Chapter 1

The Two Perspectives of Diachronic Linguistics

Whereas synchronic linguistics allows only one perspective, which is that of the language-users, and consequently only one method, diachronic linguistics has both a forward-looking perspective, following the course of time, and a backward-looking perspective, which takes the opposite direction (cf. p. [128]).

The first of these perspectives corresponds to the actual progression of events. This is the perspective necessarily adopted in writing any chapter of historical linguistics, to develop any point in the history of a language. The method is simply a matter of checking the available evidence. But in very many instances this way of going about diachronic linguistics is inadequate or inapplicable.

For in order to set out the history of a language in detail following the chronological sequence, we would have to be in possession of an infinite series of photographs of the language, taken moment by moment. But this condition is never fulfilled. Romance scholars, for example, who have the advantage of knowing Latin, which is the starting point for their investigations, and of having an impressive mass of documents extending over many centuries, constantly find great gaps in their evidence. So the forward-looking method, with its reliance on direct documentation, must be abandoned in favour of going in the opposite direction, proceeding retrospectively against the chronological sequence of events. Adopting this perspective involves selecting a given period and inquiring not what a given form subsequently became, but what earlier form was its parent.

[292]

Whereas the forward-looking method involves simple narration of events, based entirely on a critical assessment of evidence, the opposite approach requires a method of reconstruction, based upon comparison. We cannot establish the original form of just one sign in isolation. Two different forms of the same origin, like Latin pater (‘father’) and Sanskrit pitar-, or the stems of Latin ger-ō (‘I bear’) and ges-tus (‘borne’), straight away invite comparison and give a glimpse of the diachronic unity which links both to a prototype which can be reconstructed inductively. The more terms available for comparison, the more exact these inductions will be, and they will result – if sufficient evidence is available – in genuine reconstructions.

The same holds for languages as a whole. We can infer nothing from Basque: its isolation offers no possibility of comparison. But from a set of related languages, like Greek, Latin, Old Slavonic, etc., it has proved possible to identify by comparison the primitive common elements they contain, and so to reconstruct the basis of Proto-Indo-European, as it existed before becoming geographically differentiated. And what was achieved for the family as a whole has been repeated on a smaller scale – but by the same means – for each individual branch, wherever such a reconstruction was necessary and feasible. Whereas many Germanic languages are directly attested by documentary evidence, the common Germanic from which they came is known to us only indirectly through the retrospective method. The same method has been applied by linguists, with varying success, to establish the original unity of other families (cf. p. [263]).

[293]

The retrospective method, then, allows us access to the past history of a language back beyond the time of the earliest documents. Any forward-looking history of Latin could hardly begin at a point earlier than the third or fourth century B.C. But the reconstruction of Proto-Indo-European has enabled us to gain some idea of what must have happened in the period which extends from the time of that original linguistic community and the earliest surviving Latin documents. Only subsequently was it possible to outline a forward-looking sketch of those developments.

In this respect evolutionary linguistics is comparable to geology, another historical science. In geology, too, stable states are described (for example, the present state of the Léman basin) without reference to anything which preceded them; but its main concern is with events and changes which, linked in succession, constitute diachronic facts. In theory, one can imagine a forward-looking geology, but usually in practice the perspective has to be backward-looking. Before describing what happened in any given location, it is necessary to reconstruct the chain of events and find out what brought that particular part of the world to its present state.

It is not only the methods of the two perspectives which differ so strikingly. Even for didactic purposes it is not wise to use both simultaneously in the same survey. A study of sound changes will present a very different picture depending on which perspective is adopted. Looking forward, it is relevant to inquire what became of Classical Latin ě in French; and the answer is that this single sound split in the course of time and gave rise to several different sounds: Latin pędem (‘foot’)→French pye (written pied), vĕntum (‘wind’)→vā (written vent), lĕctum (‘bed’)→ li (written lit), nĕcāre (‘kill’)→ nwaye. (written noyer), etc. Looking backwards, on the other hand, if we inquire what sound in Latin lies behind the open ęof modern French, we find that this one sound is the end product of what were originally several different sounds: French tęr ‘earth’ (written terre)←Latin tĕrram, vęrž ‘rod’ (written verge)←vĭrgam, fę ‘fact’ (written fait)←factum, etc. The evolution of formative elements can similarly be presented in these two ways, and equally different pictures will result. All our earlier observations on the subject of analogical formations (p. [232] ff.) demonstrate the point a priori. Looking backwards, for example, in search of the origins of the French participial suffix -é, we go back to Latin -ātum. Latin-ātum is originally attached to Latin verbs in -are formed from nouns, mainly feminine nouns in -a (cf. Latin plantāre ‘to plant’: planta ‘twig, slip’; Greek timáō ‘I honour’: tim ‘honour’). But -ātum would not have existed if the Proto-Indo-European suffix -to- had not been alive and productive in its own right (cf. Greek klu-tó-s, Latin in-clu-tu-s, Sanskrit çru-ta-s, etc.). Latin -ātum also includes the formative element -m of the accusative singular (cf. p. [212]). Looking forward, on the other hand, if we ask in which French forms the primitive suffix -to- is to be found, we can draw up a list which includes not only the various suffixes of the past participle, whether productive or not (aimé (‘loved’)←Latin amātum, fini (‘finished’)←Latin fīnītum, clos (‘closed’)←Latin clausum for *claudtum), but also many others such as -u from Latin -ūtum (cornu (‘horned’)←cornu-tum), the learned suffix -tif from Latin -tīvum (fugitif (‘fugitive’)-fugitīvum, sensitif (‘sensitive’), négatif (‘negative’), etc.), and a number of words no longer analysable, like point (‘point’) from Latin punctum, dé (‘die’)←datum, and chétif (‘wretched’)←captīvum.

[294]