Semantics: the meaning of ‘meaning’

We are all, in a sense, walking dictionaries. In addition to a set of grammatical rules which tell us, for example, that the article comes before the noun in English and not after it (as in Danish), we have a mental dictionary or lexicon which stores the form of a lexeme, any irregularities associated with it (for example that the past tense of bring is brought and not \_bringed_), its syntactic properties (e.g. that the verb _give_ realizes a three-place predicate) and its meaning. But linguists are noticeably less confident about offering rules and generalizations in the area of meaning than they are about grammar or phonology, and have long regarded semantics*, the study of linguistic meaning, as the ‘weak point’ in our understanding of language.

This chapter addresses the thorny problem of meaning in language. We will examine ways in which linguists have attempted to understand meaning through an analysis of sense relations and semantic features, and have grappled with types of meaning which go beyond the propositional content of the words a speaker utters.

The ‘weak point’ in linguistics?

No two speakers have exactly the same lexicon. My own active lexicon, for example, does not include ontological, because I’m not convinced I actually have a clue what the word means, but a philosopher friend uses it quite regularly. Some speakers have a bigger lexicon than others, though we need to be careful here not to confuse size of lexicon with breadth of standard language vocabulary – many people have an extremely rich non-standard or dialectal lexicon, but struggle to express themselves in a standard variety, and may find themselves stigmatized by mainstream society as a consequence.

Key idea: The lexicon and lexemes

The lexicon is the mental word bank in which we store lexemes, together with their form, meaning, syntactic properties and any specific features not predictable by grammatical rule.

Lexemes may consist of single words, or full phrases where the meaning cannot be reduced to that of its component parts. For example, the colloquial expression to kick the bucket may be viewed as a phrasal lexeme or idiom by virtue of the fact that its meaning cannot be reduced to that of its component parts – taken together, this idiom means ‘to die’.

What do we mean by ‘meaning’? Philosophers and linguists agree that the concept is extraordinarily difficult to pin down:

The statement of meanings is therefore the weak point in language study, and will remain so until human knowledge advances very far beyond its present state.(Bloomfield 1933: 140)

Nearly eight decades later, Paul Elbourne seems equally pessimistic:

Despite 2,400 years or so of trying, it is unclear that anyone has ever come up with an adequate definition of any word whatsoever, even the simplest.(Paul Elbourne 2011: 1)

There are a number of reasons why meaning seems less clearly structured and less susceptible to scientific investigation than other areas of language. Firstly, Bloomfield claimed that our knowledge of the world was quite simply deficient, and not readily susceptible to the scientific analysis he craved:

Actually our knowledge of the world in which we live is so imperfect that we can rarely make accurate statements about the meaning of a speech-form. The situations (A) which lead to an utterance, and the hearer's responses (C), include many things that have not been mastered by science.(Bloomfield 1933: 74–5)

He also notes the very rough-and-ready way in which meanings are often learned. In some cases a definition will work, but in other cases it proves impractical. Rather than attempt to define, say, the word apple to a child, for example, we are likely simply to reach for an apple, show it to the child and hope that he/she can extrapolate from that example what all apples have in common and, more importantly, what distinguishes them from pears, grapes, plums and so on. This is often the only way to explain word meanings – young children cannot after all reach for the Oxford English Dictionary or equivalent for their definitions – and generally it seems to work.

But if meanings are often acquired by little more than supported guesswork, how can we be sure that the meaning of a given lexeme as stored by one individual is identical to that of another? The short, and easy, answer to that question is that if our internalized meanings were radically different, then communication would be impossible: I might be talking perhaps about ‘soccer’ and imagining a game involving 22 players and a round ball, while another person would hear that term and understand what I mean by ‘blue cheese’, and yet another would access my meaning of ‘saucepan’. That clearly does not happen, and I’d be wasting my time writing this book if it did, as any attempt to communicate ideas would be futile. But we cannot be sure that our internalized meanings correspond exactly: is my definition of cup the same as yours, for example? At what point does cup become mug for you, and is that point the same for me?

For some philosophers, notably Wittgenstein, the meaning of any word eludes abstract definition and is entirely dependent on its use. It is therefore constantly modified and reshaped by its users. The word colour, for example, has a very different meaning for a painter and a snooker player: for the latter it contrasts with red and includes black, but not white, whereas no such restrictions apply for the former. In Part I of his Philosophical Investigations (1953), Wittgenstein argues that any attempt to find a common meaning for the word game in all its uses ultimately ends in failure. He observes that not all games have a competitive element (compare chess and solitaire); some, but not all, involve the amusement of children (catch, ring-a-ring-o’-roses); while some involve skill (tennis, chess) and others chance (dice). Our knowledge of the ‘meaning’ of the word game is therefore based not on some idealized notion of what a ‘game’ is, but on our ability to use the word in different contexts. Others would argue that a common ‘core’ meaning can be identified which is shared in all contexts (I know of no one who doesn’t think cups can contain liquids, for example), but that there is a significant amount of ‘fuzziness’ or semantic indeterminacy around that core. Indeterminacy is particularly evident with new lexemes, whose meanings are often contested (see Case study below).

Case study: The semantics of ‘Chavs’

In 2011 the journalist Owen Jones published a book entitled Chavs: The Demonization of the Working Classes, which was the subject of this withering critique by Rod Liddle in the Sunday Times (12 June 2011):

‘The author, an Oxford graduate from Stockport, has based it upon this demonstrably false premise, that working-class equals chav. And that, further to this, the deployment of the word “chav” is part of a conspiracy by the ruling class and especially the Tories to keep the lower orders in their place. And while he concedes that working-class people themselves do sometimes describe those they despise as “chavs”, this is but part of the “divide and rule” strategy employed by the bourgeoisie to maintain their economic and cultural hegemony. Yes, this is a book written by the bastard offspring of Private Eye’s Dave Spart and Sue Townsend’s Adrian Mole, a sustained rant devoid of nuance and wit, one part Socialist Worker editorial and one part undergrad history essay.’

Underpinning the political/ideological critique is an argument about semantics. In Liddle’s view, Jones treats chav as a pejorative synonym of ‘working-class individual’: whereas Liddle himself views the relationship as one of hyponymy: all chavs are working-class, but not all working-class people are chavs. At the time of writing a lively debate was ongoing on internet forums about (a) whether chavs are necessarily white, and if so (b) whether the term is not only pejorative, but also racist. Not everyone, however, even agrees that the term is pejorative: Labour MP Stephen Pound sees it as a term of envy, no different in kind from style labels such as Teddy Boy or Mod, used to identify groups in the past.

Semantic relativity

Studying semantics would be more straightforward if concepts could be taken as ‘given’ and simply assigned different labels by different languages, e.g. dog (English), Hund (German), Ci (Welsh) and so on. Such equivalence is, however, the exception rather than the rule, as anyone who has ever attempted a translation, even of a very basic kind, will know: languages divide up the world conceptually in different ways. French, for example, has no word for ‘shallow’ – the nearest equivalent is something like ‘not very deep’ (peu profond); on the other hand, French has two terms covering the semantic range of English cupboard, requiring the speaker to specify whether the item is wall-mounted (placard) or free-standing (armoire). The Japanese verb suu covers both ‘to smoke’ and ‘to sip’ in English.

Words which look very similar can have very different connotations: on the political spectrum liberal in British English is usually complimentary, implying tolerance and openness, but in the United States the same word is often used pejoratively, implying an over-readiness to accept fashionable left-of-centre ideas and extend the scope of the state at the expense of personal freedom, while in French libéral has exactly the opposite connotations, implying an over-readiness to dismantle the state and extend the free market.

Colour terminology offers a good example of lexical non-equivalence between languages, as can be seen in the examples of English and traditional Welsh below:

Table 9.1: Colour terms in English and Welsh

The semantic range of Welsh glas overlaps partly with that of English green, blue and grey, while llwyd overlaps partly with grey and brown. Russian, by contrast, has two words covering English blue: goluboi corresponds broadly with sky blue or light blue while sinii is dark blue. Many languages, including Vietnamese, Kurdish and Kazakh, do not distinguish blue and green as English does.

The semantics of colour has been a focus of scholarly attention since the publication in 1969 of Berlin and Kay’s Basic Color Terms: Their Universality and Evolution. Setting aside complex colour expressions (pea-green, sky-blue-pink and so on) and focusing only on basic terms, Berlin and Kay examined a sample of 98 languages spoken across the world and argued for a universal hierarchy, acquired by languages in chronological sequence. A very small number of languages, for example Dugum Dani, spoken in western New Guinea, are still at the first stage, in which only two colours are distinguished: prototypically ‘white’ and ‘black’, but covering the semantic area of ‘light-’ and ‘dark-’coloured respectively. The next stage is the acquisition of ‘red’ as a third colour term, followed by ‘green’ and/or ‘yellow’, and then ‘blue’. Russian and Italian have most basic colour terms, with 12; English has 11.

Berlin and Kay’s hierarchy has been modified and refined since the publication of Basic Color Terms, but some critics have challenged their universalist claims, viewing colour as a culture-specific concept. These critics argue that a bias towards western assumptions and perceptions underpinned much of their methodology.

Sense relations

While native speakers are often able to offer clear and unambiguous judgements of form (e.g. ‘“I have readed” isn’t English’, ‘“They are big ones” is correct but “they are bigs ones” isn’t’), they are less confident (and less likely to agree) on judgements of word meaning, or lexical semantics. To understand word meaning, therefore, we need to look at the concrete evidence provided by the sense relationships into which lexemes enter. An important distinction needs to be drawn here between denotation, or the relationship between a lexical item and the world, and sense, its relationship with other lexemes. Thus starling denotes a subset of the set defined by bird and having the properties of being small, black and speckled in appearance, and enters into sense relations with words such as robin, sparrow, bird and so on.

Key idea: Identifying word sense

We can identify the sense of a word by examining its relations with other words, the most basic forms of which are antonymy, synonymy, hyponymy and hypernomy.

Following Saussure (see Chapter 3), we can identify both paradigmatic (or substitutional) relationships between lexemes, involving their interchangeability in a particular context, and syntagmatic ones, involving their collocational possibilities: for example, one may toast bread in English but grill meat, despite the fact that the activity involved – exposure to heat – is essentially the same. Many of the terms used by semanticists to describe these sense relations are familiar, but employed in a more precise or specialized sense.

Sense relations between lexemes can be determined by specifying the truth conditions of the sentences in which they occur, i.e. the set of conditions that must necessarily be met for a sentence to be declared true. Consider, for example, the following two statements:

• The cat ate the starling.

• The cat ate a bird.

The first is true if – and only if (for which one writes, conventionally, iff) – the second is true also. This is an implicational (or one-way) relationship of entailment, from which we can deduce that all starlings have the property of being birds. Entailments must hold true in all possible worlds, and not just in a particular set of contexts. (We will explore context-dependent, or pragmatic, meaning in the next chapter.) One can possibly imagine a science-fiction novel being written in which, as a result perhaps of a bizarre radioactive accident, all starlings were green, or had four legs, but it is impossible to imagine starlings not being birds. In cases of entailment of this kind, we can say that starling is a hyponym of bird, and that bird is the superordinate term or hypernym of starling, robin, jackdaw, ostrich, penguin and so on.

In many cases, psycholinguistic evidence suggests that a superordinate term is associated in a speaker’s mind with a prototype, i.e. a typical member of the category in question. For the superordinate term bird, for example, English speakers are more likely to think of robins as being typical of the bird class than, say, ostriches or penguins (see Spotlight below).

Spotlight: Testing prototypicality

Radford et al. (2009: 181) offer the ingenious ‘technically’ or ‘strictly speaking’ test as a way of establishing prototypicality in such cases:

(a) Strictly speaking, a penguin is a bird.

(b) Strictly speaking, a robin is a bird.

(c) Technically, a whale is a mammal.

(d) Technically, a trout is a fish.

While all of the above sentences are grammatically well formed in English, there’s something slightly odd about (b) and (d), which seem to labour the obvious, because robin and trout are prototypical hyponyms of bird and fish respectively, while penguin is not a prototypical bird, nor whale a prototypical mammal.

Key idea: Specifying truth-conditions

Sense relations can be established by specifying the truth-conditions of well-formed sentences in which lexemes occur. In the case of entailment, an implicational relationship implies in that if X is true, then Y must be true also, but the reverse relationship does not hold (i.e. if Y is true, then X need not be).

Another kind of sense relationship is synonymy, which involves identity of lexical meaning. Semanticists would argue that total synonymy is rare, if indeed it occurs at all in language. Hide and conceal, for example, might appear to be synonyms, because of their substitutability in a wide range of contexts, e.g.:

1. Was Saddam hiding/concealing weapons of mass destruction?

2. He’s been hiding/concealing the truth for some time.

3. They hide/conceal their secrets very well.

4. Finally he found the stolen necklace, hidden/concealed in an old musical box.

5. But the interchangeability is not total: conceal can’t be used, for example, as an intransitive verb (the kids are hiding/\*concealing in the understairs cupboard), and no child ever asks to play conceal and seek.

6. Occasionally, two words with a technical meaning may be described as fully synonymous (tetanus and lockjaw, for example) but, even here, one form is likely to have different connotations from the other (lockjaw is a lower register, i.e. more informal, term than tetanus in this example), and it is a sign of efficiency within language systems that where two lexemes fulfil exactly the same role one will tend to oust the other. Perhaps for this reason lockjaw is an old-fashioned term these days, the medical term tetanus having largely prevailed in everyday usage.

7. Partial synonymy, on the other hand – as demonstrated by conceal and hide above, which overlap in many of their senses – is quite common: in some cases, different lexemes of similar or identical meaning are associated with different registers. While child might be preferred to kid except in informal situations, the more elevated term minor (or youth) might be appropriate in a formal or legal context.

8. <Image src="OPS/images/common7.jpg" alt="image" />

The only words for semantic relatedness in general use in our language are synonym (word of the same meaning) and antonym (word of opposite meaning). But even the very simple illustration I have given shows up the inadequacies of this terminology, particularly in regard to contrasts of meaning. The proportions above show that there is no one answer to the question: ‘What is the antonym of woman?’: girl and man are equally suitable candidates. The trouble is that the word ‘antonym’ encourages us to think that words contrast only on a single dimension; whereas in fact they may contrast with other words on a number of dimensions at once.(Leech 1974: 99)

ANTONYMS

Antonymy involves opposition of meaning, which can take a variety of forms. In the case of gradable antonyms, for example long and short, to affirm one member of the pair is to negate the other:

• X is short entails X is not long

• X is long entails X is not short

An important property of gradable antonyms is that negation of one does not entail the other, i.e. ‘not tall’ does not entail ‘short’: it is perfectly possible to be neither. They can be used in sentence frames of the ‘X is Y-er than Z’ type, and are distinguished from non-gradable antonyms or complementaries (e.g. alive/dead; true/false) in that, for the latter, negation of one does entail the other:

• X is not true entails X is false

• X is not dead entails X is alive

Another kind of antonymy involves what are known as relational opposites. If I give you something, then you receive it; if John is Paul’s teacher then Paul is John’s pupil, and so on. Finally, there is the antonymy of reversives, in which one form means not the negative of the other, but its reverse: examples here include enter/exit (or entrance/exit), remember/forget, tie/untie and so on.

Spotlight: Gradable antonyms

In the case of gradable antonyms, one member of the pair is generally perceived as the unmarked or default option in expressions of degree, e.g. ‘20 miles long’ not ‘20 miles short’. In 2010 the American comedian Joan Rivers was unimpressed by her host’s use of a stylistically marked choice to reveal her age, and responded with one of her own:

‘I met Vanessa Feltz and she said: “Here’s Joan, she’s 77 years young,” and I wanted to say “And here’s Vanessa Feltz and she’s 350 pounds thin.”’

HOMONYMS

The term homonym will be familiar in its common meaning – ‘word pronounced or spelled in the same way as another’ – but the term is used with greater precision by linguists, for whom only words with identical pronunciation, also known as homophones, count as homonyms. Homonyms may or may not be spelled identically: see as a verb meaning to apprehend by vision and see meaning the diocese of a bishop are both homonyms and homographs, while gate and gait are homonyms but not homographs.

A related concept here is heteronym, which refers to homographs which are pronounced differently, e.g. ‘bow’ in to bow politely and in he adjusted his bow tie. Homonymy needs to be distinguished from polysemy, which refers to a single word having multiple meanings, for example set meaning a group of things with something in common, to prepare as in to set a trap or a set as in a part of a tennis match.

In practice, separating homonymy and polysemy can be a challenge and the boundaries are not always clear. Should we, for example, regard the two uses of foot in he hurt his foot playing football and she found it at the foot of the bed as separate lexemes foot~1~ and foot~2~, i.e. homomyms, or as a single, polysemous word foot? It’s fairly clear that in this case, the criterion most lexicographers would invoke is relatedness of meaning: while foot~2~ does not denote a part of the body, it shares with foot~1~ the notion of being at the end of something, and it is indeed where one’s feet go when sleeping. For this reason, most dictionaries would regard foot~2~ as a secondary, but related, meaning of foot~1~.

A secondary criterion is etymology, i.e. a word’s origins and history, though it is important not to confuse synchronic and diachronic analysis because, as we saw in Chapter 3, a native speaker does not need to know the history of his/her language to speak it fluently. For example, the term right as the antonym of left (right~1~) and in its meaning of ‘correct’ or ‘proper’ (right~2~), is often viewed by lexicographers as an example of polysemy rather than homonymy, on the grounds that right-handedness and the right side used to be associated with moral virtue (e.g. in the expression seated at the right hand of the Father), in contrast to the negative connotations of the word sinister, which retains its historical meaning of ‘left’ in heraldry. The historical link argues for a polysemic interpretation, even if few people maintain such prejudices today. Relatedness of meaning generally trumps etymology in such judgements, however: pupil as ‘schoolchild’ and in the sense of ‘part of the eyeball’ are in fact historically related, but the meanings have now diverged to the point where no English speaker readily makes a connection between the two.

METONYMS AND MERONYMS

The example of foot~2~ above illustrates a particular kind of sense relation, in which a word associated with another is used to stand for it: this is termed metonymy, and the relationship here is between a part and its whole (cf. the _head_ of the company, or indeed the _head_ of a phrase). In other cases, the relationship is between a symbol and the institution, place or person it represents, for example the White House for the US President or Downing Street for the UK Prime Minister. Other relationships of metonymy might involve, for example, a container and its contents, as in he was overly fond of the bottle meaning ‘he was partial to the bottle’s alcoholic contents’. A metonym generally has a symbolic relationship with what it denotes (a head of department is not a literal ‘head’, obviously), but the term meronym refers to something which constitutes a part of something else, e.g. arm is a meronym of body.

Spotlight: Homonymy- and polysemy-based humour

Homonymy and polysemy have always been a rich source of humour. Jokes based on homonymy are known as ‘puns’, and English, with its wealth of homonyms, provides plenty of potential for humorous word play. Puns tend to elicit laughs or groans, but rarely a neutral response: some people like being awakened to sense relations in language while others do not. One of the leading exponents of pun-based humour is Milton Jones, whose work draws on surprising or unexpected connections between homophonous (or near-homophonous) words, or different senses of polysemous ones:

‘I phoned up the spiritual leader of Tibet, and he sent a large goat with a long neck. Turned out I’d phoned Dial-a-Llama’.

‘If they make it illegal to wear the veil at work, bee-keepers are going to be furious.’

‘The pollen count. That’s a difficult job.’

‘Incredible to think, isn’t it, that every single Scotsman started off as a Scotch egg.’

‘Years ago I used to supply filofaxes to the Mafia. I was involved in very organized crime.’

Semantic features

The terminology above provides a useful toolkit for the description of sense relations between lexemes, but does not amount to anything resembling a theory of semantics or to an understanding of how meaning is constructed at the lexical level. We have seen how sentences can be broken down into constituents, morphemes, and ultimately phonemes: might meaning, too, be analysed in terms of more basic semantic components? This is the principle behind an approach to semantics known as componential analysis, which starts from the assumption that meanings can be decomposed into bundles of binary semantic features, comparable to the distinctive features of phonology (see Chapter 5). For example, dog and puppy might be distinguished by their specification for the feature ±\[ADULT], dog being +\[ADULT] and puppy –\[ADULT]; similarly, the distinction between dog and bitch could be captured by a feature ±\[MALE] or ±\[FEMALE]. These features could be used to distinguish man/woman/boy/girl; father/mother; duck/drake/duckling and so on, while ±\[HUMAN] could be used to differentiate man, woman, grandmother from animals, all of which could be distinguished from non-living things, such as book, lamp, car, by ±\[ANIMATE].

A partial feature matrix based on these features is illustrated below. Note that +\[HUMAN] entails +\[ANIMATE], and only items marked +\[ANIMATE] can have a specification for +/-\[ADULT] or +/-\[MALE]. The symbol Ø indicates that a lexeme is unspecified for a particular feature.

Table 9.1: A partial semantic feature matrix

This approach has a number of theoretical attractions. Firstly, it offers a technical definition for many of the sense relations we explored earlier. A hyponym, for example, can be said to contain all the features of its hypernym, and some more besides: while person, for example, is +\[ANIMATE], +\[HUMAN], woman is additionally –\[MALE] or +\[FEMALE] according to the feature system employed. Features can also capture common relationships between sets of lexemes in the same way as distinctive features in phonology (as we saw in Chapter 5) capture the relationships between pairs of sounds. So, just as the members of the pairs /b/-/p/, /g/-/k/, /d/-/t/ differ in their specification for the feature ±\[VOICE], so dog-puppy, woman-girl, horse-foal, pig-piglet differ in their specification for the semantic feature ±\[ADULT], the former of each pair having a positive and the latter a negative value.

Semantic features can help refine our grammatical description, too. Consider the examples below:

1. \*Much coins

2. ?Two muds

3. The pelican read the newspaper.

4. The ungrammaticality of the first example could be explained by a requirement that the quantifier much can only collocate with nouns having a negative value for the feature ±\[COUNT]. Similarly, for example 2, nouns marked –\[COUNT] cannot normally be collocated with numerals, or with many. Where they are, as in this case, the hearer will try where possible to reinterpret the noun itself as +\[COUNT] and meaning ‘type of’ (hence ‘two good wines’, ‘my three favourite cheeses’). Finally, example 3 is perfectly grammatical according to the syntax of English, but pragmatically odd. The oddity could be explained by positing a specification for the verb to read that its subject will be marked +\[HUMAN]: this requirement could only be overridden in a fictional or hypothetical world in which pelicans can and do read newspapers.

5. Similarly, we could posit within a grammar of English a feature ±\[SOLID] associated with nouns such as timber, wood, paper, glass, and require of verbs like cut, sever, rip, knock, tap or drill that they collocate only with nouns with a positive value for this feature, thereby ruling out \_he severed the water_ or _\he knocked on the gas. Regularities between verbs such as kill, cultivate, incite, inspire might be explored using a meaning component ±\[CAUSE]: kill_, for example, has been analysed as +\[CAUSE] +\[COME ABOUT] –\[ALIVE]. Hopes were raised that, as in phonology, meaning might be reducible to a basic set of semantic features or meaning primes, applicable to all natural languages.

6. For all their initial promise, semantic features soon proved to be limited in their theoretical scope. Let us return again to nouns in our table above, which we noted were only partly specified. Puppy is +\[ANIMATE], –\[HUMAN], –\[ADULT], but this specification would do also for kitten, duckling, foal, cub, and the young of other animals. Likewise, the specification –\[ANIMATE], –\[HUMAN], for lamp would fit any inanimate object we cared to name. To distinguish, for example, puppy and kitten we have to posit a specific feature for each, perhaps ±\[CANINE] or ±\[FELINE], but these begin to look more like ad hoc creations for the relevant lexical sets (dog/mongrel/bitch/spaniel… and cat/tabby/moggy/tiger… respectively), rather than genuine semantic primes with explanatory power outside a restricted domain.

7. In similar vein, a partial feature analysis for the verb assassinate might be +\[CAUSE], +\[COME ABOUT] –\[ALIVE] as suggested above for kill, but we would have to specify that its object be marked +\[IMPORTANT PUBLIC FIGURE] or suchlike to capture, however approximately, the difference between the two verbs, which again looks a little contrived and is not obviously generalizable to other lexical items. More complex still would be a feature specification to distinguish the connotative meanings of sweat and perspiration, for example, or the distinction between the verbs demand, ask and request.

8. Extending the concept of semantic features beyond a promising but small number of putative semantic ‘primes’ leads to a proliferation of features applicable only to individual lexical items, and a feature set which is in fact almost as large as the lexicon itself. We are, in effect, doing little more than offering fancy formal definitions for the words involved. Semantics ultimately seems too culture-specific for a universal feature set to be applicable and semantic features which are important in one language may not be in another. As we saw in Chapter 6, Dyirbal has a semantically motivated noun class for ‘women, fire and dangerous things’, while in Navajo, a native American language spoken in Southern USA, verbal suffixes vary according to whether the noun object is + or –\[FLEXIBLE].

9. <Image src="OPS/images/common2.jpg" alt="image" />

10. Key idea: Componential analysis

11. Componential analysis breaks down lexemes into their meaning components, e.g. man as +\[ADULT] +\[HUMAN] +\[MALE].

Other types of meaning

Not all meaning conveys propositional content, i.e. information or an opinion about the state of the world. For example, when Kermit the Frog uttered the immortal words: ‘Good grief! The comedian’s a bear!’ his opening exclamation ‘Good grief!’ conveyed no meaning which can be expressed in terms of truth conditions, but rather an expressive (or affective) meaning, indicating his feelings about the event he is reporting.

Arguably, most if not all utterances carry an element of expressive meaning, which is not always easy to disentangle from the propositional meaning. ‘He’s running for President’, for example, looks like a statement, but uttered with a rising tone at the end and perhaps a stress on the final word, it might convey incredulity or disbelief on the part of the speaker (‘He’s running for PRESIDENT?!’).

Another kind of meaning is what many linguists call phatic communion, which encompasses those seemingly meaningless pleasantries which in many societies are important ways of signalling shared membership of a community. When someone asks ‘How are you?’ in most cases he/she is not looking for a detailed account of your current state of health: it’s simply a social ritual designed to show that you matter as a human being.

Key idea: Affective and social meaning

Some types of meaning do not lend themselves readily to analysis in terms of semantic components. These include expressive or affective meaning (a speaker’s feelings about what is said), and phatic communion (social rather than propositional meaning).

Fact-check

1. [1](answers.mdx#rfn9-1)  Entailments define the sense relationship of what?

1. Hyponymy

1. Metonymy

1. Synonymy

1. Antonymy

1. [2](answers.mdx#rfn9-2)  Which of the following pairs are not gradable antonyms?

1. High/low

1. Bright/dull

1. Male/female

1. Thin/thick

1. [3](answers.mdx#rfn9-3)  Which of the following is an antonym of man?

1. Woman

1. Boy

1. Neither

1. Both

1. [4](answers.mdx#rfn9-4)  Which of the following is a hyponym of fish?

1. Perch

1. Animal

1. Bird

1. Whale

1. [5](answers.mdx#rfn9-5)  For which of the following sense relationships is it true that ‘not X entails Y’?

1. Gradable antonyms

1. Synonyms

1. Complementaries

1. Reversives

1. [6](answers.mdx#rfn9-6)  Help and aid are what?

1. Antonyms

1. Total synonyms

1. Partial synonyms

1. Meronyms

1. [7](answers.mdx#rfn9-7)  Which of these is an example of metonymy?

1. The White House has vetoed the proposal

1. Time flies like an arrow

1. The long and winding road

1. He has an inflated opinion of himself

1. [8](answers.mdx#rfn9-8)  Which of these has the partial feature specification –\[ADULT] –\[MALE]?

1. Granddaughter

1. Daughter

1. Girl

1. All of the above

1. [9](answers.mdx#rfn9-9)  Which of the gradable antonyms in bold below is stylistically marked?

1. long/short

1. empty/full

1. old/young

1. high/low

1. [10](answers.mdx#rfn9-10)  Which of the following best illustrates phatic communion?

1. Good heavens above!

1. You must be joking!

1. Good morning!

1. Delighted to hear your news!

Dig deeper

P. Elbourne, Meaning: A Slim Guide to Semantics (Oxford University Press, 2011) is clear, accessible and illustrated with excellent examples and good humour.

J. Hurford, B. Heasley & M. Smith, Semantics: a Coursebook (2nd edition, Cambridge University Press, 2007 – previous edition as Hurford and Heasley, 1983), esp. Parts 1–3

G. Leech, Semantics (Pelican, 1974) esp. Chapters 1–4, 6 & 11

F. Palmer, Semantics (2nd edition, Cambridge University Press, 1981), esp. Chapters 1, 2, 4 & 5

J. Saeed, Semantics (3rd edition, Blackwell, 2009), esp. Chapters 1, 3, 4 & 9

Online sources

Wikipedia article on ‘Linguistic relativity’: http://en.wikipedia.org/wiki/Linguistic\_relativity

Colour terms

The question of colour terminology is analysed in depth in Berlin and Kay’s classic work Basic Color Terms (1969); excellent more recent studies include V. Loreto, A. Mukherjee & Francesca Tria (2012) ‘On the origin of the hierarchy of color names’, Proceedings of the National Academy of Sciences (PANS), 109(18), 6819–24. Available online: www.pnas.org/content/109/18/6819. See also M. Dowman (2007) ‘Explaining Color Term Typology With an Evolutionary Model’, Cognitive Science 31: 99–132. (Available online: www.lel.ed.ac.uk/\~mdowman/explaining-color-term-typology.pdf)