Andrew Radford - Linguistics An Introduction [Second Edition]

X-ing is a sort of/type of Y-ing

The semantic relation of hyponymy must be distinguished from another seman-

tic relation which is illustrated by pairs such as those in (150): (150) a.

body, arm

b.

arm, elbow

c.

house, roof

d.

engine, carburettor

It is easy to see that the one-way entailment that we have seen to be characteristic of hyponymy does not obtain for cases such as these. Thus, neither (151a) nor (151b) obtains:

(151) a.

‘X is a body’ entails ‘X is an arm’

b.

‘X is an arm’ entails ‘X is a body’

174

words

The relationship between arm and body is one whereby the objects to which they refer are in a part–whole relation, and the term used for this relationship is meronymy. We also say that arm is a meronym of body and that arm, leg, etc.

are co-meronyms. As (150a, b) show, it is also possible to have meronymic structures with more than one level, as in (152):

(152)

body

arm

leg

elbow

wrist

knee

ankle

Note, however, that structures such as this are not to be confused with taxonomies –

as we move up such a structure, we encounter ‘larger’ entities, not more general categories (exercise 5).

While large sections of the vocabulary of a language can be analysed in terms of relations such as hyponymy and meronymy, such analysis is not always straightforward. For instance, consider the set of verbs in (153): (153)

think, believe, hope, wish, know, realise

These verbs (and the set could be extended) are known as propositional attitude verbs, i.e. they are all used to express something about the nature of the attitude of someone to a particular proposition, and the fact that they are labelled in this way indicates that they are perceived as having something in common semantically. However, there is no verb in English which qualifies as a superordinate for members of this class. In these circumstances, there is a lexical gap, and if we wished to represent the fact that the verbs in (153) do form a natural set, we could do so using (154), where φ indicates the position of the gap (exercise 6): (154)

ϕ

think

believe hope wish

know

realise

Synomymy or identity of meaning is a semantic relation with which most

readers will be familiar. However, it is doubtful whether there are lexemes in a language which can be regarded as completely identical in meaning. As a consequence, linguists often distinguish different types of synonymy, and among these cognitive synonymy can be defined in terms of entailment, as in (155), where S(L) means that L occurs in a sentential context S:

(155)

Lexemes L1 and L2 are cognitive synonyms if and only if S(L1) entails S(L2)

and S(L2) entails S(L1)

(Note that strictly speaking it is not lexemes which occur in specific contexts but the appropriate word forms.)

Word meaning

175

To illustrate, consider the pair of lexemes HORSE and STEED. These are

cognitive synonyms because if we consider a sentential context such as Sir

Lancelot rode a white . . ., both entailments in (156) obtain: (156) a.

‘Sir Lancelot rode a white horse’ entails ‘Sir Lancelot rode a white steed’

b.

‘Sir Lancelot rode a white steed’ entails ‘Sir Lancelot rode a white horse’

Why can we not simply drop the modifier ‘cognitive’ and say that these two

lexemes are synonyms? Because there are sentential contexts where their appearance, while not affecting the truth-value of the containing sentence, certainly affects its acceptability. For the case at hand, we feel that (157b), while sharing the truth-value of (157a), is rather odd (exercise 7):

(157) a.

The old lame horse gamely pulled the plough

b.

The old lame steed gamely pulled the plough

Obviously, synonymous lexemes exhibit considerable overlap of meaning.

Interestingly, the same is true of pairs of words opposite in meaning to which we now turn.

Meaning opposites

We have already noted properties of manage and fail which led us to

regard these items as ‘opposites’. Oppositeness of meaning is a pervasive semantic relation in the lexicons of human languages and it comes in several varieties.

Here we shall introduce two particularly important types.

Consider the pairs of dimensional adjectives in (158):

(158)

tall–short, high–low, wide–narrow, fat–thin, old–young, old–new

We can readily agree that each of these pairs illustrates oppositeness of meaning, but it is worthwhile to use our entailment relation to pursue the properties of such pairs in a little more detail. Thus, taking just tall and short (the other pairs behave identically), we have the entailments in (159) which make explicit that these are indeed semantic opposites:

(159) a.

‘X is tall’ entails ‘X is not short’

b.

‘X is short’ entails ‘X is not tall’

Now, we might expect that these entailments could be reversed, but this is not the case. The entailments in (160) do not obtain:

(160) a.

‘X is not short’ entails ‘X is tall’

b.

‘X is not tall’ entails ‘X is short’

The reason for this is easy to see. If we imagine all of those objects which can be described using tall and short, they fall into not two but three categories: there are tall things, there are short things and there are things in between which are neither

176

words

tall nor short (exercise 8). It follows that if X in (160a) designates one of these things, then ‘X is not short’ will be true, but ‘X is tall’ will not be true, i.e. the entailment does not hold. Pairs of opposites which behave like tall and short with respect to entailments are known as antonyms and they exhibit the semantic

relation of antonymy.

Remaining with adjectives, opposite pairs such as those in (161) behave rather differently to antonyms:

(161)

open–closed (of a store); married–single; dead–alive; broken–unbroken

Here we find analogous entailments to those in (159):

(162) a.

‘The store is open’ entails ‘The store is not closed’

b.

‘The store is closed’ entails ‘The store is not open’

For this case, however, the converse entailments do obtain:

(163) a.

‘The store is not closed’ entails ‘The store is open.’

b.

‘The store is not open’ entails ‘The store is closed’

This reflects the fact that for a store there is no state of being neither open nor closed but somewhere in between the two. Opposites like those in (161) are referred to as complementaries and the corresponding semantic relation is complementarity (exercise 9).

Semantic features

The semantic relations we have introduced above are clearly important

in suggesting that there may well be links of different kinds between lexical entries, i.e. the lexicon in a grammar is more than just a list of lexical entries.

However, we have not yet sought to look inside a lexical entry and see how

semantic information is represented there. We shall now see whether we can make any headway with this problem.

A proposal which many linguists have found attractive is that the meaning of a lexeme should be decomposable into a set of semantic features. The best way to illustrate what this involves is to immediately consider the triples of words in (164): (164) a.

ram, ewe, lamb

b.

bull, cow, calf

c.

stallion, mare, foal

In these triples, the first two words are opposites, and for concreteness we can regard them as complementaries. However, they are not merely opposites: for

each pair, it appears that the same fundamental distinction underlies their oppositeness. This is the distinction of gender, so we might propose a two-valued gender feature with values [male] and [female]. Such a feature can then function as part of the meaning of a word, and our intuition that ram differs in meaning