Words - статья на английском языке

We have to ask ourselves, then, quite objectively, what a word is, how it can be defined and even perhaps whether there are words in the spoken language. There have been three main approaches to this problem. The first is to see the word as a semantic unit, a unit of meaning; the second sees it as a phonetic or phonological unit, one that is marked, if not by 'spaces' or pauses, at least by some features of the sounds of the language; the third attempts to establish the word by a variety of linguistic procedures that are associated with the idea that the word is in some ways an isolable and indivisible unit.
Let us begin by looking at semantic definitions of the word. In fact these are not so much semantic as notional, and fail for all the reasons we have already discussed in dealing with form and meaning. The word is said to be a linguistic unit that has a single meaning. The difficulty, of course, is in deciding what is meant by a single meaning, for meanings are no more easily identified than words. It is easy enough to show in a variety of ways that we cannot define words in terms of units of meaning. To begin with, it is very clear that very many single words cover not one but two or more 'bits' of meaning. <...>
Nothing more than that single words do not appear to have single meanings, even if we could (which I doubt) establish single meanings. Conversely there are combinations of words in English that do not have separate meanings. Put up with, for instance, cannot be divided into the three meaning units of put, up and with, but seems to have the single meaning of 'tolerate'. There are dozens of combinations of this kind in English, mostly of two words — make up, take to, look for, put off, take in, etc. These are all idioms, of course, but the whole point about idioms is that they cannot be divided into separate units of meaning. <...> Indeed, any attempt to define the word in terms of a unit of meaning will be circular; we shall ultimately have to define 'unit of meaning' in terms of the word. <...>
What about the phonetic and phonological definitions of the word? In some languages it is indeed quite possible to recognize a word by some feature of its pronunciation. There are some languages, for instance, which have what is known as 'fixed stress': the stress always falls on one particular syllable of the word, not as in English or Russian where it falls on different syllables in different words. <...>
There are similar difficulties with any definition of the word as an 'indivisible unit'. It is sometimes suggested, for instance, that a word is a linguistic unit which does not permit the insertion of any other linguistic material. For instance, we know that little boy is two words because we can insert English — little English boy — but we cannot insert anything within, say, singing. But this too turns out to be false for we can divide singing into two parts sing-and-ing, and insert-ing and danc-to give us singing and dancing. Of course we did not insert a word or even whole words, but that is no objection. If we have not yet established what a word is we cannot require that the 'no insertion' criterion applies only to complete words! <...>
In conclusion, sadly, we have to say that word is not a clearly definable linguistic unit.
(From "Grammar" by Frank Palmer)