English Grammar: LEXICOLOGY

Introduction. Reference, sense, and referring expressions.

Lexicology is basically understood as the study of lexis (vocabulary or stock of words of a language). Stock evokes the idea of a set of words which brings the idea of the lexicon. The vocabulary is understood as meaningful units. (…)

Lexicology is closely linked to morphology. In lexicology what matters is the meaningful side of the words. And in morphology what matters is the form of these units. The form and the meaning of a word may be related so they are closely linked. Meaning takes us to semantics. Part of the semantic studies is based in meaning: lexical semantics. So lexicology has links to many other areas, but our vision of lexicology is restricted. Etymology is a branch of linguistics on its own. It's regarded with words and its history.

We should never mix lexicography with lexicology because it's not a branch of linguistics but a technique and has a very narrow aim. Lexicography deals with the making of dictionaries. So it's not a level of languages. We have a narrow concept of lexicology: it's a branch of linguistics concerned with the meaning and use of words. There is a diversity of approaches which goes from narrow conceptions to very wide conceptions.

In sentence semantics we can predict the meaning of the whole sentence because we have rules that show it to us. The limit is in lexical semantics, because here we don't have any possible combination that allows us to guess the meaning of a word. So “in linguistics is simply axiomatic that words have meanings…” (Harris).

• SENSE AND REFERENCE

(…) When we deal with the sense we deal with relationships within language. When we talk about reference we deal with the relationship between language and the world. By means of reference a speaker is pointing at something in the world by using language / a linguistic expression. Therefore with reference there is always an external reality that is being talked about.

It would wrong to think that every linguistic sign corresponds to some external reality. One single linguistic expression may point to different entities of the outside world.

Ex: Your left hand

There are different entities in the outside world that refer to this same expression.

A

• B

C

This is called variable reference. The perception of variable reference depends on a number of factors, which are there because of the relationship between language and the world. There are factors like time and place that play and important role in this.

Ex: The British Prime Minister

Does this expression have a referent? Yes. You could say Tony Blair but you could also say Margaret Thatcher. In this expression the time factor isn't expressed so both possibilities are possible. The question of place is specified so you can't say Aznar because it has to be British. If you say the current British Prime Minister, there is no variable reference.

In this relationship between languages and the world we also have expressions which only refer to one entity in the outside world:

• A

• B

• C

Ex: The moon, the sun…

They are not so frequent and they don't raise any semantic problems at all and they are not interesting for the purposes of lexical semantics. We can also find a number of linguistic expressions referring to one referent in the outside world:

1

A

2

Ex: The Morning Star

Venus

The Evening Star

In English, there are two expressions that refer to the same entity. This example is always like that, but there are cases in which this association between single entities in the world and different linguistic expressions can be also be made for specific cases.

The British Prime Minister

Tony Blair

The Leader of the Labour Party

This relationship isn't permanent. It has been made for a specific purpose.

That these expressions may be true depends on time and time factors.

Ex: John

The man at the corner

In a specific case these two expressions may have the same referent in the world.

Whenever you talk about reference there is always something in the outside world pointed out by an expression. When we talk about sense what we find is that the sense of a word is its place in a net of semantic relationships within a language. Sense is a purely inherently linguistic phenomenon. Sense is very difficult to define, but very easily perceived by native speakers. The clearest for a native speaker is `sameness of sense'.

Ex:

Almost has the same sense as nearly. These two words occupy the same place in the net of relationships in a language. The same happens with vertical and upright.

John took off his jacket

John took his jacket off

They have different syntactic structures but they have the same sense.

Bachelors prefer red haired girls.

Girls with red hair are preferred by unmarried men.

They also have the same sense.

The opposite also happens, that is, we may have two different senses and one linguistic expression. This is the case of polysemous words.

Ex:

Bank (of money)

Bank

Bank (of a river)

One single sentence can also have more than one sense. These sentences are represented in two different trees depending on the sense.

Ex:

The chicken is a living animal and is going to eat.

The chicken is ready to eat.

The chicken has become a meal and is ready for us to be eaten.

It's a lexical question.

If `with a smile' is a constituent of the higher level, the one who smiles is `He'

He greeted the girl with a smile.

If `with a smile' is a constituent of the lower level, the one who smiles is `the girl'

It's a syntactic question.

Sense is an abstraction, therefore, if you say a linguistic expression has a meaning, it's because it has sense but it may not necessarily have reference because not all words connect with an entity in the outside world. And this is a rule.

Ex:

So, in, at, over, either, onwards, and, almost: these words have no reference in the outside world.

Mind / Thought

Language World

Some make a distinction between the outside world and the world of the mind. For others it's all the same. (…)

Definitions may make us think about dictionaries. The sense these words have can be expressed in different languages. The problem with definition is that the only way we can get to the meaning of words like so, in, at… is through definitions. Why are definitions problematic? Definitions are given by using words and are circular. What happens with the sense and the words contained in these definitions? You can find some words in them you don't know the meaning so you have to look up their definition, etc. It's circular. It's possible to learn the meaning of a new word in a language through images. This is impossible with words that only have sense and no reference. The question of translations is that senses are out there, that they exist. In the same way as we can show (…) that sense can also be expressed.

Ex: Pavement and Sidewalk show the same sense.

Sense and reference may constitute the key in Frege's theory of semantics.

A referring expression is an expression used in an utterance to refer to something used with a particular referent in mind.

Ex:

Fred hit me Referring expression

There is no Fred at this address Non-referring expression, it has no referring interpretation.

Fred points at someone.

Proper names are inherently referential. But even a proper name can be referential or not depending on the utterance it's contained in. even if a word has reference that doesn't mean that that word is a referring expression. Depending on the syntax the interpretation may be referential or not or even problematic.

Ex: Indefinite noun phrases

A man was in here looking for you last night.

The first sign of the Monsson is a cloud on the horizon no bigger than a man's hand.

40 buses have been withdrawn from service.

This engine has the power of 40 buses.

A. and C. have referring interpretations.

This indefinition gives us the clue.

Nancy married a Norwegian.

Nancy wants to marry a Norwegian.

John is looking for a car.

A man with a limp killed Bob.

In A. she married a certain Norwegian, C. and B. clearly admits both interpretations and B. is has a referring interpretation.

If we want to clarify the interpretation of a noun phrase you insert `a certain'. Definite noun phrases have referring interpretations by default. However there are some cases in which they have no referring interpretations. The meaning of a pronoun is inherently a referring expression but sometimes the utterance produces non-referring interpretations.

Ex: If anyone ever marries Nancy, he is in for a bad time.

`He' has no referring interpretations because there is not anyone in our mind.

An equative sentence is a sentence that asserts the equivalence of the reference of two expressions. It always contains the verb to be: something is something else / A is B.

Ex: Margaret Thatcher is the P.M.

That Woman over there is my daughter's teacher.

John is the person in the corner.

Both expressions in each sentence have the same referent.

Cairo is a large city. Ted is an idiot.

This is not a referring expression so they're not an equative sentence.

One of the properties normally given to equative sentences is irreversibility, that is, A is B and B is A. But some sentences which are not equative can be irreversible.

Ex: A pint of beer is what I need

What I need is a point of beer

But there may be case also of equative sentences which aren't irreversible, which doesn't accept irreversibility.

Ex: That is the man who kidnapped my boss.

This isn't irreversible but it's an equative sentence.

2. Lexical units, semantic transparency, semantic opacity.

An opaque context is an utterance in which this association of one referent with two linguistic expressions doesn't occur. What is it that provokes an opaque context? This has to do with the factual world and the world of thoughts. In a situation in which someone talks about his thought, believes, what this person says is true because it doesn't belong to the world of facts.

Ex: Nancy believes that the British P.M. is the Leader of the Conservative Party.

If we take the fact that the British P.M. is the Leader of the Labour Party, this sentence is wrong but it's true because it refers to the world of thought and not to the world of facts. This is an opaque context.

Dennis thinks that the P.M. is a genius.

Dennis thinks that the Leader of the Labour Party is a genius.

These utterances may not be connected on with the other, because they are introduced by the verb to think.

The idea that we work in lexicology about lexical units it's not very different from that used in lexicography. A lexical unit in lexicography is the entry of a dictionary. The problem is that in lexicography sometimes there's no coherence, stability in the arrangement of entries in a dictionary.

Ex: The skylight was open.

She refused to open the safe.

The word `open' in the first sentence is an adjective and in the second one is a verb. They can appear as lexical entries depending on their grammatical information. But there some dictionaries that doesn't treat them like that.

Lexical units in lexicology normally follow two criteria. On one hand the basis fact is that a lexical unit is the smallest part which satisfies two things:

A lexical unit must be one word.

A lexical unit must be a semantic constituent.

Ex: We have a sentence containing the word `disobey'. The problem is that it has an internal structure: dis - obey. Both parts have their own semantic contribution to the whole sentence. It can be turned into: `…didn't obey'. The thing is that `dis-' is not a lexical unit because it is not a word.

Arthur pulled a fast one.

`To pull a fast one' is to cheat someone in a money matter. In this utterance we have the word `pull' but it's not a semantic constituent alone because it doesn't contribute to the meaning of the whole sentence. In this utterance `pull' alone means nothing. Therefore it isn't a lexical unit even if it's a word. The lexical unit would be `pulled a fast one'.

This is what happens with idioms. They're especial semantic combinations, that is, expressions whose meaning cannot be inferred from the meaning of its parts in isolated contexts. Idioms have to be lexically complex and semantically they are one single semantic constituent.

Ex: This will cook Arthur's goose

The idiom seems to be `to cook someone's goose'. So this sentence has four semantic constituents: this, will, Arthur, cook `s goose. It's obvious that `this' seldom would appear as an idiomatic expression. Proper nouns can be very often part of an idiomatic expression, but this is not the case

Idioms are elementary lexical units and they behave like single words in many respects. This is reflected in a number of characteristics they have. In many ways, idioms behave like words and they don't resist interruption and reordering of the part.

Ex:

To kick the bucket Idiom

To kick the large bucket No idiom

To pull somebody's leg Idiom

To pull somebody's left leg No idiom

Arthur apparently has a chip on his shoulder Idiom

*Arthur has a chip, apparently, on his shoulder

After a shaky start, we took them to the cleaners

*We took them, after a shaky start, to the cleaners

Neither of the following mechanisms can be applied to idioms:

• passive

• inversion

• topicalization

The idiom has to be left intact.

Ex:

*What john pulled was his sister's leg topicalization of the verb: it breaks the idiom

What john did was pull his sister's leg topicalization of the action / event: the idiom is left as such.

This also happens with the constituents of a sentence.

However, sometimes idioms show characteristics of phrases, that is, sometimes the elements inside an idiom can behave as individual words and, therefore, these words can be affected by morphological phenomena. Sometimes idioms accept pluralization. The plural mark isn't given at the end of the idiom but in the place it corresponds.

Ex: John has bees in his bonnet about many things.

*John has bee-in-his bonnets about many things

In morphology, there are mechanisms to create new languages out of words. Idioms, sometimes, enter this kind of mechanism. Idioms mostly behave as blocks but sometimes idioms show properties of phrases like the nominalization of the verb. They are semantically minimal but syntactically dual.

Ex: In English we can say `a leg-pull'.

The question of idioms takes us to the idea of degrees of opacity. Idioms are in terms of meaning opaque, but these expressions show a degree of opacity. Some are more opaque than others and some are more transparent than others.

Ex:

- Blackbird is more transparent than Ladybird

- Ladybird is more opaque than Blackbird, but more transparent than Red herring or In a brown study

- Red herring or In a brown study are the most opaque of the all.

You can say that Blackbird is a type of bird but Ladybird is not a bird although it connects with the idea of an animal. Red herring is not a type of bird, neither a type of animal, neither a type of herring. And In a brown study has nothing to do with its meaning. Blackbird has a semantic indicator but red herring has no semantic indicator.

The difficult part is to identify parts inside. It's more a question of feeling than a linguistic matter. However, there are some parts inside. Jackendorff stresses the idea that idioms have two types of components:

• predictable lexical items

• unpredictable lexical items

The thing is that it's impossible to systematize these two types.

Ex:

smile at / *take a smile at / make a smile at / *have a smile at

sniff (*of) / take a sniff of / *make a sniff of / have a sniff of

listen to / *take a listen to / *make a listen to / have a listen to

look at / take a look at / *make a look at / have a look at

think about / *take a think about / * make a think about / *have a think about

guess about / take a guess about / make a guess about / *have a guess about

ride (in/to/…) / take a ride (in/to/…) / *make a ride (in/to/…) / have a ride (in/?to/…)

drive (in/to/…) / take a drive (in/to/…) / *make a drive (in/to/…) / *have a drive (in/to/…)

fuss about / *take a fuss about / make a fuss about / *have a fuss about

try (*at) / *take a try at / make a try at / *have a try at

stab (at) / take a stab at / make a stab at / have a stab at

etc.

In Generative Grammar this doesn't have interest because there is no rule to follow so he says that there may be some generalizations. Facial expressions generally occur with make but there are many idiosyncratic variations also.

Ex: smile at / make a smile at

Sniff (*of) / *make a sniff of

The other factor affecting degree of opacity is the discrepancy between the combined contribution of the indicators, whether full or partial, and the overall meaning of the idiom. It is of course difficult to measure such a discrepancy objectively, but it does seem that, for instance, some of the so-called irreversible binomials: expressions usually of the form X and Y (where X an Y are noun phrases), whose semantic properties change when the order of the noun phrase is reversed.

Ex: `fish and chips' is taken to be as a fixed expression. It's a particular type of cooking the fish. You can say `chips and fish' but there's no longer the idiomatic expression in that. The same happens with `fish with chips'.

Other examples are: salt and vinegar

soap and water

bread and butter

Many dictionaries don't list this kind of expressions so lexicography doesn't pay attention to this.

We need the help of cognitive linguistics to find a theory that explains why the irreversible binomials are like that. Language is a cognitive skill, like smell, hearing and many others we have. There is a theory called the Theory of Markedness, which is related to our cognitive capacities as human beings. It says that there are things which are marked and things which are unmarked. This division is related to two factors: salience and frequency. Salience is: if something stands out in the world that is much more easily perceived by us than something which isn't so outstanding. And frequency is: if something happens more frequently in the outside world that is much more easily perceptible to us human beings. Salient and frequent things are filtered in much more easily than the rest.

How does this relate to the language? Salient and frequent things become unmarked in the language. In what way is the unmarked element different from the marked element? These elements cover an area of knowledge in the world. The unmarked element (positive) is representative of the whole of the domain, whereas the marked element (negative) is restricted to that part of the domain. You can't represent the whole thing.

Ex: Affirmative adjective question

How long is it? Very long / Very short

So `long' is a marked element because it's reduced to that area of scale

The theory that covers this is that the element is much more easily perceived than the rest cognitively speaking.

Ex:

Positive Negative Perceptual Property

big small ease of visual perception

tall short ease of visual perception

wide narrow ease of visual perception

loud quite ease of auditory perception

heavy light ease of weight perception

We can translate this into the question of irreversible binomials. The order of irreversible binomials follows this. The first element has to be easier for us to perceive than the second one. So the first element is the unmarked and the second the marked one.

Perceptual Saliency: something that is near is easier to perceive than something which is far.

Ex:

Near Far

Now and then (*then and now)

Here and there (*there and here)

Something which is large is easier to perceive than something which is small:

Large and small (*small and large)

Something which is singular is much better perceived than something plural:

Singular Plural

One and all (*all and one)

Ham and eggs (*eggs and ham)

Cheese and crackers (*crackers and cheese)

There can also be combinations:

Singular/large plural/small

Hammer and nails (*nails and hammer)

This theory explains these tendencies of order but it's not a recipe. We can also talk about the influence of cultural perspectives.

Ex:

Adults are more salient than non-adults:

Father and son (*son and father)

Mother and daughter (*daughter and mother)

Males are more salient than female:

Man and wife (*wife and man)

Humans are culturally more prominent than non-humans:

A man and a dog (*a dog and a man)

The possessor is more salient than the possessed thing:

John and his brother (*his brother and John)

An exception to all this could be the expression `Ladies and Gentlemen', which follows a phonological patron and that's the reason of its special order. It's a question of the suprasegmental structure of English language. The same happens with the expression `black and white', which in Spanish is the other way round, that is, `blanco y negro'.

A collocation is the tendency of two words to occur together or co-occur. In any case each of the two words are semantically independent.

Ex:

Fine weather Light drizzle

Torrential rain High winds

Collocations tend to favour specific meaning contexts:

Ex: `heavy' favours the idea of consumption and therefore it can be used in any phrase if that idea of consumption is implied. So you can say `heavy drinker', `heavy smoker', `(a car may be) heavy on petrol'

Collocations are easy to handle because they more or less behave freely. The problem comes when the link between words in the collocation gets tighter. These expressions are called bound collocations: the two words of the collocation are mutually selective and one depends on the other. In many respects they behave like idioms, but they are not considered idioms because of their syntactic behaviour.

Ex: to foot the bill

the electricity bill

all the bloody bills

These changes wouldn't be possible if it were a true idiom.

Bound collocations require the presence of the partner next to it. This means that the pronominalization is unacceptable in these collocations.

Ex:

- I've just got the bill for the car repairs.

- *I hope you don't expect me to foot it.

This is not possible because foot needs the company of bill.

You can't break the collocation with a sentence constituency.

Ex: I'm expected not only to foot, but also to add up, all the bills.

This is unacceptable.

So what distinguishes these collocations from idioms is that it can be modified like in the first examples. It can be modified one of the elements.

(Photocopy `Colours and their associations')

Ex:

Red alert: we can only have one interpretation which in this case is idiomatic. The noun `alert' would never in a normal situation accept the adjective `red', except in an idiomatic expression like this.

Green belt: we can have two interpretations:

belt cinturon. This is a literal interpretation. We have two semantic constituents.

idiomatic reading: we only have one semantic constituent green area of a city.

So we have two kinds of expressions:

expressions which only accept one interpretation

expressions which accept two interpretations: the literal interpretation and the idiomatic expression.

3. Core meaning or fuzzy meaning. Theories of the mental lexicon

Core meaning is the meaning which is at the centre of the word. And fuzzy meaning is a diffuse meaning. Meaning can also be denotative or connotative. Denotative meaning refers to the hard core of essential meaning of the word. Surrounding this core meaning there are a number of accidental facts that can be added or omitted without altering the basic meaning of the word. This is the connotative meaning.

Ex: bachelor [+human] [+male] [+adult] [- married]

Anything you know apart from this, which is the hard-core meaning, is added information, that is the connotative meaning.

Some people concentrate only on the hard core of the words, whereas some others give more importance to the extra information.

Denotation is intrinsically connected with reference. But denotation of a word is always invariance and utterance independent, it's part of language system independently of its use on particular occasions of utterance. Reference is variable and utterance dependent.

Ex:

'Dog' always has the same number of concepts that define the category of `dog'. It always defines the same class of animals. `Dog' by itself has no reference, what gives reference to it is the determiners added to it in an utterance. The lexeme DOG has denotative meaning.

G. Aitchison examined the word `tiger' and this is the information she found in two dictionaries:

• large Asian yellow-brown black-striped carnivorous maneless feline.

• type of very fierce wild cat that has yellowish fur with black bands across and lives in Asia.

The feature that these definitions share is only two: [+animal] [+feline]. Therefore, what are the basic concepts which are permanent in an object for us to call it `tiger'? then the characterization of words doesn't simply rely on a number of basic features and this is the example. This characterization needs to include some extra skills. Aitchison says: what happens with a tame tiger living on vegetables which is still called `tiger'? so there can be several types within the same category.

Componential analysis is a theory of lexical semantics which tries to characterize (…). There is a universe of atoms of meaning which are understood to be concepts. A concept is represented by a word.

Ex: the word `human' is used to represent the concept [±human].

The first problem here is that there are doubts that these atoms exist and the other that these need to be defined / limited to define the core meaning of words. Aitchison is wrong in this and Wierzwicka, a woman who represents a school of semantics, has elaborated a list of concepts.

Contrary to this idea of core meaning is that of fuzzy meaning. The meaning of lexical items isn't simple an addition of concepts. Aitchison and Lakoff believe that meaning of words is fuzzy (…). Discrete means that you can establish the limits of something, you can give an exact definition of it. Discrete meaning is the exact definition of a word and fuzzy meaning is the contrary. Aitchison says: `Words have fuzzy edges in the sense that there's no clear point at which one word end and another begin' (Words in the Mind, p. 46)

Large areas of the vocabulary of English have been described using this core meaning theory using nouns and verbs. One of Aitchison's criticisms is that not the whole of a language can be described with core meaning. Not all types of verbs can be defined, only action verbs are easily defined. That's because words don't have fix edges but fuzzy edges. Aitchison and Lakoff rely on a theory to justify their opinions about language and fuzzy meaning: Cognitive Linguistics. Experience is part of our linguistic capacities.

Rosch's Prototype Theory (P.T.) is based on cognitive linguistics and tries to explain the way we categorize the world. (Photocopy)

Ex: `bird'

Our experience of birds gives us more information than that it's an animal but the core meaning makes it difficult to find concepts exclusive of birds. For Spanish speakers the typical bird is a sparrow but English speakers' experience makes that their typical bird are robins. In a Prototype the word in the core contains all of the features of the category. Not all of the elements of the bird category have the same features and that's what hard-core meaning theory doesn't admit. The scale of prototypicality is itself a fuzzy meaning. In that scale we can talk about: core birds, less core birds, peripheral birds, more peripheral birds.

The typicality conditions are the features that the best example of a category fulfils. One of these typicality conditions must be included in the sense / meaning for the sentence to be grammatical. A linguist called Persson tried to reconcile the two options, to take the best of core meaning theory and the best of Prototype theory and put them together in a sort of coherent way. Both approaches recognize the existence of some features which are essential and other which aren't. the emphasis is what is more important. P.T. the essence is the prototypicality whereas in the core meaning theory the emphasis is made on the basic feature. We have the same facts but view from different angles.

Ex: Woman description

+human

+female

+adult

human

womb

children

This is Persson's model. We have like two boxes. One box that is close includes CC (Core / Categorial Concept). Below that (so there is no inclusion) there is another box with a dotted line which includes typical attributes. It's open because the list of features is open too.

In Core Theory we work with concepts but in P.T. we have examples. Persson's model uses plain English, natural language words because the knowledge native speaker has is manifested in plain descriptions / definitions, using natural language words. Native speakers knowledge of language follows an order which is the one his model has: first the essential features and then the typical attributes. Some people have criticized the inclusion of female in the categorical concept. It's redundant for them to say female and then the typical attributes.

Some people have criticised the inclusion of female in the categorical concept. It's redundant for them to say female and then the typical attributes. But they are wrong because there is no redundancy. They are the extension of the others.

The Network theory (called by Aitchison the Cobweb Theory) is a theory of the mental lexicon. Behind this theory what we have is a conception of the mental lexicon as a net in which each dot of the net is a word. These words are connected through others to the whole of the system. The relationships established between words have multiple routes. Not all the words are related equally. Distance: some are closer than others to the rest. Words from a same family are closer than those outside this family. There is a problem with this theory. Network theory fails to capture the overlapping of meaning. We have this when we have hyponymy or synonymy, for example. These semantic facts are difficult to reconcile with this theory.

word

Net

Person thinks that words are containers of meanings and this idea is compatible with the Cognitive Theory and Lakoff's idea about meaning. He assumes the question of distance in the network theory and the question of word families. In Persson's model we also have the common core metaphor, that is, the overlapping of meaning. Some words may share part of the meaning.

Container / word

Common core metaphor

There are psycholinguistic tests that prove this. (…)

It's obvious that the mind doesn't process words in the same way.

Ex: hammer concern

saw

car

`Hammer' and `saw' belong to the same family but they have nothing to do with the other words.

If we have words like `pretty', `handsome', `beautiful' and `nice', in a Network Theory this words form a family. Person thinks that they share at least one atom of meaning so this has to be represented in the theory. In `hammer' and `saw' there's no common feature, but in these other words we have a common core: `pleasing to the senses'

Common core

Person again tries to improve the difficulties of a theory giving a new version

The triangle of meaning (Odgen and Richards) was made for capturing the _________ of meaning.

Concept / Reference / Thought

Linguistic expression World Referent

There are two types of _________ semantics:

• Extensional semantics: the relationship between the linguistic expression and its referent and the world.

• Intensional semantics: relationship between the linguistic expression and its referent, but reference as a concept without paying attention to the world.

So we talk about the extension and the intension of words.

Ex: Figure (19)

Spinster: the person who spins this is the categorical concept that changes to the meaning it has nowadays, that is an unmarried woman. First it was used as a neutral name but in the 70's the typical attributes began to take importance and now we distinguish between an unmarried woman and a spinster. The typical attributes become defining and, moreover, pejorative. (…)

Questions related with the problem of categorizing:0

• Have you ever argued with another English speaker about whether or not to call some object blue?

• Have you ever been in doubt yourself, as an individual, about whether to call something pink or orange?

• Have you ever been in doubt whether to call something a tree or a shrub?

Questions related with the Prototype Theory:

• Could a double-decker bus (of the kind found in British cities) be a prototype for the predicate bus for a British English-speaker? Yes

• Could such a bus be a prototype for the predicate bus for an American English-speaker? No

4. Meaning Inclusion or Hyponymy

Hyponymy is the relationship of meaning inclusion between a superordinate term and one or more subordinate terms or hyponyms. We have a hierarchical relationship here. The meaning of the superordinate term is included in the meaning of the hyponym. We can also define hyponymy by using extension and intension. The superordinate term has greater extension than the hyponym and the hyponym has greater intension than the superordinate term, that is, this is more vague in meaning and therefore including more entities from outside world, whereas the hyponym has more semantic features so it's restricted to fewer entities in the outside world.

Hyponymy Tree

1

2 3

4 5 6 7 8

9 10

There is transitivity because there is a way of reaching the top from the bottom. 10 is a hyponym of 7. if 7 is a hyponym of 3 and 3 is a hyponym of 1, then 10 is a hyponym of 1 too.

Ex: Figure (1)

The two criteria along which we portray hyponymy are adulthood and sex. But it's incomplete so we should draw a tree like in figure (2). We have to combine both conceptual areas.

Boy: human, not adult, male (connected to man through the concept `male')

If this were a -tree, it would be wrong because the crossing of branches is forbidden in -theory. So this method was criticized, and Persson created a new system shown in figure (8).

Systematic Functional Grammar has as a basic tool the system which is just a set of semantic options. And through the combination of the options available in the system you get the grammar and lexical items. The basis is semantic information. You also have lexical semantic information, a hierarchy and the syntactic information include in that lexical items. The superordinate term includes the set of items denoted by the hyponym.

Ex: superordinate

subordinate

The group of all women would be inside the group of adults.

We have two types of inclusion:

- Referential inclusion is an `is a' relation. Ex: `a woman is an adult'

Superordinate

`is a'

Hyponymy

- Semantic inclusion is a `may be' relation. Ex: `an adult may be a woman'

Superordinate

`may be'

Hyponymy

Drawing the hierarchy in figures (1) and (2) isn't perhaps very convenient. But following person we arrange the same information in his conceptual model in figure (8). In this example, the concept `human' is in the core. Then we have a second level. In the third level we have lexical items. So woman is a `human' that is `adult' and `female'. `Girl' and `woman' are joined by the feature `female'.

1

2 3

2 and 3 are co-hyponyms and their relationship is that of contrast.

Problems with hyponymy:

• Use of concepts in the design of the hyponymy tree.

Ex: Figure (8)

Persson says that the concept [±adult] hasn't counterpart in languages. This is a problem because in the idea of [-adult] we have two atoms of meaning included: `child' and `adolescent'. The interesting part here is the definition of the words `boy' and `girl'. Persson says that these words include these two features. Therefore the lexemes become vague lexemes rather than ambiguous. When you say vague you mean that they can mean either only `child' or `adolescent' or both of them. Their meaning isn't discrete enough.

We deal with ambiguous items, when they mean either one thing or another but never both.

Disjunction is the relationship of `or'. In a vague word both senses can be included so vagueness is the same as inclusive disjunction, which is an `or' relation in which both elements may be included. Whereas ambiguity is an exclusive disjunction, that is, an `or' relationship in which one of the terms excludes the other.

Ex: `Stone'

• a piece of hard compact earthy or mineral matter.

• the hard central portion of a fruit.

Only one of the meanings can be activated, so the word is ambiguous; it's an exclusive disjunction.

Figure (10)

Instead of `not adult' we have `child' v `adolescent' (`v' represents inclusive disjunction).

If the meaning lexeme `male' is exhausted by the feature [-female] and if the meaning of the lexeme `child' is exhausted by the feature [-adult], then a boy would be a `male child'. But `an 18-year-old boy' is certainly not a child. That's why we have to include in the system the concepts `child', `adolescent' and `adult'. So a boy is a `male child' / `male adolescent'.

• Question of the concept `young'.

The following analysis was made by Fromkin Rodman:

Woman

+female

+human

-young

Girl

+female

+human

+young

The problem here comes with the concept `young'. It is the antonym of `old', not of `adult'. `Young' and `old' make a pair of gradable items.

Contradiction takes place whenever you state a property of something which has the opposite property. If we say the women are [-young], the phrase `young women' would be a contradiction. The same happens with the phrase `young adults', if in the idea of adult we include the feature [-young].

Tautology is when you say in the phrase a feature inherent to the definition of the word. So `young girl' is a tautologous phrase because girl is defined as [+young]. The same happens with `young children'. But all these phrases are correct. The problem is the inclusion of the feature [± young] in their definition. So this analysis is totally wrong. If figure (12) were correct a woman would obligatory be defined as an old female and a girl as a young female, which isn't necessary true.

*old boy *old girl

These two phrases are anomalous in a neutral context, that is, outside a friendship context. This reinforces the idea of the problem of using [± young] in a definition.

*an unmarried bachelor this is a tautology

All this is resumed in that we, perhaps, need a kind of scale: `young', `middle-aged' and `old'.

Persson's quotations:

`You are still young enough to think that torment of the spirit is a splendid thing, a sign of superior nature. But you are no longer a young man; you are a youngish middle-aged man...'

`Muriel was followed by a female child, unknown to me...'

`Young girl' is usually used as a `young female person' because for a `young female child' we use `little girl', which is more discrete than the first one.

(...)

`boy' `a male child from birth to puberty'

`girl' `a female child' it's not correct

The definition of boy is more neatly defined in terms of age than that given for girl. In English `youth' could cover the time span from adolescence to adulthood. However in real English, `boy' and `youth' are used indistinctively. So the definition given to `boy' is not correct in the real world.

With all this Lyons talks about Functional Aspects of meaning. The following text is a quotation by Lyon:

`By any of the most obvious criteria, girls reach what would normally be described as adulthood earlier, rather than later, than boys; and yet they are described as boys. The proposition “X is now a man” may well imply “X is no longer a boy”: but “X is now a woman” doesn't imply “X is no longer a girl.”'

So we can see that the meaning of `girl' and `woman' overlap in some aspects. So although they are co-hyponyms they don't contrast totally.

Ex: grown up girl *grown up boy

Ex: `Miss Simpson studied the girl seated across the table from her in the busy tea room. In the years since they had last met she had grown into a charming young woman...'

`Girl' and `woman' here refer to the same person.

`girl' `a young unmarried woman'

`spinster' `a woman who has passed the usual age for marrying or seems unlikely to marry'

Persson notes that the idea of marrying is also part of the knowledge we have of girl and woman.

Ex:

unmarried girl married woman *married girl

This is a contradiction.

Would it be ok to include [-married] in the definition of `girl'? No, because the example `unmarried girl' would be a tautology and it's not considered as that. So the idea of unmarrying is associated to `girl', but we cannot include it in the definition.

There is overlapping in the meanings of `girl' and `woman'.

There are no purely linguistic reasons for this. Lyons says that the kind of knowledge of the world we associate to words `girl' and `woman' is different from those associated to `boy' and `man'. It's a question of connotations. In the world of jobs `woman' gets the favourable connotations rather than `girl'.

Figure (54)

`child' v `adolescent'

`girl'

`young female adult'

The most fuzzy meaning is that of `girl' and isn't a co-hyponym as such of `woman'.

Carter concentrated basically on the question of artefacts and hyponymy. The idea behind this is that there are many words in the lexicon of a language which are defined depending on the purpose for which they are used.

Ex:

BUILDING

factory

hospital

house

museum cottage

theatre bungalow

school villa

The superordinate term is `building'; `factory', `hospital' and `house' are put one below the other in order to apply the rule `… is a type of…'. `Museum', `theatre' and `school' and `cottage', `bungalow' and `villa' are grouped in different groups because the first are private and the others public. The wrong thing in this tree is that as `museum' is below house, it implies that a museum is a type of house. It can be improved by the point of view of function. There are many cases in which buildings are defines because of the use we make of them.

`School' has lots of meanings:

`an institution for educating children'

`a place of education for children'

`attendance or study at such a place; a course of learning at such a place'

`the body of students (and teachers) at such a place'

All theses senses are related to the function. It's a fuzzy word because in the same utterance it can have several senses.

Figure (60): A functional hyponymy system.

This is a container model. All these words may be understood as an institution or as a place. And then they may be a building. In the third level we have that a house is a building. In the forth level, we have the function of the buildings. Here we have for the first time brackets, which imply optionality. But the function is always active. The problem here is that the superordinate term, that is, the core of the analysis, is optional and there's no obligation till the fourth level. But there is no solution. Either `institution' either `place' have to be activated.

Figure (62)

It's the function what makes the categorical classification. There is no optionality.

Figure (63) The same but with the container model.

Categorical hyponymy is when morphologically related words share part of the meaning. A derived lexeme contains the concept expressed by the base plus the additional semantic contribution that is added to the word. We can also establish the relation between several derivatives of the same word.

A base

B C

Derivatives

Ex: employ

employee employer

Morphology accounts for the meaning of `employer' in terms of a facet of `employ'. This facet of `employer' is [+agent +action] and of `employee' [+experiencer] or [+affected], which basically is the same as theme. These words include `employ' plus a semantic component. In the container model this would be:

employee [+theme] `employ' [+agent] employer

There are not only derived words, cases of conversion are also included here.

Ex: to drive core concept the / a drive [+location]

The verb `stand' can have several senses:

He is standing

I can't stand you

The meaning of the base `stand' is included in the second `stand' but this is more complex syntactically and semantically than the first one. It requires more thematic roles.

What is at the heart of this approach is meaning inclusion. The meaning of the superordinate term is included in the hyponyms, which are more complex than the base.

Hyponymy has the `a type of...' test, and then we also have meronymy, which uses the `a part of' test. With Meronymy we have the opposite case than in hyponymy. The meaning of the higher terms includes the meaning of the lower terms. It's a whole-part relationship between lexical items.

A

B C D

E F

`A' includes the meaning of `B', `C' and `D'; and `C' includes the meaning of `E' and `F'.

PLANT

leaf stem flower shoot

petal stamen

There can be plants that have no flower, but it's included anyway.

Meronymy trees are open to Prototype theory (some parts may be there or not, depending on the degree of prototypicality). They are useful for words that have more than one sense. The fuzziness of these words is captured in this kind of trees. Meronymy hierarchies are less clear-cut and regular than taxonomies. Meronyms vary for example in how necessary the part is to the whole.

Ex:

Day 24 consecutive hours.

sunlight scope.

Night non-sunlight scope.

`Night' may contrast with `day' but also with `morning', `evening'...

DAY

Day Night

down morning afternoon twilight evening night

The lower you come the subtle the meaning is.

Some authors emphasise the idea of lexical gaps, which has no explanation. They occur when there is no concept for that idea in the gap.

BUS

coach bus

single-decker double-decker

mini- bus trolley

5. Meaning Exclusion or Incompatibility

Jespersen distinguishes between contradictory terms and contrary terms. Contradictory terms are pairs of terms that excludes each other and they don't have middle terms, for example, like `white / non-white'. Whereas contrary terms are terms that admit a scale and middle terms, for example, like `rich / poor'.

The complexity of the question of meaning exclusion is enormous. We have several terms to refer to it: antonymy, incompatibility, opposition, contrast, polarity... There are three types of meaning exclusion:

Taxonomic Opposition: G. Leech says that there are two terms when the contrast is absolute, like a territorial boundary. The idea is that a domain is named by a property and it's divided in two parts and the two terms of the opposition cover two halves of the domain. Ex:

Territory [± live]

Terms alive / dead

One is in opposition with the other.

There are syntactic tests to prove this. They are test of entailment.

Ex: John is alive. John isn't dead

The dead animal is still alive (contradiction)

John is a bachelor but he is not married

Taxonomic opposition poses a fundamental problem in its arguments and there's no agreement in how to solve it. It has to do with the use of a concept to define the territory. In the example were we have [± live] as territory, why can't we have [± dead]? There's a question of markedness here. If we say that `alive' is simpler than `dead', we imply that `dead' is more complex than `alive'. Wierbicka says that `alive' is the negation of `dead' and not the other way round. She concludes that `dead' is the default term and `alive' the complex one. You can't mention `alive' if you haven't mention `dead' before.

The same happens with the following example:

Single not married

Married married

Persson's represents all this as follows:

He points out that probably the knowledge a native speaker has about the meaning of these two terms is not so much about what categorial concept these terms have but rather the set of typical features assigned to living and dead things. This means that the meaning lays on the typical features.

Polar Opposition: the terms in opposition in this relation allow some sort of degree modifiers. It's characterized by the existence of the idea of a middles term or a middle ground, also called norm. Who decides which is the norm? In order to establish the meaning of norm, we have to take into account the object-related norm and the speaker-related norm. Following the object-related norm, the norm is defined depending on the object:

Ex: big ant small elephant

`Big' and `small' doesn't mean the same in any case. It depends on the object they are accompanying. An small elephant has not the same seize as a big ant, although this is big and the other small.

The application of polar terms in the case of hyponymy terms makes the transitivity to stop working.

DOG

Labrador Alsatian ....... ....... ......

A small dog doesn't imply a small Albatian

In the speaker-related norm, the norm is established by the speaker. So different speakers may establish different norms. There is nothing objective in the norm. This norm is always evaluative, that is, it's a question of subjectivity.

Relative Opposition is what characterizes converse relations, that is, when a term is defined only through the converse term and vice versa.

Ex: teacher / pupil `pupil' is defined through `teacher'

own / belong to

parent / child

Cruse's ideas about oppositeness (Photocopy)

• Prototypical examples of oppositeness:

good:bad

large:small

true:false

top:down

command:obey

mother:father

town:country

tea:cofee

There is a great amount of examples like this. There is a scale of degree of oppositeness.

So long as two terms are felt to be choices in a certain context, these two terms enter an opposite relationship.

tea:cofee gas:electricity

They are opposite in some contexts. You have to choose one or the other.

Oppositeness is the semantic relationship about which native speakers have the strongest intuition, even stronger than that of synonymy. This gives way to the amount of examples and the scale of prototypical oppositeness

• Cruse's ideas about complementaries

Complementaries are exactly the same as taxonomic oppositions. We have one conceptual domain divided in two departments and the two terms fall in one or the other compartment. There is no neutral ground, no third term lying between them. Complementaries are recognize as an entailment relationship. The entailment is bidirectional.

Ex: John is not dead John is alive

The door isn't open The door is shut

The first sentence entails and is entailed by the second and vice versa.

As a result from this we have the diagnosis of denying both terms and producing an anomalous sentence.

Ex:

*The door is neither open nor shut.

*The hamster was neither dead nor alive.

This diagnosis test can be applied to other type of opposites giving correct sentences. This is because they are not complementaries but antonyms.

Ex: Her exam results were neither good nor bad

It's interesting to notice that the category of complementaries is in many cases verbs and adjectives. And, especially in the case of verbs, Cruse says that in a pair of complementaries is possible the option of a third term.

Ex: obey:disobey command command:obey

The idea of obeying or disobeying takes place only in a context of command. This is why sometimes there's a mixture of one of the terms of the complementaries pair with respect to this third term. So if you're asked what's the opposite of `command', you would say `obey', but you would never say that `obey' is the opposite of `command', but `disobey'. The thing is that `command' doesn't have a natural opposite.

Cruse also says that it's true that some complementary adjectives are odd in superlatives and comparatives and with adverbs, but it's not the case of all of them.

Ex:

*extremely true

*fairly dead

*a little shut

*more married than most

*moderately female

In complementary terms sometimes one of the terms admits grading and the other not.

Ex: dead: alive

*very dead

*moderately dead

*deader than before

very alive

moderately alive

more alive than before

open: shut

*slightly shut

*moderately shut

* more shut than before

wide open

slightly open

moderately open

more open than before

Moreover there is a term, `ajar', that refers to a grade of openness.

clean:dirty safe:dangerous

These pairs haven't been considered complementaries because precisely this grading diagnosis. Since they admit grading they aren't complementaries.

moderately clean

very clean

fairly clean

cleaner

slightly dirty

quite dirty

fairly dirty

dirtier

moderately safe

very safe

fairly safe

safer

slightly dangerous

quite dangerous

fairly dangerous

more dangerous

Many speakers reject the entailment in:

(1) `It's not clean' entails and is entailed by `It's dirty'

But they accept:

(2) It's neither clean nor dirty

Cruse says that we use `dirty' when there is a clear dirtiness and that it wouldn't be appropriate to use it when something is slightly dirty.

`It's not clean' entails and is entailed by `it's at least slightly dirty'

It's paradoxical for everyone to say: *It's neither clean, nor even slightly dirty.

This is completely different to `long' and `short'.

`It's not long' DOES NOT entail and is entailed by `It's at least slightly dirty'

It's neither long, nor is it even a little bit short this is not paradoxical since there is a region on the scale of length which exactly fits this description.

Even if the grading test is valid to identify pairs of complementaries. We have to be careful to apply `at least slightly' test to pairs which admit grading. This test works well with antonyms too.

Cruse says that antonyms are fully gradable and most of them are adjectives; there are very few verbs. The pair of antonyms express one concept: size, accuracy, length, height…

Ex:

long:short

fast:slow

hot:cold

easy:difficult

good:bad

When more strongly intensified, the members of a pair move in opposite directions along the scale representing degrees of the relevant variable property. The terms of an antonymous pair are symmetrically disposed around a neutral region of the scale, which is called the pivotal region, and which cannot be referred to by either member of the pair. In the majority of the cases, the pivotal region is not designated linguistically by any lexical item. There is one single exception. It's the case of `tepid' and `lukewarm', which refer to the pivotal region between `hot' and `cold'.

Poles give us the idea of extreme ends. Antonyms don't have extreme ends either, that is, you cannot name poles.

Ex:

The value of `slow', although it tends towards zero speed, never actually reaches it, but approaches it. This is not a physical fact, but a linguistic one: we cannot say `completely slow' when we mean “stationary”. Thus, we cannot say `completely cheap' when we mean “free of charge”, nor `completely short' when we mean “having zero length”.

It has been said that antonyms are always to be comparative.

Ex:

Expressions like `It's long' or `That's a long one' are to be understood to mean “longer than X”, where X is some explicit reference point on the scale of length.

`A tall man entered the room' is likely to refer to someone taller than the average adult male human.

`Isn't he tall?', however, may mean “tall for his age, family, class at school, tribe, or profession” or “taller than the last time the speaker saw him”, etc.

This leads us to say that the meaning of antonyms is more vague, more flexible than the meaning of taxonomic opposites.

In English there are some adjectives that are inherently superlative. Some scales, besides having a pair of gradable lexical items that are implicit comparative, also have lexical items which are better characterized as implicit superlatives.

Ex:

An obvious example of this is the scale of SIZE, which is associated not only with the antonym pair `large:small', but also with huge:tiny' and `enormous:minute', which are confined to the negative an positive extremes of the scale.

Implicit superlatives can be recognized by a number of distinctive properties:

They are, generally speaking, resistant to grading, although to varying degrees:

*very huge

*fairly huge

* This one is huger

*rather minute

*very minute

*This one is minute

*slightly enormous

*pretty tiny

They can be modified by unstressed `absolutely':

absolutely huge

absolutely enormous

absolutely minute

Simple antonyms sound very odd when spoken like this:

*absolutely large *absolutely small

Although they cannot be lexically or morphologically graded, they can be prosodically graded, that is to say by means of stress and intonation.

*gladder `glad' is the inherent superlative of `happy', so it's not gradable, although `happy' is fully gradable.

There is also a group of verbal opposites which share a great many characteristics with equipollent antonyms.

Ex: Consider `like:dislike' they represent psychological states and they are fully gradable:

I quite like it.

I like her enormously

And there is a neutral area between the opposites poles:

I neither like nor dislike her - she leaves me totally indifferent.

A further resemblance between verbs of this group and adjectival opposites is that they include what appear to be inherent superlatives.

Ex:

`love:hate' seem to be of this type. First, they are not so fully gradable as `like' and `dislike':

I quite like him.

I dislike it, a little.

*I quite love him

*I hate him, a little

Second they are modifiable by unstressed `absolutely':

I absolutely love it!

I absolutely hate it!

*I absolutely like it!

*I absolutely dislike it!

Underlying many lexical opposites there is a type of opposition which Cruse calls directional opposition. A pair of lexical items denoting opposite directions indicate potential paths, which would result in their moving in opposite directions. Cruse says there are three types:

• Antipodals: one term represents an extreme in one direction along some salient axis, while the other term denotes the corresponding extreme in the other direction.

Ex:

top:bottom

peak:foot (of mountain)

head:toe

always:never

all:none

start:finish

beginning:end

cradle:grave

We have verbs, nouns, nouns derived from verbs, adverbs…

• Reversives: pairs of words which denote motion or change in opposite directions. There are two main ways of characterising `opposite direction' for reversive verb pairs. The first applies to those verbs which refer to change between two determinate states. The reversivity of the verb pair resides in the fact that one member denotes a change from state A to state B, while its reversive partner denotes a change from B to A.

Ex:

appear:disappear

tie:untie the relevant states are “being tied” and “being untied”. The action of `untying' a shoelace is not the literal reversal of the action of `tying' it: usually one unties a shoelace merely by pulling the ends of the laces.

enter:leave

What is important is that the appropriate states should come about.

The second characterisation of opposite direction involves not absolute, but relative, states. Reversive verbs of this type denote changes between states defined merely as having a particular relationship to one another.

Ex:

ascend:descend

advance:retreat

Syntactically, the most elementary type of reversive opposites are intransitive verbs whose grammatical subjects denote entities which undergo changes of state.

Ex:

rise:fall

enter:leave

• Converses: pairs of words which express a relationship between two entities by specifying the direction of one relative to the other along some axis. So they are characterized by not representing the extreme end but the direction to that end. This definition doesn't match with the examples given by Cruse for it because they don't really show the idea of direction or movement.

Ex:

above:below

ancestor:descendat

guest:host

in front of:behind

father:son

teacher:pupil

before:after

husband:wife

master:servant

They express a relationship and depending on the perspective you give one label or the other.

Converse pairs in which the interchangeable noun phrases both occupy central valency slots are called direct converses; those where a central and peripheral noun phrase are interchanged will be called indirect converses.

Ex:

give:receive

bequeath:inherit

buy:sell

pay:charge

These are examples of indirect converses because you always have a third element between them. They have three -roles:

• you give something to someone

• you pay something to someone

• you buy something to someone

`Give' and `receive' are the same but it depends on your perspective.

6. Lexical Ambiguity: Homonymy and Polysemy

The notion of relatedness unifies homonymy and polysemy. A case of homonymy is one of an ambiguous word whose different senses are so far apart from each other and not obviously related to each other in any way. And homonyms are unrelated senses of the same phonological word.

Ex: “drinking vessel”

Mug

“gullible person”

There is no obvious conceptual connection between its two meanings.

Some authors distinguish between:

• Homographs: senses of the same written word.

• Homophones: senses of the same spoken word.

Saeed uses generally the term homonym. And he distinguishes different types depending on their syntactic behaviour, and spelling:

• Lexemes of the same syntactic category and with the same spelling:

Ex: “circuit of a course”

lap

“part of the body when sitting down”

• Lexemes of the same category, but with different spelling:

Ex: ring wring

• Lexemes of different categories but with the same spelling:

Ex: the verb `keep' and the noun `keep'

• Lexemes of different categories and with different spelling:

Ex: not knot

We can talk of prototypicality here, being the first case the most typical example of homonyms.

On the other hand, a case of polysemy is one where a word has several very closely related senses:

Ex: “of a river”

mouth

“of an animal”

The two senses are clearly related by the concepts of an opening from the interior of some solid mass to the outside, and of a place of issue at the end of some long narrow channel.

There is a traditional distinction made in lexicology between homonymy and polysemy. Both deal with multiple senses of the same phonological word, but polysemy is invoked if the senses are judged to be related. This is an important distinction for lexicographers in the design of their dictionaries, because polysemous entries are listed under the same lexical entry, while homonymous sense are given separate entries. Lexicographers to use criteria of `relatedness' to identify polysemy. This criteria include speakers' intuition, and what is known about the historical development of the items.

Ex:

Various senses of `hook' treated as polysemy and therefore listed under one lexical entry:

Hook: n.1. a piece of material, usually metal curved or bent and used to suspend, catch, hold, or pull something. 2. short for fishhook. 3. a trap or snare. 4. Chiefly U.S. something that attracts or is intended to be an attraction. 5. something resembling a hook in design or use. 6.a. a sharp bend or angle in a geological formation. b. a sharply curved spit of land. 7. Boxing. a short swinging in which the ball is hit square on the leg side with the bat held horizontally. 9. Golf. a shot that causes the ball to go to the player's left. 10. Surfing. the top of a breaking wave, etc.

Two groups of senses of `hooker' treated as unrelated, therefore a case of homonymy, and given two separate entries:

• Hooker1: n. 1. a commercial fishing boat using hooks and lines instead of nets. 2. a sailing boat of the west of Ireland formerly used for cargo and now for pleasure sailing and racing.

• Hooker2: n. 1. a person or thing that hooks. 2. U.S. and Canadian slang. a. a draught of alcoholic drink. b. a prostitute. 3. rugby. the central forward in the front row of a scrum whose main job is to hook the ball.

The most problematic one in the analysis is `Hooker2'. The distance between `Hooker1' and `Hooker2' is greater than the distance between the two senses in `Hooker1' or between senses 1 and 4, for example in `Hook'

Such decisions are not clear-cut. Speakers may differ in their intuitions, and worse, historical fact and speaker intuitions may contradict each other.

Ex:

Sole “bottom of the foot”

Sole “flatfish”

Most English speakers seem to feel that the two words are unrelated, and should be given separate lexical entries as a case of homonymy. They are however historically derived via French from the same Latin word `solea' “sandal”. So an argument could be made for polysemy.

This process is likely to happen continuously.

Exercise: Decide whether the following words are examples of homonymy (H) or polysemy (P)

Bark (of a dog vs. of a tree) Homonyms

Fork (in a road vs. instrument for eating) Polysemy (because of the shape)

Tail (of a coat vs. of an animal) Polysemy

Steer (to guide vs. young bull) Homonyms (because the words categories are different)

Lip (of a jug vs. of a person) Polysemy

Punch (blow with a fist vs. kind of a fruity alcoholic drink) This is probably the most controversial one because you can find it as two different entries or as only one, that is as a polysemous word or as a homonymous word.

These problem gets worse when these words appear in an utterance.

JAMES PUSTEJOVSKY

• The problem: Contrastive ambiguity and complementary ambiguity

The problem for Pustejovsky is that polysemy and homonymy have been treated as different semantic relationships among words. No account has been given from a unified perspective. both are considered cases of lexical ambiguity. Puestejovsky believes that ambiguity is the worst problem in language processing )language generation and language interpretation). In this dual view the generation and / or the parsing (interpretation) is the key issue.

It is certainly true that many words in a language have more than one meaning. but the ways in which words carry multiple meanings can vary. We can distinguish two types of ambiguity:

• Contrastive ambiguity: this is seen where a lexical item accidentally carries two distinct and unrelated meanings (i.e., homonymy).

Ex:

a. the bank of the river.

b. the richest bank in the city

a. The judge asked the defendant to approach to the bar.

b. The defendant was in the pub at the bar.

In the examples above the underlined items have more than one lexical sense. Whether these senses are historically related or accidents of orthographic and phonological blending, is, for Pustejovsky, largely irrelevant for purposes of lexicon construction and the synchronic study of meaning.

• Complementary ambiguity: It involves lexical senses which are manifestations of the same basic meaning of the word as it occurs in different contexts. The alternative readings are manifestations of the same core sense.

1. a. The bank raised its interest rates yesterday. (i.e., the institution)

b. The store is next to the new bank. (i.e., the building)

2. a. John crawled through the window. (i.e., the aperture)

b. The window is closed. (i.e., the physical object)

3. a. Mary painted the door. (i.e., the physical object)

b. Mary walked through the door. (i.e., the aperture)

Somehow, our model of lexical meaning must be able to account for how the word for bank can refer to both an institution and a building, how the word for window can refer to both an aperture and a physical object

• Sense Enumeration Lexicon: semanticality

Pustejovsky argues that the most direct way to account for polysemies is to allow the lexicon to have multiple listing of words, each annoted with a separate meaning or lexical sense. This is what he calls Sense Enumeration Lexicon (SEL), which is characterized as follows:

“A lexicon L is a Sense Enumeration Lexicon if and only if for every word w in L, having multiple senses s1,..., sn associated with that word, then the lexical entries expressing theses senses are stored as {ws1, ..., wsn}.”

Given this view of lexical sense organization, the fact that a word-form is ambiguous does not seem to compromise or complicate the compositional process of how words combine in the interpretation of a sentence.

Ex: The two contrastive senses of the word `bank' as used above could be listed in a straightforward fashion as below, using a fairly standard lexical data structure of category type (CAT), and a basic specification of the genus term (GENUS), which locates the concept within the taxonomic structure of the dictionary.

bank1

CAT = count_noun

GENUS = financial_institution

bank2

CAT = count_noun

GENUS = shore

`bank1' and `bank2' would be listed separately.

Pustejovsky felt the need to include the concept of semanticality, analogous to the view of grammaticality, but ranging over semantic expressions rather than syntactic structures. Semanticality refers to the well-formedness of expressions in grammar. It's a binary judgement on whether an expression is truth-functional or not.

A sentence may be judged wrong because the semantic conditions or because it's semantically wrong. So a sentence may be [± grammatical] or [± interpretable]:

- A sentence may be [+ grammatical] [- interpretable]:

Ex: The book is on the table

- A sentence may be [- grammatical] [+ interpretable]

- A sentence may be [- grammatical] [- interpretable]

- A sentence may be [+ grammatical] [+ interpretable]

Ex: Garden Path sentences: I knew the boy had been waiting for three hours it has two possibilities for processing, that is, `the boy' is first object of `knew' and then once it's processed it's also the subject of `had been waiting'.

This is what takes Pustejovsky to semanticality referring to sentences which even if they are grammatical, there is something wrong with them.

Ex:

(1) a. *Mary kicked me with her foot.

b. Mary kicked me with her left foot.

(2) a. *John buttered the toast with butter.

b. John buttered the toast with an expensive butter from Wisconsin.

Although a-sentences are not ungrammatical in any strict sense, they are semantically less acceptable than the b-sentences.

(3) a. *The house was built.

b. The house was built by accomplished builders.

(4) a. *The cookies were baked.

b. The cookies were baked in the oven.

Even if the semantic role is fulfilled, there may be something wrong.

This question of unsemanticality isn't something Pustejovsky took out of the blue. Coseriu and Kastovsky also dealt with this problem and they invent the notion of lexical solidarities. The notion of this arises out of the need of analysing the sintagmatic relations of words. Sintagmantic means the same as structural. So lexical solidarities deal with the problem of the location of the words in the structure and how that affects the semantic of theses words. Lexical solidarities are the determinations of the content of one word by a class of other words or by a specific other lexeme.

Opposed to the lexical solidarities, we have the selection restrictions, which are negatives implications. And lexical solidarities are the positive implication of the meaning of words in combination.

Ex:

- selection restriction Green ideas sleep furiously.

• lexical solidarities In the content of the lexeme `kiss' we have the content of `lips' and not the other way round. The feature `human' is contained in verbs like `marry', `apologize', `admire', `murder'...

There are two kinds of lexical solidarities:

• Unilateral

Ex:

bite - teeth

lick - tongue

kiss - lips

• Multilateral

Ex:

bark - dog

neigh - horse

bray - donkey

The syntactical semantic behaviour is different. The determining lexeme is not expressed.

Ex:

*The dog chewed the bone with his teeth.

*He kissed her with his lips.

These are semantically unacceptable. The explicit realization is normal if the determining lexeme is further modified.

Romeo kissed Juliet with chocolate smeared lips.

The dog chewed the bone with his sharp teeth.

These are semantically acceptable.

With multilateral solidarities the determining lexeme may appear or not.

Ex:

You can say: The donkey brayed.

Or simply: The animal brayed.

`Bray' includes the meaning of `donkey'

In order to build his lexicon, Pustejovsky is going to criticize the Sense Enumeration Lexicons. Basically there are two types of criticisms. One has to do with the creative use of words, that is, how words can take on an infinite number of meanings in novel contexts. The other is the permeability of word senses. He thinks that word senses aren't atomic definitions but overlap and may make reference to other senses of the world.

Ex: Consider the ambiguity of adjectives such as `good':

a. Mary finally bought a good umbrella.

b. After two weeks on the road, John was looking for a good meal.

c. John is a good teacher.

`Good' is ambiguous in the sense that for each case it has different interpretations. So within an SEL, the only way to represent distinct senses for this adjective would be by an explicit listing of senses in the received usage of the word: good1, good2, ...goodn. For the sentences above, this would correspond to the three fixed senses listed below:

good (1) to function well

good (2) to perform some act well

good (3) tasty

The conditions which make an umbrella `good for something', however, are very different from those which make John a `good teacher'. For each novel sense we encounter, the SEL approach must enter a new lexical item into dictionary, creating one entry for each new sense.

In English, there are verbs which are systematically ambiguous, requiring discrimination with respect to change-of-state vs. creation readings, depending on the context and on the internal argument.

Ex:

a. John baked the potatoes (change-of-state)

b. Mary baked a cake (creation)

a. Mary cooked the carrots (change-of-state)

b. Mary cooked a meal (creation)

Sometimes words have several senses which are logically related to each other. If we apply a SEL (Sense Enumeration Lexicon), the senses have to appear separate and this logical relation is not manifested whereas with a different model this logical relation is captured.

Ex:

Window1

CAT = count_noun

GENUS = aperture

Window2

CAT = count_noun

GENUS = physical_object

The problem with this is that the logical relation that exists between the things in the world denoted by these expressions is not expressed, and these senses are embodied in the use of the as in :

John crawled through the broken window

It activates the meaning of `physical object'

It activates the meaning of `aperture'

We have plenty of evidence against this idea of one sense, one word, because words behave in a much complex way than in this unilateral strategy.

• Constructional Polysemy: Qualia Structure

Constructional polysemy is when a word acquires multiple senses depending on the construction in which that word appears (normally more abstract meanings). This is the same as syntactic polysemy. Depending on the syntactic construction in which a word appears, this word acquires several senses. This is a question of specialization of meaning; i.e. words get more and more specialized.

Ex:

reel its meaning depends on the `of-phrase' that accompanies it. The word `reel' is different from a `film reel' or a `fishing reel'. This means that this word alone is taken to be something like a container. It's a vague word and specializes in contact with the `of-phrase' or a with a pronominal modification. We have sometimes that `reel' is used with the meaning of the content and therefore it adopts the whole of the meaning of the phrase:

I just accidentally exposed three reels [of Kodak].

We cannot omit `of Kodak' because `reel' adopts the whole meaning of the world. Sometimes `reel' specializes and adopts the meaning `reel of film':

The mystery is only resolved in the final reel.

This happens because it's a question of the construction in which in which this word enters. It's always a process of specialization. There is a degree of specialization. It goes from a general term to a concrete term.

Perhaps a clearer case of the specialization of meaning is the adjectible premodification. It happens to be that in English when you premodify a noun with an adjective, that adjective acquires multiple readings.

Ex:

a. a sad poem / poet / day

b. a fast motorway / car / driver

The adjective doesn't mean the same in all the cases.

The crucial idea for constructional polysemy is that the meaning of the head noun determines the meaning of the adjective. So we can defend that there is a close relationship between the adjective and the noun. What Pustejovsky says is that the adjective is like a higher-order predicate which establishes the link with the noun. It works as if we have a verb with its arguments. In this case the object is the noun and the adjectives are the predicates that take objects. Adverbs are always higher-order predicates, because an adverb is like the head predicate that takes the whole utterance as its argument.

Ex:

carefully higher-order level

the teacher

the lesson 1st order level

explained

`Carefully' has scope over the rest of the predication, i.e., the first order predication. It is the nucleus and `the teacher explained the lesson' is its argument.

`a fast typist' meaning `a person who types very fast' or `a typist who does other things fast'. The Generative lexicon has to interpret it as the default reading (`somebody who types fast') and then as an extra interpretation. We have the meaning `a typist who does something fast'. In order to achieve this other meaning, the context has to be informationally rich. The meaning of `typist' includes the predicate (verb) `type', which has an argument (x) and an event argument (e). this event argument expresses `fast', which is the way the event is realized. So when you combine `fast typist', it means `someone who types fast':

[x][typist (x) ^ fast (e) ^ type (e,x)]

[x][type (e,x)]

The meaning of typist includes the meaning of type, which implies a person (x) and an event (e).

The meaning of a lexical item necessarily enters on relationship with the neighbouring words. Pustejovsky believes that this relationship that is established between words has to be present in the lexical entries of words. A purely denotational approach is not enough because knowledge of the world is part of the meaning of lexical items. This is the Qualia Structure, which is the structured representation which gives the relational force of a lexicon item. It's a way of formalizing knowledge of the world. Obviously this has something to do with cognitive linguistics. This gives a formal representation of cognitive capacities. Qualia Structure specifies four essential aspects of a word's meaning (or qualia / quales):

• Constitutive: the relation between the object and its constituent or proper parts:

• Material

• Weight

• Parts and component element

• Formal: that which distinguishes it within a larger domain:

• Orientation

• Magnitude

• Shape

• Dimensionality

• Colour

• position

• Telic: it purpose and function of the object:

• Purpose that an agent has in performing an act.

• Built-in function or aim which specifies certain activities.

• Agentive: factors involved in the origin or `bringing about':

• An object

• Creator

• Artefact

• Natural kind

• Causal chain

All these kinds of information intermingle and have consequences on the meaning of the structures in which these words appear. A purely compositional analysis isn't enough. There are two general points that should be made concerning qualia roles:

• Every category expresses a qualia structure but not all the four quales are necessary in the Qualia Structure of words:

Ex:

- cookie and beer

We recognize that they are `foodstuff' and `beverage', respectively. While `cookie' is a term that describes a particular kind of object in the world, the noun `foodstuff' denotes by making functional reference to what we `do with' something, i.e., how we use it. In this case, the term is defined in part by the fact that food is something one eats, for a specific purpose, and so on. Similar remarks hold for the information related to the noun `beer'. The TELIC quale for the nouns `food' and `beer' encodes this functional aspect of meaning, represented informally as [TELIC = eating] and [TELIC = drinking].

- novel and dictionary

Although both objects are books in a general sense, how we use them differs: while one reads a novel, dictionaries are for consulting. The respective qualia values encoding this functional information for `novel' and `dictionary' are [TELIC = reading] and [TELIC = consulting]. This distinction, of course, is not the only way these concepts differ; the structure of the text in a novel is characteristically a narrative or story, while a dictionary is by definition a listing of words. This distinction is captures by the constitutive role, expressing the internal structural differences: [CONST = narration] and [CONST = listing of words]. Finally, even with their overall similarity as expressed in identical formal roles ([FORMAL = book]), novels and dictionary also differ in how they come into being, expressed in the agentive role. That is, while a novel is written, a dictionary is compiled: [AGENTIVE = written] and [ AGENTIVE = compiled].

All this is basic to understand how these words may appear in a construction.



…

CONST = …

FORMAL = …

QUALIA = TELIC = …

AGENT = …

This kind of representation is derived from the Head-driven Phrase Structure Grammar. This is a model to capture the relational form.

Ex:

novel

…

CONST = narrative

QUALIA = FORMAL = book

TELIC = reading

AGENT = writing

There you have information but there's no relation. `Reading' expresses `someone reading something'. That's what we have in the following one:

novel

…

FORMAL = book (x)

QUALIA = TELIC = read (y,x)

…

agent object (book)

In these action of `reading' you connect the agent with the object. You have to include the -roles in the action of `reading'.

break

E1 = e1: process

EVENTSTR = E2 = e2: state

RESTR = <

FORMAL = broken (e2,y)

QUALIA = AGENTIVE = break (e1,x,y)

…

The meaning of `break' includes among other things the qualia structure. The action of breaking comes about by the break_act (agentive quale) and you have three arguments an one is an event argument. In the action of breaking, you have two participants plus the idea that it's a process. What's the form of breaking? When something is broken. So the formal quale says that `break' comes about when something is broken (state). So you can't say: *The jar broke but it's not broken.

But this analysis doesn't give any specifications about possible restrictions that the participants have to fulfil. (Ex: it doesn't say who broke the object). The participants of the breaking action are not restricted.

kill

E1 = e1:process

EVENTSTR = E2 = e2:state

RESTR = < 

HEAD = e1

ind

ARG1 =

ARGSTR = FORMAL = physobj

animate_ind

ARG2 =

FORMAL = physobj

cause-lcp

QUALIA = FORMAL = dead (e2,)

Agentive = kill_act (e1,,)

Here, it specifies the semantic condition of the argument structure. The subject has to be an individual but it has to be a physical object. The object has to be also an individual but it has to be an animate physical object. `Being dead' is a state, so for example; `The stone killed John, and John didn't died' is semantically odd and this oddity can be perfectly explained with this scheme.

The qualia structure is a way of capturing the semantic oddity of many utterances.

Ex: Pustejovsky says imagine we have examples like:

a. Sam enjoyed (drinking) the beer.

b. Sam enjoyed (watching) the film.

c. Sam enjoyed (reading) the book.

d. Sam enjoyed (eating) the caviar.

A theory of lexical semantics has to explain what `enjoy' means and what `beer' is. The thing is that we know what they mean separately, but together they mean `enjoy drinking the beer'. A different model to Pustejovsky's would simply list the senses of `enjoy' when it's combined with certain words. For example `enjoy the beer' refers to `enjoy drinking the beer'. The default interpretation in these cases is possible because of the meaning of `beer', `film', `book' and `caviar'. Our knowledge of the world of these words tells us that if we apply `enjoy', `drinking' is part of the meaning of `beer', `watching' is part of `film', `reading' is part of `book', and `eating' is part of `caviar'. So in the qualia structure of `beer', it's included that it's for drinking, that `film' is for watching, that `book' is for reading, and that `caviar' is for eating.

The question of metaphor is a phenomenon called broaderning:

Ex: cloud of mosquitoes / dust cloud

We all know what a cloud is. Why then is it possible to say a cloud of mosquitoes or a dust cloud? In principle, the word `cloud' has its qualia structure. If one of the quales of its qualia structure is overridden then these examples are interpreted as acceptable. In the Constitutive quale is `water vapour'. If we override it, then we can say a cloud of mosquitoes. We substitute the information of `water vapour' by `group of mosquitoes'. In the case of `dust cloud', it is overridden by `dust'.

• Sense extensions

The qualia structure is part of the lexical entry of a word. Where as the Sense Extensions are lexically governed treatment of the phenomenon, that is, lexical rules applied to lexical items. One of this lexical rules is grinding: a process of sense extensions which creates mass nouns denoting an unindividual substance from count nouns denoting an individual physical object of the same kind.

Ex:

Count Noun Mass Noun

a lamb lamb

a rabbit rabbit

a chicken chicken

A dictionary doesn't list `lamb', `rabbit' and `chicken' as mass nouns but as count nouns.

Grindings nouns are not, of course, listed in a dictionary because they are the result of a primary rule. So unless grinding process has unpredictable consequences, there is no list. This rule is extremely productive. One of the syntactic consequences is that you cannot coordinate a ground sense of a word with a non-ground sense of that word.

Ex:

(1) a) *Sam fed and carved the lamb

b) **Sam fed and enjoyed the lamb

Here we have Aronoff's `blocking': the non-existence of a word because of the existence of another word.

(2) a) *Sam ate pig ( pork)

b) *Sam likes cow (beef)

You don't say `pig' or `cow' because we already have `pork' and `beef'.

Pustejovsky points out that processes of grinding are largely conventionalised and they vary cross-linguistically. Every language has its own way of doing grinding.

Ex: There is no pragmatic reason to explain why `orange' admits grinding and `pear' and `apricot' doesn't.

(3) a) *I drink pear rather that peach

b) I drink orange for breakfast

c) I drink *apricot / apricot juice for breakfast.

The other process we have is portioning: a mass noun is converted into a countable noun denoting a portion of that substance. Here we also have the question of convention. This is a very productive rule specially in English.

Ex: beer

You can have three beers, which is the portioning. It's the convention which says if it's three pints of beers or three cans of beers, or three glasses of beers…

coffee

It's not a countable noun but through this process of portioning you can say four coffees meaning four cups of coffee for example.

lamb

From `a lamb' we have `lamb', the substance, and from that we have the portioning, that is, three lambs meaning three portions of lamb

• Nominal Metonomy

Another type of Sense Extension involves objects standing for people, representing people.

Ex:

(4) a) The third violin is playing badly

The violin cannot play by itself but here it refers to the person who plays it.

b) The Armani suit lounging gracefully at the bar looks bored.

The question that the subject is a man or a woman is a question of conventions and associations of words.

(5) The French fries is getting impatient

Grammatical agreement is not necessarily because in this case it's obvious that the French fries is only one person. The agreement is determined by the referent and not by the syntax of the phrase.

You can't coordinate the literal senses with the Sense Extensions.


Ex:

(6) **The ham sandwich wants a coke and has gone stale

**The French fires is getting impatient and are getting cold

• Lexical disambiguation

The problem of polysemy was the question of disambiguating the ambiguities. Pustejovsky says that we have to disambiguate words so that discourse incoherence is avoided; and disambiguate words so that rhetorical connections are reinforced.

(1) a) The judge asked where the defendant was.

b) The barrister apologised, and said he was at the pub across the street.

c) The court bailiff found him slumped underneath the bar.

The word `bar' is ambiguous. He argues that in the second example it is disambiguated to its `drinking establishment' sense on the basis of constraints on coherent discourse. The interpretation of c) in a discourse context, that is, taking into account a) and b), determines which is the most likely continuation of the story.

d) He took him to coffee before returning to the courtroom.

d') *He took him out of the courtroom to get coffee.

d) is the correct one

This question of taking into account the trail of events is basic to the disambiguation of words. Discourse markers lead us to specific interpretations.

Ex:

2) The judge asked where the defendant was, his barrister apologised, and said he was at the pub across the street. But in fact, the court bailiff found him slumped underneath the bar.

The discourse marker `but in fact' changes the sense of the last `bar'

Disambiguation can give different results in the sentence scenario to those we can get in the discourse scenario.

7. Synonymy

We say that two words are synonymous if substituting one for the other in all contexts does not change the truth value of the sentence where the substitution is made. Synonymy dictionaries include something that native speakers have very clear intuitions about. They have the intuition that a number of words may express the same idea.

Ex:

You can find `kill' as a synonym of `murder', and `strong' as a synonym of `powerful', but not the other way round:

*powerful:strong

*murder:kill

When you say they A and B are synonymous because they express the same object, you expect also that if A is synonymous of B, B is also synonymous of A. but this isn't reflected in dictionaries. If A is a synonym of B and B is a synonym of A, these are true or absolute synonyms. They are interchangeable. But there are no absolute synonyms, it's an intellectual creation. Native speakers feel that some pairs of synonyms are more synonymous than others. This gives us the idea of a scale of synonymy.

Ex:

+ settee:sofa die:kick the bucket boundary:frontier -

Obviously, the idea behind synonymy is that of sharing meaning that is that two words share (part of) their meaning. It has become a problem to establish how much overlapping do we need for two words for being considered synonyms.

Ex:

• truthful:honest they are synonyms although they share only part of their meaning

• truthful:purple they are not at all synonyms

Cruse says that an important thing here is contrast. When a speaker uses them indistinctively, he emphasizes their similarities not their differences.

Ex: kill:murder they share part of their meaning

The greater the number of features two words share, the more synonyms they are.




A

1

2

3

4

5

B

1

2

3

4

6




A and B share almost all of their meaning components.

Ex:




- creature

animal

dog

+ Alsatian

philosophy

tree

cat

Spaniel




`Alsatian' and `Spaniel' share more atoms of meaning than `creature' and `philosophy' but they are not synonyms. So this claim is wrong, because we need two things for synonymy: we need overlapping of meaning and, at the same time, the two words do not have to be contrastive.

Cruse says that synonyms must not only share high degree of semantic overlapping but also a low degree of implicit contrastiveness. So, a high degree of semantic overlap results in a low degree of implicit contrast.

Ex:

- John is honest

John is truthful

- He was cashiered, that is to say, dismissed.

He was murdered, or rather executed

`Cashiered' and `dismissed' are synonyms, while `murdered' and `executed' are contrastive synonyms

*Arthur's got himself a dog -or more exactly, a cat.

The inherent relationship between `cat' and `dog' is that of contrast, for that reason this sentence is odd.

It is impossible to put an end in the scale of synonyms.

Ex:

+ rap:tap

rap:knock

rap:thwack

- rap:bang They are not prototypical synonyms. They are peripheral synonyms

Behind any study of synonymy is the idea of the quest for the establishment of true synonyms. Cruse reviews some apparently true synonyms.

Ex:

begin:commence

munch:chew

hate:loathe

Cruse takes into account the question of the contextual relations. For two words to be true synonymous we need two conditions: equivalence of meaning and equivalence of contextual relations. This is highly problematic because words don't behave like that. They tend to specialize in their contextual relations.

Ex:

`Begin' and `commence' mean exactly the same but in terms of contextual relations they are not.

Johnny, tell Mummy when Playschool begins and she'll watch it with you.

*Johnny, tell Mummy when Playschool commences and she'll watch it with you.

Arthur is always chewing gum (+)

Arthur is always munching gum (-)

I don't just hate him, I loathe him (+)

I don't just loathe him, I hate him (-)

Apart from this there are minus aspects we have to take into account

- Syntax: two syntactic terms have to behave syntactically the same

Ex:

Where is he hiding?

*Where is he concealing?

`Conceal' needs an argument (DO)

Johnny, where have you hidden Daddy's slippers? (+)

Johnny, where have you concealed Daddy's slippers? (-)

- Sense: you have to choose the correct sense of the word if you want to prove that two words are synonymous.

Ex:

Arthur's more recent car is an old one (+)

*Arthur's most recent car is a former one (-)

He had more responsibility in his old job

He had more responsibility in his former job

• Conceptual Synonymy

Words are felt to be synonymous independently of their contextual relations. Leech makes the distinction between synonymy and conceptual synonymy. The equivalence of meaning of synonymy has to adhere to the equivalence of concepts, independently from the stylistic overtones.

Ex:




Steed (poetic)

Horse (general)

Nag (slang)

Gee-gee (baby language)




The concept `horse' is evoked by these words. So these words are synonymous although they are different in their stylistic overtones. This has been strongly criticized because to prove that we all have the same concept is very doubted. Our system of conceptualisation may be different from one speaker to other. The most evident example of this is baby language. When a baby says gee-gee he may be saying it to any animal that moves.

So conceptual synonymy is alright but it has faults and objections.

Wierzwicka says that it isn't possible to distinguish semantic meaning and factual meaning. Her lexicographic descriptions are very lengthy because she has into account all knowledge of the world, that is, the habitat, size, appearance, behaviour, relation to people…

• Componential analysis

It is an analysis very popular in the 1970's and turned itself to be very useful in the identification of atoms of meaning of words. One of the applications of componential analysis is in the identification of synonyms, because if two words share atoms of meaning, they are synonymous.

Ex:

John is a bachelor

John is an unmarried man

Componential analysis serves quite well for the analysis of fairly uncompleted words (nouns, adjectives, some verbs), but there are whole areas of the vocabulary of the language that don't lend themselves for componential analysis.

Barbara Warren makes a distinction between synonyms and variants. She says that we have synonyms if the words have similar meaning and if they are interchangeable without affecting meaning in some context or contexts. Variants are words which have similar meaning but without the interchangeability in some contexts.

Ex:

`extending

Deep far below Profound

the surface'

Contexts Contexts Contexts

`Deep' and `profound' has always been considered synonyms and it's true they are interchangeable but it's also true that in some contexts one cannot replace the other.

• He had a deep / profound understanding of the matter

• This river is deep / *profound They are not interchangeable in this context.

Ex:

Sweet:candy dialectal variants

Decease:pop off stylistic variants

Lady:woman connotative variants

In one context you use one word and in the other you use the other one.

`human

lady adult woman

female'

Context Context

The point here is to try and prove that synonyms exist. The result of this research is quiet distressing. There are no synonyms following Warren's definition. What Persson did was to scrutinize the use of `deep' and `profound'. His research is especially valid because he bases his research on lexicographic words, corpus data and informance. The wide range of sources and the number of them is what makes this valid.

The conclusions: `Deep' and `profound' show a difference in collocability, that is, they tend to collocate with different words. `Deep' tends to collocate with words of affection, conviction, feeling, regret, satisfaction, sorrow… Whereas `profound' tends to collocate with words of difference, distaste, effect, failure, influence… They enter different collocations because they mean slightly different things. They specialize in certain areas of meaning and that makes them slightly different. He also talks about metaphorical status. Metaphorically speaking, they can mean position on the one hand or quality of depth on the other. Only `deep' enters for the position metaphor, but the quality of depth can be expressed by both of them.

Ex:

deep structure (*profound structure)

He was deep (*profound) in thought

It was deep (*profound) in the Middle Ages

Deep / profound learning

Deep / profound sleep

Intellectual - emotive dichotomy: `deep' and `profound' tend to relate respectively to intellectual and emotive words. The idea is that `deep' tends to collocate with emotive nouns, whereas `profound' tends to collocate with intellectual words.

There is a difference in the degree of depth and intensity of these words. `Profound' is deeper that `deep'. When both are possible, then there is a distinction.

Ex:

He has a deep understanding of the matter (`pretty good')

He has a profound understanding of the matter (`very good')

English words associations give us a very useful way to prove this. There are nouns whose inherent meaning is superlative. With such a noun you can only have `profound' because it means deeper.

Ex:

profound distaste *deep distaste

profound repugnance *deep repugnance

Of course in terms of truth-conditions one entails the other one but not vice versa, that is `profound' includes `deep' but not vice versa.

Ex:

His profound insight into human nature has stood the test of centuries His deep insight into human nature has stood the test of centuries.

His deep insight into human nature has stood the test of centuries. * His profound insight into human nature has stood the test of centuries

Synonymy is understood within mutual entailment (A ! B) but `deep' and `profound' doesn't correspond to this.

Native speakers feel that `profound' is stylistically more elevated or more formal that deep. So with all this evidence it is impossible to say that they are synonymous. This is why Persson gives the following figure as the analysis for them.

Deep CC Profound

TA TA

`concrete' `situated, coming `abstract',

`abstract' from, or extending `intellectual',

`emotive' far below the `strongly

surface' `emotive'

SA

`informal' SA

`formal'

(SA: Stylistic Attributes)

In Persson's model we have three categories: CC, TA, SA. The thing is that not all words include SA box, so it's left open. Persson also reviewed other examples analysed by Warren.

Ex: child / brat

child

CC brat

TA

`Child' and `brat' are an example of connotative variant in Warren. They are given as variants but if we apply the test of hyponymy we see that it works. `Brat' is a kind of `child' but not vice versa. `Brat' includes `child' plus the feature `bad-mannered'. Persson finds the collocation in which `brat' appears; it tends to appear with adjectives that reinforces this feature of `bad-mannered' what proves that that atom of meaning (…)

The same happens with `woman' and `lady'.

Ex:

She is a woman, but she is not a lady.

*She is a lady, but she is not a woman

Persson questions the fact that two words can be synonymous out of the blue. He defends contextual information as the key yo determine if two words are synonymous or not.

Ex: readable:legible

At to what extent can we say that they are synonyms?

- readable:

1. (of handwriting or point) `able to be read easily'

2. `pleasurable or interesting to read'

- legible:

1. (of handwriting or print) `able to be read easily'

They are only synonymous when they mean `able to be read easily'

“The child, quite obviously, would not be expected to produce a composition, but would be expected to know the alphabet, where the full stops and commas are used, and be able to write in a readable / legible manner, something like, `The cat sat on the mat'.”

“It is not easy to see why her memory should have faded, especially as she wrote a most readable / *legible autobiography which went quickly through several editions.”

legible

readable

1 2

`able to `with pleasure

be read' and /or

interest'

They share senses number 1 but to `readable' it's also added sense number 2. This claims that in some contexts they are fully interchangeable, but we have also to take into account their stylistic feature and the register.

In principle, scientific words have discrete meanings.

Ex: mercury:quicksilver

They appear as full synonyms because they say that their relationship is that of mutual inclusion (A! B)

mercury quicksilver

`metal'

`mercury'

Conceptually, the concept `mercury' can be expressed with both words. However, style draws the line between both words. Native speakers and corpora of data give us what we have in the following figure:

Mercury

`formal', quicksilver

`scientific' `whitish

fluid `informal'

metal'

Mercury formal, scientific (Romance origin)

Quicksilver informal (Saxon origin)

However something peculiar has happened with this words. The popular word `quicksilver' is starting to disappear and what usually happens is that the formal words is the one that disappears. But in this case, it is the contrary.

-cigarette:fag

cigarette fag

`tube with

`general' tobacco in `slang'

it for

smoking'

`narrow, made

of finely cut

tobacco

rolled in

thin paper'

This figure contains not only CC but typical attributes too.

• Synonymy and collocative meaning

Ex: pretty:handsome

They have been considered similar in meaning but never fully synonyms. They belong to the same categorical concept

- Collocations by Leech:




girl

boy

woman

flower

pretty garden

colour

village

etc.

boy

man

car

vessel

Handsome overcoat

airliner

typewriter

etc.




- Collocations found in the Lob and the British Corpora:




Pretty

Batman

Case

Co-ed

Dress

Headdresses

Girl

Piece of seamanship

Quilt

Range of pram sets

Shoe

Shop

Sophy

Street

Teacher (female ref..)

Trick

Woman

Handsome

Cocktail cabinet

Connor Winslow

Face (male ref..)

Man

Mayor

Offer

Pair of salad servers

Person (male ref.)

(Red brocade) curtains

Son

Staircase

Sub-Alpine gloom

Trees

Vessel

Volume (book)

Woman




`pretty' female nouns

`handsome' male nouns

This is the first division we could make but there are more differences. It cannot be based on terms of male / female words.

The idea, then, is that if an adjective tends to collocate to certain nouns means that its partner is slightly different to it. So when they are applied to the same noun, the same rule is applied.

Ex: pretty:handsome

Mary is a pretty woman

Mary is a handsome woman

A handsome woman is more elegant that a pretty woman. She also has stronger facial features. A handsome woman isn't a pretty woman at the same time and vice versa. So they are exclusive terms.

`pretty street' but `handsome avenue'

If they are exclusive terms, they are nor synonyms but co-hyponyms

If two items are closely synonymous, a coordination test will lead to a tautology.

Ex:

* Scientists have so far failed to find for this deadly and fatal disease.

However if we coordinate `pretty' and `handsome' what we have is a contradiction:

That woman is pretty and handsome

(Photocopy of definitions of `deep', `profound', `handsome', `lovely' and `beautiful')

Some of the dictionaries specialize more deeply that others.

`Profound' in the Longman is defined as deep but not vice versa. This also happens in `lovely' and `beautiful'.

• Uninformative; it doesn't give really the sense of the words.

• This isn't correct because `profound' emphasizes stronger that `deep' and this isn't true. There is a contradiction there.

• Introduction of the notion of `delicacy' for defining a pretty woman.

• This is the only dictionary which says that something pretty isn't something beautiful. They exclude each other. `Grand' is a feature of `handsome'.

handsome

`making a pleasant

lovely impression on the pretty

senses'

beautiful

Here, `beautiful' and `pretty' appear as co-hyponyms so they have to exclude each other. The CC is actually the definition given for `beautiful', so it's the generic word for the four words. `Lovely' is slightly more intense than `beautiful'. (It's the same relationship `deep' and `profound' have)

This shows how language establishes degrees of intensity.

8. Semantic Fields

Key Words:

• Semantic Fields

• Semantic Space

• Theory / Field Approach

• Association

• Lexical Field / Conceptual Field / Mosaic

• Model of Local Points.

Semantic fields are the answer to the problem / question of structuring the lexicon of a language. Those who defend the existence of semantic fields believe that the language is structured. They say that the words can be classified in sets, which are related to conceptual fields and these words divide the semantic space / domain in different ways. It's to be preferred that the label to use here is field rather than theory because theories are supposed to be complete and have explicit definitions of the matter in question, and this isn't what happens in the semantic field approach. We just have ideas of how things seem to be. Moreover, the semantic field approach isn't formalized and it was born on the basis of just a handful of ideas of how words work.

The basic notion behind any semantic field approach is the notion of association: words are associated in different words. We also have the idea of a mosaic. The words form it in such a way that for it to be complete you need all the words in their correct place. We also have to distinguish between lexical and semantic fields. Semantic fields have something to do with prototypicality. One of the main difficulties in the semantic field approach is to establish the exact number of words that are part of a set. Here is where Prototype Theory enters because it defines the basic features of a category.

• Model of Focal Points

Martin and Key concluded that the basic words of a category are very easy to identify by a native speaker but they say that the interesting point is the area a native speaker doubts whether to call something A or B. There are concepts which cannot be expressed in words. From the psychological point of view there are concepts which cannot be verbalized but that really exist in the mind. The aim of this model is to identify the relationship between the lexical fields and the semantic fields. And there are fields where the relationship doesn't exist.

The idea behind semantic fields is the arrangement of words in sets depending on the organizing concepts. Many semantic linguists say that it's difficult to think of a word outside a semantic field because if you say that a word is outside a semantic field, you say it's outside the lexicon. The problem with this is what happens with words which don't evoke a concept. Many words in English are meaningful but don't have a concept.

Ex:

Even / only

These words clearly make a semantic contribution to the sentence. It's not the same to say:

Only John drinks milk

than:

Even John drinks milk

So, the conclusion is that some words of a language don't lend themselves well to the analysis in terms of semantic fields. Other important idea is the difficulty of finding finite sets of words. In any case, there's an internal contradiction between the idea of a set with the structuring of words of a language. A set is a close set. A word can belong to several fields depending on the organizing concept.

Speakers of the language clearly identify the central example but not the peripheral ones. This doesn't mean that it would never happen that. The degree of flexibility in the discrepancy of the categorization of words is smaller.

Ex:

- Please give me some more table

`Table' is here a mass noun meaning `space in a table'.

- Two rices are grown in India

Here `two rices' refers to `two types of rice'

The idea behind this is that the dynamic character of a vocabulary cannot be reflected in the static character of the semantic fields, which are a static way of organizing the vocabulary of a language.

Ex: verbs of cooking (photocopy)

- cook, bake, boil

• appear in two places in the taxonomy

• only basic words show this characteristic.

- cook1, bake1

• most general terms

• refer to human activities

• only these freely occur intransitively with human subjects

I cook, he bakes

*John simmered yesterday

*Helen is frying

• cook2 and all the terms under it are process words which can be analysed grammatically as causatives

John boils artichokes John causes [artichokes boil]

cause *boil

John boil John artichokes

artichokes

- boil1 and its subordinates

• water or water-based liquid must be used

• the absence of water is necessary for fry, broil, roast and bake

• simmer / boil2:

simmer: the liquid is just below the boiling point

boil2: rolling bubbles

• poach: slowly cooked in water so that the shape is preserved

• stew: slowly cooked for a long time until it's soft

• braise: the food is browned and then cooked slowly in a tightly covered pot with a small amount of water

The more specific the meaning of the word, the fewer collocational possibilities there are (as you move down in the diagram the collocational possibilities are reduced)

Ex:




boiled meat

boiled eggs

boiled vegetables

*poached vegetables

*stewed eggs




Transitive implications may break down:

`poach' implies `boil' but `poached egg' doesn't implies `a boiled egg'

poach boil

poached egg * boiled egg

- steam / boil1

• steam contrasts with boil in that the food, which must be solid, isn't submerged

- fry and its subordinates

• require the presence of fat or oil

• deep-fry (French-fry): a large amount of oil or fat

• sauté: a small amount of fat (in a frying-pan)

- broil and its subordinates

• cook something directly under a heating unit or over or under an open fire

• grill: cook food on an open grill (sometimes synonymous with broil)

• barbecue - charcoal: cook food over hot coals

- bake2

• cook food in an oven such that the heat is indirect

- roast

• may appear to be a basic word and belongs at the same level as fry, bake, boil and broil

• most of its meaning is covered by bake and broil

• a relevant factor in the meaning of roast is the use of a spit

Componential analysis (Photocopy)

10. Lexical Entries. Semantic representations in dictionaries

• Word Meaning (Hudson)

Ex:

`A bicycle is a vehicle with two wheels which you ride by sitting on it and pushing two pedals with your feet'

We have to be careful with the question of terminology. Is this the definition of the word `bicycle' or of its meaning `bicycle'? It's the definition of the meaning `bicycle'. `Bicycle' is a word with seven letters which you use when you want to refer to a bicycle. Are we trying to define the sense or are we trying to describe the referent? Part of the definition of bicycle refers to object in the outside world that we call bicycle, in order to distinguish it from other objects. A definition defines one of the senses of a word and that sense of a word is a prototype. You don't define `bicycle' but its category.

Parts of the definition:

a) A bicycle is a vehicle

b) It has two wheels

c) You ride it

d) You sit on it when riding it

e) It has two pedals

f) You make it move by pushing its pedals

There are two types of information in a definition. It must contain at least one classifier and then at least one distinguisher.

a) is a classifier but that isn't enough. We have to include the number of distinguishers (b-f) in order to distinguish it from other words of the category. When you define a word you take into account neighbouring words of the same family, which are going to give you the hints for the distinguishers.

• Criteria for a successful definition of a concept C

• The definition must show the similarities between C and other concepts classifying C correctly.

• The definition must show the differences between C and other concepts supplying enough distinguishers to make it unique.

The idea of a semantic field takes us to the idea of a mosaic. A word forms part of a mosaic. So a definition is good only if it forms part of that mosaic distinguishing the word from the others. A definition can never be made in isolation because the meaning of a word depends largely on the meaning of other words.

Ex: A carrot is a vegetable

This is a bad definition because it has no distinguishers

Normally the information contained in the distinguishers has to do with knowledge of the world.

Ex.

• A two-pedal bicycle tautology

• John rides a bicycle this is not a tautology because you can do more things with a bicycle.

A good definition for a word includes a classifier, the name of a more general category, and at least enough facts to distinguish this word's sense from the senses of any other words that share the same classifier. These facts are distinguishers.

Following Aitchinson's theory in her book `Words in the mind', dictionary definitions aren't true and are misleading. All dictionaries are inevitably limited in the amount they contain, just because it would be quite impracticable to include all possible data about each word.

Ex: paint

One popular dictionary suggests that the verb `paint' means `cover surface of (object) with paint'. But if you knock over the paint bucket, thereby covering the surface of the floor with paint, you have not thereby painted the floor. Nor can one patch up the definition of the dictionary by suggesting that one must intentionally cover something with paint: for consider that when Michelangelo dipped his brush into Cerulian Blue, he thereby covered the surface of his brush with paint and did so with the primary intention that his brush should be covered with paint in consequence of his having so dipped it. But Michelangelo was not, for all that, painting his paintbrush. (Words in the mind, p. 13)

All this suggest that people have a much more detailed knowledge of the meaning of words than any book dictionary would have the space to specify.

Ex: week

Dictionaries usually define it as a seven-day cycle. But this underrepresents what most people know about a week: in England, it is thought of as five working days, labelled Monday, Tuesday, Wednesday, Thursday, Friday. These are followed by the `weekend', Saturday and Sunday, a sequence of two days off. People maintain this model, even if does not correspond to their own personal week. The week is an intangible cultural artefact - and does not even agree with the `official' week, which starts on Sunday. Compare it with an Inca week: this contained nine working days, followed by market day, when the king changed his wives. (Words in the mind, p. 69)

• Robert Ilson (`Journal of Lexicology')

Any good dictionary definition tires to give four types of information:

1. Syn-1: its syntactic categorization: is it noun, verb, etc?

2. Syn-2: its syntactic subcategorization: if verb, transitive, intransitive, etc?

3. Sem-1: its semantic categorization: January is a month / To be gorgeous is to be beautiful

4. Sem-2: its semantic subcategorization: January is first month / To be gorgeous is to be strikingly beautiful.

Ilson accuses lexicographers of providing wrong definitions by compressing information and using it as a synonym.

Ex: gorgeous adj: STRIKING

This is wrong because `striking' appears as a synonym.

Sometimes Syn-1 and Sem-1 are merged in one part of the definition and then Sem-2 is given in the rest.

Ex:

gorgeous adj: strikingly (Sem-2) beautiful (Syn, Sem-1

ugly adj: not (Sem-2) beautiful (Syn, Sem-1

and so forth: [and others or more] (Syn, Sem-1) [of the same or similar kind] (Sem-2)

Sometimes Syn-1 and Sem-1 appear separate.

Ex: gorgeous adj: having striking beauty

This kind of definitions are called formulaic definitions because they include opening gambits (having / marked by / characterized by / full of / of / with / that has…)

Ex:

- full of striking beauty

full (Syn, Sem-2a) of striking (Sem-2b) beauty (Sem-2)

- marked by striking beauty

marked (Syn, Sem-2a) of striking (Sem-2b) beauty (Sem-2)

- characterized by striking beauty

characterized (Syn-1, Sem-2a) [of striking (Sem-2b) beauty (Sem-1)] (Syn-2)

- having striking beauty

having (Syn-1, Sem-2a) [striking (Sem-2b) beauty (Sem-1)] (Syn-2)

- of striking beauty

[of (Sem-2a) striking (Sem-2b) beauty (Sem-1)] (Syn)

- with striking beauty

[with (Sem-2a) striking (Sem-2b) beauty (Sem-1)] (Syn)

gorgeously adv.: [in (Sem-2a) a gorgeous (Sem-1) way (Sem-2b)] (Syn)

41

women

adult

breast

womb

children

+human

+female

+adult

CC

TA

`human female adult'

`has breasts and a womb, can bear children, has a high-pitched voice, has feminine feature'

`y o u n g' `middle-aged' `o l d'

`child' and `a d u l t'

`adolescent'

X Y

G i r l

W o m a n

B o y

M a n

`alive'

[+live]

`dead'

[-live]

dead `not' alive

alive

dead

dead

`breathing', `warm', `able to move', `able to respond'...

`not'

dead

`not breathing', `cold', `stiff', `unable to move', `unable to respond'...

John is the child of James

James is the parent of John

`child'

`bad-man-nered'