It is often said nowadays that we have no theory of historical linguistics, or that the existing theories are unworkable (though it should be added that some of such statements are prompted by genuine modesty and by a desire to disclaim the ability to provide a fresh theory). But to the present writer it has often seemed that the objections raised to previous theories are pseudo-objections, and that it is unduly pessimistic to say we have no theory at all. Admittedly, a theory that could predict details would be impossible, since the interpenetration of extra- and intralinguistic factors is largely a matter of chance. But it is possible to form a theory that is somewhat less ambitious – one that will, for example, make general predictions, indicate what kinds of evidence to look for, and suggest how it is to be interpreted. The present book results from an attempt to provide a theory of this latter kind. Since it uses previous theories as its starting point, it makes no claim to provide either a revolutionary clean sweep or a panacea; but the resulting principles will, it is hoped, give a fair reflection of the ‘why’ as well as of the ‘how’ of linguistic change.

In chapter 2, the examples used are for the most part familiar ones from the better-known European languages. It would have been possible to add others from African and American-Indian languages, the study of which has advanced so greatly since 1945; but this might have run the risk of obscurity for all but specialists in those particular languages.