Polysemy and meaning division
A number of the analyses presented so far in this chapter necessitate the associated claim that the word under analysis is polysemous (Greek ‘many meanings’), i.e. that it possesses several distinct senses (as discussed in 10.3, constructions, as well as words, can be polysemous, but we will not pursue this possibility here). To give just one out of several possible examples, the componential analysis of the English furniture terms in Table 5.1 can only be considered valid if certain additional senses of words like chair are first excluded from consideration. For example, as well as the meaning in which it refers to an item of furniture, chair may also mean ‘professorship’ and ‘head of a committee’, meanings to which features like [+ for sitting on] clearly do not apply. A similar point could be made about the description of the transfer verbs buy and give. These verbs show a constellation of ‘metaphorical’ uses like those in (27) which contradict the feature assignments in Table 5.3, since there is no price involved in (27a), and no change of possession in (27b):
These discrepancies are naturally explained by the contention that chair, buy and give have several distinct polysemous senses, and that the componential analysis does not apply to all of them.
QUESTION Do any other analyses in the preceding parts of this chapter implicitly require the postulation of polysemy? Which?
This example of the necessity to postulate polysemy is quite typical of semantic analysis. In fact, for many semanticists it is a basic requirement on semantic theory to show how many senses are polysemously associated with a single lexeme: if a lexeme is thought of as the union of a particular phonological form with a particular meaning or meanings, then it is clearly essential for the analysis to specify, for any given word, what it is for a word to have one meaning, and what it is to have several meanings. If a theory of semantics cannot do this, it will be open to the charge that its conception of one of its basic terms is intolerably vague. As Kilgarriff (1993: 379) puts it, ‘without identity conditions for word senses the concept remains hazardously ill-defined’.
But polysemy is not required simply for the purposes of technical linguistic theorizing. The informal description of meaning in ordinary language would also be impossible without the recognition of separate senses within the same word. Consider for example the French noun pièce. This has at least five separate senses, as illustrated in (28):
It would seem impossible to give any accurate definition of pièce that did not separate out these five meanings. This is because any definition which tried to cover all the meanings simultaneously would be excessively broad, and would apply to many referents for which pièce itself is not used. Virtually the only definition that will embrace the notions of a piece, a coin, a document, a play and a room is ‘thing’, but this definition will admit many referents to which pièce itself will not ordinarily apply, such as aircraft, stationery items and meals, to name only three out of the infinite number of possibilities. This excessive breadth disqualifies ‘thing’ as a possible definition of pièce, and imposes its division into a number of different senses, each of which can then receive a separate definition.
