Andrew Radford - Linguistics An Introduction [Second Edition]

doggie – [medium-sized, hair-covered, four-legged]

Of course, if this is the case, cats, sheep and various other creatures will satisfy the conditions in (184) and a child, confronted with such a creature, will refer to it as a doggie.

Another example is provided by ball. Here we might suppose that the semantic representation in the adult lexical entry is along the lines of (185): (185)

ball – [round object, used in games, … ]

By contrast, the child relying entirely on perceptually based features, and therefore not having access to [used in games], which concerns the function of balls, has (186):

(186)

ball – [round object]

We can immediately see why a child will use ball to refer to the moon on the basis of (186).

It will come as no surprise that we regard the above proposals as flawed in

certain respects. Most obviously, the reliance on definition-like constructs as providing word meanings has been examined in section 12 and, we believe, found to be wanting. To set the child off on the acquisition road with a construct not employed in the adult system, while not totally unintelligible, would require extensive justification. More importantly, overextension of children’s early nouns is a fairly short-lived and limited phenomenon. The majority of children’s concrete nouns are not overextended, and the truly remarkable aspect of the acquisition of words is the accuracy of children’s use. Of course, we tend not to notice appropriate usage, but the fallacy of building a theory of lexical development on a

194

words

minority of aberrant cases should be apparent. Finally, there is an alternative way of thinking about overextension, which in our view is more plausible.

Take the case of doggie. Small children with limited lexical resources may find themselves in situations where they wish to draw attention to, say, a sheep. They know that the creature in front of them is not a dog, but they lack a lexical item for referring to it; in these circumstances, they may resort to the strategy of finding the word in their lexicon which most nearly matches in meaning what they are looking for. The plausibility of this way of looking at things is increased if we consider the case of an adult confronted with a novel type of creature. Such an adult may well resort to something along the lines of, ‘there’s a sort of X over there’, where X is an item in this adult’s lexicon. In these circumstances, we would not conclude that the adult’s meaning of X was too general; rather, we would say that they were doing their best in the face of inadequate lexical resources. We remain uncertain about what the ‘matching’ required by this account might amount to, because we do not have an adequate theory of the semantic representations appearing in

lexical entries. However, this view does not require that the child’s semantic representation for doggie should be different to the adult representation, and this is consistent with the overwhelming accuracy in child usage to which we have drawn attention.

As a final issue in this section, we would like to sketch what may prove to be a more promising approach to some aspects of early lexical development. In the

previous section, we introduced the semantic relation of hyponymy and indicated how it defined taxonomies in certain areas of vocabulary. Part of a taxonomy appears in (187):

(187)

furniture

chair

table

bed

kitchen

armchair garden coffee dining occasional double single

camp

chair

chair

table

table

table

bed

bed

bed

In (187), we refer to the level occupied by furniture as the superordinate level (note that this is a slightly different use of ‘superordinate’ to that which was introduced in the previous section), the level occupied by chair, table and bed is the basic level, and the lowest level is the subordinate level.

Taxonomies such as the one in (187) are very interesting for a number of reasons. For example, it will not have escaped notice that the move down the taxonomy from the basic to the subordinate level is accompanied by the appearance of morphological complexity; kitchen chair, armchair, etc. are compounds (cf. a similar observation in connection with extensions to 144, p. 172). There is no logical reason why such complexity appears at this level. The hypothetical (even more partial) taxonomy in (188) categorises the world in exactly the same

Children and words

195

way as the relevant portion of (187), but here morphological complexity appears at a higher level:

(188)

furniture

sitting furniture

sleeping furniture

kitchen

lounge

garden

sitting

sitting

sitting

furniture

furniture

furniture

Now, of course, it may be entirely accidental that the level we are referring to as basic has the property of being the most specific level of categorisation which has morphologically simple labels, but what is intriguing is that this linguistic observation (which could be extended by considering further taxonomies in

English and other languages) is linked to a set of psycholinguistic observations (see section 14) and some rather interesting facts about lexical development. Here, we focus on the latter.

The question we raise is at what level do children ‘enter’ taxonomies such as (187)? The answer is very clear. Children acquire words such as chair, table and bed before they acquire furniture or any of the subordinate terms. Of course, the subordinate items include the basic-level morphemes as components, so this

observation ought not to be too surprising, but if, as some have supposed,

children’s early words are ‘too general’ in their meaning, we might expect superordinate terms to be early acquisitions. But this is not the case.

Now, consider the fact that a child confronted with a chair is, inevitably,

confronted with a piece of furniture (a child being given a carrot is also being given a vegetable, a child eating an apple is also eating a fruit, etc.). Why is it that in these circumstances children inexorably home in on the basic-level items?

Easy, you might say: this is because adults use basic level items in such situations and obviously the child must be exposed to an item in order to acquire it. This is true, but we can continue to ask why this should be, i.e. why do adults label basic-level categories rather than superordinate ones? To answer this question, we need to consider the ‘information’ which categories at different levels in taxonomies contain. To get a sense of what is involved here, we will ask you to conduct a short ‘thought experiment’.

Referring to (187), try to think of as many properties as you can which you reliably associate with the category of furniture – note that you are not being asked to come up with a definition of furniture. You should soon admit to being stumped by this request: there simply are not very many properties that all (or, indeed, most) items of furniture typically have in common. Next, do the same thing for the categories of chairs, tables and beds. You should do better – there are quite a few properties that are reliably associated with chairs (‘is used for sitting on’, ‘has a back’, ‘has a smallish flat part’, etc.). Furthermore, these properties reliably

196

words

distinguish chairs from tables and beds, which are other categories at the basic level. Finally, try the same process for the category of kitchen chairs. Naturally, as kitchen chairs are chairs, all your chair properties will carry over to kitchen chairs; it is, however, unlikely that you will be able to come up with very much that is new about kitchen chairs (beyond, ‘usually found in a kitchen’!) which distinguishes them from other varieties of chair.