Andrew Radford - Linguistics An Introduction [Second Edition]

exercise 6).

Furthermore, the worry we are pursuing here also arises in connection with

our brief account of how a theory of semantic features might enable us to deal with hyponymy. We noted that additional features would appear in the meaning of snake when comparing it to the meaning of reptile, but we did not offer any clues as to what these features might be. Obviously, something like

[having the characteristics of a snake] would do the job, but this is hardly enlightening.

An analysis such as that in (169) can be seen as providing a definition of the meaning of ram with the features providing necessary and sufficient conditions for something being a ram. That is, if anything is a ram, then it is male, it is adult and it is ovine (the features are individually necessary), and if anything is male, adult and ovine, then it is a ram (the features are jointly sufficient). However, we have noted that some of the features emerging from this analysis (e.g. [ovine]) have unattractive properties. Of course, we are all familiar with the idea that dictionaries contain definitions of word meanings, so we shall close this section by looking briefly at familiar monolingual dictionary entries to see whether they provide any additional perspectives on the semantic components of lexical entries.

180

words

Dictionaries and prototypes

Consider a typical dictionary entry for octagon as in (170): (170)

octagon – a plane figure of eight sides and eight angles

This has all the characteristics of a definition, with the expression following the dash providing necessary and sufficient conditions for something being an octagon. We confirm this by noting that the entailments in (171) hold, indicating that the conditions are individually necessary:

(171) a.

‘X is an octagon’ entails ‘X is a plane figure’

b.

‘X is an octagon’ entails ‘X has eight sides’

c.

‘X is an octagon’ entails ‘X has eight angles’

Furthermore, (172) holds, showing that the conditions are jointly sufficient: (172)

‘X is a plane figure and X has eight sides and X has eight angles’ entails ‘X

is an octagon’

For the case of octagon, then, we can conclude that (170) provides a good definition and that it is plausible to regard the expressions which appear in the definition (eight, side, angle, etc.) as being unlike [ovine], in that they are conceptually more ‘primitive’ than the item they are being used to define.

It is no accident, perhaps, that octagon is an expression used in plane geometry, a branch of mathematics. When we move outside this highly formal and precise domain, we soon begin to encounter difficulties. Consider the example of spaniel in (173):

(173)

spaniel – a kind of dog, usually liver-and-white or black-and-white, with long pendent ears

An immediate observation on (173) is that the phrase introduced by usually does not even introduce a necessary condition: if spaniels are usually coloured in one of these ways, it presumably is the case that the occasional spaniel comes differently turned out. Such an occasional spaniel will be sufficient to falsify the entailment in (174):

(174)

‘X is a spaniel’ entails ‘X is liver-and-white or black-and-white’

If it is definitions we are after, we may as well remove this condition, leaving (175): (175)

spaniel – a kind of dog, with long pendent ears

It seems uncontroversial to say that if anything is a spaniel it is a dog, so being a dog looks like a good necessary condition for being a spaniel; what now of long pendent ears?

It is not inconceivable (indeed, it seems highly likely) that sometime in the history of spaniels there have been examples lacking the relevant attributes. This spaniel has short ears because it was born like this, or because its ears have been

Word meaning

181

bitten in a fight, or because its ears have been surgically shortened for cosmetic purposes. Such a spaniel remains a spaniel, thereby demonstrating that possessing long pendent ears is not a necessary condition for spanielhood. Accordingly, we must remove this condition from the definition, leaving us with (176):

(176)

spaniel – a kind of dog

But (176), consisting of a single necessary condition, does not approach sufficiency. If it were sufficient, (177) would hold:

(177)

‘X is a kind of dog’ entails ‘X is a spaniel’

Any whippet suffices to show that (177) does not obtain.

What we have found for spaniel is that there is at least one condition, that of being a dog, which counts as a necessary condition, and again without further argument here, it is usually possible to locate conditions which are individually necessary in this sense (see the relation of hyponymy discussed above); it is the provision of a set of conditions which are jointly sufficient which gives rise to the difficulties we have encountered.

So much for spaniels. The position we have arrived at is that whereas for some nouns dictionaries do indeed provide definitions, for others they do not, and this raises the question as to what the status of (173) is. In fact, the appearance of the word ‘usually’ is revealing, as it suggests that what (173) does is provide a description of a typical or normal spaniel, and this might lead us to wonder whether the semantic representations of at least some lexemes have similar characteristics.

We shall see in section 14 that there is a range of psycholinguistic evidence which suggests that lexical semantic representations are prototypical in that they supply descriptions of typical members of categories. For our purposes here, we can simply note that there is some linguistic evidence which points in the same direction. Consider the appropriateness of the adverbial expressions such as strictly speaking or technically in the following examples:

(178) a.

Strictly speaking, an ostrich 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

In our view, all of these sentences are true and syntactically well formed, but, whereas (178a, c) are entirely appropriate, there is something odd about (178b, d),

this oddness being signalled by the prefixed question marks. We can account for this oddness if we propose that the appropriate use of expressions like strictly speaking and technically is partly determined by prototypicality or ‘goodness’ of category membership. We have already noted that both ostrich and robin are

hyponyms of bird, but in the taxonomy (144), there is no indication that robins are somehow more representative of the class of birds than are ostriches. We are now suggesting that the taxonomic structure requires elaboration if it is to adequately represent the structure of the mental lexicon. For instance, we might suppose that

182

words

our lexical entry for BIRD, rather than containing a set of features which provide necessary and sufficient conditions for something being a bird, consists of a description (perhaps in the form of a set of features) of a prototypical bird. This description will approximate a description of a robin but not an ostrich, with the consequence that BIRD and ROBIN will be ‘closer’ to each other than will BIRD

and OSTRICH. Evidence suggesting that this is not entirely fanciful will be

introduced in section 14 (exercise 11).

Exercises

1.

In the text (pp. 170f.), we discuss the entailment properties of sen-

tences containing manage and fail. You are to extend that discussion in

two ways:

(a) determine the entailments of manage/fail sentences that them-

selves contain negation (e.g. Max didn’t manage to finish the book).

(b) consider additional verbs (V) that appear in the context Max