can often produce all thats needed for
semantic interpretation even in the face of ungrammatical input.*
The Earley parser is a type of chart parser mainly used for parsing in
computational linguistics, named after its inventor, Jay Earley
(1970s). Earley parsers are appealing because they can parse all
context-free languages

S-ar putea să vă placă și