The functionalist approach
The notion of ‘drift’
Simplification and repair
Avoidance of mergers
Gradual and discrete change
Consideration of a language’s structure brings one automatically to the question of the function of elements in a language. One school of thought in linguistics maintains that language change serves the ultimate purpose of optimising a particular language. According to this view there is an implicit goal in language change; for this reason it is frequently termed teleological from Greek teleos ‘goal’. To consider the functionalist view one must first of all distinguish motivation and interpretation of language change. If language change is motivated then this presupposes that individual speakers (or at the very least the community as a whole) have an unconscious knowledge of what constitutes an optimising change. Going on this, speakers will on the whole avoid changes which either 1) involve redundancy or 2) lead to the loss of a distinction made in the language in question.
The functionalist view is, however, fraught with anomalies. To begin with, if all languages change in such a way that they are optimised, then this implies that to begin with, all languages were inefficient and that the end stage for all languages will be one of perfection after which there will be no language change at all — unless of course one assumes that there are changes which do not lead to optimisation and hence prevent the system from reaching full equilibrium. Another difficulty is defining what one is to regard as optimal in a language. The standard wisdom on this point is to consider optimisation as referring to communication; a second and equally valid standpoint considers optimisation as referring to elimination of redundancy which can be seen as the equivalent to system optimisation. Needless to say, one can adopt an intermediary position between these two stances.
Consider first of all the point of departure for a functionalist view of language change. To begin with one assumes that there is a natural tension between the tendency amongs humans to inertia (inactivity) and the need for communication. According to this, language systems should show a move towards a reduction in complexity, this being constrained by the necessity to guarantee communication. At first sight this would appear to be true. Take English as a typical example. Throughout its history there has been a tendency to simplify English grammar (loss of inflections, and decrease in verb morphology). Various scholars have studied this phenomenon; from Edward Sapir we have the term drift which is intended to describe the gradual and largely imperceptible direction of language change. For drift to work it must be the case that with each generation of speakers an (unconscious) awareness of the direction of the drift must be present, otherwise how would speakers know what direction to change their language in?
The notion of ‘drift’
The problem with the notion of drift is not so much showing that it exists but rather explaining why it occurs in some cases and does not in others. To take another example within the Germanic family: standard German only shows very slight simplification of the original complex morphology of Germanic. Why should English (and Swedish, Dutch, etc.) have changed so rapidly while German evinces practically no drift towards an analytic language type?
No doubt the reasons here must be sought in the use of language in society. While this is speculative it offers the only plausible account for differences in language drift between genetically related languages. Consider one of the basic tenets of sociolinguistics (originally expounded by M. A. K. Halliday) that language is an instrument for the transmission of social order. Now one might maintain that the strict grammar of German runs parallel to the desire for orderliness which is a prominent characteristic of German social structure. Given such a structure, it is not surprising that a society would strive to maintain its language (at least its standard variety) in a rigidly encoded form. This is obvious in the notion of ‘correctness’ which exists in German and in the continuous appeals to socially acceptable authorities such as the Duden works as the guardians of the proper form of the language.
Now in England social order is equally transmitted via language, but most of all in its pronunciation. The morphology of English is not a real indicator of social prestige. In addition, English is spoken in many parts of the world so that English speakers are more tolerant of different forms of their language than say Germans (after all, an Englishman cannot criticise a New Zealander for using forms not acceptable in England; these may well be permissible in that person’s variety of English).
Note that this account of the sociolinguistic positions of English and German does still not tell us why there is little morphology in English and more in German. It simply describes what has happened in each language as a result of the decline in morphology — or lack of this.
The notion of drift raises another interesting question, particularly in connection with the Indo-European language family. Reviewing the history of these languages one notes that the morphology of all these languages has changed from a more complex to a simpler type. The question is thus valid: is morphological change always simplification? The answer is probably ‘yes’ but that in certain cases new fused forms can lead to more complex structures arising again. This has happened in the development of French from Vulgar Latin where various verb forms (future, for example) have arisen from a fusion of a basic verb and a form of the verb to have (originally habere).
Simplification and repair
Language change in the sense of simplification is blocked by the need for communication. Changes which would seriously impair individuals understanding each other are avoided, or if they occur, are very often ‘repaired’. The classical example of this in the history of English is the borrowing of third person plural pronouns from Scandinavian in late Old English as the native forms had become homophonous in many varieties of the language at that time (i.e. /hi:/ was, at least for some areas, both he and they, the situation was remedied by borrowing Scandinavian forms in th- [θ] from the north of England).
Reparational changes are of interest in their own right. Their basic motivation is the maintainance of system consistency. This can be illustrated with a simple example: in the course of the development of Germanic a phonological change occurred known as rhotacism where /z/ < /s/ changed to /r/ under certain well-defined circumstances, see Verner’s Law above). In verb paradigms these changes were partially reversed (see the preterite forms of the German verb sein) so as to maintain the regular appearance of paradigms. This and similar retrospective changes would seem to offer evidence for the contention that speakers tend to favour language change which preserves overall regularity. Indeed in the situation just outlined there was an obvious conflict between phonetic motivation for the change of /z/ to /r/ and the desire for system conformity which became dominant later.
Avoidance of mergers and functional load
While the avoidance of mergers no doubt plays a role in blocking language change one should not over-emphasise its importance. Above all one should remember that it is not possible to predict language change on the basis of this principle. Thus to take the example of English again, many instances of language change have given rise to very considerable homophony, just consider the outcome of the merging of Middle English /ai/ and /a:/ to /ei/ as in tail, tale; sail, sale or the vocalisation of /r/ which lead to homophones like caught, court; horse, hoarse.
Communication which is the goal of language use; with examples such as those just quoted comprehension is not impaired. The context is in all cases enough to clarify the intention of the speaker.
The degree to which a distinction is required in a language is known as its functional load, a term which derives from the Prague school of phonology and which was later applied by such structuralist authors as Martinet. The smaller the functional load of a distinction, the easier it may be subject to merger or loss.
Gradual and discrete change
If language change is something which takes places slowly over at least one generation then it is obvious that it is gradual by nature. This poses a problem for the interpretation of certain attested types of language change. In some cases it is clear that the change is discrete by nature. For example the change from Old English niman to take cannot have had intermediary stages as can, for example, the shift of /u:/ to /au/ in house. The answer to the question posed by discrete change is that there must have been a period (of at least one generation) in which both the older and the newer form were valid alternatives and that, in the course of time, the newer variant became dominant, finally suppressing the older one entirely.