While we can be optimistic about the gains linguistics has had from comparative philology, it is not to be thought that philology provides all the answers, even within its own field. In the Indo-European family, for example, for a variety of reasons, Basque, Sumerian and Etruscan have no obvious place. And the relationship of Indo-European to any other of the world's great language families (for example, the North American Indian languages, or the Malayo-Polynesian family) is impossible to ascertain in the present stage of study, though attempts, some misguided, have been made. Nor is philology likely to make much progress in this field: with primitive cultures there are rarely any written records and hence no basis for historical reconstruction. It is highly probable that many languages of non-Indo-European families have already disappeared, leaving no trace, and there are many hundreds of tongues that remain unanalysed to date.
There are also some important limitations to the comparative method as such, which should make us wary of relying too uncritically upon it. It concerns itself overmuch with dead languages, and with letters, rather than sounds. Secondly, it is preoccupied with the superficial similarities existing between languages, as opposed to the underlying differences. For example, the method does not allow for independent changes arising within a language once it has left its parent, which might not affect the parent at all; and an attempt to read such new features into the structure of the parent language (as the method is bound to do) can only produce distortion. Thirdly, and more important, there is the charge that the theory embodies a fundamental inconsistency in comparative procedure. The method characterizes a language as, say, Indo-European, by pointing to certain linguistic changes that have occurred in the course of its subsequent history; but in doing this it ignores other changes that have also occurred, which may be equally characteristic. Nor is there any criterion or principle furnished by which we can explain which type of change is relevant for deducing a parent language, and which is not, and this is dissatisfying. English may be Germanic in one sense, but it is Romance in another, especially when we consider it from the point of view of vocabulary. Fourthly, the method assumes that as soon as two languages split off from a parent, they no longer influence each other formally - which is by no means necessarily true -witness the influence of English on a variety of languages. Fifthly, the method fails to consider a variability in the degree of precision attainable at various periods of reconstruction: the further back we go in history, the more time and space we allow in between language states, and thus the more unknown influencing factors. It is not possible to talk of Indo-European with the same degree of certainty as of Old English, but the sound-laws are poker-faced, and equate all ages in their formulas. And finally, there is the assumption that Proto-Indo-European was a single language which can be deduced from all the forms evidenced in daughter languages. It is rather more likely (in view of certain contradictory pieces of evidence in the reconstructions, and in view of what we know about the nature of language) that the parent language involved many dialects, not just one which has been miraculously preserved in extant languages. But to determine the dialectology of Proto-Indo-European would be a task to wither even the most ardent German philologist's spirit.
The twentieth century, as we know, brought a reaction to purely historical studies, and today the most valuable and alive aspect of comparative linguistics is the subject of dialectology (or linguistic geography), which studies variation in speech forms of a language, and thus deals in the state of contemporary languages emphasizing speech to the almost total exclusion of writing. It is at this point, then, that we can take up the trail of modern linguistics, beginning with the work of Ferdinand de Saussure, whose contribution to our subject remains outstanding.
We are fortunate in having any of Saussure's theoretical ideas to" read at all. The chief book we have under his name was, in a way, unpremeditated. The Course in General Linguistics (Cours de Linguistique Generale) was published in 1916, three years after his death. It is a collection and expansion of notes taken by Saussure's students during various lecture courses that he gave. Understandably, it is rather fragmentary in character, and in many places there are hints only of the theoretical position which subsequent exegesis has concluded Saussure must have held. There is also very little in the way of detailed illustration of his views. But its influence has been unparalleled in European linguistics since, and it had a major formative role to play in the shaping of linguistic thought in Europe over the thirty or so years which followed its publication. In particular, it moved the subject away from the nineteenth-century emphases in language study. In Saussure, we can see a clear reaction against many of the ideas raised in my preceding section: again and again he emphasizes the importance of seeing language as a living phenomenon (as against the historical view), of studying speech (as opposed to written texts), of analysing the underlying system of a language in order to demonstrate an integrated structure (in place of isolated phonetic tendencies and occasional grammatical comparisons), and of placing language firmly in its social milieu (as opposed to seeing it solely as a set of physical features). The tradition of study which has grown up around Saussure has been to extract various theoretical dichotomies from his work and to concentrate on the clarification of these. I shall follow this tradition, and look briefly at the more important of them.
In opposition to the totally historical view of language of the previous hundred years, Saussure emphasized the importance of seeing language from two distinct and largely exclusive points of view, which he called synchronic and diachronic. The distinction was one which comparative philologists had often confused, but for Saussure - and, subsequently for linguistics - it was essential. Synchronic linguistics sees language as a living whole, existing as a 'state' at a particular point in time (an etat de langue, as Saussure put it). We can imagine this state as the accumulation of all the linguistic activities that a language community (or some section of it) engages in during a specific period, e.g. the language of the present-day working-class in Manchester. In order to study this, linguists will collect samples within the stated period, describing them regardless of any historical considerations which might have influenced the state of the language up to that time. Once linguists have isolated a focus-point for synchronic description, the time factor becomes irrelevant - whatever changes may be taking place in their material while they are collecting it, they consider trivial. To consider historical material is to enter the domain of diachronic linguistics. This deals with the evolution of a language through time, as a continually changing medium - a never-ending succession of language states. Thus we may wish to study the change from Old English to Middle English, or the way in which Shakespeare's style changes from youth to maturity: both would be examples of diachronic study. Saussure drew the inter-relationship of the two dimensions in this way:
Here AB is the synchronic 'axis of simultaneities', CD is the diachronic 'axis of successions'. AB is a language state at an arbitrarily chosen point in time on the line CD (at X); CD is the historical path the language has travelled, and the route which it is going to continue travelling.
(Crystal D. Linguistics. Second ed. Penguin Book, 1990. – pp. 140-157).